UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Twisted energy ladder : complexities and unintended consequences in the transition to modern energy services Kowsari, Reza 2013

You don't seem to have a PDF reader installed, try download the pdf

Item Metadata

Download

Media
24-ubc_2013_fall_kowsari_reza.pdf [ 3.11MB ]
Metadata
JSON: 24-1.0071978.json
JSON-LD: 24-1.0071978-ld.json
RDF/XML (Pretty): 24-1.0071978-rdf.xml
RDF/JSON: 24-1.0071978-rdf.json
Turtle: 24-1.0071978-turtle.txt
N-Triples: 24-1.0071978-rdf-ntriples.txt
Original Record: 24-1.0071978-source.json
Full Text
24-1.0071978-fulltext.txt
Citation
24-1.0071978.ris

Full Text

TWISTED ENERGY LADDER: COMPLEXITIES AND UNINTENDED CONSEQUENCES IN THE TRANSITION TO MODERN ENERGY SERVICES  by Reza Kowsari  B.Sc., Sharif University of Technology, 1996 M.A.Sc., The University of British Columbia, 2006  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF  DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE STUDIES (Resource Management and Environmental Studies)  THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)  April 2013  © Reza Kowsari, 2013  Abstract More than one-third of the world’s population has inadequate access to modern energy services and suffers as a consequence. A better understanding of energy transition is vital for improving future programs. This thesis investigates the challenges of transitioning to modern energy services with the goal of informing policy-making. Chapter 2 is a review of the literature on energy analysis and energy system uptake by the households with a particular focus on rural regions. Building upon the findings in the literature, a new conceptual framework is developed that can act as a basis for developing new empirical and theoretical models of household energy system uptake and use. In Chapter 3, the private sector based approach to rural energy provision (i.e. electrification and Improved Cookstove (ICS) dissemination) is investigated by examining two enterprises in India. Results indicate that the private sector based approach to rural energy provision cannot be a universal solution. It further shows that enterprises are facing many challenges that are beyond their capacity to address, necessitating alternate approaches to private sector involvement. Chapter 4 and 5 investigate the uptake of ICS by commercial kitchens in Bangalore and its potential health implications. The attributes of ICS and Liquefied Petroleum Gas (LPG) stove, as perceived by stove users and owners, are explored and the reasons for purchasing ICS are assessed. Results indicate that both groups mostly prefer LPG stoves and that ICS uptake is mainly motivated by economic factors. The potential health implications of this switch in the commercial kitchens are explored by investigating the stove’s emission characteristics (based on both secondary data and direct emission measurements), the stoves smokiness (as perceived by users and owners); and some health symptoms associated with stove smoke. Uptake of ICS by these commercial kitchens is found to potentially have adverse health implications. In brief, this thesis concludes that rural energy provision policies can be improved through a greater emphasis on the human dimension, comprehensive assessment of the target population, and ongoing evaluation of the programs’ outcomes given the major ii  challenges in improving rural energy access and possibilities for spillovers into other market segments.  iii  Preface With the exception of chapters 1 and 6, all other chapters were originally written as manuscripts for peer-reviewed publication. Details for co-authorship of Chapters 2 through 5 are outlined below. Chapter 2. A version of this chapter has been published: Kowsari, Reza; Zerriffi, Hisham. Three dimensional energy profile: A conceptual framework for assessing household energy use J Energy Policy 2011, Volume 39, Issue 12, 7505-7517. The development of this chapter originated from the literature review I conducted for my comprehensive exam. My advisor Dr. Hisham Zerriffi contributed to this chapter through development of ideas and research questions. Dr. Zerriffi provided comments and edits to the manuscript. Chapter 3: A version of chapter 3 was developed as a manuscript and is in the process of being submitted to an academic journal. I developed the research questions and the data collection method under the supervision of Dr. Hisham Zerriffi. I designed the interview protocol and collect data through semi-structured interviews with a number of individuals in India. I transcribed the interviews, performed the analysis and drafted the manuscript. Dr. Zerriffi provided comments and edited the manuscript. The UBC Behavioral Research Ethics Board (BREB) designated the research ‘Minimal Risk’, and certificate number H11-01211 was awarded on July 18, 2011. Renewal of the certificate was obtained on June 7, 2012. Chapter 4: A version of chapter 4 was developed as a manuscript and is in the process of being submitted to an academic journal. I developed the research questions and the data collection method under the supervision of Dr. Hisham Zerriffi. I designed the survey instrument and IMRB international translated into Hindi. Data was collected by IMRB Iinternational, based on Chennai. I supervised the survey and modified the questions in the preliminary round. I analyzed the survey data and wrote the manuscript, and Dr. Zerriffi provided comments and edits to the manuscript. The UBC Behavioural Research Ethics Board (BREB) designated the research ‘Minimal Risk’, and iv  certificate number H11-03132 was awarded on December 8, 2011. Renewal of the certificate was obtained on November 27, 2012. Chapter 5: A version of chapter 5 was developed as a manuscript and is in the process of being submitted to an academic journal. I developed the research questions and the data collection method under the supervision of Dr. Hisham Zerriffi. I designed the survey instrument and IMRB international translated into Hindi. Data was collected by the IMRB international based on Chennai. I supervised the survey and modified the questions in the preliminary round. I designed the experiments and sampling method for the direct emission measurement. The emission measurement was done by a technician from Resource Optimization Initiative (ROI) India, and the measurement equipment and the test protocols were developed by Dr. Andy Giershop (Assistant Professor at North Carolina State University) and Dr. Julian Marshall (Assistant Professor at the University of Minnesota). The filters were processed and their weight before and after the experiment were measured by The Occupational and Environmental Health Laboratory in the School of Population and Public Health (UBC). I analyzed the survey data and wrote the manuscript, and Dr. Zerriffi provided comments and edits to the manuscript. The UBC Behavioural Research Ethics Board (BREB) designated the research ‘Minimal Risk’, and certificate number H11-03132 was awarded on December 8, 2011. Renewal of the certificate was obtained on November 27, 2012.  v  Table of Contents Abstract .............................................................................................................................. ii Preface ............................................................................................................................... iv Table of Contents ............................................................................................................. vi List of Tables ..................................................................................................................... x List of Figures ................................................................................................................. xiii List of Abbreviations ..................................................................................................... xvi Acknowledgments .......................................................................................................... xix Dedication ....................................................................................................................... xxi Chapter 1  Introduction ............................................................................................... 1  1.1  Overview ............................................................................................................. 1  1.2  Human welfare and energy system ..................................................................... 2  1.3  The current energy situation in the developing world and what is expected ...... 5  1.4  Providing access to modern energy services ...................................................... 8  1.4.1  Electrification .................................................................................................. 9  1.4.2  Clean cooking energy system ......................................................................... 9  1.5  Understanding uptake and usage of energy systems......................................... 10  1.6  Thesis structure and overview .......................................................................... 12  Chapter 2 Three dimensional energy profile: a conceptual framework for assessing household energy use ...................................................................................... 15 2.1  Introduction ....................................................................................................... 15  2.2  Energy analysis and underlying assumptions ................................................... 17  2.2.1  Physical-Technical-Economic models (PTEM) ........................................... 17  2.2.2  Psychology based approaches ....................................................................... 19  2.2.3  Sociological and anthropological models ..................................................... 20  2.2.4  Integrated approaches ................................................................................... 20  2.2.5  Studies on rural household energy use in developing world ........................ 22  2.3  Energy transition in the developing world ........................................................ 23  2.3.1  Energy ladder ................................................................................................ 23  2.3.2  Energy stacking model .................................................................................. 25 vi  2.4  Factors determining household energy choice .................................................. 28  2.4.1  Endogenous factors (household characteristics) ........................................... 28  2.4.2  Exogenous factors (external conditions)....................................................... 31  2.5  An integrated approach to understanding household energy choice ................ 35  2.5.1  Shortcomings of the literature on rural household energy analysis .............. 35  2.5.2  A new approach to household energy analysis ............................................. 42  2.5.3  Energy transition in a three dimensional energy profile model .................... 46  2.6  Conclusions ....................................................................................................... 47  Chapter 3  Private sector involvement in rural energy provision ......................... 49  3.1  Introduction ....................................................................................................... 49  3.2  Background ....................................................................................................... 50  3.2.1  Electrification ................................................................................................ 50  3.2.2  Improved cookstove ...................................................................................... 53  3.3  The root of the private sector based approach to energy access provision ....... 54  3.4  Case study ......................................................................................................... 58  3.4.1  Biomass energy and energy poverty in India ................................................ 58  3.4.2  Background information on the commercial entities .................................... 61  3.4.3  Area of operation .......................................................................................... 67  3.4.4  Data collection .............................................................................................. 68  3.5  Results and discussions ..................................................................................... 70  3.5.1  Private sector based approach ....................................................................... 70  3.5.2  Challenges faced by the commercial entities ................................................ 73  3.5.3  Challenges of enterprises in India ................................................................. 78  3.5.4  Capital and finances ...................................................................................... 78  3.5.5  Policy and institutional issues ....................................................................... 80  3.5.6  Market ........................................................................................................... 83  3.5.7  Operational issues ......................................................................................... 87  3.6  Conclusions ....................................................................................................... 90  Chapter 4 Descending the "energy ladder": the case of biomass cook stove uptake by restaurants in Bangalore, India ................................................................... 94 vii  4.1  Introduction ....................................................................................................... 94  4.1.1. Energy system attributes ............................................................................... 94 4.1.2. Switching back .............................................................................................. 96 4.1.3. Energy system uptake by commercial entities .............................................. 96 4.2  Case study ......................................................................................................... 98  4.2.1  ICS provision in India ................................................................................... 99  4.2.2  The cook stoves........................................................................................... 100  4.3 4.3.1 4.4  Data collection ................................................................................................ 101 Survey ......................................................................................................... 102 Results and discussions ................................................................................... 104  4.4.1  Perception of stove attributes ...................................................................... 105  4.4.2  Reasons to buy and not to buy .................................................................... 115  4.4.3  Economic propositions................................................................................ 119  4.4.4  Health .......................................................................................................... 122  4.5  Conclusions ..................................................................................................... 125  Chapter 5 Potential health implications of switching from LPG to improved biomass cook stoves: the case of commercial kitchens in Bangalore, India ............ 128 5.1  Introduction ..................................................................................................... 128  5.2 Background: IAP health implications, stove emissions, and exposure assessment ................................................................................................................... 129 5.2.1  Health implications of exposure to biomass combustion ........................... 130  5.2.2  Biomass combustion ................................................................................... 130  5.2.3  Exposure assessment ................................................................................... 132  5.3 5.3.1 5.4  Data collection ................................................................................................ 133 Direct emission measurement ..................................................................... 133 Results and discussion .................................................................................... 135  5.4.1  Secondary data on cook stove characteristics ............................................. 136  5.4.2  Direct emission results ................................................................................ 138  5.4.3  Health effects .............................................................................................. 144  5.5  Conclusions ..................................................................................................... 157 viii  Chapter 6 6.1  Conclusions ............................................................................................ 160  Summary of findings and concluding remarks ............................................... 160  6.1.1  Knowledge gaps in understanding the adoption of energy systems ........... 160  6.1.2  Private sector based approach: great potential, but not universal ............... 162  6.1.3  The improved cookstove: a device to ascend and descend the energy ladder? 165  6.1.4 Replacing LPG use with biomass, even in advanced ICSs, has resulted in increased indoor air pollution, which potentially has adverse health impacts ........ 167 6.2  Limitations and direction of future research ................................................... 168  6.3  Incorporating the research findings into policies and programs ..................... 171  References ...................................................................................................................... 173 Appendices ..................................................................................................................... 195 Appendix A  Supporting documents for chapter 3 ................................................... 195  A.1 Interview protocol for stove developers .............................................................. 195 A.2 Interview protocol for power enterprises ............................................................. 217 Appendix B  Supporting documents for chapter 4 ................................................... 237  B.1 Survey instrument for enterprises in power sector ............................................... 237 B.2 Analysis of Stanford survey on uptake of stove X by residential consumers ...... 332  ix  List of Tables Table 1-1: Population lack access to clean cooking facilities now and in 2030 (millions) (IEA 2012) .................................................................................................................. 6 Table 1-2: Population currently lacking access to electricity and population expected to be without electricity access under the “New Policies Scenario” in 2030 (in millions) (IEA 2012) .................................................................................................................. 7 Table 2-1: Summary of factors determining household energy choice ............................ 28 Table 3-1: Private entities included in the case study ....................................................... 62 Table 3-2: Topics covered in the interview protocol ........................................................ 68 Table 3-3: Interviewee’s position, number of interviews, and method of each of interview .................................................................................................................................. 69 Table 3-4: Factors affecting businesses in low-income energy markets .......................... 74 Table 4-1: Survey questions and respondents ................................................................. 103 Table 4-2: Stove attributes, categories, and short name ................................................. 105 Table 4-3: LPG stove attributes according to users and owners of LPG kitchens ......... 106 Table 4-4: Descriptive statistics of LPG stoves and stove X attributes as reported by owners ..................................................................................................................... 109 Table 4-5 Descriptive statistics regarding LPG stoves and stove X attributes as reported by users ................................................................................................................... 111 Table 4-6: Paired sample test of score differences between LPG and stove X users and owners (Wilcoxon matched paired signed-rank test).............................................. 112 Table 4-7: Reasons to buy stove X ................................................................................. 115  x  Table 4-8: The price of each MJ delivered by LPG stoves and stove X ........................ 120 Table 4-9: Frequency of owners and users in stove X and LPG kitchens being bothered by stove smoke........................................................................................................ 123 Table 4-10: A summary of the findings .......................................................................... 125 Table 5-1: CO, CO2, PM emissions,1 and water boiling (WB) efficiency for stove X and LPG stoves, as reported in the literature. ................................................................ 137 Table 5-2: Emission concentrations of three tests .......................................................... 139 Table 5-3: CO to CO2 ratio (molecular ratio to CO2) .................................................... 140 Table 5-4: PM2.5 mass (pre- and post-test filter weight differences), temperature as recorded during the tests ......................................................................................... 142 Table 5-5: CO and CO2 emission factors (grams per kg of fuel) for stove X/pellet and LPG stove/fuel combinations.................................................................................. 142 Table 5-6: Summarized statistics of participants and work environment (i.e. kitchen) characteristics .......................................................................................................... 144 Table 5-7: Bivariate correlation test between continuous variables ............................... 145 Table 5-8: Mean scores of stove smokiness s perceptions (1 (strongly disagree) to 5 (strongly agree)) ...................................................................................................... 147 Table 5-9: Summarized statistics of health symptoms reported by stove users ............. 147 Table 5-10: Cross tabulation of being troubled by a cough and bringing up phlegm .... 148 Table 5-11: Bivariate correlation coefficient of two health symptoms and stove use category ................................................................................................................... 150 Table 5-12: Summarized statistics of variables used in the model ................................. 153  xi  Table 5-13: Regression model A) eye symptom, (eye irritation) ................................... 154 Table 5-14: Mean comparison of three cluster solutions ................................................ 155 Table 5-15: ANOVA table of the means of each cluster in the three cluster solutions based on four variables: kitchen volume, total number of meals prepared per day, number of exhaust fans in the kitchen (the room in which stoves are mainly used), and the time per day that the user spends cooking in the kitchen ........................... 155 Table 5-16: Regression model B) Respiratory symptom, (coughing) ............................ 156 Table 5-17: Concluding remarks .................................................................................... 158 Table 6-1: Recent major trends in electrification and clean cooking energy provision . 162  xii  List of Figures Figure 2-1: The classic energy ladder ............................................................................... 24 Figure 2-2: The illustration of energy stacking- including few examples of energy systems. Adopted from (IEA 2002) and “rural energy ladder” (Barnes and Floor 1996), Note: ICT is information and communication technology. ........................... 26 Figure 2-3: Endogenous and exogenous factors influencing energy profile .................... 44 Figure 2-4: Energy profile transition through a three dimensional path ........................... 47 Figure 3-1: Schematic of a forced-draft biomass gasifier stove and an actual picture of stove X captured during fieldwork ........................................................................... 63 Figure 3-2: Schematic of gasification process and a picture of an actual mini power plant in rural Bihar ............................................................................................................. 64 Figure 3-3: Locations associated with case study ............................................................. 67 Figure 4-1: Photos of stove X and the LPG stove that was tested .................................. 100 Figure 4-2: Owners’ perception of stove attributes ........................................................ 108 Figure 4-3: Users’ perception of stove attributes ............................................................ 110 Figure 4-4: Comparison of the mean scores of stove X attributes as seen by owners and users ........................................................................................................................ 114 Figure 4-5: The means of adjusted scores given by owners regarding reasons to buy stove X. The columns in the graph show the adjusted means of scores given to each reason by the respondents. The columns are coded by categorizing the reasons to buy as: 1) reasons associated with stove convenience and functionality (orange), 2) health (yellow), 3) economic proposition (purple), 4) fuel factors (cyan), and 5) others (green). The vertical dash line indicates the highest mean score and the horizontal dash line indicates the median of scores. ................................................................ 116 xiii  Figure 4-6: Frequency with which owners mentioned each reason. The columns in the graph show the adjusted means of scores given to each reason by the respondents. The columns are coded by categorizing the reasons to buy as: 1) reasons associated with stove convenience and functionality (orange), 2) health (yellow), 3) economic proposition (purple), 4) fuel factors (cyan), and 5) others (green). The vertical dash line indicates the highest mean score and the horizontal dash line indicates the median of frequencies. ............................................................................................ 117 Figure 4-7: Reasons that owners of LPG kitchens aware of stove X did not buy stove X. The columns in the graph show the adjusted means of scores given to each reason by the respondents. The columns are coded by categorizing the reasons to buy as: 1) reasons associated with stove convenience and functionality (orange), 2) health (yellow), 3) economic proposition (purple), 4) fuel factors (cyan), 5) knowledge about stove X (patterned red), and 6) others (green). The vertical dash line indicates the highest mean score and the horizontal dash line indicates the median of scores. ................................................................................................................................ 119 Figure 4-8: Price of each unit of energy delivered to pot by LPG, stove X pellets, and wood........................................................................................................................ 121 Figure 4-9: Users’ and owners’ perception of the statement “smoke is bad for health” 122 Figure 4-10: Reasons stove owners and users do not plan to take any action to reduce smoke in the kitchen ............................................................................................... 124 Figure 5-1: Average CO concentration for three tests done on stove X and LPG stoves. Region I indicates the lighting phase, when kerosene is poured on the pellets to light the initial fire. After the fire has been propagated enough, the fan is turned on and the main combustion phase starts (Region II). Finally, after the flame is out, the smoldering phase (Region III) begins; it lasts until the red char is darkened. ........ 139 Figure 5-2: Restaurant owners’ perception of the smokiness of LPG stoves versus stove X.............................................................................................................................. 146 xiv  Figure 5-3: Restaurant stove users’ perception of the smokiness of LPG stoves vs stove X ................................................................................................................................ 146 Figure 5-4: Frequency table for when the eye burning sensation occurs ....................... 149  xv  List of Abbreviations AAC  Aerosol Carbon to CO2 ratio  ABE  Advisory Board on Energy  AFR  Air to Fuel Ratio  HIV/AIDS  Human immunodeficiency virus / acquired immunodeficiency syndrome  ALRI  Acute Lower Respiratory Infection  ARI  Acute Respiratory Infection  BoP  Bottom of Pyramid  CDM  Clean Development Mechanism  CH4  Methane  CO  Carbon Monoxide  CO2  Carbon Dioxide  COPD  Chronic Obstructive Pulmonary Disease  CSTEP  Center for Study of Science, Technology and Policy  CVD  Cardiovascular Diseases  DOI  Diffusion of Innovation  EC  Elemental Carbon  EF  Emission Factor; the mass of an emission (exhaust constituent) emitted per  EPA  Environmental Protection Agency  ESMAP  Energy Sector Management Assistance Program  FAO  Food and Agriculture Organization  FC  Fuel Carbon  GACC  Global Alliance for Clean Cookstoves  GBD  Global Burden of Disease  GHG  Greenhouse Gas xvi  H2O  Water  HAP  Household Air Pollution  IAP  Indoor Air Pollution  ICS  Improved Cookstove  IEA  International Energy Agency  IMF  International Monetary Fund  IPCC  Intergovernmental Panel on Climate Change  LED  Light Emitting Diode  LPG  Liquefied Petroleum Gas (mostly comprised of propane)  LSMS  Living Standard Measurement Surveys  MDG  Millennium Development Goal  MNES  Ministry of Non-Conventional Energy Sources  MNRE  Ministry of New and Renewable Energy  NG  Natural Gas  NGO  Non-Governmental Organization  NIPC  National Programme on Improved Cookstoves  NMHC  Non-Methane Hydrocarbons  NOx  Oxides of Nitrogen (NO plus NO2)  OAP  Outdoor Air Pollution  OC  Organic Carbon  OECD  Organization for Economic Co-operation and Development  OEH  Occupational and Environmental Health  OPEC  Organization of the Petroleum Exporting Countries  PEM  Proton Exchange Membrane  PIC  Prodcut of Incomplete Combustion  PM  Particulate Matter. PM is differentiated by aerodynamic diameter (see PM2.5 and PM10) xvii  PM10  PM with an aerodynamic diameter of less than 10 microns; (i.e. Coarse Fraction PM)  PM2.5  PM with an aerodynamic diameter of less than 2.5 microns; (i.e. Fine Fraction PM)  PTEM  Physical-Technical-Economic models  PTFE  Polytetrafluoroethylene  PV  Photovoltaic  SME  Small, Medium Enterprises  STEMS  Stove Emission Measurement System  TPB  Theory of Planned Behavior  TSP  Total Suspended Particles  UBET  Unified Bioenergy Terminology  UNDP  United Nations Development Programme  UNEP  United Nations Environmental Programme  UNFCCC  United Nations Framework Convention for Climate Change  USAID  United States Agency for International Development  VBN  Value, Believe, Norm  VLE  Village Level Entrepreneurs  WBT  Water Boiling Test  WHO  World Health Organization  xviii  Acknowledgments The research for my dissertation could not have been completed without the passionate help and collaborative work of many extraordinary people. First of all, I would like to thank my advisor, Dr. Hisham Zerriffi, whose mentorship, encouragement, and advice throughout my PhD not only guided me through my research, but also broadened my understanding of its applications. Working with Dr. Zerriffi at the Liu Institute for Global Issues transformed me from a practitioner interested in running development projects to a practitioner focused on actively pursuing how scientific approaches can be incorporated into lasting solutions. Thanks are due in particular for his constant and inspiring support and his patient and coherent answers to my endless questions. I would also like to extend thanks to Hadi Dowlatabadi, Simon Donnor, and Grieesh Shrimali, who made up my supervisory committee, for their insights, constructive critiques, and guidance throughout my research. Their advice and unique approaches to my problems greatly improved my research. Dr. Hadi Dowlatabadi not only shared his vast knowledge and experience, but also financially supported my work. His kind attention and invaluable advice were great assets to me both during my PhD work and even before I started my degree. I owe particular thanks to Dr. Martin Davy, whose encouragement and support directed me toward my PhD studies. I also would like to thank Dr. Milind Kindlikar for all his support throughout my PhD from the very beginning. This research was funded by: the University of British Columbia’s University Graduate Fellowship; the International Development Research Centre’s Doctoral Fellowship Award and a grant to Dr. Zerriffi; a number of grants, including a Hampton Fund grant through Dr. Milind Kandlikar and Dr. Zerriffi as well as a CFIS HSS grant through Dr. Zerriffi; the Pacific Century Graduate Scholarship; a Bottom Billion Travel Grant from the Liu Institute for Global Issues; and the University of British Columbia’s UBC Mobility Award. I am grateful to these institutions for their support.  xix  My fieldwork in India would not have been possible without the support of many people and institutions. Many thanks to Dr. Megha Shenoy at Resource Optimization Initiative (ROI), Bangalore, and Dr, Anshu Bharadwaj at the Center for Study of Science, Technology, and Policy (CSTEP) for hosting me and providing logistical support during my stay in Bangalore. Thanks also to Dr. Andrew Grieshop (Assistant Professor at North Carolina State University) and Dr. Julian Marshall (Assistant Professor at the University of Minnesota) for lending me the emission measurement equipment and helping me arrange for research assistants for the emission measurements. I am especially grateful for the dedication and hard work of research assistants Grishma Jain and Karthik Sethuraman, who helped us complete a large number of measurements in a short period of time. I am also grateful for the help and support we received from company A in terms of information, facilities, and equipment. Moreover, I would like to extend my thanks to the Occupational and Environmental Health Laboratory in the School of Population and Public Health at UBC. The research undertaken during my dissertation was supported by numerous sources. I offer my enduring thanks to the faculty, staff, and my fellow students at the Institute for Resources, Environment and Sustainability and the Liu Institute for Global Issues. I have to mention the great coffee machine in the sitting area in the AERL, since it hosted so many inspiring and warm conversations with colleagues. I would also like to thank all my friends for offering such a welcoming and inspiring environment to work on my research. At UBC I would especially like to thank Emily Anderson, Ther Aung, Shane Joshua Barter, Christian Beaudrie, Sara Elder, Kieran Findlater, Olivia Freeman, Julia Freeman, Tom Green, Megan Mach, Robin Naidoo, Meg O’Shea, Conor Reynolds, Maryam Rezaei, Mahbod Rohani, Gerald Singh, and Jordan Tam. My deepest gratitude is toward my family: my dad, who left me in the beginning, but whose memory has always inspired me and warmed my heart; my mom, who has been my role model all of my life and has always driven me to achieve beyond my capabilities with her support and encouragement; and my sister, whose love and encouragement have always made everything easier. Thank you all for being such a wonderful family; I couldn’t be where I am now without you. xx  Dedication  To my parents  You need to become a pen In the Sun´s hand Hafez  xxi  Chapter 1  Chapter 1 1.1  Introduction  Overview  While energy is not generally considered a basic human need, the provision of adequate, reliable, and affordable energy is a precondition for meeting these needs. The strong links between energy services, economic growth, and achieving social objectives make energy provision a cornerstone of development. Access to modern energy systems (e.g. electricity, LPG) has been found to positively impact human wellbeing by reducing the health and safety risks associated with traditional energy use (Agarwal 1986, Smith 1987, Bruce, Perez-Padilla et al. 2000, IEA 2002, IEA 2006); decreasing the time constraints on household members, particularly women and children; increasing labor productivity and income (Barnes and Floor 1996, IEA 2002, ESMAP 2003); and positively impacting social issues such as gender inequality and literacy (Cecelski 2002, Rukato 2002, ESMAP 2004) . Despite the well-recognized and significant benefits of adopting modern energy systems, this transition has been very slow in almost all developing countries, except China,1 due to a range of socio-economic and political factors(Goldemberg, Johansson et al. 2004, Bailis and Hyman 2011, GEA 2012). The results of programs and policies designed to motivate a transition from traditional to modern energy systems have not been satisfactory. More than one-third of the world’s population still does not have adequate access to modern energy (Smith 2012) . A better understanding of the process of energy system uptake, the drivers affecting it, and its implications for energy users is vital for increasing future programs’ chances of success. The overall objectives of my research were to explore the challenges of transitioning to modern energy services and to contribute to the body of research that seeks a better solution to the current energy crisis in the developing world. My research investigates the factors affecting the uptake of energy systems by energy users and the knowledge gap in this area. Further, it examines strategies and policies aimed at promoting the uptake of cleaner energy systems and  1  A detailed discussion of energy transition is presented in Chapter 2.  1  Chapter 1  the potential implications these systems have for energy users. My research looks into some understudied areas and provides grounds for further research on these topics based on in part, on an extensive field work conducted in India. This introductory chapter provides the necessary background information on which this dissertation is built and justified. 1.2  Human welfare and energy system  Modern energy services are considered a prerequisite for development (IEA 2010). It is well recognized that access to reliable and efficient lighting, heating, cooking methods, mechanical power, transportation, and telecommunication services is crucial to improving human livelihood as these technologies provide clean water, sanitation, healthcare, and other general benefits. Eradicating poverty through achieving the eight Millennium Development Goals (MDGs), set in 2000, requires access to modern energy systems that can facilitate the delivery of povertyreducing services (Modi, McDade et al. 2005). Furthermore, as has been realized ever since the MDGs were published by the United Nations Development Programme UNDP, many of the goals cannot be reached if energy services are not provided (ESMAP 2002). The adverse implications of relying on the environment and traditional energy systems for socio-economic development can be clearly seen by the example of relying on traditional biofuels for cooking. Access to electricity is considered an indicator of rural development and can significantly improve human welfare by creating an enabling environment that allows both a range of economic benefits and wider developmental benefits (World Bank 1993, Goldemberg and Johansson 1995, IEA 2002, Cabraal, Barnes et al. 2005, Modi, McDade et al. 2005, UNDP 2005, IEA 2010, Zerriffi 2010, IEA 2012). At the household level, electricity is mainly used for lighting, IT and communication services, and powering household appliances such as fans and fridges. A range of social benefits are associated with access to electricity. The majority of these benefits are derived from the fact that lighting provides users with longer days. Among other things, access to electricity: provides access to information, improves communication, improves healthcare facilities, facilitates the completion of household chores, facilitates education, increases comfort levels, and improves the services provided by community buildings (Niez 2010). In addition to these social benefits, access to electricity brings a range of economic benefits by increasing productivity and facilitating the growth of businesses. Electricity is 2  Chapter 1  considered an important input for rural businesses as well as for farms and other rural structures (Niez 2010). However, it should be noted that these benefits cannot be provided solely by electrification and often require the presence of other enabling conditions (Barnes and Floor 1996, Elias and Victor 2005, Niez 2010) In developing countries, particularly in poor communities that are already struggling with the numerous negative impacts of poverty and other socio-economic problems, many people use traditional biofuels as their main cooking fuel. Users of traditional biofuels are exposed to a range of adverse socio-economic effects associated with both the consumption and supply of these fuels. Exposure to Indoor Air Pollution (IAP) caused by the burning of traditional biofuels has numerous direct and indirect impacts on population health, accounting for more than 4 million deaths worldwide (Smith 2012). These recently released numbers are a significant jump from the prior best estimate of 1.5 million deaths annually (IEA 2010), which was widely cited and used to justify stove interventions. Even the prior (lower) estimate indicates that the number of premature deaths associated with indoor air pollution (IAP) surpasses the estimated number of premature deaths currently connected to malaria and tuberculosis and the amount that will be linked to HIV/AIDS by 2030. (IEA 2010). Numerous studies have pointed out the significant health risks related to the use of traditional biofuels, particularly for women and children (Agarwal 1986, Smith 1987, Ellegård 1993, Mishra, Retherford et al. 1999, Ezzati, Bailis et al. 2004). However, despite significant improvements in our understanding of IAP resulted from using the traditional cooking system and its health implications, further study is needed if we are to have an appropriate understanding of both energy device (i.e. stove) performance and emission characteristics (Berkely 2012) and IAP exposure and its health implications (Ludwinski, Moriarty et al. 2011, Lim, Vos et al. 2012). Moreover, most studies so far have solely focused on the residential sector, meaning that the potential adverse health implications of using traditional energy systems in occupational environments (e.g. restaurants) have not received attention. While the use of traditional biofuels is associated with a range of adverse socio-economic impacts, lack of access to a sufficient supply of biofuels not only prevents people from benefiting 3  Chapter 1  from energy services, but also intensifies the negative impacts of using these fuels. Impoverished communities dedicate a high percentage of their income to acquiring biomass cooking fuel (either directly or in the form of the very high opportunity cost they pay in terms of collection and preparation) (Haines and Kammen 2000). Researchers have identified various implications of traditional biofuel shortages, including: reduction in agricultural output from the reallocation of labor (Cooke 1998, Amacher, Ersado et al. 2004); reduction in agricultural output from increased use of farmland to produce fuel wood (Cooke, Kohlin et al. 2008); reduction in agricultural output from the use of dung and/or crop residues as fuel rather than fertilizer (Mekonnen 2000, Wardle, Jansky et al. 2003, Van’t Veld, Narain et al. 2006); negative impacts on health and/or nutrition (Brouwer, Hoorweg et al. 1997, Wardle, Jansky et al. 2003); increases in the labor burden of women and children (Agarwal 1986, Cooke 1998, Rukato 2002, Amacher, Ersado et al. 2004); and increases in environmental damage (Kohlin and Parks 2001, Foster and Rosenzweig 2003). These impacts often have larger negative welfare implications for poorer households than they do for richer ones (Heltberg, Arndt et al. 2000, Linde-Rahr 2003). Whether through ensuring the supply of biomass material or reducing demand for it by improving the efficiency of cooking systems, it is of the utmost important to ensure that those relying on this type of fuel, particularly subsistence users, have sufficient access to it while reducing the adverse social implications of traditional energy system usage. Moreover, there has been a growing concerns over the global warming implications of using traditional cooking energy systems (Arnold, Kohlin et al. 2003 , Smith, Mehta et al. 2004, Bailis, Cowan et al. 2009) (Zhang, Smith et al. 2000, Ezzati, Bailis et al. 2004, Smith and Haigler 2008). Many researchers have pointed out the effects of climate-forcing emissions and reduced carbon uptake due to unsustainable wood fuel harvesting, both of which are caused by the use of traditional energy systems (Smith, Khalil et al. 1993, Smith, Uma et al. 2000, Zhang, Smith et al. 2000, Grieshop, Marshall et al. 2011); recently, the strong climate-forcing impact of black carbon (BC), which is mainly emitted from the combustion of traditional biofuels, has called much attention to the significant impact that traditional energy systems have on global warming (Bond 2007, Aunan, Berntsen et al. 2009).  4  Chapter 1  The wide range of adverse social and environmental impacts linked to using traditional energy systems – coupled with the significant benefits associated with access to modern energy services – have encouraged a large number of governments and agencies to implement programs that encourage the replacement of traditional energy systems with modern energy systems that are cleaner and more efficient. However, the situation in the developing world is not promising and a significant number of people still rely entirely on traditional energy systems. 1.3  The current energy situation in the developing world and what is expected  Based on IEA estimates, household energy demand in developing countries was about 1,090 Mtoe1 in 2004, equal to approximately 10% of the world’s total primary energy demand. In households, energy is mainly used for cooking, followed by heating and lighting (IEA 2006). Although motivating the transition from traditional to modern energies has been on government and development agency agendas for decades, this transition has been very slow in almost all developing countries, except China,2 due to a range of socio-economic and political factors (Goldemberg, Johansson et al. 2004, Bailis and Hyman 2011). Almost 40% of the world’s population still relies entirely on traditional biofuels3 for cooking fuel (2.6 billion people still suffer from lack of access to clean cooking fuel facilities4) (IEA 2012). The latest projections from the International Energy Agency (IEA) are based on the “New Policies Scenario” (IEA 2012), which shows that on average 635 million USD per year will be invested in the provision of clean cooking fuel and that the number of people without access to  1  Mtoe: million tons of oil equivalent is a unit of energy. One ton of oil equivalent is the amount of energy released by burning one ton of crude oil, which is approximately 42 GJ (gigajoules). 2 A detailed discussion of energy transition is presented in Chapter 2. 3 Biofuel refers to a variety of modern or traditional fuels that are derived from biomass material harvested from different biomass production sites. This paper focuses on a subdivision of biofuels called traditional biofuels. Traditional biofuels are commonly referred to as fuel wood, charcoal, agricultural residues, and animal dung. Many other labels – such as biomass fuels, non-commercial energy, traditional fuels, rural energy, and combustible renewables and waste – are used by different researchers and organizations. None of these labels, however, ideally reflects these fuels (Victor et al. 2002). This paper uses the term “traditional biofuel,” which is defined by the FAO as a part of Unified Bioenergy Terminology (UBET). Different forms of traditional biofuels are also named according to UBET (FAO 2004). 4 The World Energy Outlook (IEA 2012) defines having cooking energy access as possessing reliable and affordable access to clean cooking facilities.  5  Chapter 1  clean fuel facilities will slightly increase, even as the percentage of people who lack access to these facilities drops. See Table 1-1 (IEA 2012). Table 1-1: Population lack access to clean cooking facilities now and in 2030 (millions) (IEA 2012) 2010 2030 Rural  Urban  Total  Share of population  Rural  Urban  Total  Share of population  2,155  433  2,588  49%  2,139  456  2,595  39%  518  180  698  68%  629  257  886  56%  516  179  696  81%  627  256  883  65%  1,580  234  1,814  51%  1,458  182  1,640  39%  China  345  42  387  29%  220  20  240  17%  India  698  75  772  66%  680  55  735  50%  Rest of developing Asia  538  117  655  61%  558  106  664  50%  Latin America  47  18  65  14%  45  18  62  11%  Middle East  9  1  10  5%  8  0  8  3%  2,155  433  2,588  38%  2,139  456  2,595  31%  Developing countries Africa Sub-Saharan Africa Developing Asia  World  Approximately 1.3 billion people have no access1 to electricity (IEA 2012), 85% of whom reside in rural areas of the developing world (IEA 2010). Another 1 billion have only irregular and unreliable access to electricity (GEA 2012). Sub-Saharan Africa and South Asia have the highest number of people without access to electricity; electrification is lowest in countries such as Burkina Faso, Mozambique, the Democratic Republic of Congo, and Afghanistan, where less than 7% of the population has electricity. While there was a large drop in the number of people without access to electricity globally from 1990 to 2005 (from 2 billion in 1990 to 1.6 billion in 2005), this fall was mainly caused by swift development in China. Table 1-2 presents statistics on electricity access. The projections in this table are based on the “New Policies Scenario,” through which on average 14 billion USD per year will be invested in  1  For rural households, the International Energy Agency’s World Energy Outlook (IEA 2012) defines electricity access as having an electricity supply connection with a minimum consumption level of 250 kilowatt-hours [kWh] per year; for urban households the minimum consumption level is 500 kWh per year.  6  Chapter 1  electrification; the estimates presented show that the number of people in the world who lack access to electricity will decrease by 2030. Improvement in electricity access is expected to be significant in developing Asia, where lack of access to electricity is expected to be halved in both absolute numbers and in proportion to the population. In some regions, including Latin America, the Middle East, and China, the already positive trend of increasing electricity access is expected to reach universal access by 2030. It should be noted that while India will not reach universal access, progress there is expected to be significant; see Table 1-2 (IEA 2012). Table 1-2: Population currently lacking access to electricity and population expected to be without electricity access under the “New Policies Scenario” in 2030 (in millions) (IEA 2012) 2010 2030 Rural Urban  Total  Share of population  Rural  Urban  Total  Share of population  1,081  184  1,265  24%  879  112  991  15%  475  114  590  57%  572  83  655  42%  474  114  589  68%  572  83  655  48%  566  62  628  18%  305  29  334  8%  China  4  0  4  0%  0  0  0  0%  India  271  21  293  25%  144  8  153  10%  Rest of developing Asia  291  40  331  31%  161  20  181  14%  Latin America  23  6  29  6%  0  0  0  0%  Middle East  16  2  18  9%  0  0  0  0%  1,083  184  1,267  19%  879  112  991  12%  Developing countries Africa Sub-Saharan Africa Developing Asia  World  As shown in Table 1-1 and Table 1-2, by 2030 there will still be 2.6 billion people without access to clean cooking fuel and 1 billion people without access to electricity, despite the fact that for the overall population the share of people without access to modern energy systems (i.e. electricity, clean cooking fuel) will drop by about 7%. By 2030, the amount of the population in developing Asia without access to electricity will be halved compared to 2010, while universal access will be achieved in Latin America, China, and the Middle East; however, in Sub-Saharan Africa the situation will worsen, at least until about 2025 (IEA 2012). In terms of providing clean cooking energy, the situation is less positive. Although in recent decades there has been a 7  Chapter 1  significant improvement in developing Asia, the amount of people without access to clean cooking energy in India will still be about 735 million by 2030. While that represents a 16% drop in relative terms, it means roughly the same number of people will lack access in 2030 as in 2010 (as a result of the world’s growing population). The situation is also worsening in the case of Sub-Saharan Africa (IEA 2012). Moreover, the trend of energy transition does not always favor adopting modern energy systems. There is evidence of a “switching back” phenomenon whereby users at least partly replace modern energy systems with traditional ones (IEA 2006, Lallement 2008, Elgarah 2011), mainly due to increases in the cost of modern energy devices (Barnes, Openshaw et al. 1994, Masera, Saatkamp et al. 2000, Pachauri, Spreng et al. 2012). While fuel stacking (using both traditional and modern energy sources) is widely studied, it is generally assumed that when it occurs, modern energy sources are added onto traditional ones, either as substitutes or complements. However, the usage patterns of traditional and modern systems and the degree to which those with modern energy systems add on traditional systems rather than vice versa is less well understood. This trend can intensify when the price of modern energy carriers and appliances increases; it can also hinder attempts to transition toward modern energy services. As yet, this tendency has not received much attention from research communities and requires more exploration. Without a significant improvement in policies and programs aimed at providing modern energy services in developing countries, this pressing challenge cannot be addressed and a significant number of people will continue to suffer from its various adverse consequences. 1.4  Providing access to modern energy services  Energy provision is widely considered the cornerstone of development, and the energy and development communities consider universal access to clean cooking fuel and electricity primary goals (IEA 2010). Using a wide range of institutional and technological approaches, various actors have made substantial efforts to address these two primary energy access problems.  8  Chapter 1  1.4.1  Electrification  Electrification has traditionally been accomplished by grid extension through centralized power generation, transmission, and distribution systems (Zerriffi 2010). However, due to huge capital requirements, particularly in remote regions, centralized electrification programs are not likely to be able to achieve the target of universal access to electricity (Zerriffi 2010). Distributed generation is seen as an appropriate alternative to centralized systems where grid extension is not economically feasible (Chaurey, Ranganathan et al. 2004, Cust, Singh et al. 2007, Reddy, Balachandra et al. 2009, Zerriffi 2010), as is indicated in the most recent IEA report (IEA 2012). A range of technological options have been tried out for distributed generation (Banerjee 2006), with biomass-based electricity generation technologies being widely promoted, particularly for rural areas (Kartha, Leach et al. 2005). Electricity generation based on biomass gasification that uses agricultural residue as its feedstock is considered one of the most promising bioenergy options (V. Siemons 2001, Dasappa 2011, Mahapatra and Dasappa 2012), especially in rural areas with a high availability of agricultural residue (Romijn, Raven et al. 2010). For a long time, electricity provision was traditionally the domain of the state, partly due to the large capital requirements of centralized power delivery and the strategic value of energy provision in government policies (Eberhard 2004). In the past two decades there has been a growing trend toward furthering the involvement of private enterprises and minimizing the role of the state in operating and regulating the power sector (Bacon and Besant-Jones 2001). These trends largely continue the expansion of centralized systems with new actors (private firms) and new roles (e.g independent regulators). However, the implications of this trend, particularly in the case of electrification programs in the developing world, are still largely unknown, especially when considering the major role decentralized systems will play in increasing access. 1.4.2  Clean cooking energy system  The provision of clean cooking energy systems is believed to have the potential to significantly improve the living condition of almost 2.6 billion people (IEA 2012). Although substantive efforts have been made to provide clean cooking fuel and appliances, estimates indicate that a massive transition to non-traditional cooking systems is not likely in the next few decades (Foell, 9  Chapter 1  Pachauri et al. 2011). Improved biomass-burning cook stoves that can burn cleaner and more efficiently are considered a promising intermediary alternative (Foell, Pachauri et al. 2011) that can significantly reduce the adverse impacts of traditional cooking systems (Smith, Uma et al. 2000, Zhang, Smith et al. 2000, Arnold, Kohlin et al. 2003, Smith, Mehta et al. 2004, MacCarty, Ogle et al. 2008 {Smith, 2008 #560, Aunan, Berntsen et al. 2009, Bailis, Cowan et al. 2009, World Bank 2011)}. Despite recognition of the potential positive impacts of the adoption of improved cookstoves1 (ICSs) and decades of efforts to disseminate ICSs, only 30% of people relying on biomass as their main cooking fuel are using these stoves (Bailis and Hyman 2011). Programs have often had limited success and only few have achieved a substantial dissemination of ICSs (Bailis and Hyman 2011). Similar to electrification, there is an increasing interest in the role that the private sector can play in disseminating ICSs, and huge market potential for stove developers has been perceived in this field (Hoffman, West et al. 2005). However, the outcome of programs motivated by market-oriented approaches has yet to be evaluated. 1.5  Understanding uptake and usage of energy systems  The limited success of programs attempting to achieve universal access to modern energy services is also partly the result of ineffective design due to our limited knowledge of energy transition processes. Our understanding of non-industrial energy use patterns and the variables associated with them is very limited, particularly in the context of the developing world (Leach 1992, Masera, Saatkamp et al. 2000, ESMAP 2003, Heltberg 2004, Elias and Victor 2005, Farsi, Filippini et al. 2007, Pachauri 2007). Researchers from different disciplines have adopted various models to analyze energy use as well as the process of energy system uptake. Examples include Physical-Technical-Economic  1  In this thesis, the term “improved cook stove” refers to a range of non-traditional biomass-burning stoves that often vary significantly in their performance and characteristics. Depending on their technology, they do not necessarily bring improvement to health, efficiency, and global warming. Some researchers have recently started using the term “advanced improved cook stove” to identify stoves that are significantly cleaner and more efficient (Mobarak 2012, Venkataraman 2010).  10  Chapter 1  models (PTEM) that focus on the physical and technological dimensions of energy use as well as the effects of economic factors on energy use and energy system choice (Hosier and Dowd 1987, Leach 1987, Tiwari 2000, Saboohi 2001, Hart and de Dear 2004, Stokes, Rylatt et al. 2004, Heltberg 2005, Farsi, Filippini et al. 2007, Laitner 2007, Gundimeda and Köhlin 2008); psychology-based models that try to understand the behavior and variables that influence individual decisions (Ajzen 1991, Wilhite, Shove et al. 2000, Rogers 2003); and sociological and anthropological models that investigate the social determinants of individuals’ decisions (Lutzenhiser 1993, Guagnano, Stern et al. 1995, Shove 2003). However, so far energy provision policies have mainly been informed by PTEMs (Cherfas 1991). Integrated approaches, although still in their infancy, incorporate various dimensions of energy use and energy system uptake to create a more realistic view; they have been promoted by researchers (Wilson and Dowlatabadi 2007). Many researchers believe that incorporating the human dimension into energy analysis is a necessary step toward improving our understanding of the dynamics of energy system uptake and use (Ruiz-Mercado, Masera et al. 2011, Takama, Lambe et al. 2011, Johnson and Takama 2012, Mobarak, Dwivedi et al. 2012). Furthermore, almost all studies on clean cooking energy system adoption have focused on the residential sector, meaning that commercial entities have been left understudied. Restaurants and food vendors in particular are major cooking device users and the energy systems they use not only have environmental implications, but also may significantly impact workers. Street food vendors, despite their significant presence, often do not belong to a formal sector and thus receive hardly any attention from the development and research community (Tedd, Liyanarachchi et al. 2001). Gaining an understanding of commercials kitchens’ energy use patterns and the process through which they adopt energy systems is not only necessary for gaining knowledge of that sector’s energy use, but also for shedding more light on the energy use of small industries (i.e. households and artisanal industry), which compromise the lion’s share of industrial activities and are a major source of employment and income in many developing countries (Anderson 1982, Mead and Liedholm 1998).  11  Chapter 1  1.6  Thesis structure and overview  I have achieved my main research objective, which was “to explore the challenges of transitioning toward modern energy services and to contribute to the body of research that seeks a better solution to the current energy crisis in developing world,” through closely investigating a case in India, looking at 1) the process through which households take up energy systems (Chapter 2), 2) the shortcomings and potential advantages of the market-based approach, examined through a case study (Chapter 3), 3) the dynamics of ICS uptake by commercial kitchens (Chapter 4), and 4) the potential health implications of switching back in commercial kitchens (Chapter 5). This thesis is written as a paper-based dissertation and includes one introductory chapter, four research chapters, and a concluding chapter. All of the manuscripts are prepared for publication in scientific journals. Chapter 2 provides a comprehensive review of the literature on energy analysis and the factors affecting the selection of energy systems. I investigate the evolution of energy analysis, focusing particularly on energy research in the context of the developing world. The current state of knowledge on energy transition and the drivers of energy system adoption are both investigated in detail. Particular attention is paid to research on energy use in rural regions of the developing world. Further, by identifying major gaps in the literature and building on current knowledge of household energy use in rural regions, a new conceptual framework is developed. This framework acts as a basis for building new theoretical and empirical models of rural household energy. In Chapter 3, I investigate market-based interventions to electrification and the provision of clean cooking energy systems (i.e. ICSs) by examining two enterprises active in rural energy provision in India. I explore the evolution of programs that address the provision of energy to rural households, particularly focusing on the positive and negative aspects of the market-based approach. Through a comprehensive literature review, I examine the potentials and drawback of the market-based approach to energy provision in the developing world. Building upon this understanding, two hypotheses are put forward and examined through a case study. A semi12  Chapter 1  structured interview questionnaire was designed and was used as the basis for my interviews with people working for enterprises actively engaged in providing energy in India. These semistructured interviews allowed a range of issues to be raised by people at different levels of organization in these enterprises. In Chapter 3, I identify enterprises’ reasons for shifting their focus from the residential market to the commercial market; in Chapter 4, I investigate this topic from the perspective of consumers. The focus is on one of the enterprises studied earlier; it provides ICSs (stove X) and fuel made from agricultural residue. This company, which initially distributed residential ICSs in rural regions of India, is now focused on providing the larger version of its stove to commercial kitchens in major urban regions, such as Bangalore, Hyderabad, Chennai, and Pune. Building upon our understanding of gaps in the current literature on the uptake of energy systems – discussed in Chapter 2 – I investigate the attitude of energy system owners and users with regard to the two types of energy systems they are using (LPG stoves and biomass ICSs) and look at their reasons for switching from modern energy systems to those that are less modern. I pay particular attention to health and fuel savings, which are the two main objectives of many current ICS dissemination programs; I also look at how these two factors are seen from the perspective of energy system users and owners and how these factors impact the adoption of particular energy systems. The data was collected through a survey of commercial kitchens in Bangalore, India. The survey involved both kitchen owners and the main stove users of 200 (100 LPG-using kitchens and 100 LPG- and stove X-using kitchens) commercial kitchens in Bangalore, India. This paper makes a brief comparison of commercial entities’ and residential consumers’ perceptions of energy system attributes based on a similar survey of residential consumers conducted by Stanford University. In Chapter 5, I investigate the potential adverse impacts associated with switching back from LPG to biomass ICSs in commercial kitchens, focusing on the potential health implications of IAP for people working in kitchens. I look at the change in IAP due to the switch from LPG to stove X through a mixed method design that combines different assessment methods: 1) by investigating secondary data on the emission characteristics of stove X and LPG stoves, 2) by measuring the direct emissions of stove X and a commercial-grade LPG stove (i.e. CO, CO2, and 13  Chapter 1  PM) in the field using emission measurement equipment, 3) by investigating the perceptions of stove users and kitchen owners with regard to the smokiness of LPG stoves and stove X using the data collected in the survey of commercial kitchens in Bangalore, and 4) by investigating the correlation between self-reported health symptoms (i.e. coughing and eye irritation) and using LPG stoves and stove X. The results of this study improve our understanding of the complex process of energy transition in developing countries. The study indicates the shortcomings of the market-based approach and the necessity of government involvement in energy provision programs. Further, it sheds light on the less-studied switching back process through which energy users using modern energy systems adopt less efficient and more dirty energy systems. It is one of the few studies that looks at the energy use of small commercial entities in developing countries. Our findings indicate that policies that promote energy technology in a particular sector may have undesirable spillover effects in that sector. Through the concluding chapter (Chapter 6), the overall conclusion, and the summary of findings and their policy implications, the strengths and limitations of this work are presented and a path for future work is indicated.  14  Chapter 2  Chapter 2 Three dimensional energy profile: a conceptual framework for assessing household energy use 2.1  Introduction  While energy has not been generally considered as a basic human need, the provision of adequate, reliable, and affordable energy is a precondition for meeting these needs. The strong links between energy services, economic growth and achieving social objectives make energy service provision a cornerstone of development. Having access to modern energy systems (e.g. electricity, LPG) is found to have major impacts on human wellbeing by reducing health and safety risks associated with traditional energy use (Agarwal 1986, Smith 1987, Bruce, PerezPadilla et al. 2000, IEA 2002, IEA 2006); decreasing the time budget constraints on household members, particularly women and children, as well as increasing labor productivity and income (Barnes and Floor 1996, IEA 2002, ESMAP 2003), and positively impacting social issues such as gender inequality and literacy (Cecelski 2002, Rukato 2002, ESMAP 2004). More than one third of the world’s population has a very limited access to modern energy services. Almost two and a half billion people still rely entirely on traditional biofuels as their cooking fuel and suffer from its various negative socioeconomic impacts such as adverse health impact, accounting for more than 1.6 million deaths annually. Projections show both the number of biofuel dependents and energy demand will increase, even as the share of traditional biofuel dependents drops (IEA 2006). Approximately 1.6 billion people, or one quarter of the world population, have no access to electricity, 80% of whom reside in rural areas of the developing world. Sub-Saharan Africa and South Asia have the highest number of people without access to electricity. Despite the apparent large drop in the number of people without access to electricity globally (2 billion in 1990 to 1.6 billion in 2005), this fall is mainly caused by swift development in China. For the rest of the developing world, the number of people without access to electricity has steadily grown in the same period, and without any new policy implementation, this number will be around 1.4 billion people by 2030 (Saghir 2005, IEA 2006). 15  Chapter 2  Household energy consumption in developing countries counted for approximately 10% of total world primary energy demand which was about 1,090 Mtoe1 in 2004. The main use of energy in this sector is for cooking, followed by heating and lighting (IEA 2006). Household energy use has unique characteristics that make it harder to assess and analyze compared to other sectors. Some of these characteristics are: Diversity of energy system in both household energy use and energy supply; Lack of sufficient data due to lack of recorded transactions in household energy; Dominance of traditional biofuels sector, particularly in rural areas, that are part of a complex and interrelated household production system that produces food, fodder, construction material, income, and fuel (Leach 1987). Researchers have been exploring various dimensions of household energy use in order to design and implement strategies not only to provide secure access to energy services, but also to facilitate the transition to modern fuels, eradicating energy poverty, addressing environmental concerns and mitigating greenhouse gas emission. However, despite more than three decades of effort, our understanding of household energy use patterns and the variables associated with household energy use is very limited, particularly in the developing world (Leach 1992, Masera, Saatkamp et al. 2000, ESMAP 2003, Heltberg 2004, Elias and Victor 2005, Farsi, Filippini et al. 2007, Pachauri 2007). Addressing such issues requires better knowledge of how people make decisions related to their energy use. A lack of information on this decision making process is the most challenging problem facing energy analysts, particularly when trying to predict response to major perturbations in energy systems and considering new alternative interventions that are very dissimilar to the past experiences (Stern 1986). This chapter builds upon the current knowledge of household energy use to develop a new conceptual framework to guide the analysis of household energy choices with a particular focus on the household energy transition process. The past and current trends in the field of energy analysis are investigated. Due to the importance of improving energy systems in the developing world, particular attention has been paid to the energy transition process, which is also reflected  1  Mtoe: Million tons of Oil equivalent is a unit of energy. One tones of oil equivalent is the amount of energy released by burning one tone of crude oil which is approximately 42 GJ (Giga Joule)  16  Chapter 2  in the proposed framework. It should be noted that the objective of this paper is not to provide a detailed household energy analysis; rather it reviews current knowledge on energy analysis, the identified factors affecting the energy requirement and energy use patterns among households with particular focus on habitants of rural regions of developing world. Organization Section 2 provides a brief overview of the literature on household energy use and a review of different types of energy analysis used to understand household energy. Section 3 focuses on the process of energy transition in developing world, with particular attention paid to analyses of rural household energy use. The factors affecting household energy use are categorized and listed. In the following section the shortcomings in the literature are identified and a new conceptual framework is presented. 2.2  Energy analysis and underlying assumptions  Following the 1970s’ oil crisis, energy scholars have devoted substantial effort to understanding household energy usage. In the developed world much of these efforts have focused on reducing oil consumption, ways to increase energy efficiency, and investigating the impacts of oil price fluctuation on the economy. Energy research in the developing world focused on the perception of a fuelwood crisis. Increasingly research in the developing world is conducted with a view to understanding the impacts of energy use on the environment, the welfare implications of energy, and security of the energy supply (ESMAP 2003). In academia, four disciplines including engineering, economics, psychology, and sociology or anthropology, have been the main contributors to the field of household energy use. Each of these disciplines approach this issue in its own unique way based on its biases, frameworks, techniques, etc (Keirstead 2006).. 2.2.1  Physical-Technical-Economic models (PTEM)  Energy analysis has focused largely on the physical and technological dimensions of energy use as well as the effects of fluctuations in energy prices; human dimension of energy use has been largely overlooked (Laitner 2007). Schipper professes that “Those of us who call ourselves 17  Chapter 2  energy analysts have made a mistake ... we have analyzed energy. We should have analyzed human behavior”(Cherfas 1991). As a result, the field of energy analysis has been dominated by physical-technical-economic models (PTEM) of consumption. In such models, the change in consumer demand and energy use pattern is determined by changes in technologies, which are mainly driven by the cost of energy relative to consumer income (Lutzenhiser 1993). Technical models make detailed estimations of energy flows through physical systems and calculate the energy requirements based on physical laws. The studies on overall household energy (Leach 1987), lighting (Stokes, Rylatt et al. 2004), and appliances (Hart and de Dear 2004) are examples of technology based studies. However, their estimates do not match well with real world measurements because they fail to recognize humans as “active” energy users who manipulate energy devices, interacting with energy devices and energy flows at all stages of energy systems (Wilhite, Shove et al. 2000). Economic models try to go further in exploring human decisions regarding energy usage. These models often try to understand the implications of energy price, taxes, income and expenditure on household energy use (Hosier and Dowd 1987, Tiwari 2000, Saboohi 2001, Heltberg 2005, Farsi, Filippini et al. 2007, Gundimeda and Köhlin 2008).Microeconomic decision models of utility1 maximization have been used widely in economic approaches. They are based on the economic theory of rational choice, in which individuals are considered as rational actors who seek constantly to maximize their utility under certain budget constraints, have a set of ordered, well-defined, invariant and consistent preferences, and who choose the alternative with the highest utility among the consumption choices. Discrete choice modeling and economic engineering analysis are the two major applications of this theory that are relevant to energy analysis (Wilson and Dowlatabadi 2007). Yet a large body of empirical evidence shows that people do not respond rationally to economic and technical opportunities (Cogoy 1995, Kooreman 1996, Fernandez 2001, Frederick, Loewenstein et al. 2002). While PTEMs generally assume that energy usage is relatively homogenous and shifts in energy consumption are mainly  1  Utility is a measure that represents the satisfaction people derive from consuming goods and services (Frank, 2001)Frank, Robert and Ben Bernanke (2001). Principles of Microeconomics, McGraw-Hill.  18  Chapter 2  driven by changes in the price of fuels, empirical evidence shows variation and heterogeneity in energy consumption, and the influence of non-economic motives as well as social context on consumption. For instance, studies indicate that personal commitment is a key determinant of household decision on adopting energy saving measures (Lutzenhiser 1993). As the result of reliance on these assumptions, policy analyses have tended to exaggerate the importance of energy price and technological solutions while undermining the importance of non-economic factors (Stern 1986). As Stern described, heavy reliance on economic theories has not only misinformed many policies but also mislead analysts by focusing attention on economic concepts and away from many critical and important social and psychological concepts (Stern 1986). 2.2.2  Psychology based approaches  The most routine forms of energy use involve complex behavioral, cognitive, and social processes; yet, the non-technical and non-economic aspects of energy use in households are poorly understood (Stern 1986, Lutzenhiser 1993, Wilhite, Shove et al. 2000, Keirstead 2006). In response to such shortcomings, energy research has started to focus more on social science. Incorporating social variables into economic and technical models has been found to significantly improve the accuracy of these models in estimating household energy demand. Social psychology is the first discipline that tried to address human behavior in an energy context by focusing on individual motivation and information. Technology adoption theories are the result of sociologists’ and psychologists’ attempts to understand why people adopt or not adopt new and more efficient conversion technologies (Wilhite, Shove et al. 2000). The dominant model, Diffusion of Innovation (DOI) has a broad empirical basis and describes a social communication process through person to person and media channels that influence individuals’ decisions in adoption of new technologies. The underlying assumption in this theory is that the progression of knowledge, awareness, intention, and behavior that results in the adaptation of technologies is linear (Rogers 2003). DOI models are not able to properly address the cases where “adoption of new technologies is constrained by situational factors such as lack of resources and lack of access to these technologies” (Wilson and Dowlatabadi 2007); as is the case where there is no access to a particular energy carrier. Similarly, other behavioral models try to describe how knowledge will result in action; in the theory of cognitive dissonance, 19  Chapter 2  individuals strive for consistency between their knowledge, attitudes, and actions; in the theory of planned behavior (TPB), attitudes are formed from an individual’s believe about a behavior as well as the evaluation of the outcomes of that behavior (Ajzen 1991). 2.2.3  Sociological and anthropological models  People do not simply change behavior or adopt new technology based on awareness and attitudes (Wilson and Dowlatabadi 2007). Sociologists and anthropologists believe that human behavior is social and collective, thus energy models that intend to include behavioral dimensions should consider the social context of individual actions (Lutzenhiser 1993). They believe individuals decisions are determined by social and technological systems and any transformation in energy use is caused by broader social transformation; for example the work of Shove on changing our perception of comfort and cleanliness (Shove 2003). Therefore, focusing on individual behavior as the only predictor of energy use is oversimplistic and counterproductive. Household energy demand is not the product of individual decisions but rather a product of social demand (Guagnano, Stern et al. 1995). Up until now, most theories and models do not account for the relationships between individual behavior and socio-technical arrangements (Wilhite, Shove et al. 2000). 2.2.4  Integrated approaches  Although disciplinary approaches made a significant contribution to the field of household energy analysis, their narrow area of focus limits their capability to properly assess energy use. Inability of energy conservation programs to meet their anticipated energy saving targets was the first sign of such limitations (Keirstead 2006). Researchers have been looking for integrated approaches to household energy use analysis in that are more realistic and more comprehensive than isolated and disciplinary studies. Such an approach needs to simultaneously address the social and behavioral determinants of energy use as well as economic and technological aspects of energy use. It should consider individuals and institutions and their complex relationships as well as adequately account for social networks which contribute to change (Stern 1986, Wilhite, Shove et al. 2000). Many researchers have tackled the integrated approach to decision making; examples of such models are the behavioral model (Van Raaij and Verhallen 1983), the 20  Chapter 2  multigenic model (Wilk 2002), and the model of environmentally significant individual behavior (Stern 2000) . The integrated model proposed by Stern incorporates both personal variables (including attitudinal and habit and routine) and contextual variables (including external conditions and personal capabilities) that influence environmentally significant behavior (Stern 2000, Wilson and Dowlatabadi 2007). However, there has been limited progress in developing such models(Wilson and Dowlatabadi 2007). Despite the almost equal number of interdisciplinary and disciplinary studies in the past two decades, there has been a clear growth in the recent years in disciplinary studies (Keirstead 2006). A number of underlying factors hindering the development of integrated approaches have been identified by researchers including: 1. The institutional barriers in integration of various disciplines - while most energy research is considered “too applied” by social scientists, among technically trained energy analysts, social science research is often perceived as “too theoretical” (Lutzenhiser 1993); Achieving integration between these disciplines requires “greater openness on the part of the dominant economic-engineering tradition and a more applied focus on the part of behavioral scientists” (Wilson and Dowlatabadi 2007). 2. Poor communication; the integrated models often communicate in qualitative terms while a successful approach should be able to communicate quickly and transparently to policy options as well as to inform the other models (Keirstead 2006). 3. Limited scope and scale: the integrated approaches have been mainly concerned about small scale issues that limits them in providing insight at a larger scale (e.g. at the sectoral level); A proper approach requires being inclusive to be able to address broad scale issues whilst flexible to be able to address the specific issue at the small scale (Keirstead 2006). 4. Lack of objective behavioral data: There is significant lack of qualitative studies on household energy use in developing countries. This will be discussed in further details in section 4 under insufficient data. Insufficient understanding of household behavior with  21  Chapter 2  regard to energy use hinders researchers attempting to test such integrated models This limits the applicability and the development of such models. Despite evidence that purely economic and technical models perform poorly, they are still the dominant models in energy analysis and the policy makers are mainly influenced by the estimations of such models (Wilhite, Shove et al. 2000, Lutzenhiser, Cesafsky et al. 2009, Webler and Tuler 2010). Further improvement in integrated approaches is vital to make them more credible and make them a viable alternative to the existing dominant PTEMs based decision making (Keirstead 2006). 2.2.5  Studies on rural household energy use in developing world  Most of the approaches discussed so far are drawn from research conducted in the developed countries. Energy analysis, particularly at a disaggregated level (e.g. household level), that incorporates a behavioral dimension has been conducted almost exclusively in the developed world. In the developing world, on the other hand, most analyses to date have been conducted at an aggregate level, and a very limited number of these studies have focused on household energy use. The number of behavioral studies regarding energy usage in rural households is even more limited1. Even among the limited number of studies focused on the developing world, a large part of the literature comes from studies in urban areas; see (Fitzgerald, Barnes et al. 1990, Barnes and Qian 1992, ESMAP 1999, Tiwari 2000, Dube 2003, Chambwera 2004, Cabraal, Barnes et al. 2005, Gupta and Kohlin 2006, Farsi, Filippini et al. 2007) . Although these findings provide insight on household energy use patterns, applying the findings of these studies to rural areas should be done with caution due to the unique characteristics of rural regions. Rural areas differ from urban regions in that: a) The prevalence of freely gathered traditional biofuels and their cost (in monetary term) is zero, b) Modern fuels are not available and in many cases their distribution is often relatively unreliable, c) The price of modern fuels as well as their transaction cost is  1  The author was not able to find any behavioral research that focused on energy consumption in rural households in poor developing nations.  22  Chapter 2  usually high, d) A large portion of income in rural regions is non-cash and often the cash income of rural households is too low to offer upfront payments associated with modern energy systems, e) Income in rural regions is more uncertain and variable (e.g. seasonal) and therefore regular payments that require commercial energy sources or pay-back of loans are difficult to manage, and f) Local cultures relating to cooking practices and methods are stronger in rural regions than in urban centers (Masera, Saatkamp et al. 2000). Studies on energy use in rural households in developing countries are mainly based on descriptive statistic analysis, see (Masera and Navia 1997, Davis 1998, Masera, Saatkamp et al. 2000, ESMAP 2002, ESMAP 2003, Bhatt and Sachan 2004), econometric analysis, see (Hosier and Dowd 1987, Fitzgerald, Barnes et al. 1990, Barnes and Qian 1992, Heltberg, Arndt et al. 2000, ESMAP 2003, Leiwen and O’Neill 2003, Heltberg 2004, Jiang 2004, Heltberg 2005, Wuyuan, Zerriffi et al. 2008), computational models, see (Alam, Bala et al. 1997, Howells, Alfstad et al. 2002, Howells, Alfstad et al. 2005) and, to a much lesser extent, interdisciplinary approaches, see (Masera, Saatkamp et al. 2000, Jacobson 2004, Williams and Ghanadan 2006, Bailis and Hyman 2011). However, most of the quantitative studies in this field are based on economic theories, such as the household energy model developed by Howell (Howells, Alfstad et al. 2005). 2.3  Energy transition in the developing world  One of the larger questions that analyses of household energy have tried to address is the nature of the transition process from one set of energy choices to another. This is a particularly important area of inquiry as moving from traditional fuels to modern energy carriers is associated with welfare improvement and has been on the agenda of planner and policy makers. 2.3.1  Energy ladder  The “energy ladder” hypothesis was the prominent model of explaining household energy choice in developing countries (Hosier and Dowd 1987, Leach 1992), until a decade ago (Elias and Victor 2005). The energy ladder describes a pattern of fuel substitution as a household’s economic situation changes (Hosier and Dowd 1987). The model was developed based on the correlation between income and uptake of modern fuels (e.g. electricity). The energy preference 23  Chapter 2  ladder ranks fuels: modern fuels such as electricity and LPG are considered superior fuels due to their high efficiency, cleanliness and convenience of storage and usage and are located higher up the ladder than traditional fuels, or inferior fuels, see Figure 2-1(Leach 1992).  Figure 2-1: The classic energy ladder  According to this concept, households switch from traditional energy systems to modern energy systems up the ladder at the speed and to the extent allowed by factors such as household income, fuel and equipment costs, availability and accessibility of fuels as well as reliability of modern fuel distribution, and to a lesser extent relative fuel prices (Masera, Saatkamp et al. 2000). The energy ladder concept relies on the microeconomic theory of rational choice. It assumes that all forms of fuel (traditional to modern) are available, that there is a universal set of fuel preferences, and that households will choose to move up the ladder as soon as they can afford to do so. The major achievement of the energy ladder is its ability to capture the strong income dependency of energy choice in households, particularly in urban areas. However, the energy ladder concept assumes a linear progression of fuel adoption that implies moving up the ladder means a corresponding abandonment of the lower level fuels. This assumption is inconsistent with the findings from field research (Hosier and Dowd 1987, Barnes 1992, Masera, Saatkamp et 24  Chapter 2  al. 2000, Heltberg 2004); thus, the energy ladder concept can only provide a very limited view of reality. 2.3.2  Energy stacking model  During the past decade, a growing number of empirical studies on household energy consumption have shown that linear progression along the energy ladder through a series of simple and discrete steps does not actually happen in real life; fuel switching is not unidirectional and people may switch back to traditional biofuels even once they have already adopted modern energy carriers. Fuels are imperfect substitutes for each other, and often specific fuels are preferred for specific tasks. Rather than simple switching between fuels, households choose to use a combination of fuels and conversion technologies located across the energy ladder depending on budget, preferences and needs(Masera and Navia 1997, Davis 1998, Masera, Saatkamp et al. 2000, ESMAP 2003, Leiwen and O’Neill 2003, Pachauri and Spreng 2003, Heltberg 2004). Based on this concept, once a modern fuel is adopted, traditional fuels and devices are normally kept (e.g. charcoal stove) and households only partially switch, See Figure 2-2. Empirical studies such as the study done by Leiwen in rural China (Leiwen and O’Neill 2003) indicate that some forms of traditional energy are still used by the wealthiest households.. In the case of cooking, studies have demonstrated that despite the common perceptions, LPG is not a perfect substitute for traditional biofuels and that there are clear fuel preferences based on cooking practices (Masera, Saatkamp et al. 2000). Even in places such as Brazil where the share of traditional biofuel in overall energy consumption has dropped as income has risen, the complete switch to modern fuels (fossil fuel and electricity) has occurred only at the highest income level (ESMAP 2003).  25  Modern fuels  Chapter 2  ICT Air conditioning  Electricity  Other appliances  Electricity, kerosene, LPG  Refrigeration  Refrigeration Electricity, Batteries  Basic appliances  Basic appliances  Traditional fuels  Electricity, Batteries  Cooking  Traditional biofuels  Space heating Lighting  Cooking  Traditional biofuels, Kerosene,  Space heating Candles, batteries, and kerosene  Lighting  Cooking  Traditional biofuels, Kerosene,LPG, coal, electricity  Space heating Candles, batteries, and kerosene  Lighting  Kerosene, electricity, gasoline  Income Low  High  Figure 2-2: The illustration of energy stacking- including few examples of energy systems. Adopted from (IEA 2002) and “rural energy ladder” (Barnes and Floor 1996), Note: ICT is information and communication technology.  Barnes proposed a “rural energy ladder” that illustrates the steps through which rural households generally move from traditional biofuels and human and animal power to a mix of traditional and modern fuels (Barnes and Floor 1996). A study of Mexican households by Masera et al. confirms this model by showing that as households get wealthier, the change in energy use can be characterized as an “accumulation of energy options” rather than as a linear switching between fuels. This process is termed “fuel stacking”(Masera, Saatkamp et al. 2000). Fuel stacking that includes the regular use of both traditional and more modern energy systems is found to be commonly practiced in rural regions of the developing world and, to a lesser extent, in urban centers (Heltberg 2004). In some countries, such as Ghana and Nepal, it is valid for almost the entire population (ESMAP 2003, Heltberg 2005). Causes of fuel stacking include: keeping traditional energy systems as an insurance against modern energy supplier failure (ESMAP 1999); decreasing vulnerability to modern energy price fluctuations by diversifying energy use (e.g. use electricity for lighting and fuelwood for 26  Chapter 2  cooking) (Leach 1992, Thom 2000); inapplicability of alternative energy systems to cooking methods and preferences (Masera, Saatkamp et al. 2000, ESMAP 2003); high costs associated with using modern energy sources such as electrical wiring and LPG containers preventing people from fully adopting such energy systems (Davis 1998); not having capital available to purchase modern energy conversion technologies, particularly when spending significant capital on traditional ones (e.g. wood stove) (Elias and Victor 2005); ability to trade-off time/convenience with fuel costs at particular times of day or for particular cooking practices (Mekonnen and Köhlin 2008). The energy stacking model is based on empirical evidence and is more realistic than the classic energy ladder hypothesis. Although the energy ladder and fuel stacking models are different with regards to how energy sources are adopted, they both assume the existence of hierarchies in household energy services (IEA 2002). According to this assumed hierarchy, cooking and heating are the first services to be met followed by lighting and later entertainment. An underlying assumption is that using electricity for uses other than lighting only happens after core demands are met and the first kilowatt of electricity is almost always dedicated to lighting while traditional energy is still used for cooking and heating. However, studies show that households progress along different pathways and also there is no step-like progress and households often use, if available, higher ranked energy carriers (e.g. electricity) even in small quantity while satisfying their bulk of demand by lower ranked fuels (Victor 2002). Moreover, although the energy stacking model suggests households adopt a portfolio of energy systems instead of complete switching and also considers a range of factors affecting the household energy portfolio, it still relies heavily on income as a major determinant, as well as the universal hierarchy of fuels and energy services.. However, the energy portfolio of households depends on household decisions based on a complex interaction between economic factors (e.g. fuel price), social factors (e.g. reliability of household cash income), cultural factors (e.g. cooking practices) and environmental factors (e.g. access to natural resources) (Leach 1992, Masera, Saatkamp et al. 2000).  27  Chapter 2  2.4  Factors determining household energy choice  What the existing literature shows is that there are a number of factors that affect household energy choices. The major factors identified in the literature are reviewed here, shown in Table 2-1. Although these factors are presented in isolation from each other, in the real world these factors are closely interrelated. Table 2-1: Summary of factors determining household energy choice Factors determining household energy choice Categories Endogenous factors (household characteristics) Economic characteristics Non-economic characteristics Behavioural and cultural characteristics Exogenous factors (external Conditions) Physical environment Policies Energy supply factors Energy device characteristics  2.4.1  Factors Income, expenditure, landholding, Household size, gender, age, household composition, education, labor, information Preferences (e.g. food taste), practices, life style, social status, ethnicity Geographic location, climatic condition, Public policy, energy policy, subsidies, market and trade policies Affordability, availability, accessibility, reliability of energy supplies Conversion efficiency, cost and payment method, complexity of operation,  Endogenous factors (household characteristics)  Three sub-categories are defined for household characteristics. The first two sub-categories, economic and non-economic characteristics reflects the capabilities of household and the third sub-category, behavioral and cultural characteristics, is mostly concerned about more personal aspects of households including attitude, preferences and experiences. Household economic characteristics As discussed earlier, a large body of literature points to income as the major driver of fuel choice and suggest that there is a strong correlation between an increase in income and uptake of modern fuels (Leach 1988, Leach 1992, Barnes and Floor 1996, ESMAP 2000, Victor and Victor 28  Chapter 2  2002, ESMAP 2003, Leiwen and O’Neill 2003, Barnes, Krutilla et al. 2004, Goldemberg, Johansson et al. 2004, Pachauri 2004, Wuyuan, Zerriffi et al. 2008). In addition to fuel choice, studies find that there is a strong positive correlation between income and the amount of final energy used (Fitzgerald, Barnes et al. 1990, Elias and Victor 2005). However, the energy consumption for basic services such as cooking and lighting normally remains almost unchanged (ESMAP 2003) until basic demands are met. A study in Vietnam shows that although wealthier households use almost 10 times more electricity, they continue to use approximately the same amount of energy for cooking (Tuan and Lefevre 1996). In low income groups a large percentage of income is dedicated to meeting basic energy needs (e.g. cooking, heating, illuminating). Once people have access to all basic energy services, a fraction of the additional income goes toward switching to more efficient and cleaner energy systems as well as other services such as entertainment, refrigeration, and so on (Tiwari 2000, Dube 2003). Primary energy use is expected to follow a path of inverse-U due to adopting more efficient fuel/device combination by households (Leach 1992, Foster, Tre et al. 2000, Elias and Victor 2005). However, studies indicate that although people start using more efficient energy systems as income rises, their energy consumption increases, both primary energy and useful energy (Roy 2000, Mestl and Eskeland 2009). In Brazil, energy use follows a U shape figure and although energy consumption decreases at middle incomes, it then starts to increase again at higher income levels (WEC 2004). In India, a study on the effect of increasing the efficiency of lighting appliances showed an increase in overall energy consumption (Roy 2000) . This “rebound effect” is mainly due to having access to new energy services at the same time that there is a high level of unmet demand in households. This is combined with the relaxation of household budgetary constraints since more energy can be used without additional cost (Roy 2000). In many cases, due to the absence of income data, researchers use expenditure and income interchangeably as a measure of wealth (Heltberg 2005). However, such approximating should be done cautiously since it might not be universally applied (Elias and Victor 2005), as shown in the studies by Leiwen in rural China 29  Chapter 2  Non-economic household characteristics Household characteristics such as household size, gender, age, composition and educational attainment have been found to influence energy use (Sathaye and Tyler 1991, ESMAP 2000, ESMAP 2003, Leiwen and O’Neill 2003, Heltberg 2005, Gupta and Kohlin 2006, Farsi, Filippini et al. 2007). Household size directly affects energy use by influencing the amount of energy consumed. Although total energy consumption is higher in larger households, they consume less per capita energy due to returns to scale (Fitzgerald, Barnes et al. 1990, ESMAP 1991, ESMAP 2000, Heltberg 2004, WEC 2004, Cabraal, Barnes et al. 2005, Wuyuan, Zerriffi et al. 2008). Labor has been found by many researchers to be a determining factor of household energy use, particularly in rural households (Dewees 1989, Bluffstone 1995, Hyde and Köhlin 2000, Kohlin and Amacher 2005, Modi, McDade et al. 2005, Arnold, Köhlin et al. 2006, Cooke, Kohlin et al. 2008). Household size can also indirectly influence energy use by changing income and resource availability; it also triggers fuel stacking rather than fuel switching; larger households are more likely to use multiple fuels (ESMAP 2003, Heltberg 2004, Cabraal, Barnes et al. 2005, Heltberg 2005). Gender is also found by many researchers to be a determining factor in household fuel choice and consumption (Farsi, Filippini et al. 2007, Pachauri 2007, Schlag and Zuzarte 2008). Information, education, and social learning are also described as determining factors in adopting energy systems. Lack of information regarding the alternative energy systems (e.g. improved cookstoves, LPG) and the benefits associated with using them is found to be a barrier in adoption of these systems (ESMAP 2003, Jack 2006, Schlag and Zuzarte 2008). Behavioral and cultural characteristics Household preferences and habits such as food tastes and cooking practices also influence the choice of energy system (Fitzgerald, Barnes et al. 1990, ESMAP 1991, Heltberg 2005, IEA 2006). A number of studies have identified household food preferences as a determinant of energy choice (Masera, Saatkamp et al. 2000, Cabraal, Barnes et al. 2005). For instance, a study in rural Mexico by Masera et al. found that people continue to use traditional biofuels even when they can afford to use modern fuels because tortillas are hard to cook using LPG and need to be 30  Chapter 2  cooked over a direct flame (Masera, Saatkamp et al. 2000). Similarly, Indian households prefer to use wood stoves for baking traditional bread (IEA 2006). Lifestyle and other cultural factors strongly influence fuel choice (Heltberg 2005, Gupta and Kohlin 2006). In Guatemala, indigenous ethnic groups have a energy portfolio that is significantly different from others due to their cultural differences (Heltberg 2005). Moreover, household status also influences household energy use patterns (Masera, Saatkamp et al. 2000, Xiaohua and Zhenmin 2003). In rural Mexico, LPG stoves play a role as “status symbols” and purchasing them is perceived as parallel with higher social status (Masera, Saatkamp et al. 2000). 2.4.2  Exogenous factors (external conditions)  External conditions influence household decisions regarding their energy system. This concept is explored here in more detail. Physical environment A number of geographic factors are closely related to household energy use patterns. People, particularly poor people, living in colder climates consume more energy than others in warm regions (Jiang 2004, Elias and Victor 2005). Bhatt’s study of mountain villages in India confirmed both seasonal variation and altitudinal variation of traditional biofuel consumption (Bhatt and Sachan 2004). Policies and regulations Government policies to control distribution of energy carriers and production and distribution of energy appliances indirectly affect household energy choices (Cabraal, Barnes et al. 2005). However, following the prevalent perception of income and access as the major determinants of adopting modern fuels, most energy intervention strategies have focused on increasing the availability of modern fuels, the reliability of the fuel distribution network, reducing the price of modern fuels through subsidies, and dissemination of end-use technologies. The cost of modern fuels is often too high for the poorest households to afford. However, highly subsidized markets in which price signals are absent often cause wasteful energy use (ESMAP 31  Chapter 2  2000, ESMAP 2004). Various pricing policies such as lifeline rates, cross subsidies, and blanket subsidies have been implemented with widely different outcomes. Research has shown that the subsidies often do not target the poor properly and end up benefiting middle income and even wealthy households (ESMAP 2000). In recent decades, there has been a growing trend toward market strategies rather than subsidies and government or NGO led initiatives. This is due to a variety of factors, including the economic effect of subsidies on state budgets, the growing realization that profits can be made in the household energy sector, the distorting incentives that subsidies can have on consumer behavior, benefiting rich people more etc. (ESMAP 2000, Dube 2003, Elias and Victor 2005, Victor 2009). Other forms of market distortion such as regulating production and distribution of both energy carriers and energy appliances affect household energy choice. In China, a restriction on traditional biofuel consumption caused people to adopt coal as their main fuel (Jiang 2004). Rationing energy carriers in Hyderabad discouraged people from choosing kerosene as their main energy carrier because the ration was not sufficient, distribution was not reliable, and the ration cards have not reached to all members of the society. At the same time, however, government policies that increased access to LPG through both public and private distribution channels caused a significant transition from woodfuel to LPG (Cabraal, Barnes et al. 2005). The failure rate of such policies in reaching to the target population is relatively high, signaling the vital need for overhauling the existing policy measures and proposing innovative and effective and well informed strategies. Energy supply factors Energy supply factors including the affordability, the availability, accessibility and reliability of energy supplies are found to influence household fuel choice. The affordability of a fuel is determined by its price which is an important factor in household energy use, in terms of both fuel choice and also quantity of fuel consumed (Leach 1992, ESMAP 2003, Cabraal, Barnes et al. 2005, Heltberg 2005, Schlag and Zuzarte 2008, Wuyuan, Zerriffi et al. 2008). The fuel price has been found to affect shifts between fuels in households 32  Chapter 2  using multiple fuels (Fitzgerald, Barnes et al. 1990, Leach 1992). Studies indicate the price of a modern fuel has a stronger impact on a backward substitution rather than an upward transition(Leach 1992). This implies that households gather traditional biofuels freely and continue relying on such fuels until the cost of using such fuels exceeds adopting other alternatives. For instance, in Indonesia, when kerosene was heavily subsidized even poor households switched from woodfuel to kerosene (Fitzgerald, Barnes et al. 1990). In rural areas, estimating the price of fuel is far more difficult due to the prevalence of freely gathered traditional biofuels. Self-collected traditional biofuels do not have any monetary costs and their usage depends instead on the opportunity costs of collection. The shadow price of fuel collection is unobservable and unknown except to the household itself (Heltberg, Arndt et al. 2000). Therefore, an analysis based on the relative price of modern fuels to freely gathered traditional biofuels is prone to price overestimation or underestimation. Method of payment also makes a difference. In some cases the price of modern fuels such as LPG is not higher than woodfuels, but must be bought in large amounts, unlike kerosene or woodfuels that can be purchased in small amounts on a daily basis (Leach 1992, Foster, Tre et al. 2000, Masera, Saatkamp et al. 2000). The availability, accessibility and reliability of energy supplies are major contributing factors to fuel choice. For instance, traditional biofuel use is prevalent in rural regions of the developing world, particularly in places where these fuels are available locally (Fitzgerald, Barnes et al. 1990). Scarcity of traditional biofuels strongly affects energy use and household welfare. Households generally respond to woodfuel shortages by spending more time and labor on collection, by purchasing more of their supply, by substituting straw, dung, and less favorable fuels, by economizing on woodfuel use, and by switching to commercial fuels (Agarwal 1986, Brouwer, Hoorweg et al. 1997, Cabraal, Barnes et al. 2005, Cooke, Kohlin et al. 2008). Access to reliable sources of modern fuels is also recognized as a major factor affecting fuel choice (Fitzgerald, Barnes et al. 1990, Leach 1992, Cecelski 2002, ESMAP 2002, Cabraal, Barnes et al. 2005). The effect can be observed particularly in the pattern of household energy use based on settlement size and distance from major trading routes and large cities as well as 33  Chapter 2  reliability and availability of energy distribution channels (ESMAP 2002, Chaurey, Ranganathan et al. 2004). Generally, the problem of access to modern fuel is more intense in rural areas, particularly in remote and low density areas where the distribution of modern fuels are either insufficient or unreliable (Elias and Victor 2005). The problem of access to fuels, particularly those that cannot be self collected, is closely linked to characteristics of the location’s infrastructure such as roads, distribution channels, and access to markets (Fitzgerald, Barnes et al. 1990, Wuyuan, Zerriffi et al. 2008). Factors such as electrification have been used by researchers to demonstrate the quality of infrastructure or the level of development in the region. (ESMAP 2003, Heltberg 2004). Availability of power grids not only affects the adoption of electricity and changes the service demand of the household but also correlates with uptake of other modern fuels (Davis 1998, Heltberg 2005). Although, fuel price is sometime considered a primary indicator of accessibility, studies in urban Java indicate that it is not a proper determinant of fuel accessibility and availability (Fitzgerald, Barnes et al. 1990). Reliability of the energy supply is another factor affecting the household’s energy use. Unreliable modern energy supply in many regions forces households to adopt multiple fuels and resort to woodfuel that is locally. For example, unreliability of kerosene in Myanmar causes people to rely on woodfuel although its price is three times the price of kerosene (Cabraal, Barnes et al. 2005). Energy device characteristics Energy conversion technology is a key factor in household energy usage. Despite the advantages of using more efficient technologies (cleaner and more efficient combustion), the high capital cost (e.g. energy end-use appliances as well as additional costs such as connection cost) associated with using modern energy conversion technologies is a major barrier to choosing to use modern energy systems. A survey in Kenya, Nairobi, in 1986 found that the cost of a LPG stove and half deposit for a gas cylinder is about 60 times more than a standard charcoal stove (Leach 1992). The cost of electricity connection is also frequently identified as limiting adoption of this energy form (Barnes 1988, Barnes and Floor 1996, IEA 2002, ESMAP 2004). Barnes et al 34  Chapter 2  note that service availability and the initial cost of service is far more important than monthly electricity payments or the income level of consumers (Cabraal, Barnes et al. 2005). Even when energy devices are affordable, people may still not adopt them due to their incompatibility with their existing energy service equipment. For instance, when the energy device is not compatible with widely used cooking pots, adoption of such devices is hindered (Barnes and Floor 1996). 2.5  An integrated approach to understanding household energy choice  Despite the large number of analyses conducted on household energy choice overall and the more limited set of rural household energy studies, there remains much that is unknown about the household energy transition process. While a number of factors that determine household choice have been examined through such studies (discussed above), there are a number of shortcomings in the literature that require a more integrated approach to overcome. 2.5.1  Shortcomings of the literature on rural household energy analysis  In the past three decades a lot of effort has been put into exploring the relation between human agency and energy consumption, yet our understanding of human behavior in regard to energy use is still very limited. The literature on this topic is spread across various academic disciplines and the quantity of research have declined since the mid 1980s (Lutzenhiser 1993). As Wilhite stated about ten years ago which looks valid today, “we do not know much more about the nature of energy demand today than we did in 1980” (Wilhite, Shove et al. 2000).There is a general consensus among researchers that household energy use patterns is poorly understood and further theoretical and empirical studies is required to formulate meaningful policies and intervention strategies (Leach 1992, Masera, Saatkamp et al. 2000, ESMAP 2003, Heltberg 2004, Elias and Victor 2005, Farsi, Filippini et al. 2007, Pachauri 2007). The shortcomings identified in the literature can be grouped to a) Methodological issues including “Predicting Micro Trends through Macro Analysis”, “Insufficient Data”, “Correlation versus Causation”; b) Drivers of energy use and transition including “Overemphasis on Income” and “Human Dimension of Energy Use”; c) Energy systems 35  Chapter 2  including “Focus on Energy Quantity” and “Reducing the Complexity of Household Energy use”. These issues are orderly explored in further details. Methodological issues A number of shortcomings identified in the literature are due to the methodological issues that are described here. Predicting micro trends through macro analysis A large number of studies explore the correlation between energy use and a number of factors such as population growth, technical advancement, urbanization and economic growth (see (Ghosh 2002, Victor and Victor 2002, Leiwen and O’Neill 2003, Jumbe 2004). Although macro level factors influence energy use patterns, micro trends can not be accurately extrapolated from national figures; particularly in the case of poor rural households, energy use pattern may easily be obscured by large industrial patterns (Elias and Victor 2005) The actual determining factors of household energy use can be found at the household level. The aggregate level of energy demand is made up of day-to-day decisions at the household level that are affected by a variety of socioeconomic factors. Insufficient data Where micro-level data are used, they are often not of the quality necessary to answer many questions. Much of the research in this area uses disaggregated data taken from large scale surveys such as the Living Standard Measurement Surveys (LSMS) or similar surveys (e.g. Nepal Living Standard Survey 1995/6, South Africa Integrated Household Survey 1993/4). These surveys include some information on household energy use, but not sufficient data to accurately describe household energy use patterns. A study by ESMAP on a number of such surveys showed that normally only the main and secondary types of cooking fuel and only the main lighting fuel is asked; the only information on energy quantity asked is about fuel expenditure associated with each fuel. Almost none of these surveys asked specifically about energy conversion technologies adopted by households and the energy service demand (ESMAP  36  Chapter 2  2003). The field of energy analysis in developing countries, still requires significantly better both quantitative and qualitative data. Correlation versus causation Even when detailed surveys are conducted, their estimations are often based on economic theories and various forms of regression analysis. A lion’s share of studies on domestic energy use is drawn from econometric analyses that identify the correlation between different variables and their vector of progression (Leach 1992). Econometric analyses are generally more prone to problems of causality than in-depth qualitative studies. Many of these studies ,although, show significant correlations, yet fail to firmly establish a causal relations between these factors (Stern 1986). For instance, while the strong correlation between economic and energy growth is confirmed; yet no consensus on the casual relationship between these two variable has been reached (Elias and Victor 2005). Moreover, many of these studies omit variables, interrelation between variables, and their simultaneity. Such simplifications, in turn, limit the usefulness of these studies in policy analysis. It may result in overestimation of the impact of certain variables; for example, if access to information about the health costs of traditional biofuels is positively correlated with income but omitted from the analysis, then the coefficient for income would have an upwards bias. The impact of some variables (e.g. electrification) might be attributed to unobserved household factors that are jointly correlated with this factor. For instance, income and access to fuel may both be a function of infrastructure quality and/or proxy to markets (ESMAP 2003). Drivers of energy use and transition Overemphasizing or underestimating the determinants of households’ decisions on energy use as well as energy transition is identified as a cause of a number of shortcomings in the literature. Overemphasis on income Although the impact of different variables on household energy use pattern has been explored, many researchers consider income as the main determining factor (Leach 1992, ESMAP 2003, Pachauri, Mueller et al. 2004). As Whitfield professed “three decades of research into the 37  Chapter 2  determinant of fuel choice has failed to advance our understanding beyond an association between higher incomes and cleaner fuel”(Jack 2006). Overemphasizing income or expenditure may obscure other determinants of wealth, particularly in rural areas where a large proportion of households rely on free sources of energy and home production and where the barter economy is prevalent. The dominant role of traditional biofuel as well as lack of access to modern fuels in a large part of rural developing world resulted in a limited impact of income on energy choice in such regions. Traditional biofuels that are often freely gathered depends mostly on resource availability, labor availability, land and livestock holding rather than income or expenditure. A study in India, by studying the impact of income on commercial and non-commercial fuel consumption concluded that “there is little evidence of correlation between noncommercial energy use and income” (Pachauri, Mueller et al. 2004). The key determinants of energy system choice are internal to households, and also include but are not limited to the desire for more flexible energy sources, the desire for time free from fuel collection and fire attending to spend on other activities, the desire to reduce the adverse health impacts of traditional energy systems as well as the unwillingness to abandon current practices and traditions.(Elias and Victor 2005) Human dimensions of energy use Household decision making on energy use is a complex process mainly due to the interrelations between economic, technical, social and cultural issues as well as physical environment (Masera and Navia 1997). Social and cultural factors such as cooking habits and household characteristics may make households behave contrary to economic predictions based on income and relative fuel prices (Foster, Tre et al. 2000). There are a range of non-economic variables that are central to explaining household decisions regarding energy use; yet such variables are often not included or are underestimated in energy analyses. Although, there is a large number of studies that point out the importance of such variables, most of these studies do not provide any more insight on the dynamics of such factors and why and how these factors influence the energy use. The impacts of income itself on energy 38  Chapter 2  use are not well understood; while it appears to relate to energy consumption directly, it also affects energy usage through other socio-demographic variables in many ways. For instance, income is an aspect of social class and can be treated as a rough proxy for “social location,” which affects energy usage and household choice (Lutzenhiser 1993). There is limited behavioral research on energy usage in the developing world. Variables such as gender perspectives on expenditure priorities and social power relations within households are not usually considered in quantitative analyses and require qualitative study. Even in the context of the developed world, behavioral aspects of energy use are little understood. An energy model should accurately predict the ways in which individuals (and households) behave and consume energy, react to perturbations and historical events, and respond to interventions. Such accuracy can be achieved only when individual (household) behavior is properly understood. More in depth studies on the behavioral aspects of energy use in rural households of developing countries is necessary not only to enhance existing models and theories but also to formulate proper policy and design effective intervention. Moreover, behavioral research conducted in a developed world context should be adapted and customized to inform future research in developing countries whilst accounting for the structural differences in both energy systems and socioeconomic systems (Urban, Benders et al. 2007). Energy systems Focusing on different aspects of household energy use rather than having a holistic view of household energy system is a contributing factor to a number of shortcomings. Energy services versus energy quantity Energy services have received limited attention and research often relies on the quantity of energy demand, often an aggregate number that is the lump sum of various energy requirements. However, people do not use energy but consume the services provided by energy. The term energy services is used to describe these benefits, such as illumination or cooked meals (Pachauri and Spreng 2003).As Whitile described, “graphs of increasing energy consumption are, in fact, graphs of the societal appropriation or increasingly intensive use of technologies such as cooking 39  Chapter 2  devices, lighting systems, refrigeration, and so on” (Wilhite, Shove et al. 2000). Estimations that are based solely on the quantity of energy demand are problematic when considering the differences in the services that are provided by technologies. The illumination of different lighting devices may differ significantly due to inherent differences between these sources. For instance, although 60 candles emit as much light as a 60W incandescent bulb based on luminous flux, their lights are not comparable (ESMAP 2000). In addition, many studies assume energy service requirements are constant and estimate energy requirements based solely on fuel and conversion technology. These methods do not consider the variation in energy consuming behaviors for different energy systems. Evidence shows that adopting different forms of energy systems accompanies changes in patterns of energy use such as power setting, amount of time for cooking, the types of food cooked, and nighttime lighting hours. These shortcomings often result in underestimation of household energy requirement (Fitzgerald, Barnes et al. 1990). Changes in household behavior followed by changes in energy service demand should be incorporated in energy analysis to have a realistic view of household energy use. These changes may be triggered by change in lifestyle that transforms household energy use pattern. Pointing to impact of lifestyle on energy use, Schipper states: "lifestyle changes could eat into everything you think you've saved [by adopting more efficient systems]" (Cherfas 1991) Reducing the complexity of household energy use Households adopt different energy systems to satisfy their various energy service demands. An energy system is a series of processes through which primary energy1 is extracted, converted in one or more steps to final energy2, and then converted again through energy end use devices to useful energy3. The useful energy provided by energy systems benefits consumers through  1  Primary energy: The energy carrier prior to any form of conversion (e.g. crude oil) Final energy: An energy carrier that is suitable for use in end use energy devices such as stoves (e.g. electricity, kerosene, diesel) 3 Useful energy: also called utilized energy, energy output, end-use delivered energy, or available energy. The energy that is needed for a specific task, such as heat needed to cook a meal. Depending on the efficiency of the 2  40  Chapter 2  energy services such as cooked food and indoor lighting. They often use a single energy device for satisfying multiple needs (e.g. cook stoves for cooking, water heating, space heating), adopt multiple energy systems for satisfying a service requirement (e.g. lighting by fire in three stone stove, kerosene wick lamp, torches), and produce end use energy through some sort of conversion (e.g. use batteries and dry cells to produce electricity). The energy required to meet household energy service demands varies significantly among households. Engineering-based approaches have been used by researchers to estimate the energy required to satisfy the basic needs of a household; Goldemberg estimated household requirement of direct primary energy per unit of time to satisfy their basic needs is about 500 Watts per person be 43 MJ per day per capita, the Advisory Board on Energy (ABE) in India estimated approximately 33 Watts of useful energy per capita is needed. .. Such estimations are based on a number of assumptions regarding energy systems adopted by households, such as the energy conversion technology used, its efficiency and intensity of use. They are additionally based on normative definitions of basic needs, which can be problematic as basic needs depend on complex factors such as climate, culture, region and period in time in addition to subjective wants (Goldemberg 1996, Pachauri, Mueller et al. 2004). Moreover, household energy systems are often oversimplified and their different aspects examined in isolation of each other to simplify the models and analysis. A majority of existing research has focused on fuel choice and to a lesser extent on the combination of fuel and energy device used. Their focus is mainly on the energy carriers (Masera and Navia 1997, Masera, Saatkamp et al. 2000, ESMAP 2003, Farsi, Filippini et al. 2007, Pachauri 2007) and to lesser extent on fuel/device mix (Tuan and Lefevre 1996, Chambwera 2004, Edwardsa, Smith et al. 2004) However, household energy systems are really a combination of three factors: energy carriers, energy conversion technologies, and energy service demands. For instance, the energy  conversion technology it may be as little as 5-8% of primary energy input or as high as 95-100% in the heat delivered (Leach 1987)  41  Chapter 2  requirement for cooking, which can account for up to 90% of household energy demand among poor populations in the developing world (Elias and Victor 2005), depends on the specific cooking service requirements (i.e. boiling water for tea and slow-cooking a stew are two very different requirements), the cooking device, and the cooking fuel. Cooking service requirements change depending on multiple factors such as household size, number of meals cooked per day, meal ingredients, and cooking methods. Energy requirements for cooking also depend on the energy carrier and the energy conversion technology (i.e. cooking stove efficiency). The energy delivered to a cooking pot per kilogram of fuel differs significantly, ranging from 3 MJ/Kg of fuel for a three stone wood fire up to 25-30 MJ/Kg of fuel for an LPG stove.(Barnes and Floor 1996)1. Similar dimensions of energy systems can be explored for lighting, space heating, water heating and other energy consuming activities. Addressing these three interrelated concepts together is necessary for a realistic approach to household energy analysis. Furthermore, these three dimensions of household energy systems need to be investigated for each energy consuming activity to be able to create a detailed profile of household energy use. Doing so will also illuminiate the inter-relationships between these varying activities since devices and carriers can be used to meet multiple energy services (e.g. cookstoves provide cooking, heating and lighting services). If treated in aggregate or only according to fuel or technology, such interdependencies are not captured. 2.5.2  A new approach to household energy analysis  The gaps in the literature on rural household energy analysis provide a basis for developing an alternative model that can create a more realistic view of household energy use. The three dimensional energy profile is presented as a new conceptual framework for assessment of household energy use, see Figure 2-3. This framework acts as a basis for building new theoretical and empirical models of rural household energy use. At the center of the framework is  1  Energy delivered to pot based on fuel’s energy content and the efficiency of conversion technologies typically used in developing countries (Barnes and Floor 1996)  42  Chapter 2  the relationship between energy services, devices and carriers. Around that are the various factors influencing the energy profile of the household. The framework addresses energy use at the most disaggregated level, the household, and can be used with both quantitative and qualitative data Therefore, the framework is able to capture the micro trends based on quantitative information as well as objective qualitative data. It also provides a template for capturing the casual relations between different aspects of household energy use. As discussed earlier, household energy use patterns depend on simultaneous decisions on the type of energy carriers and energy conversion technologies as well as the energy service requirements, which altogether can be considered the Household Energy System. The proposed framework addresses all three dimensions of these systems simultaneously while considering the close interrelations between them and their dynamics (Figure 2-3). The framework focuses particular attention on the energy services instead of energy quantity to shed more light on the human side of energy use and create a more realistic view of household energy use. Therefore, the framework seeks to provide a realistic view of household energy consumption by considering energy service requirement instead of the quantity of energy demand, as well as capturing all three dimensions of an energy system and their interlinkages. The three dimensional energy profile model places the energy systems as the central focus for assessment, while recognizing a range of influential drivers affecting these three dimensions. Household energy use can be affected by a range of drivers through complex, interacting, and reciprocal linkages. Meeting the cooking service requirement of a household, for instance, depends on the availability and affordability of energy carriers and conversion technologies, household characteristics (e.g. income, household size), household preferences, and so forth. The framework, while considering income as a determining factor under the category of capabilities, avoids over emphasis on income by drawing equal importance to other sets of variables and determinants. Moreover, the three dimensional energy profile framework draw together different dimensions of human behavior in order to capture the relationship between human agency and energy consumption. 43  Chapter 2  Figure 2-3: Endogenous and exogenous factors influencing energy profile Notes: a) Only examples of variables are shown in the boxes. This is not a complete list of variables related to energy use. b) The categories of variables affecting energy use is adopted from (Wilson and Dowlatabadi 2007)  44  Chapter 2  Determining factors affecting household decision regarding energy consumption, located in the circle, are strongly interrelated and simultaneously affect the three dimensions of household energy use pattern. This part of the framework is adopted from the integrated model of proenvironmental behavior that is described in section 2.4. Four types of casual variables are identified by Stern (Stern 2000) under two main domains; attitudinal factors and habits and experiences under the personal domain, and personal capabilities and external conditions under the contextual domain. Attitudinal factors include behavior-specific beliefs, values and personal norms. Several socialpsychological theories such as Value, Belief, Norm (VBN) theory, cognitive dissonance theory, norm-activation theory, and planned-behavior theory are proposed to explain the specific behavior. Habits and experiences such as standard operating procedures and routines are the key determinant in human behaviors since behavior change often requires breaking routings and creating new ones. Personal capabilities include socioeconomic status, technical skills and resources required for an action. Sociodemographic factors such as gender, age, ethnicity, education, and income, etc, can all be considered as personal capacities. External conditions include factors such as market actors, technologies, regulations, formal and informal institutions, economy, supply chain, social interactions and norms, etc. Policies are under this category. (Stern 2000, Wilson and Dowlatabadi 2007). Most of the literature on human dimension of energy use has addressed the factors described under the contextual domain that includes personal capabilities and external conditions. In this paper, instead of the set of variables suggested by Stern and Wilson et al., the determining factors reviewed in the previous chapter is used, except for the attitudinal variable. However, there are a number of variables associated with each category that need to be identified and incorporated in the framework. It provides a basis for further qualitative studies on energy use in rural households.  45  Chapter 2  As a final note, a full assessment of these drivers and interactions between them, as well as their causal relationship with energy systems, requires a multi-scale approach in order to understand how changes at different scales (e.g. individual household members, the household as a unit, community, nation, international) affect household energy use. 2.5.3  Energy transition in a three dimensional energy profile model  Based on the three dimensional energy profile framework, a new method of identifying household energy transition is proposed here. Energy transitions in this model are not limited to switching between fuels, stacking multiple fuels, or adopting improved cookstoves. Instead they include all three dimensions of household energy systems. The 3d graph below provides a holistic view of household energy system characteristics and the shifts occur to it due to changes in any of three dimensions of household energy system (i.e. energy service demand, energy carrier, energy conversion technology). This graph is a representation of the “social appropriation”(Wilhite, Shove et al. 2000) of energy use along with increasing use of a more efficient and modern energy system. A change along the path shown in Figure 2-4, can be seen as an improvement in overall household energy usage. For instance, the improvement which is shown by the arrow in Figure 2-4, can be the result of: increase in energy service demand that is often associated with a higher welfare level (e.g. more warm meals, brighter room, more comfortable room temperature); fuel switching to higher quality and cleaner fuel (e.g. switch from kerosene to LPG); adopting higher efficiency energy appliance (e.g. replacing a traditional cook stove with an improved one). Studies confirmed that the improvement in any of these dimensions directly or indirectly has positive impact on household welfare level.  46  Chapter 2  Figure 2-4: Energy profile transition through a three dimensional path  The graph of energy transition is not necessarily a smooth and continuous line. Rather, it is most likely in the shape of steps in each of the three dimensions. There are a number of drivers affecting the three dimensions of energy use, described earlier in this section that triggers such transition. The cumulative effect of a range of drivers or the effect of each factor on the transition of energy use can be visualized through the proposed graph. . 2.6  Conclusions  As frequently mentioned here, there is a large gap in the literature regarding the behavioral aspects of energy use. Although some research has attempted to include cultural and habitual factors and has confirmed their importance, there is almost no research that explores these variables and their dynamics in detail. For instance, while an econometric analysis looks at the differences between energy use in households headed by males and females, there is no explanation of why these differences exist and how they may change. As a final note, the growing concern about energy and household welfare, impacts of climate change, and energy security requires a more realistic understanding of household energy use. An in depth study of 47  Chapter 2  the human dimension of energy use is a vital step for improving our understanding of household energy use in rural regions of developing countries. The proposed conceptual framework provides a basis for both qualitative and quantitative researches. The framework addresses the gaps identified in the literature by focusing on household level qualitative and quantitative data in order to be able to capture the micro trends based on appropriate information and identifying the casual relationship between variables. By incorporating a range of personal and contextual variables the framework avoids overemphasizing income as the major determinant of energy consumption and also addresses the human dimension of energy use. Finally, the three dimensional energy profile consider all three dimensions of energy system (i.e. energy service, energy carrier, and energy conversion device) and their interrelations to have a realistic view of household energy use. The three dimensional energy profile framework can act as a basis for building new theoretical and empirical models of rural household energy use. Such models along with in depth study of rural energy usage are precondition for formulating any meaningful strategy.  48  Chapter 3  Chapter 3 Private sector involvement in rural energy provision 3.1  Introduction  The close relationship between energy and development has long been acknowledged by the international community. Addressing energy poverty1 has been on the agenda of governments and development organizations for more than three decades. Energy poverty is not only considered a factor that hinders socioeconomic development, but is also thought to be associated with a wide range of adverse social and environmental implications. Governments, international organizations, donors, and non-governmental organizations have all made substantial efforts to address two primary energy access problems: electrification and cleaner cooking energy. Various institutional and technological approaches have been adopted and tried out, but as yet programs to address the developing world’s energy poverty have had limited success. Recently, there has been a growing emphasis on private sector based solutions to energy access problems in the developing world. That is partially due to some high-profile failures in government-led energy poverty alleviation programs (e.g. the first national cook stove program in India), the financial burden on governments and donors (particularly in the case of fuel subsidies), and the private sector’s growing realization that there may be business opportunities in serving such a large population. The result is that a wide range of players, including the donor community (e.g. the World Bank and bilateral donors), non-governmental organizations, private and “social” enterprises, and governments, have expressed interest in promoting the private sector as at least a partial solution, particularly for distributed generation technologies aimed at reducing the impacts of traditional energy systems by improving access to electricity and clean cooking system. This paper investigates private sector based interventions related to electrification and the provision of clean cooking energy systems by examining two major enterprises active in rural energy provision in India. We hypothesize that:  1  Although the precise definition of energy poverty has not yet been agreed upon, a widely-used description that has been adopted by the International Energy Agency is lack of access to modern energy (i.e. lack of access to electricity and reliance on traditional cooking systems).  49  Chapter 3  I)  The business-based approach to rural energy provision cannot be a universal solution and is not appropriate for all segments of market.  II)  Even if enterprises are addressing a more profitable segment of the market, they face many challenges that are beyond their capacity to address.  Our evidence justifies a more balanced approach to energy provision instead of putting energy provision solely in the domain of the government or the private sector. Our case studies indicate that while the private sector can successfully provide energy to certain segments, government or not-for-profit sector support is necessary for energy provision to other segments. However, appropriately segmenting the market may be quite difficult. Moreover, it indicates that some challenges inherent to rural energy provision in the developing world would require the government to make a long-term commitment to facilitate the operation of private entities in energy provision even in segments that are currently assumed to be amenable to private sector supply. Section 2 provides a brief background on the programs that try to address the two major energy access issues (electrification and clean cooking system provision). In Section 3 we lay out our case studies, and in Section 4 we examine hypothesis I and II. Section 5 presented some concluding remarks. 3.2 3.2.1  Background Electrification  Electricity provision, including power generation, transmission, and distribution, has predominantly been achieved through centralized systems, mainly due to economies of scale and the high capital cost of energy” (Zerriffi 2010). Although this setup has performed well in many situations, its economic feasibility is highly restricted in the case of electricity provision to remote regions with low population density. Centralized electrification of rural areas faces many challenges, mainly due to: 1) high capital costs and low rates of return for both power generation capability and grid expansion (due to rural households’ inability to make cash payments as well as low consumption from rural households that are dispersed over a large geographical area) and 2) the low quality of the electricity being delivered (the result of generation shortages, long transmission lines, and poor maintenance) (Zerriffi 2010). 50  Chapter 3  Despite these shortcomings, most electrification programs are implemented based on centralized systems. Recently, there has been growing interest in providing electricity to rural areas using small-scale distribution methods. In the “New Policies” section of the IEA’s latest publication, a range of technological options, such as grid, mini-grid, and off-grid solutions, are incorporated (IEA 2012). There is a recognition that for both technical and institutional reasons, universal access via grid extension may not be feasible or may take a significant amount of time to reach some rural areas. Distributed generation has been found to be a suitable intermediary alternative to grid extension for rural electrification because of its scalability and ability to operate autonomously; those qualities are well suited to low population density, low consumption levels, and the specific consumption characteristics of rural areas (e.g. there is a greater imbalance in the consumption levels of rural households in comparison to urban populations since most rural households only use electricity for lighting and few use appliances, such as refrigerators, that consume large amounts of energy) (Zerriffi 2010). Distributed generation may also allow local control (Zerriffi 2010) and may decrease the overhead costs associated with centralized electricity provision systems (Banerjee 2006). A number of different technologies have been adopted for distributed generation, including: 1) non-renewable resources such as diesel engines, natural gas engines, micro-turbines that run on natural gas, and Proton Exchange Membrane (PEM) fuel cells; 2) renewable resources such as wind turbines, solar photovoltaic (PV) systems, biomass gasifiers that produce gas to run an internal combustion engine, Bagasse cogeneration in sugar factories, and small hydropower, geothermal, ocean thermal, tidal, and solar thermal power generation options (Banerjee 2006). Biomass-based energy systems, particularly biomass gasification, have received increasing interest worldwide in recent years. A World Bank report notes that some of the key advantages such systems have over fossil fuel systems and other renewable energy sources (Kartha, Leach et al. 2005) are that biomass-based energy systems use fuel that: 1) Is “widely available.” Biomass fuels are diverse and in principle can be found anywhere that agricultural or forestry activities are taking place.  51  Chapter 3  2) Is “available on demand.” Similar to fossil fuels, biomass material does not require expensive storage systems. Different from other renewable sources – like wind, which can only be supplied intermittently – fossil fuels contain energy. 3) Is “convertible to convenient forms.” Biomass material can be converted to all energycarrying forms, such as electricity, gases, and liquid fuels. It is also appropriate for decentralized power generation. 4) Has the “potential to contribute to greenhouse gas reductions and other environmental objectives.” Biomass-based energy is more climate friendly than fossil fuels, particularly if it is based on renewable resources. 5) Is a “source of rural livelihoods.” Bioenergy systems can operate locally and thus can result in income-generating activities by allowing local residents to run the systems and help supply raw material, both of which can potentially promote rural development. Bioenergy systems have many other advantages that can directly and indirectly contribute to rural development. However, researchers also pointed some of the drawbacks of agricultural residue based energy systems such as low energy content in compare to fossil fuel, time needed to process fuels for using in biomass gasification (e.g. sizing, drying) (Dasappa 2011), high cost of biomass based power plants (Buragohain, Mahanta et al. 2010, Dasappa 2011), complexities of biomass gasification power plants (Ghosh, D Sagar et al. 2006) requiring expertise and local manufacturing capacity (Dasappa 2011), limited availability of engines that run on producer gas (Banerjee 2006), low load factor of stand-along biomass gasification based power plants (Buragohain, Mahanta et al. 2010), and the potential impact on agricultural productivity (MA 2005) Agricultural residue as a raw material for bioenergy systems has received much attention as it does not cause land competition with food production and also because in many instances it is considered waste (Kartha, Leach et al. 2005). Biomass can be converted to combustible gas through: 1) high-temperature thermochemical processes (i.e. through a biomass gasifier that generates “producer gas”) and 2) low-temperature biological processes (i.e. via an anaerobic 52  Chapter 3  digester that produces “biogas”). Electricity generation through the combination of a gasifier and a gas engine (or through a modified diesel engine) has been considered one of the most promising bioenergy options for distributed generation and rural electrification (V. Siemons 2001, Dasappa 2011, Mahapatra and Dasappa 2012), particularly in areas with high biomass material availability, such as developing areas of Asia, including India (Romijn, Raven et al. 2010). 3.2.2  Improved cookstove  Globally, biomass material is most prominently used for cooking (GEA 2012). It is mostly used in traditional and inefficient energy systems, and switching to modern cooking fuels is considered vital for improving the welfare of traditional biomass users, of which there are 2.6 billion worldwide (IEA 2012). But despite three decades of attention to household cooking energy systems, it is unlikely that we will observe a massive transition toward modern energy in the short to medium term. Foell et al. estimate that even providing access to non-traditional energy systems (i.e. improved cookstoves (ICSs), ethanol stoves, LPG, electricity) to only half of the population currently relying on traditional biomass-based cooking methods by 2015 (recommended by the United Nations Millennium Project (Modi, McDade et al. 2005)) would require providing 800,0001 people per day with access to modern energy systems (Foell, Pachauri et al. 2011). That means transitioning to solely modern energy carriers, such as liquid or gaseous fossil fuels, is simply impossible in the short term, and alternatives and more immediate solutions are necessary. Many researchers consider ICSs (Venkataraman, Sagar et al. 2010) a promising intermediary alternative that can reduce the adverse social, economic, and environmental impacts of current traditional energy systems (Foell, Pachauri et al. 2011). Programs addressing household cooking energy systems started as early as the 1970s and were motivated by a variety of factors, from the assumed links between deforestation and household energy (the so-called fuel wood crisis during 70s) (Arnold, Kohlin et al. 2003) to recent concerns about the health of stove users (Smith, Mehta et al. 2004, Bailis, Cowan et al. 2009) and global  1  The number is based on an earlier International Energy Agency (IEA) (IEA 2010) estimate that places the number of people using biomass as cooking fuel at 2.7 billion; the most recent estimate is 2.6 billion (IEA 2012).  53  Chapter 3  warming (Smith, Uma et al. 2000, Zhang, Smith et al. 2000, MacCarty, Ogle et al. 2008, Aunan, Berntsen et al. 2009). There has been an ongoing attempt to promote the dissemination of ICSs around the world; currently, more than 160 programs with a variety of objectives, technology, dissemination techniques, and financial mechanisms are running (Gifford 2010). In September 2011, the major initiative “Global Alliance for Clean Cookstoves (GACC) was launched by the United Nations Foundation; the initiative is a public-private sector partnership aimed at promoting a global market for clean cooking technologies (Martin, Glass et al. 2011). The United Nations Environment Programs (UNEP) was the main implementer of this initiative and is working with more than 250 other organizations to achieve the goal of “100 by 20,” which means providing 100 million homes with clean cooking fuel by 2020 (Bailis and Hyman 2011). Bailis estimates, based on UNDP/WHO data, that there are around 200 million stoves in use around the world. Considering the fact that the average household contains four to five people, that means 30% of the overall population that relies on traditional fuels is using ICSs; China holds 75% of that 30% (Bailis and Hyman 2011). So far, most ICS dissemination programs have been unable to achieve their targets, and only a few programs have been able to disseminate more than a few thousand ICSs (Bailis, Cowan et al. 2009). Research has often been focused on the technical aspects and the efficiency of ICSs as well as on stove emission characteristics and pollution exposure (Barnes, Openshaw et al. 1993, Smith, Dutta et al. 2007, Takama, Lambe et al. 2011); less attention has been paid to the programs themselves that are trying to achieve adoption and sustained usage of these stoves which is critical in formulating the successful programs (Agarwal 1983, Hessen, Schei et al. 2001, Ruiz-Mercado, Masera et al. 2011, Takama, Tsephel et al. 2012). 3.3  The root of the private sector based approach to energy access provision  The challenges of providing access to modern energy services in rural regions of developing countries – for example, by implementing biomass-based distributed generation projects – are not just technological; policies and institutional factors both play a significant role in the success or failure of energy provision programs (Turkson and Wohlgemuth 2000, Dubash 2003, Haanyika 2006, Verbong, Christiaens et al. 2010{Zerriffi, 2010 #572)}. As such, efforts to use 54  Chapter 3  the power of the market to disseminate cleaner cooking technologies and distribute electricity generation need to be placed in the broader context of the last few decades’ energy sector reforms and their relationship to the pressure that has been placed on developing country governments to restructure their economies. Since the 1970s, when the issue of energy access in the developing world found its place on development agendas, energy provision has mostly been in the control of governments (Bailis, Cowan et al. 2009). In many parts of the developing world, energy has been and still is the domain of the state. The state-led model is based on a top-down approach that is focused on “centralized service delivery” and “subsidies” run by central and sub-national governments (Williams and Ghanadan 2006, Zerriffi 2011). In electrification, technical, centralized systems are often accompanied by centralized institutions and organizations. Even many decentralized approaches only address technical decentralization without considering the decentralization of the necessary institutions. The large capital requirements and the strategic value of energy provisions in government policies and plans are some of the drivers of the state-based model (Eberhard 2004). In the late 70s and early 80s, following two major economic crises, neo-liberal discourses around the role of the state vis-à-vis the private sector emerged and became the dominant model (Heynen and Robbins 2005, Schmidt 2011). Putting neo-liberal policies into action and moving toward the restructuring of various economic sectors in the UK and USA resulted in unbundling, privatizing, and marketization (Victor, Heller et al. 2007); this shift further caused the transformation of major donor institutions such as the World Bank and the International Monetary Fund (IMF) (Ahmed 2006). In turn, development practices shifted toward private sector based approaches (McCarthy and Scott 2004); examples of this type of approach include attempts to commercialize ICSs (Bailis, Cowan et al. 2009) and power sector reforms (Haanyika 2006, Victor, Heller et al. 2007). Adherents of this shift believed that state-led models could have only very limited success in expanding energy access due to their low-quality services, unsustainable enterprise practices, and lack of innovation (ESMAP 2000, Bacon and BesantJones 2001). Minimal role of state in functioning and regulating the market was considered by pro-market advocates as the most efficient way of allocating resources (Peet and Watts 1993) 55  Chapter 3  and providing services, even those that were traditionally the domain of governments, such as energy provision (Williams and Ghanadan 2006). The shift toward reform and structural adjustment concurred with the debt crisis of the 1980s, which was also partly the impetus for the shift. At that time, the strain on states’ financial resources and governments’ inability to balance their books limited the expansion of energy access, the improvement of infrastructure, the maintenance of energy subsidies, and promoted the creation of immediate revenue through the sale of state assets (Bacon and Besant-Jones 2001, Bailis, Cowan et al. 2009). Projects initiated by civil society, international NGOs, aid organizations, and a range of foundations emerged in response to governments’ diminishing role in providing public services and development programs (Raustiala 1997). Nevertheless, since the 1990s, inter-country issues – in conjunction with the political economy’s shift toward neoliberal doctrine – have placed more emphasis on private sector based approaches focused more on investing strategies than on charities and aids. Additionally, major international financial institutions such as the World Bank (World Bank 1993) have advocated for structural adjustment and reform as well as for the creation of lending policies that require governments to make progress at market liberalization before loans are disbursed (Bacon and Besant-Jones 2001). During the 90s, many developing countries, including India, were coerced into adopting these private sector based policies by the IMF (Ahmed 2006). The assumptions supporting private sector based approaches to infrastructure development – or energy sector reform, as Bacon describes it – are that the private sector is a) more efficient at resource allocation due to cost recovery, b) performance enhancing due to the profit motive, and c) cost reducing due to competitive pressures (Bacon and Besant-Jones 2001). The political shift toward liberalized markets has also reached development projects that are more broadly based both on the assumptions mentioned above and on the idea that the elimination of poverty is connected to economic growth and economic growth is connected to private sector based economies (Wamukonya 2003). Donors, as well as multi-lateral and bilateral agencies, have put increasing pressure on developing countries to adopt private sector based approaches to their development projects using the “carrot and stick” policy of conditional 56  Chapter 3  aid and loans (Lefevre and Todoc 2000, Hoffman, West et al. 2005). As a result, the past two decades have seen a decrease in government involvement in the energy sector, and reform in the power sector has been picking up pace (Bacon and Besant-Jones 2001). The renewed attention to ICS dissemination programs also happened in the same period when donor communities pressed the stove developers to adopt the business-like programs and be profitable (Hoffman, West et al. 2005). A large number of projects whose sustainability should be guaranteed by a healthy return on investment have been launched. However, there is evidence pointing to failures of private sector based (sometime also referred to as market-based approaches)1 approaches to energy provision due to significant market imperfections in many developing countries (Boberg 1999, OECD 2004, WEC 2004, Schlag and Zuzarte 2008, Howells, Jonsson et al. 2010, Mink 2010) . A number of market distortions are identified in the literature, of which Howells listed a number of them under two broad categories: “informational failures” due to the poor consumer, supplier and government informationt; and “management failures” including government intervention, land mismanagement, prevalence of a barter economy, externalities (i.e. carbon costs associated with climate change and health costs) that are not included in the energy price, monopoly power of energy device/carrier suppliers, lack of financial institutions and services, and labor market failure (Howells, Jonsson et al. 2010). Moreover, studies indicate that huge population in both urban and rural regions are excluded from the market due to poverty (WEC 2004) Researchers both for and against private sector based approaches to energy poverty have studied the method in detail, though mostly in theory or based on experiences in the developed world. Private sector based approaches to energy poverty are still in their infancy, and accurate evaluation of their outcomes has yet to occur. Achieving the potential of private sector based approaches while avoiding adverse impacts for poor populations necessitates a more  1  It is important to note that in this dissertation I investigated the issues that caused the failure of private sector based approaches in the Indian context (so called market-failures). This does not imply that the “market” failed but rather that the private sector was unable to play the role in the market that it was assumed they could.  57  Chapter 3  comprehensive understanding of their advantages and deficiencies. Much further investigation of these projects is needed. 3.4  Case study  The aim of this study was to explore the challenges and opportunities of commercial approaches to energy poverty. It was completed by investigating two of the most significant private energy enterprises in India, both of which have been active in biomass-based energy in rural areas; one in the provision of clean cooking energy, and the other in the electrification of rural areas through distributed generation.1 The selection of these two companies was based on the fact that they a) have the largest consumer base among enterprises that provide energy in rural regions of India and b) both use agricultural residue, which has been promoted as a promising energy provision alternative in rural areas. Selecting entities active in providing two different energy services (electrification and clean cooking energy system) enabled us to explore issues that are common among biomass-based energy providers. Following the first round of the field study and interviews, follow-up interviews to confirm our initial findings were done with these companies as well as two other enterprises. 3.4.1  Biomass energy and energy poverty in India  India is facing the pressing challenge of meeting the fast-growing energy needs of its large population, of which around 40% is classified as below the poverty line (Parikh and Parikh 2011). IEA estimates indicate that almost 293 million people suffer from lack of access to electricity, while 772 million rely on traditional biomass for cooking; 92% and 90% respectively reside in rural regions (IEA 2010). Electricity generation has remained a challenge, as has distribution; currently, India’s power generation capacity is far less than its energy demand (Buragohain, Mahanta et al. 2010) and government policies on rural electrification are highly criticized for their inability to provide a reliable supply of electricity (Kishore and Ramana 2002). Meeting rural electricity demand, which is mainly limited to domestic lighting, irrigation,  1  As per UBC Behavioural Ethics Review Board approval of the research protocol, the identities of both firms and individual respondents has been kept anonymous.  58  Chapter 3  and small-scale commercial activities, is one of the Indian government’s top priorities (Buragohain, Mahanta et al. 2010). Providing cleaner cooking fuel has also been a major concern for a long time (Buragohain, Mahanta et al. 2010). LPG and kerosene are supplied to households by the government at subsidized prices. However, LPG (44% subsidy) diffusion has been limited to urban areas and has been constrained by supply shortages and the high upfront cost of LPG cylinders. No specific government program is in place to improve LPG accessibility in rural areas (Lambe and Atteridge 2012). Kerosene is also highly subsidized (i.e. 69% of the market price) and is distributed in rural areas through a public distribution system. However, it is mainly used as a lighting fuel and as fire starter. Moreover, households using subsidized LPG are only entitled to half of the normal ration of kerosene at the subsidized price (Lambe and Atteridge 2012). To reduce its dependence on imported fossil fuels, India, along with many other developing countries, has been promoting modern energy provision based on locally produced biofuels (Takama, Lambe et al. 2011). Biomass-based energy constituted about 10% of the world’s primary energy supply in 2008 (IEA 2011). With a growth rate of 8.2% per year, it is anticipated that renewable sources such as biofuels will be the fastest growing fuel in the coming decades and will gain a larger share in the energy mix (2010-2030) (BP 2011). However, most of the world’s biomass-based energy usage occurs through traditional and inefficient methods, such as the combustion of wood fuel in three stone stoves (Ghosh, D Sagar et al. 2006). An increasing number of people are relying on traditional biofuels for their cooking. By 2030, there will be 2.8 billion people using these fuels (though by 2030 that will represent a slightly lower share of the world’s population than it does today); most of them will live in Sub-Saharan Africa, India, and Southeast Asia. (IEA 2010). Indian biomass use in various forms was about 110 Mtoe in 2000, of which 81% went toward domestic energy use, primarily cooking. The industrial sector also widely uses these fuels, mainly for heat generation (Parikh and Parikh 2011). However, at the same time, biomass is also 59  Chapter 3  considered a promising alternative energy source that has the potential to not only meet India’s rapidly-increasing energy demand, but also to address rural energy poverty where it is widely available (Romijn, Raven et al. 2010). India’s biomass-based energy potential is estimated as being above 100,000 MW, of which 16,000 MW are from agricultural residue (Bhattacharya and Jana 2009). At present, however, the major sources of biomass in India are agricultural residues and by-products of agro-industries. Biomass-based energy potential estimates often deal with the ratio of various crop residues and useful products. The National Productivity Council of India provides the updated estimates for state-wise and crop-wise agricultural residue production. For instance, for the agricultural production for 2006-2007, the council’s estimates show that 90 million tons of rice production results in 27 million tons of rice husks, based on a 0.3 crop to husk ratio. However, that is a raw estimate of residue production; the amount of residue that can be used for power generation without changing the current usage – for example, as cattle feed and cooking fuel – is estimated to be about 15-20% of the total residue production (Buragohain, Mahanta et al. 2010). The estimates of potentially available crop residue are nevertheless constrained by available land, which is a relatively scarce resource in India (Parikh and Parikh 2011). India began promoting renewable energy in 1982, when a Commission for Additional Sources of Energy was established under the country’s Department of Science and Technology. The Ministry of Non-conventional Energy Sources (MNES) started work in 1982; it was renamed the Ministry of New and Renewable Energy (MNRE) in 2006 (Bhattacharya and Jana 2009). The government of India has addressed biomass-based energy as well as rural energy provision through a number of policies and programs, such as the Renewable Power Purchase Guideline (1993), the Electricity Act (2003), the National Electricity Policy (2005), the National Tariff Policy (2006), the Rural Electrification Policy (2006), and the Improved cookstove Program(2009). However, the impact of these government initiatives and policies has been quite limited. Despite the Indian government’s efforts in recent years, rural electrification remains a major challenge, particularly in India’s poor states. The ICS programs also have not shown very promising results. Though India’s massive stove program is the second largest of its kind after China’s, current estimates show that fewer than 6 million households are using ICSs 60  Chapter 3  (Bhattacharya and Jana 2009) compared to the approximately 193 million households1 that are using biomass for cooking (IEA 2012, Shrinivasan 2012) As in other parts of the developing world, government inefficiency, donor-driven programs, and concerns about long-term replicability and sustainability have led to an increased interest in the role that the private sector can play in addressing energy poverty, and huge market potential has been perceived in rural energy. A report prepared by the Center for Development Finance estimates that 76% of rural households – and almost 60% of the population – are at the Base of the Pyramid (BoP) of wealth. These 114 million rural households spend almost 224 billion INR per year on energy, based on 2004/2005 expenditure data (Bairiganjan, Cheung et al. 2010). The potential market value per year for providing energy access only to BOP rural households is estimated to be around 94.06 billion INR (2.04 billion USD) for decentralized biomass-based and micro hydro energy; 1.26 billion INR (27.39 million USD) for solar home services2; 885 million INR (18.58 million USD) for solar lanterns3; and 1.11 billion INR (24.13 million USD) for ICSs (Bairiganjan, Cheung et al. 2010). Despite the existing potential, very few commercial entities are providing energy to rural households, let alone the BoP population. In this paper we discuss the challenges that are preventing commercial entities from getting more involved in the rural energy market by looking more specifically at two cases in India. 3.4.2  Background information on the commercial entities  Initially, two commercial entities that use agricultural residue to provide energy to rural households were selected. Both companies are prominent enterprises active in rural energy provision and had the largest customer base in their field of operation. In the second round of interviews, follow-up interviews with the same companies were done. Additionally, a number of other companies active in rural energy provision were interviewed; of those companies, two were  Rukmini (2012). Median Household Size Drops Below 4 in Cities. The Times of India. New Delhi.  1  Based on the 2011 census, the median household size in India is four Shrinivasan,  2  Solar Home Systems provide multi-usage electricity through solar panels (photovoltaic). Solar Lanterns are portable LED lanterns that are powered by solar panels (photovoltaic) and can provide light for four to eight hours.  3  61  Chapter 3  included in the analysis (see Table 3-1). Further detail is presented throughout the section. Their areas of operation are shown in Figure 3-3 at the end of this section. Table 3-1: Private entities included in the case study  Company  Energy service  Technology/Product  Company A  Cooking energy system  Advanced ICS and pellet fuel  Company B  Electrification  Electricity through biomass gasifier and generator set  Company C  Cooking energy system  ICS  Company D  Electrification  Electricity through biomass gasifier and generator set  Company A (Stove Developer) Company A was founded in 2006 by a large multinational oil company’s emerging consumer market division. In 2009, the company was sold to Indian entrepreneurs who were members of the local management team. It is based in Pune and its residential market operations are mainly focused on rural Karnataka and Maharashtra. Company A’s stove, X, has been promoted as a low-smoke, forced-draft cook stove that burns pellets, a low-cost fuel made from agricultural residues (See Figure 3-1) In order to build and develop the market, the company, relying on its financial resources, initially provided stoves and fuel at prices that could not recover costs. However, after the separation, company A could no longer follow that strategy and had to double the price of fuel in order to reach an appropriate cost recovery level and margin of profit. The initial stove price in 2007 was around 657 to 950 INR, but that has been increased to around 1150 INR. Fuel prices have increased almost twofold, from 6 to 6.5 INR per kg to around 10 INR per kg (Shukla and Bairiganjan 2011). Company A’s business model was based on two different distribution channels. The first was the NGO channel, through which local NGOs distributed and marketed stoves and fuel; the second channel was a retail network through which products were distributed to a series of distributors (wholesalers) and then to Village Level Entrepreneurs (VLEs) and local retailers (Shukla and Bairiganjan 2011). 62  Chapter 3  Figure 3-1: Schematic of a forced-draft biomass gasifier stove and an actual picture of stove X captured during fieldwork  Company B (Electrification) Company B was founded in 2007 by four individuals: two local engineers and two business students in the United States. It provides decentralized electricity to villages with 300-500 households via 35-100 kW mini power plants that usually run 6-8 hours per day (MNRE 2011). The power plants generate electricity using a gen set that runs on producer gas, which is the product of biomass gasification. The schematic of the process plus a picture of a power plant taken during the field visit are shown in Figure 3-2. 63  Chapter 3  Figure 3-2: Schematic of gasification process and a picture of an actual mini power plant in rural Bihar  Currently, 80 of these mini power plants are operating in rural regions of Bihar state. They supply electricity for different connection capacities (i.e. 30 Watts, 100 Watts) at the fixed monthly charge of 100 INR per 30 Watts of power for six hours a day. The price excludes the first-time connection charge of 100 INR as well as the cost of internal wiring and appliances, which cost around 300 INR per household. Company B is based in Patna and operates mainly in Bihar. Its initial capital came from its founders’ personal savings and from the multiple awards it received for its business plan. Later, company B established a strategic alliance with a foundation that is part of a multinational oil company; as part of that relationship, company B receives various forms of support from the foundation (IFMR 2010). Moreover, the company receives about 50% of the capital cost of the power plants it operates and built as a subsidy from the government. Its business plan initially focused on providing low-cost electricity to rural populations through the gasification of rice husks, which used to be a very cheap commodity. Company B’s entrance to a market depends on the potential customers in the area; when there are over 300 clients that are willing to pay the 100 INR initial connection fee, the company surveys households and assesses the watt-hour demands 64  Chapter 3  for the site (MNRE 2011). The company has developed three business models: I) an integrated model through which the company builds, owns, operates, and maintains power plants (called the BOOM model by the company) , II) a split-responsibility model through which the company builds, owns, and maintains power plants, but leaves operations to an entrepreneur who agrees to pay a nominal fee to the company (called the BOM model by the company), and III) a sales model through which the company builds and maintains power plants that are sold to another entity (called the BM model by the company). A limited number of follow-up interviews were done with the members of two other companies. Company C (stove developer) Company C is an ICS manufacturer that was formed in India in 2007 as a part of US-based nonprofit organization. The company’s capital comes from the non-profit organization, which receives funding from a number of foundations. Moreover, the US-based organization has partnered with a foundation associated with a multinational energy company in order to support the design and marketing of its ICS. The stoves are manufactured in China and go through further assembly in India. They are distributed and sold through trade channels comprised of dealers, distributors, VLEs, and NGOs. Company C claims that its stoves consume 50% less fuel and generate 80% less pollution than traditional wood stoves. Although company C initially offered four stove types priced between 570 and 2500 INR, (Shukla and Bairiganjan 2011), currently it provides only two types of stoves at 1,000 and 1,600 INR. So far, the company has sold around 300,000 stoves in India’s southern states, including Karnataka, Tamil Nadu, and Andhra Pradesh. Initially, company C adopted two business models: rural retail and NGO based. The former focused on local multi-brand distributors, dealers, and local retailers. Each of those levels kept a margin of profit and sold the stoves at a fixed price. The latter was based on cooperation with NGOs and non-profit organizations, which distributed the stoves to their clients at subsidized rates.  65  Chapter 3  Company D (Electrification) Company D is also active in providing electricity to villages through biomass-gasification-based power plants. The company initially emerged as a partnership between an Indian energy sector professional and a Swiss energy technology company and was based on a design the partners acquired from the Indian Institute of Science Bangalore. In 1992, the partners founded a technical firm that provided gasifiers and engines. In 1996, company D was founded, and through collaboration with a Swiss non-profit the partners built their first power plant with a capacity of 80 kW. In 2001, they started their first non-captive power plant in a remote unelectrified village in Bihar. Company D has built six power plants based on biomass gasification in villages, technical universities, and industrial areas. Its funding comes from multiple sources, including international organizations and a government agency under a bilateral aid scheme. It is also funded by a foundation associated with a multinational oil firm. It later began a partnership program that was registered under the Clean Development Mechanism (CDM) (IFMR 2010). Company D’s power plants are gasifier-based power plants with a capacity of 30-100 kW. The primary fuel for these gasifiers is wood chips; they can run on a variety of other agricultural residues, but doing so lowers their efficiency. Company D’s business model is based on I) building biomass-based power plants and II) creating and empowering microenterprises to purchase its electricity. The company’s ground operations are taken care of by a local entrepreneur who is in charge of pricing, collection, and daily power plant operations. Company D often uses meters to determine what to charge for electricity, but it also provides households with fixed rates. Further, the company promotes two non-profit organizations aimed at empowering villagers and microenterprises. However, the initial demand generation plan was gradually replaced by another scheme that required power plants to be built in areas with functioning enterprises (IFMR 2010). Currently, company D is running three power plants with capacities of 100, 75, and 32 kW. The two larger plants are providing electricity to the commercial sector and run 10-11 hours a day starting at 8:00 am. The smaller one provides electricity to rural households and runs during the evening.  66  Chapter 3  3.4.3  Area of operation  The companies investigated in my case study operate in different location in India. The following figure indicates the locations these companies were operating (Figure Figure 3-3).  : Bangalore, where I was based in the Center for Study of Science, Technology and Policy (CSTEP). AH: Headquarters of company A in Pune (Capital of Mahrashta state). A: Regions in which company A distributes stov stoves/fuel, es/fuel, including Maharashta, Karnataka, Andhra Pradesh, and Tamilnadu. BH: Headquarters of company B in Patna (capital of Bihar state). B: Rural regions of Bihar state where company B’s power plants are located. CH: Headquarters of company D in Bangalor Bangalore. C: Rural regions of Bihar state where company D’s power plants are located. DH: Headquarters of company C in Bangalore. D: Rural regions of Karnataka, Tamil Nadu, and Andhra Pradesh where company C distributes its products. Figure 3-3: Locations associated with case study  67  Chapter 3  3.4.4  Data collection  Guided by our review of previous studies on business model characterization, we identified a range of factors in order to both investigate the opportunities and challenges these companies are facing and characterize the business models they use to supply agricultural residue and provide energy (Osterwalder and Pigneur 2009, Shrimali, Slaski et al. 2011, Winrock 2011). These data were used to develop a context-appropriate questionnaire, which was then pre-tested and revised in order to identify and develop the indicators relevant to the activities of the companies being examined. These indicators, along with specific questions regarding the activities of both companies, were then used to create semi-structured questionnaires to guide the interviews with the members of these enterprises. The semi-structured interview protocol was designed so that it could be used as guidance for interviews with individuals in different operational and management positions at the businesses being considered. Table 3-2 shows the topics that were covered in the interview protocol. Table 3-2: Topics covered in the interview protocol Topic  Information  Business model  Value propositions, customer segment (target market), distribution channels, customer relationships, revenue streams, key resources, key activities, key partnerships, and cost structure.  Business environment  Market, policy, social and environmental impacts and determinants.  Business operation  Organization, residue supply, product and raw material supply, energy production, power plant or stove.  Raw material specifics  Agricultural residue general information, residue market situation, policies targeting agricultural residue supply.  Energy producing specific  Technology, production system, cost structure.  Since the interviews were aimed at exploring various dimensions of these businesses at different scales, the protocol was used mainly to stimulate conversation with the interviewees. A number of individuals at different organizational levels were interviewed (Table 3-3). Most of the interviews were done in person at the interviewee’s place of work; a few interviews were done  68  Chapter 3  over the phone. Some follow-up interviews were done for further clarifications on the points raised in the first interview. Table 3-3: Interviewee’s position, number of interviews, and method of each of interview Number of Company Interviewee position Method interviews CEO 2 Phone and in-person Head of operation 1 In-person Head of technical 2 In-person Head of distribution in In-person 3 Karnataka Company A Manager of fuel In-person production plant 1 (Daharwad) Manager of fuel In-person production plant 1 (Islampur) CEO 2 Phone and in-person Head of finances 1 In-person Head of training 1 In-person Company B Head of marketing 1 In-person Consultant 1 In-person Power plant manager 1 In-person Company D CEO 2 In-person CEO 1 In-person Company C Head of operation 1 In-person  These interviews were complemented by field observations and short interviews with individuals active in different field roles at the companies. For example, field visits were made to a company B power plant in rural Bihar and to a company A residue supply system and fuel-processing plant in northern Karnataka and along the supply chain. These short interviews and field visits are not included in the table, but their information is used in the paper. Moreover, the information was supplemented by the enterprises’ records. Since most of the interviewees were fluent in English, most of the interviews were done in English. However, a few short interviews done during the field visits were facilitated and translated with the help of enterprise personnel. The interviews then were transcribed, coded, and qualitatively analyzed, after which the key areas discussed by multiple interviewees were 69  Chapter 3  highlighted. The results are described in further detail in the next section; direct quotes from the interviews are included throughout the discussion. 3.5 3.5.1  Results and discussions Private sector based approach  Using interviews with the members of companies A and B, we were able to assess the enterprises’ business sustainability as well as the opportunities and challenges they are facing in India’s rural energy market. In brief, neither of the two companies, nor any of the other interviewed entities, has achieved sustainable business practices based solely on providing energy to rural households. The companies’ initial success was potentially due to the early stage of the specific rural markets that they could reach and supply with their products. Sustaining this growth and expanding to other regions has proven more difficult. Company A Despite company A’s initial success in disseminating ICSs, after it increased the price of both its stove and fuel to a profitable level, demand in the residential segment dropped sharply and the company lost sales in many areas. To solve the situation, the company searched for a market segment with more purchasing power and higher demand for its products. Commercial entities were considered likely buyers since they have to buy LPG at non-subsidized prices and also have sufficient financial resources. Since 2010, company A has introduced a line of commercial-grade stoves that can be used by enterprises such as hotels, caterers, restaurants, and food vendors. The company started with three stove versions priced at 9,950 to 19,950 INR. Fuel was provided at a much higher price than it was to the residential market, 15-16 INR per kg. The company’s success rate in the new market, combined with decreased fuel and stove sales in the residential market, pushed company A to not only abandon its residential market expansion plans, but also to start pulling out of the residential market by reducing the number of its distributors, service centers, and fuel/stove sales. Currently, although the company still sees very high potential in the residential market, it has no plans to go back into that market unless special circumstances guarantee that it will sell stoves and fuel. 70  Chapter 3  Company B Company B now believes that selling only electricity cannot be profitable: “You cannot just sell electricity and be very happy about it, you will have problem” (Company B). As a result, the company has initiated a number of other revenue streams, including product channeling, through which it supplies various products to its customers through its distribution channels, and making incense sticks from husk char. Recently, the company stopped its expansion of electricity provision services (the BOOM model) despite its initial target. It has also faced many difficulties in managing model II (the BOM model) due to issues with operators and an inability to legally enforce contracts. As a result, the company now focuses on the BM model, through which it sells power plants to a third party and provides maintenance services. However, company B continues to operate 80 power plants through which it sells electricity to consumers. Companies C and D Although company D is still operating its three power plants, none of these power plants has been economically sustainable. Currently, company D is in a transition phase; it hopes to attract funders and investors in order to build new power plants based on what it has learned through its experiences. Despite large investments, particularly in awareness campaigns, company C has not yet been profitable and has struggled to recover its costs. Therefore, in recent months, the company has adopted a new business model that is based more on B2B (business to business) than on B2C (business to consumer). The company is seeking partnerships with corporations, plantations, non-profit organizations, and the government and sells its stoves to these organizations through a corporate social responsibility scheme. The stove buyers will then give away or offer the stoves to their workers or members at highly subsidized rates. Although the company’s focus has shifted away from household consumers, it tries to maintain minimum activity in retailing.  71  Chapter 3  Summary of results None of these companies achieved profitability solely by providing energy to rural households. All the companies has been faced sufficient challenges in the rural energy business, and despite the huge potential they see in the residential market, they have been moving away from it and are shifting toward more profitable commercial customers. These challenges will be discussed in further detail in the next section. Even when providing energy to households in rural regions, the companies’ focus was mainly on mid- to high-income households and their service costs were well beyond the reach of lowincome households. For instance, for rural households, the ongoing cost of procuring pellets combined with buying the stove itself makes company A’s service seven to ten times more expensive than using traditional cook stoves that burn all sorts of biomass material (Bairiganjan, Cheung et al. 2010). Affordability of these technologies and the scheme through which these technologies are available to consumers is found to have a significant influence on the success and failure of these energy access provision attempts. Our findings indicate that the “private sector based approach to rural energy provision cannot be a universal solution and is not appropriate for all segments of market,” which is in line with recent literature on the success and shortcomings of the private sector based approach. Even in the case of solar PV in Kenya, where it reaches a large population; there was early donor involvement in establishing the market that mainly targeted the middle-income rural households (Jacobson 2007). Although private sector based approaches hold much promise with regard to addressing energy poverty, they cannot be considered the best universal choice, particularly when applied to projects related to poor populations and projects aimed at behavioral change and health interventions (Bailis, Cowan et al. 2009). There is not a single universal solution for these issues; instead, combinations of different technological and institutional options are often the best approach (Chaurey, Ranganathan et al. 2004). Social enterprises and hybrid models are now being discussed in the literature and put in place in practice. One example is the model being used in Bangladesh for disseminating solar home systems, which is based on a combination of a credit mechanism and market delivery with an institution overseen by the government that handles donor financing and is able to offer 72  Chapter 3  below-market loans while assuring quality. However, further understanding of private sector based approaches is required if well-informed hybrid models are to be designed. Moreover, the success of market-oriented approaches depends on identifying and addressing the challenges enterprises are facing in providing energy in rural regions of developing world. 3.5.2  Challenges faced by the commercial entities  Investigating these companies – the most prominent enterprises active in rural energy provision in India – in conjunction with other cases pinpoints private sector based approaches’ advantages and deficiencies with regard to addressing energy poverty in rural regions of the developing world. None of the companies discussed attained profitability exclusively through the residential market, despite the fact they approached higher-income groups in rural India. The hindrances faced by these companies are discussed in detail in this section. Through a comprehensive literature review, I reviewed the challenges faced by private entities involved in energy provision in developing countries. Building upon our findings, I investigated the factors affecting their business based on qualitative interviews with the members of these enterprises. Some of the factors limiting the success of private sector based approaches A number of studies have examined the businesses involved in the low-income energy market, most of which are located in rural areas. Their results indicate that in general, the issues affecting these businesses fall into four major categories: 1) financial, 2) market, 3) policy, and 4) operational (Table 3-4).  73  Chapter 3  Table 3-4: Factors affecting businesses in low-income energy markets Theme  Topics  Description  Financial constraints  - Inability to raise seed fund (Shrimali, Slaski et al. 2011) - Lack of access to capital through third-party financial institutions (ESMAP 2000, Love and Mylenko 2003, Monroy and Hernandez 2008, Dinh, Mavridis et al. 2010) - Low rates of return that are dissatisfying for investors and require “patient capital” (UNEP 2003) - Potential revenue from carbon finances overlooked (Howells, Jonsson et al.)  Demand side  - Products and services may simply not be affordable for the majority of customers (Slaski and Thurber 2009) - Households cannot afford the initial cost of technologies due to the weak lending market (Fernando 2007, Brew-Hammond) - Lack of banking facilities prevents people from depositing finances in secure, convenient, and appropriate locations (Fernando 2007, World Bank 2010)  Access to lower income  - Commercial entities unable to reach lower-income households (Jacobson 2007)  Business environment  - Unfavorable business environments limit the growth of private enterprises (Dethier, Hirn et al. 2008, Dinh, Mavridis et al. 2010) - Inefficient bureaucracy (Dethier, Hirn et al. 2008, Dinh, Mavridis et al. 2010)  Competition and market regulations  - Lack of competition adversely impacts productivity, business growth and expansion (Bastos and Nasir 2004)  Poor infrastructure  - Poor infrastructure makes business practices prohibitively challenging, especially for SMEs (Dethier, Hirn et al. 2008, Dinh, Mavridis et al. 2010, Howells, Jonsson et al. 2010, Zerriffi 2011)  Lack of information and awareness  - Rural areas lack awareness about the socioeconomic benefits of switching to modern energy services (Baris and Ezzati 2007, Slaski and Thurber 2009, Howells, Jonsson et al. 2010, Mondal, Kamp et al. 2010)  Finances  Market  74  Chapter 3  Theme  Topics  Target customer characteristics  - Target customers often live in remote, rural, low-population-density areas that make distribution systems complicated and costly (Martinot, Ramankutty et al. 2000, Zerriffi 2011) - Rural households’ irregular consumption patterns are challenging for businesses (Zerriffi 2011) - The energy-poor are often also financially poor, so the money they have to invest in new technologies is very limited (Brew-Hammond , Zerriffi 2011) - Cultural and social norms and beliefs that are stronger in rural areas may impede the uptake of modern energy systems (Eberhard 2004) - Prevalence of barter economies in rural regions, which can limit the penetration of new goods and services (Marin and Kaufmann 2000)  Competitive energy sources  - Modern energy services often impose a contentious energy cost on households that rely on common (financially free) resources – such as fuel wood – for their energy needs (Zerriffi 2011)  Intellectual property rights  - Weak patent protection in developing countries disincentivizes technological innovation (Reichman, Rai et al. 2008)  Law enforcement and legal rights  - Weak contracting institutions and weak legal systems lead to an inability to enforce contracts and a lack of property rights (Pande and Udry 2005)  Institutional support  - Lack of institutional support for SMEs in developing countries hinders the potential of these businesses (Bailis, Cowan et al. 2009, Brew-Hammond 2010)  Policy implementation  - State is often not able to implement policies properly (Ribot 2004, Brew-Hammond) - State is downwardly accountable for ensuring the accurate implementation of policies (Ribot 2004)  Market  Policy and institution  Description  75  Chapter 3  Theme  Topics  - Description  Policies  - In many instances, pro-poor policies such as ill-planned subsidies and labor regulations act as barriers for businesses (Martinot, Chaurey et al. 2002, Bhattacharyya and Srivastava 2009, Howells, Jonsson et al. 2010, World Bank 2012) - Many politically-charged barriers need effective policies based on accurate assessments (Radulovic 2005, Monroy and Hernandez 2008, Mondal, Kamp et al. 2010) - Regulatory systems may act as barriers to the development of businesses involved in energy provision (Birner and Martinot 2005, Bhattacharyya and Srivastava 2009) - Lack of coordination between beneficiaries and governments (Monroy and Hernandez 2008) - Long-term stability and consistency in policy is necessary (Foster-Pedley and Hertzog 2006)  Corruption  - Widespread corruption hinders business growth in developing countries (Transparency International 2006, Honorati and Mengistae 2007)  Cultural norms of business  - Strong cultural norms can impede business growth (Lee and Peterson 2000)  Research and development  - R&D is usually underinvested in the private sector (Reichman, Rai et al. 2008, Mondal, Kamp et al.)  Complicated distribution  - Distribution to bottom of the pyramid poses lots of challenges due to factors such as poor infrastructure and low population density (Shukla and Bairiganjan 2011)  Marketing  - Marketing, particularly when it targets behavioral change, is very costly and is often beyond the financial capacity of SMEs in developing countries (Gan and Yu 2008, Bailis, Cowan et al. 2009) - Lack of marketing experience also makes marketing a challenging task for SMEs in developing countries (Schumacher, Frank et al. 2011)  Monitoring and evaluation  - Monitoring and evaluating the social outcomes of energy interventions is very difficult and often beyond the capacity of local NGOs and private enterprises (Smith, Dutta et al. 2007)  Maintenance  - Modern energy technologies require contentious maintenance, which is challenging in rural areas because maintenance is needed for discrete events and does not follow a regular pattern (Zerriffi 2011)  Policy and institution  Operation  76  Chapter 3  Theme  Topics  - Description  Quality assurance and quality control  - Lack of proper monitoring and quality assurance standards allows private entities to try to achieve cost convergence so that they can be competitive and reach low-income households. The market-spoiling effect makes that potentially harmful to businesses’ long-term sustainability (Jackson and Oliver 2000, Bailis, Cowan et al. 2009)  Poor assessment by businesses  - The energy systems proposed by companies may not be well-suited to consumers’ needs, habits, and preferences (Barnes, Openshaw et al. 1993, Slaski and Thurber 2009, Howells, Jonsson et al. 2010)  Operation  77  Chapter 3  Although these issues have been placed in separate categories and are under different headings for the sake of simplicity and comprehension, they are interconnected; often there is a two-way relationship between them. These issues differ in dimension and scale and their importance and priority varies from case to case, but that is not discussed here. 3.5.3  Challenges of enterprises in India  We analyzed our data to gain insight regarding the four key issues pointed out by the literature review as well as the challenges identified through the interviews; the results are discussed below. The information presented here is based on the interviews and field observations. Some quotes are included as examples. 3.5.4  Capital and finances  Financial constraints have been found to be a major obstacle facing businesses in unconventional energy systems (e.g. renewable, ICS) (Zerriffi 2011). Our findings also confirm that accessing capital and sustaining a strong financial flow are two of the major obstacles these businesses are up against. Both companies enjoyed initial capital funding without which they could not have expanded beyond their initial pilot size. They achieved most of their R&D, infrastructure, and organizational development using their seed funds. While company B still enjoys a strategic alliance with the foundation of a multinational oil company that provides it with various types of support, company A is not receiving any other funding in the form of grants. Securing further funding from third-party financial institutions is very challenging for these businesses. “We are still trying to get some bank, so far we don’t have it, because it’s very hard, and they are very tough to deal with” (Company B)” , “The money is not easy to get in this industry...” (Company D). Moreover, providing electricity in rural areas does not provide a satisfying revenue stream. That is perhaps partly due to low profit margins; businesses reported a low margin of profit due to their products and consumers: “You cannot expect to reach huge margin just on electricity sale” (Company B), “We don’t keep much margin right now, because the 78  Chapter 3  product itself is new to the customers and the pricing is a very sensitive issue” (Company C). Weak revenue streams and lack of access to financial institutions are both factors that can limit these companies’ available capital; in turn, that can result in an inability to expand and invest in new revenue streams. Company A reported that lack of capital hindered its efforts to support sustained growth and scale up. “We are short of capital, it’s a new industry, maybe we add infrastructure in time, and there are lots of bottlenecks” (Company A). According to the literature, social impact investors or investors with more modest expectations on rates of return are most likely to invest in these businesses (UNEP 2003). Indeed, the initial investors in these companies were foundations, other grantors, and divisions of large firms that were willing to accept lower returns. Due to low rates of return compared to normal market trends – and therefore a very long payback period – getting access to capital through investors or financial institutions is prohibitively hard for these companies. None of the studied cases were successful in attracting capital from investors or financial institutions such as banks and venture capitals. As mentioned by an interviewee, investors are looking for a higher rate of return: “Everybody wants at least 15% IRR, you can’t give it. It’s a social enterprise “ (Company D). The other major financial issue reported by these companies was revenue stream inconsistencies, which cause unpredictable cash flow. These inconsistencies are seen mainly because most of the companies’ customers rely on agricultural economies that are based on highly variable environmental factors. “These are the months when people are completely broke, a lot of them. And no matter how hard you try your collection cannot be the same.... If there is drought, one very basic thing is people will not have enough to eat, they don’t need electricity” (Company B). This issue can be a major concern in India, where most crops are rain-fed (Sharma, Rao et al. 2010); reliance on the weather leads to uneven earning patterns depending on 79  Chapter 3  harvest time and the quality of the harvest. Other destabilizing factors – such as exchange rate fluctuation, world crude oil, and other political and economic turmoil – were also reported by interviewees as sources of inconsistencies. Considering the fact that these companies have weak revenue streams and lack access to capital from third-party financial institutions, it can be predicted that any revenue stream fluctuations will put huge pressure on their business practices and will make them very susceptible to market fluctuations. Our findings are consistent with the literature and our interviewees pointed to capital and financial constraints as major limiting factors. Accessing patient capital has been identified as necessary for developing businesses in low-income energy markets (UNEP 2003). Our study indicates that financial and capital constraints are one of the major challenges that these businesses are facing. For the interviewees, major concerns included lack of access to capital for improvement, insufficient revenue and low margins of profit, their business being unattractive to investors and financial institutions, inconsistent revenue due to customers’ irregular income, and the instability of the exchange rate and fossil fuel prices. 3.5.5  Policy and institutional issues  Policies and institutional issues were found to influence energy enterprises. Prior research indicates that in some cases, pro-poor policies such as government subsidies can act as a barrier to business expansion (Martinot, Chaurey et al. 2002, Bhattacharyya and Srivastava 2009, Howells, Jonsson et al. 2010, World Bank 2012). Our case study also shows that one of the major obstacles companies A and D faced in dealing with the rural residential market was that the subsidized LPG price for households was too low for a biomass system to be able to compete in a sustainable way. “As long as the [LPG] subsidy there, we won’t be successful in domestic market” (Company A). However, it should be noted that while subsidizing LPG for residential customers hinders the development of the market for ICSs, it provides an incentive for households to adopt cleaner fuel (i.e. LPG). Although the Indian government has considered the removal of the subsidy on petroleum products, petrol prices were only liberalized until mid-2012. 80  Chapter 3  The next step, which was diesel price deregulation, has been put on hold due to a rise in the level of food inflation; as a result, the subsidy has only been slightly decreased. Moreover, since 2009, the government has intensified its efforts to better targeting subsidized LPG in rural areas (IISD 2012). The Indian government’s emphasis on biomass-based electricity generation is reflected in its range of programs promoting biomass gasification (Ravindranath and Rao 2011). As reported by interviewees from company B and company C, the government of India pays a total of 40-50% of the capital cost of biomass-gasification power plants. “There is subsidy on the engine, the gasification part...There is now a subsidy on the distribution too,” (Company D). Other companies that consume agricultural residue, but do not receive a subsidy expressed concerns over this policy. They reported that this policy has already caused a rapid increase in the number of biomass-based power plants in the country, and could increase the demand for agricultural residue. The result of that may be an increase in the price of agricultural residue. “So when you have this type of subsidy on your capital cost so they put up the plant, but they start pulling raw material and then what happens is the price goes up” (Company A). However, this policy, along with license-free rural electrification policies, is one of the main factors that allowed companies B and D to expand to their current level. “That helps us to generate electricity without the need of license 2003 electricity act. If it was not for that we have not been able to do this”(Company B). The policy regime looks encouraging for biomass-gasification power plants; however, company B reported that under the current policy regime, decentralized power plants with a capacity of less than 1 megawatt cannot sell electricity to national grids. That limits their ability to compensate their costs when local demand fluctuates. “The grid doesn’t allow me to sell them less than 1-2 MW. So that cannot be decentralized” (Company D). 81  Chapter 3  That is an example of a policy that directly provides incentives for a target group to expand (i.e. by promoting biomass-gasification-based power plants), but indirectly affects another group (i.e. by increasing the price of the agricultural residue that is used as raw material for pellet making). The literature reports that the policies achieved mixed results due to lack of coordination between the government and beneficiaries (Monroy and Hernandez 2008) The interviewees mentioned that their companies could not invest and plan long-term in part due to uncertainty about government policies and programs. “The natural gas could come into India. That can impact the amount of LPG which get used and that can, in fact, affect our business” (Company A). “the market base approach does not last too long if the energy supply constrained by state, how long will stay, we don’t know” (Company D). The appropriate implementation of these policies, however, was seen as more of a concern. As pointed out by one interviewee, even programs started by the government either weaken as time passes or lose their effectiveness due to a lack of coordination among different institutional scales. “Normally what happens is that the governments schemes are not done with kind of importance that it should be done, it gets diluted at different levels...What is seriousness by the central government may not be the same in Teluka level and Panchaat level” (Company C. Weak law enforcement, institutional structure, and legal support were also mentioned by the interviewees. “There is no way to enforce contracts, there is no legal, legal resources are very limited” (Company B). Lack of institutional and legal support were mentioned as factors that contribute to creating businesses that operate riskily and outside of normal business practices. They also prevent businesses from expanding beyond the local region where they have a strong network.  82  Chapter 3  “I have a contract but nobody cares, so the only way to makes it work is personal relation” (Company C); “The model will work, and there is money, but it has to be very localized” (Company B). Other complicating factors raised by the interviewees include governmental organizations’ non-transparent activities, the influence of lobbies on government programs, and the cooperation of these organizations through their strong government connections. As is indicated by the literature, policies that are poorly planned, implemented, and evaluated hinder the development of businesses in rural energy markets. India is not an exception. As is pointed out by our case study, significant improvements to policy formulation and implementation are vital for creating an enabling business environment. However, some issues raised by the interviewees, such as weak legal and institutional support, require fundamental transformations. 3.5.6  Market  Studies indicate that the rural energy markets where these businesses operate are very challenging. The literature review points out some major issues that commercial entities are facing such as ability to reach to the low income population, difficult business environment to operate, issues with competition and market regulations, poor infrastructure, lack of information and awareness, competitive energy sources and the characteristics of rural habitants. Many of these issues were raised by interviewees and some new dimensions of these problems were revealed. Issues regarding the supply of residue have been pointed out by companies that are using agricultural residue as raw material (i.e. biomass gasification, pellet making). The underdeveloped residue market and supply chain have caused unpredictable fluctuations in both the price and supply of residue. “There is some 15-20 varieties of crops which can be used. Two to five of them have natural supply chain which has been created to some extent” (Company A); “The price of residue goes high and low, mainly because the market is not developed” (Company A). 83  Chapter 3  One interviewee noted that government interference, particularly subsidies (i.e. LPG subsidies and biomass-based electrification incentives), prevents the market from functioning. Additionally, government-initiated monopolies over some residues (i.e. bagasse) have affected these businesses. “Bagasse which is the fairly highly used biomass which is in the hand of cooperative sector not the market which is semi controlled” and its price “Currently it’s not really dictated by the market” (Company A) Information from our interviewees indicates that the agricultural residue market is strongly influenced by international crude oil prices as well as by Indian government policies on importing and distributing fossil fuel. The impact of crude oil prices extends beyond implications for transportation costs, one of the major expenses for all companies. Oil price increases also result in small, local industries switching to biomass sources, and that increases the market price of residue. “Price of residue... it’s generally attached to crude market” (Company B); “What happens is, the moment oil prices go high; smaller industries feel the pressure who uses oil and coal and they try to use biomass and biomass demand goes up.” (Company A). On the supply side, the interviewees pointed out a number of issues associated with the agricultural residue market. The residue market is an undeveloped market that is made less efficient by government interference and the price of commodities (i.e. agricultural residue) does not correctly reflect the supply and demand balance and may fluctuate unexpectedly. The situation is not much better on the demand side (i.e. energy provision). All of the interviewees pointed out the challenges in dealing with an undeveloped, poorly functioning energy provision market; they also noted the heavy financial burden of market development. “Half of our energy goes into educating these people” (Company C) Company A’s initial business plan was to develop the market, but financial constraints ultimately made that impractical.  84  Chapter 3  “…was willing to try develop the market and then make the profit later; so the approach was different, so develop the market for 2-3 years then can get into scale and then profit” (Company A) A part of this market development, as the studied ICS companies mentioned, is creating awareness and marketing, both of which are huge burdens for stove manufacturers. They found that convincing consumers to spend a large portion of their monthly income on a stove that is many times more expensive than the stove they are currently using requires sustained and costly campaigns that are often beyond their capacity. “Our stove is 1000 INR which is 1000 times more expensive than the stove they are currently using” (Company C). “Any company wants to work in improved stove business, needs a very deep pocket to raise the awareness” (Company C). Marketing was also reported to be even more difficult because the consumers these companies are trying to reach are difficult to approach through media. “They don’t watch TV, they don’t read, how do you reach them?” (Company C) Another interesting issue raised through the interviews with company B executives is the lack of competition that the scale of demand has created. As described by an interviewee, the number of villages without access to electricity is so large that even if other companies are providing electricity in an area, there are still so many more villages that require service that there is no need to compete to find customers. “The need is humongous. For instance, if you see a village there is lots of solar lantern why you go there, there is another village that you can go” (Company B) Although it might at first seem like taking over the market would be positive, most of the interviewees expressed concern over the absence of competitors since it means that they are alone in creating and developing markets and distribution networks.  85  Chapter 3  “If you have competition the market actually grows. The whole business of creating market and growing it is on us and that’s a bit of challenge actually” (Company A). “If it comes to that, it is good because the overall awareness will really go up if there are four or five competitors then everybody will looking for major volume” (Company C). The socio-economic characteristics of rural energy market consumers were also pointed out by our interviewees, who noted that they see great potential in domestic markets in rural regions. “It is not profitable business but it got the potential to become one of the largest business segment.”(Company C). However, they have found residential consumers in rural regions to be relatively low in disposable income, which makes for a weak market for domestic energy. “the thing is the value is so different between domestic and commercial; it’s barely less than one dollar for one month family use, unless it will improve a bit it [the business] will not survive”(Company A) Moreover, these commercial entities have not found any incentives to go to the market at the bottom of the pyramid, simply because that group of people could not afford their services anyway. “ two lowest income group people...are so poor that they are not able to afford these stoves right now” right now it’s not possible. That’s the reason why we are not even really trying to make a attempt to get into this particular segment” (Company C). ” I don’t think that people should get electricity for cheaper, should be a free market thing, electricity is not a basic survival thing it’s needed for things that is beyond this”,” we are a commercial entities at the end of it and we need to show money as well” (Company B). Our case studies confirm previous research that indicates that rural energy markets in the developing world – in this case, India – are very challenging. Our research adds another 86  Chapter 3  layer to what is already understood about these complex markets. For example, creating awareness to develop the market and informing consumers appear to be the prohibitive issues for entering these markets. Further, the time and the capital requirements of such activities are often beyond the capacity of private entities. Moreover, although such markets have huge potential (i.e. a large number of households that need access to energy services), the low disposable cash income of rural households makes them unattractive consumers for commercial enterprises. More importantly, commercial entities do not or cannot reach or provide affordable products and services to the BoP population. This issue is very important and needs to be highlighted when discussing the potential of private sector based approaches to energy provision. While the areas these companies are aiming at have huge potential as they contain a large number of energy-poor people, the fact that these people lack disposable income – particularly in the lowest income brackets where disposable income does not exist – makes tapping into this market very challenging. 3.5.7  Operational issues  The operational challenges affecting commercial enterprises in rural energy markets include research and development issues; complicated and difficult marketing and supply; and the ability to provide post-purchase services, such as monitoring, evaluation, poor assessment for product design as well as quality assurance and quality control. All of these issues have been identified by previous studies; however, operational issues are very context specific and may vary from company to company or between different technologies and geographical regions. Therefore, while our findings confirm some of the main operational issues commonly identified by other researchers, they also point out specific challenges that these businesses are facing with regard to technology, services, and their area of operation. Many of the operational challenges mentioned by the companies are directly linked to the scarcity of financial resources. Nevertheless, they are also related to other business-specific issues. For example, the environment a business is operating in might make it uneconomic to deal with regardless of its financial resources.  87  Chapter 3  Issues with securing a sustainable supply of raw material were pointed out by all biomass-based energy enterprises that were interviewed. They found securing such supply very difficult due to: a. The difficulty of collecting small quantities of biomass materials that are spread across large areas. “Trying to makes large-scale biomass sourcing from directly from farmers won’t be very easy” (Company A). b. The climate dependency of residue supply and usability. “During Monsoon it is very difficult the relative humidity itself is 150%. If the relative humidity itself is 150%, then maintaining moisture level below 15% is next to impossible” (Company A). c. Poor infrastructure. “Because due to the rain, there are issues with roads, people may not be able to go to their fields and collect the raw material, there will be flood” (Company A). d. Supply shortages due to government policies. “There is a shortage, I could put it like that, as more plants are coming there is a more pressure on the demand and supply” (Company A). “Boiler based big plants in the area that are more accessible we have competitors, suddenly, 3-4 years ago there were no competitors but now becoming more”(Company B). These businesses, particularly the ICS manufacturers, found that distribution and transportation issues complicate the supply of raw material and the distribution and production of energy carriers (i.e. fuel pellets). “... material handling is the bottle neck,”(Company A) Moreover, our case study indicates that distribution in remote areas in particular requires very complex supply channels and management, both of which are costly. Distribution issues can further complicate companies’ market-building activities; reaching a large customer base was described by one interviewee as: “very tough, because the marketing cost will go really high to actually tap to that amount of customer base” (Company C).  88  Chapter 3  As is pointed out by the interviewee, financial constraints prevent companies from expanding and upgrading their business and infrastructure. Such issues also create some operational challenges, as is highlighted by an interviewee who mentioned the quality of his company’s infrastructure. “Unfortunately it’s [storage] uncovered, and they [residue] will be useless, since the moisture level goes high, we can use partly, we have to use fresh ones, due to infrastructure constraints” (Company A) Our field visits and interviews indicate that operating in India’s rural and remote areas is very challenging due to weak infrastructure, the absence of industry, and the difficulty of attracting skilled and professional personnel. It is also hard to manage and monitor supply and distribution lines as they are spread over a large geographical region. Another major challenge pointed out by an interviewee from company B is the difficulty of acquiring industrial supplies and services in remote rural areas. “We have to create everything, pretty much everything that we need we build ourselves,” “our management, our systems are all made here, we don’t have a luxury of getting it from somewhere else,” (Company B) The interviewee described finding support and supplies industry, as well as hiring professional manpower, as difficult. “People are very unprofessional,” ” we need we build ourselves, from machine to gadget to expertise and man power” (Company B). Difficulty in monitoring business makes management complex and costly, as was indicated by a company B interviewee. “Serious disadvantages that is management overhead. Just, huge, you know … people to manage that scattered in … locations is very hard” (Company B). That is one of the main issues that caused company B to walk away from direct involvement in residential electricity provision. Electricity theft, for instance, is a major challenge that company B tried to address through transferring operations and control  89  Chapter 3  tasks to local individuals. Company D also loses 10-15% of its production to electricity theft. “Usually our capacity is 32, our engine is peaking. The peak of the engine is about 20 kw which is the combination of theft and distribution loss” (Company B) In addition to the aforementioned issues, as a new technology in India, start-up companies with limited experience in such markets may face many challenges. As pointed out by an interviewee from company A: “So when we started we didn’t have any idea what’s the cost of making pellet in India” (Company A). In general, these challenges are often due to weak infrastructure in remote and rural areas, lack of support and access to the service industry, and the issues inherent to operating in large, sparsely populated geographical areas. These issues not only hamper market expansion to distant regions due to the high cost of transportation, but also increase the capital investment that companies must make if they want to build manufacturing and storage facilities or set up sophisticated distribution and transportation channels. 3.6  Conclusions  Our study confirmed our hypothesis regarding the private sector based approach to energy provision and showed that despite the potential advantages of the private sector based approach, it cannot be seen as a universal solution. While it may function for a certain segment of the market, it might not be appropriate for another. It should be mentioned again that these enterprises did not aim to provide universal access. Moreover, even if private entities are addressing the appropriate market segment, they face many challenges in addressing rural energy provision in the developing world. Interviews with members of India’s most prominent commercial entities in this field, combined with field observations and data collection, provide a more accurate view of both energy poverty and strategies for addressing this issue. All of the companies interviewed were not able to build sustainable businesses through the rural residential energy market, despite their best efforts. Although they all see huge market potential in residential rural regions, they have left and stopped expanding their services in such areas, mainly due to the issues discussed 90  Chapter 3  here. Nevertheless, private sector based approaches have many positive aspects that should be integral to strategies aimed at addressing energy poverty. Our findings both from our case study and the literature review point out some common issues. The three main deficiencies that need to be considered in engaging private entities in addressing energy poverty in rural regions of developing countries are: 1. Business environments in the developing world, particularly in the energy sector 2. Challenges inherited through operating in rural and low-income markets in the developing world 3. Motives for addressing the lowest-income brackets Hybrid policies may be the best way to tackle these issues. The governments of most developing countries have failed to create an enabling environment for businesses and their involvement has often been haphazard, poorly planned and implemented in ways that adversely affect the market. That means that in most cases, existing business environments are not facilitating operations. Further, government policies and programs are usually ungrounded and by design often bring mixed results. The implementation of these government programs is even more problematic and deficient. 1. Business environment Our study provides grounding for the fact that structural factors have the highest implications for the business climate, and thus are a large part of what needs to be improved. However, some of these issues, such as corruption and weak institutions, are very hard to address and require a long-term approach. Strengthening institutional framework is vital for any public or private initiatives on residential energy provision. Even so, other options to help businesses overcome these barriers may be applicable. While the business climate needs to be appropriate if businesses are to function and overcome barriers, addressing lower-income households is a different problem and needs different approaches. 2. Operation in rural regions and low-income segments The challenges facing businesses in the developing world are daunting, particularly in rural areas where most energy-poor people reside. Among the many factors that make 91  Chapter 3  operating a business overwhelming are access to capital and financial institutions, low and unstable revenue streams, products and services that require behavioral change, low population density in rural areas, the target market’s dependency on geographical and climatic factors, and lack of a skilled workforce. While some of these issues can only be addressed through innovative and progressive business solutions, government and donor sector involvement in easing some of these barriers is necessary. For example, creating awareness requires capital that is well beyond the capacity of these businesses; thus, the government is the appropriate candidate to take over this task. Also, where the social good is of concern, as in the case of alleviating energy poverty, the government should be involved to some degree in order to ensure that the social good is met (Bailis and Hyman 2011, Kolk and van den Buuse 2012). Moreover, even where viable and cheap solutions are available for commercial entities, often starting and setting up the business requires donor-based and/or patient capital from socially oriented investors who can accept longer payback periods (Kolk and van den Buuse 2012). 3. Motives of private entities Surprisingly, one of the frequently ignored parts of the challenge of tapping the power of the market for alleviating energy poverty is corporations’ inherent desire to profit. Due to their focus on using corporate practices to address the energy market, companies often have no desire to engage in the lower-income market. Past experiences in Kenya’s solar market (Jacobson 2007) and with the insecticide-treated bed nets in Kenya and Nigeria (Kyama and McNeil Jr. 2007) show that the poor population can sometimes not even afford subsidized prices. So, while private sector based models may address the needs of the rural, middle-class population, as Bailis et al. describe, they may also “price poor consumers out of the market” (Bailis, Cowan et al. 2009). It is a place where public and civil needs meet. As is pointed out in the literature, products that have a direct impact on social welfare should not be left to the commercial sector; instead, financial assistance from governments, NGOs, and donors is required (Bailis and Hyman 2011). One further issue that needs to be considered is the sensitivity of working with products that are closely linked to social welfare, particularly public health (Kolk and van den 92  Chapter 3  Buuse 2012). Providing such services not only requires massive awareness campaigns, but also careful monitoring and evaluation, all of which can be beyond private enterprises’ ability to handle (Bailis, Cowan et al. 2009). Corporate culture often dictates the pursuit of profit and can undermine the improvement of social welfare. Even if these enterprises are willing to do so, they may lack the expertise necessary to deal with sensitive issues such as poverty, as was shown by some of the assumptions made by company members during the interviews. “99% of the cases [poor households in rural India] the reasons could be because the guy could be a drunker, or there would be something about it, like the guy is not able to work-unhealthy”(Company B) Addressing these sensitive issues while benefiting from the effectiveness of business approaches could be an optimum solution. The relatively new concept of social enterprise is one promising alternative. However, the fragile situation of households in the lowest income brackets makes it too risky to try out innovative solutions without strong, secure backup plans. Integrated planning is a must for ensuring an appropriate and efficient business climate while providing energy to the poor households that are isolated by businesses. Leaving the whole market to businesses endangers poor households by leaving them to perhaps be neglected altogether, and any solutions aimed at tackling household energy issues require the direct involvement of national governments, whether that involvement be by creating an enabling environment for the private sector or by providing a safety net for the most vulnerable groups.  93  Chapter 4  Chapter 4 Descending the "energy ladder": the case of biomass cook stove uptake by restaurants in Bangalore, India 4.1  Introduction  Adopting modern energy systems, which are often characterized by higher energy density, higher combustion and heat-transfer efficiency, and controllability (Takama, Lambe et al. 2011), is considered a key step in enhancing the living conditions of billions of people in developing countries. Improved cookstoves (ICSs) are promoted as an efficient intermediary solution to traditional cooking energy systems. However, the results of the programs trying to disseminate ICSs have been far from satisfactory thus far. In recent years, private sector based approaches to ICS dissemination have been promoted widely, particularly by major donor organizations. However, those results are also still far from satisfying. As our case study indicates (Chapter 3), despite all the programs aimed at encouraging uptake by residential consumers in India, the only market in which ICSs have been successful in India is in commercial kitchens in major urban centers. Researchers have been studying the reasons for the failure and success of these programs by investigating various stove uptake determinants. These determinants are discussed in detail in Chapter 2 (Kowsari and Zerriffi 2011). 4.1.1. Energy system attributes This study extends the previous literature by focusing on three areas of stove uptake and diffusion that have been under-researched until now. The first is the impact that nonsocio-economic factors have on consumers’ stove choices. Namely, we look at user perceptions of stove attributes in order to understand how different types of users view the positive and negative characteristics of their stoves and to determine which features are key to stove purchasing decisions. The second area we look at is the phenomenon in which users of clean and modern energy systems (e.g. LPG) “switch back” to fuels and technologies that are considered lower on the “energy ladder” – namely biomass-burning ICSs. The final gap we focus on is the predominance in the literature of studies on household energy systems. This thesis addresses the lack of research on commercial 94  Chapter 4  entities’ usage of cooking technologies by studying a different class of users (i.e. restaurants). Similar to other energy system adoption studies, the research on ICSs mainly focuses on socio-economic factors (e.g. adopter income, education, household size, etc.) and stove performance characteristics (e.g. fuels, efficiencies, etc.) (see for example (Heltberg 2005, Farsi, Filippini et al. 2007, Pachauri 2007, Wuyuan, Zerriffi et al. 2008, Pine, Edwards et al. 2011). Insufficient attention has been paid to the human dimension of stove uptake, which holds significant practical importance (Ruiz-Mercado, Masera et al. 2011, Mobarak, Dwivedi et al. 2012). As Johnson et al. describe, the “product-specific”1 energy system attributes that are perceived by users have not received much attention, despite the fact that they may have a strong influence on stove uptake (Johnson and Takama 2012). While Takama admits that the socio-economic determinants that have often been the main focus of the literature are important for market characterization and consumer profiling, the product-specific attributes that emphasize the agency of household consumers in fuel/stove selection are major determinants of product design, demand forecast, and policy formulation. Product-specific factors are also easier to transform than socioeconomic factors (e.g. education, household income, household size), which often are only influenced over a longer timeframe (Takama, Lambe et al. 2011). Moreover, a better understanding of product-specific factors and their strength in different socio-economic groups (e.g. wealth or income groups) provides useful  1  The term “product-specific” is defined by Takama et al. as “product-specific factors are  attributes of the particular stoves and fuels that are potentially available to the households faced with such choices such as fuel price, stove price and they are potentially have different influence across a range of socioeconomic characteristics (Takama et al., 2011). They differ from socio-economic factors in that they are not characteristics related to the individuals or households making stove or fuel choices but rather are specific to the technology. (Takama et al. 2012). 95  Chapter 4  information for stove dissemination programs, which often target particular segments with stoves with a particular characteristics (Johnson and Lambe 2009). 4.1.2. Switching back There is evidence that populations that have already adopted modern energy systems sometimes switch back to traditional energy systems, often due to increases in modern energy prices due to the removal of subsidies (Barnes, Openshaw et al. 1994, Masera, Saatkamp et al. 2000, Pachauri, Spreng et al. 2012); switching back has been reported in Brazil (IEA 2006), Ethiopia (Kassa 2009, cited in (Takama, Tsephel et al. 2012)), Morocco (Elgarah 2011), Senegal (Lallement 2008), and recently in restaurants in India’s major urban centers. Both adopting modern energy systems and switching back are, however, rarely binary processes; often users adopt multiple energy systems, which is called energy stacking (Masera, Saatkamp et al. 2000). As projected by OPEC, the price of oil will rise further in the coming decades, as will shares of LPG, both of which could result in an average increase in LPG prices, though regional variation will be significant (OPEC 2011). Considering the projected LPG prices, it can be expected that the trend of switching back will become more frequent and will be seen in different parts of the world. That may exacerbate the already daunting task of transitioning to modern energy systems in the developing world. However, the literature on this trend is quite limited. Implementation programs and research have exclusively focused on issues around removing barriers and creating incentives for households to replace their traditional stoves with ICSs. However, these programs sometimes create incentives for modern energy users to move to lower-grade energy systems, as was the case in our case study (Chapter 3). Almost all of the commercial kitchens we studied and most of the households studied in a parallel study by Stanford University have replaced their LPG use with stove X. While development and academic communities are still pushing for the removal of barriers to ICS dissemination, their efforts can also promote switching from cleaner, more efficient technology to lower-grade energy systems. 4.1.3. Energy system uptake by commercial entities As almost all studies on ICS adoption focus on the residential sector, commercial entities have been left understudied. Restaurants and food vendors in particular are major 96  Chapter 4  cooking device users and the energy systems they use not only have environmental implications, but also may significantly impact workers and consequently their families. Street food vendors, despite their significant presence, often do not belong to a formal sector and receive hardly any attention from the development and research community (Tedd, Liyanarachchi et al. 2001). Our field study reveals the growing trend of restaurants using ICSs to replace some part of their LPG use (Chapter 3). Chapter 3 shows that ICS companies’ failure to achieve sustainable business in the residential sector has forced them to explore new markets. So far, the market of restaurants in major urban centers has been promising, and our case study indicates that in that market, company A (see Chapter 3) has achieved a growing margin of profit. That is in contrast to its experience in the residential market, where it ran on a net loss. Similar trends have been seen in other enterprises that once focused on the residential market (Chapter 3). While Chapter 3 focuses on the supply side (enterprises attempting to provide a technology/service), this chapter is more focused on the demand side (the purchasing and usage decisions of the commercial customers of company A). We studied the uptake of ICSs by commercial kitchens in Bangalore, India. Bangalore is the fastest-growing metropolis in India and is home to many wellrecognized technical schools and major international IT companies (e.g. Yahoo, Google); it has gained fame as the “Silicon Valley of India” thanks to its leading IT exports (Siskin 2009). The uptake of biomass-burning stoves in this city, which is a major technology hub in India and does not lack access to modern energy services, is an interesting case study since it captures the dimension of energy system choice regardless of access constraints. We further explored some of the less-studied areas of ICS adoption by: 1. Investigating the underlying factors affecting the adoption of ICSs, focusing on the perceived attributes of ICSs (i.e. product-specific factors) as reported by both kitchen owners, who are mainly in charge of purchasing decisions, and the cooks that mainly use the stoves.  97  Chapter 4  2. Examining the uptake factors that owners reported as being the most valued and the most influential on their purchasing decisions. This study not only shed more light on the dynamics of ICS adoption by commercial entities, but also improved our understanding of the switching back trend, which is mainly stimulated by the price of fossil fuel. By taking this approach we were also able to examine two of the main drivers of ICS dissemination programs that were frequently mentioned by the researchers: (a) improvement of health conditions (i.e. stove smokiness and the impact of stoves on health) and (b) fuel savings (i.e. fuel expenditure savings) (Barnes, Openshaw et al. 1993, Masera, Diaz et al. 2005, Rehfuess 2006, Foell, Pachauri et al. 2011, Mobarak, Dwivedi et al. 2012). The Global Alliance for Clean Cookstoves lists these issues as two of their three primary objectives, the third being environmental protection (Global Alliance for Clean Cookstoves 2013). Another issue of concern in many ICS adoption studies is split incentives, which are caused by intra-household decision-making inequalities and the differential impacts of stove changes (particularly along gender lines); they may significantly affect household decisions on energy systems (Wamukonya 2000, Lallement 2008, Mink 2010). While decision-making authority in households may be divided and unequal, it could be argued that divisions and inequality would be even more prevalent in commercial operations where an owner/manager would have clear authority over purchasing decisions while not necessarily bearing the decisions’ non-financial impacts faced by those in the kitchen. Therefore, we also paid attention to the different stove attributes favored by stove users and decision makers. We had access to the results of a very similar survey, conducted in India by Stanford University, on households that had adopted the same type of ICS but the residential version. The results of this survey were also reviewed so that they could be compared with our findings about the commercial sector. 4.2  Case study  Although a fair number of studies have investigated the implications of IAP caused by biomass combustion in households in India, the recent trend of commercial kitchens 98  Chapter 4  switching from solely LPG stoves to ICSs has not been studied. This study focuses on commercial kitchens that have partly switched back from LPG to ICSs in the urban area of Bangalore (the capital of the southeastern state of Karnataka, India). 4.2.1  ICS provision in India  Providing cleaner cooking fuel has been a major concern for a long time in India, and is addressed by government programs aimed at ICS and LPG service expansion (Buragohain, Mahanta et al. 2010). IEA estimates indicate that 800 million people still rely on traditional biomass fuel for cooking; 80% reside in rural regions (IEA 2010). In 1985, India’s Ministry of Non-conventional Energy Sources (MNES) launched the National Programme on Improved Cookstoves (NIPC), a national program that was aimed at improving cook stove dissemination (Kishore and Ramana 2002). The program oversaw the installation of 28 million ICSs, but by 1995-96 only 60% of the cook stoves were in use (Sinha 2002). By 2009, fewer than 6 million households were using the ICSs (Bhattacharya and Jana 2009). The NIPC program was dismantled by 2002, and research indicates that its impact was both far lower than initial goals had anticipated and less than was reported by MNES (Kishore and Ramana 2002). Currently, the Indian government is developing another ICS distribution program. In India, similar to other parts of the developing world, government inefficiency, donor-driven programs, and concerns about long-term replicability and sustainability have caused an increasing interest in the role that the private sector can play in ICS dissemination, and huge market potential has been perceived in rural energy. A report prepared by the Center for Development Finance estimates that 76% of rural households and almost 60% of India’s population are considered to be at the base of the pyramid (BoP) of wealth. These 114 million rural households spend almost 224 billion INR per year on energy, based on 2004/2005 expenditure data (Bairiganjan, Cheung et al. 2010). The potential market value per year for providing BoP rural households with access to ICSs is estimated to be around 1.11 billion INR (24.13 million USD) (Bairiganjan, Cheung et al. 2010). Despite the existing potential, very few commercial entities are active in the provision of energy to rural households, and even fewer are providers for the BoP population. Previous research on commercial entities indicates that they have abandoned or limited their activity in 99  Chapter 4  residential markets and shifted toward the more profitable market of commercial entities (Kowsari, 2012_chapter 3). The largest distribution of ICSs by a private sector company was completed by company A, which has now abandoned its residential market distribution and is focusing on commercial entities in urban centers. Our case study focuses on company A’s ICSs. 4.2.2  The cook stoves  Restaurants in major urban centers in India mostly use LPG as their main cooking fuel. Therefore, two stoves were chosen to be investigated: A) stove X, the ICS made by company A that has recently attracted the interest of restaurant owners and B) the commercial-grade LPG stove most commonly used in commercial kitchens (see Table 4-1). Stove X  LPG Stove  Figure 4-1: Photos of stove X and the LPG stove that was tested  Stove X is a forced-draft gasifier stove that burns biomass pellets and was designed and developed by the Indian Institute of Science, Bangalore. Gasifier stoves have achieved better performance in comparison to normal ICSs because they have better control of combustion. In this kind of stove, biomass materials turn into combustible gases (via the pyrolisis process) controllably in a separate zone, and the resulting “syngas” (wood gas) moves to the top of the stove where it mixes with oxygen and burns. The process results 100  Chapter 4  in a higher flame temperature and more complete combustion by controlling the input heat and air at different stages of combustion and by operating near a stoichiometric airto-fuel ratio (AFR). The wood is converted to char and eventually turns into ash through oxidization (Roth 2011, Just 2012). Stove X is a scaled-up version of the primary smaller design that was intended for residential kitchens. The larger stove was developed based on the same design principles as the smaller one and maintains the same operational AFR to achieve maximum flame temperature without losing efficiency. The stove has two separate fans, one for gasification and one for combustion. However, there are no comprehensive reported emission and performance characteristics for this stove; the only published study so far reports that the commercial-grade stove X can achieve 65% efficiency and a CO emission factor of 0.4 grams per MJ (Varunkumar 2012). As a result, the numbers reported for the residential version can be considered more reliable. LPG stoves are characterized extensively in the literature (Basu, Saha et al. 2008). Although there is a wide range of LPG stoves, due to their relative uniformity in combustion characteristics, fuel composition, and container pressure, it is reasonable to extend the numbers reported in the literature to the commercial-grade LPG stoves that are used in restaurants, keeping in mind that the stove characteristics reported in the literature are often the traits of residential stoves. One key advantage that LPG stoves have over stove X is their controllability: users can stop and start combustion at any time as well as increase or decrease the rate of combustion. 4.3  Data collection  Commercial kitchens in Bangalore, India were chosen as study subjects. The sample chosen for this study consists of two groups of commercial entities: 1. Restaurants that purchased stove X and use a combination of LPG stoves and stove X (in this paper referred to as stove X kitchens). 2. Restaurants that solely use LPG as their cooking fuel (in this paper referred to as LPG kitchens). 3. All the samples located in the urban area of Bangalore.  101  Chapter 4  The commercial kitchens selected varied in size from small restaurants that serve fewer than 100 meals per day to large restaurants that serve over 600 meals per day. The total number of restaurants in Bangalore that use LPG stoves is unknown. However, a random sampling was found using a restaurant database available to the Bangalore marketing firm IMRB. Stove X users were also selected randomly from a customer database that was available through company A’s office in Bangalore. At the time of the study, the total number of stove X buyers in the Bangalore region was around 526. From this number, 286 cases were contacted; of that number, 100 responded (the number of samples required for the survey), so the response rate for stove X users was about 35%. To achieve 100 respondent samples of LPG users, 220 commercial kitchens were contacted, meaning that the response rate was about 45%. The commercial kitchens were a range of establishments, such as restaurants, hotels, and dormitories; however, the majority of respondents were medium-sized restaurants that are called hotels. 4.3.1  Survey  Data on stove X uptake factors, business information, demographic information, and health symptoms were collected through structured interviews and a survey of commercial kitchens over a period of two months in 2011 and 2012. The structured survey questionnaire was completed in Bangalore through consultation with local researchers and a marketing research firm that was contracted for manpower, translation, interviews, and data entry. The questionnaire was then pre-tested, revised, and translated from English to Hindi – and subsequently translated back into English (verbatim responses were requested in very few places and there are very few responses to them in 200 samples). Moreover, during the first few interviews (as part of the survey), the questions were evaluated and some modifications were made to the survey questionnaire. Further, the survey questionnaire was designed in two parts and was given to two respondents from each entity. The questions regarding stove X adoption were only asked of stove X kitchens (see Table 4-1):  102  Chapter 4  •  Part I (kitchen owner/manager1,2 ): Cook stove attributes, motivation for adopting ICSs as well as health perceptions  •  Part II (main stove user3): Experience using ICSs in comparison to LPG stoves, qualitative health measures, and health perceptions. Table 4-1: Survey questions and respondents Questions Kitchens Respondent Kitchen owners and Both kitchens Cook stove attributes stove users Stove X adoption Stove X kitchens Kitchen owners reasons Kitchen owners and Both kitchens Health perceptions stove users Both kitchens Stove users Health symptoms Kitchen owners and Both kitchens Demographics stove users Kitchen owners Business characteristics Both kitchens  Health-related questions were based on the health symptoms reported in the literature and were adopted from the survey questionnaire used by a USAID study on household energy use, indoor air pollution, and health perceptions in the southern Philippines (Saksena, Subida et al. 2005). The health symptoms of exposure to IAP that are frequently reported in the literature are coughing, phlegm, and eye irritation (Berglund, Brunekreef et al. 1992, Ellegård 1997, Chengappa, Edwards et al. 2007, Ludwinski, Moriarty et al. 2011). The binary yes/no response questions asked whether the respondents had experienced these symptoms. The questions on kitchen characteristics were consistent with the literature on ventilation and space characteristics in exposure studies, and were adopted partly from (Saksena, Subida et al. 2005) and partly from a survey conducted by Stanford University on the residential model of stove X adopted by rural households in Karanaka,  1  Kitchen owners (called “owners” in the rest of the text) are owners or managers in charge of purchasing decisions. 2 Among the “owner” respondents, 6.5% were the decision maker, owner and main stove user (i.e. cook); 0.5% were the decision maker and cook, but not the owner; 72.5% were the decision maker and owner, but not the main user; 20.5% of decision makers were neither the owner nor the main user (i.e. managers). 3 Main stove users (called “users” in the rest of the text) are the main operators/users of cookstoves and are often the chef.  103  Chapter 4  India. Commercial kitchen owners were asked questions regarding the type of ventilation in their kitchen, the number of exhaust fans in their kitchen, the dimensions of their kitchen, and the number of meals prepared at their restaurant. Information regarding their finances, business profile, and other enterprises was also collected. The questions were adopted from the Enterprise survey, which was designed and implemented by the World Bank (World Bank 2011). The survey also collected demographic data on each respondent, including age, gender, education level, and smoking habits. 4.4  Results and discussions  The literature also often looks at various user characteristics to assess the drivers of energy system adoption. In this study, we tried to improve our understanding of stove attributes and ICS uptake factors from the user perspective. Moreover, we investigated two more of the major drivers of stove programs – fuel savings and health – from the perspective of users. We used survey results to understand the adoption process and the influence that various stove attributes have on purchasing decisions. As indicated in the literature (Barnes, Openshaw et al. 1993, Baris and Ezzati 2007, Gifford 2010, World Bank 2011), ICS programs often focus mainly on fuel savings and health, so we paid particular attention to these two factors and how they affect enterprises’ decisions. Further, we studied functionality (whether stoves are culturally appropriate and match current cooking practices) and convenience (how easy stoves are to work with) (Takama, Tsephel et al. 2012), among many other factors. Through the qualitative and quantitative analysis of data, we also investigated the issue of split incentives, which we expected to have more influence on commercial entities than on households. Identical questions were asked of households using the residential version of stove X, and those results are presented in Appendix B. In Section 3.1 we investigate both kitchen owners’/managers’ (i.e. decision makers’) and stove users’ (i.e. cooks’) perceptions of the attributes of stove X and LPG stoves. Section 3.2 explores the reasons to buy and not to buy stove X as reported by kitchen owners/managers. Section 3.3 focuses on economic factors that influence the uptake of stove X. Section 3.4 discusses the extent to which kitchen owners’ and stove users’ knowledge and perception of the health risks associated with stove smoke influence 104  Chapter 4  decisions to purchase stove X. Appendix B, which includes the supporting information for Chapter 4, presents the highlights of the Stanford study on residential users of stove X. 4.4.1 Perception of stove attributes A number of stove attributes were investigated by asking respondents to rank the statements presented in Table 4-2: Stove attributes, categories, and short name from 1 to 5 (with 1 indicating “strongly disagree” and 5 indicating “strongly agree”). These statements are grouped by some of the categories of stove uptake that we considered in this study. For each statement, a short abbreviation is used; the abbreviations are mentioned in Table 4-2. Each statement’s average score for both groups of respondents – owners and users – regarding stove X and LPG stoves is presented in Figure 4-2, Figure 4-3. Table 4-2: Stove attributes, categories, and short name Category Convenience Functionality Economic proposition Health concerns Fuel availability Reliability Aesthetic  Statement Simple and convenient to use  Short Simple  Cooks quickly Able to cook most dishes  Cooks quickly Most dishes  Food tastes good  Tastes good  Stove is inexpensive  Stove cheap  Fuel is inexpensive Produces no smoke  Fuel cheap No smoke  Has fuel that is easily available and quick to obtain  Fuel ez to get  Fuel is always available whenever needed Stove is reliable (does not break)  Fuel avail Stove reliable  Does not favorably impress colleagues/friends/community  Impress  LPG stove attributes as seen by stove X and LPG owners and users All stove X kitchens used to be LPG kitchens and all retain at least one LPG stove (i.e. stove X kitchens should be considered stove X/LPG kitchens while LPG kitchens are 100% LPG). Given the universal use of LPG among the sample, we initially investigated the perception of LPG stove attributes among LPG and stove X kitchen users and owners (See Table 4-3). Our results indicate that stove attribute perceptions differ between the two groups. In other words, people who own/use the LPG and stove X combination have 105  Chapter 4  different perceptions of LPG stove attributes than people who only use LPG stoves. That may be due to stove users’ point of reference or their experience using stove X influencing their perception of LPG stoves. Due to the fact that the data are not normally distributed, the standard T-test to compare means is not applicable. Instead, the MannWhitney U (two samples) test was run as it is a non-parametric test for independent samples that does not assume normality of the data. Table 4-3: LPG stove attributes according to users and owners of LPG kitchens  OWNER USER Stove X LPG MW test Stove X LPG MW test kitchen kitchen Sig kitchen kitchen Sig 4.59 4.61 4.64 4.57 .843 .120 Simple (.534) (.510) (.504) (.685) 4.47 4.48 4.58 4.55 .558 .430 Cooks quickly (.577) (.689) (.629) (.543) 4.58 4.52 4.63** 4.38** .634 Most dishes .047 (.606) (.674) (.566) (.786) 4.23* 4.38* 4.27 4.30 .088 .609 Tastes good (.737) (.776) (.643) (.738) 4.55 4.35 4.53* 4.26* .192 No smoke .091 (.609) (.869) (.712) (.998) 4.25** 3.77** 4.25** 3.84** Stove cheap .002 .029 (.770) (1.102) (.838) (1.078) 3.99 3.76 4.01** 3.6** .279 Fuel cheap .020 (.759) (1.102) (.792) (1.134) 4.31 4.18 4.35** 4.15** .113 Fuel ez to get .034 (.787) (.702) (.809) (.744) 4.39 4.16 4.33 4.19 .113 .479 Fuel avail (.695) (.929) (.805) (.988) 4.54** 4.36** 4.40 4.43 .037 .897 Stove reliable (.642) (.674) (.690) (.635) 4.02 3.93 4.03 4.10 .793 .442 Impress (1.146) (1.265) (1.115) (1.155) The level of significance using the Mann-Whitney U (two samples) test is shown in the column “MW test sig”. Attribute mean scores and (standard deviations) are shown. A * in front of the mean number indicates a significant level of mean difference. * = significant at 0.1 level, ** = significant at 0.05 level, and *** = significant at 0.001 level. Attributes  Both owners and users of LPG stoves in different kitchens (i.e. LPG-only kitchens and LPG/stove X kitchens) had different perceptions of some LPG stove attributes. Owners of LPG kitchens agreed less that “food tastes good” on LPG stoves; in stove X kitchens, that may point to perceptions of how stoves impact food taste, which is identified in the literature (IEA 2006, Bailis and Hyman 2011, Mukhopadhyay, Sambandam et al. 2012). 106  Chapter 4  Data indicate that with regard to LPG stoves, owners of stove X kitchens agreed more with “stove is inexpensive” and “stove is reliable” than owners of LPG kitchens did. That may reflect the higher price of stove X in comparison to LPG stoves; it may also be a reflection of maintenance issues with stove X. Stove users in LPG kitchens and stove X kitchens also saw LPG stoves attributes differently. Compared to LPG kitchens, more users in stove X kitchens agreed with “LPG stoves can cook most dishes.” This response may reflect the cooking characteristics of these two stoves; that is, stove X is better at cooking batches, while LPG stoves offer full control of the rate of combustion and can be turned on and off at any time. Users and owners had similar perceptions of the price of LPG stoves, although more stove X/LPG kitchen users than LPG kitchen users agreed that “LPG stoves are inexpensive.” Moreover, those working in LPG kitchens agreed less with “LPG fuel is available” and “LPG fuel is cheap” than LPG/stove X kitchen users. Such responses may point to the fact that the price of pellet fuel and its distribution system, appear cheaper – and superior – in comparison to LPG. As discussed, some of these differences may be due to the presence of stove X and its impact on users’ and owners’ perceptions of LPG stove attributes. As a result of these different perceptions of LPG stove attributes, we focused solely on stove X kitchens in order to compare how owners and users perceive the attributes of LPG stoves as compared to those of stove X. Attributes of stove X and LPG stoves as seen by owners and users of stove X kitchens Our comparative analysis of stove X and LPG stove attributes includes only the answers from stove X kitchens that use both LPG stoves and stove X. The questions asked regarding each stove’s attributes are listed in Table 4-2: Stove attributes, categories, and short name . Respondents were asked to indicate their attitude toward particular attributes by choosing one of the following scores: 1) strongly disagree, 2) somewhat disagree, 3) neither agree nor disagree, 4) somewhat agree, and 5) strongly agree. The attributes of both stoves from the perspective of owners is shown in Figure 4-2. 107  Chapter 4  Figure 4-2: Owners’ perception of stove attributes  The graph above shows that on average, respondents scored LPG stove attributes slightly higher than those of stove X, except in terms of food taste and fuel price, which was expected. There was a strong difference in their views on the attribute “stove produces no smoke.” To have a better understanding of the spread of data, Table 4-4 presents some descriptive statics of the stove attributes as reported by owners.  108  Chapter 4  Simple  Cooks quickly  Most dishes  Tastes good  No smoke  Stove cheap  Fuel cheap  Fuel easy  Fuel avail  Reliable  No impress  Table 4-4: Descriptive statistics of LPG stoves and stove X attributes as reported by owners  N  100 4.59 5.00 5 3 5 4.00 5.00 100  100 4.47 5.00 5 3 5 4.00 5.00 100  100 4.58 5.00 5 3 5 4.00 5.00 100  100 4.58 5.00 5 3 5 4.00 5.00 100  100 4.55 5.00 5 3 5 4.00 5.00 100  100 4.25 4.00 5 2 5 4.00 5.00 98  100 3.99 4.00 4 2 5 4.00 4.75 100  100 4.31 4.00 5 2 5 4.00 5.00 100  100 4.39 4.50 5 2 5 4.00 5.00 100  100 4.54 5.00 5 3 5 4.00 5.00 100  100 4.02 4.00 4 1 5 4.00 5.00 100  Mean  4.14  4.03  4.16  4.16  3.59  4.06  4.31  4.33  4.36  4.10  3.84  Median  4.00  4.00  4.00  4.00  4.00  4.00  5.00  4.00  5.00  4.00  4.00  Mode  4  4  4  4  4  5  5  5  5  5  4  Minimum  1  1  1  1  1  1  1  1  1  1  1  Maximum  5  5  5  5  5  5  5  5  5  5  5  25  4.00  4.00  4.00  4.00  3.00  4.00  4.00  4.00  4.00  4.00  3.00  75  5.00  5.00  5.00  5.00  4.00  5.00  5.00  5.00  5.00  5.00  5.00  LPG stove  Type of Stove N Mean Median Mode Minimum Maximum  Stove X  Percentile  Percentile  25 75  Comparison of the spread of data associated with stove X and LPG stoves indicates that, in general, respondents considered LPG stoves superior to stove X. The median scores for stove attributes associated with LPG stoves are generally higher than those linked to stove X. Also, while for most cases no respondents strongly disagreed with statements, mentioned in the Table 4-2with regard to LPG stoves, that was not the case for stove X. The 25th and 75th percentile scores given to the stove attributes do not show a strong difference, except for “no smoke,” which indicates that the respondents agreed more that LPG stoves produce no smoke. Further statistical analysis is presented below to investigate any significant differences between these two stoves as perceived by owners. It is worth noting that the responses are all above three (neither agree nor disagree). That is, both stoves are viewed favorably and the differences between them appear minimal for most. Stove users were asked the same set of questions; the mean scores of their responses are shown in Figure 4-3.  109  Chapter 4  Figure 4-3: Users’ perception of stove attributes  Overall, Figure 4-3 is very similar to Figure 4-2. Excluding the categories “food tastes good” and “fuel is cheap,” LPG stoves were seen as superior to stove X, but only slightly so in many categories. However, the two graphs show that concern about the smokiness of stove X was more significant among users than owners. Fuel availability and ease of obtaining it are almost the same because both fuels are delivered to kitchens by delivery people and both are fairly available in Bangalore. We can see also that both groups noted that LPG stoves perform better with regard to the attribute “able to cook most dishes,” which could mainly be due to the batch cooking characteristics of stove X. Table 4-5 shows descriptive statistics that indicate the spread of scores given by users to stove attributes associated with LPG stoves and stove X.  110  Chapter 4  Stove X  No impress  95 4.64 5.00  95 4.58 5.00  95 4.63 5.00  95 4.27 4.00  94 4.53 5.00  95 4.20 4.00  95 4.01 4.00  95 4.35 5.00  95 4.33 5.00  95 4.40 5.00  5 3 5 4.00 5.00 95  5 3 5 4.00 5.00 95  5 3 5 4.00 5.00 95  4 3 5 4.00 5.00 95  5 2 5 4.00 5.00 94  4 2 5 4.00 5.00 95  4 2 5 4.00 5.00 95  5 2 5 4.00 5.00 95  5 2 5 4.00 5.00 95  5 2 5 4.00 5.00 95  Stove cheap  Reliable  Most dishes  Fuel avail  N  5 4.00 5.00 95  Fuel easy  Percentile  25 75  Fuel cheap  Maximum  No smoke  95 4.59 5.00 5 3  Tastes good  LPG stove  Type of Stove N Mean Median Mode Minimum  Cooks quickly  Simple  Table 4-5 Descriptive statistics regarding LPG stoves and stove X attributes as reported by users  Mean  4.14  4.19  4.12  4.12  4.47  3.32  4.06  4.31  4.37  4.35  4.29  Median  4.00  4.00  4.00  4.00  5.00  3.00  4.00  5.00  4.00  4.00  4.00  4  4  5  4  5  5  5  5  5  2  2  2  1  1  1  1  3  2  Mode  4  4  Minimum  1  2  Maximum  5 4.00  5  5  5  5  5  5  5  5  5  5  4.00  4.00  4.00  3.00  3.00  4.00  4.00  4.00  4.00  4.00  5.00  5.00  5.00  5.00  4.00  5.00  5.00  5.00  5.00  5.00  5.00  Percentile  25 75  The spread of scores given to stove attributes associated with LPG stoves and stove X confirms the findings based on the mean scores. The fact that the means and medians of scores given to LPG stoves are generally higher than those assigned to stove X indicates that LPG stoves are generally considered superior to stove X. Similar to our findings regarding scores reported by owners, for the most part, more people “strongly disagreed” or “disagreed” with the statements in regard to stove X. As with the owners’ responses, the 25 and 75 percentile scores do not show a strong difference, except for “no smoke,” which indicates that respondents were more in agreement with the idea that LPG stoves produce no smoke. Further statistical analysis was done to point out significant differences between owners’ and users’ perspectives of the attributes of LPG stoves and stove X. Since the questions were asked of the same people, the Wilcoxon matched-pair signed rank test was done; it is a non-parametric test for matched-pair and does not assume normality of data (See Table 4-6). 111  Chapter 4 Table 4-6: Paired sample test of score differences between LPG and stove X users and owners (Wilcoxon matched paired signed-rank test) OWNER USER Stove X Stove X Attributes LPG mean Wilcoxon LPG mean Wilcoxon mean mean (STDEV) Test Sig. (STDEV) Test Sig. (STDEV) (STDEV) 4.14*** 4.59*** 4.19*** 4.64*** .000 .000 Simple .932 .534 .789 .504 4.03*** 4.47*** 4.12*** 4.58*** Cooks .000 .000 quickly .926 .577 .784 .629 4.16*** 4.58*** 4.12*** 4.63*** .000 .000 Most dishes .861 .606 .770 .566 4.35 4.23 4.47** 4.27** .258 .031 Tastes good .833 .737 .712 .643 3.59*** 4.55*** 3.32*** 4.53*** .000 .000 No smoke 1.055 .609 1.013 .712 4.21 4.25 4.12 4.25 .609 .438 Stove cheap 1.313 .770 1.210 .838 4.31** 3.99** 4.31** 4.01** .003 .007 Fuel cheap .759 .907 .912 .792 4.33 4.31 4.37 4.35 Fuel ez to .748 .781 get .805 .787 .459 .809 4.36 4.39 4.35 4.33 .937 .884 Fuel avail .894 .695 .665 .805 4.10*** 4.54*** 4.29 4.40 Stove .001 .308 reliable 1.049 .642 .770 .690 3.84 4.02 3.96 4.03 .147 .394 Impress 1.237 1.146 1.100 1.115 The non-parametric test Wilcoxon matched-pair signed rank (two samples) test was done and the significance level is shown in the table. Mean scores and (standard deviations) for attributes are shown. A * in front of the mean number indicates a significant level of mean difference. * = significant at 0.1 level, ** = significant at 0.05 level, and *** = significant at 0.001 level.  Both kitchen owners and stove users found LPG stoves significantly more simple and convenient to use. They also found that LPG stoves cook quickly, cook most dishes, and produce no smoke. Regarding convenience, the process of starting up and damping char is clearly more cumbersome than using LPG. LPG stoves also potentially have some impact on cooking time. The batch cooking characteristics of stove X may have contributed to the superiority of LPG stoves in the “cooks most dishes” category. The results regarding stove smokiness were also anticipated, considering that LPG stoves are very clean-burning, while stove X emits more smoke despite the fact that in comparison to other ICSs it has better characteristics. Both owners and users consider fuel for stove X cheaper than fuel for LPG stoves, which correlates to the fuel price information, 112  Chapter 4  presented in Figure 4-8. While owners did not see a significant difference in food taste between the two stoves, users noted that stove X makes food taste significantly better. Also, owners ranked LPG stoves as more reliable, while users did not see any difference. That could be because owners have a better understanding of maintenance and are more conferenced due to the fact that they pay for it and potentially arrange for it. In general, LPG stoves were ranked as superior to stove X in most categories, and in some performed significantly better. The mean scores of stove attributes were similar for both users and owners except in the case of “taste of food” and ”stove is reliable.” That makes sense considering users are more familiar with food taste and owners are more aware of stove reliability. In addition to the above mentioned differences, no particular difference in the perception of users and owners can be seen in Table 4-6 Figure 4-2and Figure 4-3, as well as Table 4-6, show how owners and users independently perceive LPG stoves versus stove X. It is also possible to compare how owners and chefs differ in their perceptions of the LPG stove and stove X. The difference between the mean scores given to each attribute by users and owners is shown in Figure 4-4.  113  Chapter 4  Figure 4-4: Comparison of the mean scores of stove X attributes as seen by owners and users  114  Chapter 4  Figure 4-4 indicates that user and owner perceptions of stove attributes are very similar except regarding user perceptions of the smokiness of stove X. In that case, the MannWhitney U (two samples) test found the difference to be significant at a level of p<.05 (P=.04). The Wilcoxon matched-pair signed rank (two samples) test found the difference significant at p<.05 (P=.011). While in general the LPG stove was considered cleaner and more convenient by users and owners, stove X is being purchased and has partly replaced LPG use. The reasons this is happening are investigated in the next section. 4.4.2  Reasons to buy and not to buy  In addition to tracking owners’ and users’ stove attribute perceptions, we also gauged how much various factors influence owners to purchase stove X. We provided 21 reasons to buy stove X and instructed respondents to pick the ones applicable to their decision and give each reason a score between 1 and 20. The questions and their short version that are used in figures and analysis are shown in Table 4-7. Table 4-7: Reasons to buy stove X Statement Simple and convenient to use Cooks quickly Able to cook most dishes Food tastes good Produces no smoke Improves health Fuel is inexpensive Stove is inexpensive Good value for money Fuel is easy and quick to obtain Fuel is available when it is needed Alternative fuels are not easily available Availability of payment installment plan Stove is reliable (does not break) Attractive design/looks good Favorably impress colleagues/friends/community Recommended by colleagues or friends Purchased for backup/emergency use Problems with previous cooking method  Short buy_simple buy_quick buy_dishes buy_taste buy_smoke buy_health buy_fuelcheap buy_stovecheap buy_value buy_fueleztoget buy_fuelavail buy_noaltfuel buy_financing buy_reliable buy_looks buy_Impress buy_recommend buy_backup buy_problem  115  Chapter 4  Four main concerns that we wanted to pay particular attention to were: 1) convenience and functionality, 2) health concerns, 3) economic propositions, and 4) fuel availability. Other reasons are all grouped under the category of “other.” Respondents were asked to provide a score for their chosen reasons; a score of zero was assigned to reasons that they did not mention. Further, scores were adjusted based on repeated measure design;1 their averages are shown in Figure 4-5.  Figure 4-5: The means of adjusted scores given by owners regarding reasons to buy stove X. The columns in the graph show the adjusted means of scores given to each reason by the respondents. The columns are coded by categorizing the reasons to buy as: 1) reasons associated with stove convenience and functionality (orange), 2) health (yellow), 3) economic proposition (purple), 4) fuel factors (cyan), and 5) others (green). The vertical dash line indicates the highest mean score and the horizontal dash line indicates the median of scores.  1  Repeated measure design tries to understand the differences between two conditions by eliminating the systematic differences. It was chosen here to understand the real differences between different attributes by comparing the mean scores given by respondents. This method allows for the elimination of any betweensubject differences (some generally give higher scores, while some generally give lower scores).  116  Chapter 4  The mean of all the adjusted scores is 0.05 and the dotted line is the median, which is 0.022. When a respondent did not choose a reason we considered the score for that reason zero. The figures indicate that the respondents gave the highest scores to economic propositions, including “good value for money,” ”fuel is cheap,” and to a lesser extent, “stove is cheap.” That is consistent with the attributes associated with the stoves. Functionality and convenience were also seen as reasons for purchase, but were ranked significantly lower than fuel price and overall economic value. Furthermore, smokiness and health issues ranked very low. Figure 4-6 shows the frequency with which particular factors were identified as reasons to buy stove X and is aimed at identifying whether the differences in Figure 4-5 are due to just a few strong scores.  Figure 4-6: Frequency with which owners mentioned each reason. The columns in the graph show the adjusted means of scores given to each reason by the respondents. The columns are coded by categorizing the reasons to buy as: 1) reasons associated with stove convenience and functionality (orange), 2) health (yellow), 3) economic proposition (purple), 4) fuel factors (cyan), and 5) others (green). The vertical dash line indicates the highest mean score and the horizontal dash line indicates the median of frequencies.  117  Chapter 4  Figure 4-6 indicates that the frequency with which reasons were mentioned shows a trend almost identical to that of the average scores; most respondents selected “good value for money” as a reason. The mean score for frequency is 30. Similarity in two figures indicates consistency of data and supports the average scores as indicators of owners’ reasons to buy. Comparing these two graphs with the comparison of attributes between stoves shows that although owners generally believe LPG stoves are significantly better in all aspects, they replaced them mainly for economic reasons. Out of 100 LPG-using kitchens, only 29 had heard of stove X, but did not purchase it. Their reasons for not purchasing stove X are shown in Figure 4-7. The figure is based on adjusted values and the questions are the opposite of those asked to owners about their reasons for purchasing the stove. The mean is 0.044 and the median is much lower, but there are a number of factors that appeared to have less influence on purchasing decisions. Owners of LPG kitchens who were aware of stove X did not buy stove X mainly due to “lack of sufficient knowledge of stove.” The next group of reasons for not buying stove X is more concerned with convenience and the stove’s functionality, which is consistent with the stove attributes discussed earlier. Interestingly, stove smokiness ranked relatively high among reasons owners did not buy stove X in comparison to the views of those that did buy the stove, but was ranked much lower than the categories of functionality, convenience, and knowledge of stove X. Stove price was also considered a reason not to buy the stove, but had a relatively low score.  118  Chapter 4  Figure 4-7: Reasons that owners of LPG kitchens aware of stove X did not buy stove X. The columns in the graph show the adjusted means of scores given to each reason by the respondents. The columns are coded by categorizing the reasons to buy as: 1) reasons associated with stove convenience and functionality (orange), 2) health (yellow), 3) economic proposition (purple), 4) fuel factors (cyan), 5) knowledge about stove X (patterned red), and 6) others (green). The vertical dash line indicates the highest mean score and the horizontal dash line indicates the median of scores.  Comparing the two graphs shows that there are a range of purchase reasons that neither group considered influential. In particular, health and some of the “other” stove attributes were not considered important reasons for either purchasing or not purchasing stoves (though smokiness did score somewhat higher for non-purchasers). The three major influences on respondents’ stove purchasing decisions appear to be: a) stove functionality and convenience, b) economic proposition, and c) fuel availability. 4.4.3  Economic propositions  The information we gathered about stove attribute perceptions indicates that users consider biomass-based fuel (i.e. pellets) cheaper than LPG fuel, but do not see stove prices as different. These findings are evident in owners’ decisions about purchasing stove X. The scores associated with economic propositions are significantly greater than 119  Chapter 4  those of all other categories. Table 4-8 shows a very simple calculation of fuel costs for three different types of fuel. The calculation is based on the information on the stove X company website. Based on current market prices for LPG and pellet fuel, which is used in stove X, and information reported in the literature on stove efficiency and the energy content of fuels, a rough estimate of the price of each MJ delivered to the pot has been calculated. Table 4-8: The price of each MJ delivered by LPG stoves and stove X Fuel/Stove  Energy content (MJ/kg)  Efficiency  Price Rs/MJ  100.7 (unsubsidized) 28.5 (subsidized) 5 (2008 launch) Pellet/Stove X 16 51 15.5 (current) Wood 16 20 2.5 (average) [1] Energy content numbers and efficiency are based on the numbers reported by Munkunda et al. (Mukunda, Dasappa et al. 2010) [2] Pellet and LPG prices are based on news articles published in 2012 (Banerjee 2012, Ravi Kumar 2012). [3] Wood prices are as reported in a Stanford presentation (Thurber, Phadke et al. 2012). LPG  45  72.5  These numbers are somewhat simplified because the amount of energy delivered to the pot and overall stove efficiency both vary according to the condition of fuel, fire tending, the type and size of the pot, stoves’ specific characteristics and condition, and many other factors. These efficiencies, however, are reported in the literature as the result of water boiling tests. Based on the prices indicated in Table 4-8, Figure 4-8was constructed to provide a better visual comparison between fuel combinations and each stove’s fuel costs. The subsidized price of LPG and the price of wood fuel are included to create an understanding of the situation in the residential market.  120  Chapter 4  Price of unit of energy delivered to the pot 3.5 3.1 3.0  Rs./MJ  2.5 1.9  2.0 1.5 1.0  0.8  0.9  0.6 0.5 0.0 Pellet_price at 2008  Wood_average LPG_subsidized price price (residential)  Pellet_current price  LPG_market price (commercial)  Figure 4-8: Price of each unit of energy delivered to pot by LPG, stove X pellets, and wood  Respondents’ answers to the questions about stove attributes and their reasons for choosing a particular stove indicate that despite the perception that stove X is inferior to LPG stoves, owners still purchase it, mainly because of the price of fuel and because they believe they are getting “good value for their money.” Stove price is not considered an important factor because stove X and the LPG stove have similar prices and/or at this level, commercial kitchens in general are insensitive to the upfront cost of stoves (i.e. price point). Figure 4-8 shows these ideas clearly. It indicates that when stove X initially came to the residential market, the price of its fuel was fairly low (5 Rs per kg of pellet, or approximately 0.006 Rs/MJ). At that price, the cost of cooking a meal with stove X would not only be cheaper than it would if wood was used as fuel, but would also be cheaper than using subsidized LPG. Stove X adoption increased rapidly during this period. The current pellet price of 15-16 Rs per kg is three times higher than the initial price, and that has resulted in the rapid disadoption of the stoves among residential users. Adoption by commercial users that purchase LPG cylinders at market prices is shown in  121  Chapter 4  the column on the left. The difference between these different fuel price levels is significant and indicates that economic propositions outrank all other concerns. 4.4.4  Health  Health is a major concern that stove programs have increasingly started trying to address. A number of questions were asked to identify respondents’ perception of the health implications of smoke emitted from cook stoves. Respondents were asked to rank their attitude toward the statement “smoke is bad for health” from 1) strongly disagree to 5) strongly agree. Figure 4-9 shows that both owners and users believe smoke is bad for health or is bothersome; most strongly agreed with this statement. Users  Owners  Figure 4-9: Users’ and owners’ perception of the statement “smoke is bad for health”  Our study indicates that users and owners are both aware of the negative impacts of smoke from cook stoves. The question about smoke’s effects on health was asked of users and owners at both stove X kitchens and LPG kitchens, and the Mann-Whitney test for independent samples (i.e. users vs. owners in each type of kitchen as well as users and owners of LPG kitchens vs. users and owners of stove X kitchens) found no significant difference between the mean scores of any of these groups. Similar questions about the potential adverse health implications of stove smoke yielded similar results.  122  Chapter 4  Respondents were also asked whether they have been bothered by the smoke from cook stoves; the results of that question are shown in Table 4-9. No significant difference can be seen between owners and users. Table 4-9: Frequency of owners and users in stove X and LPG kitchens being bothered by stove smoke Bothered by smoke Stove X kitchens LPG kitchens Kitchen owners 17% 9% Stove users 18% 8%  While no significant association was found between being bothered by smoke and being the owner of an LPG or stove X kitchen, the association between kitchen type and whether stove users were bothered by smoke is significant at .05 level (Chi-square test alpha level .036). However, recall from Figure 4-2and Figure 4-3 that users also considered stove X smokier compared to what owners suggested. Yet as noted earlier, despite being aware of the negative health impacts of stove smoke and despite being bothered by smoke, potential health impacts and stove smokiness received the least attention from owners when considering what stove to purchase. That could point to split incentives between users and owners. This issue is highlighted by the response to the question, “why did you not take any action to reduce the smoke in the kitchen?” Figure 4-10 shows users’ and owners’ responses to this question.  123  Chapter 4  Stove X kitchens  LPG kitchens 3% 4% 2%  6% 10%  10%  Kitchen Owner  4% 70% 91%  6.1% 6.1% 27%  Stove  42.9 %  Users 42.9 %  73%  2.0%  Figure 4-10: Reasons stove owners and users do not plan to take any action to reduce smoke in the kitchen  Figure 4-10 clearly shows the split incentive in decision making in the kitchens. In stove X kitchens, 43% of users stated that they are not in charge of deciding whether smoke is reduced, while only 10% of owners (potentially managers) stated the same reason. Similarly, in LPG kitchens, 27% of users stated that they do not reduce the smoke because they are not in charge, while only 4% of owners stated the same. The graph 124  Chapter 4  further points out that stove X kitchens are generally smokier than LPG kitchens. 73% of LPG kitchen users stated that there is no need to reduce smoke because there is no smoke, while 43% of stove X kitchen users stated the same; for owners these numbers came in at 91% and 70% respectively. Chapter 4 presents stove owners’ and users’ perceptions of the importance of stove attributes and reasons to buy stoves, placing particular focus on economic propositions and health concerns in regard to commercial entities. The highlights of a similar Stanford University study on the adoption and disadoption of stove X by residential consumers is presented in Appendix B for comparison with our findings in the commercial sector. 4.5  Conclusions  This case study explores the attitudes towards the attributes of LPG stoves and stove X as reported by users and owners in both LPG and LPG/stove X kitchens. Further, to identify the prominent factors influencing purchasing decisions, kitchen owners were asked about their reasons for purchasing stove X. Some highlights of our finding are summarized in Table 4-10. Table 4-10: A summary of the findings LPG/Stove X Kitchen Owners  LPG/Stove X Kitchen Users • View LPG stoves as superior in most categories, particularly in that they produce no smoke • Believe food tastes better on stove X  Stove Attributes  • View LPG stoves as superior in most categories, particularly in that they produce no smoke • Do not view stoves as being as smoky as users see them  Purchasing Decision  • Rank economic proposition highest in making stove purchasing decisions  N/A  Health Concerns  • Although aware of the adverse health outcomes of smoke, neither health concern nor stove smokiness is found influential in their purchasing decision • See little need to improve kitchen conditions  • Aware of the adverse health outcomes of stove smoke • Lack of agency partially responsible for not improving kitchen conditions  125  Chapter 4  Our study indicates that, similar to findings in the literature (Pohekar, Kumar et al. 2005), both stove users and owners in commercial kitchens still adopt stove X – mainly because of fuel price differences – even though they clearly prefer almost all the attributes of LPG stoves. It also shows that businesses are very sensitive to fuel price and that as a result, economic factors are more important than all other attributes. However, the extent to which price differences result in the prioritization of economic propositions cannot be concluded from this study. Economic propositions, stove functionality, and ease of use are the top three priorities of kitchen owners when purchasing stoves. A similar trend can be seen in the residential sector: “adoption in rural areas appears based on economic proposition” (Thurber, Phadke et al. 2012). The household survey results indicate that convenience also ranks very high in importance. That may mainly be due to responses from households that are switching from traditional stoves to ICSs, a trend not present in the commercial sector we studied. Our results suggest that different external driving forces (e.g. fuel pricing) affect consumer perceptions of attributes and trade-offs, meaning that contrary to what socioeconomic determinants might predict, other factors play a role in consumers’ stove uptake decisions. In contrast to our findings, Gupta et al.’s study on the socio-economic attributes of stove uptake indicates that fuel price is not a significant factor (Gupta and Kohlin 2006). The subsidy on modern fuels is still seen as necessary by many, though its failure to address the right target group is a subject of debate. However, removing it might be an incentive for users to at least partially give up their use of modern fuels. Stove X was seen as smokier than LPG stoves by both owners and users. However, the stove is still being purchased, and smokiness – along with health concerns – ranks fairly low in buyers’ reasons for purchasing stove X. Moreover, questions regarding health perceptions indicate that most respondents are well aware of the health issues associated with stove emissions and are also bothered by smoke from the stove. That clearly indicates that health outcomes and stove smokiness are not considered major concerns; instead, many other factors are seen as more important by both residential and commercial users. This conclusion is consistent with recent findings in the literature (Gupta and Kohlin 2006, Pine, Edwards et al. 2011, Lambe and Atteridge 2012, 126  Chapter 4  Mobarak, Dwivedi et al. 2012) and it has important implications for current programs and policy designs that focus on informing stove users about the potential health risks associated with indoor air pollution. The focus on educating users about the issues that matter most will likely increase the chance of ICS adoption by users. In the reverse trend, as seen in commercial kitchens, health information has not convinced people not to adopt dirtier fuel.  127  Chapter 5  Chapter 5 Potential health implications of switching from LPG to improved biomass cook stoves: the case of commercial kitchens in Bangalore, India 5.1  Introduction  Improved cookstoves (ICSs) have been promoted as an intermediate solution that can immediately and significantly impact the efficiency and pollution of the energy systems adopted by households (Venkataraman, Sagar et al. 2010, Foell, Pachauri et al. 2011) (see Introduction). Programs promoting ICSs have been motivated by concerns over various issues, including deforestation, stove user health, and global warming (Arnold, Kohlin et al. 2003) (Smith, Mehta et al. 2004, Bailis, Cowan et al. 2009) (Zhang, Smith et al. 2000, Ezzati, Bailis et al. 2004, Smith and Haigler 2008). Despite decades of attempts to provide clean cooking energy to households including ICS dissemination projects, the results have been far from satisfactory thus far. Private sector based approach to ICS dissemination has been promoted in the recent years. However, despite the promises of this type of approach, its results have yet to be determined; there is some evidence that private sector based approaches to ICS dissemination have failed (Bailis, Cowan et al. 2009), as happened with the commercialization of ICSs in India (Chapter 3). Our previous study indicates that due to the wide range of challenges these businesses are facing, they are currently shifting from the residential market toward more profitable markets, namely other commercial entities. Our field study identifies the growing trend of commercial kitchens (restaurants) using ICSs to replace some part of their LPG use (Chapter 4). The uptake of energy systems considered lower on the “energy ladder” may potentially have socio-economic impacts on stove users and other employees working in commercial kitchens; that is the focus of this final study. A fair number of studies have investigated the various dimensions of indoor air pollution (IAP) and cook stoves in the residential sector. However, issues associated with IAP in occupational environments – for example, its implications for workers’ health – have not received sufficient attention despite the potentially significant impact that IAP may have on ICS-using workers. Restaurants, where the cook stoves are 128  Chapter 5  being used most intensely, are one occupational environment that has not been investigated by IAP researchers. The adoption of ICSs by restaurants in urban centers in India has the potential to increase the IAP in kitchens, meaning that workers would be exposed to various hazardous materials, primarily biomass smoke. This study, which covers ICS uptake by restaurants in Bangalore, India, is one of the first steps toward assessing the potential human health impacts of cook stove use in commercial kitchens. The first part of this study uses direct emission measurements to compare the emission characteristics of stove X, an ICS model that is being adopted by restaurants, with those of the conventional commercial-grade LPG stoves it is displacing. The second part investigates user perceptions of kitchen smokiness and compares the responses of LPGonly kitchen users to those of LPG/ICS users. The third part uses statistical analysis, including regression analysis, to analyze factors relevant to the self-reported health symptoms of workers exposed to IAP, including stove technology. 5.2  Background: IAP health implications, stove emissions, and exposure assessment  Burning fuels, particularly solid fuels, in poorly ventilated environments (i.e. indoors without a chimney) causes a high level of air pollution to which stove users or other people working in the same environment are most heavily exposed. Biomass stove users in the developing world are usually exposed to very high levels of IAP caused by the combustion of biomass material, particularly when cooking is taking place (Bruce, PerezPadilla et al. 2000, Cynthia, Edwards et al. 2008). Exposure to these pollutants has been found to be associated with significant mortality and morbidity rates; it also correlates with a range of adverse health impacts (Naeher, Brauer et al. 2007, Torres-Duque, Maldonado et al. 2008, WHO 2010, Kim, Jahan et al. 2011). The most recent Global Burden of Diseases (GBD 2010) study estimates that air pollution caused by cooking fuel causes 4 million direct premature deaths annually. Of that amount, 3.5 million direct premature deaths a year are due directly to household air pollution (HAP), which is referred to as IAP, while 0.5 million are caused by outdoor air pollution (OAP) due to cooking fuel combustion (Lim, Vos et al. 2012, Smith 2012). Child pneumonia (ALRI)  129  Chapter 5  deaths account for 0.5 million of those deaths, and the rest are adult deaths from lung cancer, COPD, and cardiovascular diseases (CVD) (Smith 2012). 5.2.1  Health implications of exposure to biomass combustion  Research indicates that a range of ailments are associated with IAP, including upper and lower respiratory infections and the subgroups acute respiratory infection (ARI), acute lower respiratory infection (ALRI) (Bruce, Perez-Padilla et al. 2000, Smith, Samet et al. 2000, Ezzati and Kammen 2001, Mishra 2003), chronic obstructive pulmonary disease (COPD) (Pandey 1984, Albalak, Frisancho et al. 1999, Chapman, He et al. 2005, Idolor, De Guia et al. 2011), tuberculosis (Pérez-Padilla, Pérez-Guzmán et al. 2001, Mishra 2003, Pokhrel, Bates et al. 2011), lung cancer (Hernández-Garduño, Brauer et al. 2004, Straif, Baan et al. 2006, Engels, Shen et al. 2009), and asthma (Mishra 2003, Arbex, Martins et al. 2007). Non-respiratory illnesses, such as low birth weight, infant mortality (Boy, Bruce et al. 2002, Mishra, Dai et al. 2004, Rinne, Rodas et al. 2007), cardiovascular disease (Ray, Basu et al. 2003, Pope, Burnett et al. 2004, McCracken, Smith et al. 2007), cataracts and eye diseases (Mishra, Retherford et al. 1999, Saha, Kulkarni et al. 2005, Rozanova, Heilig et al. 2009), and nasopharyngeal and laryngeal cancer (Franco, Kowalski et al. 1989, Pintos, Franco et al. 1998, Zhong, Goldberg et al. 1999), are also linked with IAP. Although there is enough evidence to indicate the strong links between high exposure to IAP and adverse health outcomes, many health studies do not measure subjects’ level of exposure to IAP (Bruce, Perez-Padilla et al. 2000{Cynthia, 2008 #476)} . Exposure level is defined as the “concentration of pollution in immediate breathing environment during a specific period of time” (Bruce, Perez-Padilla et al. 2000). That is, ventilation and space characteristics as well as the function of, type, and source of emission; the duration of exposure to emissions and individual behavior are also taken into consideration. 5.2.2  Biomass combustion  Biomass fuels are mainly lignocellulosic biomass (biomass from plants), which contains cellulose, hemi celluloses, lignin, and extractives; of these, celloluse is the major dry component (Carroll and Finnan 2012). Through combustion, cellulose pyrolyzes and the resulting volatile gases oxidize and release heat. In the meantime, carbon in charwood 130  Chapter 5  oxidizes with O2 and produces CO2 and H2O. Although the complete combustion of biomass materials should result only in water and carbon dioxide, incomplete combustion and the existence of non-organic material in the fuel cause a large number of products of incomplete combustion (PICs) to be emitted from stoves (Smith, Uma et al. 2000). Biomass smoke contains a wide range of solid, liquid, and gaseous compounds that react and transform – sometimes quickly – due to time, temperature, sunlight, other emitted substances, water vapor, etc. (Naeher, Brauer et al. 2007). The main gaseous compounds are carbon dioxide (primarily a concern as a major greenhouse gas (GHG)); carbon monoxide (the main PIC in biomass combustion by volume); and other by-products such as methane, formaldehyde, benzene, nitrogen oxide, and ozone (of concern for their climate impacts and/or health impacts). Importantly, biomass smoke also includes particulate matter (PM), which is grouped based on particle size: total suspended particles (TSP) and its sub-measurements, PM10 (particles smaller than 10 µm) and PM2.5 (the particles smaller than 2.5 µm), are the most common measurements used for reporting PM (De Koning, Smith et al. 1985, Just 2012). PICs are produced due to incomplete combustion, the presence of certain chemicals in fuel (e.g. chlorinated organics produced when chlorine exists in the biomass), the condensation of combustion gases, the entrainment of biomass material and ash fragments (PM), and chemical reactions beyond combustion (e.g. when the oxidization of nitrogen in a high temperature yields nitrogen oxides (NOx)) (Naeher, Brauer et al. 2007). Recently, particular attention has been paid to PM due to increasing evidence of its significant adverse impacts on both health and global warming (Bond, Streets et al. 2004, Roden, Bond et al. 2006, Bond, Doherty et al. 2013). While many substances emitted from stove combustion are considered hazardous to human health (Naeher, Brauer et al. 2007), carbon monoxide and PM are widely considered to be two of the major health-damaging pollutants emitted from the combustion of biomass materials and are the ones commonly measured in studies of cookstoves (Edwards, Smith et al. 2003). ICSs were developed to improve the combustion efficiency and emission characteristics of traditional stoves. A growing number of studies have attempted to characterize ICS emissions and the performance of ICSs (Smith, Uma et al. 2000, Zhang, Smith et al. 2000, Smith, Mehta et al. 2004, Jetter and Kariher 2009). MacCarty et al. (2008) 131  Chapter 5  investigated the performance of different type of ICSs; they assessed the emissions and performance of three stone fires, rocket stoves, fan-assisted stoves, gasifier stoves, and charcoal stoves, and found that fan-assisted stoves significantly outperform all the others. Although studies indicate that ICSs mostly perform better and have better emission characteristics compared to their traditional counterparts, ICSs are outperformed by stoves that run on liquid and gaseous fuels (i.e. LPG, NG, kerosene, biogas), not only with regard to thermal efficiency, but also in terms of PIC and CO2 emissions per unit of energy delivered (Smith, Uma et al. 2000, Zhang, Smith et al. 2000, Smith, Mehta et al. 2004, Jetter and Kariher 2009). However, most studies on stove characteristics have been done in laboratories under controlled conditions, and researchers have indicated that these characteristics are significantly different than those that would be encountered in realworld situations (Roden, Bond et al. 2006, Bailis, Berrueta et al. 2007, MacCarty, Ogle et al. 2008, Roden, Bond et al. 2009, Johnson, Lam et al. 2011, Chen, Roden et al. 2012). Roden et al. (2009), report that when actual cooking is taking place, the PM emissions of improved biofuel cook stoves are, on average, three times greater than those measured in kitchen simulated tests in the laboratory. While the emission characteristics and performance of the stove/fuel combinations seen in laboratories may reduce fuel consumption and thus satisfy environmental concerns, the health implications of ICSs are mainly determined by the exposure of individuals to harmful emissions in real-life situations. 5.2.3  Exposure assessment  A fair number of studies assess the emission and performance characteristics of stoves; however, the relationship between stove characteristics and pollutant concentration has not been well investigated (Bailis, Berrueta et al. 2007, Johnson, Bond et al. 2011). Many studies show that lower pollutant concentrations can be achieved with the adoption of cleaner fuel/stove combinations (Smith, Mehta et al. 2004, Chengappa, Edwards et al. 2007) and/or by enhancing ventilation (Saksena, Thompson et al. 2003, Chengappa, Edwards et al. 2007). Yet further field studies and pollutant concentration measurements are necessary to assess the success of ICS programs in addressing health issues (Edwards, Smith et al. 2003) (Smith, Dutta et al. 2007). LPG stoves have frequently been cited as having the best emission and performance characteristics in both stove/fuel assessments 132  Chapter 5  and exposure studies (Kandpal, Maheshwari et al. 1995). Direct measurement of emission concentration is feasible and has been done by many researchers (Agarwal 1986, Pandey, Neupane et al. 1990, Raiyani, Shah et al. 1993, Armendariz-Arnez, Edwards et al. 2010, Johnson, Lam et al. 2011), but its implementation on a large scale is costly and difficult. Researchers have used other methods to estimate IAP, such as surveying the health symptoms caused by IAP exposure. Studies indicate that some health symptoms are closely associated with exposure to IAP, particularly eye symptoms – such as red eye, eye irritation, and eye watering (Berglund, Brunekreef et al. 1992, Ellegård 1997, Zhong, Goldberg et al. 1999, Chengappa, Edwards et al. 2007) – and respiratory symptoms, including coughing, breathing problems, and trouble with phlegm (Berglund, Brunekreef et al. 1992, Ludwinski, Moriarty et al. 2011). 5.3  Data collection  This study addresses one of the understudied areas in the literature on IAP in commercial entities and explores the increase in exposure to health-damaging pollutants that may be occurring due to the combustion of biomass in commercial kitchens as LPG stoves are replaced with biomass-burning ICSs. It compares the emission characteristics of LPG stoves to those of a clean-burning ICS and looks at self-reported IAP-related symptoms in kitchens that use only LPG stoves versus kitchens that use both ICSs and LPG stoves. Building upon the information collected through emission measurements and the survey, the potential health implications of partly replacing LPG stoves with ICSs in commercial kitchens are investigated. A survey of kitchen owners and main stove users was used to assess 1) perceptions of kitchen smokiness and 2) self-reported health symptoms associated with exposure to IAP. The data collection and sampling method is described in Chapter 4. Stove emissions were estimated through the measurement of direct emissions from one type of ICS and the prevalent type of LPG stove. 5.3.1  Direct emission measurement  The direct emissions of stove X and a common model of a commercial-grade LPG stove were measured. Measurements were made of the three pollutants, CO, CO2, and PM2.5, that are frequently considered the most concerning substances emitted from stoves in 133  Chapter 5  terms of both health and global warming (Edwards, Smith et al. 2003). Three tests were done to measure stove X’s emissions during one full combustion cycle, which started with the ignition of the stove and lasted until the end of the smoldering phase of combustion (see Chapter 4 for a description of both stoves and their operational characteristics). Similarly, tests were done three times with the LPG stove for a period of one hour. The shorter period of time for LPG stove measurements was due to the consistency of observed data and the uniformity of stove performance (unlike the ICS, the LPG stove’s burning characteristics are consistent over time and there is little or no ignition or smoldering phase). The LPG stove was a commercial-grade stove that was rented from an event equipment rental agent in the area and stove X was provided by the distributor of the stove. Background emissions were recorded for 10 minutes before and after each test, and were later deducted from emissions recorded during the actual combustion. The equipment was set up at company A’s warehouse in Bangalore, and research assistants from Bangalore and Canada helped with the measurement process. The equipment used for emissions measurement was developed and provided by Dr. Andrew Grieshop of North Carolina State University and is called the Stove Emission Measurement System (STEMS) and is able to collect the following information: -  Real-time (1-2s) CO2, CO, and PM light scattering and absorption at three wavelengths o  Carbon monoxide (through an electrochemical cell with a range of 0-1000 ppm and a resolution of 1 ppm)  o Carbon dioxide (through non-dispersive infrared (NDIR) with a range of 0-10,000 ppm and a resolution of 2 ppm) o Real-time PM (PM 2.5) (through a red laser-scattering photometer with a range of 0-60,000 µg/m3) -  Flow rate and temperature in exhaust duct  -  Total PM2.5 mass (through a Teflon filter that is weighed before and after measurement)  -  Organic/Elemental Carbon ratio (through a quartz filter) 134  Chapter 5  An in-plume, eight-armed probe was used to collect exhaust from the plume that was located 1.5 meters above the stove. CO and CO2 were measured in real time. Polytetrafluoroethylene PTFE filters (also called Teflon) were used to collect samples for offline determination of PM mass and quartz filters were used to determine the partitioning of PM between the organic and elemental phases (OC/EC). The Teflon filters were weighed before and after the tests in the Occupational and Environmental Health Laboratory (OEH) at UBC and the results are presented below. Quartz filter measurements are still pending and, therefore, we cannot report specifically on black carbon emissions. 5.4  Results and discussion  As indicated through the literature review above, a full exploration of the causal chain linking stove combustion to health outcomes would characterize stove combustion performance, the resulting emissions and indoor air quality levels, the exposure of relevant populations, and the resulting health impacts. Such a comprehensive study design was beyond the scope of this thesis; its focus is instead on characterizing emissions and on looking at indicators of potential health outcomes. In this chapter, we bring together three lines of evidence regarding the potential health impacts of switching from LPG stoves back to biomass-fueled stoves: 1) Secondary data on stove emissions: As discussed above, residential LPG stoves and some ICSs are well characterized in the literature, at least under laboratory settings. We review the existing data on both LPG stoves and the residential version of stove X, for which some measurements are reported, and discuss their relative performance in terms of the pollutants of interest. 2) Direct emission measurements: The fact that the stoves examined in this study are designed for commercial kitchens and are larger may result in some differences between the residential measurements reported in the literature and the emissions from these stoves. Therefore, the direct emission measurements of a typical commercial-grade LPG stove and of stove X were taken and compared to the literature.  135  Chapter 5  3) Smokiness perception among users: While it was not possible to directly measure indoor air quality in commercial kitchens, our survey did ask users about their perceptions of smokiness. We used this as an indicator of possible differences in air quality between LPG and stove X kitchens. 4) Self-reported health outcomes: The switch to stove X in commercial kitchens is relatively recent and we did not necessarily expect to see some of the associated health outcomes among users. Furthermore, it was beyond the scope of this study to conduct direct health measurements. Instead survey questions asked whether they were experiencing a few key health indicators associated with increased smoke, namely eye irritation and coughing. Both are more immediate outcomes of smoke exposure and could indicate the potential for other health outcomes to occur. Each of these four lines of evidence is discussed below. 5.4.1  Secondary data on cook stove characteristics  The results of some of the studies on stove X and LPG stove characteristics are reported in Table 5-1. The few reports on stove X’s emission measurements were collected, and its emission factor (g/MJ of fuel) was calculated based on an energy content of 16 MJ/kg for the pellets that it uses as fuel (shown in the first part of Table 5-1). The first three rows in Table 5-1 are based on in-lab measurements reported by the company and/or affiliates. The emission factor reported by company A’s website is very low. Varunkumar (Varunkumar 2012) identifies two modes of combustion in stove X: I) flaming mode and II) char mode (smoldering); the study found that CO emissions are higher during char mode (smoldering), meaning that the emission results shown on company A’s website may correspond to the flaming mode. Just (Just 2012), tested the residential version of stove X with pellets sourced in Canada.  136  Chapter 5 Table 5-1: CO, CO2, PM emissions,1 and water boiling (WB) efficiency for stove X and LPG stoves, as reported in the literature. Stove/Fuel  CO g/MJ fuel  CO2 g/MJ fuel  PM mg/MJ fuel  WB efficiency  Stove X Residential/Pellet2  0.95  N.R3  N.R  45%  N.R  51.7%  (Varunkumar 2012)  Stove X Residential/Pellet  1.1  4  78.36  Stove X Residential wt Pellet  1.2  N.R.  1.2  64%  0.5  N.R.  N.R.  50-59%  1.35  91.77  4.6  N.R.  (Just 2012)  3.13  103.81  120  N.R.  (Johnson, Bond et al. 2011)  0.55  71.68  2.25  70%  (Smith, Khalil et al. 1993, Smith 1994)  0.05  63.1  10.7  45.2%  (Zhang, Smith et al. 2000)  0.38  62.45  0.21  42.13%  (Zhang, Smith et al. 2000)  0.33  67.36  11.2  53.6%  (Smith, Uma et al. 2000)  Stove X Commercial wt pellet7 Stove X with Canadian-made pellet  13  Stove X Residential/Pellet (in-home) LPG9,10 LPG/ traditional stove11 LPG/ stove with infrared head LPG/Stove combination  12  11  8  5  Reference  (Mukunda, Dasappa et al. 2010) Bryden and Taylor 20086 Company A website  [1] This calculation was made based on fuel energy content. Where applied, stove efficiency was not considered in transformation. [2] Pellet biomass: 16MJ/Kg. Emissions were based on input energy (fuel energy content). [3] Not reported. [4] The best emission achieved under a different stove design. [5] This number was calculated based on the reported number of 815 g of CO2 per 650 grams of fuel (16MJ/KG). [6] Reported in (Mukunda, Dasappa et al. 2010). [7] Company A’s website mentions 0.5 mg/MJ. The CO/CO2 ratio is 0.04, most likely volumetric, and PM is reported as TSP equal to 2 mg/m3. [8] The study consisted of an uncontrolled kitchen test done in homes from the plume during normal daily cooking events. There were 18 samples, PM4.0 was measured, and emission factors were based on carbon balance. [9] LPG energy content: 45.5 MJ/Kg (WHO 2010). [10] The study was an uncontrolled kitchen test done in homes directly above the stove during normal daily cooking events. There were two samples, PM was measured in RSP Respirable suspended particulates considered to be 75% carbon, and emission factors were based on carbon balance. [11] Stove combinations from China were tested in a simulated village kitchen house in a lab. Three tests were done on each fuel/stove combination. The LPG used in China consists of 26.7% propane, 18.8% butane, 43.4% butane, and 11.1% other hydrocarbons. LPG energy content is reported as 49 MJ/kg, PM is reported as Total Suspended Particulate. “Infrared head” refers to a circular device located around the burner under the pot that is designed to transform some of the waste heat into infrared radiation that in turn heats up the pot. [12] Stove combinations from India were tested in a simulated village kitchen house in a lab. Three tests were done on each fuel/stove combination. The LPG in India that is distributed by Indian Oil Corporation and Bharat Petroleum consists of about 80% butane and 20% propane. LPG energy content is reported as 45.837 MJ/kg and PM is reported as Total Suspended Particulate. [13] The study was a lab-controlled test, the emission factor was reported as g/kg of fuel; it has been converted to g/MJ by assuming 18.6 MJ/kg, as reported on the website of Premium Pellet Ltd., BC, Canada.  The last row of stove X data consists of measurements associated with 18 sample field studies and measures in-home emissions directly from the plume. Since these results reflect the real operational situation much better, much higher emission rates were recorded. They were three times higher for CO, about 100 times higher for PM (PM 4.0 is measured instead of PM2.5), and 1.5 times higher for CO2. These results clearly 137  Chapter 5  indicate that CO2 emissions depend on the quantity and quality of fuel and show that inhome operating conditions significantly impact stove emissions. LPG stove characteristics were generally more uniform in comparison to stove X. The first row of LPG stove data is from Smith et al.’s study on six different fuel/stove combinations in Manila, Philippines (Smith, Khalil et al. 1993, Smith 1994). A series of stove characterization studies were done by East-West Center (EWC) and funded by the US Environmental Protection Agency (USEPA) in China and India. The LPG stove results mentioned in this table are sourced from the publications that resulted from these studies. Zhang et al. studied 28 fuel/stove combinations in China (Zhang, Smith et al. 2000). A similar study of 28 fuel/stove combinations commonly used in India is reported by Smith et al. (Smith, Uma et al. 2000). 5.4.2  Direct emission results  Gaseous emissions were corrected by subtracting the average of background emission concentration measurement data from during operation recorded measures. Net CO2 and CO from combustion is the recorded PPM of these gases during actual combustion minus the average of the PPM readings 10 minutes before and 10 minutes after the combustion. The combustion patterns of stove X and LPG stoves are quite different. The graph of CO concentration during combustion in LPG stoves (Figure 5-1) shows a sharp increase in CO concentration, which indicates the start of combustion, and a rapid decrease, which indicates the end of combustion. This graph is very different in the case of stove X. For stove X, the sharp initial increase (region I) represents the CO being emitted due to the combustion of kerosene (used as a fire starter) and indicates that the fan is not on yet. That is then followed by a relatively uniform region (region II), which indicates the flaming mode, when both fans are working. The final increase in CO emission (region III) is the smoldering phase before the combustion completely finishes.  138  Chapter 5  LPG Vs Stove X CO Graph (PPM) 30  Region I: Fire starter (Fan off)  Region II: Flaming Mode (Fan on)  Region III: Smoldering (Fan on)  25  20  PPM  15 Stove X AVG LPG Stove AVG  10  5  0 0  1000  2000  3000  -5  4000  5000  6000  7000  8000  9000  10000  Time (second)  Figure 5-1: Average CO concentration for three tests done on stove X and LPG stoves. Region I indicates the lighting phase, when kerosene is poured on the pellets to light the initial fire. After the fire has been propagated enough, the fan is turned on and the main combustion phase starts (Region II). Finally, after the flame is out, the smoldering phase (Region III) begins; it lasts until the red char is darkened.  Table 5-2 shows the average CO and CO2 emissions in PPM over one whole combustion cycle. Table 5-2: Emission concentrations of three tests Fuel/Stove  CO (PPM)  CO2 (PPM)  1.0  936  1.2  788  2.1  945  7.9  2140  4.0  2230  4.4  1414  Mean LPG (STDEV)  1.4 (0.6)  890 (88)  Mean ICS (STDEV)  5.4 (2.1)  1928 (448)  LPG stove  STOVE X/Pellet  139  Chapter 5  The relatively high variability seen in LPG stoves was caused by the LPG flow rate, which could not be controlled accurately by the existing facilities. Stove X’s variability is potentially a characteristic of the fuel/stove combination used, since the amount of fuel, the fire starter, and the flow rate were identical. Based on the emission concentrations, the dimensionless emission ratio (K) as a molar ratio to CO2 can be calculated for each test run; that is shown in Table 5-3. Table 5-3: CO to CO2 ratio (molecular ratio to CO2) Fuel/Stove K=CO/CO2 0.0017 LPG stove  0.0024 0.0034 0.0058  STOVE X/Pellet  0.0028 0.0049  Mean LPG (STDEV) Mean ICS (STDEV)  0.0025 (0.0009) 0.0045 (0.0015)  Using these emission ratios, we were able to construct emission factors (grams of pollutant per kg of fuel) and a rough carbon balance for these two fuel/stove combinations (see (Smith, Khalil et al. 1993) for more details on these methods). The carbon content of fuels is reported in the literature as 50% for pellets (Obernberger, Brunner et al. 2006) and 86% for LPG (Smith 1987). We assumed that the fuel was the only source of the net carbon measured during the test and as it is burnt it is mostly converted to CO and CO2. Disregarding the other carbon species such as methane, nonmethane hydrocarbons, carbonaceous aerosol and ash is estimated to add 1 to 4% error to the overall calculation (Smith, Khalil et al. 1993, Zhang, Smith et al. 2000). With these assumptions, the carbon balance is: CO + CO2 + ε =FC (Fuel Carbon),  140  Chapter 5  Where ε is about 1-4% of overall carbon content and so is assumed to be zero for calculating the emission factors (though the CO and CO2 emissions are then used to calculate the particulate matter emissions, which are a key factor for health impacts). So based on the K calculated in Table 5-3: CO2=FC/(1+K) Considering pellet carbon content, which is 50%, in each 1 Kg of fuel there are 500 grams of carbon. Therefore: CO2=500/1.0058=497 g carbon per kg of fuel, or 1822 g CO2 / Kg fuel Assuming the energy content of fuel is close to what was reported, pellet energy content is 18.6 MJ/Kg and LPG energy content is 46 MJ/Kg. Table 5-5 shows the average for both types of emission factors. To calculate PM, we used the equation proposed by Roden et al. (Roden, Bond et al. 2006): Equation 1: Calculating the PM emission factor (Roden et al. 2006)  ‫ܨܧ‬ሺܲ‫ܯ‬ሻ ൬  ݃‫݉ܽݎ‬ ൰ൌ ‫݈݁ݑ݂ ݂݋ ݃ܭ‬  ‫ܯܲ ݎ݁ݐ݈݅ܨ‬ଶ.ହ ݉ܽ‫ ݏݏ‬ሺ݃‫݉ܽݎ‬ሻ 1 1 ൈ൬ ሺ ଷ ሻ൰ ଷ ܸ‫ ݈݀݁݌݉ܽݏ ݁݉ݑ݈݋‬ሺ݉ ሻ ൈ ሺ∆‫ܱܥ‬ଶ ൅ ∆‫ܱܥ‬ሻ ݉ ‫ܱܥ‬ଶ , ‫ܱܥ‬ 8.34 ൫ܶ௔௩௚ ൅ 273.15൯ ݉ଷ ‫ܱܥ‬ଶ , ‫ܱܥ‬ ൈቆ ሺ ሻቇ ൈ ሺܿܽ‫݈݁ݑ݂ ݂݋ ݐ݊݁ݐ݊݋ܿ ݊݋ܾݎ‬ ݇݃ ‫ܥ‬ 0.012 ൈ ܲ ݇݃ ‫ܥ‬ ൈ൬ ൰ሻ ݇݃ ݂‫݈݁ݑ‬ 141  Chapter 5  Where filter mass was calculated by weighing the Teflon filters before and after the tests, the difference shows the amount of PM2.5 that was collected during combustion (shown in Table 5-4). The volume samples were based on the flow rate measured after the filter. The total volume of the samples was calculated by multiplying the duration and the average flow rate and is shown in Table 5-4. ∆CO2+∆CO are the mole fraction of CO and CO2 and were calculated by dividing their concentration in PPM (Table 5-2) by 10E6. Tavg is the average temperature of the environment recorded during the tests (shown in Table 5-4). The low temperature of one of stove X tests was due to the fact that it was the first test done in the morning. 0.012 is the molar mass of carbon, which is 0.012 kg/mol. P stands for pressure, which was assumed to be the standard pressure of 101 kPa. The carbon content of fuel is 0.5 kg of carbon per kg of pellet and 0.86 kg of carbon per kg of LPG. Table 5-4: PM2.5 mass (pre- and post-test filter weight differences), temperature as recorded during the tests Volume sampled Fuel/Stove Temp (°C) PM (mg) (m3) 0.142 32.2 0.035 LPG stove 0.033 0.141 32.5 0.037 0.154 31.8 0.333 32.2 0.518 STOVE X/Pellet 0.437 0.309 25.6 0.354 0.402 31.2 Mean LPG (STDEV) Mean Stove X (STDEV)  0.035 (0.002) 0.436 (0.082)  0.146(0.007) 0.348(0.048)  32.2 (0.3) 29.7(3.5)  The emission factor of PM2.5 was then calculated using Equation 1. The emission factors in (gram of PM/kg of fuel) and (gram of PM/MJ of fuel) are shown in Table 5-5, along with other emissions. Table 5-5: CO and CO2 emission factors (grams per kg of fuel) for stove X/pellet and LPG stove/fuel combinations CO CO2 PM EF EF EF (gram/ EF (gram Fuel/Stove EF (gram/ EF (gram/ (gram/kg (gram/MJ kg fuel) /MJ fuel) MJ fuel) kg fuel) fuel) fuel) 5.0 0.11 3161 68.7 0.48 0.01 LPG STOVE 5.2 0.28 1825 98.1 0.67 0.04 X/Pellet  142  Chapter 5  Emission factors based on gram/MJ fuel are more informative because they indicate the amount of fuel to be used to cook, which depends on both the energy content of fuel and the stove’s overall efficiency. The efficiency of LPG stoves is generally higher than that of ICSs, so it is safe to assume that with the inclusion of that efficiency, stove X is even more polluting. The calculated CO emission factor of stove X is somewhat lower than the EFs reported in the literature for this stove. That may be due to factors including the commercial-grade stove’s improved emission characteristics, the fuel/stove combinations that were in company A’s warehouse (i.e. using fresh fuel (lower humidity) and brand new stoves), direct plume measurement (we did not cover the flame), and the calibration of our instrument. However, the CO/CO2 ratio of our test (0.45%) is very close to the ratio that was similarly calculated (i.e. Carbon Balance Method) and reported in a study of a comparable stove, 0.35-0.4% (MacCarty, Ogle et al. 2007). Just (2012) reports the emission factors for CO and CO2 as 25.1 and 1707 respectively, which is close to our emission factors for CO2, but much higher than the CO emission factor we calculated (Just 2012). The PM emission factor is somewhat higher than what MacCarthy reports for a fan stove, 0.2 g/kg (MacCarty, Ogle et al. 2007), and very close to what Just reports as the PM emission factor of the residential version of stove X, 0.86 g/ kg fuel (Just 2012). Although our calculated emission factor shows some degree of variation from others that are reported in the literature, it is sufficiently informative for the purpose of comparison with LPG stove emission factors measured and calculated by similar devices and methods, and indicates almost three and four fold increases in CO and PM emissions respectively. It should be noted that the fan used on these forced draft gasifier stoves has a significant impact on the stoves’ pollution characteristics, and if not operational, often resulting in much higher rates of pollution (Just 2012). Field observations indicate that the breakdown of fans is very common in real settings. Moreover, although the quantity of PM emitted by this type of stove is lower than that of natural draft stoves, the emissions’ particle size characteristics are another cause for concern. According to lab measurements, the gasifier stoves emit proportionally more particles in the smaller size range (i.e. 2.5 microns or smaller), and it is these particle sizes that are of greater concern 143  Chapter 5  for health impacts (as they cannot be filtered out by the human body as larger particles can) (Just 2012). 5.4.3  Health effects  Data on health symptoms and the demographics of stove users and owners in both kitchens were collected through a survey that is explained in detail in Chapter 4. Information regarding the participants’ characteristics, perceptions of stove smokiness, and health symptoms was collected. Participant characteristics Kitchen characteristics and some related demographics regarding the respondents are displayed in Table 5-6. Data indicates that both respondents’ demographics and commercial kitchens’ physical attributes are comparable between the two groups. The survey indicates that all of the restaurants using stove X also use LPG stoves; in other words, they have all switched part of their cooking from LPG stoves to stove X. Generally, the LPG is bought in 19 kg cylinders, the standard size of commercial-grade LPG cylinders in India. Table 5-6: Summarized statistics of participants and work environment (i.e. kitchen) characteristics Space and Ventilation characteristics  STOVE X/LPG users (n=100) 168.3 (146.3) 1.7 (1.6)  Only LPG users (n=100) 145.8 (146.6) 1.6 (1.4)  Mean kitchen volume in cubic meters (SD) Mean number of exhaust fans (SD) User demographics1 Mean age of the user in years (SD) 35.97 (9.8) 34.02 (9.1) Regular smoker (%) 25.0 20.0 Emission and exposure duration estimates Mean of hours spent cooking by the user/day (SD) 6.1 (1.64) 5.7 (2.0) Mean of number of meals served/day in the kitchen (SD) 399.4 (178.8) 365.3 (169.8) Type of stove 2 100% 100% [1] “User” refers to the main chef who is the main user of the stove in the kitchen. “Owner” refers to the owner or manager of the kitchen who is in charge of stove purchase decisions. [2] The survey sample constituted only 100 LPG users and 100 stove X/LPG users.  The mean number of hours spent cooking was slightly higher for stove X users, as was the number of meals they prepared. Further, kitchen sizes were larger for the stove X sample population, though the size difference is very small and the confidence interval 144  Chapter 5  (95%) of the two kitchen types overlaps significantly. The Pearson correlation test for continuous variables indicates a significant, but not strong, association between these variables (Table 5-7). Table 5-7: Bivariate correlation test between continuous variables Cooking Number of Kitchen Number of User age hours per meals per volume (m3) exhaust fans day day Pearson 1 .191** .125 .282** .304** Kitchen Correlation volume Sig. (2-tailed) .008 .090 .000 .000 (m3) N 189 189 186 189 182 Pearson Number of .191** 1 .190** .255** .242** Correlation exhaust Sig. (2-tailed) .008 .008 .000 .001 fans N 189 199 196 199 192 Pearson .125 .190** 1 .056 .008 Correlation User age Sig. (2-tailed) .090 .008 .437 .913 N 186 196 196 196 189 Pearson Cooking .282** .255** .056 1 .503** Correlation hours per Sig. (2-tailed) .000 .000 .437 .000 day N 189 199 196 199 192 Pearson .304** .242** .008 .503** 1 Number of Correlation meals per Sig. (2-tailed) .000 .001 .913 .000 day N 182 192 189 192 192 **. Correlation is significant at the 0.01 level (2-tailed).  Mean comparison tests also did not reveal any significant difference between the two groups. Perception of Smokiness Questions regarding owners’ and main users’ (i.e. chefs) perception of the smokiness of both stove types were asked. LPG stoves were seen as less smoky compared to stove X. Respondents were asked to answer the question of whether the stoves produce no smoke using a Likert scale: 1) strongly disagree, 2) somewhat disagree, 3) neither agree nor disagree, 4) somewhat agree, and 5) strongly agree. For commercial kitchens using stove X, that question was asked regarding both LPG stoves and stove X. Owners’ answers to the questions are shown in Figure 5-2.  145  Chapter 5  Figure 5-2:: Restaurant owners’ perception of the smokiness of LPG stoves versus stove X  The scale on the vertical rtical axis moves from strongly disagree (1) to strongly agree (5). It shows that while 60% of respondents strongly agreed that LPG stoves produces no smoke, only 20% feel that way about stove X. Moreover, while no respondents disagreed to any extent that LPG stoves produce no smoke, some respondents disagreed with the statement that stove X produces no smoke. It is clear that owners perceive stove X as smokier than LPG stoves. Figure 5-3 shows similar responses from main in stove users (i.e. chefs).  Figure 5-3:: Restaurant stove users’ perception of the smokiness of LPG stoves vs stove X  146  Chapter 5  Figure 5-3 shows a very similar trend, which means that both owners and stove users believe stove X is smokier compared to LPG stoves. The similarity of users’ and owners’ perceptions of stove smokiness also can be seen in Table 5-8, which is based on the mean of scores (1 to 5). The Wilcoxon matched-pair signed rank test was done on user and owner perceptions of the statement “the stove produces no-smoke” with regard to LPG stoves and stove X, and in both cases the difference was significant at the .000 level. Table 5-8: Mean scores of stove smokiness s perceptions (1 (strongly disagree) to 5 (strongly agree)) User  Owner  Mean of LPG stove score (STDEV)  4.55 (.60)  4.53 (.71)  Mean of stove X scores (STDEV)  3.59 (1.00)  3.32 (1.06)  .000  .000  Wilcoxon-test significant  Health symptoms A number of the eye and respiratory symptoms reported by the users are shown in Table 5-9. Except for “burning sensation in the eyes,” stove users working in LPG and stove X kitchens reported no significantly different health symptoms. Descriptive analysis Table 5-9: Summarized statistics of health symptoms reported by stove users Users in stove X Kitchen  Users in LPG kitchen  (n=100)  (n=100)  Diagnosed with asthma (%)  0.0  2.0  NA  Troubled by cough in the last one year1 (%)  11.0  14.0  .521  Bringing phlegm from the chest in the last one year (n)  6.0  9.0  .421  Often having burning sensation in the eyes (%)  35.0  15.0  .001  Noticed discharge on eyelids in the morning (%)  24.0  18.0  .386  Eyes look red often (%)  21.0  14.0  .264  Often have watering eyes  24.0  17.0  .147  Health symptoms  Chi-Square test p  [1] The interviewer communicated the concept of intense coughing that is troubling  147  Chapter 5  Only two cases reported being diagnosed with asthma and both were LPG stove users. Due to the very low response rate, the data on this respiratory symptom was not be conclusive and we could not run any further analysis on it. The number of users who reported being troubled by a cough during the last one year was about 11% for stove X users and 14% for LPG stove users. The number is very close and is not conclusive. The people who reported bringing up phlegm from the chest were all people who reported being troubled by coughing. Table 5-10 shows the cross tabulation of these two questions. Table 5-10: Cross tabulation of being troubled by a cough and bringing up phlegm In the last one year have you been troubled by a cough?  Total  Count  No  Yes  Did you ever bring phlegm No up from your chest in the Yes last year?  175  10  185  0  15  15  Total  175  25  200  In terms of the respiratory symptoms, the only variable with a large number of observations is the experience of being troubled by a cough. Since the results for that symptom are very similar between LPG and stove X users, no association between stove use and these symptoms can be drawn using this data. We were not able to see any significant difference between stove X and LPG users regarding respiratory-related symptoms. Considering the relatively short time since stove X has been used in kitchens (mean 16.2 months (sd=2.1)), chronic symptoms due to overall exposure to this smoke could not be identified. Moreover, the sample number was not large enough to be able to indicate small and medium effects. We further used this symptom as a dependent variable in regression analysis to explore the associations controlling for other variables. Regarding eye symptoms, the difference is more significant. 35% of stove X users and 15% of LPG stove users reported often having a burning sensation in their eyes. As mentioned in the literature review, eye irritation can be considered an appropriate indicator of IAP (Ellegård 1997). Within the group that reported often experiencing a 148  Chapter 5  burning sensation in their eyes, all of them experienced it during and right after cooking sessions (Figure 5-4). Among the respondents who experienced the eye burning sensation, 56% reported that the eye burning stops after a period without cooking. 38% didn’t know if that is the case, and 6% reported that the symptoms do not disappear after a period without cooking. The responses indicate that the eye burning sensation is associated with cooking. The statistically significant difference in the number of LPG users and stove X users who experience the eye burning sensation could be indicative of higher IAP due to the usage of stove X.  Figure 5-4: Frequency table for when the eye burning sensation occurs  Three other eye-related problems were probed: discharge on the eyelids, eyes that often look red, and eyes that often water. Although no statistically significant association was found between LPG users and stove X users regarding these three symptoms, the chisquare test shows a statistically significant association between these three symptoms and experiencing a burning sensation in the eyes. Often, respondents who reported experiencing these three symptoms (discharge on the eyelids was reported to a lesser extent), belonged to the same group of respondents who reported a burning sensation in 149  Chapter 5  their eyes. Bivariate correlation analysis also shows a significant and strong correlation between these symptoms (Table 5-11). Since the burning sensation question is the only one that was followed up by further questions, such as “When do you experience the eye burning sensation?” and is clearly associated with cooking activity – and because most of the other symptoms were reported by respondents who had already reported experiencing a burning sensation in their eyes – we decided to use the burning sensation as one of the dependent variables to be analyzed further through a regression model. Bivariate correlation analysis was done on both respiratory symptoms and eye irritation; the non-parametric correlation of Kendall’s tau-b, as well as that of spearman rhaw, was calculated, and the two were found to be identical. Table 5-11shows the results. Table 5-11: Bivariate correlation coefficient of two health symptoms and stove use category In the last one year have you been troubled by coughing?  Do you get a burning sensation in the eyes often?  1.000  .045  -.231**  -  .522  .001  N (observation)  200  200  200  Correlation Coefficient  .045  1.000  .061  Sig. (2-tailed)  .522  -  .389  N (observation)  200  200  200  -.231**  .061  1.000  Sig. (2-tailed)  .001  .389  -  N (observation)  200  200  200  Kendall’s correlation coefficient τ  User category (stove X/LPG vs LPG only)  In the last one year have you been troubled by coughing? Do you get a burning sensation in the eyes often?  Correlation Coefficient Sig. (2-tailed)  Correlation Coefficient  User category  The correlation coefficient and significant test indicate a strong association between stove use and experiencing an eye burning sensation. The association between stove use and being troubled by a cough was not statistically significant.  150  Chapter 5  Regression Analysis Building upon our findings from the descriptive statistical analysis, a binary logistic model was designed to investigate the association between energy system use (LPG and stove X/LPG) and the eye burning sensation and coughing. The model controls for factors with a potential impact on IAP, including exposure-dose variables such as the number of meals cooked per day as a proxy of energy use, the number of exhaust fans in the kitchen, kitchen volume, and user age (a demographic factor that in the literature is found to be associated with health symptoms). Binary logistic regression model: We used the logistic regression equation that expresses the probability of success (Agresti and Finlay 2009, Field 2009). The logit model is defined as: Pi describes the probability of the occurrence of dependent variable Y over the value of x: Pi=Pr(Y=1|X=xi) ‫ ݃݋ܮ‬൬  ܲ௜ ൰ ൌ ‫ݐ݅݃݋ܮ‬ሺܲ௜ ሻ ൌ ܻ௜ ൌ ߚ଴ ൅ ߚଵ ‫ݔ‬ଵ ൅ ‫ ڮ‬൅ ߚ௡ ‫ݔ‬௡ 1 െ ܲ௜ ܲሺܻ ൌ 1ሻ ൌ  1 1 ൅ ݁ ି௒೔  Where P(Y) is the probability of Y occurring (Y is the dependent variable of interest, in this case the eye burning sensation and cough attacks, which are measured by the dichotomous response question in this study) given known values of independent variables. The relevant questions in the survey required a yes or no response and are as follows: 1) Do you often experience an eye burning sensation? and 2) Have you been troubled by coughing in the past year ? (the interviewer explained that a cough attack means an intense experience of coughing) β0 is the intercept that is the log odds of Y (P/1-P) when the value of all independent variables is 0, and β1 to β8 indicates the multiplicative effect 151  Chapter 5  of the associated independent variables on the odds of Y occurring, controlling for other variables (Agresti and Finlay 2009, Field 2009). The odds ratio, given by eβ, expresses the likelihood of change in the dependent variable given a one-unit increase in the independent variable (Kutner et al., 2005). A binary logistic regression model was designed for each dependent variable. The model includes all observations and explanatory variables and indicates whether the person is a stove X user or not. The odds ratios are shown in the model results under the heading EXP(B). Yi=β0+ β1UC+ β2smker + β3userage + β4cooktime + β5kchnvol+ β6fanno+ β7mealsum  The independent variables selected were used as the proxies of demographic factors (age, smoking behavior) and exposure, including emission concentrations (source of emission, amount of emission, ventilation) and duration of exposure. UC is the user category which is 0 for LPG kitchens and 1 for stove X kitchens. Age is the age of the main stove user (i.e. chef) being interviewed, kchnvol is the volume of the kitchen in cubic meters (it should be noted that the interviewer explained the kitchen as the place where the main cook stoves, including stove X, are located), fanno is the number of exhaust fans in the kitchen (this measurement was found to be a better approximation of forced ventilation since even in kitchens with modern chimneys, stove X was not necessarily located under the chimney), cooktime is the amount of time the respondent indicated spending on cooking (the minimum amount of time the user reported being exposed to the kitchen’s emission concentration), mealsum is the number of meals served per day in the kitchen (used as a proxy of the amount of fuel used in the restaurant), and smker indicates whether the respondent is smoker or not (a variable that can have a causal relationship with respiratory symptoms). Each model has two variants: I) Without controlling for energy consumption (as proxied by the total number of meals served per day (mealsum)), in order to investigate the presence of stove X regardless of energy use intensity and II) Controlling for the effect of total energy consumption on IAP (i.e. including mealsum). Table 5-12 shows the sample statistics for the logit regression model described above.  152  Chapter 5  The difference in variable scales may have an impact on the magnitude of coefficients. As such, all the variables are standardized and calculations were based on the standardized variables, which are the z score of the variable responses. The z score was calculated as: ܼ௜ ൌ  ܺ௜ െ ܺത ܵ  in which Xi is the score of the case for each particular variable, ܺത is the mean of the  variables’ scores, and S is the standard deviation of the scores of the variable of interest. Table 5-12: Summarized statistics of variables used in the model Dependent Variables Category  Variable  Obs  Yes (%)1  No (%)  Yes count  No count  Health symptom  Coughing  199  12.6  87.4  25  174  Eye irritation  199  25.1  74.9  50  149  Independent Variables Continuous Variables Category  Variable  Obs  Min  Max  Mean  STDV  Demographic  User age  196  18  68  35.5  9.5  Time  Cooking time  199  2  10  5.9  1.8  Dose-emitted  # meals per day  189  90  980  379.5  174.9  Doseventilate  Kitchen vol (m3)  189  5.09  815.0  156.8  146.5  Number of fans  199  0  10  1.6  1.5  Categorical Variables Variable  Obs  Yes (%)  No (%)  Yes count  No count  Health background  Smoker  199  22.1  77.9  44  155  Dose-emitter  Stove X kitchen  199  50.3  49.7  99  100  [1] valid percentage, the percentage of answered, excluding not replied  153  Chapter 5  Regression model A results: Dependent variable eye irritation Table 5-13 presents the results of both iterations of regression Model A Eye irritation. Table 5-13: Regression model A) eye symptom, (eye irritation) Iteration I Variable  Iteration II  B  Sig.  Exp(B )  95% C.I.for EXP(B) Lower Upper  1.031  .006  2.804  1.349  5.828  Smoker  -.195  .676  .823  .329  2.054  User age  .038  .846  1.039  .707  Cooking time  .276  .163  1.318  Kitchen vol  -.209  .404  .235 Number fans #meals per NA day Overall model fit  UC  95% C.I.for EXP(B) Lower Upper 1.698 7.908  B  Sig.  Exp(B )  .001  3.664  .550  .749  .291  1.930  1.525  1.29 9 .289 .081  .692  1.084  .727  1.618  .895  1.941  .551  .019  1.735  1.093  2.753  .812  .497  1.352  .837  .948  .569  1.580  .235  1.265  .858  1.867  .053 .525  .194  1.316  .870  1.991  NA  NA  NA  NA  .480  .028  .591  .371  .944  Chi-square significant test 0.052  .003  0.065  .114  Cox & Snell R square[4]  .168 0.098 Ngelkerke R Square [1] UC (user category): 1 is assigned to stove X and LPG users; 0 is assigned to LPG-only users [2] Smoker: 1 is assigned to smokers; 0 is assigned to non-smokers [3] Results significant at a level of 0.05 are shown in bold fonts [4] An equivalent statistic to R-squared does not exist when using logistic regression. Instead, the pseudo R-squared is used instead to indicate the goodness of fit of the model. There are multiple methods of calculating pseudo R-squared. As with the R-squared, their values range from 0 to 1, with higher values indicating a better model fit. However, the value cannot be interpreted as the Ordinary Least Square (OLS) R-squared. SPSS uses Cox and Snell’s R2sc and Nagelkerke’s R2n. The benefit of the latter is that unlike Cox and Snell’s method, Nagelkerke’s method can theoretically reach a maximum of one (Field 2009)  Both iterations of the model indicate a significant association between stove X uptake and the experience of eye irritation. Controlling for the energy, the model indicates a strong (odds ratio of 3.7 with CI 95 1.7 to 7.9) and significant association (at P<0.001) between stove X use and an increase in the experience of a burning sensation in the eyes. Cooking time was also found to be statistically significant, with an odds ratio of 1.735. That indicates that the longer a person spends working in the kitchen, the higher chance they will have of experiencing eye irritation. The number of meals served per day is also significant at the 0.05 level, but the direction of association is counterintuitive since it can be expected that more meals would equal more emissions, and thus higher chances of eye 154  Chapter 5  irritation. To understand whether results were impacted by an association between kitchen characteristics, number of meals, and number of hours spent for cooking, a cluster analysis on the cases has been done. Hierarchical clustering method based on the ward method and the squared Euclidean distances between the clusters was implemented. The mean comparison and the ANOVA table associated with these clusters are shown in Table 5-14 and Table 5-15. Table 5-14: Mean comparison of three cluster solutions Ward Method  1  2  3  Total  Mean N Std. Deviation Mean N Std. Deviation Mean N Std. Deviation Mean N Std. Deviation  Kitchen Meals cooked per volume day 65.78 251.87 56 56 43.25 75.50 164.97 368.03 91 91 159.45 128.11 295.16 635.31 32 32 118.52 143.21 157.20 379.47 179 179 148.19 174.89  Number of Cooking time per exhaust fans day .64 4.18 56 56 .67 1.20 1.77 6.15 91 91 1.30 1.45 2.66 7.84 32 32 1.54 .92 1.58 5.84 179 179 1.38 1.81  The ANOVA Table 5-15shows the variance between the clusters. The significant tests point out the statistically significant different means for these clusters. Table 5-15: ANOVA table of the means of each cluster in the three cluster solutions based on four variables: kitchen volume, total number of meals prepared per day, number of exhaust fans in the kitchen (the room in which stoves are mainly used), and the time per day that the user spends cooking in the kitchen Ward Method Clusters ANOVA Table  Sum of  df  Mean  Squares Kitchen volume Meals cooked per day Number of exhaust fans Cooking time per day  F  Sig.  Square  Between Groups  1082570  2  541285  33.7  .000  Between Groups  3018182  2  1509091  109.5  .000  Between Groups  89.5  2  44.8  31.5  .000  Between Groups  291.7  2  145.8  87.3  .000  The analysis of clusters indicates that three significantly distinct clusters can be formed between cases. These clusters indicate that a larger kitchen volume, the more exhaust 155  Chapter 5  fans, more meals cooked per day, and a longer cooking period. That makes the relationship between IAP and the number of meals cooked complicated because it is clear that more meals are cooked (resulting in higher emission rates) in larger kitchens (which allow for lower IAP concentrations) with better ventilation (a higher number of fans). That may explain the negative association between the number of meals cooked and the experience of eye irritation. Regression model B results: Dependent variable respiratory symptom (coughing) Table 5-16 presents the results of both iterations of regression Model B. The dependent variable is experiencing coughing attacks. Table 5-16: Regression model B) Respiratory symptom, (coughing) Iteration I Variable  Iteration II 95% C.I.for EXP(B) Lower Upper  B  Sig.  Exp(B)  -.774  .124  .461  .172  Smoker2  .890  .107  2.436  Userage  -.066  .812  Cooktime  0.817  Kchnvol Fanno  UC1  Mealsum  95% C.I.for EXP(B) Lower Upper .136 1.112  B  Sig.  Exp(B)  1.235  -.945  .078  .389  .824  7.202  .855  .136  2.352  .765  7.236  .937  .546  1.608  -.121  .692  .886  .487  1.611  .001  2.263  1.397  3.667  .494  .081  1.638  .941  2.852  -.575  .120  .563  .272  1.162  -.609  .145  .544  .240  1.231  -.821  .017  .440  .225  .861  -.666  .059  .514  .257  1.027  NA  NA  NA  NA  NA  .334  .299  1.396  .744  2.620  Model overall fit Chi-square significant test  0.001  .053  Cox & Snell R square  0.110  .076  Ngelkerke R Square  0.208  .150  [1] UC (user category): 1 is assigned to stove X and LPG users, 0 is assigned to only LPG users [2] Smoker: 1 is assigned to smokers, 0 is assigned to non-smokers [3] The significance level of 0.05 is shown in bold font  [4] An equivalent statistic to R-squared does not exist when using logistic regression. Instead, the pseudo R-squared is used instead to indicate the goodness of fit of the model. There are multiple methods of calculating pseudo R-squared. As with R-squared, their values range from 0 to 1, with higher values indicating a better model fit. However, the value cannot be interpreted as the Ordinary Least Square (OLS) R-squared. SPSS uses Cox and Snell’s R2sc and Nagelkerke’s R2n. The benefit of the latter is that unlike Cox and Snell’s method, Nagelkerke’s method can theoretically reach a maximum of one (Field 2009)  The first iteration indicates the significant positive association between cooking time and coughing as well as the negative association between the numbers of exhaust fans and 156  Chapter 5  being troubled by coughing. These two variables can be seen as the proxy of the duration of exposure and the quality of ventilation. The results are consistent with our expectation that increases in exposure would be positively associated with respiratory symptoms, while enhancing ventilation would be negatively associated with them. However, after controlling for total meal number, neither the overall model nor the independent variables have been found to have a significant association. In particular, no significant association between using stove X and coughing was found using this model. That was expected based on descriptive analysis since almost the same number of people were troubled by coughing in both user groups. 5.5  Conclusions  Through this study, we have tried to assess the implications of the uptake of ICSs with regard to human exposure to potentially harmful emissions. This study adopted a unique mixed method of research, combining 1) physical measurements of direct emissions, 2) a qualitative assessment of users’ perception of smokiness and their awareness of the health hazards associated with it, and 3) the assessment of health symptoms associated with IAP. The resulting assessment brings together these lines of evidence along with the existing evidence in the literature on emissions based on household sized stove measurements (Table 5-17). First, the review of the emission data of comparable stoves reported in the literature confirmed that in general stove X (residential version) is much smokier than any residential LPG stove. Second, direct emission measurements (CO, CO2, and PM emission factors for the commercial variant of stove X and a commercial LPG stove) were estimated based on the direct emissions of these stoves in the field,. The results indicate that the commercial version of stove X is more polluting than a comparable LPG stove. Third, our analysis shows that in general, both users and owners are in greater agreement with the statement that “LPG stove produces no smoke” than they are in terms of the statement “stove X produces no smoke.” These results can be interpreted a perception that stove X is smokier than LPG stoves. Finally, the data on self-reported health symptoms (i.e. experiencing cough attacks and experiencing an eye burning sensation) were collected through the survey; they indicate early effects of smoke exposure for one symptom (eye irritation). Through binary regression analysis, the correlation between using stove X instead and experiencing these symptoms was 157  Chapter 5  estimated. Models indicate a statistically significant relation between experiencing a burning eye sensation and using stove X, but no significant relation between having coughing attacks and stove X use. The conclusions from these four lines of evidence are presented in Table 5-17. Table 5-17: Concluding remarks Results Higher PIC emission rate for stove X compared to LPG stoves, Secondary Data both in direct measurements and concentration measurements. Our study indicates a CO emission rate that is almost double for Direct Emission stove X as well as higher PM emission rates in comparison to LPG stoves. Relative emission rates would be higher if stove efficiencies were included. Stove Smokiness Perception Both kitchen owners and stove users consider stove X smokier than LPG stoves. No significant association between coughing and the uptake of Health Symptom stove X was found. Respiratory A significant association between eye irritation and usage of Health Symptom Eye stove X relative to LPG stoves was found. Irritation  Based on our results it can be concluded that the LPG to ICS switch in commercial kitchens will likely expose kitchen workers, and perhaps the people in the rooms attached to these kitchens, to higher levels of pollutants. Our results regarding health symptoms indicate some association between the variables; however, the variance between individuals and reported symptoms creates a statistical power issue. The analysis of our sample size of 100 respondents in each category can only indicate strong associations. A larger sample size is required to point out weak associations (Field 2009). It should be noted that there is a range of volatile compounds (carbonyl compounds) that have significant adverse health effects and exist only in biomass smoke, not LPG smoke. Therefore, many factors point to the increased health risks associated with this switch. Moreover, the contribution of cooking fuel to outdoor air pollution (OAP) has also been shown by researchers; 25-50% of OAP in India, for example, is contributed by cooking (Smith 2012). The growing number of commercial kitchens switching partly to ICSs in major urban centers, which are already facing the challenge of OAR, may exacerbate the situation and its associated health risks.  158  Chapter 5  This study is a primary step in investigating the occupational health hazards associated with cook stoves. However, its results can be extended to situations in which households switch back to lower-grade energy systems due to increased costs or limited availability of modern energy systems such as LPG stoves. It is necessary to confirm the extent of these impacts with additional research along all three lines of primary evidence collection in this thesis: 1) Kitchen concentration and exposure assessment studies under realistic operating conditions and incorporating a wider range of emissions; 2) Additional data collection on user perceptions of stoves, emissions and user priorities in decision-making and operation of stoves; and 3) More comprehensive health studies that include a more elaborate assessment and diagnosis of symptoms through both self-reporting and medical examinations conducted over a time period long enough to observe potential health impacts. Further policy studies may look at the pros and cons associated with LPG subsidies and the leakage impacts of energy pricing on other sectors. Keeping that in mind, moving toward modern energy systems with higher performance and cleaner combustion is a necessary step for achieving sustainable development in the poor and underdeveloped areas of the developing world.  159  Chapter 6  Chapter 6 Conclusions The research presented in this dissertation contributes to an improved understanding of the energy transition process in developing countries. This concluding chapter includes a summary of each chapter’s findings as well as discussions on the limitations of this research, the way forward, and some overarching concerns about policy formulation. 6.1  Summary of findings and concluding remarks  6.1.1  Knowledge gaps in understanding the adoption of energy systems  In chapter 2, I explore current knowledge of household energy use and energy transition, focusing in particular on households in rural regions of the developing world. Many disciplines, including engineering, economics, psychology, sociology, and anthropology, are seen as contributing to the field of household energy use. These disciplines approach this issue uniquely, often in isolation of each other (Keirstead 2006). However, the Physical-Technical-Economic Model (PTEM) is the model that is most often used to address energy issues and inform policies and programs (Wilhite, Shove et al. 2000, Lutzenhiser, Cesafsky et al. 2009, Webler and Tuler 2010). Integrated approaches are seen as the best way of combining the knowledge of these disciplines and others (e.g. environmental science) to create a more realistic picture of the situation at hand (Wilson and Dowlatabadi 2007). However, such integration is hindered by a number of factors, including: 1. Institutional barriers. Different disciplines perceive each other differently, and that can make integration difficult (Lutzenhiser 1993). 2. Poor communication. The output from integrated models is often qualitative, not quantitative, meaning that it is harder for the policy making community to comprehend or use to justify action (Keirstead 2006) 3. The limited scope and scale of integrated approaches (Keirstead 2006). 4. Limited objective behavioral data – the result of a lack of qualitative studies on household energy use in developing countries (Urban, Benders et al. 2007). One of the major issues that analyses of household energy have tried to address is the process of transition from one set of energy systems to another. Our knowledge of 160  Chapter 6  energy transitions and household decisions regarding which energy system to adopt is quite limited; income is often seen as a major determinant (Leach 1992, ESMAP 2003, Pachauri, Mueller et al. 2004) as is the universal hierarchy of fuel and energy services. Nevertheless, household energy portfolios (rather than a single energy system) depend on complex interactions between economic, social, cultural, and environmental factors (Leach 1992, Masera, Saatkamp et al. 2000). The factors that the literature reports as determining household energy choices can be categorized as 1) endogenous factors, including households’ socio-economic characteristics and 2) exogenous factors, such as physical and environmental conditions, policies, and supply factors, as well as energy device attributes. Following an overview of these factors, a number of shortcomings in the literature on rural household energy analysis were identified, including: 1. Shortcomings in methodologies, including predicting micro trends through macro analysis (Elias and Victor 2005), insufficient data (ESMAP 2003), and claiming causality where only a statistical correlation has been established (Stern 1986). 2. Limited understanding of the drivers of energy use and transition, including overemphasizing income as the main driver of energy choice (Jack 2006) and paying insufficient attention to the human dimension of energy use (Lutzenhiser 1993). 3. Limited view of energy system, such as focusing on the amount of energy used rather than the services provided by energy use (Wilhite, Shove et al. 2000) , 4. Addressing energy use simplistically without sufficiently addressing the interactions between energy services, energy devices, and energy carriers (Fitzgerald, Barnes et al. 1990). Building upon these findings, a conceptual framework was proposed. It attempts to integrate three dimensions of household energy systems (i.e. energy device, energy carrier, energy service) with the exogenous and endogenous factors that influence them and their interactions. Informed by this framework, the energy transition can be seen beyond the energy ladder and energy stacking models (the prevalent models of energy transitions) in a three-dimensional space that incorporates all three dimensions of energy use. 161  Chapter 6  This framework can act as a basis for building new theoretical and empirical models of household energy use, which in turn can be used to formulate more effective programs and policies by improving our understanding of energy use. While these programs can be helpful in addressing the issues associated with households’ adoption of energy systems, their design and the approach they take contributes to whether they will fail or be successful. In the next chapter, recent energy provision programs are explored. 6.1.2  Private sector based approach: great potential, but not universal  Chapter 3 investigates the private sector based approach to energy provision and reviews the literature on electrification and clean cooking energy provision programs. The major technological and institutional trends in programs attempting to address these two issues are identified that are summarized in Table 6-1. Table 6-1: Recent major trends in electrification and clean cooking energy provision Energy service provision Technological Institutional Growing interest in distributed generation and renewable energy technologies, particularly biomassElectrification based energy systems, instead of conventional fossil-fuel based central Moving from state-led, power generation and grid extension. central, and donor-driven energy provision programs Trend toward providing an to decentralized and marketintermediary solution, such as oriented approaches. biomass-burning ICSs, as a more Cooking energy system immediate action in compare to provision modern energy cooking systems.  Biomass-based energy systems (i.e. biomass gasification-based electricity generation and pellet-burning ICSs) have been receiving lots of attention due to the various advantages that are believed to be associated with them, including: 1) availability of raw material, 2) ease of storage, 3) easy conversion to other forms of energy, 4) fewer impacts on global warming, and 5) contribution to rural development(Kartha, Leach et al. 2005). While our literature review confirms some advantages of the private sector based approach to providing rural energy access through biomass-based energy systems, it also points out a number of issues. Two hypotheses were created based on these issues: 162  Chapter 6  I)  The private sector based approach to rural energy provision cannot be a universal solution as private sector firms generally do not aim for universal access and is therefore not appropriate for all segments of market.  II)  Even if enterprises are addressing a more profitable segment of the market, they are facing many challenges that are beyond their capacity to address.  The first hypothesis was examined through a comprehensive literature review as well as a case study. Two enterprises active in rural energy provision through biomass-based energy systems were selected. Members of these two entities, including senior executives, were interviewed using a semi-structured protocol. To confirm our findings, two more enterprises were investigated. In brief, whilst all of the interviewees mentioned the huge market potential in residential energy provision in rural India, none succeeded in building a sustainable business solely through providing energy to households, despite the support and the funding they had received. Our study confirms that although there are some advantages associated with the marketoriented approach to rural energy provision, it cannot be considered a universal solution (Schlag and Zuzarte 2008, Bailis, Cowan et al. 2009); also, while energy provision through the for-profit private sector may function for a certain segment of the market, it may not be appropriate for others. Nonetheless, market-oriented programs have many positive aspects that should be integral to strategies aimed at addressing energy poverty. Close collaboration between the private and public sectors can be seen as a prerequisite for formulating and successfully implementing meaningful policies and programs. Using both secondary data and the case study, a range of challenges that private enterprises are facing in addressing rural energy provision were identified that could be categorized into I) Financial constraints and weak cash flow, II) Rural energy market specific characteristics and market deficiencies, III) policy and institutional issues, and IV) operational challenges. Our study confirms that even if the enterprises are addressing the right market segment, they face a range of challenges that may prohibitively affect their performance. 163  Chapter 6  Three major common deficiencies that need to be addressed in order to effectively engage the private sector in rural energy provision were identified. These include: I)  The business environment in the developing world. Institutional and structural issues have been found to have significant implications for success of business operations.  II)  The rural and low-income market segments. There are a range of challenges inherent to operating in this segment. For example, while operating costs in rural and remote regions are high, businesses operating there often face weak cash flow due to the inability of low-income consumers to pay for their services. Income characteristics specific to rural regions of the developing world are also an issue.  III)  The profit motive inherent in commercial operations. Finally, and perhaps the most importantly, businesses are sustainable only when profitable, and they have no motive to pursue low-income consumers unless the government can provide them with enough incentive to do so.  Addressing these issues is well beyond the capacity of private entities and strong government dedication and involvement is required. On their own, private entities may not only face many challenges in providing energy to rural regions, particularly to lowincome consumers, but also may adversely affect the situation in such areas. Therefore, careful attention is an essential part of creating an enabling environment for businesses and providing the appropriate safety net for the BoP population, which cannot afford the services provided by businesses. However, as mentioned earlier, while these businesses were not successful in addressing this particular market segment, at least one of them (i.e. the ICS developer) received very positive feedback when dealing with commercial entities in major urban centers. While the dissemination of ICSs has faced prohibitive challenges in rural regions, the stoves have received lots of interest from commercial entities; this topic was investigated in chapter 4. 164  Chapter 6  6.1.3  The improved cookstove: a device to ascend and descend the energy ladder?  As discussed earlier, while enterprises were not successful in developing a sustainable ICS business in the residential segment, commercial kitchens proved to be interested customers for this type of stove. Our case study on commercial kitchens in Bangalore was aimed at exploring some of the under-researched areas of stove uptake by: 1) focusing on user perceptions of various stove attributes to explore how stove characteristics are viewed by buyers and to discover which characteristics most influence them to adopt these stoves, 2) investigating the switching back phenomenon, in which the energy users at least partly replace their modern and clean energy systems with less modern systems (i.e. partly replacing LPG use with biomass ICS use), and 3) exploring the commercial segment’s adoption of cooking energy systems, a topic that has not received much attention. The case study explored the two main foci of stove dissemination programs: a) the improvement of health conditions and b) fuel saving. These factors’ influences on adoption decisions are examined. Furthermore, the split incentives inherent in the commercial segment due to separation between owners/managers and workers are studied. Analysis of data from kitchens using only LPG stoves and kitchens using a combination of LPG stoves and ICSs indicates variation between both users’ and owners’ perceptions of the attributes of LPG stoves. The most significant difference is in their perception of stove smokiness; that is, both owners and users perceive the LPG stove as less smoky in LPG/ICS kitchens than they do in only LPG kitchens. These differences may be due to the presence of stove X as a point of comparison. Further comparison of the respondents views on attributes of LPG stoves and stove X (ICS) was done solely based on data from kitchens using both stoves. Our analysis reveals that both owners and users perceive LPG stoves as being significantly better (p<.001) in a number of attributes including, simple and convenient to use, cooks quickly, ability to cook most dishes, smokiness. They also perceive stove X fuel as cheaper (p<.05); that can be seen in the price difference between commercial (unsubsidized) LPG and pellet fuel that is burnt in stove X. Moreover, only owners consider the LPG stove more reliable (p<.05), which could be due to fact that they take 165  Chapter 6  care of maintenance and pay repair bills. On the other hand, the users (i.e. cooks) believe the food tastes better on stove X (p<.05), although that is to be expected considering cooks are more concerned about how food tastes. The results indicate that while the LPG stove is considered a better stove in regard to many attributes, kitchens still purchase and use stove X (ICS) to partially replace their LPG usage. To determine why stove X is being purchased, owners were asked about factors such as convenience, functionality, health concerns, economic proposition, and fuel availability. Analysis of kitchen owners’ responses reveals that economic propositions are the dominant reason for purchasing stove X. While functionality, convenience, and fuel availability were also stated as reasons for purchasing stove X, health concerns did not rank highly. To further investigate owners’ reasons for adopting stove X, owners of LPG only kitchens were asked why they did not buy stove X. Lack of knowledge about stove X was found to be the dominant reason. Functionality and convenience were also stated as reasons for not buying stove X. Further investigation was done in the case of economic proposition and health concerns. Regarding the economic proposition, findings from the case study in conjunction with data collected from the field indicate that significant fuel price differences may be the determining factor in purchase decisions; that is, in the case study, economic proposition outranked all other reasons. Regarding health, the analysis shows that although both owners and users are aware of the adverse health impacts of smoke in the kitchen – and regardless of the fact that those in LPG/stove X kitchens were more bothered by smoke – health concerns did not influence purchasing decisions. This discovery may have a significant impact on ICS dissemination programs that focus on raising awareness about the health outcomes of IAP and put less attention on economic proposition. The split incentive was explored by investigating owners’ and users’ reasons for not taking any action to reduce smoke. A large portion of owners stated that there is no smoke (70%) in comparison to a smaller portion of users who hold that view (42.9%). Also, about 43% of users stated that the reason they have not taken action is that they are not in charge.  166  Chapter 6  The findings of our case study were also compared to the findings of a similar Stanford University study on residential customers. The reasons that residential and commercial customers gave for the uptake of stove X are not wholly different. Economic propositions are the dominant determinant, followed by functionality and convenience; health concerns are not considered significant by either group. 6.1.4  Replacing LPG use with biomass, even in advanced ICSs, has resulted in increased indoor air pollution, which potentially has adverse health impacts  Our field study reveals the growing trend, mainly driven by economic propositions, of commercial customers taking up ICSs. Moving toward less modern energy systems – also referred to as moving down the ladder – may have socioeconomic impacts, particularly on the people working in these kitchens. The move toward removing fossil fuel subsidies may trigger a similar trend in different sectors. Moreover, although a fair number of studies have explored the impacts of IAP on the residential sector, this issue has not received much attention in the context of occupational environments such as restaurants (Tedd, Liyanarachchi et al. 2001). This study examined the potential impacts of such a switch by 1) examining secondary data on stove X and LPG stove emissions (primarily based on measurement of household stove versions), 2) directly measuring the commercial version of stove X’s direct emissions and comparing them to the commercial grade LPG stove that is commonly used in restaurants, 3) analyzing users’ and owners’ perceptions of the smokiness of these two types of stoves, and 4) examining the selfreported health symptoms associated with IAP and comparing these symptoms between the two kitchen types. The secondary data indicates a much higher emission rate for stove X in particular and ICSs in general in comparison to LPG stoves. The analysis of direct emission measurements shows that stove X has a higher emission factor, particularly when its emission factor is calculated based on the per energy content of fuel. As discussed in the previous section, both owners and users considered stove X significantly smokier than LPG stoves. Finally, the correlation between two health symptoms associated with IAP (i.e. eye irritation and coughing) and stove X usage was investigated through binary logistic models that also considered other potentially influential factors, including 167  Chapter 6  respondent age, time spent by the stove, number of meals cooked in the kitchen per day, volume of kitchen space, number of exhaust fans, and whether respondents smoke. The analysis shows a significant association between eye irritation and stove X usage, while no significant correlation was found in regard to coughing. In brief, our study indicates that switching from LPG to stove X (ICS) in commercial kitchens can have a potentially adverse impact on the health of the workers in these kitchens. There is enough evidence pointing out the higher levels of IAP in kitchens that use stove X in comparison to kitchens that solely use LPG stoves to see that IAP may have harmful effects on the health of individuals spending time in these kitchens. However, it is necessary to confirm the extent of these impacts through a more comprehensive exposure assessment. 6.2  Limitations and direction of future research  My dissertation, which investigates the energy transition in the developing world is an important step toward improving current understanding of this process. However, by exploring understudied areas, it has ultimately as many questions as answers. While all the chapters identify gaps in the literature and provide some answers based on both primary and secondary data, they also further point out understudied areas and signal the need for further research. In addition to presenting a comprehensive literature review on energy system adoption and use, chapter 2 identifies a number of knowledge gaps in the literature. Further, the conceptual framework proposed at the end of chapter can be used as a theoretical basis for further development of research in this field; applying such theoretical framework to an empirical model requires further research. Various factors that influence whether or not individuals adopt the three interconnected energy system components (i.e. energy service, energy device, energy carrier) are conceptually described. Mixed research methods that meaningfully integrate quantitative and qualitative methods from different disciplines are necessary and the framework points to the data that would need to be collected for an integrated assessment. The chapter calls for deeper integration of the human dimension of energy systems into current models and mindsets in order to 168  Chapter 6  formulate programs and policies that promote the transition toward cleaner and more efficient (i.e. modern) energy systems. External conditions – such as programs and policies that address energy provision – were identified as one of the major factors affecting individual decisions on energy system adoption. In chapter 2, the private sector based approach, which is currently a major trend,was explored through a case study in India. The study improved our understanding of the promises and drawbacks to taking a private sector based approach to rural energy provision. To extend our findings beyond this specific case, it will be necessary to undertake a meta-study of case studies on how market-oriented approaches can provide rural regions of the developing world with energy as well as other public goods; that is necessary for evaluating the potential drawbacks of using the private sector based approach to provide services, particularly those that are closely associated with human welfare. Additionally, our study points out many important issues that affect private sector based approaches to rural energy provision. Among these issues, some are unique to India, some are unique to the regions in which the businesses examined operate, and some are unique to the specific industry that was studied. Further research is required to further isolate the factors that are unique to biomass-based energy provision businesses (e.g. electrification, clean cooking fuel provision) and to identify issues that are universal to India, the developing world, or to tech-oriented businesses in the developing world. Enterprises’ shifting focus from rural, residential consumers to urban, commercial customers has important policy implications and provides a unique opportunity to investigate and compare the dynamics of how these two distinct customer segments adopt energy systems. Chapter 4 discusses the adoption of ICSs by commercial kitchens in urban centers, an understudied area in energy transition; this research created grounds for further study in this area. By focusing on the human dimension of stove choice, which extends beyond socio-economic factors, we studied how stove attributes are perceived by users. However, similar research on a larger sample size would be beneficial. Incorporating various 169  Chapter 6  external factors could also further our understanding of the interlinked factors that influence energy system adoption as outlined in the framework from Chapter 2. The switching back phenomenon not only has very important policy implications, but also has not been very well studied. Rising energy prices, coupled with the availability of cheaper and dirtier fuel, yet bearably convenient, as is the case for advanced ICS in commercial kitchens, may trigger a larger-scale shift toward switching back; this possibility needs to be carefully studied. The prominence of economic propositions could also provide a foundation for further interesting research. For instance, understanding the extent to which fuel expenditure differences would result in fuel costs overtaking other energy system attributes is an area worthy of research. Commercial entities in developing countries have also not received much attention in comparison to the residential segment. However, a better understanding of the dynamics of energy system choices by commercial entities that has significant socio-economic as well as environmental implications is necessary. The implications that such a switch could have on the health of stove users is explored in chapter 5. In chapter 5, a mixed method approach was used to identify the potential health implications of switching from LPG to ICSs. These implications were studied through the physical measurement of a few harmful pollutants, users’ perception of the smokiness of the space where they work, and the self-reported health symptoms of those exposed to IAP. The study was constrained by financial and logistical limitations. A more accurate method for physical measurement would be to assess the concentration of a wider range of harmful pollutants in an actual kitchen in working condition. Further exposure assessment could be done to provide a better estimate of exposure to IAP among workers. A larger sample size with more qualitative interviews could also provide a deeper understanding of the perception of smokiness by the users and the people who are working in the spaces with high IAP. Measurement of the health symptoms associated with exposure to IAP could also be much improved through larger sample sizes, inclusion of more symptoms, and medical examinations (e.g. blood pressure). In brief, this dissertation created new grounds for future research while digging deeper into understudied areas of the energy transition literature. Nevertheless, the findings of 170  Chapter 6  this research contribute significantly to the knowledge necessary for formulating wellinformed – and more efficient – policies and programs. 6.3  Incorporating the research findings into policies and programs  Despite wide recognition of energy’s key role in human welfare, as well as decades of efforts to provide universal access to sufficient, reliable, and clean energy services, the situation is far from satisfactory(Goldemberg, Johansson et al. 2004, Bailis and Hyman 2011). Formulating and implementing policies and programs that promote transitioning from traditional energy systems to modern ones is expected to not only improve the living conditions of almost 2.6 billion people (IEA 2012), but also to have significant positive environmental implications(Agarwal 1986, Smith 1987, Barnes and Floor 1996, Bruce, Perez-Padilla et al. 2000, Cecelski 2002, IEA 2002, Rukato 2002, ESMAP 2003, ESMAP 2004, IEA 2006). Today, this transition is a top priority for many governments and development agencies. Improving our understanding of the various dimensions of energy system adoption, the drivers affecting it, and its implications for energy users is necessary. One of the main objectives of this dissertation as a whole was to identify knowledge gaps and improve our understanding of the process of energy transition and its implications in order to formulate better-informed policies and programs. Various policy implications of the research are discussed in each chapter. Some of the overarching concerns that need to be incorporated into policies were identified by looking at different dimensions of energy transition in the developing world. This dissertation calls for: 1. Greater emphasis on the human dimension of energy systems and the incorporation of this dimension into policies and programs through an integrated framework that considers the various factors affecting decisions on energy system uptake. In both chapter 2 and chapter 4, the significance of the human dimension of energy system uptake is demonstrated. Even the technical attributes of energy systems can be perceived differently by users and can ultimately influence their decisions. The split incentive in energy choice in both commercial and residential consumer segments needs to be included in the analysis. 171  Chapter 6  2. Moving toward strong public and private collaboration is vital. Neither private sector based nor state-led models can achieve the goal of universal access to modern energy services on their own. Collaboration between the private and public sectors is vital for the successful formulation and implementation of energy provision programs and policies. This will require careful consideration of the differing consumer segments (commercial versus residential and differing levels of incomes within those two classes), as well as appropriate regulatory models that ensure social goals are being met. 3. Many of the issues limiting the success of these programs and policies can only be addressed by long-term dedication and large amounts of capital, both of which are often beyond the capacity of developing countries. A long-term financial commitment and close collaboration between the industrialized and developing worlds is necessary if effective programs are to be implemented instead of oneshot projects. 4. Although a large body of literature is focused on energy transition, our understanding of the various dimensions of this complicated issue is lacking. In order to conduct comprehensive research and use the findings from that research to formulate policies and programs, both research and development communities and also various disciplines must become better integrated.  172  References  References  Agarwal, Bina (1983). "Diffusion of Rural Innovations: Some Analytical Issues and the Case of Wood-Burning Stoves." World Development 11(4): 359-376. Agarwal, Bina (1986). Cold Hearths and Barren Slopes. New Delhi, Allied Publishers private limited. Agresti, Alan and Barbara Finlay (2009). Statistical Methods for the Social Sciences (Fourth Edition). New Jersey, PEARSON, Prentice Hall. Ahmed, Waquar (2006). "Global Discourses and Local Politics in the Production of Power Policy in India." Development 49(3). Ajzen, Icek (1991). "The Theory of Planned Behavior." Organizational Behavior and Human Decision Processes 50(2): 179-211. Alam, M. S., B. K. Bala and A. M. Z. Huq (1997). "Simulation of Integrated Rural Energy System for Farming in Bangladesh." Energy 22(6): 591-599. Albalak, R., A. R. Frisancho and G. J. Keeler (1999). "Domestic Biomass Fuel Combustion and Chronic Bronchitis in Two Rural Bolivian Villages." Thorax 54(11): 1004-1008. Amacher, Gregory S., Lire Ersado, Donald Leo Grebner and William F. Hyde (2004). "Disease, Microdams and Natural Resources in Tigray, Ethiopia: Impacts on Productivity and Labour Supplies." Journal of Development Studies 40(6): 122145. Anderson, Dennis (1982). "Small Industry in Developing Countries: A Discussion of Issues." World Development 10(11): 913-948. Arbex, MA, LC Martins, RC de Oliveira, LA Pereira, FF Arbex, JE Cancado, PH Saldiva and AL Braga (2007). "Air Pollution from Biomass Burning and Asthma Hospital Admissions in a Sugar Cane Plantation Area in Brazil." Journal of Epidemiology and Community Health 61(5): 395-400. Armendariz-Arnez, Cynthia, Rufus D. Edwards, Michael Johnson, Irma A. Rosas, F. Espinosa and Omar R. Masera (2010). "Indoor Particle Size Distributions in Hom