AN ECONOMETRIC ANALYSIS OF THE DEMAND FOR WOOD PRODUCTS IN J A P A N BY PRODUCT TYPE, SPECIES, AND S O U R C E by CHRISTOPHER WILLEM GASTON B . S c , The University of British Columbia, 1979 M . S c , The University of Guelph, 1982 A THESIS SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE D E G R E E OF DOCTOR OF PHILOSOPHY in THE FACULTY OF G R A D U A T E STUDIES Department of Forest Resources Management We accept this thesis as conforming to the requjred standard THE UNIVERSITY OF BRITISH COLUMBIA July, 1997 © Christopher W. Gaston, 1997 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Department of ^f^r^h r ^ J o ^ ^ - 1 Ae»-ey±-^e-J^~ The University of British Columbia Vancouver, Canada Date py t DE-6 (2/88) Gaston Abstract Page ii Abstract This thesis investigates the Japanese demand for wood by product type, by country of origin, and by species, over the period 1965 to 1993. The product types include softwood and hardwood logs, softwood and hardwood lumber, and wood-based panel products (plywood, fibreboard and particle board). In addition to estimating the own-price effects on quantity demanded for individual wood product imports, substitution effects within these product categories are documented to the degree possible, including Japanese substitution with domestic product and non-wood alternatives. The research makes two important contributions. The first is to offer Japanese demand descriptors at a level of wood product detail which is not found in the existing literature. The second is to review and critique the existing methodologies available for investigating substitution effects among disaggregated products (such as softwood lumber by species). As it was discovered that the available approaches are inadequate for properly dealing with product detail, strong recommendations for further research are made for improving our ability to document cross-price effects. The primary conclusion of the study is that individual wood products, by product type, by country of origin, or by species, behave as distinct economic units. This suggests that studies which aggregate wood products into broad categories such as "softwood lumber" risk obscuring important dimensions of both forest products' trade and forest policy. Gaston Table of Contents Page iii TABLE OF CONTENTS A B S T R A C T ii T A B L E OF C O N T E N T S iii LIST OF T A B L E S v LIST OF F IGURES vi A C K N O W L E D G E M E N T S vii 1.0 INTRODUCTION 1.1 Motivation for the Research 1 1.2 Background 3 1.3 Scope of the Research 6 1.3.1 The Research Problem 8 1.3.2 Objectives 10 1.3.3 Hypotheses 10 1.4 Organization of Thesis 11 2.0 L ITERATURE REVIEW ON L O G A N D L U M B E R SUBSTITUTIONS A N D IMPLICATIONS FOR F U R T H E R R E S E A R C H 2.1 The Econometric Estimates of the Price Elasticity of Demand 12 2.1.1 Estimates of Wood for Wood Substitutes 12 2.1.2 Estimates of Non-Wood for Wood Substitutes 16 2.2 Parametric Demand Elasticity Estimation Techniques 20 2.3 Implications for the Present Study 37 3.0 THE M A R K E T FOR W O O D P R O D U C T S IN J A P A N 3.1 The Japanese Domestic Timber Resource 40 3.2 The Use of Japanese Domestic Timber Production 44 3.3 Imports of Wood Products into Japan 49 3.4 Japanese Processing of Domestic and Imported Logs 57 3.5 Summary 60 4.0 T H E O R E T I C A L FOUNDATIONS A N D IMPLICATIONS FOR THE EMPIRICAL Gaston Table of Contents Page iv A N A L Y S I S OF THE J A P A N E S E D E M A N D FOR W O O D P R O D U C T S 4.1 Theoretical Foundations 63 4.2 The Empirical Model 71 4.3 The Data Sources Used in the Empirical Analysis 73 5.0 EMPIRICAL R E S U L T S 5.1 Direct Estimation of Japanese Price Elasticities of Demand for Wood Imports . . 85 5.2 Estimation of the Armington Two-Stage Model of the Japanese Demand for Total Wood Imports 93 5.3 Comparison of the Two-Stage and Direct Estimates of the Own-Price Elasticities of Demand for Wood Product Imports in Japan 100 5.4 Non-Wood Substitution in Japan 103 5.5 Summary 105 6.0 DISCUSSION OF R E S U L T S 6.1 Japanese Wood Product Imports, Aggregated by Product Type 108 6.2 Japanese Softwood Lumber Imports, Aggregated by Source 112 6.3 Japanese Softwood Lumber Imports from Canada, Aggregated by Species . . . 116 6.4 Japanese Softwood Lumber Imports from Non-Canadian Sources, Aggregated by Species 122 6.5 Japanese Softwood Log, Aggregated by Source 127 6.6 Japanese Softwood Log, Aggregated by Species 130 6.7 Japanese Hardwood Lumber and Log Imports, Aggregated by Source 134 6.8 Japanese Panel Product Imports, Aggregated by Source 136 7.0 CONTRIBUTIONS, LIMITATIONS A N D IMPLICATIONS FOR F U R T H E R R E S E A R C H 7.1 Research Contributions and Implications for the BC Forest Industry 141 7.1.1 Summary of the Results 142 7.1.2 Implications of the Research for BC Wood Product Marketing 144 7.1.3 Implications of the Research for BC Forest Policy 150 7.2 Limitations 151 7.3 Implications for Further Research 154 B I B L I O G R A P H Y 158 Gaston List of Tables Page v L I S T O F T A B L E S Table 2.1 Own-Price Elasticity of Demand for Softwood Lumber in N.A 13 Table 2.2 Cross-Price Elasticity of Demand for Similar Lumber in Different Regions . . . . 15 Table 2.3 Cross-Price Elasticity of Demand for Different Lumber 16 Table 2 . 4 Own-Price and Cross-Price Demand Elasticities for Construction Materials: McKillop, etal 17 Table 2.5 Own-Price and Cross-Price Demand Elasticities for Construction Materials: Rockel and Buongiorno 18 Table 2.6 Own-Price and Cross-Price Demand Elasticities for US Softwood Lumber . . . 19 Table 2.7 Own-Price and Cross-Price Demand Elasticities for Selected Canadian Construction Materials 19 Table 2.8 Elasticities of Demand of US Hardwood Plywood Imports by Country of Origin 33 Table 2.9 Elasticities of Demand of US Softwood Lumber Imports from Canada By Species 37 Table 4 .1 Japan Tariff Association Data, Converted Codes 75 Table 4 . 2 B.C. Offshore Lumber Exports Relative to the Whole of Canada (000s m3) . . . 83 Table 5.1 Estimates of the Japanese Demand for Aggregated Wood Imports 86 Table 5.2 Estimates of the Japanese Demand for Selected Disaggregated Wood Products 90 Table 5.3 Estimates of the Constant Elasticity of Substitution Over Varying Degrees of Wood Import Aggregation, Correcting for Serial Correlation 94 Table 5 . 4 Calculated Constant Elasticity of Substitution Weights from Table 5.3 96 Table 5.5 Cochrane-Orcutt Estimates of the Japanese Demand for Selected Aggregations of Wood Imports, Utilizing CES Quantity and Price Indices 96 Table 5.6 Calculated Own- and Cross-Price Elasticities of Demand for the Japanese Imports of all Wood Products by Type 98 Table 5.7 Calculated Own- and Cross-Price Elasticities of Demand for the Japanese Imports of Softwood Lumber by Country of Origin 99 Table 5.8 Calculated Own- and Cross-Price Elasticities of Demand for the Japanese Imports of Canadian Softwood Lumber by Species 100 Table 5.9 Cochrane-Orcutt Estimates of the Japanese Demand for Aggregated Wood Imports, With the Inclusion of a Non-Wood Regressor 105 Table 7.1 Destination of Canadian Softwood Lumber and Log Exports, 1992 146 Gaston List of Figures Page vi L I S T O F F I G U R E S Figure 1.1 Random Lengths S-P-F Lumber Futures 4 Figure 1.2 PNW Douglas Fir Lumber Prices 7 Figure 3.1 Distribution of Man-Made Forest by Age Class (Japan) 42 Figure 3.2 Japanese Domestic Log Production by Species 43 Figure 3.3 Japanese Domestic Log Production by Ownership 44 Figure 3.4 Japanese Domestic Log Supply by Utilization 46 Figure 3.5 Japanese Housing Starts by Number 48 Figure 3.6 Japanese Housing Starts by Area 48 Figure 3.7 Japanese Industrial Wood Supply 49 Figure 3.8 Japanese Self-Sufficiency in Logs 50 Figure 3.9 Japanese Self-Sufficiency in Lumber 53 Figure 3.10 Japanese Self-Sufficiency in Panel Products 53 Figure 3.11 Japanese Imports of Softwood Lumber and Logs, 1993 55 Figure 3.12 Japanese Imports of Softwood Lumber and Logs, 1965 55 Figure 3.13 Japanese Imports of Hardwood Lumber and Logs, 1993 56 Figure 3.14 Japanese Imports of Hardwood Lumber and Logs, 1965 56 Figure 3.15 Japanese Lumber Shipments by Use 58 Figure 4.1 Nominal Price of Japanese Imports of Canadian Sitka Spruce Lumber, By Size 80 Figure 4.2 Nominal Price of Japanese Imports of Canadian Yellow Cedar Lumber, By Size 80 Figure 4.3 Non-wood Housing Starts in Japan 53 Figure 5.1 Observed versus Predicted Values of Quantity Demanded of Aggregated Softwood Lumber Imports by Japan 88 Figure 5.2 Number of Non-Wood Housing Starts in Japan 104 Figure 6.1 Japanese Imports of Wood Products by Major Product Types 108 Figure 6.2 Japanese Imports of Softwood Lumber by Source 113 Figure 6.3 Japanese Imports of Canadian Softwood Lumber by Species 117 Figure 6.4 Japanese Imports of US Softwood Lumber by Species 123 Figure 6.5 Japanese Imports of Former USSR Softwood Lumber by Species 125 Figure 6.6 Japanese Imports of NZ/Chile Softwood Lumber by Species 126 Figure 6.7 Japanese Imports of "Other" Softwood Lumber by Species 128 Figure 6.8 Japanese Imports of Softwood Logs by Source 129 Figure 6.9 Japanese Imports of US Softwood Logs by Species 131 Figure 6.10 Japanese Imports of Former USSR Softwood Logs by Species 133 Figure 6.11 Japanese Imports of Hardwood Lumber by Source 135 Figure 6.12 Japanese Imports of Hardwood Logs by Source 137 Figure 6.13 Japanese Imports of Veneer Sheets by Source 138 Figure 6.14 Japanese Imports of Plywood by Source 139 Figure 6.15 Japanese Imports of Particle Board and Fibreboard 140 Gaston A c k n o w l e d g m e n t s P a g e vii A C K N O W L E D G M E N T S I gratefully acknowledge Dr. David Haley as my supervisor, for his continued guidance, encouragement and support. As a side, special appreciation is extended for Dr. Haley's faith in my teaching abilities, arranging for me to instruct F R S T 319 (Forestry Economics) while he was on sabbatical. I would also like to express my appreciation to the members of my advisory committee, being Dr. Clark Binkley, Dr. David Cohen and Dr. Russell Uhler. Special thanks are also due to Dr. lian Vertinsky and Dr. Casey Van Koonten at the Forest Economics and Policy Research Unit (FEPA), Dr. William Stanbury at the Faculty of Commerce and Business Administration, and Dr. Bill Wilson at the Canadian Forest Service for their comments and advice on many aspects of my research. As for the many friends and colleagues that have helped make the long process of doing a Ph.D. bearable, I can only say that I could never have done it without your moral support. Although the people that I have had the pleasure to get to know over the years are too numerous to mention, I wish to single out two individuals which have made particularly strong impressions on me, both in a professional and a friendship capacity. Ramvir Singh and Paul Mitchell-Banks, I thank you! I am more than grateful for the financial assistance I have received to reduce the burden of doing a graduate degree. Appreciation goes to the University of BC for the Donald S. McPhee Fellowship, to F E P A for the Forest Economics and Policy Analysis Research Grant, and for the Canadian Forest Service for the FRDA Research Grant. Finally, I wish to express my strongest appreciation of all to my parents, Lloyd and Suzanne. Without their love, support and encouragement, I would never have dreamed of such an ambitious undertaking. Gaston Chapter One Page 1 Chapter 1 Introduction This thesis investigates an aspect of Pacific Rim log and lumber trade which has received surprisingly little attention to date: factor demand estimation with recognition that wood inputs are imperfect substitutes in production. While there have been many studies which have estimated demand parameters for wood inputs, virtually all of them have used highly aggregated trade data (such as "softwood lumber"). The present study investigates demand substitution by product, by region, and by species. 1.1 Motivation for the Research Hiding wood characteristics through data aggregation tends to obscure important dimensions of both forest products' trade and forest policy. A good example of this problem is illustrated by US allegations that BC export restraints on softwood logs constitute a subsidy for BC sawmills. By aggregating all softwood logs, there is a danger of obscuring the log export pattern which might exist in the absence of export restrictions. Kalt's (1994) submission to the US Department of Commerce in the Canada/US softwood lumber countervail case offers a good example of how to improve trade analysis with less aggregated data. He argues that British Columbia (BC) export restraints, which primarily affect coastal logs, including a significant proportion of logs from which clear and merchantable grade lumber can be extracted, do not constitute a subsidy for interior sawmills producing mostly lower grade construction lumber. Another example is offered by the determination of allowable annual cuts within the Gaston Chapter One Page 2 context of BC's sustained yield policy. Haley and Luckert (1994) and van Kooten (1993), for example, argue that meeting the objectives of sustained yield does not simultaneously meet the objectives of sustainable development. In short, choosing forest rotations and/or silvicultural regimes which maximize volume, without any reference to value, does not necessarily promote a strong, forest-based economy. If one adds goals to incorporate non-timber values in forest management, the inherent problems in focusing on physical volume alone become further amplified. There are a number of important questions which require an investigation of trade related to wood species and sources of origin. For example, will BC 's second growth Douglas-fir (Pseudotsuga menziesii (Mirb.) Franco) be able to compete with New Zealand's plantation produced clear radiata pine (Pinus radiata D. Don)? More generally, where have BC's comparative advantages lain in the past, and where are they likely to lie in the future? What will be the economic consequence of BC's transition to a second growth resource, particularly in light of increased environmental pressures to reduce the forest land base? To what degree will non-wood materials substitute for existing forest products produced in BC, and what will be the economic and environmental consequences of such substitutions? How does the emergence of engineered wood products affect demand substitution for BC timber resources? There are two potential situations which will have to be faced as BC makes the transition toward a forest industry that is wholly dependent on second-growth and subsequent forest crops: 1) according to the most recent BC Ministry of Forests timber supply reviews, Gaston Chapter One Page 3 B C is going to witness a significant reduction in the volume of available timber over the next couple of decades; 2) according to Constantino (1986) and Constantino and Haley (1988), without appropriate changes in BC forest policy, the quality^ of timber is going to be significantly lower. If the forest industry in BC is to minimize these potentially negative impacts on the provincial economy, it will be necessary to examine marketing opportunities for the future, and translate these into appropriate land use plans, levels of silvicultural activities, and forest rotations. In other words, it is time for the forest sector to make the transition from a production-oriented to a market-oriented industry. This can only be accomplished by a detailed analysis of which BC wood products have historically contributed most to net revenues, and which are most likely to do so in the future. This need will become increasingly important as old-growth timber becomes scarcer and price increases lead to accelerated substitution. 1.2 Background As can be seen in Figure 1.1, cash prices for lumber more than doubled in the first couple of months of 1993 (following decades of limited price growth). Since then, prices have been extremely volatile, making any forecast of future price trends difficult. There is some debate as to the significance of this price spike. While some believe 1lt is not easy to define quality in a general way. For example, one definition of quality might be the presence of attributes in wood that are related to appearance. In the case of softwood lumber, this would include such characteristics as clear grain, large dimensions, and narrow ring width. Another definition of quality might be structural strength. In other words, quality must be related to the intended purpose of the lumber. Constantino gets around defining quality in terms of specific wood characteristics by using a price index, where quality is related to the buyer's aggregate willingness to pay. Gaston Chapter One Page 4 500 1989 1990 1991 1992 1993 1994 Figure 1.1 Random Lengths S -P-F Lumber Futures, Chicago Mercantile Exchange, Spot Contract*. Compiled from various issues of The Financial Post. * A s this chart always tracks the nearest delivery month, prices are analogous to the cash market. that this occurrence was not all that unusual (see Sohngen and Haynes, 1994), others suggest that the market is displaying a structural change (see Sutton, 1994; and Michaelis, 1994). The latter opinion would suggest that prices will either stabilize at a new plateau at some point in the future or continue to demonstrate real price growth. Historically, the demand for construction lumber has been price inelastic2. This can be explained by one or more of the following: there have been few substitutes (this has not likely been the case); price has not been an issue (e.g., lumber has represented a small 2 A review of the literature which reports historical lumber elasticities, both own-price and cross-price, is offered in Chapter 2. Gaston Chapter One Page 5 portion of the cost of a home); or, there have historically been no inexpensive available substitutes relative to the price of lumber. However, demand for construction lumber may become price elastic (i.e. a structural change) if an increased price level leads to reduced wood consumption through the building of smaller homes and/or lumber substitution. The economic implications of the potential for such wood product substitutes translates into the central theme of this thesis. Substitutes for logs, lumber or further processed wood products can take many forms. The most obvious is substitution with the same basic product, but from a different location. In the literature review offered in Chapter 2, it will be seen that such cross-price elasticities of demand for lumber are significantly higher (even elastic) as compared to the own-price elasticities. In other words, while the quantity of local lumber demanded is not very price responsive (such as the demand for US Midwest lumber given the price of Midwest lumber), the quantity demanded is responsive to the price of similar wood from a different area, such as imports from Canada. For example, a 1% decrease in the price of Canadian lumber may cause the quantity of US Midwest lumber demanded to decrease by more than 1%. Further, the review in Chapter 2 illustrates that cross-price elasticities may be high even for dissimilar types of wood, such as imports of hardwood from Indonesia to replace US consumption of local softwood. Some studies also show that non-wood materials may substitute for logs and lumber. While the apparent willingness to substitute seems rather straight forward, it must be noted that no mention has been made of the specific characteristics of "similar" products. Due to an apparent lack of trade data broken down by grade, little can be found Gaston Chapter One Page 6 in the literature to document this potentially important aspect of substitution. Figure 1.2, showing the prices of three grades of PNW Douglas-fir lumber over the past two decades, illustrates the danger of describing lumber (or logs) as a single homogeneous commodity. Note that these lumber prices are in real terms, not nominal. Over the time period indicated, clear grade Douglas-fir export prices rose roughly 3.5% per annum, the merchantable grade price trend was virtually flat, and the price of the structural grade fell. Given these distinct differences in price trends, it is obviously not rational to expect that construction grade lumber, for example, can fully substitute for clear grades. Prices can also vary tremendously within a grade. For example, prices for clear grade coastal BC lumber of certain species, when the timber from which it is cut is "hand picked" by Japanese buyers, have been reported to exceed $15,000 C D N per thousand board feet (Currie, 19943). 1.3 Scope of the Research Analysis of silvicultural regimes, forest practices, land-use and trade policies are all negatively affected by the lack of information on wood product demand by some measure of disaggregation. However, before research can be carried out which addresses the policy and trade ramifications of using aggregated wood product data, significant background research is needed. This must begin with a quantification of the uniqueness of individual product types (logs, lumber and further processed products), species and source as distinct economic goods. 3Personal communication, Valuation Branch, Timber Pricing Section, BC Ministry of Forests, Victoria. Gaston Chapter One Page 7 2500 2000 . £1500 ihooo 500 . 1972 ' ' 1975 ' ' 1978' ' 1981 ' ' 19'84' ' 19lB7 ' ' 19'90' ' 1993' —*— Clear —•— Merchantable —*— Structural Figure 1.2 PNW Douglas Fir Lumber Prices ($US per thousand board-feet, real, PPI adjusted, 1992=100) Source: Complied from Random Lengths, Various Yearbooks. Grade Definitions: - Clear Douglas-Fir, green, #2 Clear, 15% #3; 2% X 6 and wider; export price, f.o.b. dock, Or. and Wash, (prior to 1985 prices f.a.s.). - Merch. Douglas-Fir, Merch., #1, 15% #2; 6 X 12 and wider; export price, as above (prior to 1985 prices f.a.s., based on #1, 25% #2). - Struct. Douglas-Fir, green, #1 and better, random 10/20; 2 X 4 ; domestic price, f.o.b. mill. Gaston Chapter One Page 8 As the central theme of this thesis is to quantify the degree of substitution of wood products, the sole focus will be on demand descriptors. Further, to keep the analysis manageable, the thesis focuses on a single market—Japan. The Japanese market was chosen as 1) it represents the largest importer of forest products today (Sedjo, 1994); 2) it has an interesting history of evolving from reliance on domestic production, then importation of whole logs, and most recently importation of lumber (which allows for quantification of the substitution between these alternative inputs); and 3) it has been a significant buyer of both construction grade and appearance grade wood products. The balance of this chapter is devoted to defining the problem to be investigated, leading to the research hypotheses. 1.3.1 The Research Problem Figure 1.2 (page 7) helps put the research problem into perspective. By comparing the three grades of Douglas-fir lumber over the past two decades, the growing market premium for the clear grade (and, to a lesser extent, the merchantable grade) is obvious. Although international trade data do not offer such grade detail, they do offer species detail, from which grade measures can often be deduced. For example, the species mix spruce-pine-fir (S-P-F) lumber, which is exported primarily from North America, is known as a construction or structural grade commodity. North American lumber exports to Japan of such species as yellow cedar (Chamaecyparis nootkatensis (D. Don) Spach) and Sitka spruce (Picea sitchensis (Bong.) Carr.), on the other hand, are primarily of clear and merchantable grades. Further, the source of the wood also offers an association with Gaston Chapter One Page 9 grade. New Zealand log and lumber exports, for example, have historically been known to provide sub-structural grades, which have been used in Japan primarily as packaging materials. The research problem is best addressed through a number of questions. For instance, how do the own-price elasticities of demand differ in Japan by product type, source and by species? How do the cross-price elasticities for these wood products differ, both with other wood (type, species and source) and non-wood substitutes? Will scarcity in North American S-P-F lumber lead to real price growth as evidenced in higher grade lumber, or will price rises be met with reduced demand through substitution—both wood and non-wood (Perez-Garcia, 1993; Prins, 1993a and 1993b). Will there be less substitution in the future in species and/or source typical of clear grades as compared to structural? The second ramification of the price trend distinctions shown in the figure is that coastal BC relies on old-growth timber stands for the vast majority of its present lumber production. Timber yielding clear grades is exploited in such stands. Given existing silvicultural efforts and harvest rotations, the supply of clear timber will be significantly reduced as old-growth availability declines. This leads to the question of whether it is possible to generate this high-grade material economically from second growth stands (although this will largely be an implication for further study). Finally, related to the substitution questions above, it must be asked to what extent clear lumber from second growth can compete with clear lumber from old-growth timber (again, this question is posed as a motivation for the present research; the answer can only come from research which Gaston Chapter One Page 10 extends beyond the scope of this study). 1.3.2 Objectives The research problem described in the previous section can be translated into the following three objectives: 1 . To determine own-price and cross-price demand elasticities in Japan for logs, lumber and other wood products by region and species. Cross-price demand elasticities include substitution with Japanese domestic logs, and substitution with non-wood products. 2. To qualitatively extend objective one to explore Japan's demand for broad grade categories (construction versus appearance). 3. To explore the implications of the above for BC forest industry strategy and public forest policy. 1.3.3 Hypotheses The research hypotheses to be tested are as follows: 1 . BC wood species, in the form of logs, lumber or further processed products, behave as distinct economic goods, as measured by own-price and cross-price elasticities of demand, in the Japanese market. 2. The market share of BC wood species in Japan, in the form of logs, lumber or further processed products, is dependent on the individual prices relative to other species and wood products in Canada and around the world. 3. Japan's wood product import mix is affected by Japan's domestic log supply and non-wood alternatives. 4. Structural changes in international markets for logs, lumber and panels have affected price levels and trends for these products. Gaston Chapter One Page 11 1.4 Organizat ion of Thesis The objectives/hypotheses of the previous sections are addressed and presented in this thesis as follows. Chapter 2 offers a literature review of North American studies which have estimated log or lumber elasticities of demand, as well as a discussion of the implications of such studies for the methodological approach to be used in the present study. In Chapter 3 the Japanese wood products market is described. Chapter 4 describes the theoretical foundation, data and the empirical model in detail, while Chapter 5 presents the results of this quantitative analysis. Finally, Chapter 6 offers a discussion of the results, followed by a summary, including limitations and implications for further research in Chapter 7. Gaston Chapter Two Page 12 Chapter 2 Literature Review on Log and Lumber Substitutions and Implications for Further Research This chapter reviews the literature on log and lumber demand elasticity estimates, including a cross-section of the methodological techniques used. The chapter concludes with methodological implications for the present study. 2.1 The Econometric Estimates of the Price Elasticity of Demand There are a number of potential responses to a rise in the price of domestic lumber. These responses can be placed in one or more of the following categories: 1) increased efficiency of wood use and/or reduced consumption of finished products; 2) substitution by wood from another location, different form (for example, lumber for logs) or different species; and 3) substitution of non-wood inputs for wood inputs. In this section, these possibilities are examined separately in Sections 2.1.1 through 2.1.3, respectively. 2.1.1 Estimates of Wood for Wood Substitutes There has been a considerable amount of research which investigates demand elasticities for timber products, primarily for US lumber. Tables 2.1 through 2.3 highlight some of this research. The summary information is largely adapted from the review by Phelps (1993). The most obvious consistency found in Table 2.1, which presents own-price demand elasticity studies, is that softwood lumber demand is shown to be inelastic, with Gaston Chapter Two Page 13 TABLE 2.1 Own-price Elasticity of Demand for Softwood Lumber in N.A. Elasticity Time Frame Author Comments -0.173 1947-1974 McKi l lopefa/ . (1980) US softwood lumber wholesale price index -0.35 1950-1974 Waggener et al. (1978) US softwood lumber price -0.38 07/79-12/84 Gellner etal. (1991) US softwood lumber price -0.075 1947-1974 Adams (1977) US softwood lumber price index, 1 year lag -0.285 -0.162 -0.130 -0.111 1950 (Point) 1960 ( " ) 1970( " ) 1980( " ) Spelter (1985) US softwood lumber price -0.88 -0.39 1950-1954 1970-1974 Spelter (1985) US softwood lumber price -0.91 01/68-12/77 Rockel and Buon-giorno (1982) US Douglas-fir wholesale price index -0.88 1947-1967 Robinson (1974) Douglas-fir -0.667 -0.149 01/77-12/87 Lewandrowski (1992) <- Southern pine *- Douglas-fir -0.55 -1.15 1950-1987 Adams etal. (1992) <- Residential construction <- Non-residential construction Canada .0,. i -0.29 01/71-02/82 Jacques etal. (1982) Domestic softwood lumber purchases -0.023 1970-1982 Sharma (1986) Softwood lumber for residential construction -0.05 01/71-02/92 Prins(1993) Total shipments of softwood lumber less exports a range of-0.023 to -1.15, and an average of roughly -0.4. In other words, these studies suggest that a 10% increase in the domestic price of lumber will decrease the local quantity Gaston Chapter Two Page 14 demanded by an average of 4%. Not surprisingly, demand elasticities which are smaller in value are from studies which estimate short-run demand responses. In the very short run substitutes do not, for all practical purposes, exist. Measurements of elasticities must also consider the specific years being investigated. Spelter's (1985) results illustrate this by showing that elasticities (in this case share elasticities) have fallen over time. This trend, according to Spelter, may be attributable to improved technology and utilization, both of which have helped to alleviate scarcity. One must also note that elasticity estimates are affected by the price range over which they are measured. For example, elasticity estimates derived from data where prices tended to be low will likely be significantly different from a comparable study over a different time period where prices tended to be high. The range of elasticity estimates shown can also be partially explained by what is being measured. For example, Adams, et a/.'s (1992) results show that the demand for lumber used in residential construction is less elastic than for non-residential construction, supporting the point that home buyers are not greatly influenced by the price of an input which makes up a relatively small portion of the total purchase price, as well as the fact that in non-residential construction more substitutes in the form of non-wood materials are available and acceptable. Lewandrowski's (1992) results show that elasticities can vary by species, here suggesting that southern pines (e.g., Pinus taeda) have more substitutes (i.e., the quantity demanded is more price sensitive) than Douglas-fir. This point is paramount to the main theme of this thesis, and will be further explored throughout much of this chapter. Gaston Chapter Two Page 15 T A B L E 2.2 Cross-price Elasticities of Demand for Similar Lumber in Different Regions Elasticity Time Frame Author Comments United States 1.283 1947-1974 Adams (1977) j Imports from Canada; ratio of US to I Canadian import prices. 0.81 01/71-02/82 Jacques, et al. (1982) I Demand for Canadian shipments; US I lumber price index. 1.48 01/74-01/86 Buongiorno et al. j Demand for imports from Canada; prices (1988) j of softwood lumber in the US. 0.56 1950-1982 Singh and Nautiyal | Demand for Canadian lumber; US price (1986) | index for all lumber. -0.80 -1.95 1963-85 Flora etal. (1991) i <- Offshore demand facing the US in 1987; performance grade, j <- Off shore demand facing the US in 1987; construction grade. -3.088 2.27 1965-1985 Chen et al. (1988) I <- Demand for Canadian softwood lumber; import price from BC. j <- Price of US softwood lumber. 4.39 Sawnwood 12.30 Plywood 1975-1985 Constantino (1988) i World imports of hardwood from | Indonesia; importing country's price of | hardwood. Note: When interpreting the sign of the elasticity values, it must be noted which price is being considered. With Chen, ef a/.'s (1988) results, for example, a. 10% decrease in the BC price of lumber will increase US import demand by over 30%. Conversely, a 10% decrease in the US price of lumber will decrease US import demand from Canada by over 22%. Table 2.2 shows some estimates of cross-price elasticities of demand for softwood lumber in different regions, in this case primarily the US demand for imports from Canada. The most obvious point is that the demand response is now elastic (greater than 1), or at least more elastic than for own-price. This clearly shows, as would be expected, a willingness to substitute for a similar commodity from another geographic area. One might expect Gaston Chapter Two Page 16 T A B L E 2.3 Cross-price Elasticity of Demand for Different Lumber Elasticity Time Frame \ Author | Comments 1.30 Sawnwood 0.75 Plywood 1975-1985 | Constantino (1988) j World imports of hardwood from j Indonesia relative to importing j country's price of softwood. 1.45 1971-1991 j Brooks (1993) | US imports of tropical lumber relative I to US price of softwood lumber. 1.06 1970-1989 I Youn and Yum I (1992) I Korean imports of hardwood logs ! relative to the import price of softwood I logs. these elasticity values to be higher than indicated. That they are not suggests that while the imports are good substitutes for domestic production, they are far from being perfect substitutes. This may be partially due to the fact that aggregate imports in these studies are not equivalent to aggregate domestic production, neither by species nor other characteristics. Finally, Table 2.3 shows some estimates of cross-price elasticities of demand for different types of lumber. Although the estimates are mostly greater than unit, the demand is generally less elastic than for similar wood products in different regions. This suggests that consumers are less willing to substitute hardwoods for softwoods than Canadian S-P-F for US southern pine. 2.1.2 Est imates of Non-Wood for Wood Substi tutes Another possible reaction to higher lumber prices is, of course, substitution for wood products with non-wood commodities. For construction lumber, this includes steel, Gaston Chapter Two Page 17 concrete, bricks, plastics, etc. Recent price increases in lumber have already initiated an extensive program by the American Iron and Steel Institute to promote the replacement of wood structural and non-structural members in construction with steel members (Haws, 1994). Surveys undertaken in 1993 indicate that 45% of builders in California would consider switching to steel due to the high and unstable price of wood products. There are a number of studies which have examined the cross-price demand elasticities between wood and non-wood materials. A better understanding of the substitution impact is achieved when the extent to which non-wood materials substitute for wood is examined, and the extent to which wood substitutes for non-wood materials. Table 2.4 summarizes the results of McKillop, et al. (1980) for the US. First, it should be noted that all of the own-price elasticities (the diagonal from top-left to bottom-right) have the expected negative signs and are all less than 1, indicating inelastic demands. Second, the table indicates that the price of lumber influences the quantity demanded of non-wood materials. By contrast, however, the price of substitutes T A B L E 2.4 Own-price and Cross-price Demand Elasticities for Construction Materials Q Lumber Q Plywood Q Steel Q Aluminum Q Concrete P Lumber -0.17 0.79 0.24 -0.54 P Plywood 0.14 -0.67 -0.4 0.54 P Steel 0.37 -0.93 0.74 P Aluminum 0.02 0.47 -0.83 P Concrete -0.51 Source: McKillop, etal. (1980) Gaston Chapter Two Page 18 T A B L E 2.5 Own-price and Cross-price Demand Elasticities for Construction Materials Q Lumber Q Plywood Q Non-Wood P Lumber -0.91 0.05 0.12 P Plywood 0.09 -0.95 0.12 P Non-Wood 0.09 | 0.05 -0.88 Source: Rockel and Buongiorno (1982) does not influence the quantity demanded of lumber to the same degree. For example, a 10% increase in the price of steel will increase the demand for lumber by 3.7%. Conversely, a 10% increase in the price of lumber will increase the demand for steel by 7.9%. Rockel and Buongiorno (1982) specifically examined the demand for wood products for residential construction (as opposed to total US demand) (Table 2.5). The non-wood substitutes included in the study were structural steel, cement, bricks, plumbing and heating fixtures, and selected fabricated metal products. The extremely low cross-price elasticities indicated in the table are the result of aggregating all these inputs into a single basket of goods. In Table 2.6, the result of time on elasticities (primarily technological change) is demonstrated by Spelter for the US. As can be seen, both the own-price and the cross-price elasticities fell (with the exception of concrete) from the 1950s to the 1980s. Finally, Prins (1993) examined wood/non-wood substitution in Canada (Table 2.7). Note that while the results suggest that a 100% increase in the price of lumber will cause Gaston Chapter Two Page 19 T A B L E 2.6 Own-price and Cross-price Demand Elasticities for US Softwood Lumber 1950 1960 1970 1980 P Lumber -0.285 -0.162 -0.13 -0.111 P Plywood 0.109 0.04 0.009 0.004 P Steel 0.026 0.017 0.012 0.005 P Concrete 0.006 0.006 0.006 0.006 Source: Spelter (1985) T A B L E 2.7 Own-price and Cross-price Demand Elasticities for Selected Canadian Construction Materials Q Lumber Q Brick Q Cement Q Steel P Lumber -0.05 0.51 0.15 0.32 P Brick 0.49 -0.3 0.73 P Cement 0.78 0.71 -0.5 P Steel 0.55 0.56 -2.09 P Gypsum -0.31 P Panels 0.09 0.08 Source: Prins (1993) only a 5% reduction in the demand for lumber, it also creates a 51% increase in the demand for bricks, a 15% increase in cement and a 32% increase in steel. This demonstrates the dominance of wood use in construction: 5% of all lumber used in construction is a significant volume of material relative to 32% of all steel used. Also, note that price increases in non-wood substitutes have a greater impact on the demand for wood than suggested by the previous studies. Gaston Chapter Two Page 20 2.2 Parametric Demand Elasticity Estimation Techniques From the outset, it should be emphasized that only a small number of the demand elasticity studies listed in the previous tables made any attempt to disaggregate the data beyond softwood logs or lumber. As mentioned in the previous chapter, the likely reason for this is the lack of published disaggregated data. All of the studies reported to this point have involved econometric techniques in estimating the price elasticities of demand. While the estimation techniques employed were not unusual (normally ordinary least squares, two or three stage least squares, non-linear least squares or generalized least squares), a few studies utilized an approach of interest to the present study. Flora and his colleagues at the USDA Forest Service, Pacific Northwest Research Station in Seattle have conducted North American wood product demand studies which have made direct reference to quality or grade (Flora and Lane, 1994; Flora, et al., 1993; Flora, 1993; Flora, 1992; Flora, etal., 1991-a; Flora, etal., 1991-b; Flora, 1991; Flora, et al., 1990; Flora and McGinnis, 1989; Flora, 1986). In one of the studies (Flora, et al., 1991-b), the researchers developed export supply functions for Pacific Rim log suppliers (US, Canada, Chile, New Zealand and the Soviet Union), and import demand functions for Pacific Rim buyers (Japan, Korea, China and Taiwan). These supply and demand functions were then summed across quantities to yield aggregate demand and supply functions. To estimate trade flows pertinent to an individual region, the demand or supply facing that region is developed by netting out all of the other regions' demand and supply functions. The individual equations used by Flora tend to be very simple, with quantity Gaston Chapter Two Page 21 demanded typically a function of price, G D P and housing starts, and quantity supplied a function of plantation area, timber harvest, and possibly a sawmilling cost index. Flora's methodology suffers from three specific limitations. First, where price projections are made, projections of all variables except price and volume are done outside of the model (by making assumptions relative to a present-day "base case"). This means either heavy reliance on other studies, use of other modelling techniques, or significant personal judgement. The second limitation is Flora's method of dealing with disaggregated trade data. Notes Flora, etal. (1991-b, page 6): Because log-trade data are rarely reported by grade, quality class, or economic category, it was necessary to judge the proportions and relative values. ...Future volume-share shifts among grades were assumed... This, unfortunately, offers little guidance for dealing with such data over a wide range of applications. Finally, Flora's models are limited to a single grade at a time. This does not allow for the measurement of cross-price elasticities of demand across grades or species. However, the own-price elasticities which resulted do suggest that elasticities of demand are negatively correlated with quality (as quality increases, demand becomes more inelastic). Flora (1991-b) categorized logs into one of four grades: Select logs whose value derives from "appearance" grade lumber; Performance Coast and Cascade Grade No. 2 sawlogs; second- and old-growth logs with scaling diameters between 12 and 24 inches; Gaston Chapter Two Page 22 Construction Coast Grade No. 3 sawlogs; second-growth logs with scaling diameters between 6 and 12 inches; Utility submerchantable in the export market. Flora's conclusions suggest that the performance grade will have rising real price growth through the turn of the century, and that the construction grade will see declines due to international competition. As shown in Table 2.2, the authors pegged the price elasticity of demand facing the US in 1987 for the performance grade at -0.80 as compared to -1.95 for construction (the author did not analyse the two extremes in grades, reasoning that selects will always be scarce and that the utility grade is unimportant for the export market). Haynes and Fight (1992), also working with the USDA Forest Service in the PNW, conducted a study on projecting prices of selected grades of Douglas-fir, Coast Hem-fir, Inland Hem-fir and ponderosa pine lumber. Working with historical volumes and prices from representative invoices submitted to the Western Wood Products Association for the P N W region4, the authors estimated the relationships between the prices of the selected lumber grades and the price of the dominant lumber grade for each species in the general form: Sjt = by + b2j Sdt + bZj Wfi (2.1) where: S j t = regional lumber price for the j t h species and grade in year t; S d t = price of the dominant species and grade in year t, and; W j t = the proportion of total lumber production in year t that comes from j t h species and grade. "Unlike the other methodologies reported in this section, Haynes and Fight (1992) are using domestic market data (as opposed to export data). Given the noted premium for export markets (see Flora, et al., 1993), this will underestimate the price premium for higher grades. Gaston Chapter Two Page 23 With the values of the by's estimated, the authors predicted the price for each grade by independently projecting the price of the dominant grade (S d t) and grade production proportions. Their results supported the notion that increasing scarcity of high-grade material will result in higher prices. However, given the confines of their analysis, projections out to the year 2040 showed that the relative price spread for each grade remains virtually unchanged. This is in contrast to historical changes in price spreads over grades. As noted by Flora, for example, in Japan in 1978, Alaska Prime Spruce cants were worth 3 times as much as US #3 hemlock logs; by 1992, the multiple increased to 20. Sedjo, et al. (1994) offer a study which specifically investigates cross-price elasticities of wood inputs. Noting the effect of the price of US logs on the log import behaviour of Japan, the authors reason that imports from any region will be a function of that region's timber price to Japan, the price of Japanese domestic timber, the level of construction activity in Japan, the price of US timber to Japan, and the price of timber to Japan from any other source that may substitute for the region in question's timber. A multiple regression analysis was used, with the quantity of timber demanded from region "A" as the dependent variable, and each of the above factors taken as independent variables, over the period 1970-1991: QA = b., + b2 PA + bz PUS + bA PJ + b5 HS + £ c,P(. <2-2) where: QA = quantity imported from region "A"; PA = price of region A 's timber in Japan; P U S = price of US logs to Japan; PJ = Japanese domestic price of logs; HS = number of Japanese new wooden housing starts; Pi = prices of timber to Japan from regions other that "A" or the US. Gaston Chapter Two Page 24 Interestingly, the results showed no significance for the domestic price in Japan ("sugi" conifer logs), suggesting either that import decisions were made independently of domestic production (possibly suggesting different end uses), or that perhaps there was too little variation in domestic production to estimate the effect. The price parameter on imports from Canada (Douglas-fir lumber) was also found to be insignificant, with the authors reasoning a high degree of multicollinearity of Canadian lumber exports (roundwood equivalent) to Japan and US Douglas-fir logs exported to Japan 5 . The other regions investigated, all of which were found significant, included the Philippines, Indonesia, New Guinea, and Malaysia, collectively called "tropical" (lauan veneer tropical logs); Russia (larch logs); and Chile/New Zealand, collectively called "radiata". The elasticities of Japanese imports from these three regions/wood type with respect to the independent variables were then calculated. Due to the objective of the study, the authors focused on the cross-price elasticity of Japanese imports with respect to the price of US logs. This cross-price elasticity of wood quantity from region "A" (QA) with respect to the US log price (PUS) is the percentage change in Japanese imports from A for each unit percentage change in the US price, given by: E, = b 3 (PUS/QA) (2.3) QA.PUS where: b 3 = the regression coefficient of P U S in the previous equation; (PUS/QA) = value using the mean over the study period. 5As will become apparent in the following review of Armington (1969) and applications by Chou and Buongiorno (1983) and Hseu and Buongiorno (1992), such multicollinearity is a major limitation of trade studies which have many price regressors of "similar" products in the same equation. Gaston Chapter Two Page 25 Using ordinary least squares, the recovered cross-price elasticities for Japanese imports with respect to the US log price were 0.58, 5.0, and 0.84 for tropical, radiata and Russia, respectively. Unfortunately, none of the methodologies reviewed to this point are totally adequate for addressing the main objective of this thesis, that is to estimate the substitution possibilities among a significant number of disaggregated wood inputs in the Japanese market. The problem which needs to be addressed is how to estimate own- and cross-price elasticities for a number of potentially unique, yet similar, price series. Armington (1969) offers a potential solution. Recognizing the heterogenous nature of products, even when of a similar "kind", Armington relaxes the assumption that a particular "good" produced in a particular country is a perfect substitute for the "same" good produced in another country (relaxing the assumption that the elasticity of substitution between, say, softwood lumber from Canada and softwood lumber from New Zealand is infinite). This is accomplished by assuming that an importer performs a two-stage optimization process. In stage one, the importer decides the total amount of the commodity "kind" to import from all sources (say, softwood lumber, or even all wood products in aggregate). In the second stage, the importer determines the optimal levels of "good" imports (say, softwood lumber from a number of different sources). Armington makes four assumptions, "systematically simplifying the product demand functions to the point where they are relevant to the practical purposes of estimation and forecasting" (Armington, 1969; page 160). The first is that importer preferences are homogeneously separable. This assumption is necessary to incorporate two-stage Gaston Chapter Two Page 26 optimization. It is next assumed that market shares depend only on relative prices of the products in the market, not on the size of the market itself. In the absence of these assumptions, a country would have mn demand functions in the general form: Xij = Xjff),P^vP^ ••• 'P-inr ^21 ' ^22 '^2m' ••• ' ^n1'^>n2' •••^nrr) where: m = number of supplying countries (specific product); n = number of goods (groups of products); Xjj = specific product demand; D = income; P. = specific product price. With Armington's first two assumptions, this is reduced to m+n demand functions: Xij ~ Xij ( p.. p.. p.. \ im (2.5) where X, = X / ( D , P 1 , P 2 P„) (2-6) and: X; is any good, and is any product. From Armington's first two assumptions, Equation 2.5 requires that a linear, homogeneous quantity index function, cp, be utilized for each market, such that X; = cp (XM, X i 2 , X i m ) . The reason for this is that if imports of products of the same kind from different countries are considered to be imperfect substitutes, the arithmetic sum of various imports would not be an appropriate index of total imports. Recognizing that equation 2.5 is still too complicated to be of practical use when more than a few countries are identified, Armington proposes two further simplifying Gaston Chapter Two Page 27 assumptions. These are that the elasticities of substitution are constant for each market, and that the elasticity of substitution between any two products competing in a market is the same as that between any other pair of products competing in the same market. These assumptions are equivalent to specifying that the cp's above are constant elasticity of substitution (CES) functions, having the general form 6: =-