Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Terroir and reputation : the economics of British Columbia wine industry in three essays. Pankowska, Katarzyna 2017

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2017_november_pankowska_katarzyna.pdf [ 10.17MB ]
Metadata
JSON: 24-1.0357184.json
JSON-LD: 24-1.0357184-ld.json
RDF/XML (Pretty): 24-1.0357184-rdf.xml
RDF/JSON: 24-1.0357184-rdf.json
Turtle: 24-1.0357184-turtle.txt
N-Triples: 24-1.0357184-rdf-ntriples.txt
Original Record: 24-1.0357184-source.json
Full Text
24-1.0357184-fulltext.txt
Citation
24-1.0357184.ris

Full Text

  Terroir and Reputation: The Economics of British Columbia Wine Industry  in Three Essays.   by  Katarzyna Pankowska   MFRE, The University of British Columbia, 2011  B.Sc. in GRS, The University of British Columbia, 2010   A THESIS SUBMITTED IN PARTIAL FULFILMENT OF  THE REQUIREMENTS FOR THE DEGREE OF  DOCTOR OF PHILOSOPHY  in The Faculty of Graduate and Postdoctoral Studies (Integrated Studies in Land and Food Systems)   THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)     October 2017   © Katarzyna Pankowska, 2017  	 ii	Abstract   Terroir and collective reputations are two principal and interconnected elements believed to influence wine price and sales. In this dissertation, I examine the role of terroir (measurable features of the grape land) and collective reputation (eligibility for Vintners Quality Alliance, VQA) in determining the price, volume, and revenue of wine sales in British Columbia (BC). My research is highly relevant because this New World wine-producing region is currently altering its terroir-based geographical organization and sub-regional collective reputation, and plans to introduce new appellations and sub-appellations.  My first chapter provides an empirical overview of the BC wine industry including market structure, market shares, and regulations. My first analytical chapter on terroir consists of using hedonic regression to connect wine prices and terroir. By matching grape and wine production at a micro level, I examine how agronomic characteristics of grape land affect the price of wine due to variation in grape quality. In this analysis, I make an extensive use of a detailed dataset consisting of vineyards' terroir characteristics. In my second analytical chapter of collective reputation, I use a three-stage endogenous dummy variable regression model to identify the average effect of VQA status on the average volume share, the average revenue share, and the average price of wine.  I find somewhat limited evidence that vineyards' natural elements are important determinants of the price of BC wine. In my hedonic regression, the factors that seem to matter more are wine variety and brand. I also find that a relatively large number of wine brands represent VQA and that VQA certification positively influences the volume of sales for BC-made wines. My results also show that VQA certification has an insignificant impact on the average price and the average sales revenue of BC-made wines. Therefore, my results imply that VQA certification allows rent dissipation via over-certification. This over-certification allows arbitraging away of producers' rents.  	 iii	Lay Summary  In this dissertation, I research the BC wine industry. I analyze the influence of natural endowments of vineyards also known as terroir in the formation of wine prices for BC VQA wines. I also verify how VQA certification influences the average volume, average revenue, and average price for BC-made wines. The results of my research suggest that terroir has limited importance in the formation of wine prices for BC VQA wines. In my research, I also prove that while VQA certification has a positive influence on the average volume of wine sales, it shows an insignificant impact on the average price and average sales revenue. This dissertation contributes to wine economics, especially to literature that analyzes the influence of terroir and regional reputation on the formation of wine prices.                        	 iv	Preface  This dissertation is an original, unpublished intellectual product of the author, Katarzyna Pankowska.                                   	 v	Table of Contents Table of Contents Abstract ........................................................................................................................................... ii Lay Summary ................................................................................................................................ iii Preface ............................................................................................................................................ iv Table of Contents ........................................................................................................................... v List of Tables ................................................................................................................................ vii List of Figures ................................................................................................................................. x List of Equations ......................................................................................................................... xiv List of Abbreviations .................................................................................................................... xv Acknowledgements ...................................................................................................................... xvi Dedication .................................................................................................................................. xviii Chapter 1: Introduction ................................................................................................................ 1 1.1. Background ................................................................................................................... 1 1.2. Research Problem and Research Questions ..................................................................... 2 1.3.  Research Rationale .......................................................................................................... 3 1.4. Dissertation Outline and Content ..................................................................................... 4 Chapter 2: Overview of the Canadian and British Columbia Wine Industry ......................... 8 2.1. Introduction ...................................................................................................................... 8 2.2. Liquor Policies in Canada and British Columbia ........................................................... 12 2.3. Recent Developments in Wine Policies in BC ............................................................... 15 2.4. Selected Statistics for Wines Sold in the BC Market ..................................................... 22 2.5. BC VQA Wines and Brands in the BC Wine Market .................................................... 28 2.6. Conclusion ...................................................................................................................... 37 Chapter 3: Does Terroir Matter for BC-made Wines? ............................................................ 39 3.1. Introduction .................................................................................................................... 39 3.2. Literature Overview ....................................................................................................... 43 3.3. Data Sources and Construction of Variables .................................................................. 45 3.4. Methodology, Empirical Model Specification and Estimation Method ......................... 68 3.5. Empirical Results and Discussion .................................................................................. 82 3.6. Robustness Checks ......................................................................................................... 96 3.7. Conclusion ...................................................................................................................... 99 Chapter 4. Does VQA Certification Matter for BC-made Wines? ........................................ 102 4.1. Introduction .................................................................................................................. 102 4.2. Literature Overview ..................................................................................................... 110 4.3. Stylized Facts and Conceptual Framework .................................................................. 112 4.4. Methodology, Empirical Model Specification and Estimation Method ....................... 118 4.5.  Data Sources and Data Transformation ....................................................................... 127 4.6. Results and Discussion ................................................................................................. 133 4.6. Conclusion .................................................................................................................... 146 Chapter 5. Conclusion ............................................................................................................... 149 5.1. Research Aims .............................................................................................................. 149 5.2. Research Contributions ................................................................................................ 152 	 vi	5.3. Strengths and Limitations ............................................................................................. 154 5.4. Research Applications .................................................................................................. 155 Bibliography ............................................................................................................................... 157 Appendices .................................................................................................................................. 167          Appendix A: Chapter 2. .................................................................................................... 168          Appendix B: Chapter 3 ..................................................................................................... 187          Appendix C: Chapter 4. .................................................................................................... 226                 	 vii	List of Tables  Table 2.1. Canadian liquor commissions and boards, April 2017.……………………....13 Table 2.2. Old versus new BCLDB wholesale pricing formula, April 2017…………….17 Table 2.3. BC VQA versus BC non-VQA wines………………………………………...29 Table 3.1. Soil classes …………………………………………………………………...60 Table 3.2. Soils well-suited for grape cultivation.……………………………….............60 Table 3.3. Soils moderately well-suited for grape cultivation.…………………………..60 Table 3.4. Average elevation indicator variables ……………………………………….62 Table 3.5. Aspect indicator variables ……………………………………………………63 Table 3.6. Row direction indicator variables.……………………………………………65 Table 3.7. Distance to lake from vineyard indicator variables...………………………...67 Table 3.8. Data descriptive statistics.…………………………………………………….74 Table 3.9. Descriptive statistics continuation.…………………………………………...75 Table 3.10. Distribution of wines (SKU#) per origin of grapes used for their   production (total over the whole sample).……………………………………….76 Table 3.11. Level-level model. SE clustered on proposed   sub-appellations (15).……………………………………………………………84 Table 3.12. Log-level model. SE clustered on proposed   sub-appellations (15).……………………………………………………………85 Table 3.13. Grapevines/wines varieties significant estimates …………………………..88 Table 3.14. Brand significant estimates.…………………………………………………91 Table 3.15. Heat significant estimates.…………………………………………………..94 Table 4.1. Summary statistics -part 1.…………………………………………………..129 Table 4.2. Summary statistics -part 2.…………………………………………………..130 Table 4.3. Summary statistics -part 3.…………………………………………………..132 Table 4.4. Binomial probit estimation results.……………………………………….....134 Table 4.5. Regression A, 2SLS first and second stage results. Dependent   variable: logarithm of the average volume share……………………………….138 Table 4.6. Regression B, 2SLS first and second stage results. Dependent   variable: logarithm of the average price………………………………………..141 	 viii	Table 4.7. Regression C, 2SLS first and second stage results. Dependent   variable: logarithm of the average revenue share……………………………....144 Table A.1. The BC Wine Appellation Task Group recommendations and   plebiscite results ………………………………………………………………..172 Table A.2. Identified wine brands present in the BC market in  2011-2015………………………………………………………………………174 Table A.3. Identified BC VQA wine brands present in the BC market in   2011-2015.……………………………………………………………………...181 Table B.1. List of wineries that participated in the research presented in   chapter 3.………………………………………………………………..............190 Table B.2. Volume of sales (litres) for selected BC VQA wines per variety,   2011- 2015.………………………………………………………………….....198 Table B.3. Volume of sales (litres) for selected BC VQA wines, per winery,   2011- 2015.………………………………………………………………….....199 Table B.4. Level-level model. SE clustered on sub-appellations (15)..………………...214 Table B.5. Log-level model. SE clustered on sub-appellations (15)..……………….....218 Table B.6. Regressions results (full level-level and log-level models (6), with and   without inclusion of capacity dummy. Capacity dummy variable=1   if winery belonged to  the top 10 market players in BC in 2011-2015…………222 Table B.7. Regressions results (full level-level and log-level models (6), with and   without inclusion of capacity dummy. Capacity dummy variable=1   if winery belonged to  the top 5 market players in BC in 2011-2015…………..224 Table C.1. Average prices of BC grown grape varieties, year 2015.…………………..226 Table C.2. List of estate wineries……………………………………………………….228 Table C.3. Explanatory power of the binomial probit model…………………………..232 Table C.4. Binomial probit -full estimation results..…………………………………...233 Table C.5. 2SLS estimation results. Dependent variable: logarithm of the average   volume share..…………………………………………………………………..237 Table C.6. Table C.6. 2SLS estimation results. Dependent variable: logarithm   of the average price..……………………………………………………………242  	 ix	Table C.7. 2SLS estimation results. Dependent variable: logarithm of the   average revenue share...…………………….......................................................247 Table C.8. First-stage regression summary statistics. Dependent variable:   logarithm of the average volume share. ..............................................................252 Table C.9. First-stage regression summary statistics. Dependent variable:   logarithm of the average price..………………………………………………...253 Table C.10. First-stage regression summary statistics. Dependent variable:   logarithm of the average revenue share………………………………………...254                 	 x	List of Figures  Figure 2.1. Top grapevine varieties planted in BC, April 2017…………………...…….21 Figure 2.2. Grapevine acreage in BC, April 2017..………………………………….…..21 Figure 2.3. Total volume of wines sold in BC (Canadian and imports),  2011-2015 (in‘000L)……………………………………………………………..23 Figure 2.4. Total value of wines sold in BC (Canadian and imports),   2011-2015 (in ‘000 CAD$, real CAD$ 2015=base)……………………………..23 Figure 2.5. Percentage composition of BC wine sales per wine colour (total volume  of wine sales), 2011-2015………………………………………………………..24 Figure 2.6. Percentage composition of BC wine sales per wine colour (total value  of wine sales), 2011-2015.……………………………………………………….25 Figure 2.7. Classification and number of domestic wine brands sold in the province   of BC in 2011-2015……………………………………………………………...27 Figure 2.8. Volume of BC VQA versus BC non-VQA wines sold in BC in 2011-2015 ……………………………………………………………………….31 Figure 2.9. Value of BC VQA versus BC non-VQA wines sold in BC in  2011-2015.……………………………………………………………………….31 Figure 2.10. VQA brands present in the BC wine market in 2011-2015..……………….32 Figure 2.11. Volume market shares (%) of BC VQA wines in 2011-2015,  per brand category..………………………………………………………………33 Figure 2.12. Value market shares (%) of BC VQA wines in 2011-2015,  per brand category..……………………………………………………………...33 Figure 2.13. Share of the total volume of sales of BC VQA wines sold in BC in  2011- 2015, per top brands...…………………………………………………….34 Figure 2.14. Share of the total value of sales of BC VQA wines sold in BC in  2011- 2015, per top brands...…………………………………………………….34 Figure 2.15. The biggest players in the BC VQA market in 2011-2015 in  terms of the % share in the total volume of sales of BC VQA wines……………35 Figure 2.16. The biggest players in the BC VQA market in 2011-2015 in terms   of the % share in the total value of sales of BC VQA wines. …………………...35 	 xi	Figure 2.17. Herfindahl-Hirschman Index (HHI) for all BC made wines/brands……….36 Figure 3.1. Terroir versus wine pricing……………………………………………….....51 Figure 3.2. Levels of vineyard’s climate………………………………………………...53 Figure 3.3. The relationship between heat and grapevine growth ………………………56 Figure 3.4. Division of variables used in the empirical model based on type  of variable.……………………………………………………………………….70 Figure 3.5. Division of variables used in the empirical model based on their  variability over time.……………………………………………………………..71 Figure 3.6. Price vs grape variety, separated by winery/brand.………………………….77 Figure 3.7. Price vs winery/brand, separated by grape variety..…………………………78 Figure 3.8. Price vs alcohol content, separated by grape variety..……………………….78 Figure 3.9. Price vs distance to lake, separated by grape variety..………………………79 Figure 3.10. Price vs average elevation on vineyard, separated by grape variety….……80 Figure 4.1 Distribution of average prices for VQA wines (red and white),   2011-2015………………………………………………………………………113 Figure 4.2. Distribution of average prices for Non-VQA wines (red and white),   2011-2015………………………………………………………………………114 Figure 4.3. Mean volume of wine sales, 2011-2015……………………………………114 Figure 4.4. VQA price premium before industry expansion...…………………………116 Figure 4.5. VQA price premium after industry expansion...…………………………...117 Figure A.1. The BCLDB old versus new mark-up formula……………………………168 Figure A.2. Proposed demarcation of sub-appellations…………....…………………...170 Figure B.1. The map with locations of the estate wineries that cooperated on the  research presented in this chapter.……………………………………………...191 Figure B.2. The map with locations of the vineyards that sourced grapes of the BC  VQA wines analyzed in this chapter..…………………………………………..192 Figure B.3. Old and new BCLDB wine mark-up formulas ……………………………193 Figure B.4. Price vs vineyard’s aspect, separated by grape variety.……………………195 Figure B.5. Price vs row’s direction in the vineyards, separated by grape variety.…….195 Figure B.6. Price vs soil, separated by grape variety.…………………………………..196 Figure B.7. Price vs volume of wine sales, separated by grape variety.………………..196 	 xii	Figure B.8. Price vs volume of wine age, separated by grape variety.………………....197 Figure B.9. Histogram wine prices.………………………………………………….....200 Figure B.10. Histogram logarithmic transformation of wine prices.…………………...200 Figure B.11. Wine price versus sub-appellation, by soil type..…………………...........201 Figure B.12. Wine price versus sub-appellation, by row direction..…………………...201 Figure B.13. Wine price versus sub-appellation, by average elevation..……………….202 Figure B.14. Wine price versus sub-appellation, by distance to lake..…………………202 Figure B.15. Wine price versus sub-appellation, by variety..…………………..............203 Figure B.16. Wine price versus sub-appellation, by brand..…………………................203 Figure B.17. Wine price versus sub-appellation, by soil if winery belongs   to top 10 biggest producers……………………………………………………..204 Figure B.18. Wine price versus sub-appellation, by soil if winery   doesn’t belong to top 10 biggest producers..…………………...........................204 Figure B.19. Wine price versus sub-appellation, by row direction if winery   belongs to top 10 biggest producers.…...............................................................205 Figure B.20. Wine price versus sub-appellation, by row direction if winery   doesn’t belong to top 10 biggest producers..…………………..........................205 Figure B.21. Wine price versus sub-appellation, by average elevation if winery   belongs to top 10 biggest producers..…… ….....................................................206 Figure B.22. Wine price versus sub-appellation, by average elevation if   winery doesn’t belong to top 10 biggest producers……………………………206 Figure B.23. Wine price versus sub-appellation, by aspect if winery belongs to   top 10 biggest producers..……… …………………………………………..…207 Figure B.24. Wine price versus sub-appellation, by aspect if winery   doesn’t belong to top 10 biggest producers……….……………..……..............207 Figure B.25. Wine price versus sub-appellation, by distance to lake if winery   belongs to top 10 biggest producers…………………………………………….208 Figure B.26. Wine price versus sub-appellation, by distance to lake if winery   doesn’t belongs to top 10 biggest producers……………………………………208 Figure B.27. Wine price versus sub-appellation, by soil if winery belongs   to top 5 biggest producers………………………………………………………209 	 xiii	Figure B.28. Wine price versus sub-appellation, by soil if winery doesn’t belong   to top 5 biggest producers………………………………………………………209 Figure B.29. Wine price versus sub-appellation, by row direction if winery   belongs to 5 top biggest producers.…………………..........................................210 Figure B.30. Wine price versus sub-appellation, by row direction if winery   doesn’t belong to 5 top biggest producers.…………………..............................210 Figure B.31. Wine price versus sub-appellation, by average elevation if winery   belongs to top 5 biggest producers..………………….........................................211 Figure B.32. Wine price versus sub-appellation, by average elevation if winery   doesn’t belong to top 5 biggest producers ………..……………………………211 Figure B.33. Wine price versus sub-appellation, by aspect if winery belongs to   top 5 biggest producers..………………………………………………………..212 Figure B.34. Wine price versus sub-appellation, by aspect if winery doesn’t belong to   top 5 biggest producers ……….……………..……............................................212 Figure B.35. Wine price versus sub-appellation, by distance to lake if winery   belongs to top 5 biggest producers……………………………………………..213 Figure B.36. Wine price versus sub-appellation, by distance to lake if winery   doesn’t belong to top 5 biggest producers……………………………………...213 Figure C.1. Distribution of prices for red VQA wines, 2011-2015…………………….230 Figure C.2. Distribution of prices for red non-VQA wines, 2011-2015………………..230 Figure C.3. Distribution of prices for white VQA wines, 2011-2015………………….231 Figure C.4. Distribution of prices for white non-VQA wines, 2011-2015……………..231              	 xiv	List of Equations  Equation 3.1.………………………………………..........................................................71 Equation 4.1.………………………………………........................................................119 Equation 4.2.………………………………………........................................................121 Equation 4.3.………………………………………........................................................122                    	 xv	List of Abbreviations  AAFC    Agriculture and Agri-Food Canada BC    British Columbia BCLDB   British Columbia Liquor Distribution        Branch BCWI    British Columbia Wine Institute BCWA   British Columbia Wine Authority CR    Concentration Ratio CUSFTA   Canada-US Free Trade Agreement  GDD    Growing Degree Days GI    Geographic Indication GIS    Geographic Information System HSI    Heat Summation Index  IV    Instrumental Variable NAFTA   North American Free Trade Agreement OLS    Ordinary Least Squares PARC    Pacific Agri-Food Research Centre PPAR    Potential Photosynthetically Active Radiation RUE    Radiation Use Efficiency US    United States VQA    Vintners Quality Alliance              	 xvi	Acknowledgements  Many people in various ways assisted in the successful completion of this dissertation. It is not possible to list them all, but below you can find an attempt to do so.  The biggest thanks go to my doctoral thesis supervising committee, especially to Dr. Sumeet Gulati and Dr. James Vercammen.   This research would not be viable without funding support received from the Social Sciences and Humanities Research Council (SSHRC), the Canadian Dairy Commission, Douglas McRorie Memorial Scholarship (Agricultural Institute of Canada Foundation), as well as internal funding from the Faculty of Land and Food Systems (LFS). My sincere thanks go to all these funding entities, Canadian taxpayers, all LFS people that wrote the required reference letters, as well as to those that handled and processed my many scholarship applications: Shelley Small and Lia Maria Dragan.   My particular thanks go to Dr. Patricia Bowen from the Agriculture and Agri-Food Canada (AAFC/PARC Summerland) who furnished valuable scientific advice and necessary clarifications. I also thank the British Columbia Liquor Distribution Branch as well as all British Columbia winemakers that kindly agreed to cooperate on this dissertation, shared wine knowledge and provided essential data and suggestions.  Lastly, I would like to thank my family and friends who stayed with me when I was implementing this crazy research idea and were forgiving when I had to cancel my participation in social gatherings, just because I had to clean my pricing data or suddenly go to the Okanagan to verify row directions in vineyards. Therefore, my thanks are allocated to the following people: My husband, Artur Stanisz, who has consistently been supporting me and who has been patiently waiting for my Ph.D. finale and my "return to living," as he calls it; My sister, Ewelina Bednarek, and my brother-in-law, Wojciech Bednarek, for providing their home in Port Moody as a useful escape from my reality;  	 xvii	My friends and wine enthusiasts, Charles McArthur, Andrew Norden, and Iwona Michalak, for moral support, encouragement, and wine consumption when it was urgently needed.                	 xviii	Dedication 	I dedicate this dissertation to my grandmother, Zofia Sacha, a small-scale farmer from Sufczyn, Poland. I have never seen her growing vinifera, but I am sure that with her farming wisdom and skills she would have made an excellent grape grower. That strong and independent woman full of rare qualities always believed in me, offering support and encouragement for my scholarly endeavours. All those years spent on her farm not only shaped my personality and my future, but they were also the best time of my life. She proved to me that it is not required to have an extensive formal education to be wise, make a positive influence in the world and be remembered well. But it is necessary to have an open mind, unstoppable hunger for knowledge, strength to openly stand against evil, and courage to cultivate human decency regardless of circumstances. I sincerely doubt that I could get this type of schooling and wisdom from any school or university degree. So it is to you, Grandma, and I am sorry for not becoming a real doctor as you always suggested.      	 1	“Anyone who tries to make you believe that  he knows all about wines is obviously a fake.” [Leon Adams, The Commonsense Book of Wines]  Chapter 1: Introduction 	1.1. Background 	Wine is a complex commodity from production, marketing, and analytical perspectives. On the production side, its creation involves a multi-step process that starts at the agronomic level, where the interplay between a vineyard’s natural endowments, also known as terroir,1 and a winemaker’s specific management decisions jointly impact the quality of the final product. The importance of terroir in the winemaking process tends to be emphasized further via development of a collective reputation associated with a given wine region. The collective reputation, in turn, is built via the establishment of wine appellations and sub-appellations that divide wine regions into smaller, terroir-dependent sub-regional wine-producing units. On the marketing side, terroir and collective regional recognition, together with individual and brand-specific reputation, form the basis for wine marketing, with wine marketing strategies and sales built through the establishment of a wine’s esteem. The underlying concept behind terroir recognition and collective reputation implies that wines produced in different areas encompass distinct taste characteristics because their primary input, grapes, is sourced from vineyards with varying natural attributes. Also, what tends to be insinuated is that different wine regions possess different winemaking traditions and skills. All this implies that both region-specific terroir and winemaking skills influence wine taste and quality.   																																																								1 “Terroir” comes from a French word “terre,” meaning “land.” The term itself has various definitions. Some define terroir as natural endowments of the vineyard (soil, elevation, climate, etc.). Others also include elements like “experience” that wine-producing villages offer to wine tourists, idyllic landscape, specific architecture, history, local know-how, etc. (Gergaud & Ginsburgh, 2008).   	 2	However, it is possible that regional, terroir-dependent wine differentiation in the form of wine appellations or sub-appellations is not solely established to emphasize differences in terroir or winemaking know-how (and therefore in wine taste and quality) but to bring to wine an additional level of heterogeneity and enable other marketing avenues. Analyses that point out the marketing role of terroir and appellations have been previously pursued in the literature. They sought to investigate the influence of terroir specifics on prices of vineyards (Cross, Plantingan, and Stavins (2011)) or to find out if site attributes of vineyards influence prices of Bordeaux wines (Gergaud and Ginsburgh (2008)).  The role of terroir and collective reputation remains a compelling research topic in wine economics. This type of the investigation can be especially interesting and relevant in the case of young wine industries that are still growing and trying to establish long-term expansion paths. One such wine industry that is in the process of official legalization of new appellations and sub-appellations is in BC, and constitutes the main research topic in this dissertation. Of specific interest in this thesis are interactions between wine pricing and terroir, as well as the influence of collective reputation on wine pricing, volume share, and revenue share for BC’s locally sourced and made wines.  While wine production, marketing, and analytical complexity are likely to introduce systematic obstacles, they also bring an invitation to face the challenge. This invitation can be especially tempting when one analyzes the economics of terroir and collective reputation of a relatively small, unknown, sparsely researched, young and still developing wine industry like the one located in BC.  1.2. Research Problem and Research Questions  	The BC wine industry is a rare example among young wine industries in the world because of its strongly manifested attachment to the concept of terroir and weak grape-based wine industry specialization. This positions BC as a New World wine-producing region by the label and as an Old World wine-producing region by its love affair with terroir. The importance of terroir in the production of grapes is undeniable. But the actual 	 3	influence of terroir elements on the pricing of BC-sourced and made wines, as well as the impact of BC's appellation (VQA) on pricing, the volume of wine sales, and sales revenue, remain a bit enigmatic. Therefore, the primary goal of this dissertation is to find answers to these two research questions:  Research question 1: Does terroir influence the pricing of BC VQA wines from the Okanagan and Similkameen Valleys? (Chapter 3) Research question 2: What is the average impact of VQA certification on the average volume, average revenue and average price of wines produced by the estate wineries from the Okanagan and Similkameen Valleys of BC? (Chapter 4).  1.3.  Research Rationale  The fact that the BC wine region is currently in the process of policy changes and aims for the introduction of new appellations (four) and sub-appellations (16) constitutes the first research rationale for this dissertation. This new policy will probably be implemented by January 1, 2019.2 Therefore, it is interesting to analyze what the current relationships are between terroir, collective reputation, wine pricing, and wine sales to be able to envision what might happen in the BC wine industry when the new policy comes to life.   Also, the BC wine industry is very young and developing.3 While in the literature the relationships between wine pricing, terroir, and reputation in the Old World wine-producing regions are frequently analyzed; it seems to be less the case with young and relatively small wine regions. Therefore, it is likely that the research presented in this dissertation will be able to shed some light on dynamics between terroir, reputation, wine pricing, and wine sales in the world's youngest wine-producing regions.  																																																								2 In fact, one of these sub-appellations, the Golden Mile Bench sub-appellation, was established in 2015. For simplicity the total number of sub-appellations that will be officially set up by January 1, 2019, is used here (hence 16 instead of 15 sub-appellations). 3 The origins of the BC wine industry go as far back as the 19th century, but the modern BC wine industry started to develop in the late 1980s/early 1990s. Chapter 2, Subsection 2.1.1 below presents more information on this topic. 	 4	Finally, the research on the BC wine region is sparse. This fact brings an opportunity to find answers to some important terroir- and reputation-related questions that have not been addressed but might be of interest to academia, local policymakers and the BC wine industry.  1.4. Dissertation Outline and Content  This dissertation is divided into three separate but thematically interconnected chapters. In the next Chapter 2, I present an overview of the Canadian and BC wine industry. The primary goal of this chapter is to bring forward some important specifics of the BC wine industry that help set the stage for the empirical analyses of chapters three and four that follow. Therefore, in this chapter, I outline a short history of the BC wine industry highlighting the joint role of Canadian federal and provincial governments in setting wine policies in BC. Additionally, I describe the most significant past policies that helped shape and modernize the BC wine industry, such as "the great pull out" law which resulted in the government subsidized re-planting of Vitis labrusca grapevines with Vitis vinifera, which helped establish the BC wine industry in its current form. I also discuss the most recent wine policy developments in Canada and BC, which provides an understanding of the current wine policy climate at the national and provincial levels. This discussion includes the most recent markup formula change for liquor products that was officially brought to life in BC by the British Columbia Liquor Distribution Branch (BCLDB) on April 1, 2015.  I also outline the most current wine industry proposal and plebiscite (2016) that aim to introduce new wine appellations (four) and sub-appellations (16) in BC by January 1, 2019. Also, I outline the basic characteristics of the BC wine industry regarding grape acreage, grape varieties and a number of wineries in various sub-regions of BC. What these statistics point towards is that the BC wine industry is heterogeneous and not specialized in the production of any particular grape or wine type. Via the use of the BCLDB wholesale scanner wine sales data set for years 2011–2015 I show descriptive statistics for all wine sales (domestic and imports) in the entire province of BC. Also, I establish the number and division of domestically produced wine brands sold in the BC market between 2011 and 2015, and estimate brand shares (volume and 	 5	value) for BC-produced wines, with emphasis put on the BC VQA wines. My estimations of the BC VQA brand shares show that while in the BC wine market there are numerous VQA brands, five companies grasp about 59% of the volume and about 52% of the value of the VQA wines sold in BC in years 2011–2015. Also, my calculations of the Herfindahl-Hirschman Index (HHI) prove that in years 2011–2013 a moderate level of industry concentration characterized the BC wine market, but in years 2014–2015 the BC wine industry showed the HHI values characteristic for a competitive industry.  In Chapter 3, I study how terroir elements influence the pricing of selected BC VQA wines from the Okanagan and Similkameen Valleys of BC. The BCLDB scanner wholesale level data set maintains a basis for the analysis in this chapter. I use data on sales of selected BC VQA wines present in the BC market between 2011 and 2015. Then, I match each of these wines with self-collected micro level data (winery level data from 33 wineries) on exact locations of vineyards that sourced the grapes used to produce these VQA wines (71 different vineyards located within the Okanagan and Similkameen Valleys of BC). Additional information regarding natural elements (terroir) specific for each of these 71 vineyards like soil type, average elevation, row direction, vineyard aspect, distance to the lake, as well as temperature during the growing season enriches this data set. The terroir-specific variables come about because of actual verification of the location of these 71 vineyards using Google Earth satellite imagery or by physical visits on these plots. The climate variable is self-constructed using the Environment Canada (EC) temperature database. The combination of all these data sets (in the form of a panel data set with N=6785 observations on BC VQA wines) allows the inclusion of terroir elements that are unique for each of the vineyards that supplied grapes used in the production of selected BC VQA wines in the hedonic pricing modelling of this chapter. The results of my analysis show that terroir elements have somewhat limited importance in the formation of prices of the BC VQA wines, with soil, average elevation, row direction, and climate showing some significant results. What seems to be more important in the formation of wine prices for these wines are grape variety and the winery brand.  	 6	In Chapter 4, I estimate the impact of VQA certification on the average volume share, average prices, and average revenue share of wines sold in BC in years 2011–2015. The data set used in this analysis also comes from the BCLDB scanner wholesale level data set for years 2011–2015 and consists of all wholesale wine sales pursued by BC wineries located in the Okanagan and Similkameen Valleys that possess estate location. For this analysis, I transformed the available panel data set into a cross-sectional data set with N=3450 observations on different wines with monthly wine sales on a per wine basis for various VQA and non-VQA wines (straight average over each SKU). The modelling process in this chapter follows the three-stage procedure developed by Woolridge (2010, Subsection 21.4.1, page 937) and consists of an approach that corrects for the inclusion of an endogenous dummy variable. The VQA certification constitutes the endogenous dummy variable in my modelling setup that calls for a correction procedure. In the stage one of the endogenous dummy variable method, the binomial probit model, I use two types of indicator variables as my instruments for the VQA certification. They are winery age (four indicator variables) and a set of indicator variables for proposed sub-appellations, based on the estate winery location (15 indicator variables).  Other explanatory variables used in stage 1 of this procedure include all explanatory variables that are later used in stages 2 and 3 and come from my wine sales data set: wine colour, wine variety, reserve, sweetness, alcohol content, and a proxy control for winery capacity. I then use fitted values of the VQA certification obtained in stage 1 of this procedure as instruments in stages 2 and 3. In stages 2 and 3 I employ the Two Stage Least Squares (2SLS) method, to estimate a set of three different regressions, with three different dependent variables: the logarithm of the share of the average volume of wine sales, the logarithm of the average price and the logarithm of the average revenue share. The explanatory variables in the 2SLS modelling stage include wine colour, wine variety, reserve, sweetness, alcohol content, and a proxy control for winery's capacity. The fitted values of the VQA certification obtained in stage 1 of this procedure (binomial probit) are used here as an instrument for the VQA indication. The results obtained in Chapter 4 show that after controlling for the endogeneity of the VQA certification, there exists a positive influence of VQA certification on the share of the average volume of wine sales. 	 7	At the same time, the impact of VQA certification on the average price and the average sales revenue of BC-made wines remains insignificant.  Finally, in Chapter 5 of this dissertation, I summarize my findings, discuss limitations, and form recommendations for further research.                        	 8	Chapter 2: Overview of the Canadian and British Columbia Wine Industry  This chapter presents an overview of the Canadian and BC wine industry. Specifically, in Section 2.1 I give an introduction to a short history of the wine industry in Canada and BC. In Section 2.2 I outline the organization and governance of liquor-related policies at the national and provincial level. In Section 2.3 I present the most recent liquor and wine policy developments in BC. In Section 2.4 I outline important wine sales statistics for BC in years 2011–2015, for all wines sold in the province (produced domestically and imported). In Section 2.5 I define types of locally made wines, present classes of BC VQA wines sold in BC, outline wine sales statistics, show estimated brand shares for the BC VQA wines sold in BC in years 2011–2015, and verify the level of industry concentration. Finally, in Section 2.6 I present conclusions.  2.1. Introduction 	The fact that Canada domestically grows Vitis vinifera and produces various types of table wines may come to some as a surprise, yet it is true. The geographic location and common association of Canada with a cold climate, the relatively small size of the Canadian wine industry (especially in comparison to wine giants like France or the United States (US), for example) are the main reasons why the industry still lacks international exposure. Because of that, to the average wine consumer in the world, Canada still is not known as being able to grow vinifera and supply domestically made table wines. If anybody in the world happens to know that Canada produces wines, it is usually because of ice wines. Canadian ice wines remain the most frequently recognized in the world as being Canadian-made and associated with Canada (Canadian Vintners  Association Website statistics accessed on December 5, 2016: http://www.canadianvintners.com/info-centre/wine-statistics/).   	 9	Regardless of this still rather low level of world recognition for Canada-made wines, the Canadian wine industry shows a dynamic growth. The most recent, comprehensive wine industry economic study, "The economic impact of the wine and grape industry in Canada, 2015" prepared by A Frank, Rimerman + Co. LLP and published in March 2017, estimated the full economic impact of the wine and grape industry in Canada in the year 2015 to be about CAD 9.04 billion. The report shows that the full economic impact of the wine and grape industry in Ontario, the biggest wine-producing Canadian province, was at the level of about CAD 4.4 billion. In the same year, in BC the economic impact reached about CAD 2.8 billion (Frank, Rimerman + Co. LLP, 2017).  In comparison to total world wine production, the production volume of Canadian wines remains insignificant, and Canada is considered a small wine producer, with total production volume accounting for about 0.5% of all world wine production of about 28.2 billion litres. Regarding exports to the global marketplace, in 2015 Canada exported about 72.9 million litres of wine valued at about CAD 73.9 million, with premium wine (non-bulk) maintaining about 1.8 million litres of the total volume, valued at CAD 32.8 million. This was a significant increase in the volume of wine exports from past years, with a 237% increase between 2011 and 2015, as well as in the value of exports, which increased by 101% over the same period. Regardless of this growth in wine exports, Canada is still only ranked as the 27th biggest wine exporter in the world (by value of wine exports) (Canadian Vintners Association Website accessed on December 5, 2016: http://www.canadianvintners.com/info-centre/wine-statistics/).   Six Canadian provinces produce wine: Ontario, British Columbia, Quebec, Nova Scotia, New Brunswick and Prince Edward Island. Traditionally the highest volumes are generated in Ontario. British Columbia is Canada's second biggest wine-producing region. Currently, Canadian winemaking provinces show no specialization in any specific wine or grape variety. This lack of specialization and the practice of producing comparatively low volumes of numerous wine types make Canadian wine production 	 10	dispersed, negatively influencing international recognition of Canada as a wine-producing country.4   2.1.1. From plonk to Decanter’s platinum Pinot Noir5 	Although modest in size, the beginnings of Canadian winemaking can be traced as far back as the first half of the 19th century. The Canadian wine industry is a young one, and Canada belongs to the group of New World wine-producing countries. The birth of the wine industry in Canada is attributed to retired German corporal Johann Schiller who received land near Toronto and started to cultivate grapevines and sell wine to his neighbours. While the first vineyards in the province of BC were planted at the Oblate Mission of Father Charles Pandosy near Kelowna in 1860 (The Canadian Encyclopedia),6 the first winery in BC started to operate much later, in 19317 (BCWI, accessed on April 1, 2017: http://www.winebc.com/discover-bc/okanagan-valley).  Initially, BC cultivated grapevines belonged to the variety Vitis labrusca, a native species known as more suitable for BC and Canada due to its ability to withstand harsh winters. At that time, many agronomists doubted there was potential for Vitis vinifera cultivation resulting in labrusca as the primary grapevines species option for Canada in general and BC in particular. The lack of substantial domestic supply of vinifera, considered as superior for winemaking, was one of the reasons why the first BC-made wines were not 																																																								4 Canada remains a wine-producing country with no specialization in specific wines based on grape type like Malbec, for example, that is a crown grape/wine type associated with Mendoza in Argentina. The Canadian wine production approach is frequently called a “fruit salad” approach, where many different wineries produce many different wine types. The Canadian wine regions do not specialize in the cultivation of any particular grape variety. This fact negatively influences recognition of Canadian wines and the Canadian wine regions in export markets. Nevertheless, some BC wine industry members claim that such an approach is better for the BC wine industry in mitigating risks associated with the specialization that could negatively influence the survival rate of wineries in situations when there is a drop in prices for a particular wine type, for example. 5 Plonk is a derogatory name for wine of low quality, with high alcohol content. Such wine was produced in BC before the modernization of the wine industry in the early 1990s. The Platinum Decanter award was given to Mission Hill (a BC winery) for its Pinot Noir in 2013. Mission Hill’s Pinot Noir was considered the best in the world in 2013. The link below presents more information on this topic. Accessed on April 1, 2017: http://www.missionhillwinery.com/media/24718/Decanter_Trophy-PressRelease2013.pdf  6 The BC Wine Institute Website lists 1859 as the year of planting of first grapevines in BC. Accessed on April 1, 2017: http://www.winebc.com/wines/wine-101). 		 11	considered premium and superior quality. These wines had high alcohol content and usually constituted a component for port or sherry.  2.1.2.  Government intervention and “the great pull out”  The groundbreaking change in the BC wine industry came because of free trade agreements in the late 1980s and early 1990s, especially the Canada-US Free Trade Agreement (CUSFTA) negotiated in 1987 (ratified in 1989) and the North American Free Trade Agreement (NAFTA) that came to life in 1994. These trade agreements enforced and allowed industry modernization and development. At the time of ratification of these trade agreements, predictions for the future of the BC wine industry were very pessimistic. Many suggested that opening borders for trade with the US would end the BC wine industry, as it wouldn't be able to compete with premium quality wines from California. But, these gloomy predictions about the inevitable deterioration of the BC wine industry did not come to fruition. Instead, the BC provincial government came to the rescue, introducing the British Columbia Wine Act (Bill 58-1990) that reformed the whole BC wine industry. Among many rules brought in that act, probably the most important one was the fact that the BC government offered a subsidy to all grape growers in the province of BC that were willing to remove labrusca grapevines and replace them with vinifera plants superior for premium wine production. This government initiative is called in the BC wine industry as “the great pull out.” When the government introduced this support, a group of BC grape growers decided to accept the payment 8 and leave the industry. Others stayed and switched to vinifera cultivation. The grape growers that decided to stay received a subsidy of CAD 8,100 per acre for replacing labrusca with vinifera plants. “The great pull out” diminished the number of industry participants and made the BC wine industry more compact and profitable. This policy enforced changes in the sector size and ordered replanting of grapevines. There were also some other elements that influenced industry development: the introduction of the BC Wine Act and quality norms for BC-made wines (VQA) as well as the establishment of the BC Wine Institute 																																																								8 Those that decided to exit the industry also received payment. The government spent about CAD 27 million for this purpose (Hira, 2013). 	 12	(BCWI).  All these elements jointly modernized the BC wine industry and made it more suitable to compete with other modern wine industries in the world (Hira, 2013).  2.2. Liquor Policies in Canada and British Columbia  In this section, I outline details regarding Canadian liquor laws. While in Subsection 2.2.1 I discuss alcohol policies at the national level, in Subsection 2.2.2 I describe details about BC liquor laws at the provincial, BC level.  2.2.1. Liquor policies in Canada  Canadian liquor laws are multidimensional and complex. Because of the Canadian national organization, with federal and provincial governments that jointly govern in each province or territory, Canadian liquor policies are geographically heterogeneous. While each of the Canadian provinces or territories is left with autonomy for the organization of its internal management of liquor distribution and development of regionally specific alcohol policies, at the national level, federal laws bind all Canadian regions. Regarding jurisdiction that overlooks all alcohol related issues like liquor control, distribution and sales, each of the 13 Canadian provinces and territories have a liquor board or commission. The prerogatives of these liquor boards or commissions differ regionally, but they cooperate at the federal level in unifying the vision for the Canadian liquor status quo at the national level.  Table 2.1 below presents details on the 13 Canadian liquor commissions and boards. The joint mandate of all Canadian liquor boards and commissions is to:  1.  “Promote and encourage frank, open and ethical practices concerning the control, purchase and/or sale of alcoholic beverages; 2.  Co-operate with all provincial, territorial and federal agencies concerned with the control, sale and taxation of alcoholic beverages; 3. Improve the provinces’ and territories’ systems of control and distribution of alcoholic beverages by co-operation and free flow of information among the 	 13	members of the Association and by regular meetings or conferences of the members of the Association and comparable jurisdictions outside Canada” (Canadian Association of Liquor Jurisdictions website, accessed on April 15, 2017: http://www.calj.org/AboutUs.aspx).  Table 2.1. Canadian liquor commissions and boards, April 2017. Province/Territory Board/Commission Alberta Alberta Gaming and Liquor Commission (AGLC) British Columbia British Columbia Liquor Distribution Branch (BCLDB) Ontario Liquor Control Board of Ontario (LCBO) Manitoba Manitoba Liquor & Lotteries  Northwest Territories Northwest Territories Liquor Commission Newfoundland and Labrador Newfoundland and Labrador Liquor Corporation New Brunswick New Brunswick Liquor Corporation  Nova Scotia Nova Scotia Liquor Corporation Nunavut Nunavut Liquor Commission Prince Edward Island Prince Edward Island Liquor Control Commission Saskatchewan Saskatchewan Liquor and Gaming Authority  Québec Société des alcools du Québec Yukon Yukon Liquor Corporation    (Source: Canadian Association of Liquor Jurisdictions website accessed on April 15, 2017:  http://www.calj.org/CALJMembers.aspx)  The regional differences in Canadian liquor laws can be profound and significantly influence the level of board or commission's engagement in alcohol management within each province. Alberta, for example, has a fully privatized liquor industry and the sole role of the AGLC is to regulate the manufacture, importation, sale, purchase, possession, storage, transportation, and consumption of liquor in the province, oversee the industry and collect markup from alcohol sales. In Alberta, privately owned retail stores and licensed premises are in charge of all retail alcohol sales. British Columbia in turn, has a mixed private-public liquor distribution model with the BCLDB being a sole purchaser of alcohol within BC and from outside the province, by the federal Importation of 	 14	Intoxicating Liquors Act (BCLDB Website accessed on April 1, 2017: http://www.bcldb.com/about/who-we-are)  2.2.2. Liquor policies in British Columbia  All alcohol produced and sold in the province of BC must comply with numerous federal and BC specific policies. At the national level, the Importation of Intoxicating Liquors Act contains the primary liquor rules.9 Other federal laws concerning wine include the Canada Agricultural Products Act,10 Consumer Packaging and Labelling Act,11 Food and Drugs Act (Food and Drugs Regulations Part B-Alcoholic Beverages).12 As Section 2.2.1 above implies, the BC Liquor Distribution Act additionally enforces federal liquor laws. This act, together with the Importation of Intoxicating Liquors Act, outlines the BCLDB's mandate. The BC Liquor Distribution Act gives the BCLDB an exclusive right to purchase liquor for resale and reuse in the province of BC. This fact makes the BCLDB one of the biggest alcohol purchasers in the world. Besides being the sole buyer and reseller of all liquor in BC, the BCLDB also runs its own liquor stores, BC Liquor Stores (as of April 2017, there were 198 stores in the province).13 Therefore, the BCLDB also remains one of the biggest liquor retailers in BC. The BCLDB is responsible for reporting all liquor sales in the province. All alcohol producers and sellers, including all wine producers are required by law to report their direct sales information to the BCLDB.14    																																																								9Source: https://www.canlii.org/en/ca/laws/stat/rsc-1985-c-i-3/latest/rsc-1985-c-i-3.html	accessed on December 5, 2016. 10Source: http://laws-lois.justice.gc.ca/eng/acts/C-0.4/), accessed on 5, December 2016. 11Source: http://laws-lois.justice.gc.ca/eng/acts/C-38/index.html), accessed on 5, December 2016.	12Source:http://laws-lois.justice.gc.ca/eng/regulations/C.R.C.,_c._870/page-160.html#s-B.16.100), accessed on December 5, 2016. 13Source: http://m.bcliquorstores.com/m/stores), accessed on 5, December 2016. 14Since 2013, the Direct Sales Web-Reporting (DSWR) –Internet based reporting is being used. (BCLDB Website accessed on April 15, 2017: http://www.bcldb.com/about/who-we-are) 	 15	2.3. Recent Developments in Wine Policies in BC  In this section, I outline details regarding the developments in the BC liquor policies. While in Subsection 2.3.1 I describe the BC’s liquor markup formula that was in place before April 1, 2015, in Subsection 2.3.2 I discuss the most recent change in the BC’s liquor markup method. In Subsection 2.3.3 I outline details regarding the most current wine specific policy developments in the province of BC.  2.3.1. Liquor wholesale pricing in BC up to April 1, 2015 	Before April 2015 wine wholesale prices in BC were based on a formula where the retail price of wine, as seen in the government run liquor stores constituted a basis for the wholesale price formation. The official BCLDB markup method at that time included a 117% of markup on the first CAD 10.25 of the wholesale cost of wine plus 51% of markup on the remaining value. From that wine price, various retailers received different levels of discounts. In British Columbia, five classes of wine retailers were getting the following discounts off the government-run liquor stores’ retail price: 1. Independent wine stores: 30% discount off the LDB display price, 2. Private liquor stores: 16 % discount off the LDB display price, 3. Rural agency stores: 10 % discount off the LDB display price, 4. VQA wine stores: 30% discount off the LDB display price, 5. Restaurants and bars: 0% discount off the LDB display price. The BCLDB wholesale pricing calculators available to all wine suppliers in BC were used to establish wholesale prices for wines in BC. The wine vendors were using these official calculators to input their primary costs. The BCLDB calculator automatically applied the BCLDB markup formula to come up with a liquor store display price. This price before 2015 constituted a wholesale price in BC to which retailer-specific discounts were applied (as described above) (BCLDB website accessed on December 15, 2016: http://www.bcldb.com/files/BCWI%20presentation%20-%20Doing%20Business%20with%20the%20LDB%20-%2016Nov16.pdf).                      	 16	The BCLDB employs the theory of Social Reference Pricing (SRP). The SRP states that if prices of alcohol are set high, society consumes less alcohol. The SRP for BC wines is connected to minimum wine prices. Currently these minimum prices (including tax) are: • CAD 7.20/litre if wine size is <10 litres  •  CAD 6.45/litre if wine size is  >10 litres  (CALJ website accessed on July 24, 2017: http://calj.org/Articles/Publications/tabid/106/ArticleId/42/Minimum-Pricing-in-Canadian-Alcohol-Jurisdictions.aspx   The main issue associated with the official wine markup formula prior to 2015 was that on lower-priced products (that were a subject of the SRP policy, as per minimum price threshold presented above), the BCLDB official wholesale price was frequently higher than the price that would be offered by the producer. It is likely that these SRP price floors were binding mainly for wines from the category Cellared in Canada (CIC) (made from mixes of domestic and foreign wines or grape juice). These wines belong to the group of the least expensive BC-made wines (at retail), and they are likely candidates for the SRP price floors. The level of costs of production for BC-made wines can additionally support this statement. As mentioned above, the costs of production of other BC wines that are produced in BC and are made from 100% BC-grown grapes tend to be much higher. Lee Cartier, in his report on the BC wine industry, estimates that the average cost per litre of BC VQA wine was at the level of about CAD 5.91. At the same time, the average cost per litre of CIC wine was at the level of about CAD 3.20 (Cartier, 2013).  2.3.2. Liquor wholesale pricing in BC after April 1, 2015 	On April 1, 2015, a new liquor wholesale pricing formula was officially implemented in the province of BC. The aim of this new method for the calculation of wholesale liquor prices was to simplify the old way, which was based on the application of complicated and retailer-specific discounts. The new formula introduced a standard rate for all 	 17	commercial vendors in BC. Table 2.2 below outlines differences between the old and new pricing method.  Table 2.2. Old versus new BCLDB wholesale pricing formula, April 2017. Old BC wholesale prices formula New BC wholesale prices formula (April 1, 2015) Discount off display price Liquor Distribution Branch Price Less PST and GST15 =Retail Price Less applicable wholesale discount =Wholesale Price for that Customer Type, plus GST    Duty paid costs plus mark-up =Wholesale Price (tax excluded), plus GST     (Source: BCLDB Website accessed on April 1, 2017: http://www.bcldb.com/files/Wholesale_Pricing_Changes-Overview.pdf )  The detailed calculations involved in the formation of the new wholesale price markup formula are presented in detail in Figure A.1 in Appendix A: Chapter 2.  Together with the new wholesale price markup model, the BCLDB introduced some additional operating changes: 1. Eligible grocery stores were allowed to sell BC VQA wines16, 2. BCLDB-run liquor stores expanded hours of operation, 3. Refrigeration was introduced to the government-run liquor stores. There were no changes to the BC VQA program, and BC VQA wines are still exempt from the BCLDB markups. There were also no significant changes in pricing for the hospitality industry. As was the case with the old markup formula (before April 2015), under the new pricing formula, the BCLDB provides for BC wine vendors wholesale pricing calculators where wine suppliers input their primary costs and the BCLDB calculator automatically applies the BCLDB markup method. The new markup formula is based on the graduated markup 																																																								15PST (Provincial Sales Tax) and GST (Goods and Services Tax) are two charges present in BC.  16The BC VQA wines (BC Vintners Quality Alliance wines) are wines that are 100% BC-made. The full description of conditions that need to be met for VQA wines is outlined in detail in Subsection 2.5.1 below. This element of the policy change that concerns BC VQA wine sales in grocery stores like Save-On-Foods, for example, has already caused opposition, especially in the US. The US wine industry claims that the introduction of BC VQA wines in grocery stores in BC is not following ratified free trade agreements, especially NAFTA. While the BCLDB argues that its policy does not introduce preferential treatment for BC-made wines and the volumes of sales of these wines in grocery stores are minimal, the US wine industry claims the opposite. On January 18, 2017, the US government filed a formal complaint to the World Trade Organization. More details here in this link accessed on January 20, 2017: https://ustr.gov/about-us/policy-offices/press-office/press-releases/2017/january/Challenges_Canadian_Trade_Measures_That_Discriminate_Against_US_Wine.  	 18	calculated in the following way: supplier cost plus 89% wholesale level tax on the first CAD 11.75 wholesale cost + 27% markup on the remaining cost. The price of wine calculated in this way constitutes the wholesale price of wine. As of 2015, the wholesale pricing is the same for all retailers and retailer specific discounts are no longer in place  (as per BCLDB website accessed on December 15, 2016: http://www.bcldb.com/files/BCWI%20presentation%20-%20Doing%20Business%20with%20the%20LDB%20-%2016Nov16.pdf ). For more details, please refer to Figure A.1 in Appendix A: Chapter 2 and the text that follows under that figure.  2.3.3. The BC wine industry turning point  In November 2015, a significant development in the BC wine industry took place. The BC Wine Appellation Task Group constituted of members of the BC wine industry and coordinated by the BC Wine Institute board, in partnership with the BC Minister of Agriculture as well as the BC Wine Authority (BCWA), prepared a set of recommended changes to regulations of BC Wines of Marked Quality and delivered it to the BC Ministry of Agriculture. The recommendations from the BC Wine Appellation Task Group (Table A.1 Appendix A: Chapter 2 outlines these results in detail) were further reviewed and revised, resulting in final recommendations presented on April 28, 2016.  The BC wine industry plebiscite that took place between May 20 and July 1, 2016, followed the release of the revised version of recommendations. The results of this vote are also presented in Table A.1, in Appendix A: Chapter 2.   Participation in the plebiscite was not solely the prerogative of wineries that were current members of the BCWA, but instead, all producers of BC-made grape wines were asked to participate. All licensed wineries in BC could not only participate in this plebiscite, but they were encouraged to do so. About 71% of all licensed wineries operating in BC at that time (180 wineries out of 252 total) voted, making the results of the plebiscite valid and binding.   	 19	The official results of the plebiscite were delivered to the BC Ministry of Agriculture on July 8, 2016, with a request for rapid processing of these recommendations and introduction of proper policies. As of October 2017, the post-plebiscite recommendations have not been officially amended into binding legal rules and await the completion of the legislative process, but it is expected that this new policy will be formally introduced on January 1, 2019.   The most significant change coming from this wine industry proposal concerns the introduction of new appellations (four) and sub-appellations (16) (BCWI BC Wine Industry Plebiscite on Recommended Changes to the British Columbia Wines of Marked Quality Regulation as Proposed by the BC Wine Appellation Task Group, accessed on August 1, 2016: http://bcwinetaskgroup.ca/wp-content/uploads/2016/06/Plebiscite-Cover-Letter.pdf    When fully implemented these policies will likely open a new chapter in the history of the BC wine industry, making it more transparent as well as more oriented towards wine origin and wine quality. It is also likely that in the future the BC wine industry will put more emphasis on the varietal specialization of sub-regions (sub-appellations). But it remains to be seen what kind of influence the new policies will have on the future of the BC wine industry and how they will impact its numerous players.   While we must wait to see how these new policy developments will affect the wine industry in BC, several BC wine comparative statistics for the period 2011–2015 are presented and discussed in the next sections of this chapter, which will help to set a stage for the analyses in Chapters 3 and 4.      	 20	2.3.4. Status quo in the BC wine industry   Currently, six main wine-producing areas compose the BC wine industry (with five of them recognized officially as Geographic Indications (GI)), with a total of 299 wineries: 1.    The Okanagan Valley GI, with 172 licensed wineries, 2.    The Similkameen Valley GI, with 19 licensed wineries, 3.    The Fraser Valley GI, with 36 licensed wineries, 4.    The Vancouver Island GI, with 37 licensed wineries, 5.    The Gulf Islands GI, with 13 licensed wineries, Emerging regions (Lillooet (1), Kootenays (6), Shuswap (10), Thompson Valley (4), Northern BC-fruit winery (1)) account for 22 wineries17 (BC Wine Institute Website accessed on May 17, 2017: http://www.winebc.com/discover-bc ).18  Out of these six regions, the Okanagan Valley constitutes British Columbia’s biggest grape-growing area, with over 80% of the total grape acreage in the whole province coming from there.   Regarding grape varieties, red grape varieties constitute about 52% of all vines grown in BC; the remaining 48% are white varieties. Figure 2.1 below lists the top planted varieties in BC, together with their acreage. Overall, the most commonly grown grape varieties in BC include Pinot Gris and Merlot (BC Wine Institute Website accessed on May 17, 2017: http://www.winebc.com/wines/varietals).          																																																								17The number of wineries is somewhat flexible as they go out of business or merge with other wineries. This number of wineries was taken on May 17, 2017, from the BC Wine Institute website: http://www.winebc.com/discover-bc/okanagan-valley. 18Not all these wineries are grape wineries. Some are fruit wineries. The BC Wine Institute lists the number of all licensed facilities, including those that produce non-vinifera made wines. 	 21	Figure 2.1. Top grapevine varieties planted in BC, April 2017.   Regarding the BC’s total planted acreage of grapevine (per sub-region), as of April 2017, the following grape planting statistics hold:  Figure 2.2. Grapevine acreage in BC, April 2017.  Merlot (1600 acres) Shiraz (547 acres)Pinot Noir (949 acres) Cabernet Sauvignon (755 acres)Cabernet Franc (518 acres) Pinot Gris (1065 acres)Chardonnay (916 acres) Gewurztraminer (706 acres)Riesling (440 acres) Sauvignon Blanc (393 acres)Pinot Blanc (266 acres)Source: BC Wine Institute Website accessed on April 1, 2017: http://www.winebc.com/wines/varietals April 2017Top grapevine varieties planted in BCOkanagan Valley (8619 acres) Similkameen Valley (691 acres)Vancouver Island (432 acres) Gulf Islands (94 acres)Fraser Valley (200 acres) Kootenays (68 acres)Lilooet (54 acres) Shuswap (85 acres)Thompson Valley (95 acres)Source: BC Wine Institute Website accessed on April 1, 2017:http://www.winebc.com/wines/varietals April 2017Grapevine acreage in BC	 22	2.4. Selected Statistics for Wines Sold in the BC Market  In this section, I outline several descriptive statistics related to wholesale wine sales within the province of BC between years 2011–2015. In Subsection 2.4.1 I describe the volume and value of all wines traded in BC (domestic and imports), and in Subsection 2.4.2 I present the composition of wine sales regarding wine colour, and show the volume and value of sales for Canadian-made wines.  2.4.1. Total volume and value of wine sales in the BC market between 2011-2015   Wine sales in the BC market consist of sales of domestic (Canadian- and BC-made wines) and wine imports. Between years 2011–201519 the total volume of wines sold in the province of BC was at the level of about 300 million litres, with red, white and rosé wines maintaining respectively about 54.2%, 43.3%, and 2.5% of the total volume of wine sales in the province. The domestic wine supply in that period was mainly from BC and Ontario and maintained on average about 52% of the total volume of all wines sold in BC (about 156 million litres). The supply of imported wines (about 144 million litres or about 48% of the total volume of wines sold in the province of BC) came from about 20 different wine-producing countries (depending on the year). As the statistics show, there is an overall increasing trend in the volume of wine sales in BC. Figure 2.3 below presents the volume of sales for domestic and imported wines found in the province of BC in years 2011–2015.  The total value of wine sales in the BC market in 2011–2015 was at the level of about CAD 4.6 billion (real value, 2015=base year).20 Canadian wines captured on average about 46% of this total value of all wine sales in BC. Figure 2.4 below presents the value of all wine sales in BC in years 2011–2015. 																																																								19Data used for the analysis in this chapter comes from the BCLDB and consists of wholesale scanner data for April 1, 2011 to March 31, 2015. Canadian and BC wines include wines recognized in Canada as “Cellared in Canada (CIC).” As mentioned in section 2.3.1 above, these wines are made from a mix of Canadian and foreign grapes (or grape juice). 20The CPI deflator was calculated using Statistics Canada Table 326-001, http://www5.statcan.gc.ca/cansim/a26?id=3260021 accessed on January 15, 2016. 	 23	Figure 2.3. Total volume of wines sold in BC (Canadian and imports), 2011-2015 (in ‘000L).   Figure 2.4. Total value of wines sold in BC (Canadian and imports), 2011-2015 (in ‘000CAD$, real CAD$, 2015=base).21  																																																								21In Figure 2.4 the visible decrease in the total value of wines sold in BC in years 2014 and 2015 is associated with the change in the liquor markup formula (as discussed in Section 2.3 above). Total Canadian: 28656.6Total imported: 27811Total Canadian: 29373.1Total imported: 28624Total Canadian: 30675.5Total imported: 29167Total Canadian: 32788.7Total imported: 28404Total Canadian: 35242.1Total imported: 29913020,00040,00060,00080,0002011 2012 2013 2014 2015Source: Own calculations based on the BCLDB wholesale scanner sales data for 2011-2015.2011-2015Total volume of wines sold in BC (Canadian and imports, in ‘000L)Canadian wines Imported winesTotal Canadian: 440317Total imports: 527170Total Canadian: 449891Total imports: 551900Total Canadian: 468188Total imports: 575254Total Canadian: 364588Total imports: 405629Total Canadian: 393855Total imports: 433136020000040000060000080000010000002011 2012 2013 2014 2015Source: Own calculations based on the BCLDB wholesale scanner sales data for 2011-2015.2011-2015Total value of wines sold in BC (Canadian and imports, in ‘000CAD$, real CAD$, 2015=base)Canadian wines Imported wines	 24	2.4.2. The composition of wine sales, per wine colour, in the BC market in years 2011–2015 	The composition of all wine sales (domestic and imports) in the BC market in 2011–2015, per wine colour, is presented in Figure 2.5 below.  Figure 2.5. Percentage composition of BC wine sales per wine colour (total volume of wine sales), 2011-2015.   Between years 2011–2015 Canadian-made wines maintained on average about 43.7% of the total volume of red wine sales, about 62.5% of the total volume of white wine sales, and about 53% of the total amount of rosé wine sales in BC.  The total value of wine sales in the BC market between 2011–2015 was at the level of about CAD 4.6 billion (real value, 2015=base year).22 Canadian wines captured on average about 46% of this total value of all wine sales in BC.  																																																								22 Note: The CPI deflator was calculated using Statistics Canada Table 326-001, http://www5.statcan.gc.ca/cansim/a26?id=3260021 accessed on January 15, 2016. red 54.2%white 43.3%rose 2.5%red whiteroseSource: Own calculations based on the BCLDB wholesale scanner sales data for 2011-2015.2011-2015Percentage composition of the BC wine sales per wine colour (total volume of sales)	 25	Regarding the average percentage share in the value of wine sales per wine colour, the composition of wine sales in the province of BC between 2011 and 2015 is as presented in Figure 2.6 below.  Figure 2.6. Composition of BC wine sales per wine colour (total value of wine sales), 2011-2015.   Canadian wines took on average about 37.8% of the total value of red wines sales, about 57.6% of the total value of sales of white wines, and about 57.6% of the total value of sales of rosé wines.  Between 2011 and 2015 the total volume of Canada-made wines sold in BC23 was at the level of about 156 million litres (valued at about CAD 2.1 billion (real value, 2015=base year). About 99.8% of all Canadian wines sold in the province of BC in 2011–2015 were classified as wines bottled in the province of BC. The remaining 0.2% was classified as wines bottled elsewhere in Canada or outside Canada.  																																																								23 Canadian-made wines include those made in BC, as well as those made in other Canadian provinces. red 57.9%white 39.7%rose 2.4%red whiteroseSource: Own calculations based on the BCLDB wholesale scanner sales data for 2011-2015.2011-2015Percentage composition of the BC wine sales per wine colour (total value of sales, real (2015))	 26	Because the BC locally made wines maintain the primary interest of the analysis pursued in this chapter as well as in the next chapters, a closer look at the classification of these wines and their sales statistics follows in the next sections of this chapter.  2.4.3. Canadian wine brands and their significance in the BC wine market.  In the BCLDB wholesale scanner sales pricing data for years 2011–201524 all identified Canadian wine brands that were present in the BC wine market at that time were divided into six categories,25 as per Figure 2.7 below. Table A.2 in Appendix A: Chapter 2 contains all identified Canada-made wine brands found in the BC wine sales between 2011 and 2015.  Please note the following:  1. The BC Virtual brands are brands that stated that they possessed the estate location, but the brand’s website was listing a P.O. Box or a store in Vancouver as the place of the estate winery https://www.artisanwineshop.ca, for example.). 2. The Canadian, non-BC bottled brands include names from Ontario, for instance, that in the raw scanner data set were listed as “bottled elsewhere in Canada.” 3. Miscellaneous these are entries that didn’t contain a brand name or an estate winery but were reading: “Pinot Noir” or “Gewürztraminer,” for example. This category also contains all identified hospitality brands and private labels (e.g., Four Seasons, Sheraton, etc.).       																																																								24 This data includes all wine sales, so private labels (e.g., wines made specifically for restaurant or hotel and sold as a “house wine” for example) are also included here. In the volume and value sales statistics presented below, the sales from the fruit wineries are included. 25 The identification of these brands was pursued with the use of available sources, e.g., Internet search, visits to liquor stores, etc. 	 27	Figure 2.7. Classification and number of domestic wine brands sold in the province of BC in 2011-2015.26  Source: Own calculations based on the BCLDB wholesale scanner sales data for 2011-2015.  The available sales data set proves that wine brands and wineries in BC are on the constant move. While new brands and wineries enter the BC wine market, others disappear because they go out of business, swap hands or pursue strategic rebranding and change their names. This element added to the task of brand identification. Not all brands that were identified and listed in Table A.3 in Appendix A: Chapter 2 were present in the data set in all years 2011–2015. Also, there is a chance that there were some additional 																																																								26 All values are listed in Canadian dollar (CAD $), real 2015=base year. The CPI deflator was calculated using Statistics Canada Table 326-001: http://www5.statcan.gc.ca/cansim/a26?id=3260021 accessed on January 15, 2016. Canadian wine brands sold in BC:  376 brands Volume: about 156 million L Value: about 2.1 billion CAD$ Okanagan Valley estate wineries brands: 154 brands Volume: about 54.9 million L Value: about 1 billion CAD$    Similkameen Valley estate wineries brands: 15 brands Volume: about 0.358 million L Value: about 10.5 million CAD$   Virtual brands: 85 brands Volume: about 66.25 million L Value: about 688 million CAD$   Miscellaneous  Canadian not-BC bottled brands: 41 brands Volume: about 29.1 million L Value: about 298 million CAD$ Non-Okanagan estate wineries brands: 81 brands Volume: about 3.8 million L Value: about 76.6 million CAD$    	 28	brands that could not be identified and were put in the group “Miscellaneous” because they could not be assigned to any of the defined groups of names.  2.5. BC VQA Wines and Brands in the BC Wine Market  In this section, I present statistics related to BC-made wines and brands that were present in the BC market in years 2011–2015. In Subsection 2.5.1 I define classes of the BC-made wines. In Subsection 2.5.2 I give sales statistics for the BC-made wines (VQA and non-VQA). In Subsection 2.5.3 I present volume and value of sales statistics for BC VQA wines. Finally, in Subsection 2.5.4 I show the most significant BC VQA brands and the most important market players that contributed to sales of BC VQA wines in the province of BC in years 2011–2015. In this section I also calculate the industry concentration index (Herfindahl-Hirschman Index (HHI)).  2.5.1. Classification of BC made wines 	There are two main classes of table wines produced within the province of BC: BC VQA wines and BC non-VQA wines, also known as Wines of Distinction or BC Wines of Marked Quality.27  The VQA certification is considered in BC a wine appellation. Table 2.3 below presents definitions regarding what conditions must be met to achieve either wine status, BC VQA or BC Wine of Distinction.        																																																								27 Note: As mentioned earlier, in BC there is a third class of wines called Cellared in Canada (CIC) wines. They are made from mixes of domestic and foreign grapes, grape juice or wine. CIC wines are excluded from the analyses in Chapters 3 and 4. 	 29	 Table 2.3. BC VQA versus BC non-VQA wines. BC VQA Wines BC Wines of Distinction/ BC Wines of Marked Quality 28   1. Be a BC wine of distinction Be produced entirely from grapes of the varieties that meet the requirements of section 19: Grape varieties of 100% Vitis labrusca must not be used in BC wines of distinction; 2. Be made from one or more of the grape varieties listed in Table 1 or Table 2 of Schedule 5 of this regulation: BC Regulation 79/2005 (O.C.186/2005) not from any other grape varieties Be produced entirely from fresh grapes, grape juice and grape must derived from grapes grown in British Columbia; 3. Pass a taste test assessment, administered by the authority Be entirely fermented, processed, blended and finished in British Columbia 4. Meet the other requirements for certification as a BC VQA wine in accordance with this regulation:  • 100% British Columbia grapes • 95% of grapes must come from specific region mentioned on the label • 85% of grapes must come from the vintage stated on the label • 85% of grapes must be the stated varietal Be certified in accordance with this regulation  Be prepared on the premises of the practice standards certificate holder Source: http://www.bclaws.ca/civix/document/id/loo97/loo97/11_79_2005, accessed on December 5, 2016.  In other words, the main difference between BC VQA wines and BC Wines of Distinction (Wines of Marked Quality) comes from the fact that while BC VQA wines go through a panel of expert tastings before they obtain a right to VQA recognition, BC Wines of Distinction do not. Also, there are some additional requirements related to grape origin and VQA certification (please refer to Table 2.3, cell 4, above).   Even though the BC Wines of Distinction do not possess VQA recognition on their labels, they are still allowed to differentiate from other non-locally made (CIC) wines and indicate on their labels that they are a “Product of British Columbia.”   (BC Laws website: http://www.bclaws.ca/civix/document/id/loo97/loo97/11_79_2005, accessed on December 5, 2016).    																																																								28 As per: http://www.bclaws.ca/Recon/document/ID/freeside/11_79_2005 accessed on December 5, 2016. 	 30	2.5.2. Sales statistics for BC VQA wines for 2011-201529  For years 2011–2015 the total volume of BC VQA wines sold in the BC market was at the level of about 45 million litres, which maintained the total value of about CAD 981 million (real value, 2015=base year). Figure 2.8 below shows a yearly progression of the volume of sales of BC VQA versus BC non-VQA wines in years 2011–2015. There is an apparent increasing trend in the amount of sales of BC VQA as well as BC non-VQA wines in the province of BC.   Figure 2.9, below presents the value of sales for BC VQA versus BC non-VQA wines between 2011 and 2015. A noticeable element on this graph is a drop in the total value of sales starting in 2014. The new wholesale price model that began to be implemented before it was officially announced on April 1, 2015 (BCLDB phasing in of the new pricing model) caused this drop in the value of sales. As explained in Section 2.3 above, the new pricing model introduced by the BCLDB brought a unified wholesale price for all wholesale buyers and replaced the old pricing model that was built around the idea of inclusion of various, specific discounts for different classes of wholesale vendors.  Regarding wine colour, the BC VQA red, white, and rosé wines maintained respectively an average share of about 43.3%, 53.9%, and 2.8% of the total volume of all BC VQA wine sales in the province of BC.         																																																								29 These statistics concern only BC VQA wines that were bottled in BC. The data set also contains VQA brands bottled elsewhere in Canada and outside of Canada. Such wines were excluded from these statistics. 	 31	 Figure 2.8. Volume of BC VQA versus BC non-VQA wines sold in BC in 2011-2015.   Figure 2.9. Value of BC VQA versus BC Non-VQA wines sold in BC in 2011-2015.   Total VQA: 8246Total Non-VQA: 20367Total VQA: 8200Total Non-VQA: 21128Total VQA: 8437Total Non-VQA: 22171Total VQA: 9413Total Non-VQA: 23316Total VQA: 10685Total Non-VQA: 24492010,00020,00030,00040,0002011 2012 2013 2014 2015Source: Own calculations based on the BCLDB wholesale scanner sales data for 2011-2015.2011-2015Volume of BC VQA versus Non-VQA wines sold  in BC (in '000L)BC VQA wines BC Non-VQA winesTotal VQA: 204672Total Non-VQA:233362Total VQA: 205515Total Non-VQA:241189Total VQA: 207833Total Non-VQA:254625Total VQA: 172346Total Non-VQA:188851Total VQA: 191523Total Non-VQA:19824001000002000003000004000005000002011 2012 2013 2014 2015Source: Own calculations based on the BCLDB wholesale scanner sales data for 2011-2015.2011-2015Value of BC VQA versus Non-VQA wines sold in BC (in '000CAD$, real 2015=base year)BC VQA wines BC Non-VQA wines	 32	2.5.3. VQA brand categories present in the BC wine market in years 2011-2015  The available BCLDB wine data set for years 2011–2015 allows the identification of 221 brands that were supplying VQA wines. Not all of them were present in sales in the whole period of 2011–2015. Figure 2.10 below shows all identified groups of brands, together with a number of brands in each category   Figure 2.10.  VQA brands present in the BC wine market in 2011-2015.    The full list of VQA brands that were sold in the BC wine market between 2011 and 2015 can be seen in Table A.3, in Appendix A: Chapter 2. The VQA brands that were identified in the available data set grasped the following average market shares (volume and value), as per Figures 2.11 and 2.12 below.     1.357%4.525%17.19%19.46%57.47%Canadian ( 3 brands) Similkameen (10 brands)BC Non-Okanagan (38 brands) Virtual (43 brands)Okanagan (127 brands)Source: Own calculations based on the BCLDB wholesale scanner sales data for 2011-2015.2011-2015Percentage share of VQA brands present in the BC wine market (by brand origin)	 33	 Figure 2.11. Volume market shares (%) of BC VQA wines in 2011-2015, per brand category.   Figure 2.12. Value market shares (%) of BC VQA wines in 2011-2015, per brand category.   Canadian (not originally from BC) (0.03%)Okanagan Valley Brands (79.1%)Similkameen Valley Brands (0.4%)Non-Okanagan Brands (4.1%)Virtual Brands (15.2%)Miscellaneous (1.2%)Canadian (not originally from BC) Okanagan Valley BrandsSimilkameen Valley Brands Non-Okanagan BrandsVirtual Brands MiscellaneousSource:Own calculations based on the BCLDB wholesale scanner sales data for 2011-2015. 2011-2015Volume market shares (in %) of BC VQA wines per brand category.Canadian (not originally from BC) (0.12%)Okanagan Valley Brands (84.2%)Similkameen Valley Brands (0.5%)Non-Okanagan Brands (4.3%)Virtual Brands (10.2%)Miscellaneous (0.7%)Canadian (not originally from BC) Okanagan Valley BrandsSimilkameen Valley Brands Non-Okanagan BrandsVirtual Brands MiscellaneousSource: Own calculations based on the BCLDB wholesale scanner sales data for 2011-2015. 2011-2015Value market shares (in %) of BC VQA wines	 34	2.5.5. The most important BC VQA brands present in the market in 2011-2015  While there was 221 BC VQA (BC bottled) wine brands identified in the analyzed data set, not all of them had the same weight regarding the volume and value of sales.  Figures 2.13 and 2.14 below show the most significant BC VQA wine brands (based on market share (volume and value of sales).   Figure 2.13. Share of the total volume of sales of BC VQA wines sold in BC in 2011-2015,  per top brands.   Figure 2.14. Share of the total value of sales of BC VQA wines sold in BC in 2011-2015,  per top brands.  Mission HillJackson TriggsProspectGray MonkSumac RidgeQuailsgateInniskillinCedar CreekRed RoosterSee Ya LaterTinhornGehringer Mission Hill (8.79%) Jackson Triggs (7.82%)Prospect (5.72%) Gray Monk (5.43%)Sumac Ridge (4.37%) Quails'Gate (3.78%)Inniskillin (3.60%) Cedar Creek (2.74%)Red Rooster (2.56%) See Ya Later (2.46%)Tinhorn (2.33%) Gehringer (2.33%)Source: Own calculations based on the BCLDB wholesale scanner sales data for 2011-2015. 2011-2015. Total % share: 51.94%Share of the total volume of sales of BC VQA wines sold in BC, per top brands.Mission HillJackson TriggsProspectGray MonkSumac RidgeQuailsgateInniskillinCedar CreekRed RoosterSee Ya LaterTinhornGehringer Mission Hill (8.79%) Jackson Triggs (7.82%)Prospect (5.72%) Gray Monk (5.43%)Sumac Ridge (4.37%) Quails'Gate (3.78%)Inniskillin (3.60%) Cedar Creek (2.74%)Red Rooster (2.56%) See Ya Later (2.46%)Tinhorn (2.33%) Gehringer (2.33%)Source: Own calculations based on the BCLDB wholesale scanner sales data for 2011-2015. Total % share: 51.94%Share of the total volume of sales of BC VQA wines sold in BC, per top brands, 2011-2015	 35	Knowing that in the BC market individual wineries/companies own more than one brand, the largest suppliers of BC VQA wines in years 2011–2015 are presented in Figures 2.15 and 2.16 below:30  Figure 2.15. The biggest players in the BC VQA market in 2011-2015 in terms of the  % share in the total volume of sales of BC VQA wines.    Figure 2.16. The biggest players in the BC VQA market in 2011-2015 in terms of the  % share in the total value of sales of BC VQA wines.  																																																								30 Please keep in mind that there is a fraction of the unidentified entries in the available data set (Miscellaneous entries). Some of these entries might belong to listed companies, but they could not be identified. This element might influence final shares of the market and reshuffle the order, for example. Constellation Brands (24.04%)VMF Estates Kelowna(19.19%)Andrew Peller (6.73%)Gray Monk (5.43%)Quails' Gate (3.78%)Constellation Brands VMF Estates KelownaAndrew Peller Gray MonkQuails' GateSource: Own calculations based on the BCLDB wholesale scanner sales data for 2011-2015. CR 5 = 59.17%BC VQA wines, 2011-2015The biggest market players in BC in terms of % share in volume of wine salesConstellation Brands (18.41%)VMF Estates Kelowna (18.05%)Andrew Peller (6.08%)Gray Monk (4.95%)Quails' Gate (4.46%)Constellation Brands VMF Estates KelownaAndrew Peller Gray MonkQuails' GateSource: Own calculations based on the BCLDB wholesale scanner sales data for 2011-2015. CR 5 =  51.95%BC VQA wines 2011-2015 The biggest market players in BC in terms of %share in value of wine sales	 36	2.5.6. Herfindahl-Hirschman Index (HHI)  To assess the level of market concentration in the BC wine industry, the standard measure of market concentration, the Herfindahl-Hirschman Index (HHI) is calculated. The HHI maintains a typical measure for market concentration used by the US Department of Justice and sets guidelines for horizontal mergers. The index is calculated by squaring the market share of each firm competing in the market and then summing the resulting numbers (as per the US Justice website: https://www.justice.gov/atr/herfindahl-hirschman-index accessed on July 25, 2017).  The HHI for the BC wine industry was calculated on a yearly basis for years 2011–2015. BC’s total domestic wine production including VQA and non-VQA wines are treated in these calculations as the market based on which individual market shares for BC wine brands are calculated. Figure 2.17 below shows the results obtained from the calculations of the HHI.  Figure 2.17. Herfindahl-Hirschman Index (HHI) for all BC made wines/brands.  16851588 15681442136205001,0001,5002,000HHI2011 2012 2013 2014 2015Source: Own calculations based on the BCLDB sales data set for 2011-2015.2011-2015Herfindahl-Hirschman Index (HHI) for all BC made wines/brands.	 37	The standard definition of the HHI gives the following classification for market concentration: § HHI up to 1500 indicates a competitive industry, § HHI between 1500 and 2500 indicates industry moderately concentrated, § HHI above 2500 indicates industry highly concentrated.  (as per the US Justice website: https://www.justice.gov/atr/herfindahl-hirschman-index accessed on July 25, 2017). As the Figure 2.17 above shows, the BC wine industry was moderately concentrated in years 2011–2013, and in years 2014–2015 it reached the level of the HHI that suggests a competitive industry.  2.6. Conclusion  In this section, I present the summary of the analysis pursued in Chapter 2. Specifically, in Subsection 2.6.1 I outline conclusions and research limitations, and in Subsection 2.6.2 I form recommendations for further studies.  2.6.1. Conclusions 	The overview of the Canadian and BC wine industry pursued in this chapter leads to several conclusions.  First, Canada is a New World wine-producing country with a young wine industry located in six provinces, with Ontario and BC being the biggest wine suppliers of domestically produced wines. In the world market, Canada is still known from a comparatively small wine production and exports. Canada and BC are not associated with the cultivation of any specific grape variety and production of any particular wine type (except ice wines).  As my research in this chapter proved, the Canadian and BC liquor and wine industry are government-controlled at both national and provincial levels. The BC wine industry shows a high degree of heterogeneity that is visible at numerous levels. The industry 	 38	shows multiple classes of wine producers (e.g., estate wineries from different areas within BC, various BC virtual brands) and numerous categories of wine types (VQA, Non-VQA: BC Wine of Distinction, or Cellared in Canada).  The analysis pursued in this chapter also shows that the concentration ratio for the five biggest VQA suppliers, CR5, is at the level of 59% (volume) and 52% (value).  The results of the HHI show that overall for the years 2011–2013 the BC wine industry was moderately concentrated, but in years 2014–2015 its concentration decreased reaching the HHI levels that point towards a competitive industry.  The main research limitation of this chapter comes from the available scanner pricing data obtained from the BCLDB. The nature of this data set didn’t allow the identification of all sales entries. Therefore, it was necessary to construct the group called “Miscellaneous” that included the unidentified winery and brand entries (e.g., entries that did not point towards winery/brand but presented only wine variety: Gewürztraminer, Pinot Noir, etc.). This element could have affected the estimations of brand shares. Fortunately, this group of unidentified entries (about 1.2% of the total volume of BC VQA wine sales) was relatively small, so its effect on brand shares should be minimal.  2.6.2. Recommendations 	The obvious suggestion that arises from the analysis in Chapter 2 is to pursue more in-depth research on BC virtual brands (BC VQA and non-VQA virtual brands). This group of wines sold in the province of BC constitutes a significant volume of wine sales and has a substantial impact on the BC wine market. The virtual brands are usually hard to identify with a particular location for the winery (if any) or with an actual producer of such wine. Their labels tend to disclose very little information regarding the producer of such wine. As my research in this chapter unveiled, some of the most notable market players in the province of BC (like the VMF Estates Kelowna or Constellation Brands, for example) own some of these virtual brands. More details on this topic can be seen in Appendix A: Chapter 2, Table A.3.   	 39	Chapter 3: Does Terroir Matter for BC-made Wines?                    In this chapter, I pursue an empirical analysis of terroir versus wine pricing for the selected BC VQA wines produced in the Okanagan and Similkameen Valleys of BC. Specifically, in Section 3.1 I present an introduction to this chapter and research rationale. In Section 3.2 I discuss relevant literature. In Section 3.3 I examine data sources and outline methods for construction of variables. In Section 3.4 I present methodology and empirical model specification. In Section 3.5 I show regression results and discussion. In Section 3.6 I discuss robustness checks. Finally, in Section 3.7 I form conclusions, explore research limitations and develop recommendations for further studies.  3.1. Introduction  Not long ago, in a Decanter article from August 2016, Steven Spurrier, a British wine expert and merchant, said that for him wine was about 3 Ps: the place, the people and the product (Decanter, August 3, 2016). Spurrier made this statement about his very recent visit to the BC Wine Country, where he had a chance to taste regional wines and familiarize himself with the Okanagan’s approach to winemaking. For Steven Spurrier, as well as for many other wine professionals and enthusiasts that visit this Canadian wine region, the BC Wine Country is puzzling. The mystery lies in a clear juxtaposition of the classification of British Columbia as a New World31 wine-producing region, while its winemaking and vineyard management approach bears a striking resemblance to a winemaking philosophy that is characteristic of Old World wine-producing countries.   In the winemaking universe, two main directions are shaping the credo for wine production in each wine region. The first one concerns particular geographic location, a 																																																								31The New World wine-producing countries include countries that are located outside of the traditional wine-producing regions of France, Spain, Italy, Germany, Portugal, Hungary and the Middle East. The New World wine-producing countries include the US, Argentina, Chile, Australia, New Zealand, South Africa, Canada. 	 40	vineyard that produces grapes used in the winemaking process, and its terroir,32 as well as local winemaking traditions and craftsmanship. This connection frames the Old World wine-producing countries’ winemaking model. The winemaking giants like France, Spain or Italy, for example, tend to use this modus operandi.  The second approach puts more emphasis on grape variety, regional recognition based on specific wine and grape type, together with associated wine science and wine sophistication at its centre. This winemaking model helps create a flagship grape variety that generates the basis for a region’s winemaking recognition in the world. This second strategy distinguishes the New World wine-producing countries like Argentina, for example, with its principal grape variety, Malbec, Chile with its crown variety, Carménère, or Australia with its Shiraz.  While the classification of British Columbia as a New World wine-producing, region can’t be considered as a mistake because its wine industry is relatively young when compared to the Old World wine-producing regions, with a rather short track record of about 25 years or so in modern grapevines cultivation and winemaking, some doubts may quickly arise. A troubling element is that BC is still rather far from establishing a flagship grape variety, which is a defining approach for the New World wine-producing regions. Instead, to date, the BC winemaking industry does everything, except grape-based specialization. The BC winemakers produce many wine types coming from numerous grape varieties. The BC Wine Institute (BCWI) proudly states on its website that there are over 60 different grape varieties cultivated in BC (BC Wine Institute Website accessed on May 17, 2017: http://www.winebc.com/wines/varietals). Given the comparatively small acreage of the BC wine region, which adds up to a bit over 10 thousand acres of planted grapevines, it is quite an assortment of grape varieties. Again, this points towards a lack of wine industry specialization.  																																																								32“Terroir” comes from a French word “terre,” meaning, land. The term itself has various definitions. Some define terroir as natural endowments of the vineyard (soil, elevation, climate, etc.). Others also include elements like “experience” that wine-producing villages offer to wine tourists, idyllic landscape, specific architecture, history, local know-how, etc. (Gergaud & Ginsburgh, 2008). In this dissertation, terroir is understood as natural endowments of the vineyard (soil, elevation, climate, aspect, etc.). 	 41	What the BC wine industry tends to do instead of the conventional New World wine region approach of grape variety specialization, is to emphasize the role of a vineyard, especially its terroir. Or at least this is what the BC industry typically targets in its marketing campaigns. This approach is also visible in the latest developments on the policy side, including the most recent wine industry plebiscite (May 20–June 1, 2016). In this plebiscite, one of the main matters under industry voting was the establishment of new appellations (four new appellations proposed) and sub-appellations (16 new sub-appellations proposed for the Okanagan Valley).33 The proposal for the establishment of new appellations and sub-appellations that were confirmed in the plebiscite, with 98% approval rate for new designations and 64% approval rate for sub-appellations sets British Columbia as an outlier among New World wine-producing regions. Instead of concentrating on variety specialization, the BC wine industry seems to be choosing the terroir-based winemaking road established by France and Italy.   (BCWA Website accessed on August 1, 2016: http://bcwinetaskgroup.ca/wp-content/uploads/2016/06/Plebiscite-Cover-Letter.pdf).  What is interesting about these new policy developments is a current situation in wine production in BC, with a movement to continue the status quo that contradicts the newest appellations-related efforts. Currently, many BC wineries produce wines from grapes that do not necessarily come from the same location as their estate wineries. This means that their grapes and, therefore, wines come from different “sub-appellations,” even if these sub-appellations currently do not have an official demarcation or names. Because of the lack of strict policies and associated controls regarding wine labelling that coincides with the actual origin of grapes used in the winemaking process, a winery from the Okanagan Valley, for example, can produce wines from grapes coming from different locations within the whole of BC.34 A notable exception concerns the BC VQA wines, where 95% 																																																								33Appellations of origin are country-region specific. In France, for example, there is Appellation d’Origin Controllee (AOC), in the US, there are American Viticultural Areas (AVA) in British Columbia Canada, there is Vintners Quality Assurance (VQA). Appellations (and sub-appellations) of origin allow geographical identification for wines and prevent producers from beyond the appellation to make false claims about the origin of their wines. They also aim to distinguish different terroirs. 34The official wine policy states that BC-made wines must be produced from BC-grown grapes, but this is as far as it goes regarding terroir-related specification of grape origin for BC-made wine. The exceptions are BC VQA wines. 	 42	of grapes used in wine production must come from the location stated on the label. For example, if a winemaker says on the wine label that it is a VQA Okanagan Valley wine, then 95% of grapes used for its production must come from the Okanagan Valley. But even in the case of BC VQA wines and their stricter definition of grapes’ origin, the idea of terroir in BC is still somewhat diluted. The issue comes from the simplest element, the definition of terroir. Choosing its most straightforward specification, terroir is defined as the natural endowments of a vineyard: soil, climate, aspect, elevation, etc. But using this definition in the Okanagan Valley, BC’s biggest wine-producing region raises some serious doubts. The Okanagan Valley stretches about 155 km North to South and is about 9–16 km wide (Hira, 2013). The climate, as well as soil specifics and other natural elements, differ in various locations along this 155 km stretch. Consequently, there must be differences in terroir as one moves from one vineyard to the other within the Okanagan Valley. Therefore, even in the case of BC VQA wines that state that they come from the Okanagan Valley and must be produced with 95% of grapes coming from that particular region, the idea of terroir becomes a fuzzy concept. Since the Okanagan Valley is characterized by multiple, location specific climates, soils, etc., it hosts different terroirs. Currently, in the BC Wine Country, it is rather a rule than an exception that wineries source their grapes from various locations that are not necessarily located in proximity to their estates. This brings wines produced from grapevines grown on different terroirs and in different sub-appellations under an umbrella of one winery label (brand).  Knowing that specifics of vineyard terroir are associated with grape quality (Winkler et al., 1974, among many). The quality of grapes, in turn, is correlated with quality of the wine (Ramirez, 2008). And being aware that quality of the wine is associated with its price (Noev, 2005, among many), an interesting research question arises naturally:  Does terroir influence the pricing of BC VQA wines from the Okanagan and Similkameen Valleys?   	 43	Therefore, the purpose of the analysis pursued in this chapter is to examine how terroir elements influence the wine price formation of BC VQA wines produced in the Okanagan and Similkameen Valleys. To do this, I analyze wine prices and sales of selected BC VQA wines, in connection to terroir specifics that characterize vineyards that sourced grapes for these wines.   I organized the analysis in this chapter in the following way: in Section 3.2 I present an overview of relevant literature; in Section 3.3 I outline data sources and methods for construction of necessary variables; in Section 3.4 I describe methodology and specification of the empirical model; in Section 3.5 I present and discuss empirical results; in Section 3.6 I pursue some robustness checks. Finally, in Section 3.7 I form conclusions, research limitations, and recommendations for further studies.  3.2. Literature Overview  A famous champagne producer, Johan Joseph Krug (1800–1866) once said:  “(…) a good wine comes from good grape, good vats, a good cellar and a gentleman who is able to coordinate the various ingredients” (as quoted in Gergaud & Ginsburgh, 2008).  There is an ongoing discussion in the wine industry, as well as in wine literature, regarding what makes a good wine. Some claim that production of premium wine depends on terroir (Ramirez, 2008; Ashenfelter, 2008; Ashenfelter, Ashmore, and Lalonde, 1995; Ashenfelter and Storchmann, 2008, among many). These statements to large extent confirm what has been observed in the Old World wine-producing regions where wineries have been marketing their wines with a strong attachment to the idea of terroir and its significance in the winemaking process. Others argue that terroir might be more marketing or reputation than an actual concept related to wine quality (Cross, Plantinga, and Stavins, 2011). These latter claims sympathize with the approach of New World winemaking regions, where specialization and strong regional connection to a particular grape variety and wine type replaced the idea of terroir.  	 44	Despite all these arguments and differences in opinions, it is widely recognized that winemaking is a very sophisticated and fragile process that starts with terroir and its soil components, slope, sun exposure, and microclimate. Then the fruits of terroir’s characteristics, the wine yielding grapes, are accompanied by certain management practices and winemaking knowledge to make the final product (Gergaud and Ginsburgh, 2008). The only puzzling element that remains in this discussion is the extent to which each, terroir and winemaking art, contributes to the quality and marketing success of wine, and these items are not very easy to quantify. As much as the characteristics of terroir are exogenous, to some extent static and hard to change because each terroir is naturally endowed with certain specific natural elements,35 management practices are dynamic because they can be learned and possibly improved over time. There is also another factor that comes enters this equation: marketing efforts that lead to a wine brand’s recognition. The recognition can be gained via individual wine awards and ratings by wine experts, marketing efforts (advertisement, social media, in-store promotions, etc.), wine tourism, appellations of origin and individual terroir recognition, as well as collective, region-specific reputations known as collective reputations (Schamel and Anderson, 2003; Costanigro, McCluskey and Goemans, 2010).   For some time now, wine literature has been oscillating around the idea of geographical location and terroir, yielding analyses that research different wine regions in the world. From the numerous scholarly publications that studied directly or indirectly the concept of wine pricing versus terroir, a few that seem to be directly relevant for the analysis presented in this chapter are discussed in more detail below.  In 2003, Schamel and Anderson estimated hedonic price functions for premium wines from Australia and New Zealand, and found out that the local reputation of wines from both countries differed over time. They also established that there was a significant effect of sensory wine quality ratings on wine price premia. Ashenfelter and Storchmann (2010) examined the effects of climate change on vineyard prices in the Mosel Valley finding 																																																								35Climate-related terroir elements (e.g., temperature, precipitation, etc.) are an exception here, as they can vary between vintage years. 	 45	those specific site characteristics like slope, orientation, soil type, altitude, and solar radiation influenced vineyards and grape quality. These analyses could suggest that pricing of the BC-made wines could possibly be connected to terroir specifics that in turn influence wine sensory characteristics. On the contrary, Gergaud and Ginsburgh (2008) analyzed Bordeaux appellations and found out that site attributes of vineyards in the Haute-Medoc appellation did not affect wine prices. Also, Cross, Plantingan, and Stavins (2011) examined the value of terroir via hedonic analysis of vineyard sales in the Willamette Valley of Oregon. In their analysis, the authors regressed the prices of the vineyards located in the Willamette Valley on the measurable vineyard attributes, e.g., slope, aspect, elevation, soil type, as well as on appellation, to estimate the value of terroir in the Willamette Valley. They found that appellations strongly influenced prices for vineyards in the Willamette Valley, and that the specifics of terroir were not as important for vineyard prices. Their research discovered that the concept of terroir mattered economically, but the reality of terroir while proxied by location attributes was not significant.  These two analyses, contrary to Schamel and Anderson (2003) and Ashenfelter and Storchmann (2010), suggest that terroir could have no influence on pricing for BC-made wines.  While all these scholarly publications have brought to critical elements that helped shape the empirical analysis outlined in the next sections of this chapter, the research by Cross, Plantingan, and Stavins (2011) remains the most significant source of inspiration for the analysis of this chapter.  3.3. Data Sources and Construction of Variables  In this section, I provide an overview of all data sources and methods employed for the construction of necessary variables used in the empirical analysis of this chapter. This section is composed of two subsections: in Subsection 3.3.1 I discuss all data sources used in the empirical analysis, and in Subsection 3.3.2 I explain the rationale behind the 	 46	construction of necessary terroir variables and outline in detail methods used for their creation.  3.3.1. Data sources   The analysis pursued in this chapter investigates the connection between prices of selected BC VQA wines from the Okanagan and Similkameen Valleys of BC and the terroir that yielded grapes used in their winemaking process. In this analysis I used the following data sources:  1. The British Columbia Liquor Distribution Branch (BCLDB) wholesale pricing scanner data for BC VQA wines.  This data set consists of monthly sales of all BC VQA wines in the province of BC 36 in the period between April 1, 2011, and March 31, 2015. The variables present in this data set and used in the analysis of this chapter include wine prices (wholesale), volume of sales, time of sales (year), winery brand name, alcohol content, wine (grape) variety, and vintage year.  2. The exact location of vineyards that sourced grapes for the selected BC VQA wines and are present in the BCLDB pricing data set as described in point 1 above.  I self-collected this data from the BC wineries that produce VQA wines and agreed to deliver data on the exact location of vineyards that supplied grapes used to make selected VQA wines. The process of data collection consisted of the following steps: a) I constructed a list of all BC wineries that produce VQA wines, based on the BCLDB scanner pricing data set mentioned in point 1 above.  b) I obtained contact details for wineries from two sources:  																																																								36This wine data includes sales that occurred via all government and private liquor stores, wineries, restaurants, etc. in the whole province of BC. From all BC VQA wine sales in BC all “private label wines” were excluded. Private label wines are wines that are ordered directly from a winery by hotels, restaurants, or other establishments and they are used within these facilities only. These are often hotels’ “house wines,” for example. Additionally, all ice wines were also excluded from the analysis. Wines specified as “late harvest” are included in the analysis of this chapter. 	 47	§ The Pacific Agricultural Research Centre (PARC) Summerland wineries contact list, § Self-extracted from the official wineries mailing list of the British Columbia Wine Institute (BCWI) website or wineries websites.  c) On August 15, 2015, I contacted all wineries for which I had available contact details (via mail and email). Several wineries replied and either agreed to cooperate or requested more clarifications and then decided to participate in this research. Unfortunately, many wineries that I contacted did not respond to this initial contact letter/email. d) Due to a rather low reply rate to the initial email/mail from August 15, 2015, I visited all wineries during the second week of March 2016 and presented the opportunity to cooperate in this research. During these field visits, certain wineries agreed to cooperate, but many were closed for the low season.  I contacted the wineries that were closed during field visits again via email and presented them with the initial letter describing the purpose and details of this research. Appendix B: Chapter 3 contains the text of the original study invitation letter. e) Out of all contacted wineries, 33 agreed to participate in this research.37 The wineries that decided to be a part of this study were given (either physically during visits in wineries or via email) a list of their VQA wines that are present in the BCLDB pricing data set. Wineries provided addresses, Geographic Information System (GIS) coordinates or names of specific vineyards for all VQA wines found in the BCLDB pricing data set. To control for specific terroir elements like soil, elevation, aspect, row direction, distance to lake and climate, it was important to know the origin of grapes only for wines that were produced from a single vineyard. Wines produced from the same variety of grapes, but coming from multiple vineyards, as well as all blends (wines that were derived from multiple grape varieties) and ice wines were excluded from the data set analyzed in this chapter. 																																																								37All 33 wineries that agreed to participate in this project are estate wineries meaning that they possess a physical location for their estate winery and brand. 	 48	f) In the next step, I verified the location of vineyards using the Google Earth Pro version 7.1.5.1557 satellite imagery, to ensure that grapes were present on the provided vineyards. In the case when the satellite image was unclear, I physically visited the vineyard during additional research trips in June 2016 and made sure that grapes were planted on a given plot. g) In the last step, I matched the BCLDB pricing data set on BC VQA wines (as described in point 1 above) with vineyard data provided from wineries. This task resulted in the construction of a panel data set that matched each of the selected BC VQA wines with the exact location of a vineyard that sourced grapes used to produce that wine.  3. Environment Canada (EC) historical data set on temperatures.  To obtain a control for climate on each vineyard, I extracted the EC data set on minimum and maximum daily temperatures. I did this in the following way: a) I assigned each of the vineyards for which the location was provided by the winery (as described in point 2 above) to the closest weather station in the area. I based the assessment of the nearest weather station on the distance between said vineyard and the weather station. I measured this distance in a straight line, using the Google Earth Pro version 7.1.5.1557 software. b) After I assigned vineyards to the proper weather stations, I extracted the vintage years for all wines from the BCLDB pricing data set (the match of wine-vintage was ensured). I extracted the temperature data only for vintage years presented in the BCLDB pricing data set. c) Finally, I extracted the EC temperature data only for the months that constitute grapevines growing season in BC: April 1-October 31 (seven months total).38  4. “Atlas of Suitable Grape Growing Locations in the Okanagan and Similkameen Valleys of British Columbia.”  																																																								38In the case where the closest weather station to the vineyard was missing data for a particular vintage year, I took the temperature data from the second closest weather station (assigned by using a straight line in the Google Earth Pro version 7.1.5.1557 software/satellite imagery). 	 49	The information about soil type present in each of the vineyards was obtained from the “Atlas of Suitable Grape Growing Locations in the Okanagan and Similkameen Valleys of British Columbia” and accompanying soil maps. This atlas and maps are publicly available on the BC Ministry of Environment website.39  5. Google Earth Pro version 7.1.5.1557 software.  The information regarding several terroir-specific variables necessary for the analysis in this chapter I extracted via physical examination of satellite imagery of provided vineyard locations. As a result, I obtained the following variables using Google Earth Pro version 7.1.5.1557 satellite imagery:  a) Row direction of grapevines present in the vineyard; b) Aspect of the vineyard; c) Average elevation of the vineyard (measured in row direction); d) Distance to the closest lake (shortest distance measured in a straight line).  The first two variables: row direction and aspect of the vineyard I additionally physically and randomly checked for a sample of vineyards during research trips to the area in June 2016 (I checked 20% of all vineyards from the data set, a total of 14 vineyards).  The final panel data set composed from available data sets and used in the analysis of this chapter consists of variables coming from all five data sources, as described above. In the final panel, I matched each of the selected BC VQA wines with terroir variables characteristic for the origin of grapes used in the wine’s production. I additionally enriched this data set with terroir variables that I constructed specifically for the analysis in this chapter. The Subsection 3.3.2 below presents these variables, together with detailed methods for their construction.    																																																								39Source: https://a100.gov.bc.ca/pub/acat/public/viewReport.do?reportId=25881  accessed on December 5, 2015. 	 50	3.3.2. Construction of additional variables  As I mentioned in previous sections, the leading concept for this chapter is rooted in the idea of terroir and its role in the formation of prices for BC VQA wines produced in the Okanagan and Similkameen Valleys. To pursue analysis in this chapter, a choice of a formal definition of terroir was necessary. Based on a literature review and consultations with the wine industry (winemakers from BC and PARC Summerland), for the analysis in this chapter the concept of terroir was specified in the following way:    Terroir is defined as land and climate variables that are unique to a given location where the grapevines are being grown to make wine. Therefore, terroir incorporates the following two groups of variables:  1.    Climate variables; 2.    Topographic variables.  Both groups of terroir variables, climate, and topographic variables represent production inputs that yield key wine ingredients, the grapes. These terroir-specific elements are important because they are directly correlated with the quantity and quality of grapes grown in a vineyard. They help assess which vinifera cultivars are the best choice for that location, considering a vineyard’s natural endowments. Indirectly, they also influence the selection of wines produced from the varieties planted on a plot. The climate and land variables not only jointly characterize terroir, endowing it with location-specific natural elements, but they are also essential for the future success or failure of a winemaking process and consequently the financial prosperity of a winemaker (Winkler et al., 1974; Hellman, 2003). For example: if a vineyard is populated with an inappropriate variety of grapevines, e.g., a variety given the climate and land combination present on a vineyard doesn’t reach maturity before harvest), it will affect the winemaking process and as a consequence the quality of the wine.  Knowing that a uniqueness of wine regarding its flavor and other quality traits like acidity, sweetness, body, etc. distinguishes fine wines from poor ones, the match of terroir-grapevine variety comes with consequences that affect the quality of the wine (Hellman, 2003). 	 51	From an economic standpoint, the relationships between terroir specifics that influence grape quality, wine quality and therefore wine price can be seen from two perspectives, as per Figure 3.1 below. All these elements affect wine price via terroir.  Figure 3.1. Terroir versus wine pricing.    First, the link between superior terroir and implied higher-grade of grapes, hence better quality of wine priced at a premium could arise because of more inelastic demand for these wines (demand side). It would suggest that consumers are willing to pay a price premium for wines coming from certain sub-appellations (distinct terroir) because they associate these wines with favoured sensory characteristics. Therefore, if normal market conditions hold and if grapes are cultivated on preferred terroir resulting in the production of high-quality grapes, the quality of grapes should influence the quality of wine (its specific and valued sensory attributes) and consequently wine price. This link constitutes the first way in which terroir variables can affect the process of wine price formation.  Terroir can also impact wine price via production costs (supply side). It is possible that grapes cultivated on terroir that is deemed superior are given more attention because it is anticipated that they will be used to produce high-end/boutique wines.  This element, in turn, can translate into increased vineyard management costs that are passed on to grape buyers in the form of higher prices for these grapes (if a winemaker buys grapes for the production of its wines), or directly on to consumers in the shape of higher prices for Price of wine Terroir Price of grapes Consumer demand 	 52	wines produced from these grapes (if the winemaker grows its own grapes). For example, because individual terroir is considered superior in the production of quality grapes, the work at such a vineyard is 100% manual and the use of mechanical equipment is minimal. This element suggests higher labour costs and therefore higher costs of production for grapes cultivated on that terroir. Because of increased labour costs, the wines produced from these grapes are priced at a premium.  It is not easy to disentangle the impact that these two groups of variables have on the formation of wine prices for wines produced in the Okanagan and Similkameen Valleys of BC, as they can have individual or joint influence. But matching terroir elements with sales data for specific wines that were produced from grapes grown under specific terroir conditions can help isolate the power of natural terroir elements on wine pricing.  Now that I have established the definition of terroir and its possible role in the formation of wine prices, my next step is to investigate two groups of terroir elements: climate and topographic variables. Both groups are analyzed separately from a winemaking science-based perspective. In the next two subsections (3.3.2.1 and 3.3.2.2), I discuss the most suitable and scientifically supported variables that belong to these two groups. I describe and explain the background for each climate or topographic terroir variables and outline detailed methods used in their construction. Then I use these variables in the empirical model presented in Section 3.4.  3.3.2.1. Terroir variables  Group 1: climate variables  A proper assessment of a vineyard’s climate is one of the most important elements in grape cultivation and the winemaking process, but it remains one of the most difficult tasks. The problem arises from the multilevel definition of climate and the necessity of distinguishing various levels of climatic heterogeneity. The definition of a vineyard’s climate can be constructed and understood on at least three primary levels, as presented in Figure 3.2 below. 	 53	Figure 3.2. Levels of a vineyard’s climate.   The three primary climate levels are: 1.    Macroclimate: a general type of climate associated with latitude and longitude-dependent world climatic zones.  2.    Mesoclimate: the regional climate linked to a particular vineyard and location, which has distinctive regional differences in general climatic patterns related to terrain and topography. 3.    Microclimate: the climate present on a particular plot that results from direct interactions between soil and the grapevine’s canopy (Hellman et al., 2003).  The wine-related literature employs various measures to capture a vineyards’ particular climate, but most of them evolve around the concept of the available heat. The available heat measures usually include one of two heat variables: temperature or amount of sunlight that reaches the vineyard. Examples of the rationale for using these climate measures in the literature include numerous publications in the American Journal of Enology and Viticulture (for example: Spayd et al., 2002; Berqvist et al., 2001), as well as Winkler et al. (1974), Hellman et al. (2003), Schlenker (2006), to name a few). The use of the temperature variable is methodologically straightforward and in the grapevine-related literature usually involves the construction of a Heat Summation Index (HSI) or Growing Degree Days (GDD) index (also known as Winkler’s Index). These indices can be derived directly from the observations on mean daily temperatures coming from local weather stations and their summation over the growing period. The amount (and type) of sunlight is not as easy to measure and is more problematic because it requires 1.Macroclimate 2. Mesoclimate 3. Microclimate 	 54	computational intensive calculation algorithms that are widely borrowed from physics and earth sciences. The available sunlight is usually measured via daily extraterrestrial solar radiation and a Radiation Use Efficiency (RUE) formula 40  (Ashenfelter & Storchmann, 2010) or a Potential Photosynthetically Active Radiation (PPAR) algorithm (Failla et al., 2004). While both available heat measures—temperature and amount of sunlight—are possible to use in empirical analysis of this chapter, for reasons explained below, the temperature-based variable (see heat variable below) was considered a superior one and chosen for the empirical modelling in this chapter.  Heat variable  As I mentioned above, I considered both climate measures—temperature and amount of sunlight—as potential variables for the empirical analysis of this chapter, but I chose the temperature variable as it proved to be superior for use when analyzing BC VQA wines produced in the Okanagan and Similkameen Valleys. The main reasons for the superiority of the temperature variable over the measure of the available sunlight radiation in the analysis of this chapter are as follows:  1. Vineyards used in the analysis of this chapter are located within Latitude:   49°0’27.04” N and 49°57’16.31” N; and Longitude: 119°21’12.87” W and    119°48’36.94” W. 2. This indicates that they are situated in a relatively small area, which suggests that there would be only small differences in available sunlight. Knowing that solar radiation depends mainly on the latitude and longitude, cloud cover in the area and individual topographical characteristics of a vineyard like a vineyard’s aspect (Aschenfelter and Storchmann, 2010), the inclusion of a measure of the solar radiation may not be the optimal choice for the analysis in this chapter. 3. The algorithms used for the calculation of solar radiation differ, are complicated, and their estimates may be imprecise. A large part of the solar radiation that 																																																								40The radiation Use Efficiency (RUE) measures the mass accumulation in a gram of dry matter per MJ-1m-2of intercepted solar radiation. RUE differs for different crops, but tends to be similar across the same species of plants. 	 55	reaches the surface is diffused because of the cloud cover present at a given time and place. The data that would allow putting a control on cloud cover is not readily available for BC vineyards. Therefore, the calculation of solar radiation would need to be based on a strong assumption of no cloud cover over the vineyards (Aschenfelter and Storchmann, 2010). This hypothesis would naturally lead to a measurement error and would inevitably cause imprecise estimates in the models of this chapter.  4. Agronomic research shows that temperature variation around mean influences the growth of grapevine plants. The way in which temperature varies on any given day and between days and months impacts the plant’s overall health, well-being, yield size and crop quality (Rayne & Forest, 2016; White et al. 2006; Berquist et al., 2001). This fact suggests that a temperature measurement is a better way to control for the quality of grapes that can be influenced by extreme temperatures, which can impact the region studied given its northerly location.  While points 1-3 above describe the rationale for excluding the solar radiation measurements in the analysis of this chapter, the use of standard mean-based temperature indices (HSI or GDD) to control for climate at a given vineyard is also problematic. The standard HSI and GDD temperature indices are a poor measure not only in capturing diurnal variations in temperature around the mean but also because they are not suitable for achieving any other changes in temperatures, e.g., weekly or monthly. The omission of temperature variations can cause a serious problem in the analysis pursued in this chapter, that aims to investigate how various elements of terroir like climate influence wine pricing of BC VQA wines. Since the quality of grapes depends on weather, and it has been scientifically proven that extreme temperatures can have a detrimental effect on the quality of grapes (especially a grapevine’s fruit), it is important to control for temperature variations that influence the quality of grapes (Rayne & Forest, 2016; White et al. 2006; Berquist et al., 2001). Therefore, following on grapevine-related agronomic knowledge, in this chapter I assumed the following regarding the relationships between grapevine development and temperature: 	 56	1.    The physiological development of the vine is highly dependent on temperature, and extreme heat can damage grapes. 2.    The linkage between temperature and vine growth is dynamic, rather complex and not necessarily linear (Schlenker & Roberts, 2006; Brown, 2013). Therefore, it is assumed that the relationship between grapevine growth and temperature follows a classic nonlinear form of the S-shaped curve, as illustrated in Figure 3.3 below.  Figure 3.3. The relationship between heat and grapevine growth.   Figure 3.3: The green dots visible on Figure 3.3 show thresholds that together with  red vertical lines divide the S-shaped curve into regions 1, 2 and 3. The area on the left, region 1  and the area on the right, region 3 show an environment with temperatures that are too low and too  high, respectively for vinifera to thrive. The heat-induced development of vinifera occurs in the middle  part of this figure, in the region 2 (Brown (2013), Winkler (1974), among many).    	 57	Agronomists claim that a plant’s growth and development happen between specific temperature bounds: lower and upper thresholds, which on Figure 3.3 are represented by the area 2. Beyond the upper (or lower) temperature limit, which differs among plant species, heat (or cold) might have a detrimental influence on the plant’s well-being (Schlenker et al., 2006; Rayne & Forest, 2016). In the case of grapes, it has been scientifically established that the growth of grapes starts at a temperature of about 10°C while a detrimental heat influence is associated with a temperature of about 35°C and higher (Rayne & Forest, 2016; Hellman et al., 2004).  Therefore, for the analysis in this chapter, a heat variable is constructed. This variable controls for the frequency of the occurrence of temperatures within outside the temperature bounds (lower and upper bound). These bounds are derived based on the minimum and maximum temperatures present within each month of the grape growing season in BC. The specifics regarding the construction of the heat variable are outlined below. 	Construction of the heat variable   I constructed the heat variable in the following way: 1.  I extracted from the EC database daily minimum and maximum temperatures for each month of the grapes growing season in BC (April 1–October 31) and each weather station assigned to the vineyard (based on the smallest straight-line distance, as described in Subsection 3.1.1). 2. Then I calculated and assigned the average temperature for each month, in each vintage year, for all weather stations matched with specific vineyards in the data set. 3. From the average temperature for each month I subtracted one standard deviation to form a minimum temperature bound or added one standard deviation to create a maximum temperature bound. 4. Then I assigned the frequency of occurrence of temperatures that belonged to each of the temperature bounds. For example: if a minimum temperature bucket 	 58	for April was established at the temperature less than 8°C (<8°C) and if the average temperature in that month, in Kelowna was 6°C, one frequency observation was recorded in the data. 5. Finally, I set the comparison bound for temperatures ad hoc as the middle interval. For example: if a minimum temperature bucket for April was set at temperatures less than 8°C (<8°C) and a maximum temperature bucket was set at temperatures more than 20°C (>20°C), then the reference (comparison) interval consisted of temperatures in the interval [8°C, 20°C].  3.3.2.2. Terroir variables  Group 2: topographic variables  The topography of a vineyard is essential for a proper development of grapes and therefore for wine quality. This influence comes from the interactions between temperature, soil, and canopy that interfere with the mesoclimate of the vineyard (Hellman, 2003). The most important topographic elements include soil type, elevation, and slope. The elevation and slope influence grape quality via topographic moderations in the mesoclimate of the vineyard that can be affected by the steepness of the slope or site elevation (absolute and relative). The soil, on the other hand, has a direct influence on grape quality, mainly via its mineral composition that is also able to affect the taste of the wine. The soil-type-dependent water holding capacity is another element that proves to be crucial for grape vigour and can influence grape and wine quality.  For the empirical analysis in this chapter, I investigated five possible topographic variables: soil type, average elevation, aspect, row direction, and distance to the lake.  I excluded the slope variable due to the lack of a good quality measure for slope on vineyards that are present in the available data set.    	 59	Soil variable  Soil remains one of the most important topographic elements of terroir. Its type defines the availability of nutrients and water holding capacity. Both these factors are the most important variables not only for the future well-being of grapes but also for the choice of important vineyard management strategies: trellising system, rootstock and vine spacing.  In the case of grapevines, soils characterized by moderate fertility are more beneficial for the cultivation of grapes than highly fertile soils, as moderately fertile soils allow better management of the vine canopy (Hellman et al., 2003; Winkler et al., 1974). Among many soil types that allow cultivation of grapes, there is no single soil type that is superior and able to guarantee the highest quality of grapes, hence the best wines. The most important element related to the quality of soil used for grapes cultivation is its good internal drainage (Hellman, 2013). Also, as some research shows, sandy and gravelly soils might be more desirable for grape cultivation (FAO Agribusiness Handbook, 2009). More details regarding the actual construction of the soil variable can be found below.  Construction of the soil variable  The matching of the soil type at each of the vineyards was pursued via comparison of the exact geographical location of the vineyard using Google Earth Pro version 7.1.5.1557, with a soil map of the Okanagan and Similkameen Valleys of British Columbia, as outlined in the “Atlas of Suitable Grape Growing Locations in the Okanagan and Similkameen Valleys of British Columbia” prepared by the Association of British Columbia Grape Growers (1984) (accessed on December 15, 2015: https://a100.gov.bc.ca/pub/acat/public/viewReport.do?reportId=25881) The soils in this atlas include 14 soil types, with soils classified in the following way:      	 60	Table 3.1. Soil classes. Classification in terms of suitability for grapevines cultivation Soil class Well-suited 1, 2, 3, 4 Moderately well-suited 5, 6, 7, 9 Poorly suited 10, 11 Not suited 8, 12, 13, 14  After matching of vineyards with soil maps, I identified and matched with proper vineyards the following groups of soils:  Table 3.2. Soils well-suited for grape cultivation. Soil Type Description Type 1  It is a well-drained soil, with medium to fine textured stream deposited fan material. Subsoil: gravelly sandy loam, gravelly silt loam or silt loam. Type 2 It is a well to rapidly- drained soil, with medium to moderately coarse texture stream deposited fluvial fan materials. Type 3 Soil type 3: It is a well -drained soil, with medium to moderately coarse textured unsorted till deposits. Occurs on slopes 10-30%. Type 4 It is a well-drained medium textured soil with medium to moderately fine textured glaciolacustrine sediments. Occurs on slopes 2-9%. Weak to moderate salinity.   Table 3.3. Soils moderately well-suited for grape cultivation. Soil Type Description Type 5  It is a well-drained soil developed on veneers of coarse textured melt water stream deposits overlaying moderately fine silt and clay sediments. Occurs on slopes 5-30%. Weak salinity. Type 6 It is mostly rapidly drained soil with coarse textured melt water streams, fluvial fans or recent stream deposits. Low water holding capacity, mainly sands. Occurs on slopes up to 30%. Type 7 It is rapidly drained soil with coarse textured melt water streams, stream deposited fans, or recent stream deposits. Gravels, sands and cobbles. Low water holding capacity. Occurs on slopes up to 30%. Type 9 It is a moderately well-drained soil with moderately fine to fine textured silts and clays. Gravel-free. Slow infiltration, low aeration and relatively cool soil temperature. Occurs on slopes 2-9%.  To pursue the analysis in this chapter, I grouped all soil types into two classes that formed two indicator variables: 	 61	1. Indicator variable 1: well-suited (includes soil types 1, 2, 3, and 4, as described above), 2. Indicator variable 2: moderately well-suited (includes soil types 5, 6, 7, and 9, as described above).  I coded the soil type dummy variables in the data set according to this method: Indicator variable “well-suited” 1 =1 if the soil at a vineyard is well suited and zero otherwise. Indicator variable “moderately well-suited” =1 if the soil at the vineyard is moderately well suited and zero otherwise.  Average elevation variable  The elevation of a vineyard is important mainly from the vineyard temperature standpoint (Hellman et al., 2003; Winkler et al., 1974; Failla et al., 2004). The scientific research shows that the mean temperature drops by about 0.5°C–0.6°C for each 100 metres of an increase in elevation (FAO Agribusiness Handbook, 2009). Due to this elevation-dependent temperature drop, vineyards located at higher elevations may observe lower temperatures. This element, in turn, can negatively influence grape maturation and consequently the quality of the wine. Experiments pursued on the interactions between grapes and elevation gain show that there is an observed average bud break delay of 2.3 days for an increase in elevation of 100 metres. This relationship affects the ripeness of grapes and their readiness for harvest and, therefore, wine flavour, acidity, and other quality-related wine specifics (Failla et al., 2004). The method used for the construction of the average elevation variable is presented below. 	Construction of the average elevation variable  Due to the lack of a precise measure of slope in vineyards and to put a control on the influence of altitude of the vineyard on grape quality, I constructed the average elevation variable. In the first step, I measured the average elevation using the Google Earth Pro 	 62	version 7.1.5.1557 software. I pursued the measurement of the average elevation in the direction of rows of grapes that were planted in a given vineyard. In the last step, I assigned each elevation to one of three groups. As a result, I created the following three indicator variables:   Table 3.4. Average elevation indicator variables. Indicator Variable Description avgelev1 Average elevation of [0-200 metres] avgelev2 Average elevation of (200-400 metres] avgelev3 Average elevation of (400 metres and up)  The coding of the average elevation indicator variables in the data set was pursued in the following way: Indicator variable [0–200m]=1 if the average elevation on a vineyard is in the interval [0–200] and zero otherwise, etc.   Aspect variable 	While I excluded from the list of variables used in this chapter the measurement that controls for the amount of sunlight that reaches vineyards (due to its problematic likelihood of the measurement error), I decided to include another variable that can put a control on the direction of insolation that reaches vineyards. This is the variable called “aspect,” which is a compass direction of a vineyard towards the sun. The insolation is important for the vineyard and therefore the quality of grapes because of the influences that sunlight has on the photosynthetic processes and the overall well-being of grapevine plants. Scientific research has proven that grapevines that are exposed to sunlight show higher levels of total soluble solids, anthocyanins, and phenolics, and have lower titratable acidity, malate, juice pH, and berry weight, when compared to non-sunlight-exposed grapes (Berquist et al., 2001; Crippen & Morrison, 1986; Dokoozlian et al., 1996; Hale & Buttrose, 1974 to name a few). Since the exposure of grapevine plants towards sun can influence all these wine-taste-related elements, it can be concluded that there is a connection between sun exposure and the quality of the wine. Therefore, a control for the aspect on each of the vineyards in the data set is justified. The variable 	 63	aspect is considered as a topographic variable, even though it directly influences the mesoclimate of a given vineyard. The literature on this topic considers Southern (S), South-Eastern (SE) and South-Western (SW) aspects as preferable vineyard directions towards the sun in the Northern Hemisphere. Vineyards with North-Western (NW), Northern (N) and North-Eastern (NE) aspects are considered to have inferior facing for grape maturation and an overall negative influence on grape quality (Hellman et al., 2003). Further details regarding the construction of the aspect variable are presented below.  Construction of the aspect variable  As I mentioned above, the aspect variable shows a compass direction of a vineyard towards the sun. For the analysis in this chapter, I constructed the aspect variable via observation of the satellite images and sun-facing direction for each of the vineyards present in the data set. These observations were pursued using Google Earth Pro version 7.1.5.1557 satellite images. The process resulted in the creation of eight indicator variables that match vineyard directions towards sunlight:   Table 3.5. Aspect indicator variables. Indicator Variable Description E Eastern aspect W Western aspect S Southern aspect SW South-West aspect SE South-East aspect NW North-West aspect NE North-East aspect FLAT Undistinguishable aspect   	 64	 In the available data set, the Northern (N) aspect direction wasn’t observed. The coding of aspect indicator variables in the data set was pursued according to this example: Indicator variable E=1 if the aspect on a vineyard is E and zero otherwise, etc.  Rows variable  The direction of rows on a plot is one of the most significant elements for the optimal functioning of a vineyard (Berquist et al., 2001). It not only influences the quality of grapes and therefore the quality of wine,41 but row direction is also a fundamental business decision. Once rows are put in place on a vineyard, it is costly and labour intensive to make any changes in the way they are set on the plot. The decision about the direction of rows usually depends on the shape of the vineyard, its topography, microclimate and prevailing winds (Greenspan, 2008). Row direction is especially important in vineyards located on slopes steeper than 30%. In such cases, rows influence the ability to use machinery introducing the risk of machinery tipping over, especially when rows are directed down the slope instead of across the slope (Hellman et al., 2003). The scientific literature that concerns row direction claims that the North-South (NS) direction of rows is preferable in the Northern Hemisphere as in this orientation all grapes receive a similar amount of heat and sunlight, which in turn positively influences grape’ quality (Hellman et al. 2003). This NS direction can additionally be improved in the case of vineyards located in the “cool climate” areas by tilting row direction by about 10–15 degrees West of North (Greenspan, 2008). The direction of rows is a justified variable for inclusion in the modelling of this chapter, as it can reinforce the availability of sunlight and its diurnal distribution across the canopy of a given vineyard. The availability of sunlight can influence the quality of grapes and, therefore, the quality of the wine. While row direction is a management decision of a winemaker or grape grower, in this chapter row direction is treated as a variable that reinforces and belongs to the Terroir Group 2 																																																								41Row-direction-dependent absorption of sunlight and heat facilitates or impedes the uniform maturation for grapes. The uniform maturation (ripening) of grapes positively influences the quality of the wine (especially wine flavour) as per Greenspan, 2008. 	 65	variables: topographic variables. It is argued that the direction of rows on a vineyard strongly depends on the topography of a vineyard and when it is chosen, it is usually unchanged because it requires pulling off grapevine plants and re-planting, which constitute expensive and radical management steps. The direction of rows influences access of sun rays to the vineyard and their diurnal distribution. This, in turn, affects the climate that is present in a given vineyard. More details regarding the actual construction of this variable are shown below.  Construction of the rows variable  The assignment of row direction to each of the vineyards presented in the data set was pursued via an inspection of satellite images of specific vineyards using the Google Earth Pro version 7.1.5.1557 software.42 To capture the influence of row direction on the quality of grapes and therefore the quality of the wine, for the analysis in this chapter the following indicator variables on row direction were constructed:   Table 3.6. Row direction indicator variables. Indicator Variable Description NS North-South rows direction EW East-West rows direction SE-NW South East-North West direction SW-NE South West-North East direction  The coding of row direction in the data set was pursued according to this example: Indicator variable NS=1 if the direction of rows in a vineyard is NS and zero otherwise, etc.   																																																								42Many vineyards presented in the data set were also physically inspected and row direction was confirmed during numerous visits to the Okanagan and Similkameen Valleys. 	 66	Lake variable  The last topographic variable considered for the analysis in this chapter is the distance from a vineyard to the closest lake. The literature on the topic suggests that the distance to a lake can influence the mesoclimate of a particular vineyard. This power comes from the lake’s ability to moderate nearby land temperatures due to the high heat capacity of the body of water (Cohen et al. 2012; Ashenfelter and Storchmann, 2010). This heat management ability of lakes is crucial for grapes, as the proximity to a lake can cool grapes during hot days and warm them up during colder nights, diminishing the possibility of plant stress that could affect the plants’ optimal growth and the quality of fruit. The distance to the lake could also be an important variable if vineyards present in the data set were not equipped with irrigation systems. This influence would be associated directly with water availability in the vineyard. Since all vineyards in the data set used in this chapter are equipped with irrigation systems, the proximity to the lake is not as important regarding water availability as it could be. Further details related to the construction of this variable are explained below.  Construction of the lake variable  To put a control on a vineyard’s distance from a large water reservoir like a lake, for example, a measure of the distance of the vineyard from the closest lake was established. I pursued the measure using Google Earth Pro version 7.1.5.1557. All measured values of distance to the lake were recorded and assigned to three range groups. Consequently, three indicator variables, one for each  “distance group,” were created (as per Table 3.7 below):   	 67	Table 3.7. Distance to lake from vineyard indicator variables.43 Indicator Variable Description lake1 Distance from vineyard to lake: [67m-700m] lake2 Distance from vineyard to lake: (700m-3000m] lake3 Distance from vineyard to lake: (3000m and up)  The coding of the lake indicator variables in the data set was pursued according to this example: Indicator variable [67m, 700m] =1 if the distance of the vineyard to lake belongs to the interval [67m, 700m] and zero otherwise, etc.   Non-terroir variables  While most of the variables used in the analysis of this chapter are constructed and directly associated with the terroir of specific vineyards, there are also variables that come directly from the BCLDB data set. These latter variables are not terroir variables. These variables include alcohol content, variety, brand, and wine age. The alcohol variable is indirectly associated with terroir (climate) and puts control on the alcohol content of a specific wine. The variety variable controls for grape type/wine type (and for wine colour). The brand variable controls for wine label (winery). The wine age variable controls for the age of the wine. The wine age squared variable controls for a possible nonlinearity in the wine age. The year variable puts control on time trend.      																																																								43Because all vineyards present in data set are equipped with irrigation systems, the variable that could control for water scarcity and its possible detrimental effect on grapes quality on the plot was omitted in this analysis. For the same reason, I also excluded the variable controlling for rainfall. I assumed that due to the presence of irrigation system on the plot, each vineyard had an abundant water supply. Also, the field interviews with winemakers in the area and representatives of the Agriculture and Agri-Food Canada in Summerland (AAFC/PARC), a research body responsible for extension services in wine industry confirmed that extensive rainfall or unfavorable winds are not problematic in this research.  	 68	3.4. Methodology, Empirical Model Specification and Estimation Method  In this section, I provide an overview of the methodology used in the empirical analysis of this chapter, economic theory that rationalizes the choice of this method, and empirical model specification. Specifically, in Subsection 3.4.1 I discuss the methodology and its theoretical economic background, in Subsection 3.4.2 I outline details regarding the empirical model specification, and in Subsection 3.4.3 I present the estimation method.  3.4.1. Methodology  From an economic theory standpoint, the methodology I chose for the analysis in this chapter seems to belong to the stream known by economists as a hedonic price method. This type modelling approach goes as far back as 1928 when Fredrick V. Waugh pursued an analysis regarding the quality factors influencing the price of asparagus. Waugh published a research paper where he regressed the price of asparagus sold in the Boston market between May-July 1927 on three asparagus quality measures: colour, size of stalks, and uniformity of spears (as cited in Nerlove, 1995). Even though Waugh was the first to use the hedonic specification, the term “hedonic pricing method” is attributed to Court (1939) who applied this method to automobiles (as cited in Combris, Lecocq, and Visser, 1997). From that time, the methodology has gained momentum and has frequently been used to estimate consumers’ valuation of certain quality attributes for many different consumer products, agricultural commodities, housing, and even air quality (Nerlove, 1995). The theoretical basis for the hedonic pricing method was laid by Sherwin Rosen, who in 1974 published a seminal paper: “Hedonic Prices and Implicit Markets: Product Differentiation in Pure Competition” (Journal of Political Economy 82, 34–55). The hedonic method analyzes price-quality relationships, and as Rosen presented in his paper, it can be pursued from the consumers and producers’ perspectives (Rose, 1974). The formal steps in the hedonic analysis include the use of the observations on prices of a differentiated good, together with attributes of the good, and a construction of a regression that estimates the hedonic price function. The regression results yield the implicit prices for the good’s characteristics. The ratios of these implicit prices provide 	 69	the consumers’ marginal rates of substitution among attributes (Rosen, 1974; Nerlove, 1995). The hedonic price method is not without its problems. The most commonly stated issue concerns the identification problem when one tries to draw inferences about consumer preferences from the hedonic regression. The problematic part is related to the fact that the quantities of attributes associated with each variety and the number of units sold are in general jointly determined by supply and demand (Rosen, 1974). The identification problem is not unique to the hedonic pricing method, but it is a problem of much other demand and supply modelling approaches, as price-quantity observations jointly represent demand and supply equilibria (Working, 1927).  Regardless of the identification issues, the hedonic pricing method is particularly popular and frequently employed for analyses in wine economics. Wine economists usually use the hedonic pricing method to estimate consumers’ valuation of wine attributes, either sensory (aroma, acidity, etc.) or objective (wine origin, region, vintage, etc.), to find the implicit prices for these attributes. Examples of publications that use the hedonic pricing methodology in the analyses of wine include: Oczkowski (1994), Combris, Lecocq and Visser (1997), Landon and Smith (1997), Schamel and Anderson (2003), Noev (2005), Costanigro, McCluskey and Mittelhammer (2007), among many.   While in wine economics the estimation of the wine hedonic price function from the consumer side is very popular, it is much less common from the producer side. In the case of the methodology used in this chapter, the closest publication regarding the approach is from the research pursued by Cross, Plantinga, and Stavins (2011). In their analysis, the authors regressed the prices of vineyards located in the Willamette Valley (Oregon, US) on the measurable vineyard attributes, e.g., slope, aspect, elevation, soil type, as well as on appellation, to estimate what was the value of terroir in the Willamette Valley. They found out that prices for vineyards in the Willamette Valley were strongly determined by appellation, but not by the specifics of terroir.  	 70	In the analysis of this chapter I use the hedonic pricing methodology, but instead of a usually seen approach where the price of wine is regressed on the wine’s various sensory (sweetness, aroma, etc.) and objective (vintage, variety, etc.) attributes, in this chapter I regress the price of wine on terroir elements associated with a specific wine that was produced from grapes grown on said terroir. As I described in the preceding section, I ensured the existence of the match of the wine price with the origin of grapes used in the process of wine production and specifics of the vineyard that yielded these grapes. It is likely that because of this modelling approach and because of the research on the BC wine region, which is sparse in this type of analyses, the analysis in this chapter constitutes a unique approach in wine hedonic literature. 	3.4.1.1. Empirical model specification  In the classic hedonic pricing model, the price of a good is regressed on the good’s attributes to find an implicit valuation of these attributes (implicit prices). I followed this methodology in the development of the empirical model for this chapter. The available data set that I constructed for this analysis includes two groups of variables that can be classified as per Figure 3.4 or Figure 3.5 below:   Figure 3.4. Division of variables used in the empirical model based on type of variable.      Empirical Model Variables Terroir Variables:  HEAT, SOIL, ROWS, ASPECT, AVGELEV, LAKE  Wine Specific Variables: WINEAGE, WINEAGESQ, ALCOHOL, BRAND, VARIETY, YEAR 	 71	  Figure 3.5. Division of variables used in the empirical model based on their variability over time.   3.4.1.2. Empirical model   Since the primary goal of the analysis pursued in this chapter is to establish if (and how) terroir elements influence the formation of wine prices in the case of the selected BC VQA wines produced in the Okanagan and Similkameen Valleys, the estimation of the hedonic pricing model, as per specification outlined below is employed:                                     𝒚𝒊𝒕 = 𝜶+ 𝑿𝒊𝒕! 𝜷+ 𝒁𝒊𝒕! 𝜸+  𝜺𝒊𝒕                         Equation 3.1                      Where:  yit is a wholesale price of wine “i” in year “t” (in Canadian dollars), either in the level-level or log-level form as these two specifications are tested, α is a regression intercept, Xit’ is a matrix of explanatory variables including the following variables: § WINEAGE: the age of wine based on the wine’s vintage year and calculated in the following way: wine sales year minus wine vintage year = WINEAGE (a continuous variable); Empirical Model Variables Time Invariant:  SOIL, ROWS, ASPECT AVGELEV, LAKE, BRAND, VARIETY, ALCOHOL Time Varying: HEAT, WINEAGE, WINEAGESQ, YEAR 	 72	§ WINEAGESQ: squared WINEAGE, a variable constructed to control for a possible non-linearity in the WINEAGE (a continuous variable), § BRAND: wine/winery brand (33 indicator variables), § VARIETY: type of wine based on the grape variety e.g.: Merlot, Malbec, etc. (an indicator variable). Note: the variety also indicates the wine color: either red or white. Therefore, the variable “color” was excluded in this analysis because it would be redundant, § ALCOHOL: wine alcohol content (a continuous variable organized in 3 groups (3 indicator variables), § YEAR: year of wine sales (an indicator variable: 5 indicator variables for years 2011-2015). This is time effect/trend.  Zit’ is matrix of explanatory variables associated with terroir and includes the following variables: § AVGELEV: average elevation on the vineyard (a continuous variable organized in 3 groups (3 indicator variables), § ASPECT: vineyard’s direction towards sun (8 indicator variables), § ROWS: rows’ direction on the vineyard (4 indicator variables), § LAKE: distance of a vineyard from the lake (3 indicator variables), § HEAT: frequency of extreme temperatures in the upper and lower bound (continuous variable), § SOIL: soil type (2 indicator variables),  β and γ are vectors of regression estimates, εit is the regression error term, where εit ≈IID (0, σv 2).      	 73	3.4.2. Estimation method  The available sample of the selected BC VQA wines from the Okanagan and Similkameen Valleys of BC used in the analysis of this chapter consists of 252 different wines (different SKU numbers) together with their volumes of sales and wholesale prices observed between April 1, 2011 and March 31, 2015. The total data set used in the analysis is composed of N=6785 observations on prices and sales of these 252 wines (repeated monthly purchases of these wines over 2011–2015). The actual presence of these wines in the data set varies between years 2011 and 2015 as some wines go out of sales and others enter the market. All wines that are present in this data set have standard wine bottles with the volume of 0.75 litres, and they were produced by 33 different brands (estate wineries) located in the Okanagan and Similkameen Valleys of BC. The list of all wineries that participated in this research can be seen in Appendix B: Chapter 3, Table B.1 and Figure B.1. All wines used in this analysis are either red or white, with the age of wines between 0-15 years (16 vintages in total). There are 24 different grape varieties/wine types present in this data set. The origin of grapes used to produce these 252 wines can be traced to 71 different vineyards located in the Okanagan and Similkameen Valleys. All these 71 vineyards are mapped and presented in Appendix B: Chapter 3, Figure B.2.  Some of the vineyards coincide with the location of the estates of the 33 wineries (brands), but numerous are in different, sometimes quite distant areas in comparison to the location of the estate wineries. The 71 vineyards that supplied grapes to produce these wines are located within 14 different proposed sub-appellations (as per demarcation suggested by the BC Wine Appellation Task Group), plus one area (Similkameen Valley) that was not included in the sub-appellations proposal (called in the analysis of this chapter: “Beyond sub-appellations demarcation (Similkameen Valley)”). All wines present in this data set are associated with the same origin of grapes in years 2011–2015 meaning that if a wine A was produced from grapes coming from the vineyard X in 2011, the grapes from the same vineyard were used to produce this wine in the next years, 2012–2015. There was only one exception to this rule when the same SKU/wine between 2011 and 2015 was produced from grapes coming from two different vineyards.  	 74	Tables 3.8 and 3.9 below present additional summary statistics for this data. Table 3.10 below outlines the distribution of wines per origin of grapes used for their production. Table 3.10 below outlines the distribution of wines per origin of grapes used for their production.  Due to changes in the wholesales pricing model that officially came to life in BC, in 2015, the BCLDB BC VQA wine scanner pricing and sales data available for this research is composed of two groups of pricing data: 1.    2011–2013 pricing data that shows prices constructed based on the pre-wholesale pricing model changes, with wholesale prices (Liquor Distribution Board (LDB) display prices) formed under the old pricing model. 2.    2014–2015 pricing data that shows wholesale prices created under the new wholesale pricing model.   To correct for these differences between these two pricing models, prices from 2011–2013 were adjusted by the Provincial Sales Tax (10%) to put them on a comparable level with prices from 2014–2015. More details about the differences between these two groups of prices can be seen in Appendix B: Chapter 3, Figure B.3 and in comments under that chart  Table 3.8. Data descriptive statistics.         N=6785 obs.  (252 wine SKU)  Variables  Modalities  Mean Standard Deviation  Minimum  Maximum  Price  continuous variable 19.49 8.59 9.61 90.66 Wineage " 3.83 2.95 0 15 Brand indicator variable   1 33 Variety "   1 24 Year "   2011 2015 Vineyard (source of grapes)     1 71 	 75	Table 3.9. Descriptive statistics continuation.     Variables Frequency Percent  Alcohol (total 3 groups)   Alcohol below 12% 733 10.8 Alcohol [12-14%] 4523 66.66 Alcohol above 14% 1529 22.54 Sub-appellations (total 15)   Alluvial fans and flood plains 22 0.32 East side mixed sediments 263 3.88 Glaciofluvial terraces 517 7.62 Golden Mile Bench  898 13.24 Kettled outwash and fans  896 13.21 Mission Creek terraces  589 8.68 Mixed sediments and fans 624 9.2 NE side lacustrine bench  687 10.13 SE side lacustrine bench 378 5.57 Sandy outwash lakeside terraces East side 298 4.39 Sandy outwash lakeside terraces West side  60 0.88 Sandy outwash terrace and deposits  123 1.81 West side lacustrine bench  46 0.68 West side mixed sediments  957 14.1 Beyond sub-appellations demarcation (Similkameen Valley)  427 6.29    TERROIR VARIABLES Soil (total 2 groups)   Well-suited 3322 48.96 Moderately-well suited 3463 51.04 Rows (total 4 groups)   North-South 2299 33.88 East-West 1887 27.81 Southeast-Northwest 1387 20.31 Southwest-Northeast 1221 18 Aspect (total 8 groups)   East 391 5.76 Flat 2516 37.08 North-East 253 3.73 North-West 508 7.49 South 488 7.19 South-East 1177 17.35 South-West 743 10.95 West 709 10.45 Average Elevation (total 3 groups)   Average elevation [0-200m] 2027 29.87 Average elevation (200-400m] 1432 21.11 Average elevation (400m and up) 3326 49.02 Distance to lake (total 3 groups)   Distance to lake [67-700m] 1870 27.56 Distance to lake (700-3000m] 2396 35.31 Distance to lake (3000m and up) 2519 37.13       SKU #  252 N   6785 * All variables in this table are indicator variables.        	 76	Table 3.10. Distribution of wines (SKU#) per origin of grapes used for their production (total over the whole sample). Proposed Sub-appellation Number of SKU (Wines), which grapes came from specific sub-appellation Alluvial fans and flood plains 3 East side mixed sediments 8 Glaciofluvial terraces 17 Golden Mile Bench  25 Kettled outwash and fans  34 Mission Creek terraces  17 Mixed sediments and fans 25 NE side lacustrine bench  30 SE side lacustrine bench 15 Sandy outwash lakeside terraces East side 19 Sandy outwash lakeside terraces West side  1 Sandy outwash terrace and deposits  4 West side lacustrine bench  1 West side mixed sediments  39 Beyond sub-appellations demarcation (Similkameen Valley)  15 *Note: There is one SKU (wine) that was produced between 2011-2015 from grapes coming from two different vineyards. Therefore, a total of SKU numbers in this table adds up to 253, not 252.                              	 77	3.4.3. Scatter plots  To present more details regarding the available wine pricing data set, I created several scatter plots. They show additional relationships that characterize the BC VQA wine data set analyzed in this chapter. Below I show and discuss five of these scatter plots. The other plots I present in Appendix B: Chapter 3 (Figures B.4–B.8).  Figure 3.6. Price vs grape variety, separated by winery/brand.   This figure shows that there is a visible variability in the wholesale pricing across wine varieties and wine brands, but most varieties/brands in this data set tend to be located at the price level of CAD 10-30. 102030405060708090PRICE in CAD $1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24VARIETY (24 in total)8TH GENERATION ANCIENT HILLBENCH 1775 BLACK HILLSBLACK WIDOW CROWSNESTD'ANGELO/QUAILS GATE FAIRVIEW/ROBIN RIDGEGEHRINGER BROTHERS/ROLLINGDALE HAINLE/SERENDIPITYHAYWIRE/SPERLING HESTER CREEK/ST.HUBERTUS &OAK BAYHILLSIDE/SUMMERHILL HOUSE OF ROSE/THORNHAVENHOWLING BLUFF/TINHORN LANG/UPPER BENCHLITTLE STRAW/VOLCANIC HILLS MEYER/WILD GOOSEMISCONDUCT NOBLE RIDGEPOPLAR GROVESource: BCLDB pricing data set for 2011-2015 plus self-collected data on terroir.2011-2015Wine price versus grape variety, separated by winery/ brand	 78	Figure 3.7. Price vs winery/brand, separated by grape variety.    Figure 3.7 shows that there exists a variability in the wine pricing across grape varieties and brands, but it is not clear how the price of wine depends on the variety.  Figure 3.8. Price vs alcohol content, separated by grape variety.  102030405060708090PRICE in CAD $1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33WINERY (33 in total)BACO NOIR CABERNET FRANCCABERNET SAUVIGNON CARMENERECHARDONNAY EHRENFELSERGAMAY NOIR/TREBBIANO GEWURZTRAMINER/VIOGNIERMARECHAL FOCH/ZWEIGELT MERLOTPINOT AUXERROIS PINOT BLANCPINOT GRIGIO PINOT GRISPINOT MEUNIER PINOT NOIRRIESLING SANGIOVESESAUVIGNON BLANC SYRAHTEMPRANILLOSource: BCLDB pricing data set for 2011-2015 plus self-collected data on terroir.2011-2015Wine price versus winery/brand, separated by variety102030405060708090PRICE in CAD $9 10 11 12 13 14 15 16ALCOHOL in %BACO NOIR CABERNET FRANCCABERNET SAUVIGNON CARMENERECHARDONNAY EHRENFELSERGAMAY NOIR/TREBBIANO GEWURZTRAMINER/VIOGNIERMARECHAL FOCH/ZWEIGELT MERLOTPINOT AUXERROIS PINOT BLANCPINOT GRIGIO PINOT GRISPINOT MEUNIER PINOT NOIRRIESLING SANGIOVESESAUVIGNON BLANC SYRAHTEMPRANILLOSource: BCLDB pricing data set for 2011-2015 plus self-collected data on terroir.2011-2015Wine price versus alcohol, separated by variety	 79	Figure 3.8 shows that there exists variability in the wine pricing versus alcohol content, but it is not entirely clear how the alcohol content influences wine prices.  Figure 3.9. Price vs distance to lake, separated by grape variety.   Figure 3.9 shows that most wineries in this sample source grapes from the vineyards that are located within 0–5000 metres (m) from the closest lake. These wines tend to have prices in the interval of CAD 10-30. But there exist wines that were made from grapes cultivated on the vineyards located more than 1000 m from the closest lake, and these wines tend to have much higher prices. This might suggest that the distance to the lake may not be a major factor in the pricing of BC-made wines.         102030405060708090PRICE in CAD $0 5000 10000 15000 20000 25000LAKE in MetresBACO NOIR CABERNET FRANCCABERNET SAUVIGNON CARMENERECHARDONNAY EHRENFELSERGAMAY NOIR/TREBBIANO GEWURZTRAMINER/VIOGNIERMARECHAL FOCH/ZWEIGELT MERLOTPINOT AUXERROIS PINOT BLANCPINOT GRIGIO PINOT GRISPINOT MEUNIER PINOT NOIRRIESLING SANGIOVESESAUVIGNON BLANC SYRAHTEMPRANILLOSource: BCLDB pricing data set for 2011-2015 plus self-collected data on terroir.2011-2015Wine price versus distance to lake, separated by variety	 80	Figure 3.10. Price vs average elevation on vineyard, separated by grape variety.   Figure 3.10 shows that most of the grapes used to produce the selected wines come from the vineyards that are located on the plots with the average elevation of 300–500m. There exists variability between the pricing, variety, and the average elevation, but it is not obvious how the average elevation influences the prices of wines.  3.4.4. Necessary assumptions for model estimation 	To estimate the empirical model outlined in Subsection 3.4.1.2, I used the Ordinary Least Squares (OLS) estimation method with fixed effects. The two hedonic price equations were estimated: one with the dependent variable, the price of wine in the level form, and the other with the price of wine in the logarithmic form. The dependent variable, the price of wine, is not normally distributed (please refer to Figures B.9 and B.10 in Appendix B: Chapter 3); therefore, it is suspected that the model with the logarithmic transformation of the dependent variable might have a better fit. The independent variables consist of terroir and non-terroir (or time-varying versus time-invariant) variables, as described in Subsection 3.4.1.1 above. 102030405060708090PRICE in CAD $0 100 200 300 400 500AVGELEV in MetresBACO NOIR CABERNET FRANCCABERNET SAUVIGNON CARMENERECHARDONNAY EHRENFELSERGAMAY NOIR/TREBBIANO GEWURZTRAMINER/VIOGNIERMARECHAL FOCH/ZWEIGELT MERLOTPINOT AUXERROIS PINOT BLANCPINOT GRIGIO PINOT GRISPINOT MEUNIER PINOT NOIRRIESLING SANGIOVESESAUVIGNON BLANC SYRAHTEMPRANILLOSource: BCLDB pricing data set for 2011-2015 plus self-collected data on terroir.2011-2015Wine price versus average elevation, separated by variety	 81	Clustering of errors 	I suspect that the specifics of the available for this chapter wine data require an assumption of correlated standard errors (SE). The available observations on wines are associated with different areas of the Okanagan and Similkameen Valleys (different proposed sub-appellations). I suspect that wines coming from the same proposed sub-appellation can be correlated in some unknown way (inter-group correlation) introducing correlation in the error term within that group. The assumption of correlated errors implies that the observations within group i are correlated in some unknown way inducing correlation in the error terms within group i, but that groups i and j do not have correlated errors. In the presence of correlated SE, the OLS estimates are still unbiased, but the SE may be quite wrong leading to incorrect inference in a surprisingly high proportion of finite samples. Therefore, in the estimation process, the clustering of SE based on the proposed sub-appellations is pursued. The proposed sub-appellations with their demarcation frontiers seem like a good choice for clustering variables for BC VQA wines. The boundaries of these proposed sub-appellations were chosen by the industry, with the help of the PARC Summerland (scientific background). This suggests that the elements that could induce clustering of errors like region-specific wine styles, grape production techniques, vineyard management, and region-specific winemaking know-how are enclosed by these proposed sub-regional boundaries. Therefore, the SE clustering on sub-appellations should mitigate problems associated with correlated errors.  Additional scatter plots that visualize why clustering of SE based on sub-appellations is justified can be seen in Appendix B: Chapter 3, in Figures B11–B36.  3.4.5. Software used for model estimation  All empirical model specifications were estimated using Stata 13 Special Edition software. The results are presented and discussed in Section 3.5 below. The tables with full results from these specifications are shown in Appendix B: Chapter 3 (Tables B.4 and B.5).  	 82	3.5. Empirical Results and Discussion  In this section, I present results obtained from the empirical analysis and discuss their overall significance. Specifically, in Subsection 3.5.1 I outline tables containing significant regression results, and in Subsection 3.5.2 I discuss the significance of the results.  3.5.1. Regression results  As I mentioned above, for the empirical analysis in this chapter I implemented two forms of the empirical model specification:  1. Model 1: level-level—with the dependent variable, the price of wine in the level form, 2. Model 2: log-level—with the dependent variable, the price of wine in the logarithmic form. The significant OLS estimates for terroir related variables, for both the level-level and log-level models, are presented in Tables 3.11 and 3.12 below. The results from the full specifications for these models can be seen in Appendix B: Chapter 3 (Tables B.4 and B.5). The specifications of each of these models, either in the level-level or log-level form present results coming from six regressions, with terroir variables being added in a sequence. Details regarding the specifications of these regressions are discussed in  Subsection 3.5.2.1.         	 83	3.5.2. Discussion  3.5.2.1. General comments about regressions   For each of the two model specifications: level-level and log-level, six different regressions were implemented. In both cases the first regression is always a regression with price as a dependent variable, with the following independent variables: wine age, wine age squared, sales years, variety (24 indicator variables), brand (33 indicator variables), and alcohol content (two indicator variables). In the case of regressions 2-6, terroir variables are being added in a sequence to check how the model behaves.  Therefore: 1. Regression 2 has all the same variables as regression 1 plus soil (two indicator variables);  2. Regression 3 has all the same variables as regression 2 plus row direction (four indicator variables);  3. Regression 3 has all the same variables as regression 2 plus aspect (eight indicator variables);  4. Regression 4 has all the same variables as regression 3 plus average elevation (three indicator variables);  5. Regression 5 has all the same variables as regression 4 plus distance to lake (three indicator variables);  6. Regression 6 has all the same variables as regression 5 plus temperature bucket variables. The discussion about results that follows in the next subsection concentrates on results obtained from the full model (as per regression 6 described above). 	 84	Table 3.11. Level-level model. SE clustered on proposed sub-appellations (15).      (1) (2) (3) (4) (5) (6)   price price price price price price year_2014 -2.961*** -2.971*** -3.163*** -3.132*** -3.231*** -2.523**  (0.336) (0.335) (0.347) (0.298) (0.275) (0.751) year_2015 -2.818*** -2.806*** -3.037*** -3.014*** -3.075*** -2.327*  (0.369) (0.379) (0.423) (0.358) (0.357) (0.895) soil well-suited  1.731 2.43 3.928* 4.524* 1.972+   (1.130) (1.632) (1.720) (1.760) (1.101) rows NS   -0.796 -2.577+ -2.682* -2.319*    (0.738) (1.315) (1.166) (0.974) aspect S    0.251 1.833 6.145+     (3.287) (3.513) (3.212) avgelev (200m-400m]     -2.32 -3.779*      (1.834) (1.570) april<11C      0.304*       (0.110) april>19C      0.893**       (0.271) june>29C      0.294+       (0.158) july<25C      -0.624+       (0.330) july>33C      -0.542**       (0.166) august<24C      -0.458**       (0.142) october<10C      -0.404*       (0.168) may>11C      -0.617*       (0.210) june>15C      -0.776**       (0.240) july>18C      0.311*       (0.117) august<11C      -0.409**       (0.134) Constant 15.17*** 14.46*** 13.43*** 13.11*** 12.08** 30.40**   (1.896) (2.095) (1.994) (2.644) (3.464) (10.120) N 6785 6785 6785 6785 6785 6785 R-sq 0.677 0.68 0.686 0.701 0.706 0.751 adj. R-sq 0.674 0.677 0.683 0.698 0.702 0.747 Standard errors in parentheses      + p<0.10, * p<0.05,  ** p<0.01,  *** p<0.001     SE clustered on 15 sub-appellations.      These are results obtained after controlling for variety (24), brand (33) and year (5) fixed effects.  Comparison Groups: Soil: moderately well-suited, Rows: EW, Aspect: E,   Elevation: [0-200m], Heat: middle interval for each month.    Alcohol above 14%, Lake distance [67-700m], Elevation [0-200m]. Only results that yielded significant estimates in Model 6 are presented here.    Full results can be seen in Appendix B: Chapter 3, Table B.4. Wineries/brands were coded for privacy.  	 85	Table 3.12. Log-level model. SE clustered on proposed sub-appellations (15)        (1) (2) (3) (4) (5) (6)   lnprice lnprice lnprice lnprice lnprice lnprice year_2014 -0.168*** -0.168*** -0.174*** -0.173*** -0.179*** -0.160***  (0.017) (0.017) (0.018) (0.014) (0.013) (0.018) year_2015 -0.162*** -0.161*** -0.168*** -0.167*** -0.169*** -0.151***  (0.018) (0.018) (0.019) (0.015) (0.014) (0.021) rows NS   -0.0126 -0.08 -0.0829* -0.0817*    (0.028) (0.051) (0.033) (0.028) avgelev (200m-400m]     -0.197** -0.229**      (0.066) (0.057) april<11C      0.0104*       (0.005) april>19C      0.0321**       (0.010) july>33C      -0.0187**       (0.006) august<24C      -0.0115+       (0.006) october<10C      -0.0167*       (0.006) may>11C      -0.0237**       (0.007) june>15C      -0.0270*       (0.010) july>18C      0.00957+       (0.005) august<11C      -0.0147*       (0.006) Constant 2.747*** 2.723*** 2.681*** 2.758*** 2.776*** 3.219***   (0.089) (0.099) (0.087) (0.100) (0.098) (0.263) N 6785 6785 6785 6785 6785 6785 R-sq 0.745 0.747 0.751 0.768 0.78 0.815 adj. R-sq 0.742 0.744 0.749 0.766 0.777 0.812 Standard errors in parentheses      + p<0.10, * p<0.05,  ** p<0.01,  *** p<0.001      SE clustered on 15 sub-appellations      These are results obtained after controlling for variety (24), brand (33) and year (5) fixed effects.   Comparison Groups: Soil: moderately well-suited, Rows: EW, Aspect: E,    Elevation: [0-200m], Heat: middle interval for each month.     Alcohol above 14%, Lake distance [67-700m], Elevation [0-200m], Heat: middle interval for each month.  Only results that yielded significant estimates in Model 6 are presented here.    Full results can be seen in Appendix B: Chapter 3, Table B.5. Wineries/brands were coded for privacy.           	 86	3.5.2.2. The overall fit of the models  The results coming from regressions with the full specification (Model 6) show that the model with the logarithmic transformation of the dependent variable proves to have a better fit than the model in the level-level form, with the adjusted R2=0.81, versus adjusted R2=0.75 in the case of the model with the dependent variable in the level form. The discussion about the significant regression estimates for each of these models is presented below.  3.5.2.3. Estimates on wineage and wineage squared  While the signs on the estimate for the variable wine age that controls for the age of wine have the expected positive signs (in the case of both models, level-level and log-level) because older wines tend to be sold with a price premium due to their scarcity and an investment potential, the estimates are insignificant at all conventional significance levels. The estimates on the wine age squared also have expected signs (negative), but they are also insignificant at all conventional significance levels. I included the wine age squared in the regressions to control for the possible nonlinearities that the age of wine could have on the price of wine. I expected its negative sign, as its role was to correct and adjust the value of the estimate on the wine age. The obtained results might suggest that the age of wine may not be an essential element in the formation of prices of BC VQA wines.  3.5.2.4. Estimates on the sales years  The sales years (five indicator variables) were included in the regressions to control for time trends. Two out of five years, years 2014 and 2015 yielded significant estimates. The significance of these estimates differs per model type, e.g., the estimate in the level-level model shows that year 2014 has a negative estimate that is significant at 1% and year 2015 has a negative estimate that is significant at 5%. In the case of the log-level 	 87	model, both years 2014 and 2015 have negative estimates that are highly significant (at 0.1%). The negative signs on these years were expected, as these are the years when the change in the wholesale pricing model came to life. The change in the pricing model has been discussed earlier in Chapter 2 of this dissertation, with some more details on this topic presented in Appendix B: Chapter 3 (Figure B.3). The new pricing model changed the method for the wholesale wine pricing in the province of BC. This, in turn, introduced changes in the level of wholesale prices for the BC VQA wines. In the modelling process of this chapter some adjustments were pursued e.g.: the Provincial Sales Tax of 10% was taken off the prices in years 2011–2013 to make them more levelled with the 2014–2015 prices, but it is not possible to trace what other changes in the wholesale prices of BC VQA wines were made, as prior to 2014 the pricing model was highly dependent on various, not apparently available to the public, discounts given to different wine market players in BC (or taxes assigned to various wine classes).  3.5.2.5. Estimates on wine varieties  The regressions estimates on the wine variety differ regarding signs and significance (on per variety basis) when compared to the base variety, the Gewürztraminer. Such results were largely anticipated. The white wine varieties like Gewürztraminer, for example, tend to be sold at a discount when compared to red wine varieties due to the perceived lower investment potential for white wines and a customary association of red wines with superior quality wines. The statistically significant estimates are present on 15 (in the case of both model specifications) out of 24 varieties (total). Table 3.13 below shows all statistically significant results on wine varieties.  Full results can be seen in Appendix B: Chapter 3, Tables B4 and B5.       	 88	Table 3.13. Grape/wine varieties significant estimates.   price lnprice BACO NOIR 6.216*** 0.306***  (0.869) (0.033) CABERNET FRANC 7.573* 0.352**  (2.800) (0.103) CABERNET SAUVIGNON 13.51*** 0.434***  (2.388) (0.069) EHRENFELSER 11.54* 0.412**  (4.916) (0.114) GAMAY NOIR 3.459+ 0.07  (1.676) (0.100) MARECHAL FOCH 3.675*** 0.148**  (0.857) (0.038) MERLOT 3.921** 0.181**  (1.036) (0.047) PINOT BLANC -2.906* -0.231**  (1.202) (0.069) PINOT NOIR 5.266** 0.238***  (1.284) (0.055) RIESLING 3.408* 0.110+  (1.497) (0.053) SANGIOVESE 14.29** 0.656***  (4.643) (0.114) SYRAH 6.632** 0.299**  (2.173) (0.099) TEMPRANILLO 3.667 0.136*  (2.231) (0.057) TREBBIANO -3.025*** -0.190***  (0.618) (0.026) VIOGNIER 2.876** 0.116*  (0.867) (0.049) ZWEIGELT 6.881* 0.262**   (2.324) (0.077) Standard errors in parentheses  + p<0.10, * p<0.05,  ** p<0.01,  *** p<0.001 SE clustered on 15 sub-appellations  Comparison Grape/wine variety: Gewurztraminer.    	 89	Out of the 16 wine varieties that have significant estimates, only five varieties belong to the group of white wines (Ehrenfelser, Pinot Blanc, Riesling, Trebbiano, and Viognier). As expected all red varieties (with the statistically significant estimates) have positive signs on their regression estimates in comparison to Gewürztraminer. From the white varieties with statistically significant estimates, two varieties have negative signs on their estimates in comparison to Gewürztraminer: Pinot Blanc and Trebbiano. The other white wine varieties with statistically significant estimates Viognier, Riesling, and Ehrenfelser, have positive estimates in comparison to Gewürztraminer. The statistical significance of estimates for the 16 out of 24 varieties that come from the estimated models might suggest that the grape variety is an important variable in the pricing of BC VQA wines produced in the Okanagan and Similkameen Valleys. Another observation that can be concluded from the obtained results may suggest that the “exotic-sounding” wines tend to have higher values on their estimates, e.g., Sangiovese or Ehrenfelser. This might be associated with the low planting acreage of these grapes in BC and therefore scarcity of BC VQA wines that are made from these grapes. Or, it can simply suggest that consumers enjoy foreign-sounding wine varieties, perceive them as unique, and are willing to pay a price premium for such wines. This behaviour, in turn, might be well known to BC VQA producers and they might use this knowledge to increase the price premia on such “exotic wines.”44  3.5.2.6. Estimates on brand  The estimates on brands are primarily an empirical exercise, as it could not be predicted from theory what results would be obtained. The brand recognition in wine depends on many elements, and the quality of wine alone may not be the most important factor. Some of the most important factors influencing the recognition of the wine brand in the market include: a longevity of the brand in the market, with older brands having more chances to be valued higher; individual winery brand marketing skills (promotion strategies and advertisement channels); brand-specific taste of wines, volume of wines sold (e.g., a 																																																								44 Certain winemakers from BC confirmed that the “exotic-sounding wines” are priced at a premium. 	 90	strong presence in the liquor stores that can suggest to consumers their recognition, hence superiority); a presence at hospitality venues e.g., wineries, hotels, restaurants, etc. that allow brand recognition via tourism. All these elements can mutually reinforce brand recognition in the market and, therefore, the valuation of wines produced by that brand. Regarding the significance of the regression estimates on wine brands, the obtained results show that out of the 33 brands: 1.    In the case of the level-level model, nine brands show significant results, 2.    In the case of the log-level model, 12 brands show significant results.  The significance of estimates on brands varies per brand, as well as per model (level–level vs. log-level). The signs on brand estimates differ, too, when compared with the base brand, WINERY 22. This outcome was expected, as brands have different levels of recognition in the market and are associated with various locations, as well as a “different winery experience.” Some of the brands/wineries present in this data set are estates with well-known restaurants or estates that are frequently visited by tourists. Others are very active promoters of their wines. Detailed results on the estimates for brands are presented in Table 3.14 below.  The general conclusion coming from the obtained estimates on wineries/brands suggests that the winery/brand recognition effects seem to constitute an essential element in the formation of prices of BC VQA wines. The obtained statistically significant estimates on brands are also rather significant in their magnitudes. Such results were anticipated, as individual brand recognition tends to constitute an essential element in the formation of wine prices.        	 91	Table 3.14. Brand significant estimates.     price lnprice WINERY 2 -10.58* -0.418**  (4.209) (0.117) WINERY 4 27.69*** 1.121***  (2.894) (0.118) WINERY 6 -7.707* -0.368**  (2.679) (0.122) WINERY 7 -6.121 -0.267+  (3.666) (0.127) WINERY 9 -9.406+ -0.449*  (4.751) (0.156) WINERY 10 17.65*** 0.606***  (3.573) (0.121) WINERY 13 -5.933+ -0.230+  (2.953) (0.115) WINERY 14 -6.601* -0.286*  (3.059) (0.122) WINERY 15 3.911* 0.194*  (1.673) (0.085) WINERY 21 3.202 0.187*  (1.999) (0.076) WINERY 26 4.005 0.361*  (4.829) (0.162) WINERY 30 15.34*** 0.692***   (3.215) (0.109) Standard errors in parentheses  + p<0.10, * p<0.05,  ** p<0.01,  *** p<0.001  SE clustered on 15 sub-appellations  Comparison Winery/brand: Winery 22  Wineries were coded for privacy.           	 92	3.5.2.7. Estimates on alcohol  The estimates of the alcohol levels: alcohol below 12% and on alcohol of [12-14%], in comparison to the alcohol level of [14% and up] are insignificant. This is true in the case of all six model specifications for both level-level and log-level models. This might suggest that the alcohol content in the case of BC VQA wines may not be a major factor in the formation of wine prices.  3.5.2.8. Estimates on terroir variables  The estimates on terroir variables are significant for the analysis of this chapter that aims to verify what the influences of terroir elements are on the pricing of BC VQA wines. All estimates on terroir variables used in the modelling process of this chapter are discussed below. 1.    Soil: The estimate on the well-suited soils (in comparison to moderately well-suited soils) is positive and significant at 10% in the case of the level-level model type but is insignificant in the case of the log-level. The outcome when the well-suited soil has a positive impact on wine pricing was expected as soil that is well suited for grapes cultivation should have a positive influence on the quality of grapes (when compared to the moderately well-suited soils) and consequently on the quality of the wine.  2.    Rows: The estimates on the row direction are all insignificant, except the NS row direction that in the case of both models (level-level and log-level) have a negative and significant estimate (at 5%), in comparison to the EW row direction (base group). This result is a bit puzzling as much of the literature points towards the NS direction of rows as a superior in the Northern Hemisphere when compared to all other row directions. Since the NS row direction is considered superior for grape cultivation, it should positively influence the quality of grapes and therefore the prices of wines made from such grapes. The obtained results show that it is not true in the case of BC VQA wines as the NS row direction yielded a negative estimate when compared to the EW row direction. The reasons for such status quo might be associated with the 	 93	perception that the EW row direction is superior. Most of the row installations in the 1970s and 1980s in California opted for that direction and perceived it as superior for the quality of grapes. Later, this perception changed, and now many grape growers claim that there is no reason to consider the EW rows direction as a superior for grape cultivation in the Northern Hemisphere (Greenspan, 2008). As the field interviews with BC winemakers revealed, numerous BC winemakers take a lot of knowledge about cultivation of grapes from California. It is possible that following California’s example, BC winemakers consider the EW row direction as superior for grape cultivation and value it more in the wine pricing. This, in turn, might influence the negative sign in the estimates on the NS row direction in the model when compared to the EW row direction. 3.    Aspect: All estimates on the aspect are statistically insignificant, except the S aspect in the case of the level-level model that is positive and significant at 10%. The direction of this estimate agrees with the expectations as south-facing vineyards are considered to be those that can produce higher quality grapes. But, as the results show, aspect seems to be largely an insignificant variable for the wine pricing of BC VQA wines.  4.    Average elevation: The estimates on the average elevation are significant only in the case of the average elevation between (200–400m]. The sign is negative when compared to the elevation [0–200m] and significant at 5% in the case of the level-level model, and significant at 1% in the case of the log-level model. This may suggest that a lower elevation produces in the Okanagan and Similkameen Valleys better quality grapes. Grapes grown at lower elevations can mature on time, and there is a diminished risk for the occurrence of lower temperatures that are associated with higher elevations. This, in turn, could translate into a better quality of wines that come from grapes grown at lower elevations.  5.    Lake: The estimates on the distance to a lake came out insignificant in all models’ specifications, suggesting that the distance to a lake in the case of BC VQA wines has no influence on the formation of wine prices. One reason for this status quo might be that after it was controlled for climate (as per variable “heat” below) the proximity of the lake that could potentially mitigate climate influence on the 	 94	grapevines lost its significance. Since all BC vineyards are equipped with irrigation systems, the access to water that could also potentially impact the significance of the “lake” variable lost its potential. 6.    Heat: The specifics of the construction of the heat variable that shows the frequency of the temperature occurrences within a given temperature bucket make estimates on the heat variable a bit more complicated to interpret. Therefore, a more detailed analysis of these estimates is pursued. The estimation results from both model specifications, the level-level and the log-level, are presented in Table 3.15. The discussion regarding these estimates is outlined below.  Table 3.15. Heat significant estimates.     price lnprice april<11C 0.304* 0.0104*  (0.110) (0.005) april>19C 0.893** 0.0321**  (0.271) (0.010) june>29C 0.294+ 0.0109  (0.158) (0.007) july<25C -0.624+ -0.015  (0.330) (0.012) july>33C -0.542** -0.0187**  (0.166) (0.006) august<24C -0.458** -0.0115+  (0.142) (0.006) october<10C -0.404* -0.0167*  (0.168) (0.006) may>11C -0.617* -0.0237**  (0.210) (0.007) june>15C -0.776** -0.0270*  (0.240) (0.010) july>18C 0.311* 0.00957+  (0.117) (0.005) august<11C -0.409** -0.0147*   (0.134) (0.006) Standard errors in parentheses  + p<0.10, * p<0.05,  ** p<0.01,  *** p<0.001  SE clustered on 15 sub-appellations  Comparison group: middle heat interval for each month. 	 95	The heat variable is composed of two groups of temperature buckets:  1. The maximum temperature frequency buckets.  2. The minimum temperature frequency buckets.  In the case of both bucket groups, a middle-temperature interval maintains a comparison group for that class of heat variables (and is excluded from the regression to avoid perfect collinearity). For example: In the case of the maximum temperature buckets: two temperature buckets for October are used in both models (level-level and log-level): the bucket with the frequency of the occurrence of temperatures in the interval (-∞, 10°C) is called in the regressions: October <10, and the bucket with the frequency of the occurrence of temperatures in the interval (18°C, +∞)- is called in the regressions: October >18. The comparison interval [10°C, 18°C] is omitted from the regressions to avoid perfect collinearity. The obtained regressions results show significant estimates only in a fraction of used temperature buckets. In the case of the maximum temperature frequency buckets, the significant results are observed for temperature buckets in April, June, July, August, and October. In the case of the minimum temperature frequency buckets, the significant results (in the case of both models, level-level and log-level) are for temperature buckets in May, June, July, and August. The magnitudes in the statistically significant estimates of temperature buckets are rather low, especially when compared to estimates obtained on brands or wine varieties. Nevertheless, these results suggest that the best growing conditions in terms of temperature for grapes in July in the Okanagan and Similkameen Valleys are in the interval of [25°C, 33°C], and temperatures that are above 33°C have a negative influence on grapevines and therefore a negative impact on wine prices. This result agrees with the literature that claims that there are negative effects of high temperatures on the quality of grapes. In the presence of high temperatures, grapes go through a heat stress and accumulate sugar at a higher rate, which in turn may lead to the higher alcohol contents for wines that are made from such grapes. This, in turn, may negatively affect the wine production process, wine quality, and wine prices.  	 96	3.5.2.9. Comparison of general results from regressions 1-6 (as described in the Subsection 3.5.2.1.)   Some interesting comments arise when I compare in sequence the results of all six regressions (either in the level-level or the log-level form), as presented in Tables B.4 and B.5 in Appendix B: Chapter 3. The explanatory power of regression 1 that regresses the price of wine (with price of wine in either the level-level or the log-level form) on the following independent variables: sales year, variety, brand, and alcohol shows an adjusted R2=0.67 (in the case of the level-level model) and R2=0.74 (in the case of the log-level model). Comparing these results to the regression 6, with the full model specification, including all terroir variables, with the adjusted R2=0.75 (level-level model) and R2=0.81 (log-level model), it can be noticed that the explanatory power of the volume, brand, and variety is rather high. It could suggest that the influence of terroir elements in the wine pricing of BC VQA wines might be relatively modest.  3.6. Robustness Checks  Following on results obtained from the analysis pursued in Chapter 2, Subsection 2.5.5 that showed that certain BC companies/wineries had high market shares in the BC VQA wine market, additional analysis is introduced in this section. The role of this analysis is to address concerns that market power in the BC wine market could influence results obtained in hedonic regressions (as per main specifications of this chapter that are presented in Section 3.5.2 above).  To pursue these “market power” robustness checks, I constructed two types of dummy variables for a company’s significance in the BC market to control for potential market power. I called these dummies “capacity.” Then, I included each of these two types of dummy variables separately in each of the regressions from main specifications of this chapter.    	 97	3.6.1. Construction of capacity dummy variables to control for possible market power  3.6.1.1. Capacity dummy variable 1:  I constructed this capacity dummy variable in the following way: In the first step I calculated the individual winery’s total sales revenue in years 2011–2015. I did this for each of the 33 wineries that were present in my data set. In the next step I assigned to each of these 33 wineries either the value of 1 (if the winery was among the top five market players in BC by sales revenue between 2011 and 2015) or the value of zero otherwise. In my data set there was only one winery that was assigned a value of 1 meaning that there was only one winery that belonged to the group of top five market players in BC.  3.6.1.2. Capacity dummy variable 2:  I constructed this capacity dummy variable in the following way: In the first step I calculated the individual winery’s total sales revenue in years 2011–2015. I did this for each of the 33 wineries that were present in my data set. In the next step I assigned to each of these 33 wineries either the value of 1 (if the winery was among the top 10 market players in BC by sales revenue between 2011 and 2015) or the value of zero otherwise. In my data set there were only three wineries that were assigned a value of 1 meaning that there were only three wineries that belonged to the group of top 10 market players in BC.  3.6.2. Results of the use of “market power dummy variables” and regressions  In the first set of robustness check regressions, I added to my main models (level-level and log-level) a dummy variable that controls for the fact that a winery was among the top five market players in BC in years 2011–2015. As I explained in subsection 3.6.1 	 98	above, this dummy variable was constructed based on a winery’s total sales revenue between 2011 and 2015. All other explanatory variables remained the same as was the case in the main specifications of this chapter. Similarly, I clustered SE on 15 proposed sub-appellations. For details, please refer to Table B6 in Chapter 3: Appendix B.  In the second set of robustness check regressions, I constructed another dummy variable, but this time I controlled for the fact that a winery was among the top 10 market players in BC in years 2011–2015 (also based on a winery’s total sales revenue between 2011 and 2015). All other explanatory variables remained the same as was the case in the main specifications of this chapter. I also clustered SE on 15 proposed sub-appellations. For details, please refer to Table B7 in Chapter 3: Appendix B).  In the second set robustness check regressions, I constructed another dummy variable, but this time I controlled for the fact that winery belonged to top 10 market players in BC in years 2011-2015 (also based on winery’s total sales revenue between 2011-2015). All other explanatory variables remained the same as it was the case in the main specifications of this chapter. I also clustered SE on 15 proposed sub-appellations. For details, please refer to Table B7 in Chapter 3: Appendix B).  3.6.3. Robustness checks results  The results obtained in these robustness checks closely match the results that were obtained in the main specifications of this chapter (as per Tables 3.11 and 3.12 above). In both cases, the estimates on dummies for potential market power due to large capacity (either capacity dummy variable 1 or capacity dummy variable 2) are insignificant suggesting that the inclusion of these dummies had no impact on an intercept for prices.   Overall, the conclusions regarding the influence of terroir variables on the pricing of BC VQA wines are the same as were presented in the main specifications of this chapter (as per analysis presented in Section 3.5 above). 	 99	For full results coming from these additional regressions, please refer to Tables B6 and B7 in Appendix B: Chapter 3.  Additionally, Figures B17–B36 in Appendix B: Chapter 3 show comparison of scatters of distribution of terroir characteristics between wineries that ranked among the top five or top 10 biggest producers versus other wineries from my data set. These scatters show that there are no visible relationships between specifics of terroir and a winery’s membership in the top five or top 10 biggest market players. The distribution of terroir specifics between different wineries does not show any distinctive patterns that could suggest that wineries belonging to the top five or top 10 group of market players in BC source their grapes from specific terroir from which no other wineries from my data sample are allowed to source their grapes.   3.6.4. Robustness checks limitations  These market power robustness checks have their limitations that are associated with the size of the available data set. The data set used for this analysis did not allow for inclusion of all interactions, as there was only one winery that ranked among the top five biggest market players and only three wineries that belonged to the top 10 market players in BC. Therefore, the available data set did not allow the full control for possible market power.  3.7. Conclusion  In this section I present conclusions, discuss research limitations, and form recommendations for further studies in this area. Specifically, in Subsection 3.7.1 I outline research limitations and form conclusions, and in Subsection 3.7.2 I make recommendations for further research.   	 100	3.7.1. Conclusions  The analysis pursued in this chapter sheds some light on relationships between terroir and wine pricing for BC VQA wines produced by the estate wineries located in the Okanagan and Similkameen Valleys of BC. This research constitutes the first scholarly attempt and analysis of this type coming from the BC wine region. The results of the empirical analysis of this chapter point towards grape variety and wine brand as two important variables influencing prices of BC VQA wines from the Okanagan and Similkameen Valleys. Another observation suggests that “exotic-sounding” wine varieties seem to obtain higher price premia.   The obtained results also indicate that while there exist some terroir variables that show significant results, many terroir variables yielded insignificant estimates. This fact may suggest that terroir has limited importance in the formation of wine prices for BC VQA wines.  The BC VQA wines are marketed and considered by many BC wine industry representatives and consumers as premium quality wines because to receive VQA certification they must undergo a unique accreditation process. They are also advertised and marketed with a strong emphasis on the idea of local grape origin, particular geographic location, and terroir. It is widely assumed that because of this specific treatment the BC VQA wines are of higher quality. There is also an underlying assumption that BC VQA wines might be priced with price premia based on the specifics of terroir that supplied grapes used for their production. The analysis pursued in this chapter shows that the importance of terroir variables in the pricing of BC VQA wines may not be as large as one would have expected.   This research acknowledges some limitations. The most severe study limitation arises from the fact that the available dataset consists of a non-random sample of BC VQA wines coming from 33 estate wineries. Even though considerable efforts were undertaken to include all VQA wines produced by the Okanagan and Similkameen Valleys wineries 	 101	and repeated advertisements of this research were announced, only 33 wineries provided data on their VQA wines and the origin of grapes used for their production. This research could also benefit from data on specific management practices used in the vineyards of the BC Wine Country, as well as from information regarding winemaking costs. Regrettably, such data was not available for this research.  Another limitation comes from the available pricing data set. There might be some results sensitivity associated with the change in the wholesale pricing formula that took place in 2015.  3.7.2. Recommendations  The repetition of similar research with the participation of all Okanagan and Similkameen wineries that produce BC VQA wines is advised. Such analysis could shed more light on terroir versus wine pricing relationships in the BC wine industry, bringing an extra robustness check for results obtained in this study. It could also help emerging BC wine regions (e.g., Kootenays, Lillooet-Lytton, Shuswap, Thompson Valley), the ones that will obtain new appellations, in suggesting factors that influence wine pricing in more established regions like the Okanagan Valley. These emerging regions and their winemakers could benefit from this knowledge, as it would help them in their strategic management and investment decisions. The knowledge about terroir-wine pricing dependencies is especially important in times when the new terroir-related wine policy changes, the new appellations, and sub-appellations are coming to the wine industry in British Columbia.         	 102	Chapter 4. Does VQA Certification Matter for BC-made Wines?  The purpose of the analysis pursued in this chapter is to investigate the role and importance of the BC VQA certification program. Specifically, I aim to establish what the influence of VQA certification is on the share of the average volume, the share of the average revenue, and the average price of wines produced by the estate wineries located in the Okanagan and Similkameen Valleys of BC. Therefore, in Section 4.1 I present an introduction to this chapter and research rationale. In Section 4.2 I discuss relevant literature. In Section 4.3 I present stylized facts and develop a conceptual framework. In Section 4.4 I show methodology, discuss the use of explanatory variables, outline model specification, and present estimation method. In Section 4.5 I discuss data sources and explain necessary data transformations. In Section 4.6 I present research results. Finally, in Section 4.7 I explore research limitations, form conclusions, and develop recommendations for further studies.  4.1. Introduction  Wine belongs to the group of consumer products that show a significant level of product heterogeneity. This heterogeneity reveals itself through wine’s vertical and horizontal differentiation. This is why wine marketing and sales rely on numerous strategies that are built around two main reputation-related concepts: individual (wine brand) and collective (wine region) recognition. The establishment of individual reputation to a large extent remains in the hands of particular wine producers. It is directly associated with consumers’ appreciation of the brand’s unique quality of the wine.  At the same time, the construction of collective reputation is usually set up on a broader level, bringing together and representing all wine producers in a specific geographic location (Schamel and Anderson (2003); Panzone and Simoes (2009), Landon and Smith, (1997), among many).  The concept of geographic location that connects product origin with region-specific quality or taste has previously been recognized as an essential element able to influence business profitability or economic success. It remains valid in various areas of business, 	 103	but this concept is probably most frequently associated with Geographic Indications (GI) attached to food products like Parmigiano Reggiano or Roquefort Cheese (accessed on December 5, 2017: http://ec.europa.eu/trade/policy/accessing-markets/intellectual-property/geographical-indications/). The common association of GI with food products is not the only area where the concept of geographic location finds its way into business and economics. Even Frank Underwood, one of the main characters in the Netflix original series “House of Cards,” says: “Power is a lot like real estate. It is all about location, location, location. The closer you are to the source, the higher your property value.” (“House of Cards”, Season 1, Episode 1, 2013).  As distant as they may initially seem, Underwood’s words are remarkably applicable to wine analyses as a geographic location in the case of wine is associated with the peculiarity of natural endowments of a vineyard (terroir) that supplies winemaking grapes. The connection between the uniqueness of terroir and terroir-implied grape exceptionality continues to be recognized as one of the most significant building blocks for many wine industries worldwide.  The British Columbia wine industry, which is the main research topic of this dissertation in general and this chapter in particular, also emphasizes the role of regional collective reputation in its development strategies. Historically, in the province of BC, it has been implemented via the introduction of the VQA certification program that constitutes BC’s wine appellation.45 The BC VQA appellation guarantees that wines labeled as BC VQA are made from 100% BC-grown grapes, with at least 95% of grapes coming from a stated sub-region (e.g., the Okanagan Valley, Similkameen Valley, Vancouver Island, etc.) and with at least 85% of grapes being of reported variety and vintage year. Also, to obtain VQA certification BC wines have to go through the quality-tasting panel (BCWI Website accessed on May 17, 2017: http://www.winebc.com/wines/bc-vqa). In the BC market, VQA-certified wines are perceived and advertised as superior quality products with a strong emphasis on their origin, terroir and local sourcing of grapes. The VQA 																																																								45VQA certification was officially introduced in 1990. Subsection 2.1.2 (ch. 2) has more details on this topic. The BC VQA certification is somewhat similar to appellations in the US (the American Viticultural Area (AVA)) or France (Appellation d'Origine Contrôlée (AOC)). 	 104	certification is also used for export markets where wines with this recognition represent products “made in BC”46  (BCWI website: http://www.winebc.com). Therefore, this regional recognition also helps place the BC wine region on the map of the New World wine-producing areas.   The role of the VQA program does not end when the certification is granted, and a winemaker is allowed to put a VQA indication on its wine labels. Instead, VQA certification facilitates common marketing strategies bringing supplemental and more cost-efficient group marketing opportunities for its wine producers. More importantly, it also introduces for VQA-certified wines extra marketing channels. These additional sales channels include special VQA wine stores and more recently (starting from 2015) also certain supermarkets.47   A large number of BC estate wineries produce VQA wines,48 but there exist wineries that produce only non-VQA wines, and numerous wineries produce both VQA and non-VQA wines. The choice regarding VQA certification of a specific wine or becoming a strictly VQA, non-VQA or mixed VQA and non-VQA wine-producing winery is a voluntary course of action by individual winemakers. It remains an important internal management decision that is based on a strategic, long-term development vision and it is influenced by expected benefits and costs that are associated with the adoption of this certification.    On the benefits side, VQA accreditation lends credibility to certified wines suggesting their superiority in comparison to wines that lack such recognition (non-VQA wines). This advantage of the VQA-certified wines could find its marketing realization in two forms: an increase in the wine price and an increase in the volume of sales.  																																																								46Exports of Canadian and BC wines remain low. Canada remains the 27th biggest world wine exporter in terms of value of wine exports. It is likely that this regional recognition is currently useful in export markets mainly in the case of icewines. This class of wines constitutes the most popular wine exports group in Canada. More details on this topic can be found in Chapter 2 of this dissertation. 47The policy that allows sales of BC VQA wines in certain supermarkets officially came to life on April 1, 2015. More details on this topic are presented in Chapter 2, Subsection 2.3.2 of this dissertation. 48BC VQA wines are also produced by virtual brands. The analysis in this chapter concerns only BC VQA wines produced by estate wineries. VQA wines produced by virtual brands are excluded. 	 105	Both the increased price and the increased volume of sales can be associated with consumer demand for VQA wines, that lies to the right of consumer demand for non-VQA wines.  The first possible source of VQA benefits, the one related to price premium, can accrue as a direct result of an association of VQA wines with quality products. Because of the requirement for a quality tasting component, VQA wines are advertised and perceived as superior-quality products in the flavour of a classic wine quality terminology used by sommeliers.49 This element, in turn, can positively influence the prices for these wines, putting them in the price interval that they would not be able to achieve if VQA certification was not obtained. The ability to charge a price premium on VQA wines that are based on the underlying wine sensory and quality characteristics might also positively influence a winemaker’s brand esteem. This, in turn, can bring additional long-term marketing advantages associated with the recognition of such wines as higher-quality products and a brand itself as the one supplying superior wines.  The second source of VQA benefits that can be linked either to an increased price or an augmented volume of wine sales can come about because of the identification of VQA wines as products that are locally sourced and made. This association tends to be helpful as a marketing tool in the “buy local” advertising campaigns. The positive influence of VQA certification on wine prices that is related to local sourcing of grapes is relatively straightforward to understand as it can be due to passing on costs of a more expensive input to consumers.50 Simply, if the locally sourced grapes required to produce VQA wines incur additional expenditures to VQA winemakers, these costs can be passed on to consumers in the form of higher wine prices.   																																																								49Certain BC winemakers that were consulted during field interviews claimed that the mandatory VQA tasting is highly skewed towards New World wines in the sense that the VQA tasting panel favours wines that possess characteristics of New World wines. 50The higher costs of the locally sourced grapes may come because of input scarcity associated with the limited availability of specific grape varieties or lack of land suitable for grapes production, for example. 	 106	Linking VQA wines with locally sourced inputs can also have a positive influence on the volume of sales of these wines. The underlying logic is also straightforward. Wine consumers might prefer locally made wines and buy them more often and/or in higher volumes. Their willingness to buy locally sourced and made wines can be an effect of personal preferences that can be additionally enforced by the “buy local” marketing campaigns. This element on its own could positively influence the volume of wine sales, but in the case of BC VQA wines, it is additionally supported by the presence of VQA-particular wine marketing channels (VQA wine stores and more recently also certain supermarkets). Due to a higher number of sales channels, VQA wines can benefit from the higher volumes of sales. Therefore, these two sources of possible increase in the volume of sales either individually or jointly can benefit the winemakers that adopt VQA certification.  Switching to the costs side of VQA certification, the production of BC-made wines is associated with two main costs: the cost of VQA certification and the cost of the primary input in wine production, namely the cost of grapes.51 While VQA wines are obliged to bear both costs, the production costs of non-VQA wines does not include the cost of VQA certification. In 2016, the cost of VQA certification included: the cost of VQA registration (about CAD 10/tonne of grapes used to produce VQA wines), the cost of SKU registration as a VQA wine (about CAD 110 per SKU), and the annual fee for inspection purposes (about CAD 65 overall). To put it all in a winery perspective, the estimated annual cost of VQA certification of 5000 wine cases (11 SKUs) in 2016 was at the level of about CAD 2000 (as per field interviews with BC wineries pursued in March 2016). As the evidence shows, the cost of VQA certification in BC is not particularly prohibitive, implying that there are no obvious administrative barriers to entry into this certification scheme.   																																																								51Of course, there are some winery specific costs associated with its business model like marketing costs, labour, etc. This subsection aims to shed some light on costs of main input (grapes) and VQA certification only, leaving other winery specific costs out of this discussion. 	 107	The second and likely more significant expenditure associated with VQA certification is linked to the cost of the primary input, the cost of locally sourced grapes. Since VQA certification requires that grapes used for the production of VQA wines must be locally sourced (e.g.: 100% BC-grown grapes, with at least 95% of grapes coming from a stated sub-region and with at least 85% of grapes being of reported variety and vintage year), it immediately suggests that the cost of such input might depend on its demand and supply at any given point of time. Knowing that BC estate wineries do not always source their grapes from their estates, but also purchase them from local grape growers, this might suggest that the cost of grapes might constitute a significant factor in the winemakers’ decision to VQA-certify its wines.52   The BC VQA and non-VQA wines can be produced from estate-grown grapes, contracted grapes (long-term contracts between grape growers and wineries) or from grapes that are traded in the BC market in any given year, via posts on the website of the BC Grape Growers’ Association, for example, but not exclusively (Buy & Sell section: http://www.grapegrowers.bc.ca/grapes accessed on April 1, 2017).53   These three different supply sources might bring different costs to both VQA and non-VQA winemakers in any given year. The costs of grapes might also be winery-specific and might depend on a winery’s internal business model. For example: in the province of BC some winemakers produce wines only from grapes that are estate-grown; others produce wines from both estate-grown and contracted grapes; and there are those that produce wines only from purchased grapes. The cost of grapes is also varietal and terroir 																																																								52This suggests that different grape varieties can sell at different prices. The same grape variety, but sourced from different vineyards can sell at different prices. This can be associated with the scarcity of each variety in any given period and location. Additionally, there might be certain sub-regions that produce better grapes than others. This might additionally influence prices of grapes, hence overall costs of the VQA certification. 53The non-VQA wines might also include wines “Cellared in Canada” that are made from a mix of BC and foreign grapes (or juice). While the possibility of the production by the estate wineries of wines “Cellared in Canada” is acknowledged, it is more likely that these types of wines are produced mainly by virtual brands. These virtual brands are sometimes owned by the actual estate wineries. For example, Mission Ridge is a virtual brand belonging to Mission Hill (http://johnschreiner.blogspot.ca/2010/08/mission-hill-and-its-alter-ego-and.html). In the process of data preparation, particular attention was given to make sure that wines “Cellared in Canada” were not included in the data set used in this chapter. Therefore, unless estate wineries that produced non-VQA wines sell wines “Cellared in Canada” under their estate winery brand name, the “Cellared in Canada” wines were excluded. 	 108	dependent meaning that there might be grape varieties that are more expensive due to their varietal scarcity (low planting acreage, for example) or terroir shortages (lower acreage on terroir that produces higher-quality grapes).54   While the significance of interplays between demand and supply of locally sourced grapes and their influence on the production costs of BC winemakers might remain a major factor in VQA adoption, these fundamental economic “push-pull” interactions can be partially mitigated by the long-term input supply contracts between winemakers and grape growers. The field interviews with BC winemakers verified that such contracts exist in the BC wine region.   The average prices for BC grapes on a per variety basis are available from the annual survey report funded by the BC Grape Growers’ Association and BCWI. While these reports outline an overall level of grape prices in the province of BC, they do not present the actual costs of grapes that are incurred by individual wineries. Table C.1 in Appendix C: Chapter 4 shows average prices for certain BC-grown grape varieties for the year 2015.   After accounting for all possible benefits and costs, the BC winemakers decide in favour of the VQA certification when the expected benefits of VQA adoption outweigh expected costs. Intuitively, the size of these advantages and costs will depend on many factors that are winemaker-particular, like the long-term business development plan, individual brand building strategy, period that a winemaker has been present in the market, winemaker’s production capacity, winemaker’s estate location, or access to locally supplied grapes, to name a few.    																																																								54Some such locations mentioned by winemakers included Golden Mile Bench and Black Sage. Also, during field interviews, the BC winemakers stated that grapes that are being used for blends tend to be maintained with less care. This might make such grapes cheaper than those used for single varietal wines. 	 109	In the analysis pursued in this chapter, I aim to verify if and how the VQA certification influences the average volume and average revenue of wine sales and if there exists a price premium on BC VQA wines. Of particular interest in this chapter is to find an answer for the following research question:  What is the average impact of VQA certification on the average volume, average revenue and average price of wines produced by the estate wineries from the Okanagan and Similkameen Valleys of British Columbia?  To my best knowledge to date, there has only been one known scholarly publication that attempted to estimate the influence of VQA recognition on the prices of BC wines: “Does VQA Certification Matter? A Hedonic Analysis” written by Danielle Rabkin and Timothy Beatty in 2007. The analysis pursued in this publication used the standard hedonic pricing method on the BCLDB wine pricing data set for years 2002–2004. The results of this research showed that there existed a price premium on BC VQA wines.   There are a couple of reasons that set the research question and analysis of this chapter as an interesting empirical exercise.  First, the BC wine industry is currently undergoing a set of significant policy changes and is aiming for the establishment of new appellations (four) and sub-appellations (16).  The analysis regarding the influence of VQA certification on the average volume, average revenue, and average price for BC wines could bring some insights into the possible future effects that the introduction of new appellations and sub-appellations might have on the BC wine industry. The results of this research could also point towards some important factors that influence winemakers’ decisions to adopt VQA certification. This, in turn, could suggest what the future impact of the upcoming policy change would be on the BC wine industry if goals for the new appellations and sub-appellations mimicked those set by the VQA example.   	 110	Also, the research pursued in this chapter will add to the rich literature on the adoption of third-party certifications in the food and beverage industries. This highly active field of research in industrial organization is still somewhat low in analyses that concern third-party certification of wine55.  The rest of this chapter is organized in the following way: Section 4.2 discusses relevant literature. Section 4.3 presents stylized facts and a conceptual framework. Section 4.4 shows methodology, discusses the use of explanatory variables, and outlines model specification and estimation method. Section 4.5 shows data sources and explains necessary data transformations. Section 4.6 presents results. Section 4.7 explores research limitations, forms conclusions, and develops recommendations for further studies.  4.2. Literature Overview  There are two main literature streams that build the theoretical basis for the analysis pursued in this chapter.  The first stream is associated with literature that analyzes product certification.  A particularly relevant publication is by Alain de Janvry, Craig McIntosh and Elisabeth Sadoulet (2015): “Fair Trade and Free Entry: Can a Disequilibrium Market Serve as a Development Tool?” In this publication, the authors analyze fair trade certification in the coffee market. They show that fair trade certification in a competitive market is unlikely to benefit coffee producers as the current system allows complete arbitrage and rent dissipation due to over-certification of coffee. The situation in the coffee market as described by the authors of this publication resembles the status quo in the BC VQA wine market where a large part of wine producers VQA-certifies their wines. Also, similarly to the fair trade coffee certification, the BC VQA certification, due to relatively inexpensive certification costs also facilitates over-certification. The BC winemakers compete among themselves in the market, possibly enforcing the dissipation of the VQA rent. The obvious analogy between both certification schemes makes this publication an important 																																																								55A possible exception is organic versus standard wine, and usually with an emphasis on relationships between such certification and price. 	 111	source of inspiration for the analysis pursued in this chapter. It also helps set the basis for the development of the conceptual framework that is presented in Section 4.3 below.  The second stream of relevant literature comes from wine-specific scholarly research that investigates wine pricing and concerns the interactions between wine pricing, wine sales, and regional, collective reputation. This stream of literature delivers an important message as to the possible role that the BC VQA certification could have on wine pricing and wine sales for BC-made wines. Among numerous publications on this topic, one of the most relevant for this analysis is by Luca A. Panzone and Orlando M. Simoes. In 2009, the authors analyzed the importance of regional and local origin in consumers’ choice of Portuguese wines. The results of their analysis show that consumers are willing to pay a premium for wines coming from the recognized and reputable regions, but the recognition of appellation per se (e.g., AOC on the label) does not bring a price premium. Only the interaction between the region and the AOC allows for a price premium on these wines. They also established that there are regional differences in the contribution of the AOC recognition to wine prices. Landon and Smith (1997) analyzed how consumers use quality and reputation indicators for Bordeaux wines in the formation of their willingness to pay for these wines. Their research shows that reputation has a substantial impact on the consumers’ willingness to pay for wine, with the long-term reputation being more important than short-term quality adjustments. The authors also established that the collective reputation has as significant an impact on the consumers’ willingness to pay for wine as individual reputation. Noev (2005) analyzed the Bulgarian wine industry, concluding that wine quality, regional and varietal reputation influence wine prices. The author also shows that the regional reputation in the Bulgarian wine industry strongly depends on the regional wine specialization.  Costanigro, McCluskey, and Goemans (2010) analyzed a joint effect of the wine-specific name, individual reputation (wine brand), and collective wine region’s reputations. Their results show that a relative importance of reputation differs with the change in wine 	 112	prices. Specifically, the premia associated with reputation move from collective to individual reputation as the price of wine increases. Alessandro Corsi and Steinar Strom (2013) estimated the hedonic pricing function for Piedmont wines from the supply side. The authors used the Heckman correction method for sample selection bias to establish the price premium for organic wines in Piedmont. Their results show that after correcting for the sample selection bias, there exists a price premium on organic wines coming from the Piedmont region of Italy.  Finally, Danielle Rabkin and Timothy Beatty (2007) supplied an analysis that is directly associated with the research pursued in this chapter. Their publication: “Does VQA Certification Matter? A Hedonic Analysis” researched the BC wine industry and the significance of VQA recognition on prices of BC VQA wines. Their results show that VQA certification has a positive impact on wine prices and consumers are willing to pay a price premium for VQA recognition, but VQA certification is less important for expensive wines.  4.3. Stylized Facts and Conceptual Framework  The purpose of this section is twofold: to present a set of stylized facts that show pricing and sales volume of BC-made wines and discuss some supporting anecdotal evidence obtained during field interviews, and to develop a conceptual framework for the empirical analysis that follows. Specifically, in Subsection 4.3.1 I outline some basic pricing and sales statistics for BC VQA and non-VQA wines produced by the estate wineries from the Okanagan and Similkameen Valleys and sold in the BC market from 2011 to 2015. In Subsection 4.3.2 I develop a conceptual framework that I use as a theoretical basis for the empirical analysis pursued in this chapter.     	 113	4.3.1. Stylized facts  The available data set of all VQA and non-VQA wines from the Okanagan and Similkameen Valleys estate wineries that were sold in the province of BC in years 2011–2015 paints an interesting picture. While it could be expected that VQA-certified wines would be priced at a premium due to VQA certification and association of these wines with locally sourced and higher-quality products, it does not seem to be the case. As two histograms of average prices for VQA and non-VQA wines show (please refer to Figures 4.1 and 4.2 below), the price distribution for both wine types is almost identical. It suggests that for years 2011–2015 there was not much of a difference in pricing of both VQA and non-VQA wines.    While the pricing histograms seem to send a clear message that VQA certification in BC may not influence wine pricing, the situation looks different regarding the volume of wine sales. Figure 4.3 below shows the volume of sales for BC VQA and non-VQA wines for years 2011–2015. It is visible that the volume of sales of VQA wines is much higher than the volume of wine sales for non-VQA wines. This suggests that VQA certification in BC has a positive influence on the volume of wine sales. Some additional pricing histograms can be seen in Chapter 4: Appendix C, Figures C1–C4.  Figure 4.1. Distribution of average prices for VQA wines (red and white),  2011-2015.  0.01.02.03.04.05.06.07.08Density0 10 20 30 40 50 60 70 80 90 100 110 120Average Price in CAD $Source: Own calculations based on the BCLDB sales data.SKU# = 2104VQA Wines from the Okanagan and Similkameen Valleys (Estate Wineries Only)'Distribution of Average Prices for VQA Wines, 2011-2015	 114	Figure 4.2. Distribution of average prices for non-VQA wines (red and white),  2011-2015.   Figure 4.3. Mean volume of wine sales, 2011-2015.    This rather interesting picture of pricing and volume of sales of BC VQA and non-VQA wines produced by the estate wineries from the Okanagan and Similkameen Valleys is additionally supported by information obtained from wineries during field interviews.  None of the consulted wineries claimed that VQA certification allowed them to price their wines higher in comparison to non-VQA wines. At the same time, all these wineries mentioned two main reasons for VQA adoption: the ability to sell a higher volume of wines thanks to additional marketing channels for VQA wines and some brand-building 0.01.02.03.04.05.06.07Density0 10 20 30 40 50 60 70 80 90 100 110 120 130Average Price in CAD$Source: Own calculations based on the BCLDB sales data.SKU# =1346Non-VQA Wines from the Okanagan and Similkameen Valleys (Estate Wineries Only)Distribution of Average Prices for Non-VQA Wines, 2011-2015139.638551.186141.312528.31128.955504.616128.815466.496140.136464.7060100200300400500600Mean of Volume in Litres2011 2012 2013 2014 2015NON-VQA VQA NON-VQA VQA NON-VQA VQA NON-VQA VQA NON-VQA VQASource: Own calculations based on the BCLDB sales data.VQA and Non-VQA Wines from the Okanagan and Similkameen Valleys (Estate Wineries Only)Mean Volume of Wine Sales, 2011-2015	 115	benefits from the VQA common marketing program. Some other reasons for VQA adoption that were sometimes mentioned were: a belief that VQA certification was bringing some winemaking standards to the BC wine industry and that VQA certification was giving some recognition for BC wines in export markets. When asked about the differences in wine production between VQA and non-VQA wines, the wineries claimed that their production process did not differ and both VQA and non-VQA wines were produced with the same attention to grape sourcing and wine quality. The most common reason that wineries gave to explain why they did not VQA-certify some (or all) of their wines was a small volume of wine production.  The winemakers explained that if they produced a low volume of a certain wine type, they were able to sell such wine via their winery store and did not need access to VQA marketing channels. This is why they were not VQA-certifying such wines.  While the initial data analysis and anecdotal evidence obtained from wineries during field interviews (as described above) suggest that VQA certification might have a significant influence on the volume of wine sales but a negligible (or non-existent) impact on wine pricing, a more thorough analysis of this issue is necessary. The conceptual framework outlined in Subsection 4.3.2 below is the first step in this process. It develops a theoretical framework for the empirical analysis that follows.  4.3.2. Conceptual framework  As mentioned earlier, the main goal of the analysis in this chapter is to find out if BC VQA certification influences the average volume, average revenue, and average prices of wines produced by the estate wineries located in the Okanagan and Similkameen Valleys of BC.  It is suspected that the role of VQA certification in BC has evolved over the years since its introduction in 1990. As it is rather unlikely that the VQA system stayed unchanged, it is expected that its introduction initially brought to the BC wine industry an unprecedented guarantee of wine origin in the form of an official certification for locally produced wines. This, in turn, translated into an increased demand for VQA wines. Knowing that in the short run the supply of grapes could not be increased, as grapes normally require three to five years to produce fruits suitable for winemaking, it is likely 	 116	that the initial growth in demand for VQA wines positively influenced the prices of these wines. This brought about a price premium for VQA-certified wines. This situation is visualized in Figure 4.4 below.   Figure 4.4 shows the BC wine market just after the introduction of VQA certification. Initially, due to a comparatively small supply of locally grown grapes required for VQA certification, the volume of VQA wines remained low, resulting in a high price premium for VQA wines.  Figure 4.4. VQA price premium before industry expansion.         Where:  PVQA = price of VQA wines  PNon-VQA = price of Non-VQA wines  QVQA = quantity of VQA wines  Qmax = quantity of all BC made wines  	 117	However, since 1990, the BC wine industry grew in both the number of wineries and the supply of locally sourced grapes. Keeping in mind that the costs of VQA certification in BC are relatively small and there are no barriers to entry into VQA certification, with time the VQA certification system produced a high supply of VQA wines.  Figure 4.5 below shows the current situation in the BC wine industry where, with time and a growing supply of local grapes and VQA-certified wines, the additional supply caused a movement down the demand curve, reducing the price premium on VQA wines. Concurrently, the costs of locally grown grapes increased. Consequently, the price premium on VQA wines has been driven down to the unit cost differential between VQA and non-VQA wines.  Figure 4.5. VQA price premium after industry expansion.     Where:  PVQA = price of VQA wines  PNon-VQA = price of Non-VQA wines  QVQA = quantity of VQA wines  Qmax = quantity of all BC made wines  	 118	Therefore, the current situation in the BC wine market suggests that VQA rents that were initially present in the BC wine market have been fully dissipated. The perfectly elastic demand for non-VQA wines (flat blue line on Figures 4.4 and 4.5) is constructed because of an assumption that BC non-VQA wines are forced to compete with an inflow of highly competitive wine imports. Therefore, the demand is flat at the world price.  The central hypothesis that motivates the empirical analysis that follows is that VQA certification in BC currently earns a negligible price premium. This is similar to what was presented in the publication about fair trade certification in the coffee market, discussed in Section 4.2. At the same time, because of the expansion of the BC wine industry that is driven by rent dissipation, it is hypothesized that VQA wines observe higher volume and possibly higher revenue share. It is also expected that wineries will continue to enter the VQA program due to the relatively low costs of entry to this certification. It is also unlikely that wineries will be switching from VQA certification to non-VQA as the costs of certification are relatively small and sunk.   4.4. Methodology, Empirical Model Specification and Estimation Method  This section provides an overview of methodology that I used in the empirical analysis of this chapter, economic theory that rationalizes the choice of this method, choice, and construction of variables and empirical model specification. Specifically, in Subsection 4.4.1 I make general comments on the selection and rationale for the modelling approach; in Subsection 4.4.2 I explain empirical model specification; and in Subsection 4.4.3 I outline details regarding the choice of explanatory variables.     	 119	4.4.1. Methodology  The choice of the methodology used in the analysis of this chapter was dictated by the specifics of the research question, available data, and anecdotal evidence obtained from BC wineries during field interviews. Because the research question and data used to answer it suggested a presence of an endogenous dummy variable (VQA certification dummy), correction for this endogeneity issue from the beginning of the modelling process became one of the most important matters that required particular attention. Therefore, to mitigate the problem with the endogenous dummy variable I chose the three-stage procedure described by Woolridge (2010, Subsection 24.4.1, page 937) as a proper modelling approach. The three-stage procedure belongs to instrumental variable (IV) procedures, where fitted values obtained from the estimation of a binary response model for treatment are used in the next stages (first and second stage) of the Two Stage Least Squares (2SLS) estimation as instruments, not as regressors. This procedure itself allows a consistent estimation of the Average Treatment Effect (ATE) with usual 2SLS errors and statistics being asymptotically valid (Woolridge, 2010). Details on the specific use of the three-stage modelling approach employed in the analysis of this chapter can be seen in Subsection 4.4.2 below.  4.4.2. Empirical model specification  4.4.2.1. Primary equation  To specify the endogenous dummy variable method formally, let us assume that the structural equation (primary equation) of interest for the analysis in this chapter is of this form:                                                                       𝑦! =  𝑥!!𝛽 + 𝑉𝑄𝐴!+ 𝜀!                                                  Equation 4.1  	 120	Where: 𝑦! is either a logarithm of share of the average volume of wine sales (or logarithm of the average wine price; or logarithm of the average revenue share, respectively as 3 different specifications of dependent variable are being tested in this chapter);   𝑥!! is a matrix of observable, explanatory variables that include: § Alcohol content (continuous variable), § Sweetness (indicator variable), § Variety (indicator variable), § Color (indicator variable), § Reserve (indicator variable), § Winery capacity (indicator variable), § Sub-appellations (indicator variable), § Winery age (indicator variable).  VQAi is an indicator variable that controls for VQA certification, where VQA =1 if wine is VQA-certified and zero otherwise. The estimate on this variable is of highest interest for this research, β is a vector of regression estimates, εi  is an error term of this regression.  The problem in Equation 1 is that the VQAi variable is an endogenous dummy variable that necessitates the special correction procedure as described in Woolridge, 2010. Further details concerning the three-stage procedure are described in Subsections 4.4.2.2 and 4.4.2.3 below.       	 121	4.4.2.2. Step 1 of the three-stage endogenous dummy variable estimation procedure- the binomial probit  Let us suppose that the Equation 4.2 below represents the binomial probit regression used in stage 1 of the three-stage endogenous dummy variable procedure:                                                                    𝑧!∗ =  𝑤!! 𝛾 +  𝑢!                                                                Equation 4.2  Where: 𝑧! = 1 𝑖𝑓 𝑉𝑄𝐴 𝑖𝑛𝑑𝑖𝑐𝑎𝑡𝑖𝑜𝑛 = 10 𝑜𝑡ℎ𝑒𝑟𝑤𝑖𝑠𝑒    zi is an indicator dependent variable determining winery’s choice that is observable (choice to VQA-certify wine (VQA=1), or not to VQA-certify it (VQA=0)),  𝒘𝒊! constitutes a vector of the following observable variables (The reasons for choice of these variables and their constructions are outlined in the Subsection 4.3.4 below).  § Alcohol content (continuous variable), § Sweetness (indicator variable), § Variety (indicator variable), § Color (indicator variable), § Reserve (indicator variable), § Winery capacity (indicator variable), § Sub-appellations (indicator variable), § Winery age (indicator variable). ϒ is a vector of regression estimates, ui is an error term. The main goal in the estimation of stage 1 of this three-stage endogenous dummy variable procedure, the binomial probit, is to obtain fitted values for VQA certification that will be used in stages 2 and 3 of the 2SLS procedure as IV for VQA certification. 	 122	4.4.2.3. Stage 2 and 3- the Two Stage Least Squares (2SLS) procedure  Stages 2 and 3 of the three-stage endogenous dummy variable procedure involve a classic 2SLS IV estimation of the model presented below, with an inclusion of VQA-fitted values from stage 1 (binomial probit) as instruments for VQA certification.                                                                        𝑦! =  𝑥!!𝛽 + 𝑉𝑄𝐴!+ 𝜀!                                                  Equation 4.3  Where: 𝑦! is either a logarithm of share of the average volume of wine sales (or logarithm of the average wine price; or logarithm of the average revenue share, respectively as 3 different specifications of dependent variable are being tested in this chapter);  𝑥!! is a matrix of observable, explanatory variables that include: § Alcohol content (continuous variable), § Sweetness (indicator variable), § Variety (indicator variable), § Color (indicator variable), § Reserve (indicator variable), § Winery capacity (indicator variable), § Sub-appellations (indicator variable), § Winery age (indicator variable). 𝑉𝑄𝐴! is a fitted value for VQA from stage 1 (binomial probit). It is an IV for VQA certification. The estimate on this variable is of highest interest for this research, β is a vector of regression estimates, εi  is an error term of this regression.    	 123	4.4.3. Choice and construction of variables  As it was discussed in the introduction to this chapter, it is hypothesized that the individual winemaker’s choice to VQA-certify wine depends on an interplay of forces of supply and demand, as well as the expected costs and benefits associated with this certification. Therefore, the most important variables that influence the choice of VQA certification are associated with the winemaker’s expected costs and benefits that will arise because of subscribing to the group of VQA-certified wines. Unfortunately, data was not available on the specific costs and benefits observed by each winemaker that decides to VQA-certify its wines. Therefore, the concept of latent variable is invoked in this analysis. In due course of the model specification, it was hypothesized that there exist certain observed variables that could mimic the variables that directly represent costs and benefits coming from the choice of VQA certification and that could be used in lieu of those variables for which the measure was not available.  4.4.3.1 Stage 1 model-the binomial probit  Therefore, in stage 1 of the three-stage endogenous dummy variable estimation procedure, the binomial probit model (as per Equation 4.2 above), I chose the following variables: § WINERY AGE: This variable aims to control for the unobserved elements associated with the expected benefits and cost that the winemakers obtain from VQA certification. Wineries that have been present longer in the market might have more (or less) willingness to participate in the VQA program, as they know more about its benefits (or lack thereof). It is possible that wineries that have been in the market longer might also have easier access to the locally supplied grapes because of two reasons: 1. They have had more chances to purchase local vineyards; 2. They have had more chances to secure contracts with the local grape growers. Therefore, it is speculated that winery age influences winery choice regarding the production of VQA versus non-	 124	VQA wines in any given year. 56  This variable was constructed in the following way: Year 2015 which is the last year of the available wine data minus the year of winery establishment = winery age. Then, winery age was divided into four groups (four indicator variables):  • indicator variable 1: winery age [1932–1990),  • indicator variable 2: winery age [1990–2000),  • indicator variable 3: winery age [2000–2010),  • indicator variable 4: winery age [2010–2014].57 § SUB-APPELLATION: An indicator variable on the sub-appellation associated with the location of the estate winery. This variable aims to control for some of the estate winery location-specific elements that could be related to the costs of production of VQA wines. These elements can be linked to terroir-specific variables like the quality of soil, climate, or the sub-appellation specific know-how that can directly influence wine taste (wine quality), for example. As the wine tasting panel is required prior to the assignment of VQA certification, the natural elements of terroir might constitute an important factor in VQA adoption. In addition, it is stipulated that the location of the estate winery (based on sub-appellation) might influence the access to local grapes; hence, it might have a role in the winery’s decision to enter the VQA program. Therefore, these indicator variables account for the possible regional differences that could influence a winery’s decision to certify wine as VQA versus its resignation from such certification. The sub-appellations indicator variables used in this model follow the sub-appellations demarcation proposed in 2015 by the BC Wine Appellation Task Group. More details on the topic of proposed sub-appellations can be found in Chapter 2 of this dissertation.  § WINERY CAPACITY: The role of this variable is to put a control on the winery’s production capacity. As an actual production capacity on a per 																																																								56This logic can also hold in the opposite way, meaning that wineries that have been longer in the market might see that the VQA certification does not bring expected benefits and might withhold from the VQA certification of their wines. It is also possible that wineries that have been longer in the market are not more likely to adopt VQA because their business model has been set on the production of non-VQA wines and they did not secure access to locally sourced grapes. 57The youngest wineries in this data set were established in 2014. 	 125	winery basis wasn’t available, a proxy variable was used. I constructed this variable by calculating the total wine sales over 2011–2015 on a per winery basis and assigning each winery to one of three groups according to the total volume of wine sales. Therefore, three indicator variables were constructed: Capacity Large (equal to 1 if winery produced [500,000; ∞) litres of wine and zero otherwise), Capacity Medium (equal to 1 if winery produced: [100,000; 500,000) litres of wine and zero otherwise), and Capacity Small (equal to 1 if winery produced below 100,000 litres of wine and zero otherwise). It is hypothesized that the volume of production that is implied by the volume of wine sales influences a winery’s decision to certify its wine as VQA or not. The field interviews with BC wineries confirmed that if a batch of wine is low in volume, wineries frequently decide not to supply such wine for VQA certification. There are two main reasons for doing so: 1. VQA certification incurs additional certification costs for wineries and if the wine batch is small in volume sometimes it is not worth certifying it; 2. If the volume of wine is low a winery is more likely to sell such a wine batch directly via its wine store, and usually there is no need for VQA certification and use of additional marketing channels (VQA stores or supermarkets).  § SWEETNESS—composed of five indicator variables characterizing sweetness of the wine (0, 1, 2, 3, and NA (where NA constitutes unspecified sweetness level)); § ALCOHOL—a continuous variable indicating wine alcohol content (in %); § RESERVE—an indicator variable that shows if wine was labelled as reserve or not; § COLOUR— an indicator variable that shows if wine was white or red; § VARIETY— an indicator variable on wine type (52 varieties); § VQA indication—an indicator variable that states if an SKU (wine) was VQA-certified or not. This variable constitutes a dependent variable in the binomial probit and shows if a winery chose to VQA-certify its SKU (wine) or not. 	 126	4.4.3.2. Stages 2 and 3 model- the 2SLS estimation  In stages 2 and 3 of the endogenous dummy variable procedure (as per Equation 4.3 above), the following variables were used: § Dependent variable:  • Step 2 Regression A: LOG_AVG_VOLUME_SHARE,  • Step 2 Regression B: LOG_AVG_PRICE,  • Step 2 Regression C: LOG_AVG_REVENUE_SHARE,  • These variables were constructed in the following way:  ü LOG_AVG_VOLUME_SHARE: constructed by formulating the ratio of the average individual volume of sales (per SKU basis) to the total industry volume of sales, and taking the logarithm of that number, ü LOG_AVG_PRICE: constructed by formulation of the weighted  average of prices (per SKU), where weight constituted all monthly sales  per SKU over the total aggregated sales for that SKU, and taking the  logarithm of that number, ü LOG_AVG_REVENUE_SHARE: constructed by formulating the  ratio of  the average revenue (per SKU) to the total revenue of the industry,  and  taking the logarithm of that number; § SWEETNESS—composed of five indicator variables characterizing sweetness of the wine (0, 1, 2, 3, and NA (where NA constitutes unspecified sweetness level)); § ALCOHOL—a continuous variable indicating wine alcohol content (in %); § RESERVE—an indicator variable that shows if wine was labelled as reserve or not; § COLOUR— an indicator variable that shows if wine was white or red; § VARIETY— an indicator variable on wine type (52 varieties); § VQA INDICATION—a variable that constitutes fitted values for VQA that were obtained in step 1 (binomial probit). It is used as an IV for the VQA certification. The estimate on this variable will show the average effect of VQA certification. 	 127	4.4.3.3. Estimation method for the three-stage endogenous dummy variable specification  The estimation of the three-stage endogenous dummy variable model was pursued in two stages using Stata 13 Special Edition (SE) software. The primary goal of the estimation of the binomial probit model in stage 1 was to obtain fitted values for VQA indication. Stages 2 and 3 consisted of the 2SLS regression estimation that included VQA-fitted values obtained in stage 1, and treated them as IVs for VQA certification to correct for the VQA endogenous dummy variable bias.  4.5.  Data Sources and Data Transformation  This section presents data sources and descriptive statistics for the data set used in the analysis of this chapter. Specifically, in Subsection 4.5.1 I discuss data sources, and in Subsection 4.5.2 I outline data descriptive statistics.  4.5.1. Data sources  There were two main data sources used for the empirical analysis of this chapter: 1. The BCLDB wholesale pricing data set for all wines sold in the province of BC between April 1, 2011 and March 31, 2015.58 This data set maintains the core data set for all specifications of the empirical modelling process, as described in Section 4.4 above. The full available data set is composed of the scanner data of all wines sold in BC between 2011 and 2015 (domestically produced and imports). From this data set, all BC-made and BC-bottled wines (VQA and non-VQA) were extracted. For each of the extracted wines, the following variables were available:  § Winery (brand) name,  																																																								58The measure of the wine age (vintage) was not included as an explanatory variable in this analysis. The reason is that for the majority of SKUs (wines) the vintage year was missing in the data set obtained from the BCLDB, and it was not possible to recover it. 	 128	§ Grape/wine variety, § Wine colour, § Alcohol content (in %),  § Sweetness (scale 1-7 plus, N/A-not stated), § Monthly volume of sales (in litres and units (0.75l bottles)),  § Wholesale price of wine, § Year of sales,  § Indication if wine was a VQA or a non-VQA.   2. The self-collected (from online sources) information on the year of the establishment of the estate wineries. This data on the age of wineries was collected from winery websites, articles on wine, and John Schreiner’s blog (http://johnschreiner.blogspot.ca, accessed on February 5, 2017). While searching for the age of a winery, it was assumed that the age of the winery was calculated from the year it stated that its estate winery became operational. In case such information was lacking, the first vintage year was assumed as the year when the estate winery started to exist in the market.59  4.5.2. Data descriptive statistics  As I described in the previous sections, in the analysis pursued in this chapter I employ the three-stage endogenous dummy variable modelling approach and test three model specifications that are distinguished because of different dependent variables. All these empirical models use the same primary data source, as described immediately below.    																																																								59If a winery was out of business as of 2015, but its wine was still in the sales data, it was included in this analysis, and the winery age was established, as described in Subsection 4.3.4. 	 129	Main data set  For the empirical modelling of this chapter, I used the data set consisting of all red and white VQA and non-VQA wines sold by the BC wineries that possess real estate locations in the Okanagan and Similkameen Valleys. This means that I excluded all VQA and non-VQA wines made by the virtual brands (brands that did not have a physical estate location at the time of this analysis). I also eliminated from my data set all rosé, organic, ice wines, and all private label wines.  The final data set consists of wines (VQA and non-VQA) produced and sold by 139 different estate wineries located in the Okanagan and Similkameen Valleys of BC.   Table 4.1 below shows more details on this topic. The list of all wineries can be seen in Appendix C: Chapter 4, Table C.2.   Table 4.1. Summary statistics -part 1.  Winery type Number of estate wineries Number of associated SKU VQA only 23 423 Non-VQA only 19 326 Both (VQA and Non-VQA) 97 2701 Total 139 3450  Initially, the data set consisted of 3490 different wines (3490 different SKUs). Out of this total, 40 SKUs were removed as in various sales periods they were inconsistently listed as either VQA or non-VQA wines. The following could cause this situation: 1.    A data input mistake, 2.   Over time the wine changed its status from a non-VQA to VQA but it remained listed under the same SKU number.  After the removal of the problematic SKUs, the data set diminished from N=87,512 to N=85,986 observations (repeated monthly observations on wine sales). In terms of the number of SKUs, the total number diminished from 3490 to 3450 SKUs. Out of the total 	 130	of 3450 different SKUs, 2104 were listed as VQA SKUs and 1346 as non-VQA SKUs.  Detailed statistics of this data set can be seen in Tables 4.2 and 4.3 below.60  Table 4.2. Summary statistics -part 2.   VQA  Non-VQA      SKU Count Mean St. Dev. Min. Max. SKU Count Mean St. Dev. Min. Max. Total SKU # 2104         1346         Red wine 1118      723      White wine 986         623         Price (CAD $)                     Red wine   24.16 12.93 6.91 129.03   24.58 12.59 6.25 125.93 White wine  16.00 4.57 5.67 54.01  16.58 5.01 5.58 54.00 Unit (0.75L bottles)                   Red wine  582.00 1285.08 -10754.00 31575.00  147.96 298.95 -99.00 7159.00 White wine   756.25 1478.04 -72.00 20494.00   217.05 440.77 -173.00 15708.00 Volume (litres)                     Red wine  436.50 963.76 -8065.50 23681.25  110.97 224.21 -74.25 5369.25 White wine   567.19 1108.53 -54.00 15370.50   162.79 330.57 -129.75 11781.00 Alcohol (%)   13.27 0.96 8 15.4   13.3 0.93 9.5 15.6 Winery Age (years) 18.88 13.33 1 83   14.05 9.04 1 47 SKU (total)                   3450 N (VQA)         62075   N (Non-VQA)     23911 N (total)  85986  As Table 4.2 above shows, some unusual patterns arise.   First, it appears that non-VQA wines are priced higher on average than VQA wines. This confirms what was visible on histograms presented in Subsection 4.3.1 where the pricing distribution of VQA and non-VQA wines was almost identical. The initial descriptive statistics results of this chapter differ from the results presented in the previous research 																																																								60The negative values in Table 4.2 constitute wine returns. 	 131	on BC wines (Rabkin and Beatty, 2007) where it was stated that VQA-certified wines observed higher average prices.61 Also, when comparing VQA and non–VQA wines, the average number of units (or average volume) of wine sales is higher in the case of VQA wines. In the case of both wine classes, VQA and non-VQA, the average sales of white wines are higher than the average sales of red wines. The count of SKUs (number of different wines) is also higher in the case of VQA than non-VQA wines (in the case of both red and white wines).  Also, in the case of both wine groups, the number of red wine SKUs is higher than the number of white wine SKUs. This suggests that the product differentiation is higher in red than in white wines.  Table 4.3 below presents some additional descriptive statistics that characterize this data set. As the data shows, the most frequently observed sub-appellation associated with both VQA and non-VQA wines in this data set is the area of the proposed sub-appellation that is called “NE side lacustrine bench.” This is roughly the region of Naramata, BC. The second most frequently observed sub-appellation in the case of both VQA and non-VQA wines is the area of the proposed sub-appellation that is called “West side mixed sediments.” This is essentially the area of West Kelowna. None of these results are surprising, as both these areas are known for a high number of estate wineries. For the map of proposed sub-appellations, please refer to Appendix A: Chapter 2, Figure A.2.   The results outlined in Table 4.3 also show that the “Reserve” indication is seen only in about 9.5% of VQA and 5% of non-VQA wines. It may suggest that such recognition may not be currently important for the marketing of BC wines. Regarding wine colour, the available data set shows that in the case of both VQA and non-VQA wines, the sales of red wines are more frequently observed than the sales of white wines. This means that even though the volume of sales of white wines is higher when compared to red wines, there is a greater number of red wines (different red wine SKUs) than white wines (different white wine SKUs) observed in this data set. 																																																								61It is acknowledged that differences in wholesale prices between VQA and non-VQA wines can to an extent be caused by differences in the markup formula used when establishing prices of each wine type. 	 132	Table 4.3. Summary statistics -part 3.   Indicator Variable VQA  Non-VQA    Frequency Percent Frequency Percent Sub-appellations       Alluvial fans and flood plains 5202 8.38 1103 4.61 East side mixed sediments 2513 4.05 135 0.56 Golden Mile 6561 10.57 1062 4.44 Kettled outwash and fans 2070 3.33 1888 7.9 Lakeside alluvial fans 3113 5.01 391 1.63 Mission Creek terraces 5266 8.48 886 3.71 Mixed sediments and fans 5211 8.39 2165 9.05 NE side lacustrine bench 9394 15.13 6279 26.26 Sandy outwash lakeside terraces East side 1623 2.61 468 1.96 Sandy outwash lakeside terraces West side 1021 1.64 210 0.88 Sandy outwash terraces and deposits 6091 9.81 2202 9.21 SE side lacustrine bench 2807 4.52 2232 9.33 Similkameen Valley 2007 3.23 2214 9.26 West side lacustrine bench 1342 2.16 18 0.08 West side mixed sediments 7854 12.65 2658 11.12 Reserve        Reserve=1 5875 9.46 1227 5.13 Reserve=0 56200 90.54 22684 94.87 Color        Red 32,527 52.4 12,710 53.16 White 29,548 47.6 11,201 46.84 N(VQA)   62075   23911 N(total)   85986  For the empirical modelling of this chapter the available sales panel data set was transformed into a cross-sectional data set. The monthly observations on the volume of wine sales were averaged over individual SKUs (straight average over 2011–2015 on a per SKU basis). This process yielded a final cross-sectional data set that consisted of N=3450 observations on different SKUs (different wines). This cross-sectional data is a final data set that was used in all three stages of the empirical modelling of this chapter.  	 133	4.6. Results and Discussion  This section presents results that I obtained from the empirical analysis and discusses their overall significance. Specifically, in Subsection 4.6.1 I outline the regression results obtained in stage 1. In Subsection 4.6.2 I present results obtained in stages 2 and 3.  4.6.1. Stage 1 results  The model used in stage 1 of the three-stage endogenous dummy variable procedure, the binomial probit (as per Equation 4.2 presented above), assumes that a winery’s decision to VQA-certify its wines depends on the following set of explanatory variables:   § Winery age indicator variables (four age groups),   § Sub-appellation, area where the estate winery is located (15 sub-appellations, as per demarcation of sub-appellations proposed by the BC Wine Appellation Task Group), 62 § Winery capacity proxy (three capacity groups), § Wine specific variables from the wine sales data that constitute explanatory variables in stages 2 and 3 of this 3-stage estimation procedure: wine variety, wine colour, reserve, alcohol content, and wine sweetness.  The primary role of the binomial probit model is to provide an IV for VQA certification in the form of VQA-fitted values obtained in the post-estimation process of stage 1. The VQA-fitted values are used later in stages 2 and 3 to correct for the endogenous dummy variable problem. The results obtained from the estimation of the binomial probit (stage 1 of the procedure) that are of highest interest for this research are presented in  Table 4.4 below. The full set 																																																								62There is one more region that is not included in proposed sub-appellations. This is the Similkameen Valley. It was added to this research as an additional sub-appellation, to control for the location of wineries from that area. The official proposal for sub-appellations consists of 16 sub-appellations. Two of these sub-appellations (Valley Bottom Systems and Glaciofluvial Terraces) were not present in the available data set as none of the estate wineries were located in these sub-appellations. 	 134	of results of the binomial probit model can be seen in Tables C.3 and C.4 in Appendix C: Chapter 4.  Table 4.4. Binomial probit estimation results.   Binomial Probit VQA indication (VQA=1) Winery Age [1990, 2000) 0.266+  (0.149) Winery Age [2000, 2010) 0.746***  (0.145) Winery Age [2010, 2014) 0.879***  (0.151) East Side Mixed Sediments 0.769**  (0.242) Golden Mile 0.305*  (0.152) Kettled Outwash and Fans -0.680***  (0.149) Lakeside Alluvial Fans 0.433*  (0.182) Mission Creek Terraces 0.460**  (0.150) Mixed Sediments and Fans -0.165  (0.125) NE Side Lacustrine Bench -0.188  (0.118) Sandy Outwash Lakeside Terraces East Side -0.127  (0.207) Sandy Outwash Lakeside Terraces West Side 0.251  (0.224) Sandy Outwash Terrace and Deposits -0.186  (0.126) SE Side Lacustrine Bench -0.605***  (0.145) Similkameen Valley -0.325*  (0.130) West Side Lacustrine Bench 0.816+  (0.449) West Side Mixed Sediments 0.0811  (0.136) Capacity Medium -1.599***  (0.153) Capacity Small -2.107***  (0.154) N 3419 Standard errors in parentheses   + p<0.10, * p<0.05, ** p<0.01, *** p<0.001  Comparison groups:  Winery Age [1932, 1990), Capacity: Large  Sub-appellation:  Alluvial fans and flood plains  All results after controlling for variety, sweetness, reserve, color, alcohol content. Full results can be seen in Chapter 4: Appendix C, Table C.4.  	 135	Overall the binomial probit model correctly classifies 71.10% of observations. The obtained results show that the influence of winery age on the probability of choosing VQA certification tends to be significant. The estimates on all age groups of the wineries are positive and significant in comparison to the group of the oldest wineries that were combined in the age interval [1932, 1990). These estimates are significant either at a 10% significance level in the case of the wineries from the interval [1990, 2000) or a 0.1% significance level in the case of wineries placed in age intervals [2000, 2010) and [2000, 2014]. The obtained results suggest that younger wineries are more likely to VQA-certify their wines when compared with the base group, the wineries from the age group [1932, 1990). This situation might be caused by the fact that with time the VQA certification program became more prevalent and easy to access as the supply of local grapes increased. These results agree with the theory developed in Subsection 4.3.2 (Conceptual Framework) where it was hypothesized that with time the entry to VQA certification became easier as the supply of locally grown grapes increased.  The probit results also show that winery capacity is a major factor in a winery’s choice to VQA-certify its wines. As described in Subsection 4.3.4 above, winery capacity is controlled for via a proxy variable (total volume of wine sales in years 2011–2015 on a per winery basis). The wine sales were used here as a proxy variable for capacity because actual production capacities were not available. The estimates on both capacity indicator variables—winery capacity medium and capacity small—yield negative estimates when compared with the base group, winery capacity large. The results are strongly significant at 0.1%. These results suggest that wineries with large capacity are more likely to adopt VQA certification. This is hardly a surprise. Wineries with higher wine sales (and production) are bigger, and it is likely that they have better access to local grapes, as they might own more local land. When a winery is being established, it usually plans and decides on its production possibilities. This, in turn, is likely to be associated with the size of the winery’s vineyards or the number and size of contracts with local grape growers. The large-capacity wineries might be more willing to adopt VQA certification because they have higher production volumes and therefore need access to extra marketing channels for their wines (e.g., VQA wine stores or, recently, the Save-On-	 136	Foods supermarkets). The anecdotal evidence obtained from wineries during field interviews also agrees with the results achieved in this specification.  The results of the binomial probit model also show that the location of the estate winery that in this model is controlled by the sub-appellations indicators is not always significant in the winery’s choice to VQA-certify its wines. These results are not surprising, as it was anticipated that a decision to adopt VQA certification might depend on the location of the estate winery and associated terroir.63 Such decision might come about because of sub-appellation-dependent differences in the availability of land suitable for grape production.    4.6.2. Stage 2 and 3 results  In stages 2 and 3 of the endogenous dummy variable modelling, the 2SLS IV procedure (Stata command: ivregress 2SLS), three different dependent variables were implemented: the share of the average volume of wine sales, the share of the average revenue and the average price of wine (Regressions A, B and C, as seen below). Except for different dependent variables, all three 2SLS IV regression types used the same second stage instrument for VQA certification (the fitted values of the VQA dummy variable obtained in stage 1, the binomial probit) as well as the set of the same explanatory variables: wine variety, wine colour, reserve, alcohol content, wine sweetness, winery age, proxy for winery capacity, and indicator variables on proposed sub-appellations (based on the location of the estate winery). All regressions used the full cross-sectional data with N=3450 observations on individual SKUs (wines).  The estimate of the highest importance and interest for the analysis in this chapter is the estimate on the VQA indication as it estimates the average effect of VQA certification. The full results associated with each regression (A, B, and C) can be seen in Appendix C: Chapter 4 (Tables C.5, C.6, and C.7).    																																																								63Terroir in the meaning of grape quality and availability, as both terroir-specific (sub-appellation-specific) quality and quantity of grapes might influence VQA adoption. 	 137	Regression A results: Dependent variable -logarithm of the average volume share  The obtained results show that after controlling for the endogeneity of the VQA certification, there exists a positive and significant (at 10%) influence of VQA certification on the average volume of sales for BC wines produced by the estate wineries from the Okanagan and Similkameen Valleys of BC.  The results also show that alcohol level and sweetness N/A have a negative and significant (at 0.01%) influence on the average volume of wine sales.   Similarly, winery capacity medium and small (in comparison to winery capacity large) have negative and highly significant (at 0.01%) impact on the average volume of wine sales.   The results also prove that the age of winery has a significant impact on the average volume of wine sales. Younger wineries (in comparison to wineries from the age interval [1932, 1990)) all show positive and significant (at 0.01%) estimates.  Regarding estimates on sub-appellation dummies, the results are mixed regarding sign and significance. Such results on these regional dummy variables could be expected, as it was anticipated that the volume of wine sales is region-specific and depends on many elements, of which one of the most important would be sub-appellation-specific availability of agricultural land suitable for grape cultivation.  Overall, the results obtained in this specification agree with expectations, the theory developed in the conceptual framework (Subsection 4.3.2), and anecdotal evidence obtained from BC wineries during field interviews.   The results of the highest interest for this chapter that were achieved in stages 2 and 3 of the three-stage endogenous dummy variable procedure with the logarithm of the average 	 138	volume share of wine sales as dependent variable are presented in Table 4.5 below. The full results can be seen in Chapter 4: Appendix C, Table C5.  Table 4.5. Regression A, 2SLS first and second stage results. Dependent variable: logarithm of the average volume share.  First stage Second Stage   logarithm average volume share logarithm average volume share VQA Indication  0.655+   (0.349) VQA probability 1.015322***   (0.00)  Sweetness N/A 0.0049971 -0.414***  (0.039) (0.116) Alcohol -0.0034282 -0.155***  (0.011) (0.032) East Side Mixed Sediments -0.0143846 -0.039  (0.061) (0.183) Golden Mile -0.0023718 -0.385**  (0.042) (0.126) Kettled Outwash and Fans -0.0049115 0.311+  (0.053) (0.162) Lakeside Alluvial Fans -0.0108049 0.564***  (0.054) (0.160) Mission Creek Terraces -0.0009846 -0.0566  (0.046) (0.139) Mixed Sediments and Fans -0.0128443 0.236*  (0.038) (0.116) NE Side Lacustrine Bench -0.0015179 0.470***  (0.036) (0.109) Sandy Outwash Lakeside Terraces East Side -0.0108636 0.476**  (0.061) (0.183) Sandy Outwash Lakeside Terraces West Side -0.0066681 0.188  (0.069) (0.207) Sandy Outwash Terrace and Deposits -0.0044024 0.167  (0.039) (0.117) SE Side Lacustrine Bench 0.0007355 -0.275+  (0.050) (0.151) Similkameen Valley -0.0081657 0.155  (0.044) (0.132) West Side Lacustrine Bench -0.0075759 -0.344  (0.080) (0.241) West Side Mixed Sediments -0.0106201 -0.419*** 	 139	Table 4.5. Regression A, 2SLS first and second stage results. Dependent variable: logarithm of the average volume share.  First stage Second Stage   logarithm average volume share logarithm average volume share  (0.038) (0.115) Winery Age [1990, 2000) 0.0037946 0.461***  (0.034) (0.104) Winery Age [2000, 2010) 0.0007705 0.653***  (0.043) (0.129) Winery Age [2010, 2014) -0.0054993 1.177***  (0.049) (0.147) Capacity Medium 0.0043983 -1.237***  (0.044) (0.130) Capacity Small 0.0125084 -1.875***  (0.062) (0.183) Constant -0.7498799 -12.24***   (0.466) (1.376) N 3365 3365 R-sq 0.24 0.28 adj. R-sq 0.23 0.26 Standard errors in parentheses    + p<0.10, * p<0.05, ** p<0.01, *** p<0.001   Comparison groups:   Winery Age [1932, 1990), Capacity: Large. Sub-appellation:  Alluvial fans and flood plains. Sweetness: Sweetness 0,  Above results come after controlling for wine variety, reserve and wine color.  Full results can be seen in Chapter 4: Appendix C, Table C5.                 	 140	Regression B results: Dependent variable -logarithm of the average price  The obtained results show that unlike in the case of Regression A above, in the case of Regression B the estimate on the VQA indication yields a negative but insignificant estimate. This suggests that VQA certification does not have a significant impact on pricing of BC wines produced by the estate wineries from the Okanagan and Similkameen Valleys of BC.  The results also show that wine colour white (in comparison to red) has a negative and significant (at 0.01%) impact on wine pricing. At the same time, alcohol level and sweetness level 3 both have a positive and significant impact on pricing of BC-made wines (significant at 0.01% and 5%, respectively).   The results obtained on winery capacity indicator variables suggest that the capacity of the winery that implies its production possibilities influence pricing of BC wines, with medium- and small-sized wineries having a positive and significant impact on wine pricing, in comparison to wineries with a large capacity (significant at 0.01% and 5%, respectively).  The results of Regression B also prove that the age of winery has a significant impact on the average price of BC wines with younger wineries having a negative and significant impact (at 0.01%) on wine prices, in comparison to wineries from the age interval [1932, 1990).   Regarding estimates on sub-appellation dummies, as was the case with Regression A above, the results are mixed regarding sign and significance. Such results on these regional dummy variables could be expected as it was anticipated that the volume of wine sales is region-specific and depends on many elements, of which one of the most important would be sub-appellation-specific availability of agricultural land suitable for grape cultivation. 	 141	Overall, as was in the case of regression A above, the results obtained in this specification agree with expectations, theory developed in the conceptual framework (Subsection 4.3.2), and anecdotal evidence obtained from BC wineries during field interviews.  The results of the highest interest for this chapter that were obtained in stages 2 and 3 of the three-stage endogenous dummy variable procedure with the logarithm of the average price of wine as dependent variable are presented in Table 4.6 below. The full results can be seen in Chapter 4: Appendix C, Table C6.  Table 4.6. Regression B, 2SLS first and second stage results. Dependent variable: logarithm of the average price.  First stage Second Stage   logarithm average price logarithm average price VQA Indication  -0.0657046   (0.084) VQA probability .9944108 ***   (0.00)  Color White 0.0038534 -0.274***  (0.029) (0.021) Sweetness 3 0.0244488 0.243*  (0.166) (0.120) Alcohol 0.0002667 0.107***  (0.010) (0.008) East Side Mixed Sediments 0.0027112 0.157***  (0.060) (0.043) Golden Mile 0.0104331 0.245***  (0.042) (0.030) Kettled Outwash and Fans 0.0025931 0.0715+  (0.053) (0.038) Lakeside Alluvial Fans 0.0079825 0.0648+  (0.053) (0.039) Mission Creek Terraces 0.0007252 0.194***  (0.046) (0.033) Mixed Sediments and Fans -0.0018092 0.0698*  (0.038) (0.027) NE Side Lacustrine Bench 0.0058863 0.155***  (0.036) (0.026) Sandy Outwash Lakeside Terraces East Side -0.0006699 0.0645  (0.060) (0.044) Sandy Outwash Lakeside Terraces West Side 0.0041342 0.542*** 	 142	Table 4.6. Regression B, 2SLS first and second stage results. Dependent variable: logarithm of the average price.  First stage Second Stage   logarithm average price logarithm average price  (0.069) (0.050) Sandy Outwash Terrace and Deposits 0.0028915 0.207***  (0.038) (0.028) SE Side Lacustrine Bench 0.0068515 0.124***  (0.050) (0.036) Similkameen Valley -0.0000881 0.149***  (0.043) (0.031) West Side Lacustrine Bench 0.0040074 0.0622  (0.080) (0.056) West Side Mixed Sediments 0.0002643 0.220***  (0.038) (0.027) Winery Age [1990, 2000) -0.0038217 -0.118***  (0.034) (0.025) Winery Age [2000, 2010) -0.0030698 -0.127***  (0.042) (0.030) Winery Age [2010, 2014) -0.0033919 -0.178***  (0.049) (0.035) Capacity Medium -0.0018829 0.129***  (0.043) (0.031) Capacity Small -0.0011772 0.107*  (0.061) (0.044) Constant -0.7788785 1.584***   (0.465) (0.328) N 3413 3413 R-sq 0.24 0.36 adj. R-sq 0.23 0.352 Standard errors in parentheses    + p<0.10, * p<0.05, ** p<0.01, *** p<0.001   Comparison groups:   Winery Age [1932, 1990), Capacity: Large. Sub-appellation:  Alluvial fans and flood plains. Sweetness: Sweetness 0. Color: Red.   Above results come after controlling for wine variety and reserve.  Full results can be seen in Chapter 4: Appendix C, Table C6.        	 143	Regression C results: Dependent variable -logarithm of the average revenue share  The obtained results show that unlike in the case of Regression A above, in the case of Regression C the estimate on the VQA indication yields a positive but insignificant estimate. This suggests that VQA certification does not have a significant impact on the average revenue of BC wines produced by the estate wineries from the Okanagan and Similkameen Valleys of BC.  The results also show that wine colour white (in comparison to red) has a negative and significant (at 5%) impact on the average revenue. At the same time, sweetness level 3 has a positive and significant (at 5%) impact on the average revenue of BC-made wines, but sweetness N/A has a negative and significant (at 0.01%) impact on the average revenue of BC-made wines.  Winery capacity medium and small (in comparison to winery capacity large) have a negative and highly significant (at 0.01%) impact on the average revenue of wine sales.  The results of Regression C also prove that the age of winery has a significant impact on the average revenue of wine sales. Younger wineries (in comparison to wineries from the age interval [1932, 1990)) all show positive and significant (at 0.01%) estimates.  The estimates on sub-appellation dummies are mixed regarding sign and significance. As mentioned in the case of Regressions A and B, such results on these regional dummy variables could be expected as it was anticipated that the volume of wine sales is region-specific and depends on many elements, of which one of the most important would be sub-appellation-specific availability of agricultural land suitable for grape cultivation.  Overall, the results obtained in this specification agree with expectations, the theory developed in the conceptual framework (Subsection 4.3.2), and anecdotal evidence obtained from BC wineries during field interviews.  	 144	The results of the highest interest for this chapter that were obtained in stages 2 and 3 of the three-stage endogenous dummy variable procedure with the logarithm of the average price of wine as dependent variable are presented in Table 4.7 below. The full results can be seen in Chapter 4: Appendix C, Table C7.  Table 4.7. Regression C, 2SLS first stage results. Dependent variable: logarithm of the average revenue share.   First stage Second Stage   logarithm average revenue share logarithm average revenue share VQA Indication  0.635   (0.521) VQA probability 1.015303***   (0.117)  Color White 0.0041057 -0.292*  (0.030) (0.134) Sweetness 3 0.0152561 1.575*  (0.166) (0.750) Sweetness N/A 0.0049815 -0.896***  (0.039) (0.174) East Side Mixed Sediments -0.0144911 0.207  (0.061) (0.273) Golden Mile -0.0023701 -0.263  (0.042) (0.188) Kettled Outwash and Fans -0.0047116 0.241  (0.053) (0.242) Lakeside Alluvial Fans -0.010542 0.688**  (0.054) (0.240) Mission Creek Terraces -0.000515 0.113  (0.046) (0.207) Mixed Sediments and Fans -0.0125779 0.0557  (0.038) (0.173) NE Side Lacustrine Bench -0.0014762 0.588***  (0.036) (0.163) Sandy Outwash Lakeside Terraces East Side -0.011025 0.620*  (0.061) (0.274) Sandy Outwash Lakeside Terraces West Side -0.0064667 0.737*  (0.069) (0.310) Sandy Outwash Terrace and Deposits -0.0044848 0.131  (0.039) (0.174) SE Side Lacustrine Bench 0.0009897 -0.258  (0.050) (0.226) 	 145	Table 4.7. Regression C, 2SLS first stage results. Dependent variable: logarithm of the average revenue share.   First stage Second Stage   logarithm average revenue share logarithm average revenue share Similkameen Valley -0.0086697 -0.000648  (0.044) (0.198) West Side Lacustrine Bench -0.0079211 -0.188  (0.080) (0.360) West Side Mixed Sediments -0.0087799 -0.395*  (0.038) (0.172) Winery Age [1990, 2000) 0.0047147 0.438**  (0.034) (0.156) Winery Age [2000, 2010) 0.002112 0.691***  (0.043) (0.193) Winery Age [2010, 2014) -0.0049955 0.862***  (0.049) (0.220) Capacity Medium 0.0039414 -1.365***  (0.044) (0.195) Capacity Small 0.0125965 -2.259***  (0.062) (0.273) Constant -0.7497545 -14.32***   (0.466) (2.058) N 3366 3366 R-sq 0.24 0.25 adj. R-sq 0.23 0.24 Standard errors in parentheses    + p<0.10, * p<0.05, ** p<0.01, *** p<0.001   Comparison groups:   Winery Age [1932, 1990), Capacity: Large. Sub-appellation:  Alluvial fans and flood plains. Sweetness: Sweetness 0. Color: Red.   Above results come after controlling for wine variety and reserve.  Full results can be seen in Chapter 4: Appendix C, Table C7.    Some additional tests for instrument relevance (for Regression A, B, and C, as described above) can be seen in Chapter 4: Appendix C, Tables C8–C10.    	 146	4.6. Conclusion  In this section I present conclusions, discuss research limitations, and form recommendations for further studies in this area. Specifically, in Subsection 4.6.1 I outline research limitations and form conclusions, and in Subsection 4.6.2 I make recommendations for further research.  4.6.1. Conclusions  The results obtained in the analysis of this chapter point towards some interesting and important findings for the BC wine industry.  First, the obtained results answer research question posed in this chapter and show that VQA certification positively influences the share of the average volume of wine sales of BC-made wines. This is good news as the VQA program has been in place since 1990 and it is reassuring to see that it has some positive impact for BC winemakers. While the results in this chapter show that VQA certification positively influences the share of the average volume of wine sales, the situation looks different in the case of VQA’s influence on the average prices of wine and average sales revenue. The obtained estimates on VQA indication in both cases were insignificant and suggest that VQA indication does not influence the average price and average sales revenue of BC-made wines.  The results that show VQA’s positive influence on average volume share and simultaneously point to the lack of influence that VQA certification has on the average price and average revenue of BC-made wines prove the correctness of theory developed in Subsection 4.3.2.   One could have expected that VQA certification that is an official BC appellation would have a positive influence on wine prices. The earlier research on this topic (Rabkin and Beatty, 2007) suggested that there existed a price premium on VQA certification of BC-	 147	made wines. The results of this chapter that account for the endogeneity of VQA certification show that such price premium does not exist.  Keeping in mind that more research on the relationships between VQA and pricing for BC-made wines might be necessary (e.g., with an inclusion of BC VQA virtual brands), the results of this chapter seem to give a clear picture of the VQA’s role.  The results from the analysis of this chapter might be important for the official introduction of new appellations (four) and sub-appellations (16) that are expected to come to life in BC no later than January 1, 2019. As the analysis of the significance of VQA indication shows, if the proposed appellations and sub-appellations adopt similar strategies to those that are currently used by VQA certification, they will possibly influence the volume of wine sales but not the prices of BC-made wines or the revenue of BC winemakers.  The analysis pursued in this chapter has some limitations. One limitation of this analysis is associated with using a proxy for winery capacity (in the form of total wine sales over the years 2011–2015), instead of an actual winery production capacity that was not available for this research. A better variable to be used in this model would be a real production capacity on a per winery basis.  Another limitation might come from the lack of other plausible choice-influencing variables that could be used as the explanatory variables in the binomial probit model. Besides the variables that were used and were possible to collect (winery age, sub-appellation and a proxy for winery capacity), the binomial probit model in step 1 could benefit from other explanatory variables that were winery-specific, e.g., some details on the winemaker’s education level.      	 148	4.6.2. Recommendations  It would be interesting to pursue similar research but on a larger data set with more sales years and with the inclusion of virtual brands of BC wine. Such analysis could outline a full evolution of the importance of VQA certification on BC-made wines. It could also show if differences exist in the role of VQA certification between virtual and non-virtual brands of BC wine.                        	 149	Chapter 5. Conclusion 	In this chapter, I revisit primary research goals set up in this dissertation and outline how the evidence presented in each chapter helped find answers to research questions described in this thesis. This chapter is a concluding chapter that I have divided into four subsections. In Subsection 5.1 I summarize research aims; in Subsection 5.2 I discuss research contributions; in Subsection 5.3 I examine research strengths and limitations. Finally, in Subsection 5.4 I outline research applications.  5.1. Research Aims  The overreaching aim of this dissertation was to pursue research on the British Columbia wine region and its wine industry, with a particular emphasis put on the significance of terroir and collective reputation in pricing and sales of locally sourced and made wines. I achieved this goal via analyses presented in three separate but interconnected chapters that constitute the core of this dissertation. Each of these chapters maintains its independence regarding the central analytical theme and research approach, but all three chapters combined shed light on the BC wine industry in its entirety. The leading reason that influenced my decision to pursue analyses regarding the BC wine region is associated with the most recent wine policy developments that aim for the introduction of new wine appellations (four) and sub-appellations (16). This industry’s turning point that per definition intends to strengthen the role of regional recognition for BC-made wines introduced an opportunity to verify the current function of BC’s terroir and collective reputation (VQA) in pricing and sales of locally made wines. I envisioned that such analysis could be used as a benchmark for the comparison of terroir and collective reputation influences on wine pricing and sales after new appellations and sub-appellations are established.   	 150	The analyses presented in this dissertation are interesting not only from a strictly academic point of view, but they can also assist the BC wine industry and local policymakers in their micro level decisions relating to the wine industry. The overview of the BC wine region and wine industry as outlined in Chapter 2 of this dissertation was used to set up a stage for the analyses pursued in Chapters 3 and 4. Therefore, in Chapter 2, I presented an analysis of the BC wine industry from the organizational, historical, and policy points of view. In the outline of Chapter 2, I placed particular emphasis on the most current wine policy developments in BC: the change in the liquor markup formula from 2015, as well as the proposal for the establishment of new appellations and sub-appellations and the industry plebiscite that followed. Based on the available scanner sales data obtained from the BCLDB, I presented statistics regarding all wine sales in BC (domestic wines and imports). Also, I outlined the types of domestically sourced wine brands found in the BC wine market during the years 2011–2015. To bring more clarity to an actual number and significance of domestic wine brands in the BC market, I estimated market shares (volume and value) for all main types of brands sold in the BC wine market. I also estimated market shares for the most significant VQA brands in terms of volume and value of wine sales that were selling wines in the BC market during the years 2011–2015.  Overall, my analysis in this chapter shows that the BC wine region is not specialized in the production of any grape or wine type. The BC wine market seems to be heterogeneous at various levels (e.g., heterogeneity of grape and wine types, a large number of wine brands, a relatively large number of estate wineries), but about 59% and 52% of the total volume and value market share, respectively, belong to just five companies. At the same time, a calculated industry concentration index (Herfindahl-Hirschman Index (HHI)) shows that the BC wine industry was characterized by a moderate level of concentration in years 2011–2013 and by a competitive level of concentration in 2014–2015.   	 151	After I outlined the status quo in the BC wine industry, I moved to the analysis of Chapter 3. The primary goal of Chapter 3 of this dissertation was to find an answer for its leading research question:  Does terroir influence the pricing of BC VQA wines from the Okanagan and Similkameen Valleys?  To answer this question, I matched scanner sales data on the selected BC VQA wines from the BCLDB wholesale scanner sales data for years 2011–2015 with micro level data collected from 33 BC estate wineries located in the Okanagan and Similkameen Valleys of BC. This allowed me to link each of the selected BC VQA wines with its actual origin, a vineyard that sourced grapes used for its production. In the next step, I collected data on the terroir specifics of each of these vineyards from Google Earth Pro (satellite images) and the Environment Canada weather database, and included in my data set and analysis the following terroir/vineyard-specific variables: soil type, row direction, aspect, average elevation, distance from vineyard to the nearest lake, and a temperature-based climate measure. Since the primary goal of this chapter was to establish what the influence of terroir variables was on the pricing of BC VQA wines, I employed the hedonic pricing method in the modelling stage of this analysis. Specifically, I regressed the price of wine on the terroir variables (as described above) and non-terroir variables available in the BCLDB pricing data set: volume of wine sales, variety, brand, alcohol content, age of wine, and year of sales.  The results of my analysis in this chapter show that terroir elements have some importance in the pricing of BC VQA wines, but they may not constitute the most significant pricing variables. The wine variety and wine brand seem to have more significance in the formation of wine prices for BC VQA wines.   In Chapter 4 of this dissertation I asked a different research question:  What is the average impact of VQA certification on the average volume, average revenue, and average price of wines produced by the estate wineries from the Okanagan and Similkameen Valleys of British Columbia? To answer this question, I also used the data obtained from the BCLDB scanner sales data set for years 2011–2015. Specifically, I employed the data on wine sales pursued by 	 152	the BC wineries located in the Okanagan and Similkameen Valleys that possess a physical estate location. The modelling process in this chapter was based on the three-stage approach, with the correction for the endogenous dummy variable (VQA certification dummy). In stage 1 of this procedure (binomial probit model), I used a control on winery capacity, winery age, and a set of indicator variables for sub-appellations (based on the estate winery location) to calculate VQA-fitted values that were used in stages 2 and 3 of the 2SLS. I estimated three different model specifications that used the same explanatory variables but differed in the dependent variables: logarithm of a share of the average volume of wine sales, logarithm of average price, and logarithm of the share of average revenue.  The results that I obtained show that while VQA certification has a positive and statistically significant impact on the share of the average volume of wine sales, it doesn’t have a significant effect on the average price of wine and the share of the average revenue of wine sales.    5.2. Research Contributions 	This research constitutes the first analysis of this type and magnitude that concerns the economics of the BC wine region. The uniqueness of the studies presented in this dissertation is a result of various elements that are associated with the particular data sets used in the empirical modelling process, the modelling approach, and the first attempt of such analysis in a young, developing, and sparsely researched wine region.   In terms of strictly scholarly contributions, the studies pursued in this dissertation contribute to three main fields: wine economics, wine business, and wine marketing. With regards to wine economics, this research adds to the stream that investigates the role of terroir and collective reputation in the formation of wine prices and wine sales. In respect of wine business and wine marketing, the empirical analyses pursued in this dissertation show the industry’s status quo outlining an overall marketing situation in the province of BC, together with types and number of brands and estimations of individual brand and wine industry market shares. This sort of analysis can be helpful for 	 153	winemakers that are already established in the market as well as for new entrants into the BC wine market, to guide them on strategies used for wine pricing and wine sales.  Specifically, in the analysis of Chapter 2, I estimated the number of brands that were present in the BC market, outlined brand division, calculated brand shares for VQA brands, estimated volume and value market shares on a per VQA label basis, and determined industry concentration index (HHI). This analysis brought a previously unseen insight into the organization and functioning of the BC wine industry from the wine business and wine marketing sides.  The empirical modelling approach employed in Chapters 3 and 4 of this dissertation is also mostly unique.  In the case of Chapter 3, I used a standard hedonic pricing methodology, but instead of the usual approach where the price of wine is regressed on various sensory or objective wine characteristics, in this dissertation I regressed the price of wine on the unique, terroir-specific variables. In the empirical modelling of Chapter 3, I used a self-constructed panel data set composed of the following data sets: wine wholesale data that constitutes all wine sales in BC in years 2011–2015 (for the selected BC VQA wines), micro level (winery level) data on the locations of vineyards that sourced grapes used in the production of these wines, the Environment Canada climate (temperature) data, and agronomic data that was self-collected from Google Earth Pro satellite images or from physical visits to the vineyards. The construction of this data set and the hedonic modelling approach allowed me to control for terroir elements that were characteristic for the vineyards that sourced grapes used in the production of each of the selected wines. Therefore, in my hedonic model, I could establish what the influence of terroir variables (and therefore an implied quality of grapes) was on the pricing for BC VQA wines.  In Chapter 4, I used the three-stage endogenous dummy variable modelling specification to estimate the influence of VQA certification on the share of the average volume of wine sales, average price of wine, and average revenue share. To the best of my knowledge, 	 154	this approach has not been used previously in estimations related to wine appellations. Therefore, it is likely that the analysis pursued in Chapter 4 also constitutes unique research.  5.3. Strengths and Limitations  The strengths and limitations of this dissertation are chapter-specific, and they have already been discussed in the “Conclusion” subsections of the proper chapters of this dissertation. Regardless, there exist the overall strengths and limitations that apply to all analyses presented in this thesis.   The most apparent strength of the research presented in this dissertation is associated with the fact that this is the first attempt of a rigorously pursued empirical analysis and modelling of this type coming from the BC wine region.  This element makes it pioneering research.  Also, the analyses pursued in this dissertation, especially their empirical modelling, are laid out straightforwardly so they can easily be reproduced elsewhere if there exists access to the necessary data.  One of the possible weaknesses of the analyses of this dissertation is associated with the nature of wine as a highly complex product in terms of production (heterogeneous terroir and production costs), consumption (consumer-specific tastes), and marketing process (various levels and options for marketing and brand building). There exists a risk of a hidden endogeneity that could influence empirical modelling and results but could not be accounted for in these analyses due to data unavailability.   Another possible weakness is associated with the lack of a random sample of wineries that provided data for Chapter 3 of this dissertation.    	 155	An additional limitation might come from the fact that the empirical results presented in this thesis could be region-specific and apply only to the BC wine region and its winemaking industry. Therefore, the interpretation of results presented in this dissertation might be contextually limited to the BC wine region.  5.4. Research Applications  The results of the research pursued in this dissertation point towards a couple of interesting implications and applications.  First, the results obtained in Chapter 3 of this thesis suggest that in the BC wine region, wine variety and wine brand are currently the two most important variables in the formation of prices for VQA wines. Also, the obtained results suggest that “exotic-sounding” varieties are priced at a price premium (e.g., Sangiovese). These results might be relevant from the perspective of BC winemakers and might suggest that variety specialization might be a “way to go” for the BC wine industry. Therefore, it is possible that the attention of the BC wine industry should be focused more on the sub-regional varietal specialization that would build on the specifics of the sub-regional terroir differences, based on their superior fit for the cultivation of particular grape varieties.   Currently, the BC wine region and its wine production resemble a buffet on a “specials night” or a potluck soirée. A wine customer that visits the BC wine region can get a wide selection of different wine types that are derived from multiple grape varieties. This status quo applies to the whole BC wine region as well as to individual wineries. Unfortunately, the potluck or buffet-like abundance rarely guarantees a quality, consistent sensory experience between the dishes. In other words, buffets usually do not feature any exceptional dishes on which they could build a reputation, but they offer a large choice of dishes instead. This analogy applies to the BC wine region; at the moment nobody associates its winemaking with any particular wine variety.    	 156	Also, the results obtained in Chapter 4 of this dissertation show that VQA certification has a positive and significant impact on the average volume share, but it does not have a significant impact on the average price and average revenue share. As I explained in Chapter 4, this situation might be associated with the issue of VQA over-certification that allows rent dissipation.   The results obtained in Chapter 4 might help the BC wine industry in understanding what has happened to VQA certification over time. This, in turn, might provide the BC wine industry and policymakers with guidance on how to properly design future wine policies related to collective reputation.                      	 157	Bibliography  Aggelogiannopoulos, D., Drosinos, E.H., and Athanasopoulos, P. (2007): “Implementation of a quality management system (QMS) according to the ISO 9000 family in a Greek small-sized winery: A case study.” Food Control, 18: 1077-1085.  Ali, H.H. and Nauges, C. (2003). Reputation and Quality Effects on Wine Prices: A   Comparison Between En Primeur and Bottled Bordeaux Wine. GREMAQ   Touluse.  Ali, H.H., Lecocq, S. and Visser, M. (2008). The Impact of Gurus: Parker Grades and En Primeur Wine Prices. The Economic Journal, 118: F158-F173.  Anderson, K. (2009). Terroir Rising? Varietal and Quality Distinctiveness of Australia’s Wine Regions. The University of Adelaide School of Economics, Research Paper No. 2009-18  Angulo, A. M., Gil, J. M., Gracia, A., & Sánchez, M. (2000). Hedonic prices for Spanish red quality wine. British Food Journal, 102(7), 481–493.   Ashenfelter, O. (2008). Predicting the quality and prices of Bordeaux wine. Economic Journal, 118(529), 174–184.   Ashenfelter, O. and Storchmann, K. (2010). Using Hedonic Models of Solar Radiation  and Weather to Assess the Economic Effect of Climate Change: The Case of  Mosel Valley  Vineyards. The Review of Economics and Statistics, 92(2): 333- 349.  Barham, E. (2003) Translating Terroir: The Global Challenge of French AOC Labeling. Journal of Rural Studies 19 (2003): 127–138   Barla, P. (2007). ISO 14001 certification and environmental performance in Quebec’s  pulp and paper industry. Journal of Environmental Economics & Management,  53, 291-306.  BC Wine Appellation Task Group. (2015). Wine Industry Turning Point. Recommended Changes to the British Columbia Wines of Marked Quality Regulations. Accessed on January 15, 2016: http://bcwinetaskgroup.ca/report/   BC Laws Website. Accessed on December 15, 2015: http://www.bclaws.ca     BCLDB Website. Accessed on December 15, 2015: http://www.bcldb.com   	 158	BC Wine Institute Website. Accessed on December 15, 2015:  http://www.winebc.com/discover-bc/okanagan-valley Benfratello, L., M. Piacenza, et al. (2009). Taste or Reputation: What Drives Market   Prices in the Wine Industry? Estimation of a Hedonic Model for Italian Premium   Wines. Applied Economics 41(16): 2197-2209.  Berard, L. and Marchenay, P. (2007). Produits de Terroir. Comprendre et Agir. Ressources des terroirs – Cultures, usages, sociétés UMR Eco-Anthropologie et Ethnobiologie Centre national de la recherche scientifique: 1-59.   Bombrun, H. and Sumner, D. (2003). What Determines the Price of Wine? The Value of   Grape Characteristics and Wine Quality Assessments. AIC Issues Brief, 18:1-6.  Bowen, P.A. et al. (2005). Use of Geographic Information System Technology to Assess  Viticulture Performance in the Okanagan and Similkameen Valleys, British  Columbia. Geoscience Canada, 34(4): 161-176.  Bramley, R. G. V, & Hamilton, R. P. (2007). Terroir and precision viticulture: Are they compatible? Journal International Des Sciences de La Vigne et Du Vin, 41(1), 1–8.  Brentari, E., Levaggi, R., & Zuccolotto, P. (2015). A hedonic price analysis for the Italian wine in the domestic market. Quality & Quantity, 49(3), 999–1012.   British Columbia Wine Appellations Task Group Website. Accessed on January 15,  2017:  http://bcwinetaskgroup.ca   British Columbia Wine Authority Website. Accessed on January 15, 2017:  http://www.bcvqa.ca   Cameron, A.C. and Miller. D.L. (2015). A Practitioner’s Guide to Cluster-Robust  Inference. The Journal of Human Resources, 50(2): 317-373.  Canadian Association of Liquor Jurisdictions Website. Accessed on January 15, 2016:  http://www.calj.org   Canadian Vintners Association Website. Accessed on December 5, 2016:  http://www.canadianvintners.com   Cardebat, J.-M., & Figuet, J.-M. (2013). Expert opinions and Bordeaux wine prices: an attempt to correct the bias of subjective judgments. AAWE, (129).  Carew, R. and Florkowski, W. (2012). Wine Industry in the Pacific Northwest: A  Comparative  Analysis of British Columbia, Washington State and Oregon.  Journal of Wine Research, 23(1): 27-45.  	 159	Cartier, L. (2014). The British Columbia Wine Industry: Can It Compete with the Big  Guys? AAWE Working Paper No. 147.  Castriota, S., & Delmastro, M. (2009). Individual and Collective reputation: Lessons from the wine market. Available at SSRN 1349992, (30), 25.   Chang,H.S., Campbell, G.,  and Sniekers, P. (2007). Assessing the Use of Geographical   Indications for the New England Wine Region in NSW. Paper presented to 51st   AARES Annual Conference 2007, Queenstown, New Zealand, 13-16 February   2007.  Combris, P., Lecocq, S., & Visser, M. (2000). Estimation of a hedonic price equation for Burgundy wine. Applied Economics, 32(8), 961–967.   Corsi, A. and Steinar, S. (2013). The Price Premium for Organic Wines: Estimating a  Hedonic Farm-Gate Price Equation. Journal of Wine Economics, 8(1): 29-48.  Costanigro, M., Mccluskey, J., and Goemans, Ch. (2010). The Economics of Nested  Names: Name Specificity, Reputations, and Price Premia. American Journal of  Agricultural Economics, 92(5): 1339-1350.  Costanigro, M., Mccluskey, J., and Mittelhammer, R.C. (2007). Segmenting the Wine  Market Based on Price: Hedonic Regression when Different Prices Mean  Different Products. Journal of Agricultural Economics, 58(3): 454-466.  Cross, R., Plantinga, A.J., and Stavins, N. (2011). What is the Value of Terroir? The  American Economic Review, 101(3): 152-156.  Court A.T. (1939). Hedonic Price Indexes with Automotive Examples. In The Dynamics  of Automobile Demand. New York: The General Motors Corporation.  Decanter, Aug 3, 2016. Accessed on December 15, 2016:   http://www.decanter.com/wine-reviews-tastings/steven-spurriers-british-columbia-wines-323102/   Deconinck, K., and Swinnen, J. (2014). The Political Economy of Geographical I ndications. AAWE Working Paper No. 174.  Defrancesco, E., Orrego, J.E, and Gennari, A. (2012). Would “New World” Wines  Benefit From Protected Geographical Indications in International Markets? The  Case of Argentinean Malbec. Wine Economics and Policy, 1(1): 63-72.  Delmas, M.A., and Grant, L.E. (2014). Eco-Labeling Strategies and Price-Premium: The  Wine  Industry Puzzle.” Business & Society, 53(1): 6-44.  	 160	Deloire, A., et al. (2005). Grapevine Responses to Terroir. A Global Approach.  International Journal of Vine and Wine Sciences, 39(4):149-162   Durrieu, F. (2008). Impact of Brand Identity on Labeling: The Case of Regional   Branding. 4th International Conference of the Academy of Wine Business   Research, Siena, 17-19 July, 2008: 1-15.  Deconinck, K., and Swinnen, J. (2014). The Political Economy of Geographical  Indications. AAWE Working Paper No. 174.  De Janvry, A. McIntosh, C. and Sadoulet, E. (2015). Fair Trade and Free Entry: Can a  Disequilibrium Market Serve as a Development Tool? Review of Economics and  Statistics, 97(3): 567-573.  Delgado, M., Porter, M.E., and Stern, S. (2012). Clusters, Convergence and Economic  Performance. NBER Working Papers Series, Working Paper 18250.  Florkowski, W.J., Carew, R., and Senhui, He. (2008). British Columbia Consumer   Preferences for Italian Wines: Reputation and Vintage Effects on Wine Quality   and Prices. Paper prepared for presentation at the 12th EAAE Congress ‘People,   Food and Environments: Global Trends and European Strategies’, Gent   (Belgium), 26-29 August 2008.  Fogarty, J. J., & Jones, C. (2011). Return to wine: A comparison of the hedonic, repeat sales and hybrid approaches. Australian Economic Papers, 50(4), 147–156.   Fogarty, J. J. (2006). The return to Australian fine wine. European Review of Agricultural Economics, 33(4), 542–561.   Galbreath, J. (2014). Does “Liquid Geography” Facilitate Cooperation or Competition? A  Study of Sub-Clusters, Knowledge Exchange, and Innovation in the Wine  Industry. AAWE Working Paper No. 158.  Gergaud, O. and Ginsburgh, V. (2008). Natural Endowements, production Technologies  and Quality of Wines in Bordeaux: Does Terroir Matter? Economic Journal,  118(529): F142-157.  Gergaud, O. and Livat, F. (2010). Collective Reputation Effects: An Empirical Appraisal.  AAWE Working Paper No. 73.  Getz, D. and Brown, G. (2006). Benchmarking Wine Tourism Development. The Case of  the Okanagan Valley, British Columbia, Canada. International Journal of Wine  Marketing, 18(2): 78-97.  	 161	Giacomarra, M. et al. (2016). The Integration of Quality and Safety Concerns in the Wine  Industry: The Role of Third-Party Voluntary Certifications. Journal of Cleaner  Production, 112:267-274.  Golan, A. and Shalit, H. (1993). Wine Quality Differentials in Hedonic Grape Pricing.  Journal of Agricultural Economics, 311-321.  Greene, W. H. (1981). Sample Selection Bias as a Specification Error: Comment.  Econometrica, 49(3): 795-798. Greenspan, M. (2008). Row Direction-Which End is Up? Wine Business Monthly, July  2008 Issue. Accessed on January 1, 2017:  https://www.winebusiness.com/wbm/?go=getArticle&dataId=58458   Gustafson, Ch. R., Lybbert, T. and Sumner, D. (2011). Consumer Characteristics,  Identification, and Hedonic Valuation of Wine Attributes: Exploring Data from  Field Experiment. RMI-CWE Working Paper Number 1102.   Harrison, B., Kelley, m., and Gant, J. (1996). Innovative Firm Behavior and Local  Milieu: Exploring the Intersection of Agglomeration, Firm Effects, and  Technological Change. Economic Geography, 72 (3): 233-258  Heckman, J.J. (1979). Sample Selection Bias as Specification Error. Econometrica, 47(1):  153-161.  Hellman, E. W. (2003). Oregon Viticulture. Oregon State University Press.  Hira, A. et. al. (2013). What makes Clusters Competitive? McGill Queens University  Press.  Huang, C. L. (1996). Consumer preferences and attitudes toward organically grown  produce. European Review of Agricultural Economics, 23: 331-342.  House of Cards, Season 1, Episode 1, 2013. Netflix Original Series.  Jones, G.V. (2004). Geology and Wine 8. Modeling Viticultural Landscapes: A GIS Analysis of the terroir Potential in the Umpqua Valley of Oregon. Geoscience Canada, 31(4).  Jones, G. V., White, M. a., Cooper, O. R., & Storchmann, K. (2005). Climate change and global wine quality. Climatic Change, 73(3), 319–343.   Justice Laws Website. Accessed on January 1, 2016:    http://laws-lois.justice.gc.ca/eng/acts/C-0.4/   Klein, B. and Leffler, K.B. (1981). The Role of Market Forces in Assuring Contractual  Performance. Journal of Political Economy, 89(4): 615-641. 	 162	 Kwong, L.M.K. et al (2011). A Semiparametric hedonic pricing Model of Ontario Wines.  Canadian Journal of Agricultural Economics, 59: 361-381.  Lancaster, K.J. (1966). A New Approach to Consumer Theory. Journal of Political  Economy, 74: 132-157.  Landon, S. and Smith, C.E. (1997). The Use of Quality and Reputation Indicators by  Consumers: The Case of Bordeaux Wines. Journal of Consumer Policy, 20: 289- 323.  Lecocq, S. and Visser, M. (2006). What determines Wine Prices: Objective vs Sensory Characteristics. Journal of Wine Economics, 1(1): 42-56.  Loureiro, M. L. (2003). Rethinking new wines: Implications of local and environmentally  friendly labels. Food Policy, 28, 547-560.  Luppe, R., & Belfiore, P. (2009). Hedonic Pricing Models and the Evaluation of Attributes: The Case of Wines from Brazil, (December), 631–652.  Luppe, R., Lopez Favero, M., and Prado belfiore, L.P. (2009). Hedonic Pricing Models   and Evaluation of Attributes: The Case of Wines from Brazil, Argentina and   Chile. ESIC Market, 09: 631-651.  Lutzeyer, S. (2008). Estimating Hedonic Prices for Stellenbosch wine, 1–23. Accessed on December 15, 2016: http://www.ekon.sun.ac.za/wpapers/2008/wp152008/wp-15-2008vers2.pdf   Malorgio, G., Camanzi, L., & Grazia, C. (2007). Effectiveness of European Appellations of Origin on the International Wine MArket. Contributed Paper prepared for presentation at the 105th EAAE Seminar ‘International Marketing and International Trade of Quality Food Products’, Bologna, Italy, March 8-10, 2007.   Mercado, L., Dalcero, a., Masuelli, R., & Combina, M. (2007). Diversity of Saccharomyces strains on grapes and winery surfaces: Analysis of their contribution to fermentative flora of Malbec wine from Mendoza (Argentina) during two consecutive years. Food Microbiology, 24(4), 403–412. Accessed on December 15, 2016:  http://doi.org/10.1016/j.fm.2006.06.005   Mullins MG, Bouquet A, Williams LE (1992). “Biology of grapevine.” Cambridge  University Press, Cambridge. 1992.  Nerlove, M. (1995). Hedonic Price Functions and the Measurement of Preferences: The  Case of Swedish Wine Consumers. European Economic Review, 39: 1697- 1716.  	 163	Noev, N. (2005). Wine Quality and Regional Reputation. Eastern European Economics,  43(6):5-30.  Oczkowski, E. (1994). A Hedonic Price Function for Australian Premium Table Wine.  Australian Journal of Agricultural Economics, 38(1): 93-100.  Oczkowski, E. (2001). Hedonic Wine Price Predictions and Nonnormal Errors.  Agribusiness,  26(4): 519-535.  Panzone, L.A. and Simoes, O.M. (2009). The Importance of regional and Local Origin in  the Choice of Wine: Hedonic Models of Portuguese Wines in Portugal.  Journal  of Wine Research, 20(1): 27-44.  Peiris, G.P. (2010). Una Primera Estimacion de Un Modelo De Precios Hedonicos Para El Vino Catalan.Departamento de Economia Aplicada, Universidad Autonoma de   Barcelona: 1-56  Priilaid, D. a., & Rensburg, P. Van. (2006). Non-linearity in the hedonic pricing of South African red wines. International Journal of Wine Marketing, 18(3), 166–182.  Accessed on January 15, 2015: http://doi.org/10.1108/09547540610704747   Rabkin, D.E. and Beatty, T.K.M. (2007). Does VQA Certification Matter? A Hedonic  Analysis. Canadian Public Policy, 33(3): 299-314.  Ramirez, C.D. (2008). Wine Quality, Wine Prices, and the Weather: Is Napa Different?  Journal of Wine Economics, 3(2): 114-131.  Rickard, B., J., McCluskey, J., and Patterson, R. (2012). Reputation Tapping. AAWE  Working Paper No. 119.   Rimerman, F. &Co. LLP Report (2013). The Economic Impact of the Wine and Grape  Industry in Canada 2011. Accessed on December 15, 2016:  http://www.canadianvintners.com/wp-content/uploads/2013/03/Canada-2011-Wine-Industry-Economic-Impact-Report-FINAL-3-7-13.pdf   Rimerman, F. &Co. LLP Report (2017). The Economic Impact of the Wine and Grape  Industry in Canada 2015. Accessed on April 15, 2017:   http://www.canadianvintners.com/wp-content/uploads/2017/03/Canada-Economic-Impact-Report-2015.pdf   Roma, P., Di Martino, G., & Perrone, G. (2013). What to show on the wine labels: a hedonic analysis of price drivers of Sicilian wines. Applied Economics, 45(19), 2765–2778. Accessed on April 15, 2017: http://doi.org/10.1080/00036846.2012.678983   	 164	Rosen, S. (1974). Hedonic prices and Implicit Markets: Product Differentiation in Pure  Competition. Journal of Political Economy, 82(1): 34-55.  Rosenbaum, P.R. and Rubin, D.B. (1983). ”The Central Role of Propensity Score in  Observational Studies for Causal Effects.” Biometrika, 70(1): 41-55.  Rossel, J. and Beckert, J. (2012). Quality Classification in Competition: Price Formation  in the German Wine Market. AAWE Working Paper No. 114.  Schamel, G and Anderson, K. (2003). Wine Quality and Varietal, regional and Winery  Reputations: Hedonic Pricing for Australia and New Zealand. The Economic  Record, 79(246): 357-369.  Schamel, G. (2003) A Hedonic Pricing Model for German Wine. Agrarwirtschaft, 52(5):  247-254.  Shapiro, C. (1982). Consumer Information, Product Quality, and Seller Reputation. Bell  Journal of Economics, 13(1): 20-35.  Skuras, D., & Vakrou, A. (2002). Consumers’ willingness to pay for origin labelled wine: A Greek case study. British Food Journal, 104(11), 898–912. Accessed on April 15, 2017:  http://www.emeraldinsight.com/doi/abs/10.1108/00070700210454622   Smith, A. (1776). An Inquiry into the Nature and Causes of the Wealth of Nations.  Accessed on January 15, 2016: http://www.econlib.org/library/Smith/smWN1.html   Soderbom, M. (2011). Lecture 9: Sample Selection Bias. Econometrics II. Department of  Economics, University of Gothenburg.  Steiner, B.E. (2009). Old World Wines Revisited: Consumers’ Valuation of Spanish and   German Wines In the UK Wine Market. Staff paper 09-03, Department of Rural   Economy, Life and Environmental Sciences, University of Alberta, Edmonton,   Canada:1-89.  Steiner, B.E. (2004). Australian Wine in the British Wine Market: A Hedonic Price Analysis. Agribusiness, 20(3): 287-307.  Szolnoki, G. (2010). Expensive and Colour Intensive –Are These the Factors of Success?   A Hedonic Liking Model Using the Example of Red Wine. Geisenheim Research   Center, Section of Economics and Market Research, 1-7.  Thrane, Ch. (2004). In Defence of the Price Hedonic Model in Wine Research. Journal of  Wine Research, 15(2): 123-134.    	 165	Tirole, J. (1996). A Theory of Collective Reputations (With Applications to the   Persistence of Corruption and to Firm Quality). The Review of Economic Studies  Limited, 63: 1-22.  The Canadian Encyclopedia: Accessed on January 15, 2016:   http://www.thecanadianencyclopedia.ca/en/  The Globe and Mail: Accessed on July 22, 2016:  https://www.theglobeandmail.com/news/national/canadas-premiers-reach- internal-free-trade-pact/article31089692/   Thornton, R. V. (2000). Understanding the Benefits of Third-Party Audits and Certification. Environmental Quality Management, 9: 99-102.  Troncoso, J. L., & Aguirre, M. (2006). Short communication. Price determinants of Chilean wines in the US market: A hedonic approach. Spanish Journal of Agricultural Research, 4(2), 124–129.  Unwin, T. (1999). Hedonic price Indexes and the Qualities of Wines. Journal of Wine research, 10(2): 95-104.  Van Leeuwen, C., & Seguin, G. (2006). The concept of terroir in viticulture. Journal of Wine Research, 17(1), 1–10. Accessed on January 5, 2017: http://doi.org/10.1080/09571260600633135  Van Leeuwen, C. et al. (2004). Influence of Climate, Soil, and Cultivar on Terroir.  American Journal of Enology and Viticulture: 55(3): 207-217.   Wilson, J.E. (1999). "Terroir: The Role of Geology, Climate, and Culture in the Making  of French Wines" University of California Press 1999.  WineLaw.ca Blog: Accessed on April 1, 2017: http://www.winelaw.ca/cms/   Winkler, A.J. et al. (1974). General Viticulture. University of California Press.  Wood, D., & Anderson, K. (2006). What Determines the Future Value of an Icon Wine? New Evidence from Australia. Journal of Wine Economics, 1(02), 141–161. Accessed on January 15, 2016: https://www.cambridge.org/core/journals/journal-of-wine-economics/article/what-determines-the-future-value-of-an-icon-wine-new-evidence-from-australia/762B1BAB2B1ACB9229FF77ECC85EBC46   Woolridge, J.M. (2010). Econometric Analysis of Cross Section and Panel Data. Second Edition. The MIT Press. Cambridge, Massachusetts.  Woolridge, J.M. (2013). Introductory Econometrics: A Modern Approach. Fifth Edition.  South-Western Cengage Learning. 	 166	 Yoo, V. Florkowski, W.J. and Carew, R. (2015). Pricing Attributes of Wines from  Emerging Suppliers on the British Columbia Market. Review of Agricultural and  Applied Economics, 2: 28-39.  Zeithaml, V.A. (1988). Consumer Perceptions of Price, Quality and Value. A Means-end  Model  and Synthesis of Evidence. Journal of Marketing, 52(3): 2-21.	 167	Appendices                                           	 168	 Appendix A: Chapter 2.  Figure A.1. The BCLDB old versus new mark-up formula.   Source: Based on the BCLDB website accessed on April 1, 2017: http://www.bcldb.com/files/Wholesale_Pricing_Changes-Overview.pdf    Old mark-up formula Prime cost + domestic charges (optional) = In Bond Cost +Excise Tax&Customs =Duty Paid Costs +Volume Markup/Distribution Charge +%Markup for Wine, Spirits  or +Volume mark-up for beer +Cost of Service Fee +Container recycling fee =LDB Retail Price +GST and PST = LDB Display Price before applicable container deposit New mark-up formula Prime cost + domestic charges (optional) = In Bond Cost +Excise Tax&Customs =Duty Paid Costs + Distribution Charge (beer only) + % Markup for Wine, Spirits, Refreshment Beverages OR + Per Litre Markup by Supplier annual production for beer +Container recycling fee =LDB Wholesale Price +GST = LDB Wholesale Price before applicable container deposit 	 169	The new wine wholesale pricing markup formula works upwards from supplier’s cost of production (winery’s prime costs) and brings following markups for wine: 89% markup on the first CAD $ 11.75/litre and graduated markup of 27% on any amount over  CAD $ 11.75/litre. Source: BCLDB website accessed on January 15, 2016: http://www.bcldb.com/files/Wholesale_Pricing_Changes-Wholesale_Customer_Presentation.pdf?v=1 ).  The old (prior to April 1, 2015) provincial wholesale wine markup was at the level of 117% on the first CAD$ 10.25/litre and  51% on the reminder cost to generate the  retail prices as seen in the government run liquor stores. From these government run liquor stores prices various discounts were offered to different retailers to come up with a wholesale price for such retailers:  1. Independent wine stores: 30% discount off the LDB display price, 2. Private liquor stores: 16 % discount off the LDB display price, 3. Rural agency stores: 10 % discount off the LDB display price, 4. VQA wine stores: 30% discount off the LDB display price, 5. Restaurants and bars: 0% discount off the LDB display price,  When the new wholesale pricing formula came to life, the provincial markup was lowered to compensate previous retailers for discounts (as seen in point 1-5 above) that were removed and replaced with the common wholesale pricing formula for all retailers.  Source: WineLaw.ca website accessed on January 15, 2015: http://www.winelaw.ca/cms/legal-info-industry/retail-distribution/298-liquor-changes-chart         	 170	Figure A.2. Proposed demarcation of sub-appellations            Source: dr. Patricia Bowen, AAFC/PARC Summerland.       	 171	Figure A.2. Proposed demarcation of sub-appellations     Source: Dr. Patricia Bowen, AAFC/PARC Summerland (used with permission).                          	 172	Table A.1. The BC Wine Appellation Task Group recommendations and plebiscite results.64 BC Wine Appellation Task Group Recommendations (revised version from April 28, 2016) Plebiscite Results 1. In order to have a winery license producers making wine from 100% BC grown grapes must become members of the BC Wine Authority (BCWA) and be subject of audits conducted and enforced by the Wines of Marked Quality regulations    APPROVED 2 a). Change the “Wines of Distinction” category to British Columbia Wines.    APPROVED 2 b). After the change, both wine types, BC VQA and British Columbia Wines will be allowed to use geographic indication on their labels   REJECTED 3. Taste panels should be put to a review by the Wine Industry Advisory Committee and should use as a reference a survey pursued by the BC Wine Appellation Task Group in the wine industry, in June 2015   NOT INCLUDED IN PLEBISCITE 4. After sub-appellations are established (not later than January 1, 2019), the BCWA should be given the authority to prohibit the use of unregulated geographical indicators on wine labels   APPROVED 5. All wines made 100% from BC grapes must register as, either BC VQA wines or British Columbia Wines   APPROVED 6. Wines of British Columbia that use geographic indication (sub-appellation) will need to show on their label region and sub-region (appellation and sub-appellation)   APPROVED 7. Four new appellations in the emerging regions (Thompson Valley, Shushwap, Lillooet-Lytton and Kootenays) should be established. Boundries of these appellations will require demarcation upon consultations in each of these regions.   APPROVED 8. The set of sub-appellations is proposed for the Okanagan Valley (for details, please refer to the Appendix). The naming of sub-appellations should include the name of town, village or historical place.   APPROVED 9. Three separate audits currently pursued by the Liquor Control and Licensing Branch, BC Liquor Distribution Branch and BC Wine Authority should be harmonized   NOT INCLUDED IN PLEBISCITE 10. BCWA should establish a flat fee for small wineries that covers cost of membership, grape levies, audits and wine certification (with threshold for definition of small winery not exceeding 50 tons)   APPROVED 11. Section 29(3)(c) of the Wines of Marked Quality regulations should be amended to: At least two thirds of the vote measured by registrants of productive wine grape acreage in a proposed    																																																								64 Note: Some recommendations were omitted from the plebiscite because they either recommended continuation of existing practices/requirements, or they were accepted by the BCWA and didn’t require industry voting. 	 173	Table A.1. The BC Wine Appellation Task Group recommendations and plebiscite results.64 BC Wine Appellation Task Group Recommendations (revised version from April 28, 2016) Plebiscite Results geographical area or subdivision, who produce at least two thirds of the total production of wine made from grapes grown in that area or subdivision, must have voted, by ballot, in favour of the proposed geographical area or subdivision; APPROVED 12 a). Section 29(3)(e) of the Wines of Marked Quality regulations should be deleted. Additional review of section 29 should be pursued by the BCWA and WIAC  APPROVED 12 b). Additional review of section 29 should be pursued by the BCWA and WIAC  NOT INCLUDED IN PLEBISCITE Source: The BC Wine Appellation Task Group Website & British Columbia Wine Authority Website, accessed on January 1, 2017: http://bcwinetaskgroup.ca/report/                                  	 174	Table A.2. Identified wine brands present in the BC market in 2011-2015.    Brand Name Town Region/Brand Classification 1 ALIGOTE N/A CANADIAN NON BC BOTTLED 2 ANDREWPELLER N/A CANADIAN NON BC BOTTLED 3 BENJAMIN BRIDGE  N/A CANADIAN NON BC BOTTLED 4 CAVE CELLARS N/A CANADIAN NON BC BOTTLED 5 CAVE SPRING N/A CANADIAN NON BC BOTTLED 6 CHATEAU DES CHARMES N/A CANADIAN NON BC BOTTLED 7 CHILL WINSTON N/A CANADIAN NON BC BOTTLED 8 CLOSSON CHASE N/A CANADIAN NON BC BOTTLED 9 CSP N/A CANADIAN NON BC BOTTLED 10 COYOTE'S RUN N/A CANADIAN NON BC BOTTLED 11 DAN AYKROYD N/A CANADIAN NON BC BOTTLED 12 EAST DELL N/A CANADIAN NON BC BOTTLED 13 EQUIFERA N/A CANADIAN NON BC BOTTLED 14 EQUULEUS N/A CANADIAN NON BC BOTTLED 15 G. MARQUIS N/A CANADIAN NON BC BOTTLED 16 GENERATIONSEVEN N/A CANADIAN NON BC BOTTLED 17 GIGGLE JUICE N/A CANADIAN NON BC BOTTLED 18 HENRY OF PELHAM N/A CANADIAN NON BC BOTTLED 19 INN. NIAGARA N/A CANADIAN NON BC BOTTLED 20 KONZELMANN N/A CANADIAN NON BC BOTTLED 21 LAILEY WILEY N/A CANADIAN NON BC BOTTLED 22 LE CLOS N/A CANADIAN NON BC BOTTLED 23 LE CLOS JORDANNE N/A CANADIAN NON BC BOTTLED 24 LIAISON WINES N/A CANADIAN NON BC BOTTLED 25 MAGNOTTA N/A CANADIAN NON BC BOTTLED 26 MIKEWEIR N/A CANADIAN NON BC BOTTLED 27 NAKED GRAPE N/A CANADIAN NON BC BOTTLED 28 PELEE ISLAND N/A CANADIAN NON BC BOTTLED 29 PILLITTERI N/A CANADIAN NON BC BOTTLED 30 RED HERRING N/A CANADIAN NON BC BOTTLED 31 SCHONMARKE N/A CANADIAN NON BC BOTTLED 32 SO KITTLING RIDGE N/A CANADIAN NON BC BOTTLED 33 SO MONDE N/A CANADIAN NON BC BOTTLED 34 SO PELEE N/A CANADIAN NON BC BOTTLED 35 SO STREWN N/A CANADIAN NON BC BOTTLED 36 SO VIDAL N/A CANADIAN NON BC BOTTLED 37 STRATUS N/A CANADIAN NON BC BOTTLED 38 TAWSE N/A CANADIAN NON BC BOTTLED 39 THIRTY BENCH N/A CANADIAN NON BC BOTTLED 40 WAYNEGRETZKY N/A CANADIAN NON BC BOTTLED 41 TRIUS CANADA CANADIAN NON BC ESTATE 42 1STROW SURREY NON-OKANAGAN ESTATE 43 22OAKS DUNCAN NON-OKANAGAN ESTATE 44 40KNOTS COMOX NON-OKANAGAN ESTATE 45 ALDERLEA DUNCAN NON-OKANAGAN ESTATE 46 AVERILLCREEK DUNCAN NON-OKANAGAN ESTATE 47 BACCATA RIDGE GRINDROD NON-OKANAGAN ESTATE 48 BACKYARDVINEYARD LANGLEY NON-OKANAGAN ESTATE 49 BAILLIEGROHMAN CRESTON NON-OKANAGAN ESTATE 50 BEAUFORT COURTENAY NON-OKANAGAN ESTATE 51 BLACKWOODLANE ALDERGROVE NON-OKANAGAN ESTATE 52 BLOSSOM RICHMOND NON-OKANAGAN ESTATE 53 BLUEGROUSE DUNCAN NON-OKANAGAN ESTATE 54 CANADABERRIES RICHMOND NON-OKANAGAN ESTATE 55 CARBREA HORNBY ISLAND NON-OKANAGAN ESTATE 56 CELISTA CELISTA NON-OKANAGAN ESTATE 	 175	Table A.2. Identified wine brands present in the BC market in 2011-2015.    Brand Name Town Region/Brand Classification 57 CHALETESTATE NORTH SAANICH NON-OKANAGAN ESTATE 58 CHASEWARREN PORT ALBERNI NON-OKANAGAN ESTATE 59 CHATEAUISABELLA RICHMOND NON-OKANAGAN ESTATE 60 CHERRYPOINT COWICHAN VALLEY NON-OKANAGAN ESTATE 61 COLUMBIAGARDENS TRAIL NON-OKANAGAN ESTATE 62 DAMALI COBBLE HILL NON-OKANAGAN ESTATE 63 DE VINE SAANICH NON-OKANAGAN ESTATE 64 DEOL DUNCAN NON-OKANAGAN ESTATE 65 DIVINO COWICHAN VALLEY NON-OKANAGAN ESTATE 66 DOMAINEDECHABERTON LANGLEY NON-OKANAGAN ESTATE 67 DOMAINE JASMIN THETIS ISLAND NON-OKANAGAN ESTATE 68 DOMAINE ROCHETTE SIDNEY NON-OKANAGAN ESTATE 69 DRAGONFLY HILL VICTORIA NON-OKANAGAN ESTATE 70 EDGE OF THE EARTH ARMSTRONG NON-OKANAGAN ESTATE 71 EMERALD COAST PORT ALBERNI NON-OKANAGAN ESTATE 72 ENRICO MILL BAY NON-OKANAGAN ESTATE 73 FORTBERENS LILOOET NON-OKANAGAN ESTATE 74 GARRYOAKS SALT SPRING ISLAND NON-OKANAGAN ESTATE 75 GLENTERRA COBBLE HILL NON-OKANAGAN ESTATE 76 GODFREY BROWNELL DUNCAN NON-OKANAGAN ESTATE 77 GRANITECREEK TAPPEN NON-OKANAGAN ESTATE 78 HARPERSTRAIL KAMLOOPS NON-OKANAGAN ESTATE 79 HIGHLAND HOUSE FARM SAANICH NON-OKANAGAN ESTATE 80 KERMODE DEWDNEY NON-OKANAGAN ESTATE 81 LARCHHILLS SALMON ARM NON-OKANAGAN ESTATE 82 LITTLE TRIBUNE HORNBY ISLAND NON-OKANAGAN ESTATE 83 LOTUSLAND ABBOTSFORD NON-OKANAGAN ESTATE 84 LULUISLAND RICHMOND NON-OKANAGAN ESTATE 85 MAPLE CREEK SURREY NON-OKANAGAN ESTATE 86 MIDDLE MOUNTAIN HORNBY ISLAND NON-OKANAGAN ESTATE 87 MILLSTONE NANAIMO NON-OKANAGAN ESTATE 88 MISTAKENIDENTITY SALT SPRING ISLAND NON-OKANAGAN ESTATE 89 MONTECREEK MONTE CREEK NON-OKANAGAN ESTATE 90 MORNING BAY PENDER ISLAND NON-OKANAGAN ESTATE 91 MTLEHMAN ABBOTSFORD NON-OKANAGAN ESTATE 92 MUSE NORTH SAANICH NON-OKANAGAN ESTATE 93 NECKOFTHEWOODS LANGLEY NON-OKANAGAN ESTATE 94 NORTHERN EXPRESSIONS PRINCE GEORGE NON-OKANAGAN ESTATE 95 OVINO SALMON ARM NON-OKANAGAN ESTATE 96 PACIFICBREEZE NEW WESMINSTER NON-OKANAGAN ESTATE 97 PRIVATO KAMLOOPS NON-OKANAGAN ESTATE 98 RECLINERIDGE TAPPEN NON-OKANAGAN ESTATE 99 RIVERSBEND SURREY NON-OKANAGAN ESTATE 100 ROCKYCREEK COWICHAN BAY NON-OKANAGAN ESTATE 101 SAGEWOOD KAMLOOPS NON-OKANAGAN ESTATE 102 SALTSPRING SALT SPRING ISLAND NON-OKANAGAN ESTATE 103 SANDUZ RICHMOND NON-OKANAGAN ESTATE 104 SATURNA SATURNA ISLAND NON-OKANAGAN ESTATE 105 SEA STAR PENDER ISLAND NON-OKANAGAN ESTATE 106 SEMPER GRAND FORKS NON-OKANAGAN ESTATE 107 SINGLETREE ABBOTSFORD NON-OKANAGAN ESTATE 108 SKIMMERHORN CRESTON NON-OKANAGAN ESTATE 	 176	Table A.2. Identified wine brands present in the BC market in 2011-2015.    Brand Name Town Region/Brand Classification 109 SOUTHEND FARM QUADRA ISLAND NON-OKANAGAN ESTATE 110 STARLING LANE VICTORIA NON-OKANAGAN ESTATE 111 SUNNYBRAE TAPPEN NON-OKANAGAN ESTATE 112 SUNSHINE COAST SECHELT NON-OKANAGAN ESTATE 113 SYMPHONY SAANICHTON NON-OKANAGAN ESTATE 114 THECELLARSATRISE VERNON NON-OKANAGAN ESTATE 115 UNSWORTH MILL BAY NON-OKANAGAN ESTATE 116 VANCOUVERURBANWINERY VANCOUVER NON-OKANAGAN ESTATE 117 VENTURI SCHULZE COBBLE HILL NON-OKANAGAN ESTATE 118 VIGNETI ZANATTA DUNCAN NON-OKANAGAN ESTATE 119 VISTADORO LANGLEY NON-OKANAGAN ESTATE 120 WESTHAM ENDERBY NON-OKANAGAN ESTATE 121 WYNWOOD CELLARS DELTA NON-OKANAGAN ESTATE 122 WATERSIDE ENDERBY NON-OKANAGAN ESTATE 123 50THPARALLEL CRESTON OKANAGAN ESTATE 124 8THGENERATION SUMMERLAND OKANAGAN ESTATE 125 ADEGA ON 45TH OSOYOOS OKANAGAN ESTATE 126 ANCIENTHILL KELOWNA OKANAGAN ESTATE 127 ANTELOPERIDGE OLIVER OKANAGAN ESTATE 128 ARROOWLEAF LAKE COUNTRY OKANAGAN ESTATE 129 BARTIERBROS OLIVER OKANAGAN ESTATE 130 BEAUMONT WEST KELOWNA OKANAGAN ESTATE 131 BENCH1775 NARAMATA OKANAGAN ESTATE 132 BLACK DOG CELLARS OKANAGAN FALLS OKANAGAN ESTATE 133 BLACKHILLS OLIVER OKANAGAN ESTATE 134 BLACKWIDOW NARAMATA OKANAGAN ESTATE 135 BLASTEDCHURCH OKANAGAN FALLS OKANAGAN ESTATE 136 BLUE MOUNTAIN OKANAGAN FALLS OKANAGAN ESTATE 137 BONITAS SUMMERLAND OKANAGAN ESTATE 138 BURROWINGOWL OLIVER OKANAGAN ESTATE 139 CCJENTSCH OLIVER OKANAGAN ESTATE 140 CALLIOPE OLIVER OKANAGAN ESTATE 141 CALONA KELOWNA OKANAGAN ESTATE 142 CAMELOT KELOWNA OKANAGAN ESTATE 143 CANA OLIVER OKANAGAN ESTATE 144 CASSINICELLARS OLIVER OKANAGAN ESTATE 145 CASTORODEORO OLIVER OKANAGAN ESTATE 146 CEDARCREEK KELOWNA OKANAGAN ESTATE 147 CHANDRA OLIVER OKANAGAN ESTATE 148 CHURCHSTATE OLIVER OKANAGAN ESTATE 149 COVERTFARMS OLIVER OKANAGAN ESTATE 150 CULMINA OLIVER OKANAGAN ESTATE 151 DANGELO PENTICTON OKANAGAN ESTATE 152 DAYDREAMER NARAMATA OKANAGAN ESTATE 153 DEEP ROOTS NARAMATA OKANAGAN ESTATE 154 DESERTHILLS OLIVER OKANAGAN ESTATE 155 DIRTYLAUNDRY SUMMERLAND OKANAGAN ESTATE 156 DOMAINECOMBRET OLIVER OKANAGAN ESTATE 157 ELEPHANT ISLAND NARAMATA OKANAGAN ESTATE 158 EXNIHILO LAKE COUNTRY OKANAGAN ESTATE 159 FAIRVIEW OLIVER OKANAGAN ESTATE 160 FIRSTESTATE PEACHLAND OKANAGAN ESTATE 161 FOXTROT NARAMATA OKANAGAN ESTATE 162 FREQUENCY WINE AND SOUND KELOWNA OKANAGAN ESTATE 163 GEHRINGER OLIVER OKANAGAN ESTATE 	 177	Table A.2. Identified wine brands present in the BC market in 2011-2015.    Brand Name Town Region/Brand Classification 164 GOLDHILL OLIVER OKANAGAN ESTATE 165 GRAYMONK LAKE COUNTRY OKANAGAN ESTATE 166 GREATA PEACHLAND OKANAGAN ESTATE 167 HAINLE PEACHLAND OKANAGAN ESTATE 168 HAYWIRE SUMMERLAND OKANAGAN ESTATE 169 HEAVEN'S GATE SUMMERLAND OKANAGAN ESTATE 170 HESTER OLIVER OKANAGAN ESTATE 171 HIDDEN CHAPEL OLIVER OKANAGAN ESTATE 172 HILLSIDE PENTICTON OKANAGAN ESTATE 173 HOUSEOFROSE KELOWNA OKANAGAN ESTATE 174 HOWLINGBLUFF PENTICTON OKANAGAN ESTATE 175 INNISKILLIN OLIVER OKANAGAN ESTATE 176 INTERSECTION OLIVER OKANAGAN ESTATE 177 INTRIGUE LAKE COUNTRY OKANAGAN ESTATE 178 JACKSONTRIGGS OLIVER OKANAGAN ESTATE 179 JOIE NARAMATA OKANAGAN ESTATE 180 KALALA KELOWNA OKANAGAN ESTATE 181 KANAZAWA PENTICTON OKANAGAN ESTATE 182 KETTLE VALLEY NARAMATA OKANAGAN ESTATE 183 KISMET OLIVER OKANAGAN ESTATE 184 KRAZELEGZ KALEDEN OKANAGAN ESTATE 185 LA FRENZ PENTICTON OKANAGAN ESTATE 186 LAKEBREEZE NARAMATA OKANAGAN ESTATE 187 LANG NARAMATA OKANAGAN ESTATE 188 LARIANACELLARS OSOYOOS OKANAGAN ESTATE 189 LASTELLA OSOYOOS OKANAGAN ESTATE 190 LAUGHINGSTOCK PENTICTON OKANAGAN ESTATE 191 LEVIEUXPIN OLIVER OKANAGAN ESTATE 192 LIONELLO PENTICTON OKANAGAN ESTATE 193 LIQUIDITY OKANAGAN FALLS OKANAGAN ESTATE 194 LITTLESTRAW KELOWNA OKANAGAN ESTATE 195 LIXIERE KALEDEN OKANAGAN ESTATE 196 LOCK &WORTH PENTICTON OKANAGAN ESTATE 197 LUSITANO OKANAGAN FALLS OKANAGAN ESTATE 198 MARICHEL NARAMATA OKANAGAN ESTATE 199 MAVERICK OLIVER OKANAGAN ESTATE 200 MEYER OKANAGAN FALLS OKANAGAN ESTATE 201 MISCONDUCT PENTICTON OKANAGAN ESTATE 202 MISSION HILL WEST KELOWNA OKANAGAN ESTATE 203 MOCOJO NARAMATA OKANAGAN ESTATE 204 MONEY PIT OLIVER OKANAGAN ESTATE 205 MISTRAL PENTICTON OKANAGAN ESTATE 206 MONSTER PENTICTON OKANAGAN ESTATE 207 MONTAKARN OLIVER OKANAGAN ESTATE 208 MOONCURSER OSOYOOS OKANAGAN ESTATE 209 MORAINE PENTICTON OKANAGAN ESTATE 210 MTBOUCHERIE KELOWNA OKANAGAN ESTATE 211 NICHE KELOWNA OKANAGAN ESTATE 212 NICHOL NARAMATA OKANAGAN ESTATE 213 NKMIP OSOYOOS OKANAGAN ESTATE 214 NOBLERIDGE OKANAGAN FALLS OKANAGAN ESTATE 215 OLIVERTWIST OLIVER OKANAGAN ESTATE 216 OSOYOOSLAROSE OSOYOOS OKANAGAN ESTATE 217 PAINTEDROCK PENTICTON OKANAGAN ESTATE 218 PARADISERANCH PENTICTON OKANAGAN ESTATE 219 PENTAGE PENTICTON OKANAGAN ESTATE 	 178	Table A.2. Identified wine brands present in the BC market in 2011-2015.    Brand Name Town Region/Brand Classification 220 PERSEUS PENTICTON OKANAGAN ESTATE 221 PHASION OKANAGAN FALLS OKANAGAN ESTATE 222 PLATINUMBENCH OLIVER OKANAGAN ESTATE 223 POPLARGROVE PENTICTON OKANAGAN ESTATE 224 QUAILSGATE KELOWNA OKANAGAN ESTATE 225 QUIDNI PENTICTON OKANAGAN ESTATE 226 QUINTAFERREIRA OLIVER OKANAGAN ESTATE 227 REDROOSTER PENTICTON OKANAGAN ESTATE 228 RIVERSTONE OLIVER OKANAGAN ESTATE 229 ROAD13 OLIVER OKANAGAN ESTATE 230 ROLLINGDALE KELOWNA OKANAGAN ESTATE 231 RUBYBLUES PENTICTON OKANAGAN ESTATE 232 RUSTICO OLIVER OKANAGAN ESTATE 233 SAGEHILLS SUMMERLAND OKANAGAN ESTATE 234 SANDHILL KELOWNA OKANAGAN ESTATE 235 SAXON SUMMERLAND OKANAGAN ESTATE 236 SCORCHED EARTH KELOWNA OKANAGAN ESTATE 237 SEEYALATER OKANAGAN FALLS OKANAGAN ESTATE 238 SERENDIPITY NARAMATA OKANAGAN ESTATE 239 SILK SCARF SUMMERLAND OKANAGAN ESTATE 240 SILVERSAGE OLIVER OKANAGAN ESTATE 241 SOARINGEAGLE PENTICTON OKANAGAN ESTATE 242 SONORAN ESTATE SUMMERLAND OKANAGAN ESTATE 243 SPERLING KELOWNA OKANAGAN ESTATE 244 SPIERHEAD KELOWNA OKANAGAN ESTATE 245 SQUEEZEDWINES OLIVER OKANAGAN ESTATE 246 STHUBERTUS EAST KELOWNA OKANAGAN ESTATE 247 STABLE DOOR PENTICTON OKANAGAN ESTATE 248 STAGSHOLLOW OKANAGAN FALLS OKANAGAN ESTATE 249 STONEBOAT OLIVER OKANAGAN ESTATE 250 STONEHILL PENTICTON OKANAGAN ESTATE 251 SUMACRIDGE SUMMERLAND OKANAGAN ESTATE 252 SUMMERGATE SUMMERLAND OKANAGAN ESTATE 253 SUMMERHILL KELOWNA OKANAGAN ESTATE 254 SYNCHROMESH OKANAGAN FALLS OKANAGAN ESTATE 255 TANGLEDVINES OKANAGAN FALLS OKANAGAN ESTATE 256 TANTALUS KELOWNA OKANAGAN ESTATE 257 TERRAVISTA PENTICTON OKANAGAN ESTATE 258 THWINES SUMMERLAND OKANAGAN ESTATE 259 THEHATCH KELOWNA OKANAGAN ESTATE 260 THEVIEW KELOWNA OKANAGAN ESTATE 261 THERAPY NARAMATA OKANAGAN ESTATE 262 THORNHAVEN SUMMERLAND OKANAGAN ESTATE 263 TIGHTROPE PENTICTON OKANAGAN ESTATE 264 TIME OLIVER OKANAGAN ESTATE 265 TINHORN OLIVER OKANAGAN ESTATE 266 TOPSHELF KALEDEN OKANAGAN ESTATE 267 TOWNSHIP7 PENTICTON OKANAGAN ESTATE 268 TWISTEDTREE OSOYOOS OKANAGAN ESTATE 269 UPPERBENCH PENTICTON OKANAGAN ESTATE 270 VANWESTEN NARAMATA OKANAGAN ESTATE 271 VIBRANT KELOWNA OKANAGAN ESTATE 272 VOLCANICHILLS KELOWNA OKANAGAN ESTATE 273 WILDGOOSE OKANAGAN FALLS OKANAGAN ESTATE 274 WORKING HORSE WINERY PEACHLAND OKANAGAN ESTATE 275 YOUNGWYSE OSOYOOS OKANAGAN ESTATE 	 179	Table A.2. Identified wine brands present in the BC market in 2011-2015.    Brand Name Town Region/Brand Classification 276 ZEROBALANCE PENTICTON OKANAGAN ESTATE 277 CERELIA CAWSTON SIMILKAMEEN VALLEY EST 278 CLOSDUSOLEIL KEREMEOS SIMILKAMEEN VALLEY EST 279 CORCELETTES KEREMEOS SIMILKAMEEN VALLEY EST 280 CROWSNEST CAWSTON SIMILKAMEEN VALLEY EST 281 EAUVIVRE CAWSTON SIMILKAMEEN VALLEY EST 282 FORBIDDEN FRUIT CAWSTON SIMILKAMEEN VALLEY EST 283 HERDER KEREMEOS SIMILKAMEEN VALLEY EST 284 HUGGING TREE CAWSTON SIMILKAMEEN VALLEY EST 285 K MOUNTAIN CAWSTON SIMILKAMEEN VALLEY EST 286 LITTLEFARM CAWSTON SIMILKAMEEN VALLEY EST 287 OROFINO CAWSTON SIMILKAMEEN VALLEY EST 288 ROBINRIDGE KEREMEOS SIMILKAMEEN VALLEY EST 289 SAGE BUSH KEREMEOS SIMILKAMEEN VALLEY EST. 290 SEVENSTONES CAWSTON SIMILKAMEEN VALLEY EST. 291 STLASZLO KEREMEOS SIMILKAMEEN VALLEY EST. 292 SIRENSCALL N/A VIRTUAL BRAND 293 BOUNTYCELLARS N/A VIRTUAL BRAND 294 EARLCO N/A VIRTUAL BRAND 295 NAGGINGDOUBT N/A VIRTUAL BRAND 296 SONORAN RANCH N/A VIRTUAL BRAND 297 _49NORTH N/A VIRTUAL BRAND 298 _9ACRES N/A VIRTUAL BRAND 299 ACES N/A VIRTUAL BRAND 300 ANDRES N/A VIRTUAL BRAND 301 BLACK CELLAR N/A VIRTUAL BRAND 302 BLACK CLOUD N/A VIRTUAL BRAND 303 BLACKSAGE N/A VIRTUAL BRAND 304 BLACKSWIFT N/A VIRTUAL BRAND 305 BODACIOUS N/A VIRTUAL BRAND 306 BONAMICI N/A VIRTUAL BRAND 307 BROKENSHADOW N/A VIRTUAL BRAND 308 CAIRN&YORK N/A VIRTUAL BRAND 309 CAPISTRO N/A VIRTUAL BRAND 310 CARSON N/A VIRTUAL BRAND 311 CLOUD CHASER N/A VIRTUAL BRAND 312 COOLSHANAGH N/A VIRTUAL BRAND 313 COPPERMOON N/A VIRTUAL BRAND 314 DIBELLO N/A VIRTUAL BRAND 315 DIABOLICA N/A VIRTUAL BRAND 316 DOMAINE D'OR N/A VIRTUAL BRAND 317 EDIBLEMARKET N/A VIRTUAL BRAND 318 ENOTECA N/A VIRTUAL BRAND 319 ENTRE LACS N/A VIRTUAL BRAND 320 ERRO N/A VIRTUAL BRAND 321 ESCAPOLOGIE N/A VIRTUAL BRAND 322 FORKINTHEROAD N/A VIRTUAL BRAND 323 FULL PRESS N/A VIRTUAL BRAND 324 HELIOS N/A VIRTUAL BRAND 325 HOCHTALER N/A VIRTUAL BRAND 326 INCLUDE N/A VIRTUAL BRAND 327 KINDLE N/A VIRTUAL BRAND 328 L'AMBIANCE N/A VIRTUAL BRAND 329 LINDEN BAY N/A VIRTUAL BRAND 330 LITTLEDOE N/A VIRTUAL BRAND 331 MACFITZ N/A VIRTUAL BRAND 	 180	Table A.2. Identified wine brands present in the BC market in 2011-2015.    Brand Name Town Region/Brand Classification 332 MARVELOUS ADVENTURES N/A VIRTUAL BRAND 333 MCWATERS N/A VIRTUAL BRAND 334 MISSION RIDGE N/A VIRTUAL BRAND 335 MONTAIGNE N/A VIRTUAL BRAND 336 NATHALIEDECOSTER N/A VIRTUAL BRAND 337 NOBLE BEAST N/A VIRTUAL BRAND 338 OKANAGANVINEYARDS N/A VIRTUAL BRAND 339 ONEFAITHVINEYARDS N/A VIRTUAL BRAND 340 OPEN N/A VIRTUAL BRAND 341 PAINTED TURTLE N/A VIRTUAL BRAND 342 PEMBERTON N/A VIRTUAL BRAND 343 PROSPECT N/A VIRTUAL BRAND 344 RAFTER N/A VIRTUAL BRAND 345 REDBARN N/A VIRTUAL BRAND 346 RIGAMAROLE N/A VIRTUAL BRAND 347 ROCHE N/A VIRTUAL BRAND 348 SAINT AND SINNER N/A VIRTUAL BRAND 349 SAWMILL N/A VIRTUAL BRAND 350 SCHLOSS LADERHEIM N/A VIRTUAL BRAND 351 SCRAPBOOK N/A VIRTUAL BRAND 352 SCREW IT N/A VIRTUAL BRAND 353 SEVENDIRECTIONS N/A VIRTUAL BRAND 354 SHIFT IT N/A VIRTUAL BRAND 355 SKINNYGRAPE N/A VIRTUAL BRAND 356 SOAHC N/A VIRTUAL BRAND 357 SOLA NERO N/A VIRTUAL BRAND 358 SOMMET N/A VIRTUAL BRAND 359 STOMPING GROUND N/A VIRTUAL BRAND 360 STONEROAD N/A VIRTUAL BRAND 361 STRUT N/A VIRTUAL BRAND 362 THREEBEARRANCH N/A VIRTUAL BRAND 363 TOSCANO N/A VIRTUAL BRAND 364 TROVE N/A VIRTUAL BRAND 365 VINDICATION N/A VIRTUAL BRAND 366 VINTAGEINK N/A VIRTUAL BRAND 367 WHISTLER N/A VIRTUAL BRAND 368 WHITEBEAR N/A VIRTUAL BRAND 369 WILDHORSECANYON N/A VIRTUAL BRAND 370 WILLOW HILL N/A VIRTUAL BRAND 371 WILDTHYME N/A VIRTUAL BRAND 372 WINEOCLOCK N/A VIRTUAL BRAND 373 WINE4YOU N/A VIRTUAL BRAND 374 XOXO N/A VIRTUAL BRAND 375 YOLO N/A VIRTUAL BRAND 376 ZIRALDO N/A VIRTUAL BRAND 377 MISCELLANEOUS N/A MISCELLANEOUS Source: The BCLDB wholesale scanner sales data for 2011-2015.        	 181	Table A.3. Identified BC VQA wine brands present in the BC market in 2011-2015.    VQA BRAND ESTATE LOCATION CLASSIFICATION 1 MISCELLANEOUS N/A MISCELLANEOUS  CANADIAN BRANDS NON-BC WITH ESTATE LOCATION 2 MIKE WEIR CANADA CANADA NON-BC 3 GENERATION SEVEN CANADA CANADA NON-BC 4 TRIUS CANADA CANADA NON-BC  NON- OKANAGAN OR SIMILKAMEEN VALLEY ESTATES 5 1ST ROW SURREY NON_OKANAGAN ESTATE 6 40 KNOTS COMOX NON_OKANAGAN ESTATE 7 ALDERLEA DUNCAN NON_OKANAGAN ESTATE 8 AVERILL CREEK DUNCAN NON_OKANAGAN ESTATE 9 BACKYARD VINEYARD LANGLEY NON_OKANAGAN ESTATE 10 BAILLIEGROHMAN CRESTON NON_OKANAGAN ESTATE 11 BEAUFORT COURTENAY NON_OKANAGAN ESTATE 12 BLACKWOOD LANE ALDERGROVE NON_OKANAGAN ESTATE 13 BLOSSOM RICHMOND NON_OKANAGAN ESTATE 14 BLUE GROUSE DUNCAN NON_OKANAGAN ESTATE 15 CANADA BERRIES RICHMOND NON_OKANAGAN ESTATE 16 CELISTA CELISTA NON_OKANAGAN ESTATE 17 CHALET ESTATE NORTH SAANICH NON_OKANAGAN ESTATE 18 CHATEAU ISABELLA RICHMOND NON_OKANAGAN ESTATE 19 CHERRY POINT COWICHAN VALLEY NON_OKANAGAN ESTATE 20 COLUMBIA GARDENS TRAIL NON_OKANAGAN ESTATE 21 DOMAINE DE CHABERTON LANGLEY NON_OKANAGAN ESTATE 22 FORT BERENS LILOOET NON_OKANAGAN ESTATE 23 GARRY OAKS SALT SPRING ISLAND NON_OKANAGAN ESTATE 24 GLENTERRA COBBLE HILL NON_OKANAGAN ESTATE 25 GRANITE CREEK TAPPEN NON_OKANAGAN ESTATE 26 HARPER'S TRAIL KAMLOOPS NON_OKANAGAN ESTATE 27 LARCH HILLS SALMON ARM NON_OKANAGAN ESTATE 28 LULU ISLAND RICHMOND NON_OKANAGAN ESTATE 29 MISTAKEN IDENTITY SALT SPRING ISLAND NON_OKANAGAN ESTATE 30 MONTE CREEK MONTE CREEK NON_OKANAGAN ESTATE 31 MORNING BAY PENDER ISLAND NON_OKANAGAN ESTATE 32 NECK OF THE WOODS LANGLEY NON_OKANAGAN ESTATE 33 PACIFIC BREEZE NEW WESMINSTER NON_OKANAGAN ESTATE 34 PRIVATO KAMLOOPS NON_OKANAGAN ESTATE 35 RECLINE RIDGE TAPPEN NON_OKANAGAN ESTATE 36 RIVER'S BEND SURREY NON_OKANAGAN ESTATE 37 ROCKY CREEK COWICHAN BAY NON_OKANAGAN ESTATE 38 SALT SPRING SALT SPRING ISLAND NON_OKANAGAN ESTATE NON- OKANAGAN OR SIMILKAMEEN VALLEY ESTATES 39 SATURNA SATURNA ISLAND NON_OKANAGAN ESTATE 40 THE CELLARS AT RISE VERNON NON_OKANAGAN ESTATE 	 182	Table A.3. Identified BC VQA wine brands present in the BC market in 2011-2015.    VQA BRAND ESTATE LOCATION CLASSIFICATION NON- OKANAGAN OR SIMILKAMEEN VALLEY ESTATES 41 VANCOUVER URBAN WINERY VANCOUVER NON_OKANAGAN ESTATE 42 VISTA D'ORO LANGLEY NON_OKANAGAN ESTATE OKANAGAN VALLEY ESTATES 43 DOMAINE COMBRET OLIVER OKANAGAN ESTATE 44 50 TH PARALLEL CRESTON OKANAGAN ESTATE 45 8TH GENERATION SUMMERLAND OKANAGAN ESTATE 46 ADEGA OSOYOOS OKANAGAN ESTATE 47 ANCIENT HILL KELOWNA OKANAGAN ESTATE 48 ANTELOPE RIDGE OLIVER OKANAGAN ESTATE 49 ARROOWLEAF LAKE COUNTRY OKANAGAN ESTATE 50 BARTIER BROS OLIVER OKANAGAN ESTATE 51 BEAUMONT WEST KELOWNA OKANAGAN ESTATE 52 BENCH 1775 NARAMATA OKANAGAN ESTATE 53 BLACK HILLS OLIVER OKANAGAN ESTATE 54 BLACK WIDOW NARAMATA OKANAGAN ESTATE 55 BLASTED CHURCH OKANAGAN FALLS OKANAGAN ESTATE 56 BONITAS SUMMERLAND OKANAGAN ESTATE 57 BURROWING OWL OLIVER OKANAGAN ESTATE 58 C.C. JENTSCH OLIVER OKANAGAN ESTATE 59 CALONA KELOWNA OKANAGAN ESTATE 60 CAMELOT KELOWNA OKANAGAN ESTATE 61 CASSINI CELLARS OLIVER OKANAGAN ESTATE 62 CASTORO DE ORO OLIVER OKANAGAN ESTATE 63 CEDAR CREEK KELOWNA OKANAGAN ESTATE 64 CHURCH & STATE OLIVER OKANAGAN ESTATE 65 COVERT FARMS OLIVER OKANAGAN ESTATE 66 CULMINA OLIVER OKANAGAN ESTATE 67 D'ANGELO PENTICTON OKANAGAN ESTATE 68 DAYDREAMER NARAMATA OKANAGAN ESTATE 69 DESERT HILLS OLIVER OKANAGAN ESTATE 70 DIRTY LAUNDRY SUMMERLAND OKANAGAN ESTATE 71 EX NIHILO LAKE COUNTRY OKANAGAN ESTATE 72 FAIRVIEW OLIVER OKANAGAN ESTATE 73 FIRST ESTATE PEACHLAND OKANAGAN ESTATE 74 GEHRINGER OLIVER OKANAGAN ESTATE 75 GOLD HILL OLIVER OKANAGAN ESTATE 76 GRAY MONK LAKE COUNTRY OKANAGAN ESTATE 77 GREATA PEACHLAND OKANAGAN ESTATE 78 HAINLE PEACHLAND OKANAGAN ESTATE 79 HAYWIRE SUMMERLAND OKANAGAN ESTATE 80 HESTER OLIVER OKANAGAN ESTATE 81 HILLSIDE PENTICTON OKANAGAN ESTATE 	 183	Table A.3. Identified BC VQA wine brands present in the BC market in 2011-2015.    VQA BRAND ESTATE LOCATION CLASSIFICATION OKANAGAN VALLEY ESTATES 82 HOUSE OF ROSE KELOWNA OKANAGAN ESTATE 83 HOWLING BLUFF PENTICTON OKANAGAN ESTATE 84 INNISKILLIN OLIVER OKANAGAN ESTATE 85 INTERSECTION OLIVER OKANAGAN ESTATE 86 INTRIGUE LAKE COUNTRY OKANAGAN ESTATE 87 JACKSON TRIGGS OLIVER OKANAGAN ESTATE 88 JOIE NARAMATA OKANAGAN ESTATE 89 KALALA KELOWNA OKANAGAN ESTATE 90 KANAZAWA PENTICTON OKANAGAN ESTATE 91 KISMET OLIVER OKANAGAN ESTATE 92 KRAZE LEGZ KALEDEN OKANAGAN ESTATE 93 LAKE BREEZE NARAMATA OKANAGAN ESTATE 94 LANG NARAMATA OKANAGAN ESTATE 95 LARIANA CELLARS OSOYOOS OKANAGAN ESTATE 96 LASTELLA OSOYOOS OKANAGAN ESTATE 97 LAUGHING STOCK PENTICTON OKANAGAN ESTATE 98 LE VIEUX PIN OLIVER OKANAGAN ESTATE 99 LITTLE STRAW KELOWNA OKANAGAN ESTATE 100 LIXIERE KALEDEN OKANAGAN ESTATE 101 LUSITANO OKANAGAN FALLS OKANAGAN ESTATE 102 MARICHEL NARAMATA OKANAGAN ESTATE 103 MAVERICK OLIVER OKANAGAN ESTATE 104 MEYER OKANAGAN FALLS OKANAGAN ESTATE 105 MISCONDUCT PENTICTON OKANAGAN ESTATE 106 MISSION HILL WEST KELOWNA OKANAGAN ESTATE 107 MISTRAL PENTICTON OKANAGAN ESTATE 108 MONSTER PENTICTON OKANAGAN ESTATE 109 MONTAKARN OLIVER OKANAGAN ESTATE 110 MOON CURSER OSOYOOS OKANAGAN ESTATE 111 MORAINE PENTICTON OKANAGAN ESTATE 112 MT. BOUCHERIE KELOWNA OKANAGAN ESTATE 113 NICHE KELOWNA OKANAGAN ESTATE 114 NK'MIP OSOYOOS OKANAGAN ESTATE 115 NOBLE RIDGE OKANAGAN FALLS OKANAGAN ESTATE 116 OLIVER TWIST OLIVER OKANAGAN ESTATE 117 OSOYOOS LAROSE OSOYOOS OKANAGAN ESTATE 118 PAINTED ROCK PENTICTON OKANAGAN ESTATE 119 PARADISE RANCH PENTICTON OKANAGAN ESTATE 120 PENTAGE PENTICTON OKANAGAN ESTATE 121 PERSEUS PENTICTON OKANAGAN ESTATE 122 PLATINUM BENCH OLIVER OKANAGAN ESTATE 123 POPLAR GROVE PENTICTON OKANAGAN ESTATE 	 184	Table A.3. Identified BC VQA wine brands present in the BC market in 2011-2015.    VQA BRAND ESTATE LOCATION CLASSIFICATION OKANAGAN VALLEY ESTATES 124 QUAIL'S GATE KELOWNA OKANAGAN ESTATE 125 QUIDNI PENTICTON OKANAGAN ESTATE 126 QUINTA FERREIRA OLIVER OKANAGAN ESTATE 127 RED ROOSTER PENTICTON OKANAGAN ESTATE 128 RIVER STONE OLIVER OKANAGAN ESTATE 129 ROAD 13 OLIVER OKANAGAN ESTATE 130 ROLLINGDALE KELOWNA OKANAGAN ESTATE 131 RUBY BLUES PENTICTON OKANAGAN ESTATE 132 SAGE HILLS SUMMERLAND OKANAGAN ESTATE 133 SANDHILL KELOWNA OKANAGAN ESTATE 134 SAXON SUMMERLAND OKANAGAN ESTATE 135 SEE YA LATER OKANAGAN FALLS OKANAGAN ESTATE 136 SERENDIPITY NARAMATA OKANAGAN ESTATE 137 SILVER SAGE OLIVER OKANAGAN ESTATE 138 SOARING EAGLE PENTICTON OKANAGAN ESTATE 139 SONORAN ESTATE SUMMERLAND OKANAGAN ESTATE 140 SPERLING KELOWNA OKANAGAN ESTATE 141 SPIERHEAD KELOWNA OKANAGAN ESTATE 142 SQUEEZED WINES OLIVER OKANAGAN ESTATE 143 ST. HUBERTUS EAST KELOWNA OKANAGAN ESTATE 144 STAG'S HOLLOW OKANAGAN FALLS OKANAGAN ESTATE 145 STONEBOAT OLIVER OKANAGAN ESTATE 146 STONEHILL PENTICTON OKANAGAN ESTATE 147 SUMAC RIDGE SUMMERLAND OKANAGAN ESTATE 148 SUMMERGATE SUMMERLAND OKANAGAN ESTATE 149 SUMMERHILL KELOWNA OKANAGAN ESTATE 150 TANGLED VINES OKANAGAN FALLS OKANAGAN ESTATE 151 TANTALUS KELOWNA OKANAGAN ESTATE 152 TERRAVISTA PENTICTON OKANAGAN ESTATE 153 TH WINES SUMMERLAND OKANAGAN ESTATE 154 THE HATCH KELOWNA OKANAGAN ESTATE 155 THE VIEW KELOWNA OKANAGAN ESTATE 156 THERAPY NARAMATA OKANAGAN ESTATE 157 THORNHAVEN SUMMERLAND OKANAGAN ESTATE 158 TIME OLIVER OKANAGAN ESTATE 159 TINHORN OLIVER OKANAGAN ESTATE 160 TOP SHELF KALEDEN OKANAGAN ESTATE 161 TOWNSHIP 7 PENTICTON OKANAGAN ESTATE 162 TWISTED TREE OSOYOOS OKANAGAN ESTATE 163 UPPER BENCH PENTICTON OKANAGAN ESTATE 164 VAN WESTEN NARAMATA OKANAGAN ESTATE 165 VIBRANT KELOWNA OKANAGAN ESTATE 	 185	Table A.3. Identified BC VQA wine brands present in the BC market in 2011-2015.    VQA BRAND ESTATE LOCATION CLASSIFICATION 166 VOLCANIC HILLS KELOWNA OKANAGAN ESTATE 167 WILD GOOSE OKANAGAN FALLS OKANAGAN ESTATE 168 YOUNG & WYSE OSOYOOS OKANAGAN ESTATE 169 ZERO BALANCE PENTICTON OKANAGAN ESTATE SIMILKAMEEN VALLEY ESTATES 170 CLOS DU SOLEIL KEREMEOS SIMILKAMEEN VALLEY ESTATE 171 CORCELETTES KEREMEOS SIMILKAMEEN VALLEY ESTATE 172 CROWSNEST CAWSTON SIMILKAMEEN VALLEY ESTATE 173 EAUVIVRE CAWSTON SIMILKAMEEN VALLEY ESTATE 174 HERDER KEREMEOS SIMILKAMEEN VALLEY ESTATE 175 LITTLE FARM CAWSTON SIMILKAMEEN VALLEY ESTATE 176 OROFINO CAWSTON SIMILKAMEEN VALLEY ESTATE 177 ROBIN RIDGE KEREMEOS SIMILKAMEEN VALLEY ESTATE 178 SEVEN STONES CAWSTON SIMILKAMEEN VALLEY ESTATE 179 ST. LASZLO KEREMEOS SIMILKAMEEN VALLEY ESTATE VIRTUAL BRANDS WITH UNIDENTIFIED ESTATES LOCATION 180 WILD HORSE CANYON VIRTUAL BRAND    ARTISAN WINE SHOP 181 49 NORTH VIRTUAL BRAND   ARTISAN WINE SHOP 182 9 ACRES VIRTUAL BRAND   183 ACES VIRTUAL BRAND  ACES WINE GROUP 184 ANDRES VIRTUAL BRAND   BELONGS TO ANDREW PELLER 185 BLACK SAGE VIRTUAL BRAND  BELONGS TO CONSTELLATION BRANDS 186 BONAMICI VIRTUAL BRAND  BELONGS TO BONAMICI CELLARS CONSULTING GROUP 187 BOUNTY CELLARS VIRTUAL BRAND  BOUNTY CELLARS (RON PENNINGTON) 188 BROKEN SHADOW VIRTUAL BRAND   ARTISAN WINE SHOP 189 CALLIOPE VIRTUAL BRAND  BELONGS TO BURROWING OWL 190 COOLSHANAGH VIRTUAL BRAND  BELONGS TO SKIP AND JUDY STOTHERT. GRAPES CRUSHED IN THE OKANAGAN CRUSHPAD 191 COPPER MOON VIRTUAL BRAND  BELONGS TO ANDREW PELLER 192 DIABOLICA VIRTUAL BRAND   ARTISAN WINE SHOP 193 EDIBLE MARKET VIRTUAL BRAND   194 FORK IN THE ROAD VIRTUAL BRAND   ARTISAN WINE SHOP 195 HELIOS VIRTUAL BRAND  BELONGS TO TERRABELLA WINERIES LTD. 196 KINDLE VIRTUAL BRAND   ARTISAN WINE SHOP 197 LITTLE DOE VIRTUAL BRAND   198 MAC & FITZ VIRTUAL BRAND    ARTISAN WINE SHOP 199 MCWATERS VIRTUAL BRAND  BELONGS TO ENCORE VINEYARDS (HARRY MCWATERS) 200 NAGGING DOUBT VIRTUAL BRAND  BELONGS TO ROBERT WESBURY 201 NATHALIE DECOSTER VIRTUAL BRAND  BELONGS TO VMF KELOWNA 202 OKANAGAN VINEYARDS VIRTUAL BRAND   203 ONE FAITH VINEYARDS VIRTUAL BRAND  BELONGS TO BILL LUI 	 186	Table A.3. Identified BC VQA wine brands present in the BC market in 2011-2015.    VQA BRAND ESTATE LOCATION CLASSIFICATION VIRTUAL BRANDS WITH UNIDENTIFIED ESTATES LOCATION 204 OPEN VIRTUAL BRAND   BELONGS TO CONSTELLATION BRANDS 205 PROSPECT& GANTON VIRTUAL BRAND   ARTISAN WINE SHOP (BELONGS TO VMF KELOWNA) 206 RAFTER VIRTUAL BRAND  BELONGS TO BILL AND DARLENE FREDING 207 RED BARN VIRTUAL BRAND   ARTISAN WINE SHOP 208 RIGAMAROLE VIRTUAL BRAND   ARTISAN WINE SHOP (BELONGS TO VMF KELOWNA) 209 ROCHE VIRTUAL BRAND  BELONGS TO DYLAN AND PENELOPE ROCHE 210 SAWMILL VIRTUAL BRAND   BELONGS TO CONSTELLATION BRANDS 211 SCRAPBOOK VIRTUAL BRAND   212 SEVEN DIRECTIONS VIRTUAL BRAND  BELONGS TO DANIEL BONTORIN (CONSULTING WINEMAKER) 213 SIREN'S CALL VIRTUAL BRAND  BC WINE STUDIO 214 SONORAN RANCH VIRTUAL BRAND   ARTISAN WINE SHOP 215 STONE ROAD VIRTUAL BRAND   216 STRUT VIRTUAL BRAND  NIAGARA PENINSULA BRAND 217 THREE BEAR RANCH VIRTUAL BRAND   ARTISAN WINE SHOP 218 VINTAGE INK VIRTUAL BRAND    BELONGS TO CONSTELLATION BRANDS 219 WHISTLER VIRTUAL BRAND   220 WHITE BEAR VIRTUAL BRAND  ARTISAN WINE SHOP 221 WILD THYME VIRTUAL BRAND   222 WINE O'CLOCK VIRTUAL BRAND   ARTISAN WINE SHOP Source: The BCLDB wholesale scanner sales data for 2011-2015. Note: Virtual brands were defined as those that didn’t have physical location for their estate (actual address with tasting room, estate location that could be found while searching for their brand names online). It is acknowledged that certain brands classified as virtual brands could become estate wineries later and open physical tasting room, but at the time when this research was pursued they weren’t identified as such. Whenever possible, virtual brand was assigned to its actual owner (physical person(s) or company).                  	 187	Appendix B: Chapter 3  Letter B.1. Initial letter and email send out to wineries in August 2015.  Dear Sir/Madam,  I am a 3rd year PhD student in the Faculty of Land and Food Systems at the University of British Columbia in Vancouver and would deeply appreciate your assistance acquiring data for my research.  An agricultural economist by training, I have undergraduate and graduate degrees from the University of British Columbia. More information about my background and experience can be found here: https://www.linkedin.com/profile/view?id=145572124&trk=nav_responsive_tab_profile  My PhD research is focused on the British Columbia wine industry and specifically the influence of winery location in the Okanagan and Similkameen valleys on wine value.   I hope to have your cooperation in my research. Below are short descriptions of my proposed research; a description of the data I have access to; and the data I am seeking from you.  Research description: Robert Mondavi once said:” One bad wine in the valley is bad for every winery in the valley. One good wine in the valley is good for everyone.” This is one of the statements that led me towards my PhD thesis topic.  In researching wine industries around the world and specifically the BC wine industry I noticed a gap in the economic literature and understanding related to spatial relationships and spatial clustering among wineries in wine regions worldwide and specifically in the BC Wine Country.  Therefore, I propose to combine a wine pricing dataset with geographical information system (GIS) data in order to estimate wine price and location relationships for wines and wineries in BC. Specifically, my research aims to test following hypotheses:  Hypothesis 1:  Fruit from different locations in the Okanagan and Similkameen Valleys produces wines that differ in quality related variables. The cause is fruit 	 188	quality differences resulting from the combined effects of terroir (soil, slope, climate etc.) and management practices. Therefore, the value of terroir is likely location specific and differs regionally.  Research question 1:  What is the value of terroir in different regions in influencing wine price?   Hypothesis 2: The economic theory usually claims that close proximity to a well -established and recognized neighbor brings the recognition to the whole sub-region and as a consequence it is beneficial to all lesser known neighbors in the same area. But there might be instances when a well-known and recognized neighbor (or neighbors) negatively impacts the sale of products of lesser-known neighbors.  Research Question 2: Is there always a positive value gained from a location near a well-recognized winery with a well-known, well -established reputation?  Hypothesis 3: An easy access to the point of sales is one of the most important elements influencing business and sales. It is especially important in cases of  EX-factory (or EX-winery) sales. Therefore, it is hypothesized that wineries located near main roads and closer to the wine route are rewarded with higher benefits regardless the quality of wine. Research Question 3: How does the distance from the main wine route/main road influences wine prices?  Accessible Data  I have access to wine pricing data from the BC Liquor Distribution Branch, for all wine types sold in British Columbia between 2011 and 2015. This includes sales data on a selection of wines produced by your winery.   Data Needed from Your Winery I would like to obtain the exact location of the vineyard or vineyard block where the grapes used to produce a certain selection of your wines were grown.  Please note: Not all of your wines will be used for the purpose of this research so I won’t need GIS data on all grapes growing plots.  I aim to use a selection of varietal wines, not 	 189	more than 6 wines depending on how many wine types produced by your winery are in the BCLCB dataset).   If you are willing to provide the information I am requesting, please send me an email to xxx@gmail.com  and I will provide you with a short table for you to identify the vineyard block location to specific wines that I chose to use in my research. The exact GIS positioning and associated agriculture-related variables I will be able to obtain from Dr. P.B. from PARC/AAFC Summerland after your confirmation on willingness to cooperate on this research. Having fruit production location data will enable proper estimation of variables associated with the value of terroir. Until now almost all economics research related to wine terroir assumed that grapes are grown in close proximity or at the estate winery. This assumption is often not true and as a consequence such research can yield biased estimates. I would deeply appreciate your assistance in helping me acquire the data needed for my research.  I believe this research will be valuable to the wine industry in recommending locations or clustering that will benefit marketing strategies and economics. In exchange I will offer summary results that will have estimates clearly visible for your winery and coded results for your winery neighbors (to fulfill confidentiality requirements).  Please be assured that all information I receive from you will remain confidential. I would be happy to arrange for a confidentiality agreement if you request one. All results coming from this research will have general character and they won’t be showing any specifics related to the exact data information I am asking from you. If you have any suggestions, questions or comments related to this project, please do not hesitate to contact me. Your insights on this subject would be valuable to me.   Please contact me by email: xxx@gmail.com  or phone: XXXX Thank you very much, and I hope to hear from you soon. Sincerely, Kate Pankowska   	 190	 Table B.1. List of wineries that participated in the research presented in chapter 3. WINERY/ BRAND ESTATE LOCATION 8TH GENERATION 6807 BC-97, Summerland, BC V0H 1Z9 ANCIENT HILL 4918 Anderson Road, Kelowna, BC V1X 7V7 BENCH 1775 1775 Naramata Rd, Penticton, BC V2A 8T8 BLACK HILLS 4318 Black Sage Rd, Oliver, BC V0H 1T1 BLACK WIDOW 1630 Naramata Rd, Penticton, BC V2A 8T7 CROWSNEST 2035 Surprise Rd, Cawston, BC V0X 1C2 D'ANGELO 979 Lochore Road, Penticton, BC, V2A 8V1 FAIRVIEW 989 Cellar Road, Just off Old Golf Course Road, Oliver, BC  GEHRINGER BROTHERS 876 Road #8, Oliver, BC V0H 1T1 HAINLE 5355 Trepanier Bench Rd, Peachland, BC VOH 1X2 HAYWIRE 16576 Fosbery Rd, Summerland, BC V0H 1Z6 HESTER CREEK  877 Road 8, Oliver, BC V0H 1V5 HILLSIDE 1350 Naramata Rd, Penticton, BC V2A 8T6 HOUSE OF ROSE 2270 Garner Rd, Kelowna, BC HOWLING BLUFF 1086 Three Mile Rd, Penticton, BC V2A 8T7 LANG 2493 Gammon Rd, Naramata, BC V0H 1N0 LITTLE STRAW 2815 Ourtoland Rd, Kelowna, BC V1Z 2H7 MEYER 4287 McLean Creek Rd, Okanagan Falls, BC V0H 1R1 MISCONDUCT 375 Upper Bench Rd N, Penticton, BC V2A 8T2 NOBLE RIDGE 2320 Oliver Ranch Rd, Okanagan Falls, BC V0H 1R2 POPLAR GROVE 425 Middle Bench Rd N, Penticton, BC V2A 8S5 QUAILS GATE 3303 Boucherie Rd, West Kelowna, BC V1Z 2H3 ROBIN RIDGE 2686 Middle Bench Rd SS 2, Keremeos, BC V0X 1N2 ROLLINGDALE 2306 Hayman Rd, West Kelowna, BC V1Z 1Z5 SERENDIPITY 990 Debeck Road, Naramata, BC V0H 1N0 SPERLING 1405 Pioneer Rd, Kelowna, BC V1W 4M6 ST. HUBERTUS &OAK BAY 5205 Lakeshore Rd, Kelowna, BC V1W 4J1 SUMMERHILL 4870 Chute Lake Rd, Kelowna, BC V1W 4M3 THORNHAVEN 6816 Andrew Ave, Summerland, BC V0H 1Z7 TINHORN 537 Tinhorn Creek Rd, Oliver, BC V0H 1T1 UPPER BENCH 170 Upper Bench Rd S, Penticton, BC V2A 8T1 VOLCANIC HILLS 2845 Boucherie Rd, West Kelowna, BC V1Z 2G6 WILD GOOSE 2145 Sun Valley Way, Okanagan-Similkameen D, BC V0H 1R2 **All winery-specific results from this analysis have been coded to assure privacy.                   	 191	 Figure B.1. The map with locations of the estate wineries that cooperated on the research presented in this chapter.   Source: Own mapping using Google maps: https://www.google.com/maps/about/mymaps/                	 192	 Figure B.2. The map with locations of the vineyards that sorced grapes of the BC VQA wines analyzed in this chapter.   Source: Own mapping using Google maps: https://www.google.com/maps/about/mymaps/    	 193	  Figure B.3. Old and new BCLDB wine mark-up formulas.    Source: Based on the BCLDB website accessed on April 1, 2017: http://www.bcldb.com/files/Wholesale_Pricing_Changes-Overview.pdf  Old mark-up formula Prime cost + domestic charges (optional) = In Bond Cost +Excise Tax&Customs =Duty Paid Costs +Volume Markup/Distribution Charge +%Markup for Wine, Spirits  or +Volume mark-up for beer +Cost of Service Fee +Container recycling fee =LDB Retail Price +GST and PST = LDB Display Price before applicable container deposit New mark-up formula Prime cost + domestic charges (optional) = In Bond Cost +Excise Tax&Customs =Duty Paid Costs + Distribution Charge (beer only) + % Markup for Wine, Spirits, Refreshment Beverages OR + Per Litre Markup by Supplier annual production for Beer +Container recycling fee =LDB Wholesale Price +GST = LDB Wholesale Price before applicable container deposit 	 194	 The BCLDB pricing data set for 2011-2013 includes BC VQA wine prices as per the green box in the Chart 1, above. They include GST and PST taxes.  The BCLDB pricing data set for 2014-2015 includes BC VQA wine prices as per the red box in the Chart 1, above. They include GST tax only.  Therefore, to make them more comparable, the prices for 2011-2013 were corrected to exclude the PST tax. Please compare “yellow” boxes on Chart 1.  Please note: According to the official statements, the BC VQA wines DO NOT go through a standard BCLDB mark-up process. So the “blue” boxes in the Chart 1, above don’t apply to the BC VQA wines. There is a chance that there were some other pricing adjustments done by the BCLDB to the BC VQA wines between 2011-2015. Unfortunately the information on such possible pricing adjustments is not available.                                	 195	  Figure B.4.  Price vs vineyard’s aspect, separated by grape variety.   Figure B.5.  Price vs row direction in the vineyard, separated by grape variety.    102030405060708090PRICE in CAD $E FLAT NE NW S SE SW WASPECTBACO NOIR CABERNET FRANC CABERNET SAUVIGNONCARMENERE CHARDONNAY EHRENFELSERGAMAY NOIR/TREBBIANO GEWURZTRAMINER/VIOGNIER MARECHAL FOCH/ZWEIGELTMERLOT PINOT AUXERROIS PINOT BLANCPINOT GRIGIO PINOT GRIS PINOT MEUNIERPINOT NOIR RIESLING SANGIOVESESAUVIGNON BLANC SYRAH TEMPRANILLOSource: BCLDB pricing data set for 2011-2015 plus self-collected data on terroir.2011-2015Wine price versus aspect, separated by variety102030405060708090PRICE in CAD $EW NS SE-NW SW-NEROWSBACO NOIR CABERNET FRANCCABERNET SAUVIGNON CARMENERECHARDONNAY EHRENFELSERGAMAY NOIR/TREBBIANO GEWURZTRAMINER/VIOGNIERMARECHAL FOCH/ZWEIGELT MERLOTPINOT AUXERROIS PINOT BLANCPINOT GRIGIO PINOT GRISPINOT MEUNIER PINOT NOIRRIESLING SANGIOVESESAUVIGNON BLANC SYRAHTEMPRANILLOSource: BCLDB pricing data set for 2011-2015 plus self-collected data on terroir.2011-2015Wine price versus rows direction, separated by variety	 196	 Figure B.6.  Price vs row soil on the vineyard, separated by grape variety.     Figure B.7.  Price vs volume of wine sales, separated by grape variety   102030405060708090PRICE in CAD $moderately well-suited well-suitedSOILBACO NOIR CABERNET FRANCCABERNET SAUVIGNON CARMENERECHARDONNAY EHRENFELSERGAMAY NOIR/TREBBIANO GEWURZTRAMINER/VIOGNIERMARECHAL FOCH/ZWEIGELT MERLOTPINOT AUXERROIS PINOT BLANCPINOT GRIGIO PINOT GRISPINOT MEUNIER PINOT NOIRRIESLING SANGIOVESESAUVIGNON BLANC SYRAHTEMPRANILLOSource: BCLDB pricing data set for 2011-2015 plus self-collected data on terroir.2011-2015Wine price versus soil, separated by variety102030405060708090PRICE in CAD $0 1000 2000 3000 4000 5000 6000VOLUME in LitresBACO NOIR CABERNET FRANCCABERNET SAUVIGNON CARMENERECHARDONNAY EHRENFELSERGAMAY NOIR/TREBBIANO GEWURZTRAMINER/VIOGNIERMARECHAL FOCH/ZWEIGELT MERLOTPINOT AUXERROIS PINOT BLANCPINOT GRIGIO PINOT GRISPINOT MEUNIER PINOT NOIRRIESLING SANGIOVESESAUVIGNON BLANC SYRAHTEMPRANILLOSource: BCLDB pricing data set for 2011-2015 plus self-collected data on terroir.2011-2015Wine price versus volume, separated by variety	 197	 Figure B.8.  Price vs wine age, separated by grape variety.                          102030405060708090PRICE in CAD $0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15WINEAGE in YearsBACO NOIR CABERNET FRANCCABERNET SAUVIGNON CARMENERECHARDONNAY EHRENFELSERGAMAY NOIR/TREBBIANO GEWURZTRAMINER/VIOGNIERMARECHAL FOCH/ZWEIGELT MERLOTPINOT AUXERROIS PINOT BLANCPINOT GRIGIO PINOT GRISPINOT MEUNIER PINOT NOIRRIESLING SANGIOVESESAUVIGNON BLANC SYRAHTEMPRANILLOSource: BCLDB pricing data set for 2011-2015 plus self-collected data on terroir.2011-2015Wine price versus wine age, separated by variety	 198	  Table B.2. Volume of sales (litres) for selected BC VQA wines per variety, 2011-2015.   VOLUME IN LITRES      Grape Variety Observations Mean Std. Dev. Min Max 1 BACO NOIR 59 503.78 512.09 42.75 2029.5 2 CABERNET FRANC 352 198.22 262.91 0.75 1412.25 3 CABERNET SAUVIGNON 242 157.46 292.11 0.75 2080.5 4 CARMENERE 51 175.74 407.74 4.5 2358.75 5 CHARDONNAY 747 153.73 243.45 0.75 1899.75 6 EHRENFELSER 60 1529.41 1024.39 174 4227 7 GAMAY NOIR 237 197.63 267.47 0.75 1633.5 8 GEWURZTRAMINER 755 396.42 712.11 0.75 5846.25 9 MARECHAL FOCH 317 396.14 562.02 0.75 2875.5 10 MERLOT 482 136.28 220.01 0.75 2365.5 11 PINOT AUXERROIS 120 710.19 709.69 15.75 3415.5 12 PINOT BLANC 145 722.67 831.7 0.75 3199.5 13 PINOT GRIGIO 101 113.92 109.95 0.75 527.25 14 PINOT GRIS 791 166.38 211.21 0.75 2532 15 PINOT MEUNIER 75 66.32 112.4 0.75 563.25 16 PINOT NOIR 985 125.16 192.19 0.75 2706.75 17 RIESLING 506 119.38 193.13 0.75 2393.25 18 SANGIOVESE 5 54.9 79.77 0.75 191.25 19 SAUVIGNIN BLANC 299 225.04 190.63 0.75 1413 20 SYRAH 130 143.01 238.98 0.75 234.75 21 TEMPRANILLO 50 88.05 63.94 0.75 318 22 TREBBIANO 58 276.06 319.8 5.25 1182.75 23 VIOGNIER 112 165.84 215.87 0.75 1281 24 ZWEIGELT 106 46.26 99.62 0.75 780 *Note: Gewurztraminer was used as the base/comparison group in regressions  because of its highest volume of sales observed in the data set used in this chapter. Source: The BCLDB wholesale scanner sales data for 2011-2015.                       	 199	  Table B.3. Volume of sales (litres) for selected BC VQA wines, per winery, 2011-2015.    VOLUME IN LITRES    WINERY Observations Mean Std. Dev. Min Max WINERY 1 286 120.01 175.02 0.75 906 WINERY 2 203 117.2 126.81 0.75 663.75 WINERY 3 129 153.55 182.03 0.75 1413 WINERY 4 51 175.74 407.74 4.5 2358.75 WINERY 5 78 292.3 235.78 3 852.75 WINERY 6 260 65.8 79.49 0.75 699 WINERY 7 91 59.51 61.71 0.75 318 WINERY 8 315 47.49 71.11 0.75 480.75 WINERY 9 176 645.82 619.69 2.25 3415.5 WINERY 10 189 10.21 23.36 0.75 183.75 WINERY 11 370 138.26 173.3 0.75 1188 WINERY 12 369 524.94 598.36 1.5 3199.5 WINERY 13 82 175.18 127.94 14.25 534 WINERY 14 93 50.63 41.31 0.75 171.75 WINERY 15 81 100.61 120.77 0.75 553.5 WINERY 16 129 351.76 412.06 0.75 2393.25 WINERY 17 292 179.98 149.13 1.5 783.75 WINERY 18 118 166.18 142.93 1.5 604.5 WINERY 19 6 21 19.63 1.5 48 WINERY 20 236 123.64 131.73 0.75 742.5 WINERY 21 374 204.53 317.9 0.75 2365.5 WINERY 22 261 987.05 1022.99 0.75 5846.25 WINERY 23 167 47.17 52.96 0.75 287.25 WINERY 24 91 133.15 155.26 0.75 572.25 WINERY 25 62 139.81 134.92 0.75 482.25 WINERY 26 251 79.98 67.11 0.75 345.75 WINERY 27 207 216.14 173.65 0.75 891 WINERY 28 516 312.44 621.82 0.75 4227 WINERY 29 493 224.64 384.24 0.75 2551.5 WINERY 30 34 356.45 610.47 6.75 2706.75 WINERY 31 176 51.2 63.36 0.75 284.25 WINERY 32 253 180.86 270.31 0.75 1633.5 WINERY 33 346 272.49 285.26 4.5 1931.25 *Note: Winery 22 was used as the base/comparison group in regressions because of its highest  volume of sales in the data set used in this chapter. Source: The BCLDB wholesale scanner sales data for 2011-2015.            	 200	  Figure B.9. Histogram wine prices .  Source: Based on the BCLDB wholesale scanner sales data for 2011-2015.   Figure B.10. Histogram  logarithmic transformation of wine prices.  Source: Based on the BCLDB wholesale scanner sales data for 2011-2015.     0.02.04.06.08.1.12.14Density10 20 30 40 50 60 70 80 90PriceHistogram 10.511.522.5Density2 2.5 3 3.5 4 4.5Log PriceHistogram 2	 201	 Figures B11-B16 concern the full data set.  Figure B.11. Wine price versus sub-appellation, by soil type.    Figure B.12. Wine price versus sub-appellation, by rows direction.    102030405060708090PRICE in CAD $1 2 3 4 5 6 7 8 9 10 11 12 13 14 15SUBAPPELLATIONmoderatelywellsuited wellsuitedSource: BCLDB pricing data set for 2011-2015 plus self-collected data on terroir.Wine price vs sub-appellation separated by soil type102030405060708090PRICE in CAD $1 2 3 4 5 6 7 8 9 10 11 12 13 14 15SUBAPPELLATIONEW NSSE-NW SW-NESource: BCLDB pricing data set for 2011-2015 plus self-collected data on terroir.Wine price vs sub-appellations separated by rows direction	 202	  Figure B.13. Wine price versus sub-appellation, by average elevation.   Figure B.14. Wine price versus sub-appellation, by distance to lake.      102030405060708090PRICE in CAD $1 2 3 4 5 6 7 8 9 10 11 12 13 14 15SUBAPPELLATION0 286289 296313 319346/409/490 357/413/492365/414/494 367/417/525370/420 373/421383/429 385/430387/437 388/438396/443 398/456399/462 403/474408/484Source: BCLDB pricing data set for 2011-2015 plus self-collected data on terroir.Wine price versus sub-appellation separated by avg.elevation102030405060708090PRICE in CAD $1 2 3 4 5 6 7 8 9 10 11 12 13 14 15SUBAPPELLATION66.83 73.42 112.05129.46 138.02 174195.88/648.14/1448/3986.34/11086 200.91/655.6900000000001/1656.74/4489/11682.33 247.43/665.79/1726.56/4751/19561.91251.13/696.75/1754.92/4775.69/25803.7 271.16/733.5599999999999/1823.43/5807.29 312.72/814.74/2045.39/5835.93378.72/932.3200000000001/2086.33/6366.15 415.09/1172.12/2271.15/7185.03 439.02/1172.35/2476.28/7559.68452/1209.15/2492.98/7883.13 491.46/1211.29/2958.07/7975.08 516.08/1214.6/3046/9161.540000000001551/1300.38/3624.5/9198.74 565.0700000000001/1307.67/3669.26/9715.42 611/1325.11/3776/10912.14Source: BCLDB pricing data set for 2011-2015 plus self-collected data on terroir.Wine price versus sub-appellations separated by lake distance	 203	  Figure B.15. Wine price versus sub-appellation, by variety.   Figure B.16. Wine price versus sub-appellation, by brand.       102030405060708090PRICE in CAD $1 2 3 4 5 6 7 8 9 10 11 12 13 14 15SUBAPPELLATIONBACO NOIR CABERNET FRANC CABERNET SAUVIGNON CARMENERECHARDONNAY EHRENFELSER GAMAY NOIR/TREBBIANO GEWURZTRAMINER/VIOGNIERMARECHAL FOCH/ZWEIGELT MERLOT PINOT AUXERROIS PINOT BLANCPINOT GRIGIO PINOT GRIS PINOT MEUNIER PINOT NOIRRIESLING SANGIOVESE SAUVIGNON BLANC SYRAHTEMPRANILLOSource: BCLDB pricing data set for 2011-2015 plus self-collected data on terroir.Wine price versus sub-appellation separated by variety102030405060708090PRICE in CAD $1 2 3 4 5 6 7 8 9 10 11 12 13 14 15SUBAPPELLATION8TH GENERATION ANCIENT HILL BENCH 1775BLACK HILLS BLACK WIDOW CROWSNESTD'ANGELO/QUAILS GATE FAIRVIEW/ROBIN RIDGE GEHRINGER BROTHERS/ROLLINGDALEHAINLE/SERENDIPITY HAYWIRE/SPERLING HESTER CREEK/ST.HUBERTUS &OAK BAYHILLSIDE/SUMMERHILL HOUSE OF ROSE/THORNHAVEN HOWLING BLUFF/TINHORNLANG/UPPER BENCH LITTLE STRAW/VOLCANIC HILLS MEYER/WILD GOOSEMISCONDUCT NOBLE RIDGE POPLAR GROVESource: BCLDB pricing data set for 2011-2015 plus self-collected data on terroir.Wine price versus sub-appellation separated by brand	 204	  Figures B17-B26 use grouped data set: group 1: if winery belonged to top 10, group 2: all other wineries.  Figure B.17. Wine price versus sub-appellation, by soil if winery belongs to  top 10 biggest producers.   Figure B.18. Wine price versus sub-appellation, by soil if winery doesn’t belong to  top 10 biggest producers.      10121416182022242628PRICE in CAD $1 2 3 4SUBAPPELLATIONmoderatelywellsuited wellsuitedSource: BCLDB pricing data set for 2011-2015 plus self-collected data on terroir.2011-2015Wine price versus sub-appellation, separated by soil if winery belongs to top 10 biggest wine producers102030405060708090PRICE in CAD $1 2 3 4 5 6 7 8 9 10 11 12 13 14SUBAPPELLATIONmoderatelywellsuited wellsuitedSource: BCLDB pricing data set for 2011-2015 plus self-collected data on terroir.if winery doesn't belong to top 10 biggest wine producers, 2011-2015Wine prices versus sub-appellations, separated by soil type 	 205	 Figure B.19. Wine price versus sub-appellation, by rows direction if winery belongs to top  10 biggest producers.   Figure B.20. Wine price versus sub-appellation, by rows direction if winery doesn’t belong  top 10 biggest producers.     10121416182022242628PRICE in CAD$1 2 3 4SUBAPPELLATIONEW NSSE-NW SW-NESource: BCLDB pricing datas set for 2011-2015 plus self-collected data on terroir.if winery belongs to top 10 biggest wine producers, 2011-2015Wine price versus sub-appellations, separated by rows direction102030405060708090PRICE in CAD $1 2 3 4 5 6 7 8 9 10 11 12 13 14SUBAPPELLATIONEW NSSE-NW SW-NESource: BCLDB pricing data set for 2011-2015 plus self-collected data on terroir.if winery doesn't belong to top 10 biggest producers, 2011-2015Wine price versus sub-appellations, separated by rows direction	 206	  Figure B.21. Wine price versus sub-appellation, by average elevation if winery belongs to  top 10 biggest producers.   Figure B.22. Wine price versus sub-appellation, by average elevation if winery doesn’t belong to  top 10 biggest producers.      10121416182022242628PRICE in CAD $1 2 3 4SUBAPPELLATION0 357385 388413 421Source: BCLDB pricing data set for 2011-2015 plus self-collected data on terroir.if winery belongs to top 10 biggest producers, 2011-2015Wine price versus sub-appellations, separated by avg. elevation102030405060708090PRICE in CAD$1 2 3 4 5 6 7 8 9 10 11 12 13 14SUBAPPELLATION0 286289 296313 319346/417/525 365/420367/421 370/429373/430 383/437387/438 396/443398/456 399/462403/474 408/484409/490 413/492414/494Source: BCLDB pricing data set for 2011-2015 plus self-collected data on terroir.if winery doesn't belong to top 10 biggest producers, 2011-2015Wine versus sub-appellations, separated by avg.elevation	 207	 Figure B.23. Wine price versus sub-appellation, by aspect if winery belongs to  top 10 biggest producers.   Figure B.24. Wine price versus sub-appellation, by aspect if winery doesn’t belong to   top 10 biggest producers.     10121416182022242628PRICE in CAD $1 2 3 4SUBAPPELLATIONE FLATS WSource: BCLDB pricing datas set for 2011-2015 plus self-collected data on terroir.if winery belongs to top 10 biggest producers, 2011-2015Wine price versus sub-appellations, separated by aspect 102030405060708090PRICE in CAD $1 2 3 4 5 6 7 8 9 10 11 12 13 14SUBAPPELLATIONE FLATNE NWS SESW WSource: BCLDB pricing data set for 2011-2015 plus self-collected data on terroir.if winery doesn't belong to top 10 biggest producers, 2011-2015Wine price versus sub-appellations, separated by aspect	 208	 Figure B.25. Wine price versus sub-appellation, by distance to lake if winery belongs to  top 10 biggest producers.   Figure B.26. Wine price versus sub-appellation, by distance to lake if winery doesn’t belong to  top 10 biggest producers.      10121416182022242628PRICE in CAD $1 2 3 4SUBAPPELLATION200 312516 648932 36699715Source: BCLDB pricing datas set for 2011-2015 plus self-collected data on terroir.if winery belongs to top 10 biggest producers, 2011-2015Wine price versus sub-appellaitons, separated by lake distance 102030405060708090PRICE in CAD $1 2 3 4 5 6 7 8 9 10 11 12 13 14SUBAPPELLATION66.83 73.42 112.05129.46 138.02 174195.88/733.5599999999999/2045.39/6366.15 247.43/814.74/2086.33/7185.03 251.13/1172.12/2271.15/7559.68271.16/1172.35/2476.28/7883.13 378.72/1209.15/2492.98/7975.08 415.09/1211.29/2958.07/9161.540000000001439.02/1214.6/3046/9198.74 452/1300.38/3624.5/10912.14 491.46/1307.67/3776/11086551/1325.11/3986.34/11682.33 565.0700000000001/1448/4489/19561.91 611/1656.74/4751/25803.7655.6900000000001/1726.56/4775.69 665.79/1754.92/5807.29 696.75/1823.43/5835.93Source: BCLDB pricing data set for 2011-2015 plus self-collected data on terroir.if winery doesn't belong to top 10 biggest producers, 2011-2015Wine price versus sub-appellations, separated by distance to lake	 209	  Figures B27-B36 use grouped data set: group 1: if winery belonged to top 5, group 2: all other wineries.  Figure B.27. Wine price versus sub-appellation, by soil if winery belongs to  top 5 biggest producers.   Figure B.28. Wine price versus sub-appellation, by soil if winery doesn’t belong to  top 5 biggest producers.     121314151617181920212223PRICE in CAD $11 12 13 14SUBAPPELLATIONmoderatelywellsuited wellsuitedBCLDB pricing data set for 2011-2015 plus self-collected data on terroir.2011-2015Wine price versus sub-appellation, separated by soil if winery belongs to top 5 biggest producers.102030405060708090PRICE in CAD $0 5 10 15SUBAPPELLATIONmoderatelywellsuited wellsuitedBCLDB pricing data set for 2011-2015 plus self-collected data on terroir.2011-2015Wine price versus sub-appellation, separated by soil if winery doesn’t belong to top 5 biggest producers	 210	  Figure B.29. Wine price versus sub-appellation, by rows direction if winery belongs to  top 5 biggest producers.    Figure B.30. Wine price versus sub-appellation, by rows direction if winery doesn’t belong to  top 5 biggest producers.       121314151617181920212223PRICE in CAD $11 12 13 14SUBAPPELLATIONEW NSSW-NEBCLDB pricing data set for 2011-2015 plus self-collected data on terroir.2011-2015Wine price versus sub-appellation, separated by rows direction if winery belongs top 5 biggest producers102030405060708090PRICE in CAD$0 5 10 15SUBAPPELLATIONEW NSSE-NW SW-NEBCLDB pricing data set for 2011-2015 plus self-collected data on terroir.if winery doesn’t belong to top 5 biggest producers, 2011-2015Wine price versus sub-appellation, separated by rows direction 	 211	   Figure B.31. Wine price versus sub-appellation, by average elevation if winery belongs to  top 5 biggest producers.    Figure B.32. Wine price versus sub-appellation, by average elevation if winery doesn’t belong to  top 5 biggest producers.       121314151617181920212223PRICE in CAD $11 12 13 14SUBAPPELLATION0 385388 421BCLDB pricing data set for 2011-2015 plus self-collected data on terroir.if winery belongs to top biggest producers, 2011-2015Wine price versus sub-appellation, separated by average elevation 102030405060708090PRICE in CAD $0 5 10 15SUBAPPELLATION0 286289 296313 319346/414/494 357/417/525365/420 367/421370/429 373/430383/437 387/438396/443 398/456399/462 403/474408/484 409/490413/492BCLDB pricing data set for 2011-2015 plus self-collected data on terroir.if winery doesn’t belong to  top 5 biggest producers, 2011-2015Wine price versus sub-appellation, separated by avg. elevation	 212	  Figure B.33. Wine price versus sub-appellation, by aspect if winery belongs to  top 5 biggest producers.    Figure B.34. Wine price versus sub-appellation, by aspect if winery doesn’t belong to   top 5 biggest producers.       121314151617181920212223PRICE in CAD $11 12 13 14SUBAPPELLATIONFLAT SBCLDB pricing data set for 2011-2015 plus self-collected data on terroir.2011-2015Wine price versus sub-appellation, separated by aspect if winery belongs to top 5 biggest producers102030405060708090PRICE in CAd $0 5 10 15SUBAPPELLATIONE FLATNE NWS SESW WBCLDB pricing data set for 2011-2015 plus self-collected data on terroir.if winery doesn’t belong to top 5 biggest producers, 2011-2015Wine price versus sub-appellation, separated by aspect 	 213	  Figure B.35. Wine price versus sub-appellation, by distance to lake if winery belongs to  top 5 biggest producers.    Figure B.36. Wine price versus sub-appellation, by distance to lake if winery doesn’t belong to  top 5 biggest producers.     121314151617181920212223PRICE in CAD $11 12 13 14SUBAPPELLATION200.91 312.72516.08 648.14932.3200000000001BCLDB pricing data set for 2011-2015 plus self-collected data on terroir.if winery belongs to top 5 biggest producers, 2011-2015Wine price versus sub-appellation, separated by distance to lake 102030405060708090PRICE in CAD $0 5 10 15SUBAPPELLATION66.83 73.42112.05 129.46138.02 174195.88/733.5599999999999/2045.39/5835.93 247.43/814.74/2086.33/6366.15251.13/1172.12/2271.15/7185.03 271.16/1172.35/2476.28/7559.68378.72/1209.15/2492.98/7883.13 415.09/1211.29/2958.07/7975.08439.02/1214.6/3046/9161.540000000001 452/1300.38/3624.5/9198.74491.46/1307.67/3669.26/9715.42 551/1325.11/3776/10912.14565.0700000000001/1448/3986.34/11086 611/1656.74/4489/11682.33655.6900000000001/1726.56/4751/19561.91 665.79/1754.92/4775.69/25803.7696.75/1823.43/5807.29BCLDB pricing data set for 2011-2015 plus self-collected data on terroir.if winery doesn’t belong to top 5 biggest producers, 2011-2015Wine price versus sub-appellation, separated by distance to lake 	 214	 Table B.4. Level-level model. SE clustered on sub-appellations (15).  (1) (2) (3) (4) (5) (6)   price price price price price price wineage 0.395 0.386 0.491 0.47 0.352 0.360  (0.453) (0.458) (0.526) (0.538) (0.602) (0.390) wineagesq -0.0285 -0.03 -0.0357 -0.0341 -0.024 -0.0374  (0.035) (0.035) (0.040) (0.042) (0.046) (0.032) year_2012 -0.271 -0.244 -0.374 -0.391 -0.405 -0.0795  (0.226) (0.225) (0.328) (0.394) (0.423) (0.348) year_2013 -0.234 -0.242 -0.434 -0.385 -0.468 0.184  (0.378) (0.385) (0.537) (0.625) (0.645) (0.323) year_2014 -2.961*** -2.971*** -3.163*** -3.132*** -3.231*** -2.523**  (0.336) (0.335) (0.347) (0.298) (0.275) (0.751) year_2015 -2.818*** -2.806*** -3.037*** -3.014*** -3.075*** -2.327*  (0.369) (0.379) (0.423) (0.358) (0.357) (0.895) BACO NOIR 3.765+ 3.971* 3.902* 4.450*** 5.359*** 6.216***  (1.937) (1.815) (1.669) (0.639) (0.718) (0.869) CABERNET FRANC 8.024+ 7.707+ 7.082+ 6.755 6.271 7.573*  (3.816) (3.744) (3.856) (4.123) (4.122) (2.800) CABERNET SAUVIGNON 11.96* 12.18* 12.38* 13.69** 12.75** 13.51***  (4.842) (4.575) (4.282) (3.517) (3.467) (2.388) CARMENERE 28.79*** 29.51*** 28.04*** 33.13*** 0 0  (1.364) (1.821) (1.320) (3.373) (.) (.) CHARDONNAY 0.619 0.676 0.613 0.436 0.433 1.117  (0.939) (0.964) (0.948) (0.874) (0.815) (0.793) EHRENFELSER -1.632 -1.219 1.511 7.664* 9.503* 11.54*  (1.795) (1.759) (2.990) (3.373) (3.483) (4.916) GAMAY NOIR 0.105 0.182 0.373 0.508 0.584 3.459+  (1.989) (1.953) (1.903) (2.054) (2.000) (1.676) MARECHAL FOCH 2.639+ 2.740+ 2.660+ 2.853+ 2.405* 3.675***  (1.498) (1.528) (1.324) (1.359) (0.994) (0.857) MERLOT 3.421+ 3.208+ 3.291+ 2.778+ 2.247 3.921**  (1.663) (1.597) (1.672) (1.518) (1.303) (1.036) PINOT AUXERROIS -0.0169 -0.014 -0.0803 -0.128 -0.504 -0.837  (1.133) (1.105) (1.150) (1.428) (1.379) (2.217) PINOT BLANC -2.698* -2.652* -2.412* -2.648* -3.174** -2.906*  (1.045) (1.024) (1.012) (1.076) (0.997) (1.202) PINOT GRIGIO -1.301 -1.067 -0.851 -2.227 -2.023 0.0982  (1.562) (1.328) (1.220) (2.688) (2.163) (1.569) PINOT GRIS -0.987 -0.893 -1.223 -0.891 -0.828 -0.155  (0.918) (0.920) (1.298) (1.239) (1.203) (1.014) PINOT MEUNIER 0.48 0.534 0.315 0.436 0.62 0.718  (0.913) (0.917) (1.007) (0.989) (0.991) (1.501) PINOT NOIR 3.691* 3.722* 3.690* 3.860* 3.902* 5.266**  (1.302) (1.300) (1.321) (1.381) (1.433) (1.284) RIESLING 1.831 1.823 1.589 2.729+ 2.288 3.408*  (1.148) (1.279) (1.347) (1.462) (1.318) (1.497) SANGIOVESE 20.00*** 20.43*** 20.34*** 20.12*** 18.72*** 14.29**  (1.907) (1.737) (1.671) (1.574) (2.181) (4.643) SAUVIGNON BLANC -2.852 -3.221 -2.273 -1.415 -0.727 1.414  (2.912) (3.118) (2.120) (1.574) (1.241) (1.130) SYRAH 5.111* 4.734* 4.448* 5.084* 4.811+ 6.632**  (2.093) (2.125) (2.070) (2.296) (2.307) (2.173) TEMPRANILLO 2.56 2.564 2.568 3.161 2.886 3.667  (1.876) (1.886) (1.835) (1.867) (2.023) (2.231) TREBBIANO -2.133* -2.276** -2.237** -2.395** -2.962*** -3.025***  (0.780) (0.695) (0.703) (0.709) (0.664) (0.618) VIOGNIER 2.344 1.749 1.721 1.843 2.267+ 2.876**  (1.754) (1.804) (1.628) (1.507) (1.258) (0.867) ZWEIGELT 6.645* 6.138* 5.652+ 5.504+ 5.247+ 6.881*  (2.409) (2.637) (2.852) (2.841) (2.790) (2.324) WINERY 1 1.469 2.008 3.062 2.312 2.434 -1.188 	 215	Table B.4. Level-level model. SE clustered on sub-appellations (15).  (1) (2) (3) (4) (5) (6)   price price price price price price  (1.547) (1.958) (1.884) (1.849) (3.307) (2.323) WINERY 2 1.24 1.905 -0.0151 -1.354 -5.061 -10.58*  (1.390) (1.664) (2.523) (2.085) (3.628) (4.209) WINERY 3 1.701 1.034 1.221 7.538* 4.548 1.503  (1.425) (1.461) (1.234) (2.947) (3.085) (2.886) WINERY 4 0 0 0 0 33.70*** 27.69***  (.) (.) (.) (.) (3.566) (2.894) WINERY 5 3.873* 4.537* 3.279* 8.118* 7.535* -0.501  (1.577) (1.898) (1.385) (2.949) (2.794) (2.384) WINERY 6 -0.0883 -0.929 -3.23 -3.922 -7.214+ -7.707*  (1.297) (1.560) (2.377) (2.359) (3.457) (2.679) WINERY 7 -1.59 -2.566 -4.925 -1.903 -2.659 -6.121  (2.212) (2.765) (4.021) (4.020) (3.848) (3.666) WINERY 8 20.37*** 19.51*** 16.95** 19.30*** 14.74** 8.607+  (2.407) (2.750) (4.142) (3.077) (4.411) (4.211) WINERY 9 -1.839 -2.77 -5.493+ -7.96 -10.82 -9.406+  (1.447) (1.577) (2.804) (5.233) (7.189) (4.751) WINERY 10 10.93*** 11.63*** 12.65*** 13.13*** 13.21** 17.65***  (1.230) (1.602) (1.826) (3.015) (3.637) (3.573) WINERY 11 4.363+ 4.824* 6.138* 7.998* 5.545+ 0.427  (2.144) (2.236) (2.143) (2.740) (3.037) (2.766) WINERY 12 4.629** 3.756* 4.921** 11.57 6.282 0.383  (1.458) (1.641) (1.540) (7.144) (7.830) (6.852) WINERY 13 -0.395 0.357 -0.0877 0.863 -0.985 -5.933+  (3.549) (3.525) (3.222) (4.472) (4.526) (2.953) WINERY 14 -0.527 -0.891 0.574 1.05 -1.298 -6.601*  (2.018) (2.167) (2.319) (2.539) (2.244) (3.059) WINERY 15 6.060*** 5.079** 6.161** 10.04** 10.22** 3.911*  (1.443) (1.660) (1.641) (2.958) (2.646) (1.673) WINERY 16 1.462 0.446 1.24 4.386+ 1.675 0.707  (1.853) (1.983) (1.769) (2.413) (3.410) (4.488) WINERY 17 1.422 0.714 1.046 6.717 1.725 4.265  (1.820) (1.654) (1.505) (7.109) (9.333) (9.129) WINERY 18 8.256*** 8.977*** 10.79*** 12.78*** 11.48** 4.676  (1.377) (1.683) (1.884) (1.736) (3.471) (3.070) WINERY 19 6.851*** 7.500** 6.200** 7.166* 4.321 -0.769  (1.517) (1.834) (1.554) (2.468) (2.858) (2.652) WINERY 20 2.177 1.103 2.047 12.36** 7.274 1.905  (1.607) (2.126) (1.410) (3.644) (4.875) (3.494) WINERY 21 4.831** 5.169** 4.371** 5.239* 6.338* 3.202  (1.501) (1.642) (1.317) (2.203) (2.609) (1.999) WINERY 23 1.548 2.311 3.955* 5.282* 4.176+ -3.842  (1.856) (2.023) (1.752) (2.050) (2.013) (2.831) WINERY 24 3.602 3.675 4.694 10.45 9.801+ 5.383  (2.839) (2.696) (2.808) (5.977) (5.521) (3.622) WINERY 25 3.533 4.739 5.097+ 11.38* 9.654+ 2.505  (2.313) (2.844) (2.552) (4.312) (4.771) (3.203) WINERY 26 5.590** 6.301** 4.335 1.557 3.903 4.005  (1.579) (1.940) (2.782) (4.726) (5.082) (4.829) WINERY 27 -0.599 0.161 -2.145 -3.09 -1.214 -3.867  (1.786) (2.102) (2.288) (1.990) (2.565) (3.384) WINERY 28 4.958* 5.296* 3.754* 4.801+ 3.587 -0.00427  (1.956) (1.997) (1.458) (2.462) (2.567) (1.927) WINERY 29 0.415 -0.555 -0.0471 2.34 1.183 -0.928  (1.452) (1.743) (1.424) (2.394) (3.200) (3.858) WINERY 30 9.533*** 10.32*** 11.99*** 20.48*** 19.62*** 15.34***  (1.681) (1.890) (2.124) (4.303) (4.009) (3.215) WINERY 31 2.941+ 1.985 2.306 7.129** 3.864 2.899  (1.659) (2.109) (1.629) (1.949) (2.955) (3.977) WINERY 32 0.624 1.349 2.636 3.091 1.665 0.794  (1.730) (2.008) (2.006) (2.404) (2.385) (3.186) 	 216	Table B.4. Level-level model. SE clustered on sub-appellations (15).  (1) (2) (3) (4) (5) (6)   price price price price price price WINERY 33 1.872 2.178 2.216 3.607 3.137 -0.733  (1.848) (2.202) (2.247) (2.219) (2.136) (1.356) alcohol below12% -0.402 -0.461 -0.526 -1.545 -0.771 -0.811  (2.135) (2.144) (2.172) (2.353) (2.465) (2.512) alcohol [12%,14 %] -0.871 -0.862 -0.894 -1.314 -1.04 -0.391  (2.094) (2.058) (2.019) (2.096) (2.165) (1.954) soil well-suited  1.731 2.43 3.928* 4.524* 1.972+   (1.130) (1.632) (1.720) (1.760) (1.101) rows NS   -0.796 -2.577+ -2.682* -2.319*    (0.738) (1.315) (1.166) (0.974) rows SE-NW   3.114 5.153 5.039 4.834    (3.120) (5.548) (6.411) (5.408) rows SW-NE   2.554 1.445 2.688 1.672    (2.167) (2.087) (2.286) (1.588) aspect FLAT    1.053 1.413 2.804     (1.674) (3.026) (3.423) aspect NE    -8.995** -4.175 -0.326     (3.000) (3.562) (4.076) aspect NW    -0.441 1.149 3.774     (4.186) (4.414) (4.456) aspect S    0.251 1.833 6.145+     (3.287) (3.513) (3.212) aspect SE    -5.473 -2.38 2.341     (5.389) (7.096) (6.360) aspect SW    -3.234 -3.971 0.115     (2.276) (2.418) (3.213) aspect W    -6.114** -3.863+ -2.734     (2.037) (2.045) (3.081) avgelev (200m-400m]     -2.32 -3.779*      (1.834) (1.570) avgelev (400m and up)     1.02 0.197      (1.878) (2.591) lake (700m-3000m]     0.954 -1.313      (1.628) (1.979) lake (3000 m and up)     2.174 3.037      (2.294) (2.472) april<11C      0.304*       (0.110) april>19C      0.893**       (0.271) may<16C      -0.114       (0.408) may>25C      0.0029       (0.182) june<20C      -0.119       (0.208) june>29C      0.294+       (0.158) july<25C      -0.624+       (0.330) july>33C      -0.542**       (0.166) august<24C      -0.458**       (0.142) august>33C      0.137       (0.150) september<18C      -0.0974       (0.281) september >27C      -0.252       (0.262) october<10C      -0.404* 	 217	Table B.4. Level-level model. SE clustered on sub-appellations (15).  (1) (2) (3) (4) (5) (6)   price price price price price price       (0.168) october>18C      -0.437       (0.410) april<10C      0.0133       (0.228) april>16C      -0.194       (0.144) may<4C      -0.0139       (0.191) may>11C      -0.617*       (0.210) june<9C      -0.248       (0.342) june>15C      -0.776**       (0.240) july<12C      0.133       (0.167) july>18C      0.311*       (0.117) august<11C      -0.409**       (0.134) august>17C      -0.537       (0.323) september<6C      -0.338       (0.331) september>13C      -0.113       (0.185) october<1C      0.203       (0.260) october>8C      0.0914       (0.270) _cons 15.17*** 14.46*** 13.43*** 13.11*** 12.08** 30.40**   (1.896) (2.095) (1.994) (2.644) (3.464) (10.120) N 6785 6785 6785 6785 6785 6785 R-sq 0.677 0.68 0.686 0.701 0.706 0.751 adj. R-sq 0.674 0.677 0.683 0.698 0.702 0.747 F . . . . . . Standard errors in parentheses      + p<0.10, * p<0.05,  ** p<0.01,  *** p<0.001     SE clustered on 15 sub-appellations      These are results obtained after controlling for variety (24), brand (33) and year (5) fixed effects.  Comparison Groups: Soil: moderately well-suited, Rows: EW, Aspect: E,   Elevation: [0-200m], Heat: middle interval for each month.    Alcohol above 14%, Lake distance [67-700m], Elevation [0-200m], Heat: middle interval for each month.             	 218	 Table B.5. Log-level model. SE clustered on sub-appellations (15).    (1) (2) (3) (4) (5) (6)   lnprice lnprice lnprice lnprice lnprice lnprice wineage 0.0151 0.0148 0.0176 0.015 0.00571 0.00686  (0.014) (0.015) (0.017) (0.015) (0.017) (0.015) wineagesq -0.00127 -0.00132 -0.00145 -0.00126 -0.000489 -0.00093  (0.001) (0.001) (0.001) (0.001) (0.001) (0.001) year_2012 -0.00605 -0.00514 -0.0092 -0.00867 -0.00832 0.00134  (0.007) (0.008) (0.010) (0.012) (0.013) (0.013) year_2013 0.000814 0.000527 -0.00575 -0.00221 -0.00585 0.00918  (0.012) (0.013) (0.016) (0.018) (0.018) (0.013) year_2014 -0.168*** -0.168*** -0.174*** -0.173*** -0.179*** -0.160***  (0.017) (0.017) (0.018) (0.014) (0.013) (0.018) year_2015 -0.162*** -0.161*** -0.168*** -0.167*** -0.169*** -0.151***  (0.018) (0.018) (0.019) (0.015) (0.014) (0.021) BACO NOIR 0.234* 0.241* 0.233* 0.211* 0.264*** 0.306***  (0.100) (0.098) (0.092) (0.093) (0.060) (0.033) CABERNET FRANC 0.364* 0.354* 0.330* 0.326* 0.291+ 0.352**  (0.135) (0.130) (0.131) (0.138) (0.140) (0.103) CABERNET SAUVIGNON 0.391* 0.399** 0.404** 0.458*** 0.392** 0.434***  (0.140) (0.131) (0.123) (0.098) (0.103) (0.069) CARMENERE 1.036*** 1.059*** 1.006*** 1.200*** 0 0  (0.057) (0.073) (0.066) (0.160) (.) (.) CHARDONNAY 0.0136 0.0155 0.0159 0.0102 0.00316 0.0237  (0.047) (0.048) (0.048) (0.041) (0.039) (0.033) EHRENFELSER -0.0868 -0.073 0.0177 0.278* 0.360** 0.412**  (0.083) (0.083) (0.095) (0.099) (0.104) (0.114) GAMAY NOIR -0.0347 -0.0321 -0.0275 -0.0233 -0.016 0.07  (0.108) (0.107) (0.107) (0.112) (0.111) (0.100) MARECHAL FOCH 0.147* 0.150* 0.147* 0.144* 0.120* 0.148**  (0.065) (0.068) (0.057) (0.056) (0.045) (0.038) MERLOT 0.171+ 0.164+ 0.165+ 0.137 0.105 0.181**  (0.088) (0.085) (0.087) (0.083) (0.075) (0.047) PINOT AUXERROIS 0.0158 0.0159 0.0174 0.00999 -0.0124 -0.0223  (0.051) (0.050) (0.050) (0.060) (0.059) (0.076) PINOT BLANC -0.213** -0.212* -0.204* -0.220** -0.253** -0.231**  (0.071) (0.072) (0.072) (0.070) (0.072) (0.069) PINOT GRIGIO -0.0824 -0.0746 -0.0683 -0.111 -0.121 -0.0419  (0.073) (0.067) (0.064) (0.100) (0.095) (0.061) PINOT GRIS -0.0555 -0.0524 -0.0583 -0.0527 -0.055 -0.0376  (0.038) (0.039) (0.055) (0.053) (0.052) (0.046) PINOT MEUNIER 0.0157 0.0175 0.0118 0.02 0.0283 0.0256  (0.040) (0.041) (0.044) (0.045) (0.047) (0.051) PINOT NOIR 0.181** 0.182** 0.182** 0.192** 0.195* 0.238***  (0.058) (0.059) (0.061) (0.062) (0.066) (0.055) RIESLING 0.0836 0.0833 0.0771 0.121+ 0.0972 0.110+  (0.054) (0.058) (0.061) (0.062) (0.058) (0.053) SANGIOVESE 0.786*** 0.801*** 0.785*** 0.800*** 0.727*** 0.656***  (0.085) (0.083) (0.067) (0.065) (0.086) (0.114) SAUVIGNON BLANC -0.123 -0.135 -0.111 -0.0851+ -0.0478 0.0283  (0.088) (0.093) (0.063) (0.048) (0.038) (0.038) SYRAH 0.272** 0.259* 0.245* 0.258* 0.233* 0.299**  (0.091) (0.089) (0.087) (0.102) (0.093) (0.099) TEMPRANILLO 0.109+ 0.109+ 0.111+ 0.138* 0.122+ 0.136*  (0.054) (0.056) (0.056) (0.053) (0.060) (0.057) TREBBIANO -0.155** -0.160** -0.159** -0.169*** -0.206*** -0.190***  (0.042) (0.039) (0.041) (0.038) (0.035) (0.026) VIOGNIER 0.106 0.0859 0.0811 0.0891 0.0974 0.116*  (0.087) (0.085) (0.080) (0.071) (0.058) (0.049) ZWEIGELT 0.259** 0.242** 0.221* 0.198* 0.183* 0.262**  (0.080) (0.079) (0.082) (0.087) (0.083) (0.077) WINERY 1 0.057 0.075 0.117 0.0546 0.027 -0.067 	 219	Table B.5. Log-level model. SE clustered on sub-appellations (15).    (1) (2) (3) (4) (5) (6)   lnprice lnprice lnprice lnprice lnprice lnprice  (0.073) (0.087) (0.078) (0.079) (0.096) (0.093) WINERY 2 0.0248 0.047 0.00878 -0.124 -0.269+ -0.418**  (0.068) (0.077) (0.112) (0.083) (0.139) (0.117) WINERY 3 0.0427 0.0205 0.0303 0.229* 0.0638 -0.0419  (0.060) (0.067) (0.053) (0.101) (0.107) (0.097) WINERY 4 0 0 0 0 1.308*** 1.121***  (.) (.) (.) (.) (0.161) (0.118) WINERY 5 0.154* 0.176* 0.126+ 0.317* 0.293** 0.0527  (0.063) (0.073) (0.063) (0.146) (0.098) (0.111) WINERY 6 -0.0523 -0.0803 -0.163 -0.224+ -0.361* -0.368**  (0.063) (0.073) (0.096) (0.108) (0.151) (0.122) WINERY 7 -0.119 -0.152 -0.237 -0.126 -0.138 -0.267+  (0.088) (0.104) (0.140) (0.180) (0.147) (0.127) WINERY 8 0.651*** 0.622*** 0.562*** 0.588*** 0.395* 0.181  (0.094) (0.103) (0.135) (0.122) (0.177) (0.169) WINERY 9 -0.213** -0.244** -0.311* -0.431+ -0.510+ -0.449*  (0.065) (0.075) (0.113) (0.203) (0.251) (0.156) WINERY 10 0.438*** 0.461*** 0.503*** 0.448*** 0.434** 0.606***  (0.063) (0.075) (0.077) (0.106) (0.109) (0.121) WINERY 11 0.214* 0.229* 0.272** 0.314** 0.204+ 0.0169  (0.093) (0.096) (0.083) (0.101) (0.103) (0.108) WINERY 12 0.222** 0.193* 0.227** 0.36 0.0927 -0.12  (0.068) (0.076) (0.064) (0.227) (0.240) (0.244) WINERY 13 -0.0042 0.0209 0.00497 -0.00168 -0.0876 -0.230+  (0.129) (0.126) (0.114) (0.150) (0.149) (0.115) WINERY 14 -0.0792 -0.0913 -0.047 -0.0469 -0.153 -0.286*  (0.091) (0.098) (0.088) (0.090) (0.093) (0.122) WINERY 15 0.236** 0.204* 0.235** 0.378* 0.400*** 0.194*  (0.064) (0.076) (0.069) (0.138) (0.086) (0.085) WINERY 16 0.0361 0.00226 0.0317 0.123 -0.0245 -0.0705  (0.092) (0.094) (0.080) (0.086) (0.105) (0.149) WINERY 17 0.0208 -0.00284 0.0114 0.109 -0.183 -0.122  (0.072) (0.073) (0.064) (0.236) (0.286) (0.298) WINERY 18 0.311*** 0.335*** 0.390*** 0.476*** 0.342* 0.151  (0.067) (0.079) (0.071) (0.061) (0.119) (0.120) WINERY 19 0.346*** 0.367*** 0.317*** 0.317* 0.183 0.0738  (0.070) (0.080) (0.076) (0.108) (0.122) (0.128) WINERY 20 0.0959 0.0601 0.0863 0.318* -0.0285 -0.114  (0.074) (0.088) (0.067) (0.108) (0.158) (0.140) WINERY 21 0.224** 0.235** 0.205* 0.206+ 0.304** 0.187*  (0.075) (0.078) (0.069) (0.102) (0.101) (0.076) WINERY 23 0.0876 0.113 0.163+ 0.177+ 0.119 -0.0994  (0.087) (0.095) (0.081) (0.086) (0.075) (0.100) WINERY 24 0.158 0.161 0.196 0.348 0.333+ 0.196  (0.143) (0.137) (0.139) (0.202) (0.187) (0.140) WINERY 25 0.122 0.162 0.191+ 0.412** 0.256+ 0.066  (0.077) (0.095) (0.092) (0.130) (0.123) (0.096) WINERY 26 0.258** 0.282** 0.242+ 0.166 0.351* 0.361*  (0.077) (0.090) (0.128) (0.189) (0.160) (0.162) WINERY 27 -0.0744 -0.049 -0.0976 -0.222* -0.0842 -0.146  (0.086) (0.096) (0.113) (0.092) (0.089) (0.097) WINERY 28 0.213* 0.225* 0.182* 0.168 0.118 0.00218  (0.092) (0.093) (0.065) (0.100) (0.097) (0.080) WINERY 29 -0.0352 -0.0675 -0.0438 0.0361 -0.00436 -0.116  (0.066) (0.080) (0.067) (0.114) (0.092) (0.132) WINERY 30 0.428*** 0.454*** 0.506*** 0.805*** 0.810*** 0.692***  (0.078) (0.085) (0.083) (0.136) (0.141) (0.109) WINERY 31 0.183* 0.151 0.171* 0.331** 0.164 0.102  (0.075) (0.087) (0.073) (0.085) (0.118) (0.152) WINERY 32 -0.00348 0.0207 0.0672 0.0461 -0.0314 0.0103  (0.086) (0.095) (0.091) (0.095) (0.086) (0.113) 	 220	Table B.5. Log-level model. SE clustered on sub-appellations (15).    (1) (2) (3) (4) (5) (6)   lnprice lnprice lnprice lnprice lnprice lnprice WINERY 33 0.0593 0.0695 0.0767 0.113 0.106 -0.027  (0.097) (0.107) (0.109) (0.093) (0.085) (0.058) alcohol below12% 0.00789 0.00591 0.00263 -0.0326 0.00672 0.0188  (0.077) (0.077) (0.079) (0.078) (0.085) (0.081) alcohol [12%,14 %] 0.00239 0.00272 -0.000126 -0.0162 -0.00168 0.0296  (0.065) (0.064) (0.062) (0.062) (0.063) (0.054) soil well-suited  0.0577 0.0813 0.142* 0.152* 0.0852   (0.045) (0.057) (0.054) (0.056) (0.050) rows NS   -0.0126 -0.08 -0.0829* -0.0817*    (0.028) (0.051) (0.033) (0.028) rows SE-NW   0.0861 0.0984 0.0663 0.0563    (0.114) (0.198) (0.192) (0.158) rows SW-NE   0.0993 0.0528 0.105 0.0753    (0.086) (0.087) (0.083) (0.066) aspect FLAT    -0.0085 -0.00823 0.0858     (0.043) (0.092) (0.105) aspect NE    -0.276* 0.0283 0.131     (0.106) (0.132) (0.152) aspect NW    0.0515 0.131 0.254     (0.169) (0.143) (0.156) aspect S    -0.0894 -0.0045 0.191     (0.124) (0.110) (0.120) aspect SE    -0.176 0.0199 0.208     (0.175) (0.208) (0.195) aspect SW    -0.207+ -0.260* -0.0844     (0.107) (0.089) (0.125) aspect W    -0.296** -0.173* -0.0939     (0.083) (0.074) (0.103) avgelev (200m-400m]     -0.197** -0.229**      (0.066) (0.057) avgelev (400m and up)     0.011 0.025      (0.070) (0.086) lake (700m-3000m]     0.0433 -0.012      (0.057) (0.068) lake (3000 m and up)     0.0688 0.0689      (0.085) (0.076) april<11C      0.0104*       (0.005) april>19C      0.0321**       (0.010) may<16C      -0.00447       (0.015) may>25C      -0.0000888       (0.007) june<20C      -0.00166       (0.008) june>29C      0.0109       (0.007) july<25C      -0.015       (0.012) july>33C      -0.0187**       (0.006) august<24C      -0.0115+       (0.006) august>33C      0.00634       (0.006) september<18C      -0.00145       (0.009) september >27C      -0.00287       (0.006) october<10C      -0.0167* 	 221	Table B.5. Log-level model. SE clustered on sub-appellations (15).    (1) (2) (3) (4) (5) (6)   lnprice lnprice lnprice lnprice lnprice lnprice       (0.006) october>18C      -0.0104       (0.012) april<10C      0.00398       (0.009) april>16C      -0.00661       (0.006) may<4C      0.000944       (0.008) may>11C      -0.0237**       (0.007) june<9C      -0.00943       (0.012) june>15C      -0.0270*       (0.010) july<12C      0.000689       (0.007) july>18C      0.00957+       (0.005) august<11C      -0.0147*       (0.006) august>17C      -0.0107       (0.008) september<6C      -0.00872       (0.011) september>13C      -0.0031       (0.007) october<1C      0.00673       (0.009) october>8C      -0.00529       (0.009) _cons 2.747*** 2.723*** 2.681*** 2.758*** 2.776*** 3.219***   (0.089) (0.099) (0.087) (0.100) (0.098) (0.263) N 6785 6785 6785 6785 6785 6785 R-sq 0.745 0.747 0.751 0.768 0.78 0.815 adj. R-sq 0.742 0.744 0.749 0.766 0.777 0.812 F . . . . . . Standard errors in parentheses      + p<0.10, * p<0.05,  ** p<0.01,  *** p<0.001      SE clustered on 15 sub-appellations      These are results obtained after controlling for variety (24), brand (33) and year (5) fixed effects.   Comparison Groups: Soil: moderately well-suited, Rows: EW, Aspect: E,    Elevation: [0-200m], Heat: middle interval for each month.     Alcohol above 14%, Lake distance [67-700m], Elevation [0-200m], Heat: middle interval for each month.              	 222	 Table B.6. Regression results (full level-level and log-level models (6), with and without inclusion of capacity dummy.  Capacity dummy variable=1 if winery belonged to top 10 market players in BC in 2011-2015.    Main specification Main specification Specification with capacity dummy Specification with capacity dummy   price lnprice price lnprice year_2014 -2.523** -0.160*** -2.523** -0.160***  (0.751) (0.018) (0.751) (0.018) year_2015 -2.327* -0.151*** -2.327* -0.151***  (0.895) (0.021) (0.895) (0.021) capacity   4.265 -0.0738    (9.129) (0.128) soil well-suited 1.972+ 0.0852 1.972+ 0.0852  (1.101) (0.050) (1.101) (0.050) rows NS -2.319* -0.0817* -2.319* -0.0817*  (0.974) (0.028) (0.974) (0.028) aspect S 6.145+ 0.191 6.145+ 0.191  (3.212) (0.120) (3.212) (0.120) avgelev (200m-400m] -3.779* -0.229** -3.779* -0.229**  (1.570) (0.057) (1.570) (0.057) april<11C 0.304* 0.0104* 0.304* 0.0104*  (0.110) (0.005) (0.110) (0.005) april>19C 0.893** 0.0321** 0.893** 0.0321**  (0.271) (0.010) (0.271) (0.010) june>29C 0.294+ 0.0109 0.294+ 0.0109  (0.158) (0.007) (0.158) (0.007) july<25C -0.624+ -0.015 -0.624+ -0.015  (0.330) (0.012) (0.330) (0.012) july>33C -0.542** -0.0187** -0.542** -0.0187**  (0.166) (0.006) (0.166) (0.006) august<24C -0.458** -0.0115+ -0.458** -0.0115+  (0.142) (0.006) (0.142) (0.006) october<10C -0.404* -0.0167* -0.404* -0.0167*  (0.168) (0.006) (0.168) (0.006) may>11C -0.617* -0.0237** -0.617* -0.0237**  (0.210) (0.007) (0.210) (0.007) june>15C -0.776** -0.0270* -0.776** -0.0270*  (0.240) (0.010) (0.240) (0.010) july>18C 0.311* 0.00957+ 0.311* 0.00957+  (0.117) (0.005) (0.117) (0.005) august<11C -0.409** -0.0147* -0.409** -0.0147*  (0.134) (0.006) (0.134) (0.006) Constant 30.40** 3.219*** 30.40** 3.219***   (10.120) (0.263) (10.120) (0.263) N 6785 6785 6785 6785 R-sq 0.751 0.815 0.751 0.815 adj. R-sq 0.747 0.812 0.747 0.812 Standard errors in parentheses    + p<0.10, * p<0.05,  ** p<0.01,  *** p<0.001   	 223	Table B.6. Regression results (full level-level and log-level models (6), with and without inclusion of capacity dummy.  Capacity dummy variable=1 if winery belonged to top 10 market players in BC in 2011-2015.    Main specification Main specification Specification with capacity dummy Specification with capacity dummy   price lnprice price lnprice SE clustered on 15 sub-appellations    These are results obtained after controlling for variety (24), brand (33) and year (5) fixed effects.  Alcohol(3), Lake distance(3).    Comparison Groups: Soil: moderately well-suited, Rows: EW, Aspect: E,  Elevation: [0-200m], Heat: middle interval for each month.   Capacity: in top 10 market players.    Only results that yielded significant estimates in Model 6 are presented here.                                      	 224	Table B.7. Regression results (full level-level and log-level models (6), with and without inclusion of capacity dummy.  Capacity dummy variable=1 if winery belonged to top 5 market players in BC in 2011-2015.    Main specification Main specification Specification with capacity dummy Specification with capacity dummy   price lnprice price lnprice year_2014 -2.523** -0.160*** -2.523** -0.160***  (0.751) (0.018) (0.751) (0.018) year_2015 -2.327* -0.151*** -2.327* -0.151***  (0.895) (0.021) (0.895) (0.021) capacity   0.769 -0.0738    (2.652) (0.128) soil well-suited 1.972+ 0.0852 1.972+ 0.0852  (1.101) (0.050) (1.101) (0.050) rows NS -2.319* -0.0817* -2.319* -0.0817*  (0.974) (0.028) (0.974) (0.028) aspect S 6.145+ 0.191 6.145+ 0.191  (3.212) (0.120) (3.212) (0.120) avgelev (200m-400m] -3.779* -0.229** -3.779* -0.229**  (1.570) (0.057) (1.570) (0.057) april<11C 0.304* 0.0104* 0.304* 0.0104*  (0.110) (0.005) (0.110) (0.005) april>19C 0.893** 0.0321** 0.893** 0.0321**  (0.271) (0.010) (0.271) (0.010) june>29C 0.294+ 0.0109 0.294+ 0.0109  (0.158) (0.007) (0.158) (0.007) july<25C -0.624+ -0.015 -0.624+ -0.015  (0.330) (0.012) (0.330) (0.012) july>33C -0.542** -0.0187** -0.542** -0.0187**  (0.166) (0.006) (0.166) (0.006) august<24C -0.458** -0.0115+ -0.458** -0.0115+  (0.142) (0.006) (0.142) (0.006) october<10C -0.404* -0.0167* -0.404* -0.0167*  (0.168) (0.006) (0.168) (0.006) may>11C -0.617* -0.0237** -0.617* -0.0237**  (0.210) (0.007) (0.210) (0.007) june>15C -0.776** -0.0270* -0.776** -0.0270*  (0.240) (0.010) (0.240) (0.010) july>18C 0.311* 0.00957+ 0.311* 0.00957+  (0.117) (0.005) (0.117) (0.005) august<11C -0.409** -0.0147* -0.409** -0.0147*  (0.134) (0.006) (0.134) (0.006) Constant 30.40** 3.219*** 29.63** 3.293***   (10.120) (0.263) (9.063) (0.294) N 6785 6785 6785 6785 R-sq 0.751 0.815 0.751 0.815 adj. R-sq 0.747 0.812 0.747 0.812 Standard errors in parentheses    + p<0.10, * p<0.05,  ** p<0.01,  *** p<0.001   SE clustered on 15 sub-appellations    	 225	Table B.7. Regression results (full level-level and log-level models (6), with and without inclusion of capacity dummy.  Capacity dummy variable=1 if winery belonged to top 5 market players in BC in 2011-2015.    Main specification Main specification Specification with capacity dummy Specification with capacity dummy   price lnprice price lnprice These are results obtained after controlling for variety (24), brand (33) and year (5) fixed effects.  Alcohol(3), Lake distance(3).    Comparison Groups: Soil: moderately well-suited,  Rows: EW, Aspect: E,  Elevation:[0-200m], Heat: middle interval for each month.   Capacity: not in top 5 market players.    Only results that yielded significant estimates in Model 6 are presented here.                                     	 226	Appendix C: Chapter 4.  Table C.1. Average prices of BC grown grape varieties, year 2015.  Varietals  Average Price/Tonne White Pinot Gris $2,076 Chardonnay $2,033 Gewürztraminer $1,866 Sauvignon Blanc $1,799 Riesling $1,790 Pinot Blanc $1,822 Viognier $2,267 Bacchus $1,951 Muscat $2,121 Auxerrois $1,906 Ehrenfelser $1,816 Semillon $2,348 Icewine Riesling $2,305 Siegerrebe $1,939 Kerner $1,994 Müller Thurgau $1,559 Vidal $1,046 Misc. White Vinifera $2,175 Roussanne $2,415 Misc. White Hybrid $1,378 Schönburger $1,733 Madeleine Angevine $1,420 Optima $2,012 Ortega $1,825 Rotberger $2,000 Siegfriedrebe $1,665 Red   Merlot $2,466 Cabernet Sauvignon $2,563 Red 	 227	Table C.1. Average prices of BC grown grape varieties, year 2015.  Varietals  Average Price/Tonne  Pinot Noir $2,270 Cabernet Franc $2,563 Syrah/Shiraz $2,683 Malbec $2,713 Gamay Noir $2,063 Petit Verdot $2,681 Maréchal Foch $1,786 Zweigelt $2,400 Misc. Red Hybrids $2,156 Tempranillo $2,393 Zinfandel $2,002 Misc. Red Vinifera $2,187 Lemberger/Blaufränkisch $2,128 Sangiovese $2,664 Pinot Meunier $1,768 Carmenere $2,741 Chancellor $1,305 Dunkelfelder $2,204 Mourvedre $3,225 Source: 2015 British Columbia Wine Grape Report, accessed on April 1, 2017: http://www.grapegrowers.bc.ca/sites/default/files/resource/2015%20-%20Crop%20Report%20-%20Public/files/2015%20BC%20Wine%20Grape%20Crop%20Report%20-%20Public.pdf                  	 228	Table C.2. List of estate wineries.       Winery     1 CASSINI CELLARS 72 GREATA 2 QUINTA FERREIRA 73 HILLSIDE 3 COVERT FARMS 74 LE VIEUX PIN 4 DESERT HILLS 75 JOIE 5 SAXON 76 KETTLE VALLEY 6 MISCONDUCT 77 HAYWIRE 7 CHURCH&STATE 78 ST. HUBERTUS 8 OROFINO 79 RED ROOSTER 9 HAINLE 80 THE VIEW 10 WILD GOOSE 81 SUMMERHILL 11 RIVER STONE 82 LANG 12 NICHOL 83 QUAILSGATE 13 PENTAGE 84 TANTALUS 14 POPLAR GROVE 85 SPERLING 15 ROBIN RIDGE 86 HERDER 16 PERSEUS 87 CROWSNEST 17 SONORAN ESTATE 88 TINHORN 18 LASTELLA 89 KISMET 19 ADEGA 90 SYNCHROMESH 20 INTERSECTION 91 MORAINE 21 LAKE BREEZE 92 MT. BOUCHERIE 22 ARROWLEAF 93 NK'MIP 23 D'ANGELO 94 BENCH 1775 24 INTRIGUE 95 CORCELETTES 25 8TH GENERATION 96 FOXTROT 26 SILK SCARF 97 LAUGHING STOCK 27 FAIRVIEW 98 BARTIER BROS 28 ROLLINGDALE 99 DEEP ROOTS 29 HESTER 100 LITTLE STRAW 30 BURROWING OWL 101 CULMINA 31 HOUSE OF ROSE 102 SEE YA LATER 32 RUBY BLUES 103 MISTRAL 33 VOLCANIC HILLS 104 GOLD HILL 34 CAMELOT 105 ANTELOPE RIDGE 35 YOUNG & WYSE 106 SILVER SAGE 36 ZERO BALANCE 107 TWISTED TREE 37 ROAD 13 108 BEAUMONT 38 MISSION HILL 109 MOCOJO 39 JACKSON TRIGGS 110 50TH PARALLEL 40 CLOS DU SOLEIL 111 TH 41 MOON CURSER 112 CALONA 42 HEAVEN'S GATE 113 RUSTICO 43 NOBLE RIDGE 114 BONITAS 44 EXNIHILO 115 EAUVIVRE 45 VIBRANT 116 SUMAC RIDGE 46 SPIERHEAD 117 BLACK HILLS 47 SEVEN STONES 118 OSOYOOS LAROSE 48 SERENDIPITY 119 PAINTED ROCK 49 PLATINUM BENCH 120 CERELIA 50 HIDDEN CHAPEL 121 BLUE MOUNTAIN 51 OLIVER TWIST 122 KRAZE LEGZ 52 VAN WESTEN 123 MEYER 53 MAVERICK 124 ANCIENT HILL 54 HOWLING BLUFF 125 BLACK WIDOW 55 INNISKILLIN 126 GEHRINGER 56 UPPER BENCH 127 MARICHEL 57 LIQUIDITY 128 STONEBOAT 58 STAG'S HOLLOW 129 FIRST ESTATE 59 TIGHTROPE 130 ST. LASZLO 60 GRAY MONK 131 DIRTY LAUNDRY 	 229	Table C.2. List of estate wineries.       Winery     61 SANDHILL 132 HUGGING TREE 62 CEDARCREEK 133 BLACK DOG 63 CASTORO DE ORO 134 CANA 64 TOWNSHIP 7 135 TOPSHELF 65 KALALA 136 DAYDREAMER 66 C.C. JENTSCH 137 THE HATCH 67 THORNHAVEN 138 LARIANA 68 LA FRENZ 139 MONTAKARN 69 BLASTED CHURCH   70 THERAPY   71 NICHE                              	 230	Figure C.1. Distribution of prices for red VQA wines, 2011-2015.   Figure C.2. Distribution of prices for red non-VQA wines, 2011-2015.        0.01.02.03.04.05.06.07Density0 10 20 30 40 50 60 70 80 90 100 110 120 130PRICE (CAD $)Source: Own calculations based on the BCLDB sales data.SKU# =1118VQA Red Wines from the Okanagan and Similkameen Valleys (Estate Wineries Only)Distribution of Prices for Red VQA Wines, 2011-20150.01.02.03.04.05.06.07Density0 10 20 30 40 50 60 70 80 90 100 110 120 130PRICE (CAD $)Source: Own calaculations based on the BCLDB sales data.SKU# = 723Non-VQA Red Wines from the Okanagan and Similkameen Valleys (Estate Wineries Only)Distribution of Prices for Red Non-VQA Wines, 2011-2015	 231	Figure C.3. Distribution of prices for white VQA wines, 2011-2015.   Figure C.4. Distribution of prices for white non-VQA wines, 2011-2015.        0.02.04.06.08.1.12.14.16.18Density5 10 15 20 25 30 35 40 45 50 55PRICE (CAD $)Source Own calculations based on the BCLDB sales data.SKU# = 986VQA White Wines from the Okanagan and Similkameen Valleys (Estate Wineries Only)Distribution of White VQA Wines, 2011-20150.02.04.06.08.1.12.14Density5 10 15 20 25 30 35 40 45 50 55PRICE (CAD $)Source: Own calculations based on the BCLDB sales data.SKU# = 623Non-VQA White Wines from the Okanagan and Similkameen Valleys (Estate Wineries Only)Distribution of White Non-VQA Wines, 2011-2015	 232	Table C.3. Explanatory power of the binomial probit model.      -------- True --------  Classified  D            ~D   Total + 1588             491 2079 - 497           843   1340 Total   2085      1334 3419       Classified +if predicted Pr(D) >= .5   True D defined as vqaindic != 0     Sensitivity Pr( +|D)  76.16%  Specificity Pr( -|~D) 63.19%  Positive predictive value   Pr( D|+)    76.38%  Negative predictive value  Pr(~D|-)   62.91%   False + rate for true ~D      Pr( +|~D) 36.81%  False - rate for true D         Pr( -|D) 23.84%  False + rate for classified +Pr(~D|+) 23.62%  False - rate for classified -  Pr( D|-) 37.09%   Correctly classified 71.10%                          	 233	 Table C.4. Binomial probit -full estimation results.   Binomial Probit VQA indication (VQA=1) Winery Age [1990, 2000) 0.266+  (0.149) Winery Age [2000, 2010) 0.746***  (0.145) Winery Age [2010, 2014) 0.879***  (0.151) East Side Mixed Sediments 0.769**  (0.242) Golden Mile 0.305*  (0.152) Kettled Outwash and Fans -0.680***  (0.149) Lakeside Alluvial Fans 0.433*  (0.182) Mission Creek Terraces 0.460**  (0.150) Mixed Sediments and Fans -0.165  (0.125) NE Side Lacustrine Bench -0.188  (0.118) Sandy Outwash Lakeside Terraces East Side -0.127  (0.207) Sandy Outwash Lakeside Terraces West Side 0.251  (0.224) Sandy Outwash Terrace and Deposits -0.186  (0.126) SE Side Lacustrine Bench -0.605***  (0.145) Similkameen Valley -0.325*  (0.130) West Side Lacustrine Bench 0.816+  (0.449) West Side Mixed Sediments 0.0811  (0.136) Capacity Medium -1.599***  (0.153) Capacity Small -2.107***  (0.154) Reserve -0.163  (0.112) Color White 0.0338  (0.096) Sweetness 1 0.202  (0.225) Sweetness 2 -0.133  (0.425) Sweetness 3 0.249  (0.587) Sweetness 4 0  (.) Sweetness 5 0  (.) Sweetness 6 0  (.) Sweetness N/A -1.104*** 	 234	Table C.4. Binomial probit -full estimation results.   Binomial Probit VQA indication (VQA=1)  (0.095) Baco Noir -0.0387  (0.906) Barbera 0  (.) Blaufrankisch 0  (.) Blend -0.41  (0.337) Cabernet Franc -0.344  (0.362) Cabernet Sauvignon -0.714*  (0.360) Carmenere 0.0524  (0.846) Chardonnay -0.372  (0.356) Chasselas 0  (.) Chenin Blanc -0.563  (1.031) Ehrenfelser -0.806  (0.573) Gamay Noir -0.745+  (0.387) Gewurztraminer -0.396  (0.360) Grenache 0  (.) Gruner Vetliner 0  (.) Kerner -0.442  (0.651) Lemberger 0  (.) Malbec -0.271  (0.407) Marechal Foch -0.523  (0.449) Merlot -0.528  (0.344) Mourvedre 0  (.) Muscat Ottonel -0.215  (0.674) Optima 0  (.) Oraniensteiner -0.63  (0.640) Petit Verdot -1.078+  (0.601) Pinot Auxerrois -1.011  (0.743) Pinot Blanc -0.242  (0.386) Pinot Grigio -0.228 	 235	Table C.4. Binomial probit -full estimation results.   Binomial Probit VQA indication (VQA=1)  (0.435) Pinot Gris -0.573  (0.358) Pinot Meunier 0  (.) Pinot Noir -0.494  (0.341) Pinotage -0.156  (0.546) Riesling -0.276  (0.365) Sangiovese 0  (.) Sauvignon Blanc -0.09  (0.374) Schonburger -0.775  (0.841) Semillon -0.798+  (0.484) Siegerrebe 0  (.) Sovereign -1.887  (1.688) St.Laurent 0  (.) Syrah -0.457  (0.352) Tannat -0.236  (0.973) Tempranillo 0.541  (0.587) Tokay 0  (.) Touriga 0  (.) Trebbiano 0  (.) Vidal -0.412  (0.680) Viognier -0.468  (0.371) Voros 0  (.) Zinfandel -0.469  (0.550) Zweigelt 0  (.) Alcohol 0.103**  (0.034) Constant 1.420*   (0.585) N 3419 Standard errors in parentheses   + p<0.10, * p<0.05, ** p<0.01, *** p<0.001  Comparison groups:  Winery Age [1932, 1990), Capacity: Large,   	 236	Table C.4. Binomial probit -full estimation results.   Binomial Probit VQA indication (VQA=1) Sub-appellation:  Alluvial fans and flood plains  Variety: Arneis, Sweetness=0, Reserve=0   	 237	Table C.5. 2SLS estimation results. Dependent variable: logarithm of the average volume share.  First stage Second Stage   logarithm average volume share logarithm average volume share VQA Indication  0.655+   (0.349) VQA probability 1.015322***   (0.00)  Reserve=1 -0.0039349 0.0851  (0.031) (0.093) Color White 0.0042477 -0.0302  (0.030) (0.089) Sweetness 1 0.0060714 0.0939  (0.044) (0.132) Sweetness 2 -0.0094689 -0.293  (0.112) (0.339) Sweetness 3 0.0151871 0.684  (0.166) (0.501) Sweetness 4 0 0  (.) (.) Sweetness 5 0 0  (.) (.) Sweetness 6 0 0  (.) (.) Sweetness N/A 0.0049971 -0.414***  (0.039) (0.116) Baco Noir 0.7668957 -0.173  (0.484) (1.478) Barbera 0 0  (.) (.) Blaufrankisch 0 0  (.) (.) Blend 0.7819821+ 0.11  (0.433) (1.321) Cabernet Franc 0.7773784+ -0.196  (0.436) (1.329) Cabernet Sauvignon 0.7766538+ -0.316  (0.436) (1.323) Carmenere 0.7775433 0.388  (0.496) (1.521) Chardonnay 0.7851669+ -0.29  (0.433) (1.323) Chasselas 0 0 	 238	Table C.5. 2SLS estimation results. Dependent variable: logarithm of the average volume share.  First stage Second Stage   logarithm average volume share logarithm average volume share  (.) (.) Chenin Blanc 0.7660962 -0.417  (0.467) (1.422) Ehrenfelser 0.7745429+ 0.0533  (0.449) (1.360) Gamay Noir 0.782682+ 0.289  (0.439) (1.329) Gewurztraminer 0.776901+ -0.0608  (0.434) (1.323) Grenache 0 0  (.) (.) Gruner Vetliner 0 0  (.) (.) Kerner 0.7725572+ -0.487  (0.464) (1.411) Lemberger 0 0  (.) (.) Malbec 0.7810404+ -0.5  (0.439) (1.340) Marechal Foch 0.7779297+ -0.219  (0.444) (1.352) Merlot 0.7882533+ -0.12  (0.434) (1.323) Mourvedre 0 0  (.) (.) Muscat Ottonel 0.7610193 -0.585  (0.474) (1.445) Optima 0 0  (.) (.) Oraniensteiner 0.7827361+ -0.325  (0.468) (1.419) Petit Verdot 0.7879901+ -0.834  (0.455) (1.375) Pinot Auxerrois 0.8784982+ 0.561  (0.475) (1.444) Pinot Blanc 0.7884337+ 0.144  (0.435) (1.333) Pinot Grigio 0.7751416+ 0.645  (0.439) (1.344) 	 239	Table C.5. 2SLS estimation results. Dependent variable: logarithm of the average volume share.  First stage Second Stage   logarithm average volume share logarithm average volume share Pinot Gris 0.7778133+ 0.267  (0.434) (1.319) Pinot Meunier 0 0  (.) (.) Pinot Noir 0.7838458+ -0.174  (0.434) (1.323) Pinotage 0.771444+ -0.000491  (0.452) (1.385) Riesling 0.7693093+ -0.332  (0.434) (1.326) Sangiovese 0 0  (.) (.) Sauvignon Blanc 0.772091+ -0.131  (0.434) (1.330) Schonburger 0.9338849+ 0.374  (0.530) (1.605) Semillon 0.7676919+ -0.317  (0.445) (1.348) Siegerrebe 0 0  (.) (.) Sovereign 0.7815046 -0.987  (0.532) (1.603) St.Laurent 0 0  (.) (.) Syrah 0.782937+ -0.092  (0.435) (1.325) Tannat 0.7864829 0.532  (0.526) (1.605) Tempranillo 0.7870768+ -0.273  (0.455) (1.407) Tokay 0 0  (.) (.) Touriga 0 0  (.) (.) Trebbiano 0 0  (.) (.) Vidal 0.8508183+ -0.393  (0.474) (1.448) Viognier 0.7837274+ -0.12 	 240	Table C.5. 2SLS estimation results. Dependent variable: logarithm of the average volume share.  First stage Second Stage   logarithm average volume share logarithm average volume share  (0.435) (1.324) Voros 0 0  (.) (.) Zinfandel 0.7755698+ -1.054  (0.451) (1.373) Zweigelt 0.7732222+ -0.571  (0.443) (1.358) Alcohol -0.0034282 -0.155***  (0.011) (0.032) East Side Mixed Sediments -0.0143846 -0.039  (0.061) (0.183) Golden Mile -0.0023718 -0.385**  (0.042) (0.126) Kettled Outwash and Fans -0.0049115 0.311+  (0.053) (0.162) Lakeside Alluvial Fans -0.0108049 0.564***  (0.054) (0.160) Mission Creek Terraces -0.0009846 -0.0566  (0.046) (0.139) Mixed Sediments and Fans -0.0128443 0.236*  (0.038) (0.116) NE Side Lacustrine Bench -0.0015179 0.470***  (0.036) (0.109) Sandy Outwash Lakeside Terraces East Side -0.0108636 0.476**  (0.061) (0.183) Sandy Outwash Lakeside Terraces West Side -0.0066681 0.188  (0.069) (0.207) Sandy Outwash Terrace and Deposits -0.0044024 0.167  (0.039) (0.117) SE Side Lacustrine Bench 0.0007355 -0.275+  (0.050) (0.151) Similkameen Valley -0.0081657 0.155  (0.044) (0.132) West Side Lacustrine Bench -0.0075759 -0.344  (0.080) (0.241) West Side Mixed Sediments -0.0106201 -0.419***  (0.038) (0.115) Winery Age [1990, 2000) 0.0037946 0.461***  (0.034) (0.104) 	 241	Table C.5. 2SLS estimation results. Dependent variable: logarithm of the average volume share.  First stage Second Stage   logarithm average volume share logarithm average volume share Winery Age [2000, 2010) 0.0007705 0.653***  (0.043) (0.129) Winery Age [2010, 2014) -0.0054993 1.177***  (0.049) (0.147) Capacity Medium 0.0043983 -1.237***  (0.044) (0.130) Capacity Small 0.0125084 -1.875***  (0.062) (0.183) Constant -0.7498799 -12.24***   (0.466) (1.376) N 3365 3365 R-sq 0.24 0.28 adj. R-sq 0.23 0.26 Standard errors in parentheses    + p<0.10, * p<0.05, ** p<0.01, *** p<0.001   Comparison groups:   Winery Age [1932, 1990)   Capacity: Large   Sub-appellation:  Alluvial fans and flood plains   Sweetness: Sweetness=0   Color: Red   Reserve: Reserve=0   Variety: Arneis   Instrumented: VQA Indication                    	 242	Table C.6. 2SLS estimation results. Dependent variable: logarithm of the average price.    First stage Second Stage   logarithm average price logarithm average price VQA Indication  -0.0657046   (0.084) VQA probability .9944108 ***   (0.00)  Reserve=1 -0.0017288 0.00224  (0.030) (0.022) Color White 0.0038534 -0.274***  (0.029) (0.021) Sweetness 1 -0.0019688 -0.0567+  (0.043) (0.031) Sweetness 2 -0.0045814 -0.000643  (0.112) (0.081) Sweetness 3 0.0244488 0.243*  (0.166) (0.120) Sweetness 4 0 0  (.) (.) Sweetness 5 0 0  (.) (.) Sweetness 6 0 0  (.) (.) Sweetness N/A 0.000036 -0.00914  (0.039) (0.028) Baco Noir 0.7697206 -0.127  (0.484) (0.353) Barbera 0 0  (.) (.) Blaufrankisch 0 0  (.) (.) Blend 0.7803659+ -0.0554  (0.433) (0.316) Cabernet Franc 0.7763052+ 0.0543  (0.436) (0.318) Cabernet Sauvignon 0.7784848+ 0.119  (0.436) (0.316) Carmenere 0.7788005 0.256  (0.496) (0.363) Chardonnay 0.775539+ 0.0454  (0.433) (0.316) Chasselas 0 0 	 243	Table C.6. 2SLS estimation results. Dependent variable: logarithm of the average price.    First stage Second Stage   logarithm average price logarithm average price  (.) (.) Chenin Blanc 0.766925+ 0.111  (0.462) (0.336) Ehrenfelser 0.7741154+ -0.067  (0.449) (0.325) Gamay Noir 0.7779764+ -0.318  (0.439) (0.318) Gewurztraminer 0.7763291+ -0.0894  (0.434) (0.316) Grenache 0 0  (.) (.) Gruner Vetliner 0 0  (.) (.) Kerner 0.771672+ 0.0144  (0.464) (0.337) Lemberger 0 0  (.) (.) Malbec 0.7803657+ 0.0898  (0.439) (0.320) Marechal Foch 0.7801818+ -0.168  (0.444) (0.323) Merlot 0.7819632+ -0.122  (0.434) (0.316) Mourvedre 0 0  (.) (.) Muscat Ottonel 0.7731595 0.22  (0.474) (0.346) Optima 0 0  (.) (.) Oraniensteiner 0.7812458+ -0.0684  (0.468) (0.339) Petit Verdot 0.7809098+ 0.218  (0.455) (0.329) Pinot Auxerrois 0.7733528+ -0.266  (0.468) (0.339) Pinot Blanc 0.7847109+ -0.0902  (0.435) (0.318) Pinot Grigio 0.7775192+ -0.171  (0.439) (0.321) 	 244	Table C.6. 2SLS estimation results. Dependent variable: logarithm of the average price.    First stage Second Stage   logarithm average price logarithm average price Pinot Gris 0.7749337+ -0.0578  (0.434) (0.315) Pinot Meunier 0 0  (.) (.) Pinot Noir 0.7761917+ -0.0273  (0.434) (0.316) Pinotage 0.7801885+ -0.0787  (0.452) (0.331) Riesling 0.7778449+ 0.0784  (0.434) (0.317) Sangiovese 0 0  (.) (.) Sauvignon Blanc 0.7739112+ -0.0269  (0.434) (0.318) Schonburger 0.7784991 -0.0118  (0.500) (0.361) Semillon 0.7697902+ -0.0622  (0.444) (0.322) Siegerrebe 0 0  (.) (.) Sovereign 0.7769405 -0.404  (0.532) (0.383) St.Laurent 0 0  (.) (.) Syrah 0.7774509+ 0.0501  (0.435) (0.317) Tannat 0.783551 0.193  (0.526) (0.384) Tempranillo 0.7921102+ -0.0792  (0.455) (0.336) Tokay 0 0  (.) (.) Touriga 0 0  (.) (.) Trebbiano 0 0  (.) (.) Vidal 0.7629139 -0.0371  (0.467) (0.340) Viognier 0.7790443+ 0.0293 	 245	Table C.6. 2SLS estimation results. Dependent variable: logarithm of the average price.    First stage Second Stage   logarithm average price logarithm average price  (0.435) (0.317) Voros 0 0  (.) (.) Zinfandel 0.7758777+ 0.088  (0.451) (0.328) Zweigelt 0.7717516+ 0.185  (0.443) (0.325) Alcohol 0.0002667 0.107***  (0.010) (0.008) East Side Mixed Sediments 0.0027112 0.157***  (0.060) (0.043) Golden Mile 0.0104331 0.245***  (0.042) (0.030) Kettled Outwash and Fans 0.0025931 0.0715+  (0.053) (0.038) Lakeside Alluvial Fans 0.0079825 0.0648+  (0.053) (0.039) Mission Creek Terraces 0.0007252 0.194***  (0.046) (0.033) Mixed Sediments and Fans -0.0018092 0.0698*  (0.038) (0.027) NE Side Lacustrine Bench 0.0058863 0.155***  (0.036) (0.026) Sandy Outwash Lakeside Terraces East Side -0.0006699 0.0645  (0.060) (0.044) Sandy Outwash Lakeside Terraces West Side 0.0041342 0.542***  (0.069) (0.050) Sandy Outwash Terrace and Deposits 0.0028915 0.207***  (0.038) (0.028) SE Side Lacustrine Bench 0.0068515 0.124***  (0.050) (0.036) Similkameen Valley -0.0000881 0.149***  (0.043) (0.031) West Side Lacustrine Bench 0.0040074 0.0622  (0.080) (0.056) West Side Mixed Sediments 0.0002643 0.220***  (0.038) (0.027) Winery Age [1990, 2000) -0.0038217 -0.118***  (0.034) (0.025) 	 246	Table C.6. 2SLS estimation results. Dependent variable: logarithm of the average price.    First stage Second Stage   logarithm average price logarithm average price Winery Age [2000, 2010) -0.0030698 -0.127***  (0.042) (0.030) Winery Age [2010, 2014) -0.0033919 -0.178***  (0.049) (0.035) Capacity Medium -0.0018829 0.129***  (0.043) (0.031) Capacity Small -0.0011772 0.107*  (0.061) (0.044) Constant -0.7788785 1.584***   (0.465) (0.328) N 3413 3413 R-sq 0.24 0.36 adj. R-sq 0.23 0.352 Standard errors in parentheses    + p<0.10, * p<0.05, ** p<0.01, *** p<0.001   Comparison groups:   Winery Age [1932, 1990)   Capacity: Large   Sub-appellation:  Alluvial fans and flood plains   Sweetness: Sweetness=0   Color: Red   Reserve: Reserve=0   Variety: Arneis   Instrumented: VQA Indication                   	 247	Table C.7. 2SLS estimation results. Dependent variable: logarithm of the average revenue share.  First stage Second Stage   logarithm average revenue share logarithm average revenue share VQA Indication  0.635   (0.521) VQA probability 1.015303***   (.1173954 )  Reserve=1 -0.0041384 -0.00851  (0.031) (0.139) Color White 0.0041057 -0.292*  (0.030) (0.134) Sweetness 1 0.006316 -0.0844  (0.044) (0.198) Sweetness 2 -0.0091668 -0.376  (0.112) (0.507) Sweetness 3 0.0152561 1.575*  (0.166) (0.750) Sweetness 4 0 0  (.) (.) Sweetness 5 0 0  (.) (.) Sweetness 6 0 0  (.) (.) Sweetness N/A 0.0049815 -0.896***  (0.039) (0.174) Baco Noir 0.7662217 0.321  (0.484) (2.210) Barbera 0 0  (.) (.) Blaufrankisch 0 0  (.) (.) Blend 0.7815424+ 0.737  (0.433) (1.975) Cabernet Franc 0.78215758+ 0.481  (0.436) (1.988) Cabernet Sauvignon 0.7762864+ 0.54  (0.436) (1.978) Carmenere 0.7773415 1.348  (0.496) (2.274) Chardonnay 0.7848753+ 0.466  (0.433) (1.977) Chasselas 0 0 	 248	Table C.7. 2SLS estimation results. Dependent variable: logarithm of the average revenue share.  First stage Second Stage   logarithm average revenue share logarithm average revenue share  (.) (.) Chenin Blanc 0.76592 0.454  (0.467) (2.126) Ehrenfelser 0.7740857+ 1.175  (0.449) (2.033) Gamay Noir 0.7820952+ 0.598  (0.439) (1.987) Gewurztraminer 0.7763698+ 0.669  (0.434) (1.978) Grenache 0 0  (.) (.) Gruner Vetliner 0 0  (.) (.) Kerner 0.7722179+ 0.231  (0.464) (2.110) Lemberger 0 0  (.) (.) Malbec 0.7808552+ 0.202  (0.439) (2.003) Marechal Foch 0.7768535+ 0.321  (0.444) (2.021) Merlot 0.78787+ 0.505  (0.434) (1.978) Mourvedre 0 0  (.) (.) Muscat Ottonel 0.7606186 0.884  (0.474) (2.161) Optima 0 0  (.) (.) Oraniensteiner 0.7817366+ 0.528  (0.468) (2.122) Petit Verdot 0.787856+ -0.208  (0.455) (2.056) Pinot Auxerrois 0.8784145+ 1.666  (0.475) (2.158) Pinot Blanc 0.7880064+ 0.823  (0.435) (1.992) Pinot Grigio 0.7747998+ 1.262  (0.439) (2.010) 	 249	Table C.7. 2SLS estimation results. Dependent variable: logarithm of the average revenue share.  First stage Second Stage   logarithm average revenue share logarithm average revenue share Pinot Gris 0.7774812+ 1.006  (0.434) (1.973) Pinot Meunier 0 0  (.) (.) Pinot Noir 0.7832359+ 0.55  (0.434) (1.978) Pinotage 0.7707695+ 1.139  (0.452) (2.070) Riesling 0.7688285+ 0.482  (0.434) (1.982) Sangiovese 0 0  (.) (.) Sauvignon Blanc 0.7717954+ 0.511  (0.434) (1.988) Schonburger 0.9338304+ 1.609  (0.530) (2.400) Semillon 0.7672159+ -0.0613  (0.445) (2.016) Siegerrebe 0 0  (.) (.) Sovereign 0.7821661 -1.787  (0.532) (2.397) St.Laurent 0 0  (.) (.) Syrah 0.7825821+ 0.642  (0.435) (1.981) Tannat 0.7859317 0.725  (0.526) (2.400) Tempranillo 0.78677+ 0.215  (0.455) (2.103) Tokay 0 0  (.) (.) Touriga 0 0  (.) (.) Trebbiano 0 0  (.) (.) Vidal 0.8502812+ 0.496  (0.474) (2.165) Viognier 0.7835296+ 0.573 	 250	Table C.7. 2SLS estimation results. Dependent variable: logarithm of the average revenue share.  First stage Second Stage   logarithm average revenue share logarithm average revenue share  (0.435) (1.980) Voros 0 0  (.) (.) Zinfandel 0.7751668+ -0.467  (0.451) (2.053) Zweigelt 0.7719931+ 0.3  (0.443) (2.031) Alcohol -0.0034776 -0.072  (0.011) (0.047) East Side Mixed Sediments -0.0144911 0.207  (0.061) (0.273) Golden Mile -0.0023701 -0.263  (0.042) (0.188) Kettled Outwash and Fans -0.0047116 0.241  (0.053) (0.242) Lakeside Alluvial Fans -0.010542 0.688**  (0.054) (0.240) Mission Creek Terraces -0.000515 0.113  (0.046) (0.207) Mixed Sediments and Fans -0.0125779 0.0557  (0.038) (0.173) NE Side Lacustrine Bench -0.0014762 0.588***  (0.036) (0.163) Sandy Outwash Lakeside Terraces East Side -0.011025 0.620*  (0.061) (0.274) Sandy Outwash Lakeside Terraces West Side -0.0064667 0.737*  (0.069) (0.310) Sandy Outwash Terrace and Deposits -0.0044848 0.131  (0.039) (0.174) SE Side Lacustrine Bench 0.0009897 -0.258  (0.050) (0.226) Similkameen Valley -0.0086697 -0.000648  (0.044) (0.198) West Side Lacustrine Bench -0.0079211 -0.188  (0.080) (0.360) West Side Mixed Sediments -0.0087799 -0.395*  (0.038) (0.172) Winery Age [1990, 2000) 0.0047147 0.438**  (0.034) (0.156) 	 251	Table C.7. 2SLS estimation results. Dependent variable: logarithm of the average revenue share.  First stage Second Stage   logarithm average revenue share logarithm average revenue share Winery Age [2000, 2010) 0.002112 0.691***  (0.043) (0.193) Winery Age [2010, 2014) -0.0049955 0.862***  (0.049) (0.220) Capacity Medium 0.0039414 -1.365***  (0.044) (0.195) Capacity Small 0.0125965 -2.259***  (0.062) (0.273) Constant -0.7497545 -14.32***   (0.466) (2.058) N 3366 3366 R-sq 0.24 0.25 adj. R-sq 0.23 0.24 Standard errors in parentheses    + p<0.10, * p<0.05, ** p<0.01, *** p<0.001   Comparison groups:   Winery Age [1932, 1990)   Capacity: Large   Sub-appellation:  Alluvial fans and flood plains   Sweetness: Sweetness=0   Color: Red   Reserve: Reserve=0   Variety: Arneis   Instrumented: VQA Indication                    	 252	Table C.8. First-stage regression summary statistics. Dependent variable: logarithm of the average volume share.                  Adjusted Partial     Variable R-sq. R-sq. R-sq. F(1,3302) Prob > F vqaindic  0.2418 0.2275 0.0222 74.8055 0.0000             Minimum eigenvalue statistic = 74.8055                 Critical Values  #of endogenous regressors  1 Ho: Instruments are weak   #of excluded instruments   1   5% 10% 20% 30% 2SLS relative bias     (not available)       10% 15% 20% 25% 2SLS Size of nominal 5% Wald test 16.38 8.96 6.66 5.53 LIML Size of nominal 5% Wald test 16.38 8.96 6.66 5.53                                    	 253	Table C.9. First-stage regression summary statistics. Dependent variable: logarithm of the average price.                 Adjusted Partial     Variable R-sq. R-sq. R-sq. F(1,3302) Prob > F vqaindic  0.2427 0.2287 0.0214 73.1809 0.0000             Minimum eigenvalue statistic = 73.1809                  Critical Values  #of endogenous regressors  1 Ho: Instruments are weak #of excluded instruments   1   5% 10% 20% 30% 2SLS relative bias     (not available)       10% 15% 20% 25% 2SLS Size of nominal 5% Wald test 16.38 8.96 6.66 5.53 LIML Size of nominal 5% Wald test 16.38 8.96 6.66 5.53                              	 254	Table C.10 First-stage regression summary statistics. Dependent variable: logarithm of the average revenue share.                 Adjusted Partial     Variable R-sq. R-sq. R-sq. F(1,3302) Prob > F vqaindic  0.2416 0.2274 0.0221 74.7979 0.0000             Minimum eigenvalue statistic = 74.7979                Critical Values  #of endogenous regressors  1 Ho: Instruments are weak #of excluded instruments   1   5% 10% 20% 30% 2SLS relative bias     (not available)       10% 15% 20% 25% 2SLS Size of nominal 5% Wald test 16.38 8.96 6.66 5.53 LIML Size of nominal 5% Wald test 16.38 8.96 6.66 5.53                          

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.24.1-0357184/manifest

Comment

Related Items