Household Energy Supply and Use in Yemen: Volume II, Annexes Report 315/05 December JOINT UNDP / WORLD BANK ENERGY SECTOR MANAGEMENT ASSISTANCE PROGRAMME (ESMAP) PURPOSE The Joint UNDP/World Bank Energy Sector Management Assistance Program (ESMAP) is a special global technical assistance partnership sponsored by the UNDP, the World Bank and bi-lateral official donors. Established with the support of UNDP and bilateral official donors in 1983, ESMAP is managed by the World Bank. ESMAP's mission is to promote the role of energy in poverty reduction and economic growth in an environmentally responsible manner. Its work applies to low-income, emerging, and transition economies and contributes to the achievement of internationally agreed development goals. ESMAP interventions are knowledge products including free technical assistance, specific studies, advisory services, pilot projects, knowledge generation and dissemination, trainings, workshops and seminars, conferences and roundtables, and publications. ESMAP work is focused on three priority areas: access to modern energy for the poorest, the development of sustainable energy markets, and the promotion of environmentally sustainable energy practices. GOVERNANCE AND OPERATIONS ESMAP is governed by a Consultative Group (the ESMAP CG) composed of representatives of the UNDP and World Bank, other donors, and development experts from regions which benefit from ESMAP's assistance. The ESMAP CG is chaired by a World Bank Vice President, and advised by a Technical Advisory Group (TAG) of independent energy experts that reviews the Programme's strategic agenda, its work plan, and its achievements. ESMAP relies on a cadre of engineers, energy planners, and economists from the World Bank, and from the energy and development community at large, to conduct its activities under the guidance of the Manager of ESMAP. FUNDING ESMAP is a knowledge partnership supported by the World Bank, the UNDP and official donors from Belgium, Canada, Denmark, Finland, France, Germany, the Netherlands, Norway, Sweden, Switzerland, and the United Kingdom. ESMAP has also enjoyed the support of private donors as well as in-kind support from a number of partners in the energy and development community. FURTHER INFORMATION For further information on a copy of the ESMAP Annual Report or copies of project reports, please visit the ESMAP website: www.esmap.org. ESMAP can also be reached by email at esmap@worldbank.org or by mail at: ESMAP c/o Energy and Water Department The World Bank Group 1818 H Street, NW Washington, D.C. 20433, U.S.A. Tel.: 202.458.2321 Fax: 202.522.3018 Household Energy Supply and Use in Yemen Volume 2: Annexes December 2005 Energy Sector Management Assistance Program (ESMAP) Table of Contents Abbreviations and Acronyms............................................................................ix Annex 1............................................................................................................. 1 The Household Energy Survey......................................................................... 1 The Consultation Process in Survey Design ................................................ 1 Survey Coverage and Sample Frame........................................................... 1 International Comparisons ............................................................................ 7 Questionnaire Design and Field Survey ....................................................... 9 Training of Supervisors and Enumerators .................................................... 9 Data Processing.......................................................................................... 10 Remarks and Lessons Learned.................................................................. 11 Annex 2........................................................................................................... 13 Participatory Rapid Assessment..................................................................... 13 Research Sites and Methodology............................................................... 13 Socio-Economic Context and Household Eenergy Use in Yemen............. 18 Implementation of the PRA Field Research: What Worked?...................... 19 Instruments for Field Research................................................................... 19 Data Analysis .............................................................................................. 20 Annex 3........................................................................................................... 21 Kerosene......................................................................................................... 21 Consumption Patterns................................................................................. 21 The Retail Price of Kerosene ...................................................................... 23 Annex 4........................................................................................................... 33 Diesel .............................................................................................................. 33 Household Consumption Patterns .............................................................. 33 Diesel Expenditure...................................................................................... 38 The Retail Price of Diesel............................................................................ 38 Diesel Smuggling ........................................................................................ 39 Reducing the Diesel Subsidy: Direct Effects .............................................. 42 Reducing Diesel Subsidies: Indirect Effects ............................................... 44 Impact of Past Increases in the Diesel Price .............................................. 47 Annex 5........................................................................................................... 51 Impact of Diesel Subsidies on Water, Qat and Food Prices ......................... 51 Water........................................................................................................... 51 Food ............................................................................................................ 56 Annex 6........................................................................................................... 59 LPG................................................................................................................. 59 Consumption Patterns................................................................................. 59 Reconciliation with Supply Side Data ......................................................... 68 The Retail Price of LPG .............................................................................. 73 LPG Subsidies............................................................................................. 77 Policy Option: Reduce the Subsidy on LPG ............................................... 78 Policy Option: Subsidize LPG Cylinders..................................................... 82 Annex 7........................................................................................................... 85 Gasoline and Fueloil....................................................................................... 85 Fueloil.......................................................................................................... 85 iii Gasoline ...................................................................................................... 86 Subsidizing the Refinery ............................................................................. 87 Electricity......................................................................................................... 89 Patterns of Electricity Consumption............................................................ 89 Tariff Structure............................................................................................. 94 Willingness to Pay for Electricity............................................................... 101 Productive Use of Electricity ..................................................................... 104 Annex 9......................................................................................................... 107 Biomass ........................................................................................................ 107 Fuelwood Collection.................................................................................. 108 Fuelwood Consumption ............................................................................ 115 Charcoal.................................................................................................... 118 Crop Residues........................................................................................... 120 Annex 10....................................................................................................... 123 Survey Data Reconciliation........................................................................... 123 Expenditure and Income Data in the HES Survey Data........................... 123 Calculation of Income Deciles................................................................... 125 Comparison with the Last Household Expenditure Survey ...................... 126 Food Expenditure Data ............................................................................. 128 General Patterns of Income Distribution................................................... 128 Poverty Distribution and Comparison with the Poverty Update Report.... 130 List of Figures Figure A2.1: Map of the Study Areas ............................................................. 15 Figure A3.1: Kerosene Use by Income Decile ............................................... 22 Figure A3.2: Monthly Kerosene Demand ....................................................... 23 Figure A3.3: Frequency Distribution of Retail Kerosene Prices..................... 23 Figure A3.4: Average Kerosene Price by Governorate.................................. 24 Figure A3.5: Kerosene Price v. Distance of Village to Nearest Paved Road. 24 Figure A3.6: Kerosene Price v. Distance to Nearest District HQ (km)........... 25 Figure A3.7: Distribution of Monthly Kerosene Expenditure .......................... 26 Figure A3.8: Average Monthly Household Kerosene Expenditure................. 27 Figure A3.9: Fraction of Total Kerosene Subsidy to Households Captured by Each Income Decile........................................................................................ 29 Figure A4.1: Proportion of Households using Diesel...................................... 34 Figure A4.2: Percentage of Households in Each Governorate Reporting Diesel Use....................................................................................................... 37 Figure A4.3: Monthly Diesel Consumption (All Uses, Including PEC and Non Households).................................................................................................... 37 Figure A4.4: Distribution of Reported Diesel Prices, YR/liter......................... 38 Figure A4.5: Average Price of Diesel, by Governorate .................................. 39 Figure A4.6: Diesel Consumption and GDP................................................... 41 Figure A4.7: Diesel Consumption v. GDP and Diesel Prices......................... 42 Figure A4.8: Fraction of Total Diesel Subsidy to Households, Captured by Each Income Decile........................................................................................ 43 Figure A4.9: Impact of Diesel Smuggling on Food Price Increases............... 46 Figure A4.10: Quarterly Inflation Rates (CPI) v. Diesel Price ........................ 48 iv Figure A6.1: The LPG Supply Chain .............................................................. 62 Figure A6.2: Urban v. Rural Access to LPG................................................... 63 Figure A6.3: LPG Penetration by Income Decile............................................ 63 Figure A6.4: LPG Expenditure by Income Decile........................................... 64 Figure A6.5: Percentage of Households in Each Governorate Reporting LPG Use.................................................................................................................. 67 Figure A6.6: Monthly LPG Demand................................................................ 68 Figure A6.7: Distribution of Reported Cylinders/month.................................. 69 Figure A6.8: Bottles per Month v. Household Size......................................... 69 Figure A6.9: Apparent 2002 LPG Consumption per Household v. Household Income ............................................................................................................ 71 Figure A6.10: LPG Consumption/Household v. Household Income as per 2003 HES........................................................................................................ 72 Figure A6.11: Expected Household LPG Consumption ................................. 72 Figure A6.12: Distribution of Reported Consumer LPG Prices...................... 74 Figure A6.13: Average Price of 11kg LPG Cylinder, by Governorate............ 74 Figure A6.14: Cost of LPG Cylinder as a Function of Income Decile and Location .......................................................................................................... 75 Figure A6.15: LPG Penetration v. Average Price........................................... 75 Figure A6.16: LPG Penetration v. Average Price, Rural Households............ 76 Figure A6.17: LPG Use per Household v. Average LPG Price (all Households and all Income Deciles) .................................................................................. 76 Figure A6.18: LPG Use v. Price, Rural Households, Bottom Income Quintile77 Figure A6.19: Fraction of Total LPG Subsidy to Households, Captured by Each Income Decile........................................................................................ 78 Figure A6.20: Net Effect of the Mitigation Scheme ........................................ 80 Figure A7.1: Fueloil Price Differentials, Rotterdam ........................................ 85 Figure A7.2: Difference between Leaded and Unleaded Gasoline, Spot Cargoes, Italy.................................................................................................. 87 Figure A7.3: Crude Yields v. Yemen Domestic Product Market Slate........... 88 Figure A8.1: Access to Electricity by Income Decile ...................................... 90 Figure A8.2: Hours of Service, All Isolated Systems and Self-gen Sets........ 92 PEC national grid Isolated systems and self-generation ... 92 Figure A8.3: Electricity Consumption, kWh/month (Top Income Decile, Rural PEC Grid Customers)..................................................................................... 93 Figure A8.4: Electricity Consumption v. Income............................................. 94 Figure A8.5: Ratio of Monthly Bills, Rural to Urban, as a Function of kWh.... 98 Figure A8.6: Distribution of Household Electricity Expenditure.................... 100 Figure A8.7: Fraction of Income Spent on Electricity ................................... 101 Figure A8.8: Demand for Lighting Service.................................................... 102 Figure A8.9: Expenditure on Electricity and Electricity Substitutes, by Type of Electricity Access.......................................................................................... 104 Figure A9.1: Method of Obtaining Fuelwood................................................ 108 Figure A9.2: Distribution of Fuelwood Collection Distances ........................ 109 Figure A9.3: Average Fuelwood Collection Distances by Governorate....... 109 Figure A9.4: Frequency Distribution of the Number of Collections per Month ...................................................................................................................... 110 v Figure A9.5 Monthly Household Time Budget for Fuelwood Collection....... 110 Figure A9.6: Relationship between Monthly Collection Time Budgets and Family Size ................................................................................................... 111 Figure A9.7: Labor Input of Girls .................................................................. 111 Figure A9.8: Labor Input of Boys.................................................................. 112 Figure A9.9: Comparison of Labor Inputs by Governorate .......................... 112 Figure A9.10: Labor Input of Women ........................................................... 113 Figure A9.11: Labor Input of Women as a Function of Collection Distance 113 Figure A9.12: Labor Input of Men................................................................. 114 Figure A9.13: Labor Contribution of Men as a Function of Collection Distance ...................................................................................................................... 114 Figure A9.14: Comparison of Labor Inputs, Adult Men v. Adult Women ..... 115 Figure A9.15: Consumption of Fuelwood By Decile, kg/HH/month.............. 117 Figure A9.16: Fuelwood Prices, YR/kg......................................................... 117 Figure A9.17: Fuelwood Consumption v. Price............................................ 118 Figure A9.18: Use of Charcoal by Income Decile ........................................ 119 Figure A9.19: Charcoal Prices, YR/kg.......................................................... 119 Figure A9:20: Use of Crop Residues............................................................ 120 Figure A10.1: Expenditure v. Income ........................................................... 123 Figure A10.2: Expenditure v. Income, Outliers Removed........................... 124 Figure A10.3: Frequency Distribution of Income v. Expenditure Discrepancies ...................................................................................................................... 124 Figure A10.4: Al Hodeida [Monthly Expenditure Per HES] .......................... 125 Figure A10.5: Frequency Distribution of "Other Expenditure"..................... 127 Figure A10.6: Distribution of Food Expenditure Fractions ........................... 128 Figure A10.7: Food Expenditure as Fraction of Total Household Expenditure ...................................................................................................................... 128 Figure A10.8: Income Distribution, Selected Governorates......................... 130 List of Tables Table A1.1: Total Number of Households in the Survey (Population Frame). 2 Table A1.2: Urban/Rural Categories ................................................................ 3 Table A1.3: Sample Design for Rural Strata ................................................... 4 Table A1.4: Sample Design for Urban Strata.................................................. 5 Table A1.5: Total Number of Households Surveyed....................................... 6 Table A1.6: Distribution of Households by Region and Income Decile............ 7 Table A1.7: Urbanization and Household Size................................................ 8 Table A1.8: Training ....................................................................................... 10 Table A2.1: Summary of Location Characteristics ......................................... 14 Table A2.2: PRA Instruments and Objectives................................................ 16 Table A2.2: PRA Instruments and Objectives................................................ 16 Table A2.3: Total Number of Interviewees ..................................................... 17 Table A3.1: Kerosene Use by Income Decile................................................. 21 Table A3.2: Kerosene End-uses (Liters/HH/month)....................................... 22 Table A3.3: Kerosene Expenditure (For Households Using Kerosene)......... 25 Table A3.4: Monthly Kerosene Expenditure Reported by the PRA................ 26 vi Table A3.5: Monthly Kerosene Expenditure by Income Decile and Governorate.................................................................................................... 27 Table A3.6: Comparison of PRA and 2003 HES Kerosene Expenditures..... 28 Table A3.7: Kerosene Subsidies by Income Decile (2003)............................ 29 Table A3.8: Impact of Bringing Kerosene Price to its Economic Value (YR 36.8/liter)......................................................................................................... 30 Table A3.9: Impact of Reducing Kerosene Subsidies.................................... 30 Table A3.10: Energy Costs Per Meal ............................................................. 31 Table A4.1: Diesel Use by Decile................................................................... 33 Table A4.2: Distribution of Household Energy Consumption by Use, Per Month .............................................................................................................. 34 Table A4.3: Diesel End-uses: Liters Per Household Per Month [In Households Using the Fuel for the Stated Use] ................................................................. 35 Table A4.4: Proportion of Households Reporting Diesel Use ........................ 35 Table A4.5: Monthly Diesel Expenditures (For Households Reporting Diesel Use)................................................................................................................. 38 Table A4.6: Diesel Prices ............................................................................... 39 Table A4.7: Diesel Price Comparison, in US cents/liter ................................. 40 Table A4.8: Diesel Consumption, GDP and Price Differential ....................... 41 Table A4.9: Subsidies by Income Decile........................................................ 43 Table A4.10: Effect of Diesel Price Increases on Direct Diesel Use by Household....................................................................................................... 44 Table A4.11: Impact of Diesel Subsidy Elimination........................................ 45 Table A4.12: Direct and Indirect Impacts of Increasing Diesel to Economic Price, All Households...................................................................................... 47 Table A4.13: Monthly Inflation Rates in 2001................................................. 49 Table A5.1: Expenditure on Purchased Water............................................... 52 Table A5.2: Impact of Diesel Price Increases on Water Intensive Fruit and Vegetables...................................................................................................... 57 Table A6.1: LPG Salient Statistics.................................................................. 60 Table A6.2: 2003 LPG Consumption.............................................................. 61 Table A6.3: Cost per Unit of Energy............................................................... 61 Table A6.4: LPG use by Income Decile.......................................................... 62 Table A6.5: Usage of LPG, as % of Households ........................................... 64 Table A6.6: LPG End-uses............................................................................. 65 Table A6.7: LPG for Cooking.......................................................................... 66 Table A6.8: LPG for Lighting .......................................................................... 67 Table A6.9: Range of Reported Number of Bottles........................................ 70 Table A6.10: Reconciliation of LPG Use (from YGC Data)............................ 71 Table A6.11: LPG Price Structure .................................................................. 73 Table A6.12: Subsidies by Income Decile..................................................... 77 Table A6.13: Effect of LPG Price Increases (on Households) ...................... 78 Table A6.14: Impacts of an LPG Price Increase to 60% of the Economic Price ........................................................................................................................ 81 Table A8.1: Electricity Access ........................................................................ 89 Table A8.2: Electricity Access and Income .................................................... 90 Table A8.3: Hours of Service.......................................................................... 92 vii Table A8.4: Average Monthly Consumption, kWh/HH ................................... 93 Table A8.5: International Comparisons of the First Tariff Block..................... 95 Table A8.6: Monthly Consumption, PEC Grid Customers ............................. 97 Table A8.7: Electricity Consumption and Expenditure Data ........................ 100 Table A8.8: Spending on Electricity Substitutes (all Households) ............... 103 Table A8.9: Incremental Expenditure ........................................................... 103 Table A8.10: Home Business in Rural Households by Electricity Access ... 104 Table A9.1: % of Households using Biomass Fuels..................................... 107 Table A9.2: How do Households Obtain Biomass Fuels, as % of Households Using Fuels?................................................................................................. 108 Table A9.3: Calculated Monthly Fuelwood Usage by Decile and Governorate (1000kg)........................................................................................................ 115 Table A9.4: Uses of Fuelwood...................................................................... 116 Table A9.5: Average Monthly Consumption in Households using Firewood (kg/month)..................................................................................................... 116 Table A9.6: Charcoal Use............................................................................. 118 Table A10.1: Income Deciles........................................................................ 125 Table A10.2: Income and Expenditure by Income Decile ........................... 126 Table A10.3: 1998 Household Expenditure Survey ..................................... 126 Table A10.4: Comparison by Expenditure Categories, All Households....... 127 Table A10.5: Distribution of Households by Income Decile ......................... 129 Table A10.6: Distribution of Households by Region and Income Decile...... 129 Table A10.7: Number of (Weighted) Households in Each Governorate and Decile ............................................................................................................ 131 Table A10.8: As Percent of the Total Households ....................................... 131 Table A10.9: Distribution of the Poor (Defined as the Bottom 30% of all Yemeni Households) .................................................................................... 132 Table A10.10: Share of Poor by Governorate.............................................. 133 Table A10.11: Number of Persons per Household....................................... 133 Table A10.12: Wage Earners in Each Household, By Governorate and Household Income Decile............................................................................. 134 List of Boxes Box A4.1: Comparison with results of 1999 Nationa...................................... 36 Box A4.2: The Economics of Petty Smuggling............................................... 40 Box A5.1: Impact on Diesel Price Increases on Water Sellers ...................... 53 Box A5.2: Impact on Water Prices.................................................................. 54 Box A5.3: Cost of Water Pumping.................................................................. 55 Box A5.4: Assumptions and Calculations of Qat Price Increases.................. 55 Box A5.4: Assumptions and Calculations of Qat Price Increases.................. 56 Box A6.1: The Deepam Scheme in Andhra Pradesh, India........................... 83 Box A7.1: Impact on Fisheries........................................................................ 86 Box A8.1: Inequality of Access ....................................................................... 91 Box A8.2: PEC Tariff....................................................................................... 96 Box A8.3: Costs of Isolated Systems ............................................................. 98 Box A8.3: Costs of Isolated Systems ............................................................. 99 viii Abbreviations and Acronyms ARC Aden Refinery Company BPL below-poverty-line cif cost insurance freight CPC Ceylon Petroleum Corporation CPI Consumer Price Index CSO Central Statistical Organisation DFID Department for International Development, UK ERR economic rate of return ESMAP Energy Sector Management Assistance Programme fob free on board FRR financial rate of return GAREWS General Authority of Rural Electrification and Water Supply GDP Gross domestic product GOY Government of Yemen HH household HBS Household Budget Survey (1998) HES Household Energy Survey (ESMAP 2003) IRR internal rate of return (to equity investors) LPG liquefied petroleum gas LRMC long run marginal cost MDG Millennium Development Goals MoF Ministry of Finance MOM Ministry of Oil and Minerals MOPIC Ministry of Planning and International Cooperation MRC Marib Refining Company NGO Non-government Organization PEC Public Electricity Company PPP purchase power parity PRA Participatory Rapid Assessment (ESMAP 2003) PRSP Poverty Reduction Strategy Program SWF Social Welfare Fund YGC Yemen Gas Company YOGC Yemen Oil and Gas Company YPC Yemen Petroleum Company YR Yemeni Rial ix Annex 1 The Household Energy Survey The Consultation Process in Survey Design A1.1 The Chairman of the CSO requested that the Household Energy Survey (HES) be implemented in co-operation with the CSO. This cooperation took the form of technical advice provided to the local consultants and use of the CSO facilities for data entry. The sampling frame was prepared from a database provided by the CSO. Maps prepared by the CSO were used to define the location of the primary sampling units. The local consultants consulted with Dr. Mohamed Al-Mansoub, the statistics expert in the Poverty Alleviation Unit of MOPIC and with Mr. Bakhbazi (Survey Research Specialist at the CSO) on several occasions to discuss the survey methodology. A1.2 Workshops were organized by MOPIC at which participants from other sector ministries and NGOs were invited to contribute to the design of the HES. They were held on July 17 and 29, 20031. Survey Coverage and Sample Frame A1.3 The survey was conducted in both urban and rural areas of the country covering all governorates except for Al-Jawf, Marib, and Al-Marah for reasons of the expense of surveying in these areas. The population frame was constructed from the 1994 census that contains a list of all villages by governorate, district, sub-district, and urban and rural areas. 1Participants included representatives of the Ministry of Electricity, Public Electricity Corporation, PEC General Dept for Rural Electricity Projects, Ministry of Oil and Minerals, Yemen Gas Corporation, Yemen Petroleum Corporation, the Central Statistical Organisation, Ministry of Planning and International Cooperation (MOPIC), Local Government. 1 2 Household Energy Supply and Use in Yemen Volume 2: Annexes Table A1.1: Total Number of Households in the Survey (Population Frame) Number of Households Name of Governorate Covered in Excluded the Survey From the Survey Ibb 298,662 Abyan 48,086 Sana'a City 140,483 AL-Baida 56,661 Taiz 294,591 AL-Jawf - 38,747 Hajja 155,796 AL-Hodeida 267,641 Hadramout 148,779 Dhamar 211,494 Shabwh 46,671 Saadah 67,901 Sana'a Governorate 154,711 Aden 65,651 Lahj 86,540 Mareb - 22,138 AL-Mahwit 57,990 AL-Mahrah - 8,475 Amran 100,694 Ad-dala 46,822 Total Number of 2,249,173 69,360 Households Source: Census 1994 Sample Size and Sample Design A1.4 The total sample size for the household survey was 3,625 households. In addition to the household interview, another 135 interviews were also conducted with the village head of the sampled village in the rural areas to collect additional village level data. The sample design is based on a stratified two-stage random sampling technique. The stratification is based on the following geographical classification. A1.5 Urban areas Sana'a city - considered to be a major urban area Aden city - considered to be another major urban area Other urban areas - including households in the main city of every governorate except Sana'a and Aden A1.6 Rural areas - grouped into five regions Annex 1: The Household Energy Survey 3 Table A1.2: Urban/Rural Categories Area No. Areas Categories 1 Capital Secretariat Urban __ 2 Aden Urban __ 3 Saddah, Sana'a, Amran Urban Rural 4 Hajja, Hodeida, Dhamar, Urban Rural Mahwait 5 Ibb, Taiz Urban Rural 6 Abyan, Al-Baida, Lahj, Al- Urban Rural Dhala 7 Hadramout, Shabwa Urban Rural A1.7 These classifications were adopted because it is assumed that variability in energy usage among households within the two main and sub-strata is small. Within each stratum, a two-stage random sampling technique was employed. The first stage of sample selection involved selecting villages within each stratum at random. Due to the fact that the list of villages is based on the 1994 census, the total number of households in each village is assumed to be outdated and is recognized as a shortcoming. Therefore, at the second stage, a fixed number of households within each sampled village were randomly selected from a current list of households in the village obtained from village or neighborhood head. A1.8 For the rural strata and urban strata (other than Sana'a and Aden), 15 households per stratum were randomly selected. A larger number of households per selected neighborhood were randomly selected from Sana'a and Aden. This is because it is believed that there is a larger variability of energy usages among households within neighborhoods in major urban areas than in the villages in rural areas. A total number of 20 households were randomly selected from each sampled neighborhood in Sana'a and Aden. A1.9 Given the fixed sampling rate and fixed number of households to be sampled in the final stage and to ensure that every sampled household within each stratum would have an equal chance of being selected ­ each sampled element in the final stage has equal probability of selection method (epsem) property ­ the number of villages (or neighborhoods) sampled from each stratum will vary from stratum to stratum. Using the following formula for sampling rate, the number of villages (or neighborhoods) to be sampled (b) can be solved: f = fa x fb = n/N = (a/A) x (b/B) N = Total number of household within the strata n = Sample size A = Total number of villages within the strata a = Number of villages to be sampled B = Total number of households within the village b = Number of households within the village to be sampled 3 4 Household Energy Supply and Use in Yemen Volume 2: Annexes A1.10 Tables A1.3 and A1.4 show the planned sample design including sample size, sampling rate, total number of village and households expected to be sampled from each sampled village. Table A1.5 shows the actual sample size collected from the field. Table A1.3: Sample Design for Rural Strata Total Number of Planned House- Villages Avg. no. Sample No. of No. of Name of holds of HHs Size Sampling HH Villages Governorate per Rate per Village Village Saadah 60,097 1,139 53 86 14 6 Sana'a 151,898 3,006 51 217 15 14 Amaran 85,540 1,552 55 122 15 8 Region 1 297,535 5,697 425 0.00143 28 Hajjah 141,609 3,709 38 106 15 7 Al Hodiedah 184,783 2,255 82 138 15 9 Dhamar 188,631 3,252 58 141 16 9 Al-Mahwit 54,354 1,176 46 41 14 3 Region 2 569,377 10,392 425 0.00075 28 Ibb 265,008 2,707 98 223 15 15 Taiz 240,168 1,907 126 202 17 12 Region 3 505,176 4,614 425 0.00084 27 Abyan 38,125 2,369 16 72 14 5 Al-Baida 47,961 1,404 34 90 15 6 Lahj 82,939 3,733 22 156 16 10 Adalah 43,738 1,286 34 82 16 5 Region 4 212,763 8,792 400 0.00188 26 Shabwh 41,884 2,842 15 109 16 7 Hadramout 111,172 3,387 33 291 15 19 Region 5 153,056 6,229 400 0.00261 26 Total 2,075 135 Annex 1: The Household Energy Survey 5 Table A1.4: Sample Design for Urban Strata Total Number of Planned House- Urban Avg. Sample No. of No. of Governorate Name holds Villages No. of Size Sampling HH per Villages HH per Rate Village Village Saadah 7,804 15 520 13 13 1 Sana'a 2,813 16 176 5 5 1 Amaran 15,154 20 758 26 13 2 Region 1 25,771 51 505 44 0.00172 4 Hajjah 14,187 27 525 24 12 2 Al Hodiedah 82,858 28 2,959 143 14 10 Dhamar 22,863 9 2,540 39 13 3 Almahwit 3,636 8 455 6 6 1 Region 2 123,544 72 1,716 213 0.00172 16 Ibb 33,654 19 1,771 58 14 4 Taiz 54,423 20 2,721 94 12 8 Region 3 88,077 39 2,258 152 0.00172 12 Abyan 9,961 9 1,107 17 17 1 Al-Baida 8,700 11 791 15 15 1 Lahj 3,601 4 900 6 6 1 Adalah 3,084 4 771 5 3 2 Region 4 25,346 28 905 44 0.00172 5 Shabwh 4,787 7 684 8 8 1 Hadramout 37,607 14 2,686 65 8 8 Region 5 42,394 21 2019 73 0.00172 8 9 Other urban- combined 305,132 211 1446 525 0.00172 46 Sana'a City 140,483 46 3,054 625 0.00445 19 33 Aden (combine all) 65,651 19 3,455 400 0.00609 45 19 5 6 Household Energy Supply and Use in Yemen Volume 2: Annexes Table A1.5: Total Number of Households Surveyed Governorate Urban Rural Total Sana'a City 600 600 Aden City 400 400 Sa'da 14 86 100 Sana'a Gov. 9 212 221 Amran 28 122 150 Region 1 50 422 472 Hajjah 27 100 127 Al-Hudayda 149 138 287 Dhamar 42 140 182 Al-Mahwit 7 41 48 Region 2 225 419 644 Ibb 60 224 284 Taiz 98 202 300 Region 3 158 426 584 Abyan 18 72 90 Al-Bayda 15 91 106 Lahij 6 156 162 Ad-Dala 6 82 88 Region 4 45 401 446 Hadramawt 96 265 361 Shabwa 10 109 119 Region 5 106 374 480 Total 1585 2040 3625 A1.11 Table A1.6 shows the distribution of sample households across income deciles and across governorates. Since each income decile has (roughly) the same number of households, it follows that the sample weights show large variation. For example, there are only 240 households sampled in the lowest income decile, so, on average, one sampled household represents around 1,000 actual households; but in the top decile, one sampled household represents only 500 households. Interpretation of results cross-tabulated by income decile and governorate therefore requires caution, particularly where access rates are small. Annex 1: The Household Energy Survey 7 Table A1.6: Distribution of Households by Region and Income Decile Monthly 0 9,001 12,001 15,001 19,801 22,501 27,001 33,001 42,701 61,001 income -9000 - - - - - - - - > 12,000 15,000 19,800 22,500 27,000 33,000 42,700 61,000 decile bottom D[2] d[3] d[4] d[5] d[6] d[7] d[8] d[9] top YR/month 10% Ibb 68 55 40 16 38 26 14 7 13 6 283 Abyan 4 2 3 4 3 11 8 2 23 30 90 Sana'a City 14 30 31 24 42 49 79 77 111 141 598 Al-Baida 3 7 2 8 15 5 18 21 26 105 Taiz 32 30 32 29 38 30 31 36 21 21 300 Hajjah 10 8 11 15 10 11 23 13 16 15 132 Al-Hodeida 54 79 21 23 27 23 26 13 9 12 287 Hadramout 4 7 20 27 29 52 31 38 51 23 282 Dhamar 10 13 32 22 23 23 20 21 10 8 182 Shabwah 3 2 2 6 10 27 13 19 23 13 118 Sa'adah 1 8 9 13 15 16 13 10 7 8 100 Sana'aGovern 16 13 29 14 24 17 24 21 24 38 220 Aden 8 28 26 29 38 41 51 67 78 33 399 Lahj 2 9 22 25 25 23 17 15 11 11 160 Al Mahweet 5 3 3 6 2 4 6 15 1 2 47 Amaran 9 12 21 13 15 9 15 30 14 11 149 Adelah 3 5 5 6 11 11 14 26 7 88 Total 240 305 314 273 353 388 387 416 459 405 3540 International Comparisons A1.12 The results of the 2003 Yemen HES are compared to the eight countries examined by a 2003 ESMAP study.2 This study examined a diverse set of countries ­ Brazil, Vietnam, South Africa, Ghana, Vietnam, Guatemala, India and Nepal ­ with a view to drawing general conclusions about the determinants of household energy use and fuel substitution patterns. The countries studied did not include any from the Middle East ­ an omission that the Yemen survey can now rectify. A1.13 When making international comparisons, it is proper to note at the outset the main cultural characteristics and differences, and the degree of development and urbanization. Table A1.7 shows how Yemen compares to the other countries in degree of urbanization and household size. 2 ESMAP, Household Energy Use in Developing Countries: A Multi-country study, Washington, October 2003. 7 8 Household Energy Supply and Use in Yemen Volume 2: Annexes Table A1.7: Urbanization and Household Size Urbaniz­ation Persons/HH 2003 HDR Poverty Index for developing countries income decile [%] urban rural all 0-9000 9.4 5.9 5.5 5.6 9001-12000 16.4 6.6 6.2 6.3 12001-15000 14.6 6.9 6.6 6.6 15001-19800 20.3 6.7 7.2 7.1 19801-22500 19.4 6.7 6.7 6.7 22501-27000 25.4 6.9 7.2 7.2 27001-33000 28.9 7.4 8.4 8.1 33001-42700 28.9 7.2 8.9 8.4 42701-61000 32.3 7.6 9.6 9.0 61001>0 33.1 10.0 11.3 10.8 average, Yemen 22.7 7.4 7.6 7.5 40.3 Brazil 80.7 3.7 4.3 3.9 11.8 Nicaragua 56.7 5.2 5.7 5.4 18.3 South Africa 53.3 3.9 5.1 4.5 31.7 Vietnam 24.1 4.4 4.8 4.7 20.0 Guatemala 43.1 4.7 5.7 5.2 22.5 Ghana 36.7 3.9 4.5 4.3 26.0 Nepal 7.3 5.4 5.7 5.7 41.2 India 27.3 4.5 5.0 4.9 31.4 Average 41.2 4.5 5.1 4.8 Source: ESMAP. Poverty Index from 2003 Human Development Report A1.14 Average household size across all income deciles is significantly higher in Yemen than in any other country. Indeed, household size increases with income decile, and there is little difference between urban and rural households in this regard. (As explained further in Annex 2, this correlation of income and family size is largely determined by the number of wage earners per family: the average in the bottom decile is one wage earner per household, while the average in the top decile is 2.7.) The overall average of 7.5 persons per household is significantly greater than the average of the other countries, which explains one of the reasons why energy consumption per household (and in particular LPG consumption) is much greater than elsewhere. Yemen is also less urbanized than most of the comparative countries: only Nepal is less urbanized. Yemen is also among the poorest countries: only Nepal has a comparable level of poverty. Completion of Fieldwork A1.15 Fieldwork was carried out between 7 and 22 December 2003. 3,625 forms were returned to the offices of NHL Consulting by January 27, 2004 where the international and local consultants carried out quality control procedures (e.g. the records of supervisors were used to add missing village identifier codes in the case of a small number of forms). Annex 1: The Household Energy Survey 9 Questionnaire Design and Field Survey A1.16 The draft questionnaire was designed by local consultants with the international survey consultant in an advisory role. A preliminary field test of the questionnaire was carried out by the local and international consultants in the Taiz area between July 13 & 15, 2003. Subsequently further field testing of the survey was integrated into the training program of supervisors and enumerators. The field testing led to modifications in the questionnaire. The main modifications related to phrasing of the questions. The field testing also suggested changes to interview techniques that were incorporated in the instructions for supervisors and enumerators. The finalized questionnaire was transmitted to MOPIC by the local consultants on October 7, 2003. Training of Supervisors and Enumerators A1.17 Supervisors were recruited from Sana'a and Aden universities, the faculty of Hajjah and other third-level educational institutes. They were associate professors, lecturers, researchers and postgraduates. Enumerators were students recruited from the universities. In all 15 supervisors and 116 enumerators were recruited. A1.18 Training documents No. 1 and No. 2 were provided by the local consultants to supervisors and enumerators. (These documents are contained in the project file). Training document No. 1 contains the methodological basis of the survey, a general background of the project, sample size and the contents and characteristics of the questionnaire form. A1.19 Training document No.2 contains a detailed explanation of the following: Objectives of the questionnaire form. Supervisor's tasks. General instructions concerning the form completion. Instructions about how to select households to be surveyed. Statement about how to fill the questionnaire form question by question. Instructions related to certain questions that require explanations. Instructions related to dealing with problems that surface during field implementation and how to overcome them. Instructions about how to make use of the relevant parties and councils. Instructions related to preparation of field reports. Instructions about how to deal with information not required by the form. A1.20 Maps were provided to enumerators and supervisors. A1.21 The training program covered: General background, major objectives and methodology basis of the project. Presentation and discussion of the economic and social characteristics and relation to each of the questionnaire chapters. 9 10 Household Energy Supply and Use in Yemen Volume 2: Annexes Detailed presentation and discussion of the questionnaire form by using modern technology means (PowerPoint). Identification of the locations of the randomly selected villages on maps prepared for this purpose. A1.22 The training consisted of five stages: Preliminary two-days' training for supervisors at the World Bank building with participation of the international survey expert and the local consultant team on July 12, 2003. The training focused on presentation and discussion of the project objectives and the questionnaire form in its preliminary version. All workers of the field survey were trained in Taiz Governorate for one day prior to undertaking the first experimental survey in the presence of the international survey expert. Extensive training for supervisors and surveyors in small groups at different locations by the local consultants. Workshops for each survey group with its supervisor. Final meetings with certain supervisors prior to implementation. Table A1.8: Training Governorate Place of No. of Training Trainees Sana'a + Al-Mahweet + Amran Sana'a 20 Capital Secretariat Sana'a 24 Hadramout + Shabwa Al-Mukalla 14 Aden + Lahj Aden 19 Taiz + Abyan Aden 15 Hajjah + Al-Hodeida Al-Hodeida 17 Ibb + Dhamar + Al-Baida Dhamar 18 Data Processing A1.23 The data entry program was programmed using Microsoft Access software. The Terms of Reference of the local consultant called for the use of an SPSS data entry program. When the local consultant was unable to obtain a copy, the international survey consultant gave permission to use MS Access. The local consultant had prepared a draft program by January 12, 2004. This was tested in trials at the CSO and was further developed by the international consultant. The program was finalized on January 24, 2004. A1.24 The international consultant supervised training of the data entry staff at the CSO on January 24 & 25 2004, and the data entry was done at the CSO laboratory and supervised by designated staff from the CSO. Annex 1: The Household Energy Survey 11 Remarks and Lessons Learned A1.25 As noted above, the sampling frame for the HES is based on the 1994 census. Since 1994, new villages will have been established and the population in villages existing at the time of the 1994 census will have changed. The population of Yemen is thought to have grown from 15 million 1994 to over 19 million at the time of the HES. Any bias introduced into the survey results by not taking account of new villages in the sampling frame will only be significant if energy use by households in the new villages has unique characteristics. The potential bias of using outdated lists of households in villages from the 1994 census was addressed by obtaining a current list of households in the village from the village or neighborhood head. This, rather than the list from the 1994 census, was then used in the second stage of sampling a fixed number of households within each sampled village. A1.26 Another source of bias arises since each income decile contains (roughly) the same number of households and it follows that the sample weights show large variation. For example, there are only 240 households sampled in the lowest income decile, so, on average, one sampled household represents around 1,000 actual households; but in the top decile, one sampled household represents only 500 households. A1.27 Other potential sources of error in the reported data derive from difficulty in ensuring accurate responses from respondents, careful completion of questionnaires by enumerators and punctilious data entry and data cleaning. While there are some aspects, mainly relating to questionnaire design, that would be done differently with the benefit of hindsight, data quality assurance showed that data entry and data cleaning were of a high standard. A1.28 With hindsight, the design of the HES questionnaire and the conduct of the survey could have been improved in the following ways. Respondents had better recall of the expenditure amounts on purchases of electricity and various fuels than of the quantities consumed (very few households were able to show the enumerator their electricity bill). Additional questions on expenditures made by households might have been useful (e.g. by querying last month's expenditures as well as average expenditures per billing period). In addition, questions on energy expenditures could have been included in the expenditure section of the questionnaire as a way of cross-checking energy expenditures in the individual fuel and electricity modules. Questions to probe the seasonality of fuel use may have been useful in the HES as the PRA findings indicated that energy purchases are seasonal with winter spikes of heating fuels in the highlands and increased use of fans during summer in the lowlands as well as increased use during religious festivals. The survey was too long. Questions which turned out to be of peripheral interest included, for example, the purchase of gensets (i.e. amount and payment terms of loan) and on their condition when purchased and characteristics of the household dwelling. It is thought that more reliable 11 12 Household Energy Supply and Use in Yemen Volume 2: Annexes data on quantities of fuels and electricity consumed would be obtained from respondents if the questionnaire were of shorter length and if these questions were towards the front of the questionnaire. Training of field interviewers was carried out over two days in several locations. This may not have been sufficient time for interviewers to fully familiarize themselves with the questionnaire and to gain experience in testing it under field conditions. A larger budget for training would have allowed for more time for all interviewers to test the questionnaire in the field and for the training consultants to fully satisfy themselves as to the proficiency of all interviewers. The PRA that provided qualitative information and the HES that provided quantitative data of household energy use are complementary instruments. The PRA, for example, is a better instrument to explore attitudes to fuel use and the inclusion of attitude questions in the HES is now considered superfluous. It is noted that it is important that the PRA results be written up and fully assimilated prior to undertaking the design of the household survey. The local cost of implementing the HES was approximately $150,000 and for the PRA was approximately $50,000. With a larger budget, a longitudinal HES would have permitted collection of more detailed information from households (for example by more use of electricity bills) and more detailed study of seasonal changes in household energy demand. The survey work was delayed on several occasions because of political events. From January through June 2003, it was not possible for World Bank staff to visit Yemen. In addition, for several months in the run-up to the general election in Yemen in April 2003 it was not possible to carry out survey field work. In planning similar surveys, it is important to take account of the political calendar that may impact on the timetable of implementation. Specialized energy surveys (such as the HES), which are mainly interested in fuels and electricity use for lighting and cooking, have not usually queried gasoline use as it is not normally used within the household for these purposes. Gasoline subsidy removal however will undoubtedly have an impact on household expenditure through the increased cost of gasoline purchased by households and indirectly through the increased cost of transport services used by households. In the case of poor households, it is assumed that the indirect impact would be dominant. Whereas direct use of gasoline by households could have been queried in the HES (and it is recommended that this be done in future surveys of this type), it is not a well adapted survey instrument to probe use of transport services by households. For this, a separate specialized transport survey would have been required. Annex 2 Participatory Rapid Assessment Research Sites and Methodology A2.1 Using a purposive sampling method, twelve sites were selected in coordination with the CSO. The criteria for selecting the sample included the following key indicators that shape energy consumption: geographic variation: (i) Western Coastal Zone, (ii) Eastern Plains, (iii) Highlands, (iv) Southern Plains (v) urban & rural locations level of poverty (range of extremely poor and medium poor areas distributed across the four key geographic zones) electricity availability (areas with and without any grid access and local/cooperative off-grid service) physical accessibility (areas with no road access, dirt roads and paved roads) presence or absence of fossil fuels presence or absence of biomass A2.2 A stratified sample frame: The above criteria were applied to the CSO's sampling frame for the 1998 Household Budget Survey to ensure diversity of national conditions. The sample was stratified on the basis of geographic representation as defined by the government's Strategy for Rural Development. That strategy divided the country into four main geographic areas, based on common socio-economic features, rather than the arbitrary political boundaries of administrative units (i.e. governorates). The areas are Western Coastal Zone, Eastern Coastal Zone, Highlands, Southern Plains. The eastern and southern coasts are lowlands with high temperatures and desert conditions; but some areas also have rich agriculture, mostly fruits and vegetables, fishing and livestock. With mild to cold temperatures, the Highlands are the bread basket of Yemen, renowned for terraced agricultural fields which grow cereals and vegetables. The research areas also included an equal number of sites from the former socialist south and the former north which had developed distinct political economic systems that continue to operate today. Table A2.1 summarizes the characteristics of each of the study sites. A2.3 Within each of these geographic areas the poorest and average income districts were identified using the results of the CSO's 1998 Household Budget Survey. Since the CSO only had data at the district level, the research team then consulted with the 13 14 Household Energy Supply and Use in Yemen Volume 2: Annexes district's Local Council to further identify villages on the basis of the availability of electricity, roads and biomass. The research was conducted in 12 locations (nine rural and three urban) in the four main geographic zones of the country. Table A2.1: Summary of Location Characteristics District and Poverty Electricity Accessibility Petroleum Collected Governorate of the District products Biomass Western Marawa, Poor Not Highly Available Available Coastal HOUDEIDAH available accessible Zone Bajil, Medium Not Medium Available Scarce HOUDEIDAH available accessibility Meena, Medium Grid Highly Available Scarce HOUDEIDAH accessible CITY Eastern Qatn, Medium Grid Medium Available Available Plains HADRAMOUT Poor accessibility Sah, Poor Rural Medium Available Scarce HADRAMOUT cooperativ accessibility e Highlands Utmah, Poor Not Medium Scarce Available DAMAR available accessibility Jabl yal yazid, Medium Not Highly Scarce Available AMRAN available difficult to access Mutheikhra, Poor Not Highly Scarce Available IBB available difficult to access Southern Mozaa, TAIZ Poor Not Highly Scarce Available Plains available difficult to access Musseimeer, Poor Not Medium Scarce Available LAHJ available accessibility Urban Alnuba, LAHJ Poor Grid Medium Available Scarce accessibility Alhuk, LAHJ Poor Not Highly Available Scarce available accessible Annex 2: Participatory Rapid Assessment 15 Figure A2.1: Map of the Study Areas 15 16 Household Energy Supply and Use in Yemen Volume 2: Annexes Table A2.2: PRA Instruments and Objectives PRA Instruments Objectives and expected output 1. Rapport building: meeting the sheikhs, Preliminary community profile established local council members and community Socio-economic groups identified leaders Preliminary identification of energy-related stakeholders 2. Community data sheet Community-wide quantitative information collected, such as population size, number of boys and girls enrolled in schools, type of social services available in the village (schools, health centers etc), number of energy-related service 3. Guided visit of village with key Key community institutions and energy services identified and informant geographically located Socio-economic groups geographically located 4. Community-wide meeting and Community profile finalized stakeholder analysis using a guide prepared Identification of socioeconomic groups finalized for the purpose Profile of community institutions finalized Key stakeholders identified Preliminary assessment of quality of energy services for community service providers (schools, health centers, public institutions, private businesses, etc) 5. Mapping exercise Map of the village drawn by community members identifying key energy related landmarks (energy suppliers, wood fuel collection sites), social services (schools, health centers etc), private businesses using energy (grain mills, shops etc). 6. Semi-structured interviews with males: Energy uses, supply and constraints described by members of the one-on-one interviews with key male local councils, suppliers and users of energy sources. informants identified during stakeholder analysis and using interview guides prepared for the purpose 7. Semi-structured interviews with Gain an in-depth understanding of household energy related behaviors, women: one-on-one interviews from each coping strategies and attitudes. Special attention given to division of of the three social groups, using interview labor within the household for supplying energy sources and its guides prepared for the purpose associated constraints. More detailed assessment of cooking practices and constraints faced at the household level. 8. Men's focus groups: meetings with Gain an in-depth understanding of household energy-related men organized separately for each of the behaviors, coping strategies and attitudes faced by households from a three socioeconomic categories male perspective. Assess quality of energy services for community service providers (schools, health centers, public institutions, private businesses etc) from a male perspective. 9. Women's focus groups: meetings with Gain an in-depth understanding of household energy-related women organized separately for each of behaviors, coping strategies and attitudes faced by households from a the three socioeconomic groups female perspective. Assess quality of energy services for community service providers (schools, health centers, public institutions, private businesses etc) from a male perspective. Annex 2: Participatory Rapid Assessment 17 A2.4 Twelve teams composed of four researchers (two men and two women), led by a Yemeni consulting firm, NHL, were deployed in each of the locations. They conducted focus group and in-depth interviews with both men and women and carried out geographic and poverty mapping exercises, stakeholder analysis and participant observation. Prior to the field work, they tested and refined the instruments. The teams lived in each of the research locations for four days. In each village they held at least one community-wide meeting, where communities defined their notions of well- being and grouped themselves in to three categories: well-off, poor and very poor. Following the initial community-wide meetings, gender segregated focus group interviews were held with each of the three social categories. In-depth interviews were conducted with key men and women informants representing different social categories, sheikhs, community leaders, elected representatives, shop owners, other energy suppliers, teachers, health workers, etc. The study also examined energy use by local institutions, such as health centers and schools, and locally-based private businesses. It also assessed constraints that energy providers and decision-makers faced at the local level in making different energy sources more widely accessible. A2.5 In total, approximately 795 individuals were interviewed in the PRA survey. The large number of interviews, combined with data gathered through the different instruments above and the field teams' own observations, permitted triangulation of information. Table A2.3: Total Number of Interviewees Focus groups with Total community energy users Number of Focus Groups Women Men Total Well-off 108 123 231 12 Poor 115 125 240 12 Very poor 126 115 241 12 Total 349 363 712 36 In-depth interviews Totals Women users of energy services 36 Community leaders 15 Energy service suppliers (shop owners, LPG distributors etc) 32 Total 83 A2.6 Data Analysis: The field data were coded and entered into a Microsoft Access and Excel database which allowed information to be compared by gender, research site and socio-economic group. This report thus presents data which reflect both consistency and variation by gender, research site and socioeconomic group. 17 18 Household Energy Supply and Use in Yemen Volume 2: Annexes Socio-Economic Context and Household Eenergy Use in Yemen Understanding Poverty A2.7 The research places energy use within the social and economic context of the household and also the community within which it is located. An attempt was made to understand the meaning of poverty from the perspective of the people living in the research locations. In nearly all locations there was a consensus that people could be categorized into three wealth or income groups, the "well-off," "poor," and "very poor." Perceptions of the characteristics of these three groups were also remarkably consistent in the 12 locations. These categorizations are used throughout the paper. How Did Communities Define Who is Well-Off? A2.8 In rural areas, households considered well-off typically own their own homes and have land or livestock to rent to others. They also have one or more of the following resources: (i) fixed income through public or private sector employment which provides a predictable and consistent cash flow to households in cash poor areas (ii) a commercial activity (iii) remittances from abroad or (iv) several employed members of the household. Well-off households have assets and a diversity of income sources that enable them to mitigate risks and uncertainties. On average, they constitute 10-20% of the total population of a research site. "Being well-off means shifting completely from wood to LPG and from kerosene to electricity." Poor woman from Nuba How Did Communities Define Who is Poor? A2.9 Those considered poor do not own the house they live in, have limited land ownership and livestock and have no assets to rent out. In some areas, returnees of the 1990 Gulf War who remain unemployed are included in this category. Poor households rely on a single source of income. In the urbanized areas of the Eastern and Southern Plains, low-ranking government employees, especially those without job-related benefits (such as teachers who have no housing allowance), are considered poor. They constitute between 30-60% of a village population in the research sites. How Did Communities Define Who is Very Poor? A2.10 The very poor are utterly destitute and generally depend on others to meet their basic needs. They have no steady source of income. In general, they have at least one or often more of the following attributes: (i) the main provider (generally male) is either dead or disabled; (ii) they are female headed households; (iii) they have no land or livestock; (iv) they work as agricultural sharecroppers; (v) they have no means of production; (vi) they are unskilled; (vii) they are recipients of social security or other alms; (viii) they are elderly with no source of support; (ix) they are daily wage laborers; (x) they are nomads without livestock, homeless or beggars. They represent anywhere between 30-60% of a village population. Annex 2: Participatory Rapid Assessment 19 Implementation of the PRA Field Research: What Worked? A2.11 First, special attention was paid to the composition of the field research team. The importance of ensuring that teams have men and women interviewers cannot be emphasized enough, given significant gender segregation in Yemen. Recruiting women field workers though is a challenge given the country's gender inequalities and the cultural restrictions on women's mobility. Fortunately, there are well- qualified women field workers, but finding them requires a conscious effort. This paid off in the wealth of information that was amassed from women respondents (which could not have been collected by male interviewers due to cultural restrictions on the interaction between men and women). Field workers from the same rough geographic areas were selected, taking Yemeni social relations into account, in order to generate trust and collect reliable information. Second, a one week training and pilot exercise were conducted, in view of the multiplicity of research instruments even though the field teams were generally familiar with PRA methods. This was especially important since the energy issue was entirely new to them. To ensure the collection of valid results, for instance, field workers were trained in encouraging all focus group participants to speak; the teams also learned techniques for politely limiting those who dominate conversations (for example, one technique includes inviting such individuals to the side for one-on-one interviews while the focus group meetings are being conducted). Third, research sites were selected using specific criteria applied to a sample frame for a nationally representative household budget survey with the assistance of the National CSO and further refined through consultations with the district offices of the CSO and local councils. An added lesson from this PRA is that the method can indeed by used as a monitoring and evaluation tool if sites are selected through a combination of statistical methods and purposive sampling to identify localities that are nationally representative. The PRA research instruments can then be used for data collection from which larger inferences can be drawn. Instruments for Field Research A2.12 (i) Field instruments ­ since energy touches upon a wide array of issues (poverty, human development, natural resource management, community cohesion, etc) survey instruments that collect such diverse data need to be designed. The most effective instruments for capturing the energy-related constraints that people face were the case studies using semi-structured interviews with key informants and focusing on a particular energy-related issue. These provided powerful illustrations of coping mechanisms in the face of energy-related constraints. Since they were open ended and one-on-one, they also allowed respondents to tell their personal stories in their own way and encouraged them to speak openly. The interview guides for the focus groups were excessively long and some questions, therefore, went unanswered or were incomplete. In particular, questions dealing with income or the history of the village, designed to identify changes over time that may impinge on energy use, were often only partially answered. In part this was because historical reflection, especially in the south, was perceived as straying into political territory and was therefore sensitive. There is a challenge in designing field instruments that capture diverse data needs yet remain simple and concise. 19 20 Household Energy Supply and Use in Yemen Volume 2: Annexes A2.13 (ii) Training of field staff ­ so they understand the significance of each category of questions and learn to distinguish relevant from irrelevant data. A lot of marginally relevant data were collected and promising information that could have been documented through probing techniques was neglected. A2.14 (iii)Nature of the questions ­ some energy-specific questions need to be deepened: for instance, the PRA has shown that interviewees recognize the negative health impacts of indoor smoke from wood fuel, however, it did not explore if people also realized that it shortens life expectancy. Data Analysis A2.15 The most significant challenge in undertaking the PRA was conducting the data analysis in a way that was rigorous, particularly for data entry and interpretation. Should it be aggregated and disaggregated? Although analysis forms were designed in parallel with the questionnaires, these proved unwieldy and cumbersome when it came to using them for treating field data. For future work, data entry and analysis programs should also be tested as part of the piloting of the field instruments. The analysis focused on information that was triangulated, i.e. data which were found to be consistent across social category and space; consistent across different social categories within the same locality; or consistent within the same social categories but different across space. However, it would have also been important to analyze data that were not confirmed, ie. non-conclusive findings. One such example is the frequent discrepancy in the price, quantity consumed and frequency of use as reported by men and by women. There was discrepancy too in data collected through focus groups and semi-structured interviews with both men and women. These were not significant variations, but a discrepancy nonetheless. The information used in the report is that where there was consistency across different instruments. Annex 3 Kerosene Consumption Patterns A3.1 Kerosene is used primarily by the household sector, and there is little use by industry and commerce. About one third is used for cooking and two-thirds for lighting. However, the expectation that kerosene is used primarily by the poor is not confirmed by the 2003 HES. 92% of the poorest households report kerosene use, but 57% of the richest decile also report kerosene use. Indeed, as shown in Table A3.1, monthly consumption of households using kerosene varies very little across income deciles (between 8 and 11 liters/month)3. Table A3.1: Kerosene Use by Income Decile Income per %HH reporting use Consumption Total Consumption decile [liters/month] [10^6 liters/year] (YR/month) Urban Rural All Urban Rural All Urban Rural All 1 0-9000 79% 94% 92% 8 10 10 1 17 18 2 9001-12000 66% 93% 89% 10 11 11 2 16 18 3 12001-15000 50% 85% 80% 10 11 11 2 16 17 4 15001-19800 56% 90% 83% 6 11 11 1 12 13 5 19801-22500 52% 86% 79% 6 9 8 1 11 12 6 22501-27000 46% 80% 71% 7 10 9 1 11 12 7 27001-33000 52% 80% 72% 9 11 11 2 12 14 8 33001-42700 40% 74% 64% 8 11 10 1 10 11 9 42701-61000 27% 71% 57% 7 11 10 1 9 10 10 61001>0 35% 68% 57% 7 11 10 1 9 10 average 46% 83% 75% 8 10 10 14 121 135 3As in the case of LPG, the quantities reported in the survey significantly exceed the total reported as supplied by YPC. Quantities have therefore been adjusted in order to reconcile with the YPC total, and which is used as the basis for calculating subsidies. If one assumes that the expenditure data is reasonably reliable, the corollary is that reported prices are significantly underestimated. Again a variety of possible explanations could be given, including use of smaller 750ml water bottles sold as "litres". (see Annex 6 for a full discussion of this problem in the case of LPG) 21 22 Household Energy Supply and Use in Yemen Volume 2: Annexes Figure A3.1: Kerosene Use by Income Decile 1 RURAL enesoreKgnisu 0.8 0.6 HH URBAN of onitcarF0.4 0.2 d[1] d[2] d[3] d[4] d[5] d[6] d[7] d[8] d[9] d[10] URBAN 79% 66% 50% 56% 52% 46% 52% 40% 27% 35% RURAL 94% 93% 85% 90% 86% 80% 80% 74% 71% 68% A3.2 As shown in Table A3.2, the key insight is that what varies by income decile is not the quantity of kerosene used, but the percentage of households using kerosene. Thus, for example, 40% of households in the bottom decile use kerosene for cooking, but only 25% of households in the top decile do so. But for those households that do use kerosene for this purpose, monthly consumption increases only from 11 liters/HH/month in the bottom decile, to 13 liters/HH/month in the top decile. Table A3.2: Kerosene End-uses (Liters/HH/month) Cooking & Baking Lighting Home Other Business [%ofHH] [%ofHH] [%ofHH] [%ofHH] bottom Decile 11 [40%] 12 [80%] 2 [0.5%] 3 [1.9%] ALLincome 12 [33%] 12 [59%] 5 [1.0%] 7 [4.6%] top decile 13 [25%] 11 [45%] 32 [0.5%] 12 [4.7%] URBAN bottom Decile 11 [48%] 7 [68%] 2 [4.9%] 1 [8.3%] ALLincome 11 [22%] 8 [32%] 6 [1.8%] 6 [10.6%] top decile 6 [12%] 10 [25%] 32 [1.4%] 9 [10.6%] RURAL bottom Decile 11 [39%] 13 [82%] [0.0%] 4 [1.2%] ALLincome 12 [37%] 12 [67%] 4 [0.8%] 8 [2.9%] top decile 14 [31%] 12 [56%] [0.0%] 19 [1.8%] A3.3 The demand for kerosene is highly seasonal, as shown in Figure A3.2. Kerosene consumption peaks in winter, with a sharp peak in December. Over the past few years, Ramadan has fallen in January (1997, 1998), gradually receding by one Annex 3: Kerosene 23 week or so each year: in 2003, Ramadan was from October 25-November 25, the month immediately before the survey was taken. Estimates of the total annual consumption for 2003 have therefore been adjusted for this seasonal effect. Figure A3.2: Monthly Kerosene Demand 18 m onthly value 16 14 h ontm/serlit 12 lionlim 10 six-m ont h m oving average 8 6 jan 1997 jan 1998 jan 1999 jan 2000 jan 2001 jan 2002 jan 2003 july july july july july july july Source: Ministry of Oil and Minerals, Planning and Statistics Department. The Retail Price of Kerosene A3.4 The PRA reports kerosene retail prices ranging from 20 YR/liter in urban areas and in rural areas with easy access (paved road, close to district capital) to 25-35 YR/liter in less accessible rural areas (with poor roads or far from a district capital), to 30-35 YR/liter during the rainy season in rural areas. The overall national average retail price reported in the 2003 HES is 23 YR/liter, some YR7 more than the nominal ex-YPC price of 16 YR/liter: the most common price is 20 YR/liter (Figure A3.3). Figure A3.3: Frequency Distribution of Retail Kerosene Prices 800 600 uency 400 eqrf 200 0 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 >45 kerosene price, YR/litre 23 24 Household Energy Supply and Use in Yemen Volume 2: Annexes A3.5 The national 2003 HES data reveal no significant difference in retail price paid across the national income deciles, which range from 22 to 24 YR/liter. However, regional differences are more pronounced, as shown in Figure A3.4: Sana'a has the highest at 31 YR/liter, and Shabwah the lowest at 20 YR/liter. Figure A3.4: Average Kerosene Price by Governorate average retail kerosene price 0 10 20 30 40 Ibb 22 Abyan 24 SanaaCity 31 AlBaida 21 Taiz 21 Haja 22 AlHodeidah 22 Hadramouth 29 Dhamar 29 Shabwah 20 Sadah 25 SanaaGovern 23 Aden 27 Lahj 24 Al Mahweet 26 Amaran 22 Adelah 23 A3.6 The hypothesis that kerosene prices increase with increasing distance from paved roads or district capitals is intuitively plausible and was a finding of the PRA survey. However, it is not confirmed by any statistical evidence, as shown in Figures A3.5 and A3.6. The least-squares fit is not statistically significant.4 Figure A3.5: Kerosene Price v. Distance of Village to Nearest Paved Road 60 etr/liecirp 40 enesorek20 0 0 20 40 60 80 100 120 distance to paved road, km 4However, there are problems with the dataset used for these regressions: only half of the rural households sampled have distance data that could be extracted from the village database. Annex 3: Kerosene 25 Figure A3.6: Kerosene Price v. Distance to Nearest District HQ (km) 60 e /litre 40 icrp enesorek20 0 0 20 40 60 80 distance to district HQ Kerosene Expenditure A3.7 As shown in Table A3.3, survey reported kerosene prices show relatively little variation, either by income group or by urban/rural location. Consequently, monthly expenditures closely track consumption (shown in Table A3.1): rural households spend about 33% more than urban households. Both kerosene consumption and expenditure are dependent upon electricity access. The reported average retail price of 24 YR/liter compares to the official ex-distributor price of 17 YR/liter, and reflects local retail markups. Table A3.3: Kerosene Expenditure (For Households Using Kerosene) Income per decile Expenditure [YR/month] Reported price [YR/liter] (YR/month) Urban Rural All Urban Rural All 1 0-9000 282 394 386 24 22 22 2 9001-12000 368 446 436 23 24 24 3 12001-15000 313 433 422 22 23 23 4 15001-19800 233 475 442 20 24 23 5 19801-22500 268 367 354 23 28 27 6 22501-27000 259 397 374 23 24 24 7 27001-33000 360 415 404 23 24 24 8 33001-42700 304 418 398 24 22 23 9 42701-61000 230 377 354 27 25 25 10 61001>0 303 406 384 24 23 23 average, all deciles 297 414 397 23 24 24 A3.8 The PRA reported monthly kerosene expenditures as indicated in Table A3.4 and observed that: "in nearly all geographic districts, poor and very poor households outspent the well-off on monthly kerosene consumption: this is because the well-off use electricity and generator power for lighting." 25 26 Household Energy Supply and Use in Yemen Volume 2: Annexes Table A3.4: Monthly Kerosene Expenditure Reported by the PRA DISTRICT, village Very Poor Well-off poor Western Coastal Zones MARAWA Altour - Rural 500 500 400 BAJIL - Rural Al Kharsha 640 800 500 average 570 650 450 Eastern Plains QATN Batina 800 500 SAH Sah 900 600 300 average 850 550 300 Highlands JABAL YAL YAZID Beit Al 500 500 750 Harethi UTMAH Khiara 200 200 200 MUTHEIKHRA Saaha 500 500 625 average 400 400 525 Southern Plains MOZAA Al Hud 1000 800 700 MUSSEIMEER Mareeb 1080 1400 TIBN AlNuba 1100 ALHUK Al Rabsa 1800 1500 200 MEENA - Alziadiah Al 1800 1500 1350 Shamaliya average 1356 1300 750 Source, PRA. A3.9 However, as suggested by Figure A3.7, the kerosene expenditures reported by the PRA for the Southern Plains are not representative; while there are indeed some households that report monthly kerosene expenditure of 1,500 YR/month, these account for only 1.5% of the sampled households. Figure A3.7: Distribution of Monthly Kerosene Expenditure 400 300 ycneuqefr200 100 0 0 200 400 600 800 1000 1200 1400 1600 1800 100 300 500 700 900 1100 1300 1500 1700 >1850 monthly kerosene expenditure Annex 3: Kerosene 27 A3.10 Table A3.5 shows the monthly kerosene expenditure by income decile and Governorate (for households using kerosene). The average monthly expenditure by decile is fairly flat (386 YR/month in the bottom decile, 384 YR/month in the top decile) Table A3.5: Monthly Kerosene Expenditure by Income Decile and Governorate Income 0 9,001 12,001 15,001 19,801 22,501 27,001 33,001 42,701 61,001 [YR/month] -9000 -12,000 -15,000 -19,800 -22,500 -27,000 -33,000 -42,700 -61,000 > bottom d[2] d[3] d[4] d[5] d[6] d[7] d[8] d[9] top 10% Ibb 262 365 249 342 304 419 323 483 319 963 Abyan 110 224 230 166 80 553 169 250 388 520 Sana'aCity 80 80 78 120 1354 160 63 114 Al-Baida 386 357 320 224 638 113 423 464 458 Taiz 733 528 665 616 435 414 415 429 379 341 Hajjah 254 250 377 327 389 366 455 464 958 470 Al-Hodeida 383 408 395 489 382 430 431 341 296 351 Hadramout 8 101 105 91 76 138 103 137 126 80 Dhamar 306 960 445 377 25 362 492 427 414 352 Shabwah 68 60 110 67 200 191 147 153 130 144 Sa'adah 1300 540 517 549 602 643 898 486 630 496 Sana'aGovern 259 326 299 564 349 316 374 540 315 504 Aden 202 164 288 122 383 201 243 244 168 433 Lahj 231 422 335 480 266 287 298 162 296 261 Al Mahweet 856 509 417 459 750 413 434 570 Amaran 368 588 644 812 364 634 835 628 605 626 Adelah 500 331 331 276 226 455 152 120 Total 386 436 422 442 308 374 404 398 354 384 A3.11 However, there is great regional variation, as shown in Figure A3.8, ranging from as little as 108 YR/month in Hadramout to 620 YR/month in Sa'adah. Figure A3.8: Average Monthly Household Kerosene Expenditure average household kerosene expenditure/month 0 200 400 600 800 Ibb 317 Abyan 399 SanaaCity 211 AlBaida 387 Taiz 534 Haja 431 AlHodeidah 401 Hadramouth 108 Dhamar 429 Shabwah 156 Sadah 620 SanaaGovern 383 Aden 247 Lahj 310 Al Mahweet 500 Amaran 611 Adelah 265 27 28 Household Energy Supply and Use in Yemen Volume 2: Annexes A3.12 Table A3.6 compares the PRA kerosene expenditure results with those of the 2003 HES. The results for the Southern Plains are notably different, with the PRA estimates over twice that of the HES. In two of four cases, the HES shows increasing expenditure with increasing income (Eastern Plains and Highlands) ­ exactly the opposite pattern to that noted in the PRA. Table A3.6: Comparison of PRA and 2003 HES Kerosene Expenditures Very poor Poor Well-off RPA Western Coastal Zone 570 650 450 2003HES: Al Hodeida 396 433 330 RPA: Eastern Plains 850 550 300 2003 HES: Sabwah 79 151 142 RPA: Highlands 400 400 525 2003 HES: Sana'a 294 401 453 Governorate RPA:Southern Plains 1356 1300 750 2003 HES: Taiz 642 470 383 A3.13 At present, there is little incentive to add kerosene to diesel fuel. But experience in other countries shows that where a significant price differential does exist, kerosene is often added to diesel fuel for transportation use. Therefore, if the diesel price is increased and that of kerosene is not, an increasing share of kerosene consumption can be expected to be diverted away from household use. For unelectrified rural households, kerosene is the only source of lighting (other than candles and dry cells): richer households tend to use LPG for non-electric lighting. Who Benefits from the Kerosene Subsidy? A3.14 As shown in Table A3.7, as a percentage of household income, the subsidy is more important to the poorest decile (13% of income) than to the richest (7%). However, the variation across deciles is relatively small (unlike that for diesel, where 43% of the total is captured by the richest decile). Annex 3: Kerosene 29 Table A3.7: Kerosene Subsidies by Income Decile (2003). Income by Subsidy For HH reporting use: decile (YR/month) consumption total captured by subsidy total subsidy each decile expenditure [10^6 liters/year] [10^6 YR] [%] [YR/month] [YR/month] [% of total expenditure] 1 0-9000 18 382 13% 208 8481 2.5% 2 9001-12000 18 377 13% 230 16021 1.4% 3 12001-15000 17 359 13% 224 18389 1.2% 4 15001-19800 13 270 9% 224 21365 1.0% 5 19801-22500 12 248 9% 178 22990 0.8% 6 22501-27000 12 253 9% 194 27591 0.7% 7 27001-33000 15 303 11% 230 32867 0.7% 8 33001-42700 11 233 8% 210 33326 0.6% 9 42701-61000 10 206 7% 209 49248 0.4% 10 61001>0 10 211 7% 206 91469 0.2% Total 137 2842 100% 212 31997 0.7% A3.15 The variation in the corresponding share of total household budget is much greater: the subsidy is equivalent to 2.5% of the total expenditure of the poorest decile, but only 0.2% of the top decile (Figure A3.9). Figure A3.9: Fraction of Total Kerosene Subsidy to Households Captured by Each Income Decile 0.15 0.1 era %total subsidy Sh 0.05 %HH budget 0 d[1] d[2] d[3] d[4] d[5] d[6] d[7] d[8] d[9] d[10] %total subsidy 13% 13% 13% 9% 9% 9% 11% 8% 7% 7% %HH budget 2.5% 1.4% 1.2% 1.0% 0.8% 0.7% 0.7% 0.6% 0.4% 0.2% 29 30 Household Energy Supply and Use in Yemen Volume 2: Annexes Direct Effects A3.16 Table A3.8 shows the impact of bringing the kerosene subsidy to its economic price of YR 36.8/liter). The additional expenditure (assuming no price response, e.g., no switching to fuelwood for cooking) represents 2.5% of total household expenditure for the lowest decile and 0.2% for the richest decile. Table A3.8: Impact of Bringing Kerosene Price to its Economic Value (YR 36.8/liter) Income per Present Additional Total Additional decile (YR expenditure expenditure HH budget expenditure /month) [YR/month] [YR/month] [YR/month] [%] 0-9000 386 208 8179 2.5% 9001-12000 436 230 15592 1.5% 12001-15000 422 224 17882 1.3% 15001-19800 442 224 20869 1.1% 19801-22500 354 178 22427 0.8% 22501-27000 374 194 26976 0.7% 27001-33000 404 230 32166 0.7% 33001-42700 398 210 32751 0.6% 42701-61000 354 209 48479 0.4% 61001>0 384 206 90548 0.2% A3.17 Table A3.9 summarizes the impact of increasing prices to 60%, 80% and 100% of the 2003 economic price, corresponding to increases of 6, 13 and 21 YR/liter respectively. Table A3.9: Impact of Reducing Kerosene Subsidies Price level [% of economic price] 43% 60% 80% 100% average effect on decile Price increase [YR/liter] 0.0 6.1 13.4 20.8 Economic price [YR/liter] 36.8 36.8 36.8 36.8 Retail price [YR/liter] 16.0 22.1 29.4 36.8 Subsidy [YR/liter] 20.8 14.7 7.4 0.0 Consumption Price elasticity -0.2 Consumption [million liters] 137 128 121 116 Impact on Government Subsidy [YRbillion] 2.8 1.9 0.9 0.0 Net gain to government [YRbillion] 1.0 2.0 2.8 Impact on households using Kerosene %HH affecte d Poorest decile [% of total present HH 92.4% 0.0% 0.7% 1.6% 2.5% 2.4% expenditure] Middle decile [% of present 75.3% 0.0% 0.2% 0.5% 0.8% 0.6% expenditure] Richest decile [% of present 57.4% 0.0% 0.1% 0.1% 0.2% 0.1% expenditure] Annex 3: Kerosene 31 A3.18 How would households be likely to respond to increases in the kerosene price? One might assume that the poor would revert to fuelwood for that part of kerosene used for lighting, but at the moment, purchased fuelwood is the most expensive form of cooking fuel, as shown in Table A3.10. LPG has the lowest variable costs per meal, but high upfront fixed costs (for purchase of the cylinder and stove). Table A3.10: Energy Costs Per Meal Cost Number Cost Unit (YR) of per meals meal Logs Hamla (60 kilos) 3,500 30 116 Wood chips Bag full 65 1 65 Kerosene 2 liter 40 1 40 LPG 1 cylinder 280 30 9 Source: Participatory Rapid Assessment Indirect Effects A3.19 In the case of kerosene, one might hypothesize two reasons why not all kerosene consumption goes to households. First, given the extent of subsidy, some quantity of kerosene may be smuggled abroad. Second, it is conceivable that some amount of kerosene is used to dilute transportation fuels. Because the kerosene price is very close to that of diesel (16 YR/liter as opposed to 17 YR/liter for diesel), the incentive to adulterate diesel is small. However, with a gasoline price at 35 YR/liter, in theory the incentive to adulterate gasoline is strong. However, this would cause maintenance (and emission) problems related to the fact that kerosene does not burn as easily, and it is not believed this is practiced in Yemen. The incentive to convert gasoline vehicles to LPG is much greater. 31 Annex 4 Diesel Household Consumption Patterns A4.1 Direct consumption of diesel reported by households in 2003 was 486 million liters per year, or 21.6% of the total (the rest is used in transportation and by commerce and industry). However, only 11% of all households reported direct use of diesel ­ and of the households that do use diesel, 52% are in the two top income deciles. More remarkably, 34% of the households in the top decile use diesel (and 45% in the rural top decile), as opposed to 1-9% of households in the bottom 50% of households. In short, (direct) diesel use by households is sharply concentrated in the well-off households. Table A4.1: Diesel Use by Decile Income per decile %HH reporting Consumption Consumption (YR/month) use [liter/month] [10^6 liters/year] Urban Rural All Urban Rural All Urban Rural All 1 0-9000 4% 3% 3% 172 35 55 2 3 5 2 9001-12000 0% 3% 2% 274 40 43 0 2 3 3 12001-15000 3% 7% 6% 3 119 111 0 21 21 4 15001-19800 1% 11% 9% 71 96 95 0 18 18 5 19801-22500 3% 10% 9% 26 130 124 0 29 29 6 22501-27000 4% 6% 6% 86 98 96 2 13 15 7 27001-33000 2% 17% 13% 45 169 163 1 56 57 8 33001-42700 5% 15% 12% 143 203 196 5 57 62 9 42701-61000 6% 23% 18% 184 175 176 10 74 84 10 61001>0 10% 45% 34% 185 217 214 16 176 192 average 4% 13% 11% 139 167 164 38 448 486 33 34 Household Energy Supply and Use in Yemen Volume 2: Annexes Figure A4.1: Proportion of Households using Diesel 0.5 RURAL 0.4 leseiD ing 0.3 us HH ofnoitcarF0.2 URBAN 0.1 0 d[1] d[2] d[3] d[4] d[5] d[6] d[7] d[8] d[9] d[10] URBAN 4% 0% 3% 1% 3% 4% 2% 5% 6% 10% RURAL 3% 3% 7% 11% 10% 6% 17% 15% 23% 45% A4.2 The largest reported use of diesel is for agriculture: 22.1 million liters of the total of 45 million liters per month is used for this purpose (Table A4.2). Diesel for self-generation is the second largest use, accounting for 14 million liters/month. Table A4.2: Distribution of Household Energy Consumption by Use, Per Month Purchased Charcoal Electricity LPG kerosene gasoline diesel Firewood 1000t 1000t GWh 1000t 10^6 10^6 10^6 liters liters liters Cooking & 48 1 53 9 Baking Lighting 6 16 Fridge 0 Ironing 0 Space heat 7 1 Water Heat Electric 140 3 14 appliances Home Business 0 0 0 1 Agriculture 22 Water Pumping 1 Transport 3 Other 1 3 1 4 Total 56 5 140 59 25 3 45 Note: diesel and gasoline use for "appliances" represents fuel for self-generation A4.3 Indeed, this use of diesel for agriculture is highly concentrated in the two top income deciles (Table A4.3): for example, in rural areas, 29.8% of households in the top decile report diesel use for agriculture (and who use 200 liters a month for this purpose), as opposed to only 1.7% of rural households in the poorest decile (who use only 25 liters a month). The rationale that diesel requires subsidy in order to assist agriculture and farmers may be correct, but in fact this benefits well-off farmers, not poor farmers. Annex 4: Diesel 35 Table A4.3: Diesel End-uses: Liters Per Household Per Month [In Households Using the Fuel for the Stated Use] Appliances Home Agriculture Water Transport Other Business Pumping(a) [%ofHH] [%ofHH] [%ofHH] [%ofHH] [%ofHH] [%ofHH] bottom Decile 207 [0.4%] [0.0%] 25 [1.6%] [0.0%] 55 [0.1%] 47 [1.0%] ALL income 218 [2.8%] 105 [0.5%] 143 [6.9%] 45 [0.7%] 80 [1.8%] 98 [1.9%] top decile 284 [8.9%] 97 [3.0%] 202 [21.8%] 61 [2.6%] 85 [9.4%] 85 [5.6%] URBAN bottom Decile 207 [4.4%] [0.0%] [0.0%] [0.0%] 40 [0.1%] [0.0%] ALLincome 120 [0.9%] 224 [0.3%] 189 [1.6%] 21 [0.2%] 66 [1.0%] 60 [1.4%] top decile 60 [1.0%] [0.0%] 261 [5.5%] 7 [0.7%] 72 [2.8%] 61 [2.4%] RURAL bottom Decile [0.0%] [0.0%] 25 [1.7%] [0.0%] 57 [0.1%] 47 [1.1%] ALLincome 226 [3.3%] 82 [0.5%] 140 [8.5%] 47 [0.8%] 82 [2.0%] 106 [2.0%] top decile 292 [12.8%] 97 [4.4%] 197 [29.8%] 66 [3.6%] 87 [12.7%] 89 [7.2%] Note: (a) water pumping for domestic water supply only (pumping for irrigation purposes is shown under "agriculture") A4.4 Table A4.4 shows the proportion of households reporting diesel consumption for the various categories. 6.9% of all households report diesel use for agriculture, but only 2.8 for self-generation. Table A4.4: Proportion of Households Reporting Diesel Use Transportation Water Agriculture Business Other Self- Any pumping generation diesel use 1 0-9000 0.1% 0.0% 1.6% 0.0% 1.0% 0.4% 3.0% 2 9001-12000 0.0% 0.1% 1.7% 0.0% 0.3% 0.0% 2.1% 3 12001-15000 0.6% 0.1% 3.5% 0.0% 2.9% 1.8% 6.4% 4 15001-19800 0.0% 0.2% 6.5% 0.5% 1.6% 1.4% 8.7% 5 19801-22500 0.5% 0.2% 5.0% 0.5% 2.3% 2.9% 8.7% 6 22501-27000 0.9% 0.1% 4.2% 0.3% 0.7% 0.6% 5.8% 7 27001-33000 1.5% 0.9% 7.0% 0.1% 1.2% 4.1% 12.9% 8 33001-42700 2.0% 1.2% 6.6% 0.0% 1.4% 5.0% 11.9% 9 42701-61000 3.1% 1.5% 12.2% 0.3% 1.9% 3.0% 17.6% 10 61001>0 9.4% 2.6% 21.8% 3.0% 5.6% 8.9% 33.6% average 1.8% 0.7% 6.9% 0.5% 1.9% 2.8% 11.0% 35 36 Household Energy Supply and Use in Yemen Volume 2: Annexes Box A4.1: Comparison with results of 1999 Nationa The 1999 National Poverty Survey estimated the % of households that had access to diesel and who irrigated with an artesian well (as cited in the Dec 2002 World Bank Poverty Update). Income per HH reporting HH reporting Total HH decile (YR / diesel use for diesel use for reporting diesel month) self-gen Agriculture use Rural urban rural urban rural urban 1 0-9000 0.0% 4.4% 1.7% 2.8% 2 9001-12000 0.0% 0.1% 3.2% 0.1% 3.6% 3 12001-15000 2.1% 0.0% 4.1% 0.0% 7.0% 3.0% 4 15001-19800 1.6% 0.5% 7.9% 0.9% 10.6% 1.4% 5 19801-22500 3.6% 0.2% 6.1% 0.5% 10.1% 3.9% 6 22501-27000 0.8% 0.1% 5.3% 0.8% 6.3% 4.2% 7 27001-33000 5.7% 0.0% 9.8% 17.3% 2.3% 8 33001-42700 6.2% 1.9% 8.4% 2.2% 14.7% 4.8% 9 42701-61000 3.5% 2.0% 17.0% 2.1% 23.4% 6.8% 10 61001>0 12.8% 1.0% 30.1% 5.5% 45.6% 9.9% average 3.3% 0.9% 8.6% 1.6% 13.1% 4.6% 1999 NPS 2.4% 0.4% 9.7% 2.8% 13.9% 6.5% 0.5 2003 HES 0.4 leseiD ngi 0.3 us HHfo 0.2 onitcarF 0.1 1999NPS 0 d[1] d[2] d[3] d[4] d[5] d[6] d[7] d[8] d[9] d[10] 1999NPS 6.4% 7.7% 9.2% 11.9% 11.1% 13.0% 15.8% 17.8% 21.9% 28.1% 2003 HES 2.8% 3.6% 7.0% 10.6% 10.1% 6.3% 17.3% 14.7% 23.4% 45.6% A4.5 Three governorates show a large percentage of households using diesel: Sana'a Governorate, Sa'adah and Adelah (Figure A4.2). Annex 4: Diesel 37 Figure A4.2: Percentage of Households in Each Governorate Reporting Diesel Use fraction of HH using diesel 0 0.1 0.2 0.3 0.4 0.5 Ibb 4% Abyan 8% SanaaCity 2% AlBaida 10% Taiz 14% Haja 11% AlHodeidah 4% Hadramouth 9% Dhamar 12% Shabwah 9% Sadah 30% SanaaGovern 41% Aden 4% Lahj 0% Al Mahweet 0% Amaran 12% Adelah 32% A4.6 Diesel demand also shows regular seasonal variation, with the annual peak observed in July (Figure A4.3). Figure A4.3: Monthly Diesel Consumption (All Uses, Including PEC and Non Households) 250 200 h ontm/sertil 150 onillim 100 50 jan 1997 jan 1998 jan 1999 jan 2000 jan 2001 jan 2002 jan 2003 july july july july july july july 37 38 Household Energy Supply and Use in Yemen Volume 2: Annexes Diesel Expenditure A4.7 Table A4.5 shows monthly expenditure on diesel fuel (for households reporting diesel use). Rural expenditure is significantly greater. Annex 8 discusses further diesel expenditures for self-generation. Table A4.5: Monthly Diesel Expenditures (For Households Reporting Diesel Use) Income per %HH reporting use Expenditure, decile (YR / [YR/month] month) Urban Rural All Urban Rural All 1 0-9000 4% 3% 3% 2 9001-12000 0% 3% 2% 3 12001-15000 3% 7% 6% 2337 2186 4 15001-19800 1% 11% 9% 2193 2202 5 19801-22500 3% 10% 9% 2957 2814 6 22501-27000 4% 6% 6% 1832 1912 7 27001-33000 2% 17% 13% 3389 3259 8 33001-42700 5% 15% 12% 2651 4101 3880 9 42701-61000 6% 23% 18% 3142 6453 6066 10 61001>0 10% 45% 34% 2930 4262 4107 average, all 4% 13% 11% 2533 3779 3652 deciles Note: expenditure where less than 5% of HH report diesel use is not reliable, and is omitted The Retail Price of Diesel A4.8 With a wide network of filling stations where diesel is sold at the official price of 17 YR/liter throughout the country, reported retail diesel prices vary little: 70% of respondents report a price of between 17 and 20 YR/liter (Figure A4.4). Where there are no gas stations, diesel is sold in village shops in small quantities at higher prices. Figure A4.4: Distribution of Reported Diesel Prices, YR/liter 100 80 yc 60 equenrf 40 20 0 10 14 18 22 26 30 34 38 42 46 12 16 20 24 28 32 36 40 44 >1850 retail price of diesel A4.9 Figure A4.5 shows the average price of diesel by governorate. Abyan and Dhamar appear to have an unusually high average price to households (but with few Annex 4: Diesel 39 households reporting diesel use in these provinces, this may not be not statistically significant). Figure A4.5: Average Price of Diesel, by Governorate average diesel price YR/litre 0 10 20 30 40 Ibb 19.4 Abyan 35.3 SanaaCity 20.7 AlBaida 24.3 Taiz 25.5 Haja 17.6 AlHodeidah 18.8 Hadramouth 17.3 Dhamar 29.9 Shabwah 26.1 Sadah 26.2 SanaaGovern 17.8 Aden 20.9 Lahj Al Mahweet Amaran 17.4 Adelah 18.7 A4.10 Nevertheless, as shown in Table A4.6, such variation as is observed in diesel prices is income dependent: the bottom quintile reports average purchase prices of 23 YR/liter, as opposed to YR18 in the top quintile (or 19 YR/liter overall). Table A4.6: Diesel Prices Income per decile (YR /month) Reported price [YR/liter] Urban Rural All 1 0-9000 18 23 23 2 9001-12000 13 23 23 3 12001-15000 24 19 20 4 15001-19800 25 20 20 5 19801-22500 19 21 21 6 22501-27000 20 21 20 7 27001-33000 16 19 19 8 33001-42700 15 16 16 9 42701-61000 16 19 18 10 61001>0 19 18 18 average, all deciles 18 19 19 Diesel Smuggling A4.11 It is widely reported that considerable quantities of diesel (and LPG) are smuggled to neighboring countries, notably across the Red Sea, motivated by the sharp difference between the domestic price in Yemen and the international price. Estimates of as much as 30% of total diesel consumption being smuggled have been quoted. As shown in Table A4.7, Yemen diesel prices are substantially below those of neighboring countries, so there is substantial incentive for smuggling. A4.12 However, as suggested in Box A4.2, while the economics of petty smuggling suggest significant incentives, it is hard to see how this could amount to 30%, which, given 2003 consumption of 2,260 million liters of diesel, implies some 700 million liters, or 670,000 tons. 39 40 Household Energy Supply and Use in Yemen Volume 2: Annexes Table A4.7: Diesel Price Comparison, in US cents/liter 1998 2000 2002 Yemen 7 6 10 Eritrea 23 33 25 Somalia(a) Djibouti 40 53 54 Saudi Arabia 10 10 10 UAE 18 26 30 Oman 29 29 26 Source: World Bank Development Indicators, 2002; GTZ (a) No official information; according to press reports, diesel prices in Mogadishu are around 30 US cents/liter, but supply is erratic (mainly by tanker from UAE to the port at El-Ma'an, and often reach 80 US cents to $1/liter). Smuggling is also reported from Djibouti (where retail prices are high) into northern areas of Somalia. It is a reasonable assumption that Somali is a substantial market for smuggled fuels. Box A4.2: The Economics of Petty Smuggling While smuggling in the face of large differentials between domestic and international prices is plausible, the evidence for the magnitudes involved appears to be entirely anecdotal: no reliable quantitative study of the subject has been found. Smuggling must either involve an exchanged commodity such as fish or cattle or dollars. The economic consequences of smuggling can be illustrated by simple example. The economic cost of diesel is 37 YR/liter, but can be bought by a smuggler in Yemen for 17 YR/liter. The smuggler's transportation cost to bring the diesel to a neighboring country is, say, 2 YR/liter (which is spent in Yemen). Assume the retail price of diesel at a coastal location in the neighboring country is also (the equivalent of) 37 YR/liter. It is reasonable to suppose that the available rent (37-(17+2)= 18) is shared based on a negotiation that is based on an allocation of the risks to the parties involved: assume that 4 accrues to the party in the neighboring country, and 14 to the Yemeni smuggler. The diesel could be exchanged for a commodity with a market value in the neighboring country of 33 YR, that can be sold back in Yemeni markets for the same price; the smuggler's expenses for bringing back the produce to the Yemeni market is 1YR (spent in Yemen). The winners and losers in this transaction are as follows: · Government of Yemen: -20 YR/liter (cost of subsidy) · Neighboring country party: +4YR/liter (obtains diesel worth 37 YR/liter for 33 YR/liter) · Economy of Yemen +3 YR/liter (smuggler's expenses) · Yemeni smuggler +13 YR/liter (revenue from produce sale 33 YR/liter, less costs of 17+2+1=20 YR/liter) Unless the diesel is sold for dollars and deposited in an offshore banking account, the likelihood is that the Yemini smuggler's surplus is spent in the Yemeni economy. Therefore, the only actual economic loss to the Yemeni economy as a whole is that (probably relatively small) share of the rent that accrues to the party in the neighboring country: the main effect is a transfer payment from the government to the smuggler (and the sectors that benefit from the smuggler's expenditures), and some (small) leakage to the foreign party. A4.13 It has been suggested by other commentators that high growth rates of diesel consumption in Yemen, substantially in excess of the growth rate in GDP, support the proposition of large quantities of diesel smuggling. Figure A4.6 also supports this proposition. Annex 4: Diesel 41 Figure A4.6: Diesel Consumption and GDP 200 180 DIESEL consumption 160 001 140 97= 19 GDP 120 100 80 1996 1998 2000 2002 2004 A4.14 One needs to exercise great care when drawing such conclusions. First, one of the main reasons why diesel consumption has increased sharply over the past few years is increased consumption by PEC for electricity generation (as well as diesel purchases by industries and households for self-generation). As shown in Table A4.8, when PEC consumption is subtracted out, the 2002 increase in diesel consumption is 16%. Table A4.8: Diesel Consumption, GDP and Price Differential 1997 1998 1999 2000 2001 2002 2003 Total diesel consumption [10^6 liters] 1271 1310 1458 1689 1891 2222 2474 Growth rate [ ] 3% 11% 16% 12% 18% 11% PEC consumption [10^6 liters] 60 127 203 257 334 Non-PEC consumption [10^6 liters] 1271 1310 1398 1562 1688 1965 2140 Growth rate [ ] 3% 7% 12% 8% 16% 9% Differential (economic price-actual price) [YR/liter] 11.9 5.4 10.9 18.2 15.4 17.3 22.8 Change [YR/liter] -6.6 5.6 7.3 -2.8 1.9 5.5 As % change [ ] -55% 104% 67% -15% 12% 32% Non-oil real GDP growth rate [ ] 6.2% 5.8% 2.7% 3.5% 5.2% 4.6% 4.0% A4.15 Second, the main conclusion of Figure A4.6 is not necessarily that the difference in diesel consumption and GDP is driven by smuggling. Indeed, many countries at Yemen's stage of economic development (and particularly rapid growth in road traffic) exhibit high income elasticities for diesel consumption, and one would expect that diesel consumption grows faster than GDP growth. While it is true that a large price differential between border price and domestic price is an incentive for diesel smuggling, a falling real (domestic) diesel price also induces additional diesel consumption. Indeed, when real price and price differential are added to Figure A4.6, as shown in Figure A4.7, a very different picture emerges: the falling real price of diesel has driven diesel consumption just as much as increasing price differential. 41 42 Household Energy Supply and Use in Yemen Volume 2: Annexes Figure A4.7: Diesel Consumption v. GDP and Diesel Prices 250 200 CONSUMPTION 0 150 10 7= GDP 199 100 PRICE 50 Incentive 0 1996 1998 2000 2002 2004 A4.16 With so short a time series, one may be reluctant to read too much into a simple regression model. Nevertheless, when one estimates the model Diesel =k GDPa PRICEb PXc the income elasticity calculates as 2.4, the own-price elasticity as ­0.2, and the elasticity with respect to the price differential as 0.1 (all statistically significant, signs and magnitudes of elasticities very much as expected, overall R2=0.99). Based on this (admittedly simplistic) model, when one sets PX=0, i.e., if the border and retail price were equal and thus no longer an incentive for smuggling, then the 2003 consumption reduces from 2,474 million liters to 2,300 million liters, a reduction of 7%. This reduction of 173 million liters calculates to one million barrels/year, or 145,000 tons per year. A4.17 Smuggling on a larger scale is also plausible. Assuming a full load of diesel, a 100 A1 oil tanker with six oil tanks totaling a 2,760 DWT vessel could move about 18,000 bbls or 2.8 million liters. The gross margin (difference between Yemen domestic and international price) is 22.8 YR/liter (see Table 5.9, Volume 1), for a total of YR62 million, or $320,000. To account for the entire 7% estimate of smuggled diesel of 145,000 tons, 56 loads in this type of tanker would be required, generating a potential gross margin of 56 x $320,000 = $19 million. Smuggling on this scale seems both plausible and profitable. Reducing the Diesel Subsidy: Direct Effects A4.18 Experience from other countries indicates that when diesel is so far below its opportunity cost, the consequences include loss of tax revenues to the Government, smuggling and corruption, and wasteful consumption. In addition, in Yemen under- pricing of diesel leads to over-pumping of groundwater for water-intensive crops (fruit, vegetables and Qat) and depletion of aquifers, whose costs are borne by all water users. A4.19 Households would experience the effect of a rise in diesel price in two ways. The first is the direct effect of higher prices on the diesel fuel purchased by households. The second is the indirect effect of higher diesel prices on other household expenditures: for example, since diesel is used for irrigation pumping and Annex 4: Diesel 43 transport of food, food prices would increase as producers and distributors pass on the cost of their higher diesel prices. Similarly, where diesel is used for power generation, the cost of electricity would also increase. A4.20 Table A4.9 and Figure A4.8 show subsidies by income decile. 57% of the total subsidy (associated with direct use) is captured by the two top deciles: the bottom three deciles capture 6% of the total ­ a simple reflection of the low direct diesel use in the low deciles. Table A4.9: Subsidies by Income Decile. Subsidy for HH reporting use: consumption total captured by subsidy Total subsidy each decile expenditure Income [10^6 liters/year] [10^6 YR] [%] [YR/month] [YR/month] [% of total decile (YR / expenditure] month) 1 0-9000 5 93 1% 1114 8481 13.1% 2 9001-12000 3 52 1% 870 16021 5.4% 3 12001-15000 21 427 4% 2247 18389 12.2% 4 15001-19800 18 375 4% 1935 21365 9.1% 5 19801-22500 29 597 6% 2507 22990 10.9% 6 22501-27000 15 305 3% 1952 27591 7.1% 7 27001-33000 57 1147 12% 3304 32867 10.1% 8 33001-42700 62 1266 13% 3982 33326 11.9% 9 42701-61000 84 1699 17% 3581 49248 7.3% 10 61001>0 192 3899 40% 4335 91469 4.7% total 486 9861 100% 3331 31997 10.4% Figure A4.8: Fraction of Total Diesel Subsidy to Households, Captured by Each Income Decile 0.5 %total subsidy 0.4 0.3 era Sh 0.2 0.1 %HH budget 0 d[1] d[2] d[3] d[4] d[5] d[6] d[7] d[8] d[9] d[10] %total subsidy 1% 1% 4% 4% 6% 3% 12% 13% 17% 40% %HH budget 13.1% 5.4% 12.2% 9.1% 10.9% 7.1% 10.1% 11.9% 7.3% 4.7% A4.21 Table A4.10 shows the direct effect of raising the diesel price to 60%, 80% and 100% of the economic price. For example, if raised to the economic price, the effect on diesel users in the poorest decile is 13.6%, and only 4.8% in the richest decile. However, since only 3% of households in the poorest decile use diesel, the average across all households is only 0.4%, while for the richest decile it is 1.6%. 43 44 Household Energy Supply and Use in Yemen Volume 2: Annexes Table A4.10: Effect of Diesel Price Increases on Direct Diesel Use by Household Price level [% of economic 46% 60% 80% 100% Average effect price] on decile Price increase [YR/liter] 0.0 5.4 12.8 20.3 Economic price [YR/liter] 37.3 37.3 37.3 37.3 Retail price [YR/liter] 17.0 22.4 29.8 37.3 Subsidy [YR/liter] 20.3 14.9 7.5 0.0 Consumption Price elasticity -0.1 Consumption [million liters] 486 473 459 449 Impact on Government Subsidy [YRbillion] 9.9 7.1 3.4 0.0 Net gain to government [YRbillion] 2.8 6.4 9.9 Impact on households using Diesel %HH affected Poorest decile [% of total present 3.0% 0.0% 3.6% 8.6% 13.6% 0.4% HH expenditure] Middle decile [% of present 7.3% 0.0% 2.4% 5.8% 9.2% 0.7% expenditure] Richest decile [% of present 33.6% 0.0% 1.3% 3.0% 4.8% 1.6% expenditure] Reducing Diesel Subsidies: Indirect Effects A4.22 The main concern of poor households regarding indirect effects of increasing diesel prices is the potential effect on food prices, given the role of diesel fuel for food production (water pumping), and food transportation (diesel trucks). Similarly great concern has been expressed that the poor would be hit by higher costs of purchased water (again because of diesel used for pumping groundwater, and diesel fuel used in bowsers operated by water sellers). The PRA notes, for example, that Respondents expressed strong opposition to paying higher diesel prices, even those who do not purchase it. They fear that an increase in diesel prices will mean an increase in the cost of transporting basic goods. Already, remote areas have significantly higher prices largely attributable to transportation costs. Respondents explained that increasing diesel prices would elicit a stronger negative public reaction than the recent increase in bread due to the far-reaching implications of diesel consumption. A4.23 It is noted (Annex 10) that the poor devote a much higher proportion of their income to food than the non-poor, so if indeed food prices rise as a consequence of higher diesel prices, the poor would be affected disproportionately: food accounts for 54% of the household expenditure in the poorest decile, but only 36% of the richest decile. A4.24 Ideally, to estimate the indirect effect of a diesel price increase on the CPI requires an input-output table. This is not available for Yemen, and therefore it is only possible to estimate an upper bound of the potential impact based on certain Annex 4: Diesel 45 assumptions that are conservative from the standpoint of estimating distributional impacts. A4.25 The calculations are shown in Table A4.11. Of the total 2003 diesel consumption, 334 million liters was used by PEC, and 486 million liters was purchased directly by households; and for the moment assume no diesel is smuggled. The remaining quantity of diesel will be used by a variety of sectors, but in the absence of an I/O table and lacking information about how much diesel is used in transportation, by industry, by Government, etc., the most conservative assumption that can be made with regard to the potential impact on the poor is that the entire residual diesel consumption, i.e. 1,655 million liters, goes into the production, transportation and distribution of food and water. Since the poor consume a much higher fraction of their income on food than the non-poor and are particularly concerned about the potential impact on water costs, this is the most conservative assumption one can make. Thus, under this assumption, the sector food and water would incur increased costs of YR33.6 billion. Table A4.11: Impact of Diesel Subsidy Elimination Diesel expenditure Baseline expenditure Price increase quantity current incremental [million liters] [YR billion] [YR billion] [YR billion] [%] Total 2475 42.1 50.2 PEC consumption 334 5.7 6.8 49.3 13.8% Smuggled 0 0.0 0.0 Direct use by HH 486 8.3 9.9 119.4% Export goods 0.0 0.0 Government consumption 0.0 0.0 Investment 0.0 0.0 Non-HH private consumption 0.0 0.0 All other items in CPI 0.0 0.0 Food&Water diesel 1655 28.1 33.6 electricity 3.0 total 36.6 417 8.8% A4.26 With regard to diesel use by PEC, elimination of the diesel subsidy would increase PEC costs by YR6.8 billion. PEC's total costs in 2003 were YR49.3 billion, so if the diesel price increase were passed to consumers (as a fuel adjustment in the tariff), its price would increase by 13.8% (assuming no adjustment of electricity consumption by consumers in the face of a price increase, or fuel switching by PEC, as discussed below). A4.27 Some of this electricity is purchased directly by consumers ­ in 2002, households consumed 55% of electricity sold by PEC.5 Therefore, assuming all 5 In 2002, the distribution of PEC sales by sector was as follows GWh Domestic 1380 55.8% Commercial 382 15.4% Industrial 281 11.4% Agricultural 39 1.6% Government 365 14.8% Mosques 27 1.1% Total 2474 100.0% 45 46 Household Energy Supply and Use in Yemen Volume 2: Annexes tariffs are equally adjusted, the remaining 45% is sold to the rest of the economy (which is here assumed to be just food). Therefore, the food and water sector faces price increases of YR28.8 million due to more expensive diesel, and YR3.0 billion due to more expensive electricity, for a total of YR36.6 million. From the 2003 household survey, total food and water expenses amounted to YR417 billion, and therefore the food prices would increase by 8.8%. A4.28 As noted, this is a very pessimistic calculation from the standpoint of the poor, and therefore should be interpreted as an upper bound of the potential increase in food and water prices. First, one of the main reasons why PEC uses diesel rather than fueloil is price: in 2002, PEC paid 17 YR/liter for diesel, but 28.50 YR/liter for fueloil, so were diesel to rise to more than 28.50 YR/liter, some fuel switching would be likely to occur. Moreover, with power generation switching to gas, the importance of diesel to generation costs will in any event decline. A4.29 Second, the calculation shown in Table A4.11 assumes that no diesel is smuggled. As shown in Figure A4.9, the greater the fraction of diesel that is now smuggled, the smaller the impact on food prices: if 7% were smuggled (as estimated above), then the price increase would fall from 8.8% to 7.9%. Figure A4.9: Impact of Diesel Smuggling on Food Price Increases 9 seic 8 study estimate prretaw 7 ood&f inesaer 6 nci% 5 4 0 10 20 30 40 percentage of total diesel consumption smuggled A4.30 With the estimates of price increases of Table A4.11 in hand, one may then estimate the overall impact of removal of the diesel subsidy on households. As shown in Table A4.12, the distribution across deciles is remarkably flat ­ at least under the assumptions made here, notably that the entire indirect impact is concentrated in food. Annex 4: Diesel 47 Table A4.12: Direct and Indirect Impacts of Increasing Diesel to Economic Price, All Households Indirect effects Direct effects Income per Food Electricity Diesel Diesel Self-gen Total increase decile (YR/month) Total HH present increase present increase present increase present increase as % of expenditure 8.8% 13.8% 119.4% 119% total E 0-9000 8179 4650 408 92 13 18 21 14 17 460 5.6% 9001-12000 15592 8342 732 223 31 26 31 1 1 795 5.1% 12001-15000 17882 9025 792 240 33 120 144 21 25 994 5.6% 15001-19800 20869 9879 867 400 55 150 180 63 76 1177 5.6% 19801-22500 22427 11456 1005 346 48 176 210 82 98 1360 6.1% 22501-27000 26976 13650 1198 575 79 116 138 7 9 1424 5.3% 27001-33000 32166 14583 1280 736 101 273 326 173 207 1914 5.9% 33001-42700 32751 15061 1322 577 79 283 338 221 264 2003 6.1% 42701-61000 48479 19739 1732 985 136 586 700 589 703 3271 6.7% 61001>0 90548 32641 2864 1375 189 1252 1495 220 263 4812 5.3% total 32401 13843 1215 550 76 298 355 138 165 1811 5.6% Impact of Past Increases in the Diesel Price A4.31 The main report examines the impact of the diesel price increases in 2001 by looking at trends in the overall monthly CPI and aggregate expenditures, and concludes that the likely impact on the CPI was about 2%. The small impact of past diesel price increases can also be confirmed by examination of the quarterly averages and by examining the 1997 price increase as well. A4.32 As shown in Figure A4.10, in the case of the first diesel price increase in mid- October 1997, the quarterly inflation rate immediately preceding the diesel price increase showed an increasing trend, from -0.5% in the first quarter of 1997 to 2.6% in the fourth quarter. In the fourth quarter, i.e. immediately following the price increase in mid-October, it increased to 4.2%, but then dropped to 2.5% and 0.15% by mid-1998. It would be hard to argue that the increase in the third quarter of 1998, to 9.3%, was connected to the diesel price increase of the previous year. In short, the only discernable evidence of the 70% increase in diesel price in 1997 is an increase in the fourth quarter rate that is 1.6% higher than the quarterly rate immediately preceding and immediately following. 47 48 Household Energy Supply and Use in Yemen Volume 2: Annexes Figure A4.10: Quarterly Inflation Rates (CPI) v. Diesel Price 0.06 18 DIESEL 16 0.04 11.6% 14 100) 9.3% 9.1% 994=1r 6.5% 12 6.1% bemeceD(IPC 0.02 5.4% 4.2% 4.5% 3.7% 10 2.6% 2.5% 2.1% 0.8% 0.7% 0.8% 0.1% -0.1% -0.3% 8 0 -0.5% -0.4% -1.8% QUARTERLY RATE 6 -3.9% -5.0% -0.02 4 jan 1997 jan 1998 jan 1999 jan 2000 jan 2001 jan 2002 jan 2003 july july july july july july july A4.33 In the case of the diesel price increase in the third quarter of 2001, the inflation rate does indeed increase sharply in this quarter with an 11.6% rise.6 But in the two previous years when there was no diesel price increase, the third quarter inflation rates were 9.3% and 9.1%. Note that in the subsequent quarters, the inflation rates drops to 5.4% and -3.9% in the first quarter of 2002 ­ a clear indication of seasonal variation and that increased food prices attributed to "diesel price increases" are computed away over time. A4.34 An examination of the detailed monthly increases in the major components of the CPI also shows little evidence of dramatic effects (Table A4.13). August shows a sharp increase of 3.5% for transport (but which only has a weight of 4.25% in the CPI). September shows sharp increases in food and qat prices, but these, as noted above, occur almost every year due to normal seasonal fluctuations. 6The Cabinet Decree increasing diesel prices was passed on July 26th, 2001. Tariffs for electricity, water and telephone were also increased on August 1, 2001, together with domestic air fares (5%) and international air fares (15%). Annex 4: Diesel 49 Table A4.13: Monthly Inflation Rates in 2001 Weight Jan01 Feb Mar April May June July Aug Sept Oct Nov Dec in CPI All items 10000.0 0.5% -0.7% 1.8% 1.2% 0.0% 1.2% 0.8% 2.4% 8.1% 1.0% 2.7% 1.7% Food and 4381.2 1.7% -0.3% 2.9% 1.8% -0.3% 1.8% -0.2% 1.2% 8.3% 1.3% 4.1% -4.0% non-alcoholic beverages Tobacco, cigarettes and 1484.5 -3.3% -4.9% 2.3% 2.1% 0.0% 1.2% 1.6% 4.3% 25.7% 2.2% 4.0% 19.8% Qat Clothing and footwear 871.5 -0.4% 0.8% 0.1% -0.5% 0.6% 0.6% 1.5% 1.8% 2.8% 0.3% 0.8% 0.5% Housing and related 1327.2 2.1% 0.5% 0.1% -0.2% 0.1% 0.0% 2.3% 5.6% 0.1% -0.0% 0.0% -0.1% items Household furnishings 405.1 -0.8% 0.0% 0.6% 0.5% 0.9% 0.5% 0.7% 1.2% 0.7% 0.4% 0.6% -0.1% and appliances Health 267.3 -1.6% 0.9% 2.9% 0.4% -0.1% 2.5% 3.0% 3.2% 1.8% -0.2% 0.2% 0.3% Transport 425.6 0.5% 0.0% 0.1% 0.7% 0.0% 0.2% 0.1% 3.5% 1.1% -0.0% 0.0% -0.0% Communications 18.9 -1.0% 0.2% 0.8% 0.1% 1.4% 0.0% 0.0% 3.5% 1.3% 1.3% 0.0% -0.3% Recreation and culture 84.1 0.1% 0.0% 0.0% -0.2% 0.4% 0.1% 1.3% 1.1% 1.3% 0.1% 0.0% -0.1% Education 52.1 1.1% 0.0% 0.0% 0.0% 0.0% 0.0% 0.7% 0.0% 24.2% 0.0% 0.0% 0.0% Restaurants & Hotels 283.3 4.0% -1.0% 0.5% 1.0% 0.9% 2.8% 1.9% 0.0% -0.2% -0.8% 1.5% -0.4% Misc. goods and services 399.3 -0.1% 0.5% 0.2% 0.8% 0.9% 0.2% 0.5% 0.2% 2.9% -0.1% 0.1% 0.0% 49 Annex 5 Impact of Diesel Subsidies on Water, Qat and Food Prices Water A5.1 Only 32% of the population has access to drinking water from a public supply system and therefore large sections of the population are dependent upon purchased water, which in turn is dependent upon diesel both for groundwater extraction and for bulk transportation. The impact on purchased water prices will therefore be one of the critical issues in the political economy of reducing subsidies. A5.2 The 2003 HES included a question on water purchases. As shown in Table A5.1, households (that buy water) spend 5% of their expenditure on water. 89% of urban households, and 53% of rural households purchase water: particularly in rural areas, the proportion of households that purchase water rises sharply with income. A5.3 In some areas, mosques play an important role in distributing water to the poorest households free of charge. The study team visited a mosque in Sana'a in a poor area (of the old city), where children came with wheelbarrows and were filling 5, 10 and 20 liter plastic containers with water. The mosque purchases water from a water seller enabled by contributions made by the better-off adherents, a practice said to be widespread and in accordance with the zakat traditions of Islam. A5.4 Elimination of the diesel subsidy will result in the cost of wholesale (bulk) water increasing by 10-15%. But demand for bulk water is likely to be elastic: it is relatively easy to use 10-15% less for washing, bathing, etc. if one wants to adjust to higher prices by reducing consumption. The survey shows the average urban purchased water bill is 3,000 YR/HH/month and thus the average household will face an increase of 300-450 YR/month. A5.5 However, much of the bulk water supply is consumed by the non-poor: for example, the top urban income decile spends 6,750 YR/month on purchased water. But if these groups adjust to higher prices by conservation, that can only be to the good of Yemen. 51 52 Household Energy Supply and Use in Yemen Volume 2: Annexes Table A5.1: Expenditure on Purchased Water Income per decile %HH reporting use Expenditure, [YR/month] purchased water as % of total HH (YR/month) Urban Rural All Urban Rural All Urban Rural All 1 0-9000 73% 29% 1.1.1 33% 621 215 253 6% 3% 3% 2 9001-12000 79% 39% 45% 976 293 405 5% 2% 3% 3 12001-15000 81% 36% 1.1.2 42% 1119 514 602 6% 3% 3% 4 15001-19800 82% 43% 51% 1613 420 663 5% 2% 3% 5 19801-22500 90% 58% 1.1.3 64% 2016 829 1059 8% 4% 5% 6 22501-27000 94% 62% 70% 1871 984 1209 7% 4% 4% 7 27001-33000 95% 58% 1.1.4 69% 2445 1359 1673 8% 4% 5% 8 33001-42700 86% 60% 1.1.5 68% 3242 1586 2064 10% 5% 6% 9 42701-61000 94% 74% 80% 4418 2317 2996 9% 5% 6% 10 61001>0 96% 85% 1.1.6 89% 6747 4608 5316 7% 5% 6% average 89% 53% 61% 3002 1208 1616 8% 4% 5% 1.1.7 1 1.1.8 ret URBAN 1.1.9 wadesahcrupginsu 0.8 1.1.10 1.1.11 RURAL 0.6 1.1.12 HH 1.1.13 of 0.4 onitcarF 1.1.14 0.2 1.1.15 d[1] d[2] d[3] d[4] d[5] d[6] d[7] d[8] d[9] d[10] URBAN 73% 79% 81% 82% 1.1.1690% 94% 95% 86% 94% 96% RURAL 29% 39% 36% 43% 1.1.1758% 62% 58% 60% 74% 85% 1.1.18 0.12 1.1.19 e 0.1 urit 1.1.20 ndepxe 0.08 1.1.21 URBAN HHlatot 0.06 0.04 RURAL of %sa 0.02 0 d[1] d[2] d[3] d[4] d[5] d[6] d[7] d[8] d[9] d[10] URBAN 6% 5% 6% 5% 8% 7% 8% 10% 9% 7% RURAL 3% 2% 3% 2% 4% 4% 4% 5% 5% 5% Annex 5: Impact of Diesel Subsidies on Water, Qat and Food Prices 53 Box A5.1: Impact on Diesel Price Increases on Water Sellers In January 2003 the study team visited two private urban water operations in the Haddah area of Sana'a (a wealthy suburb). One was pumping water from a 170-meter depth, the other from a 200-meter depth and selling water to water sellers who came to the pumping stations with bowsers. The table below shows a calculation of the probable impact of a diesel price increase (from the present price to the 2003 economic price of 37 YR/liter) on the selling price. While the information collected relied on the operators' statements and could not be verified, the indicated calculations provide an order of magnitude estimate of the probable price impact. Water Water wholesaler A wholesaler #2 (30HP pump) (40 HP pump) Water wholesaler (& storage) 1 wholesale water price [YR/tanker] 200 200 2 total daily revenue [YR/day] 20,000 14,000 3 tankers/day [#] 100 70 4 capacity [m3] 3.5 4 5 total quantity of water [liters/day] 350,000 280,000 6 diesel consumption [liters/day] 120 140 7 diesel cost [YR/liter] 17.5 17.5 8 [YR/day] 2100 2450 9 wholesale water price [YR/liter] 0.057 0.050 10 of which diesel [YR/liter water] 0.006 0.009 11 [%] 11% 18% 12 incremental diesel cost [YR/liter] 20.3 20.3 13 [YR/liter water] 0.007 0.010 14 increase in wholesale price [%] 12% 20% Water seller 15 transportation distance [km] 3 3 16 [km/liter] 3 3 17 diesel use per round trip [liters] 2.0 2.0 18 [YR/trip] 35.0 35.0 19 seller's price of water [YR/tanker] 600 600 20 [YR/trip] 5.8% 5.8% 21 22 seller's incremental diesel price [YR/liter] 20.3 20.3 23 [YR/trip] 40.6 40.6 24 increase in wholesale price [YR/trip] 24.4 40.6 25 total increase [YR/trip] 65.0 81.2 26 [ ] 10.8% 13.5% The calculation suggests a price increase of 10-15%, depending on assumptions. It may be noted that this modality of private water supply is quite inefficient. 53 54 Household Energy Supply and Use in Yemen Volume 2: Annexes Box A5.2: Impact on Water Prices A significant portion of urban water supply is in the form of bulk deliveries supplied by small bowsers. These fill up from groundwater pumping stations. Water is also widely sold in 10-liter plastic containers at most neighborhood food shops: this water is ozone-treated, and suitable for drinking purposes. The sample calculations presented here are based on information obtained from a sample of such operations visited in 2003. WHOLESALE WATER (Water Pumping) Assumptions: · Water pumping business serving 100 tankers/day, each @ 3.5m3 · Present wholesale water price: 200 YR/tanker · Diesel consumption: 120 liters @ 17.5 YR/liter · Present diesel cost: 2,100 YR/day · Total sales revenue: 20,000 YR/day Then if diesel price increases by 20 YR/liter · Diesel cost increases by 2,400 YR · Water cost increases by 12% · Cost per 3.5m3 tanker increases from 200 YR to 224 YR = 24 YR URBAN BULK WATER DELIVERY Assumptions: · Delivery distance: 3km; · Diesel fuel consumption: 3km/liter; 2 liters/round trip @ 17.5 YR/liter · Diesel cost: 35 YR/trip · Selling price of water: 600 YR/tanker (3.5m2) · Delivered cost per liter: 0.2 YR per liter! If diesel price increases by 20 YR/liter · Wholesale water cost increases by YR24 · Diesel fuel cost increases by 20 YR x 2 liters = YR40 · Total cost increase 24 + 40 = YR64 · Delivered cost increases from YR600-664 DRINKING WATER: (10-liter container in food shop) Assumptions · Present shop price 40 YR for 10 liters = 4 YR/liter (plus a refundable YR200 for the container) If diesel price goes up by 20 YR/liter · Water extraction cost increases by YR24 for 3500 liters, and the water delivery cost increases by YR40 per 3 km per 3,500 liters; · Hence for, say, 15 km distance, delivery cost increases by YR200 for 3,500 liters · Total cost increase for 3,500 liters = YR224 (=0.064 YR/liter) · Cost increase for a 10-liter container = 0.64 YR/bottle. A5.6 On the other hand, drinking water demand is inelastic and a basic human need. 79% of poor urban households buy water, spending 900 YR/month on water. If diesel increases 20 YR/liter, their water costs will increase by 10-15% (assuming they access a bulk water supply), or 90-140 YR/month. This increase represents about 1% of their total household expenditure. Similarly, 35% of poor rural households buy water, spending 330 YR/month. If diesel increases by 20 YR/liter, their water costs Annex 5: Impact of Diesel Subsidies on Water, Qat and Food Prices 55 will increase by 10-15%, or 33-66 YR/month. This increase represents 0.5% of their household expenditure. A5.7 Thus the impact of diesel price increases on (purchased) drinking water is small (Box A5.2). The most significant impact is on the urban poor, because a much smaller proportion of the rural poor purchase water. However, even for the urban poor, the impact is only about 1% of their monthly income. Box A5.3: Cost of Water Pumping The cost of water pumping can be derived from the basic equation for the power, in kW, required to lift a given flow, namely 9.81Qh kW = 1000epumpemotor where Q = pumping rate in liters/second h = hydraulic head, in meters epump = efficiency of the pump emotor = efficiency of the motor Given that 1 kWh is equivalent to 0.35 liters of diesel, then the daily diesel requirement, d, follows as d = 0.35 kW ophours Where ophours is the number of hours the pump is in operation per day. This may be applied to the data for the first water seller reported in Box A5.1. For the data given, for and for the stated amount of water per day, the hydraulic head calculates to 200 meters if the pump and motor efficiencies are taken at 75%. Number of tankers served per day [ ] 100 Tanker size cu meters 3.5 Water pumped cubic m/day 350 liters/day 350000 Operating hours hours/day 15 Flow liters/sec 6 Head meters 200 Pump efficiency [ ] effpump 0.75 Power unit efficiency [ ] effpower 0.75 Gross capacity kW capacity 23 HP 30 kWh kWh/day 339 Diesel 0.35 liters/kWh liters/day 119 liters/cuMetre 0.34 Thus for 1 cubic meter to be brought from 200-meter depth requires 0.34 liters of diesel fuel ­ a value that is used below for the calculation of the diesel input into a sample of water-intensive fruit and vegetables (Box A5.4). 55 56 Household Energy Supply and Use in Yemen Volume 2: Annexes Box A5.4: Assumptions and Calculations of Qat Price Increases Qat Production Costs Production costs per ha of Qat under well irrigation · Water requirement 700-1400mm/year = 10,000cubic meters/ha · Assume entire water requirement from groundwater (not trucked) · diesel requirement 8,000 liters/year · diesel cost/ha = 8,000 @ 17 YR/liter = 136,000 YR/year (=9.5% of revenue) · Output/ha: 7,200 bundles [1200 trees/ha; 2 harvest/year; 3 bundles/tree/harvest: 1 bundle=0.5kg] · Revenue @ 200 YR/bundle (ex-farm price, lower quality) = YR1,440,000 If diesel price increases by 20 YR/liter · Increased diesel cost = 160,000 Year for 7,200 bundles= 22 YR/bundle Qat Production Using Trucked Water Assumptions · 10% of water requirement (i.e. 1000m3 )brought by truck · Higher quality Qat, ex-farm price 300 YR/bundle · Water brought by truck over a 20km distance · Fuel consumption 3km/liter = 13 liters/trip · 1400 m3 requires 400 truck loads, hence 400x13=5,200 liters of diesel If diesel price increases by 20 YR/liter · Diesel cost for transportation increases by 20 x 5,200 = YR104,000 · Cost increase per bundle = 104,000/7,200 = 14 YR/bundle =5% of 300 YR ex-farm price Qat: Sana'a Price Assumptions for transportation · Qat transported 40km by diesel vehicle at 5km/liter = 16 liters (very conservative, average distances much lower) · Present diesel cost per round trip @ 17 YR/liter = YR272 · Assume 500 bundles per trip · Retail price 500 YR/bundle (lower quality) · Total market value per trip YR500 x 500 = YR250,000 If diesel price increases by 20 YR/liter · Transportation cost increases by YR320 for total trip, or about 1 YR/bundle · Farm price increases by YR22 · Hence total price increase 1 + 22 = YR23 Food A5.8 A detailed assessment of the impact of diesel subsidy removal requires an input-output model, which is not available for Yemen. The best that can be done is to estimate the increase in diesel price required for water pumping,7 and express this as a 7 The water inputs required per ha were taken from the 1993 Agriculture Sector Study, which contains detailed calculations of production costs for a selection of water intensive crops. While the prices for the non-diesel inputs (labour, pesticides, fertiliser, seeds, land preparation, etc.) have obviously changed significantly since the early 1990s in response to general inflation, the estimates of water requirement itself per ton (or planted area) of crop would have changed much less (and would be limited to the response to changes in the relative prices of the inputs). Annex 5: Impact of Diesel Subsidies on Water, Qat and Food Prices 57 fraction of the retail price. Even this calculation is subject to uncertainty, given lack of information about groundwater depths. Here we assume a conservative depth of 200 meters: if the actual depth is only 100m, the impact is halved. Moreover, this calculation does not include the additional impact of the transportation of crops from the farm to the consumer: however, based on the analysis of Qat transport, in the previous section, the diesel input in production far exceeds the diesel input in transportation (per unit of retail market value), and therefore the additional price increase attributable to transportation would be a few percent at most. In addition the calculation does not take into account second round effects. A5.9 The calculations are shown in Table A5.2. The increase as a fraction of the retail price lies between 6% (for tomato) and 13% (for coffee). These are significant increases, and therefore require that the diesel price be increased over a period of a few years, rather than in a single step to allow households and farmers reasonable time to adjust. Table A5.2: Impact of Diesel Price Increases on Water Intensive Fruit and Vegetables Tomatoes Potato Grapes Coffee Water requirement m3 per ha 7,600 11,000 22,000 16,600 Yield tons/ha 14 12.0 15 0.8 Retail price YR/kg 60 70 126 1050 Retail yield YR/ha 840,122 839,978 1,889,983 840,005 Diesel requirement liters/m3 0.34 0.34 0.34 0.34 liters/ha 2,584 3,740 7,480 5,644 Diesel cost increase YR/liter 20 20 20 20 YR/ha 51,680 74,800 149,600 112,880 Retail price impact [ ] 6.2% 8.9% 7.9% 13.4% YR/kg 3.7 6.2 10.0 141.1 Sources: Sana'a retail prices: Central Bank Annual Report 2003 Water requirements and yields: World Bank, Republic of Yemen, Agriculture Sector Study, Annex 1I, comparative economic analysis for crops. A5.10 Despite the large uncertainties in this calculation, the impact must necessary be bounded by what is known about the overall level of diesel use in the economy. If the total national expenditure on food is YR600 billion (the survey indicates YR417 billion, which may be understated by about one third, hence YR600 billion may be a better estimate of actual 2003 expenditure) and the total use of diesel in agriculture is, say, 1,000 million liters (total household diesel consumption is 486 million liters, plus an equal amount, say, by commercial farming enterprises not captured in the survey), then a YR20 increase in diesel price increase on the 1,000 million liters represents YR20 billion, equivalent to an increase in retail prices of just 3.3%. Fruit and vegetables are the most water (and hence diesel) intensive of all crops, so one would expect these to be more sensitive to diesel prices than all food (that includes a significant amount of imports whose production cost is not affected by Yemen diesel prices, and whose sole diesel input is for transportation. 57 Annex 6 LPG Consumption Patterns A6.1 Over the past decade there has occurred a dramatic transformation of household energy use for cooking, with a major shift from fuel wood to LPG: in the Northern governorates, for which we have data from the 1989 HES, LPG consumption has increased from 87,000 tons in 1988 to over 500,000 tons by 2003.8 This strategy of encouraging household use of LPG was adopted by Yemen on the basis of several considerations, including concerns over deforestation, the heavy time burden on rural women and children for fuelwood collection, the health impacts of using fuelwood for cooking, and the strong preference expressed by all income groups for LPG as the most desired fuel for cooking. A6.2 LPG consumption from Marib has risen dramatically during the 1990s (Table A6.1) increasing from some 7,433 tons in 1990 to 624,813 tons in 2003.9 8 The previous Yemen HES covered only the Northern governorates of the former YAR, and therefore comparisons of aggregate amounts require caution. World Bank/ESMAP, Republic of Yemen, Household Energy Strategy Study, Phase I: A Preliminary Study of Northern governorates, Washington DC, March 1991. 9 This historical presentation of LPG use by YGC may be a clue to the reconciliation problems discussed in Annex 10: the HH survey shows greater consumption of LPG than is reported sold by YGC. According to the above table, in 1990 LPG sales were 7,433 tons; yet the 1991 HES (for the Northern Governorates of the former YAR) states LPG consumption at 87,000 tons! 59 60 Household Energy Supply and Use in Yemen Volume 2: Annexes Table A6.1: LPG Salient Statistics Year Consumption Annual Bottling Annual tons/ Primary growth Plants bottling transport Rate station fleet [tons/year] [%] [#] [tons] #trucks 1990 7,433 7 1,062 5 1991 51,771 597% 8 6,471 17 1992 89,200 72% 10 8,920 56 1993 107,067 20% 16 6,692 96 1994 232,815 117% 17 13,695 113 1995 261,106 12% 21 12,434 121 1996 279,790 7% 27 10,363 141 1997 324,011 16% 29 11,173 194 1998 348,856 8% 36 9,690 215 1999 412,894 18% 48 8,602 276 2000 462,783 12% 60 7,713 339 2001 505,823 65 399 2002 587,994 68 418 2003 624,813 71 420 Source: YGC A6.3 The bulk of LPG use is for domestic cooking, but there are few data on the extent of other uses. Some LPG is delivered to larger establishments (for hotels, or heating chicken broiler houses) in bulk form, and some LPG filling stations offer larger cylinders (also used in restaurants). Over the past year, a significant number of restaurants have converted from cylinders to bulk supply replenished by small road tankers. A6.4 The largest non-domestic use is likely to be for road transport, as there is high incentive to convert gasoline cars to LPG given the difference in price. LPG consumption of the transport sector is estimated at 10% of the total.10 This is driven entirely by the present retail price differential between gasoline (35 YR/liter) and LPG (10.25 YR/liter). However, there is no reliable information on the number of conversions that have actually occurred, and the ability to infer LPG consumption from lower than expected growth in gasoline demand is beset with a number of practical difficulties. A6.5 Based on discussions with officials of the YGC, the composition of LPG consumption for 2003 can be taken as 87% household, 8% transport, and 5% for agriculture, hospitals, restaurants & hotels, government, and military. However, the transport share is increasing, and for 2004 may be taken as 10% of the total. An estimate of 2003 consumption is as shown in Table A6.2. 10According to YGC, most car filling stations are owned by owners of LPG bottling stations. Annex 6: LPG 61 Table A6.2: 2003 LPG Consumption 1000 tons Marib 625 Aden refinery 6 Total sales 100% 631 Consumption Domestic 88% 555 Transport 7% 44 other 5% 32 A6.6 Table A6.3 shows the cost of energy expressed in terms of cost per unit of calorific value. At present retail prices, LPG is by far the cheapest form of energy in Yemen, to which consumers respond by inventive new applications for LPG: at current retail prices, LPG has a 61% cost advantage per MJ over gasoline (and hence the incentive for conversion of gasoline powered automobiles) and a 13% advantage over diesel. Table A6.3: Cost per Unit of Energy LPG Gasoline Kerosene Diesel Fueloil cost per liter [YR/liter] 10.2 35.0 16.0 17.0 31.0 net calorific value [MJ/liter] 24.5 32.4 34.4 35.4 38.9 cost per MJ [YR/MJ] 0.42 1.08 0.47 0.48 0.80 advantage of LPG [YR/MJ] 0.66 0.05 0.06 0.38 [%] 61% 10% 13% 48% A6.7 For example, it is reported that farmers are running irrigation pumps with a mix of 30% LPG and 70% diesel. Increasing amounts are used for powering fridges (reflected in the household survey: 0.6% of households use LPG for this purpose). Small gasoline engines (e.g. that power pumps on water bowsers) are being converted to LPG. LPG is also being used for space heating. The LPG Supply Chain A6.8 LPG is produced by separating it from the gas produced in association with the crude oil at Marib. The gas treatment plant has a capacity of 2,200 tonnes of LPG per day and presently produces 1,860 tonnes per day. The gas treatment plant is owned by the Government, and operated as part of the Yemen-Hunt oil complex at Marib. The storage capacity is 500 tonnes. A6.9 The Aden refinery also produces LPG.11 In 2002 it produced 95,145 tons, of which 90,713 tons were exported, and the balance of 4,432 tons was sold to the domestic market.12 In 2003, the contribution from the Aden refinery increased slightly to 6,315 tons. The storage capacity at Aden refinery is 50 tons, which is currently under expansion to 100 tons. However, while LPG delivered to YGC at Marib is at no cost to YGC, the cost of LPG sold to YGC by the Aden refinery is still under negotiation (the refinery is requesting payment at the border price). A6.10 YGC owns and operates the bulk storage and tanker loading facility at Safir. There are presently 420 heavy tankers averaging 23 tons of LPG, and the tanker loading facility has a capacity of 85 tanker-trucks per day. The present rate of loading 11LPG is not produced at the Marib refinery (but only from the Marib gas processing plant). 12Oil, Gas and Mineral Statistics, Annual Bulletin 2002, Issue#2. 61 62 Household Energy Supply and Use in Yemen Volume 2: Annexes is 72 trucks/day. The tanker-trucks are privately owned (except for a few tanker- trailers owned by YGC which are leased to private operators). Figure A6.1: The LPG Supply Chain A6.11 LPG is delivered to 71 filling stations (of which 64 are privately owned, 7 owned by YGC). The typical filling station has 50 tons of storage, but presently operates at only 40% capacity: as evident from Table A6.1, average throughput has declined from some 13,500 tons/year in 1994 to only 7,700 tons/year in 2000. This follows from the sharp increase in the number of filling stations built by the private sector in response to the prospects for a rapidly growing business and guaranteed returns. A6.12 YGC is responsible for maintenance of cylinders, for which it receives a margin of 3 YR/cylinder (see Table A6.11). Filling stations are responsible for the return of damaged bottles to YGC, who repair or replace them, as necessary. A6.13 Table A6.4 shows LPG use by income decile, which ranges from 50% in the poorest decile, to 94% in the top decile (or 78% of all households). Table A6.4: LPG use by Income Decile Income per decile %HH reporting use Consumption Consumption [1000 (YR /month) [Kg/month] tons/year] Urban Rural All Urban Rural All Urban Rural All 1 0-9000 74% 47% 50% 21 20 20 4 24 28 2 9001-12000 71% 60% 62% 24 22 22 7 31 39 3 12001-15000 91% 73% 75% 23 22 22 9 41 51 4 15001-19800 90% 70% 74% 21 23 22 9 28 37 5 19801-22500 96% 78% 82% 23 24 24 12 43 55 6 22501-27000 96% 82% 86% 23 24 24 15 41 56 7 27001-33000 90% 83% 85% 24 30 28 17 48 65 8 33001-42700 91% 77% 81% 24 27 26 18 41 58 9 42701-61000 97% 92% 94% 24 32 29 20 54 75 10 61001>0 99% 92% 94% 31 38 36 27 63 91 average 92% 74% 78% 25 26 26 139 415 555 Annex 6: LPG 63 A6.14 Across all income deciles urban access is greater then rural access (Figure A6.2). Figure A6.2: Urban v. Rural Access to LPG 1.2 1 GPLlatotginsu URBAN RURAL 0.8 HH of onitcarF0.6 0.4 d[1] d[2] d[3] d[4] d[5] d[6] d[7] d[8] d[9] d[10] URBAN 74% 71% 91% 90% 96% 96% 90% 91% 97% 99% RURAL 47% 60% 73% 70% 78% 82% 83% 77% 92% 92% A6.15 In the major urban areas (such as Sana'a and Aden), LPG is used by more than 97% of households even in the poorest income groups (Figure A6.3). Most Governorates exhibit the expected increase in LPG use with income: on average just 50% of households in the poorest income decile use LPG, rising to close to 100% in the top decile. Al Hodeiah has the greatest variation across income deciles, with only 4% of the lowest income decile using LPG. Figure A6.3: LPG Penetration by Income Decile 1.2 SanaaCity 1 SanaaGovern 0.8 total 0.6 Taiz fraction Amaran 0.4 0.2 AlHodeidah 0 bottom 10% d[3] d[5] d[7] d[9] d[2] d[4] d[6] d[8] top 10% income decile 63 64 Household Energy Supply and Use in Yemen Volume 2: Annexes A6.16 The corresponding LPG expenditure increases gradually with income, reaching around 700 YR/month in the 8th decile. This then rises sharply in the 9th and 10th income deciles, reaching an average of 961 YR/month in the top decile (Figure A6.4). Figure A6.4: LPG Expenditure by Income Decile 1000 e omc 900 in HH.v ing 800 ookc G LP 700 oft osc average hly ontm 600 SanaaCity 500 bottom 10% d[3] d[5] d[7] d[9] d[2] d[4] d[6] d[8] top 10% A6.17 The bulk of LPG is used for cooking, with smaller quantities used for lighting and for fridges (in non-electrified areas). Use of LPG for fridges is concentrated in the three richest deciles. Usage in all but the two lowest deciles is relatively constant, but falls sharply for both lighting and cooking in the two bottom deciles ­ where the cost of LPG (as well as the initial cylinder cost and LPG device costs) appear to be relatively unaffordable. Table A6.5: Usage of LPG, as % of Households Lighting Cooking Fridge Any LPG use 1 0-9000 13.7% 49.4% 0.0% 50.8% 2 9001-12000 18.8% 62.1% 0.0% 63.3% 3 12001-15000 32.4% 75.7% 0.1% 77.1% 4 15001-19800 30.5% 73.6% 0.2% 75.0% 5 19801-22500 33.3% 82.3% 1.5% 84.9% 6 22501-27000 34.0% 85.7% 0.9% 86.5% 7 27001-33000 35.7% 85.9% 0.5% 86.4% 8 33001-42700 37.2% 82.5% 1.3% 83.7% 9 42701-61000 34.8% 93.9% 0.8% 94.1% 10 61001>0 41.0% 93.4% 1.8% 95.0% average 31.0% 78.3% 0.7% 79.5% A6.18 Table A6.6 shows consumption per month in households using LPG for the use in question. There is little difference between urban and rural consumption (in households that use LPG): top decile rural users in fact consume slightly more than their urban counterparts: but on average, urban and rural users consume about the same (30 kg/month, or about 5.5 bottles per month). Annex 6: LPG 65 Table A6.6: LPG End-uses Cooking & Baking Lighting Fridge [kg/month] [% of HH] [kg/month] [% of HH] [kg/month] [% of HH] Bottom Decile 24 [49%] 9 [13%] [0%] ALL income 30 [78%] 9 [30%] 24 [1%] Top decile 42 [93%] 10 [41%] 21 [2%] URBAN Bottom Decile 26 [73%] 6 [8%] [0%] ALL income 30 [92%] 7 [16%] 36 [0%] Top decile 38 [100%] 7 [19%] 34 [2%] RURAL Bottom Decile 23 [47%] 9 [14%] [0%] ALL income 30 [74%] 9 [35%] 22 [1%] Top decile 44 [90%] 11 [52%] 12 [1%] A6.19 However, the big difference between urban and rural values is in penetration rates: only 47% of the poorest decile use LPG for cooking in rural areas, as compared to 73% in urban areas. A6.20 Tables A6.7 and A6.8 show the details of consumption and expenditure for LPG for the two major uses, cooking (and baking), and lighting. The following may be noted: the unit cost of LPG for lighting is substantially higher than for cooking (by over 50%, at 48 YR/kg v. 25 YR/kg), because it is sold in small, non- standard bottles. Households therefore pay very high costs per lumen (discussed further in Annex 8 where the willingness to pay for electricity access is examined). even for LPG sold in standard containers, the reported average price is 25 YR/kg. The official price is 18.63 YR/kg (205 YR per 11kg cylinder). The distribution of these markups is discussed below. the per household consumption of LPG, between 2 and 3 11kg cylinders/month per household, is high compared to other countries. 65 66 Household Energy Supply and Use in Yemen Volume 2: Annexes Table A6.7: LPG for Cooking %HHreportinguse Consumption,[Kg/month] Expenditure,[YR/month] Reportedprice,[YR/kg] Income per decile (YR/month) Urban Rural All Urban Rural All Urban Rural All Urban Rural All 1 0-9000 74% 45% 48% 20 18 18 607 630 626 253 293 287 2 9001-12000 70% 59% 60% 23 20 20 684 715 709 267 305 298 3 12001-15000 91% 70% 73% 23 20 20 708 688 692 265 293 288 4 15001-19800 90% 66% 71% 20 20 20 621 722 696 260 309 296 5 19801-22500 93% 75% 79% 22 22 22 645 734 714 253 295 285 6 22501-27000 95% 81% 84% 22 22 22 682 750 731 265 298 288 7 27001-33000 90% 83% 85% 24 26 25 691 886 827 257 281 274 8 33001-42700 91% 75% 80% 24 23 23 673 746 722 246 282 270 9 42701-61000 97% 91% 93% 23 28 27 672 902 824 253 268 263 10 61001>0 99% 89% 92% 29 35 33 853 1077 998 249 268 261 average 91% 72% 76% 24 24 24 699 792 767 256 288 279 1.2 gni 1 ookc URBAN GPLgnisu RURAL 0.8 HH ofnoitcarF0.6 0.4 d[1] d[2] d[3] d[4] d[5] d[6] d[7] d[8] d[9] d[10] URBAN74% 70% 91% 90% 93% 95% 90% 91% 97% 99% RURAL 45% 59% 70% 66% 75% 81% 83% 75% 91% 89% Annex 6: LPG 67 Table A6.8: LPG for Lighting Income per decile %HHreporting use Consumption,[Kg/month] Expenditure, [YR/month] Reported price,[YR/2kg cyl] (YR/month) Urban Rural All Urban Rural All Urban Rural All Urban Rural All 1 0-9000 8% 11% 11% 5 7 7 203 288 282 70 89 88 2 9001-12000 10% 20% 18% 7 7 7 317 316 316 104 96 96 3 12001-15000 12% 33% 30% 3 7 7 184 316 309 109 90 91 4 15001-19800 11% 34% 30% 5 7 7 305 375 369 95 107 106 5 19801-22500 31% 31% 31% 6 8 7 261 329 316 89 89 89 6 22501-27000 14% 39% 33% 5 6 6 345 309 313 110 101 102 7 27001-33000 12% 42% 33% 4 7 7 173 388 366 91 101 100 8 33001-42700 20% 38% 33% 5 7 7 219 347 324 90 101 99 9 42701-61000 11% 41% 32% 6 8 8 259 338 329 96 99 99 10 61001>0 17% 50% 39% 6 8 8 256 368 353 95 94 94 average 15% 33% 29% 5 7 7 252 342 331 95 97 97 0.6 ginth 0.5 lig RURAL GPLgnisu 0.4 0.3 HH of 0.2 onitcarF URBAN 0.1 0 d[1] d[2] d[3] d[4] d[5] d[6] d[7] d[8] d[9] d[10] URBAN 8% 10% 12% 11% 31% 14% 12% 20% 11% 17% RURAL 11% 20% 33% 34% 31% 39% 42% 38% 41% 50% A6.21 Figure A6.5 shows the use of LPG by Governorate. Al Hodeida has by far the lowest LPG penetration, with only 27% of households reporting LPG use: Sa'adah and Al-Amran also have significantly low LPG penetration rates. Figure A6.5: Percentage of Households in Each Governorate Reporting LPG Use fraction of HH using LPG 0 0.2 0.4 0.6 0.8 1 1.2 Ibb 87% Abyan 89% SanaaCity 100% AlBaida 86% Taiz 78% Haja 68% AlHodeidah 27% Hadramouth 99% Dhamar 75% Shabwah 93% Sadah 57% SanaaGovern 99% Aden 97% Lahj 98% Al Mahweet 87% Amaran 58% Adelah 100% 67 68 Household Energy Supply and Use in Yemen Volume 2: Annexes A6.22 The demand for LPG shows a distinct winter peak (Figure A6.6), which in recent years has also coincided with Ramadan: many families report usage of an extra cylinder during Ramadan.13 Figure A6.6: Monthly LPG Demand 1.1.22 60 50 h ontm/s 40 ont 30 20 jan 1997 jan 1998 jan 1999 jan 2000 jan 2001 jan 2002 jan 2003 july july july july july july july Reconciliation with Supply Side Data A6.23 The raw data of the 2003 HES suggest annual LPG consumption of 711,000 tons /year, which is considerably higher than that reported by YGC (555,000 tons/year). When one also takes into account that there is likely to have been significant LPG consumption not covered by the survey (LPG for transportation, LPG use in commercial and industrial establishments, and smuggling), and that the YGC figure is likely to be fairly accurate, this discrepancy needs explanation. A6.24 The survey estimate is based on the following assumptions: that each recharge of the standard cylinder contains 11kg of LPG. that bottles are empty when refilled. that the number of cylinders used on average each month would be answered reasonably well. However, there are doubts regarding each of these assumptions: There is much anecdotal evidence about short-filling of bottles. YGC data suggest that "empty" bottles may contain as much as 0.5kg of LPG when returned for refilling. 13This may have had some impact on the survey, which was conducted in the month following Ramadan. Although households were requested to report the number of cylinders bought "on average" each month, recent purchases would inevitably influence their responses. Indeed, the raw data implies a much higher rate of consumption than that inferred from YGC data on actual deliveries to LPG bottling plants. Moreover, a YGC survey of 2002 showed that the average bottle contained 0.5kg of LPG when it was returned for refilling, so that the assumption that each cylinder represented 11kg of consumption leads to a 4.5% overestimate in total LPG consumed by the household sector. Annex 6: LPG 69 The answer to the question on average number of cylinders was inevitably influenced by the higher consumption during Ramadan, which immediately preceded the survey14 A6.25 The distribution of reported number of LPG bottles used per month is shown in Figure A6.7. The most frequently reported number was two per month. Figure A6.7: Distribution of Reported Cylinders/month 1200 1000 800 ncy 600 que fre 400 200 0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0 11.0 12.0 LPG bottles/month A6.26 One might suppose that the number of bottles used per month increases with household size, but one would also expect substantial scale economies. That is indeed the case, as shown in Figure A6.8. Figure A6.8: Bottles per Month v. Household Size The least squares regression relationship is [LPG bottles/month] = 1.41 + 0.15 [persons/HH]; R2=.14 14The companion question #112 (B3_2 "how many days does one cylinder of LPG last"), which would serve as an independent consistency check, is unhelpful, since the replies are generally consistent (with an implied number of days per month that ranges from 28 to 31.5!) It is not clear (from the data itself) which of the two questions was actually used, and which inferred. 69 70 Household Energy Supply and Use in Yemen Volume 2: Annexes A6.27 Addition of household income (or income decile variable) is not statistically significant. For a family of four this equation yields two bottles per month, which seems reasonable. A6.28 The data show high variability. For example, as shown in Table A6.9, for a family of four the lowest reported number of bottles per month was 0.5, the highest number eight. The low value is plausible, implying that LPG is used sparingly: According to the PRA, a typical cylinder should suffice for 30 meals in an average (rural) family. However, the highest reported value seems doubtful: a family of four that reports eight bottles a month would need to cook 240 meals, a very unlikely number, unless the individual cooked for others or had a cottage industry. Table A6.9: Range of Reported Number of Bottles Persons/HH Number of HH Average Lowest Highest 1 16 1.7 0.8 4.0 2 97 1.6 0.3 4.0 3 153 1.8 0.3 4.0 4 280 2.0 0.5 8.0 5 355 2.1 0.4 7.5 6 406 2.3 0.5 10.0 7 402 2.5 1.0 10.0 8 342 2.6 0.3 10.0 9 278 2.8 0.4 12.0 10 203 3.0 0.7 10.0 11 147 3.0 0.4 15.0 12 106 3.1 1.0 10.0 13 72 3.2 0.3 7.5 14 53 3.6 1.0 10.0 15 41 3.5 1.0 8.0 16 27 3.6 1.0 9.3 17 27 4.5 1.5 8.0 18 4 3.8 1.3 6.0 19 10 4.6 1.3 10.0 20 8 4.7 2.0 8.0 21 4 2.7 2.0 3.0 22 3 5.6 3.8 7.5 23 3 4.5 1.0 8.0 24 3 7.0 4.0 10.0 25 2 4.0 4.0 4.0 26 4 5.0 4.0 6.0 27 4 4.7 4.0 5.0 28 4 5.4 2.1 7.0 29 2 8.0 8.0 8.0 30 3 3.0 2.0 4.0 Annex 6: LPG 71 A6.29 Table A6.10 shows LPG sales by governorate as per YGC data. Using the LPG household penetration data, we can calculate from this data the average kg/HH, assuming all LPG is sold to households. The expectation would then be that in provinces where one expects large industrial & commercial use, or diversion to transport, the apparent household consumption average would be high. Table A6.10: Reconciliation of LPG Use (from YGC Data) Population HH LPG HH using Ton LPG KG/HH Kg/month Household (`000) (` 000) Use LPG (`000) Income Source> YGC YGC 2003 HES YGC 2003 HES [1] [2] [3] [4]=[2]*[3] [5] [6]=[5]/4] [7]=[6]/12 [8] Sana'a Govern 2468 329 99% 326 101980 313 26.0 44523 Sana'a City 617 82 100% 82 26493 322 26.8 62101 Aden 537 72 97% 70 20218 291 24.2 35762 Taiz 2442 326 79% 257 78300 304 25.4 26523 Hodeida 2071 276 29% 81 35726 441 36.7 21104 Lahj 686 91 97% 89 25882 292 24.3 26294 Ibb 2143 286 90% 258 64667 251 20.9 16650 Abyan 447 60 89% 53 14621 276 23.0 55583 Dhamar 1275 170 77% 131 28168 215 18.0 26019 Shabwa 484 65 96% 62 11021 177 14.8 35486 Hajja 1451 193 73% 142 26803 189 15.7 44417 Al-Baida 599 80 86% 69 22604 328 27.3 45933 Hadramout 920 123 99% 121 28872 238 19.8 38889 Sa'adah 635 85 63% 53 18522 346 28.9 26721 Amran 1027 137 59% 81 21624 268 22.3 34429 Meheit 479 64 8580 Mahara 75 10 2827 Marib 702 94 6261 Dhalee 428 57 16177 total 19486 2598 0.79 24630 559346 273 22.7 32158 A6.30 When this is plotted against average household income in the various Governorates, the results shown in Figure A6.9 are obtained ­ with the remarkable result that consumption appears to decline with household income ­ and that Al Hodeida has by far the highest consumption per household. Figure A6.9: Apparent 2002 LPG Consumption per Household v. Household Income 40 Hodeida 35 30 Sadah h/HH Al-Baida Sanaa City 25 Taiz SanaaGovern ont Lahj Aden Amran Abyan Kg/m Ibb 20 Hadramout Dhamar Hajja 15 Shabwa 10 10 20 30 40 50 60 70 Thousands HH income 71 72 Household Energy Supply and Use in Yemen Volume 2: Annexes A6.31 However, from the 2003 HES it is known that LPG consumption should increase with rising income, as shown in Figure 6.10 ­ from which we see that LPG consumption in Al Hodeida is only 20 kg/month (per household using LPG) ­ rather than the 35 kg/HH that can be inferred from the YGC data. Figure A6.10: LPG Consumption/Household v. Household Income as per 2003 HES 60 50 AlBaida Adelah SanaaGovern 40 Dhamar Al Mahweet Haja Sadah Amaran onth/HH Ibb Abyan Taiz 30 SanaaCity Kg/m Shabwah Hadramouth 20 AlHodeidah Lahj Aden 10 10 20 30 40 50 60 70 Thousands HH income A6.32 In Figure A6.11 the trend-line from the 2003 household survey data is superimposed onto the YGC data for 2002 (though displaced downward to reflect the lower overall 2002 consumption). Figure A6.11: Expected Household LPG Consumption 40 Hodeida 35 30 Sadah Al-Baida th/HH Sanaa City 25 Taiz SanaaGovern on Lahj Aden Amran Abyan Kg/m Ibb 20 Hadramout Dhamar Hajja 15 Shabwa 10 10 20 30 40 50 60 70 Thousands HH income Annex 6: LPG 73 A6.33 Four Governorates are seen to lie significantly above the expected trend line: Hodeida, Sa'adah, Taiz and Lahj. Of these, three are coastal provinces from which one might expect some degree of smuggling across the Red Sea (though the LPG filling stations are located far from the coast in Taiz and Lahj, and in these two governorates one might infer a more significant commercial LPG use in the larger cities. However, in the case of Hodeida, the level of LPG cannot reasonably be explained by household consumption, and for which, therefore, smuggling seems the most likely explanation. The Retail Price of LPG A6.34 The retail LPG price at the bottling shops in the main urban centers is regulated by the government, and presently stands at YR205 per standard cylinder containing 11kg of LPG.15 The price structure is shown in Table A6.11. Table A6.11: LPG Price Structure YR/11kg cylinder Input price 0 Primary transportation 41.5 YGC costs 6.0 Depreciation 3.0 Filling station investment return 25.0 Taxes &Royalties Central Government 99.0 Local taxes 1.5 Local authority 5.0 Ex-filling station 181.0 Bottle shop expenses+ secondary 24.0 transportation Ex-bottle shop 205.0 A6.35 The survey shows that the most common price actually paid by the consumer for an 11kg cylinder is YR250, with YR300 and YR350 being other common prices (Figure A6.12). As expected, there are significant differences between rural and urban prices: the overwhelming majority of bottles priced more than YR350 are in rural areas. The price differential reflects tertiary distribution costs (see Figure A6.1). YGC is responsible for monitoring and enforcement of the official price at the bottle shop, which it does by random inspections throughout the country. 15Some sources report the official LPG price per cylinder of 12.5Kg as YR220/cylinder. There are some 7 million LPG cylinders in circulation, made in five different countries as well as in two plants in Yemen (one in Sana'a and one in Aden). Because these have different tare weights (ranging from 14.8 to 15.2 kg) , YGC enforcement of the 11 kg LPG refill is based on an average 15kg tare weight. Although the Ministry of Industry and Trade officially controls prices and weights, enforcement is provided by YGC which conducts random sampling at filling stations to enforce weights, and has introduced coloured caps so that cylinders can be traced to specific filling stations. 73 74 Household Energy Supply and Use in Yemen Volume 2: Annexes Figure A6.12: Distribution of Reported Consumer LPG Prices 0.6 rural urban average 293 261 median 270 250 la 0.4 ott of iont ac Fr0.2 0 200 240 280 320 360 400 440 480 520 560 220 260 300 340 380 420 460 500 540 >570 Retail price, 11kg cylinder Rural HH Urban A6.36 Figure A6.13 shows the reported average consumer purchase price of LPG by Governorate. Figure A6.13: Average Price of 11kg LPG Cylinder, by Governorate Averagepriceof 11KgLPGcylinder 0 100 200 300 400 Ibb 317 Abyan 288 SanaaCity 261 AlBaida 338 Taiz 290 Haja 296 AlHodeidah 262 Hadramouth 246 Dhamar 343 Shabwah 263 Sadah 332 SanaaGovern 267 Aden 241 Lahj 250 Al Mahweet 334 Amaran 286 Adelah 293 A6.37 With the one exception of Al Baida, the price varies little with income decile (Figure A6.14). Annex 6: LPG 75 Figure A6.14: Cost of LPG Cylinder as a Function of Income Decile and Location 500 450 AlBaida rednil 400 cy GPLfo 350 st co e agre 300 av average 250 SanaaCity 200 bottom 10% d[3] d[5] d[7] d[9] d[2] d[4] d[6] d[8] top 10% A6.38 There is however some evidence that price variations affect LPG penetration, as shown in Figure A6.15: higher prices result in lower penetration rates. The simple linear least squares fit has a statistically significant R2 of 33% (if one excludes Al Hodeida as an outlier).16 Figure A6.15: LPG Penetration v. Average Price 1.2 1 SanaaCity Adelah ]HHfo Aden Hadramouth SanaaGovern Lahj Shabwah Abyan Ibb AlAlBaida Mahweet %[ 0.8 Taiz Dhamar iontarten Haja 0.6 Amaran Sadah pe GPL 0.4 AlHodeidah 0.2 220 240 260 280 300 320 340 360 Average price of 11 KgLPG cylinder A6.39 Figure A6.16 shows the same plot for rural households only: the relationship between price and uptake is much stronger. Al Hodeida has an extremely low LPG penetration rate in rural areas of only 8%, by far the lowest in any Governorate. 16 The low rate of LPG consumption in Hodeida has been known to YGC for some time. YGC has been trying to overcome local resistance to the use of LPG in the rural areas of this Governorate, which is apparently grounded in fears about safety. 75 76 Household Energy Supply and Use in Yemen Volume 2: Annexes Figure A6.16: LPG Penetration v. Average Price, Rural Households 1.2 1 SanaaGovern ]HH Lahj Adelah Shabwah Al Mahweet of 0.8 Ibb %[ Hadramouth Abyan AlBaida Dhamar iontart 0.6 Taiz Haja ne Sadah pe 0.4 Amaran GPL 0.2 AlHodeidah 0 220 240 260 280 300 320 340 360 Average price of 11 KgLPG cylinder A6.40 However, consumption per household and price are not related as one might expect: households consume more LPG in Governorates with higher LPG prices, not less, as one might expect (Figure A6.17). Figure A6.17: LPG Use per Household v. Average LPG Price (all Households and all Income Deciles) 60 h 50 AlBaida montd/ Adelah hole SanaaGovern usoH/gK 40 Dhamar AmaranHaja Al Mahweet Abyan Sadah Ibb Taiz on,i SanaaCity pt 30 Shabwah um Hadramouth onsc GPL20 Aden AlHodeidah Lahj 10 220 240 260 280 300 320 340 360 Average price of 11 KgLPG cylinder A6.41 Moreover, this is true even when one corrects for income decile and urban/rural location: Figure A6.18 shows the relationship for the bottom quintile of rural households. While the correlation is weak for this income quintile, for all rural users (and particularly the high income users), the relationship is statistically significant. Annex 6: LPG 77 Figure A6.18: LPG Use v. Price, Rural Households, Bottom Income Quintile 60 1.1.23 h ontmd/ 50 Adelah hole Al Mahweet 40 usoH/gK, 30 Ibb SanaaGovern Dhamar ion TaizAmaran AlBaida Sadah pt Abyan 20 Haja ums Hadramouth onc Lahj Shabwah AlHodeidah GPL10 0 220 240 260 280 300 320 340 360 Average price of 11 KgLPG cylinder LPG Subsidies A6.42 As noted in the introduction, LPG is the most highly subsidized of all petroleum products: in 2003, the domestic price was only 23% of the import parity price. The LPG subsidy is not included in many of the estimates of petroleum product subsidies (such as that compiled by YPC). Experience in other countries suggests that at the present level of subsidy, conversion of vehicles from gasoline to LPG will accelerate. Attempts to price domestic size bottles at a lower price than at filling stations will unlikely succeed, as again experience elsewhere shows that this simply leads to illicit conversion of vehicles to use domestic bottles. A6.43 Table A6.12 and Figure A6.19 show LPG subsidies by income decile. The top decile captures over three times the subsidy of the bottom decile (16% v. 4%).17 Table A6.12: Subsidies by Income Decile Subsidy For HH reporting use: consumption total captured by subsidy total subsidy each decile expenditure [1000 tons/year] [10^6 YR] [%] [YR/month] [YR/month] [% of total expenditure] 1 0-9000 28 1123 5% 1080 8481 12.7% 2 9001-12000 39 1537 7% 1209 16021 7.5% 3 12001-15000 51 2006 9% 1206 18389 6.6% 4 15001-19800 37 1474 7% 1213 21365 5.7% 5 19801-22500 55 2203 10% 1318 22990 5.7% 6 22501-27000 56 2227 10% 1309 27591 4.7% 7 27001-33000 65 2578 12% 1529 32867 4.7% 8 33001-42700 58 2315 11% 1422 33326 4.3% 9 42701-61000 75 2965 13% 1607 49248 3.3% 10 61001>0 91 3602 16% 1950 91469 2.1% Total 555 22029 100% 1415 31997 4.4% 17 Note that the subsidy shown here of YR30.3 billion relates only to the 88% of total LPG consumption by households, and is therefore correspondingly smaller than the total LPG subsidy shown in Table 5.9, Volume 1. 77 78 Household Energy Supply and Use in Yemen Volume 2: Annexes Figure A6.19: Fraction of Total LPG Subsidy to Households, Captured by Each Income Decile 0.2 %total subsidy 0.15 era 0.1 Sh 0.05 %HH budget 0 d[1] d[2] d[3] d[4] d[5] d[6] d[7] d[8] d[9] d[10] %total subsidy 5% 7% 9% 7% 10% 10% 12% 11% 13% 16% %HH budget 12.7% 7.5% 6.6% 5.7% 5.7% 4.7% 4.7% 4.3% 3.3% 2.1% Policy Option: Reduce the Subsidy on LPG A6.44 Table A6.13 shows the direct effect of raising the LPG price to 60%, 80% and 100% of the economic price. Raising the LPG to the economic price would have significant effects on the poor: the cost of living in the poorest decile would increase 6.5% ­ as opposed to only 2.0% for the top decile. Table A6.13: Effect of LPG Price Increases (on Households) Price level [% of economic 25% 60% 80% 100% Average price] effect on decile Price increase [YR/liter] 0.0 25.3 40.0 54.6 Economic price [YR/liter] 73.26 73.26 73.26 73.26 Retail price [YR/liter] 18.6 44.0 58.6 73.3 Subsidy [YR/liter] 54.6 29.3 14.7 0.0 Consumption Price elasticity -0.2 Consumption [1000 tons] 555 467 441 422 Impact on Government Subsidy [YR billion] 30.3 13.7 6.5 0.0 Net gain to government [YR billion] 16.6 23.8 30.3 Impact on households using LPG %HH affected Poorest decile [% of total present 49.5% 0.0% 6.1% 9.7% 13.2% 6.5% HH expenditure] Middle decile [% of total present 83.8% 0.0% 2.5% 3.9% 5.4% 4.5% HH expenditure] Richest decile [% of total present 94.2% 0.0% 1.0% 1.6% 2.2% 2.0% HH expenditure] Annex 6: LPG 79 Indirect Effects A6.45 The indirect uses of LPG are difficult to estimate, as are the related indirect effects of any price increases. Some LPG is used in transportation, replacing gasoline: if LPG prices were to increase, the incentive to convert vehicles would diminish (and in any event this would be to the disbenefit only of the top income deciles who can afford gasoline cars, removing their incentive to convert). As noted, some LPG is used in restaurants and bakeries: again the amounts are unknown, but it would be reasonable to assume that if prices at restaurants were to increase, it is the top income deciles who would be affected. A6.46 Smuggling of LPG is not likely to occur on a large scale. The need for pressurized cylinders makes smuggling much more difficult than for diesel. An official at YGC believed that no more than 2,000 tons/years were diverted in this way in 2003. In any event, whatever the quantity smuggled, raising the LPG price can only reduce whatever incentives presently exist for smuggling. In short, the indirect impacts of LPG price increases are unlikely to constitute a policy constraint for reducing LPG subsidies. Mitigating the Effect on the Poor A6.47 Reducing the LPG subsidy would bring a significant revenue gain to the Government, revenue that can be used in whole or in part to mitigate the effects on the poor. One possibility would be to pay every household a flat sum, including the mainly poor households that presently do not use LPG. This has the virtue of simplicity, because it does not require means testing, or who does and does not use LPG. By definition, a flat payment covers a larger share of the cost increase experienced by a poor household than by a large household, and therefore has the desired income redistribution effects. Moreover, for those (poor) households that do not presently use LPG, the flat sum would make a significant contribution to the up- front costs of buying the LPG stove and the first cylinder. A6.48 The most important feature of the flat sum payment is that the LPG user still experiences a cost increase on fuel purchases, which will motivate more efficient use of LPG for cooking. And therein lies the main gain to the Yemen economy, namely a more efficient use of resources. A6.49 Table A6.14 illustrates the calculations: At 60% of the economic price, LPG would experience an 88% cost increase (over the present level), which accounts for 6.4% of household income of the lowest decile, and 0.7% of the highest decile. Hence the need for returning some of the Government's additional revenue. In row [10] an assumption is made about the own-price elasticity of LPG demand ­ while there are no studies for Yemen, -0.3 would be a representative value based on experience in other countries. 79 80 Household Energy Supply and Use in Yemen Volume 2: Annexes With this elasticity, the total quantity consumed decreases from 711,000 to 588,000 tons/year, and the effective cost to Government reduces from YR28.1 billion to YR13.7 billion, a saving of YR14.4 billion, of which it is assumed 65% is returned. This results in a flat payment to all households of 4,180 YR /year. A6.50 For the lowest income decile, this YR4,180 payment [row 16] offsets the additional bill of YR4,965 [row 4], for a net increase of YR785 per year ­ which represents a tolerable 1% of total household expenditure, as opposed to 6.4% without mitigation. For the highest decile, the YR4,180 payment accounts for a much lower proportion of the increase (because this decile uses much more LPG) ­ but the net impact of mitigation is much smaller (decreasing from 0.7 to 0.4% of total household expenditure). (Figure A6.20) Figure A6.20: Net Effect of the Mitigation Scheme 0.08 e urti no mitigation ndepxe 0.06 HH of onitcarfsa,tceffeteN0.04 0.02 with flat rate compensation payment 0 bottom d[2] d[3] d[4] d[5] d[6] d[7] d[8] d[9] top 10% Annex 6: LPG 81 Table A6.14: Impacts of an LPG Price Increase to 60% of the Economic Price Monthly 0 9,001 12,001 15,001 19,801 22,501 27,001 33,001 42,701 61,001 income -9000 - - - - - - - - > total 12,000 15,000 19,800 22,500 27,000 33,000 42,700 61,000 decile bottom d[2] d[3] d[4] d[5] d[6] d[7] d[8] d[9] top per YR/month 10% year [1] present LPG use/HH/year [Kg/YR] 304 340 340 341 336 369 430 400 452 549 [2] LPG expenditure YR 5,664 6,345 6,329 6,365 6,271 6,870 8,024 7,462 8,432 10,241 /HH/year [3] total HH income/year 1000 YR 78 129 173 209 250 297 363 455 614 1,263 [4] additional expenditure Per year 4,965 5,562 5,548 5,580 5,498 6,022 7,034 6,541 7,392 8,977 [5] as % of present price [ ] 88% 88% 88% 88% 88% 88% 88% 88% 88% 88% [6] as % of HH income [ ] 6.4% 4.3% 3.2% 2.7% 2.2% 2.0% 1.9% 1.4% 1.2% 0.7% [7] Mitigation scheme [8] old LPG annual [10^6 kg] 36 50 65 48 71 72 83 75 96 116 711 consumption [9] old subsidy/ year [YR mill] 1,442 1,966 2,567 1,885 2,819 2,849 3,298 2,962 3,793 4,604 28,185 [10] assumed price elasticity -0.3 [11] new total LPG [mill Kg] 30 41 54 39 59 59 69 62 79 96 588 consumption/year [12] Subsidy at new [YR mill] 702 957 1,250 918 1,372 1,387 1,606 1,442 1,847 2,241 13,721 consumption/year [13] Savings to Govt./year [YR mill] 740 1,009 1,317 968 1,446 1,462 1,693 1,520 1,947 2,363 14,464 [14] proportion of revenue 0.65 returned [15] revenue returned/year [YR mill] 9401 [16] Flat Payment/HH/year [YR/HH] 4,180 4,180 4,180 4,180 4,180 4,180 4,180 4,180 4,180 4,180 [17] net price increase/HH [per year] 785 1,382 1,368 1,400 1,318 1,842 2,854 2,361 3,212 4,797 [18] as % of present price [ ] 13.9% 21.8% 21.6% 22.0% 21.0% 26.8% 35.6% 31.6% 38.1% 46.8% [19] as % of HH income [ ] 1.0% 1.1% 0.8% 0.7% 0.5% 0.6% 0.8% 0.5% 0.5% 0.4% A6.51 The proportion of revenue returned has been chosen in this example in such a way that the net impact on the lowest decile is no more than about 1% of total expenditure and that there remains a net price increase (needed to motivate more efficient use). However, as shown in row [18], while the poor household sees a net price increase of 13.9% in LPG purchase, the top decile sees a net LPG price increase of 46.8% ­ in other words, high income LPG users will have a greater incentive to conserve than low income users who use little LPG. That is another desirable attribute of a redistribution scheme. A6.52 There are undoubtedly some transaction costs to such a scheme, but even if these were to be in the billion YR range, if 65% of the total is returned, there remains a net fiscal gain to the Government (of some YR4 billion). 81 82 Household Energy Supply and Use in Yemen Volume 2: Annexes Policy Option: Subsidize LPG Cylinders A6.53 The high start-up costs of LPG (for purchase of the initial cylinder and for an LPG stove) are a significant obstacle to higher access rates among the poor: the initial purchase cost of a cylinder is YR2,500-3,000. And even where poor households do have access, LPG is used sparingly (a rural households in the bottom decile using LPG consumes 24kg/month, compared to 42kg/month in the top decile). Indeed, unlike kerosene, which can be bought in very small quantities, LPG must be bought in 11kg increments, which the poor often find difficult. A6.54 Therefore it has been proposed that the up front costs of moving to LPG be subsidized. However, private distributors already have a strong incentive to provide credit facilities to families in this situation and for very simple straightforward commercial reasons. According to YGC, most bottling operations presently operate far below capacity. Since the price structure reimburses bottlers according to the number of bottles they sell, recovering the up-front investment is largely dependent upon bottle throughput. It is therefore in the interest of the bottlers to move as many households to LPG as possible, so that they provide a steady stream of bottle purchases, and hence cash flow to the bottler. The costs of providing credit facilities evidently offset the increased income from higher bottle throughput. A6.55 Even if one could make a case for Government to provide a subsidy of this type to poor families, two questions need answers before such a scheme could be made effective: How are the poor to be identified? Would the recipients in fact use LPG if given a free cylinder and cook stove? A6.56 The international experience is relevant to Yemen, for such schemes have been tried elsewhere. The Deepam scheme in India (see Box A6.1) had a reasonably effective mechanism for identifying poor households by registered women's self-help schemes. However, in rural areas where free or cheap biomass is available, LPG was used by the recipients only very sparingly. The average cost of a cylinder refill is 270 Rs. for a 14.3kg cylinder (or about YR 845/11kg cylinder). Yet the maximum monthly household incomes of the recipients is 265 Rs. in rural areas, and 457 Rs. urban areas (YR1,060 and YR1,860, respectively). Thus a cylinder refill in rural areas covered by the Deepam scheme amounts to one month's income ­ clearly a very significant outlay. A6.57 Private LPG dealers selling to better-off customers report that the average household consumes about half a cylinder per month, or 7kg/HH/month (a rate that is less than half that observed in Yemen). Deepam recipients used only 2.6 kg/month in rural areas, and 4.8 kg/month in urban areas.18 18Many recipients of the free cylinder sold them (or even used them as part of dowries). A survey showed that the high cost of LPG was the main reason for discontinuing LPG use. Annex 6: LPG 83 A6.58 It is thus unclear that such schemes are sustainable, even when, as in the case of the Deepam scheme, qualified beneficiaries could be identified with reasonable certainty. Therefore the first task in Yemen, were such a scheme to be considered by Government, would be to develop a mechanism for identifying beneficiaries. In discussions held in September 2004, both Government officials and NGOs expressed skepticism that the SWF could effectively do so in the more remote rural areas where the need is greatest. Box A6.1: The Deepam Scheme in Andhra Pradesh, India Box A6.1: The Deepam Scheme in Andhra Pradesh, India The Government of India has attempted to encourage fuel switching from biomass to cleaner commercial fuels by providing large universal price subsidies to kerosene, sold through the Public Distribution System, and LPG sold in 14.2kg cylinders by dealers belonging to state- owned oil companies. A scheme providing price subsidies, however, does not address one of the barriers to household fuel switching to LPG: the high up-front cost associated with the start-up of LPG service. For example, a new LPG user in the state of Andhra Pradesh must (i) pay Indian Rupees (Rs.) 1,000 (about YR4,050) for an "LPG connection" in order to receive an LPG cylinder and (ii) purchase an LPG stove and associated accessories for a further Rs1,000 (YR4,050) or so. The combined cost of LPG connection and stove purchase makes it difficult for many poorer households to start using LPG as a cooking fuel. In order to help overcome this barrier, the Government of Andhra Pradesh launched the so-called Deepam scheme in July 1999 whereby the connection fee was paid by the Government for below-poverty-line (BPL) households possessing white ration cards. Those who do not possess white ration cards are also eligible provided that their self-help groups pass a resolution attesting to their BPL status. Deepam recipients still had to purchase their own stove, and were only given the LPG cylinder. The policy objectives of the Deepam scheme include (i) reducing drudgery among women and children from wood collection and cooking; (ii) improving the health of household members by reducing ambient concentrations of smoke and other harmful pollutants; and (iii) protecting forests from further degradation. The scheme was originally designed to cover one million rural and 0.5 million urban households. Only members of self-help groups satisfying certain criteria may participate in the scheme. There are more than 373,000 self-help groups in Andhra Pradesh with a total of more than five million members. About 150,000 of these self-help groups are in rural areas. As of February 2002, more than 1.5 million LPG connections had been released through the Deepam Scheme, including 1.2 million in rural areas. The majority of recipients were members of groups under the Development of Women and Children in Rural (or Urban) Areas (DWCRA and DWCUA, respectively). Source: S. Rajakutty and M. Kojima, Promoting Clean Household Fuels Among the Poor: Evaluation of the Deepam Scheme in Andhra Pradesh. World Bank, March 2002. 83 Annex 7 Gasoline and Fueloil Fueloil A7.1 Fueloil is not directly used by households. However, it should be included in the comprehensive revision of the system petroleum product pricing, and there are several important features of a pricing system that are well illustrated by fueloil (and that affect the overall household energy bill by virtue of the role of fueloil in electricity generation. A7.2 The price charged by the refinery to YPC (and its consumers) should bear some relationship to quality. Under the present pricing system, the cost to YPC (and PEC) for heavy fueloil is the same, regardless of sulfur content. However, as shown in Figure A7.1, prices and quality are linked: in Rotterdam, low sulfur fueloil typically trades for 1-2 $/bbl more than high sulfur fueloil (with similar differentials for Italy spot cargoes). Figure A7.1: Fueloil Price Differentials, Rotterdam 40 1% S 30 20 blb$/ 3.5% S avg.diff., $/bbl 2001 1.6 10 2002 1.1 2003 2.3 2004 0.9 0 -10 2001 2002 2003 2004 A7.3 PEC in particular has encountered problems with the fueloil that it receives from YPC, since the high sulfur content of some deliveries corrodes equipment. A rational pricing system would price fueloil according to its sulfur content, in accordance with world market differentials. 85 86 Household Energy Supply and Use in Yemen Volume 2: Annexes Box A7.1: Impact on Fisheries The Implementation Completion Report of the World Bank's 4th Fisheries Development Project includes a detailed analysis of the economics of small-scale fishing operations. The table shows the estimated costs for a six-boat, and a single-boat operation. Small fishing boats are powered by small, gasoline-powered outboard engines, predominantly the 15HP Yamaha engine, for which 1,250 units were imported by the project (of which 815 were in fact sold), in addition to 100 40HP Yamaha engines (all sold), and 350 25HP Selva engines (only seven sold!). At the present retail price of gasoline (35 YR/liter), fuel accounts for 6.9% of the total operating cost (including labor) in the six-boat operation, and 15.8% of the cost of a single-boat operation. The table also shows the costs if the price of gasoline were increased to the 2003 economic price of 41 YR/liter. Operating costs for the six-boat and single-boat operations increase by 2.4% and 4.0%, respectively. In YR/year Six-boat operation One-boat-operation at present at 2003 at present at 2003 price economic price price economic price Gross value of fish landings 16250000 16250000 1848000 1848000 Operating costs Cost of fuel, YR/liter 35 41 35 41 Cooperative charges (8% of sales) 1300000 1300000 147840 147840 Fuel: 10 gallons/boat/fishing day[100 days/year] 960000 1124571 192000 224914 (as % of total operating costs, including labor) 6.9% 8% 15.8% 18% Oil: 2 cans@YR200/fishing day 40000 40000 8000 8000 Food: 400 YR/fisherman/day x 36 fishermen 1444000 1444000 64000 64000 Engine spare parts (YR10,000/year/engine x 6) 60000 60000 10000 10000 Net Maintenance 72000 72000 12000 12000 Engine Maintenance 30000 30000 5000 5000 Boat maintenance & replacement 120000 120000 20000 20000 Net replacement (@YR 300,000/year) 300000 300000 50000 50000 Transport of fish (to market) 1625000 1625000 184800 184800 Engine replacement 300000 300000 50000 50000 (3year life, YR150,000/engine) Contingencies, 5% of above 313000 313000 54780 54780 Total operating costs 6564000 6728571 798420 831334 Increase in operating costs 164571 32914 2.4% 4.0% Gross income (before labor) 9686000 9521428 1049580 1016665 Labor cost 7264512 7264512 418548 418548 Net income 2421488 2256917 631032 598118 Decrease in margin 164571 32914 7.3% 5.5% Source: World Bank, Implementation Completion Report: Republic of Yemen Fourth Fisheries Development Project, Report 20015-YEM, March 6, 2000. Gasoline A7.4 Gasoline was also not included in the household survey. But as in the case of fueloil, gasoline should be priced on the same basis as other fuels, and prices adjusted regularly (as discussed in Volume 1). A7.5 Direct purchases of gasoline do play a role in coastal households dependent on fishing. These households would be (slightly) affected by bringing gasoline to the economic price as well. While it is true that large commercial fishing trawlers may be diesel powered, small-scale fisheries predominantly use gasoline-powered outboard Annex 7: Gasoline and Fueloil 87 motors. Box A7.1 illustrates the potential impact of a gasoline price increase on small-scale fisheries (e.g. cooperatives as may operate small fleets, or single-boat operations). The increase in operating costs were the gasoline price increased by 6 YR/liter (to bring it to the 2003 economic cost) is between 2 and 4%. While this is obviously not trivial, it hardly represents the type of devastating impact on the livelihoods of small fisherman envisaged by popular imagination. A7.6 As in the case of fueloil, the pricing basis for gasoline should be rational. Notwithstanding that Yemen sells regular leaded gasoline, it is understood that the price basis used by the refinery for sales to YPC is premium unleaded. However, as shown in Figure A7.2, there is little difference between leaded and unleaded gasoline prices in Italy, and in 2003 and 2004 leaded gasoline was slightly more expensive than unleaded. This is again a reflection of market conditions in Europe, where there is low demand for leaded gasoline. Thus it matter little whether the price basis is a European premium unleaded or leaded; neither is appropriate. Instead, the appropriate pricing basis is Platts Gulf. Figure A7.2: Difference between Leaded and Unleaded Gasoline, Spot Cargoes, Italy 60 40 Leaded (0.15gTEL/l) l Unleaded 95 bb$/ 20 0 -20 2002 2003 2004 Subsidizing the Refinery A7.7 The subsidy calculations presented in Volume I suggest that the subsidy as stated in the official Government figures is overstated, because it includes subsidies that are in effect provided to the refinery. This is not say that the entire amount of the subsidy to the refinery is attributable to uneconomic operations at the refinery per se (attributable to negative refining margins that one would normally expect at such an old refinery). Since the refinery accounts were not available this figure cannot be broken down into its actual components. However, three conclusions can be drawn: The figure recorded by MoF and MOM as subsidy on petroleum products is significantly overstated if the term subsidy is to be used in its normal meaning: i.e. as the difference that arises between the economic price based on actual border price, and the domestic retail price. 87 88 Household Energy Supply and Use in Yemen Volume 2: Annexes The corollary is that if domestic prices are raised to the notional economic price as presently defined (i.e. Rotterdam or Italy plus notional freight), then they would be too high: the border price as used in any price formula or price calculations should be based on Gulf prices plus actual freight. It is unclear that refining is economic at all. Refining margins at simple hydroskimming refineries are rarely adequate. The rationale for refining Marib crude is unclear, since the domestic product mix is a very poor match to its distillation yields, as shown in Figure A7.3 (resulting in the export of large quantities of naphtha). Figure A7.3: Crude Yields v. Yemen Domestic Product Market Slate 120 100 GASOLINE NAPHTHA 80 KEROSENE KEROSENE 60 DIESEL GASOIL 40 20 FUELOIL FUELOIL 0 YEMEN MARK MARIB LIGHT ZAKHUMLOW UMMSHAIFF LAVAN QATAR MARIN FOROZAN OMAN MASILA QATAR LAND MURBAN SYRIAN LIGHT SIRRI DUBAI BASRAH LIGH SOUEDIE Annex 8 Electricity Patterns of Electricity Consumption Access to Electricity A8.1 Table A8.1 shows access to electricity by the classification used in the survey. While 91% of urban households report access to electricity (of which 79.3% are served by PEC), only 42% of rural households are electrified, and, of these, only 23% are served by PEC's national grid. Table A8.1: Electricity Access Urban Rural All [#HH] [%] [#HH] [%] [#HH] [%] PEC national grid 402,747 79.3% 400,724 23.0% 803,471 35.7% PEC isolated system 0 0.0% 56,988 3.3% 56,988 2.5% Cooperative 21,118 4.2% 31,927 1.8% 53,045 2.4% Private 0 0.0% 2,328 0.1% 2,328 0.1% Village/community 22,603 4.4% 157,414 9.0% 180,017 8.0% Relative/neighbor 12,642 2.5% 26,771 1.5% 39,413 1.8% Family-owned 6,311 1.2% 52,484 3.0% 58,795 2.6% Other 0 0.0% 6,724 0.4% 6,724 0.3% Total non-grid 62,674 12.3% 334,636 19.2% 397,310 17.7% Total with electricity 465,421 91.6% 735,359 42.2% 1,200,781 53.4% HH with no access 42,665 8.4% 1,005,727 57.8% 1,048,392 46.6% Total HH 508,086 100.0% 1,741,087 100.0% 2,249,173 100.0% A8.2 Other notable features of access patterns include: widespread interconnection of family-owned systems to neighbors. Of 58,795 family-owned self-generation systems, 67% also serve neighboring households. there are very few households served by privately-owned systems; by far the largest number of rural households who do not have grid access are served by village/community-based systems. of the total households that do not have access (1,048,392), 96% (1,005,727) are in rural areas. As expected, lack of access to electricity is a rural issue. 89 90 Household Energy Supply and Use in Yemen Volume 2: Annexes A8.3 Access to electricity is strongly dependent on income and on the urban/rural divide (Figure A8.1). In the poorest decile, 76% of urban, but only 18.6% of rural households have electricity access. However, in the top decile, the difference between urban and rural is much smaller (95 v. 76%). Figure A8.1: Access to Electricity by Income Decile 1 URBAN sseccaytic 0.8 0.6 RURAL tricele thiw 0.4 noticarf0.2 0 Income decile 1 2 3 4 5 6 7 8 9 10 Urban 75.8% 84.2% 90.9% 93.3% 88.5% 90.4% 92.0% 95.3% 94.7% 95.6% Rural 18.6% 22.0% 35.8% 38.5% 43.5% 49.5% 55.3% 53.3% 68.1% 76.0% A8.4 Indeed, access to electricity is strongly correlated to income. Table A8.2 shows monthly household income by type of access. The average monthly income of those with electricity access, some 41,000 YR/month, is almost double that of households without access. Table A8.2: Electricity Access and Income electricity total HH income access [HH] [%HH] [YR/month] PEC national grid 803471 35.7% 43950 In urban areas 402747 17.9% 44573 In rural areas 400724 17.8% 43323 PEC isolated system 56988 2.5% 36428 Coop 53045 2.4% 34684 Private 2328 0.1% 38727 Total grid/minigrid access 915832 40.7% 42932 Village/community genset 180017 8.0% 29379 Relative/neighbor genset 39413 1.8% 27880 Family genset 58795 2.6% 50025 Other 6724 0.3% 47817 Total with electricity access 1200781 53.4% 40781 No access 1048392 46.6% 22282 Total 2249173 100.0% Annex 8: Electricity 91 Box A8.1: Inequality of Access Another way of displaying this income-dependence of electricity access is to examine the distribution of the type of access by income decile, shown in the table. If access was income-neutral (as it is in developed countries), one would expect that among grid-connected households, roughly 10% would be in each decile. But of PEC connections, only 3% are to be found in the bottom decile Distribution of Access Income per No PEC Minigrids Self-Gen decile (YR Access National Grid /month) PEC Coop Private village relative/ Family- neighbor owned 1 0-9000 18% 3% 12% 4% 0% 4% 11% 3% 2 9001-12000 16% 6% 3% 7% 9% 10% 3% 0% 3 12001-15000 14% 8% 4% 2% 9% 16% 13% 7% 4 15001-19800 9% 6% 1% 7% 0% 12% 16% 7% 5 19801-22500 11% 8% 11% 10% 18% 12% 18% 11% 6 22501-27000 9% 11% 19% 19% 18% 10% 4% 2% 7 27001-33000 8% 12% 17% 17% 9% 9% 14% 13% 8 33001-42700 8% 11% 14% 18% 18% 11% 7% 20% 9 42701-61000 5% 17% 7% 11% 9% 7% 7% 16% 10 61001>0 4% 17% 13% 7% 9% 8% 7% 21% total 100% 100% 100% 100% 100% 100% 100% 100% 0.2 No Access 0.15 la ott of income-neutral access onit 0.1 opor pr 0.05 PEC grid 0 0-9000 12001-15000 19801-22500 27001-33000 42701-61000 9001-12000 15001-19800 22501-27000 33001-42700 61001>0 income decile Service Reliability A8.5 As expected, there are significant differences in service quality between grid- connected customers and those connected to isolated systems and self-generation sets. The survey asked households to report on average hours of service per day, whose results are shown in Table A8.3. 83% of grid connected customers report 23-24 hours of service per day, whereas the bulk of self-generation and mini-grid customers report service for 4-6 hours per day. Surprisingly, there are few differences between PEC's urban and rural customers. 92 Household Energy Supply and Use in Yemen Volume 2: Annexes Table A8.3: Hours of Service PEC grid Isolated systems All Hours Urban Rural total PEC Coops Private Village Neighbor isolated of service areas areas systems 1 1% 1% 10% 5% 2% 2 1% 8% 6% 1% 2% 3 1% 7% 3% 1% 3% 4 1% 3% 17% 12% 11% 5 1% 12% 29% 49% 24% 6 34% 21% 100% 23% 23% 25% 7 1% 7% 4% 8 2% 1% 14% 5% 1% 5% 4% 9 1% 1% 30% 5% 7% 10 2% 1% 1% 1% 1% 11 7% 1% 2% 2% 12 2% 3% 2% 5% 13% 8% 13 4% 2% 14 1% 1% 15 16% 4% 3% 16 17 18 1% 3% 2% 11% 2% 19 1% 1% 20 1% 5% 3% 21 1% 1% 1% 1% 22 2% 6% 4% 23 6% 14% 10% 24 77% 64% 71% 1% 1% Figure A8.2: Hours of Service, All Isolated Systems and Self-gen Sets PEC national grid Isolated systems and self-generation 1.1.24 800 100 25% 80 24% 600 71% ncy queerF 400 ousands ycneuqe 60 nds Th Fr Thousa 40 11% 8% 200 7% 20 10% 0 0 0 2 4 6 8 10 12 14 16 18 20 22 24 0 2 4 6 8 10 12 14 16 18 20 22 24 hoursof service hoursof service Annex 8: Electricity 93 Consumption Of Electricity A8.6 The average urban household with electricity access consumed 274 kWh/month, as opposed to only 101 kWh/month in rural households. As shown in Table A8.4, consumption patterns are strongly dependent upon the type of access: consumption in the PEC national grid is typically double that of isolated systems (whether PEC or cooperative). Table A8.4: Average Monthly Consumption, kWh/HH Urban Rural PEC national grid 274 101 PEC isolated system 74 Coop 91 44 Private 45 village/community genset No data No data relative/neighbor genset No data No data family genset No data No data other No data No data A8.7 However, there is large variation in individual household consumption rates, even within narrowly defined categories. For example, as shown in Figure A8.3, monthly consumption among PEC national grid connected customers varies from less than 20 kWh to over 800kWh/month. Figure A8.3: Electricity Consumption, kWh/month (Top Income Decile, Rural PEC Grid Customers) 8 11% 6 10% ncy ds 7% queerF san 4 7% 7% ou 6% 7% 6% Th 5% 2 0 0 100 200 300 400 500 600 700 800 900 50 150 250 350 450 550 650 750 850 >925 Monthly kW h, top income decile, rural PEC national grid connected A8.8 Most governorates follow much the same patterns of consumption by income level as the national average (though the level of consumption across governorates varies more, see Figure A8.4). 94 Household Energy Supply and Use in Yemen Volume 2: Annexes Figure A8.4: Electricity Consumption v. Income 500 400 h 300 Wk SanaaCity onthlym200 average 100 Taiz 0 bottom 10% d[3] d[5] d[7] d[9] d[2] d[4] d[6] d[8] top 10% Tariff Structure A8.9 PEC has two domestic tariffs; one for its grid-connected customers, and one for its "rural" customers, which in fact does not mean rural (in its normal administrative or practical definition), but customers in its isolated systems. Box A8.2 shows provides the details of the structure. Therefore, of PEC's total number of customers (see Table A8.1), only 56,000, or 7%, pay the rural tariff. A8.10 The existing tariff structure raises several questions: to what extent the tariff recovers PEC's costs to what extent does the 1st least-cost tariff block serve as an effective "lifeline" rate for the poor? to what extent do the differences in rural and grid tariffs reflect differences in actual economic costs of the two types of service? to what extent does the high connection charge discourage formal connections? Cost Recovery A8.11 PEC does not cover its present costs, notwithstanding the subsidy on diesel fuel. At this point it is not possible to make recommendations on the structure of the tariff because this requires, as a first step, an understanding of the actual economic costs of supply at different voltage levels, properly reflecting the economic costs of generation, transmission and distribution. It is recommended such a study be undertaken as soon as possible (perhaps as one of the background studies for the proposed Rural Electrification Project). Annex 8: Electricity 95 Lifeline Rate A8.12 As shown in Table A8.5, the first block in the PEC tariff for urban customers, whose purpose should be to provide a first tranche of low cost power to poor households, is set at 200 kWh/month. This is substantially higher than in other countries, and there is no evidence that it effectively serves this role. Table A8.5: International Comparisons of the First Tariff Block kWh Indonesia 20 India (Gujrat, GSEB) 20 India (Ahmedabad, Kolkata) 25 Egypt 50 Pakistan 50 Laos 50 India (Bombay, BSES) 100 Bangladesh (BPDB/DESA) 100 Yemen: Isolated systems 100 National grid 200 Source: World Bank, Energy Sector Performance Improvement and Future Development: The Way Forward, Washington, D.C, 2002. 96 Household Energy Supply and Use in Yemen Volume 2: Annexes Box A8.2: PEC Tariff The current PEC tariff for domestic customers is as shown in the table below Urban Rural Fixed charge (1 phase) YR/month 300 300 Variable charge 0-100 YR/kWh 4 7 101-200 YR/kWh 4 17 201-350 YR/kWh 7 17 350-700 YR/kWh 10 17 `>700 YR/kWh 17 17 The corresponding average cost per kWh, as a function of monthly consumption, therefore follows as shown below: 50 40 h 30 Wk/ YR 20 RURAL 10 URBAN 0 0 200 400 600 800 1000 1200 1400 kWh/month and the total monthly bill as a function of consumption follows as 20 15 llbiyhl RURAL 10 nt mo URBAN 5 0 0 200 400 600 800 1000 1200 1400 kWh/month Annex 8: Electricity 97 A8.13 Yet as shown in Table A8.6, the average monthly consumption in rural areas is 101 kWh, and even the top decile consumes only 137 kWh. The average consumption of the bottom decile is 94 kWh, which would argue for a first block of no more than 100 kWh per month, and certainly not 200 kWh per month as at present. Table A8.6: Monthly Consumption, PEC Grid Customers Income per decile %HH reporting use Consumption, (YR / month) [kWh/month] Urban Rural All Urban Rural All 1 0-9000 51% 8% 12% 157 58 94 2 9001-12000 74% 8% 19% 228 64 174 3 12001-15000 79% 16% 25% 188 65 123 4 15001-19800 74% 16% 28% 258 94 182 5 19801-22500 72% 19% 29% 228 77 144 6 22501-27000 81% 25% 39% 264 95 181 7 27001-33000 82% 28% 43% 275 106 191 8 33001-42700 81% 24% 41% 283 81 171 9 42701-61000 90% 46% 60% 291 114 196 10 61001>0 83% 52% 63% 388 137 237 average 79% 23% 36% 273 101 183 Note: Rural refers to HH in rural areas, not HH paying the REC rural tariff! A8.14 In the absence of a cost study it is difficult to make specific suggestions for a residential tariff structure. However the HES data do suggest that the present structure requires revision, and that (as with petroleum product subsidies), the first block results in poor targeting of the implied subsidy. PEC Rural Tariff (For Isolated Systems) A8.15 As in the case of the PEC grid system, the extent of subsidy to customers of the isolated systems is not transparent, and again the economic costs of service to isolated systems need to be clearly established. The available data (see Box A8.3) suggest very high connection costs for these systems. Although many isolated systems appear to have benefited from grant aid, the value of this assistance is not included in the PEC database: this should be corrected in any proper assessment of economic costs. A8.16 Figure A8.5 shows the ratio of monthly household bills of the rural (isolated system) and normal PEC tariffs, as a function of the kWh consumed. The rationale for the very different relative block structure is quite unclear, reaching a peak of 2.5 times the normal tariff for consumption between 200 and 350 kWh/month. 98 Household Energy Supply and Use in Yemen Volume 2: Annexes Figure A8.5: Ratio of Monthly Bills, Rural to Urban, as a Function of kWh na 3 rbu:la 2.5 ruffriateg 2 raevafo 1.5 oit 1 Box A8.3: Costs of Isolated Systems ra 0.5 0 200 400 600 800 1000 1200 1400 kWh/month Annex 8: Electricity 99 Box A8.3: Costs of Isolated Systems PEC data highlight the high cost of isolated rural systems. A PEC database contains data on 410 isolated systems implemented over the past 15 years, and includes total project cost (including meters and the cost of house connections), system capacity (as kVA), and the number of households connected. The total cost of each scheme given in this database excludes the value of grants (mainly from the Govt. of Japan), and such systems are not therefore included in this analysis. The data on number of households are doubtful; when summed, it shows over 800,000 households, whereas we know from more reliable data from the commercial accounts division of PEC that the number of systems in PEC rural (isolated) systems is around 56,000. Therefore the costs per household connection are quite low (the mean is less than $400/HH). The figure given for installed capacity (as kVA) is probably more reliable: the figure below shows the frequency distribution of $/kVA. 20 15 y nc ue 10 eqrf 5 0 0 400 800 1200 1600 2000 2400 2800 3200 3600 4000 200 600 1000 1400 1800 2200 2600 3000 3400 3800 >1000 The data exhibit the classic expected economies of scale, with falling $/kVA as system size increases. The median cost is $1,072/kVA. (The average of $1,720 is distorted by a few outliers that have costs in excess of $10,000/kVA). 5 4 aVk dsna 3 $/ housT2 1 0 0 500 1000 1500 2000 2500 system capacity, 1000 kVa Thus, given an average demand per connected household of 300 watts, the cost per connection may be estimated at over $3,573/HH (including the cost of house connections and meters). Source: Engineer Waheeb, Rural Electric Projects in the Republic of Yemen, PEC, 2004 100 Household Energy Supply and Use in Yemen Volume 2: Annexes Electricity Expenditure A8.17 The distribution of household electricity expenditure closely tracks that of consumption, which follows from the single national tariff structure in the PEC grid. Thus the frequency distribution of Figure A8.6 (expenditure) follows closely that of Figure A8.3 (kWh consumption). Figure A8.6: Distribution of Household Electricity Expenditure 0.25 0.2 la ott 0.15 of onit ac 0.1 Fr 0.05 0 0 1000 2000 3000 4000 5000 6000 7000 8000 9000 500 1500 2500 3500 4500 5500 6500 7500 8500 >9250 Electricity expenditure, YR/month Rural HH Urban HH Table A8.7: Electricity Consumption and Expenditure Data %HH reporting use Consumption,[kWh/monthExpenditure, [YR/month] Reported price,[YR/kWh] Income per decile Urban Rural All Urban Rural All Urban Rural All Urban Rural All (YR / month 1 0-9000 76% 16% 22% 157 58 94 939 577 719 8 15 12 2 9001-12000 84% 19% 30% 228 64 174 1437 740 1201 8 15 10 3 12001-15000 91% 34% 42% 188 65 123 1186 804 978 7 20 14 4 15001-19800 93% 36% 47% 258 94 182 1674 1060 1374 7 15 11 5 19801-22500 88% 42% 51% 228 77 144 1386 811 1058 7 13 11 6 22501-27000 90% 48% 59% 264 95 181 1675 973 1291 8 15 11 7 27001-33000 92% 53% 65% 275 106 191 1724 1213 1470 7 13 10 8 33001-42700 95% 51% 64% 250 81 171 1591 889 1242 7 16 12 9 42701-61000 95% 67% 76% 291 114 196 1907 1325 1583 7 13 10 10 61001>0 96% 75% 82% 388 137 237 2662 1758 2122 7 16 12 average 92% 42% 53% 273 101 183 1754 1154 1433 7 15 11 1 ytiicrtcleE 0.8 RURAL 0.6 ing us HH 0.4 ofniotcarF0.2 0 d[1] d[2] d[3] d[4] d[5] d[6] d[7] d[8] d[9] d[10] URBAN 76% 84% 91% 93% 88% 90% 92% 95% 95% 96% RURAL 16% 19% 34% 36% 42% 48% 53% 51% 67% 75% Annex 8: Electricity 101 A8.18 Note that rural customers pay very high prices per kWh (as a consequence of low kWh use), and therefore the monthly fixed charge of YR300 dominates the monthly bill. The fraction of monthly income spent on electricity is also as expected, as shown in Figure A8.7: as income increases, the fraction of total household expenditure spent on electricity declines. Figure A8.7: Fraction of Income Spent on Electricity 0.2 yticir ctele Abyan 0.15 notn SanaaCity spe em 0.1 coniy average hltnomfo 0.05 Taiz noit acrf 0 bottom 10% d[3] d[5] d[7] d[9] d[2] d[4] d[6] d[8] top 10% Willingness to Pay for Electricity A8.19 Benefit-cost analysis of rural electrification requires (obviously) some measure of the benefits of electrification. If a demand curve were available, then the total economic benefits of some level of consumption, Q, follow as the area under the demand curve (i.e. the total willingness to pay), and the net benefits as the consumer surplus (which is total willingness to pay less the actual cost of consuming Q units at price P). A8.20 Thus, in Figure A8.8, a demand curve for lighting (as lumens) is depicted. Two points are shown. The first point, x, represents the demand for lumens in an unelectrified household that uses kerosene for lighting; for the kerosene consumption of the household (as revealed in a survey) and given knowledge of the type of lamp employed, one may derive the number of lumen-hours provided (QKERO), and the cost per lumen hour (PKERO). At this point the household enjoys a consumer surplus equal to the area A. 102 Household Energy Supply and Use in Yemen Volume 2: Annexes Figure A8.8: Demand for Lighting Service price A PKERO x B C y PE D E QKERO QE service level A8.21 The second point, y, represents that same household after electrification; this household consumes a far greater number of lumens (QE) at the much lower price of the grid tariff PE. Now the consumer enjoys a surplus of the area A + B + C. From this follows that the economic benefit of electrification to this household is the change (increase) in consumer surplus, namely B + C.19 A8.22 This approach works best in longitudinal surveys, where information is available from a household before and after electrification, and has been used in a number of recent World Bank studies in Vietnam (for electrification by mini-hydro), the Philippines (for electrification by diesel mini-grids), and Sri Lanka and Indonesia (for electrification of rural households in remote areas by solar homes). The method is easiest to apply for solar systems, because the quantity of electricity provided is small, and therefore gets used just for lighting and TV-viewing, for which deriving demand curves is tractable. A8.23 In the case of the Yemen energy survey, although the derivation of such demand curves is difficult, valuable insights about household behavior and preferences may still be drawn. Table A8.8 shows spending on fuels that are in theory substitutes for electricity ­ LPG and kerosene for lighting, candles, dry cells and battery charging. For example, the table shows that households connected to the PEC grid that use kerosene for lighting spend 134 YR/month on kerosene; but households with no access to electricity spend 332 YR/month on kerosene. Overall, households with access to electricity spend 418 YR/month on items (that could be replaced by electricity), as opposed to 779 YR/month in households that have no access. 19Because the prices seen by consumers are financial prices, they must be adjusted for taxes and subsidies implicit in both the price of kerosene, and in the price of grid supplied electricity. These adjustments can be complex because the final price may simultaneously embody both taxes (e.g. VAT on electricity) as well as subsidies (e.g. in subsidies for rural electrification) Annex 8: Electricity 103 Table A8.8: Spending on Electricity Substitutes (all Households) total kerosene batterycharging dryCell LPG lighting LPGfridge candles [YR/m] [YR/m] [%HH] [YR/m] [%HH] [YR/m] [%HH] [YR/m] [%HH] [YR/m] [%HH] [YR/m] [%HH] PEC national grid 307 134 36% 212 1% 224 53% 204 18% 594 0% 119 58% at PEC urban tariff at PEC rural tariff PEC isolated system 499 149 64% 353 1% 304 71% 265 48% 0% 84 28% Coop 407 191 66% 314 1% 257 58% 244 19% 484 6% 98 19% Private 1043 162 91% 0% 607 100% 198 100% 0% 141 64% total grid/minigrid access 328 141 40% 226 1% 234 54% 215 20% 512 0% 117 54% village/community gense 488 156 57% 314 4% 308 68% 247 31% 1159 1% 149 33% relative/neighbor gense 575 257 41% 277 12% 372 66% 245 35% 1250 1% 185 25% family genset 1365 306 43% 332 20% 716 80% 421 52% 918 2% 760 30% other total with electricity acce 418 160 42% 309 3% 287 58% 245 24% 703 1% 141 49% no access 779 332 78% 354 8% 439 64% 401 34% 492 0% 197 22% A8.24 Households that have access to electricity still use substantial quantities of the alternative fuels, presumably because the cost of kerosene (and LPG) is so cheap that electricity is used sparingly for lighting (because of the perception of high cost): electricity is used for services for which there are no (or no cost-effective) substitutes ­ such as TVs, fans, and other appliances. A8.25 Table A8.9 shows the incremental expenditures. Column [1] shows electricity expenditure, and column [2] expenditure on substitutes; column [3] is the total expenditure (=[1]+[2]). The incremental expenditure for electricity in column [4] is relative to households with no access: for example, in households connected to the PEC national grid, the incremental expenditure (for electricity) is 1,072 YR/month. Table A8.9: Incremental Expenditure Electricity Non- electricity Total Incremental expenditure expenditure expenditure expenditure [1] [2] [3] [4] PEC national grid 1543 307 1850 1072 PEC isolated system 663 499 1162 383 Coop 570 407 977 198 Private 1355 1043 2397 1618 Village/community genset 1166 488 1654 875 Relative/neighbor genset 1386 575 1962 1183 Family genset 3822 1365 5187 4409 No access 0 779 779 A8.26 The most interesting finding of these data is highlighted in Figure A8.9: households with the highest expenditure for electricity are those connected to private mini-grids, or own gensets ­ but these households spend more on electricity substitutes than those with no access at all, and more than those connected to the PEC grid. Evidently these households place great value not just on electricity, but on the entire bundle of energy services. 104 Household Energy Supply and Use in Yemen Volume 2: Annexes Figure A8.9: Expenditure on Electricity and Electricity Substitutes, by Type of Electricity Access 6 1000RY,eru 4 nditepxe 2 ELECTRICITY hly ntoM E-SUBSTITUTES 0 PEC national grid Coop village/community ge family genset PEC isolated system Private relative/neighbor gen no access Productive Use of Electricity A8.27 The extent of productive use of electricity in rural households is one of the enduring themes of the rural electrification debate. In Yemen, 7.6% of rural households reported some form of home business: as shown Table A8.10, small food/groceries (1.8%) and crop processing (2.3%) were the two most commonly reported business types. Table A8.10: Home Business in Rural Households by Electricity Access #HH as % of PEC grid With non-grid With no access reporting all HH access access Food/beverage grocery 31105 1.8% 14062 45% 8210 26% 8834 28% Retail sales shop, pharmacy 1544 0.1% 0 0% 1544 100% 0 0% Storage space 9504 0.5% 2921 31% 1544 16% 5038 53% Repair shop 1271 0.1% 572 45% 699 55% 0 0% Handicraft production/sales 18763 1.1% 5878 31% 8303 44% 4582 24% Furniture making, carpentry 0 0.0% 0 0 0 Hair salon/barbershop 1349 0.1% 0 0% 0 0% 1349 100% Crop processing, milling 39948 2.3% 21628 54% 3168 8% 15152 38% Laundry 0 0.0% 0 0 0 Bakery 3295 0.2% 1362 41% 1934 59% 0 0% Other 26334 1.5% 3785 14% 10013 38% 12535 48% Total 133114 7.6% 50208 38% 35415 27% 47490 36% total HH 1752551 40348 23% 33228 19% 1016777 58% 6 9 A8.28 Such home businesses are indeed concentrated in homes with electricity access. 23% of all households have PEC grid access, but 38% of households with home businesses have PEC access. Similarly, 19% of all households have non-grid access (mini-grids, self generation) but 27% with home businesses have non-grid access. Annex 8: Electricity 105 A8.29 Pharmacies are a good example of a home business that is only possible with some form of electricity access (for refrigeration) and no household without electricity reports such a business. On the other hand, barber shops (as a home business) apparently are reported only in households with no access. A8.30 Interpretation of these data requires caution: obviously there are many barbershops and pharmacies in areas served by the grid. However, in such areas these are commercial establishments and therefore not included in the HES. Nevertheless, the data do show that households that have electricity access have a higher incidence of home business than those without. Annex 9 Biomass A9.1 Four types of biomass fuel were examined in the survey: fuelwood, charcoal, crop residues, and dung. Despite the large-scale uptake of LPG, a surprising 74% of all households report use of firewood (Table A9.1). The rate of decline with increasing income is modest: while 80% of households in the lowest income decile report fuelwood use, this declines only to around 70% in the middle deciles, and 66% in the top decile. Table A9.1: % of Households using Biomass Fuels Decile Fuelwood Charcoal Crop dung residue 1 80 8 24 12 2 82 12 24 19 3 84 11 31 27 4 82 15 30 22 5 70 11 22 19 6 71 18 18 13 7 70 24 19 15 8 71 19 22 18 9 59 20 19 18 10 66 30 20 21 All 74 17 23 18 deciles A9.2 Dung and crop residues are overwhelmingly collected (at no money cost, though at the cost of family time), while substantial fractions of fuelwood are purchased (Table A9.2). 107 108 Household Energy Supply and Use in Yemen Volume 2: Annexes Table A9.2: How do Households Obtain Biomass Fuels, as % of Households Using Fuels? Decile Fuelwood Crop residue Dung Purchase Purchase Collect Purchase Purchase Collect Purchase Purchase Collect only and only only and only only and only collect collect collect 1 11 6 83 1.4 98.6 0 200 2 18 14 68 2.6 93 5 94 6 3 12 14 74 0.7 93 7 96 4 4 22 13 65 94 6 1 98 1 5 22 7 72 0.5 89 11 5 91 5 6 34 6 61 0.1 100 3 96 2 7 21 9 70 97 3 99 1 8 33 12 55 98 2 95 5 9 30 12 58 0.5 91 9 98 2 10 44 10 47 2.2 89 9 2 98 All deciles 24 10 66 1 94 5 1 96 3 Rural 19 11 70 Urban 61 7 32 As percentages of households who use the fuel in question A9.3 As expected, the extent to which fuelwood is purchased is strongly dependent upon income (Figure A9.1). In the bottom decile, only 17% purchase fuelwood, as opposed to 54% in the top decile. Figure A9.1: Method of Obtaining Fuelwood 1.2 1 collect only HHlalfonoit 0.8 0.6 acrF 0.4 0.2 purchase & collect purchase only 0 0-9000 12001-15000 19801-22500 27001-33000 42701-61000 9001-12000 15001-19800 22501-27000 33001-42700 61001>0 income decile Fuelwood Collection A9.4 Figure A9.2 shows the distribution of fuelwood collection distances: the average is 2km. However, 30% of households report collection distances more than 3km. Annex 9: Biomass 109 Figure A9.2: Distribution of Fuelwood Collection Distances 500 27% 400 y 300 19% nc queerf 15% 200 12% 8% 8% 100 4% 4% 1% 1% 0% 0% 0% 0% 0% 0% 0 0 0.2 0.5 1 2 3 4 5 6 7 8 9 10 11 12 13 collection distance, km A9.5 There is surprisingly little variation across governorates (Figure A9.3). Sana'a City is the outlier, with an average reported distance of 4.8km ­ though this is certainly unsurprising given the situation in the capital and its surroundings. However only 18 sample households reported collecting fuelwood in Sana'a city. Figure A9.3: Average Fuelwood Collection Distances by Governorate average collection distance, km 0 1 2 3 4 5 6 Ibb 1.9 Abyan 2.1 SanaaCity 4.8 AlBaida 1.6 Taiz 1.9 Haja 2.3 AlHodeidah 2.0 Hadramouth 2.6 Dhamar 2.1 Shabwah 1.9 Sadah 1.7 SanaaGovern 2.1 Aden 2.8 Lahj 1.9 Al Mahweet 1.8 Amaran 3.0 Adelah 2.7 A9.6 Simple tabulations of the time reported by each gender/group on fuelwood collection can be misleading: indeed, given that an adult woman collects fuelwood, the average time spent per collection (3.8 hours) is in fact less than the average time spent by adult males (4.6 hours).20 But this says little about the proportion of men and women engaged in collection. There is also large variation in the number of collections per month, as shown in Figure A9.4: while 50% of households collect once a week, small numbers collect once a month, and yet others collect daily. 20This apparent anomaly is explained below: there is a greater propensity for men to participate in fuelwood collections when distances are great (or perhaps if the fuelwood collected is sold). 110 Household Energy Supply and Use in Yemen Volume 2: Annexes Figure A9.4: Frequency Distribution of the Number of Collections per Month 600 49% 400 y nc ueqerf 19% 200 10% 7% 5% 4% 2% 1% 1% 1% 2% 2% 0% 0% 0% 0% 0 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 collections per month A9.7 Thus it becomes necessary to bring everything to a common basis. This is done by calculating for each household that reports fuelwood collection a total monthly family time budget, B, calculated as B = Nc (Hw Nw + Hm Nm + Hg Ng+ Hb Nb) A9.8 Where Nc is the number of collections per month, HW is the hours per collection reported by women, and NW the number of women (in the household) who on average participate in each collection, and correspondingly for adult men (subscript m), boys (b) and girls(g). The resulting frequency distribution of the monthly time budget is shown in Figure A9.5. Figure A9.5 Monthly Household Time Budget for Fuelwood Collection 800 60% 600 y nc 400 queerf 19% 200 6% 4% 1% 0% 1% 1% 1% 0% 0% 0% 0% 0% 0% 0% 0 0 100 200 300 400 500 600 700 800 900 1000 1100 1200 1300 1400 1500 monthly time input per HH, hours Annex 9: Biomass 111 A9.9 This monthly time budget is not correlated with collection distance, though it is weakly correlated to family size (Figure A9.6), as one would expect (scale economies in cooking mean that the heat input to cooking is not linear to the number of persons participating in each meal). Figure A9.6: Relationship between Monthly Collection Time Budgets and Family Size A9.10 Interestingly, the time budgets are also not correlated with income: though the regression coefficient has the correct sign (higher income brings lower time budgets), it is not statistically significant. Labor Inputs A9.11 With time budgets in hand, the input of each to the collection effort can be calculated. Figure A9.7 shows the fraction of households in which girls provide a given fraction of the labor input. For example, we see that in 73% of all households (that collect fuelwood), girls do not participate at all; in 8% of households girls provide 50% of the total labor and in 2% of households, girls provide 100% of the collection effort. There is no correlation of the labor input of girls to collection distance. Figure A9.7: Labor Input of Girls 0.8 73% 0.6 H Hfo onitcarf0.4 0.2 8% 7% 3% 3% 1% 1% 2% 1% 0% 2% 0 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% fraction of labor input by girls 112 Household Energy Supply and Use in Yemen Volume 2: Annexes A9.12 Figure A9.8 shows the corresponding labor input of boys: they contribute no labor to fuelwood collection in 92% of all households, and only in very few cases do they make significant contributions. Figure A9.8: Labor Input of Boys 1 92% 0.8 H Hfo 0.6 noi actrf0.4 0.2 1% 1% 1% 2% 1% 0% 0% 0% 0% 1% 0 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% fraction of labor input by boys A9.13 There are some significant differences across Governorates, as shown in Figure A9.9: in fact in one Governorate, Aden, the input of boys is greater than that of girls. Figure A9.9: Comparison of Labor Inputs by Governorate 0.5 GIRL 0.4 0.3 0 0.2 0.1 0 BOY 0 Ibb SanaaCity Taiz AlHodeidah Dhamar Sadah Aden Al Mahweet Adelah Abyan AlBaida Haja Hadramout Shabwah SanaaGover Lahj Amaran A9.14 The labor input of adult women is shown in Figure A9.10. In 52% of all households (that collect firewood), adult women provide 100% of the collection labor. However, in 11% of households, they contribute no labor and in 14% they contribute 50%. Annex 9: Biomass 113 Figure A9.10: Labor Input of Women 0.6 1.1.25 52% 1.1.26 0.4 1.1.27 H H of 1.1.28 iontcarf 1.1.29 0.2 1.1.30 14% 11% 1.1.31 6% 5% 5% 4% 1.1.32 Figure2%9.10: labor input of adult women 0% 1% 0% 0 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% fraction of labor input by women A9.15 However, as shown in Figure A9.11, there is a statistically significant relationship between the labor contribution of women and collection distance: as collection distances increase, the contribution of women decreases. This confirms the anecdotal evidence of men wishing to accompany women where collection distances are long. Figure A9.11: Labor Input of Women as a Function of Collection Distance A9.16 Finally, Figure A9.12 shows the labor input of men. The pattern is very similar to boys, with 75% of households reporting no contribution of (adult) men. 114 Household Energy Supply and Use in Yemen Volume 2: Annexes Figure A9.12: Labor Input of Men 0.8 75% 0.6 H H of 0.4 iontcarf 0.2 6% 7% 2% 3% 4% 1% 1% 1% 0% 0% 0 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% fraction of labor input by men Figure A9.13: Labor Contribution of Men as a Function of Collection Distance A9.17 Figure A9.14 compares the labor inputs of adult men and women by Governorate. That of men exceeds that of women in only two Governorates, Sana'a city, and Aden. Annex 9: Biomass 115 Figure A9.14: Comparison of Labor Inputs, Adult Men v. Adult Women 1 0.8 W OMEN 0.6 0.4 0.2 MEN 0 Ibb SanaaCity Taiz AlHodeidah Dhamar Sadah Aden Al Mahweet Adelah Abyan AlBaida Haja Hadramout Shabwah SanaaGover Lahj Amaran Fuelwood Consumption A9.18 Table A9.3 shows the total consumption of fuelwood by Governorate and income decile. Note that the top income deciles account for the largest total quantity of fuelwood use ­ yet these are also the deciles that consume the bulk of the LPG. Table A9.3: Calculated Monthly Purchased Fuelwood Usage by Decile and Governorate (1000kg) Monthly 0 9,001 12,001 15,001 19,801 22,501 27,001 33,001 42,701 61,001 income -9000 - - - - - - - - > 12,000 15,000 19,800 22,500 27,000 33,000 42,700 61,000 decile Bottom d[2] D[3] d[4] d[5] d[6] d[7] d[8] d[9] top total/ YR/month 10% month Ibb 1027 2595 669 679 103 831 5904 Abyan 239 120 120 5 239 5 419 168 1315 Sana'aCity 18 4 3 9 30 17 82 Al-Baida 3 14 23 24 64 Taiz 81 712 227 195 129 334 300 233 201 214 2627 Hajjah 214 97 62 90 452 58 298 566 540 1517 3895 Al-Hodeida 733 1724 190 359 258 1260 273 397 69 225 5489 Hadramout 6 144 676 334 728 1001 416 879 3236 1772 9193 Dhamar 802 462 587 407 252 160 14 947 87 347 4065 Shabwah 4 2 9 6 90 478 578 104 107 0 1378 Sa'adah 162 400 181 376 374 280 248 60 349 164 2594 Sana'aGovern 20 121 278 612 226 2097 2782 997 712 2331 10175 Aden 319 401 13 17 22 69 32 12 15 4 904 Lahj 353 102 90 183 28 27 6 789 Al Mahweet 827 16 135 977 Amaran 939 21 137 302 108 326 64 3570 1716 168 7351 Adelah 154 417 571 4547 6700 4328 3478 2953 7104 5429 7818 7928 7088 57373 116 Household Energy Supply and Use in Yemen Volume 2: Annexes A9.19 Table A9.4 shows the uses of fuelwood in households reporting wood use. The dominant use is cooking, but a significant number also use fuelwood for heating, even in the big cities (where 31% use fuelwood for heating, compared to a national average of 24% of households). Table A9.4: Uses of Fuelwood For HH reporting use % reporting use for: % HH using cooking heating Home other fuelwood business Sanna City 5 85 31 3 Aden 23 82 19 3 All urban HH 36 92 18 5 4 Rural Sa'adah,Sana'a 90 82 31 2 9 Rural Al Hodeida, Hajjahh 91 91 23 1 4 Rural Ibb&Taiz 82 94 20 2 Rural Abyan, Lahj Adalah 62 86 34 2 1 Rural Shabwah&Haramout 94 94 17 1 2 All Rural 85 90 24 1 4 All HH 74 90 24 1 3.8 A9.20 The corresponding monthly consumption per household is shown in Table A9.5. Table A9.5: Average Monthly Consumption in Households using Purchased and Collected Firewood (kg/month) Monthly 0 9,001 12,001 15,001 19,801 22,501 27,001 33,001 42,701 61,001 income -9000 - - - - - - - - > 12,000 15,000 19,800 22,500 27,000 33,000 42,700 61,000 decile bottom d[2] d[3] d[4] d[5] d[6] d[7] d[8] d[9] 10% bb 510 320 230 480 49 218 Abyan 160 240 240 10 480 10 143 72 Sana'a City 33 20 20 33 33 22 Al Beida 5 24 20 14 Taiz 86 140 33 38 94 96 59 38 112 29 Hajjah 45 50 25 45 99 32 84 116 135 99 Al Hodeida 64 69 26 37 35 93 61 42 35 55 Hadramout 80 44 89 55 74 92 131 94 218 103 Dhamar 115 117 122 103 81 32 5 186 80 78 Shabwah 4 8 10 5 43 93 133 28 44 7 Sa'adah 100 76 52 67 144 83 211 51 168 126 Sana'aGovern 60 135 68 223 97 561 263 147 94 166 Aden 661 219 13 14 12 42 17 8 11 14 Lahj 122 29 22 40 16 22 24 Al Mahweet 300 30 260 Amaran 4000 44 49 123 61 246 34 473 303 112 Adelah 400 225 Total 145 118 86 77 68 118 130 134 159 100 Annex 9: Biomass 117 A9.21 The patterns of consumption per decile are noteworthy: the poorest deciles consume larger quantities (much of it collected) than the middle deciles; consumption increases again in the upper deciles of rural areas (Figure A9.15). Evidently there is a cultural preference for wood, which the upper deciles manage by purchase (rather than spending time for collection). Figure A9.15: Consumption of Fuelwood By Decile, kg/HH/month 200 h 150 ntom/gK RURAL n,oi 100 pt um nsoC 50 URBAN 0 d[1] d[2] d[3] d[4] d[5] d[6] d[7] d[8] d[9] d[10] URBAN 139 96 46 28 69 46 68 66 64 46 RURAL 146 123 96 95 68 145 155 131 188 117 A9.22 As expected, urban fuelwood prices are higher than in rural areas (Figure A9.16). However, these differences are far smaller than the variation across Governorates. Figure A9.16: Fuelwood Prices, YR/kg 0 20 40 60 80 URBAN average Ibb Abyan SanaaCity AlBaida Taiz Haja AlHodeidah Hadramouth Dhamar Shabwah ALL urban rural Sadah SanaaGovern Ibb 5 15 4 Aden Abyan 7 10 5 Lahj Al Mahweet SanaaCity 40 40 Amaran AlBaida 59 59 Adelah Taiz 37 25 49 RURAL average Haja 17 76 14 Ibb AlHodeid 27 27 26 Abyan Hadramo 11 12 11 SanaaCity AlBaida Dhamar 27 21 27 Taiz Shabwah 33 16 33 Haja AlHodeidah Sadah 11 4 13 Hadramouth Dhamar SanaaGov 8 24 8 Shabwah Aden 49 49 Sadah SanaaGovern Lahj 25 12 26 Aden Al Mahwe 8 14 6 Lahj Al Mahweet Amaran 19 21 19 Amaran Adelah 40 40 Adelah average 21 26 19 118 Household Energy Supply and Use in Yemen Volume 2: Annexes A9.23 These price variations have the expected impact on consumption rates, as shown in Figure A9.17: the higher the price, the lower the monthly consumption per household. However, price explains only about 15% of the total variation in consumption rate across Governorates,21 suggesting that the dominant determinant of consumption rates is resource availability. Figure A9.17: Fuelwood Consumption v. Price 400 300 htno m/ H 200 H/ Kg 100 0 0 20 40 60 80 price, YR/kg urban rural Charcoal A9.24 18% of all households report uses of charcoal (Table A9.6). The dominant use is neither for heating nor cooking, but for pipes (recorded in the survey as "other"). Table A9.6: Charcoal Use For HH reporting use, % reporting use for: % HH cooking heating ironing Home Other using business (Pipes) charcoal Sanna City 54 1 16 80 Aden 54 2 19 1.5 90 All urban HH 43 14 15 2.7 77 Rural Sa'adah,Sana'a 3 16 66 2 43 Rural Al Hodeida, Hajjahh 17 36 14 2 3 58 Rural Ibb&Taiz 2 35 0 4 39 Rural Abyan, Lahj Adalah 15 10 38 3 2 61 Rural Shabwah&Haramout 30 81 20 2 6 6 All Rural 11 41 22 1 3 45 All HH 17 27 18 1 3 62 A9.25 As shown in Figure A9.18, there are sharp differences between urban and rural charcoal use. Particularly in the low rural deciles, charcoal use is very low. 21As captured by the R2 in a simple linear model. Addition of second and third order polynomial terms, or a log form, did not significantly improve the explanatory power. Annex 9: Biomass 119 Figure A9.18: Use of Charcoal by Income Decile 0.6 0.5 l URBAN oacr 0.4 hacgnisu 0.3 HH of onitcarF0.2 RURAL 0.1 0 d[1] d[2] d[3] d[4] d[5] d[6] d[7] d[8] d[9] d[10] URBAN 28.3% 45.9% 43.5% 40.7% 33.0% 39.7% 50.8% 40.8% 30.2% 44.4% RURAL 5.9% 5.1% 6.0% 8.0% 5.7% 10.2% 13.3% 10.8% 15.5% 23.5% A9.26 As expected, there are again significant differences in urban and rural prices. As shown in Figure A9.19, the average urban price is 62 YR/kg, as opposed to 37 YR/kg in rural areas. Figure A9.19: Charcoal Prices, YR/kg 0 50 100 150 200 URBAN average Ibb Abyan SanaaCity AlBaida Taiz Haja AlHodeidah Hadramouth Dhamar ALL urban rural Shabwah Ibb 27 27 Sadah SanaaGovern Abyan 10 20 9 Aden SanaaCity 67 67 Lahj Al Mahweet AlBaida 87 70 97 Amaran Adelah Taiz 50 49 53 RURAL average Haja 44 42 45 Ibb AlHodeid 40 49 28 Abyan Hadramo 54 76 50 SanaaCity AlBaida Dhamar 43 55 29 Taiz Shabwah 48 91 34 Haja AlHodeidah Sadah 20 20 Hadramouth Dhamar SanaaGov 44 5 45 Shabwah Aden 74 74 Sadah SanaaGovern Lahj 15 15 Aden Al Mahwe Lahj Al Mahweet Amaran 146 161 50 Amaran Adelah 18 50 17 Adelah average 51 62 37 120 Household Energy Supply and Use in Yemen Volume 2: Annexes Crop Residues A9.27 The use of crop residues is largely confined to rural areas and, as noted above, little is purchased. There is little variation in use by income decile (Figure A9.20). Figure A9:20: Use of Crop Residues 0.4 uediser 0.3 oprc RURAL ngi 0.2 us HH of onitcarF0.1 URBAN 0 d[1] d[2] d[3] d[4] d[5] d[6] d[7] d[8] d[9] d[10] URBAN 2.3% 4.0% 1.1% 2.6% 7.4% 4.2% 1.6% 4.2% 2.1% 6.5% RURAL 26.8% 27.9% 36.0% 36.6% 25.4% 22.6% 26.6% 29.6% 28.0% 27.3% Annex 9: Biomass 121 Annex 10 Survey Data Reconciliation Expenditure and Income Data in the HES Survey Data A10.1 Figure A10.1 plots expenditure vs. income for all observations in the 3540 households in the 2003 household energy survey (2003 HES). Four households appear with monthly income greater than YR1 million; and two households show expenditures of more than one million (though these are not in households reporting corresponding income!) These outliers were removed from the dataset. Figure A10.1: Expenditure v. Income A10.2 The result is shown in Figure A10.2. But this figure raises a different question. In theory, the difference between expenditure and income of a household in any given month is either an addition to savings, or a draw-down from savings. But does this explain the level of variation in the case of the Yemen survey? Given that the implied draw-down or addition to savings is a multiple of expenditure in many cases, this seems an unlikely explanation. 123 124 Household Energy Supply and Use in Yemen Volume 2: Annexes Figure A10.2: Expenditure v. Income, Outliers Removed A10.3 The frequency distribution of discrepancies (Figure A10.3) is odd. In just 19% of the records do expenditure and income match to within 1%. Figure A10.3: Frequency Distribution of Income v. Expenditure Discrepancies 800 19% 600 13% 400 frequency 8% 200 0 -38 -34 -30 -26 -22 -18 -14 -10 -6 -2 2 6 10 14 18 22 26 30 34 38 Income - expenditure, as % of income A10.4 But in 8% of households, expenditure exceeded income by more than 40%, and 13% of households' income exceeded expenditures by more than 40%. Under- reporting of income in household surveys is a well-documented phenomenon ­ which would explain the negative entries in Figure A10.3.22 But here there is a preponderance of over-reporting incomes (or under-reporting expenditures) ­ represented by the positive entries in Figure A10.3. A10.5 The problem of the inconsistency between income and expenditure data becomes apparent when one looks at the data for individual Governorates. For example, Figure A10.4 shows the household expenditure profile for Al Hodeida. Since the definition of deciles in the 2003 HES is by income, as revealed in this survey, and the sum of expenditures does not match that income figure, the total expenditure curve is not smooth (e.g. total expenditure in the 3rd decile is less than that in the 2nd decile). 22World Bank, Sri Lanka Energy Services Delivery Project, Implementation Completion Report, May 2003. Annex 10: Survey Data Reconciliation 125 Figure A10.4: Al Hodeida [Monthly Expenditure Per HES] 80 disposable RY 60 othe r basic 0001,e urt 40 ndi pexe food ylht on 20 m 0 bottom 10%d[2] d[3] d[4] d[5] d[6] d[7] d[8] d[9] top 10% average food 4513 8999 7238 12217 15370 18391 16092 20769 22956 35542 12359 other basic 2533 5838 5815 5751 6777 12107 10219 11580 19833 26373 7753 disposable 196 1795 1081 1704 3163 2043 2962 2062 7222 6625 2073 electricity 62 203 137 744 87 363 806 873 1139 1632 391 LPG 114 339 142 238 360 522 269 383 376 614 299 other energy 11 53 4 0 267 183 0 369 878 492 122 "Other basic" is the sum of expenditure on housing, education, medical water and transportation. Calculation of Income Deciles A10.6 The definition of income deciles is survey-based and designed so that the weighted number of households is roughly equal. With a total estimated number of households being 2,249,173, each decile should contain roughly 225,000 households. This is true of all deciles except 3 and 4, as shown in Table A10.1. In the view of the CSO and the survey consultant, a readjustment was not deemed necessary. Table A10.1: Income Deciles Income range Households 1 <9000 234,560 2 9,001-12,000 233,535 3 12,001-15,000 247,273 4 15,001-19,800 185,532 5 19,801-22,500 227,267 6 22,501-27,000 226,015 7 27,001-33,000 223,618 8 33,001-42,700 223,603 9 42,701-61,000 224,887 10 >61,000 222,882 total 2,249,173 A10.7 With deciles defined by income, the average expenditures of each decile may not necessarily fall within the income range of each decile: as shown in Table A10.2, in four out of six deciles, average expenditure lies out of the income range (in three cases below and in one case above). 126 Household Energy Supply and Use in Yemen Volume 2: Annexes Table A10.2: Income and Expenditure by Income Decile total Expenditure expenditure outside income range [YR/month] 1 <9,000 8,179 No 2 9,001-12,000 15,592 YES 3 12,001-15,000 17,882 YES 4 15,001-19,800 20,869 YES 5 19,801-22,500 22,427 No 6 22,501-27,000 26,976 No 7 27,001-33,000 32,166 No 8 33,001-42,700 32,751 YES 9 42,701-61,000 48,479 No 10 61,001>0 90,548 No Comparison with the Last Household Expenditure Survey A10.8 The expenditure data may be compared to that of the previous 1998 Household Expenditure Survey. Table A10.3 shows the expenditure data by income decile, as reported by the World Bank in the 2002 Poverty Update.23 Table A10.3: 1998 Household Expenditure Survey Income decile 1 2 3 4 5 6 7 8 9 10 All HH persons/HH 9.4 8.6 7.8 7.6 7.7 7.9 7.4 7.2 7.1 5.6 7.1 per capita, annual (1998) Food 9814 14191 17374 20101 23197 25696 29973 35347 42471 63922 28209 Housing 2608 3725 4470 5259 5733 6600 7459 8263 10246 19269 7363 Clothing 1137 1715 2190 2566 3006 3491 3921 4945 5874 9899 3874 Health 196 450 587 654 902 1082 1232 1494 2707 5834 1514 Education 150 166 188 229 274 374 313 361 476 1175 371 Transport 361 536 700 977 1237 1643 2125 2686 4370 9981 2462 Leisure 1225 2148 3058 4135 4781 6085 7504 9117 11683 23087 7282 Other 206 392 659 833 1220 1482 1755 2231 3447 9600 2183 Total 15697 23323 29226 34754 40350 46453 54282 64444 81274 142767 53257 per household, per month Food 7717 10226 11299 12676 14812 16855 18530 21269 25029 29772 16663 Housing 2051 2684 2907 3316 3661 4329 4611 4972 6038 8975 4349 Clothing 894 1236 1424 1618 1919 2290 2424 2975 3462 4610 2289 Health 154 324 382 412 576 710 762 899 1595 2717 894 Education 118 120 122 144 175 245 194 217 281 547 219 Transport 284 386 455 616 790 1078 1314 1616 2575 4649 1454 Leisure 963 1548 1989 2608 3053 3991 4639 5486 6885 10753 4302 Other 162 282 429 525 779 972 1085 1342 2031 4471 1289 Total 12342 16806 19007 21917 25765 30471 33558 38777 47896 66494 31459 2003 Energy Survey 8405 15978 18020 21135 22698 27474 32406 32780 48215 90640 31598 delta -3937 -828 -987 -782 -3067 -2997 -1152 -5996 319 24146 139 -32% -5% -5% -4% -12% -10% -3% -15% 1% 36% 0% Source: World Bank, Republic of Yemen Poverty Update, op. cit., Volume 2, Table 29. 23 World Bank Republic of Yemen Poverty Update, Report 24422-Yemen, Dec. 2002 Annex 10: Survey Data Reconciliation 127 A10.9 In all the lower income deciles (1-8), the 2003 HES gives smaller expenditures than the 1998 Survey; but in the highest decile expenditure is 36% higher. Overall the average household expenditure is essentially identical! One would have expected higher expenditures in 2003 than in 1998, given an increase in the CPI of some 55% over this same period. A10.10A comparison by expenditure category is instructive: the 2003 HES shows 20% lower food expenditure, but significantly higher expenditures in health and education (Table A10.4). The definition of "housing" is obviously different. Table A10.4: Comparison by Expenditure Categories, All Households Expenditure HES 2003 Survey 1998 Food 16663 13843 -20% Housing 4349 586 -643% Water 1616 Clothing 2289 2168 -6% Health 894 2929 +69% Education 219 1437 +85% Transport 1454 1620 +10% Leisure 4302 Agriculture & livestock 2373 Other 1289 2995 Total 31459 29567 Energy (including electricity) 2431 Total 31459 31998 A10.11However, the problem with inclusion in "other" is that 69% of the households report zero "other expenditure" (Figure A10.5), while only five households report zero energy expenditure. Figure A10.5: Frequency Distribution of "Other Expenditure" 3000 2500 69% 2000 ncy 1500 requef 1000 500 0 3000 7000 11000 15000 19000 23000 27000 31000 35000 39000 1000 5000 9000 13000 17000 21000 25000 29000 33000 37000 other expenditure 128 Household Energy Supply and Use in Yemen Volume 2: Annexes Food Expenditure Data A10.12Food accounts for the largest single expenditure: the average household spent 43% of its expenditure on food, distributed as shown in Figure A10.6. Figure A10.6: Distribution of Food Expenditure Fractions 250 200 150 cy uen reqf 100 50 0 2.5 7.5 13 18 23 28 33 38 43 48 53 58 63 68 73 78 83 88 93 98 fraction of expenditure spent on food A10.13The proportion of total household expenditure accounted for by food declines as expected with increasing household income (Figure A10.7). However, the food shares of total expenditure in the 2003 HES are significantly lower across all income deciles: this would be expected given the increase in household incomes over the five years that have elapsed since the 1998 Expenditure Survey. Figure A10.7: Food Expenditure as Fraction of Total Household Expenditure 0.7 l) (totac 0.6 fra 1998 HBS sa 2003 HES 0.5 ture ndiepxe 0.4 d foo 0.3 1 2 3 4 5 6 7 8 9 10 income decile General Patterns of Income Distribution A10.14Much of the difference can also be explained by the distribution of poverty, which is largely a rural phenomenon. As shown in Table A10.5, 91% of households in the bottom income decile are in rural areas, as opposed to an overall average of 77% of all households being in rural areas. Similarly, 33% of the households in the top decile are in urban areas, as against 23% of all households. Annex 10: Survey Data Reconciliation 129 Table A10.5: Distribution of Households by Income Decile d[1] d[2] d[3] d[4] d[5] d[6] d[7] D[8] d[9] d[10] average Urban HH 22,13 38,27 36,03 37,73 44,01 57,31 64,67 64,60 72,65 73,82 511,266 6 2 0 5 3 7 3 7 8 4 9% 16% 15% 20% 19% 25% 29% 29% 32% 33% 23% Rural HH 212,4 195,2 211,2 147,7 183,2 168,6 158,9 158,9 152,2 149,0 1,737,90 24 62 44 97 54 98 45 96 28 58 7 91% 84% 85% 80% 81% 75% 71% 71% 68% 67% 77% Total HH 234,5 233,5 247,2 185,5 227,2 226,0 223,6 223,6 224,8 222,8 2,249,17 60 35 73 32 67 15 18 03 87 82 3 A10.15Table A10.6 shows the distribution of sample households (i.e. before weighting). Table A10.6: Distribution of Households by Region and Income Decile Monthly 0 9,001 12,001 15,001 19,801 22,501 27,001 33,001 42,701 61,001 income -9000 - - - - - - - - > 12,000 15,000 19,800 22,500 27,000 33,000 42,700 61,000 decile bottom d[2] d[3] d[4] d[5] d[6] d[7] d[8] d[9] top YR/month 10% Ibb 68 55 40 16 38 26 14 7 13 6 283 Abyan 4 2 3 4 3 11 8 2 23 30 90 Sana'aCity 14 30 31 24 42 49 79 77 111 141 598 Al-Baida 3 7 2 8 15 5 18 21 26 105 Taiz 32 30 32 29 38 30 31 36 21 21 300 Hajjah 10 8 11 15 10 11 23 13 16 15 132 Al-Hodeida 54 79 21 23 27 23 26 13 9 12 287 Hadramout 4 7 20 27 29 52 31 38 51 23 282 Dhamar 10 13 32 22 23 23 20 21 10 8 182 Shabwah 3 2 2 6 10 27 13 19 23 13 118 Sa'adah 1 8 9 13 15 16 13 10 7 8 100 Sana'aGovern 16 13 29 14 24 17 24 21 24 38 220 Aden 8 28 26 29 38 41 51 67 78 33 399 Lahj 2 9 22 25 25 23 17 15 11 11 160 Al Mahweet 5 3 3 6 2 4 6 15 1 2 47 Amaran 9 12 21 13 15 9 15 30 14 11 149 Adelah 3 5 5 6 11 11 14 26 7 88 Total 240 305 314 273 353 388 387 416 459 405 3540 A10.16According to the HES, Abyan has the highest proportion of high income households (see Figure A10.8): almost a third of its households fall into the (national) top decile. 130 Household Energy Supply and Use in Yemen Volume 2: Annexes Figure A10.8: Income Distribution, Selected Governorates 0.4 Abyan 0.3 Ibb SanaaCity 0.2 0.1 Dhamar Abyan Dhamar SanaaCity Ibb 0 bottom 10% d[3] d[5] d[7] d[9] d[2] d[4] d[6] d[8] top 10% A10.17It follows from the bottom row of Table A10.7 that each sample household in the lowest decile represents more households than in the top decile. In other words, the sampling rate in the bottom decile is 240/234560, about 1 per 1000, while the sampling rate in the top decile is 405/222,882, about 2 per 1000. Poverty Distribution and Comparison with the Poverty Update Report A10.18The Poverty Update notes24: The distribution of the poor across the governorate of Yemen suggests marked disparities in poverty rates across the national territory. About half of the poor live in four governorates: Taiz (with 18.7 percent of the total poor), lbb (16.2 percent), Sana'a region (11.9 percent) and Al-Hodeida (10.2 percent). The number of poor people as a percentage of the governorate population is highest in Taiz (56 percent), Ibb (55 percent), Abyan (53 percent), and Laheg (52 percent), but is also high in Dhamar (49 percent), Hadramout, Al-Mahrah and Shabwah (43 percent). The incidence of poverty is lowest in Al-Baida (15 percent) and Saddah (27 percent) and in the two major urban centres, Sana'a city (23 percent) and Aden (30 percent). A10.19The distribution of the poor across governorates is largely a function of the general distribution of population. Table A10.7 shows the number of households by decile in each Governorate (as per the 2003 HES), and Table A10.8 as a percentage of the total number of households in Yemen (estimated at 2.25million). 24World Bank Republic of Yemen Poverty Update, Report 24422-Yemen, Dec. 2002 Annex 10: Survey Data Reconciliation 131 Table A10.7: Number of (Weighted) Households in Each Governorate and Decile Monthly 0 9,001 12,001 15,001 19,801 22,501 27,001 33,001 42,701 61,001 income -9000 -12,000 -15,000 -19,800 -22,500 -27,000 -33,000 -42,700 -61,000 > decile bottom d[2] d[3] d[4] d[5] d[6] d[7] d[8] d[9] top total YR/month 10% Ibb 84,561 60,552 40,198 14,286 39,354 29,065 8,027 6,077 14,122 2,420 298,662 Abyan 1,992 964 1,109 902 1,617 5,281 2,346 931 13,621 19,323 48,086 Sana'aCity 2,594 6,886 7,282 4,853 8,938 13,032 19,839 16,889 27,182 32,988 140,483 Al-Baida 2,378 4,092 750 3,167 6,952 3,396 9,152 12,523 14,252 56,661 Taiz 37,569 32,999 39,941 28,300 35,709 27,040 27,635 27,921 19,094 18,383 294,591 Hajjah 18,860 8,798 12,729 15,964 12,838 9,061 18,348 15,410 15,067 28,720 155,796 Al-Hodeida 51,450 65,806 20,351 21,969 31,280 22,412 23,696 13,276 6,531 10,870 267,641 Hadramout 1,792 3,784 13,232 10,536 15,896 21,292 14,737 18,053 28,306 21,151 148,779 Dhamar 16,706 11,561 40,398 31,311 19,129 19,535 26,646 23,042 9,089 14,077 211,494 Shabwah 1,495 822 1,827 1,837 3,753 14,019 4,425 8,170 6,530 3,793 46,671 Sa'adah 1,620 7,439 7,763 11,563 8,828 9,402 6,619 5,452 3,931 5,284 67,901 Sana'aGovern 6,066 5,513 18,670 5,762 14,198 11,582 23,801 17,750 24,632 26,737 154,711 Aden 907 4,613 3,659 4,575 7,009 6,345 8,578 9,810 13,583 6,572 65,651 Lahj 786 3,775 13,722 13,375 13,975 13,699 9,089 7,268 7,384 3,466 86,540 Al Mahweet 2,603 6,086 8,268 7,809 1,094 4,369 12,118 14,086 519 1,039 57,990 Amaran 5,978 10,171 11,833 9,196 7,900 6,140 8,182 22,316 8,898 10,081 100,694 Adelah 1,389 2,199 2,545 2,582 6,788 6,138 7,999 13,874 3,308 46,822 Total 234,980 233,535 247,273 185,532 227,267 226,015 223,618 223,603 224,887 222,463 2,249,173 Table A10.8: As Percent of the Total Households Monthly 0 9,001 12,001 15,001 19,801 22,501 27,001 33,001 42,701 61,001 income -9000 - - - - - - - - > 12,000 15,000 19,800 22,500 27,000 33,000 42,700 61,000 decile bottom d[2] d[3] d[4] d[5] d[6] d[7] d[8] d[9] top total YR/month 10% Ibb 3.8% 2.7% 1.8% 0.6% 1.7% 1.3% 0.4% 0.3% 0.6% 0.1% 13.3% Abyan 0.1% 0.0% 0.0% 0.0% 0.1% 0.2% 0.1% 0.0% 0.6% 0.9% 2.1% Sana'aCity 0.1% 0.3% 0.3% 0.2% 0.4% 0.6% 0.9% 0.8% 1.2% 1.5% 6.2% Al-Baida 0.0% 0.1% 0.2% 0.0% 0.1% 0.3% 0.2% 0.4% 0.6% 0.6% 2.5% Taiz 1.7% 1.5% 1.8% 1.3% 1.6% 1.2% 1.2% 1.2% 0.8% 0.8% 13.1% Hajjah 0.8% 0.4% 0.6% 0.7% 0.6% 0.4% 0.8% 0.7% 0.7% 1.3% 6.9% Al-Hodeida 2.3% 2.9% 0.9% 1.0% 1.4% 1.0% 1.1% 0.6% 0.3% 0.5% 11.9% Hadramout 0.1% 0.2% 0.6% 0.5% 0.7% 0.9% 0.7% 0.8% 1.3% 0.9% 6.6% Dhamar 0.7% 0.5% 1.8% 1.4% 0.9% 0.9% 1.2% 1.0% 0.4% 0.6% 9.4% Shabwah 0.1% 0.0% 0.1% 0.1% 0.2% 0.6% 0.2% 0.4% 0.3% 0.2% 2.1% Sa'adah 0.1% 0.3% 0.3% 0.5% 0.4% 0.4% 0.3% 0.2% 0.2% 0.2% 3.0% Sana'aGovern 0.3% 0.2% 0.8% 0.3% 0.6% 0.5% 1.1% 0.8% 1.1% 1.2% 6.9% Aden 0.0% 0.2% 0.2% 0.2% 0.3% 0.3% 0.4% 0.4% 0.6% 0.3% 2.9% Lahj 0.0% 0.2% 0.6% 0.6% 0.6% 0.6% 0.4% 0.3% 0.3% 0.2% 3.8% Al Mahweet 0.1% 0.3% 0.4% 0.3% 0.0% 0.2% 0.5% 0.6% 0.0% 0.0% 2.6% Amaran 0.3% 0.5% 0.5% 0.4% 0.4% 0.3% 0.4% 1.0% 0.4% 0.4% 4.5% Adelah 0.0% 0.1% 0.1% 0.1% 0.1% 0.3% 0.3% 0.4% 0.6% 0.1% 2.1% Total 10% 10% 11% 8% 10% 10% 10% 10% 10% 10% 100% 132 Household Energy Supply and Use in Yemen Volume 2: Annexes A10.20Table A10.9 shows the distribution of the poor (defined as income in the bottom 30%) ­ and compares the survey distribution with those of the poverty assessment. Table A10.9: Distribution of the Poor (Defined as the Bottom 30% of all Yemeni Households) Survey Poverty Total assessment Households Ibb 26% 16% 13% Abyan 1% 2% Sana'aCity 2% 6% Al-Baida 1% 3% Taiz 15% 19% 13% Hajjah 6% 7% Al-Hodeida 19% 10% 12% Hadramout 3% 7% Dhamar 10% 9% Shabwah 1% 2% Sa'adah 2% 3% Sana'aGovern 4% 12% 7%+6% (including in Sana'a Sana'aCity City=13% Aden 1% 3% Lahj 3% 4% Al Mahweet 2% 3% Amaran 4% 4% Adelah 1% 2% total 100% 100% 100% A10.21Table A10.10 shows the proportion of the poor (again defined as households in the bottom three deciles) by Governorate, comparing the survey result with that of the Poverty Assessment. For several governorates the difference is significant (notably Abyan, Hadramout, Shabwah and Lahj). Annex 10: Survey Data Reconciliation 133 Table A10.10: Share of Poor by Governorate Survey Poverty Difference assessment [1] [2] [3] Ibb 62% 55% 7% Abyan 8% 53% -45% Sana'a City 12% 23% -11% Al-Baida 11% 15% -4% Taiz 38% 56% -18% Hajjah 26% Al-Hodeida 51% Hadramout 13% 43% -30% Dhamar 32% 49% -17% Shabwah 9% 43% -34% Sa'adah 25% 27% -2% Sana'aGovern 20% Aden 14% 30% -16% Lahj 21% 52% -31% Al Mahweet 29% Amaran 28% Adelah 8% 8% A10.22One possible explanation of the differences lies in the use of households in the 2003 HES rather than population. But since the deciles (in the 2003 HES) are defined by household, that does seem the logical unit. Table A10.11 shows the number of persons per household in each income decile and Governorate Table A10.11: Number of Persons per Household Monthly 0 9,001 12,001 15,001 19,801 22,501 27,001 33,001 42,701 61,001 Governorate income -9000 - - - - - - - - > average 12,000 15,000 19,800 22,500 27,000 33,000 42,700 61,000 decile bottom d[2] D[3] d[4] d[5] d[6] d[7] d[8] d[9] top YR/month 10% Ibb 4.8 6.2 6.0 6.7 6.2 6.1 8.4 8.0 7.2 8.4 6.0 Abyan 7.3 5.9 7.6 6.5 5.7 6.8 9.6 9.0 7.0 10.7 8.6 Sana'aCity 6.6 7.3 7.0 6.1 7.0 7.6 7.1 7.2 7.6 9.9 7.9 Al-Baida 4.4 4.9 6.0 4.7 6.8 5.9 7.1 8.9 8.5 7.3 Taiz 5.7 7.1 8.3 6.9 7.1 7.9 8.3 7.4 8.8 8.7 7.5 Hajjah 7.0 7.3 7.1 8.0 6.7 9.6 8.2 9.2 13.1 12.5 9.2 Al-Hodeida 5.6 5.7 6.2 5.8 6.1 6.0 7.2 6.4 9.7 10.9 6.3 Hadramout 6.3 4.8 5.7 6.2 6.8 6.8 8.5 9.1 9.2 11.0 8.1 Dhamar 7.1 4.9 6.0 7.2 7.7 6.6 7.0 7.0 9.4 13.1 7.2 Shabwah 6.0 3.6 6.5 4.9 7.2 6.9 7.3 6.8 7.0 5.9 6.7 Sa'adah 8.0 6.2 7.4 6.4 5.5 6.9 7.7 7.3 7.9 10.1 7.1 Sana'aGovern 4.5 6.6 7.4 6.8 7.6 8.3 10.5 9.8 11.1 13.1 9.7 Aden 3.0 6.7 6.0 7.5 7.0 5.6 6.9 6.8 6.5 7.7 6.7 Lahj 7.1 3.1 5.0 6.5 6.5 7.3 6.5 8.6 8.0 8.5 6.6 Al Mahweet 5.7 10.5 5.0 11.4 6.2 9.1 8.0 7.3 7.0 16.0 8.2 Amaran 4.7 7.4 8.6 9.9 6.2 11.5 11.3 13.6 10.8 12.2 10.2 Adelah 4.3 6.8 9.0 7.5 6.7 9.0 8.4 9.3 10.4 8.4 Total 5.6 6.3 6.6 7.1 6.7 7.2 8.1 8.4 9.0 10.8 7.5 134 Household Energy Supply and Use in Yemen Volume 2: Annexes A10.23Note the increase in household size with income decile, which suggests that households in the top decile have high income (at least in part) because these are the households with most wage earners. Fortunately there is a question in the 2003 HES survey that asks how many "income earners" there are in each household (#538), with results as shown in Table A10.12. Table A10.12: Wage Earners in Each Household, By Governorate and Household Income Decile Monthly 0 9,001 12,001 15,001 19,801 22,501 27,001 33,001 42,701 61,001 Governorate income -9000 - - - - - - - - > average 12,000 15,000 19,800 22,500 27,000 33,000 42,700 61,000 decile bottom d[2] d[3] d[4] d[5] d[6] d[7] d[8] d[9] top YR/month 10% Ibb 0.8 1.4 0.8 1.7 1.0 1.2 1.2 1.7 1.9 2.7 1.1 Abyan 1.0 1.0 1.0 1.2 1.0 1.2 1.6 2.5 1.5 2.4 1.8 Sana'aCity 1.8 1.3 1.3 1.3 1.3 1.4 1.5 1.6 1.6 2.3 1.7 Al-Baida 1.2 1.4 1.8 2.2 1.1 1.2 1.9 2.0 1.9 1.7 Taiz 0.7 0.7 1.3 1.3 1.4 1.2 1.5 1.4 1.1 2.1 1.2 Hajjah 1.0 1.4 1.3 2.1 1.6 2.1 1.7 3.0 2.3 2.9 2.0 Al-Hodeida 1.2 1.2 1.4 1.5 1.2 1.5 2.0 1.7 2.8 4.0 1.5 Hadramout 1.0 0.8 0.9 1.5 1.5 1.4 1.8 2.5 2.5 3.8 2.1 Dhamar 1.4 1.3 1.2 1.3 1.3 1.4 1.8 1.7 1.7 2.8 1.5 Shabwah 1.2 1.0 1.0 1.5 1.2 1.1 1.4 1.4 1.8 1.9 1.4 Sa'adah 0.0 1.5 1.9 2.6 2.5 3.2 4.4 3.5 3.5 2.6 2.7 Sana'aGovern 1.0 0.9 1.3 1.5 1.7 1.3 2.0 1.6 2.8 2.7 1.9 Aden 0.4 1.0 1.0 1.2 1.3 1.3 1.7 2.0 2.1 3.0 1.7 Lahj 0.9 1.1 1.3 1.5 2.0 1.6 2.2 2.2 1.9 1.7 1.7 Al Mahweet 1.8 1.0 1.0 1.1 2.1 1.3 1.5 1.4 1.0 2.5 1.3 Amaran 1.0 1.6 1.3 1.3 1.4 1.3 1.7 1.7 2.0 2.3 1.6 Adelah 0.7 0.9 2.0 1.1 1.8 1.5 2.4 2.4 3.2 2.1 Total 1.0 1.2 1.2 1.5 1.4 1.4 1.8 1.9 2.1 2.7 1.6 A10.24 Indeed, this is exactly as expected: the poorest households have on average one wage earner (and a family size of 5.6 ­ i.e. one wage earner for 5.6 persons) ­ while the top income decile has 2.7 wage earners and 10.8 persons per household (i.e. one wage earner for 4 persons). A10.25 However, this question does not appear to have been answered (or asked) consistently. In some cases one can infer that the response is equal to the number of wage earners, while in others it appears to be the number of individuals contributing to income.