ts"~~~~~~~~~~~~~1 r. ~~~ ~ ~tE - - -'7~|- 9s ;r B = ^~ ZX L E E~~~ ENERGY SIrCiOR MANAGEMENT ASSISTANCE PROGRAM PURPOSE The World Bank/UNDP/Bilateral Aid Energy Sector Management Assistance Program (ESMAP) was launched in 1983 to complement the Energy Assessment Program which had been established three years earlier. The Assessment Program was designed to identify the most serious energy problems facing some 70 developing countries and to propose remedial action. ESMAP was conceived, in part, as a preinvestment facility to help implement recommendations made during the course of assessment. Today ESMAP is carrying out preinvestment and prefeasibility activities in about 60 countries and is providing a wide range of institutional and policy advice. The program plays a significant role in the overall international effort to provide technical assistance to the energy sector of developing countries. It attempts to strengthen the impact of bilateral and multilateral resources and private sector investment. The findings and recommendations emerging from ESMAP country activities provide governments, donors, and potential investors with the information needed to identify economically and environmentally sound energy projects and to accelerate their preparation and implementation. ESMAP's policy and research work analyzing cross-country trends and issues in specific energy subsectors make an important contribution in highlighting critical problems and suggesting solutions. ESMAP's operational activities are managed by three units within the Energy Strategy Management and Assessment Division of the Industry and Energy Department at the World Bank. - The Energy Efficiency and Strategy Unit engages in energy assessments addressing institutional, financial, and policy issues, design of sector strategies, the strengthening of energy sector enterprises and sector management, the defining of investment programs, efficiency improvements in energy supply, and energy use, training and research. - The Household and Renewable Energy Unit addresses !ech-nical, economic, financial, institutional and policy issues in the areas of energy use by urban and rural households and small industries, and includes traditional and modern fuel supplies, prefeasibility studies, pilot activities, technology assessments, seminars and workshops, and policy and research work. - The Natural Gas Development Unit addresses gas issues and promotes the development and use of natural gas in developing countries through preinvestment work, formulating natural gas development and related environmental strategies, and research. FUNDING The ESMAP Program is a major international effort supported by the World Bank, the United Nations Development Programme, and Bilateral Aid from a number of countries including Australia, Belgium, Canada, Denmark, Finland, France, Iceland, Ireland, Italy, Japan, the Netherlands, New Zealand, Norway, Portugal, Sweden, Switzerland, the United Kingdom, and the United States. FURTHER INFORMATION For further information or copies of the completed ESMAP reports listed at the end of this document, contact: Energy Strategy Management OR Division for Global and Interregional Programs and Assessment Division United Nations Development Programme Industry and Energy Department One United Nations Plaza The World Bank New York, NY 10017 1818 H Street N.W. USA Washington, D.C. USA 20433 IDONESIA URBAN HOUSEHOID ENERGY STRATEGY FEBRUARY 1990 Household Enera Unit Industry and Energy Department The World Bank Washington, DC., USA. This document has a resticted distibution. Its contents may not be disclosed without authoriztion from the Government, the World Bank or the ULDP. ABDREVIATIONS AND ACRONYM BPS Biro Pusat Statsi ¶Central Bureau of Statistics) BAPPES Badan P_n n enn an Nasonal (Natonal Pho sin and Dlopment Board) BAKOREN Badan dn_s Eneri (Minil Energy Coordination Board) DJLED Direktorat lenderal Listrik darn Energa Bw (Direorate Gonera of Eecit and New Energy) LEMIGAS Lembaga Minyak dan Gas (ui and Gas Researh Institute) LKE Liters of Kerosene Equialent MIGAS Diretorate nral of Oi and Ga ME Msy of Mie and Enery NUDS National Urban Development StrateVy Peutamina Perwsahaun Tamban Micyk Nepm (National Oil Company) PLN Perusah_n Umum Listrk Nepa (National Electricity Company) PLN.LMK National Electricy Compan Laboratory and Testing Facility MTE Tedinical Committe on Eney SUSENAS National Soil and Economic Survey UHESS Urban Household Energy Strate% Study YLT Yayasan Lembega Konsumen Indonesia (Indonesian Consumers Union) k thousands kWh KiloWatt hour CWh GigaWa hour m milion ml Megaloules TPA tons per annum EXCHANGE RATE 1700 Rupiah = US$ 1 ENERGY CONVERSION FACIORS MJ BOE LE A/ Kerosene 35.2/l 5.99/kd 1.0 LPG 45.77/kg 7.79/ton 1.69/kg Wood/Biomass 14/kg 2.38/ton 0.235/kg Charcoal 25/kg 4.25/ton Electricity (final) 3.6/kWh 0.61/M-Wh 0.133/kwh (primaty) 10.68/kWh 1.82/MWh 1 BOE = 1.4 million kcal = 5.56 million BTU = 5.87 GJ / Whereas, MJ and BOE are measures of heat content, liters of kerosene equivalent (LKE) in the cooking end-use was Jerived from survey data and expresses the quantity of each fuel required to meet the same cooking demand as one liter of kerosene, according to the behavior of urban households on Java. Because electicity is mainly used in rice cookers its substitution ratio could not be calculated, but was assumed to be the same as the ratio for LPG. SUPPORTING DOCUMENTS Final Report on Urban Household Energy Survey - Biro Pusat Statistik Tests and Evaluation of Kerosene and LPG Stoves - LEMIGAS Kerosene and LPG Stoves in Indonesia: Performance, Safety Aspects, and Improvements - Ernst Sangen Kerosene and LPG Distribution and the Economics of Substitution - LEMIGAS UHESS Database Development Report - LEMIGAS Efficient Household Electricity Use in Indonesia - Dr. Lee Schipper, Lawrence Berkeley Laboratory Urban Household Kerosene Consumption: A Sociological Interpretation - Dr. Kisdaijono and Ir. Saptari Changing Patterns of Household Energy Consumption: an Econometric Analysis - Dr. Aris Ananta PREFACE This report is based on the work of an Urban Household Strategy Study team Which was resident in Indonesia for 15 months from September, 1987 to November, 1988. The study was administered by the Joint World Bank/UNDP/ Bilateral Aid Energy Sector Management Assistance Program in cooperation with the Directorate General of Electricity and New Energy of the Government of Indonesia. Field work was coordinated by the resident study manager Gordon Mcoranahan, Ph.D. Dr. McGranahan was capably assisted in study design, execution, and data analysis by Frans Nieuwenhout who was seconded to the study by the government of the Netherlands. The study was carried out in the offices of LEMIGAS. Dr. Umar Said and Chairul Aswan made major contributions not only to analysis of petroleum supply and distribution, kerosene stove improvement work, and database development, but also to the progress of the study as a whole. Thanks to the professional support of BPS, and notably Mr. Suwandhi and Mr. Aritonang, a survey instrument was developed, enumerated, and encoded that resulted in a database which yields clear insight into urban residential energy use patterns. The study team enjoyed the support and guidance of DJLEB during all phases of the study, in particular of Prof. Dr. Arismunandar, Director General of Electricity and New Energy and of Dr. A. J. Surjadi, Director of New Energy Development, DJLEB. Special thanks goes to the PFE subcommittee on household energy which provided a formal channel for study results to enter the energy policy making process. The study team was supported by a number of consultants during the course of its work: Dr. Aris Ananta (Econometrician, University of Indonesia), Martin Bussink (Economist), Dr. Kisdarijono and Ari Saptari (Sociologist Team, Bandung Institute of Technology), Ernst Sangen (Stove Expert), and Dr. Lee Schipper (Electric Appliance fficiency Specialist, Lawrence Berkeley Laboratoiy). The study was managed by ESMAP staff including Kevin Fitzgerald (Task Manager), Willem Floor (Senior Energy Planner), and Joe Leitmann (Initial Task Manager). This activity completion report was written by Gordon McGranahan and Kevin Fitzgerald. TABLE OF CONTE NTS SUMMARY ...................................................... . i The Context of Energy Policy in Indonesia ................................... i Fuel Distribution Systems and Policy ..................................... ii Fuel End-Use Patterns . .............................................. v Urban Household Energy Strategy ............... .......................... ix Substitution of LPG for Kerosene ....................................... viii Kerosene Stove Improvement Pilot Program ............................... xi End-Use Electricity Conservation Program ................................. xiii Information Base for Further Policy Development ........................... xv Rural Household Energy Strategy Study ................................... xv Summary of Recommendations ........................................... xvi 1. INTRODUCTION ......- .... ... 1 The Economic Situation . ............................................ 1 Energy Policy for the Residential Sector and REPELITA V ..... ..... ........ I The Importance of the Urban Residential Sector in Energy Planning ..... ....... 3 Objectives of the Urban Household Energy Strategy Study .................... 3 H. FUEL SUPPLY AND DISTRIBUTION ................................... 6 Kerosene ....................................................... 6 Liquified Petroleum Gas ................... ......................... 11 Electricity ..................................................... 15 Wood and Charcoal . .............................................. 18 HI. URBAN HOUSEHOLD ENERGY USE PAlTERNS ........................ 21 Urban Household Fuel Use - All Indonesia: 1981 to 1987 ................... 21 Urban Residential Fuel Consumption and End Use - Java 1988 ..... .......... 25 Electricity ................................................. 26 Kerosene ................................................. 28 Liquid Petroleum Gas . ........................................... 28 Wood and Crop Residues . ........................................ 30 Cooking .................................................... 30 Lighting .................................................... 37 Urban Household Fuel Use Projections .......... ....................... 39 IV. URBAN HOUSEHOLD ENERGY STRATEGY .43 Substituting LPG For Kerosene .43 LPG Substitution Scenarios .46 Price Reform .48 Other Measures to Stimulate Substitution of LPG for Kerosene .50 Equipment Cost .50 Education-Promotion .52 LPG Promotion Program Costs and Institutional Responsibility .52 Kerosene Stove Improvement Pilot Program .53 The Context of Kerosene Stove Improvements in Indonesia .54 Technical Modifications .54 Sociological Dimensions of Household Stove Choice .56 Kerosene Stove Production and Marketing .57 Kerosene Stove Improvement Pilot Program Design ..................... 58 Program Economic Evaluation ................................... 59 End-use Electricity Conservation Program ............................... 60 Electricity Conservation Program Design .............................. 62 Information Intervention ....................................... 63 Technical Interventions ........................................ 64 Market Adjustment ... 64 Economic Evaluation of the Program ..................................... 64 Institutional Arrangements ........................................ 66 TABLES Table 2.1: Kerosene Sales to Urban Households by Region 1981, 1984, and 1987 ...... .. 7 Table 2.2: Economic and Financial Costs of Urban Kerosene Supply .......... .... . 9 Table 2.3: Kerosene Price by Source ........................................ 11 Table 2.4: Economic and Financial Costs of Urban LPG Supply ........ ........... 12 Table 2.5: LPG Cost Estimates Before and After 1987 Price Increase ..... .......... 13 Table 2.6: PLN Residential Electricity Tariffs ............. .. .................. 16 Table 3.1: Urban Household Energy Use Patterns by Island Group, 1987 .... ........ 24 Table 3.2: Percentage of Urban Households Using Each Fuel by Income ..... ........ 26 Table 3.3: Urban Household Fuel Use by Income .. 27 Table 3.4: Percentage of Households Using Each Fuel by Urban Area Size ..... ...... 29 Table 3.5: Household Fuel Use by Urban Area Size ...... ...................... 29 Table 3.6: Attitudes Toward Kerosene for Cooking ............................. 31 Table 3.7: Attitudes Toward LPG .......................................... 31 Table 3.8: Attitudes Toward Wood ......................................... 32 Table 3.9: Lighting Electricity Use by Lamp Type (% of kWh) ...... .............. 39 Table 3.10: Base Case Fuel Use Projections for the Urban Residential Sector ... ....... 40 Table 4.1: Economic Effect of Substituting LPG for One Liter of Kerosene .... ....... 44 Table 4.2: Daily Cooking Cost Comparison of Kerosene and LPG ....... .......... 45 Table 4.3: LPG Substitution Scenarios ............... ..................... .. 47 Table 4.4: Evaluation of LPG Promotion Scenarios ............................. 48 Table 4.5: Relative and Absolute Value of Kerosene and LPG Use by Income .. ...... 49 Table 4.6: Summary Evaluation of Kerosene Stove Pilot Program ....... ........... 60 Table 4.7: Outline of Electric Appliance Conservation Program ......... ........... 62 Table 4.8: Program Targets for 1999 ................... ..................... 65 Table 4.9: Summary Evaluation of Urban Residential Electricity Conservation Program . . 66 Table A3.1: Percentage of Households Using Fuel by Income and Year ..... .......... 72 Table A3.2: Average Combustible Fuel Use per Household by Fuel, Income, and Year . . . 72 Table A3.3: Average Fuel Use per Using Household by Income and Year ..... ........ 73 Table A4.1: Cooking Fuel Use by Income and Urban Area Size .................... 80 Table A4.2: Lighting Practices and Income - Java 1988 ........................... 82 Table A5.1: Daily Cooking Cost Comparison of Kerosene and LPG ..... ............ 85 Table A6.1: Cooking Practices and Income - Java 1988 ........................... 86 Table A6.2: Principal Cooking Utenisils ....................................... 88 Table A7.1: Correlation of Income, Family Size, and Urban Area Size ..... ........... 95 FIGURES Figure 1.1: National Conventional Ene;gy Consumption by Sector ...... ............ 4 Figure 2.1: Shares of National Fuel Use Consumed in Urban Households ..... ....... 6 Figure 2.2: Domestic Depot Kerosene Price 1973 - 1987 ........ ................. 8 Figure 2.3: Representative System Load Curve (Java) ......... ................. 15 Figure 2A: Electricity Supply Source ....................................... 17 Figure 2.5: Fuelwood Supply Source ....................................... 19 Figure 3.1: Percent of Urban Househulds Using Each Fuel: 1981 - 1987 ..... ........ 22 Figure 3.2: Urban Residential Combustible Fuel Use: 1981 - 1987 ............... .. 22 Figure 3.3: Percent of Household Expenditures on Fuel: 1981 - 1987 ...... ......... 23 Figure 3.4: Cooking Fuel Shares by Income ................ .................. 35 Figure 3.5: Cooking Fuel Shares by City Size .............. ................... 35 Figure 3.6: Cooking Fuel Use by Income . ................................... 36 Figure 3.7: Cooking Fuel Use by City Size ............... .............. 36 Figure 3.8: Base Case Combustible Fuel Projections ........... ................ 40 Figure 3.9: UHESS and PLN Urban Electricity Demand Projections ...... ......... 41 Figure A7.1: Distribution of Households by Expenditure Category .92 Figure A7.2: Flow Chart: Cooking Fuel Demand Model .93 FigureA7.3: Logistic Curve.95 Figure A7.4: Logit Transformation of Cooking Fuel Shares by Household Income .96 Figure A7.5: Logit Transformation of Cooking Fuel Shares by Urban area Size .97 Figure A7.6: Projections of Official and Informal Urban Electrification ............... 106 Figure A7.7: Projections of Lighting Fuel Use .. . 107 TEXT BOXES Box 3.1: Derivation of "Liters Kerosene Equivalent" for Cooking Fuels ... 34 Box 3.2: Lighting Analysis: Electric.ity Substitution for Kerosene .. . 38 ANNEXES Annex I Petroleum Product Prices and Sales .......... ................... 68 Annex II Distribution Systems for Kerosene and LPG ....... ............... 69 Annex Im Analysis of Kerosene Price Changes and Urban Electrification ..... .... 71 Annex IV Derivation of Liters Kerosene Equivalent for Cooking and Lighting .................. ...................... 75 Annex V Cooking Cost Comparisons: Kerosene and LPG ....... ............. 85 Annex VI Cooking lmplements and Practices .............. ................ 86 Annex VI Urban Residential Energy Projections ........... ................ 91 AnnexVlI LPG Promotion Scenarios .................................... 108 Annex IX Draft Energy Module for Susenas .............. ................ 113 Annex X Rural Household Energy Strategy Study Proposal ....... ............ 115 Annex XI Economic Evaluation of Kerosene Stove Pilot Program ...... ........ 116 Annex XII Economic Evaluation of Electricity Conservation Program ..... ....... 117 MAPS IBRD 21884 1988 Urban Housenold Energy Survey IBRD 21889 Population Urban By Major Island Groups IBRD 21883 LPG Facilities APPENDICES Appendix I Urban Household Energy Survey Questionnaire and Summary Tables Appendix II Kerosene Stove Improvements Program Appendix III Electric Appliance Efficiency Program SUMMARY The Context of Energy Policy in Indonesia 1. In the 1970's and early 1980's, Indonesian energy policy was formulated in an environment of high international oil prices. Petroleum products such as kerosene and diesel oil for domestic consumption were subsidized as a means of sharing some of the profits from high export prices with the population and to stimulate productive enterprise. Kerosene was priced low to: i) provide a substitute for fuelwood and thereby reduce deforestation; ii) provide a reliable supply of fuel for the poor at affordable prices, and; ill) to be made available to all residents at the same price in every region to serve equity objectives. During this period of low domestic fuel prices, technologies were employed and patterns of use developed without due emphasis on energy effciency. 2. In an attempt to insulate the economy and the government budget from external shocks due to fluctuations in international oil prices, the government initiated a series of reforms in 1983 designed to stimulate non-oil exports through deregulation. Due to the fall of international oil prices in early 1986, net oil and LNG export earnings fell from US$ 12.3 billion in 1985/86 to US$ 8.6 billion in 1987/88. Over the same period, non-oil merchandise exports soared from US$ 6.2 billion to US$ 9.5 billion. 3. Changes in energy policy since 1978 supported national objectives and the reforms initiated in 1983. For the residential sector: kerosene prices were tripled in real terms between 1981 and 1984 and urban electrification was ambitiously pursued resulting in the percent of urban households electrified nationwide growing from 41% in 1981 to 74% in 1987. These measures were taken for a number of purposes including: i) reducing government expenditures on the kerosene price subsidy and; ii) displacing kerosene used for lighting, thereby making more kerosene available for export and generation of needed foreign exchange. These two policies resulted in a decline in kerosene use nationwide of about 2% per year from 1981 to 1987. Nonetheless, by 1987, urban households still consumed about 40% of national kerosene sales, as well as similar shares of LPG and electricity. 4. In the current five year plan, the government continues to stress the objectives from the past five years in energy: diversification of domestic energy use away from exportable petroleum products and the initiation of programs designed to encourage energy conservation. In an effort to support these objectives, the office of the Director General of Electricity and New Energy (DJLEB) requested ESMAP assistance in studying residential fuel use patterns and recommending a consistent strategy for the sector that would meet basic fuel needs while effilciently allocating domestic resources. As a study of rural energy use was underway when ESMAP assistance was requested, this study focusses on the urban residential sector. This report presents the major findings and recommendations of the ESMAP/DJLEB Urban Household Energy Strategy Study. The study benefitted greatly from the competent support and substantial technical analysis provided by LEMIGAS, BPS, and DJLEB. -U- Fuel Distribution Systems and Policy 5. One of the objectives of Indonesian energy policy is to ensure that an adequate supply of fuel is made available, at affordable prices, to meet the basic needs of Indonesian households. This forms part of the justification for: 1) requiring Pertamina to ensure adequate supplies of kerosene (and most other petroleum products) in all parts of Indonesia; 2) supporting a rapid electrification program; 3) subsidizing kerosene, and; 4) subsidizing low-power electricity users. Neither wood nor LPG receive commensurate support. LPG is a fuel of upper income households. Though wood is a fuel of low income households, neither prices nor supply are controlled by the central government. 6. Part of this objective has come close to being met. As mentioned above, there has been rapid urban electrification in recent years, and kerosene is widely available in urban areas at close-to-uniform prices. An energy budget of 300 Rupiah a day is sufficient to meet the minimal energy needs of an urban household in Indonesia. This level of expenditure is undoubtedly a burden on the poor. However, the urban households which still do not meet their basic energy needs are suffering from poverty, rather than an energy shortage in particular. Subsidizing kerosene and electricity sold through low power connections is not an economically efficient means of combatting poverty. Regardless, being both visible and simple, it is likely to remain a politically attractive option. 7. Kerosene price increases and electrification between 1981 and 1987 not only resulted in reductions in kerosene use, but were also accompanied by an increase of 50% in the average share of household expenditures devoted to fuel purchases over the period. Even though average Percent of Urban Household Expenditures by Expenditure Group and Year % of Household Expenditures by Fuel LOw- MMLPG 12%- _ Electricity Moderate E Charcoal 8% - Kerosene = woo 4% High 1%981C 1984' 1987u 1981 1984 1987 1981 19841 1987 Soitrce: SUSENAS 8l, '84. '87. household kerosene use by 1987 fell to 65% of 1981 levels and wood fuel use increased slightly, the poorest 20% of urban households spent over 10% of their budgets on fuel in 1987. 8. An annual update on distribution costs by region, accompanying a more general assessment of the opportunity costs of various fuels, would be very beneficial for future policy formulation. Kerosene provides the mainstay of cooking fuel needs for urban households in Indonesia. While kerosene is widely available, there is insufficient information within the government regarding distribution costs for accurate policy analysis. One of the advantages of having fuels distributed by a state company, rather than the private sector, should be ready access to such information. The official estimate of average distribution costs for kerosene has not changed for several years, and in any case masks regional variation. Using Third Quarter 1989 Singapore posted prices, it is estimated that kerosene is still priced at, on average, 70% of the cost of supply. 9. The government should closely monitor kerosene demand to ensure that adequate supplies are made available from low cost domestic or imported sources. Kerosene use projections based on 1988 patterns of household energy use and growth of income and population, indicate that kerosene use will continue to grow slowly through the end of the 20th century, meeting the main combustible fuel needs of urban dwellers in Indonesia. Based on declining kerosene consumption since 1981, the current five year plan projects kerosene demand to remain steady through 1994 and, consequently, no new investments in refinery capacity for distillation of kerosene are planned for the period. If the projected income growth assumptions are realized, moderate growth in kerosene use can be expected. 10. LPG is widely available in urban Java, but less so in the rest of Indonesia. Despite the relative flexibility Pertamina has over LPG pricing and distribution, a uniform LPG price is maintained nation-wide. This policy of uniform national pricing is difficult to justify as LPG is a fuel of upper income households. Since transport and distribution costs are higher for LPG than for other petroleum products, uniform pricing leads to greater distortion when applied to LPG. The average LPG price is above the estimated economic cost of supply by approximately 38%. Thus, the price/cost difference is greatest where transport and distribution costs are low. Alternatively, since there is no directive to supply LPG to all parts of Indonesia, the distortion is likely to include supply inadequacies in areas where the national price does not cover costs. By allowing regional variation in LPG prices to reflect different distribution costs, these economic distortions could be avoided, without placing undue burden on lower income households. To ensure that prices reflect actual distribution costs, the distribution system will require periodic auditing. 11. Recent investment in additional LPG production facilities provide Indonesia an opportunity to become a major exporter of LPG. Nonetheless, LPG remains a relatively minor domestic fuel: in energy terms, total domestic LPG sales in 1987 were less than 5% of national kerosene sales. Replita V explicitly supports investments in LPG capacity expansion to provide a high quality fuel .o the domestic market. Installed LPG capacity should surpass 3 MPTA in the early 1990s. 12. Though much of Indonesia's LPG production is committed for export under long- term contracts, the distribution system appears to place a more serious constraint on residential LPG growth than production capacity. Developing a technically adequate distribution system for - iv - LPG has been difficult given the extremely rapid sales growth (35% annual growth in sales to urban households from 1981 to 1987). Increased private sector involvement is an appropriate response, though care must be taken to ensure safety standards are met. The recent decision to develop a series of mini-filling-plants, to be operated privately, should improve the economics and enhance the capacity of the LPG distribution system. It is also a more economically efficient response to distribution inadequacies than LPG price increases. A majority of households not using LPG cite fuel and equipment cost as the major reason for non-use. As one result of a perceived LPG bottle shortage, the survey undertaken for this study found that the average price for bottles on Java was 40% above the official price. The production and marketing of small bottles and other ways of reducing equipment costs to households as a means to stimulate LPG use are discussed in Chapter IV. 13. Widespread access to electricity among low income urban households is in part the result of shared meters (of the 85% of households in urban Java using electricity, fully 25% obtain electricity from their metered neighbor though the low average demand of these households results in their consuming only 6% of total urban residential electricity). Shared metering deserves monitoring, but not discouragement. Electric fires are reportedly the major cause of urban fires, but there is as yet no evidence that shared metering contributes significantly. In any case, the projections indicate that shared metering is likely to decline in importance over the coming years. In short, shared metering should be acknowledged by PLN and viewed as an opportunity to enhance PLN's service. 14. Electricity tariffs were recently revised for the first time in four years. The new tariff structure moves toward recovery of the costs of supply, with higher power consumers cross- subsidizing low power consumers on energy charges, but with all consumers still paying well below estimates of capacity costs for their service. Urban electrification in the 1980's has been very successful resulting in rapid growth of percent of households electrified. However, because a significant share of urban households are electrified informally, new urban connections in the 1990's will likely have less influence on raising demand than was evident in the 1980's. Nonetheless, because informally connected households consume far less, on average, than officially connected households, both income growth and connection policy will be important in stimulating electricity use in the next ten years. 15. Wood and other biofuels have not received the full attention they deserve in this study. This was due in part to the fact that previous urban household energy surveys suggested that wood fuel use was rare in urban areas (a result of the bias towards very large cities in virtually all previous surveys). Most indications are that, at least in Java, the wood-fuel system is not only very significant, but is operating effectively. Almost 1/4 of urban households on Java use wood, more than 2/3 of which collect at least some of their wood. Since home-gardens on private land are the most common source of collected wood-fuel, it is important to ensure that government regulations do not undermine the home-garden system. The impact of regulations forbidding development on sawah land, for example, should be re-assessed. There may also be opportunities for combining green-belt environmental policy, with enhanced urban wood-fuel supplies. On a small scale, projects undertaken on government land near Bandung indicate considerable potential. Finally, as part of the ongoing Department of Forestry study of fuelwood and rural villagers, forest use by urban dwellers in the smallest towns should be examined. -V. Fuel End-Use Patterns 16. Before this ESMAP/DJLEB study, several surveys and studies of fuel use by rural and urban households had been undertaken. However, they did not result in proposed plans of action regarding policies and programs in which the government might invest to encourage rationalization of fuel use in the sector. The survey results have provided a much improved basis for quantifying these relations. 17. Electricity and LPG use increase sharply with income, and are more commonly used in larger uarban areas. Wood fuel use, almost exclusively for cooking, falls rapidly with income and also with urban area size. Kerosene use rises slowly with income, with increasing cooking demand (especially among lower income groups) moderated by declining kerosene lighting. Charcoal is widely used, but only for minor energy requirement, such as ironing. Cooking and lighting are clearly the most important energy using activities, with lighting accounting for half of electricity use, and cooking dominating the other major end-uses of kerosene, LPG, and wood. 18. Cooking is carried out predominantly with kerosene, with wood also important in low income levels and LPG important at high income levels. Even more than other household energy-using activities, cooking is primarily the responsibility of women. One of the major energy- using activities nation-wide, cooking is also central to the well-being of the population. However, like many women's activities, cooking has received relatively little attention from the policy community. There is little recognition that energy policy affects not only how much households must pay for cooking fuels, but often affects the cooking process itself. 19. The income related patterns of cooking fuel use confirm the expected hierarchy in the quality of cooking fuels, with LPG at the top, kerosene in the middle, and wood at the bottom. Household attitudes toward cooking fuel choice also reflect these preferences. When asked why they do not use fuels higher in this hierarchy, households tend to respond in financial terms. High equipment cost was the major reason given for not using LPG, followed closely by high fuel cost. In-depth sociological surveys uncovered that many households purchase kerosene stoves (at an average price of 5000 Rp) on an installment plan with implicit annual interest rates as high as 150% (much of which is required to cover the transaction costs of collecting small payments). Such severe cash flow constraints make the higher investment costs of LPG use well beyond the reach of most households: for an average household, estimates of daily financial cooking costs with LPG are twice as high as with kerosene when discounted at 20%, and three times as expensive when discounted at 100%. When asked why they do not use fuels lower in the hierarchy, households tend to respond in terms of quality: speed, cleanliness, and ease of use. There is some dissension, however, and even fuelwood has many devotees. 20. In support of the government objective to diversify domestic fuel use, the study team analyzed the potential of LPG substitution for kerosene for cooking in urban households. The UHESS data set afforded a unique opportunity to quantify how much LPG (or wood) is required to substitute for one liter of kerosene for cooking based on the actual behavior of urban households. A statistical procedure, applied to the data to account for variation in income and family size, resulted in the following: for each liter of kerosene used for cooking by an urban household in Indonesia, a similar household cooking with LPG (or wood) would use .6 kg of LPG (or 4.25 kg of wood). These substitution ratios were used to evaluate the economics of LPG substitution I vi I Cooking Fuel Shares by Income Category Urban Java Cooking Fuel Shares Family Size & LKE/ HH/ day 10,000 100,000 1,0,0 7 75%6 50%~~~~~~~~~~~~~~~~ 25% -2 1 10,000 100,000 1,000,000 Monthly Household Expenditures Souce: UIW 198. _, programs and for making cooking fuel use projections. This method, developed by the study team, to quantify fuel substitution based on household behavior overcomes one of the major problems with previous estimation methods: changes in fuel use behavior that may accompany a fuel switch are not captured when the utilized energy (heating value of each fuel converted through an assumed average eficiency) is assumed to remain constant through the fuel switch. This statistical method does not rely on end-use efficiency assumptions at alL For this reason, the substitution estimation method presented herein should be treated as a distinct output of the study, as well as a significant contribution to the literature. 21. The majority of households (85% on Java) use elactrdcity for lighting, and virtually all of the remaining households use kerosene. Electric lighting is superior to kerosene lighting in almost every respect Once a household has an electric connection, electric lighting almost invariably follows. 'Me only barrier to electric lighting is that an electric connection is required. Many low income urban households circumvent this barrier by sharing a metered connection of a neighbor. 22. A simiar statistical procedure to that used to derive cooking fuel substitution ratios was used to anabze the costs and benefits of electricity substitution for kerosene lighting. Results indicate that, on average, urban households using electicity for lighting not only use 1/6 of the energ and spend 30% less on lighting fuel than their kerosene-using counterparts, but also increase lighting levels by roughly nine times. Applied to lighting, this statistical analysis of the behavior of urban households on Java provides a compelling rationale for urban electrification, not only on the grounds of reduced costs, but also due to major increases in lighting. vii- 23. About two thirds of electricity consumption for lighting is used in incandescent lamps, with the rest used in fluorescent lamps. Given the four to six-fold higher efficiency of fluorescent lamps, this implies that most lighting in urban households is actually fluorescent. For households using electricity, the choice between fluorescent and incandescent lighting is less dear- cut than the choice between kerosene and electricity. Fluorescent lamps are considerably more efficient, last longer, and are less affected by voltage fluctuations than incandescent bulbs. Conversely, the ballast and tube for fluorescent lamps are more expensive than incandercent lamps. Given severe cash shortages, this can make incandescent lamps more attractive. Also, the color of fluorescent light is less 'natural! and sometimes felt to be less pleasing. In urban Java, fluorescent lighting provides a larger share of the eectric lighting among upper income than among lower income households, indicating that the color of fluorescent lamps is probably less of a barrier in urban Indonesia than elsewhere. 24. Projections of fuel use, based on 1988 end-use patterns and constant real prices, indicate that even with displacement of kerosene lighting by urban electrification, kerosene use in urban households is expected to grow annually at over 3% through the year 2000, continuing to provide the bulk of combustible fuel needs. Though LPG consumption is projected to grow at over 10% per year, LPG will still meet less than 15% of combustible fuel needs in urban households by the end of the century. Due to LPG affordability constraints in middle income households, the high rates of LPG growth of the 1980s should not be expected to continue through the 1990s. Even wood use is projected to rise through the period, but meet a declining share of urban household fuel needs. Lighting will remain the most impcrtant end-use of electricity in the urban residential sector, though electricity for refrigeration and teleaision, di second and third major end-uses, are exected to grow rapidly. Base Case Combustible Fuel Projections Urban Residential Indonesia Rillions of LKE/Year o SLISENAS Kieroserne 5. ........ ......... .._................... . ° SLSENAS LPG. TotalI Kerosen e , - 4L-- . S U S.ENS.---.----- - ....... ............ _---K eroseri~~~~~see for cookSing 3 2 ............................ ................. . -... --...................... . ..... 1980 1985 1990 1995 2000 Year Sowce: Anhox Vil and Vm. -vi - Urban Household Energy Strategy 25. There appears to be significant scope for diversification and conservation of fuel use in the urban residential sector. Namely: i) stimulating LPG consumption and continuing urban electrification efforts wili serve to displace kerosene use for cooking and lighting, respectively; ii) low cost technical modifications to kerosene stoves, coupled with a cooking practices sensitization campaign and appropriate pricing policy could result in significant kerosene savings, and; iii) a phased program designed to raise the average efficiencies of key electric appliances on the market and encourage the adoption of fluorescent lighting presents a low cost means to reduce peak demand requirements and lead to residential electricity conservation. These components of the proposed urban household energy strategy are designed to reduce government expenditures (principally through displacing and conserving subsidized kerosene) and improve the balance of payments while increasing effective energy services obtained by urban households. Substitution of LPG for Kerosene 26. Using the derived substitution ratio of 0.59 kg of LPG substituting for 1 liter of kerosene in the cooking end-use and cost build-ups from 1989 international prices, substitution of LPG for kerosene shows robust beneficial effects on net economic benefits, the government budget, and balance of payments. Significant improvements in the government budget and balance of payments and marginal economic cost reductions from each liter of kerosene displaced by LPG provide a strong rationale for programs and policy reform designed to stimulate LPG consumption in urban households. 27. The benefits to households of substituting LPG for kerosene lie in the services LPG delivers, not its relative costs. In economic terms, the fuel cost of LPG use by urban households for cooking is only slightly less expensive than kerosene. When the higher costs of LPG equipment are included, urban household LPG use is equal to if not slightly more expensive to the economy than kerosene use. However, the willingness of upper income households to pay nigh prices for LPG fuel and equipment indicate that they apparently value the additional convenience, cleanliness, and flexibility provided by LPG enough to offset the estimated doubling in daily financial cooking costs (over the costs of cooking with kerosene). 28. Fuel use projections were run modelling the effects of price reform and other programs designed to stimulate LPG growth. Of various scenarios, kerosene price reform appears to be the most effective approach toward reducing government expenditures and improving the balance of trade. However, the costs and benefits of price reform would be born and obtained by different segments of the population. Low income households spend a far higher share of their household budget on kerosene than do high income households, while the situation is exactly reversed for LPG. While low to middle-income households consume the highest total amounts of kerosene, over half of all LPG used in the urban residential sector is consumed by households in the highest income decile. Hence, in absolute terms, the burden of a kerosene price increase would fall most heavily on middle income groups, while in relative terms it would fail most heavily on the poor. Conversely, an LPG price reduction would benefit primarily the rich, both absolutely and relatively. Alone, removal of the kerosene subsidy and of the LPG tax, would benefit the urban rich at the cost of the urban middle class and poor. - ix- Economic Effect of Substituting LPG for One Liter of Kerosene Kerosene LPG Net Rpilt Rp/0.59 kg Rp Change in: Economic Benefit M/ -203 348 145 Economic Cost -290 253 -37 ket Economic Benefits 87 95 182 Geverruent Revenue 92 73 165 Ewperts m -97 136 pj Economic Benefit a 1988 retail price. Source: Tables 2.1 and 2.4, and substitution ratio of 0.59 kg LPG per liter of kerosene displaced. 29. The potential effects of other measures designed to reduce initial investment costs in WPG equipment and allay concems over the safety of LPG (two of the most important barriers to LPG choice) were also evaluated with the projection modeL Because elements of such a program would be aimed at new users, the large cooking fuel market of middle income households could be tapped and most of the increase in LPG use would be due to new users switching from kerosene rather than increases by households already using LPG. The net economic benefits of the scenario modelling such a program are as high as the kerosene price reform scenarios, and the budgetary and balance of trade effects are both robustly positive. It is estimated that an ambitious program could incur annualized costs as high as US$ 14.2 million, remain budget neutral, and still have annualized net benefits of US$ 1.6 million. 30. Various means for reducing investment costs in LPG equipment faced by middle income kerosene users were evaluated: i) production of enough 11 kg LPG bottles to meet demand is necessary to ensure that bottles are actually sold at prices close to official prices; ii) the introduction of small (6 kg) bottles would lower initial costs and allow marketing of LPG to more closely resemble current kerosene marketing practices with which all households are familiar; iii) a subsidized equipment package could be offered to new users that would cut household investment costs by 1/3. In addition to these possible elements in an LPG promotion program, Pertamina should provide safety information to consumers through its LPG dealers and add this requirement to its list of rules for LPG dealers. It would also be worthwhile for Pertamina to promote LPG through advertizing targeted to mid-to-upper income consumers (e.g. in the more expensive movie theaters). The advertizing and safety information distributed through dealers also provides an opportunity to emphasize how LPG can be used efficiently. 31. In evaluating price reform or other measures to promote LPG, it should be noted that even with ambitious programs, LPG is likely to remain an upper-income urban fuel in the medium term, while kerosene will continue to be used at all income levels. The policy initiatives evaluated in this study do dot result in LPGN displacing kerosene as the major cooking fuel in urban households Rather, it is shown that various means of stimulating LPG adoption would serve to .X- Relative and Absolute Value of rosene and LPG Use by !ncbm Auawl Fuel U4e In Urban Households on Java Household Expenditures X ot .0H Extn for: prcse LPG (Rupiah/HHINonth) X HH Kerosene LPG (N It) (Rp Nfti7on)i/ ('000 Tons) (Rp Nitlion) ay < 75,000 26.8M 9.0X 0.1X 515 105 1.0 0.65 75,000 - 120,000 23.8X 6.3X 0.2X 570 11S 5.5 3.25 120,000 - 185,000 23.6X 4.8K 0.3X 680 140 14.5 8.60 185,000 - 295,000 15.7K 3.5X 0.6" 490 100 24.5 14.50 ) 295,000 10.2X 1.8X 0.9X m 60 58.5 34.50 All 100.0K S.8X 0.3 2.550 So 101.0 61.50 1/ Valued at 1968 retail prices. Source: UHESS 1988. kuber of urban households an Java in 1988 estimated at 6.58 million. reduce expected growth rates of kerosene consumption. While moderate to upper income households will obtain most of the direct benefits of LPG promotion, the extent to which these households switch from kerosene wilt serve to target government expenditures on the kerosene subsidy to low and middle income households. 32. Pertamina and MIGAS would be responsibe for implementing any program of LPG promotion. As significant potential has been demonstrated for LPG to moderate growth in kerosene consumption in the urban residential sector and, thereby, contribute to government objectives of fuel diversification, export enhancement, and expenditures reduction, the Ministry of Mines and Energy should be involved in an integral fashion in the desig and supervision of policies and programs aimed at displacing kerosene by LPG. Kerosene Stove ImProvement Pilot Progm 33. Kerosene sales accounted for over one quarter of domestic petroleum product sales in 1988, with as much as 45% of this (worth roughbly USS 420 million in export value) consumed in kerosene stoves. Few other energy conversion technologies, if any, convert such large quantities of fuel The results of technical and sociological work undertaken for this study indicate that there are significant opportunities for improving existing kerosene stoves and encouraging more fuel effcient cooking practices. Evaluation of a kerosene stove improvement pilot program, drafted on the basis of this work, exhibits significant potential for improving the balance of payments by reducing demand for kerosene at a very low cost to the government and to households. 34. With regard to percent of household expenditures on kerosene, it is evident that poor households would benefit most from kerosene stove improvements that raise the efficiency of use without raising stove prices sipgificantly. As poor households spend as much as 8% of their budgets on kerosene fuel purchases, they have direct financial incentives to conserve kerosene. By reducing household expenditures on kerosene while increasing effective energy services delivered, a kerosene stove improvements program would also serve to improve the quality of life, especially for women. xi- 35. The challenge facing any kerosene stove program will be in transating the large technological and econowic opportunities into actual improvements in the efficiency of cooldng with kerosene. There are four reasons why the fuel use of kerosene stoves can be expected to be uneconomically high. First, though stoves with a wide range of efficiencies are on the market, buyers have no means of identifying which are more efficient, either through physical inspection or reputation. Second, stove producers do not have sufficient equipment to test the efficiency or power of their product and are unaware of what measures could be taken to improve the fuel economy of their stoves. Even the largest stove producers only capture a small share of the overall market, and have little incentive to fund research efforts. Third, most kerosene stove models were developed and the habits of consumers were acquired without emphasis on fuel efficiency under conditions of heavily subsidized kerosene prices. Fourth, kerosene is still subsidized. In addition to overcoming these barriers, an effective stove improvement program must not result in large increases in stove prices. Severe cash flow constraints on many urban households make consumers extremely averse to higher equipment costs as signaled by credit schemes commonly applied to kerosene stoves. 36. The proposed pilot program, presented in detail in Appendix II, is designed to raise the quality of existing stoves, instead of attempting to identify and promote the "best" stove. This strategy was selected because: i) there are numerous obstacles to marketing a new stove; ii) tastes in stoves vary so that no single stove can satisfy the diverse requirements of consumers; and iii) simple modifications should be able to raise the efficiency of most existing stoves. Technical tests on popular stoves available on the market indicated that modifications required to save kerosene and/or change the power of most existing kerosene stoves are simple, and need not be expensive. The proposed program involves four components: i) factory stove improvement; ii) artisanal stove improvement; iii) kerosene stove research, and; iv) Consumer awareness / advocacy, promoting high performance kerosene stoves and fuel efficient cooking techniques. No radical stove design changes or cooking changes are envisaged. Rather, the program emphasizes minor changes which would almost certainly be implemented given better information. 37. The program design includes the formation of a series of regional stove testing and improvement units, each with sufficient manpower and equipment to test, evaluate and modify locally produced kerosene stoves. During the first eighteen months two such units would operate in two large cities (eg., Jakarta and Surabaya). After this first phase, the number of units in the two areas would double, and four additional units would be established in large cities outside Java. 38. Assuming that "improved" stoves can penetrate at most 25% of the factory stove market and 15% of the artisanal market by 1999, the economic present value of kerosene saved would be about seven times program cost, resultng in annualized net economic benefits of over US$ 2 million over the ten year program. The program would be budget positive and the annualized value of increased net exports due to saved kerosene is estimated at roughly US$ 2 mfllion. Assuming no change in the real retail price of kerosene over a ten year program, net financial savings to households would approach the Rupiah equivalent of US$ 1 million per year. 39. Guidance and monitoring for the energy aspects of the program would be provided by the Directorate General for Electricity and New Energy (DJLEB), based on its prior experience with energy conservation. Brochures outlining actions which can be taken to conserve fuel, expanding on those already printed and distributed by DJLEB, would be an integral part of an -xi - Sumnary Evaluation of Kerosene Stove Pilot Program Not Present Value a 105 Anmualized NPV (US$ Millions) (USS Millions) Kerosene Saved 16.65 2.71 Program cost 2.03 0.33 Increased stove cost 0.21 0.03 Total Costs 2.25 0.37 Net Econmic BenefIts 14.41 2.34 Goverrnant Budget 2.74 0.45 Salane of PaWMents 12.30 2.00 1700 Rupiah a I US Dollar. Econmic Costs of Kerosene Supply (Rupiah/liter): 290 Goverrment Kerosene Subsidy (Rupiah/liter): 92 Export Value (Rupiah/tfter): 233 Conversion factor for goverroent expenditures: 0.8 Foreign exchange component of goverrment expenditures: 43X Source: See Annex XI. intensified information campaign. The Ministry of Industry, with its regional offices, contacts with stove producers, and laboratory facilities for testing stoves, would be responsible for the technical side of the program. The Indonesian Consumer Organization (YLKI) and local women's Education for Family Welfare groups (Pendidikan Kesejahteran Keluarga) would assist in disseminating information, and provide feedback. A research group could be located at one of the technical universities. End-Use Elricit Conservation Program 40. Urban households currently consume approximately 28% of total electricity nationwide. In addition to savings in generation costs, because the residential daily peak demand is coincident with the system peak, a program that effectively encourages the marketplace and consumers to reduce peak demand would also have the effect of reducing or delaying the need for scarce capital resources to be used in power generation capacity expansion. 41. In the medium term, lighting will remain the major electricity end use in the urban residential sector. Because fluorescent lighting is far more efficient than incandescent lighting, a moderate switch to fluorescent lighting holds out the most promise for increasing the overall efficiency of electricity use in urban households. However, electricity consumption for refrigeration, television, air conditioning, and other end-uses will grow rapidly in significance by the year 2000. Consequently, the proposed electricity conservation program concentrates on efficiency improvements in these major end-use devices. 42. A review of the electric appliance market in major urban centers revealed that many locally produced appliances use technologies developed in the 1970's and have not benefitted from advances in efficiency made over the past decade. Hence, there is scope for improvement in the efficiency of the stock of appliances currently used by Indonesian households. However, as with kerosene stoves, there is little relevant information available to consumers on the energy performance, of appliances when making a purchase. No ind&cation of how much electricity the particular appliance can be epected to consume annually or how it compares to others on the market is available. Though the government has waged a campaign reminding consumers to 'save electricity" by labelling certain appliances (Hemat LL0ik), if appropriate information is made available on the financial benefits of efficient devices, consumers may be far more responsive. 43. The proposed conservation program, presented in detail in Appendix m, has three elements that have been designed to supplement previous conservation campaign experience in Indonesia to effectively move toward efficiency improvements in the stock of household electrical appliances. i) Appliances are to be tested and labelled with information on expected operating costs and energy performance relative to other models on the market to allow consumers to make effective purchasing decisions. The ongoing campaign to sensitize consumers about the need to save electricity would complement this effort. ii) Through setting efficiency standards for each appliance type and monitoring the stock of appliances in households, the GOI could monitor progress toward a more efficient utilization of electricity in the residential sector. iii) To ensure that consumer incentives are consistent with the economic cost of electricity supply, electricity tariffs should reflect actual costs of supply. 44. The proposed End-use Electricity Conservation Program has been designed building on analysis of UHESS data, a review of appliances on the market, and experience from successful programs implemented elsewhere. Conservative assumptions have been used to evaluate the economics of the program: eg., by the end of a ten year program, 35% of refrig4rator sales will be energy efficient models (each using 75% of the electricity required by standard models) and 10% of the incandescent bulb market will be displaced by fluorescent lamp purchases (resulting in only 30% energy savings per installed bulb) due to program interventions. Under these and similar assumptions for television and air conditioning, Indonesia could save over 300 OWh/year by the year 2000, representing savings of approximately $US 10 million per year (at LRMC estimates of 9i\kWh) for a total program cost of roughly US$ 2.3 million/year. 45. Given that a compelling demonstration has been made that potential benefits for households and the national economy are obtainable at low cost (estimated under conservative assumptions and on the basis of other successful programs), such an end-use electricity conservation program should be an integral part of Indonesia's capacity expansion plan. The design of any program that is to be implemented along the lines of that presented herein, including efficiency targets, phasing and types of standards, and appliances to be focused on, must be developed through an active dialogue between representatives of government, appliance manufacturers, and consumer groups. It is recommended that the Ministry of Mines and Energy, through DJLEB, in coordination with the Ministry of Industry, take the lead in building a consensus among lamp and appliance manufacturers (through their respective associations), the consumers union (YLKI), and PLN- LMK on goals and means for an effective program. Such an effort should be fully coordinated with similar programs targeted at efficiency improvements in generation, transmission, and distribution systems. Ixiv- Summary Evaluation of Residenttal Electricity Conservation Program Net Present Value B 1O% Anrualized NPV (US$ Nitlion) CUS$ Millions) Total Proarm Cost CMSS mIllions) 14.7 2.3 Etletricity Cowetvatfon (GSh) 729 112 Lighting 384 59 Refrigeration 295 45 Afr-conditfonfng 28 4 Television 21 3 Ecenomic Value of Sawivgs (USi mllions) 66.3 9.9 Lighting 33.9 5.2 Refrigeration 26.1 4.0 Air-conditioning 2.5 0.4 Television 1.9 0.3 Net Economic Benfits 49.6 7.6 Source: See Ameex XII. Information Base for Further Policy Development 46. Information is required both to devise and to execute an urban household energy use strategy. During the course of this study, existing sources of information have been tapped, and much new information has been collected. Some of the information collected was designed to support particular study activities. The stove testing and modification, fcr example, were undertaken in order to better develop and assess a proposal for a kerosene stove program. The UHESS survey, however, was intended to provide a data base for use after the end of this project. It is hoped that, through the workings of the Technical Committee on Energy (PTE), the data base developed from the UHESS survey will be disseminated to government and research institutions to allow a more fully informed development of energy policy. 47. There are several reasons for creating the data base, and not simply a more extensive set of tables. First, it is impossible to predict which tables users are likely to want. For example, while the existing tabulation emphasizes income/expenditure and urban area size, for some purposes the most appropriate focus might be household size. Second, the data base allows statistical analysis. This can be useful both to assess the confidence with which the results should be treated, and to provide a more detailed analysis of particular inter-relationships. Third, there are many detailed questions which cannot be tabulated, but require interactive analysis. Thus, for example, to identify the most appropriate medium for promoting high efficiency refrigerators, the data base could be used to examine which media reach the principal actors in households owning or recently purchasing, a refrigerator. Fourth, the user may be interested in a particular target group, such as, for example, female-headed households. Finally, a data base provides the user with a far better means of assessing data quality than do summary tables. - xv - 48. To provide an ongoing source of policy-relevant information on household energy use, a 1-2 page energy module should be added to the SUSENAS survey in years during which the expenditure module is also administered (recently one year in three). A proposed set of questions for this purpose is provided in Annex IX. Such a module enumerated every three years would provide an invaluable source of information for monitoring national progress on policies an programs designed to encourage diversification and conservation of energy use in the residential sector. Rural Household Energy Strategy Study 49. Though the fuel use projections upon which these strategy components are evaluated are for urban households only, program implementation could have a significant effect on energy use in rural households as well. However, without detailed analysis of energy use patterns in rural households, neither quantification of these effects nor design of complementary piograms for rural areas is possible. If the patterns of energy use in urban households are extended to rural households, two hypotheses can be formed: i) wood and other biofuels probably meet a dominant share of rural household cooking needs, and; ii) rural electrification could displace a significant amount of kerosene for lighting while raising lighting levels by as much as tea times. For these reasons, the effects of policies and programs on rural households will necessarily be different from effects on urban households. There appears to be significant scope for raising the living standards of the rural poor through well-targeted programs and policies designed to encour-.ie fuel conservation and substitution. It is therefore recommended that a rural household energy strategy study be undertaken for Indonesia to complement the urban household energy strategy presented herein. A preliminary work program and budget for the proposed activity is outlined in Annex X. -xvi - Summan of Recommendations Responsible Agengy Rationalization of LPG Supp,lv Svstem Institute reporting system for distribution costs by region. MIGAS and Pertamina Phase in regional LPG price variation, reflecting distribution MIGAS and Pertamina costs. Remove non-price barriers to LPG sales, ensuring adequate Pertamina supply and variety of LPG bottles Phase in cost-based LPG and kerosene prices and/or promote MIGAS and Pertamina LPG use through non-price means. Disseminate safety and efftciency information through MMB, Pertamina distributors. Kerosene Stove Improvement Program Develop standards and endorsement system for factory st'ves. DJLEB, Ministry of Industry Set up centers for regional testing and stove improvement to: DJLEB, Kanwil, Ministry of Industry a) assist factories in meeting standards and introducing design improvements and. b: assist artisanal producers to upgrade their production process. Undertake generic stove research at technical institute. DJLEB, Ministry of Industry Develop consumer awareness/advocacy progam. DJLEB, YLKI End-use Blectricity Conservation Propam Set efficiency standards and testing procedures. DJLEB, Appliance Manufacturers, PLN, YlK Set up testing and labelling facilities. PLN-LMK, DJLEB, Manufacturers Undertake tests and monitoring of appliance sales. PLN-LM,K DJLEB, Manufacturers Develop and initiate information campaign. DJLEB, YLKI Other Institute continuously updated urban household energy data BPS, FIM, DJLEB, MIGAS base by adding energy module to SUSENAS survey. Disseminate existing household energy data-base. PTE, BPS, DJLEB Carry out nral household energy strategy study. DJLEB, MIGAS, Ministry of Forestry, ESMAP I. INTRODUCTION The Economic Situation 1.1 Indonesia, an archipelago in Southeast Asia, is the fifth most populous nation in the vworld with an estimated population of 175 million in 1988 and GDP of roughly US$ 400 per capita. In the early 1980s, the national population grew at 2.15% per year while the urban population was growing at the rate of 3.5%. By 1988, over 23% of the population were living in urban areas. The most populous island, Java, with 61% of the national population also has the highest concentration of urban dwellers, and an estimated 70% of the national urban population. 1.2 As a major oil producer, the economy of Indonesia grew rapidly in the 1970s, benefitting from high international oil prices. Contributions to GDP from oil and gas production changed dramatically after the collapse of the oil market in early 1986: peaking at 22% of GDP in 1981, contributions from oil and gas production to GDP fell to around 17% in 1986. In 1981/82, export earnings from oil and LNG generated 70% of total export earnings. Due to the oil market collapse and rapid growth of non-oil exports, by 1986/87, only 28% of total export earnings were contributed by oil and LNG exports. 1.3 In an attempt to insulate the economy and the government budget from external shocks due to fluctuations in international oil prices, the government initiated a series of reforms in 1983 designed to stimulate non-oil exports through deregulation. The adjustment program consisted of: i) fiscal and monetary policy, most importantly cuts in public investment; ii) Rupiah devaluations in 1983 and 1986; iii) financial sector and tax reforms; and iv) trade and industrial policy reform. These measures, taken together, have presented strong incentives for non-oil export development. As a result, non-oil export earnings soared to US$ 9.5 billion in 1987/88 (an increase of 41% in real terms over 1986/87) and non-oil tax revenues rose from 6.7% of GDP in 1983/84 to 9.4% in 1987/88. .1/ Energy Polic for the Residential Sector and REPELITA V 1.4 In the 1970s and early 1980s, Indonesian energy policy was formulated in an environment of high international oil prices. Petroleum products such as kerosene and diesel oil for domestic consumption were subsidized as a means of sharing some of the profits from high export prices with the population and to stimulate productive enterprise. Kerosene was priced low to: i) provide a substitute for fuelwood and thereby reduce deforestation; ii) provide a reliable supply of fuel for the poor at affordable prices, and; iii) to be made available to all residents at the same price in every region to serve equity objectives. During this period of low domestic fuel prices, technologies were employed and patterns of use developed without due emphasis on energy efficiency. 1.5 One of the objectives of Indonesian energy policy is to ensure that an adequate supply of fuel is made available, at affordable prices, to meet the basic needs of Indonesian households. This forms part of the justification for: 1) requiring Pertamina to ensure adequate v/ Sowze: 7ndonesJa SWegyforGmwth and Stnsalwv Change County Economic Aemomndwn, The Wodd Bank, 1989. -2- supplies of kerosene (and most other petroleum products) in all parts of Indonesia; 2) supporting a rapid electrification program; 3) subsidizing kerosene, and; 4) subsidizing low-power electricity users. Neither wood nor LPG receive commensurate support. 1.6 Changes in energy policy since 1980 supported national objectives and the reforms initiated in 1983. For the residential sector: kerosene prices were tripled in real terms between 1981 and 1984 and urban electrification was ambitiously pursued resulting in the percent of urban households electrified growing from 41% in 1981 to 74% in 1987. 2/ These measures were taken for a number of reasons including reducing government expenditures on the kerosene price subsidy and displacing kerosene used for lighting. 1.7 Institutions influential in formulating energy policy for the residential sector include: the Ministerial Energy Coordination Board (Bakoren) 2/ which issues general policy guidelines on energy matters; the Technical Committee on Energy (PTE) 4/ which reports to Bakoren and provides an active forum for consideration of policy and technical issues on energy; and the Permanent Working Group on Energy E/ which conducts analysis of energy supply and use on a quarterly basis, serves as a forum for exchange of data and information, and prepares technical documents for the consideration of the Technical Committee. During the course of this study, the Technical Committee formed a PTE household energy subcommittee for guidance on the formulation of energy policy in the residential sector. While in the field, the study team reported regularly to the TEM household energy subcommittee. It is hoped that the recommendations of this report, and programs presented in the appendices, will be acted on swiftly by the subcommittee and that the database analyzed herein will be used extensively by the subcommittee and the Permanent Working Group for future policy formulation. 1.8 In Repelita IV (1983/84 - 1988/89), the objectives of government energy policy were: to secure continuity of supply of energy for domestic consumption at prices affordable to the public to enhance the quality of life and to stimulate economic growth; and to secure an adequate supply of energy for export so that it will remain to be an important source of foreign exchange earnings to fund national development. To meet these objectives, four broad policy measures were supported: i) increasing and expanding geological surveys and exploration for energy resources; ii) diversification of energy use by reducing dependence on petroleum through substitution of other domestic energy resources; iii) conservation and energy efficiency improvements; and iv) indexation or matching each energy need to the most appropriate energy source available in the country. The energy section of Repelita V (1989/90 - 1993/94) opens by stating that 'the development and use of energy will be directed towards management of energy as economically and efficiently as possible with [a view toward enhancing] export prospects and the long-term conservation of energy sources". Hence, efficiency improvements and measures to encourage a/ Soue: SUSENAS, 1981 and 1987. 3J Bakoevncldes teMusterof isandEnoy (Chainuan), MtnistenforPublic Works, Fmanxce, Indusy, Defence, Deveopment Planning Commuica, Reah and Technolo, Population and Environment, Fomsy, Duo,t General of Atomi Eneig, Oil and Gas, and Electncity, and the Pesident Director of Petamma 1/ PTE Membea include Dictows Generl and Duitos of insiuons mpresented in Bakown MThe Pennanent Woding Goup on Enegv consist of enev aq,ts fm many govemment instituions. -3 - substitution of indigenous fuels to free up petroleum fuels for export are emphasized. Each of the recommended programs which constitute the urban household energy strategy presented in Chapter IV have been formulated in support of these objectives and broad policy measures. The Importance of the Urban Residential Sector in Energy Planning 1.9 The urban population of Indonesia is projected to grow at the annual rate of 4.1% through the year 2000 while the national population grows at just below 2%. As mentioned above, roughly 23% of Indonesia's population lived in urban areas in 1988. Between 30% and 40% of national consumption of kerosene, LPG, and electricity is consumed in urban households. According to official statistics, households consume significant shares of the nation's petroleum products and electricity (Figure 1.1) Although no official statistics are available on use of woodfuels by urban and rural households, this and previous studies show that woodfuels meet roughly 20% of urban household cooking needs (just over 10 MBOE) and probably more than 80% of rural household cooking needs (in excess of 150 MEBOE). 1.10 Because the residential sector consists of many small consumers, formulation of effective policy for the sector depends most heavily on data that can yield insight into how households of different income levels and locations may respond to interventions under consideration. The survey undertaken for this study has shown that urban households face similar problems and use similar appliances and stoves throughout Java. Before this ESMAP/DJLEB study, several surveys and studies of fuel use by rural and urban households had been undertaken. However, they did not result in proposed plans of action regarding policies and programs in which the government might invest to encourage rationalization of fuel use in the sector. 1.11 Even before the survey undertaken for this study, it was evident that cooking and lighting are the most important energy using activities in urban households, with lighting accounting for half of electricity use, and cooking dominating the other major end-uses of wood, kerosene, and LPG. The survey results have provided a much improved basis for quantifying these relations. 1.12 Cooking is carried out predominantly with kerosene, with wood also important in low income households and LPG becoming the fuel of choice in high income households. Even more than other household energy-using activities, cooking is primarily the responsibility of women. One of the major energy-using activities nation-wide, cooking is also central to the well-being of the population. However, like many women's activities, cooking has received relatively little attention from the policy community. There is little recognition that energy policy affects not only how much households must pay for cooking fuels, but also often affects the cooking process itself. The study team paid particular attention to the technical and sociological aspects of cooking in formulating recommended interventions. Objectives of the Urban Household Energy Strategy Stud, 1.13 In support of government initiatives in energy conservation and diversification of domestic energy use, the office of the Director General for Electricity and New Energy (DJLEB), requested the assistance of the World Bank/UNDP/ Bilateral Assistance Energy Sector Management Assistance Program (ESMAP) in studying fuel use in the residential sector. ESMAP mobilized funding for a thorough analysis of fuel supply and use in urban households and for -4 - Eigure 1.1: National Conventional Eneray Consumption by Sector Conventional Energy Consumption, 1988 By Sector and Fuel Group (MBOE) 120 - 100 . . .... ...................-._._ 80 - .... :~ 7- - '' ............................ 80 . ...... ..... ...... 40 . ...... ......... ...... ........ .../ . ...... ... ..i . .. ....l .. gl 20 -..... . . .. ... .... ...... 40 . - . . . .... 11 ... . 20 -.k X inrt6 Transport HoLseholds M Petroletum : Natural Gas _ Coal 1 Electricity S5uce: WJLEB. preparmg an urban household energy strategy. A similar study aad strategy preparation for rural households was postponed pending the result of a study of rural energy use in West Java completed in late 1987. / 1.14 The objectives of the Indonesia Urban Household Energy Strategy Study (UHESS) presented herein are: (a) to provide relevant data for evaluation of policies and programs aimed at the urban residential sector; (b) to propose a coherent set of costed and scheduled programs and policies that: (i) enhance the security of energy supply to urban households; (ii) increase the efficiency of fuel use in household applications; (iii) reduce government expenditures and increase export value by encouraging fuel substitution and conservation; and (c) to develop tools and skills for continued policy development and analysis. 1.15 Toward these ends, ESMAP placed a project manager in Jakarta for 18 months to oversee an urban household energy survey, analysis of resulting data, a review of urban fuel supply systems, kerosene stove improvement work, electric appliance analysis, and to prepare an integrated gi 'RWd West Java EaeV Use Sady', ETA-179, fiwded by *e Govnme of *e Nandu ad eeced in coordiuadon **h D1 . -5 - set of policies and programs for the urban residential sector. This report presents the integrated findings and recommendations of the strategy study team. -6- IL FUEL SUPPLY AND DISTRIBUTION l/ 2.1 Each of the major urban household fuels raises different distribution problems and issues. Characteristics of each fuel distnbution system are described and assessed in this chapter. Estimates of fuel consumption in the household sector, as a percent of national fuel consumption in 1987 for each fuel, are presented in Figure 2.1. 30% of national kerosene use, at least 40% of national LPG use, and 28% of total electricity sales is consumed in urban households, nationwide. Because of the quantities of fuel consumed and the economic and budgetary implications of pricing and distribution policies, the need for careful and efficient energy planning for the urban residential sector is readily apparent. Kerosene 2.2 Until the 1980s Indonesia imported a significant share of its kerosene for domestic use, despite being a net petroleum exporter. Since the mid 1980s, however, Indonesia's domestic kerosene demand has been roughly equal to domestic supply. The next five year plan (REPELITA Eire 2.1: Shares of National Fuel Use Consumed in Urban Households Urban Residential Fuel Use as Percent of National Demand 50% Lkban Residential: % of National Demand 40% - 30%- 20% - 10% = 0% Kerosene Electricity LPG Wood ESIAP/WEM 4 JUe 99. Sotire: WA, DAME, ard SUSENS. Z/ T1i Ch*te Is based on a Lemiga working paper entited 7(semne and LPG Disbo and die Ec_mic of Subsdkon -7- Table 2.1: Kerosene Sales to Urban Households by Region 1981, 1984, and 1987 Nillion Liters of Kerosene per Ami 1981 1984 1987 Sumatra 38? 411 372 Java 2,373 2,179 2.070 Kmus Tenggara 52 48 46 Kalimantan 94 91 96 Sulewesi 129 133 106 Maluku & Irian Jaya 29 27 28 Inronesia 3.06 2,a8 2.718 pource: SUSENAS 1981, 1984 ad 1987. V 1989/90 - 1993/94) projects no growth in kerosene demand. With the existing refinery configuration and sufficient lead time to adjust the product mix, there is considerable flexbility in the level of kerosene production. Should kerosene demand grow sharply, however, Indonesia could again become a kerosene importer. 2.3 Kerosene accounts for over one quarter of petroleum product sales within Indonesia, with about 7 billion liters sold in 1988. At USS 23/bbl (average Singapore posted price for the third quarter of 1989), 1988 domestic kerosene consumption represents over US$ 1 bilion or roughly 13% of estimated 1988/89 earnings from oil and LNG exports. Sales grew at an average of 10.6% per year between 1971 and 1981, up to a high of 8.4 billion liters in fiscal year 1981. Subsequently, sales fell at 3.2% per year up to 1986. Ibis decline was the result of both rural and urban electrification and price induced kerosene savings As illustrated in Figure 2.2, the real (depot) price of kerosene increased threefold between 1981 ahid 1984. Nevertheless, and despite falling international prices, the retail price of kerosene remains about 90 Rupiah less per liter than its estimated economic cost (30% subsidy). Since 1987, national kerosene sales have grown steadily at 6%, per year. 2.4 About 3 billion liters of kerosene, roughly 40% of total sales, were sold to urban households in 1988. A/ Regional estimates of kerosene sales in 1981, 1984 and 1987, based on the SUSENAS expenditure survey, are given in Table 2.2. Kerosene sales to urban households fell at about 2% per year during this period. 2/ Regional differences closely reflect the number of urban households in each region. Monthly kerosene sales to the average urban household in 1987 was between 20 and 27 liters in all regions, with the exception of Nusa Tenggara, where kerosene sales are exceptionally low. y/ There ad iepancies between e SUSEVA4S svy and ce*polato of the UHESS swvey concng e ovmill level of kerosene consumed by wb ho olds i 1987. Te SUSENeSE sed esmte of 27billion ilte bis about 15% less than the UHESS-basedesdmate of 3.2 bilo Whie the UHESSbased estmate is more likey to be acaue, SUSENASfigures are useful forpesenng intesand differences, and changes between 1981 and 1987 2/ Rouh( one th&d of the decline in koee use by ab househoksoverthe piod can be asuted to ectif lcation, one half to csatiov, and the rsdua to ubstakm by LPG and wood in rsponse to puce inreass. SeeAntnx Mforan alis. -8. Eigure2.2: Domestic Depot Kerosene Price 1973 - 1987 Domestic "Depot" Kerosene Price INDONESIA 200 Rupiah per Liter Real- 87 150 - Nominal 100 73 75 77 79 81 83 85 87 Year E9P/ JLEB 4 Jun, 1989. 2.5 Estimated build-ups of the average financial and economic costs of providing a liter of kerosene to the urban consumer are given in Table 2.2. IO/ The "average" urban household pays about 2.03 Rupiah for a liter of kerosene that cost roughly 290 Rupiah to supply. The cost-price differential, 87 Rupiah/liter, on average, will be higher in areas whose depot is serviced by more distant refineries, and falls to some 71 Rupiah in areas where refinery-depot transport is negligible. 2.6 The opportunity cost of using kerosene domestically is based on the alternative of upgrading and exporting the kerosene via Singapore. The most promising future market for exported kerosene is the growing Japanese demand for aviation fuel. Because meeting this demand requires a higher quality of kerosene from that sold domestically, a "quality discount" is applied. This opportunity cost is only appropriate if Indonesia produces at least enough kerosene for the domestic market. If domestic kerosene demand grows to the point where Indonesia again becomes a kerosene importer, then the opportunity cost of domestic kerosene use will grow by roughly twice the freight and insurance costs (line 3), or about 10 Rupiah per liter. /gl For both kevsene and LPG, te mainal cos of snqplyg wban ousehold demand is gley a fimcdon of the geogmphy of the dsiwbudon system, and where the "mginademand is located. As such, it is notposible to estimate moga costs whout knowing the soume of the demand dgange. For the pwposes of cost estimato, differences beten avemge and manal disbution costs are ignored, exept where they dea4fy affect the anaysis. As these cost estmates r not based on oficalstatistcs but mtheronprevaignagtioalpnices, and disribvtaon maxupsfrom Petamina, they shaoud be teated as indicave, only. .9. Table 2.2: Economic and Financial Costs of Urban Kerosene Supply Cost Item Financial Conversion Economic Cost Factor Cost (Rp/lt) (Rp/lt) 1. FOB Singapore 242 1.0 242 2. *Quality discount 3 1.0 3 3. -Freight & Insurance _ 1.0 _ FOB Ifdaresiun Port 233 233 4. *Port Handling - 3 0.8 - 2 Ex-Refinery 230 231 S. Port->Depot Transport 22 1.0 22 6. Wholesale Margin 5 0.8 4 7. -Subsidy -92 0.0 Depot 165 257 8. Depot->Retail Transport 16 1.0 16 9. Retail Margin 22 0.8 17 Consmer 203 290 Line Item Sources: 1. Based on average Singapore posted prices for third quarter 1989 of US$ 22.95/bbt, and currency exchange rate of Rp 1700/USS. 2. Based on price differential of US$ 0.40/bbt for kerosene with smoke point of 18 (Indonesian Domestic) and smoke point of 23 (International Standard). This roughly reflects the cost of upgrading Indonesian kerosene to international standard. 3. Assumed to be 2.5% of Singapore price. 4. Assumed to be 1.5% on FOB Refinery cost. 5. Based on Pertamina charges. 6. Regulation mark-up. 7. Financial subsidy is calculated to balance financial costs with the fixed depot price of 165 Rupiah per liter. 8. Assuaed cost of RupIah 0.64/titer-kilometer and distance of 25 kilometers. 9. Retail margin is calculated to balance financial costs with UHESS survey finding that urban households in Java pay an average of 203 Rupiah per titer for kerosene. Cost Factors: Mission and Lemigas Estimates. 2.7 National kerosene pricing and distribution policy are set in the President's Office, on the advice of BAKOREN. (In this kerosene is similar to other petroleum products, but different from LPG for which Pertamina and the Ministry of Mines and Energy are more directly responsible.) A policy designed to ensure kerosene availability at "equitable" prices throughout Indonesia has led to a long history of kerosene subsidies. The recent kerosene price increases reflect a growing emphasis on economic criteria, but other criteria are still influential, as evidenced by the continued subsidy. 2.8 Pertamina is responsible for refining, transporting and marketing kerosene. It is obliged to supply, at the government set depot price, sufficient quantities of eight fuel products (Bahan Bakar Minyak or BBM), including kerosene, to meet demand in every region and is reimbursed by the government for any cost shortfall. The government paid Pertamina the rupiah equivalent of US$ 236 million for the shortfall between sales and costs of these eight products in 1987/88. Using the subsidy estimate from Table 2.2, the value of the kerosene subsidy for kerosene consumed in urban households in 1988 was about US$ 154 million (about 1.5% of government current expenditures for 1987/88). To meet demand and prevent supply disruptions arising from demand fluctuations, Pertamina stores large reserves of kerosene in tankers. As a result, the - 10 - economic costs of the kerosene subsidy are reflected in higher kerosene use and higher storage costs than would otherwise be the case, rather than supply problems or black market prkcing. 2.9 The bulk of kerosene for household use is distributed from the Pertamina refineries to consumers via depots, agents, dealers and retailers. Licenced agents and consumer cooperatives contract with Pertamina to purchase and collect a specified volume of kerosene from a given depot each day. U/ Dealers are subordinate to a particular agent, and can number over 200 for a large agent. In Jakarta, for example, there are 11 Agents, and 1,194 dealers. Retailers include both small stores (warung) and peddlers (tukang pikul). 2.10 The most recent Presidential Decree affecting the price of kerosene, issued in 1985, set the depot price at 165 Rupiah per liter. Provincial governors issue detailed guidelines specifying the maximum mark-ups allowed per liter for: 1) intra-city agent and transport fees; 2) inter-city transport fees by destination (for destinations more than 40 kilometers from the depot), and; 3) dealer fees. In Central Java, for example, the governor has set the intra-city agent and transport mark-up to 7 Rupiah per liter, and the dealer mark-up to 5 Rupiah. The allowed inter-city transport mark-up is specified for every major town, based on distance from the nearest depot, and cost guidelines established by the Department of Transport and Communications. As in other provinces, the final mark-up by peddlers is not controlled. 2.11 As such, the price control system can be divided into two components: 1) Presidential directives on Pertamiina's depot price intended to set the basic kerosene price level and ensure regional price equality, and; 2) Regional controls on agent and dealer mark-ups designed to prevent overprcing once the kerosene enters the private sector. Some form of Pertamina price controls are necessary as long as Pertamina remains a monopoly seller. Setting all depot prices equal regardless of transport costs is an economically inefficient, but visible and hence politically attractive, means of promoting regional equity. The ex-depot controls are more unwieldy, but necessary where prices would otherwise be unduly influenced by agent or dealer monopolies. 2.12 In competitive conditions there would be no need for controlling agent or dealer mark-ups. Given agent monopolies, however, such price controls are necessary. Standard economic theory indicates that monopolists will restrict sales to drive prices up above their free market level - an undesirable situation. Even in the largest urban market, Jakarta, there are only eleven agents, and the potential for overpricing in the absence of price controls is present. This tendency may be mitigated by the regulation giving priority to consumer cooperatives to act in lieu of agents, should they deem the price too high, but the danger remains. 2.13 In the current climate of deregulation, this observation has important implications. Under existing circumstances, to deregulate the last stages of the kerosene distribution system, it would be advisable for Pertamina first to solicit the entry of more agents and/or allow dealers to purchase directly from Pertamina. Having ensured that competitive conditions prevail, the regional price controls could be lifted without the risk of overpricing. Currently, while the Provincial Government has considerable leeway in creating and enforcing the ex-depot price controls, it is Pertamina policy which determines whether or not such controls are needed. 11/ See ke dsibti, sc/ac in Anna 1 - 11 2.14 F r o m t h e Table 2.3: Kerosene Price by Source UHESS survey of Java, the average price paid for of KeroseXe Purchasing Price/liter Li teru/purchase kerosene by urban consumers in early 1988 was 203 Rupiah Sall Shop 75X 203 4 Peddler 222 20 7 per liter, with 80% paying 200 Dealer 7x 193 10 Rupiah. Three quarters of --- kerosene !;onsumers purchase S 42 of households purchase from a csibnat ion ot souces. kerosene from small stores, 22% from travelling salesmen, and only about 7% ever purchase directly from dealers. As shown in Table 2.3, households purchasing direct from a dealer paid the lowest prices, but also bought in the largest quantities while households having kerosene delivered paid the highest prices. Households going to a local shop or "warung" paid intermediate prices (on average 203 Rupiah per liter), and bought in the smalle , quantities. Of the households not using kerosene (about 13% of urban households), almost none claimed this was because of supply or distribution inadequacies (see Chapter m, Table 3.6). liquified I etroleum Gas 2.15 With the commissioning of two LPG fractionating plants at Arun and Bontang in mid-1988 (2,250 kTPA total capacity) to supply Japan with 1,950 kTPA under a ten year contract, Indonesia became a major LPG exporter. 12/ Recently commissioned LPG plants and projects under construction are expected to raise Indonesia's total installed LPG production capacity to 3 MTPA in the early 1990s, with a large share of production to be exported under long-term commitments. The bulk of domestic LPG sales are from oil refineries. Two of the three refineries producing LPG for the domestic market are in Sumatra (Rantau and Musi), and the other is in Java (Cilacap). (See map in Annex II). 2.16 Despite the recent boost in production capacity, LPG remains a relativel minor fuel in the domestic market. Domestic sales in 1988 were 235,000 tons, of which at least a quarter was destined for industrial users. In energy terms, total domestic sales of LPG in 1988 were less than 5% of national kerosene sales. Growth has been very rapid, however, averaging over 17% per annum since 1978. (See Domestic LPG Sales Table in Annex L) Sales to urban households in 1988 are estimated at 170,000 tons, with Java accounting for over eighty percent of these sales. / 21 So "The Rem Repovi Indonesia' Embag of te Unid Sta ofAmeli 1989 Ul Ts estiate of wban sidtalnsW nisn as hed on sppvf8wvs and is equivlent to roAg 70% of doesmac LPG saes Suppy-based estates ar used because possb wsder.samplng of vwy hg iome houehols in de UHESSsVmaykad towidgestmatesofhigh-no comeiuse,suchasLPG, H dwied fms wvdatL Using the shaw of national wban r=sidei LPG use connmed on Java (83*) ftum Suswena 1987 to Wsend UHESS esties to wban onesia as a sole (125 AJ/HA,r * 658 million wban RH on Java = 820 tons of LPG conmed by wban houshods on Java in 1988), leads to estmates ofjust wuder KOO Wms of LW comed by wbn houshols nationde in 198& As the sppb estiae isjudged to be mw iiablk, it is ed an theM p_jectio model pnd later. - 12 - Table 2A: Economic and Financial Costs of Urban LPG Supply Cost Item Financiat Conversion Economic Cost Factor Cost Rp/kg Rp/kg 1. FOS Indonesian Port 164 1.0 164 2. Intra-indonesia Tanker 50 1.0 50 3. Inland Transport to Depot 29 1.0 29 4. Bottling, etc. 136 0.8 109 5. Effective Tax 1Z3 0.0 Q Dealer 502 352 6. Dealer Transport 29 1.0 29 7. Dealer Margin 59 0.8 47 Consumer 590 428 Line Item Sources: 1. Based on the July 1989 FOB price of Indonesian LPG shipped under long term commitments from Arun and Bontang of USS 96.26/MT. Almost 90X of total LPG exports are shipped from these two sources. 2. Pertamina Charge. 3. Pertamina Charge - corresponds to an average distance of 60 km at a cost of Rp 0.48/(kg-km). 4. Pertamina Charge - for breakdown see Table 2.5 below. 5. The effective tax is calculated to balance financial costs with the official dealer price. 6. Assumes average distance of 25 km at a cost of Rp 1.16/(kg-km). 7. The dealer margin is calculated to equate the sun of dealer transport and margin to the regulation dealer mark-up of 88 Rp/kg. Cost factors: Missfon and Lemigas estimates. SUSENAS survey results suggest that sales to urban households may have been increasing even more rapidly than overall sales: over 35% per year since 1981. 14/ 2.17 In October 1987, partly in response to rapid demand growth, the ex-depot price of LPG was increased from 379 to 590 Rupiah per kilogram. The official bottle price also rose to 55,000 Rupiah for the 11 kilogram bottles which dominate the residential market (45 kg bottles are also sold). Prior to these changes, prices had remained constant for about four years. The current fuel price is 590 Rupiah per kilogram. 2.18 A summary of the financial and economic costs of supplying a kilogram of LPG to an urban consumer, analogous to that for kerosene above, is provided in Table 2.4. As indicated, the "average" consumer pays about 40% above economic costs for LPG, as opposed to 30% below economic cost for kerosene. As with kerosene, the difference between the set retail price and economic costs depends on location. Though LPG is less widely distributed, its transport costs are higher. The effective financial tax on LPG is 123 Rupiah/kg of which Pertamina takes a significant share. 11 35% annauawan wtsidendat LPG growth mtespom SUSENAS are not necessauy inconsistent wth gn?wth in total domic salesfrm officialfpgs of 20% overthe same pepnod Ts meaely indicates that LPG has moved from a reativeis bignfcant household fuel in the late 1970s to the fuel of choice in uwer income urban homes. Analysis of UHESS data pasented later, indicates that unless the initial cost of LPG use (costs of stoves and bottles) and/or the pkice of LPGfalts substntially, wban esidential LPG gvwth rtes over 30% expenenced in the 1980s should not be eaected to contiue in the 1990s. - 13 - Table 2.5: LPG Cost Estimates Before and After 1987 Price Increase Component Old Price New Price Increase Rp/kg Rp/kg Production Cost 108.9 199.2 83X Transport to Bottler 50.3 50.3 0X Filling Fee 33.3 new item Handling Fee 20.6 new Item Bottle Maintenance 49.2 49.2 0X Depreciation 14.5 24.9 ?2X Overhead 5.0 5.0 0X Losses 3.0 3.0 0X Inland Transport 29.0 2920 0O Depot Cost 259.9 414.5 60X Pertamina Margin 10X of Depot 25.9 41.5 60X Sales Tax 10X 28.? 45.5 60X Deater price 314.5 501.5 604 Dealer Margin 15X of Retail 55 5 .j 60X Retait Price 370.0 590.0 602 Source: Pertamina. 2.19 The opportunity cost of LPG use was taken to be the cost of not exporting LPG under long term commitments: Rp 164/kg average FOB price in July, 1989. There is some uncertainty about how well this represents the actual opportunity cost. / There is also uncertainty in the estimates of distribution costs. As indicated, many of the distribution costs are derived from Pertamina charges. However, Pertamina charges are not free to fluctuate with changing market conditions, and cannot reflect incurred costs. For kerosene, with distribution costs to the depot estimated at about eight percent of overall economic cost, such uncertainty is tolerable. For LPG, with distribution costs to the depot (including bottling) estimated at over 40% of economic cost, the resulting error can be significant. This problem is exemplified by the abrupt changes in cost estimates which accompany intermittent price changes, as illustrated in Table 25. Until the inevitable lack of correspondence between actual and charged costs is recognized, and work is undertaken with Pertamina's cooperation to ascertain actual costs, official cost figures will not be sufficiently accurate for detailed policy analysis. 2.20 The distribution and pricing of LPG is controlled directly by Pertamina, acting under the supervision of the Ministry of Energy and Mines. Pertamina sells part of the LPG directly to industrial users, but some 180 dealers, concentrated on Java, handle all of the household LPG. Dealers are contracted by Pertamina to sell at a specified price and within a specified area. The J~/ Since LG produced by recent maor os in LPG prducion pacy s commted for xwpon uidercon&c die longicam contrwt pice if used as dt opponwui cost of LPG producto As idcated in Annea I, the bulk of Indonesia'sLPGiexpoated, *iw demajshareshippedfromA,=, Aceh and Bont& Kmn porndwerlng, tern conmnucu The Juty 1989 FOB pne of contaed LPG shipnts from these two pon, US$ 96.6/MT, is siagfcanty kss than avec posed pnces for LPG in die Singapore maa dungn then d quoer of 1989, US$ 6I/MT, and Singapore posted pcs have been faidy steady since 198 If m npron of Indonesan LPG is traded on int wnatonw*ets in the future, the opponunity cost shoudd be based on netback FOB pnces at Ind esimpom Mowv, PeainawhaseimathedcostofLPGproducaion f n rfnaiesatalmnostRp 2(g (US IISIMT). Howew, this estimate is based on Talocated costm designed forsimple accounting in a muti-product enironvment at the fineay, and may not jec actalos If acual main producion cots exceed exportpnces sigcn, there should be sucintscopefrcuing backn es, despite some longlem cnmta oblgaons, and the opporbr cost should be based an the potential savngs of not producing LP. - 14 - provincial governments are not involved in price setting or enforcement, and the regulation is less rigorous than for kerosene. As a result, the dealers' bottle prices are virtually uncontrolled. 2.21 From the refineries or LPG plant, LPG for domestic use is transported by tanker or tank truck to one of the five bottling plants. 16/ The bottled LPG is then transported, at a considerable cost per ton-kilometer, to depots and on to consumers. To reduce transport costs, Pertamina is considering allowing private companies to build and operate mini LPG bottling plants, each with a capacity of about 25 tons per day. 2.22 In Java, the LPG distribution system is already quite extensive. Even households in small urban areas can generally obtain LPG. Less than 20% of households surveyed not using LPG cited distribution inadequacies as the reason for non-use, and the surveyors estimated that less than 4% of the households lived in neighborhoods where it is impossible to get LPG (though a further 36% live in neighborhoods where it is difficult to get LPG). Of the households using LPG, about 60% have it delivered to the house, and the remainder pick it up themselves. 2.23 Though it is possible for most urban households on Java to purchase LPG at near official prices, the average price consumers pay for an 11 kg bottle in Java (Rp 75,500) is roughly forty percent higher than the official price per bottle (Rp 55,000). High bottle prices, at least partly a result of a widely rumored bottle shortage, contribute to the most often cited reason households do not use LPG: high investment costs. To alleviate the shortage, Pertamina plans to produce 360,000 11 kilogram bottles in 1988/89. 2.24 In the long term, the possibility of soliciting private companies to produce LPG bottles is being considered. Particular attention is being given to 5-6 kilogram bottles, which have been proposed in order to eliminate two additional barriers to LPG use: 1) the high investment cost of LPG use; and 2) the difficulty of delivering 11 kilogram bottles down narrow (kampung) alleys. The UHESS survey indicates that the former barrier is very significant, while the latter is hardly significant at all. Of the households not using LPG, more than one third cited equipment cost as the main barrier, while only 2% cited delivery problems due to narrow roads (see Table 3.7). 2.25 Despite the small number of agents, there is little evidence of serious dealer overpricing of LPG itself. It is widely rumored that the delivery trucks often sell underweight bottles. This may be because, within 60 kilometers of the Pertamina depot, no additional charge can be made for home delivery. Seals could be improved to eliminate at least part of this pilferage, but a better long term solution would be to allow a delivery charge, while promoting dealer competition and/or enforce the requirement that dealers sell each charge on the scale. 2.26 The issue of whether or not LPG should be promoted as a household fuel is dealt with in Chapter IV. The results suggest that while the economic cost of LPG use in urban households is roughy equal to that of kerosene, significant benefits to consumers, the government budget, and balance of trade are to be had from stimulating the domestic use of LPG as a cooking fuel. IV See map and LPG dtibution schematic in Annex 11. - 15 - Figm 23: Representative System Load Curve (Java) Daily Load Curve Characteristics Indonesia Representative Figures % of Peak Load 75X%.< 25% ...... -. - ~HV ConsumerrLoadd 0%I II 0 6 12 18 24 Hours Source: PLN 9sm Oata, Power Sysem Efficiency Wtojwt Worldwa, 1987. 2.27 According to PLN statistics, 1/ it is estimated that about 28% (5600 GWh) of the electricity consumed nationwide in 1988 (19,930 GWh) was consumed in urban households, while roughly 9% (1830 GWh) was consumed in rural households. The percentage of urban households nationwide using electricity grew from 47% in 1981 to 74% in 1987. 1/ Sectoral growth rates in PLN projections indicate that urban residential demand is expected to grow at roughly 8% per annum to 14,600 GWh in the year 2000, whfle rural residential demand is expected to grow at 12% to 7000 GWh. Valued at recent estimates of the Long Run Marginal Cost (LRMC) of electricity supply to low voltage consumers of approximately 150 Rupiah/kWh, 12/ the annualized cost of projected residential electricity consumption through the year 2000 amounts to Rupiah 1.95 trillion or US$ 1.15 billion. 2Q/ 2.28 As residential demand peaks in evening hours, in coincidence with the system daily peak load (Figure 23), investments required to meet the projected growth of peak capacity can be L/ PLstaistics and projeconsfrm eanfy 1988 W Sozwe: SUSENAS. t Indoes pow erJifcc PX Eletiity Tanffs, Maion Rn, Wodd Pank, Se_Teb, 198. &9/ LRMC of residentdal ekdcty conp troughout dte peWod are diaowted ad annuaizd at 10% - 16 - Table 2.6: PLN Residential Electricity Tariffs PLN Residential Tariff Structure: April, 1984 - March 1989 Tariff Power Demand Energy Average Marainal Cost Estimate a/ Shares of Class Range* Charge Charge Price Capacity Cost Energy Cost Average Urban Residential OkVA) (Rp/VAJI4o) (Rp/kWh) (Rp/kWh) (Rp/VA/Io) (Rp/kWh) (Rp/kWh) Consumption tI Si Low-0.2 25 0 NA RI 0.2-0.5 2.1 70.5 85 8.9 77.8 160 50X R2 0.5-2.2 2.1 84.5 98 8.9 70.3 143 35X R3 2.2-6.6 3.68 126.5 156 8.9 62.8 156 7% R4 6.6 + 3.68 158.0 184 8.9 59.05 141 8X PLN Residential Tariff Structure: April, 1989 Tariff Power Demand Energy Charge Marsinal Cost Estimate a/ Class Range* Charge (Rp/kWh) Capacity Cost Energy Cost Average Cost (kVA) (Rp/VA/Mo) (Hours/month) (Rp/VA/Mo) (Rp/kUh) (Rp/kWh) 0-60 60+ RI 0.2-0.5 3.16 63.5 86.0 8.9 77.8 160 R2 0.5-2.2 3.16 76.0 115.5 8.9 70.3 143 R3 2.2-6.6 5.52 155.5 155.5 8.9 62.8 156 R4 6.6 + 5.52 196.5 196.5 8.9 59.05 141 N Note: Ranges exclude lower bound. LRNC estimates from Mission Report, World Bank Indonesia Power Efficiency Project: Electricity Tariffs, September, 1987. Shares are percent of GWh sold to each tariff class based on PLN Budget forecast for 1987/88. Source: Tariffs from PLN documents. directly attributed to generating capacity requirements for meeting residential electricity needs. At US$ 1200/kW of installed capacity, the annualized present value of the total capital investment package to meet increased demand through the year 2000 is approximately US$ 1.35 billion per year. Ll/ Because residential demand contributes significantly to the daily peak demand, residential electricity conservation could be a cost effective part of the capacity expansion plan. A program to encourage electricity conservation is presented in Chapter IV. 2.29 The real price of electricity for residential connections has grown at an annual rate of less than 3% in the 1980s, and has actually been declining since 1984. In 1987, the average nominal price was about 100 Rupiah per kWh. In Aprit 1989 a new tariff structure was instated to recover the real price decline since 1984. A summary of PLN tariff structures for residential consumers, before and after the change, along with planning figures of overall residential consumption shares by tariff class, are given in Table 2.6. Before April, 1989, energy charges for most residential consumers (RI tariff) were significantly below marginal costs of supply (as estimated in 1987) with higher power consumers cross-subsidizing low power consumers, and capacity charges were well below capacity costs for all residential consumers. The new tariff a/ PLN's capacity eypasion plan calls for 12% annual growth in installed capacity from 8,0o MW in 1988/89 (5,600 on Java) toan estimated 20,00MWin l999/00(15,000on Java). The present valueof capacityexpansion isestimated at US$ 1,200 per kW and discounted at 10%. - 17 - Figum 2A: Electricity Supply Source Source of Electricity on Java by Urban Area Size and Income 100% Percent of Households Using Electricity j BD I M HTM 0~~~~~~II 80% -M 60%- 40%- 20%- 0% Vilge Town City Jakarta Low Moderate High Urban Area Size Household Income = PLN Drect 3 PLN Indirect ffl Other Sour.: IUHESS SuIvey, l988 W=pendh l tor Categots. structure moves closer to recovery of marginal costs, but all consumers still pay well below 1987 estimates of capacity costs for their service. 2.30 Under the existing PLN tariff, most residential consumers pay a small monthly connection charge plus an energy charge, with the rates determined solely by the maximum load (VA) of their connection. Low-load consumers pay considerably less than (marginal) cost, while the very iigh-load consumers pay more. As a result of the preponderance of low-load consumers, residential electricity is generaUly priced below (marginal) cost. However, given the structure of the tariff, consumers face an extremely high marginal cost at the point where additional electricity use requires shifting to a higher tariff. A household shifting from Rl to R2, for example, must pay not only the cost for increasing the power rating, but at least an additional 12 Rupiah for every kWh consumed. In effect, purchasing an appliance which leads a household to require a higher power rating can be extremely costly. For this reason, many consumers are more concerned with the wattage of new appliances than with their overall electricity use. This feature of the distribution system is recognized by appliance manufacturers, who adjust their sales strategy accordingly by offering low wattage devices labelled "Hemat Listrik" (Save Electricity). 2.31 While PLN generates most of the electricity, a very significant number of households do not purchase directly from PLN, but rather obtain PLN generated electricity from their neighbors. The UHESS survey of Java, conducted in early 1988, indicates that of the 85 percent of urban households electrified, 25 percent obtain electricity in this fashion. Figure 2.4 shows that, households obtaining neighbors' electricity tend to be low income households who wish to avoid both connection costs and the monthly power charge. As households that obtain their electricity indirectly also tend to be low power users, they account for only about 6% of urban household electricity consumption. - 18- 2.32 Given the relatively high cost of metering electricity, this informal system of shared connections may be economically desirable. The three disadvantages of shared metering are that it: 1) dilutes the financial incentive for electricity conservation; 2) can lead to unsafe wiring; and 3) can mask electricity thefL 22/ 2.33 The significance of "shared connections' has important implications for PLN urban electrification plans, however. Currently, PLN places a priority on increasing the share of urban households with an official connection, and the number of official connections is what drives PLN's residential demand projections. According to the UHESS survey, less than one third of the truly non-electrified households cannot obtain electricity: the majority find the cost too high for their budget. In short, growth in electricity consumption in urban households is more a function of income growth than of growth in official connections. 2.34 There is considerable regional variation in the percentage of urban households using -lectricity. 22/ At the provincial level the percentage electrified ranges from 41% in Central Kalimantan to 91% in Jakarta. At the island level, the percentage ranges from 57% in Maluku and Irian Jaya to 76% in Java. 2A/ This variation cannot be explained by regional variation in the households alone, and reflects differences in the distribution system. Despite these differences, since roughly 3/4 of the total urban households are on Java, the average urban electrification over all of Indonesia was 74%: only 2% less than on Java. Wood and Charcoal 2.35 A vast, decentralized distribution system for wood, !mportant especially in the smaller urban areas, exists in Indonesia. When the system is operating well, as it generally appears to do in Java, wood provides a renewable energy source at a low cost. The success of the existing system is at least in part due to the importance of home-gardens. In urban Java, the UHESS survey reveals that more than one third of households using fuelwood collect at least some of their fuel from their own home-gardens. 2.36 Fuelwood, or in a few cases crop residues, is used by about one quarter of urban households in Indonesia (SUSENAS, 1987). 2 Since the distribution system is both informal and decentralized, and since SUSENAS does not gather information on the amounts of fuelwood used, U/ More speaicalay: shured metng dilutes the fin incentive for ekectdcity conseation in at elevicity cost are shared and each household does not pay the fidl cost of its own use. Also, shared metedng does not necessaily mask ekectiicity theft, but uthea, inhibits effective actions toward eaimating such theft as it is not cwrently known if an un- metered household is stealing ekcticity or paying a neighbor for electuicity that acual does pass through a meter. Z3 See SUSENAS data presented in Table 3.1. W The UHESS suvey in 1988found that 85% of the households in Java used some electricity. This higherfiggur may resultt fm SUSENAS not including housholds using eleckicityfwm neighbor. 5 This figure is far higher than the 4%found in a 1985 suvey of utban households in Indonesia (DILEB-EDI, 1985). The 1985 sun included & targe k rban areas, howver, and the low paenetage reflects this bias. See Appndix I for a compson of the UHESS suvey apeience with that of previous swveys. - 19 - Figure 2.5: Fuelwood Supply Source Source of Fuelwood on Java by Urban Area Size and Income 60% Percent of Households Using Fuelwood 50% i gm Collect M Both 40% . i 2 Purchase 30%- 20%- 10% Villge Town Ciy Jakarta Low Moerate High Urban Area Size Household Income SSMAP/ ULEB 4 June, 1989. see S umay TableS for Cateo9ies. it is difficult to estimate total consumption from SUSENAS data alone. From the UHESS survey, Urban households using fuelwood on Java were found to use an average of 140 kilograms per month. Extending this average using the SUSENAS estimate of the number of fuelwood using households in all of urban Indonesia, yields a total fuehwood consumption of about 3.8 million tons per annum, two thirds of which is consumed in Java. The quantities "distributed' are thus roughly the same as for kerosene, though the energy content of the wood is less than half that of the kerosene, and the useful energy in cooking even less. 237 Experience in other countries has shown that, since urban demand tends to be more concentrated and more commercial, urban fuelwood consumption can typicaly be far more responsible for environmental damage than equivalent levels of rural fuelwood use. Combined with the higher transport costs to urban consumers, this typically results in a higher share of easily- transportable stem-wood, requiring more trees to be cut per ton of fuelwood (rural users can tap the large supply of small branches, causing less damage). Furthermore, it leads to a higher portion of biofuel derived from forests, rather than agricultural waste. 2.38 On Java, however, the UHESS survey indicates that urban consumers obtain a large share of their biofuel from their own or neighbors' land. Figure 2.5 shows that even though far fewer high income households and those in large cities use fuelwood than low income and small town households, of the 23% of households using fuelwood, more than two thirds collect their own fuelwood, and more than one third obtain part of their fuelwood from their own land. Overall, 12% of households surveyed claim that they own land from which fuelwood can be obtained (see UHESS Summary Table 13 in Appendix 1). - 20- 2.39 The evidence does suggest that urban consumers are increasing their use of biofueL According to SUSENAS data, the percentage of urban households using fuelwood grew from 20% in 1981 to 24% in 1987 and the total urban population has been growing rapidly. This shift is probably the result of kerosene price increases in the early 1980s. Even in Java, up to one half of the fuelwood used in urban households may be forest-wood. It will be important to monitor developments in the urban fuehvood situation, and to ensure that uncontrolled fuelwood use does not damage the forests. 2.40 Charcoal, like wood, has a large decentralized distribution system. With lower transport costs (per unit energy), charcoal can be moved further, and there is reportedly some inter- island charcoal trade. Charcoal production is generally carried out under the supervision of the forest service, and requires a special licence. Charcoal is a minor urban fuel in Indonesia, in contrast to Thailand, for example, where charcoal is the most important urban household cooking fuel. Although about 15% of urban households use some charcoal, total urban household consumption in 1987 amounted to only about 60 thousand tons. This reflects a very low average consumption among users (less than 4 kilogram per month). Charcoal use varies widely between regions, but rarely exceeds one kilogram per household/month on average. - 21 - m. URBAN HOUSEHOLD ENERGY USE PATTERNS 3.1 In this chapter, the focus shifts to the urban households' use of fuels. The chapter begins with an overview of recent changes in urban household fuel use patterns, as evidenced by the tri-annual national household expenditure survey (SUSENAS), covering roughly 15,000 urban households each year for 1981, 1984 and 1987. This is followed by a more extensive analysis of fuel use patterns in Java in 1988, drawing on a detailed re-survey of 2700 SUSENAS households undertaken for this study. 2k/ Next, cooking, the most important household use of fuel, is examined with regard to fuel mix, cooking technology, and the factors affecting fuel choice. For this, the survey data are complemented by laboratory work on kerosene and LPG stoves, and in- depth surveys and stove testing in twenty households in Jakarta. The third and final section covers lighting, both electric and kerosene. This chapter is intended to provide background for ensuing discussions of energy policy. E7/ Urban Household Fuel Use - All Indonesia: 1981 to 1987 3.2 Energy policy has played an important role in changing urban household energy use patterns between 1981 and 1987. PLN's ambitious urban electrification strategy and the government's decision to increase the price of kerosene were the dominant policies influencing urban residential energy use patterns displayed in Figures 3.1 and 3.2. 2ff These changes have contributed significantly to the national decline in kerosene consumption, noted in Chapter IL Urban household kerosene use fell by about 2% per year, despite urban population growth of about 5% per year over the period. Simultaneously, urban electrification contributed to the national increase in electricity demand, and kerosene price increases have slowed, and even reversed, the overall shift away from biofuels in urban households. 3.3 According to SUSENAS data, the percent of urban households electrified grew from 47% in 1981 to 74% in 1987. Over the same period, kerosene consumption in urban households declined from 46 MJ/Household/day to 30 MJ/HH/day on average. Roughly 5 of the 16/MJ/HH/day decline in kerosene consumption can be attributed directly to displacement of 4/ The swy upon hich denand anaysis is based was undetake by die Buo Pusat Statisk ndonesia as patt of the Indonesia Uban Household Ene& Strategj Study (UHESS), administered by the Wodd Bank n coopemon wth the Directorte Genemi forNew Energ. The sample of 2702 househords in urban Java was ddmwn fom the SUSENAS sample flume, sutifled pAponionadly by utban area size. Fxendstum data is based on the SUSENAS household eenu moduk, wi eneaey quetons wae undenaken for the UHESS. A detaied account of the suvey design and sunmay tables of sults are pmsented in Appendix L ,/ Tables and teAt can only begin to indiate the relevance of household eneagy data. The UHESSsrwmeyadtshave been assembed in a data base, since many nantpoiey issues cannot be anidpated, but wise unexectedly in the cowe of government. In thefuture, a mited eneigymoduk should be added once eveay threeyeas to the SUSENASswuvy, along with the Expenditure Module. A &ahft of such a module is included as Annexv I Thiu wodd pmvde an invaluable sowre of infornation, and the basis for monitoring the govnmnts household enetV strtv. j8/ Consmption estmates h Fgure 3.2should be inteeted with care as they are de,ivedfmm expdue ww The avesuge wood use by wood-usinghoweholdsfrum the UHESS suvey is applied to the sham of SUSENAS households nporing wood use. Eecuidiy coWnp is not displayed as elecy consumption data in SUSENASisunrelible. -22 - Figure-1: Percent of Urban Households Using Each Fuel: 1981 - 1987 % of Urban Households Using Each Fuel 100% - 80% 1981 60%- 1987 40% - 20%- 0% Electricity LPG Kerosene Charcoal Wood Soirce: SUSENAS 81, 84, *87. Egurl 3.2: Urban Residential Combustible Fuel Use: 1981 - 1987 Urban Household Use of Combustibles by Expenditure Group and Year MegaJoules/ household/ day 60 Low Moderate Hi h 40 - 20- 0- 1981 1984 1987 1981 1984 1987 1981 1984 1987 = Wood 3 Kerosene Charcoal g LPG Sotrce: SUSENAS 81, '84, *87. - 23 - kerosene lighting by electricity in newly electrified homes. In addition, rapid electrification may have been the primary factor contributing to the declining share of households using charcoal from 28% in 1981 to 16% in 1987, as electric irons replaced charcoal irons. 3 MJ/HH/day of the residual reduction in kerosene use (11 MJ/HH/dck) can be attributed to substitution by LPG and wood, while 8 MJ/HH/day are due to outright conservation of kerosene in response to the 150% increase in the real price of kerosene faced by households over the period. 22/ 3A Despite the decrease in combustible fuel use, households were spending on average a 50% larger share of their income on energy in 1987 than in 1981. Again, this is the combined effect of increased electrification and rising kerosene prices. As illustrated in Figure 3.3, increased expenditures on electricity are highest in low income households. Increased expenditures on combustibles are most striking in the middle income groups, where households did not switch to wood as frequently as the low income households. 30/ 3.5 Since most of the detailed analysis summarized in this report is based on the UHESS survey conducted only in Java, it is important to ascertain the extent to which Java can be taken to reflect Indonesia as a whole. The regional variation in fuel use patterns in 1987 are summarized Figure 3.3: Percent of Household Expenditures on Fuel: 1981 . 1987 Percent of Urban Household Expenditures by Expenditure Group and Year % of Household Expenditures by Fuel 12% - 1 °LPG _ Electricity Moderate - Charcoal 8% ~~~~~~~~~Kerosene 21wood 4% -High 1981 1984 1987 1981 1984 1987 1981 1984 1987 Soirce: SUSENAS 81, 84, *87. &/ See Annea Ilfor detailed anasis. ,/ Theem is some doubt as to whete the lower bIome howholds 'swktched to wood, or s.mplj did not swith to kemsene despite wbanizaton. The 1987 samplk includes a consderbe number of houweods lng in areas which, in 1981, wem stll nuuI. .24 - Table 3.1: Urban Household Energy Use Patterns by Island Group, 1987 1. Percentage of Households Using Each Fuel: Electricity LPG Kerosene Charcoal Wood Sumatra 692 3X 942 26% 24X Java ?62 52 932 162 222 Nusa Tenggara 64X 2X 86X 10X 46X Kalimantan 692 1X 97X 4X 33X Sulawesi 75X 6X 902 2X 332 Haluku & Irian Jaya 572 02 962 82 362 Indbnesla 74X 4X 932 162 242 It. Average Fuel Use per Using Household: Electricity LPG Kerosene Charcoal Uood* kg/mo It/mo kg/mo kg/mo Sumatra NA 18 25 4.1 140 Java MA 18 29 3.4 140 Nusa Tenggara NA 15 17 3.4 140 Kalimantan VA 17 22 7.6 140 Sulawesi VA 16 22 2.2 140 Maluku & Irian Jaya VA MA 26 2.1 140 Indonesia NA 18 28 3.6 140 -11. Average Fuel Use per Household: Electricity LPG Kerosene Charcoal WodO kg/mo lt/mo kg/mo kg/mo Sumatra HA 0.48 24 1.1 34 Java MA 0.94 27 0.5 31 Nusa Tenggara HA 0.23 14 0.3 64 Kalimantan HA 0.14 22 0.3 46 Sulswesi HA 0.94 20 0.1 46 Maluku & Irian Jaya HA 0.00 25 0.2 50 IndNnesia NA 0.81 26 0.6 34 IV. Total Urban Household Fuel Use in Region: 2 Electricity LPG Kerosens Charcoal Wood* Households Ton/mo kIt/mo Ton/mo Ton/mo Sumatra 152 HA 600 31,000 1,400 45,000 Java 72m MA 5.90 172m500 3.500 198.000 Nusa Tenggara 32 HA 100 3,800 100 17.000 Kalimantan 42 IA 100 8,000 100 17,000 Sulawesi 5X HA 400 8,900 0 21,000 Maluku & Irian 12 HA 0 2,300 0 5,000 Indonesia 1002 NA 7.100 226,500 5.100 302.000 * Assumed average wood use of 140 kg/mo per wood-using household derived from UHESS survey results. Source: SUSENAS, 1987. -25 - in Table 3.1. The data suggest that regional variation is significant, and that urban households in Java tend to use more 'modern" fuels (electricity and LPG), and less wood fueL It would, therefore, be preferable to include all the island groups in the analysis. This notwithstanding, the differences between Java and All Indonesia are small. Far larger differences are evident between the energy use patterns of All Indonesia in 1981 and those in 1987 and between the energy use patterns of different income groups than between regions. Similarly, as will be demonstrated below, there is considerably more variation between the energy use patterns of large cities (greater than 500,000) and other urban areas in Java, than there is between Java and the rest of urban Indonesia. Consequently, projections of energy use in urban residential Indonesia, presented in this chapter, are based on patterns from urban Java. 3.6 Neither electrification nor kerosene price increases can be expected to play as important a role in the future as they have in the 1980s: urban electrification is nearing saturation and it is not likely that the multifold kerosene price increases of the early eighties will be repeated. By implication, changes in energy use patterns over the past decade ought not be mechanically extrapolated into the future. In the 1990s, rapid growth in urban electrification can be expected to have less influence on kerosene demand due to lighting substitution than it did in the 1980s. Likewise, consumption of kerosene in urban households will probably grow slowly, unlike the declining consumption from 1981 to 1987. Urban Residential Fuel Consumption and End Use - Java 1988 3.7 In the preceding section, the effects of electrification and kerosene price increases on urban residential fuel use and household expenditures were illustrated using data from national expenditure surveys covering a six year period. In this section, summary results of the UHESS survey undertaken for this study are presented. i/ The UHESS survey was designed to examine the effects of price, income, and access to fuels on fuel choice and use to inform pricing and distribution policy in the urban residential sector. In addition, it surveyed consumer preferences and attitudes toward each household fuel. Analysis on which this summary is based can be found in Annex IV and Appendix I. 3.8 First, use of each fuel in households in various income groups and urban area sizes is briefly summarized. Then, as almost 90% of combustible fuels 32/ consumed in urban households are used for cooking, estimates of the costs of cooking with each fuel are compared and consumer preferences are summarized. In addition, a statistically derived measure of interfuel substitution in the cooking end-use is employed to display the effects of income and urban area size ,U/ Two of the diffenences between the rsults of SUSENASforJava and the UHESS wanatmention. F&sM, in the UHESS swv, 85% of the househoUds were using elec#icity, while in the SUSENASxSM ofJava ayeareadrie, only 76% used elecniay. Whilk peduaps in pan the esukt of elecinfication, a more lke eq'lanadon is #hat SUSENAS does not capwr all eectricity use, padicadwfy if it is not paid for by the user. Second, the avenzge household in the UHESS swvey used one ler of keosene per day, while the avewge SUSENAS househod in wban Java used slightly kss than .9 lk per day. Again, the most likely explanation is that SUSENAS is not an energy orented swvey, and may have udestmated kerosene use. .U/ tn tenns of enev content. .26- abkl 3.2: Percentage of Urban Households Using Each Fuel by Income Total Nznthly Expenditure per Household 000Rup1ah Hissing / <75 75-120 120-185 185-295 )'295 ALL X Valid Households 27X 24X 23X 16X 10 100X ELECTRICITY 86t 65X UK 95 97 1lOOK 85X Cooking 1X OX 1X 3X 3X 14X 2X Lighting 85X 63X 83K 94K 96K 98X 84X TV 40K 22K 47K 67X 73n 89 52X Ironing 37X 16X 34X 57X 68K 80K 44X Refrigerator 7X 1X 3X 9 21K 46X 11X Air Condftioning OX OK OX OX 1K 2K OX Purping 6a 1X 3X 7X 16X 35X 9X Washing Wachine OK OX 1K 0K 1K 11K 1X _ ROSEIIE 81X 87K 87X 92X 91K 78X 87K Cooking 73X 53 78 89K 90K 77K 75K Lfghting 17K 39K 18K 7K 5K 2K 17X Econ.Act 2X 3X 2X 2X 3K 2X 2X Other OX 2 X 1% OX OX 1X 1X LPG 2X OK 2 43 ex 29KX 5 Cooking 2X 0O 2X 4K 8X 29X 5S W.Heater 0O 0 0 0 OX 0 0 utoRIEL 14X UX 26X lKX 8X 7X 22X CARCAL 17m 33X 39X 24K 16X 12X 27S Cooking 2X 4X 2X 2X 3K 3K 3X Ironing 15K 30K 36X 22K 13X 8X 24X Other 1X 2 X 2X 1K 1K 2X 1K al Expenditure infonmation was not available for households In the "Missing" category. Percentage base for all fuel use percentages is the nrumer of households in each expenditure category. Source: UHESS 1988. on cooking fuel use. Finally, summaty results of analysis done on lighting. the second largest household energy use, are presented. Again, statistically derived estimates are employed to show that though the economic costs of lighting with kerosene and electricity are similar, lighting levels obtained with electricity approach ten times that of kerosene. 3.9 Electricity is widely used at all income levels, but, predictably, is consumed in much larger quantities by the rich. Thus, almost two thirds of the households in the lowest income group use some electriity, but households in the highest income group use on average about ten times as much electricity in a month as a poor household: 170 kWh as compared to 15 kWh. Even lighting, which accounts for about 50% of overall electricity use, increases by a factor of more than four across the income groups. Most of the increase, however, is the result of new uses of electricit, adopted as income rises. The only electric devices used by more than a quarter of the households in the lowest income group are lamps. As income rises, lamps are joined by television, ironing, refrigeration and, finally, water pumping. The order of importance of the different -27- Table 3.3: Urban Household Fuel Use by Income Total Monthly Expenditure per Household 000'Rupiah Missing 75 75-120 120-185 185-295 v295 ALL Family Size 4.2 3.5 4.8 5.1 5.7 6.; 4.7 ELECTRICITY (Wh/m) a/ 38.5 14.9 30.0 50.9 80.7 169.4 51.6 Cooking OX 0 0 OX 02 2X 1X Lighting 56X 90X 74X 61X 462 35X 532 TV 9X 92 11X 12X 92 7X 92 ironing 72 72 8X 9X 8X 5X 7X Refrigerator 102 2X 52 8X 122 142 102 Air Conditioning 02 02 02 02 12 3X 12 Pumping 12 2X 2X 42 62 92 52 Uashing Nachrne OX 02 0 0 0 22 1X Residual 17X -10O -1X 7X 18X 24X 132 KEOSEIE (ltldoy) / 0.9 0.8 1.0 1.2 1.3 1.2 1.0 Cooking 842 612 842 892 94X 902 85s Lighting 9X 222 112 32 2 1X 8X Econ.Act 62 9X 52 7X 42 9X 72 Other 02 0 0 OX OX 1X 02 U6 (kgid4y) 0.01 0.00 0.01 0.03 0.0? 0.24 0.04 Cooking 1002 1002 1002 1002 1002 912 952 Uster Neater 02 02 OX OX OX 9X 52 BIOREL (oIWdoy) O.S 2.1 1.1 0.5 0.4 0.3 1.0 CNCAL tkg/dsy) gM 0.04 0.07 0.08 0.05 0.05 0.03 0.06 Cooking 30X 34X 132 272 382 702 272 Ironing 352 482 562 592 482 392 512 other 332 152 302 112 122 502 192 Fuel use figures represent average household fuetl use of households in each income category ihi t percentages indicate percent breakdown of these means for each end-use. / Breakdown for electricity I. based on a combination of survey responses regarding appliamce wattage and use-levels, and statistical analysis of relation between appliance ownership and electricity use of metered households. Breakdoun for kerosene Is based on direct physical masrement. Breakdown for charcoal is based on average charcoal Ironing use among households only ironing, extrapolated to households with more than one use. Sourse: UHESS 1988. electricty uses is almost identicaL with the exception of refdgeration which, due to its high per unit consumption, is the second most important appliance in terms of overall electriity use. Sources of electricity also vary significantly with income. As shown in Figure 23, almost half of the low income households obtain electricity via their neighbor, but the fraction fails to 5% by the uppermost income group. 3/ U/ At uper kome lkwbs, te use of elecg via a neighbors met its most Xely to be for med howing wes de ownerhas dse mtetr. Gme d t deko*i tffdicoga mul4owuehodmeting if die combined ad isso high as to teub dt the subsiberpay one of &e highload twiffs - 28 - 3.10 Larger cities tend to be more electrified, with eiectrification reaching 95% in the cities over one million in population. Even in the smallest urban areas (<50,000 inhabitants), however, more than two thirds of the households are electrified. Some of the variation with urban area size is due to the effect of lower average incomes in smaller urban areas. Nevertheless, there are clear indications that electricity is somewhat less easy to obtain in the smaller towns, and that this does affect electricity use. Households in small urban areas that do not use electricity are more likely not to live in the vicinity of the electric grid. Moreover, urban area size explains a statistically significant share of the variation in electrification, even controlling for income. Kerosene 3.11 Overall, kerosene use does not appear to be closely related to income: the percentage of households using kerosene peaks in the middle income group, and kerosene use in the highest income group is only 50% higher than in the lowest. Beneath this moderate increase, however, lie two stronger but countervailing tendencies: kerosene lighting falls rapidly to zero by the middle income group, while kerosene use for cooking doubles between the lowest and highest income groups. Overall, cooking accounts for 85% of kerosene use, with a further 8% and 7% each for lighting and economic activities using kerosene. Most households (75%) obtain kerosene from small shops in their neighborhood, with a further 22% receiving deliveries and 7% purchasing directly from dealers (4% purchase from a combination of sources). Presumably to avoid transport mark-ups, poor households are less inclined to receive deliveries. On the other hand, being unable to purchase sufficient quantities, the poor are also the least inclined to go directly to dealers. 3.12 Average kerosene use grows with urban area size, but again lighting moderates the prevailing tendency, which is set by kerosene cooking. Kerosene lighting, as expected, mirrors electrification and declines with growing urban area size. Kerosene delivery, the only way of obtaining kerosene for more than a third of the kerosene using households in the largest cities, is relatively insignificant in the smaller urban areas. Liquid Petroleum Gas 3.13 Virtually no households with income below 100,000 Rupiah per month use LPG, and, despite the small sub-sample of households using LPG (138 of 2702 households surveyed), there is clearly a greater tendency to use LPG as household income increases. Almost all of this LPG use is for cooking, M/ although water heaters were found in a few households in the uppermost income group. Having chosen to use LPG, the income dependent variation in fuel use is less than among kerosene users: among LPG users the average cooking fuel use only grows from .6 to .8 kilograms per day between the lowest and the highest income groups (see UHESS Summary Table S in Appendix I). The poorer households are more likely to go to a shop to collect their LPG than to have it delivered (see UHESS Summary Table 8 in Appendix I), but this is probably as much due to location as to attempts to economize. In addition, purchase from a shop avoids the perceived problem of delivery trucks selling underweight bottles (see para. 2.25). ,V Cooling mfeis to the use of stoves and owvs, en if water heated on the stove s for bahing. Table 3A: Percentage of Households Using Each Fuel by Table 3.5: Household Fuel Use by Urban Area Size Urban Area Size Urban Area Size ('000 persons) Urban Area Size - 000 Persons ASO 50-200 200-1,000 1,000+ ALL <50 50-200 200-1,000 1,000+ ALL ELECTRICITY 67X 751 8DX 95X 85X ELECTRICITY-bih/mo f 22.1 32.3 38.6 78.1 51.5 Cooking 1X 21 2X 4X 2X Cooking 0X 1X 1X 1X 1X Lighting 63X 741 88X 94X 84X Lighting 851 751 701 421 531 TV 36X 421 521 621 521 11X 11% 10X 8X 9X Ironing 25X 34X 43% 561 441 Ironfng 8X 9X 71 7X 7X Refrigerator 31 5X 71 181 11X Refrigerator 6X 4a 81 12X 101 Air Condition 0X 01 01 1X 0X Air Condition 0X 01 0X 1X 1X Pusping 31 3X 5X 151 91 Puping 31 31 3X 6X 5X aasl, achine 01 01 1X 21 1X Wash Zahfne 01 0X 1X 1% 1X Econ. Ac 21 3X 61 8X 61 Residuat -131 -2X 11 221 131 XEXSElEE 91X 7YX 93X KEROiE-lt/dey a/ 0.8 1.0 0.9 1.2 1.0 Cooking 491 76X 691 911 751 Cooking 681 831 821 911 851 Lighting 401 291 141 6X 171 Lighting 231 131 81 31 81 Other 1X 41 1X 01 1X Other 9X 41 101 61 71 LPG 3 4X as 5X LPG-.bJday 0.01 0.03 0.04 0.06 0.04 Cooking 1X 3% 4X 8X 5X Cooking 1001 951 911 96X 951 Water Heater 0X 01 0X 01 01 Water Neater 01 5X 81 41 5X 8IORMEL 551 30X 25X 3X m SIORfEL-kgldsy 2.7 1.3 0.9 0.1 1.0 CUAWMAL 331 40% 32X 171 27X C3 AL kgidwy a 0.06 0.8 0.10 0.02 0.06 Cooking 4X 2X 7X 01 31 Cookifns 31X 141 411 21 271 Ironing 29X 361 261 171 24X Ironi-mg 49% 51X 381 841 511 Other AC 1X 3X 2X 1X 1X other 201 35% 211 141 221 Percentage base is the nuater of households in each urban area size Percentages are breakdowns of mean fuel use of households in each urban category. area size category. Source: UNESS 1988. pj Breakdown for electricity is based on a combination of survey responses regarding appliance wattage and use-levels, and statistical analysis of the relation between appliance ownership and electricity use of metered households. Breakdown for kerosene is based on direct physical measurement. source: UHESS 1988. - 30 - 3.14 There is also a greater tendency to use LPG among households in larger urban areas, but there is some use of LPG even in the smallest urban areas. Deliveries are relatively uncommon in small urban areas, but become the major mode of obtaining LPG in cities of over 500,000 (see UHESS Summary Table 13 in Appendix I). Wood and Crop Residues 3.15 Wood is used almost exclwusively for cooking. The fraction of households using wood declines sharply with increasing urban area size: from more than half the households in the small urban areas to 3% in the largest. Even among wood-using households, wood use declines as urban area size grows. About half of the wood-using households collect all their wood, while almost a third purchase all their wood. Purchasing wood is the least important source of wood in the largest urban areas, where the little wood that is used is scavenged from a variety of minor sources, including construction projects. Fuelwood sales becomc more important in middle-sized urban areas, but in the smallest urban areas, own-collection again dominates (see Figure 2.5). 3.16 Wood use falls off rapidly with income, though a small share of households use wood even at upper income levels. As with LPG, the change in wood use with income is largely a function of fuel choice, with wood-using households showing little variation in quantities used across income groups (see UHESS Summary Table S in Appendix I). Surprisingly, there is no significant relation between income and whether or not a household collects its own wood (Figure 2.5). Cookin 3.17 The foregoing discussion indicates that cooking is not only the major energy using activity in urban households, but that pricing policy, income growth, and access to fuels significantly influence cooking fuel choice and consequently, overall fuel use patterns. Cooking is an extremely complex actMity, difficult to distill in tables or text. Cooking techniques and ingredients vary across Indonesia, and even within Java. Yet at least a minimal understanding of the cooking process, and its context, is needed if a household energy strategy is to be developed. A summary of survey results pertaining to cooking practices, utensils, and stove ownership can be found in Annex VI. It contains pertinent sociological information, for targeting effective kerosene conservation policy and has been used in designing the proposed kerosene stove improvement program summarized in Chapter IV and detailed in Appendix HI. Below, the daily cost of cooking and consumer attitudes toward each fuel are briefly analyzed and variation over income groups and urban area sizes in the shares of cooking demand met by each fuel are displayed. Kerosene 3.18 The average daily cooking fuel cost of households using only kerosene for cooking is 235 Rupiah. A kerosene stove can cost between 3,500 Rupiah for a inexpensive artisanal stove, to 15,000 Rp for a local factory stove of good repute, to upwards of 50,000 Rp for an imported Japanese kerosene stove. Traveling stove peddlers offer a variety of credit schemes, and return weekly or monthly for payments. While the implicit interest rate is likely to be as high as 150% -31 - Table 3.6: Attitudes Toward Kerosene for Cooking Table 3.7: Attitudes Toward LPG Main Reasons for Cooking with Kerosene (Households Using Kerosene for Cooking) Naln Reasons for Using LPG (Households Using LPG) Kerosene is easy to get 48X Kerosene Is cheap 13X Stove came with house 2X Kerosene is easy to use 31X Easy to get LPG 13X Kerosene is clean 6o LPG is cheap 14X other 3X LPG cooks quickly 51X LPG is clean 19% Total Users 100O Total Users 100O Nain Reasons for Not Cooking with Kerosene (Households Using Other Fuels for Cooking) Cooking Fuel Used Wood LPG Either Nain Reasons for Not Using LPG (Households Not Using LPG) Difficult to obtain 2X oX 1X Equipment expensive 22K 1K 15X No distributor 16K Fuel expensive 32K 14K 26K Road too narrow 2% Dirty 6K 41K 18K Equipment expensive 34X Time consuming 8K 33K 16K Fuel Expensive 23X Unsafe 13K 2K 9° Unsafe 11X Other 16K 9K 14X Other 13X Total Non-users 100K 100 100 Total Non-users 100 Source: UHESS 1988. Source: UHESS 1988. per annum, 3/ this allows households to replace broken stoves even when they are facing a severe cash shortage. 3.19 A summary of the main reasons why kerosene users choose to cook with kerosene is given in Table 3.6. Almost four households in five cite the ease with which kerosene can be obtained or used, and most of the remaining households cite the low cost of kerosene. These answers provide little insight into the process of fuel choice: for the most part they simply confirm the widespread availability of kerosene, and how routinely it is used. More interesting are the reasons why non-users choose not to use kerosene. Here, it is important to distinguish between those who use wood and those who use LPG. The main reason most wood users give for not using kerosene is either that kerosene stoves (22%) or kerosene itself (32%) is too expensive. While one would expect wood users to find cost the major obstacle to kerosene use, with cheap kerosene stoves available at less than 4,000 Rupiah, it is revealing that the cost of a kerosene stove is felt to be almost as significant an obstacle as the cost of kerosene itself. Most LPG users claim not to use kerosene primarily because it is too dirty (41%) or too slow (33%). In short, they believe they are using a higher quality fuel W!~/ A common sdcem forpurc g a SW Rqpiah kme stow is to pay IO PRsquah uq p t fnw md lXV) in five mNt ksalmenb. - 32 - Liquified Petroleum Gas 3.20 The average daily cooking fuel cost of households using only LPG is 480 Rupiah. Even the most basic LPG system is likely to cost upward of 130,000 Rupiah. For two bottles, two burners and a regulator, a household can expect to pay at least 250,000 Rupiah. A more extensive analysis of the costs of using LPG as compared to kerosene is provided in Chapter IV. Even a cursory examination indicates, however, that LPG is considerably more expensive than kerosene. 3.21 The reasons most LPG using households give for choosing LPG mirror their main reason for not using kerosene. The fact that LPG cooks quickly is cited by most households (51%), with cleanliness the second most common response (19%). Households that do not use LPG for cooking cite equipment expense as its major disadvantage (34%), followed by fuel costs (23%). Contrary to kerosene, equipment costs are seen as the larger obstacle. Wood 3.22 The financial costs of using fuelwood are not a reliable indication of the costs a household incurs to use fuelwood, but more appropriate cost estimates are difficult to obtain. Imperfections in the fuelwood market are as important as supply costs in price determination. In any case, wood quality varies widely and in practice most wood users consume a number of alternative biofuels when tree-wood is not available. Collection times are not available, and would be difficult to interpret. A traditional wood-stove can easily be commissioned in areas where wood use is common, but little is known about the costs except that they are almost certainly low. Generally, it is reasonable to assume that wood is a relatively inexpensive fuel in areas where it is widely used. 3.23 F o r Table 3.8: Attitudes Toward Wood most wood-using ouseho ld the in Main Reasons for Using Uood Main Reasons for Not Using Uood reaon fos, he maing Wood Using Households) (Households Not Usfng Wood) reasson for choosing fuelwood is either that Easy to get fuetuood 412 Too difficult to get 192 it ises obn wood is cheap/free 37X Too expensive 2X it iS easy to obtain Kero stove expensive 3X Too dirty 282 (41%) or that it's Prefer wood-cooked food 8X No proper kitchen 15X financial cost is low Wood gives hot flame 8 Too time corning 322 (37%). This should not be taken to Total Users 100% Total Non-users 100X indicate that wood- Source: UHESS 1988. users dislike fuelwood: more than a third of wood fuel users also cited, as secondary reasons for using wood, that food cooked on wood tastes better and/or that wood can produce a hotter flame than kerosene. On the other hand, households not using fuelwood do cite inferior qualities of fuelwood as the major reason for not using it. - 33 - Cooking Fuel Choice and Inter-Substitution 3.24 The supply infrastructure, daily cooking fuel costs, and attitudes toward each fuel help to explain why households choose the fuels they do and how much of their chosen fuel a household will use. These insights do not, however, provide a basis for quantifying either the inter- fuel substitution ratio (i.e. the amount of LPG or fuelwood used to substitute for a given amount of kerosene and vice versa) or fuel choice. For this, empirical analysis of household behavior is necessary. 3.25 Controlled cooking and efficiency tests undertaken for this study, suggest that efficiency, maximum power, the power turn-down ratio (the ratio of maximum to minimum power), and even the amount of water used to steam rice, can all have a significant effect on fuel use. These complexities preclude a physically-based estimation of, for example, how much wood or LPG is required to replace a given amount of kerosene. Instead, a fuel-neutral measure of household cooking fuel use: a "liter of kerosene equivalent (LKE)" (the quantity of LPG or wood required to substitute for one liter of kerosene) can be estimated from observed patterns of fuel use in the data. 3.26 The statistical derivation of LKE measures for wood and LPG is presented in Box 3.1. Essentially, survey results indicate that, controlling for income and family size, an urban household on Java that uses wood for cooking uses, on average, 4.25 kg of wood for each liter of kerosene used for cooking by a similar household that cooks with kerosene. Likewise, households using LPG for cooking use, on average, 0.59 kg of LPG for each liter of kerosene used by similar households cooking with kerosene. Hence, in the cooking end-use: 4.25 kg of wood = 0.59 kg of LPG = 1 liter of kerosene. The LPG-kerosene substitution ratio indicates that LPG use demands somewhat more LPG than otherwise predicted on the h-sis of relative efficiencies alone. The difference is statistically insignificant, but in the expected direction, given the higher maximum power of the LPG stoves. The wood-kerosene substitution ratio indicates that wood users consume significantly less wood than expected on the basis of the relative efficiencies often cited, perhaps because urban households use their wood stoves at a lower power than is typically used in laboratory tests of wood stoves (laboratory powers are probably selected with rural users in mind). 3.27 With this measure, the shares of cooking demand met by each fuel can be displayed against household income and urban area size as in Figures 3.4 and 3.5. These figures give compelling support to the claim that LPG is a fuel of upper income groups. Wood use appears to be as much a function of income as of access and availability (as gauged by city size). Quantities of each fuel (valued in LKE) consumed by households in each income group and city size category are also estimated and illustrated in Figures 3.6 and 3.7 Cooking fuel use in urban Indonesia was estimated by extending these fuel use patterns from Java to the urban population of Indonesia. It is evident that both income growth and urbanization will likely lead to changing fuel use patterns. The advantage of having the results in this form is that they allow precise quantification, which becomes particularly important in the context of predicting future fuel use patterns, and assessing inter-fuel substitution strategies. .34- Box 3.1: Derivation of "Liters Kerosene Equivalent" for Cooking Fuels The susitt is d fti sdy ae s d fHom e S ey ts on : schol in- only one fel for kin ( 0% of the smple). Mos of th,ese houblds:: :(76%)cok*uivel: ~it ker... ~Vu& ok wihod and 4%: coo w;th LPG0. Tie: analsis below is intded: to ie imte-, the ct- of fuel choice on f uel use, while controlling for e -ineesof h""( sa.l income. results of a staistica (0C1) esiiat of substitutin :ratos are give be : ; -~~~~i .01( . .....,..-, ooM ...t In(e) , 2.14,-+ 39nfm)+ .O?In(exe .21(L,PO + .4(W?odYr)'''". t 0.225 Significance .00OM Number of Cases 2204 w-ere- en . houlsehiold cooki fel use xpressed in MeJ /day Ap :hiousehold total expenditre per mon . : f. household size :- -:: Y =lbouseitholduses$ or cooki.g 0 if household does not ue IP for cooing "WoodY I if houehld uses wood for ci-g .0.> f hos does n:t use ::w;hood fr cking , ~~~~~~~~~~:. __ ..f >.: .' .The c ient t Y es tha'thoso ilds usLPg consumeo on average, 19% | e-rg fo :,oin than. noP usinosholds(1-e, ,.19)., Houise3holds *cookin wih ::od consume, ci average, 62% mere energy for cking ltban oiher h olds (eM.1=.62). T ransormed:into ph~lc units, thse coefcientscan be loosely interp.reted to iniply on -f roi~ghy..62 kIlogra ofL or:4.1 I fwood per:liter of kesne replce A sl-y mor mple frmution hc eff ely treas the*ood-krosene sb on and the : e.O.btt i eparate quations(presented in xI) resltMs i more accate * es atesof.$:ikgofLPG(95%@onfidW nceinterval .S2kgto.7 :kg and 4.2S kgowood (95% i& . 19 kg to Li k) per liter fI kerosne. ee preci substittion ratos ar - stdinS4jz of oo fuel harcs, and for evaluan LPG 1 ...:... . ..... .. . :., >.. .. : .:. ...... ,- - 35 - Figure 3.4: Cooking Fuel Shares by Income Cooking Fuel Shares by Income Category Urban Java Cooking Fuel Shares Family Size & LKE/HH/day 75% 6 50%-MLP4 3 25% -2 10,000 100,000 1,000,000 Monthly Household Expenditures Source, UHESS 1988. Figyre 3.5: Cooking Fuel Shares by City Size Cooking Fuel Shares by Urban Area Size Urban Java Stares of Cooking Demand 100X-_ - 75%- 0% 5,000 50,000 500,000 5,000,000 Urban Area Size Source UIESS 1958. - 36 - Figure 3.6: Cooking Fuel Use by Income Annual UJrban Cooking Fuel Use on Java by Household Expenditures Annual Cooking Fuel Use (LKE Millions) 350 - Kerosene 300l Wood 250 ° LPG 200- lcrct 150 100- 50- 14,000 45,000 140,000 450,000 Household Monthly Expenditures (Log) Source: UHESS 1988. Eigre 3.7: Cooking Fuel Use by City Size Annual Urban Cooking Fuel Use on Java by Urban Area Size Annual Cooking Fuel Use (LKE) (Millions) 700 600 500 400- 300 200 -Wo 100 vllfge Town Cly.2m Jakarta Urban Area Size (Population) Source: UHSS 1988. - 37 - Lighting 3.28 Lighting is the second largest household energy use, accounting for 10 percent of kerosene use and 50 percent of electricity use. The shift to electric lighting in urban areas has been a factor behind declining kerosene use in recent years. With electricity being used by 85% of urban households on Java (according to the UHESS survey), and probably by over 75% of urban households on other islands (see Table 3.1), the importance of this shift for displacing kerosene in the future is diminishing. Nonetheless, lighting remains important, not only because of continued electrification, but because the large differences in energy efficiency between fluorescent and incandescent lamps may offer significant opportunities for electricity conservation in the near future. 3.29 For households that must use kerosene for lighting, three main types of lamps are available: the pressurized lamp ("petromax"), the glass-shielded wick lamp ("semprong"), and the small open wick lamps ("teplok" and "sentir"). A pressurized lamp costs some ten times as much as a shielded wick lamp, while the price of an open lamp is much smaller still. A pressurized lamp can be 4 to 8 times as efficient as a shielded wick lamp, but typically gives off more than 8 times the light leading to higher rather than lower kerosene use per hour. Similarly, the open lamps are the least efficient, but use the least quantity of kerosene per hour. Only the pressurized lamp gives out a comparable amount of light to an electric bulb. The low-flux open lamps function rather like "night-lights", revealing little more than the outlines of nearby objects. 3.30 Once a household is connected to the electric grid, electric lamps compare favorably to kerosene lamps in almost every respect. They have higher energy efficiency, cost effectiveness, technical performance, and practical adaptability. Lighting data analysis, summarized in Box 3.2, similar to that undertaken to determine substitution ratios for cooking, result in an estimate that a household using electricity for lighting will use, on average 1.38 kWh for each liter of kerosene used for lighting by a similar household without electricity. Both the economic cost of supplying 1.38 kWh to urban households and the average price faced by households in 1988 was below the respective cost estimates and average prices of one liter of kerosene. Statistical analysis also indicates that households lighting with electricity enjoy, on average, nine times more light than their kerosene using counterparts. 3.31 For households with electric lighting, the principal choice is between fluorescent and incandescent lighting. Modem fluorescent lamps are four to six times more efficient than ordinary incandescent lamps. Furthermore, incandescent lamps are more sensitive to voltage fluctuations: insufficient voltage decreases efficiency and overly high voltage reduces the expected lifetime. 3.32 The disadvantages of fluorescent lamps are the higher fixture and bulb costs and, for some users, the lower color quality. However, the color does not seem to be a major obstacle in Indonesian households as the share of fluorescent lighting almost doubles from the lowest to the uppermost income group (See Table in Annex IV). This strongly supports the view that poorer households are choosing incandescent lamps because they bear lower up-front costs. - 38 - Box 3.2: Lighting Analysis: Electricity Substitution for Kerosene The a i pbed b s t t te ) t amount of elericity quired to displace kero ighti ad ii) te ra n in amoun ofg obtained by such a switch Surve dait Iit that a l bti a electricity conneion, it stops glight with kerosene. A statistical (OLS) emation p e is performed on tJd ESS dita from urban Java, while controlling for thle influenes of l*nisbhold size, income, number of rooms, and city s;iz, ial of which affect lighting fuel choice and levels iofuse. The, results are given below: ..en 2 . he,64n(#o ) + R.23njp + .04n(C:ity) + ....................... 9ln(E;am)"- t9(leei)' ..........o t: (0.84 (225) . (0.7) (6.2 (3.) (-52.5) - 39 :Sigificance Numbr Of. Cases m 2379.. -where; en : househadiold lightiig fuel use expressed in Megaioules/month #Rooms nubeof oms in ibe household Exp . household total expenditure per month : City:: popWationof contous urban area :::Pam : - household size -.Elecht . l if electricity isused for lighting -:. w0 if electrity is not sd ker is used for lighting) The coeffiien associat with-ElecIndicates that households lightig with elecricity use, on avera,oly/6 of the eneiht by households ihat Ig with kerosene (86% less energy: 1 8). Trasfmred into oi usnit, this im a li sub on ratic of 138 kWh/liter of kerosene. Using estimates of the eco cost of supply to. urban residental consumers of ISO Rp/kWh and 290 Rp; the economic cost of supplying electricity to displace one -liter of eroe for:a¢tn (207 Rpt38 kwh) is roughy 71% o0f thie cost of suppling one liter of tkero Mre w av es faced bhseholds at the tme of the survey of 105 Rp/kVWh o e ii d 23Rp/l a household lih with electricit saends, on average, *z1y 71% of what a s hoehl lighting with kerosene would spend on lightfuel. .:r ._ ..- . y;SS ' S .' 'W: Fuel use es es by lm , derw from i e surey and presented in Annex IV, are employed with e aip efficaces ted i a rect Bain wkign paper ' to estimate the differential amountiof ligt otd by ouseholds using electricity to light relative to those using kerosenie. Results of astatistical (O:`S) procdure to estimat amount of light are given below; lnQ(m) .192 + .41n(#)koos +-.321n() + .061n(Cio) + .lo0l(Faia) + 2.31( ht) t: (6.) (21.0) (10.5) - (6.9) (2.8) (44.4) R . .68 Signifince: .0000 NumberofCases 2376 :where: 1nt r ho-uehold ighig expressd in kioLtumen-hours/moah The coetficient associtd witht in t that households lithting with elecicity enjoy, on average, nin' times more ligha those lighi with kerosene (e -1e-9 .). A/ "A Comparison of at tn in Dleveoping Counttiee, World Bank Industry and Enetr DPepartment Wld Paper, Ee Series Paper No. 6, June 1988. .39, Table 3.9 Lighting Electricity Use by Lamp Type (% of kWh) Hours/Day Ituminrated 4 3-12 12+ Total Ineandescent 5-25 W 1X 2XX 242 522 40+ w OX 102 S2 152 Fluorescent 420 U 02 72 42 112 20+ 0 2X 142 6 212 Total 12 59M 402 1002 Source: UHESS 1968. 3.33 A large fraction (40%) of the electricity used for lightig is for lamps buming 12 hours or more (Table 3.9). Among these long-burning lamps, the faction of kWh used in fluorescent bulbs (27%) is less than the fraction of kWh used in fluoresent bulbs overall (34%). This indicates considerable potential for electricity conservation. Economically, fluorescent lamps are more attractive for long-burning lamps, since saving are generated more rapidly and the initia' investment in fixture and bulb can quickly be recovered. Urban Household eWl Use Eections 3.34 The significant relationships in the UHESS data between income and fuel use were modelled to project urban residential energr demand through the year 2000. Details of the model construction are presented in Annex VII. Base case projections of fuel use in the urban residential sector of Indonesia are shown in Table 3.10. M/ 3.35 Results in Table 3.10 project both LPG and electricity use to grow rapidly at a rate of almost 10% per year over the twelve year period. However, these projections, based on variation in fuel use patterns with income, are very similar to offcial electricity and LPG projections in Repelta V. UHESS data shows that rapid growth in urban residential LWG consumption in the 1980s was due to penetration of a new and preferred fuel in upper income households and that further growth in the 1990s can be expected to be moderated by an affordability barrier in middle income households. Remarkably, even the total amount of biomass use is expected to grow. Nonetheless, kerosene consumption by the urban residential sector is projected to increase by 60% over current levels by the year 2000, providing the bulk of combustible fuel demand through the period. The growth rates in the last column of Table 3.10 should be compared to the projected urban population growth rate of 4.1%. M/ De base case p.jecdon aswiaes no *se thro Ae year 2 in i) mea pies ofjfel n4", i) access to, aaUabiy of, and preferna fr houehod fu& De bs cmase d,v. by *wo wgmous vwibks: i) kw-one 8ow4 of 2% p ecita pr anwn sa4 II) go" h e wberf a h lso4.1%p (awning no chmge It me houshd siz). * 40 - Table 3.10: Base Case Fuel Use Projections for the Urban Residential Sector Fuel 1988 2000 Annual Growth Kerosene ('000 kiloliters/year) 3,310 4,960 3.4X LPG ('000 Tons/year) 171 559 10.4X Wood ('000 Tons/year) 3,130 3,730 1.5X Electricity (GWh/year) 5,576 14,380 8.3X Source: Annexes VII and VIII. 3.36 Base case projections of combustible fuel use, denominated in JLKE, are displayed in Figure 3.8: both forward to the year 2000 and back to 1981 for comparison to SUSENAS derived estimates. The backward estimates were used to check how well the model fits past data. Given the inherent uncertainties in both UHESS derived estimates and those from SUSENAS, the model estimates of residential consumption of kerosene wood and LPG in 1981, 1984, and 1987 are remarkably similar to SUSENAS estimates. This close resemblance strengthens the conviction that the econometric equations modelling relations between household characteristics and energy use can be a useful tool in assessing future household energy demand. 3.37 Figure 3.8 shows that even though electrification can be expected to reduce urban residential demand for lighting kerosene, use of kerosene in the urban residential sector nationwide, Figure 3.8: Base Case Combustible Fuel Projections Base Case Combustible Fuel Projections Urban Residential Indonesia Billions of LKE/Year ° SUSENAS Kerosene O SUSENAS LPG _ 4 . SUSENAS Wood.....Wc.x. ............. erosene ~~for Cook ing 3. .... ...... 2.- ................................................................................................................................................. 1..O.ueIwood.-. 1980 1985 1990 1995 2000 Year Sce: Annexes Vil and Vill. * 41. Fgure 3.9: UHESS and PLN Urban Electricity Demand Projections Total Indonesia Urban Residential Electricity Growth by End Use (Gwh/yr) G2 h (Thousanct) 2-UHESS Projection O-i ottner 16- M water Pumping E ironing _ 12- E Teievsion PLN Projection 1= Refrigeation 2 4 1988 1990 1992 1994 1996 1998 2000 Sotrce: Arrex VII. dwelers. Even with LPG growth projected at 10% pe9 9 ear from 1988 to 2000, it wi meet less than 15% of projected urban household combustible fuel needs (in LKE terms) by the year 2000. A number of policies and programs that could serve to stimulate domestic consumption of LPG, displacing kerosene, and thereby contributing to the government's objective of diversifying domestic energy use, are presented in the next chapter along with fuel use projections under each scenario. 3.38 Figure 3.9 displays base case projections of urban residential electricity demand by major end use along with official PIN projections based on past sales and future electrification plans. Sensitivity analysis conducted on the urban residential electricity demand model yielded results pertinent to electrification policy. Changing LAe per capita income growth rate assumption from 2% to 3% per year resulted in an increase in urban residential demand in the year 2000 to 4.7% over the base case projection (and 4% over PLNs projection). Another input parameter which was varied was the growth rate of PLN direct connections. If the official connection rate is doubled from 4.3% per year (as in the base case) to 8.6% per year, the model projects urban residential consumption in the year 2000 to be only 5.7% over the base case projection (5% over PLN's projection). This suggests that future growth in - ban residential demand for electricity from PLN is more a function of income growth than of PiN's urban electrification efforts. Details of the electricity projection model, derived from econometric analysis of UHESS data, are presented in Annex VII. -42 - 3.39 In addition to this insight, Figure 3.9 shows that lighting will remain the most important end-use of electricity in the urban residential sector, with refrigeration and television following. Because of the large share of residential demand used for lighting. coincidence with the system peak load, and large efficiency differences between incandescent and fluorescent lights, the study team has identified a signfict potential for electicity savings and reduced peak load (demand management) at low per unit costs by stimulating the switch from incandescent to fluorescent lighting. A program including such incentives is briefly summarized in the following chapter which sets out components of the recommended urban household energy strategy for Indonesia. * 43 - IV. URBAN HOUSEHOLD ENERGY STRATEGY 4.1 Given the energy policy context in Indonesia, characteristics of fuel supply systems for urban residential consumers, recent changes in patterns brought on by kerosene price changes and electrification, and projections of demand based on household characteristics, there appears to be significant scope for diversification and conservation of fuel use in the urban residential sector. Namely: i) stimulating LPG consumption and continuing urban electrification efforts will serve to displace kerosene use for cooking and lighting, respectively; ii) low cost technical modifications to kerosene stoves, coupled with a cooking practices sensitization campaign could result in significant kerosene savings, and; iii) a phased program designed to raise the average efficiencies of key electric appliances on the market and encourage the adoption of fluorescent lighting presents a low cost means to reduce peak demand requirements and lead to residential electricity conservation. These components of the proposed urban household energy strategy are designed to reduce government expenditures (principally through displacing and conserving subsidized kerosene) and improve the balance of payments while increasing effective energy services obtained by urban households. 4.2 Though the fuel use projections upon which these strategy components are evaluated are for urban households only, program implementation could have a significant effect on energy use in rural households as well. However, without detailed analysis of energy use patterns in rural households, neither quantification of these effects nor design of complementary programs for rural areas is possible. If the analysis based on patterns of energy use in urban households is exterded to rural households, two hypotheses can be formed: i) wood and other biofuels probably meet a dominant share of rural household cooking needs, and; ii) rural electrification could displace a significant amount of kerosene for lighting while raising lighting levels by as much as ten times. For these reasons, the effects of policies and programs on rural households will necessarily be different from effects on urban households. There appears to be significant scope for raising the living standards of the rural poor through well-targeted programs and policies designed to encourage fuel conservation and substitution. It is therefore recommended that a rural household energy strategy study be undertaken for Indonesia to complement the urban household energy strategy outlined below. A preliminary work program and budgc for the proposed $350,300 activity is outlined in Annex X. Substituting LPG For Kerosene 4.3 Toward diversification of energy use in the residential sector and reduction of government expenditures on domestic oil product subsidies, the study team investigated the economic effects of encouraging LPG use to reduce the rate of growth of kerosene use in urban households. Analysis of data from the UHESS survey results in the first reliable estimate of how much LPG will substitute for a liter of kerosene for cooking based on the actual behavior of urban households and a firm understanding of both the barriers to and preferences for the use of LPG at the household level. .44 - Table 4.1: Economic Effect of Substituting LPG for One Liter of Kerosene Kerosene LPG Net Rp/Ilt Rp/0.59 kg Rp Change in: Economic Benefit A/ -203 348 145 Economic Cost -290 253 -37 Net Economic Benefits 87 95 182 6ovennment Revenue 92 73 165 Exports 233 -97 136 9/ Economfc Benefit = 1988 retail price. Source: Tables 2.1 and 2.4, and substitution ratio of 0.59 kg LPG per Liter of kerosene displaced. 4.4 Using the LKE measure for LPG (0.59 kg LPG substitute for 1 lite- kerosene for cooking), the effects of substituting LPG for kerosene are summarized in Table 4.1. Though there are significant uncertainties in cost estimates for each fuel, L/ the findings that there are positive net effects on economic benefits, the government budget, and balance of payments are robust. The significant improvements in the government budget and balance of payments from each liter of kerosene displaced by LPG provide a strong rationale for programs and policy reform designed to stimulate LPG consumption in urban households. However, the estimates of resource costs and economic benefits of LPG substitution deserve a closer look. 4.5 The standard economic estimate of the (marginal) benefit of a simple commodity is the price of that commodity on the market. The 'average" urban consumer can buy LPG at slightly above 590 Rupiah per kilogram, kerosene at about 203 Rupiah per liter, and faces few supply constraints. To an approximation, consumers currently indifferent to usirg LPG rather than kerosene for cooking, at existing prices, are implicitly equating the value I liter of kerosene with the value of .34 kilograms of LPG. Prevailing prices are the most appropriate measure of the benefits of providing LPG to substitute for the kerosene used by these (marginal) consumers. As additioinal consumers switch to LPG, as the result of price changes, the appropriate measure becomes the prices at which the new LPG consumers become indifferent. Regional price variation and cooking equipment costs complicate precise measurement, but fuel prices still provide the best basis for assessing the (marginal) benefits of commercial cooking fuels. 4.6 If kerosene and LPG were perfect substitutes, their relative value to the marginal consumer (ie. their relative prices) would equal the inverse of their substitution ratio. Using the 'Basic Energy Demand' method, it would also be expected that their substitution ratio, in energy units, would be to equal the inverse of their average end-use efficiencies. None of these ratios are equal because kerosene and LPG are imperfect substitutes. They do not provide identical services to the user households, and LPG is clearly the preferred fueL Using the LPG substitution ratio J/ Unceraknis at dwe to vawason in intonaolprodupices as well as the fact hat distibution costs are necessaly based on rugh estimates not accouing for rgional vadaio -45- Table 42: Daily Cooking Cost Comparison of Kerosene and LPG Economic financial Annual Discoant/interest Rate 10X 10 20X Daily Cost to Average Hosehold (Rupiah) 2 Stove Kerasens System FueL 342 240 240 Stoves 21 21 25 Total 363 261 265 1 Burner LM System wth I s ckap Kerosene Stove Fuel 298 411 411 Equipment 38 38 46 11 kg LPG Bottle 20 20 37 Total 356 469 494 2 Bumer LP6 System Fuel 298 411 411 Equipment 48 48 58 11 kg LPG gottles 39 39 75 Total 385 498 544 Source: See Annex V. Average daily household cooking fuel use a 1.2 LKE. Economic and financial costs of LPG bottles reflect the daily interest foregone by the average household due to investment in LPG bottles. derived in Chapter m, Table 4.1 shows that the economic cost of LPG is about 37 Rupiah less per liter of kerosene replaced. M/ In financial terms, LPG users who would otherwise use one liter of kerosene a day are paying almost 150 Rupiah more per day on fueL as well as covering higher equipment costs, to use .6 kilograms of LPG. They apparent'ty value the convenience, cleanliness and flexibility of LPG at least this high. Though the posi,ve net benefits in Table 4.1 are robust with regard to uncertainties in the cost estimates, they are sensitive to the assumption that the switch occurs for "marginal" consumers, indifferent to LPG and kerosene at current prices. For consumers who would be indifferent to LPG and kerosene even if daily fuel costs were equaL the net economic effect would be slightly positive. 4.7 On the basis of fuel costs alone, there appears to be a small economic cost advantage to stimulating LPG as a substitute for kerosene for cooking in urban households. When cooking equipment costs are included in the analysis, the economic costs of cooking with LPG are roughly equal to or slightly more than the costs of using kerosene. Table 42 shows that were the daily cooking fuel needs of an average household from the UHESS survey to be met by LPG, this could be roughly 6% more costly, in economic terms, than using kerosene (almost entirely due to more costly LPG equipment) and it would be roughly twice as expensive for the household, in financial terms (due mosdy to the difference in 1988 retail prices of kerosene and LPG). ,V Givn the kvdl of wacmah, dis 13% ot ,duedom i only mwgin* sipifiwut -46 - 4.8 Financial cooking cost comparisons at high interest rates, simulating the severely constrained cash flow situation of poor households, indicate that because of the high investment cost for LPG equipment, costs of cooking with LPG could appear to a poor to middle-income household to be three times as high as the costs of cooking with kerosene (see Annex V). Even at a 20% annual interest rate, the daily cost of cooking with LPG would require over 15% of the budget of a household with monthly expenditures of 100,000 rupiah. This explains why the UHESS survey found virtually no LPG use in households with monthly expenditures below 100,000 rupiah. LPG Substitution Scenarios 4.9 A Base Case projection of urban-household fuel use for cooking and water heating was presented in Chapter III. In this Section, alternative scenarios modelling the effects of cost based pricing and alternative measures designed to lower household investment costs for LPG equipment are developed and the economic implications are assessed. Given the uncertainty of future oil prices, it would be overambitious to predict the future opportunity costs of domestic kerosene and LPG use. In addition, LPG costs are estimated for urban consumers nationwide and do not include regional variation of LPG distribution costs which must be incorporated in appropriate policy measures for effective LPG promotion. Therefore, the scenarios are intended to be indicative, only. The scenarios do indicate, however, some of the benefits and costs of price reform as well as of alternative measures to promote LPG. 4.10 While the base case scenario assumes that the existing kerosene subsidy and LPG tax are maintained in real terms, Scenarios II to IV illustrate the effects on the urban household sector of changing the price(s) of kerosene and/or LPG to their economic cost(s). L9/ Scenario V provides the basis for a brief analysis of alternative measures of achieving levels of LPG use similar to that resulting from cost-based pricing. More specifically: Base Case: Constant real kerosene price (203 Rp/liter) and LPG Price (590 Rp/kg). 4.1% per year urban population growth and 2% per year income growth per capita from 1988 to 2000. Scenario II: Fuel price set to equal economic cost for kerosene (290 Rp/liter) and LPG (428 Rp/kg). Other features as in Base Case. Scenario m: LPG price set to equal economic cost (428 Rp/kg). Other features as in Base Case. Scenario IV: Kerosene price set equal to economic cost (290 Rp/liter). Other features as in Base Case. V/ As desanbed in Annev Vl, the pAce elastcies implict in the projection model ar lawly conitent with both the histonc develmt of w*an household fuel demand between 1981 and 1987 (a penod of huge kerosene pice changes), and the cross sectoa vaniation in klrosene demand obsened among the UHESS survey households fang diffemntkerosenepnces. Howrev, datad cand the compleityoffuelswitchingbehaviorpievented dewvaion of the appropnate model parameten dkv* ftm either data seL * 47- Table 4.3: LPG Substitution Scenarios Annual Growth 1990-2000 Pemand 2000 ('000 kl.1000 TS Kerosene LPG Wood Kerosene LPG Wood Base Case 3.43X 10.38X 1.47X 4,964 559 3,729 Scenario It 2.30X 13.06X 2.40K 4,349 745 4,159 Scenario III 3.182 11.89M 1.482 4,820 658 3,734 Scenario IV 2.61X 11.622 2.39% 4,507 639 4,155 Scenario V 2.99M 12.552 1.47% 4,715 706 3,729 Source: See Annex VIII. Scenario V: Income threshold below which no households willing to invest in LPG equipment falls by 15% (ie. shifts from 100,000 Rupiah per month to 85,000 Rupiah per month). Other features as in Base Case. 4.11 Even without any policy changes, LPG can be expected to continue to substitute for kerosene over the coming years, and to capture a growing share of the urban cooking-fuel market. LPG is still a new fuel whose use is expanding as people become more familiar with its advantages. As hired cooks become less common, the convenience of LPG wil make it even more attractive. Furthermore, as income grows a larger proportion of households will be willing to pay a premium to use LPG. The Base Case projection for kerosene and LPG, summarized in Table 4.3, shows LPG use growing between 1990 and 2000 at an average annual rate of 10.4% as compared to 3.4% for kerosene. Projected LPG demand growth is even stronger in Scenarios II to V. With full cost- based pricing (Scenario II), LPG use by the end of the century is more than 30% higher than the Base Case, while kerosene use is approximately 10% less than the base case projection. As expected, cost based pricing for LPG alone (Scenario HI) and kerosene alone (Scenario IV) are less influential with kerosene price rises being relatively more effective in lowering kerosene demand. Interestingly, the final scenario (V), which models a 15% decline in the income threshold below which households do not use LPG thereby simulating a lowering of investment costs for LPG equipment, achieves one of the highest levels of LPG consumption, low levels of kerosene consumption, and no increase in wood consumption. There are several reasons for this apparent success. First, LPG is currently popular only at the upper income end of the income distribution. By shifting the income threshold downward, the large cooking fuel market of middle income households can begin to be tapped. Second, all of the increase of LPG use is substituting for kerosene - there is no increase in fuel use among households already using LPG. 4.12 Though the level of LPG consumption in the year 2000 would be only a small fraction of Indonesia's expected LPG production, the LPG distribution infrastructure would stil need to be expanded rapidly to avoid supply disruptions. None of these LPG promotion scenarios result in growth rates of LPG consumption similar to those experienced in the 1980s. Hence, LPG cannot be expected to displace kerosene as the major cooking fuel in urban households by the year 2000. At most, LPG might meet 20% of urban residential cooking needs. In this light, further penetration of LPG should be seen as a way to reduce expected growth in kerosene use. - 48 - Table 4.4: Evaluation of LPG Promotion Scenarios Annualized Costs & Benefits 1990-1999 of Base Case and Scenarios Relative to Base Case In Constant 1988 USS (Nillions) Scenario Base II III IV V Value of Fuels to Consumers I/ 584 -30.0 8.0 -33.4 12.4 Fuel Supply Costs 758 -47.4 0.5 -47.4 -3.4 Net Eceaie bnefits -174 17.4 7.5 14.0 15.8 k/ net Govevuet Revenues -193 192.8 -17.8 219.3 14.2 hI Export Value of Urban Residential Corluption 576 -50.8 -6.2 -43.5 -11.8 l/ Change in quantity of each fuel valued at mid-point between Scenario and Base Case fuel prices. b Estimates do not include program costs. Source: See Annex VIII. Arnnalized a 10 and 1700 Rupiah/US Dollar. 4.13 The economic implications of these scenarios are summarized in Table 4.4. 4Q/ The net present value of all costs and benefits at 10% discount rate for the ten year period between 1990 and 1999 are annualized, and expressed relative to the Base Case. ed Reform 4.14 Of the scenarios, kerosene price reform appears to be the most effective approach toward reducing government expenditures and improving the balance of trade. LPG price reform entails foregone government (Pertamina) revenues and is less effective at displacing kerosene for export than all other scenarios. 4.15 Though all scenarios show positive net economic benefits, the incidence of response is very different across income categories. Scenarios with kerosene price reform are dominated by sharp reductions in kerosene use (and increases in wood use) by poor and middle income households. Conversely, the net benefits of LPG price reform are obtained mostly by upper income households increasing LPG use and those who switch to LPG due to the price change. Data from the UHESS survey displayed in Table 4.5 shows that low income households spend a far higher share of their household budget on kerosene than do high income households, while the situation is exactly reversed for LPG. While low to middle-income households consume the highest total amounts of kerosene, over half of all LPG used in the urban residential sector is consumed by households in the highest income decile. Hence, in absolute terms, the burden of a kerosene price increase would fall most heavily on middle income groups, while in relative terms it would fall most heavily on the poor. An LPG price reduction, on the other hand, would benefit primarily the rich, both absolutely and relatively. Alone, removal of the kerosene subsidy and of the LPG tax, would benefit the urban rich at the cost of the urban middle class and poor. Though the striking positive effect on net government revenues of scenarios II and IV arise chiefly from foregone benefits to low and middle income households, the overall distributional consequences depend on how this S/ etaikd cash flow anaysis ispmsentd in Anw VIJ. - 49 - Table 4.5: Relative and Absolute Value of Kerosene and WPG Use by Income Aniusl Fuel Use in Urban Households an Java Household Expenditures X of HH Exmend for: Kerosene LPG (Rupfah/HH/Ionth) X HH Kerosene LPG (M tt) (Rp Miltion)#/ (8000 Tons) (Rp fMIllIon)pj < 75,000 26.8% 9.0% 0.1% S15 105 1.0 0.65 75,000 - 120,000 23.8% 6.3% 0.2% 570 115 5.5 3.25 120,000 - 185,000 23.6% 4.8% 0.3% 680 140 14.5 8.60 185,000 - 295,000 15.7% 3.5% 0.6% 490 100 24.5 14.50 > 295,000 10.2% 1.8% 0.9% 295 60 58.5 34.50 Att 100.0X 5A8 0.3% 2,550 520 106.0 61.50 g/ Valued at 1988 retail prices. Source: UHESS 1988. Number of urban households on Java in 1988 estimated at 6.58 million. (marginal) revenue is used. For these reasons, this brief analysis of price effects and price reform scenarios are intended to inform a more broad based energy pricing strategy, such as that now being developed through the Energy Pricing Policy Study. 4.1 4.16 Before moving on to an evaluation of interventions not based on fuel price reform, a compelling rationale for allowing regional price variation for LPG is briefly summarized. LPG pricing and distribution is the responsibility of energy sector institutions. The price of kerosene, on the other hand, is set by presidential decree, and Pertamina is directed to ensure that kerosene is available throughout Indonesia. This distinction in administrative approach reflects the perception that LPG is a relatively peripheral fuel used predominantly by upper income households. and whose price and availability need not be based on the broader goals of national integration and interregional equity. Despite the expanding LPG market, there are good economic grounds for extending this distinction to allow regional price variation for LPG. 4.17 Much of the preceding discussion has ignored regional variation in distribution cost and availability. However, since distribution costs are far higher for LPG than for kerosene, the economics of substitution are more favorable in areas where distribution costs are low. Unless LPG prices reflect differences in distribution costs, a policy which provides the appropriate incentive for switching to LPG in low cost areas will be inappropriate to high cost areas, and vice versa. 4.18 Currently, the price of LPG at a Pertamina depot or bottling station is fLxed at the same level in all parts of the country. This makes it difficult to implement an LPG promotion program efficiently. Furthermore, since Pertamina is not obliged to make LPG available throughout Indonesia, this pricing procedure creates a disincentive for Pertamina to supply LPG to the high distribution cost areas, even where there would be a market for very high priced LPG. Ultimately, neither economic efficiency nor equality is served. Given that LPG will remain a luxury cooking fuel for the foreseeable future, it would be appropriate to emphasize economic efficiency and include region-specific distnbution costs in the depot price of LPG. As outlined in a previous I/ Cwwied ow by GOI uwh financing wuler Bak Lon 2690IND. - 50- report on LPG in Indonesia, 4/ it would also be economically advantageous to set different fuel prices for different quantity purchases. 4.19 Even if distribution costs are not to be reflected in the regional price of kerosene, however, it would be very useful for the development of policies and programs to add estimates of the distribution costs for each region, along with a breakdown of these costs by sources (e.g. refinery, transportation, administration, etc.) to the regularly published statistics on petroleum production and distribution. Other Measures to Stimulate Substitution of LPG for Kerosene 4.20 Scenario V can be used to gauge how costly a promotion program could be, and still be ecmnomically desirable, if existing taxes and subsidies are to be maintained. The results must be interpreted with care: no program costs have been included, while current prices have been used to evaluate the fuel switching. Under such assumptions, full supply costs decrease marginally while the economic value of the fuel provided increases substantially. The net economic benefits are as high as the kerosene price change scenarios, and the budgetary and balance of trade affects are both positive. A sizeable annual program cost could be born in order to shift the LPG income threshold down from 100,000 Rupiah to 85,000 Rupiah per household per month. 4.21 Because the high cost of LPG equipment and concerns over the safety of use are major barriers to LPG choice in urban Indonesia, an effective LPG promotion program should contain components designed to lower both barriers. Equipment Cost 4.22 The most common reason households in Java give for not using LPG is the high cost of LPG equipment. At a minimum, a household must purchase the LPG bottle, regulator and stove. The survey indicated that most 11 kilogram LPG bottles in Java were sold for about 75,000 Rupiah (despite an official price at the time of 55,000). Even the simplest of LPG burners costs upward of 30,000 Rupiah, and a regulator at least 10,000 Rupiah. Thus to obtain just one working burner, a household must spend at least 115,000 on equipment plus 6,500 on LPG, and still retain a kerosene stove for back-up. For a complete LPG system, including two full bottles and two burners, a household must invest about 230,000 Rupiah in equipment and 13,000 Rupiah in 22 kilograms of LPG. By way of contrast, two of the more expensive factory made kerosene stoves cost about 30,000 Rp and a liter of kerosene another 203 Rp. For many households, the difference in daily fuel costs, which households do not know with certainty, is insignificant compared to this clear difference in investment costs. 4.23 As the household discount rate is considerably higher than the government discount rate, lowering the price of LPG equipment sold by the government (currently the bottles) can be a more attractive means of lowering overall LPG system costs than lowering the fuel price. Simply lowering the bottle price will induce some LPG users who would otherwise use only one bottle to buy a back-up bottle, as well as inducing kerosene users to switch. As described below, however, W/ 7idonesia LPG Feasibility Study, AD. Uktt and PPIR, 1986. - 51 - there are a variety of possibilities for lowering equipment cost, at least one of which would enable the program to focus on current kerosene users. Analysis for the following discussion is presented in Annex V. 4.24 Ensuring that the bottles are sold at their official price (or even with a competitive mark-up), would lower LPG system costs considerably. The initial investment required to start using the full LPG system would fall by 40,000 Rupiah. The savings would amount to 2%-10% of the daily system costs, depending upon the household discount rate (see Annex V). The overpricing of LPG bottles may already be on the decline: there is reportedly an excess supply of 11 kilogram bottles at present, bottle production is high, and Pertamina is currently attempting to build up its bottle filling capacity, thereby enabling them to release more bottles on the market. Even under such conditions, however, a lack of dealer competition could sustain the overpricing. 4.25 Another means of lowering equipment costs is to introduce smaller LPG bottles. If six kilogram bottles were being sold at a price of 30,000 Rupiah, the initial outlay required for a full LPG system would be about 136,000 Rupiah: 90,000 Rupiah less than the current outlay needed. The daily cost of a full LPG system would be less than current daily costs by between 5% and 23%, depending on the discount rate. Furthermore, local retail trade on a cash-and-carry basis would circumvent the suspicion with which some households view truck delivery (fueled by reports of underweight bottles and of bottle theft by false delivery men). 4/ 4.26 A final option was also evaluated: targeting new users by offering 2 six kilogram bottles with a two burner stove and regulator for 80,000 Rp (the cost of the bottles plus 20,000 Rupiah). This subsidized package (roughly 50,000 Rp subsidy) would lower the investment cost to about 87,000 Rupiah. Overall daily costs would be 12-37% less than currently, again depending on the discount rate. 4.27 There are two advantages to selling the stove-bottle combination at a low price rather than subsidizing the bottles directly. First, virtually all households wanting both a new stove and a bottle will be new users, particularly if the stoves are of the simple variety. Second, by ensuring that the stove(s) selected have a good turn-down ratio, a more effective use of LPG can be promoted. 4.28 At the current LPG price and effective tax (Table 2.4), the government (through Pertamina) could recoup this fifty thousand Rupiah subsidy, given to a household for switching from kerosene to LPG, in less than two years in LPG taxes alone. The government would also benefit from the kerosene subsidy not paid out. In sum, this option, as well as ensuring that adequate bottles are supplied at official prices, and offering smaller botdes, appear to be cost effective options for stimulating LPG demand. !W In he ndoesia LPG FeasibiV Sudy-, A.D. Likle and PP)R 1986, it was sugested that small bottles woidd also extend the availability of LPG to 7campuwgw (ve) Jwung whose pathways wmm deemed too narrw for LPG delivey. The UHESS swvey indicates, hower, that loca access is not a sigmficant pwbkm (see Table 3.7). - 52 - 4.29 Even if fuel prices more closely reflected costs, there would still be an economic rational for cross-subsidizing LPG equipment with an LPG fuel tax, as long as the household discount rate is well above the government discount rate. Education-Promotion 4.30 In more than a third of the households not using LPG, respondents were concerned that LPG was unsafe. More than 10% gave this as their main reason for not using LPG. Special devices are sold, which purportedly prevent the stove flame from igniting the LPG in the bottle via the gas outlet (a virtually non-existent danger). Safety outweighed other quality related reasons for not using LPG, despite a general tendency for people to prefer the fuel they are used to cooking with. 44/ Employed properly, however, the typical LPG system in Indonesia is safer than the typical kerosene system. As described in the Stove Report (Appendix H), many popular kerosene stoves allow the kerosene in the storage tank to reach unacceptably high temperatures after a relatively short period of use. LPG stoves, on the other hand, are perfectly safe to use indefinitely, so long as the bottle is kept sufficiently distant from the heat source. 4.31 The somewhat exaggerated fear of LPG is probably the effect of its relative novelty as a cooking fuel in Indonesia, combined with the accurate judgement that an exploding LPG bottle is very dangerous. As LPG becomes more popular, this concem is likely to lessen, and does not constitute a serious impediment to LPG dissemination. It is important, however, to ensure that LPG users are aware of how to minimize the danger of accidents, and that LPG does not justify its unsafe reputation. Currently, safety booklets are often not provided to new consumers, and some dealers will not provide safety instructions to users, even on request. While Pertamina advises verbally that such information should be given to all new users, providing safety information for consumers is not mentioned in Pertamina's list of rules for LPG dealers (Syarat-Syarat Dealer Elpiji, Pertamina, 1987). 4.32 If the price differential between kerosene and LPG is to be maintained, it would also be worthwhile for Pertamina to promote LPG through advertizing targeted to the upper income consumers (e.g. in the more expensive movie theaters). The advertizing could provide some safety information, and emphasize how LPG can be used efficiently. LPG Promotion Program Costs and Institutional Responsibility 4.33 The extent to which an LPG promotion program with some of the elements above would result in an effective lowering of the LPG use income threshold to 85,000 Rupiah per month (Scenario V) is uncertain. Measures such as introducing 6 kg bottles and making safety information freely available to all consumers are warranted even if fuels were to be priced closer to economic cost. An ambitious program could incur annualized costs as high as US$ 8.6 million, remain budget neutral, and still have annualized net benefits of US$ 1.6 million. By way of comparison, if evely new LPG user between 1990 and 1999 purchases a stove-bottle system with a 50,000 Rupiah (US$ 30) subsidy, the annualized cost would amount to less than US$ 5 million. O/ Mun asked to name theirprefeed cooldngfulw (disagding cost and avaWlabili4y), most respondents to the UHESS suey smpy gave the type of fuel they wa cumntly uin& vAether it was wood, kerosene orgas. -53 - 4.34 Installed LPG production capacity in Indonesia should reach 3 MTPA in the early 1990s, well above projected urban demand in the year 2000 in the most effective LPG promotion scenarios. Though much of Indonesia's LPG production is committed for export, significant investments in LPG bottling capacity and distribution infrastructure as well as measures to reduce initial costs to households for LPG equipment appear to be warranted on the basis of improvements in the government budget, balance of payments, and increased economic benefits to urban households. Nonetheless, in evaluating price reform or other measures to promote LPG, it should be noted that even with ambitious programs, LPG is likely to remain an upper-income urban fuel in the medium term, while kerosene will continue to be used at all income levels. Referring to Figure 3.4, it can be seen that even if average urban household income rises well beyond 200,000 Rp/mo., 4_/ LPG will remain beyond the reach of most households. While moderate to upper income households will obtain most of the direct benefits of LPG promotion, to the extent these households switch from kerosene will serve to target government expenditures on the kerosene subsidy to low and middle income households. 4.35 Price reform has the considerable advantage of not relying on a knowledge of price elasticities to be effective. For reasons outlined above, there is some economic justification for a limited cross subsidy on LPG equipment with revenue from an LPG tax There are also some opportunities for promoting LPG which are warranted, such as introducing small bottles. Beyond these measures, LPG promotion to compensate for LPG taxes and/or kerosene subsidies is an economic palliative, but not economic justification for such taxes and subsidies. 4.36 Pertamina and MIGAS would be the institutions responsible for implementing any program of LPG promotion. As a significant potential has been demonstrated for LPG to reduce growth in kerosene consumption in the urban residential sector and, thereby, contribute to government objectives of fuel diversification, export enhancement, and expenditures reduction, the Ministry of Mines and Energy should be involved in an integral fashion in the design and supervision of policies and programs aimed at displacing 'cerosene by LPG. Kerosene Stove Improvement Pilot Program L6/ 4.37 As stated in Chapter IL, kerosene sales of 6.9 billion liters nationwide in 1987 accounted for over one quarter of domestic petroleum product sales. As much as 45% of this (3.1 billion liters worth roughly US$ 450 million @ US$ 23/barrel) is consumed in kerosene stoves. 42/ Few other energy conversion technologies, if any, convert such large quantities of fuel. It is evident from Figure 33 and Table 4.5 that with regard to percent of household expenditures on kerosene, 4£/ t 2% projected growh in per capita income, average household income is projected to grow from 156,6O0 Rp/mo in 1988 to neady 200,000 Rp/mo (in 1988 nupiah) by the year 200t. f/ A detailed prognm proposal is pnsented in AppendixHn Wemsene Stove Inprovement Pilot Progm' 17/ According to SUSENAS 1987, 40% of total kerosene demand was consumed in wban households. Using the VHESS esdimate that 75 of kerosene used by wlan dwelrs is for cooking and an asswnption that a mawtum of 25% of the mmainng natonal kerosene sales is used for cookin&, resutts in an overaml estimate of 45% of national krsene sales being used in kerosene stoves. -54 - the poor would benefit most from kerosene stove improvements that raise the efficiency of use without raising stove prices significantly. Modifications of existing kerosene stoves and cooking tests undertaken by the project team at LEMIGAS with a visiting stove expert have uncovered important technical possibilities for stove improvement and resulted in a kerosene stove improvement pilot program summarized below. The full program proposal, presented in Appendix II, shows a significant potential for improving the balance of payments by reducing demand for kerosene at a very low cost. There are numerous non-technical obstacles to improving kerosene stoves, however. The challenge facing any kerosene stove program will be in translating the large technological and economic opportunities into actual improvements in the efficiency of cooking with kerosene. 4.38 Even with recent kerosene price increases, there re four reasons why the fuel use of kerosene stoves can still be expected to be uneconomically high. First, consumers purchasing a stove cannot ascertain its fuel economy, either through physical inspection or reputation. Second, stove producers are also ignorant of their stoves' fuel economy, and are generally too small scale to engage in research and development to improve fuel economy. Third, the habits of both con- sumers and producers have developed under low price conditions, and may not have fully adjusted to existing prices. Fourth, kerosene remains subsidized. 4.39 Test results indicate that the efficiency and quality of the kerosene stoves used in Indonesia vary considerably, although the majority are wick stoves of similar design. Many stoves are of low efficiency, and there is little relation between stove price or producer type and efficiency. The technical changes required to save kerosene and/or change the power of most existing kerosene stoves are simple, and need not be expensive. 4.40 Prior to this study, research on kerosene stoves already indicated the diverse performance of stoves on the Indonesian market. With this study, it has been possible to better identify some of the key determinants of stove performance. Furthermore, a better understanding of household needs and desires has been obtained. Rather than attempting to identify and promote the "best" stove, the objective --if the proposed program is to raise the quality of existing stoves. This strategy was selected because: 1) there are numerous obstacles to marketing a new stove; 2) tastes in stoves vary so that no single stove can satisfy the diverse requirements; and 3) simple modifications should be able to raise the efficiency of most existing stoves. To provide an overview of the context in which the proposed program would operate, some aspects of kerosene stove technology, determinants of stove choice, and production are briefly summarized below. The Context of Kerosene Stove Improvements in Indonesia Technical Modifications 4.41 The nature of kerosene stoves is such that both power and efficiency can vary significantly among stoves. LPG stoves, on the other hand, can be expected to vary considerably in power, but relatively litde in efficiency. For this reason, and given the dominance of kerosene as the major urban cooking fuel, the bulk of the work was undertaken on kerosene stoves. 4.42 Past work on testing kerosene stoves in Indonesia already supplied information about the characteristics of the types available in the market. Kerosene cookers all have a comparable basic design. Differences are not essential and relate to cooking speed and type of material used. 55- But these apparently minor differences cause performance to vary widely between stove models. Fires due to exploding stoves are common and can be prevented by better designs. Past tests also reveal the existence of a wide range in power and efficiency. The wide variation in their fuel use implied by these results suggests a large scope for improvements. 4A3 Prior to the Urban Household Energy Strategy Study (UHESS), these test results were never thoroughly analyzed with respect to how performance is influenced by the stove design. In a recent World Bank report 4/ a first attempt was made to link design features to performance for 20 stoves from all over the world. For this study, the stove work was aimed at extending these insights into the determinants of the performance of Indonesian stoves. 4.44 Lemigas conducted extensive testing on the performance of 18 kerosene and 2 LPG stoves, the results of which are fully documented in Appendix II. An analysis of the stove test results indicates that, controlling for pan size, less compact burners are less efficient. Also, pan supports with rings touching the pan appear to result in lower efficiency. Conversely, no strong relationship appears to exist between: i) efficiency and power; ii) price and efficiency; iii) producer type and efficiency, or; iv) age and efficiency. 4.45 In collaboration with a visiting stove expert, several attempts were made to improve the kerosene stoves tested. Two modifications were found to improve the efficiency of a number of stoves. Fist, restricting the amount of air entering through the inner flame-holder was found to increase the efficiency of those stoves which did not already have a built-in flow restricter. Second, inverting the pan support was found to increase the efficiency of stoves whose pan supports had rings touching the pans. The average percentage change in efficiency resulting from these modifications were 8% and 13% respectively. 42/ The results are robust. Flow restricters improved the efficiency of all of the four different stoves to which they were added. Inverting the pan support was found to improve efficiency in all four stoves for which the pan support previously touched the pan, and to decrease the efficiency of two stoves for which the pan support did not touch the pan except after being inverted. 4.46 These two simple changes would have a signfificant effect on kerosene consumption if adopted. Furthermore, as they were uncovered in a relatively short time, it likely that similar savings could be obtained from further design changes. A program to improve kerosene stoves can be expected to uncover a variety of kerosene saving possibilities during the course of its work, especially if sufficient attention is given to applied research (as opposed to simply enforcing standards, for example). 4.47 The LPG stove analysis was limited to two stoves. One was deemed typical of current LPG stoves used, while the other was chosen because of its low cost and expected low power. A wide power variation was indeed found, with the high power LPG stove being more twice as powerful, at maximum power, than the average for kerosene stoves. The low power gas stove, / est Results on Keme and OterStowsforDeloping Cawuies, E*D m PaperNo. 27. Washington: Wodd Banl4 1985. M2/ AM changes In eBcin i dis Ae am mlve changes. (e.g. a 10% effcinc cinve means chmgingfim 40% to 44%). - 56 - on the other hand, had a power comparable to most common kerosene stoves. Given the relation between power and fuel use described below, this has important implications regarding the amount of kerosene displaced as the use of LPG for cooking increases. 4.48 Laboratory boiling water tests provide technical parameters such as the efficiency and power of the stove. These results do not show how much fuel is actually used for a certain cooking task. To obtain an insight in how efficiency and power are related to fuel use, controlled cooking tests were conducted during which the same standard meal was prepared on different stoves. 4.49 The cooking tests demonstrate that using a high power stove can lead to considerably higher levels of fuel use, even for stoves with the same efficiency. They also demonstrate the potential fuel savings which could be obtained from turning the stove power down during selected cooking operations, and from minimizing the water used for rice steaming. 4.50 A model of household cooking and water heating was developed, showing how different factors affect daily kerosene consumption for a typical household starting with a typical stove. Looking to stove quality alone, roughly 1/3 of kerosene consumption depends on efficiency and roughly 1/3 on power (the rest being unexplained). This implies, for example, that if efficiency increases by 30%, kerosene consumption will decrease by 10%, and cooking time will decrease somewhat. However, if this efficiency increase is accompanied by a power decline of 30% (leaving the heat delivered to the pan roughly constant), kerosene consumption will decrease by 20%, and cooking times will not have altered. 4.51 The two more user-dependent factors, stove power adjustment while cooking and the quantity of water employed for steaming rice, were also modeled. If power is decreased by two thirds during both water simmering and rice steaming (after the water has already been brought to boil), the households daily kerosene use would fall by about 10%. Alternatively, if the amount of water used for steaming rice is decreased from 1.3 to .3 liters, 4% of the daily kerosene use could be saved. 4.52 A savings of roughly 25% in kerosene consumption can be achieved by combining efficiency improvement and maximum power reduction with a better turn-down ratio and the use of less water for rice steaming. Specifically, this would involve a decline in maximum power from 2.0 to 1 5 KW, and increase of efficiency from 41% to 45%, turning the stove to half power during simmering and steaming, and decreasing the amount of water used for steaming rice to .3 liters. Sociological Dimensions of Household Stove Choice 4.53 Stove efficiency is not perceived as an important criterion in purchasing a stove. Only 10% of the households surveyed in urban Java selected fuel economy as an important factor hifluencing their purchase of a new stove. In-depth interviews revealed that stove features associated with higter efficiency are often desired, but not because they save fuel. For example, blue fimes are generaily desired because they do not dirty the pans: the fact that they are more tfficiert is not geinerally understood. -57 - 4.54 Stove power is viewed ambiguously. Some women prefer low power stoves, because this allows then to work on other tasks while cooking. Others prefer high-powered stoves, because they allow faster cooking. High-powered stoves are recognized as fuel-guzzlers, and for at least some households the choice of stove is guided by a tradeoff between the faster cooking of a high- powered stove and the lower kerosene use of a low-powered stove. 4.55 Superficially, purchasers' neglect of efficiency (and more generally, fuel economy) appears economically irrational. Even at a discount rate of 100%, the present value of kerosene savings of 5% for an average urban household cooking with kerosene is over 4000 Rupiah. This is roughly one half the cost of a moderately priced stove. Given the wide variation in efficiency of stoves on the market, this would suggest that efficiency should be a very important parameter in choosing a stove. Even under pessimistic assumptions, a 15% change in stove efficiency should lead to a more than 5% change in kerosene consumption, and such differences in efficiency are commonplace. 4.56 Rather than irrationality, however, the most obvious explanation is a lack of information. It is not possible to ascertain a stove's efficiency simply by examining its features. LQ/ Once in use, the large variation in the amount of cooking done, easily hides a 5% change in average kerosene consumption. Even when the Project provided measuring devices to monitor households' kerosene consumption, differences among stoves were swamped by other factors, along with remaining measurement problems. Briefly, under existing circumstances it would not be rational for a purchaser to use efficiency as a criterion for selecting stoves since she cannot know the efficiency. 4.57 Four important policy-relevant observations can be made concerning household stove choice. First, attempts to promote more efficient stoves will '.ot be supported by consumers unless something is done to help consumers ascertain which stoves are efficient. Second, consumer preferences vary considerably, and no single stove design can satisfy all users. By implication, a program to improve existing stoves is likely to be more successful than one aimed at propagating particular models. Third, some consumers are likely to be attracted to efficiency improvements in part because they alic v faster cooking, although faster cooking dissipates some of the potential fuel economy of a more efficient stove. Ihis has important implications for the promotion of efficient stoves. Finally, improvements in the flame adjustment mechanism should make stoves both more attractive and save kerosene. Kerosene Stove Production and Marketing 4.58 Both factory and artisanal stove production are amenable to the modifications uncovered during the course of this study. Factories, being both larger scale and having more ready access to capital, are a more obvious target. Even the smallest artisanal producer should, however, be able to add a flow-restricter and produce stoves with efficient pan supports. Furthermore, artisanal producers are often clustered in one area, sometimes all making the same model of stove. V/ AU otherthingsbeingequal ab:ueflame ismore eficient than ayellowflame, and blueflamesaieprefemd. Flame color is not a good indicator of stove eJfidency, however. - 58 - 4.59 Materials are the major cost for both factories and artisans. For artisans, reliability of material supply is also cruciaL This could provide an opportunity for government assistance, linked to a stove improvement program. (Factory production, being more standardized, could be affected through official stove standards.) 4.60 A small survey of the Jakarta market indicates that artisanal stoves account for about 80% of sales, or, given shorter lifetimes, about 70% of the stoves used. The UHESS household survey shows that 50-60% of the stoves do not have name brands. More than half of the artisanal stoves are sold door-to-door, making Standard Product Information sheets inappropriate. The payment schemes for the door-to-door sales tend to involve very high implicit interest rates, suggesting that costly improvements are unlikely to be popular among artisanal stove buyers. Kerosene Stove Improvement Pilot Program Design 4.61 The kerosene stove improvement pilot program presented in detail in Appendix II and summarized here-in moves beyond kerosene stove analysis, towards direct involvement with kerosene stove producers and users. The results described above demonstrate the large technical potential for low cost improvements in kerosene stoves. Furthermore, they indicate a need for government to bridge the 'information gap', and enable kerosene producers and users to make informed, rational choices affecting kerosene use. An effective program must provide both technical support for improving kerosene stoves, and the information to enable users to demand better stoves. 4.62 The proposed program involves four components: 1) Factory stove improvement; 2) Artisanal stove improvement; 3) Kerosene stove research; and 4) Consumer awareness / advocacy, promoting high performance kerosene stoves and kerosene-saving cookirg techniques. No radical stove design changes or cooking changes are envisaged. Rather, the program emphasizes minor changes which would almost certainly be implemented given better information. 4.63 Work with factories would emphasize standards and endorsements, whereas -wok with artisanal producers would emphasize direct support in improving the production process. There are several reasons for distinguishing between artisanal and factory stoves. Artisans produce the majority of kerosene stoves, but are not readily amenable to standards, and are of too small scale to receive individual attention. Work with artisans would, at least initially, be limited to the large complexes of artisans, which may already be involved in government supported activities. Factories are more amenable to standards and endorsements, and produce on a larger scale. That some factories are already willing to go to some effort to obtain the little publicized "SIP" suggests that they will be very interested in endorsements promoted through a publicity campaign. 4.64 A su rgrm to i sumer and solicit consumer participation in the program, will also be necessary. Improved consumer awareness will be very important. Without pressure from consumers it will be far more difficult to convince producers to improve stove quality. Standards and endorsements must be understood by the consumers as well as producers, so that consumers can assess the costs of purchasing a sub-standard stove. Simultaneously, the knowledge and preferences of consumers (or consumer representatives) must be tapped to insure that the standard/endorsement program is appropriate to their needs. * 59 - 4.65 Appied Lesearc based on "trial and error" would be an integral part of the effort to advise on kerosene stove improvements. In addition, it would be worthwhile to support a small scale research program, complementary to the overall stove program. This research would draw on the stove test results, but its objective would be to better understand the general determinants of stove performance, rather than possible improvements in particular models. This research should also explore the possibility of improving the flame control mechanism of kerosene stoves, and the effect of different wicks on stove performance. 4.66 Guidance and monitoring for the energy aspects of the program would be provided by the Directorate General for Electricity and New Energy (DJLEB), based on its prior experience with energy conservation. The Ministry of Industry would be responsible for the technical side of the program. The regional offices of the Ministry of Industry (Kanwil) would foster contacts with producers. The Ministry's laboratories (Balai Besar Penelitian Bahan dan Barang Teknik), which have considerable experience testing kerosene stoves for safety, would provide the facilities. The Indonesian Consumer Organization (YLKI) and local women's Education for Family Welfare groups (Pendidikan Kesejahteran Keluarga) would assist in disseminating information, and provide feedback. A research group could be located at one of the technical universities. 4.67 The program design includes the formation of a series of regional stove testing- improvement units, each with sufficient manpower and equipment to test, evaluate and modify locally produced kerosene stoves. Ll/ During the first eighteen months two such units would operate in two large cities (e.g. Jakarta and Surabaya). After this first phase, the number of units in the two areas would double, and four additional units would be established in large cities outside Java. Each unit is expected to test on average 55 stoves per year: about 20 factory made stoves and 35 artisanal stoves. Initially, each unit is expected to bring on average 5 factory stoves and 7 artisanal stoves up to "improved" standard. An "improved" stove is assumed to be 15% more efficient than the original model. Average production runs are assumed to be 5,000 and 2,000 units for stoves from factories and artisanal complexes, respectively. Progam Economic Evaluation 4.68 At first, success in the program should come relatively easily, as popular, low efficiency models can be targeted. Eventually, improvements may become harder to identify, and the more reluctant producers will have to be approached. To capture these effects, it is assumed that "improved" stoves can penetrate at most 25% of the factory stove market and 15% of the artisanal market. In later years, a significint share of the work is assumed to involve repeat tests, with, in effect, lower success rates. A more optimistic assumption would be that a demonstration effect would lead initially reluctant producers to improve the quality of their stoves, and that over time a large share of the stove market would consist of improved stoves. However, given the attractive economics under conservative assumptions presented below, this more optimistic scenario was not evaluated. / lwo ewt and tm tecuiin would wwc in each nit. Eq4tn would inude a pewnal computer and electunic scale, and a suitabk wo&hop. - 60 - TIae 4.6: Summary Evaluation of Kerosene Stove Pilot Program Net Present Value G 10% Annualized NPV (USS Millions) (US$ Millions) Kerosene Saved 16.65 2.71 Program cost 2.03 0.33 Increased stove cost 0.21 0.03 Total Costs 2.25 0.37 Not Econoic Beneffits 14.41 2.34 Gowvrment Budget 2.74 0.45 Batance of Payments 12.30 2.00 1700 Rupiah a 1 US Dollar. Economic Costs of Kerosene Supply (Rupiah/liter): 290 Government Kerosene Subsidy (Rupiah/liter): 92 Export Value (Rupiah/titer): 233 Conversion factor for goverrnent expenditures: 0.8 Foreign exchange component of government expenditures: 43% Source: See Annex XI. 4.69 The bare case kerosene use projections presented in Chapter III and the conservative program targets specified above were used to evaluate the economics of the proposed pilot program. Table 4.6 summarizes the econol,iic evaluation of the program. Discounted at 10%, net returns over an extended program period of ten years would be about seven times cost. The present value of net benefits are calculated to be over fourteen million US dollars. In terms of government budget, the project is expected to show a NPV of US$ 2.7 million (assuming that kerosene remains subsidized at current levels). The balance of payments is also expected to benefit. Over the program period, the program costs are assumed to have an import content of about 40%. This cost is dwarfed by the export potential of the kerosene saved. Over the course of the program, the NPV of net exports increases by roughly twelve million US dollars. During the same period, the NPV of net savings to households (in financial terms) is estimated between five and six million US dolars (Rupiah equivalent). End-use Electricity Conservation Program L2/ 4.70 As stated in Chapter II, because the residential daily peak demand is coincident with the system peak, a program that effectively encourages the marketplace and consumers to conserve electricity and/or reduce peak demand would have the net effect of reducing and/or deiayir.g the need for scarce capital resources to be used in power generation capacity expansion. This section summarizes an End-use Electricity Conservation Program detailed in Appendix IH. Under the conservative assumptions used to evaduate the economics of the program, Indonesia could save an equivalent of 110 GWh annually over 10 years, representing savings of approximately $US 10 Q/ A ded 4ledp mr design and radonade ispesented in: 'A Policyand Progmn So>we&gjforMo'E EfficientHousehold Ekaici(y Usre in Indonesia: A>pendwv HI. - 61 - million per year. The program includes elements of programs that have been successfully implemented elsewhere. Q/ Program cost estimates amount to only 25% of the LRMC of supplying the saved electricity over the period. 4.71 Appliance ownership patterns from the UHESS survey have been used to focus on appliances that currently use significant amounts of electricity in households and those which will become more prominent by the turn of t ie century. As low to moderate "income" households dominate the UHESS sample, it is apparent that lighting is the dominant end-use for electricity in the urban residential sector. According to the electricity projections in Figure 3.9, it is evident that lighting will remain a major end use in terms of electricity consumption, but also that electricity consumption for refrigeration and by "other" appliances will grow rapidly in significance by the year 2000. Consequently, the proposed electricity conservation program concentrates on efficiency improvements in these major end-use devices. Characteristics of the Market 4.72 Information obtained from manufacturers and other sources indicate that both imports and locally-manufactured appliances are sold in Indonesia. Duties on imported appliances range from 0 to 60%, but most are over 30%. Imported appliances commonly use efficient technologies available internationally while those locally produced often use technologies from the 1970s and have not benefitted from advances in efficiency made over the past decade. Hence, there is room for improvement in the efficiency of the stock of appliances currently used by Indonesian households. 4.73 Overall, there is little relevant information available to consumers on the energy performance of appliances when making a purchase. Though brochures for air conditioners do give the energy efficiency ratio (EER) and wattages of light bulbs and televisions are certainly not hidden, no indication of how much electricity the particular appliance can be expected to consume annually or how it compares to others on the market is available. The government has recently sponsored a program to sensitize consumers to the need to conserve electricity principally through reminders such as a "Hemat Listik" (Save Electricity) label on new appliances. Apparently, the label does remind consumers of the need to conserve, but actually indicates that a particular device is of low wattage. For refrigerators, the wattage can bear little resemblance to electricity consumption in actual operation. Because households are connected through VA limited supply, they are interested in low wattage devices and manufacturers have responded by making low wattage product lines available. If appropriate information is made available on the financial benefits of efficient devices, consumer response could be dramatic. In addition, the tariff for Rl consumers is still below the cost of supply, thereby not assisting in the conservation effort. L3 E.&g in Sweden, the United States, Bnzl (Pil, 1985, &Env Cnevo in Home Appianes Infome No. 20, Ro Jdeio: CEPEL), and in Jamaic (&serUwJaica Pgnn Plan, Wodd Bk 198). - 62 - Table 4.7: Outline of Electric AppLiance Conservation Program ACTIVITY/PHASE 1990-1991 1992-1994 1995 aw. IMFOUIATION Labelling Available info Results of tests liproved labels used for brochures Labels (Japnse tests) Data in catalogues Testing Establish Procedures Manufacturers Independent body Prelim. tests (LNK) carry out tests carries out tests Sub-setering LUlI initiates Regular panel Repeat every 2-3 survey survey years Outreach First round of Clinics, brochures Consumer centers brochures with with test, sub- limited info metering results Goverrnent Buy only 25% most Buy only 1OX mast procuremnt efficient appliances efficient appliances TECHIICAL INTERVVUAION Targets, SO, consurs, SWEFs used to 2nd set of test designs manufacturers, etc. follow progress SWEF's in force set goals for 91+ 94 Efficiency Studied for Progress examined(1992 Standards imploeted standards 1994,5 Publishing the 1994/95 if not achieved: standaids 1994/5 standards tightened R & D Support of "localizedl Field testirg Marketing technology Consumption metering PRICES Cost-based 1 Increases announced Increases implemented etectr. pres1 Elquctnik Conservation Program Design 4.74 The proposed conservation program has three elements that have been designed to supplement previous conservation campaign experience in Indonesia to effectively move toward efficiency improvements in the stock of household electrical appliances. i) Relevant information would be made available to allow consumers to make effective purchasing decisions regarding energy efficiency and resulting operating costs. The campaign to sensitize consumers about the need to save electricity would complement this effort. ii) Through setting effieiency standards for each appliance type and monitoring the stock of appliances in households through the Sales Weighted Efficiency Factor (SWEF), 4/ the GOI could monitor progress toward a more efficient utilization of electricity in the residential sector. iii) To ensure that consumer incentives are / The Saes Weig*d EffikiyFac*orisa measwv ued in many coun&esWtomoniow *e orvemU eff of ie. ntia elecqk;ty&su. SWEJ7 f the s e makshams ofeach pwwesd, mapdbyits ap geffic jwndr ae coandWwo - 63 - consistent with the economic cost of electricity supply, electricity tariffs should be revised through a phased plan to reflect actual energy costs Consumer subsidies to (partly) offset the increased cost of more efficient designs and improved production methods are thought not to be practicable in the present circumstances. An outline of the proposed End-use Electricity Conservation Program is presented in Table 4.7. Information Intervention 4.75 Information interventions are designed to stimulate conservation through intorming consumers about product choices and use. This requires that the energy consumption of new appliances be tested; actual consumption in users' homes be measured, and; results displayed prominently in catalogues, advertising and labels that are attached to new appliances. A complementary measure would be a wide publicity campaign to inform consumers, officials, and manufacturers about the policy focus on efficient appliances and to educate consumers about efficiency. Information intervention components of the program include: (a) Available information on television sets and air conditioning units (A/C) can be collected rapidly by YLKI or PLN and reproduced in a brochure to be made avaiiable to customers ;n appliance stores as well as through contacts both organizations have with consumers. (b) Information on careful use of existing najor appliances can be made available within one year. A committee, composed of representatives of manufacturers, YLKI and PLN/LMK should develop a list of simple strategies for major appliances and lighting. Both these steps could be implemented in 1990. (c) Information on electricity consumption of appliances tested in the producers' home countries or other markets should be required by 1990 of all imported products. This information is useful for determining efficiency goals and in comparing foreign and future Indonesian test results. (d) With the required testing of existing and new appliances, information should be available to consumers within a few months of the testing. If testing were started early 1990, results could be m. 1e public in 1991. YLKI and PLN should also prepare a brochure for each type of major appliance, listing models available, tested energy use, and other important properties. (e) The consumer awareness outreach activities of both organizations should start in 1990. ITis would initially involve leaflets to inform appliance buyers and other electricity consumers of the energy cost of using major appliances. When better information becomes available, clinics that show consumers how to choose and use appliances effectively could be set up. (f) Government should set an example through appropriate adjustments in its procurement policy. It should start to buy only the more efficient types of Air Conditioners. Already, Government policy requiring fluorescent lighting in office buildings is a step toward energy efficiency. By the time testing and labelling is -64- completed, the GOI will be in a position to procure only the more efficient models of almost all major appliances. Technical Interventions 4.76 The development of test procedures (and the organizations selected to undertake the testing) is estimated to take two years. It will require two to three independent engineers who will represent the consumers and utility in negotiations with manufacturers and judge manufacturer capability for testing. To get testing off to a quick start, manufacturers are the logical place to start the testing. This would only be an interim measure. If the program is started in 1989, preliminary tests could take place in 1990 with the results published in that same year. By the end of 1991, it should be required to inform the buyer if an appliance's tested energy use. 4.77 A "gentlemen's agreement" between the GOI and manufacturers/ importers on efficiency targets could be developed roughly one year after testing, labelling, and information programs have been launched. The targets should be specified to take effect 1-2 years after announcement and be tightened 4-5 years after announcement. 4.78 With the start of the program, the search for an appropriate institution for independent testing should begin. It should be possible to have this body take over the responsibility for the testing from the manufacturers within a period of 1-2 years. A board composed of representatives of all parties concerned would oversee the independent institution. (a) Standards for 1994 or 1995 could be announced in 1992, i.e. after enough testing has been undertaken to determine the actual and optimal features of major products on the market. A second round of standards should follow 2-3 years later. (b) If research on developing technologies specifically for the domestic market began in 1992, using local companies, the first marketable results could probably be available by 1995. Market Adjustment 4.79 Electricity tariffs should be kept in line with the real cost of supply while still a%counting for equity considerations. Price adjustments interact with the other interventions. As prices rise, consumers get greater benefits from investing in more efficient appliances and also see greater returns from modifying how they use appliances. Information advises consumers on which appliances and usage patterns are more efficient. At the same time, the suppliers of appliances see a larger and safer market for improved, efficient appliances and will tend to be more willing to invest in improvements. Economic Evaluation of the Program Investrnent Costs for Production of Efficient Appliances 4.80 Improving the design and production of appliances usually involves cost increases. These would be passed on to the consumer. However, the improvements that are needed do not - 65 - involve expensive leading edge technology. In fact, they are commonly found in the appliances that are imported into Indonesia: appliances which were made to standards prevailing in major consumer markets such as Western Europe, Japan and the USA (See Appendix D). Domestic manufacturers on the other hand, often work with obsolete designs. Most bewtuse they are subsidiaries of foreign companies which put out much more advanced models in the:k' home market. The parent company, however, has no reason to change the designs and production methods used in their Indonesian plants unless consumers or the GOI demand it. The small minority of unaffiliated local producers lacks both incentive and, generally, access to more recent technology. The investment costs related to increased efficiency are estimated to add not much more than 5% to 10% to the cost of major appliances. S/ Program Cost Table 4.8: Program Targets for 1999 4.81 The actual costs of the proposed energy conservation program in Light bulbs Refrigerators A/C TV Indonesia wil depend on how many Potential Savings 30X 25X 25X 5K appliances are ultimately selerAed and which Market Share 10X 35K 35K 20K interventions are initiated. For a similar kind Savings: Potential kWh savings per inproved of energy conservation program proposed for appliance expres,sed as percent of kWh consuned in Jamaica, ESMAP has estimated the cost of ordinary appliance. K Share: Share of total market for each appliance (manufacturer) testing and labelling, the captured by improved appliances by the year 2000. major cost components of the program, to amount to less than US$ 2 per appliance sold. Program cost estimates for test facilities, training and personnel, an information campaign, and administering the testing and labelling of refrigerators, air-conditioners, and televisions are shown in Annex XII. These estimates are based on experience with similar facilities in other countries, including Jamaica, the Philippines, and Brazil. Program Benefits 4.82 To evaluate the benefits of the proposed program, projections of electricity consumption in the urban residential sector with and without the program were made. Projections of base case electricity consumption, without the program, are presented in Chapter m and displayed in Figure 3.9. For estimating potential savings due to the use of more efficient appliances, projected sales to urban households of each major appliance type were estimated and are shown in Annex XII. Current patterns of appliance ownership from the UHESS survey were used to project appliance holdings by appliance type as a function of income growth and age of connection to the grid. These results were used to calculate annual sales for each important appliance type. The program targets in Table 4.8, expressed in terms of market share captured and kWh savings obtained, were considered to be feasible for a program to achieve over a ten year period. Potential savings refer to the amount of electricity that could be saved (as a percentage of electricity consumed by an ordinary device) by switching from a common device to a mora efficient design. In the case of lighting, the potential savings are entirely due to a switch from incandescent M/ Apliwaces such as refiigers, Jfreis, air cdadtions, and pi is (See Appendi HI). -66 - Table 4.9: Summary Evaluation of Urban Residential Electricity Conservation Program Net Present Value a 10X Annualized NPV (USS Nitlions) (USSt llions) Total Progm Cost (USS .1111cm) 14.7 2.3 Electricity Conservation (Ch) 729 112 Lighting 384 59 Refrigeration 295 45 Air-conditioning 28 4 Television 21 3 Economic Value of Swvings (USS .wittfm) 64.3 9.9 Lightins 33.9 5.2 Refrigeration 26.1 4.0 Air-conditioning 2.5 0.4 Television 1.9 0.3 Net Economic o8nef Its 49.6 7.6 Source: See Annex XII. bulbs to fluorescent tubes. M/ Assumptions of potential savings from efficient refrigerators, air conditioners, and televisions are derived from analysis of the technical specifications of common models on the market in Indonesia in comparison to efficiency ratings of those available internationally (see Appendix III). 4.83 The annualized net present valbe of estimated program costs and total GWh savings due to efficient appliances capturing the targeted market shares are shown in Table 4.9. By the year 2000, savings could amount to over 300 GWh per year or roughly 3% of total projected urban residential electricity demand. When valued at the LRMC estimate of US 90/kWh (150 Rp/kWh) and discounted at 10%, projected savings over ten years are worth about US$ 64 million (US$ 10 million per year) for a total program cost of US$ 15 million (US$ 2.3 million per year). The breakdown of annual electricity savings shows that roughly 50% of program targeted benefits would be generated by savings in electricity for lighting due to a switch from incandescent to fluorescent bulbs amounting to 10% of the incandescent market by the year 2000. Institutional Arrangements 4.84 Any effective residential electricity conservation program along the lines summarized here and presented in more detail in Appendix II, requires the active participation of electric lamp and appliance manufacturers (through exsting associations), consumers groups (YLKI), government, and national laboratories (PLN-LMK). The Directorate General for Electricity and New Energy &6 Ah fluorscent lmps are, on awvge, S times mo efien tan incancen butbs (80% meIz,X savings can be obtained per men-hour output by lighdtng with J1uou er than incdcent), a mom consave 30% eeV saing has been assmed her because iteis ewdthat housholds switcig toluorescet mps wU ight more than they woudd have with cndesent bulb * 67 - and the Minisy of Induatty would coordinate the dialogue and provide an appropriate forum for bringing these major actors together. Such an effort should be fully coordinated with similar programs targeted at efficiency improvements in generation, tasmission, and distribution systems. Anne"I ANNEX I Petmreum Product Prices and Sales DOMESTIC PRICES OF PETROLEM PRODUCTS, 1978-1987 (Rupiah/titer) 1978 1979 1960 1961 1962 1963 1964 1985 1986 1987 Aviation ¢s 70 100 150 150 240 300 300 330 250 250 Aviation Turbo 70 100 150 150 240 300 300 330 250 250 Premium Gaoline 90 140 220 220 360 400 400 440 440 440 Regular Gasotine 70 100 1so 1SO 240 320 350 385 385 385 Kerosene 18 25 38 38 60 100 150 165 165 165 Motor Diesel 25 35 53 53 85 145 220 242 200 200 Industrial Diesel 22 30 45 45 75 125 200 220 200 200 Fuet Oil 22 30 45 45 75 125 200 220 200 200 DOMESTIC SALES OF PETROLEU PRODUCTS I978-1988 J ('o0 bbts) 1978 1979 1960 1961 1962 1963 1964 1985 1986 1967 1968 Avfatifanas 134 134 130 110 103 83 73 66 63 56 60 Aviation Turbo 3,494 3,656 4,355 4,869 4,899 3,666 4,374 4,442 3,806 4,199 4.445 Premium Gasoline 728 618 466 392 238 247 523 738 1,024 1,836 1,431 Regular Gasoline 19,608 21,295 23,321 25,648 25,709 24,380 24,909 25,206 27,083 29,048 30,855 Kerosene 41,717 45,457 48,975 52,497 51.778 48.224 45.213 43,954 43,618 43,352 44.664 Notor Diesel 31,7G9 34.595 40,116 44,737 48,918 49.790 48,567 47,662 47,421 54,075 59W143 Industrial Diesel 6,744 7.581 7.829 9,391 9,311 9,978 10,285 10,329 8,855 8,319 8.809 Fuel Oil 11,061 13.626 15,739 17,S87 19.341 21,149 23.625 22,863 18,004 19,054 18,097 TOTAL 115,195 126,962 140,931 155,231 160,297 157,537 157,569 155,280 149,874 159n941 167,504 £1 Excluding lubrtcating oil and similar products. LPG: DOMESTIC SALES AND EXPORTS, 1978-1968 (NT) 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 Domestic Sales 46,545 53,048 59,597 69,470 74,537 86,522 110,578 145,617 180,268 216,161 235,099 Exports 444,957 365,178 520.433 484,312 442,251 364,689 725,148 635.438 524,443 525,915 972,320 Source: Ministry of Mines and Energy. -69- Annex ANNEX II Dlstribution Systems for Kerosene and LPG Kerosene Distribution System Schematic Import and Domestic Refinery Processing Deal Portamina Seafed Depots |Upcountry Depots l I ' ~~~ - J ~Agents X , s ~~~~~~Pangkalan Door-to-door Vendors Final Consumer -70- Am11 LPO Distnbution System REFINERY/ BOTTLING DEPOT AGENT CONSUMER PLANT PLANT North Sumatra TT ------- RANTAU TANDEM i i I ~~~~~~~~~~~~H South Sumatra 0 mUSI SUNGAI . _ S GERONG E H CT I O0 JAMBI/LAMPUNG - L Jakarta D TK CILACAP TANJUNG PRIOK A N r ] ~~~~~~D TK I SUKABUMI West Java TK I MUNDU N D - - CT ~~~SEMARANG U L ~~TEGAL __ S CILACAP T East Java R SURABAYA > SOLID YOGYAKARTA MALANG TK KEDIRI MADIUN _ JEMBER Sulawesi |BALIKPAPAN UJNPADG|__ ~~7yyyjcT ____ _____ - 71- Anna III ANNFX III Analysi of Kerosene Price Changes and Urban Elctrifcaton / 1. Fuel use patterns have been changing very rapidly in recent years, but not always in the most obvious directions. The changing fraction of households using each fuel between 1981 and 1987 is presented in Table A3.1. The fraction of urban households nationwide using electricity has grown rapidly, reaching 74% by 1987. The fraction using LPG has also grown rapidly, though LPG remains a minor fueL still used by less than 5% of urban households in 1987. Kerosene remains by far the most widely used fuel, but the fraction using kerosene falls significantly: from 98% in 1981 to 93% in 1987. The fraction of households using wood actually increases from 20% to 24% over the same period, while the fraction of households using charcoal follows the more typical pattem of declining traditional fuel use, falling from 28% to 16%. Piped gas, restricted to a few of the larger cities, is not used by a significant share of urban households, and is not included in this discussion. 8/ 2. These changes have not been simply, or evei. predominantly, the result of income growth. Table A3.1 also presents fractions of households using each fuel for different expenditure groups denominated in constant 1988 Rupiah. D2/ For most fuels, income growth, which was relatively slow during this period, merely acted to accentuate the shifts within income groups. The fraction using electricity has grown rapidly in all but the highest income group. The fraction using LPG has more than tripled, even in the top income group. By and large, the increase in the fraction using LPG is probably the result of LPG's relative novelty: more and more householders, particularly at upper income levels, are coming to recognize the advantages of LPG, and it is simultaneously gaining a reputation as a 'modern" fuel, conferring status on its users The fraction using charcoal, which is negatively related to income, fell in all income groups. For wood, on the other hand, the overall fraction has moved against the income trend: the increase in the fraction of households in the three poorest income groups more than compensated for the declining share of households in the lowest income group. Finally, the fraction using kerosene, which until 1987 showed little relation to income, has fallen in all groups (though particularly among the vety rich and very poor, for reasons clarified below). 3. The fuel quantities are harder to estimate than the fractions of user households. For kerosene, LPG and charcoal, the SUSENAS survey provides a useful, if rough, approximation. For a/ Anasis is based on SUSENAS oen_ditaws wfrmo 1981,1984, wad 1987. L/ Piped gas for houhold use is likely to nmmc smal for die it of die cea. Whue dwe may be inweWesng passbies for wban hou d piped gas in can nstd z*oncs, dey wdd be best eamined on a pec by project bais, and ae not included In die UESS ana(ysi L/ Alto de groups am delineated by expcnd perhouseho" they at a rfenvd to as oe gwrups in die tat and tables sxne, in hffec et.dcex is beng used as a susrogate for inome. The groups wv csn t be compmble with Tables dmwn fimn die 1988 UHESS wy of Udan Jav For Indonesia in 1987, the jfve apeditua/cowne groups wppsnt diw poorest 21% of househol de nat 26*, die nea 24%, the nesx 18% and die dchest 12%. As t"h househodf also tend to be laeu, household income diffencs am 1a dan per capa income diferences between the same grp -72 - Annex Il Table AU: Percentage of Households Using Fuel by Income and Year Income Group Electricity LPGl Kerosene Charcoal Wood 1987 Very Low 44% 0.0% 94% 17% 48% Low 68% 0.2% 95% 23% 29X Moderate 82% 1.4% 96% 17% 18% High 92% 4.9% 95X 9% 11X Very Hfgh 95% 27.5x 78 7% 6% Atl 74% 4.5% 93X 161 24% 1964 Very Low 26% 0.0% 95X 16% 42% Low 51% 0.0% 97X 26% 23% Moderate 70% 0.4% 98% 21% 14XA High 85% 1.5% 98% 12% 9X Very High 95% 10.0% 92X 6% 5% Atl 61% 1.4X 96X 18X 21% 1981 Very Low 16X 0.0 98% 20% 36X Low 33% 0.0% 99X 37% 21X Moderate 54% 0.0% 98X 35X 16% Higl 76% 1.3% 98X 24% 11% Very High 86% 7.5% 95% 15% 6X Alt 471 1.0X 98X 28X 20% Income Groups are delfneated by Total Expenditure per Household in 000' 1988 Rupiah: Very Low c75c Low 4120< Noderate <185< High <295 Very High. Table A32: Average Combustible Fuel Use per Household by Fuel, Income, and Year LPG KEROSENE CHARCOAL WOOD* TOTAL (NJ/Day) (NJ/Day) (NJ/Day) (NJ/Day) (NJ/Day) Income Group 1987 Very LoW 0.0 18 0.4 31 49 Low 0.0 27 0.6 19 46 Moderate 0.3 34 0.5 12 46 High 1.3 40 0.4 7 49 Very High 7.9 38 0.5 4 50 All 1.2 30 0.5 16 48 1984 Very Low 0.0 21 0.3 27 49 Low 0.0 34 0.6 15 50 Noderate 0.1 41 0.5 51 High 0.3 47 0.4 6 54 Very High 3.3 50 0.3 3 57 All 0.4 3T 0.5 14 51 1981 Very Low 0.0 33 0.3 23 56 Low 0.0 4 0.8 14 59 Noderate 0.0 50 0.9 10 61 High 0.3 53 0.8 7 62 Very High 2.0 59 0.5 4 66 All 0.3 U 0.7 13 60 *Assunes constant use/using household of 140 kg/month. 73 - Annex m electricity, however, SUSENAS estimates are inadequate, 0Q/ and fuelwood quantities are simply not recorded. In the quantity tables which follow, electricity is omitted, and wood use estimates. when included, assume that each household using fuelwood consumes 140 kg per month (the 1988 Java average for user households). Given the inherent uncertainty attached to quantity figures from an expenditure survey, and the particular uncertainty as regards fuelwood, these tables must be interpreted with care. 4. The changing LPG, kerosene and Table A3.3: Average Fuel Use per charcoal use patterns of user households are Using Household by Income and Year summarized in Table A3.3. The quantity of kerosene Kerosn consumed by a user-household shows a clear upward Liters/month/hosehotd) trend across different income groups, and a clear 1981 1984 1987 downward trend over time, especially at the lower Very Low 28 19 16 income levels. LPG use does not appear to change Low 38 30 24 significantly between rich and poor users, or between Moderate 43 35 30 users in different years. For charcoal, the use/user VeryHigh 46 41 36 users idifrnyer.Frcacathus/er very High 54 46 42 also increases with income, but has increased with time too. The kerosene price increases are probably Ati Users 40 33 28 the major cause of the faDl in kerosene use/users. The rising level of charcoal use/user is likely to be LP6 largely the result of displacement of charcoal ironing, (kg/month/househotd) which only requires relatively small amounts of 1981 1984 1987 charcoaL by electricity leaving a bigger share of large Incan Very Low 1 charcoal users. The relatively steady LPG use/user Low * 11 is not surprising, though one would expect a positive Moderate * * 14 relation between LPG use/user and income. High 13 14 17 Very High 17 22 19 5. Average fuel use per household, ;n Alt Users 16 20 18 energy content (MJ), is summarized Table A3.2. The * Less than 3 HH/category. direction of most of the trends can be derived unambiguously from the preceding discussion. cba.rcoa Kerosene is clearly the dominant fuel, but the decline (kg/month/househoLd) in kerosene use per household is strikdng (7% per 1981 1984 1987 year between 1981 and 1987). Wood is also Very Low 2 2 3 important, and wood use per household grew at a Low 3 3 3 significant 3% per year. LPG started as a very minor Moderate 3 3 4 High 4 4 5 fuel, but with a growth rate per household of 26% per Very High 4 6 8 year, had become significant by 1987. Charcoal is shown to be a relatively insignificant fueL despite its all Users 3 3 4 widespread use. The growth in urban population accentuates the growing wood and LPG use, and leaves kerosene declining at a relatively slow 2% per year. i/ SUSENASsweyos sometimes ,coed ekeidcy 'conwnptonin kWh, somede in VA Dwing data pocesng the VA am translated into kWh uing an exesswve loadfactor (over 60%), Andeng the ekacicity quanlit data unuabk. *74 - Annex Ml 6. The combined energ content of the combustible fuels used (per household) is shown to fall by 12 MJ/day, on average. A part of this fall (approximately 5 MJ/household/day) can be explained by the substitution of electricity for kerosene lighting. Much of the rest of the decline can be attributed to household response to rising kerosene prices that led not only to fuel switching, but to considerable fuel conservation. 7. According to SUSENAS data, average urbaa household kerosene consumption declined by 16 MJ/day between 1981 and 1987. During the same period, the percent of households electrified grew from 47% to 74% of urban households. Using the average kerosene use for lighting by users: 0.5 liters/day (UHESS Summary Table 5); and noting a clear preference for electric lighting expressed by households surveyed, electrification of 27% more of total urban households by 1987 leads to an average displacement of roughly 5 MJ/day/household of kerosene for lighting by electricity. Using the cooking fuel substitution ratios derived in Chapter m, it is apparent that the increases in average household consumption of LPG and wood over the period of 0.9 MJ/day and 3 MJ/day, respectively, substitute for, on average, 1.2 and 1.8 MJ/day/household of kerosene. The residual decline in per household kerosene use over the period, 8 MJ/day, can be attributed directly to fuel conservation. 8. Treating electrification as independent, if the drop in kerosene use of 1 IMJ/HH/day from '81 to '87 due to substitution of LPG and wood and conservation is attnbuted primarily to increases in the real price of kerosene faced by households, all else being equal, a rough estimate of the kerosene own price elasticity of demand for urban households results. Over the period, the real depot price of kerosene (in 1987 Rupiah) rose from 62 Rp/l to 165 Rp/L This leads to an estimate that the real price faced by households rose from 80 Rp/l to 200 Rp/l (in 1987 Rupiah): a price increase of 150%. Kerosene use in urban Indonesia fell from 46 MJ/HH/day to 30 MJ/HH/day, 11 MJ/HH/day attributed to the price increase: a quantity reduction of 24%. The imputed kerosene own price elasticity is -0.16. .75 - Annx ly ANNEX IV Derivation of Liters Kerosene Equvalent for Cooldng and Ligting Cokin&-Fuel Demand Projections and Inter-fuel Substitution Ratios 1. The goal of the projection procedure developed for this study is to provide a tool for evaluating household energy strategy options. The projections, outlined in Chapter III, rely heavily on inter-fuel substitution ratios and fuel shares based on "liters of kerosene equivalent" (LKE). This Annex describes both the estimation and application of these quantities, as both involve innovative techniques. 2. The emphasis on inter-fuel substitution follows from the anticipated importance of fuel switching over the coming years. Cooking in urban Indonesia is carried out predominantly with kerosene, fuelwood, 61/ and LPG. Income growth is likely to cause a decline in the share of fuelwood-cooking, and an increase in the share of L-cooking. Future kerosene use depends in part on the relative speed of these two substitution processes. The projection procedure selected for this study attempts to capture these processes, and the resulting effects on fuel use, through a cross-sectional analysis of existing households. Factors not amenable to cross-sectional analysis are introduced ad hoc. Defining "Liters of Kerosene Equivalent" and "Inter-fuel Subsituton Ratios" 3. A "liter of kerosene equivalent" (LKE) of cooking fuel can be either a liter of kerosene, or a specific quantity of another fuel which would replace (or be replaced by) a liter of kerosene used in cooking. Thus, for example, if a household using a liter of kerosene each day for cooking, uses .7 kilograms of LPG per day upon switching fuels (and vice versa), then .7 kilograms of LPG are equal to 1 LKE of LPG. For simplicity, it is assumed that the average amount of LPG (or fuelwood) employed to substitute for a liter of kerosene is always the same, regardless of household size, household income, or relative fuel prices. In other words, the inter-fuel substitution rtios are assumed to be constant. This assumption is maintained throughout the analysis undertaken for UHESS. It is possible to adopt more complex formulations allowing for different stove types within a give fueL or different substitution ratios at different income levels, but this was not deemed necessary. 4. "Liters of kerosene equivalent" and "inter-fuel substitution ratios are similar concepts to the very commonly used concepts of "useful energy demand" and "stove efficiencies". This is evident from equations I and I. There are important differences in their application, however. The first equation is an empirical hypothesis about cooking fuel use, stating that interfuel substitution ratios are constant. The second equation is a physical identity, with efficiency and useful energy best measured in a laboratory. When applying the second equation to household fuel- demand analysis, it is typically assumed, not only that the inter-fuel substitution ratios are constant, but that the useful energy demanded is not affected by the households fuel choice. For reasons summarized below, this latter assumption has been shown to be incorrect iJ/ DeFined to inlde a vaieJy of biofuwhiding fedcWp sida as ml! as Orm -76 - Annex IV I. LIC; = a, LPG-kg + a, WOOD-kg + KERO-It II. UED = el LPG-hv + e2 KERO-hv +e3 WOOD-hv Where: LKE is liters of kerosene equivalent of cooking fuel. a, and a2 are inter-fuel substitution ratios. LPG-kg, WOOD-kg and KERO-It are quantities of fuel used in cooking. UED is useful energy demand for cooking. e1, e2 and e3 are efficiencies. LPG-hv, KERO-hv and WOOD-hv are quantities of fuel used in cooking, (hv indicates quantities nmeasured in heating value). 5. One of the problems with applying "useful energy demand" and stove efficiencies to cooking is described in detail in Appendix II. Briefly, controlled cooking-test results suggest that power influences fuel-economy independently of efficiency, because more useful energy is often required for high-power cooking. For example, simmering at low power cooks at roughly the same rate as boiling at high power, even though under high power more heat is transferred to the contents of the pan: the additional "useful" energy merely converts more water to steam and cannot raise the temperature of the liquid above its boiling point (unless the pan is pressurized). If cooks always adjusted the flame to deliver the same power to the pan, regardless of the stove, this might not matter. In practice, the necessary flame adjustment is often not possible, let alone practiced. As a result, the type of stove employed influences the demand for useful energy. The most important systematic differences are between stoves using different fuels. Therefore, inter-fuel substitution ratios can be expected to depend on power characteristics as well dS efficiency. L2/ 6. There are further reasons to expect useful energy demand" to vary depending on the fuel a household chooses to employ. From a social perspective, different fuels may be associated with different lifestyles, and hence different "useful energy demands". Economically, the different margjal cost of cooking with wood, kerosene and LPG, can make "useful energy demand" contingent on fuel choice, even for a household indifferent (at existing prices) as to which fuel it uses. 7. In sum, cooking habits, and hence "useful energy demands", are fuel-dependent. Laboratory measurement is insufficient to determine inter-fuel substitution ratios. Analysis of household fuel consumption behavior is more suitable. i/ LPG stow end to be higher poweed an kerosene stows: in the populat-e LPG stove tested for this stdy, even the niwnpowoftheLPGsto washigherththemanwmpow of mostofthekosenestovesteted One migfht et thewfa% for LPG stos to use moe fuel than wuld be calculated on the basis of relatw efficiencies alone. As descibed below, this does appear to be the case. -77, VAna A Procedure for Estimating Inter-fuel Substitution Ratios and Elasticities of LKE Demand 8. If a household's consumption of cooking fuel could be measured directly in liters of kerosene equivalent, the following equation could be directly applied to household data, and the coefficients estimated statistically: HI. ln(LKE1) - b + b,*In(eWp) + b?In(famJ + e; where: expe = household total expenditure per month. fam = household size The coefficient b, is the expenditure (or roughly income) elasticity of cooking fuel demand, while b2 is the household size elasticity of cooking fuel demand. While other explanatory variables could be added, this formulation would be adequate for most purposes. 9. Since LTKE of cooking fuel use cannot be measured directly, a slightly different approach is required. The following equation combines equations I and m in a form suitable for statistical analysis on households using exclusively either LPG, wood or kerosene: IV. ln(en), = b + b,*ln(expJ) + b2~*n(fami) + b,*LPGY1 + b4*WoodY1 + e, Where: en = household cooking fuel use expressed in Joules/day. LPGY - 1 if use LPG for cooking = 0 if do not use LPG for cooking WoodY - 1 if use wood for cooking = 0 if do not use wood for cooking Coefficients b, and b2 can be interpreted analogously to those in equation m: they are demand elasticities. Coeffcients b. and b4 are related to coefficients a, and a2 in equation L 10. Again, additional variables could be added. If inter-fuel substitution ratios are the principal objective, more variables pertaining to cooking habits would be appropriate (so long as these cooking habits are not the result of fuel choice). Alternatively, a more sophisticated procedure could be developed to test for and correct problems of self-seluctivity and endogenous dummy variables. Self-selectiity problems would arise if households select fuels because they have a comparative advantage in economizing on their chosen fueL A problem of endogenous dummy variables would arise if the amn unt a household cooks influences their choice of fuels. While both these problems may exist, the effort required to test and correct for their influence was not warranted given the objectives of UHESS. Estimating Inter-fuel Substitution Ratios and LKE Demand Elasticities from UHESS Data 11. Preliminary analysi of the UHESS household data indicated that fuel choice in urban Indonesia could be divided into: 1) the choice between kerosene and wood made by lower income households, and; 2) the choice between kerosene and LPG made by upper income households. The statstical analysis was conducted separately for each of these income groups. -78,- iIX Households with incomes below 100,000 Rupiah per month were included for the former, and those with incomes above 150,000 Rupiah per month for the latter. This separation allows for differences in the non-fuel coefficients for upper income and lower income households, tnd prevents questionable outliers (ie, low income LPG users and high income wood users) from having undue influence. Q/ 12. As indicated above, households using a mix of cooking fuels were not included in the statistical analysis. Such households were excluded both to simplify the estimation procedure, and because of an apparent tendency for some of them to overestimate their fuel use. i/ They were re-introduced in the later analysis of fuel shares, by applying the substitution ratios calculated from single-fuel households. The results of the analysis are summarized below. 13. Physical quantities of cooking fuel used by each household were converted into heating value using 14 MJ/kg for biofuel, 35.2 MJ/l for kerosene and 45.77 MT/kg for LPG. Regressions to determine substitution ratios were run using only households in an income range in which substitution can be expected because each of the fuels provides more than a few percent share of cooking needs. For wood-kerosene, this range was chosen to be the households with expenditure level between Rp. 30,000 and Rp. 100,000 per month. For LPG-kerosene, the chosen range was Rp. 150,000 - 1 million. In addition, only households using one fuel for cooking were included (over 90% of the sample households use only one fuel for cooking). 14. The calculated energy consumption (en) in MJ/day is regressed against the variables: expenditure level (expe), family size (fam) and KerY. The latter is a dummy variable which is set equal to 1 if kerosene is used for cooking. For the wood-kerosene substitution this becomes: ln(en) = 3.608 - 0.013 ln(expe) + 0.389 ln(fam) - 0.525 KerY Std Dev - (0.731) (0.069) (0.042) (0.041) t = (4.9) (-0.19) (9.3) (12.8) = 0.29 15. The substitution ratio can be calculated from the parameter of the dummy variable: Substitution ratio = energy use for cooking if kerosene is used divided by energy use for cooking if wood is used = exp(-0.525) = -.592. This ratio in energy units can be transformed to physical units, using the above mentioned specific energy contents, resulting in 4.25 kg wood per liter of kerosene. O/ The berf of sewuon war deened mae impotant dean the sanple selectn bias which oauld vsut hI any case, as noted In Chquterp de msui& am no peay affectd by anabzing ad hwehold at oace. i/ EVdeneh dtefield SUesSta dt mu&-fud haoholds snetmswe siwpem rasespidg to day whe du, used de4 even if bdq mwn using it mguk* dwing dte ptiod queson. -79 - Annex I 16. In case of LPG-Kerosene substitution the following regression results were obtained: ln(en) 1.647 + 0.085 In(expe) + 0.473 In(fam) + 0.263 KerY Std Dev = (0.572) (0.045) (0.038) (0.059) t (2.9) (1.9) (12.3) (4A) R m 0.18 Ihis leads to a substitution ratio of 0.768, which is equivalent to 0.591 kg LPG per liter kerosene. 17. Table A4.1 clearly indicates the influence of both urban area size and income level on the share of cooking done with each of the major fuels. On average, an urban household on Java uses 1.2 LKE equivalent per day for cooking, with higher levels in upper income or larger households. Kerosene meets 74% of this demand, wood a further 21% and W about 5%. Wood becomes the most important cooking fuel in small urban areas and at low income levels. LPG reaches its highest share (38%) in the uppermost income group of the largest cities. Kerosene displays a somewhat more complex relation to income, and for most urban area size categories the share of cooking fuel demand met by kerosene grows with income only up to a certain point, after which it begins to decline. For the most part, the direction of these tendencies is as expected. 18. Households in the highest expenditure category consume on average 56% more liters of kerosene equivalent of cooking fuel than those in the lowest expenditure category, as much because of their larger household size as because of their higher income per se. OI Based on the equation presented in Box 3.1 in which all households are included, an increase of 24% due to the larger household size in the uppermost expenditure group and 20% due to their higher income could be expected, combining to yield an overall increase of 48% (somewhat less than the observed difference). While other formulations tend to assign a smaller share of the change to household size, it may well be that less than half of the difference in total cooking fuel use (LKE) across the income groups is the direct result of income differences. Using LKE and Inter-fuel Substitution Ratios in Energy Demand Projections 19. As the basis for projecting energy demand, the LKE approach retains one of the major attractions of the 'useful energ approach: many of the assumptions and results are intuitively accessible to energy experts and policy makers. Consider, for example, the claim that over the next ten years in Country X, the overall heating value of daily per household cooking fuel use will grow from 68 to 77 Megajoules, with LPG increasing from 9.2 to 45.8 Megajoules (.2 to 1 kg), kerosene remaining constant at 17.6 Megajoules (.5 lt), and wood falling from 42 to 14 Megajoules (3 to 1 kilogram). Expressed in this way, it is difficult to judge whether these figures are reasonable, let alone interpret them. Consider, by way of contrast, the claim that the daily per household cooking fuel demand will grow from 1.5 to 2.4 liters of kerosene equivalent, with LPG growing from .3 to 1.7 LKE, kerosene staying constant at .5 LKE, wood falling from .7 to .2 LKE. From these figures is possible to note the large increase in overall cooking fuel demand per household, unlikely in any but the most extreme situations. Furthermore it is possible to calculate market shares, and demonstrate that LPG is expected to capture more than 70% of the market by O/ S&idi &wis iudicated dt khoudw size is kes sipdfta as rgw* teo fu dwice *80g IJV Table A4.1: Cooking Fuel Use by Income and Urban Area Size UHESS 1968 PCAATIN OF U AM (_u_nds) EAPENDITURE LEVEL - ALL (fp/HH/Nonth) 4 44-300 300-120 '1200-2400 Jakarta Very Low (475.000) UXMlday 1.12 .91 .0? .84 1.14 .98 X Wood 76.0X 42.1X 39.11 2.?X .0 46.51 X Kerosene 23.6X 37.9X 0.9X 97.3 96.6X 53.11 X LPG .4X .0O .05 .OX 3.4X .3X X Electricfty .OX .X .XOX OX .OX .0O Fmfly Size 3.7 3.4 3.5 3.2 3.6 3.5 Munber of Households 235 164 13J 91 29 654 Low (75,000 - 120,000) UE/day 1.12 1.18 .95 1.07 1.18 1.10 X Wood 48.21 30.61 21.21 4.9X 2.6X 23.3 X Kerosene 51.8X 68.X 7S.4X 93.31 94.8X 75.2X X LPG .01 .?X 3.31 1.8X 2.7X 1.51 X Electricity .0Q .01 .01 .0O .0Q .0O Fmi ty Size 5.1 5.1 4.7 4.7 4.0 4.8 Nlumber of Hous.holds 123 14 10 110 94 581 Moderate (120,000 - 185,000) IK/day 1.33 1.27 1.05 1.12 1.21 1.19 X Wood 26.71 15.01 15.41 .4X 1.1X 8.91 X Kerosene 72.0 81.tX ?.1X 97.6X 95.0X 87.51 % LPG 1.3X 3.11X .4A 1.91 3.9X 3.61 X Electricity .1t .OX .1X .2X .0O .1X Fami y Size 5.7 5.2 4.7 s.3 s.0 5.1 Nluber of Households 80 9 89 110 202 577 High (185,000 - 295,000) UKE/day 1.36 1.48 1.10 1.53 1.38 1.38 X Vood 24.21 12.71 3.21 .01 2.4X 5.7X X Kerosene 72.11 60.1X 75.7 91.0X 94.51 86.91 % LPG 3.7X 7.1X 20.9X 8.01 3.1X 7.21 X Electricity .0O .1X .3X 1.01 .0O .2X Fmily Sfze 6.0 5.5 5.2 5.8 5.8 5.7 Number of Households 39 56 5T 69 162 383 Very High (t2.°,000) LKE/doy 1.59 1.70 1.33 1.49 1.57 1.54 X Wood 14.71 7.41 2.3X .01 .8X 3.6X X Kerosene 75.81 78.3X W.7U 81.91 60.5X 71.51 X LPG 9.5X 14.OX 16.5X 15.21 37.6X 23.81 X Electricity .0O .3X A4X 2.91 1.1 1.1X Fdly Size 5.8 S.9 5.9 6.0 6.3 6.1 Nuber of Households 31 38 33 102 248 ALL HOUSEHOLDS UW/day 1.20 1.18 .90 1.15 1.31 1.18 XWood 53.6X 27.51 21.71 1.91 1.6X 21.11 X Kerosene 45.11 69.8X 7.71 93.7X 88.91 73.81 X LPG 1.3X 2.71 6.5X 3.91 9.3X 4.9X X Electricity .0O .01 X1X .5S .21 .2X Fanfly Sfze 4.7 4.7 4.5 4.8 5.2 4.8 Nwer of Households S08 500 422 424 59 2443 .81- AnnexIV the end of the ten years, despite sta with only 20%. In short, expressing the projection results in LICE gives them added meaning 20. Perhaps more importnt, the LIC approach incorporates important structural features of cooking fuel demand. Krosne, wood and LPG AMl alternative cooking fuels, and provide different means of performingessentially the same tasks: heatingwater, cooking rice, etc. L6/ Since the tasks exist independently of the choice of cooking fuel, estimating a fuel-neutral cooking fuel demand (e.g. LKE) is not onl intuit appealing, but analytically justifiable. Any approach which does not explicitly distingh between inter-fuel substitution and changing overall levels of cooking fuel use, risks either neglecting inter-fuel substitution or becoming overly complex. If the different fuels are analyzed separatly, it b dfficult to model, for example, the phenomenon of kerosene demand first rising with income (as it displaces wood) and then falling (as it is displaced by LPG). 21. The manner in which the inter-fuel substitution ratios were used for the UHESS projections is described in Anne VIl Applied to the survey households, they provided the means to estimate both overall cooking fuel consumption (in LKE) and shares met by each of the major fuels. As indicated in Figures 3.4 and 3.5, the share variables were found to be very "we!' behaved", and displayed an especially strong relation to household income. Separate functions were estimated for total fuel consumption (LE) and fuel choice (ie shares), at first based only on the survey results. These functions were then applied cross-temporally to the all urban households in Indonesia, using exogenous projections of income, population and fuel prices. The projections were undertaken both forward to the end of the century, and back to 1981. The backward projection was used to check the modeL and in particular the assumed fuel price "elasticities", and resulted in the incorporation of a technical change paramter to allow for the rapid growth of LPG. Estimation of Costs and Benefits of Eleriit Substitution for Kerosene in Lighting 22. As shown in Table A4.2, the fraction of households using electric lighting mirrors electricity use, with few exceptions, Amost al households with electricity use only electric lights on a regular basis, though about 2% of households combine kerosene and electric lighting. For cooking, it was demonstrated that when households switch fuels it is not possible to predict how much of the alternative fuel they will uw on the basis of technological efficiency alone. For lighting this holds a fordoni. Whereas households who switch cooking fuels are unlikely to change the amount of food they cook as a resut, when households switch from kerosene to electric lighting it is quite likely that they will engae in more light-enabled activities, such as reading. 23. The analysis prented below is intended to estimate i) the amount of electricity households use to displace kerosene ljiting and ii) the relative change in amount of light obtained by such a switch. A statistical (OLS) tmation procedure is performed on UJHESS data from urban Java, while controlling for the Iuns of household size, income, number of rooms, and city size, all of which affect lighting fuel choice and levels of use. The results are given below: W/ Wood may have imoetat bhipvawu eddip hi nal am -82 - Anne Takle 4.i2: UIghting Practces and Income - Java 1988 Income/Expenditure Group 00.0Rupfah per Household per Month 0 - 75 - 120- 185- 295 -Top Atl NH Electrified 65X 85 95f 97X 10OX 85K Number of Rooms 3.3 3.7 4.1 4.8 5.6 4.0 Lighting Source Kerosene Only 36X 16X 5X 3K oX 15S Electricity Only 60X 80K 93X 95K 99M 83X Combination 3K 4K 2K 1K 1K 2X Number of Bulbs Fluorescent .6 1.0 1.6 1.9 3.2 1.4 Incandescent 1.9 3.0 3.6 4.3 5.3 2.7 Total 2.5 4.0 5.2 6.2 8.5 4.1 Wattage (U) FLuorescent 11 19 32 39 72 28 Incandescent 26 42 60 74 104 53 Total 37 61 92 111 176 91 Lighting kWh/Month Fluorescent 3.5 6.0 9.7 12.0 23.5 8.7 Incandescent 10.0 16.2 21.1 25.4 34.6 18.8 K Fluorescent 26K 27M 31K 32K 40K 32X Electric Lighting 13.6 22.3 30.8 37.3 59.1 27.4 Total kWh/Month 16.6 32.9 55.6 85.7 171.9 54.8 X Lighting 82K 68X 55K 44K 34K 50X Note: There may be soae bulb and wattage underestimation at upper income levels and lighting electricity use overestimation at low income levels. source: UbESS survey, Java 1988. In(en) = 2.3 + .641n(#Rooms) + .231n(Exp) + .041n(City) + .09Jn(Fam) - 1.96(Eleclight) t: (.21) (22.5) (10.7) (6.2) (3.3) (-52.5) R2 = .59 Significance = .0000 Number of Cases = 2379 where: en = household lighting fuel use expressed in MegaJoules/month. #Rooms - number of rooms in the household. Exp = household total expenditure per month. City = population of contiguous urban area. Fam - household size. Eleclight = I if electricity is used for lighting. - 0 if electricity is not used (kerosene is used for lighting). -83- Annex I 24. The coefficient associated with EMeclight indicates that households that light with electricity use only 1/6 of the energy used for lighting by households that light with kerosene (1- e1 =0.86 or 86% less energy). Transformed into original units, this implies a lighting substitution ratio of 1.38 kWh / liter of kerosene. The thermal replacement value of electricity that would be used for lighting by an average household using kerosene lighting would be roughly 2/5 of the energy content of the kerosene used for lighting. 97/ 25. At the time of the UHESS survey, urban consumers were facing electricity and kerosene prices that covered about 70% and 71% of the economic cost of supply, respectively (105 Rp/kwh average price and 150 Rp/kwh average cost; 200Rp/l average price and 290 Rp/t average cost). Hence consumers were responding to a fuel price ratio that roughly reflected the ratio of economic costs of supply. In addition, as the retail market in kerosene and electric lamps is largely uncontrolled, the financial prices of these appliances are more closely related to economic cost. Using estimates of the economic cost of electricity and kerosene supply to urban residential consumers of 150 Rp/kWh and 290 RpAL the economic fuel cost of supplying electricity to displace one liter of kerosene for lighting (207 Rp/1.38 kWh) is roughly 71% of the cost of supplying one liter of kerosene. The ratio above, based on consumer behavior, also indicates that with the average prices faced by households at the time of the UHESS survey (105 Rp/kWh for electricity and 203 Rp/I for kerosene), on average, lighting fuel expenditures by households using electricity for lighting were only 71% of lighting fuel expenditures in households using kerosene for lighting. 26. Much anecdotal evidence is given in the literature indicating that households switching from kerosene to electricity for lighting obtain much higher lighting levels. The UHESS data set yields an opportunity to quantify the increase. In cleaning the UHESS electricity data, a statistical procedure was used to estimate total kwh consumption from appliance ownership and usage, including incandescent and fluorescent bulbs. In addition to other correctiorx. for air conditioning and water pumping, this resulted in multiplying household responses of bulb wattages and hours used by 1.36 (See Annex VII). Though the total amount of kerosene used for lighting was estimated directly from survey data during data preparation, a breakdown by lamp type was not undertaken. As this is a necessary first step toward obtaining an estimate of lighting levels provided by kerosene, an analysis of kerosene consumption by kerosene lamp ownership is presented below. 27. On average, households which use kerosene for lighting own about 2 lamps which they use regularly, with only one in ten using more than 4. Most households using 3 or more kerosene lamps, use at least one Petromax, while those with only one or two lamps rarely use a Petromax. Petromax ownership is more common among upper income groups. Statistical analysis of the relation between lamp ownership and lighting-kerosene consumption corroborates the notion that a Petromax uses more kerosene than a chimney lamp which in turn uses more than an open wick lamp. Summary results of an OLS regression on households using kerosene for lighting and constraining the line through the origin are shown below: LZ/ hmd raempt value of eleicky estimoated ad 10.68 MG/Wh (piay -84 - Annex IV Kerosene Use (it/day) = .29#Petromax + .18#Chimney + .10#Open Wick t = (11.6) (13.4) (12.2) * .71; F - 391; #Observations - 468; An F test comparing this formulation with one in which the total number of kerosene lamps is the only dependent variable indicates that the inter-lamp differences are highly significant (>99%). Specifically, the analysis estimates that a Petromax lamp uses on average 3 deciliters of kerosene a day, a chimney lamp 2 deciliters and an open wick lamp 1 deciliter. 28. For each household lighting with kerosene, total lighting kerosene use was weighted by lamp holdings and specific consumption estimated above. These estimates of kerosene use by lamp type and electricity use in incandescent and fluorescent lamps (estimatd from the survey and summarized in Table A4.2) were multiplied by average efficacies of standard lamps displayed in the table below to result in an estimate of lighting levels (in lumen-hours) for each household. Finally, lighting levels were regressed against various factors to determine the effect of lighting fuel choice. Typical Lamp Efficacies (Wm-hr/kWh) M/ Fluorescent 50 Incandescent 10 Petromax (kerosene pressure lamp) 0.8 Semprong (glass shielded wick lamp) 0.2 Sentir (small open wic: lamp) 0.1 Results of a statistical (OLS) procedure to estimate amount of light are given below: ln(lm) =-1.92 +.841n(#Rooms)+.321n(Exp) +.061n(City) +.lOln(Fam) +2.31(Eleclight) t - -6.6) (21.0) (10.5) (6.9) (2.8) (4.4) R - .68 Significance - .0000 Number of Cases - 2376 where: Im - household lighting expressed in kiloLume"-hours/month. The coefficient associated with Eleclight indicates that, controlling for number of rooms and household income, a household using kerosene to light would enjoy, on average, nine times the amount of light if they were to use electricity (e23-1 9.07). O/ S: 4 Casony of Lamps for Dmesic ILg in Dveloping Cow , Wodd Bank Indusy and EeV Depme Woddng Paper, Enea Sein Paper No 6, Jww 198 -85 - Anne V ANNEX V Cooking Cost Comparisons: Kerosene and LPG Table A.1: Daily Cooking Cost Comparison of Kerosene and LPG Economic Financial --> Annual Discount/Interest Rate 10% 10X 20X 50% 100% 150% Daily Cooking Cost to Average Household (Rupiah) 2 Stoa Kerosene Systm Fuel 342 240 240 240 240 240 Stoves 21 21 25 38 59 76 Total 363 260 265 278 298 316 1 Surner PG Syptm with 1 hckup Kerose Stow Fuel 298 411 411 411 411 411 Equipment 38 38 46 70 108 140 LPG Bottles 20 20 37 83 143 189 Total 355 468 494 565 661 739 2 umrner LPG Sstem Fuel 298 411 411 411 411 411 Equipment 48 48 58 90 137 178 LPG Bottles 39 39 75 167 285 377 Total 385 498 544 667 833 966 2 Burner PG Sptm with Offfcial lottte Prices Fuel 298 411 411 411 411 411 Equipment 48 48 58 90 137 178 LPG Bottles 29 29 55 122 209 276 Total 375 488 524 623 757 86S 2 Durner LPG System with 6k4 Nettles FPel 298 411 411 411 411 411 Equipment 48 48 58 90 137 178 LPG Bottles 16 16 30 67 114 151 Total 362 475 499 567 662 739 Subsidized 2 Burner LPG Systm with 6kg Bottles Fuel 298 411 411 411 411 411 Equipment 48 14 17 26 39 51 LPG Bottles 16 16 30 67 114 151 Total 362 440 457 503 564 612 Averages: family size 4.8; household expenditures 156,000 Rp/lmo; daily cooking fuel use 1.18 I kerosene a 0.7 kg LPG. Unit Cost Asswvtions: Fuel costs: Kerosene 290 Rp/l (economic) 203 Rp/l (financial), LPG 428 Rp/kg (economic) 590 Rp/kg (financial); 1 kerosene stove 15,000 Rp; 11 kg LPG Bottle 75,000 Rp (market) 55,000 Rp (official); 6 kg LPG Bottle 30,000 Rp; I LPG burner 30,000 rupiah; I LPG regulator n 10,000. Subsidized LPG package sells for 80,000 Rp (subsidy a 50,000 Rp). 5 year life for kerosene stoves and LPG equipment. Each LPG system includes one regulator, only. Economic and financial costs of LPG bottles reflect the dsily interest foregone by the average household due to investment in LPG bottles (interest rates above 20% reflect severe cash flow constraints rather than actual interest foregone). - 86 - Annex VI ANNEX VI Cooldng Implements and Practices 1. In most households, the food is cooked by the principal woman of the household. Cooking by servants is rare (less than 10% of householdr), except in the uppermost income group. In about one household in five, the woman who cooks also has a paying job outside the household. Two hot meals are prepared each day in most households, although in small households, espeially if they are poor, more often than not only one meal is cooked each day (eating only one hot meal a day is frowned on by the more educated householders). Separate kitchens are common, even among the poor. Table A6.1: Cooking Practices and Income - Java 1988 Income Group t000'Rupiah per Household per Month) 0-75 -120 -185 -295 -Top Alt RH Cook Principal Woman 83K 84K 83K 74K 64K 78X Servant oX 1X 4X 8K 25X 5X Other Fam. Member 15X 15X 12X 17K 10K 15X Kitchen Separate 56X 62K 65K 76K 77X 63X Multipurpose 37K 33K 29K 23K 19K 31X Open 7K 5K AK 2K 4 5K Meats cooked per day 1.6 1.7 1.8 1.9 1.9 1.7 Liters of Rice per day 1.3 1.6 1.7 1.8 1.9 1.6 Rice Cooking Technique Steam 64K 77 80K 85K 81K 74X KBoil 34K 22K 18X 14K 17K 23v Household Size 3.5 4.8 5.1 5.7 6.1 4.7 Source: UHESS Survey, Java 1988. 2. Almost all cooking is done on stoves: ovens are rare and grills are only used occasionally for special foods such as "sate'. Overall, four features of cooking equipment are linked to better fuel economy: higher stove efficiency, lower stove power (maximum used), greater turn- down possibilities and larger pan size, in the case of flat-bottomed pans. L2/ The actual influence of these factors depends not only on the particular practices of the cook, but also on the nature of cooking operation being performed. The major cooking operations are: 1) Heating water for drinking; 2) Cooking rice; 3) Cooking vegetables; 4) Frying, and; 5) Reheating. L9/ The effect of pan siwe isfarmo'a coap dnti statnt wduld im$y if cooUng is done w pots d pans of diffeW shapes and mamaL - 87 - Annex VI 3. In Java, water must be boiled to make it safe to drink: that this enables tea to be made is a fortunate coincidence. More than 97% of households boil water for drinking, and an average of 5 liters are boiled per day per household. Per capita water heating is not linked to income, and rises only slightly with city size. 4. Bringing water to a boil is the simplest cooking operation, and the easiest to simulate in the laboratory. The standard means of measuring stove efficienc' in a laboratory is with a water boiling test. Given the correct pan size and power setting, laboratory efficiency is an adequate measure of the fuel-economy of bringing water to a boil. Generally, less fuel is required if larger pans and/or lower power settings are used. The power setting is less crucial than in many other cooking operations, however. Perhaps the most similar other cooking operation is reheating food, which also stops once the desired temperuture has been reached. 5. Rice is the staple food, also prepared by some 97% of the households surveyed. On average, 1.6 liters of rice are cooked per household per day, The principal technique for rice preparation, used by about three quarters of the households surveyed, involves first cooking the rice in boiling water for roughly 15 minutes, and then transferring to a rice steamer, where it is steamed until cooked. Alternatively, the rice is boiled (or simmered) until cooked. Boiled rice is generally considered to have a poor texture, but requires less effort and less fuel if small quantities are being cooked. Not surprisingly, boiling rice is more common among smaller, low income households. 6. Beyond a certain point, a higher power setting has little impact on the speed of rice steamiing (or simmering), and simply leads to higher levels of fuel use. Tne additional heat converts more water to steam, but does not raise the temperature in the pan. The ability to make power adjustments depends on the fuel and even the particular stove. Even when it is possible, cooks are not inclined to take full advantage of turn-down possibilities unless they accustomed to doing so. Some are unaware, or do not believe, that rice will steam virtually as fast at lower power. Others may find it too much trouble, or simply feel that it is not the proper way to cook, and will likely lead to problems of some sort or other. Among other implications, this means that efficiency is not a suitable measure of fuel-economy during rice preparation. 7. Frying is almost invariably done in a wok. On average, the diameter across the top of the wok measures 30 centimeters. Vegetable cooking techniques are more varied. As illustrated in Table A6.2, in most households a flat bottomed pan is the principal vegetable cooking device, but a significant minority use a wok more often. The consumption of both fried foods and vegetables is enough of a luxury that the rich eat considerably more of these foods than the poor. 8. Frying and vegetable cooking can also use considerably more fuel if the (high) power is not reduced once the necessary temperature has been reached. To some extent, however, frying imposes power reduction, since overheated oil smokes rather than simply evaporating away. In addition to saving fuel, this makes simple turn-down mechanisms particularly attractive for frying. 'Though a constant but low power will also prevent burning, heat-up operations then become excessively long 1Q/ ZI/ Te dice betwem dte pa nd e heat sown can be adjuted in je of po adjusns, aan at the cost of csadmbe ful savings. -88- Annex VI 9. As part of an attempt to better Table A6.2: Principal Cooking understand the fuel use characteristics of (kerosene) Utensils cooking in Indonesia, controlled cooking tests were undertaken in a laboratory (see Stove Program Proposal: .ater Boling Device Appendix II and Lemigas Stove Testing Report). A Type X Use Avg Diameter household cook was asked to prepare a typical meal of (cm) rice, vegetables and chicken on a number of different None 2X NA kerosene stoves. From the results of these controlled Flat Pan 57X 23 cooking tests, along with the water boiling tests, a model Kettle 31X 21 of household daily fuel use for cooking and water Electric Kettle 0X 22 heating was developed. The model includes not only Other 1X 21 cooking the principal meal, but heating water and reheating food in the evening. At typical values of Rice Cooking Device efficiency and power for a kerosene stove, the model TM* x Use Avg Diameter suggests that a fifteen percent increase in efficiency has (cm) roughly the same effect on fuel use as a fifteen percent None 3X NA decrease in the stove's maximum power: both lower fuel Flat Pan 24X 22 use by about five percent. Similar savings can be Rice Steamer 6X 25 achieved by turning the power down by about 30% Electric Cooker 2X 23 Other 7X 21 during rice steaming. (As the amount of kerosene consumed during controlled cooking and water boiling Veetate Cooking Device tests appears not to be a simple function of useful Vegetable _________Device energy demand and efficiency, the same holds, afortioi, Type X Use Avg Diameter for cooking fuel use generally, including wood and LPG.) (cm) Nome 4X NA 10. The success of technical measures to Flat Pan 78X 22 Wok 16X 29 improve fuel economy may depend on how well they other 2x 23 conform to the cooking vessels. The most common traditional rice steamer (the "dandang") has a curved Source: UHESS Survey Java 1988. bottom which does not aiways accommodate pan supports designed for flat pans. The wok has an even larger curved bottom, and may also require special design adjustments. Unfortunately, the effect of these curved bottoms on fuel economy is not known, although it is likely that they do influence efficiency. 11. The following sub-sections examine each of the major cooking fuels in turn. Piped gas is not included, but has many of same cooking features as LPG. Electricity is not included because it is only used for cooking by 2% of the households, and even then is usually restricted to specialized uses. Electric cooking may become important in the future, however, and deserves careful monitoring. Kerosene 12. The bulk of urban households in Java (75%) use kerosene for cooking. As described in Chapter II, the kerosene distribution system is extensive: kerosene can be purchased along with other daily necessities at local stores or, in many areas, purchased from traveling salesmen. The principal woman of the household generally determines what kind of stove to by, when to buy it, and makes the purchase (see UHESS Summary Table 14). Kerosene stoves are ubiquitous in the -89 Anex VI local markets frequented regularly by such women. Stove repairs can be done locally, if necessary. In short, the support system enables most homemakers to obtain all the materials needed to use kerosene without disrupting their daily routine (or, perhaps equally important, that of their spouse). 13. Most of the kerosene stoves used are wick stoves of simple construction, and almost all are produced in Indonesia. Pressurized stoves are used for commercial purposes, but are used by less than 1% of households. Single wick stoves are quite popular among upper income households (18% of households with a monthly income over 295,000 Rupiah use single-wick stoves), but more than nine in ten kerosene using households use multi-wick stoves (see UHESS Summary Table 7). On average, households which cook with kerosene own 1.6 stoves. Stoves are produced by both artisans and factories, with artisanal stoves still in the majority (see Lemigas Stove Report). 14. There is considerable variation in both the power and efficiency of Indonesian wick stoves. (Most stoves tested for this study were between 30 and 50% efficient, and had a maximum power of between one and two kW.) Virtually all stoves have some means of adjusting the power, but the flames are difficult to see, and the adjustment mechanisms are often awkward. 21/ Except when the heat is clearly excessive (e.g. causing oil to burn) or the flame is no longer blue, most cooks probably do not use the power adjustment. Fuel efficiency is rarely a criterion for stove choice (Sociological Dimensions of Household Stove Choice in Chapter IV), probably because it cannot be determined by inspecting a stove or by reputation. Higher power is viewed as advantageous because it performs many cooking operations faster. However, in the absence of an adjustment mechanism, high power uses more kerosene and may sometimes prevent the cook from carrying out other tasks simultaneously. Liquified Petroleum Gas 15. LPG can be obtained in most of urban Java, but in some areas only with difficulty. Even in areas where LPG use is common, the support system is not so well suited to the every- day routines of most Indonesian women. Neither LPG stoves nor LPG bottles are readily available in most neighborhood markets, even in big cities. The bottles are, in any case, large and cumbersome, and less suited to a cash-and-carry system. LPG delivery men are more anonymous than the local kerosene peddlers, and are likely to be considered less trustworthy. It is also more difficult to verify LPG quantities, and although LPG delivery-men are officially required to carry scales, weighing the bottles upon purchase is generally discouraged. While these problems may not discourage upper-income women, who rely on the neighborhood markets less in any case, they are likely to inhibit middle and lower income women. 16. The LPG stoves marketed in Indonesia are generaly more efficient than the kerosene stoves, have higher maximum power, and a simpler power adjustment :sechanism. These latter two features help make LPG a more flexible cooking fuel for the experienced user. In the controlled cooking test, a typical LPG stove used roughly the same amount of fuel (heating value) to cook the meal as some of the more fuel-efficient kerosene stoves. A low power LPG stove used &/ Adjustment is compikated by the fact that the reladon between power and the position of the adjustor depends on how farthe wick has burnt down Somne housholds only use the adjustor to keep the flame constant over the week orso between wick-autngs. - 90- Annex-VI less fuel than any other stove. Combined with water heating and reheating food, the high efficiency of even the high power LPG stove would be expected to result in lower overall fuel use. Wood 17. Despite its regular use, especially in the smaller urban areas, fuelwood in Java has the reputation of a rural cooking fuel. Indeed, wood use is best adapted to a rural lifestyle, where wood can be collected as part of the householders' fieldwork, and the large kitchens are well suited to smoky wood stoves. In non-electrified households, the dim firelight allows chores and conversation to go on in the kitchen without the aid of a kerosene lamp. Fuelwood is not prevalent in the urban markets of Java, and is rarely available in the markets of the largest cities. Instead most urban fuelwood is collected, and almost a third of the households purchasing wood have it all delivered, quite possibly by a neighbor (see UHESS Summary Table 13). Indeed, almost half of the urban fuelwood users have land from which fuelwood can be collected, and are, in this sense, less urban than other urban dwellers, only 2% of which have such land. 18. Cooking with fuelwood is notoriously inefficient (5-20%). While no wood-stove tests were undertaken for this study, research on wood-stoves in Indonesia has generally supported this supposition. Perhaps even more than for other fuels, however, efficiency is difficult to define independently of user habits, and may be even less useful in estimating fuel use requirements. There is a wide range of power a wood-stove can achieve, and the measured efficiency depends on which power level the tests are conducted at. Altering the power is either troublesome or slow. When necessary, users are more likely to alter the distance between the pan and the heat (either by moving the pan or moving the fuel) rather than attempting to reduce the power rapidly. 22/ Generally, one can expect wood use to vary considerably depending on how willing the cook is to try to save fueL While virtually every fuelwood cook has probably been forced to stretch out a small quantity of fuelwood when an unexpected shortfall occurred, few are likely to put maximum effort into day-to-day wood conservation. ZW/ Power being defind as the eneiy use per ui of ime. - 91- Annex VII ANNEX VII Urban Residential Energy Projectlons 1. This Annex presents the econometric derivation of the model of urban residential fuel use employed for projecting demand patterns. The equations derived below model relationships in the survey data between fuel use and income, family size, urban area size, and other parameters, as appropriate. The model of residential energy use was constructed in four parts: (a) combustible fuel use for cooking; (b) electricity demand; (c) k -rosene use for lighting, and (d) LPG use in separate water heaters. As over half of energy consumed in urban households on Java (in LKE terms) is for cooking and boiling water, the transition between fuels for cooking was analyzed in most detaiL Choice of cooking fuel and amounts used were found to be strongly dependent on income, family size, and urban area size. Urban residential electricity use was determined to be largely a function of income, source of supply (PLN or other), and age of official connection. Very few households with an electrical connection use kerosene for lighting. Because of this, estimates of kerosene use for lighting were calculated only for unelectrified urban households. Water heating with a separate device is limited to the higher income households and is almost exclusively done with LPG. Results of this four part model are summed to give overall urban residential fuel use projections. L3/ Fuel Use ProiectionAssumptions 2. The income distribution of the households in the sample is shown in Figure A7.1 (logarithm of monthly household expenditures). It is evident from this histogram that the log of household income approximates a normal distribution. When using logarithmic scales, and income grows with the same rate for each household, the whole income distribution curve shifts but retains its original shape. This observation forms the basis of the key exogenous variable in the projection model: per capita income growth. As population and income grow, the entire distribution moves, allowing the number and the average fuel use of households in each category to be estimated by fitted curves modelling the income variation in demand for each fuel. 3. The distribution of the logarithm of monthly household expenditures from the LUHESS survey (Figure A7.1) was found to be distributed according to a normal curve with mean 11.673 (standard error 0.015) and standard deviation of 0.737. When used in the model, income steps of 0.2 (log-scale) were used, resulting in 26 income categories. The share of the total number of households in each of these income categories becomes: S(expe) = 0.10826 * exp(-(ln(expe)-11.673)^2/1.0863) W Chwdi mamn usedformmgandsate mangandseldomforodtecoodgactes& Becse of ted mkively small sham (kss than 2% of he eneV used for cookng) and its epected fwse decline c/ucoal is not cluded in efue wusepecons. Howver, even togh ekecicity used in eleic e cooks and ketta compnsesa n abot 1% of cooldng eawy demand, as te use of these applia is gvWng mpidly, ekecicity is iuded in the cooking fi- anws and pmjectiam -92- Ana Figue A7.1: Distribution of Households by Exenditure Category Histogram: LN of Household Nonthly Expenditures. Households (Cases) Midpoint 8 9.50* 9 9.70 : 13 9.90 *: 29 10.10 :* 43 10. 30 63 10.50 . 110 10.70 : 162 10.90 ________ :_ 195 11.10 __: 241 11.30 :_ 247 11.50 255 11.70 - _ 287 11.90 _ 244 12.10 _ : _ 162 12.30 ________________ - 131 12.50 __ __ __ __ - 97 12.70 62 12.90 : 31 13.10 17 13.30 17 13.50 *: 10 13.70 7 13.90* 0 14.10 0 14.30 1 14.50 0 14.70 2 14.90 I....,....I....,....I....,....I....4....1....+.....1 0 80 160 240 320 400 Histogram Frequency Valid Cases 2443 NMssing Cases 0 SPSS/PC+ ISurce: UHESS Survey, Java 1968. (with expe = monthly household expenditures in Rupiah) 4. For projecting urban residential fuel use through the year 2000, an average urban population growth rate of 4.1% per year (approximately 2% above the population growth for Indonesia as a whole) and an average yearly income growth rate of 2% per capita were assumed. Urban population data were derived from the NUDS study. 5. The model presented below is based on patterns of urban residential energy use on Java. Projections of residential fuel use for all of urban Indonesia have been developed by extending income related patterns from the UHESS survey of urban Java to urban Indonesia as a whole. The rationale for using urban Java data to estimate fuel consumption in urban Indonesia is the stridng similarity between means of per household fuel use by urban households on Java and those of urban Indonesia as a whole as shown in Table 3.1. -93 - Annex YU Derivation of a Cooking Fuel Demand Model 6. In Annex IV and Chapter m, evidence was presented showing the "Basic Eneergy Demand" approach to be unsuitable measure for estimating amounts of fuel needed when a household switches from one cooking fuel to another. As an alternative, substitution ratios, "Liters of Kerosene Equivalent" (LKE), have been derived from survey data and were used to calculate, from physical quantities, the share of each fuel used for cooking by each household. Shares of cooking demand provided by each fuel, averaged over households in each of 20 mincome" categories and each of 13 urban area sizes, are displayed in Figures 3.4 and 3.5. As income and population growth are the key exogenous variables used in the projection model, equations describing the cooking fuel shares depicted in Figure 3A are at the heart of the cooking fuel projection modeL Figure A7.2: Flow Chart: Cooking Fuel Demand Model Inputs ( ), Calculated Results (- ). Population Population Income distribution Income size in 1988 growth in 1988 roth Population size income distrbution in year T fn year T Average Ifuel price 2 Total conswptfon Share wood Share LPG Share 2 ~~~~per househotd L J Etectrfcity I r ' |Sha~~~I re 0Krsn Totat conuniptiron of a cooking fuets fn year T|- 7. As shown in Figure 3.4, the average quantity of fuel consumed per household (in LKE) for cooking rises with income as does family size. Tbis relationship between income and total per household cooking fuel use is incorporated into the cooking fuel model by fitting a line to the natural logarithm of average daily cooking fuel use per household (in LKE) and multiplying the result by calculated shares from the equations below. 8. Besides income, prices influence fuel use. By its very nature, cross-sectional data with little price variation across the sample is of limited use in the derivation of price elasticities -94- Anne Vl of demand. 74/ Nonetheless, analysis of the small kerosene price variation across the UHESS sample yielded a kerosene own price elasticity of -0.42 (with a standard error of 0.23). However, since the model computes the share of cooking energy provided by kerosene as a residual, this elasticity could not be incorporated directly. In addition, this elasticity, derived from 1988 data should be treated with caution. In 1988, there were no overall changes in the kerosene price. In outer areas of towns good availability of wood goes hand in hand with higher kerosene prices. Kerosene consumption of households in such an outer area is lower than average, but this is as much an effect of wood availability as of high kerosene prices. 9. In the cooking fuel model, the following elasticities were used, all given when the share of the fuel is 50%: LPG own price elasticity -0.4 LPG demand with respect to kerosene price +0.4 Wood demand with respect to kerosene price +0.3 Total cooking energy demand with respect to a weighted average fuel price 0.1 As will be shown below, with these elasticities the model simulates an implicit kerosene own price elasticity of -0.2. 10. The cooking fuel model is constructed by fitting logistic curves to the average shares of cooking fuel needs met by each fuel (in LKE) for households in each of 20 expenditure categories. Price elasticities in logistic curves are specified at a fuel share of 50%. If "al is the price elasticity at a share of 50%, the logit has the form: L - constant + b* In (expe)+ 2 * a * In (Price). The share then becomes: Share = 1 / (1 + exp(-L)). The relative change in share after a change in price is then: (dS/S) / (dP/P) - 2 * a * (1 - S) = a. 11. Access to and availability of fuels also present a strong influence on household fuel choice and use. However, robust measures of access and availability are much more difficult to derive than income and price. The UHESS survey sample frame was stratified on urban area size as a measure that incorporates many of the characteristics of variation in access to and availability of household fuels, especially with regard to availability of traditional vs. conventional fuels. Figure 3.5 illustrates the monotonic fall of woodfuel use in larger urban areas which implies that woodfuels are harder to get and/or less preferred in larger urban areas. The cooking fuel projection model is based on variation in fuel use patterns by income. However, as income and city size are correlated (see Table A7.1), the model equations are not purely based on an income effect, but a Z &Econcewic t;me-seies waaysis on SUSENAS data ftm 1981, 1984, and 1987 also pvved to be siiufifant *ih apect to deidon of priw eksticie -95- Annex VII "city size" effect as well TableA7.1: Correlation of Income, Family Size, and Urban Though these patterns are Area Size very useful for informing Corretations: LNCity Size) LN(HN Expend) LN(Famity Size) effective decisions on the _ _ _ _ _ _ _ _ _ _ importance of fuel distribution LN(City Size) 1.0000 .3732** .0829*- policies, theywere not directly LN(HH Expend) .3732** 1.0000 .43700 incorporated into the cooking fuel projection model because N of cases: 2443 1-tatted Signif: * - .01 * - .001 the shares of urban population $ource: UHESS Survey, Java 1988. living in large, medium, and small cities is not expected to change significantly in the near future. 12. In constructing the cooking fuel use model, the shares of each fuel used in 20 income categories were first transformed using a logit transformation. This technique is commonly used to describe growth phenomena like the introduction of a new technology. Households switching from wood to kerosene stoves is an example of substitution of one technology for another, which can be described conveniently by an S-curve. Figure A73 illustrates the S-shaped logistic curve which approximates exponential growth (i.e. having a constant growth rate). After a logit transformation, a fuel providing half the cooking energy in any category will have value zero while fuels providing 10% and 90%, respectively, obtain values of -1 and 1. 13. After transformation, the illustrations of variation in cooking fuel shares by income and city size from Figures 3.4 and 3.5 become "straightened out" and are displayed in Figures A7.4 Figure A7.3: Logistic Curve LOGISTIC CURVE LOGIO (Share/(1-Share)) SHARE (%_ 100% _ 80% -. . t ; . _ . . _ j 70% . .. . . . ..... - 60% . ....... .......... ..... ..... ...........-.-.... 40% _ . t | + 7Z . 4 + 4 4 30% X ' -- ¶ 20% . 10% _. ++++..... + - 2 -1L.5 -1 -O0.5 0 0.5 1 1.5 2 LOG(Share/ (1- Snare)) -96 - Anne VII EjguA7: Logit Transformation of Cooking Fuel Shares by Household Income Cooking Fuel Substitution by Income LOGIO (SHARE/(1- SHARE)) 2.00 |- WWOOD + KEhDSENE -* LPG -ELECTRICITYI 1.00 +......,.+ 0.00 2.~~~~~ ~ ~~~~~ 0 ' -1.00 -c -2.00- -3.00 4 4.2 4.4 4.6 4.8 5 5.2 5.4 5.6 5.8 6 6.2 LOG10 OF HOUSEHOLD EXPENDITURES Souce: UHESS 1988. and A7.5. This is a convenient way of representing the shares of the different fuels because logistic S-curves become straight lines. In Figure A7.4, for urban households with expenditures lower than about Rp 100,000 per month, the share of kerosene use rises with income approximately according to a straight line. This means that for different income levels a certain percentage increase in income leads to a fixed percentage increase in the share of kerosene used for cooking. Kerosene is, in practice, the only substitute for wood for these low income households. Every change in kerosene consumption is therefore mirrored in an equal change in biomass use. 14. For households with monthly expenditures higher than about Rp 100,000, LPG becomes a possible fuel. The actual shares of wood, W and electricity for cooking were fitted (by weighted least squares method) to three different logistic curves (straight lines after the data transformation), with the kerosene share, shown as a dotted line, as a residuaL Kerosene reaches its maximum share of about 90% i1 the income category of Rp 200,000 per month. Wood and LPG have an equal share of about 5% in this income category. For the higher incomes LPG becomes relatively more important. Electricity only accounts for a minor share. Even for the households with expenditures of Rp 1 million per month, the share of cooking demand met by electricity is only about 9%. These three fitted logistic curves form the core of the cooking fuel projection modeL 15. Wood and other Biofuels: In Figure A7.4, the logit transformation of the share of cooking fuel consumption met by wood is displayed. The logistic curve which fits the wood shares in this graph is calculated via "weighted least squarese (WILS): -97- .nexV Figure A7.5: Logit Transformation of Cooking Fuel Shares by Urban area Size Cooking Fuel Substitution by Urban Area Size LOG10 (SHARE/(1-SHARE)) 2.00 0.00 0.00 ~ ~ ~ ~ ~ -1.00 0 -2.00- -3.00 3.5 4 4.5 5 5.5 6 6.5 7 LOG10 OF URBAN AREA SIZE Source: UESS 1988.1 L = 23.744 - 2.194 ln(expe) Std Dev = (1.656) (0.147) t (14.3) (-14.9) R2 = 0.995 df = 18 With an elasticity of wood demand at a wood share of 50% of ea", and a price as a ratio of the current and the 1988 for kerosene of Pk, the logit which was used in the model was the following: L - 23.744 - 2.194 In(expe) + 2 * a * In (Pk) 16. Elriciq: Cooking with electricity, using rice cookers and electric kettles takes only a small share of cooking fuel consumption as can be seen in Figure A7.4. The logit which was fitted to these shares and also used in the model: L = -41.516 + 2.839 ln(expe) Std Dev = (2.236) (0.174) t = (-18.6) (16.3) R = 0.995 df - 18 17. LPG for Cooking: The logit fit to the LPG share curve in Figure A7.4 was the following: L = -27.576 + 2.038 In (expe) Std Dev = (3.732) (0.300) t = (-7.4) (6.8) R = 0.955 df = 18 -98- Annex VI 18. Because of the very rapid past growth, this is epected to give under estimated results when used for projections. Therefore a technical growth parameter was introduced. Tbis is an extra growth over and above the growth caused by population and income growth. In the model the income elasticity is allowed ta increase in time with a rate which is an input. It should be given as the ratio of the income elasticities in the year 2000, with and without technical growth. To be able to get an impression of the effects of the chosen ratio, the model calculates the share of households in the income category of Rp 300,000 in 2000 and using LPG for cooking. For example: If a ratio of 1.2 is inserted, the model calculates that the share of households will be 0.23 instead of 0.16. Including also the price effects via the own price elasticity AL and cross price elasticity for kerosene price Ak, the logit becomes: L 4.108 - EL(ln(co) - ln(expe)) + 2AL ln(PL) + 2Ak ln(Pk) with EL = (1 - Share hh using LPG) * Income Elasticity - 2.0384*(1 + (year - 1988)*((LR-1)/12) wbh LR = ratio of income elasticities in 2000 with and without technical growth. "o" is the cut- ol expenditure below which there is practically no LPG use. In 1988 this is Rp 100,000 per month. It is a variable that models the effect of lowering the barrier of high initial equipment costs. The constant of 4.108 was found by normalizing so that the share of LPG in 1988 becomes 5.74%. 19. Kerosene for Cooking: For three fuels - wood, LPG, and electricity-, the logistic curves seem to give good representations of the shares at different incomes. Systematic deviations from the fitted straight lines cannot be found. This implies that for small and large (near 1) shares, substitution takes place according to an exponental growth curve. The share of cooking energy needs, met by kerosene can simply be found by subtracting the shares of the other three fuels from 100%. Kerosene for Lighting 20. Because virtually all electrified households in the UHESS survey reported lighting with electricity only, demand for kerosene as a lighting fuel is estimated only for non-electrified households. In similar fashion to the derivation of the cook!ng fuels model separate curves were fitted to the logit transformed share of households using kerosene for lighting and to the average amount of kerosene for lighting per using household in each income category. This section of the model was constrained to fit the total share of households using kerosene for lightng to the share of non-electrified households resulting from the electricity projections presented below. The own price elasticity of kerosene for lightng in households usng kerosene for lighting was assumed to be -0.4, consistent with results of analysis on total kerosene use presented earlier. 21. Almost all electrified households use electric lighting instead of kerosene lamps. WLS on the log of the share minus one over the share of households in each income category resulted in the following. - 99 - Annex Vl L - 18.11 - 1.719 ln(expe) Std Dv - (0.61) (0.057) t (293) (-303) 0.986 df = 524 The amount of kerosene consumption for lighting per using household was also estimated as a function of income as below with K = daily kerosene consumption in liters per using household. In (K) = -5.688 + 0.452 ln(expe) Std Dev = (0.627) (0.058) t = (-9.7) (7.9) R' = 0.986 df = 524 LPG for Water Heating 22. Most households prepare their hot water, including water for bathing purposes, on their kitchen stove. In these cases the UHESS survey yields no information on actual amounts used for cooking and water heating. Hence, fuel used for heating water on kitchen stoves is treated as a cooking fuel use in the modeL The survey showed that a number of households use a separate device for water heating almost always an LPG water heater. This separate demand for LPG was modelled somewhat differently from the other household fuel demands. The average per household consumption as a function of price and income category were established. Summing over income categories the product of average household consumption and number of households yields the total LPG consumption for water heating. As in the cooking fuel model, an LPG own price elasticity of -0.4 is assumed. The demand found in kg per day was: ln(D) = -22.53 + 1.417 In(expe) Std Dev = (0.57) (0.615) t = (40) (2.3) The equation used in the model is the following ln(D) = -22.53 + 1.417 ln(expe) + 2 * AL * ln(PL) with 2 * AL the own price elasticity of LPG and PL the price relative to the 1988 price. Electricity Demand Model 23. Already a large share of urban households have access to electricity from PLN. A quarter of them however, obtain their electricity not directly from PLN, but via their neighbors (see Figure 2.4). In the official statistics, these households are considered 'not yet electrified". With the aim of assessing possible effects on future electricity demand of neglecting indirectly connected households, an electricity demand model was built from the UHESS data. Derivation of equations modeling relations in the survey data between tctal electricity demand and income, number of years connected to the grid, and appliance holdings are detailed below. Briefly, econometnic - 100 - Annex VII analysis resulted in separate equations modelling the demand of households with direct and indirect connections. For the former, electricity consumption was found to depend on household income and on the number of years the household has been connected to the grid. For indirectly connected households, consumption was found to depend only on income. Electricity use by each appliance was also estimated to gain insight into the current and future significanice of each end-use as incomes grow. Income Distribution 24. Households were classified into 17 income categories (steps of 0.3 on a log-scale) for the electricity projection model. The share of the total number of households in each of these income categories becomes (with expe monthly expenditures in Rupiah): S(expe) = 0.1624 * exp(-(In(expe)-11.673)^2/1.0863) (1) Shares of Households HavingElcticty 25. Total Share of Electrified Households: According to the SUSENAS surveys of 1981, 1984, and 1987, the share of households having access to electricity can be found by executing an Ordinary Least Squares (OLS) on the log of the shares over 1 minus the share s for these three years. The result is the following equation (in brackets the standard errors): L = -15.868 + 0.1944 * (Year - 1900) (2) Std Dev = (0.013) (0.0030) The SUSENAS electrification shares are expected to be somewhat underestimated. Therefore in the demand model the yearly growth rate has been used, and the shares of households according to UHESS having access to PLN electricity, either direct or indirect. This share is 0.836, resulting in the following logit for the total share of electrified households, which is used in the model: L = -1.629 + 0.1944 * (Year - 1988) (3) 26. For each income category the share was found by applying a Weighed Least Squares (WLS) to the UHESS data, divided into 20 income categories. In logit form it results in: L = -19.05 + 1.824 * ln(expe) (4) Std Dev - (0.95) (0.085) t - (-20) (21) R2 = 0.985 df = 18 27. Combining equations (3) and (4) results in the logit for the share of households in each income category: L = 19.15 + 1.824 * ln(expe) + 0.1944 * (year-1988) (5) Normalizing this by making the sum of the shares equal to the share of 0.8365 resulted in the value for the constant -19.15. - 101 - Annex VII 28. Share of PLN Direct Connections: Direct connections to PLN are expeced to grow with an average growth rate g = 0.043. It was found convenient to enter the growth rate of PLN's future electrification as an exogenous variable. When this growth rate is used in a logit it has to be multiplied with one over one minus the PLN share in 1988. Using the PLN share in 1988 from UHESS (0.618) resulted in the following logit for the share of directly connected households: L = OA81 + 2.618 * g * (Year- 1988) (6) 29. Not only was the total share calculated, but also the share for each income category. A WLS applied on 20 different income categories resulted in the following logit for the relation between share of households directly electrified and income: L = -19.25 + 1.709 * (expe) (7) Std Dev = (0.89) (0.077) t = (-21.7) (22.2) R2 - 0.969 df = 18 30. When equations (6) and (7) are combined, the share of households in a certain income category and having a direct PLN connection is in a logit form: L = -19.318 + 1.709 * ln(expe) + 2.618 * g * (year-1988) (8) Whereby the constant 19.318 was found by normalizing the sum of the shares for all the categories such that the total becomes 0.618, the share of directly connected households in 1988. 31. Share of Indirect PLN Connections: Indirectly connected households form a rest group. Their share can be calculated by subtracting the share of directly connected households from the total share. Because an artificial upper limit of 95% has been set as a maximum for the total share of electrified households the difference of two shares results in small negative shares for high income categories near the year 2000. In these cases, shares of indirect connections are set to zero. Consumption per Connection 32. Consumption for PLN Direct Connections: Consumption per connection was found by applying OLS to the logarithm of electricity consumption in kWh/mo, using the variables: expenditures, city size, family size and duration of the connection in years (time) for directly connected households only: Ln(el) = -3.08 + 0.442 * ln(expe) + 0.0846 * ln(city) + (9) Std Dev = (0.30) (0.027) (0.0078) t = (-10.3) (16.1) (10.8) + 0.206 * ln(fam) + 0.184 * ln(time) (0.037) (0.021) (5.5) (8.9) R = 0.38 df = 1543 - 102. AxnnexV In the model no change of city size or family size has been incorporated. The consumption per connection which was actually used is the following. CCP - exp(-1.343 + 0.442 * ln(expe) + 0.184 * In(time)) (10) Again the constant has been found by normalizin& in this case such that the weighed average is equal to the consumption per connection as was derived from the UHESS 1988 data (consumption = 77.8 kWh/mo). 33. Consumption per PLN Indirect Connection: Comparable to equation (8) the following consumption level was found via OIS for the indirect connections: Ln(el) = -3.61 + 0.427 * In(expe) + 0.074 * ln(city) +0.164 * In(fam) (11) Std Dev - (0.68) (0.063) (0.019) (0.063) t - (-5.3) (6.8) (4.0) (2.6) R = 0.17 df =478 When the duration of the connection was included, the value for the parameter did not differ signifi'cantly from 0 (t = 0.92). For the model the consumption per connection was assumed to depend only on the expenditure level CCNP = exp(-2.159 + 0.427 * In(expe)) (12) The constant was calculated based on an average consumption per connection of 15.49 kWh per month in 1988. Appliances 34. Methodologr: The demand for electricity was allocated over the main groups of appliances. In an "othere category, all the minor and new appliances are grouped. Again the relations describing the households user were found by econometric analysis of the UHESS data. However, for some appliances large growth rates can be expected. This makes them difficult to be accurately described by relations which are based on the data for 1988 only. When compared with the model for household fuel demand and the modelling of the aggregate electricity demand, the appliance shares are certainly less trustworthy. But besides the expected larger uncertainties this part of the model gives some limited insight into possible paths for future electricity demand from the different appliances. 35. Logistic curves were estimated for the share of households using a certain appliance. This gives the share as a function of expenditure leveL Except for lighting this was calculated as the share of the total number of households, not only the electrified ones. In case of incandescent lighting no shares were calculated separately, because these were assumed to be almost equal to the electrification share. The effect of the small number of electrified households, not using incandescent lighting was incorporated by using a somewhat lower average use per connection. For fluorescent lighting the shares were calculated as a percentage of electrified households. For some appliances like refrigerators and color televisions a technical growth rate was added to the equation descrnbing the shares. .103- nnex VII 36. To find the yearly urban residential electricity demand for each appliance type, the following method was used. The above mentioned shares were calcuated for each year and then multiplied by the average consumption per household owning the appliance Also this average consumption is a function of income. Again, the income distribution was used as descrbed above. After multiplying the average use per household with the caculated number of households in each income category, and summing them, the total gives the yearly consumption for each appliance. 37. kIring Econometric analysis resulted in the following logit for the share of households (of the total) using an electric iron: L = -18.62 + 1.57 * In(epe) Std Dev = (0.95) (0.08) t = (-20) (19.4) R' 8 0.955 df = 18 For the logarithm of the use per owner was found: ln(el) 0.88 + 0.37 * ln(expe) Std Dev = (0.47) (0.04) t = (1.9) (9.5) R = 0.076 df - 1089 Because the logarithm of an average is in general somewhat different than the average of logarithms, the latter equation was normalized by changing the constant such that the average becomes 8.35 kWh/month per user. This results in the following equation which was used in the model (with el electricity in kWh/mo): el - exp (-2.366 + 0.37 * 1n(expe)) 38. olor TV Relatively inexpensive color television sets are rather new. Therefore for each income category the share of households owning one is expected to increase further, which means that the increased use cannot adequately be desibed only by income changes. To overcome this an exogenous growth rate can be entered. The S on the log of the share over one minus the share gives: L -23.04 + 1.812 * (expe) Std Dev (1.75) (0.14) t (-13) (13) R = 0.947 df - 18 Including a growth rate g and normalizing to a share 0.195 leads to the following logit used in the model: L = -23.01 + 1.25 * g * (year - 1988) + 1.812 In (expe) The consumption per user was: ln(el) 5 4.70 + 0.090 ln(expe) StdeDv - (052) (0.042) t - (9.0) (2.1) R2 -0.05 df= 484 - 104 - Annex VII For use in the model it was normalized to: el = exp (1.515 + 0.09 In (expe)) leading to an average consumption per user of 13.8 kWh per month. 39. Black and White TV Part of the change in the use of black and white television is thought to be caused by substitution to color TV sets. In such a case it is better to estimate the total use of television and derive the use of black and white by subtracting the share of color from the total Despite neglecting the possibility of households owning both types, this procedure is expected to catch better the effect of substitution. But television use remains the weakest part of this modeL More or less arbitrarily, the total share of households owning TV is assumed to be increasing with the same growth rate as the color TV share. Total share is: L = -18.10 + 1.566 In (expe) Std Dev = (0.91) (0.077) t = (-20) (20) R2 = 0.958 df = 18 The following equation was actually used in the model, leading to a total share of 0.524: L = -18.155 + 1.25 * g * (year-1988) - 1.566 In (expe) Consumption per user of a black and white set is: In (el) = 3.98 + 0.102 In (expe) Std Dev = (0.38) (0.033) t = (105) (3.1) R2 = 0.01 df = 817 When normalized to 6.27 kWh per month this results to: el = exp (0.63 + 0.102 In (expe)) 40. Pumps The share of people owning a water pump was found to be: L = -22.41 + 1.68 In (expe) Std Dev = (1.54) (0.12) t = (-14.5) (13.4) 12 = 0.974 df = 18 Normalizing to a share of 0.087 resulted in the following logit: L = -22.52 + 1.68 In (expe) Electricity use per user was according to the survey answers: In (el) = 1.92 + 0.301 In (expe) Std Dev = (1.02) (0.082) t = (1.9) (3.7) R2 = 0.06 df = 213 - 105 - Annex VII The estimated constant of this equation has a low significance of 0.06. In the model the following equation was used, leading to an average use per consumer of 11.36 kWh per month: el = exp(-1.105 + 0.301 In (expe)) Econometric analysis comparing calculated and metered electricity use led to the conclusion that the above mentioned equation is under estimated by a factor 2.78. 41. Incandescent Lighting No separate share of the number of households having incandescent lighting was calculated. It was assumed that this share is the same as the share of electrified households. Remaining differences were incorporated in the use per connection. This was found to be: In (el) = -1.14 + 0.312 In (expe) Std Dev = (0.31) (0.026) t = (-3.7) (11.9) R2 = 0.066 df = 1974 In normalized form, giving an average consumption of 17.07 kWh per months: el = exp(-0.88 + 0.312 In(expe)) Again it was found by comparing metered and calculated use, that this consumption is under estimated and should be multiplied with a factor 1.36. 42. Fluorescent lighting Shares of households using fluorescent lighting were calculated as a share of the electrified households. It was found to be: L = -7.28 + 0.69 In (expe) Std Dev = (0.94) (0.08) t = (-7.7) (8.5) R2 = 0.944 df = 18 When normalized for the share of households (0.69) having access to electricity and actually owning fluorescent lighting the following logit was found: L = -7.22 + 0.69 In (expe) Consumption per user was: In (el) = -3.34 + 0.448In (expe) Std Dev = (039) (0.033) t = (-8.5) (13.5) R2 = 0.11 df 1450 When normalized to an average consumption of 10.87 per month per user, the following expression resulted: el = exp(-2.947 + 0.448 In (expe)) Also fluorescent lighting was corrected by a factor 1.36. -106- Am V1U 43. £isri By comparing the distribution of the age of the refrigerators in the stock with an estimated life time of seven years it was possible to estimate a technical growth rate. This is the growth in the share of the appliance, above the growth caused by increased incomes. This technical growth rate was found to be about 2%. The distnbution of the shares as a function of income in 1988 has the following logit: L a -25.91 + 1.983 In (expe) StdDev = (133) (0.11) t = (-19.5) (18.4) R - 0.98 df = 18 When normalized with a share in 1988 of 0.109 and with the technical growth added, the following expression was found: L -25.937 + 1.983 ln(expe) + 0.02 * (year - 1988) Consumption per user is: In (el) = 0.386 + 0.044 In(expe) Normalized with an average consumption of 83.51 kWh per month per user, this results in: el = eV (3.868 + 0.044 In (epe)) EiguaA1k: Projections of Official and Informal Urban Elecrfcation Percent of Urban Households Electrified by Official and Informal Connections % Households Using Electricity 90% _ ,-IEIlectr ified 80% - Households 70%- ~~~PLN Connections 60%9 50% . ..I., . , ..I 1988 1992 1996 2000 Year -107 - An= V Eigure A7U7: Projections of Lighting Fuel Use Lighting Fuel Projections lan Residential Indonesia Millions of Liters/Year GWh/ Year 350 7000 s Kerosene Electricity 300 for Lighting for Lighti 6000 250 - 5000 200 - 4000 150 3000 100 l 2000 1988 1990 1995 2000 Year Household Fuel Projections for Urban Indonesia M=Ift cityPoectiona 44. Figure 3.9 displays projections of urban electricity use through the year 2000 from i) the model presented above and ii) PLN, based on the number of urban households officially connected and urban electification growth rates. The resemblance between both projections is good; nowhere do they differ more than about 5%. 45. Projections of percent of urban households electrified officially by PLN and unofficialy are shown in Figure A7.6 The substitution of electricity for kerosene in the lighting end use is shown in Figure A7.7. This projected substitution results directly from the expected growth in urban electrification. As discssed in Chapter m, sensitivity analysis conducted on electricity demand projections indicates that the total urban residential demand for electricity PLN is more or less independent of PLN's urban electrification efforts. Combusaile Fuel Use Proectons 46. Base case combustible fuel projections through the year 2000 and back to 1981, along with SUSENAS estimates are shown in Figure 3.8. Base case projections are discussed in Chapter m and scenarios under various pricing regimes are disssed in Chapter IV. Projections of kerosene, LPG, and wood use in urban residential Indonesia are tabulated in Annex VmI - 108 - Annex vmI ANNEX Vm LPG Promotion Scenarios Urban Residentfat Fuel Projections: Kerosene (t000 Kltoliters); LPG and Wood ('000 Tons) Case 1988-2000 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 Growth Base 3.43X Kerosene 3,310 3,406 3,504 3,619 3,736 3,858 3,984 4,114 4,271 4,435 4,605 4,781 4,964 10.38s LPG 171 189 208 230 254 281 310 343 378 417 460 507 559 1.47K Wood 3,129 3,179 3,229 3,278 3,327 3,377 3,428 3,480 3,528 3,577 3,627 3,677 3,729 It 2.30K Kerosene 3,310 3,215 3,122 3,220 3,320 3,423 3,530 3,640 3,772 3,908 4,050 4,197 4,349 13.06K LPG 171 225 296 325 357 392 431 473 518 567 621 681 745 2.40K Wood 3,129 3,339 3,562 3,620 3,678 3,738 3,798 3,860 3,918 3,9s7 4,036 4,097 4,159 III 3.18s Kerosene 3,310 3,375 3,441 3,549 3,661 3,776 3,894 4,017 4,166 4,321 4,481 4,648 4,820 11.89K LPG 171 208 253 279 307 339 373 411 451 496 545 599 658 1.48K Wood 3,129 3,181 3,234 3,283 3,332 3,382 3,433 3,485 3,533 3,582 3,632 3,683 3,734 IV 2.61K Kerosene 3,310 3,253 3,196 3,300 3,407 3,518 3,632 3,750 3,890 4,036 4,188 4,345 4,507 11.62K LPG 171 205 246 271 298 329 362 399 438 482 529 581 639 2.39K Wood 3,129 3,337 3,558 3,616 3,675 3,734 3,795 3,856 3,914 3,973 4,032 4,093 4,155 V 2.99K Kerosene 3,310 3,357 3,404 3,508 3,614 3,724 3,837 3,954 4,095 4,242 4,394 4,552 4,715 12.55s LPG 171 214 267 295 325 359 396 438 481 530 583 641 706 1.47K Wood 3,129 3,179 3,229 3,278 3,327 3,377 3,428 3,480 3,528 3,577 3,627 3,677 3,729 -109 - Economle Value of Fuel (Sillion Constant 1988 Rupiah) NPV Equivalent a10w ArmuelPavmnt Case Price 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 Base 203 Kerosene 672 691 711 735 759 783 809 835 867 900 935 971 1,008 4,968 808 s59 LPG 101 111 123 136 150 166 183 202 223 246 271 299 330 1,137 185 11 247 Kerosene 672 644 617 636 656 676 697 718 744 770 798 826 856 4,278 696 .09 LPG 101 130 168 184 203 222 244 269 294 323 354 387 425 1,513 246 III .03 Kerosene 672 685 698 720 74 766 791 815 846 877 910 943 978 4,856 790 509 LPG 101 121 146 161 177 195 215 237 260 286 315 346 380 1,333 217 IV 247 Kerosene 672 654 635 656 677 699 722 745 m 802 832 863 895 4,432 721 590 LPG 101 121 145 160 176 194 214 235 259 284 312 343 377 1.324 215 V 203 Kerosene 672 681 691 712 734 756 779 803 831 861 892 924 957 4,784 779 590 LPG 101 126 158 174 192 212 234 258 284 313 344 378 416 1,450 236 Economic Cost of Fuel Supply (Bitlion Constant 1988 Rupiah) Case Price 1988 1989 19 1991 192 19 1994 199 1996 1997 1998 199 2000 Base 290 Kerosene 960 988 1,016 1,049 1,0t. 1,119 1,155 1,193 1,239 1,286 1,335 1,386 1,440 7,097 1,155 428 LPG 73 81 89 98 109 120 133 147 162 178 197 217 239 825 134 II 290 Kerosene 960 932 905 934 963 993 1,024 1,056 1,094 1,133 1,174 1,217 1,261 6,285 1,023 428 LPG 73 96 127 139 153 168 184 203 222 243 266 291 319 1,141 186 III 290 Kerosene 960 979 998 1,029 1,062 1,095 1,129 1,165 1,208 1,253 1,300 1,348 1,398 6,937 1,129 428 LPG 73 89 108 119 132 145 160 176 193 212 233 256 282 989 161 IV 290 Kerosene 960 943 927 957 988 1,020 1,053 1,087 1,128 1,171 1,214 1,260 1,307 6,467 1,052 428 LPG 73 88 105 116 128 141 155 171 188 206 226 249 273 960 156 V 290 Kerosene 960 974 987 1,017 1,048 1.080 1,113 1,147 1,188 1,230 1,274 1,320 1,367 6,834 1,112 428 LPG 73 91 114 126 139 154 170 187 206 227 249 275 302 1,052 171 Met Economic Benefits (Billion Constant 1988 Rupiah) Case Price 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 Base Kerosene -288 -296 -305 -315 -325 -336 -347 -358 -372 -386 -401 -416 -432 -2129 -346 LPG 28 31 34 37 41 45 S0 56 61 68 74 82 90 312 51 11 Kerosene -288 -288 -288 -297 -307 -317 -327 -337 -350 -363 -376 -391 -405 -2007 -327 LPG 28 34 41 45 50 55 60 66 73 80 88 96 106 372 61 III Kerosene -288 -294 -299 -309 -318 -328 -339 -349 -362 -376 -390 *404 -419 -2081 -339 LPG 28 32 37 41 45 50 s5 61 67 74 81 90 99 343 56 IV Kerosene -288 -290 -291 -301 -311 -321 -331 -342 -355 -368 -382 -397 -412 -2034 -331 LPG 28 33 40 44 48 53 59 65 71 78 86 94 103 363 59 V Kerosewe -288 -292 -296 -305 -314 -324 -334 -344 -356 -369 -382 -396 -410 -2050 -334 LPG 28 35 43 48 S3 5864 71 78 86 94 104 114 398 65 .110. Annx VIII Government Revenue (31f iIon ConStRnt 1988 Rupiah) NPV Equivatent a 10% Awamat Pavaent Case Price 1988 1969 1990 1991 1992 1993 1994 199S 1996 1997 19 1999 2000 Base -92 Kerosene -305 -313 -322 *333 *344 -355 -367 -378 -393 -406 -424 -440 457 -2251 *366 123 LPG 21 23 26 28 31 3S 38 42 47 S1 S7 62 69 237 39 it O Kerosene 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 O LPG 0 0 0 0 a 0 0 0 0 0 0 0 0 0 0 III -92 Kerosene -305 -310 -317 -327 -337 -347 358 -370 -383 -398 -412 428 443 -2201 .358 0 LPG 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 IV 0 Kerosene 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 123 LPG 21 25 30 33 37 40 4S 49 54 59 65 72 79 276 45 V -92 Kerosene -305 -309 -313 -323 -333 -343 -353 -364 -377 -390 -404 -419 -434 -2168 -3S3 123 LPG 21 26 33 36 40 44 49 54 S9 65 72 79 87 302 49 Export Value of Urban Residential Consumption (Bfilion Constant 1988 Rupiah) Case Price 1968 1989 1990 1991 199 1993 1994 1995 1996 1997 19 1999 2000 Base 233 Kerosene 771 794 817 843 871 899 928 959 995 1,033 1,073 1,114 1,157 5,702 928 164 LPG 28 31 34 38 42 46 51 56 62 68 75 83 92 316 S1 11 233 Kerosane 771 749 728 750 774 798 822 848 879 911 944 978 1,013 5,050 822 164 LPG 28 37 49 53 59 64 71 78 85 93 102 112 122 437 71 III 233 Kerosene 771 786 802 827 8S53 880 907 936 971 1,007 1,044 1,083 1,123 5,574 907 164 LPG 28 34 42 S0 56 61 67 74 81 89 98 108 379 62 IV 233 Kerosene 771 758 745 769 794 820 846 874 906 940 976 1,012 1,050 5,196 846 I" LPG 28 34 40 44 49 54 59 65 72 79 87 95 105 368 60 V 233 Kerosene 771 782 793 817 842 868 894 921 954 988 1,024 1,061 1,099 5,491 894 164 LPG 28 35 44 S3 S9 65 72 79 87 96 105 116 403 66 - 111- Annex NPV 1990-1999 a lox In Constant 1986 Ruptah (Ilillons) ease it i II IV V Value of Fuels to Consuiers 6.105 5,791 6,189 5,756 6,234 Fuel Supply Costs 7.922 7,426 7,926 7,427 7,867 Net Economic BneAfIts -1817 .1635 -1738 -1671 -1652 Net Governrnt Revenues -2014 0 -2201 276 -1846 Export Value of Domestic Consamption 6,018 5,487 S,9S3 S,564 5,694 V 1990-1999 8 10% In Constant 1988 US. (NIllIons) 1700Rp/SUS Base it III IV V Value of Fuels to Consumers 3,591 3,406 3,640 3.386 3,667 Fuel Supply Costs 4,660 4,368 4,663 4,369 4,639 Net Economic BenefIts -1069 -962 -1022 -963 -972 Net Government Revenues -1185 0 -1295 162 -109 Export Value of DomestIc Consmuption 3,540 3,228 3,S02 3,273 3,467 NPV 1990-1999 a 10% In Constant 1988 USS (MIlltIns) 1700Rp/SUS Base II III IV V (Dffference from Bease Case) Value of Fuels to Consauers 3,591 -185 49 -2CS 76 Fuel Supply Costs 4,660 -291 3 -291 -21 Net EconomIc Benefits -1069 107 46 86 97 Net Government Revenues -1185 1185 -110 1347 87 Export Value of Domestic Consumpt(on 3,540 -312 -38 -267 -73 - 112 - Annex VIII Annualized Costs & Benefits 1990-1999 0101 In Constant 198 Rupiah (BiLlions) Dase It III IV V Value of Fuels to Consurers 994 942 1.007 937 1,015 Fuel Supply Costs 1,289 1.209 1,290 1.209 1,283 Net Economic Benefits *296 -266 -283 -272 -269 Net Government Revenues *328 0 -358 45 -304 Export Value of Domestic Consuiption 979 893 969 905 959 Annualized Costs & Benefits 1990-1999 8 101 in Constant 1988 US$ (Militons) 1700R Base II III IV V Value of Fuels to ConsuTers 584 554 592 551 597 Fuel Supply Costs 758 711 759 711 755 Net Economic BOenefits -174 -157 -166 -160 -158 Not Government Revenues -193 0 -211 26 -179 Export Value of Domestic Consumption 576 525 570 533 564 4nnualized Costs & Benefits 1990-1999 1011 In Constant 1988 USS (Millions) 170CR Base 11 III IV V (Difference from Base Case) Value of Fuels to Consumers 584 -30.0 8.0 -33.4 12.4 Fuel Supply Costs 758 47.4 0.5 -47.4 -3.4 Not Economic Benefits -174 17.4 7.5 14.0 15.8 Net Government Revenues -193 192.8 -17.8 219.3 14.2 Export Value of Domestic Consuption 576 -50.8 -6.2 -43.5 -11.8 *113 - Annex IX ANNEX IX Draft Euerg Module fr Susenas Fuel Consumption by Use: use Cooking Lighting Water H. Space Heat Commwrcial Other Electricity Kerosene LPG Charcoal Wood/Biofuel City Gas Electricitv If don't use, why? _ non-electrified area _ other. Connection/Source: _ PLN direct _ Other household's inter B Battery (volts_) _ Non-PLN Electricity Distributor If use for Commercial purposes, specify __(code for type) If PLM: Meter Number: KVA: KWH/Mo (most recent period available): Electric Appliances used (Number): B&W TV Rice Cooker Window Air Conditioner Color TV Sound System Central Air Conditioner Radio Fan Video Iron Electric Stove Hair Dryer Refrigerator Electric Oven Toaster Kettle Washing Machine Blender/Mixer Pump Freezer Other Electric Lighting: Incandescent: Number of Bulbs: Total Wattage: Fluorescent: NMuber of Bulbs: Total Wattage: Kersn If do not use, is kerosene available? __. Source: _ Delivered _ "Warung" _ Dealer _ other. Price _ Rp/liter Use during Past week:_ liter For cooking:__ liter For lighting:__ liter Other: _ Code If cook: Stove 1: wicks - days used in last week: Stove 2: wicks - days used in last week: Stove 3: wicks - days used in last week: If light: Lamp 1: code for type - days used in last week: Lamp 2: Lamp 3: Lamp 4: Price of Kerosene: Source:_ Deliver _-Shop __Dealer _-Other -114- Alt e (this section should be both protested and calibrated.) If use for couerciel purpose specify - (code for type) Average delly use during past week:_ kg Of ihich _ kg home cooking & water heating - kg commercIal use - kg other Extra use during cold months _ kg/day for _ months (approximate) Sources of wood/biofuel used In past week: I Prfncipal Source/Type. 2 Used daily in large quantity. 3 Used at least one day in large quantity. 4 Used daily In smtll quantties. 5 Used at least one day in small quantities. Land Tvre Buy Own Neighbor Marginal Forest Fallen/Dead Branches uPruned Branches ("pollarding") "Lops and Tops" Tree felled for fuel Rice Husks Cassava Stalks Coconit tree parts Leaves Other Agricultural if do not use, is LPG available? Source: __Deliver _Shop Pealer _Other if use for comercial purposes, specify _ (code for type) Number of 11 kg bottles owned:_ Of which _ back-up; Uses include:ggt Consurption in Kg/wk: (WORK SPACE PROVIDED FOR CALCULATING FROM LENGTH OF TINE BETUEEN REPLACEMENTS) PrIce of contents (11kg): _ Rupiah Nwuier of 45 kg bottles owned:__ Of which__ back-up; Uses include:gol Consumption in Kg/ok: (WORK SPACE PROVIDED FOR CALCULATING FROM LENGTH OF TINE BETWEEN REPLACEMENTS) Price of contents (45kg): _ Rupiah Stove: _ burners Oven: _ Y/N Water Heater: _ Y/N Other: __Y/M City Gas Meter No:__ Monthly consuAption during last period metered:__ cubic meters. Stove: _ burners Oven: _ r/N Water Heater: _ Y/N Other: ___/N CharcoSl Price: _ RpWks Use over past week: _ kg Of which kg roning kg sate kg other cooking - kg camrcial purpose llS - Annex1X ANNEX X Rual Household Energy Strategy Study Proposal 1. Phase I: Analysis of SUSENAS data for rural households and review of previous rural energy studies would be conducted in Washington. hi;s would result in an isues paper and detailed terms of reference for Phase II including a detailed survey of households, data analysis, fuel use projections and policy analsis for each major region of Indonesia. Particular attention will be focussed on the impacts of rural electrification on kerosene use, expenditures on fuel by households in various income categories, and efficiency of woodfuel stoves. The rural study would complement the urban household energy strategy study presented herein, resulting in a comprehensive picture of residential energy use in Indonesia as a basis for formulating effective policy in the sector. Preliminary budget estimates are shown below. Budget Estimates for Rural Household Ener= Study Man-months LQ Inf US$ Phase I: Data Analysis and Issues Paper 5 50,000 Phase II: Regional Survey of Households 60 3 90,000 Biofuel Stoves Testing and Program Design 4 2 60,000 Analysis and Strategy Formulation 1Q , Strategy Study Total 74 20 350,000 - 116 - AnnasX ANNEX XI Economic Evaluation of Kerosene Stove Pilot Proam Economic Evaluation of Kerosene Stove Pilot Program 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 NH stock of stoves (Millions) Total 12.9 13.5 14.1 14.7 15.3 15.9 16.5 17.2 17.8 18.5 lmproved 0.0 0.6" 2.5X 5.3X 8.6X 11.9X 14.7K 16.2X 17.0K 17.2X Sales of stoves (Millions per anmnu) Total 3.8 4.0 4.2 4.3 4.5 4.7 4.9 5.1 5.3 5.5 Improved 0.0K 2.0K 6.6K 10.0X 12.3K 14.4X 16.4K 17.1K 17.1K 17.1K Costs USS (Millions) Program cost 0.25 0." 0.46 0.30 0.30 0.30 0.30 0.30 0.30 0.30 Increased stove cost 0.00 0.01 0.03 0.05 0.06 0.07 0.06 0.04 0.03 0.02 Total cost 0.25 0.44 0.49 0.35 0.36 0.37 0.36 0.34 0.33 0.32 Kerosene saved LIters (tllions) 0 1.0 4.4 9.8 '6.4 23.6 30.3 34.9 37.9 39.9 Foregone Fuel Supply Costs USS (MilLions) 0.00 0.17 0.75 1.67 2.80 4.03 5.17 5.96 6.47 6.80 Net Econemc Bnef Its -0.25 -0.28 0.26 1.32 2.43 3.66 4.81 5.61 6.14 6.48 Economfc Internal Rate of Return 128.3K ^overnment Budget Expenditures 0.31 0.54 0.58 0.38 0.38 0.38 0.38 0.38 0.38 0.38 Income 0.00 0.05 0.24 0.53 0.89 1.28 1.64 1.89 2.05 2.16 Net -0.31 -0.49 -0.34 0.16 0.51 0.90 1.27 1.51 1.68 1.78 Balanee of PaYments Imports 0.13 0.23 0.25 0.16 0.16 0.:6 0.16 0.16 0.16 0.16 Exports 0.00 0.13 0.60 1.35 2.25 3.24 4.16 4.79 5.20 5.46 Net -0.13 -0.10 0.36 1.19 2.09 3.08 4.00 4.63 5.04 5.30 Summary Evaluation of Kerosene Stove Pilot Program Net Present Value 8 10K Annualized NPV (USS Millions) (Us$ Millions) Kerosene Saved 16.65 2.71 Program cost 2.03 0.33 Increased stove cost 0.21 0.03 Total Costs 2.25 0.37 Net Economec Beneffts 14.41 2.34 ownwmmnt Budget 2.74 0.45 Belance of Paveents 12.30 2.00 1700 Rupiah a I US Doltar. Economic Costs of Kerosene Supply (Rupiah/liter): 290 Government Kerosene Sutisdy (Ruplah/liter): 92 Export Value (Rupiah/liter): 233 Conversion factor for government expenditures: 0.8 Foreign exchange component of program costs: 43K - 117 - Annex XII ANNEX XII Economic Evaluation of Electriclty Conservation Program Urban Electricity Conservation Program Economic Costs and Benefits 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 Costs (USS Millions) -Test faclities 0.3 1.5 1.5 -Training, Personnel, and Administration 0.3 0.5 0.5 0.5 0.3 0.3 0.3 0.3 0.3 0.3 0.3 -Info campain 0.3 0.5 0.5 0.5 0.4 0.3 0.3 0.3 0.3 0.3 0.3 -Testing + labelling Refrigerators B USS 2 / unit 0.45 0.50 0.55 0.61 0.67 0.74 0.82 0.90 0.99 1.10 1.21 Air-conditioners a uss 5 / unit 0.04 0.05 0.05 0.06 0.07 0.08 0.09 0.11 0.12 0.14 0.16 Television sets a uss 2 / 10 units 0.22 0.24 0.26 0.28 0.31 0.34 0.37 0.40 0." 0.48 0.52 Total Program Cost 1.61 3.28 3.36 1.95 1.75 1.76 1.88 2.01 2.16 2.32 2.50 Market Potential Share I/ Savings hi Benefits (0GA Saved) 0 9.6 26.0 55.9 90.2 129.1 166.1 207.5 253.8 305.3 362.5 Lighting 10X 301 0 5.4 14.5 30.9 49.4 69.8 88.8 109.5 132.3 15T.1 184.0 Refrigeration 351 25% 0 3.7 10.0 21.6 35.2 51.0 66.3 83.9 103.8 126.5 152.1 Air-conditioning 351 25X 0 0.2 0.7 1.6 2.9 4.4 6.1 8.1 10.4 13.1 16.1 Television 201 5X 0 0.3 0.8 1.7 2.7 3.9 4.9 6.0 7.3 8.7 10.2 Program Effectiveness I/ 01 5X 131 251 38 50X 60X 701 801 901 1001 Total Value of Sm4ngs (USS .1tt1mn a 99t/kIh) 0 0.85 2.29 4.93 7.96 11.39 14.65 18.31 22.40 26.94 31.98 Lighting 0 0.48 1.28 2.73 4.35 6.16 7.83 9.67 11.67 13.86 16.24 Refrigeration 0 0.32 0.88 1.91 3.11 4.50 5.85 7.40 9.16 11.16 13.42 Air-conditioning 0 0.02 0.06 0.14 0.25 0.39 0.54 0.71 0.92 1.15 1.42 Television 0 0.03 0.07 0.15 0.24 0.34 0.43 0.53 0.64 0.77 0.90 get Program enefits (USS miLtions) -1.61 -2.43 -1.07 2.99 6.21 9.63 12.78 16.30 20.24 2S.62 29.49 Economic Internal Rate of Return: 72.0X y/ Targetted market share of efficient appliances by the year 2000. b/ KWh savings of improved appliances expressed as percent of kWh used in standard models. pj Percent of year 2000 Improved appliance market penetration target. Sumiry Evaluation of Urban Etectricity Conservation Program Net Present Value B 10X AnnuaLized NPV (USS tillions) (USS Millions) Total Program Cost (USS millions) 14.7 2.3 Electricity Conservation (POh) 729 112 Lighting 384 59 Refrigeration 295 45 Air-conditioning 28 4 Television 21 3 Econodmc Value of Savings (USS llions) 64.3 9.9 Lighting 33.9 5.2 Refrigeration 26.1 4.0 Air-conditioning 2.5 0.4 Television 1.9 0.3 Not Economic Bnefwits 49.6 7.6 - 118 - Anna XI The evaluation above is based on the following projections of electricity use and appliance sales: UNESS Urban A4pftac. Sales Projecthns Amual sales Annual 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 Growth Light bulbs (millions) 5.9% 39.1 41.3 43.8 46.3 49.0 51.9 54.9 58.1 61.6 65.2 69.0 73.0 Refrigerators (thousands) 10.4% 204.0 225.2 248.6 274.5 303.0 334.6 369.4 407.8 450.2 497.0 548.7 605.7 Air-conditfoners (thousands) 15.0% 7.0 8.1 9.3 10.6 12.2 14.1 16.2 18.6 21.4 24.6 28.3 32.6 Television sets (millions) 9.3% 1.0 1.1 1.2 1.3 1.4 1.5 1.7 1.8 2.0 2.2 2.4 2.6 HESS Urbtn Electrfcity Oeand Projectims for Urban lIdwms1a (G1A/yr) 1988 1989 1990 1991 19M 1993 1994 1995 1996 1997 1998 1999 2000 Incandescent lighting 1,966 2,122 2,280 2,640 2,603 2,768 2,937 3,110 3,288 3,471 3,659 3,854 4,056 Fluorescent lighting 932 1,013 1,096 1,181 1,268 1,357 1,449 1,545 1,643 1,746 1,852 1,963 2,079 RefrIgeration 653 710 772 839 912 990 1,074 1,165 1,263 1,369 1,484 1,606 1,739 Ironing 400 425 451 479 509 540 573 608 645 685 726 770 817 Water pumping 305 328 352 378 406 436 469 503 540 580 622 668 716 Color television 292 312 333 355 379 404 431 460 490 522 557 593 632 SW television 225 236 247 259 271 284 297 311 325 340 356 372 389 Sales 368 384 397 406 411 413 412 403 377 350 323 295 268 Others 426 599 792 1,003 1,232 1,479 1,744 2,026 2,325 2,640 2,972 3,319 3,683 Total 2,669 2,994 3,344 3,719 4,119 4,546 5,000 5,477 5,966 6,487 7,039 7,624 8,243 ENERGY SECTOR MANAGEMENT ASSISTANCE PROGRAM Activities Completed Country Project Date Number ENERGY EFFICIENCY AND STRATEGY Africa Regional Participants' Reports - Regional Power Seminar on Reducing Electric System Losses in Africa 8/88 087/88 Bangladesh Power System Efficiency Study 2/85 031/85 Botswana Pump Electrification Prefeasibility Study 1/86 047/86 Review of Electricity Service Connection Policy 7/87 071/87 Tuli Block Farms Electrification Prefeasibility Study 7/87 072/87 Burkina Technical Assistance Program 3/86 052/86 Burundi Presentation of Energy Projects for the Fourth Five-Year Plan (1983-1987) 5/85 036/85 Review of Petroleum Import and Distribution Arrangements 1/84 012/84 Burundi/Rwanda/Zaire (EGL Report) Evaluation de l'Energie des Pays des Grands Lacs 2/89 098/89 Congo Power Development Plan 2/90 106/90 Costa Rica Recommended Technical Assistance Projects 11/84 027/84 Ethiopia Power System Efficiency Study 10/85 045/85 The Gambia Petroleum Supply Management Assistance 4/85 035/85 Ghana Energy Rationalizaton in the Industrial Sector of Ghana 6/88 084/88 Guinea- Recommended Technical Assistance Bissau Projects in the Electric Power Sector 4/85 033/85 Management Options for the Electric Power and Water Supply Subsectors 9/89 100/89 Indonesia Energy Efficiency Improvement in the Brick, Tile and Lime Industries on Java 4/87 067/87 Power Generation Efficiency Study 2/86 050/86 Diesel Generation Efficiency Improvement Study 12/88 095/88 Jamaica Petroleum Procurement, Refining, and Distribution 11/86 061/86 Kenya Power System Efficiency Report 3/84 014/84 Liberia Power System Efficiency Study 12/87 081/87 Recommended Technical Assistance Projects 6/85 038/85 Madagascar Power System Efficiency Study 12/87 075/87 Malaysia Sabah Power System Efficiency Study 3/87 068/87 Mauritius Power System Efficiency Study 5/87 070/87 Panama Power System Loss Reduction Study 6/83 004/83 Papua New Energy Sector Institutional Review: Proposals for Guinea Strengthening the Department of Minerals and Energy 10/84 023/84 Power Tariff Study 10/84 024/84 Senegal Assistance Given for Preparation of Documents for Energy Sector Donors' Meeting 4/86 056/86 Seychelles Electric Power System Efficiency Study 8/84 021/84 Sri Lanka Power System Loss Reduction Study 7/83 007/83 Syria Electric Power Efficiency Study 9/88 089/88 Energy Efficiency in the Cement Industry 7/89 099/89 ENERGY SECrOR MANAGEMENT ASSISTANCE PROGRAM Activities Completed Country Project Date Number ENERGY EFFICIENCY AND STRATEGY (Continued) Sudan Power System Efficiency Study 6/84 018/84 Management Assistance to the Ministry of Energy and Mining 5/83 003/83 Togo Power System Efficiency Study 12/87 078/87 Uganda Energy Efficiency in Tobacco Curing Industry 2/86 049/86 Institutional Strengthening in the Energy Sector 1/85 029/85 Power System Efficiency Study 12/88 092/88 Zambia Energy Sector Institutional Review 11/86 060/86 Energy Sector Strategy 12/88 094/88 Power System Efficiency Study 12/88 093/88 Zimbabwe Power Sector Management Assistance Project: Background, Objectives, and Work Plan 4/85 034/85 Power System Loss Reduction Study 6/83 005/83 HOUSEHOLD, RURAL, AND RENEWABLE ENERGY Burundi Peat Utilization Project 11/85 046/85 Improved Charcoal Cookstove Strategy 9/85 042/85 China Country-Level Rural Energy Assessments: A Joint Study of ESMAP and Chinese Experts 5/89 101/89 Fuelwood Development Conservation Project 12/89 105/89 C6te d'Ivoire Improved Biomass Utilization-Pilot Projects Using Agro-Industrial Residues 4/87 069/87 Ethiopia Agricultural Residue Briquetting: Pilot Project 12/86 062/86 Bagasse Study 12/86 063/86 The Gambia Solar Water Heating Retrofit Project 2/85 030/85 Solar Photovoltaic Applications 3/85 032/85 Ghana Sawmill Residues Utilization Study, Vol. I & II 10/88 074/87 Global Proceedings of the ESMAP Eastern & Southern Africa Household Energy Planning Seminar 6/88 085/88 India Opportunities for Commercialization of Non-Conventional Energy Systems 11/88 091/88 Indonesia Household Energy Strategy Study 2/90 107/90 Jamaica FIDCO Sawmill Residues Utilization Study 9/88 088/88 Charcoal Production Project 9/88 090/88 Kenya Solar Water Heating Study 2/87 066/87 Urban Woodfuel Development 10/87 076/87 Malawi Technical Assistance to Improve the Efficiency of Fuelwood Use in the Tobacco Industry 11/83 009/83 Mauritius Bagasse Power Potential 10/87 077/87 Niger Household Energy Conservation and Substitution 12/87 082/87 Improved Stoves Project 12/87 080/87 Pakistan Assessment of Photovoltaic Programs, Applications and Markets 10/89 103/89 ENERGY SECTOR MANAGEMENT ASSISTANCE PROGRAM Activities Completed Country Project Date Number HOUSEHOLD, RURAL AND RENEWABLE ENERGY (continued) Peru Proposal for a Stove Dissemination Program in the Sierra 2/87 064/87 Rwanda Improved Charcoal Cookstove Strategy 8/86 059/86 Improved Charcoal Production Techniques 2/87 065/87 Senegal Industrial Energy Conservation Project 6/85 037/85 Urban Household Energy Strategy 2/89 096/89 Sri Lanka Industrial Energy Conservation: Feasibility Studies for Selected Industries 3/86 054/86 Sudan Wood Energy/Forestry Project 4/88 073/88 Tanzania Woodfuel/Forestry Project 8/88 086/88 Small-Holder Tobacco Curing Efficiency Project 5/89 102/89 Thailand Accelerated Dissemination of Improved Stoves and Charcoal Kilns 9/87 079/87 Rural Energy Issues and Options 9/85 044/85 Northeast Region Village Forestry and Woodfuel Pre-Investment Study 2/88 083/88 Togo Wood Recovery in the Nangbeto Lake 4/86 055/86 Uganda Fuelwood/Forestry Feasibility Study 3/86 053/86 Energy Efficiency Improvement in the Brick and Tile Industry 2/89 097/89 SUMATERA IND)NESIA 1988 URBAN HOUSEHOLD ENERGY SUR SAMPLE CITIES D.K.I. JAKARTA OSewong *D .Iirto lcongeront2 ettw" . _s D %oOvon indrmomeu0 Jos ngo r- Jo.oog oB° °3 J A W A B A R A T C.putot --cirbonl0 Kudr%g CC-ctu9 e Worung *3 Dliumedong 0Giwi /oPemo6ng 7 Ocact (~ ~ ~~~~~QvnidonliC.mu Pagoden o bl' tng C,coaol >0$aobumi °8endu| @Kodu9o (Bontor Belong roBouni 0l Purwodod ownomia J AW A T [ N G A H Boyingbong g 9Ci.anis Wono$obo *Pirokron f*06i '28 ~~~~~~~~~~~ ~ ~~ ~ ~~Singo>gorno e\utorilett°o ewoii9 * ~oljo Iig 0 G locop NKbumni r * / O L hwon3 D.K. YOGYAKAR'iA PHILIPPINES BRUNE i I TN$ m,ap htsn Llrpar by MALAYSIA - The wbow SW*-# Setaf*S&* MALAYSIA / for VW ctAIY O of 0*e meadies and a andkMWy for goe SiN4GAPORL ingvna use of Tr*11t" 0 KALIMANTAN wnd tontOviF ince le 'I SkJLAWESI ~~~~~~~~~~~~~~Coq,oibo The dl.mwiaw&Wu Ss SULAWESI IRIAN JAYAJ usd dV b an tomupdonotp4, onIgo w o Fnce hpaib dwne on Vi lI sO$ag of any Oie"doy or alr enors*nw e or scOple of aiAiiet -i3 K_ THwAIND .w ,*u 9|$ft6 A.&- INDONESIA er f irllSrct . .} s ~~~~~~~~~~~~~ ~~~~~~PHIu - NES rPOPU;LATON r 13 MASARt 5 URBAN BY MALAYSIA\ MJ jOR ISLAND GOP \ @ WMPUR_ * r $' ' ~~12 ,(MALAYS>IA( > ~~~~P} ~~~170 ~~~~~1 m 1 , 8 c Jt.181nK 47ED4 / ~ 0 4.3eXSg < 5 ~ t7act. 2.2naAbWttq sttrd < 6g F^lSi . > + @SsSUSUR t X % < C . W :4 } - eP e MgES -%W~~~~~~~~~~~~~~~~~~~~~~~~~O FAIJIS XSA -A _E \ X X ~~~~~~~~~KUUTS , "4'WAm 13 1twnr 20f; JR14N a ># y1 SAW vTMR 1aA YA ,+St A5exi D * t&t § 4NiA7di