CURRENCY EQUIVALENTS Currency Unit – Peso($) EXCHANGE RATE $1.00 US = $11.40MXN $1.00 MXN = $0.087 US WEIGHTS AND MEASURES Metric System FISCAL YEAR January 1 – December 31 Vice President LCR: David de Ferranti Director LCC1C: Isabel Guerrero Director LCSPR: Ernesto May Acting Lead Economist: David Michael Gould Sector Manager: Mauricio Carrizosa Task Manager: Steven B. Webb - ABBREVIATIONS AND ACRONYMS AFORE Pensión fund management firms Administradoras de fondos para el retiro ALIANZA Rural Alliance Alianza para el Campo BANOBRAS National Bank of Public Works Banco Nacional de Obras y Servicios Públicos CECYT Center for Scientific and Technological Centro de Estudios Científicos y Tecnológicos Studies CEPAL Economic Commission for Latin America and Comisión Económica para América Latina y el the Caribbean Caribe CFE Federal Electric Commission Comisión Federal Eléctrica CIDE Center for Economic Research and training Centro Investigación Y Docencia Económicas CONADE National Sports Commission Comisión Nacional del Deporte CONAFE National Council for Educational Consejo Nacional de Fomento Educativo Development CONALEP National College of Professional and Colegio Nacional de Educación Profesional Technical Education Técnica CONAPO National Council on Population Consejo Nacional de Población ENIGH National Survey Of Income And Household Encuesta Nacional de Ingresos y Gastos de los Expenses Hogares FAEB Fund for Basic Education Fondo de Aportaciones para la Educación Básica y Normal FAETA The Fund for Technical and Adult Education Fondo de Aportaciones para la Educación Tecnológica y de Adultos FAIS Fund for Social Infrastructure Fondo de Aportaciones para la Infraestructura Social FAM Multiple Fund Fondo de Aportaciones Múltiples FAPPA Fund for the Promotion of Productive Projects Fondo para el Apoyo a Proyectos Productivos in the Agrarian Organizations de las Organizaciones Agrarias FARAC Trusteeship for the Bailout of Toll Road Fideicomiso de Apoyo al Rescate de Autopista Concessions Concesionadas FASSA Fund for the Supporting Health Services Fondo de Aportaciones para los Servicios de Salud FIDUCAR Rural Education Funding Financiamiento Educativo Rural FISE Fund for States Social Infrastructure Fondo de Aportaciones para la Infraestructura Social Estatal FOBAPROA Bank Fund for the Protection of Savings Fondo Bancario de Protección al Ahorro, GDP Gross Domestic Product IADB Inter American Development Bank Banco Inter Americano de Desarrollo IMJ Mexican Institute for the Young Instituto Mexicano de la Juventud IMSS Mexican Social Security Institute Instituto Mexicano del Seguro Social INAH National Institute of History and Instituto Nacional de Antropología e Historia Anthropology INBA National Institute for the Arts Instituto Nacional de Bellas Artes INEA National Institute for Adult Education Instituto Nacional para la Educación de los Adultos INEGI National Institute for Statistics, Geography Instituto Nacional de Geografía, Estadística e and Informatics Informática INI National Indigenous Institute Instituto Nacional Indigenista IPAB Institute for the Protection of Bank Savings, Instituto para la Protección del Ahorro Bancario ISSSTE Institute for the Social Security of the State Instituto de Seguridad y Servicios Sociales de Workers los Trabajadores del Estado - ITAM Autonomous Technological Institute of Instituto Tecnológico Autónomo de México Mexico IVCM Disability Old age, Unemployment in Invalidez, Vejez, Cesantía en Edad Avanzada y Advanced Age and Death Muerte NAFTA North American Free Trade Agreement OPORTUNIDADES Human Development Program – Programa de Desarrollo Humano Oportunidades ‘Opportunities’ PAC Expanding Coverage Program Programa de Ampliación de Cobertura PAFEF Program for the Strengthening of the Federal Programa para el Fortalecimiento de las Entities Entidades Federativas PAFEF Program for the Strengthening of Federal Programa para el Fortalecimiento de las Entities Entidades Federativas PEMEX Mexican Petroleum Petróleos Mexicanos PET Temporary Employment Program Programa de Empleo Temporal PIASRE Integral Program for Sustainable Agriculture Programa Integral de Agricultura Sostenible y and Productive Conversion in Zones of High Reconversión Productiva en Zonas de Risk Siniestralidad Recurrente PIDIREGAS Budget Public Investment Projects Proyectos de Impacto Diferido en el Registro Guaranteed By The Government del Gasto PROCAMPO Program for Direct Support to Agriculture Programa de Apoyos Directos al Campo PROCEDE Program for Common or Public Land Rights Programa de Certificación de Derechos Ejidales Certification y Titulación de Solares Urbanos PRODEFOR Forestry Development Program Programa de Desarrollo Forestal PRODEPLAN Development Program for Commercial Programa de Desarrollo de Plantaciones Plantations Comerciales PROMUSAG Program for Women in the Agrarian Sector Programa de la Mujer en el Sector Agrario RJP Retirement and Pension Regulation Régimen de Jubilaciones y Pensiones SAGARPA Ministry of Agriculture, Livestock, Rural Secretaría de Agricultura, Ganadería, Desarrollo Development, Fisheries and Food Rural, Pesca y Alimentación SECMAR Navy Department Secretaría de Marina SEDENA National Defense Department Secretaría de la Defensa Nacional SEMARNAT Ministry of Environment and Natural Secretaría de Medio Ambiente y Recursos Resources Naturales SEP Ministry of Public Education Secretaría de Educación Pública SHCP Ministry of Finance Secretaría de Hacienda y Crédito Público SRA Ministry of Agrarian Reform Secretaría de la Reforma Agraria SSA Ministry of Health Secretaría de Salud VAT Value Added Tax LICONSA Industrialized Milk Conasupo Leche Industrializada Conasupo - Preface At the request of the Ministry of Finance, the World Bank produced this PER that concentrates on four main issues: overall fiscal sustainability and rigidities in expenditure, the distribution of benefits of public spending across households with different income levels, the geographic distribution of the spending, and the institutions for budgeting and expenditure management. Thus, it is not a traditional PER, with extensive analysis of spending efficiency and institutions in individual sectors. Such analysis is important and is anticipated as follow-up to this report, such as with PERs on Infrastructure and Education. This report is a parallel task with the Mexico Poverty Programmatic AAA. On the critical issue of the benefit incidence of spending and poverty reduction efforts, the two reports divided the labor so that the Poverty Programmatic focuses on the effectiveness and targeting of poverty reduction programs, while the PER aims to put these programs in context by estimating the benefit incidence of all programs across all income levels. The will help the Mexican authorities to see whether the actual benefit incidence of federal spending is in line with their intended priorities for poverty reduction. The core team for the PER comprised Jose Luis Aburto (electricity), Odracir Barquera (document production) William Dorotinsky (public expenditure management), Christian Gonzalez (fiscal overview), Jonathan Halpern (water and electricity), Uri Raich (geographic distribution), Anita Schwarz (pensions), John Scott (benefit incidence analysis), and Steven Webb (task team leader). Patricia Chacon Holt processed the document and handled consultant logistics, and Elizabeth Forsythe provided valuable editorial service. We also benefited from detailed comments by Gabriela Aguilar-Martinez, Maria-Luisa Escobar, Gladys Lopez-Acevedo, Harry Patrinos, Panagiota Panopoulou, Michael Walton and the peer reviewers, Dominique van de Walle and Michael Stevens. The overall guidance of Dr. Carlos Hurtado (Subsecretario de Egresos, SHCP) and the energetic participation of his colleagues, Humberto Guzman and Deborá Schlam, were essential to the success of this endeavor. - - TABLE OF CONTENTS 1. MEXICO’S PUBLIC FINANCE OVERVIEW .............................................................. 1 Revenue and Expenditure Trends and Fiscal Balances ....................................................................................1 Revenue.......................................................................................................................................................1 Expenditures.....................................................................................................................................................3 Federal Government’s Expenditures ...........................................................................................................4 State-owned Enterprises Expenditures ........................................................................................................7 Budget Rigidities.........................................................................................................................................9 Deficits, Debt, and Fiscal Sustainability ........................................................................................................12 Conclusions and Policy Recommendations ...................................................................................................14 2. THE DISTRIBUTION OF BENEFITS FROM PUBLIC EXPENDITURE ................. 17 Overview........................................................................................................................................................18 Previous Studies ........................................................................................................................................18 Data and Methodological Assumptions.....................................................................................................21 Redistributive Demands ............................................................................................................................23 Redistributive Instruments ........................................................................................................................25 The Distribution of Benefits: 1992–2002.......................................................................................................26 Public Expenditure on Redistributive Programs........................................................................................26 Education...................................................................................................................................................28 Health ........................................................................................................................................................34 Pensions.....................................................................................................................................................38 Monetary Transfers: Oportunidades Program and Procampo ...................................................................43 Utilities: Electricity and Water..................................................................................................................48 All Benefits ...............................................................................................................................................52 Redistributive Impact .....................................................................................................................................56 All Benefits ...............................................................................................................................................56 Education and Health Resources ...............................................................................................................59 Monetary Transfers: Impact on Monetary Poverty ...................................................................................61 Cost-Efficiency of Public Services............................................................................................................63 Benefits, Taxes, and Tax Reform ..............................................................................................................66 Policy Recommendations...............................................................................................................................69 Specific Priorities ...........................................................................................................................................69 3. THE GEOGRAPHIC DISTRIBUTION OF PUBLIC SPENDING .............................. 73 Total Spending ...............................................................................................................................................73 Anti-poverty programs ...................................................................................................................................77 Education...................................................................................................................................................78 Health.............................................................................................................................................................80 Agriculture ................................................................................................................................................81 Water..............................................................................................................................................................85 Electricity .......................................................................................................................................................89 Summary and Recommendations...................................................................................................................90 4. INSTITUTIONS FOR PUBLIC EXPENDITURE MANAGEMENT .......................... 93 Background and Salient Features...................................................................................................................93 Planning, Performance Measurement, and Budget Allocation.......................................................................96 Budget Formulation and Capital Programming in Relation to Strategic Planning.......................................101 Capital Budgeting....................................................................................................................................103 Medium-Term Budgeting: International Experience...............................................................................105 Spending Rigidities and Budget Management ........................................................................................107 Budget Execution.........................................................................................................................................108 Sustaining Macro Fiscal Targets as Revenues and Cost Change (Role of Planeación Hacendaria)........108 Role of the Subsecretaría de Egresos (SSE) ............................................................................................109 Treasury...................................................................................................................................................110 Role of Secretaría de Función Pública ....................................................................................................111 Transparency and Accountability.................................................................................................................112 Budget Classification, Transparency, and Accountability.......................................................................113 Budget Documents ..................................................................................................................................115 The Role of Congress ..............................................................................................................................116 Recommendations ...................................................................................................................................118 Reform Implementation: Sequence and Pace for Managing Politics and Logistics ................................121 References....................................................................................................................................................125 Annexes........................................................................................................................................................131 List of Tables Table 1.1: Mexico – Fiscal Summary ..................................................................................................................2 Table 1.2: Mexico – Federal Government Revenues...........................................................................................3 Table 1.3: Capital Expenditures (GDP Shares)....................................................................................................4 Table 1.4: Rigidities in the 2003 Budget ...........................................................................................................10 Table 1.5: Comparison in Budget Rigidities......................................................................................................11 Table 1.6: Mexico: Fiscal Sustainability Analysis............................................................................................14 Table 2.1: Distribution of Autonomous Income and Expenditure .....................................................................23 Table 2.2: Poverty Rates with Food Poverty Line .............................................................................................25 Table 2.3: Cost of Public Redistributive Programs (millions of 2002 pesos) ....................................................27 Table 2.4: Average Schooling of Adult Population (25–65 years) ....................................................................28 Table 2.5: Distribution of Public Expenditure on Education Services, 2002.....................................................29 Table 2.6: Evolution of Public Expenditure on Education Services, 1992–2002 ..............................................32 Table 2.7: Distribution of Households Benefiting from Public Scholarship Programs, 2002 ...........................34 Table 2.8: Evolution of Federal Expenditure on Health Services, 1996–2002 ..................................................35 Table 2.9: Distribution of Public Expenditure on Health Services, 2002 ..........................................................35 Table 2.10: Distribution of Beneficiaries of Public Expenditure on Pensions, 2002 .........................................39 Table 2.11: Federal Transfers to IMSS Pensions (million MxP, base = 2002)..................................................40 Table 2.12: Deficit ISSSTE Pension Fund.........................................................................................................40 Table 2.13: Mexico: Public Sector Pensions......................................................................................................42 Table 2.14: Distribution of Monetary Transfers, 2002 ......................................................................................45 Table 2.15: Poverty Rates with the Food Poverty Line, 2002 ...........................................................................46 Table 2.16: Evolution of Oportunidades and Procampo Transfers ....................................................................46 Table 2.17: Evolution of Coverage and Private and Public Expenditure on Residential Electricity, Administrative Records, 1992–2002 .........................................................................................................49 Table 2.18: Distribution of Residential Electricity Subsidy, 1996–2002...........................................................50 Table 2.19: Consolidated Results for all Residential Tariffs .............................................................................51 Table 2.20: Distribution of Households with Access to Piped Water within.....................................................52 Table 2.21: Distribution of Total Public Expenditure, 2000–02* ......................................................................55 Table 2.22: Changes in the Distribution of Total Public Expenditure between 2000 and 2002, National*.......55 Table 2.23: Distribution of Household Expenditure before and after Public Transfers, 2002 ...........................57 Table 2.24: Inequality of Education and Health Resources before and after Public Expenditure,* 2002..........59 Table 2.25: Reduction of Poverty due to Oportunidades and Procampo Transfers (% reductions in poverty measures)...................................................................................................................................................62 Table 2.26: Students with "Sufficient" Academic Performance in Primary Schools, Estándares Nacionales (2000) ........................................................................................................................................................63 Table 2.27: Public Spending per Student in Primary Education ........................................................................64 Table 2.28: Efficient versus Progressive Taxation.............................................................................................67 Table 2.29: Benefits and Taxes: Net Impact ......................................................................................................68 Table 3.1: Water and Sanitation Investment by financing source, 1995-2002 (Millions 2002 pesos)................88 List of Figures: Figure 1.1: Oil and Non-Oil Revenues ................................................................................................................2 Figure 1.2: Composition of Federal Government total Expenditures ..................................................................5 Figure 1.3: Composition of Federal Government Expenditures by Functions.....................................................5 Figure 1.4: Economic Composition of Federal Government Expenditures .........................................................7 Figure 1.5: Expenditures of State-owned Enterprises ..........................................................................................8 Figure 1.6: Budget Rigidities, 2002 ...................................................................................................................11 Figure 1.7: Total Fiscal Balances.......................................................................................................................12 Figure 1.8: Federal Government Current Balance .............................................................................................13 Figure 2.1: Distribution of Autonomous Income, 2002 .....................................................................................24 Figure 2.2: Distribution of Public Education Services and Relevant Age Groups, National, 2002 ...................30 Figure 2.3: Evolution of the Distribution of Public Education Expenditure, Concentration Coefficients and Share of Poorest 20% ................................................................................................................................33 Figure 2.4: Distribution of Use of Public Services for Maternal Health and Infant Population, National, 2002 ...................................................................................................................................................................36 Figure 2.5: Distribution of Public Health Expenditure, 1996–2002, .................................................................37 Figure 2.6: Distribution of pensioners and benefits received: 2003...................................................................41 Figure 2.7: Distribution of Procampo Transfers, 1994–2002 ............................................................................47 Figure 2.8: Distribution of Public Expenditure on Food Programs, 1994–2000................................................48 Figure 2.9: Concentration Coefficients, National, 2000, 2002 ..........................................................................53 Figure 2.10: Concentration Coefficients, Rural, Urban, 2002 ...........................................................................53 Figure 2.11: Distribution of Oportunidades Transfers and Students in Primary Education in the Rural Population, 2002........................................................................................................................................54 Figure 2.12: Public Expenditure as a Proportion of Autonomous Expenditure, National, 2002........................58 Figure 2.13: Public Expenditure as a Proportion of Autonomous Expenditure, Urban, Rural, 2002 ...............58 Figure 2.14: Expenditure on Education and Health: Autonomous and Public...................................................60 Figure 2.15: Public Expenditure on Education and Health as a Multiple of Autonomous Expenditure on Education and Health, Urban, Rural, 2002................................................................................................61 Figure 2.16: Oportunidades and Procampo Transfers as a Proportion of Autonomous Expenditure/Income in Rural Population, Rural, 2002 ...................................................................................................................62 Figure 2. 17: Student-Teacher Ratio and Public Spending per Teacher in Primary Education, 1970–2000......64 Figure 2.18: Teacher Salaries in Basic Education..............................................................................................65 Figure 2.19: Ratio of Salary after 15 Years of Experience to GDP per Capita, 2000.......................................65 Figure 3.1: Geographic distribution of per capita resources, 2002 ....................................................................74 Figure 3.2: Geographic distribution of per capita resources, 1992 ....................................................................75 Figure 3.3: Geographic distribution of public resources, as a share of state GDP, 2002 ...................................76 Figure 3.4: Geographic distribution of targeted anti-poverty spending, 2002 ...................................................77 Figure 3.5: Geographic distribution of federal education funds, 2002...............................................................79 Figure 3.6: Geographic distribution of federal health spending, 2002...............................................................81 Figure 3.7: Geographic distribution of public spending for agriculture per capita of rural population, 2002....84 Figure 3.8: Geographic distribution of agricultural electric subsidies, 2002 .....................................................85 Figure 4.1: Relationship between Planning and Budgeting ...............................................................................97 1. MEXICO’S PUBLIC FINANCE OVERVIEW 1.1 Mexico administers its public finances and expenditures well compared with other countries, although it still faces fiscal challenges related to its contingent liabilities and limited tax income. Over the past decade, Mexico maintained modest fiscal deficits and a manageable stock of debt and other financial obligations. The government has been making explicit many of its past contingent liabilities—in the financial sector, in part of the pension system, and in the guarantees provided to private concessionaires in the transport sector—and thus its performance is even better than the numbers show. At the end of 2002, the explicit net debt of the nonfinancial public sector was about 21 percent of GDP, on top of which were off-budget debts and contingent liabilities of another 95 of GDP. With the fiscal balances of recent years, Mexico’s stock of debt and other financial liabilities is declining as a share of GDP and government revenues, so the present level of debt is sustainable. 1.2 Mexico has a relatively rigid spending budget, and about a third of its revenues depend on oil revenue. Mexico has been improving its tax system by eliminating some exemptions in its income tax code and by improving its tax administration, but it should do more to reduce its dependence on oil revenue. Public sector investment has been the main adjustable variable in Mexico’s budget, and almost all the investment has been financed with additional borrowing rather than by mobilizing government savings. Social sector expenditures constitute about 8 percent of GDP, and as discussed in the next chapter, such expenditures should be increased in order to alleviate poverty and spur economic growth. 1.3 This chapter discusses the trends in and composition of fiscal balances, revenues, expenditures, and debt as well as fiscal sustainability and contingent liabilities. It concludes by offering some policy recommendations. REVENUE AND EXPENDITURE TRENDS AND FISCAL BALANCES Revenue 1.4 From 1990 to 2002 the federal government’s total real revenues grew 40 percent (see Table 1.1). The share of tax revenue in federal revenue increased slightly during the period, from close to 68 percent in 1990 to 73 percent in 2002. In contrast, the share of non-oil revenues declined from 71 to 66 percent (see Figure 1.1). -1- Table 1.1: Mexico – Fiscal Summary Federal Government (2002 MxP Thousand Million) 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 Total Revenues 707 868 892 752 796 749 781 831 776 823 968 984 987 Tax Revenues 476 498 537 553 579 455 450 515 575 637 648 686 728 Non-Tax Revenues 231 371 355 199 217 294 331 316 201 186 319 298 259 Capital Revenues 4 3 30 22 5 5 2 25 Total Expenditures 823 732 697 736 813 789 804 902 871 921 1,061 1,045 1,122 Programmable Expenditures 323 353 385 444 532 454 462 540 559 573 657 662 743 Current Expenditures 224 263 285 363 423 366 369 441 469 487 566 565 606 Transfers 72 94 157 221 261 205 218 308 334 364 404 428 458 Wage Bill 102 114 75 91 100 77 81 70 74 79 93 96 99 Goods and Services 7 7 8 9 10 7 9 8 8 7 7 8 8 General Services and others 20 29 32 31 38 25 26 29 17 16 40 23 26 Capital Expenditures 99 90 99 81 110 89 93 100 90 85 91 97 137 Non-Programmable Expeditures 501 378 312 292 281 334 342 362 313 348 404 383 379 Financial Cost 360 212 155 111 98 188 188 188 136 177 197 175 158 Participaciones 124 131 140 146 150 131 141 156 162 172 199 206 214 Others 17 35 17 35 33 15 13 17 15 0 9 1 6 Fiscal Balance -117 137 195 16 -17 -40 -23 -71 -96 -98 -93 -61 -135 Primary Balance 218 323 334 124 78 128 149 111 37 76 99 112 23 Current Balance 45 68 -2 -28 -17 -8 34 -23 Source: SHCP and Bank Staff Calculations Figure 1.1: Oil and Non-Oil Revenues g 20% 18% 16% 14% As a share of GDP 12% Non-Oil Revenues 10% Oil Revenues 8% 6% 4% 2% 0% 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 Source: SHCP and Bank staff calculations. 1.5 Of the federal government taxes, the combined income taxes—personal income tax and corporate income tax, including the gross asset tax—are the most important sources of revenue (see Table 1.2). In 2002 income taxes accounted for 32 percent of total federal revenues. The second most important source is the value added tax (VAT), which represented 23 percent of total revenues in 2002. As a result of liberalization and, in particular, the North American Free -2- Trade Agreement (NAFTA), the relative importance of taxes on international trade has declined, falling to 2.8 percent of federal revenues in 2002. Table 1.2: Mexico – Federal Government Revenues Federal Government (2002 MxP Thousand Million) 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 Total Revenues 707 868 892 752 796 749 781 831 776 823 968 984 987 Tax Revenues 476 498 537 553 579 455 450 515 575 637 648 686 728 Income tax 198 210 246 267 263 197 193 223 241 264 288 299 319 Value added tax 160 159 129 128 139 138 143 161 171 185 211 218 218 Excise on goods and services 67 62 77 75 101 66 59 75 109 130 91 116 136 Imports 38 49 55 49 46 30 30 30 31 33 37 30 27 Luxury Goods and Services 12 17 31 34 30 24 24 26 24 25 21 22 26 Other 0 0 0 0 0 0 0 0 0 0 0 0 2 Non-Tax Revenues 231 371 355 199 217 294 331 316 201 186 319 298 259 Contributions 0 0 0 0 0 0 0 0 0 0 0 0 0 User charges 176 172 164 154 127 186 226 218 149 130 235 214 158 Products 12 12 16 7 12 24 14 14 17 10 8 7 5 Aprovechamientos 35 176 175 39 77 84 91 84 34 47 76 77 96 Accessories 8 10 0 0 0 0 0 0 0 0 0 0 0 Source: SHCP and World Bank calculations 1.6 The relative importance of income taxes fluctuated significantly during the past decade. These oscillations reflect changes in tax policy, the high sensitivity of income taxes to the business cycle, fluctuations in revenues from petroleum, a significant hike in privatization proceeds in the early 1990s, the possible effects of inflation, and administrative efforts to control tax evasion. Income taxes were as high as 36 percent in 1993 and as low as 25 percent in 1996. The relative importance of the VAT also fluctuated widely during the decade. This fluctuation reflects, among other things, the drop in the VAT rate from 15 to 10 percent in 1992 and its increase again during the 1995 recession. EXPENDITURES 1.7 Mexico’s federal public sector budget is around 22 percent of GDP, about three-fourths of which is executed by the federal government, and the rest by the state-owned enterprises. Programmable expenditures account for almost 72 percent of the total budget. For the last three years its composition has not changed much. Capital expenditures in the budget averaged about 2.7 percent of GDP up to 2002 and represent about 11 percent of the total budget. Almost all of the capital expenditure is for public investment, and about 40 percent of it is done in the social sectors. (See Table 1.3). Recently public investment, including off-budget operations (PIDIREGAS) of public enterprises, has increased to over 4 percent projected in 2004, which compares favorably to other Latin American countries. (In 2002, public investment as a share of GDP was: Argentina 0.1 percent, Brazil 3.8, Colombia 7.9, and Peru 3.1). -3- Table 1.3: Capital Expenditures (GDP Shares) 1998 1999 2000 2001 2002 Capital Expenditures 2.9 2.7 2.7 2.6 2.5 Public Investment 2.8 2.6 2.6 2.5 2.5 Other Capital Expenditures 0.1 0.1 0.1 0.2 0.0 Investment in Social Development 1.0 1.0 1.1 1.1 1.1 Education 0.2 0.2 0.2 0.2 0.2 Health 0.1 0.1 0.1 0.1 0.0 Social Security 0.0 0.0 0.0 0.0 0.0 Regional and Urban Development 0.6 0.6 0.7 0.8 0.8 Social Assistance 0.0 0.0 0.0 0.0 0.0 Labor 0.0 0.0 0.0 0.0 0.0 Source: Estadísticas de Finanzas Públicas 1992-2001 y 1993-2002. Federal Government’s Expenditures 1.8 From 1990 to 2002 the federal government’s total real expenditures grew 48 percent, and in 2002, they were about 18 percent of GDP. The composition of total expenditures changed dramatically during the decade. Programmable expenditures, which represented about 39 percent of the federal government’s budget in 1990, now account for about two-thirds of it, with the social sector being the largest component (see Figure 1.2). Nonprogrammable expenditures are about a third of the federal government’s budget, and their composition changed dramatically in the past decade. Financial costs dropped from about 43 percent of the federal government’s budget down to 14 percent, and participaciones—the proportion of federal domestic tax revenue that is shared with states and municipalities—rose from about 15 percent of total expenditures in 1990 up to about 19 percent in 2002. 1.9 Government investment has been low since the late 1980s, in the range of 1.5 to 2.25 percent of GDP for the central government until 2002, with some increase since then. The reductions were needed at the time of the 1980s debt crisis, but future growth may require increased infrastructure investment, such as in electricity and transport, which the Infrastructure PER will investigate further.1 In the public enterprises, budgeted investment remained almost frozen in nominal terms since 1996, thus falling to about 1 percent of GDP, although quasi- private investment on behalf of the enterprises, financed off-budget with PIDIREGAS, is about twice the budgeted amount. 1.10 Figure 1.3 shows the evolution of programmable expenditures by function (plus participaciones, which are non-programmable). Social development expenditures constitute the largest share of programmable spending and of the total expenditures, averaging. 45 percent of the federal government’s budget for the last five years. The main destinations for increase spending are participaciones (revenue sharing that has grown proportionally with total federal 1. Earlier studies showed that increasing investment in transport and electricity would have a significantly positive effect on growth. Mexico: Fiscal Sustainability. World Bank 2000. -4- revenue), agriculture in 2002 (mostly a one-time payment to close Banrural and cover debts built up over years), education, and social security pensions. Figure 1.2: Composition of Federal Government total Expenditures 100% 90% 35.9% 33.8% 37.8% 38.1% 36.6% 80% 70% Percentage of Total Expenditures 60% 8.6% 7.2% 11.4% 7.5% 6.8% Non-Programmable Expenditures 50% Economic Sectors Social Development 40% 43.3% 45.1% Government 45.2% 43.9% 44.6% 30% 20% 10% 10.3% 10.8% 11.8% 11.1% 10.2% 0% Source: SHCP and World Bank calculations Figure 1.3: Composition of Federal Government Expenditures by Functions 1,200 1,000 Participaciones Com m unication and Transports 800 Energy 2002 MxP Thousand Million Agriculture Regional and Urban Developm ent Social Assistance 600 Labor Social Security Health (Via State Gov.) 400 Health (Fed. Gov. Direct) Education (Via State Gov. .) Education (Fed. Gov. Direct.) 200 General Governm ent 0 1998 1999 2000 2001 2002 Source: SHCP and World Bank calculations -5- 1.11 Education expenditures—mostly transfers to cover the payroll of teachers retired from the public school system—constitute the largest share of social development expenditures and over a fourth of the federal government’s total expenditures. Mathematically, such a large sector cannot continue to grow faster than GDP without increasing tax revenues as a share of GDP. Broadening secondary education is a big item of spending that needs expansion, but this will need than money, as discussed in the next chapter. Expenditures on health, social security, labor, social assistance, and regional and urban development are small in comparison with education. Over half of the health expenditures are transfers to cover the payroll of the personnel transferred to the states in 1998. 1.12 Mexico reformed the Instituto Mexicano del Seguro Social (IMSS, Mexican Social Security Institute —the private sector pension system) in 1997. As an obligation from the reform, the government makes current payments to social security in the IMSS pension system is expected to rise to a peak of 0.8 percent of GDP by 2020 and then to decline to a steady-state level of 0.3 percent of GDP by 2075.2 The Instituto de Seguridad y Servicios Sociales de los Trabajadores del Estado (ISSSTE, Institute for the Social Security of State Workers—the federal pension system) still needs reform, and failure to do so threatens the country’s long-run fiscal sustainability. Already ISSSTE runs substantial deficits, with the government covering 60 percent of expenditures outright in addition to making the contributions it is obliged to make as employer, resulting in a cost of 0.2 percent of GDP for the deficits alone. As large cohorts in ISSSTE retire, the deficits are expected to rise to 0.5 percent of GDP by 2010 and to 1 percent by 2030. The government has a proposal to reform the ISSSTE pension system by bringing federal civil servants into the national system. 1.13 There are several options for financing the costs of the proposed transition from a public to a private pension system. The government could issue debt, and given that the administradoras de fondos para el retiro (AFOREs, Mexico’s pension fund management firms) invest in government bonds, it is likely that this will have only a small impact on interest rates. The government also could create more flexibility in its budget, allowing it to adjust allocations to finance the rest of the reform. 1.14 Figure 1.4 shows the recent evolution of the federal government’s total expenditures3 by economic classification. Transfers increased dramatically over the past decade, mostly going to 2. The 1997 reform transformed IMSS from a pay-as-you-go system to a defined-contribution system, in which individuals put their contributions into individual accounts, managed mostly by private companies. As a result, the contributions, which partly offset the expenditures that IMSS made to retirees under the old system, disappeared from the balance sheet of IMSS. This requires ongoing transfers from the central revenues to cover the gap during the (long ) transition. The long-run objective of the reform, in addition to providing a more secure retirement for workers, was to limit the government’s liabilities toward the elderly as the population ages and the elderly become more numerous. In the absence of reform, deficits in the system would have reached 2.5 percent of GDP by 2050 and 3.2 percent of GDP by 2075. Three components of government current transfers will continue indefinitely: (1) a government contribution applicable to all contributors as a percentage of their own wages, (2) the social quota (cuota social), which is a flat payment equal to 5.5 percent of the minimum wage in July 1997 that the government contributes for each day that an individual makes a contribution, and (3) the minimum pension guarantee equal to the minimum wage in July 1997 made to all contributors who have reached retirement age and completed 25 years of contributions 3. It excludes the state-owned enterprises expenditures. -6- subnational governments, pension systems and public enterprises. The first big increase came in 1992–93, when federal teachers were transferred to the state governments; accompanying this transfer was a reduction in the federal wage bill and an increase in transfers from the federal to state governments to pay for teachers’ salaries. Transfers jumped again in 1997–98, when part of the health system and social infrastructure expenditures were decentralized and Ramo 33 was created to aggregate all the earmarked aportaciones to the subnational governments. This increase in transfers was accompanied by a reduction in capital expenditures, (central government) wages, goods and services, general services, and others. Debt service declined dramatically as a consequence of debt restructuring during the Salinas administration. Goods and services, which are the smallest category in the budget, declined, but this level of expenditures probably is not sustainable, given their importance for complementing investment and personnel expenditure in order to actually deliver services. Figure 1.4: Economic Composition of Federal Government Expenditures 100% 80% O thers Participaciones As share of total expenditures 60% D ebt Service C apital Expendit ures G eneral Services and others 40% G oods and Services W age Bill 20% T ransfers 0% 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 -20% Source: SHCP and World Bank calculations State-owned Enterprises Expenditures 1.15 Net expenditures of state-owned e nterprises comprises about 4 percent of GDP. Their payroll has increased during the last decade from 20 percent of total expenditures to 33 percent in 2002. Interest payments on debt have declined in the last five years, even though interest payments on foreign debt have increased slightly. As with the federal government’s budget, capital expenditures for state-owned enterprises have been the adjustable variable and have declined since 1996 in real terms and as a share of GDP (see Figure 1.5). 1.16 Mexico’s electricity sector is large by any standard. The Federa l Electric Commission (CFE, a state-owned monopoly) is among the largest utilities in North America, and together with Central Light and Power (LFC, the government-owned distribution company serving metropolitan Mexico City and surrounding areas) provide electricity to 95 percent of the population. Mexico’s pattern of economic grow th and the high level of connections resulted in a -7- 5.2 percent annual growth in electricity demand over the last decade. Another contributory factor to rapid growth in demand has been extensive subsidies. Figure 1.5: Expenditures of State-owned Enterprises 500 450 400 350 2002 MxP Thousand Million 300 Capital Expenditures Interest-Foreign Debt Interest-Domestic Debt 250 Pensions Other 200 Wage Bill 150 100 50 0 1998 1999 2000 2001 2002 Source: SHCP. 1.17 The average tariff charged to residential customers in 2002 covered just 45 percent of costs, and the average tariff for agricultural use covered 29 percent of costs. Industry and services paid about 97 percent of costs. Total implicit subsidies amounted to MxP 45 billion in 2002. Adding direct cash transfers to LFC of MxP 17 billion to cover operating losses, the federation provided subsidies of MxP 62 billion in 2002, equivalent to 1 percent of GDP, 40 percent of the entire federal investment budget, and 164 percent of total investment in the sector from all sources. (See Annex 1 for a detailed discussion.) Chapters 2 and 3 discuss how these subsidies go mostly to upper-income households and regions. 1.18 The large direct cash transfers to LFC are merely the most visible drain on fiscal resources. The implicit subsidies reduce substantially the fiscal resources otherwise available to the federation for much needed investment in the sector and for other sectors, although that effect cannot be precisely quantified. There are several mechanisms through which electricity subsidies affect federal finances: the government does not receive the dividends or tax revenue it should on earnings the state electricity companies would otherwise make; the implicit subsidies reduce CFE’s profits and borrowing capacity; and the scope for financial leverage, central to utility financing, is attenuated. Without the subsidies, CFE would have higher and more predictable cash flow levels, and could mobilize more than proportional volumes of funds from private sector sources (without encumbering federal borrowing capacity to the same degree as -8- now). The subsidies virtually eliminate retained earnings, which undermines investment in the sector and constrains long-term GDP growth, hence tax receipts from other sectors in subsequent years. 1.19 The electricity tariff regime in Mexico is complex: Tariffs vary substantially according to a number of geographic, climatic, and end use criteria which depart significantly from costs of supply for medium and high voltage users (industrial and medium/large commercial enterprises, accounting for approximately 60 percent of electricity sales in 2002), economically rational time- of-day tariffs based on marginal cost have become the norm above 100kW demand. For other industrial and commercial users (with about 11 percent of total power use), low-voltage tariffs are based on complex criteria that do not fully reflect costs their consumption imposes on the system. For residential, agricultural and water pumping users, “special” subsidized tariffs are set largely on political criteria. (Residences use about 23 percent of total power, and agriculture 6 percent.) Without further changes to tariff structure, residential subsidies are likely to grow by at least 40–45 percent over the next 5 years, totalling up to MXP 60 billion. Chapters 2 and 3 show that the electricity subsidies go mostly to the middle and upper income families and to the regions that are already more economically developed. The vast majority of the subsidy—over 90 percent—that is not a life-line for the poor, encourages wasteful inefficiency, especially in the hot areas in the summer. The artificially low prices, encourage people to import inefficient used air conditioners, rather than finding ways to save energy. 1.20 In the water sector many subsidies arise via under-pricing that fails to cover most operation and capital costs, not to mention the value of water resources which constitutionally belong to the nation. Because the sector operates through many decentralized entities, we do not know the aggregate value of the subsidies, although it is surely large. Budget Rigidities 1.21 The rigidity of the budget in Mexico has two aspects: the first within the budget year and the second from one budget year to the next. The short-run problem is that 57 percent of the budget goes for debt service, wages, nominally specified transfers to the subnational governments, and other inflexible obligations. Although this is low compared to other countries like Peru, Colombia, Argentina, and Brazil, the short-run rigidity means that any unanticipated decline in revenue leads to declines that are a much higher percent the remaining categories. Especially investment takes a hit, for it is only about 11 percent of the budget—low by international standards. Revenue sharing with the subnational governments, while fixed in law and thus hard to change in the medium term, actually mitigates short-term rigidity, in that that the participaciones formula automatically makes the subnational governments share proportionally in the burden (or benefit) from total revenue fluctuations. 1.22 The medium term problem with rigidities arises when the government wants to cut spending in non-priority areas, in order to reallocate expenditure in line with the new priorities decided by the democratic process. In some ways the picture has already improved, as items like debt service, which cannot be changed even by law, have actually declined in the last few years to about one-fourth of outlays, although they will likely rise again as pension payments to retirees rises. Items that are fixed by law and changeable by law (other than the annual budget law), like participaciones to subnational governments and existing pay levels for federalized -9- teachers, are just over one-half. See Table 1.4 and Figure 1.6. Other not-easily reducible expenditures are those for medicines, non-earmarked transfers to subnational governments (Program for the Strengthening of Federal Entities, PAFEF, Programa para el Fortalecimiento de las Entidades Federativas,), student scholarships, and transfers to universities, which have traditionally stayed at the same level or increased every year. Table 1.4: Rigidities in the 2003 Budget Strict Obligations Obligations that are fixed Entitlements in Non-reducible Others Total (MxP thousand million) or changeable by regular law the Annual Law Expenditures in the Annual Law Contractual 873.8 279.2 0.0 0.0 2.5 1,155.5 As share of total Expenditures 57.3% 18.3% 0.0% 0.0% 0.2% 75.8% Debt Financial Cost 184.9 Federal Government 128.2 Entities under Direct Control 26.4 Ramo 34 30.2 Pidiregas and BLT's payments 38.0 Adefas 7.0 Pensions 141.2 Existing Wages and Salaries 487.6 Participaciones a entidades federativas y municipios * 226.7 Ramo 33 (Aportaciones excl. FAEB and FASSA) 52.5 Funds linked to RFP* 52.5 FAIS 22.3 Fortamun - DF 22.9 FAM 7.3 Fondos para Salud y Educación 14.8 (excluye Servicios Personales) FAEB 8.9 FAETA 0.7 FASSA 4.8 Ramo 25 0.3 Fondo para la Seguridad Publica 2.5 Reasonably ineludible 0.0 14.2 47.3 87.1 60.4 209.0 As share of total Expenditures 0.0% 0.9% 3.1% 5.7% 4.0% 13.7% Poderes y Entes Autonomos ** 12.4 PAFEF 17.0 OPORTUNIDADES 20.6 PROCAMPO 14.2 Alianza para el Campo 6.3 Indigenous Programs 1.8 Programa Ganadero 1.5 Transfers to Public Universities 4.2 Convenios de Educacion 22.5 Medicamentos (Salud, IMSS e ISSSTE) 21.7 CONALITEG 1.6 CONACYTs scholarships 1.6 PEMEX investment projects 19.3 CFEs fuels 41.1 CFEs Maintenance and labor for inv. 4.3 LyFC labor for inv. 1.7 Most Imp. Road Projects and Maint. 11.5 Hidraulic Program 5.2 Fondo de Desastres Naturales 0.6 Others 160.7 160.7 As share of total Expenditures 10.5% 10.5% Total Expenditures 873.8 293.4 47.3 87.1 223.6 1,524.8 As share of total Expenditures 57.3% 19.2% 3.1% 5.7% 14.7% * Automatically adjusted with Revenue Fluctuations Source: SHCP and World Bank - 10 - Figure 1.6: Budget Rigidities, 2002 Others 15% Non-reducible Expenditures in the Annual Law 6% Entitlements in the Annual Law 3% Strict Obligations 57% Obligations that are fixed or changeable by regular law 19% Source: SHCP and World Bank calculations. 1.23 Compared with other Latin American countries, in Table 1.5, Mexico’s strict obligations are smaller than those from Brazil, Colombia and Peru; but Mexico has the largest share of obligations and entitlements changeable by law. This means that Mexico has relatively good scope for short-term macro-fiscal management. On the other hand, to reallocate spending in the medium term will require more cooperation between Congress and the executive, to agree on the change of priorities. Table 1.5: Comparison in Budget Rigidities (2002, A a percent of the Budget) Strict Obligations Obligations/Entitlements Others changeable by law Argentina 50.21 27.6 22.19 Brazil 78.9 5.8 15.3 Colombia 64.6 11.4 24 Mexico 57.3 28 14.7 Peru 65.6 20.7 13.7 [** under Strict Obligations, put (debt service and pensions) Source: SHCP, World Bank calculations, including in other PERs. 1.24 Changing a law to reduce expenditures would require strong political leadership and clear communication of the reasons for making reallocation of the budget a higher priority. While the established spending programs have strong constituencies and a lot of inertia, Mexico’s own experience in the 1990s shows that new programs that are well-conceived and publicly presented can draw resources away from older and apparently entrenched programs. For example, the generalized food subsidy programs—including the expensive tortilla subsidy—were successfully replaced with the much better targeted Progresa/Oportunidades program. And a number of agricultural subsidy programs, which would have been impractical in the NAFTA context, were replaced with the better-targeted and less distortionary Procampo, discussed below. - 11 - 1.25 Personnel expenses, including for decentralized “federal” teachers and health workers, are the largest and fastest growing part of the rigid expenditures. As seen in appendix Table A1.1, aggregate wages and salaries increased at 6 percent a year on average in real terms from 1998–2003, although their composition varied throughout the last two administrations. This shows some upward flexibility in this area, which could be restrained although not easily reversed. Another dimension of rigidity in the wage bill is that the employees are difficult to move even within their sector, like from a school with excess teachers to one where they are needed. As mentioned in the next section, bringing civil servants into the national pension system would improve labor mobility and allow more flexibility in this budget item. 1.26 To summarize, a little over half of the budget is rigid in the short and medium term; the rest could be changed in various ways, mostly by changing laws. This would require political effort, whose success would depend partly on the consensus priority about the alternate use of the money. DEFICITS, DEBT, AND FISCAL SUSTAINABILITY 1.27 Mexico ran high deficits in the late 1980s, but in the early 1990s it achieved overall surpluses, not counting the accumulation of contingent liabilities (see Figure 1.7). Since the crisis of December 1994, the federal government has run modest deficits, largely attributable to the negative effects of the crisis, in which the government had to bail out the financial sector, some toll-road concessions, and the social security system. Since 1998 the government has made efforts to lower the deficit, although in 2002 the deficit rose once again as a result of the decision to liquidate Banrural, an agricultural sector bank that had been a large contingent liability of the government. Figure 1.7: Total Fiscal Balances 5 4 T o ta l F is c a l B a la n c e 3 2 F e d e ra l G o v e rn m e n t As share of GDP F is c a l B a la n c e 1 0 S ta te -O w n e d E n te rp ris e s a n d 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 In s titu e s F is c a l B a la n c e (S e c to r P a ra e s ta ta l, -1 in c lu d e s s o c ia l s e c u rity in s titu te s ) -2 -3 Source: SHCP and World Bank calculations 1.28 Since 1997, Mexico has financed its public investment almost completely by borrowing, rather than by mobilizing its own savings, because current expenses have been almost as high as current revenues (see Figure 1.8). This pattern of public finance has two problems alluded to - 12 - above: reducing flexibility to adjust aggregate expenditure in the short-run and a level of investment that may be too low to foster satisfactory economic growth. Figure 1.8: Federal Government Current Balance 2 .5 % 2 .0 % 1 .5 % 1 .0 % 0 .5 % As a share of GDP C u r r e n t B a la n c e 0 .0 % C a p it a l E x p e n d i t u r e s 1995 1996 1997 1998 1999 2000 2001 2002 O v e r a ll B a la n c e -0 .5 % -1 .0 % -1 .5 % -2 .0 % -2 .5 % Source: SHCP and World Bank calculations 1.29 The changes of the federal government’s deficits in the 1990s resulted from electoral cycles, the price of oil, and tax policy. The deficit in 1994, an election year, can be attributed to two factors: (1) non-tax revenues dropped significantly from 1992 through 1994, and (2) capital expenditures increased almost 36 percent from 1993 to 1994. The drop in revenues from 1992 to 1994 can be attributed not to oil prices, because oil revenues remained almost constant throughout this period, but rather to the VAT rate, which was lowered from 15 to 10 percent, as noted in Table 1.2. The deficits from 1995 through 1996 can be attributed to (1) a drop in tax revenues and (2) an increase in debt service, two consequences of the fiscal crisis. From 1996 through 1997 there was a substantial increase in transfers as well as in debt service, which led to a further increase in the federal government deficit. The transfers became larger as a result of the reform the social security system, making contingent liabilities explicit, and of changes in the Fiscal Coordination Law (Ley de Coordinación Fiscal), which made more resources available for social spending and investment by subnational governments. 1.30 Over the past decade, the net public debt declined substantially, from 45 percent of GDP in 1990 to 21 percent in 2002. During the 1995 crisis, the debt rose to slightly above 30 percent of GDP, but the Zedillo administration brought it down to 20 percent through a combination of fiscal adjustment and growth of GDP (the denominator). On top of this, the debts of the Fideicomiso de Apoyo al Rescate de Autopista Concesionadas (FARAC, Trust Fund for Bailout of Toll Road Concessions) and the off-balance-sheet debts Instituto para la Protección del Ahorro Bancario (IPAB, Institute for the Protection of Bank Savings) inherited from the Fondo Bancario de Protección al Ahorro (FOBAPROA, or the Banking Fund for the Protection of Savings) were about 20 percent of GDP, which still leaves debt in the range of 40 percent of GDP—very manageable and relatively low by international standards. - 13 - 1.31 The ratio of debt to GDP that is sustainable (debt as a share of GDP constant or declining) in the long-run depends on the difference between the real interest rate and the real rate of output growth, and the stock of debt as a share of GDP (Montes-Negret and Draaisma 2000) To see if Mexico’s debt is sustainable or not, Table 1.6 applies the same analysis to the situation today. Financial requirements include net public debt, off-balance-sheet FOBAPROA debts inherited by IPAB, and the debts of FARAC. Average growth and interest rates are for the 1993-2002. The low case pushes the interest-growth differential to 3 times the average. Even then, the actual primary balance for 1993-2002 would suffice to keep the debt ratio declining. Table 1.6: Mexico: Fiscal Sustainability Analysis Average, 1993-2002 Low Case Assumptions (annual percent) r (Real interest rate) 4.0% g (Real GDP growth) 3.0% (r-g) 1.0% 3.0% I. Domestic and External Debt Stock 21.3% 21.3% Minimum Primary Balance for Sustainability 0.2% 0.6% II. Off –Budget Liabilities 20.4% 20.4% III Financial Requirements Stock (I+II) 41.7% 41.7% Minimum Primary Balance for Sustainability 0.4% 1.2% Actual Primary Balance 3.8% 3.8% Source: Bank Staff Estimates 1.32 There are large contingent liabilities for the pension systems, especially for public sector workers. Assuming reasonable parameters for the salary trends, life expectancy and discount rate, the contingent liabilities of the ISSSTE pension system amount to about 45 percent of GDP, and those of the IMSS pension system to about 31 percent. State and municipal debt, another liability equaling about 2.5 percent of GDP, does not pose a serious treat to the federal government, although the liability is large in relation to the disposable income of a few states. This pension liability is not interest-bearing debt, so the sustainability calculation is different. For a pay-as-you-go system, which ISSSTE still is, financial sustainability depends on the growth of the number and pay of contributing workers relative to the number and expense of the retirees. Already ISSSTE requires and gets budget subsidies of 0.2 percent of GDP, and in the absence of reform these will quadruple by 2020 and would continue to grow unsustainably with present trends. Hence the impetus for reform. After that, there will be a large transition costs as with IMSS, but the debt will be finite and not growing, although the payments needed on it will grow for about a generation before declining eventually to zero. CONCLUSIONS AND POLICY RECOMMENDATIONS 1.33 Mexico probably needs to expand government investment, to alleviate poverty and foster economic growth, which the forthcoming Mexico Infrastructure PER will investigate in depth. To do this the country needs to increase non-oil revenues and reallocate expenditures toward basic priorities. Mexico ranks among the lowest in the OECD countries in tax collection and still depends heavily on oil revenue. Mexico should increase tax revenues by eliminating more tax exemptions in the income tax code, expanding the base of the value added tax, and improving the administration and collection of taxes. In particular, to obtain congressional approval, tax - 14 - collection needs to be linked with expenditures. The benefit incidence analysis in Chapter 2 could help to formulate this link. 1.34 Although spending needs to be predictable, commitments that are too rigid prevent the government from making long-run reallocations to fundamental new priorities and short-run adjustments as the nation’s revenue situation changes. Rigid spending commitments are easy to make and therefore susceptible to capture by special interest groups. Only about 24 percent of the budget consists of strict obligations like debt payments and payments for social security. Change is possible, but most of the proposed changes will require strong political leadership and a well-communicated vision of how to make spending better. 1.35 Mexico’s level of debt is fiscally sustainable, but the pension systems could become a greater burden on the government finances. Mexico needs to reform its public pension system by linking future benefits to contributions and by making benefits portable so that public employees can move easily in the labor market. The reform could be self-financed in part because most of the AFOREs invest in government bonds - 15 - - 16 - 2. THE DISTRIBUTION OF BENEFITS FROM PUBLIC EXPENDITURE 2.1 In countries with wide economic inequality, such as Mexico and other Latin American countries, the public sector can play a critical role in reducing imbalances, and most people expect it to do so. The Mexico Poverty Assessment is examining in detail the programs arising out of Mexico’s poverty reduction strategy known as Contigo (With You). Public spending has other objectives as well and responds to diverse political pressures. This chapter compares the redistributive effects of public spending on the poor and on the rich as well as the distribution of tax burden with the distribution of spending benefits. As a percentage of income, most taxes take as much or more from the rich as from the poor, so the absolute payment per person is much higher for the rich. Although the per capita benefits for the poor vary widely, the poor receive more benefits from most, but not all, programs than they pay in taxes. In this sense, the combination of public spending and taxation is highly redistributive in Mexico. 2.2 An important policy aim of the Mexican government is to direct more spending to social programs for all Mexicans, with a substantial share targeted to benefit the poor. The trend was evident in the 1990s, and the present government has aimed to accelerate it. The share of public expenditure benefiting the poor can be increased by expanding the share of the budget allocated to pro-poor programs but also by improving the targeting within programs. Improved targeted may, in turn, be achieved through supply-side reforms (for example, extending the coverage and quality of programs in poor regions) or through demand-side interventions designed to increase the participation of the poor (for example, reducing the costs of participation). This trend toward progressivity has involved complementary reforms on all three fronts: increasing the share of spending on basic education and health services for the uninsured, extending the coverage of health and education services (and notably food subsidies) in poor rural areas, and increasing the demand of the poor for these services through conditional transfers (through Progresa, which is now called Oportunidades). The benefit-incidence analysis described in this chapter is helpful for (a) measuring the gains in progressivity achieved, (b) offering a basis for estimating the relative contribution of reforms to these gains, and (c) identifying further opportunities for reform. 2.3 This chapter examines the distributive impact of spending decisions, which often are made program-by-program, in an effort to help policymakers make more informed decisions regarding where to use extra revenue or where to make budget cuts in the case of revenue shortfalls. The main analytical work concerns programs for which quantitative data on household consumption of public services can be matched with quantitative data on the government’s expenditure or costs. It is important to remember, however, that household benefits are often not proportional to the government’s cost (as in the areas of public health, regulation, and justice) and thus are not considered in the quantitative analysis. 2.4 The first section reviews previous studies on the distribution of public expenditure in Mexico, describes the data and methodological basis for the present study, and identifies the opportunities for redistribution through public expenditure by analyzing the distribution of - 17 - human capital and autonomous (pre-transfer) income and expenditures as well as the redistributive instruments available to the government. Section two presents the distribution of public expenditures in services and transfers for 2002, including education services and scholarships, health services, public contributions to pensions for formal and public sector workers, targeted monetary transfers (Procampo and Oportunidades Program), and residential electricity subsidies. This analysis is presented for national population deciles and for urban and rural quintiles. For the principal expenditure programs, this section also presents comparable results covering 1992–2002 and estimates the distribution of marginal benefits due to changes in coverage and budgetary allocations. It also compares the distribution of overall benefits from public expenditures with the distribution of tax burdens. The third section analyzes the redistributive impact of public expenditures on poverty and inequality (such as current transfers) and on the availability of education and health services assets. The final section offers policy recommendations. OVERVIEW 2.5 This section describes previous studies on the distribution of public expenditure in Mexico, presents the data and methodological basis for this study, and analyzes the distribution of human capital and autonomous (pre-transfer) income and expenditures as well as the redistributive instruments available to the government. Previous Studies 2.6 Following the pioneering study by Meerman (1979) on Malaysia and Selowski (1979) on Colombia, benefit-incidence analysis for public social spending has been undertaken for a growing number of developing countries. By the mid-1990s such studies were available for all major countries within the Latin America and Caribbean region, with the notable exception of Mexico.4 2.7 A number of pioneering studies describe the sectoral and geographic allocation of social spending in post-revolutionary Mexico, up to the early 1980s, including Wilkie (1978), Aspe and Beristáin (1984), Lustig (1989), and Maddison and others (1992). Although none of these studies estimates the distribution of social spending by income-ordered population groups,5 several draw broadly pessimistic inferences on the redistributive impact of social spending in this period. Aspe and Beristáin (1984, p. 323), for example, conclude the most detailed and comprehensive review of social spending available for the 1970s with a strong claim: “The greatest significance of this study is a negative one: the educational and health policies have not been corrective and have not diminished the disparity in income, but have, on the contrary, confirmed and reaffirmed these conditions.” However, it is difficult to see how they could have come to this conclusion without conducting even a rudimentary benefit-incidence analysis. 4. CEPAL (1994, ch. III; 2001, ch. IV), IADB (1998, ch. 8), and World Bank (2001, ch. 5) jointly report results mostly for the early half of the 1990s for 27 developing countries, 12 of them in the Latin America and Caribbean region. For comparative studies, see Yaqub (1999), Filmer and Pritchett (1998), and Scott (2002). 5. One exception is Reyes Heroles (1976), which reports incidence tables for 1968. The lack of incidence studies in this area is illustrated by the fact that studies published as recently as Lustig (1989) and Gil Díaz and Thirsk (1997) use it as a relevant reference (the only of its kind). - 18 - 2.8 What is clear is that in the aftermath of the 1968 student revolt, the allocation of public education spending shifted in favor of upper-secondary and tertiary education and against basic education (Scott 2003). Given the exceptionally regressive coverage of tertiary education in Mexico even today, the concentration curve for public spending at this level probably was well above the Lorenz curve for private income or expenditure and thus strictly unequal. The same probably was also true for health services for formal sector and state workers. What seems more difficult to imagine, however, is that social spending as a whole—including spending on basic education and health services for the uninsured, which absorbed at least half of education and health spending even in this period—could have been more unequally distributed than private income. 2.9 Since some of the conceptual confusions underlying the Aspe-Beristáin claim remain in the literature,6 they are worth pointing out: (a) regressive allocation of overall spending in education and health is inferred from inequitable budgetary allocations per beneficiary, (b) regressivity in relation to the distribution of income is inferred from regressivity in absolute terms, and (c) social security spending is considered gross of social security contributions by the recipients of benefits, when only a small fraction (5 percent) of the benefits were, in fact, subsidized by the federal government from general tax revenue (before the 1997 IMSS reform). 2.10 More recent studies have focused on the effects of the post-1982 budgetary cuts (Friedman and others 1995; Pánuco and Székely 1996) and on the distributional implications of rural development and agricultural support before and after the reforms introduced in the past decade (ejido reform, NAFTA, Procampo), including Levy and Wijnbergen (1992), Deininger and Heinegg (1995), Sadoulet and others 2001, and Dávila and others (2002). 2.11 Another relevant line of research on the distributive impact of public expenditure in recent years has evaluated the impact of the principal targeted programs in Mexico. The most important of these studies (Sedesol 2000) is the path-breaking evaluation of Progresa in its first (rural) phase (1998–2000). This has been followed by further evaluations of the urban extension of the program (as well as long-term rural impacts) currently under way. More modest evaluations have been undertaken for most other targeted programs, as evaluations have become mandatory for all such programs in Mexico since 2000. 2.12 Finally, studies of the distribution of the beneficiaries of public spending at the household level for broader expenditure programs appeared during the past decade. Castro-Leal and Dayton (1994) and López-Acevedo and Salinas (2001) analyze the distribution of public education (in 1992 and 1996, respectively). González-Pier and Parker (1999) estimate the distribution of health care financing, but not the distribution of the benefits from public health spending. Rosellón and López-Calva (2002) analyze the distribution of electricity subsidies. 2.13 Other studies focus on specific sectors of public expenditure. Scott (2003) estimates the distribution of benefits from public expenditure covering the principal programs of social expenditure (education, health, and pensions) as well as monetary transfers (Procampo and Oportunidades), food programs, and the electricity subsidy, in 2000. 6. See, for example, Corbacho and Schwartz (2002, pp. 12, 24), which quotes the Aspe-Beristáin conclusion without questioning it, drawing similar claims from more recent studies. - 19 - 2.14 This chapter uses the latest income and expenditure survey available in Mexico (2002), introducing a number of methodological refinements and taking advantage of the availability of improved survey and budgetary information (see annex A1). (a) We present a marginal benefit-incidence analysis covering the distribution of changes in benefits over the 1992–2002 decade, based on the last six editions of the household survey. (b) In addition to national population deciles, we analyze the distribution of benefits within urban and rural deciles and, in the case of monetary transfers, in terms of the official poverty lines recently adopted by the federal government in Mexico. We also analyze income distribution and gender gaps in public education. (c) We disaggregate health services by broad types of health care (primary, maternal, hospital) in addition to types of institution. (d) In the case of targeted monetary transfers (Procampo, Oportunidades), we estimate the impact of transfers on poverty measures (headcount, poverty gap, and squared gap). (e) In the case of transfers in kind, we introduce preliminary evidence relevant to a future cost-effectiveness analysis. 2.15 Benefit-incidence analysis estimates the distribution of benefits received by households from public spending. In the case of goods and services, or transfers in kind, the value of these benefits is commonly assumed to be equal to the cost of provision to the government. Their distribution among households is typically estimated from the use of services as reported in national income and expenditure surveys. The limitations of this method for policy analysis and reform have received some attention in recent years, giving rise to a number of methodological developments:7 (a) estimating the actual utility obtained by beneficiaries from the use of public services, (b) taking into account behavioral and general equilibrium effects, and (c) estimating the incidence of marginal changes in the coverage of established programs. The analysis presented here addresses the third limitation and presents evidence pertaining to the first. In the case of monetary transfers, we allow for behavioral responses by assuming a marginal propensity to consume from monetary transfers of less than 1. 2.16 We also analyze changes in the distribution of public expenditure over the past decade, using the six National Household Surveys of Income and Costs (ENIGH, Encuesta Nacional de Ingresos y Gastos de los Hogares) available for this period. Changes in distribution can be due to changes in the coverage of programs or to changes in the allocation of resources to programs. The former may reflect not only changes in the supply of programs, but also exogenous changes in their demand by households. We therefore report the evolution of public expenditure for each program and, where available, the evolution of coverage as reported by the administrative records of the programs. In addition to presenting the distribution by deciles, concentration coefficients, and shares of the poorest quintile, we summarize changes in the distribution of benefits between two points in time in two simple forms: (a) as additions (subtractions) to the share of public expenditure received by each decile and (b) as the distribution of the marginal 7. See Van de Walle and Nead (1995), Van de Walle (1998), Lanlouw and Ravallion (1999), Ajwad and Wodon (2001), Bourguignon and others (2002). - 20 - change in public expenditures. See Box 2.1 for the concepts and measures of progressivity. In this chapter we report the distribution of the marginal change in expenditures for the decade as a whole, but annex A3 includes the decomposition for intermediate years. Box 2.1. Absolute and Relative Progressivity: Concepts and Measures The concept of absolute progressivity requires transfers (taxes) to decrease (increase) with income (expenditure) in absolute terms, while the concept of relative progressivity requires transfers (taxes) to decline (increase) as a proportion of income. While the concept of relative progressivity is generally used in relation to taxes, the distribution of benefits from public expenditure is more commonly described in terms of absolute progressivity. Measures of progressivity can be derived analogously to measures of income inequality from Lorenz curves—known as concentration curves in this context—defined on the distribution of transfers (taxes) using pre-transfer (tax) income or expenditure as the relevant ordering concept. For a large set of measures (Lambert 1993), a transfer is unambiguously progressive in the absolute sense if its concentration curve lies above the diagonal and is progressive in the relative sense if its curve lies below the Lorenz curve for autonomous income or expenditure. The most common measure of absolute progressivity is the concentration coefficient (C), or quasi-Gini, which is a Gini measure derived from these curves. C is defined in the (-1, 1) interval, where, in the case of transfers, negative (positive) values represent progressive (regressive) allocations. A widely used measure of relative progressivity is Kakwani’s index (K), defined as the difference between C and the Gini coefficient for autonomous income. K is defined in the (-2, 1) interval, again with negative (positive) values corresponding to progressive (regressive) transfers. The redistributive impact of the transfer, measured by the difference between the Gini before and after the transfer, can be shown (Kakwani 1977) to be directly proportional to K and the average transfer rate (transferred resources as a proportion of autonomous household income), γ: γ ∆G = K , (1 + γ ) Redistributive efficiency (RE) can be measured as the elasticity of this impact with respect to the scale of resources absorbed by of the transfer: ∆G RE = G. γ Data and Methodological Assumptions 2.17 The data limitations—both on the use of services by households and on budgetary allocations—explain, in part, the late appearance of benefit-incidence studies for Mexico. Data have improved in recent years but still leave gaps in our knowledge. With one exception (see below, Figure 2.2.6), data reported on the use of services and transfers received by households are from ENIGH in its 1992 1994, 1996, 1998, 2000, and 2002 editions. This survey was not designed to be strictly representative for the use of services and reception of specific transfers by deciles and other partitions that we will use here (urban/rural, male/female), so some of the observed variation between years may be spurious. In addition to sampling errors, comparability - 21 - is limited by changes in the survey questionnaires over time. These will be noted in specific cases below and in the Annex. 2.18 Information on public expenditures refers to executed spending (gasto ejercido) and is obtained from various official government sources. No single, consistent source of information on national public accounts covers all the programs considered, reports information at the required levels of desegregation, or includes public spending at a level other than the federal government. In the case of education and health, we include estimated spending by state governments from own revenues and discretionary federal transfers (participaciones) as well as federal spending and earmarked federal transfers (aportaciones: FAEB and FASSA in Ramo 33). User fees are only netted out in the case of electricity services, which is the only service analyzed for which they represent a significant proportion of costs. They also exist in the case of health services for the uninsured and tertiary education, but they are generally negligible in relation to costs in both cases.8 The other mayor programs—basic education and health services for the insured--do not have legal user fees, though some might be informally charged. In the case of education there are obviously also important participation costs which are ignored here, from uniforms and school supplies to forgone income. 2.19 Given the amount of information to be presented, we use a single baseline ordering concept: population deciles (or centiles) ordered by autonomous expenditure per capita.9 Autonomous income is obtained by netting out public monetary transfers identifiable through ENIGH, which includes Procampo, Oportunidades, and other scholarships. Autonomous expenditure is obtained by netting out 50 percent of these transfers, assuming a marginal propensity to consume from government transfers of this order of magnitude.10 Given the small share of these transfers in both, aggregate household income and public expenditure, the results are essentially invariant to the fraction of transfers netted for all programs except the transfers themselves and for public expenditure as a whole. However, the transfers represent important proportions of their beneficiaries’ income.11 Given the effective targeting of Oportunidades transfers, the ordering in this case is robust at the decile level. In the case of Procampo, however, transfers significantly modify the ordering of the larger beneficiaries (in terms of transfers per household) in the upper part of the income distribution. In this case, we present the 8. The ENIGH does not distinguish between user fees and private spending on health inputs associated with a given use of public services, but acquired from private suppliers. Reported out-of-pocket health expenditures by users of services for the uninsured are not insignificant, but we assume these to involve mostly to the latter. 9. There are no agreed equivalence scales for Mexico, and estimations vary widely by method used. Scott (2003) reports four variations on unit/ordering concept: household and population, ordered by per capita and total household income. The overall distributions of public spending is quite sensitive to the ordering concept in the case of household deciles, but much less so in the case of population deciles. 10. See Van de Walle (2002). Davis and others (2002) finds the marginal propensity to consume to be 70 percent for Procampo and 40 percent for Oportunidades. Despite the difference, Davis and others (2002) cannot reject the null hypothesis that the coefficients are equal. Also the Progresa survey used in that study is only representative of the rural poor, while Procampo also benefits richer farmers, who presumably have a lower marginal propensity to consume. 11. In the case of Oportunidades this is so because they are targeted to the poor, but in the case of Procampo it is also true of non poor farmers because they receive transfers (roughly) proportional to the size of their principal asset, land. - 22 - relevant comparison, and in both cases we report an incidence analysis with income (netting 100 percent of transfers) as well as expenditure. 2.20 The survey and budgetary data and methodological assumptions used to impute benefits from specific programs to the population is described in Annex A1. Redistributive Demands 2.21 Before analyzing the supply of redistributive public expenditure, it is relevant to consider the redistributive demands motivating this effort and the instruments available to the government. Beyond standard normative arguments (and political economy explanations) for the redistributive function of the state, four factors make the case for effective redistributive expenditure especially relevant to Mexico: (a) the high level and persistence of income inequality; (b) the constraining effect of inequality on poverty, limiting both growth and the growth elasticity of poverty reduction; (c) the structure of income inequality; and (d) the economic determinants of income inequality. We consider these in turn. 2.22 First, income inequality in Mexico is high by international standards and deepened during the past two decades.12 This general trend was interrupted only during the economic recession following the 1995 crisis and stagnation following the recent downturn in the U.S. economy. Although inequality in the distribution of autonomous income and expenditure declined between 2000 and 2002 (Table 2.1), the overall pattern suggests that income inequality may increase again when the Mexican economy finally achieves a path of sustained growth. Table 2.1: Distribution of Autonomous Income and Expenditure Income Expenditure Deciles 2000 2002 2000 2002 1 1.1% 1.3% 1.3% 1.5% 2 2.2% 2.5% 2.5% 2.7% 3 3.2% 3.4% 3.4% 3.6% 4 4.1% 4.5% 4.4% 4.7% 5 5.3% 5.6% 5.5% 5.8% 6 6.7% 7.0% 6.8% 7.0% 7 8.5% 8.6% 8.5% 8.7% 8 10.9% 11.3% 10.9% 11.2% 9 15.8% 16.1% 15.5% 15.7% 10 42.2% 39.7% 41.2% 39.2% Urban 90.3% 87.8% 89.8% 86.9% Rural 9.7% 12.2% 10.2% 13.1% Gini 0.519 0.494 0.500 0.481 Source: ENIGH (2000, 2002). Population deciles ordered by same welfare concept (income/expenditure), per capita. 12. Mexico presents the thirteenth most unequal distribution among 111 countries reported in World Bank (2003a). - 23 - 2.23 Inequality in Mexico persists not only in the aggregate but also for households over time and even for families over generations. If poverty usually arose from bad luck, with a good chance of rebounding, or if the children of the poor typically had a good chance to rise in the next generation, then poverty at a point in time would be easier to bear, and the state could remain less active, just facilitating the autonomous equalizing effects for the private economy. But this has not occurred in Mexico, or in the rest of Latin America, where social mobility is limited by the distribution of assets, including human capital (World Bank 2003). Failures of both the public sector and the market economy in Mexico lead poverty to persist over time, which calls for effective policy measures to reduce poverty. 2.24 Second, the increasing inequality, combined with the low growth rates achieved over the past two decades, has demonstrated the limitations of a purely growth-based strategy of poverty reduction in Mexico.13 Over the 1984–2000 period, the impact of growth in per capita income (around 1 percent a year) was almost exactly cancelled out for the bottom third of the income distribution by the concurrent loss in income share (0.9 percent a year on average), implying a lost decade and a half for the poor. 2.25 Third, as is evident in Figure 2.1, the excess income inequality that sets Mexico—and most of the Latin America and Caribbean region—apart form the rest of the world is concentrated in the top decile. While the share in national income of the bottom decile in Mexico is close to 1 percent of national income, the share of the top decile is 40 percent and the share of the top centile is 10 percent. If we consider only the poorest 90 percent of the distribution, Mexico and the United States present similar levels of inequality (IADB 1998). Figure 2.1: Distribution of Autonomous Income, 2002 250 240 Rural 230 220 Urban 210 200 190 180 170 160 thousand pesos 150 140 130 120 110 100 90 80 70 60 50 40 30 20 10 - 1 5 9 13 17 21 25 29 33 37 41 45 49 53 57 61 65 69 73 77 81 85 89 93 97 population centil Note: Population centiles are ordered by per capita income (reordering within each sector). Source: ENIGH (2002). 13. For the general case in the Latin America and Caribbean region, see World Bank (2003b). - 24 - 2.26 This has two important implications. On the one hand, the concentration of income, the underlying concentration of assets, and the consequent concentration of political power may impose a significant structural constraint on the country’s capacity to sustain growth by limiting the pool of financial and human resources available to fuel economic growth. Such concentration limits, in particular, the national savings capacity, the pool of skilled human capital, and the capacity of the government to extract resources from the economy to finance public goods and redistributive expenditure. On the other hand, this concentration of autonomous income implies that public expenditure has an exceptionally high potential redistributive capacity (in comparison to less-concentrated distributions and to tax instruments), even if spending is not targeted with respect to household income. 2.27 Fourth, changes in the distribution of autonomous income may be analyzed, for explanatory as well as policy purposes, in terms of three factors: (a) the distribution of productive assets, (b) the economic returns to those assets, and (c) their rate of use by households. In Mexico, education is the principal (identifiable) asset explaining the evolution of earnings inequality in the past two decades.14 This, and the large share of public spending that goes for education, make it a sector of special concern in this report. 2.28 To understand the changes over time, Table 2.2 presents the evolution of poverty rates from 1992 to 2002 for the lower (“food”) poverty line officially recognized by the Government of Mexico. See the Mexico Poverty Assessment for a full discussion. During the decade, the rural population declined slowly, infant and basic school-age population began to decline in the middle of the decade, and the share of older population grew—trends that affect the potential demand of social services and transfers. See appendix tables A1.1 and A1.2. Table 2.2: Poverty Rates with Food Poverty Line Year National Urban Rural 1992 22.5 13.5 35.6 1994 21.1 9.7 36.8 1996 37.1 26.5 52.4 1998 33.9 21.3 52.1 2000 24.2 12.6 42.4 2002 20.3 11.4 34.8 Source: Cortez and others (2002); Sedesol (2003). Redistributive Instruments 2.29 Governments in mixed economies may affect the distribution of income through two channels (and time horizons): (a) by modifying the relation between autonomous and (current) disposable income and (b) by modifying the conditions determining the autonomous income of households. 2.30 The redistributive instruments available to governments in this regard may be classified as (a) monetary instruments, including taxes and direct monetary transfers, and (b) transfers in 14. Legovini and others (2001) find that these factors account for 41 percent of the increase in inequality in this period. - 25 - kind, involving the public financing and (commonly) provision of specific goods, mostly education and health services. 2.31 Although these types of instruments correspond closely with the redistributive strategies noted, the two classifications do not map perfectly one another. Monetary transfers redistribute current income, while the provision of public health and education services promotes equitable access to human capital and thus the future capacity to generate autonomous income (in addition to educational and health achievements for their own sake). However, monetary instruments may also be used to promote equitable access and improve capacity if they are conditioned on relevant investments on the part of households. Similarly, the public provision of education and health services has an impact on the distribution of current disposable income by liberating private household expenditure for other ends. 2.32 Both types of instruments play important redistributive roles in Mexico. In mature welfare states, dramatic reductions in the inequality of disposable income are achieved through massive monetary transfers. In Mexico, such transfers represent a small proportion of redistributive expenditure, barely modifying the overall distribution of income. However, their exceptionally effective targeting mechanisms, together with their conditionality on the use of basic education and health services, imply important redistributive impacts on the rural poor. Moreover, expenditures on education and health represent the government’s principal instrument for redistributing income. THE DISTRIBUTION OF BENEFITS: 1992–2002 2.33 This section analyzes the distribution of public expenditures in services and transfers for 2002 for national population deciles and for urban and rural quintiles. For the principal expenditure programs, it also presents comparable results for the previous decade (1992–2002) and estimates the distribution of marginal benefits due to changes in coverage and budgetary allocations. Finally, it compares the distribution of overall benefits from public expenditures with the distribution of tax burdens. Public Expenditure on Redistributive Programs 2.34 Table 2.3 reports executed expenditure in 2000 and 2002 for the programs analyzed in this chapter. The spending for which we have incidence data has a value equal to 52 percent of total central government spending and 9.5 percent of GDP; the spending analyzed for incidence also includes some spending outside the central budget, by public enterprises and by subnational governments from their own resources. 2.35 The bulk of these resources were used for the provision of education and health services (77.4 percent). Within these sectors, expenditures were increased proportionally more for upper- secondary than for other levels of education and more for health services for the insured population than for the uninsured. 2.36 Although pensions absorbed another 11.4 percent and, after the Oportunidades Program, grew the most between 2000 and 2002, these resources corresponded mostly to transitional costs of the IMSS reform; they are expected to start contracting and eventually disappear. If a comparable pension reform is implemented for ISSSTE, the expenditure financing ISSSTE’s - 26 - current deficit will also become transitional, although only after a prolonged transitional phase with a steep increase in the near future. Also complicating the analysis, beneficiaries made substantial contributions to the pension funds in the past. In our estimate of the impact of public expenditure, we therefore exclude public expenditure on pensions, except for current transfers to the individual pension accounts of workers under the new law. Table 2.3: Cost of Public Redistributive Programs (millions of 2002 pesos) Distribution of 2000 % 2002 % Change additional resources Total 539,456 100% 587,124 100% 8.8% 47,669 % GDP 8.8% 9.5% Education a 276,100 51.2% 298,404 50.8% 8.1% 46.8% Preschool 26,048 4.8% 29,076 5.0% 11.6% 6.4% Primary 101,445 18.8% 103,752 17.7% 2.3% 4.8% Lower secondary 56,131 10.4% 60,563 10.3% 7.9% 9.3% Upper secondary 37,964 7.0% 44,962 7.7% 18.4% 14.7% Tertiary 54,511 10.1% 60,051 10.2% 10.2% 11.6% Health a 147,331 27.3% 155,946 26.6% 5.8% 18.1% Uninsured (SSA) 52,957 9.8% 55,435 9.4% 4.7% 5.2% Insured (IMSS) 75,437 14.0% 79,737 13.6% 5.7% 9.0% Insured (ISSSTE) 14,024 2.6% 15,101 2.6% 7.7% 2.3% Insured (Pemex) 4,913 0.9% 5,672 1.0% 15.4% 1.6% Pensions 54,613 10.1% 66,714 11.4% 22.2% 25.4% IMSS active workers (1995 11,950 2.2% 12,380 2.1% 3.6% 0.9% Law) IMSS pensioners 32,476 6.0% 39,606 6.7% 22.0% 15.0% (1973 Law) ISSSTE pensioners 10,188 1.9% 14,728 2.5% 44.6% 9.5% Direct transfers 22,306 4.1% 28,854 4.9% 27.9% 12.9% Oportunidades 10,265 2.0% 16,105 2.9% 56.9% 12.4% Procampo 11,595 2.1% 11,851 2.0% 2.2% 0.5% Electricity residential 39,106 7.2% 37,206 6.3% -4.9% -4.0% subsidy b a Includes state resources from un-earmarked sources, as well as budgeted federal transfers (mostly Ramo 33) that are earmarked for education and health spending at the state level. b Not included in Federal budgeted spending. Source: SEP (DGPPP); SSA (Boletín de Información Estadística, Cuentas Nacionales y Estatales de Salud), IMSS (Coordinación de Presupuesto, Contabilidad y Evaluación Financiera); Poder Ejecutivo Federal (2003); Cuenta de la Hacienda Pública Federal (2002). 2.37 It is remarkable that Oportunidades, the principal antipoverty program of the previous administration (when it was known as Progress), not only survived the change of administration, the change of the party in power, and the change of the regime itself but was significantly expanded, doubling the number of beneficiaries between 2000 and 2002. 2.38 The only expenditure item that contracted in real terms between 2000 and 2002 was the electricity subsidy for residential consumers, which remained regressive but was significantly reduced for high-use consumers at the very top end of the distribution. - 27 - 2.39 Unless otherwise specified, the terms progressive and regressive refer to absolute levels of benefits. Any reference to benefits proportional to income is made explicit. Education 2.40 Returns to education were polarized in the past two decades, declining for basic education (as it became nearly universal) and increasing for higher education. This happened in many other countries as well, partly as a result of the rising demand for skilled labor in the increasingly globalized markets. The polarizing impact of increasing returns to education could have been compensated or even overshadowed by equalizing trends in either of the two other relevant variables—the distribution of educational assets and participation rates—and this did occur in other countries where income inequality decreased in recent decades, including Brazil.15 In Mexico, however, the marginal distribution of expanding education aggravated the polarizing price effect in 1984–94. This happened despite a slow reduction in schooling inequality (Tables 2.4 A2.1, and A2.2). The largest gains in schooling were achieved in the middle of the distribution in 1984–94, and the gain in years of schooling for the top quintile, though comparable to that achieved at the bottom quintile, entailed considerably larger gains in income, because the extra years of schooling for the rich were in higher education. Table 2.4: Average Schooling of Adult Population (25–65 years) Income Deciles 1984 1992 2000 2002 1 2.06 2.17 2.84 3.17 2 2.30 3.04 3.71 4.18 3 3.00 3.59 4.96 4.99 4 2.97 4.17 5.63 5.62 5 3.62 4.77 6.24 6.45 6 4.14 5.26 6.85 6.83 7 4.96 6.04 7.35 7.43 8 6.18 6.79 8.28 8.16 9 7.24 8.24 9.53 9.78 10 8.55 10.76 12.08 12.02 Absolute gap 6.49 8.59 9.24 8.85 1–10 Concentration 0.3453 0.3390 0.3001 0.2942 coefficient Source: ENIGH (1984, 1992, 2000, 2002). Population deciles are ordered by per capita income. Education Services 2.41 Table 2.5 presents the distribution of public education services by level of education and the distribution of total education expenditure. At the national level, coverage is highly progressive at the preschool and primary levels, close to neutral in lower-secondary levels, and moderately and highly regressive in upper-secondary and tertiary education, respectively. The distribution of total public education expenditure is almost exactly proportional. 15. See Kanbur and Lustig (1999) for a comparison of Mexico, Brazil, and Taiwan (China) in this regard. - 28 - Table 2.5: Distribution of Public Expenditure on Education Services, 2002 Presch Lower Upper Deciles Total Primary Tertiary ool secondary secondary 1 9.6% 17.8% 15.6% 9.4% 3.5% 0.0% 2 9.3% 12.6% 14.6% 10.3% 4.9% 0.7% 3 10.4% 13.7% 13.8% 13.0% 7.5% 2.4% 4 9.8% 13.0% 11.9% 10.6% 8.9% 4.5% 5 9.7% 10.3% 10.6% 10.2% 10.4% 6.7% 6 9.3% 9.2% 8.9% 10.7% 10.2% 8.2% 7 10.7% 8.8% 8.7% 11.5% 14.6% 11.1% 8 12.3% 7.2% 7.6% 10.6% 16.0% 21.9% 9 11.0% 5.0% 5.7% 9.7% 14.0% 22.1% 10 8.0% 2.3% 2.5% 4.2% 10.1% 22.3% Urban 78% 66.1% 66.9% 74.2% 87.6% 97.7% Rural 22% 33.9% 33.1% 25.8% 12.4% 2.3% Concentration coefficient National 0.009 -0.238 -0.222 -0.061 0.183 0.469 Urban 0.007 -0.250 -0.244 -0.096 0.074 0.286 Rural -0.061 -0.167 -0.121 -0.006 0.163 0.648 Source: ENIGH (2002) and table 2.3. Population deciles are ordered by per capita expenditure. 2.42 Rural households have a disproportionately large share of students in public basic education (in relation to a population share of 23.6 percent), but a small share of students in the upper-secondary and tertiary levels. Total education expenditure per capita is biased toward urban areas. 2.43 The distribution of upper-secondary and tertiary education is significantly less regressive within the urban population than nationally, implying that most of the regressivity observed at these levels is due to rural-urban inequality. However, the distribution of basic education is less progressive within the rural sector than nationally. The coverage of upper-secondary education within the rural sector appears to be less regressive than one would expect, and the participation of the first and fourth deciles is surprisingly high in relation to that of previous years.16 This may reflect the extension of Oportunidades scholarships to this educational level in 2001, but since ENIGH is not designed to be representative at this level of disaggregation within the rural sector, it could also be a statistical fluke and should be interpreted with caution. 2.44 In part, the distribution of the use of public education services may simply reflect the distribution of potential demand implied by the demographic composition of different income groups (Figures 2.2., A2.1, and A2.2). Comparing this distribution with the distribution of the relevant age groups reveals that this is true for primary education, where the two distributions are very similar. The bars represent the percentage of each age group in each income decile. If an age group were equally distributed across income deciles, then all the bars would be at 10 percent, which is approximately the case for young adults 16–18 years old. Households in the 16. Comparing 2002 with 2000, the participation of the first rural quintile in upper-secondary education increased from 7 to 11 percent, and the concentration coefficient declined from 0.38 to 0.16. - 29 - lower-income deciles have well over their proportional share (10 percent each) of children 6–12 and 13–15 years old, and therefore upper-income deciles have less than 10 percent. Young adults 19–24 years old (postsecondary age) are more prevalent in the upper-income deciles. The lines show each age group’s share of enrollment. To the extent that the line is below the bar for a certain age group, those children are not in public school at that level—either because they are not in school at all, like poor children over age 13, or because they are in private schools, like many rich children. Figure 2.2: Distribution of Public Education Services and Relevant Age Groups, National, 2002 18% 6-12 13-15 16% Primary Lower-Secondary 14% 12% 10% 8% 6% 4% 2% 0% 1 2 3 4 5 6 7 8 9 10 25% 16-18 19-24 20% Upper Secondary Tertiary 15% 10% 5% 0% 1 2 3 4 5 6 7 8 9 10 Source: ENIGH 2002 and World Bank calculations. - 30 - 2.45 In the case of primary education, only a small part of the measured progressivity of participation in education reflects effective demand, as households in the poorer deciles take advantage of public education disproportionately to their participation in the relevant age group, while richer households often opt for private services. Most of the measured progressivity is due to demographic structure. 2.46 The demographic determinant becomes less important for lower-secondary education and especially for higher levels of education. At these levels the distribution of the use of public services reflects limits in the coverage (or quality) of services, the failure of previous schooling to prepare students for the next level, and the indirect private costs of school attendance, of which the most important cost is the labor-market opportunity cost for students. 2.47 The distribution of potential demand thus measured also offers a useful minimum target of progressivity, representing the distribution required to ensure at least equal access for equal need, as measured by the relevant age group. 2.48 Over the past decade, enrollment grew proportionally the most in upper-secondary and tertiary education, remaining almost stagnant at the primary level (with close to full coverage for a declining age group). See Table 2.5. Public education expenditure increased well beyond what was required to sustain the expansion in coverage, except in tertiary education. Therefore, expenditure per student increased significantly, especially at the preschool and primary levels. 2.49 The expansion of enrollment and public expenditure in basic education was clearly pro- poor (Table 2.6 and Figures 2.3 and A2.3–A2.5). The small gain in coverage in primary education was concentrated in the first quintile, but the most striking change occurred in lower- secondary education, where the deep gap in participation of the poorest deciles was completely covered in the course of the decade (the share of the poorest quintile increased from 8 to 20 percent). Thus a regressive distribution became progressive. The large gain between 1998 and 2000 may have been due to the cash incentive that the Oportunidades Program gave for attendance. However, the share of the poorest quintile and of the rural population rose steadily from 1992 onward (Figures 2.3 and A2.5), which suggests that the program reinforced a trend that was already in place and that may have been the result of the pre-1992 gains in the coverage of primary education. 2.50 Significant gains in progressivity also occurred in higher education, although these are less apparent from the decile distributions than from the concentration coefficients. In the case of upper-secondary education, the gain was especially significant between 2000 and 2002, again consistent with both a direct effect of the Oportunidades Program (which was extended to this level in 2001) and an indirect effect of the antecedent increase in the coverage of secondary education. The gain in progressivity in the coverage of tertiary education was sufficient for public expenditure at this level to become (barely) more progressive than private expenditure. This gain reflects increases in participation between the third and eighth deciles, at the expense of the top, while the expansion in coverage completely failed to reach the poorest quintile. - 31 - Table 2.6: Evolution of Public Expenditure on Education Services, 1992–2002 Lower Upper Year Total Preschool Primary Tertiary secondary secondary ENROLLMENT (million students) 1992 22.67 2.61 13.50 3.86 1.66 1.03 1994 23.57 2.84 13.65 4.16 1.83 1.10 1996 24.48 2.99 13.72 4.47 2.09 1.20 1998 24.95 3.06 13.66 4.69 2.22 1.31 2000 25.36 3.08 13.65 4.93 2.31 1.39 2002 26.22 3.27 13.67 5.21 2.58 1.49 Growth 1992– 15.7% 25.1% 1.2% 35.0% 55.3% 44.8% 2002 PUBLIC EXPENDITURE (million pesos, base=2002) 1992 168,346 13,790 63,027 32,488 20,463 38,578 1994 237,516 18,989 85,500 47,201 35,410 50,417 1996 228,680 20,987 87,605 41,543 33,268 45,278 1998 259,592 24,370 97,669 53,419 34,900 49,232 2000 276,100 26,048 101,445 56,131 37,964 54,511 2002 298,404 29,076 103,752 60,563 44,962 60,051 Growth 77.3% 110.8% 64.6% 86.4% 119.7% 55.7% 1992–2002 PUBLIC EXPENDITURE PER STUDENT (pesos) 1992 7,427 5,279 4,668 8,417 12,297 37,498 1994 10,075 6,686 6,263 11,358 19,324 46,013 1996 9,342 7,026 6,383 9,283 15,892 37,709 1998 10,406 7,967 7,148 11,378 15,726 37,601 2000 10,889 8,456 7,433 11,391 16,422 39,245 2002 11,381 8,900 7,592 11,622 17,401 40,300 Growth 53.3% 68.6% 62.6% 38.1% 41.5% 7.5% 1992–2002 Source: SEP 200. SHCP, and World Bank calculations. 2.51 Public education expenditure overall moved over the decade from moderately regressive and biased toward urban areas to neutral in both terms. 2.52 Finally, we consider whether significant and systematic gender biases are evident in public education. This is of particular relevance for the Oportunidades Program, which was explicitly designed to overcome such biases in basic education by offering higher scholarships to female students. Parker and Pederzini (2001) find that “the overall gender gap in education in Mexico has diminished to the extent that there are no apparent gender differences in years of completed schooling for individuals under age 25,” although modest gaps remain in the rural sector. Studies on the educational impact of Oportunidades find significantly larger impacts on the enrollment of girls than boys as well as lower repetition rates and higher academic performance (Parker 2003). - 32 - Figure 2.3: Evolution of the Distribution of Public Education Expenditure, Concentration Coefficients and Share of Poorest 20% 0.600 Concentration Coefficients 0.500 1992 1994 0.400 1996 1998 0.300 2000 2002 0.200 0.100 0.000 -0.100 -0.200 -0.300 Total Preschool Primary Lower Upper Tertiary GINI Secondary Secondary 40% Share of poorest 20% 35% 1992 1994 30% 1996 1998 25% 2000 2002 20% 15% 10% 5% 0% Total Preschool Primary Lower Upper Tertiary Secondary Secondary Source: World Bank calculations. 2.53 Figures A2.6 and A2.7 in annex A2 compare the distribution of enrollment in public education for girls and boys over the 1992 to 2002 period. At present there appears to be a clear pro-male bias at the lower end of the distribution only at the preschool level, as the considerable gains in participation achieved over the decade benefited boys disproportionately. In the case of primary education, enrollment of girls in the lower half of the distribution (especially in the first - 33 - decile) lagged behind that of boys in 1992, but most of this gap was closed by 2002. In the case of lower-secondary education, both the 1992 distribution and the gains achieved by 2002 were similar for girls and boys. In the case of upper-secondary and tertiary education, a significant gap in female enrollment among the poor in 1992 was again largely overcome by 2002. At both levels, participation was pro-male in the upper part of the distribution, which may reflect higher female participation in private education as well as lower repetition rates among females. Scholarships 2.54 Scholarship programs are an important demand-side instrument for increasing the participation of lower-income groups in public (and indeed private) education services. The Oportunidades Program is by far the largest public scholarship program in Mexico, in terms of coverage as well as budget. Table 2.7 compares the distribution of beneficiaries from Oportunidades scholarships and other public scholarship programs reported in ENIGH, showing that the Oportunidades Program is highly progressive, while other programs are regressive. In both cases, progressivity is significantly lower (or regressivity higher) among rural households than among urban households. Table 2.7: Distribution of Households Benefiting from Public Scholarship Programs, 2002 Oportunidades Program Other Deciles National Urban Rural National Urban Rural 1 39.5% 52.8% 13.1% 4.1% 4.8% 4.1% 2 23.8% 18.5% 17.1% 6.7% 7.2% 3.9% 3 15.4% 11.7% 12.0% 9.2% 7.5% 6.3% 4 9.4% 4.0% 11.2% 6.3% 7.2% 9.9% 5 5.3% 3.9% 14.0% 11.2% 11.8% 5.5% 6 2.8% 4.8% 9.1% 12.9% 4.9% 10.5% 7 2.2% 3.1% 10.4% 4.9% 25.5% 17.5% 8 1.3% 1.0% 7.4% 22.1% 10.5% 13.1% 9 0.2% 0.1% 4.1% 12.4% 9.1% 16.7% 10 0.1% 0.1% 1.6% 10.1% 11.6% 12.4% Concentration coefficient National -0.615 -0.658 -0.225 0.155 0.138 0.226 Urban 19.2% 76.4% Rural 80.8% 23.6% Beneficiary 3,120,92 747,352 households 9 Note: The distributions refer to beneficiary households, not the monetary value of transfers received. In the case of Oportunidades, the distribution of monetary transfers is reported in table 2.9. Source: ENIGH (2002). Population deciles are ordered by per capita expenditure. Health 2.55 The distribution of benefits from public health services varies according to institution, with SSA serving the uninsured, IMSS serving formal sector workers, and ISSSTE and Pemex - 34 - serving public sector workers. The distribution of benefits also depends on the type of attention—primary health care, maternal health care, and hospitalization (see Tables 2.9 and A2.4). Given that an important part of IMSS health services is financed through contributions from the beneficiaries (and their employees), we also report the total distribution of expenditure considering only federal contributions to IMSS health services (see Table 2.8). Table 2.8: Evolution of Federal Expenditure on Health Services, 1996–2002 Uninsured Insured SSA IMSS Federal % of Federal contribution Year Total Total ISSTE total expenditure financed by general taxes 1996 101,762 21.5% 21,907 65,015 14,840 1998 129,876 27.8% 36,124 72,249 24,842 21,503 2000 133,136 32.8% 43,675 75,437 29,187 14,024 2002 142,451 33.4% 47,612 79,737 29,048 15,101 Growth 40.0% 117.3% 22.6% 1.8% 1996-2002 *Health expenditure as reported here differs from what is reported in table 2.3 because it does not include contributions by state governments. Table 2.9: Distribution of Public Expenditure on Health Services, 2002 By type of attention By institution Total (all public expenditure) Uninsured Insured Federal All public Decile contribution Primary Maternal Hospital SSA IMSS ISSSTE Pemex expenditure to IMSS 1 7.7% 9.6% 8.4% 4.8% 1.9% 15.9% 3.7% 0.2% 2.8% 2 7.4% 9.0% 8.0% 5.0% 1.2% 14.1% 4.2% 1.4% 2.8% 3 9.5% 11.2% 9.5% 13.0% 7.3% 16.9% 6.0% 3.2% 3.9% 4 8.6% 8.9% 8.9% 6.2% 7.6% 10.7% 8.0% 4.7% 7.1% 5 10.7% 10.8% 10.2% 10.5% 15.3% 11.1% 10.5% 6.4% 21.8% 6 10.7% 9.9% 10.4% 12.5% 16.2% 9.0% 12.5% 8.8% 7.8% 7 10.7% 9.9% 10.7% 15.7% 8.6% 7.6% 12.5% 15.3% 3.9% 8 12.2% 11.0% 12.1% 14.8% 12.3% 6.4% 14.7% 19.5% 15.1% 9 12.3% 11.2% 12.0% 12.7% 11.3% 6.3% 14.8% 19.5% 18.3% 10 10.1% 8.6% 9.9% 4.8% 18.3% 2.1% 13.2% 21.0% 16.5% Urban 70.9% 65.6% 69.3% 80.6% 77.4% 47.9% 82.0% 93.4% 81% Rural 29.1% 34.4% 30.7% 19.4% 22.6% 52.1% 18.0% 6.6% 19% Concentration coefficient National 0.076 0.008 0.060 0.093 0.247 -0.243 0.218 0.430 0.264 Urban 0.038 -0.033 0.088 0.036 0.213 -0.207 0.156 0.320 Rural 0.136 0.084 0.158 0.167 0.467 0.058 0.342 0.770 Source: ENIGH (2002) and table 2.3. Population deciles are ordered by per capita expenditure. 2.56 Public financing of health services for the uninsured is highly progressive, while public financing of health services for the insured population is regressive in absolute terms (but not - 35 - relative to income), especially in the case of state workers. The distribution of total public expenditure in health is mildly regressive but becomes neutral if we net out private contributions to IMSS. Public expenditure on primary care is also neutral, in contrast to hospital services, which are highly regressive. The first quintile benefits only from 3 percent of the public resources devoted to hospital services, while the top decile benefits from 18 percent of them. More surprising, perhaps, the poorest quintile only benefits from 10 percent of public expenditure on maternal health. Comparing the distribution of expenditure on maternal health with potential demand, proxied by the distribution of infants and young children (0–5 years of age), Figure 2.4 shows that the distribution of total public expenditure on maternal health in 2002 was exactly inverse to the distribution of demand in the first four quintiles; the first quintile was under-covered for this critical aspect of health, even for the uninsured. The contrast in progressivity between the uninsured and insured health sectors reflects similar contrasts within each of the three types of services (table A2.5a). The progressivity of the former services could be overestimated (and the regressivity of the latter underestimated), as the present analysis does not take into account the differences in costs between primary health care (which is pro poor), and maternal and hospital services (which are not), though preliminary evidence suggests this would not alter the results significantly.17 On the other hand, the gap in progressivity between the two sectors would be even larger if we took into account the distribution of resources between these types of services, as the share of resources allocated to primary health care is 46 percent in the case of SSA, but 26 percent in IMSS.18 Figure 2.4: Distribution of Use of Public Services for Maternal Health and Infant Population, National, 2002 35% 0-5 SSA 30% Total 25% 20% 15% 10% 5% 0% 1 2 3 4 5 All public expenditure. 2.57 Overall public expenditure on health is pro-rural, especially if we net out IMSS contributions by beneficiaries, but this is not true for hospital and maternal health care (which is 17. Lacking costs information on specific services in the public sector, this can be approximated by using the implicit prices revealed by private health spending in the ENIGH. Applying this method, Scott (2003) finds that the progressivity (regressivity) of public health spending for the uninsured (insured) is reduced (increased) only marginally (from a CC of -.36 to -.30). 18. Boletín de Información Estadística, Cuentas Nacionales y Estatales de Salud, SSA, 2003. - 36 - surprisingly pro-urban). Services for the uninsured are strongly pro-rural, and the reverse is true for the insured. Finally, the steep national progressivity of health services disappears completely when we consider their distribution within the rural population, where the first decile only accounts for 5 percent of users. The distribution of services for the insured is also more regressive within this sector, as is total public expenditure: the poorest decile gets 5 percent, while the richest obtains 15 percent. 2.58 For health services, the expansion of public expenditure between 1996 and 200219 favored services for the uninsured over the insured, narrowing the gap in public expenditure between these two parts of the public health system. The share of public health expenditure allocated to the uninsured, representing close to half of the population, increased from 21 to 47 percent in only five years. If we consider only federal contributions to IMSS health insurance (netting out contributions from beneficiaries and their employers), federal spending in the two systems became broadly balanced after 2000. 2.59 The gains in Ministry of Health coverage show a pattern similar to that of lower- secondary education, with a striking improvement in access for the poorest quintile (Figures 2.5 and A2.6). This suggests again the effect of the Oportunidades Program—which conditions the distribution of “food” transfers on the use of health services—along with complementary programs that extend coverage. Between 1996 and 2000 the poorest quintile increased its participation in SSA services from 18 to 31 percent, but no further gain was achieved between 2000 and 2002. Figure 2.5: Distribution of Public Health Expenditure, 1996–2002, Concentration Coefficients and Share of Poorest 20 Percent 0.6 Concentration Coefficients 1996 0.5 1998 0.4 2000 2002 0.3 0.2 0.1 0.0 -0.1 -0.2 -0.3 Total SSA IMSS ISSSTE PEMEX GINI 19. Data on the use of health services in ENIGH 1992 and 1994 are not comparable with the more recent surveys. - 37 - 2.60 The insured also experienced some gains in progressivity, which implies that services for state workers only became progressive relative to private expenditure in the past few years. In the case of IMSS, there was no observable gain in overall progressivity or in the share of the poorest quintile between 1996 and 2000, but a small gain between 2000 and 2002 reflected a significant expansion of coverage into the rural sector (increasing its share from 11 to 18 percent). 2.61 Overall, public expenditure on health services became more progressive, pro-rural (from a share of 20 to 28 percent), and pro-poor (from 8 to 15 percent for the first quintile). Despite the improved access of the poor to basic services, there remain two important gaps: for those in remote rural areas without physical access to clinics and for those with major medical emergencies. Almost five percent of Mexican households pay over 30 percent of their income out of pocket for medical services (standard definition of catastrophic health-care spending), some to the public facilities and some to the private sector, such as for medicines. The share is about 7 percent in the poorest quintile—those too poor to afford adequate nutrition (Secretaría de Salud, 2002). Pensions 2.62 The limited, largely urban, and regressive coverage of the two principal social security institutes—IMSS with 14 million right holders in the formal private sector and ISSSTE with 2.3 million government workers—makes the distribution of expenditure on pensions highly regressive, even assuming that it is approximated by the distribution of beneficiaries (rather than the distribution of income from pensions, which is even more concentrated at upper income levels). A major reform of IMSS in 1995 set up individual accounts and contribution-defined pensions for all new employees in the formal sector; the government expenditure also contributes to the accounts and pays the part (temporarily growing) of pensions outlays under the old system that exceeds the contributions by workers still on the old system. Table 2.10 presents the distribution of beneficiaries from public expenditure on pensions, distinguishing between future pensions for active workers and current pensions for retirees. 2.63 The ISSSTE right holders still working have higher average incomes than the IMSS right holders, and the ISSSTE pensioners have even higher incomes. The distribution of ISSSTE benefits is the most regressive of all programs for which we have household-level data.20 Its concentration coefficient is well above the Gini of autonomous income or expenditure, so resources allocated to the ISSSTE pension system actually increase inequality in Mexico. ISSSTE is different, however, since it is part of the compensation of government employees and is not a public-service program with autonomous objectives. It is known that Mexico’s public sector salaries are high by international standards, at least for upper officials (Guerrero and Carrillo 2003), but the net advantage or disadvantage of this is unclear and beyond the scope of the PER. 20. For PEMEX and the military schemes jointly considered, beneficiaries are reported in the ENIGH (for active workers). For 2002, the latter present a highly regressive distribution, similar to that of ISSSTE. See Annex figure A2.9. - 38 - Table 2.10: Distribution of Beneficiaries of Public Expenditure on Pensions, 2002 IMSS ISSSTE Active Active workers Pensioners Pensioners workers Pensioned Pensioned Decile Total AFORE Right holder Right holder right holder right holder* 1 0.9% 0.5% 0.9% 1.4% 0.2% 0.0% 2 1.7% 1.9% 3.3% 2.2% 1.2% 0.0% 3 4.7% 3.9% 6.2% 5.4% 1.9% 3.2% 4 6.7% 5.9% 7.8% 8.2% 4.7% 3.1% 5 7.2% 8.5% 11.1% 8.9% 4.9% 1.6% 6 9.6% 10.4% 11.3% 10.3% 7.8% 6.7% 7 11.7% 12.3% 12.6% 12.9% 11.5% 8.1% 8 15.8% 15.4% 14.1% 15.5% 14.9% 16.8% 9 17.9% 18.4% 15.4% 17.1% 23.6% 19.8% 10 23.9% 22.9% 17.3% 18.0% 29.2% 40.6% Urban 95.0% 93.4% 93.4% 94.5% 91.4% 97.9% Rural 5.0% 6.6% 6.6% 5.5% 8.6% 2.1% Concentration coefficients National 0.393 0.396 0.286 0.319 0.504 0.592 Urban 0.298 0.294 0.187 0.216 0.419 0.512 Rural 0.461 0.595 0.436 0.430 0.698 * Number of observations of ISSSTE pensioners in the rural sector are too few to obtain representative distribution. Source: ENIGH (2002) and table 2.3. Population deciles ordered by per capita expenditure. 2.64 Federal transfers to finance pensions of formal and public sector workers increased greatly over the past decade (see Tables 2.11 and 2.12). In the case of IMSS, most of this represented transitional costs of the 1997 reform, as the government was paying current pensions (under the old law) without receiving contributions from current workers. The rest were federal contributions to the individualized pension funds of current workers under the new law. In the case of ISSSTE, they represented federal transfers to finance the growing gap between current contributions and pension payments. 2.65 These transfers were highly regressive in absolute terms and, in the case of ISSSTE, became more so over the decade (see appendix Figure A2.8 and Table A2.8). Even relative to income, the ISSSTE transfers were regressive. In the case of IMSS progressivity improved slightly between 2000 and 2002, as coverage expanded in the lower half of the income distribution and contracted in the upper half. - 39 - Table 2.11: Federal Transfers to IMSS Pensions (million MxP, base = 2002) Pensioners Active workers Seguro de Current Social quota Year Total Invalidez y pensions AFOREs Vida (SIV) 1997 21,012 20,251 0 761 1998 36,353 26,895 8,690 768 1999 42,489 31,707 10,073 709 2000 44,425 32,476 11,181 769 2001 46,409 33,804 11,766 839 2002 51,986 39,606 11,511 869 Growth 147.4% 95.6% 14.2% 1997–2002 Source: Poder Ejecutivo Federal (2003). Table 2.12: Deficit ISSSTE Pension Fund (million MxP, base=2002) Year Deficit 1992 3,377 1994 4,929 1996 5,638 1998 7,690 2000 10,188 2002 14,728 Growth 1992–2002 336.1% 2.66 Although, in the short run, the IMSS pension system transfers resources to the upper levels of the income distribution, with richer individuals receiving larger pensions, in the long run, the reform will lower the overall pension burden, freeing up resources to be used for other social programs. Furthermore, the design of the new system is meant to be redistributive and to provide incentives for workers to contribute. If and when coverage expands, the remaining small expenditures will be reasonably targeted. However, if coverage never expands, the redistribution will occur, but only within the covered group. In the long run, two-thirds of the government expenditure will come from the minimum pension guarantee, which is the most targeted of the components, and the remaining one-third will come from the social quota and the government's other contribution. Reducing the social quota might be one way to reduce costs, but it would result in a higher payment for the minimum pension guarantee. The cost savings would be minimal, although the spending would be more highly focused on the poor. However, the impact of the contribution incentives in the social quota has not been adequately studied to know whether they will make a significant difference. 2.67 The question remains of how to finance the transition from one system to another and whether to focus on the distributional consequences during the transition period. It might be considered unfair to impose taxes that fall heavily on the poor to finance pensions for the relatively better off. Alternative taxes such as a tax on pensions might be more appropriate. In most countries in the world, retirement savings are taxed either when they are contributed or when they are received as benefits. In Mexico, pension benefits are taxed only if the pension - 40 - exceeds nine minimum salaries, which is a relatively high threshold. However, there are two other considerations. First, the financing needs of the pension system are largely temporary. If a progressive tax is applied to balance the regressivity of the current system, the tax will have to be reevaluated as the system becomes progressive. Second, the new pension system generates savings that pension funds invest in government bonds, essentially providing self-financing for the reform, presumably at lower interest rates than would exist in the absence of this stock of savings. Thus a large part of the cost of the reform is being passed to future generations, who will benefit from the lower government expenditure on pensions. 2.68 The above analysis of the financing and distribution of pension benefits for IMSS and ISSSTE right holders does not reveal the full extent of inequalities between private and public sector workers, as well as within the public sector. In addition to ISSSTE, the public sector includes a variety of specific pension systems, covering de armed forces (SEDENA, SECMAR), public workers employed by local governments, and the principal state companies, including Pemex, CFE, Luz y Fuerza, and IMSS (as employer). The latter cases are of particular interest in terms of financing and equity, as the historically powerful worker unions in these companies and agencies have managed over the last few decades to obtain exceptionally generous (and financially unsustainable) contractual obligations from the federal government. Considering only Pemex, IMSS and Luz y Fuerza, Figure 2.6 and Table 2.13 document the distributive implications of these achievements. The current pensioners in these three state companies represent 8 percent of all pensioners considered, but absorb almost a third of the benefits (defined as the tax-financed annual deficit of these pension schemes), and more than all ISSSTE pensioners together.21 Figure 2.6: Distribution of pensioners and benefits received: 2003 Pensioners Benefits ISSSTE PEMEX 7% 20% ISSSTE 22% IMSS-RJP 17% PEMEX 3% IMSS-RJP 5% Luz y Fuerza Luz y Fuerza 5% 1% IMSS-IVCM 71% IMSS-IVCM 49% Sources: Mexico Tercer Informe de Gobierno 2003; World Bank 2004; World Bank calculations. 2.69 Table 2.13 compares the benefits of these regimes with the pension benefits of IMSS affiliates (defined as the current federal transfers to these pensioners under the 1973 IMSS 21 Since this report was completed, Congress approved a reform of the pensions for new employees of IMSS, which will eventually help to solve this problem. - 41 - law).22 Public sector workers retire between 10 and 14 years earlier, with replacement rates of up to 100 percent of last salary for ISSSTE workers and 132 percent and 206 percent for IMSS and Luz y Fuerza workers (in case of PEMEX and Luz y Fuerza, this refers to the ratio of the average pension of retirees divided by the average salary of active workers), implying benefits per pensioner between 1.6 and 8.3 times those received by private sector workers. Although an incidence analysis of these extraordinary benefits is not feasible with available data, pensions in the state companies probably have similar a benefit distribution, and perhaps are even more regressive, given higher average salaries in comparison to ISSSTE affiliates. The majority of these benefits are paid not from employee contributions, but from general tax revenues (and social security contributions of private workers in the case of IMSS), which clearly have alternate uses in social and infrastructure investment programs. In the case of IMSS, for example, workers contribute only 7.4 percent of pension financing (implied by data reported in IMSS 2004). Table 2.13: Mexico: Public Sector Pensions Replacement Retirement Annual Benefits/pensioner+ Pensioners+ rate age Deficit+ multiples pension/last minimum or million pesos of private thousands salary average pesos sector IMSS affiliates-IVCM* 25,259 1 38-68*** 65 1,633 41,252 ISSSTE 39,372 1.6 50-100*** 55 458.5 18,052 PEMEX 99,003 3.9 63%** 55 57.8 5,722 IMSS workers-RJP 128,012 5 132%** 53 109.4 14,003 Luz y Fuerza 210,666 8.3 18.2 3,826 Sources and notes: * IMSS affiliate pensioners covered under 1973 IMSS Law by the seguro de “Invalidez, Vejez, Cesantía en Edad Avanzada y Muerte” (IVCM). ** IMSS (2003). *** Estimates from García and Villagomez (2003), refer to range of 15-30 years working period. + WB (2004), except IMSS-IVCM and ISSSTE, which are from Tercer Informe de Gobierno, Fox. Mexico Tercer Informe del Gobierno 2003; World Bank 2004; World Bank calculations. 2.70 The case of IMSS is of particular interest, as its 371,000 workers are employed to provide social security services (principally health care) for formal private sector workers, implying (on present budget constraints) a direct and rapidly growing zero-sum conflict for resources between the private clients and public workers of the Institute. In addition to the benefits established by social security law for all workers, IMSS workers benefit from a special regime known as “Regimen de Jublilaciones y Pensiones” (RJP). IMSS (2004) presents a simple simulation comparing the benefits obtained by an IMSS worker and a worker affiliated to IMSS, assuming identical salaries (5,000 pesos/month), incorporation age (25 years old) and life expectancy (78 years): the IMSS worker, who retires 12 years earlier (at 53 rather than 65), obtains a monthly pension 3 times higher (6,450 versus 2,106 pesos), and 5 times more pension income over retirement (2.56 versus 0.51 million pesos), and a similar proportion considering only resources 22. The great majority of currently pensioned IMSS affiliates are covered under the 1973 IMSS Law by the seguro de “Invalidez, Vejez, Cesantía en Edad Avanzada y Muerte” (IVCM). Following the 1997 IMSS pension reform, the federal government has absorbed these liabilities in full. - 42 - not financed by the worker’s contributions (2.45 versus 0.45 million pesos). In fact, average salaries of IMSS workers are estimated in the same source to be more than double those of IMSS affiliates, and the actual benefits received by IMSS workers are estimated to be 8.3 times higher. For IMSS, these pension rights cost an estimated 5.7 percent of GNP (IMSS 2003), equivalent to 25 percent of the costs assumed by the federal government in the IMSS pension reform of 1995. For each peso that the federal government spends for pensions on an IMSS right holder, it now spends over six pesos on pensions for each worker employed by IMSS to serve these right holders. Monetary Transfers: Oportunidades Program and Procampo 2.71 This section considers the distribution of direct monetary transfers from Mexico’s two principal antipoverty and agricultural support programs: the Oportunidades Program and Procampo. Given the innovative character of these instruments compared to the other more traditional expenditure programs analyzed here, we describe their main features before considering the distribution of their benefits. 2.72 The Oportunidades Program originated in 1997 as the Program for Education, Health, and Food (Progresa, Programa de Educación, Salud y Alimentación), offering transfers to poor rural households conditional on participation in basic education (for the relevant age group) and health services. Transfers consist of approximately equal parts education and food aid. The educational component offers scholarships with values increasing with grade level and differentiated in favor of girls (after primary education); it is designed to cover the opportunity cost of attending school for children and youngsters in poor households. The food aid component, which is provided in the form of money as well as food supplements, is conditioned on the use of health services. Until 2000 the program was targeted exclusively to poor rural communities. In 2001 coverage was extended to urban localities and medium-secondary education. 2.73 This program represents a unique and powerful redistributive instrument in Mexico for several reasons. First, it applies rigorous targeting criteria, identifying the poorest localities on the basis of a marginality index (CONAPO), and targets poor households within the localities identified using a proxy means test based on the socioeconomic characteristics and economic assets of households. Second, it exploits synergies between education, health, nutrition, and monetary transfers in the production of human capital, offering beneficiaries a long-term chance to escape chronic (intergenerational) poverty on a permanent basis as well as current income support. Third, it complements traditional public expenditure on health and education services with demand-side subsidies. As shown, between 1998 and 2000 there was remarkable improvement in the participation of the poorest deciles in lower-secondary education and health services for the uninsured. The demand-inducing impact of this program, along with complementary programs to expand coverage, is responsible for a large part of this increase. In addition to empowering the users rather than the providers of services, this strategy directly targets the principal reasons why the poor have limited access to post-primary education. 2.74 Procampo is an agricultural income support program offering fixed monetary payments per hectare of basic crops, independent of production levels. At the start of the program in 1994 the eligible beneficiaries were identified as direct producers (not owners) of a selection of - 43 - traditional crops in the three years prior to that year. According to official estimates, the program covers around 3 million agricultural producers and 90 percent of Mexico’s cultivated area and is intended to operate only during the transitional phase of agricultural liberalization under NAFTA (to be completed by 2008). 2.75 In contrast to the expensive and inequitable pre-NAFTA agricultural price support schemes, this program benefits subsistence as well as commercial farmers by offering a uniform payment per hectare, independent of yields or marketed output: 45 percent of the area covered belongs to small farmers with 5 hectares or less. For the poorer ejidatarios, the Procampo transfer represents up to 40 percent of income (Cord and Wodon 1999). 2.76 Table 2.14 presents the distribution of transfers operated by the two programs. The Oportunidades transfers are highly progressive and, despite the urban extension, still strongly pro-rural. Although this is by far the most effectively targeted program in Mexico (see figure 2.4), 15 percent of the benefits accrue to the upper half of the distribution. The poverty line used to identify beneficiaries is similar to the “food poverty” line used in Mexico at present. In 2002 the food poverty line identified as poor 20.3 percent of households nationally and 11.4 and 34.8 percent of households in the urban and rural sectors, respectively (Table 2.15). This implies inclusion errors (non-poor beneficiaries) for the Oportunidades Program in 2002 on the order of 40 percent nationally, 50 percent in the urban sector, and 55 percent in the rural sector. 2.77 Rather than failures in the actual targeting mechanisms, however, these errors may reflect differences in the welfare concepts and measurement errors used in ENIGH, versus those used in the Oportunidades Program surveys. They may arise either because income is not obtained directly in the Oportunidades surveys (but is estimated from non-income characteristics) or from limitations in the ENIGH income information at the first quintile, which, in contrast, to the Oportunidades survey, is not designed to be representative of poor rural communities. Given the well-known difficulties of obtaining accurate income reports for populations at this level of poverty, the Oportunidades indicator could be the more accurate one. 2.78 Procampo has a progressive overall distribution, which is limited because allocations are defined as a function of the area of cultivated land. Nevertheless, because poor subsistence farmers represent a large proportion of producers of basic crops, the poor (as defined by the food poverty line) benefit from 33 percent of transfers. 2.79 To understand the distributive changes implied by the creation and evolution of Oportunidades and Procampo in the past decade, it is important to consider these programs in the broader context of the instruments they were designed to replace. Procampo emerged as a transitional program to replace price support programs in the face of NAFTA. The creation and rapid growth of Oportunidades (and before it Progresa; see Table 2.16) were financed by the resources freed up by the reduction of consumer food subsidies (effectively replacing them through its food component), which was originally set up to buffer urban consumers from agricultural price supports. - 44 - Table 2.14: Distribution of Monetary Transfers, 2002 Decile Total Oportunidades Procampo NATIONAL 1 29.4% 36% 19% 2 19.5% 23% 14% 3 12.2% 15% 8% 4 9.2% 10% 8% 5 6.5% 7% 6% 6 5.1% 3% 7% 7 5.1% 3% 9% 8 5.6% 2% 11% 9 3.1% 0% 7% 10 4.4% 0% 11% Concentration -0.387 -0.581 -0.108 coefficients Urban 19.0% 20.2% 17.3% Rural 81.0% 79.8% 82.7% URBAN 1 36.5% 49.7% 14.3% 2 14.1% 18.6% 6.6% 3 8.6% 11.4% 3.9% 4 8.7% 9.1% 8.1% 5 3.6% 3.9% 3.0% 6 3.2% 4.0% 1.9% 7 1.9% 1.2% 3.2% 8 5.9% 2.0% 12.3% 9 2.2% 0.1% 5.8% 10 15.3% 0.1% 40.8% Concentration -0.308 -0.646 0.259 coefficients RURAL 1 10.8% 12.4% 8.5% 2 11.2% 14.9% 6.3% 3 9.6% 11.7% 6.7% 4 11.0% 12.6% 8.7% 5 10.1% 11.7% 7.9% 6 8.1% 9.7% 5.8% 7 9.4% 10.8% 7.5% 8 7.5% 7.7% 7.3% 9 10.4% 5.7% 17.0% 10 11.9% 2.9% 24.4% Concentration -0.013 -0.177 0.215 coefficients Source: ENIGH (2002) and Table 2.3. Population deciles are ordered by per capita expenditure. The shaded area denotes the poor population (according to the “food poverty line”). - 45 - Table 2.15: Poverty Rates with the Food Poverty Line, 2002 National Urban Rural Households 15.8% 8.5% 28.5% Population 20.3% 11.4% 34.8% Transfers received by poor Oportunidades 59% 52% 45% Procampo 33% 15% 26% Source: Sedesol (2003) and Table 2.9. Table 2.16: Evolution of Oportunidades and Procampo Transfers Beneficiaries Transfers (million MxP, base=2002) Oportunidades Oportunidades (Progresa) Year Procampo (Progresa) Procampo Total Monetary In kind* 1994 3,295 17,538 1995 2,934 15,715 1996 2,987 13,548 1997 2,850 301 12,454 376 252 124 1998 2,780 1,596 12,110 4,167 2,131 2,035 1999 2,724 2,306 11,464 7,875 6,425 1,451 2000 2,681 2,476 11,595 10,265 8,611 1,654 2001 2,695 3,238 11,558 12,231 10,181 2,049 2002 2,792 4,240 11,851 16,105 13,845 2,260 Growth 1994–2002 -32.4% 1998–2002 286.5% Note: Transfers to beneficiary households are net of administrative costs. *Transfers in kind include food supplements and medical services. 2.80 Pricing policies in agriculture benefited surplus producers of basic crops (especially corn and beans) through minimum guaranteed prices, while the real income of urban consumers was protected through a costly untargeted subsidy on these products and their derivatives, principally the tortilla. Between 1965 and 1982 these policies favored urban consumers, implying a net tax on agriculture in the context of an overvalued exchange rate in most years.23 With the severe fiscal cuts following the 1983 crisis, the generalized food subsidies to consumers became unsustainable. By the early 1990s, the internal price of corn was 70 percent above international prices, and the generalized tortilla subsidy was insufficient to compensate urban consumers for this differential (Levy and Wijnbergen 1992). The tortilla subsidy was gradually reduced beginning in the mid-1980s (and finally eliminated in 1998) in favor of targeted subsidies, principally milk (LICONSA) and tortilla rations (tortibonos). The targeted subsidies were costly to operate and badly targeted, however, and they were concentrated in urban areas, principally in Mexico City, and the criteria for selecting beneficiaries lacked transparency and targeting efficiency.24 23. Lustig (1989, p. 108). See also Levy and Wijnbergen (1992) and Friedman and others (1995). 24. 62 percent of the beneficiaries of the targeted milk food subsidy (44 percent of the beneficiaries of the tortilla food subsidy) were in the six richer deciles in 2000 (Scott 2003). - 46 - 2.81 The big losers from these policies were subsistence farmers and landless rural workers, first by being net buyers of corn and thus taxed by these pricing policies even in the later period and second by missing out on the consumption subsidies. The 1991 agricultural census reports 2.2 million farm households owning less than 5 hectares, with 1.3 million owning less than 2. Corn and coffee pricing policies in the 1980s and early 1990s imposed taxes on small agricultural producers in the poorest regions on the order of 15–30 percent, redistributing the proceeds to large farmers in richer regions (Deininger and Heinegg 1995). Most other support instruments—irrigation, electricity subsidies, credits, and technical assistance—tended to benefit richer producers in the north of the country. 2.82 Against this background, Table 2.16 reports the evolution of the two programs since their creation. Procampo began operating at full coverage but was abruptly reduced in the aftermath of the 1995 crisis, remaining essentially static since 1998. Oportunidades, in contrast, expanded in phases and by 2001 had surpassed Procampo in number of beneficiaries as well as transferred resources. 2.83 To evaluate the distributive changes in agricultural support policies implied by the introduction of Procampo would require a nationally representative estimate of the incidence of price support policies. Lacking this, Figure 2.8 presents the distribution of Procampo between 1994 and 2002. The figure suggests that the severe cuts in Procampo transfers between 1994 and 1996 affected poorer farmers disproportionately. These results must be interpreted with caution, as ENIGH is not designed to capture the small number of large Procampo beneficiaries accounting for the regressive part of the distribution at the top.25 Figure 2.7: Distribution of Procampo Transfers, 1994–2002 30% 1994 25% 1996 2002 20% 15% 10% 5% 0% 1 2 3 4 5 6 7 8 9 10 25. The years included in the figure correspond to the largest samples of Procampo beneficiaries captured in the ENIGH surveys: Observation 1994 1995 1998 2000 2002 Households reporting Procampo transfers 1,538 913 451 388 1,378 - 47 - 2.84 Figure 2.8 presents an estimate of the distributive changes implied by the introduction of Progresa on food programs in Mexico over the second part of the past decade. As the food component became the principal food subsidy program, with the elimination of the universal tortilla subsidy and gradual reduction of other targeted programs, the coverage in the first quintile jumped threefold, which was appropriate to match the distribution of undernourished children in Mexico (Scott 2003). Figure 2.8: Distribution of Public Expenditure on Food Programs, 1994–2000 35% 1994 30% 2000 25% 20% 15% 10% 5% 0% 1 2 3 4 5 6 7 8 9 10 Data: ENN (1999); ENIGH (1996, 2000); Poder Ejecutivo Federal (2001). Notes: Food subsidies include (a) general (1990) and targeted tortilla (tortivales and tortibonos in 1990, FIDELIST in 2000) subsidies, (b) targeted milk subsidy (LICONSA), (c) school breakfast (DIF), and the Progresa food component (2000). Household deciles are ordered by income per capita. Source: Scott (2003). Utilities: Electricity and Water 2.85 Between 1992 and 2002 the coverage of electricity was expanded from 15.8 million to 22.8 million households and from 91.6 to 96.5 percent of all households (Tables 2.17 and A2.9). Given that the largest gap in coverage at the beginning of the decade was concentrated in the poorest deciles, this expansion benefited poor households the most, making the distribution of residential electricity more equal (Table A2.9). Not surprising, the participation in this service remained slightly biased in favor of the urban sector. 2.86 Average monthly consumption of kilowatt-hours per covered household also increased over the decade as a whole, with dips in the two recessions (1995–96 and 2002). Average monthly household expenditure, in contrast, decreased between 1992 and 2000 and increased significantly in the 2000–02 period. This trend is observed in the ENIGH data (Table A2.10) as well as in the administrative data from the electricity companies, although the two sources differ in magnitude by a factor close to 2 (1.85, 1.97, and 1.99 in 1992, 2000, and 2002, respectively). This may happen because most households pay a bimonthly electricity bill but probably forget to divide this when reporting monthly expenditures in ENIGH. - 48 - Table 2.17: Evolution of Coverage and Private and Public Expenditure on Residential Electricity, Administrative Records, 1992–2002 Average Total User Average Average Price/cost monthly federal households price monthly consumption subsidy (million) ($/kWh) expenditure Year (kWh/user) CFE* LFC* (million $) 1992 15.8 129 0.815 105 1993 16.5 131 0.777 102 1994 17.2 137 0.770 106 0.53 0.43 21,396 1995 17.8 135 0.676 91 0.47 0.26 29,060 1996 18.3 131 0.637 83 0.42 0.25 31,930 1997 18.9 133 0.620 83 0.40 0.27 31,944 1998 19.6 137 0.623 85 0.43 0.26 31,979 1999 20.2 140 0.603 84 0.41 0.26 33,997 2000 21.1 145 0.625 91 0.41 0.24 39,106 2001 21.9 149 0.638 95 0.42 0.26 39,382 2002 22.8 145 0.774 112 0.50 0.31 37,206 *CFE =Compañía Federal de Electricidad; LFC=Luz y Fuerza del Centro. Source: Poder Ejecutivo Federal (2003); SHCP. 2.87 The contrast in the evolution of household electricity consumption and expenditure is explained by the evolution of federal electricity subsidies to residential consumers in this period, as reported by the federal government. These subsidies increased significantly, almost doubling in real terms between 1994 and 2000. Between 2001 and 2002, however, the trend reversed through a restructuring of residential tariffs, which significantly raised the infra-marginal tariffs paid by middle and especially high consumers of electricity (Figure A2.9). Tariff structures for residential consumers have become increasingly complex over the past two decades: There are now 112 different ways to calculate tariffs of residential customers based on the level of consumption, season, and local temperature. The tariff structure has contained three elements of subsidy: a nominal fixed charge at a fraction of fixed cost, several blocks of subsidized energy charges and seasonal subsidies, consisting of longer energy blocks of lower energy charges in the summer months. While the first two mechanisms were ostensibly meant to subsidize consumption of lower income households, the high correlation between income and consumption has meant that most of the subsidy accrues to non poor households. The summer subsidy, meant to offset the costs of air conditioning, has grown rapidly over time, now comprising 6 separate schedules and covering 42 percent of the population. The latter benefit non-poor households to an even greater degree, because many poor households in Mexico do not have air conditioning and those that do consume significantly less energy than non-poor households. The size of the subsidized blocks has grown, as has the duration of the summer tariff, accelerating the growth in overall consumption and hence expanding subsidies in hot- summer zones relative to temperate zones. Whereas residential customers in temperate zones received an average subsidy of 1,027 pesos in 2002, those in the warmest zones in the North got 3,471 pesos—about 15 percent of the GDP per capita in the southern-most state of Chiapas. 2.88 To approximate the impact of these changes, we analyze the distribution of household expenditure on electricity as reported in the ENIGH surveys (Tables A2.10 and A2.11). The reduction of household expenditure between 1992 and 2000 was concentrated in the extremes of the distribution: the poorest quintile and (especially) the richest decile. Although the top decile - 49 - also experienced the largest increase in household expenditure between 2000 and 2002, this just cancelled (most of) the reduction achieved in the earlier period. Over the whole decade, the poorest (decile 1) and richest (deciles 9 and 10) households presented the smallest absolute increases in expenditure. In terms of relative shares, the distribution of household expenditure on electricity became less regressive. 2.89 Table 2.18 presents the distribution of the residential electricity subsidy over this period, estimated using ENIGH data on household expenditures and the electricity tariff schedules presented in Figure A2.9 in annex A2 (see annex A1 for the method used). Between 1996 and 2000 the distribution of the subsidy became more equal, as a result not of a change in the tariff structure but of the gains in coverage in the first eight deciles. The important gain in progressivity achieved between 2000 and 2002 is explained largely by the change in the tariff structure, which lowered the subsidy significantly for the top decile. Thus, while the growth in the federal subsidy in the first period was distributed fairly equally (although biased against the poorest quintile), the contraction in this subsidy in 2000–02 was absorbed by the middle and (especially) high-use consumers in deciles eight to nine and ten, respectively. Table 2.18: Distribution of Residential Electricity Subsidy, 1996–2002 Marginal incidence Average Change of distribution incidence Distribution of change (percentage points (participation in change) gained) 2000 2002 1996–2000 2000–02 1996–2000 2000–02 Change in 22.5% -4.9% expenditure 1 2.7% 3.4% 0.5% 0.7% 6.0% -10.6% 2 4.3% 4.8% -0.5% 0.5% 2.9% -5.5% 3 6.0% 6.3% 0.7% 0.2% 10.6% 1.1% 4 6.9% 7.2% -0.3% 0.3% 6.6% 1.9% 5 8.9% 9.8% 0.6% 0.9% 12.4% -8.5% 6 9.6% 11.4% 0.7% 1.8% 14.0% -26.0% 7 11.5% 12.1% 0.8% 0.6% 15.5% -0.2% 8 13.5% 13.1% 0.4% -0.4% 15.3% 21.0% 9 16.4% 16.4% -1.6% 0.0% 6.8% 16.8% 10 20.1% 15.6% -1.3% -4.6% 10.0% 109.9% Concentration 0.294 0.242 (-9.5%) (-17.7%) coefficient Source: ENIGH. 2.90 The 2002 residential tariff revision intended to reduce the overall volume of subsidies and to target them to lower income households. Therefore, price increments were directed at only deciles 8, 9, and 10. But, subsidies remain concentrated in deciles 6-9. Beyond predetermined consumption thresholds, the full service cost was to be charged to consumers. The target was the top 5 percent of consumers but as a consequence of the introduction of tariff 1F in mid 2002 and pressures to reassign localities (still ongoing) to those summer season tariffs which provide larger subsidies, only 2.6 percent of residential consumers wound up paying the full cost (DAC) tariffs. Thus, while the 2002 tariff revision slightly addressed regressivity at the very top end, the share of subsidies going to the non-poor population (i.e., those above poverty - 50 - line 3), remains high, estimated at 64 percent. The total volume of residential subsidies in 2002 declined by 5 percent from the prior year, but the magnitude remains large, at MxP 31 billion in 2002, about equal to total revenues residential customers actually paid for electricity service. These sums place electricity subsidies among Mexico’s largest welfare programs. 2.91 The regressive effect of the electricity subsidies also shows in the detailed data directly from the billing statements of CFE (Comisión Federal Electrica). Table 2.19 shows the distribution of the subsidy according to deciles of electricity consumption. See Annex 4: Distributional and Fiscal Impact of Power Sector Subsidies (forthcoming). The overall magnitude of the subsidy is higher with the sector-data estimate, and some details of the distribution are different, due to differences in the assumption used to calculate the subsidy (a fixed cost per user of 33 pesos per month was assumed), in the nature of the sample (Luz y Fuerza in the Mexico City area is not included), and in the ordering of households by electricity consumption rather than by autonomous income. Nonetheless, the problem of this subsidy is clear from both sources—its large size, motivation of inefficiency, and distribution mainly to the middle and upper income groups. Table 2.19: Consolidated Results for all Residential Tariffs C C S C O , CO ALL RESIDENTIAL TARIFFS ANNUAL 2002 Monthly average consumption (kWh) 140.8 Number of households 23,851,010 Total annual consumption (GWh) 40,298 SUBSIDIES BY DECILE AVERAGE AVERAGE AVERAGE UNITARY INCLUDING HOUSEHOLDS WITHOUT ELECTRICITY DECILE CONSUMPTION COST INVOICE SUBSIDY Cumulate kWh/month $/hh-month $/hh-month $/kWh $/hh-year M$/year % % 1 5 39.49 21.43 3.88 217 517 1.6% 1.6% 2 24 65.36 14.38 2.10 612 1,459 4.7% 6.3% 3 55 106.19 23.73 1.49 990 2,360 7.5% 13.8% 4 86 145.91 35.39 1.29 1,326 3,163 10.1% 23.9% 5 113 181.79 47.09 1.19 1,616 3,855 12.3% 36.2% 6 138 214.83 58.58 1.13 1,875 4,472 14.3% 50.5% 7 165 249.98 74.41 1.06 2,107 5,025 16.0% 66.5% 8 202 297.82 130.69 0.83 2,006 4,784 15.3% 81.8% 9 255 367.83 216.73 0.59 1,813 4,325 13.8% 95.6% 10 365 512.37 464.21 0.13 578 1,378 4.4% 100.0% kWh/month M$/month M$/month $/kWh $/hh-year M$/year 140.8 5,203 2,592 0.78 1,314 31,339 Source: CFE and World Bank calculations. 2.92 In the case of water, coverage increased from 10 million to 15 million households between 1992 and 2002. The distribution became more equal between 1992 and 2000 if we consider access to piped water broadly defined, within the house, building, or immediate neighborhood (Table 2.20),26 but less equal if we consider access to piped water within the house only. 26. Note the distribution of households with access to piped water within the house is significantly more regressive (see Table A2.12 in the annex). - 51 - Table 2.20: Distribution of Households with Access to Piped Water within the House or in the Building, Neighborhood, or on the Land Decile 1992 2000 2002 1 5.8% 7.6% 7.1% 2 7.6% 8.8% 8.7% 3 8.9% 9.6% 9.6% 4 9.7% 9.8% 9.7% 5 10.0% 10.2% 10.3% 6 10.5% 10.3% 10.5% 7 11.2% 10.7% 10.7% 8 11.6% 10.9% 11.1% 9 12.1% 11.0% 11.2% 10 12.6% 11.1% 11.2% CC 0.112 0.056 0.065 Source: ENIGH. 2.93 Household expenditure on water showed a slight increase in regressivity between 1992 and 2000, followed by a significant reduction between 2000 and 2002 (see Table A2.13). This suggests that, in contrast to the case of electricity, water subsidies became more regressive between 2000 and 2002. All Benefits 2.94 In this section we consider the distribution of total public expenditure in all of the programs analyzed for the years 2000 and 2002. 2.95 Figures 2.10 and 2.11 present concentration coefficients for all the programs nationally and for distributions within the urban and rural sectors. We include for reference Ginis for autonomous expenditure. Perhaps the most striking observation is the wide range of coefficients, from 0.6 (ISSSTE pensions) to -0.6 (Oportunidades). The great majority of programs are regressive in absolute terms, although only public expenditure on ISSSTE pension is regressive in relative terms. Following Oportunidades at some distance, health services for the uninsured, basic education, and Procampo are the only other programs that are progressive in absolute terms. - 52 - Figure 2.9: Concentration Coefficients, National, 2000, 2002 NATIONAL ISSSTE Pensioners ISSSTE Active W orkers GINI 2002 Tertiary Edu. 2000 ISSSTE Health IM SS Active W orkers IM SS Pensioners PEM EX Health Hospital Health Electricity Subsidy IM SS Health Upper Sec. Edu. M aternal Health Prim ary Health TO TAL Low er Sec. Edu. PRO CAM PO Prim ary Edu. Preschool SSA O portunidades -0.800 -0.600 -0.400 -0.200 0.000 0.200 0.400 0.600 0.800 Note: Procampo is not reported in 2000 because the sample of beneficiaries captured by the 2000 survey is not representative. Source: ENIGH 2000 and 2002; SHCP spending data; World Bank calculations. Figure 2.10: Concentration Coefficients, Rural, Urban, 2002 ISSSTE Pensioners URBAN GINI RURAL ISSSTE Active Wkers Tertiary Edu. ISSSTE Health IMSS Active Workers Procampo IMSS Pensioners Hospital H. Care IMSS Health Other Scholarships Upper Sec. Edu. Primary H. Care Maternal H. Care TOTAL Lower Sec. Edu. SSA Primary Edu. Preschool Oportunidades -0.800 -0.600 -0.400 -0.200 0.000 0.200 0.400 0.600 0.800 1.000 Source: ENIGH 2002; SHCP spending data; World Bank calculations. - 53 - 2.96 In general, the distribution of these programs is notably less regressive (more progressive) in urban than in rural sectors, indicating that most of the equity achieved nationally is due to the intersectoral allocation of programs rather than to particular success in targeting within the rural sector. Although Oportunidades is still the most progressive program within the rural sector, the distance separating it from preschool and primary school services is marginal. The targeting efficiency of expenditure on rural primary schools is comparable to that of Oportunidades transfers (Figure 2.11). Figure 2.11: Distribution of Oportunidades Transfers and Students in Primary Education in the Rural Population, 2002 16% Primary Education 14% Oportunidades 12% 10% 8% 6% 4% 2% 0% 1 2 3 4 5 6 7 8 9 10 Source: ENIGH 2002; SHCP spending data; World Bank calculations. 2.97 Table 2.21 reports the distribution of total public expenditure. Given that public spending on pensions compensates, in part, for intertemporal interpersonal transfers in the past that we could not net out, we exclude public spending on pensions except for federal transfers to the pension accounts of active workers under the 1995 IMSS law (which are pure transfers). To ensure comparability between 2000 and 2002 we include exactly the same programs and use consistent budgetary information. The two ENIGHs are comparable for all programs considered, except for monetary transfers, where we make appropriate adjustments, described in annex A1. 2.98 The overall distribution of public spending resulting from the distribution of use of services and receipt of transfers and the distribution of the budgetary allocations to the relevant programs are broadly neutral, nationally as well as within the urban and rural sectors. Allocation is slightly pro-rural. Even though only a minority of the programs are progressive, spending is neutral overall because the rural sector absorbs almost half of the resources (48 and 47.2 percent in 2000 and 2002, respectively; see Table 2.3). The distribution becomes only slightly more regressive if we include all public expenditures on pensions (the overall concentration coefficient for 2002 increases from 0.03 to 0.054). 2.99 The neutrality within the rural and urban sectors may seem surprising, given the contrast between the progressivity of the programs in the two sectors. This is explained by the fact that the more progressive programs among rural households are also by far the most pro-rural in intersectoral allocation. - 54 - Table 2.21: Distribution of Total Public Expenditure, 2000–02* National Urban Rural Decile 2002 2000 2002 2000 2002 2000 1 9.5% 8.9% 8.3% 8.6% 8.7% 9.3% 2 8.8% 8.9% 9.2% 9.0% 10.2% 9.7% 3 9.9% 9.1% 9.1% 8.4% 8.8% 9.4% 4 9.2% 8.6% 9.3% 9.5% 9.6% 9.5% 5 9.9% 9.4% 10.0% 10.1% 10.1% 10.3% 6 9.7% 10.1% 10.4% 11.4% 10.3% 9.9% 7 10.5% 11.6% 12.0% 10.6% 10.7% 10.3% 8 12.0% 11.2% 11.3% 12.1% 10.3% 9.3% 9 11.4% 12.2% 10.8% 11.6% 9.9% 11.2% 10 9.1% 10.0% 8.3% 8.6% 10.5% 11.2% Urban 72.5% 73.8% Rural 27.5% 26.1% Concentration 0.030 0.053 0.018 0.041 0.016 0.030 coefficient *In the case of pensions, only government transfers to active workers affiliated with IMSS are included in the total. Source: Tables 2.3; 2.5; 2.9; 2.10; 2.14; 2.18 and Annex A2. 2.100 Comparing 2002 with 2000, we observe a slight gain in overall progressivity, from a concentration coefficient of 0.053 to 0.03. Table 2.22 presents a marginal incidence analysis for this period and decomposes the gains in progressivity within programs (coverage) and the budgetary reallocations between programs. There were both kinds of gains, but more equitable coverage within programs was the principal determinant. The gains in equitable coverage occurred in secondary education, health services, and residential electricity subsidy (see Figure 2.10). Table 2.22: Changes in the Distribution of Total Public Expenditure between 2000 and 2002, National* Change in relative share Additional resources Due to budgetary Millions of Distribution of Due to coverage Total allocation pesos 2002 additional resources 4.3% 2.2% 6.6% 6300 17.5% -1.2% 0.9% -0.4% 3010 8.3% 7.6% 0.1% 7.7% 7052 19.5% 6.2% -0.1% 6.1% 5939 16.5% 5.9% -0.4% 5.5% 6097 16.9% -3.3% -0.8% -4.1% 1437 4.0% -9.1% -0.5% -9.5% -1718 -4.8% 7.6% -0.1% 7.5% 8475 23.5% -5.5% -0.5% -5.9% 540 1.5% -8.2% -0.6% -8.7% -1034 -2.9% -0.019 -0.003 -0.023 *In the case of pensions, only government transfers to active workers affiliated with IMSS are included in the total. Source: Tables 2.3; 2.5; 2.9; 2.10; 2.14; 2.18 and Annex A2. - 55 - REDISTRIBUTIVE IMPACT 2.101 A strict evaluation of the redistributive impact of the programs we have analyzed would require controlling for behavioral changes, general equilibrium effects, and unobserved determinants through experimental or quasi-experimental impact evaluation, which ideally would require a comparable and randomly selected control group isolated from these benefits for a relevant period of time. This is not feasible for most of the programs analyzed here. Only in the case of the Oportunidades Program has a strict impact evaluation been undertaken, made possible by the decision to maintain a control group as the program expanded its coverage in its first two years of operation (1998–2000). 2.102 Here we consider redistributive impact only in terms of the current spending of resources on households that benefit from subsidized goods and services or receive direct monetary transfers. We consider three types of redistributive impact: (a) the impact of all benefits on the distribution of total household expenditure, (b) the impact of public education and health spending on the distribution of total (public and private) education and health spending on or by households, and (c) the impact of monetary transfers on (monetary) poverty measures. 2.103 Since the redistributive impact of a transfer is directly proportional to the average transfer rate—or the ratio of the transfer to autonomous expenditure or income27—its measurement requires consistency between household expenditure and public spending data. As with most other household income and expenditure surveys in the world, there is strong evidence of underreporting in ENIGH. Total expenditure on final private consumption as reported in Mexico’s National Accounts for 2002 is higher than total expenditure as reported in 2002 ENIGH by a factor of 1.94. Since the classifications for private consumption used in these sources are not strictly comparable, and there is no satisfactory way to allocate this difference to households along the distribution (there is much evidence to suggest that a large part of this is due to the truncation of ENIGH in the upper tail of the distribution), adjusting the ENIGH reports to make them consistent with the National Accounts is not recommended for measuring welfare (Comité Técnico para la Medición de la Pobreza 2002). However, since public expenditure is certainly not underreported, by deciding not to adjust the figures, we would obviously overestimate the average transfer rate and thus the redistributive impact. This is 19.5 percent for all the benefits considered, if we use household expenditure as reported in ENIGH, but 11 percent if we use the corresponding concept from the National Accounts. We report redistributive impact measures for both cases. All Benefits 2.104 Table 2.23 considers the potential impact on the distribution of current household expenditure of all the benefits considered. Assuming public and private spending on goods and services to be fully comparable and depending on the source of total household expenditure used, the Gini coefficient would be reduced between 10 and 18 percent. Given the very low γ ∆G = (C − G ) 27. More precisely, (1 + γ ) , where G is the Gini coefficient of autonomous household expenditure, C is the concentration of the transfer, and γ is the average transfer rate (Kakwani 1977). - 56 - participation in the autonomous expenditure of the poor, these transfers have a large potential impact on this population, doubling the participation in the total expenditure of the first decile and increasing its absolute expenditure 150 percent (with total expenditure as reported in ENIGH; see Figure 2.12). The bulk of this impact is due to basic education, health for the uninsured, and the Oportunidades Program. By contrast, despite the absolute proportionality in overall public spending, the upper deciles benefit much less in relative terms (little more than 5 percent for the top decile) and do so mostly from post-basic education and IMSS health services. Table 2.23: Distribution of Household Expenditure before and after Public Transfers, 2002 After public transfers* Autonomous Distribution Change in shares Deciles expenditure National ENIGH ENIGH National Accounts Accounts 1 1.5% 3.1% 2.4% 101.6% 57.9% 2 2.7% 3.9% 3.4% 44.2% 25.2% 3 3.6% 4.9% 4.3% 33.2% 18.9% 4 4.7% 5.5% 5.2% 18.8% 10.7% 5 5.8% 6.6% 6.2% 13.9% 7.9% 6 7.0% 7.5% 7.3% 7.5% 4.3% 7 8.7% 9.0% 8.9% 4.2% 2.4% 8 11.2% 11.3% 11.3% 1.5% 0.8% 9 15.7% 14.9% 15.2% -5.3% -3.0% 10 39.2% 33.3% 35.9% -14.9% -8.5% Gini 0.481 0.393 0.431 -18.2% -10.4% *The distribution after transfers is estimated using the ENIGH (2002) for the distribution of public expenditure and using two alternative sources for total household expenditures: (a) ENIGH (2002) and (b) National Accounts. Source: World Bank calculations. 2.105 Figure 2.13 reveals a dramatic contrast in potential impact between the rural and urban sectors, reflecting mostly the differences in autonomous expenditure but also the intersectoral allocations of the programs. This impact is larger for the first through fifth deciles in the rural sector than for the poorest decile in the urban sector. For the poorer deciles in the rural sector, the monetary transfers from the Oportunidades Program and Procampo together have the second most important redistributive impact after primary education. - 57 - Figure 2.12: Public Expenditure as a Proportion of Autonomous Expenditure, National, 2002 10 Primary Edu. SSA 9 Oportunidades Lower Sec. Edu. 8 Preschool IMSS Health 7 Procampo 6 Upper Sec. Edu. Electricity Subsidy 5 PEMEX Health IMSS Active Workers 4 ISSSTE Health Tertiary Edu. 3 2 1 0% 20% 40% 60% 80% 100% 120% 140% Source: World Bank calculations. Figure 2.13: Public Expenditure as a Proportion of Autonomous Expenditure, Urban, Rural, 2002 10 9 Primary Edu. Oportunidades SSA Preschool 8 Lower Sec. Edu. Procampo 7 IMSS Health Upper Sec. Edu. URBAN 6 Tertiary Edu. IMSS Active Workers 5 ISSSTE Health IMSS Pensioners 4 ISSSTE Pensioners 3 2 1 10 9 8 7 RURAL 6 5 4 3 2 1 0% 50% 100% 150% 200% 250% Source: World Bank calculations. - 58 - Education and Health Resources 2.106 It is clear that the primary redistributive objective of the programs we have analyzed is to promote equality not in current income or overall expenditure but in the specific dimensions addressed: education and health. It is therefore of interest to consider the impact of these programs on spending specifically for these goods. 2.107 Figure 2.14, which compares expenditures on education and health before and after public transfers in kind, shows that the public sector accounts for the majority of resources expended on these services for the poorest 80 percent of the population (in the more conservative estimate, using autonomous household expenditure as reported in the National Accounts), achieving a remarkable degree of equality in total health and educational resources. This impact is especially dramatic in the rural sector, where public expenditure multiplies the resources available for the first quintile by a factor of 11 and 6 in education and health, respectively (Figure 2.15). Education expenditure achieves a larger impact than health spending for the poor, but the health spending becomes as important as education spending for the upper deciles in the urban sector. 2.108 Quantifying the reduction of inequalities in resources expended on education and health, we find reductions in their respective concentration coefficients of 48–64 percent and 42–56 percent (Table 2.24). In other words, the scale and distribution of public expenditure on education and health services in Mexico reduce the inequalities in autonomous educational and health resources by at least a half. Table 2.24: Inequality of Education and Health Resources before and after Public Expenditure,* 2002 Autonomous Public Total Reduction Education 35% 65% Expenditure share (51%) (49%) 0.212 63.9% Concentration coefficient 0.587 0.009 (0.305) (48.1%) Health 34% 66% Expenditure share (50%) (50%) 0.226 55.9% Concentration coefficient 0.520 0.076 (0.281) (42.3%) *Values in parenthesis adjust autonomous expenditure to National Accounts. Source: World Bank calculations. - 59 - Figure 2.14: Expenditure on Education and Health: Autonomous and Public (million pesos, 2002) 200000 EDUCATION EXPENDITURE 180000 Public 160000 Autonomous 140000 120000 100000 80000 60000 40000 20000 0 1 2 3 4 5 6 7 8 9 10 100000 HEALTH EXPENDITURE 90000 Public 80000 Autonomous 70000 60000 50000 40000 30000 20000 10000 0 1 2 3 4 5 6 7 8 9 10 Note: Total household expenditures adjusted to National Accounts. Source: World Bank calculations. - 60 - Figure 2.15: Public Expenditure on Education and Health as a Multiple of Autonomous Expenditure on Education and Health, Urban, Rural, 2002 14 Education 12 Health 10 8 6 4 2 0 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 rural urban Note: Total household expenditures adjusted to National Accounts. Source: World Bank calculations. Monetary Transfers: Impact on Monetary Poverty 2.109 Unlike in education and health services, where the final redistributive impact depends on the quality and cost-effectiveness of public services, we can measure the direct impact of the monetary (or near-monetary) transfers of Oportunidades28 and Procampo on poverty with some confidence. Although these transfers represent less than 5 percent of the total public expenditure analyzed in this chapter, their allocation to the rural poor makes them highly effective as redistributive instruments. Within the rural sector, these transfers represent close to 50 percent of the autonomous expenditure of the poorest decile and more than 60 percent of their current income (see Figure 2.16). 2.110 Table 2.25 presents the impact on poverty implied by these transfers, according to three common poverty measures (headcount, poverty gap, and squared poverty gap), with the three poverty lines officially recognized by the Government of Mexico. Extreme poverty (L1) is reduced by 4.5 percent nationally and 13.7 percent in the rural sector in terms of the head count, but 17 percent nationally and 35.6 percent rurally in terms of the squared poverty gap, which gives more weight to the poorest within the poor. 28 We include food supplements distributed by the program. - 61 - Figure 2.16: Oportunidades and Procampo Transfers as a Proportion of Autonomous Expenditure/Income in Rural Population, Rural, 2002 70% Procampo Oportunidades (in kind) 60% Oportunidades (monetary) 50% 40% 30% 20% 10% 0% 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 Expenditure Income Note: Neither transfers nor autonomous resources are adjusted to public expenditure and national account data. For the Oportunidades Program, in-kind transfer refers to the food supplement and health services. Source: World Bank calculations. Table 2.25: Reduction of Poverty due to Oportunidades and Procampo Transfers (% reductions in poverty measures) Procampo Oportunidades Total Program P0 P0 P1 P2 National L1 (alimentaria) 1.47% 2.97% 4.5% 11.1% 16.8% L2 (capacidades) 1.07% 2.47% 3.5% 8.4% 13.2% L3 (patrimonial) 0.27% 0.47% 0.7% 3.6% 6.3% Rural L1 (alimentaria) 4.33% 9.49% 13.7% 26.1% 35.6% L2 (capacidades) 2.96% 6.96% 9.5% 21.2% 29.9% L3 (patrimonial) 0.97% 1.52% 2.4% 10.6% 17.2% Urban L1 (alimentaria) 0.51% 0.80% 1.4% 6.1% 10.5% L2 (capacidades) 0.43% 0.97% 1.5% 4.2% 7.7% L3 (patrimonial) 0.04% 0.12% 0.2% 1.3% 2.7% Notes: 1. The poverty measures correspond to the Foster-Greer-Thorbeck (FGT) family of poverty measures (with alpha=0-2): P0=headcount; P1=poverty gap; P2=squared poverty gap. 2. The poverty lines correspond to the three poverty lines adopted by the Government of Mexico in 2002 (Comité Técnico para la Medición de la Pobreza 2002): L1=pobreza alimentaria: L2: pobreza de capacidades, L3: pobreza patrimonial. 3. These estimates are based on the Oportunidades and Procampo transfers reported by households in ENIGH (2002). In both cases, they correspond to approximately 80 percent of transfers reported by the federal government. To this extent, the reported impacts underestimate the full impact of these transfers. Source: World Bank calculations. - 62 - Cost-Efficiency of Public Services 2.111 The estimations of distributive impact we have presented for education and health services assume that (a) the value of public services to households per dollar spent is comparable to that of private services and that (b) this value is also comparable within the public sector across all attention units (at the same level, type of attention, and serving institution). Unfortunately, public information to evaluate this assumption systematically is limited, which in itself makes the assumption improbable. This lack of information—on costs as well as quality— is partly explained by an explicit policy of secrecy in the past (SEP 2002), which the new administration replaced with a policy of transparency and a commitment to evaluating targeted social programs originating in the 1994–2000 administration (with the 1997–2000 impact evaluation of the Oportunidades Program) and the 1997–99 legislature (which made yearly evaluations obligatory for targeted programs). An evaluation of the cost-efficiency of public social services is beyond the scope of this study, but we offer some preliminary evidence to motivate such a study in the future. 2.112 Table 2.26 presents results from a standardized academic achievement test (language and mathematics) applied by SEP on a representative sample of primary and lower-secondary schools since 1997. The results were only made publicly accessible by the newly created National Institute for Educational Evaluation (Instituto Nacional de Evaluación Educativa) in 2003. A significant gap is evident between public and private schools in the urban sector, between public schools in the urban and rural sectors, and within the rural sector. Although these differences may reflect, in part, differences in socioeconomic determinants as well as school quality, at the minimum basic public education services clearly have failed to provide the necessary compensation for these socioeconomic determinants. Table 2.26: Students with "Sufficient" Academic Performance in Primary Schools, Estándares Nacionales (2000) National average 50.9% Urban private 70.8% Urban public 54.1% Rural public 42.3% Communitarian 37.8% Indigenous 25.3% Source: SEP (2002), p. 22. 2.113 On the cost side, it is interesting to analyze the evolution of public spending on basic education in the past three decades. As noted earlier, public expenditure per student in basic education expanded considerably over the past decade. The antecedent to this was the relative neglect of basic education in the previous two decades, when spending shifted in favor of upper- secondary and tertiary education (following the 1968 student revolt). Is it valid to interpret the level of spending per student as an indicator of the quality of education received? This would appear to be true in earlier decades, when spending per student on primary education in Mexico lagged behind such spending in all regions in the world in relative terms; in absolute terms, such spending was only lower in Sub-Saharan Africa and South Asia (Table 2.27). This was reflected in the number of school hours and also in the quality of those school hours. However, it is less evident that the rapid increase in spending per student at this level over the past decade was - 63 - translated into an equivalent increase in quality. This increase was absorbed principally by rapid growth in both the number of teachers and teacher salaries (Figure 2.17), which absorb 90 percent of public spending in basic education (Figure 2.18). There are large and increasing numbers of “teachers” who are not actually in classrooms, and their salaries are high relative to per cpita GDP (see Figure 2.19). Table 2.27: Public Spending per Student in Primary Education ($US PPP 1985) % GDP School hours Region 1960–90 per capita per year Mexico 175 4 780 Latin America and Caribbean 256 9.1 952 Middle East and North Africa 404 13.4 944 Sub-Saharan Africa 143 16.6 1026 East Asia and Pacific 295 9.3 1097 South Asia 101 9.1 981 Centrally planned economies 774 24.3 845 Developing countries 251 12.7 977 OECD 1656 15.7 974 Source: Scott (2003). Data from Barro and Lee (1996). Figure 2. 17: Student-Teacher Ratio and Public Spending per Teacher in Primary Education, 1970–2000 50 180 Students/Teacher spending per teacher (thousand MxP=2000) Spending/Teacher 160 45 140 40 120 students/teacher 100 35 80 30 60 40 25 20 20 0 1970 1972 1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 Source: SEP 2000. - 64 - Figure 2.18: Teacher Salaries in Basic Education (% current public education expenditure) Mexico Brasil Colombia Korea Bolivia Honduras Argentina Guatemala USA UK Uruguay Peru 40 45 50 55 60 65 70 75 80 85 90 Source: UNESCO. Figure 2.19: Ratio of Salary after 15 Years of Experience to GDP per Capita, 2000 2.5 2.0 1.5 1.0 0.5 0.0 Norway New Zealand Uruguay4 Mexico Sweden Germany Hungary Chile China Turkey Italy Russian Federation Paraguay Brazil United States Czech Republic Korea Spain England Australia Peru2 France Finland Argentina Indonesia Note: Annual statutory teachers’ salaries in public institutions at starting salary, after 15 years of experience and at the top of the salary scale, by level of education, in equivalent U.S. dollars converted using purchasing power parity. Source: OECD (www.oecd.org/els/education/eag2002). - 65 - 2.114 In the case of health, analyzing the costs and impact of services is even more challenging than in the case of education, given the heterogeneity and complexity of services delivered. However, it is worth noting that beyond high labor costs (71 and 74 percent in the insured and uninsured sector, respectively),29 the benefits gained by health-sector workers, especially in the social security institutions, suggest that the providers of services receive a disproportionate share of the of benefits from public spending. As noted earlier, the pension benefits for workers (mostly health) in the IMSS are far more generous than for the private sector workers covered by IMSS. Benefits, Taxes, and Tax Reform 2.115 Comparing the distribution of total benefits with the distribution of total tax burdens allows us to consider the net impact of the tax-transfer fiscal system. Given that we are comparing taxes with benefits mostly in kind, their aggregate impact should be interpreted with caution. This comparison is nevertheless useful for analyzing the full redistributive impact of possible tax reforms and for considering how the revenue will be used. 2.116 Both economic theory and international experience suggest that redistributive impact is strengthened by designing taxes to maximize the amount of revenue available to finance redistributive transfers rather than by enhancing the progressivity of the tax instruments themselves. For example, Stern (1987) shows under reasonable assumptions that the optimal tax structure is a linear income tax at a constant marginal rate, which becomes negative under a certain level of income. Similarly, taxation of consumption is justified for redistributive purposes, if the revenues go for spending that is progressive. 2.117 The case for designing the tax system as a source of revenue to finance redistributive spending is especially evident in the context of high initial income inequality, as in Mexico (Engel and others 1998). In this context, a strong redistributive impact is practically guaranteed by the distribution of income. This is true even if overall public expenditure is not targeted to the poor but is equal for all income levels, as is approximately the case in Mexico. It also is true if the overall tax rate is flat, although in Mexico the rich do pay a higher percentage of their income in taxes than the poor. 2.118 The same logic applies when considering the reduction of a general consumptions subsidy (think of it as a negative consumption tax) as a way to finance expenditure that is equal per capita or targeted to the poor. Given Mexico’s highly unequal income distribution, generalized consumption subsides represent particularly inefficient redistributive instruments even when applied to “basic” goods like food, medicines or residential electricity. The creation and rapid expansion of Progresa/Oportunidades in 1997–2002 was largely financed by the concurrent reduction and elimination of untargeted (and thus regressive) consumption food subsidies (tortilla) and reduction of other badly targeted (urban) food programs. 2.119 To illustrate this point with a simple example, Table 2.28 compares two polar tax policies: a highly progressive but inefficient tax with an incidence of 10 percent on the top decile 29. Boletín de Información Estadística, SSA. - 66 - only and a highly efficient but flat tax with an incidence of 10 percent on all deciles. In these two cases, we assume that tax revenues are redistributed as average public spending (table 2.19). The last column shows a third case with a flat tax from which 20 percent is spent as Oportunidades Program transfers (and the rest is spent as average public spending). Inequality as measured by the Gini is reduced more by the flat tax and still more if part of the resources are redistributed through the Oportunidades Program. The most dramatic contrast is observed in poverty reduction: the flat tax increases the income of the extreme poor (first quintile) by 55 percent, doubling the gain obtained for this group through the highly progressive tax and tripling it by redistributing 20 percent of the revenue through Oportunidades. Table 2.28: Efficient versus Progressive Taxation Autonomous Progressive tax Efficient tax (10% all) Decile income (10% decile 10) Average public 20% through expenditure Oportunidades Program Distribution 1 1.3 1.7 2.2 2.8 2 2.5 2.8 3.2 3.5 3 3.4 3.8 4.1 4.1 4 4.5 4.9 5 5 5 5.6 6 6.1 6 6 7.0 7.4 7.3 7.1 7 8.6 9.1 8.8 8.7 8 11.3 11.7 11.2 11 9 16.1 16.5 15.6 15.4 10 39.7 36.1 36.6 36.4 Gini 0.494 0.458 0.445 0.432 Net impact (% change in autonomous income) 1 32.4 71.7 119.9 2 15.5 29 40.7 3 11.3 18.4 20.2 4 8.4 11.1 11.7 5 7 7.7 5.9 6 5.6 4.2 2.3 7 4.8 2.2 0.1 8 3.5 -1.1 -2.8 9 2.7 -3.3 -4.5 10 -9.1 -7.7 -8.1 Note: Population deciles are ordered by income per capita. Tax revenue is redistributed as average public spending (table 2.19), except for the last column, where 20 percent is assumed to be distributed as Oportunidades transfers (and the rest as average public spending). 2.120 Table 2.29 compares the distribution of benefits with the current tax structure in Mexico30 and with the tax structure implied by generalizing the 15 percent VAT rate to all goods. Once again, this represents a shift to a (marginally) less progressive but more efficient overall tax structure. We assume that the effect is to increase tax revenue by 1.32 percent of GDP, which is more conservative than the SHCP fiscal expenditure estimate for VAT of 1.89 percent (see CIDE-ITAM (2003) for the reasons for the difference between the two estimates). 30. This analysis is presented in terms of household deciles because this is how the information on the distribution of tax burdens is available on a consistent basis from SHCP (Sub-Secretaría de Ingresos). - 67 - Table 2.29: Benefits and Taxes: Net Impact Distribution Net impact:** Taxes* % change in autonomous income Current VAT Generalized VAT Benefits* Current Generalized Average Average 20% through Decile VAT VAT public public Oportunidades expenditure expenditure Program 1 14.6 -0.1 0.1 81.4 93.8 100.3 2 11.7 0.9 1.2 33.8 38.4 39.6 3 11.1 1.7 2.0 23.1 26.0 26.2 4 10.8 2.5 2.9 16.1 17.9 17.6 5 9.2 3.6 4.1 9.6 10.3 10.0 6 9.8 5.0 5.6 6.9 7.1 6.7 7 9.4 7.1 7.8 2.6 2.1 1.8 8 8.6 11.1 11.6 -2.4 -3.4 -3.7 9 8.8 17.7 17.9 -6.3 -7.6 -7.8 10 5.8 50.5 46.7 -15.8 -16.9 -17.0 Note: Household deciles ordered by income per capita. *Taxes exclude social security contributions and benefits exclude pensions and in the case of public spending on IMSS health services only federal transfers are included. Total tax revenue is adjusted to coincide with total public spending. **Spending of additional resources distributed as average public spending (first column), except for last column, where 20 percent is assumed to be distributed as Oportunidades transfers. 2.121 The net impact is again to favor the poor. Without targeting through Oportunidades, however, the gain from additional expenditure obtained by the poor is comparatively small and could become negative once we take into account the asymmetry between the two types of instruments: VAT taxes reduce monetary resources with relative certainty (as informality is already taken into account in the present estimate), while the value of the compensating gain may be less if obtained through public services. For this reason, some compensation through well- targeted monetary transfers would help ensure that generalization of the VAT is not regressive in its impact. 2.122 Another, even more progressive scenario would be to eliminate the residential electricity subsidy, which now gives only about 8 percent of its benefit to the lowest 20 percent of the population, and shift these resources to a targeted program like Oportunidades. If Oportunidades were expanded with the same pattern of coverage as now, 60 percent of the benefit would go to the lowest 20 percent of the population. Focusing on increasing the coverage of the poorest would have even more progressive effects.31 The cost of the residential and agricultural electricity subsidies in 2002 was almost three times as large as Oportunidades, so even if only a part of the subsidies were recovered into the general budget, it would make a large opportunity for progressive redistribution. 31. See World Bank, Mexico Poverty, 2004, for discussion of other scenarios. As noted in Chapter 1, reducing the electricity subsidy would have its direct impact in improving the finances of CFE, allowing it to make more profits, which would partly go for reinvestment and partly come to the central government as fungible funds, usable in Oportunidades or elsewhere. - 68 - POLICY RECOMMENDATIONS 2.123 Mexico’s public expenditure programs vary widely in their degree of progressivity or regressivity, which partly reflects differences in objectives and constraints, but it also suggests opportunities for reforming the allocation of public expenditure. In general, the equity of public spending overall may be improved by expanding the participation of the poor in the programs or by reallocating public expenditure in favor of more progressive programs. Both kinds of changes contributed to the gains in progressivity achieved in public education and health services over the past decade. Some reallocations involve the complete elimination of old programs and the creation of new ones. This also occurred in the past decade, as illustrated by the elimination of the generalized tortilla subsidy and creation of programs like Oportunidades and Procampo. 2.124 In the case of competing (potential substitute) programs that pursue similar objectives using alternative instruments, as for basic health care, resources should be allocated to the most equitable alternative. In the case of non-competing (or complementary) programs, like advanced healthcare and higher education, access should be made more equitable. 2.125 The recent expenditure innovations documented here suggest that it is possible to overcome rigidities in the allocation of public spending, such as with the shift of food subsidies from the urban to the rural sector in the mid-1990s and by the redefinition of electricity tariffs (reducing some subsidies) in 2002. 2.126 Indeed, the lack of regressivity in overall public expenditure is a recent achievement. In education, where the gains came gradually over many years, the older generations of Mexicans still carry the burden of past misallocations, manifest in the exceptional degree of educational inequality of the adult population. 2.127 Finally, the distribution of marginal benefits may offer a more accurate guide to policy than the distribution of average benefits. As was the case with lower-secondary education in the early 1990s, the average incidence of upper-secondary education today underestimates the progressivity that would result from future expansions in coverage and expenditure. SPECIFIC PRIORITIES 2.128 Expanding coverage of post-basic education and reforming financing in tertiary education. The participation of the poor in tertiary education is low compared to other Latin American countries. The poorest 40 percent of the population accounts for only 9.6 percent of the students in tertiary education in Mexico, for 19 percent in Brazil, and for 39 percent in Chile. See Scott (2002, Table 6.7). Given that the largest economic returns at present are achieved at the latter educational level, this should be a first priority in reforming the allocation of public spending to achieve growth as well as equity. The participation of the poor in tertiary education will become more likely in the near future simply as a result of the rising trend in completion rates of secondary education by the poor. To ensure and accelerate this process, however, will require targeted scholarship programs designed to finance the opportunity costs for the poor. Given the costs of such programs and the limited fiscal resources available, this would require a reform in the financing of higher education in Mexico. Broadly, this would involve a gradual shift from the current system of statutory free and theoretically universal provision, to effective - 69 - universal access implemented through targeted scholarships financed in part by user fees for upper income groups. 2.129 Expanding coverage and reforming social security. In the case of IMSS, two principal challenges remain after the 1997 pension reform. First, the capacity of IMSS to provide health services to its beneficiaries is threatened by the growing pension liabilities of its own workers. Secondly, coverage is restricted by formal sector employment, leaving most of the general population in the lower half of the income distribution without access to pensions or medical insurance. In the case of ISSSTE, pension reform presents a challenge to those who would make public expenditure more progressive. Clearly the benefits of ISSSTE pensions mostly go to upper-income persons, more unequally distributed than even overall income, but that is a sunk cost of how public employees have been compensated. In its present form the system is getting more costly every year, requiring ever more current subsidies and building undisclosed contingent liabilities. Any pension reform has a substantial transition cost, but in the long-run the ISSSTE reform would reduce pressure for tax increases and for cuts in other (more progressive) spending programs. Also, with the contribution-defined benefits and individual accounts in the recent reform proposals, the cost of any future increment to the pension benefit will be more transparent. 2.130 In general, two principal lines of reforms must be pursued to achieve equitable access to social security: a) integrating the currently privileged social security schemes benefiting public sector workers (including Pemex, electricity and university workers, as well as IMSS employees and ISSSTE rightholders) into the reformed IMSS pension law, and b) increasing the coverage of IMSS in the lower deciles. The latter would require significant reductions in the cost of formal sector employment in Mexico, which in addition to social security contributions for health insurance and pensions, imposes direct taxes and restrictive labor laws. This implies reforms on all three fronts (social security, tax, and labor laws). Within social security, this could involve decoupling the health (and employment and life) insurance from the pension parts, increasing federal government contributions for poorer workers, and unifying the public health care sector as described in the next paragraph. 2.131 Unifying public health care provision and financing. The expansion of Progresa- Oportunidades in a short period of time to become the principal scholarship and food aid program in Mexico, absorbing resources previously allocated through a variety of uncoordinated, inequitable and inefficient programs in these two domains, illustrates the possibility of reallocating resources of competing programs to a single equitable and efficient allocation scheme. Even more important gains in efficiency and progressivity could be achieved by unifying public health care provision with progressive financing schemes. The recently created Seguro Popular program represents an important first step in this direction. 2.132 Deepening coverage of education, health and food programs in remote rural areas. We have seen that despite their progressivity at the national level and strong rural presence, even SSA services and Oportunidades transfers fail to reach many of the poorest of the poor in the rural sector. The progressivity of rural primary education services (comparable to Oportunidades) offers a further vehicle to target resources effectively for improvements in quality, nutrition, health, etc. Deepening the reach of the three programs to the poorest rural populations would require a coordinated effort. The Programa de Apoyo Alimentario will start - 70 - operating in 2004 and represents an initiative in this direction; its effectiveness in overcoming the logistical problems to reach remote areas remains to be seen. 2.133 Securing quality and cost-efficiency of basic education services and of health services accessible for the poor, by increasing transparency and accountability. With the spending on education and health per user being greater than initial income in the lowest deciles (over twice as large for the rural poor), funding the programs with any tax that is close to proportional with income will greatly favor the poor, if the parts of the programs serving the poor have quality and funding approximately equivalent to what the non-poor get. The geographic distribution of funding per user is not biased against poor states for education and only marginally against them for health. Realizing the redistributive potential of public education and health spending would require fundamental policy reforms to ensure the cost-effectiveness of the services benefiting the poor. Expanding the coverage of services without quality-control measures could be achieved at the price of lowering average quality of services given that facilities and personnel are fixed in the short run. Furthermore, effective participation in higher education services by the poor requires that they receive a sufficient quality of basic education in order for them to attend and benefit from higher education, i.e., to compete with upper income groups in terms of academic achievement. 2.134 In general, securing efficiency in the public provision of education and health services requires setting up mechanisms to increase transparency in these sectors, and then using this information to secure effective accountability through relevant incentive-schemes or by reallocating control from the providers to the beneficiaries (for example, through demand-side subsidies). There has been clear progress in the first part of this strategy in the current administration through such initiatives as the creation of the INEE, the quality and cost indicators and reports published by both SSA and IMSS over the last two years, and the data bases and independent evaluations developed by Sedesol for the targeted programs. Applying this newly gained information to make the programs accountable through adequate incentive mechanisms will still require further reforms. This is particularly urgent in reference to the decentralization of social spending. In contrast to the trend of increasing transparency and accountability achieved for federal programs, the decentralization process in health, basic education and anti-poverty funds has mostly decreased effective accountability in the use of the transferred resources. 2.135 Continuing the reform of electricity tariffs to reduce subsidy for upper income groups. Since the 2002 tariff reform the rate for the highest consumers is unsubsidized, if not positively taxed. But this applies only to the top percentiles of the income distribution. The subsidy on at least 95 percent of residential consumers is still substantial and regressively distributed. There is also a wide range of consumption allowed for the subsidized rates, especially in areas using heavy air conditioning. Since there is no clear justification for this subsidy beyond facilitating access to electricity for poorer households, there is a case for continuing reforming the tariff structure, gradually reducing the subsidy throughout the upper half of the income distribution. 2.136 Making agricultural support progressive. The gradual reduction of the Procampo budget since its creation in 1994 is consistent with the original design of this program to facilitate the transition to the full liberalization of agricultural markets by 2008. Given the - 71 - historically unprecedented success of Procampo in reaching poor subsistence farmers and agricultural workers, however, there is a strong case on redistributive grounds for maintaining a similarly designed agricultural support program, but targeted exclusively to small producers, and incorporating conditionality elements on the productive investment of part of the transfers received to ensure a permanent as well as current redistributive impact. 2.137 Tax reform. In countries with highly unequal income distribution, like Mexico, spending that is close to equal per capita across income groups in absolute terms becomes extremely progressive relative to autonomous income. This means that taxes that are close to equal as a share of income, or even mildly regressive in relative terms, make the overall system of public finances strongly redistributive in favor of the poor. To apply the same logic to evaluating tax policy changes, one must compare the estimate of the incidence of a new tax with the incidence of additional expenditures that the extra funds would finance. While such calculations are technically straightforward, it is politically difficult to make commitments about how the money will spent, while avoiding legalistic rigidities that undermine efficiency and accountability for quality of service delivery. That is one of the institutional challenges addressed in Chapter 4 of this report. 2.138 Also, one must consider the asymmetry that taxes reduce monetary resources with relative certainty, while public services provide a compensating gain that may be less than their cost. For this reason, some compensation through well-targeted monetary transfers may be needed to ensure that an important tax change (like broadening the base of the VAT) is not regressive in its impact. - 72 - 3. THE GEOGRAPHIC DISTRIBUTION OF PUBLIC SPENDING 3.1 This chapter analyzes the geographic distribution of federal spending, including both resources transferred to subnational governments (SNGs) and resources spent by the central government for the primary benefit of particular localities. It aims to answer questions about distributional equity and incentives for SNG behavior: • Are states and municipalities with poorer populations being treated fairly? • Are states and municipalities with large populations being treated fairly? 3.2 The chapter examines these questions in the aggregate and for the main sectors where the federal government spends directly on a geographic basis or distributes earmarked transfers to the subnational governments—education, health, agriculture, water, social investment, and the safety net.32 It also considers the geographic distribution of off-budget subsidies in the form of below-cost pricing for electricity and water to farmers and residential customers. TOTAL SPENDING 3.3 Since taxing power was centralized in the middle of the twentieth century, almost all of the resources available to Mexico’s states and municipalities (except for the Federal District) have come from the federal government. Thus the distribution of SNG resources is an active concern of the federal government, which seeks to assure that states can meet their spending obligations, especially in executing social programs decentralized since the early 1990s, and that subnational governments have appropriate incentives for efficiency, fiscal discipline, and tax effort. 3.4 The transfers from the federal level come in two main forms: discretionary participaciones and sectorally earmarked aportaciones. Participaciones—the 20–22 percent of federal domestic tax revenue that is shared with all the states and municipalities—have accounted for almost 40 percent of state and municipal revenue in recent years. Their advantage from the state’s point of view is the complete flexibility in their use. Aportaciones in the nine funds of Ramo 33 constitute almost 45 percent of state revenues, and more for most municipalities. Some are automatically transferred to the municipalities; the largest ones are transferred to the states and are earmarked for payrolls in education and health. 3.5 The volume and distribution of these decentralized resources changed substantially over the past decade, although the total amount did not increase from 2001 to 2002 (see appendix Figure A3.0). The subnational governments received transfers equal to about twice as much per capita in real terms in 2002 as in 1992, before the major sectors began to be decentralized (1993) and the last main revision of the participaciones formula was finalized (1994). Figure 3.1 shows 32. A preliminary examination of the geographic distribution of spending on roads determined that a more sophisticated analysis would be necessary, taking account of network effects, terrain, and population concentration. Such analysis is beyond the scope of this report, but will be taken up in a forthcoming Infrastructure PER. - 73 - the total resources going to states in 2002; Figure 3.2 shows the closest equivalent data for 1992, excluding the federal investment and states’ own revenues. The states are ranked by the National Council on Population (CONAPO, Consejo Nacional de la Población) index of marginality for 2002, which is the best available state-by-state indicator of poverty levels.33 Figure 3.1: Geographic distribution of per capita resources, 2002 16,000 14,000 12,000 2002 MxP per capita 10,000 8,000 6,000 4,000 2,000 0 México Oaxaca Tlaxcala Nayarit Puebla Yucatán Colima Veracruz Morelos Guerrero Hidalgo Tabasco Sinaloa Sonora Jalisco Michoacán Durango Coahuila Chiapas Campeche Querétaro Chihuahua Nuevo León Guanajuato Quintana Roo Baja California National Average Zacatecas Tamaulipas San Luis Potosí Baja California Sur Distrito Federal Aguascalientes States own revenues Participaciones Aportaciones Municipalities own revenues Sectoral Spending Source: INEGI and SHCP data and World Bank calculations 3.6 Although there is a lot of inequality between states in the resources transferred to their governments and in the direct spending of the federal government, the inequality in states’ public resources per capita is not correlated with poverty levels. Indeed, the relative position of the poorer states (higher indexes of marginality) has improved since 1992, so that, on average, they receive about as much as the national average. At all levels of marginality, the variations depend partly on the states’ population size. As judged by size of population, the five smallest states have per capita resources well above average, while the large states have less than average (except the Federal District). 34 33. Unless otherwise specified, all the figures use this ranking. The household survey (ENIGH) does not have sufficient observations to tell the poverty situation in each state. The marginality index, based on the census which has high statistical coverage for each state, reflects the percent of households lacking access to piped water and sanitation, having dirt floors, etc. It is a proxy for poverty. 34 The resources for the DF include direct federal spending for teachers, which is shown as part of aportaciones, in order to be compatible with the representation of other states. In Nayarit the state's own revenues are unusually high in 2002 because of some large improvement fees (mejoras). - 74 - Figure 3.2: Geographic distribution of per capita resources, 1992 6,000 aportaciones participaciones 5,000 own revenues 4,000 2002 MxP per capita 3,000 2,000 1,000 0 National Average Aguascalientes Nuevo León Estado de Mexico Campeche Distrito Federal Hidalgo Morelos Colima Baja California Guerrero Querétaro Quintana Roo Veracruz Chihuahua Durango Baja California Sur Sinaloa Coahuila Zacatecas San Luis Potosí Yucatán Puebla Sonora Chiapas Nayarit Tamaulipas Tabasco Oaxaca Tlaxcala Michoacán Jalisco Guanajuato Source: INEGI 1992, SHCP, and World Bank calculations. 3.7 These outcomes resulted from many separate decisions, each of which considered various sectoral concerns as well as the horizontal distribution across states. Most of the inequalities originated from the pattern of state taxation before the 1980 Fiscal Pact, which eliminated almost all remaining state tax bases and set the initial participaciones formula according to states’ own revenue foregone, and from the distribution of teachers in federal schools before 1993, when federal schools and money to pay teacher salaries were transferred to state authorities. States with high per capita tax collection before 1980 or well-funded federal school systems (often because they had not set up their own state school systems before 1993) get the most per capita transfers today. 3.8 Investment done directly by the federal government but located in the states to benefit their populations (roads, water, and so forth) declined in real pesos per capita by about 20 percent during the 1990s and remained highly unequal across states. It is positively correlated (0.62) with the transfers per capita to state governments, so it accentuates rather than offset the inequalities of the cash transfers. 3.9 Switching the geographic analysis from spending per capita to spending as a share of state GDP changes the picture dramatically (see Figure 3.3). The five states with the highest poverty rates receive public resources equivalent to 15–23 percent of state GDP. At the other end of the distribution, the eight states with the highest income per capita (except Baja California Sur, with very low population) receive below-average spending resources as a share of GDP— around ten percent or less. This shows that in low-income states the public sector is much more important than the private sector. And, of course, there are many causal relations between these facts. Since the poorer states also have a larger informal sector, public employment and contracting dominate the formal economy in marginal localities. Conversely, the large and more formal private sector in the upper-income states means that there are employment opportunities - 75 - in the private sector, that many talented people are available to take public posts for a few years and then return to corporate careers, and that the tax base is substantial relative to the funding needs for public services. Figure 3.3: Geographic distribution of public resources, as a share of state GDP, 2002 40 35 30 25 % State GDPI 20 15 10 5 0 México Oaxaca Tlaxcala Morelos Nayarit Guerrero Veracruz Tabasco Michoacán Yucatán Colima Sonora Puebla Campeche Querétaro Durango Jalisco Chiapas Zacatecas Hidalgo Sinaloa Coahuila Nuevo León Guanajuato Quintana Roo Tamaulipas Chihuahua National Average San Luis Potosí Baja California Distrito Federal Aguascalientes Baja California Sur States own reveneues Participaciones Aportaciones Municipalities own reveneues Sectoral Spending Source INEGI (2003), SHCP data and World Bank calculations 3.10 Overall then, rich states, like rich households, pay more resources into the fiscal system per capita and, on average, receive similar expenditure benefits per capita. The difference in resources received sheds light on the magnitude of interstate resource flows. If the states all contributed equal shares of their GDP to the national revenue pool for geographically distributed spending—the average spending of 8.5 percent of GDP is also the average contribution—then the richer states would contribute at least 1–4 percent of their GDP for spending in other states. Since taxation is probably progressive, the contributions for redistribution may be even larger. The poor states receive two or three times as much from the national fiscal system as the resources they put into it. 3.11 The underdevelopment—some would even say the misdevelopment—of the poorest states in Mexico highlights the problems and unfairness in the federal system. The primary problem, however, is not with the total volume of resources but with how they are used. This was one of the main finding of the World Bank’s study of a Development Strategy for Mexico’s Southern States. While recognizing that fully understanding those problems requires in-depth sectoral analyses, which are beyond the scope of this report, the rest of this chapter presents the geographic distribution of some of the main spending programs. - 76 - ANTI-POVERTY PROGRAMS 3.12 The regional analysis of targeted anti-poverty spending includes the Human Development Program Oportunidades and the state and municipal portions of the Fund for Social Infrastructure of Ramo 33 (FAIS, Fondo de Aportaciones para la Infraestructura Social). In 2002 anti-poverty spending amounted to 368 pesos per capita nationwide: 207 pesos per capita through FAIS and 161 pesos through Oportunidades. Overall, the regional distribution of spending for these two programs is clearly redistributive, as the poorer states receive more resources per capita than the better-off ones. Figure 3.4 shows this trend. The 13 poorer states receive more resources than the national average, while the 19 richer states receive less. Figure 3.4: Geographic distribution of targeted anti-poverty spending, 2002 1,200 1,000 2002 MxP per capita 800 600 400 200 0 Chihuahua Aguascalientes San Luis Potosí Distrito Federal Quintana Roo National Avg. Tlaxcala Coahuila Baja California Sur Nuevo León Tamaulipas Michoacán Yucatán México Chiapas Guanajuato Sinaloa Jalisco Baja California Oaxaca Durango Morelos Guerrero Puebla Nayarit Querétaro Hidalgo Tabasco Veracruz Zacatecas Colima Sonora Campeche FAIS OPORTUNIDADES Source: SHCP data and World Bank calculations. 3.13 In 2002 the three poorest states—Chiapas, Oaxaca, and Guerrero—received 14, 13, and 12 times more pesos per capita, respectively, than the three richest states—Nuevo León, Baja California, and Federal District. Both FAIS and Oportunidades have this redistributive tendency. For FAIS, the national average spending in 2002 was 206 pesos per capita, while Chiapas received the highest per capita spending (583 pesos per capita) and Nuevo León, the richest entity after the Federal District (excluded from the fund), received only 51 pesos per capita.35 We do not know how well the FAIS funds are targeted to poverty-reduction programs within the states. 35. While this graph addresses the question of the aggregate reallocation of resources via these programs toward lower-income states, it does not compare how well the poor are provided for in different states. For this we would need to look at resources per poor person, dividing the allocation of money by the number of people in the bottom quintile of the national income distribution or the number below the extreme poverty line. The ENIGH sample is not large enough to be representative at the state level, however, so this calculation is not possible. - 77 - Education 3.14 In Mexico, the states have carried out most public education spending since 1993, with funding mainly through two federal budgetary sources: the education segments of budgetary item 33 (Ramo 33), and the part of the education item (Ramo 11) that the Ministry of Public Education (SEP, Secretaría de Educación Pública) spends directly in the states. Education spending in Ramo 33 comprises the Fund for Basic Education (FAEB, Fondo de Aportaciones para la Educación Básica y Normal), the Fund for Technical and Adult Education (FAETA, Fondo de Aportaciones para la Educación Tecnológica y de Adultos), and the Multiple Fund (FAM, Fondo de Aportaciones Múltiples). FAETA includes the resources of both the National Institute for Adult Education (INEA, Instituto Nacional para la Educación de los Adultos) and the National College of Professional and Technical Education (CONALEP, Colegio Nacional de Educación Profesional Técnica), while FAM includes the infrastructure funds of both basic and upper education. 3.15 Through Ramo 11, the federal government spends one part directly and distributes the other part to the states. The part that is distributed geographically reaches the states through various sources: the National Council for Educational Development (CONAFE, Consejo Nacional de Fomento Educativo)36; mid-upper and higher education that is integrated by decentralized organisms of the states and by federal education institutions;37 INEA; the National Council on Science and Technology (CONACYT, Consejo Nacional de Ciencia y Tecnología); SEP-CONACYT centers; the National Council for Culture and the Arts (CONACULTA, Consejo Nacional para la Cultura y las Artes); the National Sports Commission (CONADE, Comisión Nacional del Deporte); the Mexican Institute for the Young (IMJ, Instituto Mexicano de la Juventud); the National Institute of History and Anthropology (INAH, Instituto Nacional de Antropología e Historia); and the National Institute of the Arts (INBA, Instituto Nacional de Bellas Artes). Another source of education spending is the part of budgetary item 25 (Ramo 25) that covers the salaries of teachers in the Federal District. Up to 2001 the Program for the Strengthening of Federal Entities (PAFEF, Programa para el Fortalecimiento de las Entidades Federativas) in Ramo 23 also included an allocation for education spending. 3.16 Education spending per capita is not evenly distributed among the states; the funding per capita does not strongly correlate with the marginality. Figure 3.5 shows that states with very different marginality levels—like Chiapas and Coahuila—receive about the same amount of federal spending per capita, while states with similar marginality levels—like Mexico and Colima—receive very different funding per capita. About half of the states receive allocations close to the national average of 2,394 pesos per capita, but four receive much more than average—Baja California Sur (the highest, at 5,045 pesos), Colima, Campeche, and Nayarit— 36. CONAFE encompasses a group of five programs: community organization programs, compensatory programs, the Fund for Rural Education (FIDUCAR, Financiamiento Educativo Rural), the Oportunidades Program, and the Fund for Migrant Children. 37. The decentralized organisms of the states include CONALEP, the Center for Technical and Industrial Training (CETI, Centro de Enseñanza Técnica Industrial), (Colegio de Bachilleres), the centers for scientific and technological studies (CECYTs, centros de estudios científicos y tecnológicos), the technical institutes (institutos tecnológicos), and (ICT, ). The federal institutions include the Tecnológico-SEIT, UTE of Universidad Nacional Autónoma de México, and IPN-COFFA-UPEs-CINESTAV. - 78 - while four receive substantively less than the national average—State of Mexico (only 1,166 pesos per capita), Jalisco, Guanajuato, and Puebla.38 Figure 3.5: Geographic distribution of federal education funds, 2002 6,000 5,000 4,000 2002 MxP per capita 3,000 2,000 1,000 0 Guerrero Guanajuato Nayarit Querétaro México Veracruz Campeche Durango Morelos Quintana Roo Colima Nuevo León Oaxaca Tabasco Yucatán Sonora Hidalgo Michoacán Chihuahua Coahuila National Avg. Chiapas Tamaulipas Distrito Federal Puebla Sinaloa Tlaxcala Jalisco Zacatecas Baja California Baja California Sur San Luis Potosí Aguascalientes R25+R33 R11 Source: SHCP data and World Bank calculations. 3.17 These regional differences are present in the distributions of both Ramo 33 and Ramo 11, and, with a few exceptions like Zacatecas and Durango, the pattern of regional distribution is similar in these two budgetary items (see appendix figures Education A3.2 and Education A3.3). The similarity suggests that the resources of Ramo 11—with 35 percent of total state spending on education and presumably with more flexible allocation—are not used to compensate for the unequal distribution created by Ramo 33, which is allocated according to historical and inertial criteria. Overall, the regional distribution of federal financing for education is driven by inertia and political negotiations rather than by efficiency or equity. Some states did not have their own education system before 1993; the federal government ran their entire education system.39 Thus when equivalent resources were passed to the states in 1993 along with the responsibilities (and staffing) for primary and secondary education, these states got more money than many of the states that were already funding their own system, which operated in parallel with the federal system and covered significant parts of the educational system. Consequently the more proactive states were punished fiscally, and this pattern has continued for more than a decade without correction. 38. The Federal District is an anomaly, since the federal government funds basic education directly and two-thirds of the Federal District’s educational spending comes from Ramo 11, mainly for federal educational institutes such as UNAM. 39. The State Fund for Social Infrastructure (FISE, Fondo de Aportaciones para la Infraestructura Social Estatal) also raises the aportaciones of these states, for the level of poverty is one element in the distribution formula. - 79 - HEALTH 3.18 As of 2002, about half of the population had health insurance, mainly from the IMSS, the ISSSTE, and Pemex. For the uninsured population, the financing and provision of health care come from the Fund for Health Services in Ramo 33 (FASSA, Fondo de Aportaciones para los Servicios de Salud), Program for Expansion of [basic health-care] Coverage (PAC, Programa de Ampliación de Cobertura), the Oportunidades and IMSS-Oportunidades programs, resources from the health budgetary item (Ramo 12), direct spending of the states, and some minor health expenditures from the National Indigenous Institute (INI, Instituto Nacional Indigenista). To expand the coverage of health care insurance and to reduce catastrophic health expenditure, the government is pursuing a voluntary insurance scheme (Seguro Popular) to be managed by the Ministry of Health; the scheme would allow all uninsured Mexicans—about 55 percent now—to be covered by 2006 (World Bank 2003). If Seguro Popular grows strongly, as expected, the federal and states governments will need to reassess the funding and mandate for the open system. 3.19 In 2002 total health spending in Mexico amounted to 5.8 percent of GDP, of which 2.4 percent was spent by the public sector (including IMSS, ISSSTE, and Pemex) and the remaining 3.4 percent was spent by the private sector. In 2002 Mexico’s per capita public health spending was 1,322 pesos. Health spending per capita varies widely between the insured and uninsured population, with the insured population, on average, receiving 1,720 pesos per capita and the uninsured population receiving little more than half of that (910 pesos per capita). As a percentage of the total programmable public spending, the health sector received 14.6 percent of the total; 2 percent less than in 2001. This decrease took place in the past few years, and total public health spending is once again around the levels of 1994 (SSA 2003). 3.20 Overall, the richer states receive more federal health spending per capita than the poorer states. Figure 3.6 presents the real per capita spending for the insured and the uninsured populations, with states ranked according to their marginality levels. These figures were calculated as ratios to the insured and uninsured populations, respectively. The figure also shows total spending per population, which is the weighted average of the insured and uninsured spending per capita. Per capita health spending varies widely geographically, Fourteen states, mostly the poorer ones, receive less health spending than the national average of 1,320 pesos per capita. Excluding the Federal District and the State of Mexico, which receive the highest and lowest per capita health spending, respectively,40 the state that receives the most federal health spending per capita (Baja California Sur) gets more than three times more than the state that receives the least (Michoacán). The differences are even larger if we compare per capita spending for the insured and uninsured populations: the insured population in Baja California Sur receives more than five times as much per capita as the uninsured population in Puebla. 40. The Federal District and the State of Mexico are outliers for two main reasons. First, the Federal District’s health spending data count the spending of the central national facilities and of the highly specialized hospitals (La Raza, Centro Médico Siglo XXI, Hospital 20 de Noviembre), even though people from all regions use these facilities. Second, many people in the State of Mexico—mainly the metropolitan area that forms part of Mexico City—probably use the routine health facilities of the Federal District. Also, the State of Mexico has its own program for state employees (not included in these figures for lack of 2002 data), in contrast to most other states, which use ISSTE. - 80 - Figure 3.6: Geographic distribution of federal health spending, 2002 4,500 4,000 3,500 3,000 2002 MxP per capita 2,500 2,000 1,500 1,000 500 0 Oaxaca México Michoacán Tlaxcala Chihuahua Coahuila Nuevo León National Avg. Guerrero Hidalgo Puebla Guanajuato Sinaloa Querétaro Durango Quintana Roo Nayarit Baja California Chiapas Veracruz Morelos Baja California Sur Tabasco Yucatán Colima Sonora Jalisco Aguascalientes San Luis Potosí Campeche Zacatecas Tamaulipas Distrito Federal uninsured insured total Source: SSA (Coordinación General de Planeación Estratégica) data and World Bank calculations 3.21 The component that seems most equal per capita across states is IMSS spending. FASSA spending, spending from discretionary resources, and IMSS-Oportunidades spending (for the general population) show the most variation. For both the insured and the uninsured populations, the poorer states receive less per capita than the richer states. Agriculture 3.22 The proportion of state GDP in the primary sector is only loosely correlated (0.31) with total public spending in agriculture as a share of state GDP. This means that the states for which agriculture is an important activity do not necessarily receive more federal agricultural spending, since each program has its own objectives. Evaluating the quality of these programs is beyond the scope of this chapter, but their overall distribution seems inconsistent with an objective to help the majority of farmers, or of poor farmers. 3.23 There are many agricultural programs; and geographic data are available for almost all except the central adminstration (and in 2002 the bailout of Banrural). Those administered by the Ministry of Agriculture (SAGARPA, Secretaría de Agricultura, Ganadería, Desarrollo Rural, Pesca y Alimentación) include the Support for Commercialization Program (Apoyos a la comercialización), Program for Direct Support to Agriculture (PROCAMPO, Programa de Apoyos Directos al Campo), Rural Alliance (Alianza para el Campo), Temporary Employment Program (PET), and Integrated Program for Sustainable Agriculture and Productive Conversion in Zones of High Risk (PIASRE, Programa Integral de Agricultura Sostenible y Reconversión Productiva en Zonas de Siniestralidad Recurrente). Others include programs run by the Ministry of Agrarian Reform (SRA): Fund for the Promotion of Productive Projects of Agrarian Organizations (FAPPA, Fondo para el Apoyo a Proyectos Productivos de las Organizaciones Agrarias), Program for Women in the Agrarian Sector (PROMUSAG, Programa de la Mujer en - 81 - el Sector Agrario), Program for the Certification of Ejido Rights and Titles to Urban Lots (PROCEDE, Programa de Certificación de Derechos Ejidales y Titulación de Solares Urbanos), and judicial obligations as well as programs run by the Ministry of Environment and Natural Resources (SEMARNAT, Secretaría de Medio Ambiente y Recursos Naturales): the Development Program for Commercial Plantations (PRODEPLAN, Programa de Desarrollo de Plantaciones Comerciales), Forestry Development Program (PRODEFOR, Programa de Desarrollo Forestal), forestry fires, and irrigation or hydroagriculture infrastructure. 3.24 The main programs in agriculture: • PROCAMPO (Programa de Apoyos Directos al Campo) is the main Government program for agriculture. It is a delinked subsidy consisting of a lump sum per ha paid to farmers according to the number of has cultivated in the base period. The payment is of the order of US$80 per ha and per season. It was designed as a 15 years transition program to compensate grain (specially maize) producers for decreasing protection under NAFTA. Other short cycle crops have been incorporated into the program over the years. Existing evaluations indicate that it has a significant impact on the income of subsistence farmers, favoring not only increased consumption but also investment. Larger farmers, however, benefit proportionally more. A scheme to allow farmers to capitalize the subsidy has been introduced to encourage them to carry out farm improvement projects. The subsidy benefits owners as well as tenants. • ALIANZA PARA EL CAMPO is the Government’s farm modernization program. It consists of a collection of subprograms whose purpose is to promote the technological improvement of farms. The main ones are the “Agricultural Promotion,” “Livestock Promotion” and “Rural Development” subprograms. Although originally oriented to commercial farmers, the small and subsistence farming sectors have increased their participation over the years through an increase in the Rural Development subprogram. It is, however, the medium and large farmers who benefits most from Alianza. Available evaluations indicate a positive impact of the program on farm technology and productivity parameters. • ASERCA (Programa de Apoyos a la Comercialización) is a contingent price support subsidy to grain producers. The amount and coverage of crops and regions change each year. The program targets mostly regions where agriculture is highly commercial, and hence it tends to favor the better off farmers. • PET (Programa de Empleo Temporal) is an income support program which provides wage employment in marginal rural zones to build or improve productive and social infrastructure. It clearly benefits poor communities. • PIASRE (Programa Integral de Agricultura Sostenible y Reconversión Productiva en Zonas de Alta Siniestralidad) is a lesser program to prevent and cope with emergency situations due to recurrent crop failures. It operates in regions where agricultural risks are particularly high, promoting the sustainable use of natural resources in those regions. • FAPPA (Fondo Para el Apoyo a Proyectos Productivos de las Organizaciones Agrarias) is a Secretaría de Reforma Agraria (SRA) program to supports productive projects for dwellers of social sector villages (ejidos and comunidades) who are not community/ejido members and own no land. - 82 - • PROMUSAG (Programa de la Mujer en el Sector Agrario) is another SAR program to support the economic integration of women in ejidos and comunidades through the financing of productive projects for women groups. • PROCEDE P (Programa de Certificación de Derechos Ejidales y Titulación de Solares Urbanos) is the largest SRA program. It is a land titling program for the social sector which has the purpose of providing tenure security and operates in a voluntary basis. It gives three type of titles: individual occupancy certificates for family cropping parcels, collective certificates for communal use lands, and individual ownership titles for homesteads. • SEMARNAT has two main programs: PRODEFOR and PRODEPLAN. The former aims to improve forest management, promote forest development and diversify income sources of forest dependent populations by means of both technical and financial assistance. The latter program supports the development of commercial forest plantations with a subsidy of 65 percent of the installation costs for the initial 7 years 3.25 Richer states receive more agricultural program resources per capita of rural population than poorer states, as Figure 3.7 shows. In 2002 the national average spending per capita rural was 1200 pesos, but the seven states with highest marginality received only about 500 pesos per capita rural, while Sinaloa, Taumalipas, Sonora, and Baja California received 3-4,000 pesos per capita rural. (See appendix Figure A3.6 for the total per capita distribution.) Three programs represent more than 75 percent of the public spending for agriculture: PROCAMPO (41 percent), Rural Alliance (Alianza para el Campo) (18 percent), and irrigation infrastructure (17 percent). The four high outlier states are strong agricultural producers, and they receive by far the highest allocations of the Support for Commercialization Program per rural population, while the seven states with highest marginality and large rural populations receive virtually nothing under that program. If this program, which provides direct subsidies to the producers, is so good that it accounts for the agricultural success of the four main participant states, the government should consider extending it to the poorest states.41 3.26 The bias of federal policy in favor of agriculture in the middle and upper income states appears most strongly in relation to water. The most marginal 10 states all get less than the national average in federal irrigation spending per capita rural—all but one get lees than 15 percent of the national average. The subsidies for agricultural electricity usage, mainly for pumping irrigation water, are even more skewed, with six of the eight least marginal (richest) states getting over three times the national average subsidy per capita rural, while the 10 poorest states get less than the national average, mostly under one-tenth. (See Figure 3.8 and Appendix 1) Highly subsidized tariffs have contributed to over-extraction of aquifers and excessive withdrawals from surface sources, raising costs for all users. Water is wasted, and old technologies and production patterns are not replaced with more efficient ones. These subsidies only benefit those farmers using irrigation, and are concentrated in those who engage in modern, capital intensive agriculture. In regional terms, three states, Chihuahua, Guanajuato, Sonora, account for 46 percent of all electricity subsidies for irrigation. In contrast, states in poor regions have very limited irrigation and therefore benefit little from electric power subsidies. For 41. Conversely, if the program has not contributed substantially to the agricultural success of the main recipients, it should be cut back or dropped. - 83 - example, Chiapas received subsidies of a mere 44 million pesos during 2002; Oaxaca 18 million and Guerrero 3.6 million, which taken together accounted for less that 1 percent of electricity subsidies for irrigation in 2002. Figure 3.7: Geographic distribution of public spending for agriculture per capita of rural population, 2002 5,000 4,500 4,000 2002 MxP per rural population 3,500 3,000 2,500 2,000 1,500 1,000 500 0 Guerrero Guanajuato Nayarit Oaxaca Querétaro México Veracruz Campeche Durango Colima Morelos Quintana Roo Nuevo León Tabasco Yucatán Sonora Coahuila Hidalgo Michoacán Chihuahua National Avg. Chiapas Tamaulipas Puebla Tlaxcala Sinaloa Jalisco Zacatecas Baja California Baja California Sur Aguascalientes San Luis Potosí APOYOS PROCAMPO ALIANZA FAPPA PROMUSAG OBLIGACIONES PET PROCEDE PRODEPLAN PRODEFOR INCENDIOS INF. HIDROAGRICOLA INF. HIDRAULICA PIASRE Source: SAGARPA, SHCP and World Bank calculations. 3.27 For the payment of fees for using water—a nationally owned resource, according to the constitution—we do not know the quantitative details, but qualitatively there is clearly heavy use—often overuse for growing water-intensive crops—of irrigation water in the dry states, without payment of fees equal to the value of the water in the rest of the economy. Since well- entrenched agricultural interests are the biggest users of subsidized water and the benefit of the implicit water subsidy is capitalized into the value of their land, reform will be difficult. On the other hand, the industrialization of the north has created powerful interests of non-intensive water users who now see their further development seriously constrained by the water shortages resulting from present policies. - 84 - Figure 3.8: Geographic distribution of agricultural electric subsidies, 2002 2,000 1,800 1,600 2002 MxP per rural population 1,400 1,200 1,000 800 600 400 200 0 Guerrero Veracruz Campeche Guanajuato Nayarit Durango Morelos Oaxaca Quintana Roo Colima Tamaulipas Sonora Chihuahua Chiapas Hidalgo Puebla Tabasco Zacatecas Sinaloa Coahuila Distrito Federal National Avg. Tlaxcala Jalisco Baja California Sur Baja California Aguascalientes Source: CFE data and World Bank calculations. WATER 3.28 The regional distribution of water, sanitation and sewerage service (WSS) coverage and investment is highly unequal across states. Access and quality are discussed first, as they have better data than for spending. 3.29 Regional access and quality of WSS. Nationally and within states, access to water services is higher than that of sewerage; in 2002 10–20 percent more of the population has access to water supply than to sewerage, (Appendix Table A10.1). As one could expect, the rates of access to these services are inversely correlated with marginality levels across states. In states with high incidence of marginality, these differences are often substantial; for example in Guerrero, only 70 percent of the population has access to water and just 50 percent to sewerage, while in better off states like DF, Nuevo Leon and Aguascalientes more than 90 percent of their populations have access to water and sewerage services. 3.30 Households counted as having “access” actually receive a wide range of service levels and quality. For example, aggregate statistics include households with 24/7 service in the home and those who must carry water from public standpipes with only a few hours of service/week equally as “with access”. In Mexico, the incidence of lower levels of service across states compounds the regional inequalities in WSS access. Households in poorer states tend to receive inferior levels of service to those present the richer states. For water supply service, the levels of access include: (1) in-house tap, (2) on property tap, (3) public tap & connection to tap of other house, (4) tanker, and (5) wells/surface sources. Among households that do have access to water, those in poorer states tend to receive lower levels of service than better off states—in 2002, in poorer states, less than 40 percent of the population had in-house tap water, with less than 20 percent have to obtain water from of wells and surface sources (Appendix Table - 85 - A10.3b). In contrast, in richer states more than 80 percent of people have in-house tap water, and a very small proportion of the population have to resort to wells or tankers. 3.31 Regional disparities in the level of sanitation and sewerage services are even more pronounced than for water supply. The levels of service for sanitation include (1) standard water pressure toilet, (2) pour-flush toilet, (3) other (eg. latrines), and (4) none. For example, in poorer states like Chiapas, Guerrero, and Oaxaca, less than 40 percent of the population had access to standard water pressure toilets, and a significant proportion of their populations (+20 percent) did not have access to any sanitary facility (Appendix Table A10.3c). In richer states these proportions are reversed; in mid-income states, about half the population has standard water pressure toilets, and for the richest states this proportion reaches almost 75 percent. Access to different levels of sewerage services follows a similar pattern. The levels of sewerage service include (1) public network, (2) septic tank, (3) drainage ditches, (4) indirect discharge into a water body ( lake, river, or sea) and (5) none. In richer states, more than 80 percent of the population have their wastewater removed through the public sewerage network, while in poorer states less that 50 percent of households receive this level of service (Appendix Table A10.3d). Conversely, richer states have a small proportion of people with access to only low quality sewerage services, while in poorer states this is still the norm. 3.32 The extent of access varies by size of locality as well.42 In 2002, for water supply and sewerage, rural localities had access rates of 68 percent and 40 percent, medium-size cities—92 percent and 80 percent and large cities—98 percent and 92 percent respectively.43 With medium sized cities projected to grow most rapidly over the next decade, the access gap will grow there if recent trends continue. Within localities, a much higher proportion of poor households tend to receive lower or no service relative to non-poor households (Annex 3, Figure 103ai-iii). 3.33 The above analysis of the regional WSS access and service levels underscores several general trends: Access to WSS is has been regressive because in richer states, a higher proportion of their population has access than in poorer states, and that the similar regressivity is observed in the quality of the service received: households in better off states have access to higher levels of service than poorer states. Third, access to WSS services and the level/quality of those services is significantly lower in rural areas (defined as those with <2500 hab). Lastly, poor households tend to receive lower levels of service (including no service in the cases of sanitation and sewerage) than the average. 3.34 Investment spending on WSS. Reliable information which provides a comprehensive picture of spending on WSS is not generally available, despite strenuous efforts by CNA in this area. Water companies (which serve towns and cities) are of local jurisdiction and are not required by law or otherwise to disclose comprehensive financial information to federal agencies, nor to use generally accepted accounting principles in presenting such information. Similarly, municipalities and states which receive un-earmarked federal funds (for example, Ramo 33–FAIS) do not regularly and consistently report on the use of such funds, sectorally or functionally. And as indicated below, these sources of funding have grown in importance, 42. Information presented here based on “Patrones de Acceso y Calidad de Agua, Saneamiento y Alcantarillado en las Localidades Úrbanas de México”, J.L. Ávila Martínez, 2003, mimeo. 43. Rural=<2500 inhabitants, medium cities=2500-50,000, and large cities=>50,000. - 86 - particularly for states with a high incidence of poverty. Lastly a number of municipalities (wealthier with solid credit ratings) are increasingly relying on debt financing to fund investments and these funds are passed on to water companies and other agencies without regular reporting to federal agencies. CNA therefore assembles as detailed a picture of financing sources and uses as possible, relying on the triangulation of a number of partial sources of information. The following section is based largely on this source of information. 3.35 Total annual investment spending on water and sanitation infrastructure from 1995 to 2002 oscillated around of MXP 4,000 million (constant 2002 pesos ) which is low, considering the steady growth in population (and real incomes), the mounting backlog of rehabilitation and modernization of systems and networks, and demands for higher quality of services. Investment in the sector has risen substantially since 2002. Traditionally, sectoral investments have been financed through the subsidies of the federation and the states, bank borrowing, municipal resources and by the own revenue generation (GIC-generacion interna de caja) of municipal water companies (organismos operadores). During this period, total levels of investment and the mix of sources of financing varied considerably.44 3.36 As shown in Table 3.1 and in Annex Figure [A.10.5], total investment declined from 1995 to 1999, picked up in 2000. The 1997–2000 real increase is mainly explained by the lessening of the effects of the 1995 crisis. The sharp decline in 2001 was largely due to the phase down of the Program for Water Supply, Sanitation, and Sewerage in Urban Areas (APAZU). In 2001 the total funding for this program were half of what it got in 1995. The decline of 2001 was partially compensated the following year by an increase in the funding for the newly established Program for Modernizing Water Companies (PROMAGUA), the creation of a new financing Program for the Devolution of Water Fees (PRODDER)45 and a modest program to serve rural areas (PROSSAPY).46 Also noteworthy is the recent growth in credit as a source of WSS financing. The 2002 increase appears to be due to an expansion in municipal borrowing from BANOBRAS and private commercial sources. This reflects the impact of (Steve-name of policy/program requiring sub-sovereign credit ratings). In summary, after more than a decade of decentralization of responsibility to municipalities and their water companies, the sector remains dependant on federal subsidies. At the same time, in line with federal policies to shift support to the social sectors and block transfers to sub-national governments, the absolute level of federal financing and therefore aggregate sectoral investment has remained constant at best and is projected to decline over the coming years. 44. These figures do not include certain sources of funds which CNA has only recently began to track: eg. SEDESOL programs, CONAFOVI-tertiary network expansion financed by housing developers, investments financed by State Water Commissions and FAIS (FISM and FISE). 45. PRODDER is a program through which the federation devolves the water fees paid by municipalities for the extraction of national waters. The water companies use these funds to programs to improve efficiency and development infrastructure. PROMAGUA supplies grant resources to water companies to consolidate their systems, through the collaboration with private operators in executing program to improve operating efficiency and financial solvency. 46. Programa para la Sostenibilidad de los Servicios de Agua Potable y Saneamiento en las Comunidades Rurales - 87 - Table 3.1: Water and Sanitation Investment by financing source, 1995-2002 (Millions 2002 pesos) Año Federal Estatal Créditos Municipal GIC Total 1995 1,390 1,714 1,517 1,102 5,722 1996 2,302 676 98 334 3,410 1997 2,131 850 181 838 4,001 1998 2,456 651 296 349 3,753 1999 2,024 939 203 256 3,422 2000 2,377 1,479 383 119 4,359 2001 1,103 777 639 328 2,847 2002 1,685 1,006 1,514 695 4,900 Source: CNA and World Bank calculations. 3.37 The regional distribution of investment spending on water supply and sanitation has followed a similar pattern as access rates but with some notable exceptions as discussed below. Per capita federal and state spending in urban areas in poorer states was roughly one-third that of states with low levels of marginalization. Notable exceptions being Higalgo and Quintana Roo which were significantly higher and DF47 and Nuevo Leon which were much lower than their historical averages. Per capita spending in rural areas in 2002 was a small fraction of federal and state spending in urban areas and was rather evenly distributed across states, despite the large rural access “gaps” in the poorer states (Appendix Table 10.7.6). A growing, yet not precisely quantified proportion of Ramo 33 funds transferred to sub-national governments (principally to municipalities) has gone to basic infrastructure through FAIS-FISM and the allocation of these funds reflects a high degree of progressivity across states by degree of marginalization. However, this has only partially redressed the regressive incidence of overall federal (and state) spending across states-total spending at the state level remains highly skewed toward better off states (Appendix Table 10.7.2a). 3.38 While debt financing has grown significantly in very recent years as a source of financing (share of total financing: 5 percent 1995-1999 vs 30 percent in 2002) it is concentrated in a very limited number of states, particularly those states with MDB financed WSS programs (through Banobras) and (2) those with recent private concessions or similar arrangements where commercial debt from Banobras and private sources has been mobilized with backing from the federation and from Nadbank in the case of northern states. 3.39 Most localities and their water companies remain highly dependant on subsidies (ie transfers) reflecting their still precarious financial situation, itself a product of low levels of operational (technical and commercial) and investment efficiency, generalized underpricing of water services and high turnover of management. The median share of transfers in total financing across states has remained relatively constant at 75–80 percent over the recent past, of which federal financing accounted for two-thirds and state funding one-third. 47. Investment spending by DF not fully nor consistently reported to CNA over the past 5 years. - 88 - 3.40 Similar to the geographic patterns in access and levels of service analyzed above, federal spending is heavily concentrated in urban areas, better off states and in the largest metropolitan areas within those states despite the generally inferior levels of service which prevail in smaller localities in states with higher degree of marginalization. This is partly explained by the higher levels of performance and creditworthiness of localities in better off states (appendix , 10.a): Key federal programs (eg. APAZU and PROMAGUA) are ostensibly targeted to support localities which to concerted steps to modernize their water supply and sanitation services. At the same time, the long history of CNA de facto directly administering large volumes of federal resources for unprogramed “emergency works” and other sorts of ad hoc investments in non-reforming localities in response to political pressures has diluted incentives for states and municipalities to participate in such programs. While data on the application of spending is sketchy at best, it appears that spending on water supply declined as a percentage of total spending (from 60–70 percent in the late 1990s to 30 percent in 2002), offset by large increases in wastewater treatment (10 percent to 30 percent over the same period) and very modest increases in sewerage (despite large gaps in sewerage access in all but the largest cities) and very modest investments in systems rehabilitation and modernization.48 3.41 The eligibility criteria of federally funded programs contribute to the concentration of spending by use of funds. Examples include earmarked funding for the northern border states (primarily sewerage), PROMAGUA (special dispensation for wastewater treatment), and PRODDER (primarily water supply). Very few localities participate in more than a single program, given the procedural demands each place on the local administration. This leads to a concentration of spending for a narrower set of services than would have otherwise been recommended by the analysis of service gaps. These findings should be regarded tentatively as there is currently no system in place across states and at the national level for accurately tracking the uses of funding. ELECTRICITY 3.42 Electricity supply for industry and commerce is priced on a rational cost basis for large firms, and so they receive essentially no government subsidy, and subsidies for small firms are relatively small. For residential and agricultural customers, however, there is significant underpricing and, therefore, large subsidies and off-budget spending (see Annex 1). 3.43 For residential users, the subsidies are large (0.7 of GDP) and mostly go to middle and upper income households, as summarized in Chapter 2. The data to estimate the state-by state incidence of these subsidies is not readily available, but we do know that the largest per capita subsidies go to the areas in the North where it is very hot in the summer. This is because of the special summer tariff schedules that explicitly benefit high-temperature zones. These areas are more economically developed, especially in recent years, and have the least need for subsidy from an income and development perspective. 3.44 State-by-state data on the subsidy for agricultural users of electricity shows clearly that the rural population in the more developed states benefits the most, as noted above. The 2002 48. Based on chapter 2, “Elaboración de la Estratégia para el Sector de Agua Potable y Saneamiento”, Rubén Barocio, mimeo Jan 2004. - 89 - electricity tariff reform had some measures that aimed eventually to limit the subsidies to farmers who followed the rules about how much water to extract, but it remains to be seen whether that will affect the extremely skewed geographic distribution of the subsidy. SUMMARY AND RECOMMENDATIONS 3.45 Mexico has no unified policy for the geographic distribution of public spending. Rather the overall distribution reflects separate policies made at the national level within diverse sectors, and these polices were made in response to policy issues, interest group pressures, and historical legacies particular to each sector. Although it is possible to question the wisdom of some of the individual policies, the general strategy made sense when political authority and responsibility for results were centralized at the national level. The logic of the strategy needs reconsideration now, however, since states and municipalities have political autonomy and have been given responsibility for many areas of service delivery. 3.46 For activities that aim or claim to foster local economic growth—electricity subsidies, water subsidies, local roads, and municipal services—the state and local governments have the best information and incentives to choose between these forms of spending (including the off- budget ones) and to decide how to provide them most efficiently. The same advantage of local decision-making also applies to education and health, since the subnational governments should be more responsive to their electorates and have a better understanding of the local need for public services.49 To achieve this, the federal government could allocate aportaciones for education and health on a per client basis and could liberalize the states’ authority to allocate aportaciones to their highest priorities. Of course, this could not be done instantly and would involve ending the earmarking of transfers for historical positions (such as the existing stock of teachers and health care retirees) and allocating all incremental funds toward reaching the per client targets. At the same time, the state and municipal governments need to have greater accountability for delivering quality service to their constituents. And the federal government needs to clarify its role as a monitor of the quality of service and of accountability, but not as a provider of services. The government is now starting to gather and should make publicly available the information on the performance of schools—including results to each municipality, school, and grade level. 3.47 To give them some authority to determine the priorities between spending, including off- budget subsidies, and the alternative of lower taxes and fees, subnational governments— especially the states, which have little tax authority—need more authority to raise taxes. They should also have more incentive to raise their own resources, by rationalizing, reducing, and un- earmarking the transfers they receive from the federal level. Concretely, this would involve measures like giving the states the authority to collect a sales tax on the same products as covered by the VAT, reducing the national VAT rate, reducing participaciones, and using them to some extent to compensate states that have smaller than average tax bases per capita. 49. For programs of direct transfer to households, like Oportunidades and PROCAMPO, it is appropriate for the federal government to continue its role as the primary source of financing, since the states with the most need are those least fiscally able to provide support for households. - 90 - 3.48 Some interest groups—teachers, large farmers—may oppose decentralizing the authority to allocate resources, because they have grown accustomed to receiving federal funds earmarked for them. The strong political leadership for decentralization most likely will come from the majority of governors who would see their finances and authority improve, as well as from farsighted national officials who see the wisdom of such decentralization. - 91 - - 92 - 4. INSTITUTIONS FOR PUBLIC EXPENDITURE MANAGEMENT 4.1 Mexico has an established and well-documented system of public financial management that includes the main components of a robust system. The basic building blocks are in place. The primary issues in Mexico’s public expenditure management are how to achieve better results and how to get the system to work better. To this end, The Mexican government has been pursuing reforms for many years, but withmixed results. In part this is due to the methods of formulation and implementation and in part to the scope of reforms, discussed in more detail below. 4.2 After a brief overview of Mexico’s government institutions in section 1, this chapter turns in section 2 to the issue of performance monitoring and resource allocation, because incentives are as critical in the public sector as in the private sector. Section 3 discusses how the annual budget formulation is (or should be) linked to national priorities. Section 4 discusses the institutions for budget execution, and section 5 discusses the need for transparency and ways to achieve it. Section 6 considers sequencing, timing, and political strategy, looking first at the history of public reform efforts in Mexico and then at the present efforts. Each section introduces the issue, presents the Mexican situation, discusses international evidence on the issue, and offers suggestions for Mexican policymakers. This chapter complements other analytical work such as Mexico: Fiscal Transparency ROSC (IMF, 200x) and Mexico: Country Financial Accountability Assessment (CFAA, World Bank, 2003). BACKGROUND AND SALIENT FEATURES 4.3 Mexico has a presidential system of government wherein the duties of public finance are divided between the executive and legislative branches of government. The executive branch has primary responsibility for budget formulation and national planning, although Congress has taken a more active role since 1997, when the ruling party ceased to hold the majority. Within the executive branch, the key institution for public finance is the Secretaría de Hacienda y Crédito Público (SHCP). As is common among OECD countries, SHCP encompasses a full range of public finance functions, including primary responsibility for the macroeconomic framework, public spending, tax policy, tax administration, customs administration, debt management, treasury, and intergovernmental fiscal relations. SHCP also has responsibility for water resource and energy pricing in coordination with the relevant agencies and for procurement rules and public service pay, in coordination with the Secretaría de la Función Publica (SFP). Numerous decentralized entities, including the national statistical office, are under the SHCP’s umbrella. 4.4 SFP (formerly Secretaría de Contraloría y Desarollo Administrativo, SECODAM) is another significant entity for public expenditure management within the executive branch. It audits the internal functions of government and, through its subsidiary Civil Service Unit, sets the broad employment rules and pay scales for non-union public employees. Unique among OECD countries, SFP centralizes internal control functions, reviewing transactions and monitoring to assure that proper financial management controls and procedures are in place. - 93 - SFP used to be the primary audit body of the government and is in the process of adapting to the establishment of an external audit body attached to Congress. SFP played a leadership role in establishing COMPRANET, the automated procurement system in Mexico. 4.5 Mexico has a bicameral legislature, consisting of the Chamber of Deputies and Senate. The constitution grants Congress the following powers, among others, which are relevant for public expenditure management: • To levy the necessary taxes to cover the budget • To modify and approve the annual expenditure budget, without which no expenditures are allowed. • To fix the bases on which the president of the republic may borrow on the credit of the nation; to approve such loans and to acknowledge and order payment of the national debt • To create and abolish public offices of the federation and to fix, increase, or decrease their salaries • To examine the report that the executive branch must submit to it annually, which must include not only conformity of the items expended within the budget of expenditures but also the correctness and justification of such items. 4.6 The Chamber of Deputies is more important in fiscal matters than the Senate. It also has exclusive powers granted to it under the constitution (Article 74) regarding supervision, appointments, and budgeting for the Auditor General’s Office. 4.7 The main professional body supporting the work of the Congress is the Auditoría Superior de la Federación (ASF), which began operating in January 2001. The ASF is still evolving and will take some time to become fully operational. 4.8 Since 1998, the government has embarked on a process of budget reform that includes a shift toward results-oriented budgeting and substantial changes in the classification of budget expenditures. Mexico is still striving to achieve the main objectives of the reform (see Box 4.1) The reform included three main features: • A new programmatic structure (NEP) for the budget, intended to improve the link between public policies and government offices and their activities, to improve the measurement of performance, to improve the alignment of the costs of programs with government policies, and to clarify the mission of ministries, agencies, public enterprises, and their subordinate units • A performance evaluation system (SED) for monitoring and evaluating performance and results achieved by spending • An integrated financial management information system (Sistema Integral de Administración Financiera Federal, SIAFF), aligned to the budget and allowing more timely financial management. SIAFF was intended to modernize treasury operations. - 94 - Box 4.1. Objectives of the 1998 Budget Reform The budgetary reform is on its way to achieving the following objectives: • To link the allocation of public resources to the efficiency, effectiveness, and quality of the public goods and services produced. • To relate the operation of public expenditure to the achievement of public policy goals. • To develop administrative tools and techniques for costing programs and government policies, to prepare multi-year spending projections, and to improve spending control from a strategic perspective. • To encourage the training of genuine managers of responsible units whose priority is to obtain results as efficiently as possible. • To simplify the management of spending operations. • To align the efforts of government institutions with public policy goals and purposes. Source: OECD Focus, “Examples of Recent Public Sector Reforms in Mexico: Aiming at Results: Mexico’s Budget Reform” (July 1998). http://www1.oecd.org/puma/focus/compend/mx.htm. 4.9 Mexico has better structures and operations of public financial management than most emerging market economies and it compares favorably with many other OECD countries, based on a 2003 World Bank-OECD survey of developed and developing countries (see Box 4.2).50 Its public finance institutions are broadly comparable, having a formal legal basis and operating under well-developed procedures. Notably different from other OECD countries are some of the operations of the public finance system, and these present opportunities for improving performance. Box 4.2 Public Finance Systems: Mexico Relative to Other OECD Countries • Compared to other OECD countries, Mexico is among the • 64 percent that use a cash-based budget (versus accrual or partial accrual budgeting) • 78 percent with a budget office located within the Ministry of Finance • 67 percent with a senior civil servant heading the budget office (versus a political official). There are notable differences as well. Mexico is among the • 27 percent that do not include multi-year expenditure estimates in the budget • 40 percent where the Ministry of Finance does not identify policy revisions where appropriate during budget execution • 23 percent where the budget office does not conduct routine evaluations of spending • 40 percent that give agencies more than two ceilings on spending elements (retaining more central control of spending, giving spending units less authority to manage) • 20 percent that do not allow any carryover of unused spending to future years, even with approval of the legislature (giving agencies incentive to spend all funds before year-end, whether useful or not) • 17 percent that do not allow carryover of unused appropriations for investments (such as building construction) from one year to another • 48 percent where spending units do not report to the Ministry of Finance or the treasury on their commitments (obligations) to ensure that expenditures do not exceed the budget 50. World Bank/OECD, “Survey on Budget Practices and Procedures” (2003). Available at http://ecde.dyndns.org. As of December 2003, 27 OECD and 17 developing countries are included in the database. - 95 - • 40 percent where the Ministry of Finance makes forecasts of cash spending (against which it monitors actual spending) less frequently than on a weekly or daily basis • 30 percent that do not require all public revenues to be deposited directly with the treasury in a single account under control of the Ministry of Finance or the treasury • 28 percent where ministries or government organizations are allowed to maintain interest-bearing cash accounts outside the treasury • 4 percent where information on actual expenditure is not available to the Ministry of Finance or the treasury in time for effective monitoring • 20 percent with no formal requirements for audits of nonfinancial statements (for example, performance reports) • 28 percent that do not integrate assets, liabilities, and government equity accounts into the accounting system to facilitate the preparation of financial statements • 30 percent reporting no evidence that performance results are used in determining budget allocations. PLANNING, PERFORMANCE MEASUREMENT, AND BUDGET ALLOCATION 4.10 All countries struggle to integrate information on performance and management—to use information regarding performance in decision making. Some attempt to monitor actual performance relative to prior baseline performance or benchmarks, while others seek to monitor performance relative to predetermined targets or plans. The approach chosen—and vehicles for implementing it—is influenced by the degree to which national planning and budgeting processes are integrated. 4.11 Mexico has separate planning and budget processes, with planning institutionalized in a six-year National Development Plan (PND) under the Planning Law. The National Development Plan is in theory the guide for budget formulation. Goals (metas) and objectives, as well as program structures, are supposed to articulate with the development plan. Figure 4.1 presents the intended relationship between planning and budgeting. The National Development Plan is a fixed six-year plan (not updated annually) that is roughly coterminous with the president’s term of office. Objectives, policies, and performance targets are set through this process. Programs (the mode for achieving the objective) derive from the plan. - 96 - Figure 4.1: Relationship between Planning and Budgeting Source: Subsecretaría de Egresos y Oficina de la Presidencia para la Planeación Estratégica y el Desarrollo Regional, “Proceso de Planeación, Programación y Presupuesto” (Mexico, 2001), p. 3. 4.12 The annual budget process generally takes the objectives and programs as given. After determining annual resource constraints, funds are allocated to programs and projects. Information regarding performance is incorporated into the annual budget documents, but these performance measures are not used in decisions pertaining to agency management or to resource allocation. The Office of the President monitors these indicators, but follow-up is unclear, and there are no formal reviews. Performance is not built into pay either. Moreover, the program structure has changed annually over the past few years, weakening the accountability framework. 4.13 The sectoral components of the National Development Plan are not integrated with resource constraints. Although the plan includes broad targets for the economy and public sector expenditures, these are not broken down into sectoral ceilings. As a result, sectoral plans are not crafted within resource constraints, and the objectives chosen may not be realistic. 4.14 Mexico introduced government-wide performance measurement in 1996 to allow services to be evaluated for their “quality, unit cost, usefulness, and social impact.”51 The annual budget includes performance measures for programs, totaling some 10,000 indicators. Program performance is monitored relative to the fulfillment of plan objectives rather than the origins and legal mandates of government activities or agencies. Performance is not of actual government 51. OECD, “Mexican Government Response to OECD Issues and Developments in Public Management Survey 1996–97.” Available at http://www1.oecd.org/puma/gvrnance/surveys/report97/mx.htm - 97 - offices or ongoing activities as much as programs defined relative to the plan. The programs and objectives can and do change annually, making it difficult to track progress over time and undermining the accountability of agencies and their managers. 4.15 In many OECD countries, national planning has fallen out of favor, and the annual budget process has become the government planning process. These countries have taken steps to introduce performance measurement into management and resource allocation in an effort to improve the efficiency and effectiveness of public spending. Two other large countries in the Western Hemisphere—both also presidential and federal systems—are particularly relevant to Mexico: the United States and Brazil. 4.16 In the United States, the annual budget process is also the planning process. Objectives, performance targets, policies, the mode of implementing or achieving them, and the resources for doing so are all decided through the budget process. In this environment, performance is measured relative to both planned targets and over time (relative to historical performance). The targets and performance measures are tied to standing programs and organizational structures, from the ministry or department level down to offices. Career officials are accountable, not just the political appointee. (Systems in some countries assign responsibility to the minister for all activities, and the minister is supposed to hold career officials within the ministry to account, but for political reasons it is very difficult in practice to hold a minister to account. Accountability structures within a ministry frequently are weak and not transparent.) 4.17 Brazil, like Mexico, has a national plan, separate from the budget process. The Ministry of Planning, Budget, and Management is responsible for developing the four-year plan (again roughly coterminous with the presidential term of office). The planning process is used to set priorities, objectives, and performance targets. Unlike Mexico, the structure of the plan is relatively fixed and covers all government activities. Also unlike Mexico, the national plan includes resource allocations for programs, by year, over the planning period. (But, given the fixed, multi-year nature of the plan, target allocations beyond the first year are highly uncertain.) New administrations imprint their policies by selecting the programs that they consider to be high priorities as well as the targets and resource allocations designated for them. The Cardoso Administration designated 80 programs as “priority programs” in the first year of the plan and subsequently reduced these to 50. 4.18 Brazil developed a management information system to tie program funding and performance information together, focusing on performance relative to plan targets. Programs were defined supra-organizationally, cutting across ministries and implementing agencies, and “program managers” were appointed to “manage” them. However, most programs are contained within ministries, although they are not always tied to the administrative structure of the ministry. Program managers have no formal authority, no control over resources, and no influence over the activities in ministries that hold their program, unless the head of the activity is designated the program manager. Under this structure, activity managers cannot use information on the program’s performance to manage their work, and program managers have no influence over actual management. This mismatch between authority and responsibility limits accountability. - 98 - 4.19 In Brazil, information on performance is not included in the formal budget documents, but the on-line database does allow partial matching of objectives, performance, and (marginal) resources. In developing the program concept, Brazil created separate “programs” to encompass personnel expenses, such that the other programs only contain the marginal cost of the activity. The new national plan (2004–07) does allocate corresponding personnel costs to each program, but the administrative structure does not match the program structure, so it is not clear how accurate this attribution is or how it will be monitored. 4.20 Despite these structural issues, Brazil did try to simulate the use of performance information by management. The planning office of the Ministry of Planning, Budget, and Management used the information system for quarterly performance updates, evaluating each priority program with respect to targets in the national plan and assessing financial performance relative to that year’s budget. Programs performing poorly or not likely to use all of the resources allocated to them lost resources, and these were transferred to other priority programs deemed to be performing better. This was an attempt to improve performance and to use information on importance for management and resource decisions and to improve performance. The current administration has abandoned this approach. 4.21 International experiences offer several lessons of relevance for Mexico (see Box 4.3 for a longer list). First, the units of performance monitoring, management, and resource allocation must coincide in order to have real accountability and monitoring and to motivate improvements in efficiency and effectiveness. That is, the organizational level at which performance is measured should be a distinct organizational unit under management as well as a unit to which resources are allocated. Without this coherence, it is difficult to use information on performance to improve management or to link resources. There also needs to be a link between authority and accountability. Unless the agent whose performance is measured also has some authority for changing the way the program is managed or resources are allocated, the information on performance can have little operational impact. 4.22 Linking performance and financial management is easier in countries where global or output budgeting is in place (versus Mexico’s input-based budgeting), where strategic objectives and targets are set within resource constraints (versus Mexico’s separate planning process), and where full-cost accounting is in place (the full cost for any given activity or program is in the budget for that program, including personnel, pensions, and capital outlays). 4.23 It is also important to distinguish between program evaluation and program performance. Program evaluation generally means deeper evaluations of individual programs, frequently using advanced techniques such as impact analysis to assess whether a program is having any impact on intended objectives. These evaluations can be expensive and can take 12–18 months to complete. As such, they generally are not suited to annual budget decision-making and are reserved for programs that annual performance monitoring suggests may not be operating well. Annual performance indicators, because of the short time horizon, tend to focus more on outputs. But annual performance information can serve as a broad indicator of whether a program is problematic, without diagnosing the problem. Only those programs identified as non-performing or under-performing would be referred to program evaluation or management review to assess the causes and possible solutions. Such a selective approach to full program evaluation, based on initial indicators of problems, is usually more cost-effective and realistic. - 99 - Box 4.3 Linking Performance and Budget Allocation: Lessons from International Experience Implementing effective performance monitoring requires • Setting organizational incentives to support performance monitoring • Getting performance monitoring consistent with organizational culture • Creating a central unit that plays an active and effective leadership role in defining criteria and implementing practical performance monitoring • Linking ongoing performance measurement with more periodic program and policy evaluations • Paying attention to clients, customers, and citizens in developing performance monitoring. Integrating performance and financial management is easier where • The setting of strategic and target objectives is linked to the allocation of resources • Global or output-based budgeting is in place • Full-cost activity accounting is in place • Programs consist of tangible and measurable products or services • Integration is attempted at the level of program management and operational management • The impact of programs can be seen soon after delivery • The results can be attributed to the program with a high degree of confidence. Maximizing the use and value of evaluations includes issues such as • Gaining support from the top of government • Generating effective demand for evaluations • Setting realistic expectations • Systematizing evaluation activities • Linking with the budget process • Choosing the appropriate evaluator • Planning evaluations • Timing evaluations appropriately • Meeting user needs • Ensuring relevance • Involving stakeholders • Ensuring methodological quality • Making judgments and recommendations • Communicating findings • Monitoring and following up • Recognizing the needs of staff for training and support. Source: John Mayne and Eduardo Zapico-Goni, eds., Monitoring Performance in the Public Sector: Future Directions from International Experience (London: Transaction Publishers, 1997), pp. 22–24; OECD PUMA, “Integrating Financial Management and Performance Management” (July 1999), p. 9; OECD PUMA, “Improving Evaluation Practices: Best Practices Guidelines for Evaluation and Background Paper” (January 1999), p. 24. 4.24 Even with these basic elements in alignment, it is important for there to be an imperative to pay attention—to use—the information. Simply providing the information to potential users within the government—managers and oversight agencies—is not enough. Even improved transparency—publication of performance information—is not enough. These are necessary, but not sufficient, conditions. 4.25 Countries have used different approaches to providing such an imperative, and their efforts generally fall into either “passive” or “active” groupings. Passive approaches include performance contracts (formal agreements between superiors and subordinates on targets, implying a formal review at the end of the contract period), peer pressure (making a scorecard of - 100 - performance for each unit widely available and compiling a league table to allow comparisons), public embarrassment or approval (publishing the league table), or monetary incentives (offering hope of monetary benefit if performance improves or targets are achieved, either on an individual basis, by tying senior management pay or bonuses to organizational performance, or on an organizational basis, by tying organization-wide pay or bonuses to organizational performance or linking the organization’s budget to its performance). These are passive measures insofar as they set up a structure but do not assure that performance measures will be used to affect decisions. 4.26 More active measures include formal reviews (holding regularly scheduled meetings at which performance is assessed), the engagement of senior management (either having senior managers chair the formal review or be engaged directly in monitoring and following up on performance problems), and nonmonetary rewards (generally giving public recognition with an award or honor). Many of these are blended for greater impact. Vice President Al Gore’s High- Impact Agency Initiative or the U.K. prime minister’s Delivery Unit six-month performance reviews are examples of these approaches. 4.27 As Mexico seeks to improve the efficiency and effectiveness of public expenditure— getting more from each peso spent—resources may be freed for other priority areas. A well- functioning system of performance management is one aspect of a well-functioning public expenditure system. BUDGET FORMULATION AND CAPITAL PROGRAMMING IN RELATION TO STRATEGIC PLANNING 4.28 The budget should reflect and make clear to the public the policy priorities of the government as a whole—what the president and Congress agree on. Multi-year financial and policy planning should guide the annual budget process, especially personnel policy and capital programming, which have long-lasting effects on spending. The budget should be a vehicle for steering the public sector. The manner in which the Mexican budget is formulated does not, for the most part, meet these objectives. 4.29 Mexico’s annual budget process begins in March of the prior year with the development of macroeconomic estimates and the expenditure envelope for the budget year. By the end of July, the call circular is sent to spending ministries providing guidance and formats for developing the budget. August and September are spent receiving requests from the ministries and negotiating needs. By September, a draft budget bill is available, and an updated macro forecast is used to make final adjustments to spending. By October, a final budget bill is prepared and transmitted to Congress, which has until December 15 to debate and pass the budget. In practice, however, other decisions like tax reforms are often prerequisites to the budget debate and thus defer discussion of the spending budget to the last weeks (or days) of December. This leaves too little time for full discussion. Mexico has no provision for using the previous year’s allocations or making continuing resolutions if the budget has not been approved before the year begins. This has the advantage of forcing the parties to come to agreement, but the all-or-nothing option for both sides (See Box 4.4 on the role of Congress). - 101 - 4.30 The Ministry of Finance follows a traditional mandate that focuses on detailed control of agency finances (budget formulation and execution), but gives relatively little attention to spending-policy issues and the bigger picture. The budget office reviews and financially audits spending and does not focus on the effectiveness or impact of an activity. There are few incentives for agencies to be efficient, and detailed controls of spending procedure on the part of Hacienda discourage efficiency. Policy is not a significant feature of the budget formulation process. 4.31 In Mexico, the separation of planning and budgeting into two processes and the treatment of annual budgeting as largely a technical exercise has created problems with the realism as well as with the policy content of the budget. The absence of policy content has allowed current recipients of resources to establish their claims as irreducible, thus making the budget seem ever more rigid. Existing spending programs go unchallenged, and no room remains to finance new priorities. 4.32 Generally, the President does not get involved in the budget process until the end, when reviewing the budget for transmission to Congress. This leads to limited ownership of the budget and to the view that the budget is a product of SHCP rather than the President. According to spending ministries, their discussions with SHCP do not focus on policy, programs, or objectives but on line items in the budget, use of appropriate inflation factors, and similar technical issues. 4.33 The absence of a medium-term framework limits foresight and transparency. Except for capital investment projects, where figures for up to five forward years are provided, the Mexican budget does not include information on actual expenditure in prior years, current-year estimates, or future-year projections of spending. Thus decision makers have no ready reference points with which to assess relative tradeoffs or trends, to identify and react to emerging spending crises in a timely manner, or to assess the impact of short-term policy options on long-run fiscal sustainability and service delivery. Because they lack multi-year budget estimates, line ministries prepare their multi-year sectoral plan in a resource vacuum, resulting in unrealistic and unachievable plans and targets. 4.34 The limited role of SHCP in analyzing and challenging the programs and policies in budget requests does not support greater efficiency and effectiveness of spending. During budget formulation, the Subsecretaría de Egresos (SSE) is primarily a clerical office, passing spending ceilings to the secretaries, collecting program structures and detailed budget allocations from the line ministries, and then aggregating and tabulating the results. Current “analysis” by SSE consists of reviewing requests to make sure they are in strict compliance with budget guidelines, the estimated inflation rate, proposed pay increases, and so forth. 4.35 The budget dialogue between SHCP and spending ministries concerns technical specifications for classifying and inflating estimates, not policies, programs, or results. This partly reflects the historic development of public finances in Mexico: the dominant party handled policy issues, and formal institutions of government were more administrative. 4.36 In most OECD countries, budget offices and ministries of finance have become sources of neutral competence, providing policy evaluation and advice to whatever party is in power and - 102 - to both the executive and legislative branches. They are a permanent source of expertise to support improved public spending. In 60 percent of OECD countries, the budget office identifies potential policy revisions during budget execution to support the government’s policy choices.52 In 77 percent of OECD countries, the budget office conducts routine evaluations of spending, providing analytical support to government. In the United States, the Office of Management and Budget is a source of neutral competence, supporting achievement of the president’s objectives and current law, regardless of the president’s party affiliation. In 1921, General Dawes, first director of the Bureau of the Budget, said, “Much as we love the President, if Congress, in its omnipotence over appropriations, and in accordance with its authority over policy, passed a law that garbage should be put on the White House steps, it would be our regrettable duty, as a bureau, in an impartial, nonpolitical, and nonpartisan way, to advise the Executive and Congress on how the largest amount of garbage could be spread in the most inexpensive and economical manner.” In other words, Congress decides what programs to fund, and then OMB plays an active role in deciding how to do it most effectively. Capital Budgeting 4.37 Mexico has been reforming its budgeting process for large capital projects. In 2001 the capital budget process was modified to correct several problems: projects were funded without any cost-benefit analysis or with poor-quality studies and with weak capacity of the spending ministry to undertake project evaluation; frequent cost overruns in investment projects due to weak project oversight, insufficient preparation or estimation, or poor procurement practices; absence of recurrent cost projections for projects; and weak project identification and classification (investment labeled as recurrent or vice versa). 4.38 The new process started in 2002 and includes (a) a requirement for each spending ministry to have a six-year capital plan, updated annually; (b) a whole-of-government investment coordinating committee made up of undersecretaries from line ministries; (c) external review of cost-benefit studies by an expert group; (d) an investment project portfolio with individual project identification codes (managed by spending ministries) and a requirement that only projects with a valid code be included in the budget (subsequently updated for congressional approval of projects) and that no funds be released by the treasury without a valid code; (e) clear identification of project managers for each project; (f) publication of cost-benefit studies, as well as ex post evaluation of completed projects to assess whether the actual cost-benefits match the estimated ones; (g) enhanced training for line ministry project managers and capital budget staff; and (h) direct project audits by the ASF. Previously, SSE evaluated cost-benefit studies directly. Now an external group (CEPAP, a research trust of BANOBRAS) reviews the studies. If they are negative, the project is canceled. 4.39 These reforms should bring some clarity and improved accountability to the capital budgeting process. In theory, by requiring a multi-year capital plan to be developed and maintained by each spending ministry, some linkage is expected between the sector objectives in the national plan and those in the capital plan. The degree of consistency should be monitored to assure that the relationship is sound. Improvements to the National Development Plan, such as 52. World Bank/OECD 2003 survey. Ibid - 103 - planning within multiyear sectoral resource constraints, would help to assure that capital plans are reasonable. Moreover, establishing multi-year estimates in the annual budget process would better reflect the recurrent costs of capital projects and allow a better assessment of their sustainability. 4.40 Budgeting for Direct Commitments in Brazil: To assure that the basics are covered, the Brazilian budget formulation process has two broad functional steps: (a) Define the macro-economic projections, estimates of public revenues, and plan/set expenditure and debt levels. (b) Develop indicative sectoral limits based on analysis of current operating costs, programs, and activities. The ceilings include several subsets of estimates: i. Personnel costs. This is a known number, because a detailed personnel registry provides information on wages, new hires, and early retirements. ii. Debt. Part of debt service for loans related to each Ministries programs is allocated to the Ministry and part to the Ministry of Finance. The budget office knows all contracts, and knows the agreed-upon share. iii. Activities and projects, continuing. The cost of continuing the current programs is estimated by the budget office, and refined by the line Ministry. This includes administrative costs in support of personnel (utilities, etc.), plus mandatory activities (health, pensions, etc.) iv. Activities and projects, new or expansions. Line Ministries develop these proposals and submit to the Ministry of Planning for review. 4.41 The activities and projects go through a rigorous cost-benefit test before being eligible for funding in the annual budget. The spending ministries then prioritize their approved capital projects in terms of importance to sector’s mission and to the priority programs of the multi- annual plan. 4.42 Within the available resource estimates, subset (i), (ii), and (iii) are treated as priorities in the annual budget process. Subsets (iv), (v), and (vi) are treated as residual: If resources are available to allocate, they first go to activities and then to projects (including capital spending), and within these categories, priority is given to continuation of on-going before new or expanded projects. 4.43 This process assures the basic obligations of the state—debt, personnel and fixed administrative costs, as well as transfer payments—receive funding first. It also assures on- going capital projects receive funding for completion before new projects are initiated. Overall, capital spending is still the residual, but at least this process discourages new capital commitments that the government cannot afford or complete. 4.44 Mexico budgets only one year at a time for capital projects. Although this preserves annual budget discretion, it may lead to higher project costs by complicating multi-year contracts - 104 - for construction. Further, the annual project funding often does not become effectively available until later in the year, which may increase project costs. 4.45 Capital budgets and the legislature. In countries where the legislature has an important role in the budget process, issues can arise over capital programs, where the legislature may add capital projects to the budget that have either been rejected on cost-benefit grounds or never evaluated for cost-benefit. These pose unique challenges for planning systems, orderly investment, and efficient use of limited resources. In Brazil, Congress does add new projects into the budget, sometimes funded by less than robust revisions to revenue estimates to make the new projects deficit-neutral. The Brazilian Treasury does not allow these projects to be initiated unless the revues materialize, using an authority called bloqueado. This has led to tensions between the legislature and executive that have not been resolved, with Congress threatening to enact legislation making the budget items mandatory spending (floors as well as ceilings). 4.46 With the Mexican Congress adding more specific capital projects to the annual budget, this issue could become a point of conflict between fiscal discipline, efficient use of resources, and Congressional authority over spending. As noted above in discussing changes to the capital budget process, projects not on the approved cost-benefit list would not have a project identification number, and would in theory not be funded by treasury. It is unclear if this process will work in practice, but if it did, it would be a harder constraint than the process in Brazil (which makes allowances for funding to materialize). As such, it is likely to lead to tensions between executive and legislature. 4.47 In the Untied States, where Congress has clear preeminence over budgets, it is not uncommon for capital projects to be added that were not requested by the executive. Even so, there are cases where successful project evaluation and prioritization criteria have been implemented and generally respected by the legislature. For example, the Indian Health Service, a Federal program for providing health services to American Indians and Alaskan Natives, includes capital construction for hospitals and clinics. It had been the case that the capital construction priorities of the executive and legislature generally did not match. In response, the executive worked with the relevant legislative committees and stakeholders to develop an objective means of identifying and prioritizing capital projects as a function of whether health facilities existed, age of facilities, efficiency of direct construction versus alternate health delivery means, size of the affected population, and health status of the population. With an objective set of criteria, the funding priorities of the executive and legislature were harmonized, capital budgeting became more efficient. While this can be done for individual programs, it might be more challenging to develop such criteria for capital spending across functions or sectors (e.g. health versus education). Nonetheless, a more collaborative approach and one focusing on the criteria for selecting approved capital projects might be an approach that could work in the Mexican context. Medium-Term Budgeting: International Experience 4.48 Typical OECD countries have not adopted MTEFs per se, but rather sought to introduce medium-term perspectives into their annual budget process. This includes macroeconomic forecasts of public revenue and expenditure as the basis for budget formulation and multi-year costing of ongoing or new policies and programs. Among OECD countries, three-fourths - 105 - produce multi-year cost estimates for all new spending proposals.53 These multi-year features evolved gradually over time, usually as technical improvements to the budget process rather than as big-bang reforms. They have also been pursued as iterative changes to the process, being improved on over many years. Folding a medium-term perspective into the budget process is a multi-year exercise, requiring persistent efforts to improve; the improvements may be better pursued as technical improvements than as comprehensive reform. 4.49 A 1999 survey noted that an MTEF consistent with macroeconomic targets is usually not available in the Bank’s client countries in Africa.54 At that time, only a fifth of the countries had an MTEF that projected aggregate expenditures over a three- to five-year horizon consistent with macroeconomic targets. In the 2002 HIPC expenditure tracking exercise, only one in six had a medium-term perspective well integrated into their budget formulation process, although about half did provide some kind of medium-term forecast.55 4.50 Where MTEFs are being implemented, they need to be integrated into the budget process and documents. As of 2002, only five MTEFs in Africa were integrated in a meaningful way into the budget process. In most cases MTEFs operate parallel to the general budget process. Only four countries submitted MTEFs to both the Cabinet and Parliament—and in some cases the MTEF remains a technical document of the Ministry of Finance56. Where MTEFs are separate from the budget process, they are far less effective. A proper MTEF is the budget process. 4.51 Some prerequisites for good budget formulation include a solid budget classification system, comprehensive budgets, meaningful hard budget constraints (adequate internal controls and political will), and capacity and willingness to prioritize and reallocate (make the difficult tradeoffs).57 4.52 Pursuing MTEFs as a separate reform, parallel to the budget process, is not effective. Unbundling the key features of an MTEF and pursuing them individually may be more effective given OECD experience. This suggests that the MTEF is better as a conceptual framework for seeing how elements of reform fit together than as a reform in and of itself. Regardless, introducing a medium-term perspective into the budget process will take persistent effort over a longer period of time. Adequate sequencing, including attention to basic public expenditure systems, is essential for improving budget formulation (see Box 4.4). 53. World Bank/OECD Survey. Ibid. 54. Kostopoulos, “Progress in Public Expenditure Management in Africa: Evidence from World Bank Surveys” (January 1999), p. 18. 55. World Bank, “Actions to Strengthen the Tracking of Poverty-Reducing Public Spending in Heavily Indebted Poor Countries (HIPC)” (January 2002), p. 13. 56. Le Houerou and Taliercio, “Medium-Term Expenditure Frameworks: From Concept to Practice. Preliminary Lessons from Africa.” p. 13. 57. W. Dorotinsky and R. Floyd, “Public Expenditure Accountability in Africa: Progress, Lessons, and Challenges,” draft Africa Region Working Paper (World Bank, August 2003). Malcolm Holmes with Alison Evans, “A Review of Experience in Implementing Medium-Term Expenditure Frameworks in a PRSP Context: A Synthesis of Eight Country Studies,” discussion draft (London: Overseas Development Institute, Centre for Aid and Public Expenditure, May 2003). - 106 - Box 4.4 An MTEF for Mexico? (a) Medium-term expenditure frameworks (MTEFs) gained currency as an important reform during the 1990s, with many countries adopting them. In Africa at least 19 countries have MTEFs in some stage of implementation.58 In OECD countries 73 percent incorporate multi-year expenditure estimates into their budgets, and 67 percent have a medium-term fiscal framework with ceilings for expenditures, deficits, and revenues over multiple years as part of their budget.59 (b) While the MTEF concept can encompass many features of any good budget process, its primary aims are (a) to introduce a multi-year perspective into budgeting, enabling governments to recognize the implications of current budgetary decisions for government finances in the future; (b) to create a disciplined, sequenced budget process that changes incentives to support all three objectives of any public expenditure system (macro fiscal discipline, strategic allocation of resources, and operational efficiency); and (c) to move away from input-oriented to output-oriented budgeting. (c) A medium-term expenditure framework tries to integrate policy with resource allocation decisions in a multi-year context to better reflect the effects of decisions on public finances and to strengthen the basis for decision making. It also involves structuring the decision making process to encourage economy and efficiency and to focus spending and policy on attaining the government’s objectives. Thus an MTEF is essentially concerned with establishing a process for achieving, through the budget, a strategic allocation of resources in line with government priorities, the opportunity costs of decisions, and changing incentives in decision making to favor efficiency. (d) An MTEF is not a purely technical reform. Inclusion of multiple years in the budget is only one dimension. Changing how agencies and decision makers behave, evaluate policies, and make choices is equally important. Moreover, an MTEF is fundamentally about creating a framework for making informed policy choices: what are the government’s policy priorities, what are their expected benefits and costs, and how can they be achieved through the budget? A budgeting process that takes insufficient account of the underlying policy dialogue can create a serious disconnect between government priorities, spending allocations, and budget results and outcomes. Annex B elaborates the elements of an MTEF and a good budget process Spending Rigidities and Budget Management 4.53 Chapter 1 identifies the degree of spending rigidities in the annual budget, which limit policy choices and, in many ways, prevent the administration from meeting its own policy objectives. Spending is tied up in previous commitments, political as well as legal. An MTEF can help to highlight the implications of these commitments, and the revised budget process with hard ceilings for spending ministries can change incentives and force policymakers to reexamine the contribution that these mandatory spending programs make to government and sectoral policy objectives. 4.54 Although often overlooked, an effective communication strategy has proven instrumental to the success of reforms in OECD and emerging market economies, especially significant budget management and fiscal reforms like those presented by the rigid expenditure programs. 58. Le Houerou and Taliercio, “Medium-Term Expenditure Frameworks: From Concept to Practice. Preliminary Lessons from Africa,” p. 8. 59. World Bank/OECD, “Survey on Budget Practices and Procedures” (2003). Available at http://ecde.dyndns.org - 107 - By educating citizens, policy officials, and politicians about the problems of the current system and presenting options for change and the reasons for the actions chosen, the reform is more likely to gain the necessary support to succeed. Government staff are better able to implement reforms if they understand why they are being asked to do so, and education helps citizens, civil society, and businesses adapt to change. Understanding reduces opposition and promotes working with a new direction rather than against it. Box 4.6 presents some approaches that other countries have used to improve communications. BUDGET EXECUTION 4.55 Budget execution in Mexico involves two central ministries: SHCP and SFP, plus the relevant line ministry. Within SHCP, treasury makes payments and manages cash flow; Planaeación Hacendaria sets the overall macro fiscal plan and borrowing limits; and SSE handles modifications to the budget and monitors compliance. SFP also monitors compliance with the budget, although primarily with respect to compliance with formal rules and the enacted budget. Sustaining Macro Fiscal Targets as Revenues and Cost Change (Role of Planeación Hacendaria) 4.56 Mexico’s National Program for Finance and Development (PRONAFIDE) lays out the plans for reducing the overall public sector borrowing requirement over the medium term and specifies the maintenance of fiscal targets within the budget year. This is done via aggregate spending cuts that fully offset any negative revenue shocks. Positive revenue shocks go partly to reducing debt, which gives fiscal policy a conservative bias. 4.57 Compared to other OECD countries, Mexico reported its macroeconomic and revenue forecasts over the past three years as follows:60 • GDP growth was significantly overestimated (74 percent of OECD countries reported better results) • Unemployment was significantly underestimated (83 percent of OECD countries reported better results) • Inflation was slightly underestimated (25 percent of OECD countries reported similar results) • Revenues were significantly overestimated (78 percent of OECD countries reported better results). 4.58 Sound macroeconomic forecasts are the basis for planning and budget formulation and thus are essential to sound public financial management. The survey results suggest that steps can be taken to improve the objectivity and accuracy of forecasts. Although Mexico does as well as the rest of the OECD in some ways, it lags in others (see Box 4.5). 60. World Bank/OECD, “Survey on Budget Practices and Procedures” (2003). Available at http://ecde.dyndns.org. - 108 - Box 4.5 Macroeconomics and the Budget: Mexico Compared to the Other OECD Countries Compared to good practice in other OECD countries, Mexico is among the • 85 percent of countries making economic estimates publicly available in the budget • 64 percent distinguishing ongoing policy commitments from new commitments in the budget • 76 percent providing spending reports within five weeks of the end of the accounting period. Other areas offer potential in which to strengthen transparency. Compared to OECD countries, Mexico is among the • 60 percent that do not have independent evaluation of economic estimates in their budget • 20 percent that do not have formal requirements for updating macroeconomic estimates and the budget within the year • 15 percent with no requirement for updating revenue and expenditure projections during the year. 4.59 Mexico has done relatively well in managing its macro fiscal position since 1995, but there is scope for improving the link between the strategic allocation of resources, sectoral policy priorities, and incentives for operational efficiency. Role of the Subsecretaría de Egresos (SSE) 4.60 During budget execution, as in budget formulation, the role of SSE is process oriented and does not add as much value is it could. Last year, SSE processed more than 150,000 changes to the budget. The volume of changes is directly related to the degree of detail with which SSE attempts to centrally control, ex ante, budget execution. SSE must review and approve any changes to the congressionally enacted budget, down to movements between economic classifications for individual offices or budget users. 4.61 SSE focuses on process compliance—whether the procedures for changing the budget were followed, paperwork was properly completed and signed, and so forth. SSE does not assess the policy dimensions or implications of budget changes, nor apparently even whether there is sufficient budget authority for the change. The detailed control at best generates much paperwork and clerical work and provides a false sense of security (at worst it creates a disincentive to management and efficiency and perpetuates the administrative culture that dominates the bureaucracy). In 2003, one agency submitted a budget amendment for 1 billion pesos in new spending authority to cover contingent liabilities; SSE approved the request because “all the paperwork was in order.” 4.62 Most OECD countries have moved away from centralized, ex ante control regimes. Among OECD countries, budget management has moved away from such detailed control, replacing it with active monitoring and control at a more aggregate (program) level for critical points (such as total spending, personnel spending, and capital spending).61 Among OECD 61. A. Schick, The Changing Role of the Central Budget Office (OECD, 1998). - 109 - countries, 60 percent of budget offices apply two or fewer ceilings to agency budgets during execution.62 4.63 These countries apply the time saved to exercising more aggregate control over policy and program analysis, which enables them to focus on the critical issues that otherwise go unnoticed. The budget office takes on the function of management advisor, challenging the assumptions of service delivery models and efficiency of spending, adding value to the discussion. 4.64 Although technology can make centralized control more thorough, this runs the risk of lowering program efficiency. Traditional regimes of detailed controlhad limited technology that left some freedom of action and managerial discretion to geographically remote areas. With modern communication and information technology, however, centralized control can reach a degree never known before, which could be harmful to serve efficiency in the Mexican case unless SHCP reforms its control regime. 4.65 In recent years, Thailand faced a similar challenge to revise its control regime vis-à-vis line ministries. Thailand developed a hurdle approach, where the budget office crafted series of financial management competencies that spending agencies where expected to meet in order to be granted more flexibility over their budget.63 The competency standards covered seven areas: budget planning, output costing, procurement management, budget and funds control, financial and performance, reporting, asset management, and internal audit. A lesson learned through implementation in Thailand is that the standards cannot be set too high (e.g. international best practice), or the reform will fail. Nonetheless the hurdle approach could work well in Mexico in order to increase accountability in the public finance system. Treasury 4.66 Within SHCP, treasury handles cash management and accounts payable. As noted in Box 4.2, Mexico is among the minority of OECD countries that do not receive information on agency spending commitments (which could improve cash management), that do not forecast cash spending on a weekly or daily basis, that do not require all public revenues to be deposited in a treasury single account, where spending ministries are allowed to maintain interest-bearing accounts outside the treasury, and where information on actual spending is not available from spending ministries on a timely basis for effective monitoring. 4.67 The Treasury in Mexico does, however, operate a system of zero-balance accounts and linked sub accounts in the Banco de México, in order to achieve most of the objectives of a single-treasury account. Spending agencies operate some 3,000 zero-balance accounts through a number of commercial banks. TESOFE gives agencies authority to spend (that is, a line of credit) up to an agreed ceiling, and the agencies are reimbursed for expenditures incurred. T his is standard practice in countries with developed commercial banking sectors and allows remote regions to have full treasury services. Nonetheless, the current arrangements for cash management still have a few areas to strengthen: 62. World Bank/OECD Survey. Ibid. 63. Thailand’s hurdle approach to budget reform. PREM Note 73. The World Bank, August 2002. - 110 - • As payment for the commercial bank services, and in lieu of a conventional fee structure, commercial banks are permitted to operate a three-day float—funds are held for three days before being cleared. This is inefficient and not competitive. TESOFE is reviewing the arrangements with banks and considering changing the three-day float into a fee-based system of remuneration. TESOFE noted that the relationship is already evolving, as electronic commerce lowers the costs to banks of processing transactions. • TESOFE currently maintains a two-month liquidity reserve to meet shortfalls in revenue. Even taking account of the revenue volatility in Mexico, this seems excessive in comparison with good practice in many OECD countries. It also underlines the importance of current efforts to improve revenue forecasting, broaden the revenue base, and increase tax collections through SAT and other initiatives. • The treasury function is fragmented in Mexico: in particular, social funds are handled outside TESOFE. This is inefficient. In France, for example, social funds are held outside the central government budget, but contributions to and expenditures of the funds are processed through the central treasury system. This is efficient for cash management and consolidation of fiscal information; it also gives the government greater control over the management of social funds. Role of Secretaría de Función Pública 4.68 SFP has the authority and responsibility for monitoring expenditures for consistency with the budget and National Development Plan, evaluating investment projects, modernizing public financial management, handling procurement, combating corruption, and exercising overall internal audit and control in the federal public sector. In the past, SECODAM mainly carried out its auditing function in the narrowest sense of checking compliance with procedures—ex post, in addition to SSE’s ex ante checking. It will benefit expenditure management if the new SFP carries its function into the more substantive parts of its mandate. 4.69 SFP is unique among OECD countries, in that such functions are usually the responsibility of the Ministry of Finance. SFP sets the internal audit standards for the internal control offices within each spending unit. SFP has initiated reforms to improve transparency, automate and modernize internal control, and improve accountability in public financial management. 4.70 The information generated by SFP in reviewing spending unit and project management would be a valuable input into the formulation of the annual budget, and coordination of SFP with SHCP could be improved in this regard. A more detailed review of the structure and function of SFP and SHCP—of their roles and responsibilities, complementarities, and future directions—is recommended. 4.71 As the internal audit agency, SFP will have some overlap with responsibilities of the ASF. Both will have audit responsibility over the same entities, with SFP doing so on behalf of the executive branch and ASF doing so on behalf of the legislature. Although it is a valuable standard practice, they need to coordinate audits to avoid placing excessive burdens on spending agencies. A well-functioning SFP can ease the burden on the external auditor. - 111 - 4.72 The U.S. has a system of Inspectors General, internal audit units attached to individual spending ministries. These units have evolved over time from traditional audit functions to more selective, risk based auditing, program evaluation, inspection, as well as investigations (where audits reveal more severe management problems). As SFP reviews its role, it might consider moving towards higher value added activities such as evaluations, though it should never abandon its core audit functions. TRANSPARENCY AND ACCOUNTABILITY 4.73 Accurate, timely, readily accessible information on public finances is a prerequisite of accountability and an important element of obtaining feedback for improvement. Mexico has made strides in improving transparency, but further improvement is possible (see Box 4.6). Mexico can improve transparency with better reporting of public finance, including the timeliness, frequency, and level of reporting detail. Producing more user-friendly reports—and actuarial reports on pension and other large mandatory programs—would add substantially to transparency. Some improvements to mandatory program oversight and rules on costing of new policies before their enactment might also help to limit further creation of mandatory programs that limit budget flexibility and may not always be fully financed. Box 4.6 Transparency: Mexico Compared to the Other OECD Countries Compared to other OECD countries with needs to improve their spending transparency, Mexico is among the • 26 percent that do not provide monthly in-year spending reports • 38 percent that provide in-year comparison of actual to planned spending (although only at an aggregate level in Mexico) • 37 percent that do not have a medium-term fiscal framework stating targets or ceilings for expenditures, deficits, and revenues for the medium term with targets or ceilings for each subsequent budget year within that term (they have no annual expenditure ceilings against which to reconcile actual spending) • 40 percent that do not provide a formal comparison between the medium-term fiscal policy objectives and the government’s annual budget with explanations given for any deviations • 26 percent that do not produce multi-year cost estimates for all new spending • 54 percent reporting that 60–80 percent of the annual budget is consumed by mandatory spending (spending required by legislation, such as debt service or pensions, and generally not limited by annual budget laws) • 24 percent with no sunset provisions for mandatory programs • 46 percent that do not produce a citizen’s guide to the budget • 37 percent that do not require special reports on old-age program finances • 44 percent that do not require special reports on the viability of civil service pensions • 25 percent that do not provide regular actuarial estimates for civil service pension plans 4.74 The current approach to measurement and indicators of performance in Mexico has some drawbacks: - 112 - • Although many indicators are generated, frequent changes in the budget classification make it impossible to compare movements in the indicators over a run of years, which limits the potential value of this information for illuminating trends in the economy and in the efficiency and effectiveness of spending programs. • The present system of indicators cannot link the full (capital and recurrent) cost of delivering public services to the outputs achieved. • It is unclear how the information is meant to be used by SHCP or the government in setting budget priorities; by the internal and external audit agencies, SFP and ASF, in validating the information and analyzing the performance of spending entities; and by spending units in managing their expenditure programs, projects, and activities for improved efficiency and effectiveness. • It is unclear how the information, in its present form, increases transparency and accountability and how it is intended to be interpreted by Congress and the general public. Budget Classification, Transparency, and Accountability 4.75 The essential elements of budget preparation and classification procedures in Mexico are provided in the Manual of Programming and Budget,64 the Expenditure Policy Guidelines,65 the Budgetary Code for 2003 and 2004,66 and the Guidelines for the Implementation of the Program Structure 2003.67 4.76 The budget code is organized on three bases: administrative (A), functional (F), and economic (E). The functional classification is further subdivided into functions, subfunctions, processes, projects (in two categories: investment projects and innovation projects), and several classes of activities. There is also a program classification (see more below). 4.77 Frequent changes in classification in recent years have sought to move toward performance budgeting but have also undermined accountability. The budget code was amended in 2001 to fit the requirements of the National Development Plan, the law for which (Ley de Planeación, 2002) stipulates the division of expenditures into certain programmatic categories— sectoral programs, institutional programs, regional programs, and special programs. The number of budgetary codes has increased substantially, reaching 31 digits in 2001 and 38 digits in 2002. In 2003 the number of digits was reduced to 25 by simplifying the number of codes relating to categories in the National Development Plan, program structures, activities or projects, and processes. This fluctuation reflects the frequent changes in classifications, which has made detailed comparisons over time almost impossible. 4.78 Our main conclusions on the budget classification system are as follows: 64. Manual de Programación y Presupuesto para el Año 2002. 65. Lineamientos de Política de Gasto para la Formulación del Proyecto de Presupuesto de Egresos de la Federación para el Ejercicio Fiscal del Ano 2002. 66. Formulación de la Clave Presupuestaria 2003, 2004 67. Lineamientos para la Concertación de Estructuras Programáticas 2003 - 113 - • The economic and functional coding structure for 2003–04 is broadly satisfactory in relation to the government’s objectives, bearing in mind that current legal requirements and the relationship between the budget and the National Development Plan restrict the ability of the government to drastically simplify the data and hence the “catalogue” of codes, definitions, and procedures. • Mexico’s administrative classification is sound in theory, but practical improvements would give better support to performance monitoring and costing. The administrative classification is based on an approach similar to that found in many countries, in which the responsible unit within the line agency or budget entity is the core building block of the budget system. However, it is not clear whether this focus on the responsible unit creates a satisfactory link with the major policy objectives of the ministries or with the performance measurement system. Moreover, there is no information on number of personnel per responsibility unit, making it difficult to evaluate total inputs (staff and funding) and types of expenditure (economic classification) and to link these to outputs and results. • The economic and functional classifications are based on sensible national definitions, and it should be relatively straightforward to translate these codes into the formats required by the IMF and OECD for reporting public finance statistics. More important, however, the classification has to be useful to government for strategic decision making, to spending units for managing resources to achieve results, and to society for performance and accountability purposes. • The program classification now does not support accountability. The programmatic classification and the present array of performance indicators follows from the goals in the National Development Plan, but they do not match the definition of administrative units and functions. This makes it difficult to assign responsibility for performance and hold anyone accountable. • Too much of the data has limited information value. The classification codes and budget documentation include voluminous information on processes, projects, and activities. This information may be required by the current budget law, but it has little value (a) to devolve more authority to responsible units so that they can be held accountable for results achieved in their spending programs or (b) to feed information back to the SHCP that would be1 useful in analyzing budgetary outcomes and setting spending priorities in the following year’s budget. • Although some work is being undertaken to clarify the level of detail to which spending units will be held responsible (for example, by controlling spending at the level of broad aggregates for personnel spending, other recurrent expenditures, and capital spending, rather than for each sub-element of the economic classification), such a change in approach will need to be agreed with Congress, at least informally. Such excessively detailed control prevents spending agencies from managing their activities effectively and undermines efforts to improve performance and accountability. - 114 - Budget Documents 4.79 Gathering the right information and organizing it well are only the first steps for transparency. Documents and other dissemination must make the information available in a timely and understandable fashion to Congress and the public. 4.80 Comprehensive coverage. With respect to the budget documents provided to Congress and the public, by international standards, the budget coverage is comprehensive in most respects. Few extra-budgetary funds or other accounts or transactions are recorded and controlled outside the budget itself. In addition to expenditures incurred by the federal government and transfers to subnational levels of government, the budget also includes seven important public enterprises, including Pemex, that are under direct budgetary control. Also the budget has records of many other enterprises and agencies that carry out some public functions but are not directly controlled by SHCP. The extensive financial public sector is largely outside the budget, however, and includes important institutions like the development banks, financial- sector restructuring institutions, and PIDIREGAS. The government is now including their net deficits and debts as part of the published public-sector borrowing requirement and total financial requirements (while the narrower concepts of the “budget deficit” and the “federal debt” continue in parallel).68 4.81 Multi-year information. Although the budget mainly gives information only for one year, the government has already taken some steps toward a multi-annual process. The annual budget provides figures for capital investment projects for up to five forward years, and in 2002 and 2003 the National Financial Plan for Development (PRONAFIDE) provided multi-annual projections of the aggregate fiscal framework. The government could consider preparing and presenting the full budget on a multi-annual basis. This does not mean that the annual legal base of the budget would change, because the figures for the out-years would be presented to Congress for information and review only. A multi-annual approach, followed widely in OECD countries and increasingly in developing countries, typically provides a more secure and rational basis for setting expenditure priorities. 4.82 Oversight and accountability. Formal institutions that monitor and review government finances are an important feature of modern public finance systems, serving to improve policy, attain broader social consensus, and provide accountability. Mexico has made progress in building such institutions and is broadly comparable to other OECD countries (see Box 4.7) 68. The IMF’s ROSC and the World Bank’s CFAA discuss this issue further. - 115 - Box 4.7 Role of Legislature in Budgeting: Mexico versus OECD Countries Compared to other OECD countries, Mexico has some important strengths, putting it among the • 66 percent reporting that the legislature makes only minor (3 percent or fewer) changes to the executive budget proposal • 50 percent where the legislative budget committee has five or more professional staff to assist them • 28 percent with a specialized budget research organization attached to the legislature that conducts analyses of the budget. There are also notable weaknesses as Mexico is among the • 38 percent of OECD countries where the legislature does not exercise any annual oversight of mandatory spending • 30 percent that give their legislature two months or less to consider the budget (changing in 2004) • 50 percent where the legislature does not have any opportunity to formally debate or discuss overall budget policy prior to (or just after) the introduction of the executive’s budget • 54 percent where legislative budget committee hearings are closed to the public • 5 percent where more than 25 special professional staff are serving political parties and dealing largely with budget issues • 16 percent where political parties in the legislature receive funding for a special professional staff that deals largely with budget issues from party funds • 38 percent where the internal and external audit does not coordinate audit plans • 47 percent reporting that internal audit procedures are not or are only partially clear and subject to effective process review by external auditors • 20 percent where external audits take more than one year to get to the legislature after close of the accounting period • 42 percent reporting that it is not common for politicians to use performance measures in decision making. The Role of Congress 4.83 Since 1997, when Congress was not controlled by the party of the presidency for the first time in modern Mexican history, the accountability framework and the congressional role in policy and budget formulation have been evolving. Both the executive and legislative branches are still adjusting to their roles, particularly as one party no longer provides a venue for resolving differences. 4.84 Parties still play a very important role in the Mexican governmental system, partly because individual deputies serve only 3 years with no option for reelection. Unlike other OECD countries, where the legislature has professional staff advising on budget and fiscal policy who are attached to committees or organizations that are attached to the legislature, in Mexico professional staff are attached to the parties. While this empowers each party to engage actively in the budget debate, there is an absence of institutional structures that serve the legislature as a legislature and provide institutional memory, neutral advice for good legislative practice in - 116 - budget development, and a foundation for professional-level relations between the executive and legislative branches. 4.85 Compared to other OECD countries, Mexico has been notable for giving the legislature two months or less to consider the budget, giving the legislature no opportunity to discuss overall budget policy prior to addressing the detailed budget, and allowing the legislature to exercise little or no oversight over mandatory spending programs (see Box 4.7 for more detail). The limited changes that the legislature makes to the budget could suggest that it lacks sufficient time and resources to delve deeply into the budget. This is changing in 2004, with presentations of the budget in mid-September, over three months before the end-December deadline for passage. 4.86 Research on fiscal institutions suggests that effective procedures or rules can promote agreement among various actors to the budget process and thus contribute to good fiscal performance.69 Processes that promote better performance include the following: • Comprehensive information on spending and revenues, costs, and benefits • A process centralized in a few key decision makers • Common agreement on fiscal targets, arrived at early in the budget process, negotiated by all interested parties, and with enough enforcement to make the agreement binding • A strong, professional Ministry of Finance, with effective ability to monitor implementation of the budget and to maintain budget agreements. 4.87 In the United States, the U.S. Congress modified its own budget process in 1974 to provide a vehicle for coordinated legislative action on the budget. Among other changes, Congress created a “budget resolution,” adopted at the beginning of the congressional budget process (over six months before the end of the fiscal year) and setting overall revenue and spending limits. These limits guide congressional action on the budget and give it a coherent voice as an institution). Although a parliamentary system, Sweden uses a similar device to bind the legislature and executive. At the beginning of the budget process, the Swedish government submits the fiscal policy, expenditure envelope, and sectoral allocations to Parliament for approval. This is done to bind Parliament to the expenditure totals and sector ceilings and to prevent unproductive amendments that add to expenditure rather than encourage efficient choices within sectors. 4.88 Other countries have used coalition agreements on fiscal policy among political parties (Holland), fiscal rules with specific targets to bind executive and legislative branches (US federal government in the 1990s and many states), or fiscal responsibility laws requiring greater transparency as a means to assure fiscal discipline (New Zealand, Brazil). Research indicates that the optimal choice of instrument depends on what electoral systems and governmental systems the country has. More work along this line is recommended for Mexico, to develop specific institutions within the Mexican context. 69. See James Poterba and Jurgen von Hagen, eds., Fiscal Institutions and Fiscal Performance (Chicago: University of Chicago Press and National Bureau of Economic Research, 1999). - 117 - 4.89 Mexico’s present arrangements of limited time for congressional review, historical dominance of the executive in budget matters, limited congressional institutional capacity to engage in budget debates, and limitations on budget information for the use of policy officials all combine to reduce the positive effect that congress could have on public finances. At a deeper level, the rule against reelection of congressional deputies largely greatly reduces their incentive and ability to get involved in the substantive (and complex) budget issues or to invest time in fixing the institutional constraints. Recommendations 4.90 To make the National Development Plan a useful strategic planning exercise, it should be linked to the annual budget, should have sector-wide but still concrete objectives, and should have sectoral resource ceilings that provide a realistic framework for sector ministry planning. 4.91 To achieve whatever fundamental reallocations of budget are needed to align with national priorities, it would help to involve the President and the whole cabinet in setting the initial budget priorities, prior to submissions by individual ministries in summer. It would also help to introduce a medium-term perspective into its budget and policy process and to evaluate the multi-year impact of choices before decisions are taken, including new mandatory spending laws. Using multi-year budgeting, initially especially for capital, with estimates of recurrent cost of operating and maintenance, would help make investment more sustainable and efficient. In relation to congress, a more collaborative approach focusing on the criteria for selecting approved capital projects could work well in the Mexican context. 4.92 To improve the way SHCP serves it clients in the rest of the government, it could change the budget process in ways that encourage spending ministries to improve efficiency and effectiveness. This is under the control of SHCP—or at least the executive branch—and should not require new legislation. It would require closer collaboration between SHCP and SFP, and both organizations would need a common vision of reform. 4.93 Hard budget ceilings for spending ministries, approved by the president and issued at the beginning of the budget process, can begin to shift the emphasis in spending ministries away from requesting more resources and toward improving the spending of resources they already have. This means eliminating programs with no impact and shifting resources to programs with higher marginal return. To be effective, the ceilings would need to include all revenues and spending: recurrent and capital, discretionary and mandatory. Spending agencies would be expected to identify the optimal mix of the various spending options for achieving the government’s objectives. SHCP could improve the incentives of the spending ministries, by strengthening its function as challenger with a mandate from the President—demanding public evidence of effectiveness and efficiency in spending as prerequisites for increased budget allocation. 4.94 Reformulating the approach to performance measurement, taking account of other reforms, like a shift to a multi-annual expenditure framework, should increase the efficiency and transparency of the process for managing public expenditure, and Mexico has rightly emphasized the need for spending evaluations. For them to be fully effective, they will need to affect actual allocation of resources, which will require several steps: setting objectives (now in the separate - 118 - plan) within budget constraints; aligning units of evaluation with units of resource allocation; keeping the program structure stable so that the evaluations can have a effect (as well as to develop experience and accountability)70; giving managers enough authority to achieve results; and developing indicators within sub-sectors, with higher-level indicators to be monitored by SHCP. The role of SHCP, particularly SSE, should be to establish the framework and then to leave most of the actual evaluation work to the spending secretariats. Selected indicators would be included with the annual budget for these programs, and SHCP could use the performance indicators to assess whether additional investments in the programs are warranted or whether deeper program evaluations should be commissioned to better understand issues related to program performance. There should be annual performance reviews, with less frequent in-depth program evaluations. The fiscal aspect of the reviews should account for full costs of program delivery. To enable an orientation toward performance, the revised budget process should allow spending ministries greater choice in allocating their budget within the fixed ceiling and should move the focus from input to output and policy objectives. 4.95 The role of SFP needs to be defined more clearly, in coordination with SSE, Treasury and other agencies. It needs to strengthen its core audit function and move into evaluations in a way that complements what is done by SSE and by the Auditoria Superior de la Federación. For instance, under an MTEF framework, Egresos needs to make some real-time management assessments of spending effectiveness. On the other hand, SFP can operate on a longer horizon, not necessarily driven by the budget cycle, and do deeper program evaluations and impact analysis that would feed the budget process, but not be driven by it. To be most effective, SFP should consult with Egresos to get a list of programs that SHCP thinks are most in need of evaluation, where the biggest questions of effectiveness and efficiency lie. SFP might also usefully examine cross-cutting issues, such as procurement, capital spending, pay comparability, training programs, staff skill mix, hiring practices, spending agency internal control practices, and effectiveness of IT investments. In terms of audits, which would cover the above issues as well, SFP needs to coordinate its audit programs with the Supreme Auditoria to avoid duplication. 4.96 To strengthen further the treasury function in SHCP, it would help to • improve revenue forecasting and reduce 2-month liquidity reserve; • assess the prevalence and size of positive balances in the 3000 government accounts, improve reporting, and assure that everything other than true petty cash is netted out daily; and • reduce or eliminate the 3 day float allowed to commercial banks for collecting taxes and handling disbursements. 70. Each program classification should correspond to an administrative classification that identifies the responsible unit. Each classification should include both a program and an economic classification as well as staffing levels and outputs tied to those of the responsibility unit. The level of detail will depend on the needs of the executive and Congress. Program classifications should be built around existing programs and activities of government and tied to current law. The program structure should be stable and could be tied to elements of the National Development Plan, as desired. After two budget cycles, clearer patterns of performance will emerge, and this information can be used to set future budget levels as well as to target deeper program evaluation. - 119 - 4.97 To improve budget transparency and bring Mexico more in line with the best OECD practices, useful measures would include the following: • More timely, frequent, and detailed public financial reports, including contingent liabilities, debt guarantees, quasi-fiscal activities (for example, by the development banks), and other fiscal risks • Actuarial reports on pension and other large mandatory programs • Improvements to mandatory program oversight • Requirements for costing new policies before their enactment. • Tax expenditures (that is, subsidies to citizens, companies, and other taxpayers through tax exemptions and concessions rather than directly through public expenditure) • The expenditure, debt management, and other fiscal activities of subnational governments (states and municipalities). At present, relatively little information on such activities is provided in the budget or the regular financial reports prepared by SHCP. Such information could be included, not in the budget as proposals to be approved or enacted by Congress, but rather in supplemental tables or annexes to the budget, thus providing a more complete picture of public finances • A Citizen’s Guide to the Budget or Budget in Brief for distribution to citizens and key stakeholders • Regular (though not necessarily annual) inclusion of information on the actuarial balance of trust and pension funds. • Summarize data for senior decision makers. In the context of improving the budget documents, further work should be undertaken to analyze, interpret, and present information for decision makers in more user-friendly and manageable form rather than simply to produce large quantities of raw statistical data that can, of course, continue to be made available to those interested. • Special requests for data should be handled through ad hoc reporting rather than through amendments to the budget codes. Budget information on cross-cutting issues, such as expenditure on women’s programs or support for indigenous people, could be collected through special analyses of existing data and presented in supplementary tables rather than through the addition of new budget codes (overloading the codes could become a problem as the introduction of GRP and FMIS proceeds). Such special analyses could be a retrospective of how much of last year’s spending can be attributed to these special categories, and this analysis could be used to inform decision making, without becoming a formal requirement that the budget office has to conduct during budget development. 4.98 The internal reforms in SHCP should focus on integrating its functions to serve the secretary’s role and the government’s needs. Focusing on defined outputs for the secretary is a practical way of integrating the regular workings of each component of SHCP and reducing the current stovepipe operations. For example, early in the budget formulation, production of a fiscal policy paper, based on the updated PRONAFIDE and bringing together revenue, debt, and expenditure forecasts and policy in a multi-year economic framework, could serve as a vehicle for government and congressional discussions of policy priorities and constraints. - 120 - Reform Implementation: Sequence and Pace for Managing Politics and Logistics 4.99 Experiences in Mexico and elsewhere provide some lessons for how the sequence and pace of implementing reforms can help to manage the logistics and politics of change. Mexico has a long history of public financial management reform, from creation of a Comptroller of the Treasury to record revenues and disbursements in 1824,71 through recent efforts to strengthen tax administration and external audit. The pace of reform increased beginning in the mid-1950s, with several reforms introduced per decade thereafter. Reforms over the last 25 years have addressed budget formulation and execution issues. The creation of a separate Secretary of Planning and Budget to implement program budgeting was one major reform, followed by creation of a Secretary of Controlaria (Auditing) in 1982 to monitor budget execution and assure consistency between budget formulation and execution. The early 1980s also saw the passage of new rules for planning, in an attempt to strengthen the forward-planning function generally and to improve resource allocation. The mid-1980s saw movement toward some elements of new public management, with passage of the Federal Act on Public Agencies, giving public agencies greater management autonomy in exchange for clear performance commitments. 4.100 Reforms of the late 1980s and early 1990s focused on accountability and consolidation. The creation of the Social Audit Program to enlist beneficiaries in monitoring the impact of programs in the late 1980s and the merger of the Secretary of Planning and Budget with the Secretary of the Treasury in 1992 are examples of these efforts. Complementary reforms included agency performance agreements, decentralization of public expenditure, and civil service reform. 4.101 The 1998 reforms have had mixed results. A new program structure put in place in 2001 is linked more to the National Development Plan than to the ongoing activities of government. The performance evaluation system and SIAFF have yet to be fully realized. In contrast to prior or subsequent reforms, the 1998 reform was conceived as an overarching, whole-of-government reform (a big-bang approach). It was developed centrally by SHCP, without much involvement from spending units that would implement the reforms or a focused communication and training program. While both the broad objectives of the 1998 reform and the vehicles of reform (classification, evaluation, and technology) are similar to those pursued today, current reforms are more targeted at addressing specific problems. The big-bang approach is not being repeated. The increased political competition and more independent Congress have called for increased attention to the Communication Strategy (see Box 4.8). 71. Notable also for being created as an agent of the legislature. This was modified in 1917, when an autonomous auditor was established. - 121 - Box 4.8 Communication Strategy Communications strategy can entail many activities, for example: Citizen’s guide to the budget. Many countries are adapting these “budgets-in-brief,” which outline the budget process, explain the fiscal and macroeconomic situation, and even provide a brief summary of key government policies. These are annually produced by government, with the objective of making the budget process and policy more accessible to citizens, raising awareness, and building consensus on general policies. This is an effort to communicate directly with citizens. Media training. The media will report on budget and fiscal issues, but public finances are often confusing to all but insiders. Media reports often are inaccurate or misleading, which serves to confuse rather than to inform the public. Providing a half-day seminar on budgeting, terminology, and data sources to reporters can at least improve the probability of correct reporting. Public relations. In order to improve transmission of the government’s message and policy, reduce confusion or misinformation, and promote orderly press relations, many ministries of finance establish standard operating procedures for media contact and designate a public relations office or officer to field questions. This can assure a consistent government message as well as keep the public relations officer (and through him or her, policy officials) aware of emerging issues and popular opinion. Policy promotion. Occasionally it will be necessary to prepare citizens for major changes in policy, especially where fiscal retrenchment is necessary. It may also be necessary to reduce spending on very popular programs that otherwise would never be touched (sacred cows). Such changes can be successfully pursued, but doing so requires an active public relations campaign well in advance of the proposed change, in order to prepare the ground and begin changing public opinion. Process. Editorials by cabinet members, government white papers, think-tank reports, blue- ribbon panel reports, seminars, town hall meetings, speeches to key interest groups, academic studies, international organization studies, and appearances on talk shows can all serve as vehicles for getting the facts and the government’s message into the public domain and engendering debate. These vehicles can be orchestrated to produce a continuous stream of information, raise public awareness of the problem, its causes, and consequences, and explore the options for reform. Public debate can also provide valuable feedback on perhaps unseen complications of some proposals. Over several months, public opinion can change from opposition to support for reform. When public interest has been heightened and opinion has shifted, the government can release its proposal for change. The probability of success and sustainability has increased as a result of the preparatory work. 4.102 Since 2001, public expenditure manangement (PEM) reform in Mexico has made some limited progress. SSE recognizes the need to improve the overall budget process. The current system does not support efficient use of resources in spending ministries. The 2003 budget for education, for example, increased by nearly 4 billion pesos, which went almost entirely for salaries, with minimal impact on service quality or spending for the poor. Improvements in efficiency and effectiveness cannot be driven from the center and must occur through spending ministries. SSE has improved the capital budgeting process and continues to work on improvements. It has developed some basic budget automation tools and is beginning to implement them. This is not the big-bang approach, as many of the changes under way are small and within existing structures, which makes them more likely to endure. Rarely are reforms in the public sector achieved with grand visions or sweeping reforms. More common and effective are smaller, marginal changes that improve performance or solve a specific problem faced at that - 122 - point in time.72 At some point, these small changes cumulatively are noticeable from outside. SSE has been following this model, solving specific known problems. This is sound, but there are broader challenges to be faced, reflecting SHCP’s responsibility for the functioning of the entire public finance system in Mexico. 4.103 The current administration has sought to reform public financial management within a broader effort to improve the efficiency, effectiveness, and allocation of public services. Public finance reforms over the past several years have been led by Hacienda. Reforms in 2001–02 focused on drafting legislation reforming the public finance system. The laws stalled in Congress, and no change resulted. Reacting to that experience, the current strategy is to reform internal management within SHCP. The rationale is that if the way SHCP operates is improved measurably, then change will radiate out to the rest of government. A broad reform of the Ciclo Hacendario (treasury cycle) seeks to improve the internal operations of Hacienda from budget formulation through execution and reporting. SHCP staff are engaged directly in the reform, whose substance derives directly from the operational subsecretaries. The creation of a steering committee of top management is the correct model, because only top management can implement the reforms. 72. For other OECD experiences, see the Country Event Files in Public Management Reform: A Comparative Analysis. Pollitt, Christopher and Geert Bouckaert. Oxford University Press; Oxford, 2000. - 123 - - 124 - REFERENCES Ajwad, Mohamed and Quentin Wodon,. 2001. “Marginal benefit incidence analysis using a single cross-section of data.” Mimeo, World Bank. Aspe, Pedro and J. Beristáin. 1984. “Distribution of educative and health services.” In Aspe and Sigmund, eds., The Political Economy of Income Distribution in Mexico. Homes and Meier. Barro, Robert and J. Lee. 1996. Barro-Lee Data Set. www.worldbank.org/research/growth. Barro, Robert and J. Lee. 1998. “International Measures of Schooling Years and Schooling Quality.” http://www.worldbank.org/research/growth/ddbarle2.htm, 1996. Bourguignon, Francois, L. Pereira, and N. Stern. 2002. "Evaluating the poverty impact of economic policies: some analytical challenges." Mimeo, World Bank. Cámara de Diputados. (2003).” Evolución del Gasto Público por Ramos, 1980-2003.” México: H. Congreso de la Unión-Centro de Estudios de las Finanzas Públicas. Casco Flores, A. (2001). “Gasto Público para el Desarrollo Rural en México: 1994- 2000.” Manuscript, Santiago de Chile: FAO. Castro-Leal, Florencia and J. Dayton. 1994. "The incidence of public education expenditures in Mexico." Unpublished, World Bank. CEPAL. 1994. “El gasto social en America Latina: un examen cuantitativo y cualitativo.” Cuadernos de la Cepal, No. 73. CEPAL. 2001. Panorama Social de América Latina. CEPAL ONU. CIDE-ITAM. 2003. “Análisis de las Finanzas Públicas en México.” Centro de Investigación y Docencia Económicas, A. C. (CIDE) e Instituto Tecnológico Autónomo de México (ITAM). Mimeo. Comisión Nacional del Agua. (2003). “Estadísticas del Agua en México.” México: CNA- SUIBA. Comité Técnico para la Medición de la Pobreza. 2002. “Medición de la Pobreza: variantes metodológicas y estimación preliminar.” Serie Documentos de Investigación, SEDESOL. Corbacho, A. y G. Schwartz. 2002. "Mexico: experiences with pro-poor expenditure policies." IMF WP/02/12. Cord, Louise. and Quentin. Wodon. 2001. "Do Mexico’s Agricultural Programs Alleviate Poverty? Evidence from the Ejido Sector." Mimeo, World Bank. Cortez, F, D. Hernández, E. Hernández-Laos, M. Széquely, and H. Llamas. 2002. “Evolución y características de la pobreza en México en la última década del siglo XX.” Documentos de Investigación 2, Secretaría de Desarrollo Social, México. Cuenta de la Hacienda Pública Federal. 2002. SHCP, Gobierno de México. Dávila, Enrique, Georgina Kessel and Santiago Levy. 2002. El sur también existe: un ensayo sobre el desarrollo regional de México. Economía Mexicana: Nueva Época. - 125 - Davis, B., S. Handa, Ruiz Arranz, and M. Stampini. 2002. “Conditionality and the impact of program design on household welfare: comparing the effects of two diverse cash transfer programs on the rural poor in Mexico.” Mimeo, FAO, Rome. Davis, Sadulet, and de Janvri. 2001. Deininger, Klaus W. and A. Heinegg. 1995. "Rural poverty in Mexico." Mimeo, World Bank. Dorotinsky, William and R. Floyd. 2003. “Public Expenditure Accountability in Africa: Progress, Lessons, and Challenges.” Draft Africa Region Working Paper (World Bank, August 2003). Engel, Eduardo, Galetovic and Raddatz. 1999. “Taxes and income distribution in Chile: some unpleasant redistributive arithmetic.” J. Development Economics, 59:155-192. ENIGH. 1984, 1992, 1996, 2000, y 2002. Encuesta Nacional de Ingresos y Gastos de los Hogares, INEGI. ENN. 1999. Encuesta Nacional de Nutrición, INSP. Filmer, Dion and Lant H. Pritchett. 1998. “The effect of household wealth on education attainment around the world: demographic and health survey evidence.” World Bank WP, September 1998. Formulación de la Clave Presupuestaria. 2003, 2004. Friedman, Lustig, y Legovini. 1995. In N. Lustig ed., Coping with Austerity, Brookings. García, F. and F. A. Villagomez (2003), “Reforma al sistema de pensiones del ISSSTE”, Documento de Trabajo 258, División de Economnía, CIDE, Mexico. Gil-Díaz, Francisco and Wayne Thirsk. 1997. “Mexico’s protracted tax reform.” In W. Thirsk ed., Tax Reform In Developing Countries. World Bank. Gonzales, E and S. Parker. 1999. “Equity in the finance and delivery of health care: results from Mexico.” Unpublished. Hanushek, Eric. 2002. “Publicly provided education.” NBER WP 8799. Holmes, Malcolm and Alison Evans. 2003. “A Review of Experience in Implementing Medium-Term Expenditure Frameworks in a PRSP Context: A Synthesis of Eight Country Studies.” Discussion draft, London: Overseas Development Institute, Centre for Aid and Public Expenditure. IADB. 1998. Facing Up to Inequality in Latin America. IADB Annual Report. IMSS 2003, Informe al Ejecutivo Informe y al Congreso de la Unión sobre la Situación Financiera y los Riesgos del Instituto Mexicano del Seguro Social. IMSS, 2003. IMSS 2004, “Comparecencia ante el Senado de la República”, Abril 21 de 2004. IMSS. 2003. Informe al Ejecutivo Federal y al Congreso de la Unión sobre la Situación Financiera y los Riesgos del Instituto Mexicano del Seguro Social. (http://www.imss.gob.mx/imss/informes/ejecutivos_002_2003_06.html) INEGI. (2003). “Finanzas Públicas Estatales y Municipales.” México: INEGI. Kakwani. 1977. “Measurement Of Tax Progressivity: An International Comparison.” Economic Journal, 87:71-80. - 126 - Kanbur, Ravi. and Nora Lustig. 1999. “Why is inequality back on the agenda?” Dep. of Agricultural, Resource, and Managerial Economics WP 99-14, Cornell University. Kostopoulos. 1999. “Progress in Public Expenditure Management in Africa: Evidence from World Bank Surveys” (January). Lambert, P.J. 1993. The Distribution and Redistribution of Income. Manchester: University of Manchester. Lanlouw, P. and M. Ravallion. 1999. "Benefit incidence and the timing of program capture." World Bank Economic Review, 13,2. Le Houerou and Taliercio. “Medium-Term Expenditure Frameworks: From Concept to Practice. Preliminary Lessons from Africa.” Legovini, A., C. Bouillon, and Nora Lustig. 2001. “Can education explain changes in income inequality in Mexico?” Mimeo, IADB. Levy, Santiago and Sweder van Wijnbergen. 1992. "Mexican agriculture and the Free Trade Agreement between Mexico and the United States." The World Bank Economic Review, 481-502. Lineamientos de Política de Gasto para la Formulación del Proyecto de Presupuesto de Egresos de la Federación para el Ejercicio Fiscal del Ano 2002. Lineamientos para la Concertación de Estructuras Programáticas 2003 Lopez-Acevedo, Gladys 2001. "Professional Development and Incentives for Teacher Performance in Schools in Mexico." Mimeo, the World Bank. Lopez-Acevedo, Gladys y A. Salinas. 2001. “Earnings Inequality and Education Attainment After Mexico’s Economic Reforms.” World Bank, Report No. 19945- ME. Lustig, Nora 1989. “Magnitud e impacto del gasto público en el desarrollo social de México.” Investigación Económica 187. Maddison et al. 1992. La Economía Política de la Pobreza, la Equidad y el Crecimiento: Brasil y México (FCE). Manual de Programación y Presupuesto para el Año 2002. John Mayne and Eduardo Zapico-Goni, eds., 1997. Monitoring Performance in the Public Sector: Future Directions from International Experience London: Transaction Publishers, pp. 22–24; OECD. “Mexican Government Response to OECD Issues and Developments in Public Management Survey 1996–7.” Available at http://www1.oecd.org/puma/gvrnance/surveys/report97/mx.htm OECD PUMA, “Integrating Financial Management and Performance Management” (July 1999), p. 9 OECD PUMA, “Improving Evaluation Practices: Best Practices Guidelines for Evaluation and Background Paper” (January 1999), p. 24 Pánuco and Székely. 1996. “Income distribution and poverty in Mexico.” In Bulmer- Thomas, ed., The New Economic Model in Latin America and Its Impact on Income Distribution and Poverty. MacMillan. - 127 - Parker, S. 2003. “Evaluación del impacto de Oportunidades sobre la inscripción escolar: primaria, secundaria y media superior,” WP-263, División de Economía, CIDE. Parker, S. and C. Pederzini 2001. “Gender differences in education in Mexico.” In E.G. Katz and M.C. Correia, eds., The Economics of Gender in Mexico. The World Bank. Poder Ejecutivo Federal. 2001. Primer Informe de Gobierno, Anexo estadístico. Poder Ejecutivo Federal. 2003. Tercer Informe de Gobierno, Anexo estadístico. Pollitt, Christopher and Geert Bouckaert. 2000. Country Event Files in Public Management Reform: A Comparative Analysis. Oxford University Press; Oxford. Poterba, James and Jurgen von Hagen, eds. 1999. Fiscal Institutions and Fiscal Performance. Chicago: University of Chicago Press and National Bureau of Economic Research. Presidencia de la República. “Anexo del Informe de Gobierno.” México: Presidencia de la República, several years. Reyes Heroles, G.J. 1976. “Política fiscal y redistribución del ingreso.” Undergraduate thesis, ITAM, Mexico. Rosellón J. and L.F. López-Calva. 2002. “On the potential redistributive impact of electricity reform in Mexico.” WP-223, División de Economía, CIDE. Sadoulet, E., A. de Janvry, and B. Davis. "Cash Transfer with Income Multiplier: PROCAMPO in Mexico." World Development, Vol. 29, No. 6, pp. 1043-56. Scott, John 2002. "High Inequality, Low Revenue: Redistributive Efficiency of Latin American Fiscal Policy In Comparative Perspective." Studies On Poverty and Social Protection, Regional Policy Dialogue, Poverty Reduction and Social Protection Network, Inter American Development Bank (IADB). Scott, John 2003. “Public Spending and Inequality of Oportunidades in Mexico.” Forthcoming in Wodon, Q. (ed.), Public Spending and Inequality in Latin America, Washington, DC: The World Bank. Schick, Allen. 1989. The Changing Role of the Central Budget Office. Organization of Economic Co-operation and Development (OECD). SECODAM 1998, “Análisis del pasivo por obligaciones laborales de la administración pública federal”, enero 1998 Secretaría de Agricultura, Ganadería, Desarrollo Rural, Pesca y Alimentación,. (2002). “Descripción de los sectores agroalimentario y pesquero y características del medio rural.” Mexico: SAGARPA. Secretaría de Educación Pública (SEP). (2000). “Perfil de la Educación en México.” México: SEP. Secretaría de Educación Pública. (2003). “El Sistema Educativo de los Estados Unidos Mexicanos. Principales Cifras. Ciclo escolar 200-'2003.” México: SEP-Dirección General de Planeación, Programación, y Presupuesto. Secretaría de Hacienda y Crédito Público. (2002). “Decreto de Presupuesto de Egresos de la Federación para el Ejercicio Fiscal 2002.” México: SHCP. - 128 - Secretaría de Hacienda y Crédito Público. (2003). “Diagnóstico Integral de la Situación Actual de las Haciendas Públicas Estatales y Municipales.” México: SHCP-Unidad de Coordinación Hacendaria con Entidades Federativas. Secretaría de Hacienda y Crédito Público. (2003). Cuenta de la Hacienda Pública Federal 2002. México: SHCP-SSE. Secretaría de Salud. (2003). “Salud: México 2002. Información para la rendición de cuentas.” México: SSA. Secretaría de Salud. 2002. Salud: Mexico 2002. SSA: Distrito Federal. Sedesol. 2000. “Evaluación de Resultados del Programa de Educación, Salud y Alimentación.” Secretaría de Desarrollo Social. Sedesol. 2003. “Medición del Desarrollo: México 2000-2002.” Secretaría de Desarrollo Social. Selowsky, Marcelo. 1979. “Who Benefits from Government Expenditure? A Case Study of Colombia.” (OUP-WB). SEP. 2002. “La experiencia de la dirección general de evaluación en la educación básica y normal: 30 años de medición del logro educativo.” Secretaría de Educación Pública. Stern, Nicholas.1987. “The theory of optimal commodity and income taxation.” In Newbery and Stern eds., The theory of taxation for developing countries. The World Bank. van de Walle, Dominique 1998. “Assessing the welfare impacts of public spending.” World Development, 26(3): 365-379. van de Walle, Dominique 2002. “Behavioral Incidence Analysis of Public Spending and Social Programs.” In Evaluating the Poverty and Distributional Impact of Economic Policies. The World Bank. van de Walle, Dominique and Nead. “Public spending and the poor: what we know and what we need to know.” The World Bank. Wilkie, J. 1978. La Revolución Mexicana: Gasto Federal y Cambio Social. (FCE). World Bank (2004). Mexico: Poverty Assessment. Washington, DC: The World Bank. World Bank 2004, “Reforma del financiamiento para pensiones y actividades de asistencia social: evaluación del pasivo fiscal global de los sistemas públicos de pensiones en México”, Marzo 2002. World Bank. (2003). “Universal Health Insurance Coverage in Mexico. In Search for Alternatives.” Washington, DC: The World Bank. World Bank. 2001. World Development Report 2000/2001. World Bank. 2002. “Actions to Strengthen the Tracking of Poverty-Reducing Public Spending in Heavily Indebted Poor Countries (HIPC).” World Bank. 2003a. World Development Report 2002/2003. World Bank. 2003b.”Inequality in Latin America and the Caribbean: Breaking with History?” The World Bank. - 129 - World Bank/OECD. 2003. “Survey on Budget Practices and Procedures.” Available at http://ecde.dyndns.org. As of December 2003, 27 OECD and 17 developing countries are included in the database. Yaqub, S. 1999. "How equitable is public spending on health and education?" Background paper for WDR 2000/1, World Bank. - 130 - ANNEXES - 131 - - 132 - ANNEX 1 DISTRIBUTIONAL AND FISCAL IMPACT OF POWER SECTOR SUBSIDIES BACKGROUND The establishment of national electricity tariffs 1. In 1960 the Mexican electricity industry was nationalized. Existing private utilities, serving specific regions, gradually merged into CFE, the state-owned utility72. At that time, approximately one third of all households had access to electricity; today that figure is over 95 percent. 2. At the time of nationalization tariffs were regulated and approved by COTEG73. Existing tariffs were classified according to use and between 1960 and 1964 a process of consolidation into uniform national tariffs took place. For the residential sector 3 tariffs were established in 1962, each with a fixed charge and a single energy charge. Initially, the energy charge was 40, 50 or 60 Mexican cents per kWh; by 1964 all 3 tariffs had been merged into the 50 cents per kWh tariff. Inflation was approximately 3 percent per year; therefore, consumers in the low tariff had a real increase of about 18 percent in 2 years. 3. This Tariff 1 remained without change until October 1973. Nevertheless, it was an adequate tariff until 1970, since inflation was low, while the power industry kept incorporating important economies of scale both in generation and transmission. On average there were no subsidies, although there were cross-subsidies between regions, as supply costs varied across the country while the tariff was uniform throughout. The introduction of subsidies 4. COTEG had become part of the Department of Commerce, which, during the seventies, saw the control of inflation as a priority goal. Prices of goods and services provided by the State were either left unchanged or, at any rate, declined in real terms. Electricity was no exception. 5. Inflation took off during the seventies but most tariffs were not increased until 1973. Therefore, because high inflation was not acted upon, subsidies to residential, and most other, consumers began in 1971; however, this was the result of inaction, rather than a premeditated energy policy. 6. In 1973 the structure of tariff 1 was changed, the fixed charge was eliminated and three energy blocks were established: 0 to 50 kWh, 51 to 100, and over 100. The 72. An exception was Luz y Fuerza del Centro, the largest private utility, then serving Mexico City and surrounding communities. First, legal difficulties with shareholders and, later, political considerations concerning the independent LFC union, postponed indefinitely the merging of LFC into CFE. Today LFC is fully state-owned, it maintains its territory, and has become a distribution company, buying 97 percent of its energy from CFE and generating the rest from remaining old power plants. 73. Comisión de Tarifas de Electricidad y Gas - 133 - corresponding energy charges were 0.40, 0.60, and 0.90 pesos per kWh. Therefore, with the 1973 tariff review, subsidies were formally instituted by, first, eliminating the fixed charge (although a small minimum charge was put in place), and second, explicitly subsidizing up to 100 kWh per month of energy. In that year average consumption was 76 kWh per month, which suggests possibly some 70 percent of consumers had all their consumption subsidized, while the other 30 percent had an efficient price signal at the margin, but they still received the full subsidy for the fixed charge and for the first 100 kWh of consumption. Although there are no cost figures available for that time period, it is thought that cost was reflected in the marginal price, that of the third block. This would imply all consumers were subsidized. The summer subsidy 7. A new kind of subsidy was introduced in 1974. As an answer to the tariff increase experienced in 1973, tariff 1A was established “in order to help consumers in very hot climates”. This tariff was applicable to localities where the average monthly temperature exceeds 25°C, during at least 4 months of the year. For this tariff, and for those 4 months, the second block was extended to cover from 51 to 250 kWh per month, enough to subsidize all energy to almost all consumers. Analysis of subsidies 8. Since 1974 residential tariffs in Mexico contain 3 types of subsidies: 1) there is no fixed charge74, 2) there are several blocks of subsidized energy, and 3) there are summer subsidies consisting of longer energy blocks with lower prices. 9. Since their inception in 1973, the first two types of subsidies were meant to help low income families, given the very high correlation between income and electricity consumption. However, as stated above, from the beginning those subsidies were excessive, benefiting all consumers. 10. On the other hand, the summer subsidies were established for political reasons, as a direct response to well organized complaints from the population of Mexicali, in Baja California. This city is located across the USA border; it has extremely high temperatures during the summer, and has been a free trade zone. Starting in the sixties, people in Mexicali began importing air conditioning units from the USA, some of them quite inefficient second hand units. When the tariffs were increased in 1973 well to do families led a revolt in Mexicali, mobilizing other consumers. They got the new tariff 1A, and they learned their lesson. For the last 30 years, every time the government increases tariffs they repeat the exercise, each time with a better organization, gathering the support of local authorities, as well as all sorts of political forces and civil groups. The residential tariff alphabet has been expanding ever since. 74. With one exception, for a small fixed charge was introduced in 2002 to a small segment of consumers - 134 - 11. Certainly, the summer subsidies do not respond to any economic rationale. Summer subsidies encourage energy consumption when it is more expensive to produce and supply; they translate into additional investment requirements to serve peak summer demand, (capacity which goes unused for the rest of the year75); they benefit mostly high- income families (poor people do not have air conditioning), and they encourage growth in inefficient air conditioning investment and use in the warmest regions 12. Since 1975 all three types of subsidies have persisted and increased. To this end, additional tariffs have been introduced; fixed charges have not been put in place76, blocks with price discounts have been extended, and the summer season duration has been prolonged. 13. In 1988 tariffs 1B and 1C were introduced for localities with summer temperatures exceeding 28°C and 30°C, respectively. They are extensions of tariff 1A, with longer blocks; therefore more energy is subsidized. 14. In 1990 the tariff structure was changed. Energy blocks in tariff 1 were redefined as follows: up to 75 kWh, from 76 to 200 kWh, and over 200 kWh. At that time some 40 percent of consumers had monthly consumptions under 75 kWh, 50 percent were below 200 kWh, and 10 percent were above this level. In this review, the third block was deliberately priced significantly above cost77 with the intended purpose of compensating at least some of the subsidy. Also, it was argued that the excess price addressed externalities related to environmental impacts from power plants. 15. Tariffs 1B and 1C were similarly modified and increased but, as has been the rule in tariff reviews, a new tariff, 1D, with much longer subsidized blocks, was introduced for localities with summer temperatures above 31°C. For similar reasons, and with additional subsidies, tariff 1E was introduced in 1995, for locations with summer temperatures above 32°C. Finally, in April last year; tariff 1F was established for locations with summer temperatures above 33°C. 16. By February 2002 there were 6 basic tariffs each distinguishing winter and summer seasons; except tariff 1, which is uniform throughout the year. Therefore, there were 11 different ways of billing consumers. 17. Policy goals regarding tariffs have never been made explicit. Rather, subsidies have been the result of haphazard attempts to reduce them, followed by well-organized protests and counter measures that cancel out some or most of the price increases. The 2002 tariff review 18. In February 2002 a major residential tariff review was undertaken. Its purpose was to reduce subsidies overall and to direct them to lower income groups. 75. For the Baja California load curve, the maximum demand during the winter season is lower than the minimum demand for the summer season. 76. With one exception, for tariff DAC, to be explained latter. 77 . Probably in the order of 30 percent above cost. - 135 - 19. For each of the 6 basic tariffs (1 through 1E) three thresholds of consumption were determined. They were to segregate consumers into three groups with approximately 75 percent, 20 percent, and 5 percent of consumers. The first group would maintain the existing level of subsidy; subsidies would be reduced for the second group and entirely eliminated for the third. 20. To that end, existing tariffs continued to be applied to consumers in the first group. For the second group of consumers, those beyond the 75 percent threshold, the price for the intermediate energy block was increased by approximately 25 percent78, whereas the third block, known as the excess consumption block, which was already well above supply cost, was maintained. Consumers below the first threshold of consumption do not reach this third block. The new tariff 1 varies also between summer and winter. 21. The basic structure of all the low and moderate (subsidized) tariff schedules consists of a token minimum charge79, plus a number of blocks with increasing energy charges. For the first threshold, the tariff consists of a minimum charge and 2 energy blocks80 with very low energy charges. For the second threshold, the tariff has 3 energy blocks (for basic, intermediate and surplus consumption), a first block with a very low energy charge, a second block still quite subsidized, and a third block with a very high price. This author estimates the marginal price for consumers in the second group is between 50 percent and 80 percent above true marginal cost, varying across the different tariffs. However, very few consumers reach the third block of consumption81. 22. The block structure of tariffs, particularly the very low first blocks and the feature that extends the middle block in hot climates in summer months, creates cross-subsidies within the residential customer class. The majority of customers, whose usage is no greater than the size of the second block, pay well below the marginal cost of service and face a price signal at the margin that is inefficiently low. 23. Finally, for the 5 percent of consumers with the highest consumption levels, new tariffs, called DAC (“Doméstica de Alto Consumo”), were established. For the marginal cost studies the country is divided into six regions: Baja California, Baja California Sur, Noroeste, Norte-Noreste, Sur-Peninsular, and Central. Therefore, six DAC regional tariffs were determined; these tariffs take into account the summer weather only in as much as it affects the regional and seasonal supply costs. That is, the DAC tariffs distinguish summer and winter seasons, but they do it at the regional level, not at the local level, and they attempt to reflect only seasonal and regional variations in power system supply costs. 24. DAC tariffs are supposed to represent supply cost, including marginal cost plus the proportional allocation of the revenue gap82. However, in order to avoid lowering the effective price paid by any of these consumers, the already excessive energy charges of 78. The increase was only about 15 percent for the warmest regions 79 . Equivalent to 25 kWh of monthly consumption. 80 . Except the winter tariffs 1E and 1F, which now have 3 blocks each, applicable to the first group 81. In Annex 1, the detailed structure of tariffs can be appreciated 82. The difference between the revenue requirement and the marginal cost - 136 - the third block were pretty much kept in place, with small adjustments related to variations in regional supply costs. Two energy blocks were defined, the first one covering 500 kWh for the 6 regional tariffs. Also, a fixed charge was put in place, which, in order to compensate for the excessive energy charges, is clearly much lower than the fixed cost of supply. By September 2002 the fixed charges for the DAC tariffs varied between 31 and 36 pesos per month (about 3.20 and 3.60 US dollars, respectively). 25. The following graph shows tariff 1 and tariff DAC for the central region, both for the summer season, for the month of September 2002. Tariffs 1 and DAC 3 2.5 2 pesos/kWh DAC 1.5 Tariff 1 1 0.5 0 38 63 88 110 133 148 163 188 213 244 276 350 426 487 551 600 kWh/month 26. Predictably, in April 2002, tariff 1F was created, in order to accommodate demands from the people of Mexicali, and other localities, to cushion the impact of the February tariff review. Residential tariffs today 27. There are many residential tariffs in Mexico and all of them vary monthly according to predetermined adjustment factors. Applicable billing schedules vary by temperature, season, and consumption level. 1st) Localities are distributed in 7 basic categories by maximum summer temperatures. These are the basis for the nominal tariffs: 1, 1A, 1B, 1C, 1D, 1E, and 1F. 2nd) For each temperature category 2 seasons (summer and winter), of 6 months each, are established. 3rd) Three billing schedules are applied to the populations of each of the 14 previously defined categories. These schedules are applied to “low”, “moderate”, and “high” consumption levels. - 137 - 28. The first two schedules are heavily subsidized while the third, (DAC) is purported to represent the cost function for the corresponding regions and seasonal categories. 29. In theory customers in each of the 14 tariff combinations defined before (7 basic tariffs, 2 seasons) can, once the high consumption threshold is reached, be billed in six different forms, depending on the region they are located. So, according to each customer’s consumption level, there are 8 different options for billing: low, moderate, and 6 regional high consumption categories. 30. In summary, in Mexico there are today 112 different billing possibilities, which result from considering: 7 basic tariffs, 2 seasons, and 8 billing options by consumption level. 31. In the next section we will see the impact of the 2002 tariff review on prices and subsidies, which was quite good. On the other hand, Mexico ended up with an extremely complex tariff system for the residential customer class. TRENDS IN ELECTRICITY SUBSIDIES 32. In this section we examine, first, the evolution of sales and consumption; this is followed by an analysis of the evolution of electricity prices83, and that of subsidies. 33. For the year 2002 the average number of residential consumers was 22.379 million, sales were 39,030 GWh (24.3 percent of total sales) and revenues reached 31,095 million pesos (26.7 percent of total revenues). The average price was 77.4 Mexican cents per kWh (8 US cents84), average consumption per household was 145 kWh per month, and the average monthly bill was 112.56 pesos85 (11.65 US$). Total subsidies to residential consumers were 31,336 million pesos, which means the price to cost ratio was, on average, just under 50 percent86. 83. At this point we are analyzing official figures only. For 2002 there are no separate figures by basic tariffs due to the fact that CFE is publishing statistics for the DAC tariff at the national level only. 84. This figure corresponds to an official exchange rate of 9.666 pesos per US dollar. It would become less than 7 cents per kWh if computed with power purchase parity. 85. Plus 15 percent VAT. 86. These figures are based on standard reference CFE cost; they do not include excess cost in LFC, which is covered by Federal Transfers. In 2002 such transfers, for residential consumers, amounted to 5,870 million pesos, or 92 Mexican cents for each kWh sold by LFC. - 138 - MEXICO CONSUMPTION OF ELECTRICITY IN THE RESIDENTIAL SECTOR RESIDENTIAL CONSUMERS SALES MONTHLY SALES AVERAGE AVERAGE TARIFFS YEAR (thousands) (GWh) CONSUMPTION SHARES PRICE PRICE (KWh/consumer) (%) ($ / kWh) (2002$/kWh) ALL 1993 16,192 25,511 131 0.20 0.78 1994 16,838 27,781 137 0.21 0.77 1995 17,715 28,465 134 0.25 0.68 1996 18,055 28,484 131 0.32 0.64 1997 18,621 29,641 133 0.38 0.62 1998 19,275 31,691 137 0.44 0.62 1999 19,914 33,370 140 0.49 0.60 2000 20,698 36,127 145 0.56 0.62 2001 21,512 38,342 149 0.61 0.64 2002 22,379 39,030 145 0.77 0.77 TARIFF 1 1993 9,765 12,578 107 49.3% 0.19 0.75 1994 10,129 13,669 112 49.2% 0.21 0.74 1995 10,539 13,944 110 49.0% 0.25 0.67 1996 10,833 13,680 105 48.0% 0.32 0.64 1997 11,133 13,963 105 47.1% 0.37 0.62 1998 11,471 14,744 107 46.5% 0.43 0.62 1999 11,787 15,511 110 46.5% 0.49 0.60 2000 12,224 16,610 113 46.0% 0.56 0.62 2001 12,670 17,430 115 45.5% 0.60 0.63 TARIFFS 1A to 1E 1993 6,428 12,933 168 50.7% 0.21 0.80 1994 6,710 14,112 175 50.8% 0.22 0.80 1995 7,176 14,521 169 51.0% 0.26 0.68 1996 7,222 14,804 171 52.0% 0.32 0.64 1997 7,488 15,678 174 52.9% 0.38 0.62 1998 7,804 16,947 181 53.5% 0.44 0.63 1999 8,126 17,859 183 53.5% 0.49 0.60 2000 8,473 19,517 192 54.0% 0.56 0.63 2001 8,842 20,912 197 54.5% 0.61 0.64 Electricity Consumption 34. During the last decade average consumption of electricity declined in 1995, 1996 and 2002. In the first 2 years this occurred even as the real price of electricity decreased by over 12 percent (1995) and by almost 6 percent (1996). For those 2 years the decrease in electricity consumption was the result of the income effect, which ensued after the December 1994 crisis. 35. Contrastingly, even as official figures indicate that average wages increased in real terms in 2002, electricity consumption per household declined by 2.2 percent. This is the result of the price effect, which grew by over 21 percent last year. 36. Overall we can appreciate a clear tendency for electricity consumption per household to increase in parallel with disposable income. Figures for 2000 and 2001 are interesting in that they show that small real price increases (3.6 percent and 2.1 percent, respectively) do not affect the trend growth of electricity demand. 37. When these figures are broken down by temperate (tariff 1) and warm summer zones (all others) one can see that electricity demand grows faster for the latter, its share - 139 - of sales increasing from 50.7 percent in 1993 to 54.5 percent in 2002. During the decade, electricity consumption per household grew by 7.5 percent in the temperate zones and by 17.3 percent in the warm summer zones. Electricity prices 38. Given the complex tariff structures, the average electricity price results from the combined effect of price itself and the level of consumption. Between 1993 and 1999 the real price of electricity declined every year in all zones87; the average annual rate being 3.7 percent for temperate summer zones and 4.6 percent for the warm zones. 39. For the last 3 years the real price of electricity for residential consumers has been increased, though modestly during the first two years. Then, as a direct result of the change of tariff structure undertaken in February 2002, for the year the average price grew by 21.4 percent. This increase was enough to bring the average price back to its 1994 level. Subsidies 40. Through the years, with a few isolated exceptions, subsidies have increased in real terms for a variety of reasons: • demography; an increasing number of consumers • higher disposable income; which translates into higher household consumption levels • policy: (1) the tendency to let prices decline in real terms88, (2) the introduction of new summer tariffs, with still additional subsidies, (3) more recently, the active reassignment of localities into tariffs with higher subsidies. 41. The evolution of subsidies is shown in the following table89. 42. Subsidies declined slightly in 1998, though this was possibly the result of a cost reduction, since consumption increased by 3.5 percent while prices remained essentially constant. 43. The only significant reduction in subsidies occurred last year, when they decreased by almost 5 percent in real terms. Before, subsidies had increased at an average annual rate of 6.8 percent, from 1993 to 2001. 87. Except during 1998, when real prices were essentially flat. 88 . Though, as we have seen, this policy has changed for the last 3 years. 89 . For some years, figures for LFC, and therefore for total subsidies, are own estimates. - 140 - MEXICO EVOLUTION OF ELECTRICITY SUBSIDIES TO RESIDENTIAL CONSUMERS SUBSIDIES SUBSIDIES YEAR (current (2002 million pesos) million pesos) 1993 5,019 19,422 1994 5,720 20,695 1995 8,438 22,614 1996 12,963 25,851 1997 16,260 26,882 1998 18,512 26,400 1999 23,113 28,272 2000 29,160 32,577 2001 31,373 32,951 2002 31,336 31,336 Subsidies versus other indicators90 44. At the global level, electricity subsidies for all sectors reached 44,975 million in 2002, that is 0.7 percent of GDP. Subsidies also were equivalent to 38.6 percent of aggregate electricity revenues for CFE and LFC, and to 118 percent of total investment91 in the power industry. 45. The balance of LFC shows it has been technically bankrupt for years, its debts exceeding its assets. This is so because its running costs are much higher than those of CFE, while tariffs are the same for LFC than for the central region of CFE. The Mexican government defines CFE’s costs, as standard costs for the power industry, tariffs and subsidies are determined against these standard costs. Federal transfers that, in 2002, reached a total of 17,549 million pesos cover excess costs for LFC. Of this amount, 13,091 million pesos paid for excess operating expenses while the rest went to investment, 3,332 million, and other applications. 46. In summary, because of low tariffs and LFC inefficiencies, during 2002 the power industry absorbed 62,524 million pesos of fiscal resources. This is the equivalent of 1 percent of GDP, almost 54 percent of total electricity revenues, and 164 percent of total investment in electricity. 47. In particular, subsidies to residential consumers were 31,336 million pesos, an amount roughly equal to revenues from that class of consumers, 31,095 million pesos. For every peso paid by residential consumers, they were subsidized another peso. This 90. Whenever applicable, figures refer to cost accounting for tariff purposes, which differ slightly from figures found in financial statements. 91. According to SENER’s “Prospectiva del Sector Eléctrico 2002-2011”, in 2002 total investment in power was 38,195 million pesos, including 23,338 million of private investment. - 141 - amount is equivalent to 0.5 percent of GDP and is, by far, the largest welfare program in existence in Mexico. In effect, electricity subsidies exceed the combined cost of welfare programs Progresa and Procampo, which, together, received 28,854 million pesos in 2002. In the next section the allocation of electricity subsidies among income level groups is examined. DISTRIBUTIONAL ANALYSIS OF RESIDENTIAL TARIFFS 48. This exercise is built upon the basic assumption that there is a one to one correspondence between deciles of electricity consumption and deciles of household income. 49. This assertion is supported by empirical evidence. Results of research carried out by this author during the seventies and eighties, while at CFE, and a survey by INEGI during the early nineties, indicate that households in each income category show a normal distribution of electricity consumption. The means of such electricity consumptions keep a very high correlation with income brackets; although standard deviations are somewhat large, the correlation coefficient for income level and electricity consumption is close to 1. The results were always statistically significant and the distributions well behaved, conforming to the Law of Large Numbers. 50. Statistically, even in the presence of some non-linearities in the price functions, the tails of the normal distributions tend to balance each other out. This fact is important for the purpose of analyzing allocation of subsidies by decile. However, from the policy perspective this should be a concern, since large standard deviations imply that decile subsidy distributions overlap and, therefore, tariff increases cannot be precisely targeted to income level categories. 51. From a different perspective, income and expenditure surveys indicate that, for a given tariff structure, income level is by far the most important explanatory variable for electricity consumption. Other, much less important, explanatory variables are: number of members in the household, size of the locality, and educational level. 52. Although there are no available figures for current subsidies at the State level, from previous studies we know they concentrate heavily in the 3 north-western States: Sonora, Baja California, and Sinaloa, where the summer temperatures are extreme and income levels are higher. At the other extreme, states with the poorest populations: Chiapas, Oaxaca and Guerrero, have the lowest per household electricity consumption levels and, therefore, receive lower subsidies. Distribution of subsidies by income deciles 53. The Annex describes the database and the methodology used to compute subsidies by decile; also, 14 tables are included in the Annex, illustrating all steps followed for tariff 1. - 142 - 54. After subsidies by decile were available for all basic tariffs the final results were added into a single table. An analysis of results follows, starting with the distribution of households and consumption by decile. CONSOLIDATED RESULTS FOR ALL RESIDENTIAL TARIFFS ELECTRICITY SECTOR, MEXICO ALL RESIDENTIAL TARIFFS ANNUAL 2002 Monthly average consumption (kWh) 140.8 Number of households 23,851,010 Total annual consumption (GWh) 40,298 SUBSIDIES BY DECILE AVERAGE AVERAGE AVERAGE UNITARY INCLUDING HOUSEHOLDS WITHOUT ELECTRICITY DECILE CONSUMPTION COST INVOICE SUBSIDY Cumulate kWh/month $/hh-month $/hh-month $/kWh $/hh-year M$/year % % 1 5 39.49 21.43 3.88 217 517 1.6% 1.6% 2 24 65.36 14.38 2.10 612 1,459 4.7% 6.3% 3 55 106.19 23.73 1.49 990 2,360 7.5% 13.8% 4 86 145.91 35.39 1.29 1,326 3,163 10.1% 23.9% 5 113 181.79 47.09 1.19 1,616 3,855 12.3% 36.2% 6 138 214.83 58.58 1.13 1,875 4,472 14.3% 50.5% 7 165 249.98 74.41 1.06 2,107 5,025 16.0% 66.5% 8 202 297.82 130.69 0.83 2,006 4,784 15.3% 81.8% 9 255 367.83 216.73 0.59 1,813 4,325 13.8% 95.6% 10 365 512.37 464.21 0.13 578 1,378 4.4% 100.0% kWh/month M$/month M$/month $/kWh $/hh-year M$/year 140.8 5,203 2,592 0.78 1,314 31,339 55. The unitary subsidy column shows that the subsidy per kWh is largest for decile 1 and declines for each successive decile; however, because monthly average consumption of electricity grows from 5 kWh in decile 1 to 365 kWh in decile 10, the overall distribution of subsidies by deciles is quite different. 56. Subsidies by decile are presented in 4 different columns, indicating respectively: annual subsidies by household, total subsidies in millions of pesos, percentage of subsidies allocated to each decile, and the cumulative distribution of subsidies. 57. Decile 7 captures 16 percent of total subsidies, the largest share, with deciles 5 to 9 receiving also high proportions of subsidies. The lowest share goes to decile 1, with only 1.6 percent of subsidies, while deciles 2 and 10 receive less than 5 percent each. 58. During 2000 the government, together with external consultants, undertook a study of residential tariffs and subsidies92. The allocation of subsidies by deciles showed a regressive distribution, as indicated by the following subsidy shares by decile: 5.1 percent for decile 1, 8.9 percent, 8.5 percent, 9.1 percent, 9.8 percent, 10 percent, 10.7 percent, 11.6 percent, 11.4 percent, and 14.9 percent for decile 10. These figures are not directly comparable, since they do not include households without electricity service. 92. At that time the author, in his capacity as Senior Advisor to the Secretary of Energy, participated in this study, being responsible for the technical supervision of the project. - 143 - 59. What is clear is that the 2002 tariff review, by concentrating impacts on deciles 8 to 10, is a big improvement in terms of reallocation of subsidy shares. Assessing the distribution of subsidies by decile 60. Before proceeding any further we must reflect on the fact that, as explained on the methodology, in the Annex, in order to reproduce the DAC tariffs as close as possible, the equivalent average cost function for the whole country is taken to consist of a monthly fixed charge of 33.52 pesos per customer. This, in turn, yields a national average energy charge of 1.323 pesos per kWh. This function is the weighted average of cost functions estimated for each basic tariff and for each season. Based on these estimated cost functions the allocation of subsidies by decile are undertaken and their specific values are computed. 61. This author estimates that, for 2002, the actual monthly fixed cost ought to be at least 80 pesos per customer, which implies an average energy cost of approximately 1 peso per kWh. A quick exercise using this function would leave decile 7, with 15 percent of total subsidies, still capturing the largest share. However, decile 10 would end up overpaying cost, with a cross-subsidy equal to 2.1 percent of total subsidies. The subsidy share for decile 9 would fall from 13.8 percent to 10.2 percent, with a lower decline in decile 8. Decile 6 sees a very small reduction, while subsidy shares increase for deciles 1 to 5. 62. These results ought to be much closer to reality; nevertheless it was established that, for this analysis, the estimated cost functions that more closely approximate the DAC tariffs would be used. Allocation of subsidies by tariff and decile 63. The following table shows how households and sales are distributed by basic tariff and decile. As explained in the methodology, in the Annex, tariffs 1E and 1F were merged into one, because the reassignment of localities from the first to the second that took place during 2002 impeded an isolated analysis. 64. LFC, which only sells electricity in tariff 1, shows an average consumption that is only 5 percent lower than tariff 1 for CFE. 65. Consumption patterns for tariff 1A differ from tariff 1 only in deciles 9 and 10, particularly the latter, due to the introduction of air conditioning. Tariffs 1B and 1C also display very similar consumption levels and patterns. 66. Other tariffs show much higher consumptions for all deciles. Consumption patterns for tariffs 1E and 1F are similar to those found in several developed countries. - 144 - HOUSEHOLDS AND CONSUMPTION BY TARIFF AND DECILE TARIFF CFE LFC TOTAL 1 1A 1B 1C 1D 1E and 1F 1 Households 8,911 1,808 3,355 2,811 592 1,354 5,021 23,851 Sales (GWh) 11,847 2,573 6,797 5,945 1,716 5,020 6,399 40,298 DECILE kWh/month 1 4 4 4 5 11 13 4 5 2 20 18 24 27 55 65 19 24 3 45 41 56 66 117 141 43 55 4 69 64 92 105 170 202 66 86 5 91 88 132 141 209 244 87 113 6 111 111 166 175 247 278 107 138 7 134 136 201 209 289 313 128 165 8 162 170 241 251 331 413 156 202 9 203 229 300 308 413 557 195 255 10 269 326 472 474 573 864 258 365 111 119 169 176 242 309 106 141 Includes households wthout electricity and households in DAC tariffs 67. The next table shows subsidies and subsidy shares, by tariff and decile. SUBSIDIES AND SUBSIDY SHARES BY TARIFF AND DECILE TARIFF CFE LFC TOTAL 1 1A 1B 1C 1D 1E and 1F 1 Estimated Cost (M$) 18,029 4,016 9,746 8,351 2,315 7,454 10,850 62,430 Revenues (M$) 9,305 2,054 5,442 4,955 1,268 3,189 5,531 31,094 DECILE Subsidies (million pesos) 1 179 37 70 61 18 48 104 517 2 491 100 192 175 57 159 285 1,459 3 766 152 310 295 98 295 445 2,360 4 998 200 439 408 132 407 580 3,163 5 1,188 249 587 513 154 474 690 3,855 6 1,368 291 718 603 174 522 795 4,472 7 1,555 333 862 644 190 539 903 5,025 8 1,294 331 877 683 184 663 752 4,784 9 1,011 350 645 735 192 806 587 4,325 10 310 27 30 35 9 787 180 1,378 9,160 2,069 4,730 4,150 1,208 4,699 5,319 31,339 DECILE Subsidy Shares by Decile 1 2.0% 1.8% 1.5% 1.5% 1.5% 1.0% 2.0% 1.6% 2 5.4% 4.8% 4.1% 4.2% 4.7% 3.4% 5.4% 4.7% 3 8.4% 7.3% 6.5% 7.1% 8.1% 6.3% 8.4% 7.5% 4 10.9% 9.6% 9.3% 9.8% 10.9% 8.7% 10.9% 10.1% 5 13.0% 12.0% 12.4% 12.4% 12.8% 10.1% 13.0% 12.3% 6 14.9% 14.1% 15.2% 14.5% 14.4% 11.1% 14.9% 14.3% 7 17.0% 16.1% 18.2% 15.5% 15.7% 11.5% 17.0% 16.0% 8 14.1% 16.0% 18.6% 16.5% 15.2% 14.1% 14.1% 15.3% 9 11.0% 16.9% 13.6% 17.7% 15.9% 17.2% 11.0% 13.8% 10 3.4% 1.3% 0.6% 0.8% 0.8% 16.7% 3.4% 4.4% Includes households wthout electricity and households in DAC tariffs - 145 - 68. For all tariffs, deciles 5 to 9 concentrate high shares of subsidies. However, tariffs 1E and 1F show particularly regressive allocations of energy subsidies. In contrast to other tariffs, decile 10 gets one sixth of subsidies. These two tariffs encourage intensive consumption of electricity, with deciles 8 to 10 capturing almost half the total subsidies for those tariffs. Intended versus actual impact of the February 2002 residential tariff review 69. As of June, 2003, the low, intermediate and high billing schedules comprise, respectively, 75.9 percent, 21.5 percent, and 2.6 percent of CFE customers. These percentages contrast with the 75 percent, 20 percent, and 5 percent, targeted populations. In other words, nominally, 97.4 percent of CFE customers are subsidized; furthermore, with a few exceptions, they never see the effect of the higher end block on their bills. 70. The divergence between intended and actual results is explained, first, by the introduction of tariff 1F in April 2002, and, second, by the reassignment of localities to tariffs with more extended subsidized blocks, which is still going on. 71. The leading example is the city of Monterrey, with almost 300 thousand consumers, which in March of 2002 was reclassified from tariff 1B to tariff 1C, resulting in many households paying lower bills than they did before the February tariff review. 72. As can be judged by figures in the preceding paragraphs, much of the impact of the DAC tariff already has been lost; by last June the targeted population for the DAC tariffs had been reduced by half already. Leakage rate 73. The Mexican government measures poverty by specifying poverty lines that reflect pesos per person per day of disposable income. For the year 2002, Poverty Line 1 includes 20.3 percent of the population, those with the lowest income. Line 2 covers an additional 6.2 percent, while Line 3 includes, altogether, 51.7 percent of the population. 74. Assuming a one to one correspondence between population and households, and taking deciles 1 and 2 of households as a proxy for Poverty Line 1 and deciles 1 to 5 as a proxy for Poverty Line 3, the following results concerning electricity subsidy allocation can be derived. 75. In 2002, Poverty Line 1 captured only 6.3 percent of subsidies, 1,976 million pesos out of a total of 31,339 million pesos. To Poverty Line 3 corresponded 36.1 percent of total subsidies, or 11,354 million pesos. That is, the leakage rate, the share of subsidies going to non-poor population, was 63.9 percent; a rather regressive distribution of subsidies. - 146 - Outlook for electricity subsidies 76. Analysis of subsidies in section 2 of this report shows that, in the absence of any major increase or structural change, subsidies increased in real terms at an average annual rate of 6.8 percent between 1993 and 2001. 77. This rate results from a combination of: increase in the number of customers, increase in electricity consumption per customer, and variations in cost of supply or in tariffs. 78. For the coming 5 years, the number of customers is likely to keep on growing at historical rates between 3.5 percent and 4 percent per year. Consumption per customer is closely tied to disposable income. Whenever income increases there is a surge in demand for household appliances, which translates into additional underlying electricity demand. Between 1993 and 2001 electricity consumption per household grew at rates of 0.8 percent in temperate zones and more than double that rate, at 1.7 percent, in the warm summer zones, implying the growing pursuit of comfort through some form of air conditioning in warm summer areas. 79. Unless there is a strong variation in disposable income these patterns are likely to continue for the rest of the decade. Thus, assuming there are no significant changes on costs or tariffs, subsidies are likely to continue increasing at approximately the same rate. That is, by the year 2007 subsidies might grow by nearly 40 percent in real terms, reaching between 42,000 and 45,000 million pesos (measured in 2002 currency). 80. However, subsidies might increase faster if CFE continues reassigning localities to tariffs with extended subsidized blocks, or if the government creates new tariffs with still higher subsidies. Existing summer subsidies are not only highly regressive; they have the potential to grow exponentially. Whereas customers in tariff 1 received an average subsidy of 1,027 pesos in 2002, those in tariffs 1E and 1F got 3,471 pesos. Today, there are at least 9 million households who would like to get about 2,000 pesos of additional subsidy, in order to enjoy some form of air conditioning. This represents an additional 18,000 million pesos of unsatisfied demand for electricity subsidies. 81. The long-term potential is for all the warm areas of Mexico to adopt some form of air conditioning. This better be the result of an efficient allocation of resources and higher incomes, than a continuing government give-out. SUBSIDIES IN THE AGRICULTURAL SECTOR Agricultural Customers And Sales 82. Electricity for use in the agricultural sector has been subsidized at least since the early seventies. During the seventies and eighties, after many years of high inflation, prices declined in real terms so that by 1989 the price to cost ratio was only 18 percent. At that time, in order to reduce subsidies, monthly increases of 3 percent were instituted and the price to cost ratio gradually recovered until it reached 60 percent in 1993. - 147 - 83. Following well organized protests by farmers, including road blockages, tariff increases were suspended and the price was let to fall in real terms until 1996, when the price to cost ratio was back down to 28 percent; near today’s level. 84. The following table shows the main statistics for agricultural tariffs between 1993 and 2002. MÉXICO, AGRICULTURAL SECTOR BASIC STATISTICS FOR ELECTRICITY AVERAGE MONTHLY YEAR CONSUMERS SALES REVENUES CONSUMPTION INVOICE (GWh) (thousand pesos) (KWh / month) (pesos / month) 1993 79,393 5,921 742,091 6,215 779 1994 80,788 6,552 835,941 6,758 862 1995 81,539 6,688 900,869 6,835 921 1996 82,332 7,544 1,264,057 7,636 1,279 1997 84,826 7,648 1,501,553 7,513 1,475 1998 87,665 7,745 1,749,829 7,362 1,663 1999 91,334 7,997 2,057,630 7,297 1,877 2000 93,558 7,898 2,266,159 7,035 2,019 2001 96,520 7,463 2,338,001 6,489 2,019 2002 98,653 7,644 2,566,654 6,480 2,168 avg annual rates 2.4% 2.9% 0.5% 85. The number of consumers shows a steady growth at an annual rate of 2.4 percent. Sales tend to grow with the number of customers, but they also vary yearly according to rainfall, having reached a maximum of 8 TWh in 1999. 86. Customer size, as measured by average consumption, grew until 1995, but since that time it has declined every year and is now similar to what it was at the beginning of the period. 87. This statistics are important because they may signal the initial impacts of limits in water availability. In fact, during the last few years, the Comisión Nacional del Agua (CNA) has been putting in place, in several places where water availability is becoming a constraint, local committees in charge of managing aquifers and river basins. Water rights are being revised and, where necessary, agreements are been reached, establishing programs for the gradual reductions in the volume of water rights, deemed necessary to bring sustainability to water resources. In parallel, some experiments with markets for water rights are starting. - 148 - Aggregate subsidies 88. Next we look at prices and subsidies in aggregate. MEXICO AGRICULTURAL SECTOR ELECTRICITY PRICES AND SUBSIDIES YEAR PRICE SUBSIDY INPP PRICE SUBSIDY (pesos / kWh) (million pesos) (2002 pesos / kWh) (million 2002 pesos) 1993 0.125 570 94.014 0.459 2,086 1994 0.128 1,034 100.000 0.439 3,558 1995 0.135 1,779 141.493 0.328 4,326 1996 0.168 3,224 190.093 0.303 5,836 1997 0.196 3,925 220.417 0.306 6,127 1998 0.226 4,204 250.836 0.310 5,767 1999 0.257 5,024 290.176 0.305 5,957 2000 0.287 5,869 319.151 0.309 6,327 2001 0.313 5,660 329.661 0.327 5,907 2002 0.336 5,886 344.075 0.336 5,886 avg annual rates -3.4% 12.2% subsidies for 1993-6 are own estimates 89. In this case the Producer Price Index (INPP) was used to bring figures to 2002 pesos. Prices declined every year until 1996; in particular, they fell by over 25 percent in 1995. This has been the main problem with tariffs for the last 30 years, every time inflation explodes, as it did in 1995 (with an INPP growth of 41.5 percent), heavily subsidized tariffs, in particular residential and agricultural, are not adequately corrected for inflation, losing many years of painful adjustments in the process. 90. Between 1995 and 2000 electricity prices for agriculture remained approximately constant; although they increased for the last two years, they are still well below their 1994 levels. 91. Subsidy figures are available from 1997 onwards, previous years numbers are gross estimates. Last year subsidies were somewhat below 6 thousand million pesos, a level that has been roughly maintained since 1996. 92. In the case of agricultural customers, too low tariffs mean that aquifers are degraded by excessive pumping, water is wasted, and old technologies and produce, are not replaced with more efficient ones, in terms of both energy and water use. 93. The electric tariffs for agricultural irrigation are the most subsidized of all the electricity tariffs. These low tariffs create significant distortions in the price signals, and - 149 - result not only in inefficient levels of electricity consumption but also in waste of water due to the adoption of inefficient irrigation techniques. It is not the most needy who are benefited the most by these subsidies. These subsidies only benefit those farmers who have irrigation, which frequently are those who engage in modern and profitable agriculture. Agricultural Tariff Structure And Policy 94. Until 2003 there were only 2 tariffs for agriculture; tariff 9 for low-tension customers and tariff 9M for middle tensions. They both consist of 4 blocks of increasing energy charges. 95. For January 2003 specific charges for agricultural tariffs can be seen in the following table. It can be appreciated that the tariffs structures are quite flat, and that all energy is heavily subsidized. Also, there is almost no difference between low and middle voltage prices, indicating that small consumers, in the low voltage, and accounting for approximately 10 percent of sales, are much more heavily subsidized. 96. Also, with the reintroduction of a monthly adjustment factor of 2 percent, tariffs for agriculture were on the right track again, in what could have been annual real increases greater than 20 percent. However, it should be mentioned that during 2003, tariff policy for the agricultural sector was modified. There is now a new tariff 9CU (single energy charge) that stipulates energy prices fixed at 0.30 pesos per kWh for 2003, 0.32 for 2004, 0.34 for 2005, and 0.36 for 2006. Also, there is new night-use tariff 9N, with charges equal to 50 percent those of tariff 9CU. 97. These tariffs are available for consumers who have been certified by CNA as using water for agricultural purposes and they have been allocated water rights stipulating specific volumes. Water volumes in turn are transformed into equivalent electricity use and the application of the new tariffs is limited to such electricity use. Excess electricity consumption is charged according to the old tariffs 9 and 9M. MÉXICO: ELECTRICITY TARIFFS FOR AGRICULTURE JANUARY 2003 (Prices in pesos per kWh) Voltage Energy block (kWh/month) 0-5,000 5,001-15,000 15,001-35,000 >35,000 Low tension 0.300 0.332 0.364 0.398 Middle tension 0.300 0.336 0.367 0.401 All charges have a monthly adjustment of 2% per month - 150 - 98. For all its disadvantages, the new tariffs for agriculture have some merit. They are a mechanism to purge ineligible consumers and to charge higher prices for excess energy use. Those users who cannot demonstrate the need to depend on water use rights for agricultural uses will be pay all their energy according to the old tariffs. If real annual increases of over 20 percent are maintained for the old tariffs, eventually, such customers will find it convenient to be transferred to electricity tariffs of general application. 99. At this time it appears that over 80 percent of agricultural users of electricity have been able to register for the new tariffs. Subsidies By State 100. Separate subsidy figures by State are not available. To estimate them, the total cost of supply for the agricultural costumer class was used. The national average cost per kWh was computed and, since the agricultural electricity tariff structure is quite flat, the difference between this cost and the known average price of electricity, along with sales volume by State was used to compute each State’s subsidies. 101. In the next two tables we look at electricity subsidies for agriculture disaggregated by State. The same table is presented twice only the order of States varies. In the first table order follows the magnitude of total subsidies while in the second subsidies per customer determine the order of the States. 102. From the first table it is significant that two States, Chihuahua and Guanajuato, with nearly one thousand million pesos in subsidies each, together take over one third of total subsidies. This share grows to almost half, 46 percent, when Sonora is added. Furthermore, consideration of three more states: Coahuila, Zacatecas and Durango, results in six States capturing two thirds of total subsidies. Guanajuato is at the hearth of the Bajío agricultural region in central Mexico; all other states in this most favoured group are located in Northern Mexico, and are characterized by predominantly modern agriculture. 103. Contrastingly, States in poor regions of Southern Mexico have very limited irrigation and therefore benefit little from electric power subsidies. Chiapas tops the list with only 44 million pesos in subsidies during 2002; while Oaxaca gets 18 and Guerrero 3.6 million pesos only. 104. Given that tariff structure is quite flat, subsidies per customer are very nearly proportional to energy consumed. In the second table it can be seen that while agricultural electricity consumers in Sonora, the State with the foremost modern agriculture in Mexico, got almost 200 thousand pesos of subsidy in 2002, those in Guerrero received slightly over 5 thousand pesos and those in Oaxaca only half as much93. 93. A note of caution in interpreting these numbers is warranted though. It is known that some irrigation districts, including some in the State of Sonora, buy water and pay for energy in bulk and then distribute water and allocate electricity bills to individual members of the district. - 151 - MEXICO STATISTICS FOR ELECTRICITY USE IN AGRICULTURE SALES AND SUBSIDIES BY STATE, ORDERED BY TOTAL SUBSIDIES, 2002 STATE CUSTOMERS SALES SUBSIDIES SUBSIDIES (MWh) (million pesos) (pesos / customer) Chihuahua 11,220 1,297,288 1,004 89,443 Guanajuato 13,511 1,263,532 980 72,533 Sonora 3,853 1,000,658 742 192,521 Coahuila 3,933 587,173 443 112,729 Zacatecas 6,744 460,021 357 52,948 Durango 2,842 418,767 318 111,784 Jalisco 7,396 368,493 285 38,550 Puebla 3,495 270,926 211 60,287 Aguascalientes 2,102 266,901 208 98,733 San Luis Potosi 4,150 246,163 185 44,590 Queretaro 1,359 234,331 178 131,318 Michoacan 5,148 217,454 171 33,190 Baja California 2,116 192,139 150 70,831 Baja California Sur 1,026 149,161 117 114,419 Sinaloa 1,748 101,431 78 44,670 Nuevo León 4,026 99,282 78 19,362 Tamaulipas 1,782 71,433 55 30,970 Yucatan 6,559 51,588 46 6,939 México 1,465 53,310 42 28,707 Chiapas 1,078 38,179 42 38,985 Hidalgo 540 52,079 38 70,950 Veracruz 918 50,088 38 41,725 Morelos 674 47,496 38 55,846 Colima 944 45,211 35 36,694 Oaxaca 7,136 21,853 18 2,491 Tlaxcala 516 20,028 16 30,679 Nayarit 333 5,649 4.4 13,113 Guerrero 690 4,684 3.6 5,167 Campeche 1,057 3,465 2.8 2,606 Quintana Roo 245 2,246 1.8 7,265 Tabasco 37 2,113 1.7 46,652 Distrito Federal 10 647 0.5 49,441 TOTAL 98,653 7,643,789 5,886 59,669 - 152 - MEXICO STATISTICS FOR ELECTRICITY IN AGRICULTURE SALES AND SUBSIDIES BY STATE, ORDERED BY SUBSIDY PER CUSTOMER, 2002 STATE CUSTOMERS SALES SUBSIDIES SUBSIDIES (MWh) (million pesos) (pesos / customer) Sonora 3,853 1,000,658 742 192,521 Queretaro 1,359 234,331 178 131,318 Baja California Sur 1,026 149,161 117 114,419 Coahuila 3,933 587,173 443 112,729 Durango 2,842 418,767 318 111,784 Aguascalientes 2,102 266,901 208 98,733 Chihuahua 11,220 1,297,288 1,004 89,443 Guanajuato 13,511 1,263,532 980 72,533 Hidalgo 540 52,079 38 70,950 Baja California 2,116 192,139 150 70,831 Puebla 3,495 270,926 211 60,287 Morelos 674 47,496 38 55,846 Zacatecas 6,744 460,021 357 52,948 Distrito Federal 10 647 0.5 49,441 Tabasco 37 2,113 1.7 46,652 Sinaloa 1,748 101,431 78 44,670 San Luis Potosi 4,150 246,163 185 44,590 Veracruz 918 50,088 38 41,725 Chiapas 1,078 38,179 42 38,985 Jalisco 7,396 368,493 285 38,550 Colima 944 45,211 35 36,694 Michoacan 5,148 217,454 171 33,190 Tamaulipas 1,782 71,433 55 30,970 Tlaxcala 516 20,028 16 30,679 México 1,465 53,310 42 28,707 Nuevo León 4,026 99,282 78 19,362 Nayarit 333 5,649 4.4 13,113 Quintana Roo 245 2,246 1.8 7,265 Yucatan 6,559 51,588 46 6,939 Guerrero 690 4,684 3.6 5,167 Campeche 1,057 3,465 2.8 2,606 Oaxaca 7,136 21,853 18 2,491 TOTAL 98,653 7,643,789 5,886 59,669 - 153 - Impact on federal finances 105. The main serious drawbacks of tariff subsidies in Mexico are the following: (a) The large amount of public funds directed to the electricity sector, in order to compensate for subsidies given out in low tariffs and inefficiencies in LFC, — around 62,500 million pesos in 2002 - come at the expense of other governmental programs of top priority for the country; specifically, the development of human resources through improved and expanded education, health and welfare programs. Also, these large transfers are also a big impediment to the modernization of the electric power industry. (b) Subsidies have not been carefully designed to be consistent with government policy. In practice, the subsidies are not exclusively directed to those consumers who in principle the tariff policy would like to subsidize, resulting in a corresponding negative impact in the equity of the system; furthermore, most social policy objectives can be met in more cost-effective ways than through energy subsidies. (c) Subsidized prices impose serious efficiency costs, by distorting the relative price signals to which customers react, encouraging an excessive consumption of electricity, and leading to a waste of resources. The existence of general subsidies in electric tariffs entails an important restraint on the inflow of private capital in the electric sector. This is probably the main obstacle to electricity liberalization in Mexico, both, because the population at large perceive that an eventual privatisation would have to be accompanied by a drastic increase in electricity tariffs, and because private capital would not be attracted to an industry that does not generate enough revenues. 106. What would be the effect of substantially reducing the amount of subsidies? (a) Today, the power industry suffers from an insufficient level of investment that has, over many years, resulted in a gradual degradation of facilities and operational efficiency. In generation, plant availability factors have declined and the state owned power plants are growing old. Energy losses in transmission and distribution grids are too high, and there are capability restrictions to economically transfer power amongst several regions. Technological obsolescence has been growing in all phases of the industry. All these problems refer to insufficient “investment in quality”. (b) There is here the potential for important savings. Investment to improve efficiency can, in many cases, “produce” additional energy94 at a lower cost than an alternative capacity expansion project. (c) At the same time CFE’s finances, although still manageable, exhibit an unsustainable trend, with debt increasing much more rapidly than revenues. In 94. Many projects to reduce distribution losses qualify in this category. - 154 - this context, off-balance liabilities related to the Pidiregas program are particularly worrying. Pidiregas constitute an indirect form of indebtedness that helped reduce net cash flows for the power industry for a number of years, but now that many plants financed through this skim are starting operations, large debts have to be repaid in relatively short terms. (d) It is evident that the industry badly needs additional capital. Arguably, cash generated by reductions in subsidies could be efficiently applied within the power industry to productive investments directed at: improving efficiency, reducing costs, and incorporating up to date technologies. A reduction in subsidies would also impact demand growth, with an indirect effect for reduced capital needs for capacity expansion; that is, for reduced “investment in quantity”. (e) In summary, by reducing subsidies, a virtuous circle could be established that leads to a sustainable financial path for the power industry. Detailed studies that identify the most profitable investments and allow developing a strategy for the technical and financial rehabilitation of the power industry should be undertaken. (f) A healthy power industry should eventually start generating net capital gains, in the form of taxes and dividends, which could be redirected to support welfare programs. (g) All of these problems call for the adoption of a series of political tariff measures that will help correct the most obvious problems introduced by the present subsidies, in a context of broader tariff reforms. These measures are discussed in the following section. - 155 - OPTIONS FOR SUBSIDY POLICY Fundamental Principles For Tariff Policy 107. In general, the tariff reform to reduce the distortions produced by the current system should be based on the following principles: • Efficiency: Tariffs should promote efficient consumption decisions. To achieve that, tariffs should be designed such that the marginal price a customer faces when contemplating a change in consumption reflects the marginal cost of serving that incremental load. All subsidy mechanisms create efficiency losses, however, those mechanisms that enable consumers to pay the actual incremental cost of supply (at least for marginal consumption decisions) are less distorting. • Equity: Tariff design should ensure that each customer class pays at least its marginal cost of service and also pays a fair share of the revenue gap, taking the various customers’ demand elasticity into account. • Commercial viability: Tariffs should be high enough to provide a reasonable return on investment and to generate sufficient cash flow to permit adequate maintenance and expansion of the system. • Transparency: Tariffs should help consumers understand what they are paying for, and when they are being subsidized. Targeted customers that, according to social policy objectives are to remain subsidized through electricity tariffs should see on their bills the full cost associated with their consumption and the subsidy they receive95. • Predictability: Tariffs should, to the extent possible, be predictable so that customers can make prudent long-term decisions about their energy consumption. • Administrative Feasibility: Tariffs should be consistent with the utility’s metering and billing capability, and the cost of more complex or flexible metering and billing systems should be compatible with revenues expected from each customer group. • Gradualism: In cases where a sudden reform of tariffs would result in unacceptable bill impacts, a transition plan for moving towards those marginal- cost based targets should be developed. Specific Recommendations For Tariff Structure 108. There are a number of reforms in both the structure and design of the current electricity tariffs that would help to solve the problems associated with subsidies and improve efficiency. • Any required subsidies should, preferably, be assigned to the fixed charges while the variable charges (e.g. per kWh) should be set as close to marginal cost as 95. This principle has been implemented already. - 156 - possible. Consequently, the price signal for additional consumption would be efficient. • Use seasonal tariffs for all customers in regions where this is relevant and time- of-day tariffs for all those for whom time-of-day metering is cost-effective, and install time-differentiated metering as soon as feasible. • Consolidate customer categories if a single tariff can accurately reflect their costs of service and demand elasticities. Tariffs should be geographically differentiated if the costs of the additional complexity are low relative to the differential in costs, and the welfare loss from failure to reflect these differentials. • Determine, based on the magnitude of changes required, how rapidly the transition can be accomplished from current to optimal tariffs and the necessary gradual steps. Residential tariffs 109. Many countries have residential tariffs with a more complex structure than Mexico’s current structure—with its blocked energy charges and no demand or fixed components. Most countries incorporate in their rate design some sort of fixed charge as well as a per-kWh charge. The fixed charge may be based upon contracted demand, or it may be charged to recover customer and meter costs. Some countries waive the fixed charge for low-income customers. 110. The metered per-kW option is only available to customers who have demand meters. If a customer does not have a demand meter, then there is typically a fixed charge that depends on contracted demand. Often there is a minimum consumption level required to qualify for metered demand, so customers with separate metered demand charges are typically larger residential customers. 111. There are several types of per-kWh charges. Some countries use a single rate that is applicable for the entire year. Others have seasonal (e.g., winter and summer) and diurnal (e.g., peak and off-peak) pricing periods. Chile offers a hybrid: a single rate for the entire year unless the consumer uses more than an allotted amount of energy in the winter. Any excess kWh in the winter is provided at a higher tariff. 112. The measures that need to be undertaken are the following: 1) On a monthly basis, adjust slightly above inflation the prices for the blocks priced below cost, to ensure that the subsidies at least consistently decrease over time. 2) Restructure the DAC tariffs so that they reflect marginal cost plus their fair share of the revenue gap. This calls for increasing the fixed charges and reducing the energy charges in order to provide efficiency. Total revenues from DAC tariffs seem adequate, if not excessive, and they involve only 2.6 percent of the residential consumer class with the highest consumption levels; hence, this restructuring should be easy to apply. - 157 - 3) Revise the residential subsidy to reduce its size and limit the scope of the benefited segment of the population. This involves, first, the gradual elimination of subsidies to customers in the intermediate consumption category. For them, the steps include: (1) gradually removing the intermediate consumption block at the same time that the price for the third block is reduced to eliminate a cross-subsidy, (2) gradually increasing the price (in real terms) for the basic consumption block, and (3) eventually modifying the residential tariff structures, through the introduction of fixed charges, to better reflect the structure and levels of marginal costs. 4) Through small real increases, reduce subsidies to the low consumption category, which comprises over 70 percent of consumers; while at the same time improving the structure of the tariff. Eventually, the subsidy to needy families should be given as discounts on the fixed charges of the tariff, while the variable charges (pesos/kWh) would be maintained as close as possible to the marginal costs. 5) The most important step in residential tariff policy ought to be the elimination of air conditioning subsidies. An immediate step should be to stop the reclassification of the localities so that they do not move from one summer tariff to another with higher subsidies. In spite of the level of the summer subsidies, a major part of the benefited population is unsatisfied because each locality has a unique climate; it is a question of dealing with a local problem, thus the solutions should also be local. This should be accomplished by transferring the responsibility for managing those subsidies to state governments, who have been most vocal in maintaining and increasing them. In this form air-conditioning subsidies would be eliminated from official tariffs. In their place, individual agreements between the federal and state governments would be signed, in which the state government commits to a specific program and time-table for the gradual elimination of air-conditioning subsidies, in exchange for federal funds to partially pay for those subsidies while they last. Remaining summer subsidies, if any, should be strictly a local decision paid for with local funds. 6) New tariff options should be offered, all of them without subsidies. These include a tariff with demand charge for relatively high consumption customers and a time-of-day tariff for the group with the highest consumptions. Also, appropriate regional cost differences should in time be incorporated in residential tariffs. Agricultural electricity tariffs 113. Government has already committed a program of sustained subsidies to the agricultural sector through 2006. 114. Nevertheless, it should be said that these subsidies are quite inefficient. Rather than continuing subsidization of agriculture through inefficient electricity pricing, it - 158 - would be preferable to substitute other, less-distorting forms of support to farmers. As these other support mechanisms are implemented, the electricity subsidies should be reduced in parallel. 115. Measures that should be undertaken include: • Regionalization of the tariffs: Given that the agricultural cycles change significantly depending on the region in question and that there are important differences in the load curves for different parts of the country, regional tariffs for agriculture are appropriate to reflect cost differences. • Monthly adjustments: In the meantime, it is important to make some progress toward reducing the excessive subsidies. A strategy of monthly adjustments above inflation should be reinstated as soon as possible, complemented by major increases once a year. These should be applied in the fourth trimester of the year, when the consumption of electrical energy for agricultural purposes noticeably diminishes. Other Subsidies And Tariffs Studies 116. The last general study of marginal costs was undertaken in 1996; since that time, load curves have changed, as has the structure and fuel mix of generation; while the transmission grid has also had important changes. A new study of marginal costs is needed, as is a new financial strategy that gives transparency to CFE’s financial statements and makes its fiscal regime compatible with those of independent power producers and other private generators. 117. After those studies have been completed, efficiency gains should be agreed on with CFE and, most particularly, LFC, and incorporated into tariff policy. Subsidies should then be reduced through price increases as well as cost reductions. 118. Distortions and subsidies for other tariffs are less accentuated and generally well understood. Measures to eliminate those subsidies are, for the most part, straightforward. - 159 - ANNEX 1– APPENDIX Methodology And Data Base For The Distributional Analysis Of Residential Tariffs Description Of The Problem And Available Information 119. Information provided by the 2002 ENIGH is considered insufficient to capture the complexities of the tariff schedules and the resulting allocation of subsidies. Thus, information from the power industry was used instead. 120. Whenever information was available from Secretaría de Hacienda, it was given preference over other sources. This is due to the fact that Hacienda manages official information, which is usually more aggregated and, thereby, robust. 121. Most data come from CFE, but one has to be careful because there are many sources within CFE, using information for different purposes, sometimes using the same terms for no identical concepts. For instance, financial statements contain all the income and expenditure elements, whereas tariff studies exclude those elements having to do with operations not related to electricity operations. 122. All the applicable tariff schedules are published and available. 123. From May of 2002 it became mandatory to exhibit, in each electricity bill, the subsidy granted to each customer. Therefore, from CFE records, 2002 subsidies allocated to each of the seven basic tariff categories (1, 1A, 1B, 1C, 1D, 1E, 1F) are available. However, DAC data, including revenues and subsidies, is not available for each basic tariff, but only in the aggregate 124. For each of the 7 basic categories of tariffs, CFE provided monthly histograms of customers and their consumption intervals. For each of the 7 basic tariffs there are 4 monthly histograms, in combinations: (winter / summer) and (low and moderate consumption / high consumption), the latter representing customers incorporated into DAC tariffs. That is, there are 28 histograms with a variable number of consumption intervals (from 18 to 24), indicating the percentage of households in each interval. 125. The 6 regional DAC tariff schedules have to be transformed into equivalent tariffs to be applied to the 7 basic tariff categories for which histograms exist. However, there is no separate information available for Baja California and Baja California Sur; but since this last state comprises less than 1 percent of total sales, in this study it was ignored. Hence, we are left with five regional DAC tariffs. 126. LFC applies only the basic tariff 1 throughout its territory. Even though updated histograms for LFC customers are not available, from global statistics and form old data we know that the existing CFE histogram for tariff 1 can be applied, with small adjustments, to LFC customers. - 160 - 127. In summary, our database consists of: • 42 residential tariff schedules, • SHCP provided aggregate annual data including, for each of the 7 basic tariffs, revenues, subsidies and price/cost ratios. • 28 monthly histograms of electricity consumers, 128. From CFE come histograms and aggregate data for number of customers and sales. Aggregate figures are also available for LFC. 129. All the numbers in the database are not entirely consistent and there may be several reasons for this: non-technical losses, billing cycles, and errors. Nevertheless, differences are small, thereby not important. Also, there are cross-subsidies within each tariff and non-linearities in the tariff structures, which distort partial results somewhat, particularly while making aggregations. 130. Also, some of these numbers, specifically the total number of customers, or households, may differ from other official statistics, due to the fact that CFE and LFC measure customers by meters. There are a number of dwellings containing more than one meter, and sometimes there is a single meter for more than one dwelling. Differences, however, ought to be negligible for our purposes. Methodology 131. There are 5 DAC tariff schedules, each of which may be applied to sets of consumers coming from 7 different basic tariffs. Annual subsidies granted to customers by basic tariff, by region, are known. These figures, in percentage, were applied to compute weighted average national tariffs for the DAC tariff applied to customers in each of the 7 basic tariffs. After this was accomplished, the regional dimension was eliminated and we were left with 42 applicable billing possibilities: 7 basic tariffs, 2 seasons, 3 levels of consumption. 132. The 28 histograms available were then grouped into 7 categories, corresponding to each of the 7 basic tariffs. However, since tariff 1F was established only in April 2002 and, throughout the year, many localities were reassigned from tariff 1E to tariff 1F, the statistical exercises for these two basic tariffs were not satisfactory. It was then decided to merge them into a single group comprising tariffs 1E and 1F. Statistical results for this new set were quite good. 133. For each of the six basic tariff groups (1, 1A, 1B, 1C, 1D, 1E+1F), eight steps were performed, as follows: 1) Starting with the monthly histogram for the summer season, with low and moderate consumption customers, the DAC tariff, representing the cost function, was applied (in Table 1), computing the supply cost for each consumption level; this is followed by a second cost, labeled “estimated cost” function to which we will come back latter; then, the corresponding tariff - 161 - schedules (in Table 1) for low and moderate consumptions were applied, computing payment for each consumption level; the difference between cost and payment yields the subsidy for each consumption level. Two subsidies were computed, one based on the official cost, named “subsidy based on DAC” and the other for the estimated cost, called “estimated subsidy”. Finally, monthly totals in million pesos were computed for each of the 5 variables: official cost, estimated cost, payment, official subsidy, and estimated subsidy. (Table 3) 2) Again for the summer season, the same exercise was repeated for the monthly histogram containing the high consumption customers, that is, the DAC customers coming from tariff 1. (Table 5) 3) The 2 previous histograms were merged into a single histogram, which includes all customers in a basic tariff for the summer season. To achieve this step it was necessary to compute the weighted sum of customers by interval and also the weighted payment for each interval; addition of other variables is straightforward. (Table 7) 4) Next, the resulting histogram was transformed into an equivalent histogram, normalized by deciles. This required a regrouping of intervals; initial results of this exercise show a small deviation in the average electricity consumption for the histogram as a whole. Interval average consumption levels were subsequently readjusted in order to match actual average consumption for the population. This adjustment corresponds to a rescaling of the electricity consumption variable. Here we are able to observe a first estimate of seasonal subsidies by decile. (Table 9). However, households without electricity still have to be incorporated. 5) It is estimated that approximately 5 percent of the population lacks electricity in Mexico; three further assumptions were made at this point: people without electricity are uniformly distributed across temperature regions, thereby basic tariffs; the average household size is the same as for those with electricity; and all households without electricity correspond to decile 1. Next, technical coefficients were developed that incorporate households without electricity in the first decile, and transfer an appropriate proportion of households into the next decile. Then, this reassignment of households into deciles is repeated for each decile. (Table 11). Table 11 was used to incorporate households without electricity in all 12 series of histograms, that is, for the combinations of 6 basic tariffs and 2 seasons. 6) A new normalized histogram for households in tariff 1, for the summer season, is thereby obtained which includes, in the first decile, households without electricity. All variables (cost, payment, subsidies) can be recomputed at this point. Monthly totals do not vary but their distribution by decile does. Monthly subsidies are multiplied by 6 to obtain seasonal subsidies. (Table 12) 7) The same exercise, including steps 1 to 6, was repeated for the winter season. (Tables 2, 4, 6, 8, 10, and 13) 8) The resulting histograms for summer and winter were added, obtaining annual results for tariff 1. (Table 14) - 162 - 134. At this point it was confirmed that annual official subsidies, computed by using the DAC tariff as a proxy for cost, were largely overestimated. Therefore, an estimated cost function was computed, which reproduces actual annual subsidies. In order to minimize changes from the corresponding DAC tariffs, the fixed cost was left intact, with all the cost reduction concentrated in the energy charge of the estimated cost function. This implies a national average estimated cost function consisting of a monthly fixed cost equal to 33.52 pesos per customer, which yields an average energy cost equal to 1.323 pesos per kWh. 135. In reality, this author estimates that for 2002 the monthly fixed cost ought to be at least 80 pesos per customer, which implies an average energy cost of 1.008 pesos per kWh. One should beware that such a structural change in the cost function yields a different distribution of subsidies by decile. 136. Statistical tests were performed along the way. After step 1 for both seasons has been completed, the statistical test consists of computing average consumption, as well as annual revenues and subsidies, and comparing these partial results with actual figures available from CFE’s annual statistics. For tariff 1 the errors were: 0.5 percent for average consumption, 0.2 percent for revenues, and 20.5 percent for subsidies computed with the DAC tariff as a proxy for cost. This significant deviation is consistent with our knowledge of purposeful overestimation of DAC energy charges. 137. The second statistical test was performed at the end of step 4, with the histogram normalized by deciles. Monthly results for cost, revenues, and subsidies, were compared with those obtained from the original histograms. Again, the statistical errors were very small. 138. All previous steps were then repeated for each basic tariff. For each of the 6 basic categories of tariffs and 2 seasons, histograms were similarly constructed and transformed into equivalent deciles. 139. At the end, a proxy histogram for LFC customers was constructed, based on the CFE tariff 1 histogram. This can be done because customers served by LFC are all in tariff 1, which is not subject to significant seasonal whether variations, and their average consumption is only 4.5 percent less than for CFE customers. A linear rescaling of the electricity consumption variable was performed in order to reproduce known LFC annual statistics for consumers, sales and revenues. 140. It should be noticed that final results for most variables do not match official statistics due to the incorporation of households without electricity. - 163 - Tables for Tariff 1 Table 1 TARIFF SCHEDULES COST FUNCTION DAC for T1 (SUMMER, September, 2002) For average annual consumption over 250 kWh per month Fix Charge ($/month) 35.43 35.43 Energy Charge ($/kWh) For the first 500 kWh 1.503 1.423 For each additional kWh 1.796 1.423 T1 (SUMMER, September, 2002) For consumption up to 140 kWh per month Minimum Monthly Charge 12.33 Basic Consumption (first 75 kWh) 0.493 Intermediate Consumption (each additional kWh) 0.588 For consumption over 140 kWh per month Basic Consumption (first 75 kWh) 0.493 Intermediate Consumption (next 50 kWh) 0.766 Excess Consumption (each additional kWh) 1.713 Table 2 TARIFF SCHEDULES COST DAC for T1 (WINTER, March, 2002) For average annual consumption over 250 kWh per month Fix Charge ($/month) 31.61 31.61 Energy Charge ($/kWh) For the first 500 kWh 1.331 1.260 For each additional kWh 1.599 1.260 T1 (WINTER, March, 2002) For consumption up to 140 kWh per month Minimum Monthly Charge 11.88 Basic Consumption (first 75 kWh) 0.475 Intermediate Consumption (each additional kWh) 0.564 For consumption over 140 kWh per month Basic Consumption (first 75 kWh) 0.475 Intermediate Consumption (next 50 kWh) 0.669 Excess Consumption (each additional kWh) 1.645 - 164 - Table 3 CFE TARIFF 1 SUMMER Monthly average consumption (kWh) 96.4 Number of Households 8,017,744 HISTOGRAM WITH SOURCE DATA SUBSIDY CONSUMPTION CUMULATE DAC ESTIMATED BASED ON ESTIMATED INTERVAL midpoint HOUSEHOLDS HOUSEHOLDS TARIFF COST PAYMENT DAC SUBSIDY kWh/month % % $/hh-month $/hh-month $/hh-month $/hh-month $/hh-month 0 5 2.5 7.36 7.36 39.19 38.99 12.33 26.86 26.66 6 25 15.5 8.11 15.47 58.72 57.49 12.33 46.40 45.16 26 50 38 11.24 26.71 92.53 89.50 18.73 73.80 70.77 51 75 63 13.38 40.09 130.09 125.08 31.06 99.04 94.02 76 100 88 14.7 54.79 167.66 160.65 44.62 123.04 116.04 101 125 113 13.6 68.39 205.22 196.23 59.32 145.91 136.91 126 140 133 7.08 75.47 235.28 224.69 71.08 164.20 153.61 141 150 145.5 4.12 79.59 254.06 242.48 110.39 143.67 132.09 151 175 163 7.92 87.51 280.36 267.38 140.37 139.99 127.01 176 200 188 5.31 92.82 317.92 302.95 183.19 134.73 119.76 201 225 213 3.41 96.23 355.49 338.53 226.02 129.47 112.51 226 250 238 1.96 98.19 393.05 374.10 268.84 124.21 105.26 301 400 350.5 0.39 99.87 562.09 534.19 461.56 100.54 72.64 401 450 425.5 0.05 99.92 674.79 640.92 590.03 84.76 50.88 451 500 475.5 0.03 99.95 749.92 712.07 675.68 74.24 36.39 501 600 550.5 0.02 99.97 877.45 818.79 804.16 73.29 14.63 601 750 675.5 0.01 99.98 1,101.99 996.67 1,018.28 83.71 -21.62 TOTAL (million pesos) 1,466 1,404 586 881 818 Table 4 CFE TARIFF 1 WINTER Monthly average consumption (kWh) 108.5 Number of customers 8,017,744 HISTOGRAM WITH SOURCE DATA SUBSIDY CONSUMPTION CUMULATE DAC ESTIMATED BASED ON ESTIMATED INTERVAL midpoint HOUSEHOLDS HOUSEHOLDS TARIFF COST PAYMENT DAC SUBSIDY kWh/month % % $/hh-month $/hh-month $/month $/hh-month $/hh-month 0 5 2.5 6.98 6.98 34.94 34.76 11.88 23.06 22.89 6 25 15.5 7.58 14.56 52.23 51.14 11.88 40.36 39.27 26 50 38 11.27 25.83 82.17 79.49 18.05 64.12 61.44 51 75 63 13.59 39.42 115.43 110.99 29.93 85.51 81.07 76 100 88 14.55 53.97 148.70 142.49 42.96 105.74 99.54 101 125 113 13.26 67.23 181.96 173.99 57.06 124.90 116.94 126 140 133 6.89 74.12 208.57 199.20 68.34 140.23 130.86 141 150 145.5 4.03 78.15 225.20 214.95 102.80 122.40 112.15 151 175 163 8 86.15 248.49 237.00 131.59 116.90 105.41 176 200 188 5.55 91.7 281.75 268.50 172.71 109.04 95.79 201 225 213 3.62 95.32 315.01 300.00 213.84 101.18 86.16 226 250 238 2.13 97.45 348.28 331.50 254.96 93.32 76.54 251 300 275.5 1.67 99.12 398.17 378.75 316.65 81.52 62.10 301 400 350.5 0.65 99.77 497.96 473.25 440.02 57.94 33.23 401 450 425.5 0.08 99.85 597.75 567.76 563.40 34.35 4.36 451 500 475.5 0.05 99.9 664.28 630.76 645.65 18.63 -14.89 501 600 550.5 0.04 99.94 777.63 725.26 769.02 8.60 -43.76 601 750 675.5 0.02 99.96 977.50 882.77 974.65 2.85 -91.88 TOTAL (million pesos) 1,332 1,274 587 745 688 - 165 - Table 5 CFE TARIFF 1 HOUSEHOLDS IN DAC SUMMER Monthly average consumption (kWh) 353.5 Number of Households 468,691 HISTOGRAM WITH SOURCE DATA SUBSIDY CONSUMPTION CUMULATE DAC ESTIMATED BASED ON ESTIMATED INTERVAL midpoint HOUSEHOLDS HOUSEHOLDS TARIFF COST PAYMENT DAC SUBSIDY kWh/month % % $/hh-month $/hh-month $/hh-month $/hh-month $/hh-month 0 5 2.5 0.27 0.27 39.19 38.99 39.19 0.00 -0.20 6 25 15.5 0.14 0.41 58.72 57.49 58.72 0.00 -1.23 26 50 38 0.19 0.6 92.53 89.50 92.53 0.00 -3.03 51 75 63 0.25 0.85 130.09 125.08 130.09 0.00 -5.02 76 100 88 0.31 1.16 167.66 160.65 167.66 0.00 -7.01 101 125 113 0.42 1.58 205.22 196.23 205.22 0.00 -9.00 126 140 133 0.32 1.9 235.28 224.69 235.28 0.00 -10.59 141 150 145.5 0.28 2.18 254.06 242.48 254.06 0.00 -11.58 151 175 163 0.93 3.11 280.36 267.38 280.36 0.00 -12.98 176 200 188 1.67 4.78 317.92 302.95 317.92 0.00 -14.97 201 225 213 3.50 8.28 355.49 338.53 355.49 0.00 -16.96 226 250 238 8.28 16.56 393.05 374.10 393.05 0.00 -18.95 301 400 350.5 32.77 77.55 562.09 534.19 562.09 0.00 -27.90 401 450 425.5 7.30 84.85 674.79 640.92 674.79 0.00 -33.87 451 500 475.5 4.55 89.4 749.92 712.07 749.92 0.00 -37.85 501 600 550.5 4.85 94.25 877.45 818.79 877.45 0.00 -58.66 601 750 675.5 2.94 97.19 1,101.99 996.67 1,101.99 0.00 -105.32 751 850 800.5 0.92 98.11 1,326.53 1,174.54 1,326.53 0.00 -151.99 851 900 875.5 0.31 98.42 1,461.25 1,281.27 1,461.25 0.00 -179.98 901 1,000 950.5 0.44 98.86 1,595.97 1,387.99 1,595.97 0.00 -207.98 1001 1,200 1100.5 0.49 99.35 1,865.42 1,601.44 1,865.42 0.00 -263.98 1201 2,000 1600.5 0.05 99.4 2,763.58 2,312.94 2,763.58 0.00 -450.64 2001 2500 2250.5 0.06 99.46 3,931.18 3,237.89 3,931.18 0.00 -693.29 2501 4500 3500.5 0.08 99.54 6,176.58 5,016.64 6,176.58 0.00 -1,159.94 TOTAL (million pesos) 266 250 266 0 -16 - 166 - Table 6 CFE TARIFF 1 HOUSEHOLDS IN DAC WINTER Monthly average consumption (kWh) 354.2 Number of Households 468,691 HISTOGRAM WITH SOURCE DATA SUBSIDY CONSUMPTION CUMULATE DAC ESTIMATED BASED ON ESTIMATED INTERVAL midpoint HOUSEHOLDS HOUSEHOLDS TARIFF COST PAYMENT DAC SUBSIDY kWh/month % % $/hh-month $/hh-month $/hh-month $/hh-month $/hh-month 0 5 2.5 0.42 0.42 34.94 34.76 34.94 0.00 -0.18 6 25 15.5 0.24 0.66 52.23 51.14 52.23 0.00 -1.09 26 50 38 0.33 0.99 82.17 79.49 82.17 0.00 -2.68 51 75 63 0.39 1.38 115.43 110.99 115.43 0.00 -4.44 76 100 88 0.49 1.87 148.70 142.49 148.70 0.00 -6.20 101 125 113 0.65 2.52 181.96 173.99 181.96 0.00 -7.97 126 140 133 0.50 3.02 208.57 199.20 208.57 0.00 -9.38 141 150 145.5 0.40 3.42 225.20 214.95 225.20 0.00 -10.26 151 175 163 1.35 4.77 248.49 237.00 248.49 0.00 -11.49 176 200 188 2.31 7.08 281.75 268.50 281.75 0.00 -13.25 201 225 213 4.50 11.58 315.01 300.00 315.01 0.00 -15.01 226 250 238 8.52 20.10 348.28 331.50 348.28 0.00 -16.78 251 300 275.5 25.24 45.34 398.17 378.75 398.17 0.00 -19.42 301 400 350.5 31.09 76.43 497.96 473.25 497.96 0.00 -24.71 401 450 425.5 7.26 83.69 597.75 567.76 597.75 0.00 -29.99 451 500 475.5 4.67 88.36 664.28 630.76 664.28 0.00 -33.52 501 600 550.5 5.13 93.49 777.63 725.26 777.63 0.00 -52.36 601 750 675.5 3.28 96.77 977.50 882.77 977.50 0.00 -94.73 751 850 800.5 1.04 97.81 1,177.38 1,040.27 1,177.38 0.00 -137.10 851 900 875.5 0.36 98.17 1,297.30 1,134.78 1,297.30 0.00 -162.52 901 1,000 950.5 0.51 98.68 1,417.23 1,229.28 1,417.23 0.00 -187.95 1001 1,200 1100.5 0.57 99.25 1,657.08 1,418.28 1,657.08 0.00 -238.79 1201 2,000 1600.5 0.59 99.84 2,456.58 2,048.30 2,456.58 0.00 -408.27 2001 2500 2250.5 0.08 99.92 3,495.93 2,867.33 3,495.93 0.00 -628.59 2501 4500 3500.5 0.08 100 5,494.68 4,442.38 5,494.68 0.00 -1,052.29 TOTAL (million pesos) 242 226 242 0 -15 - 167 - Table 7 CFE TARIFF 1 HOUSEHOLDS IN 1 AND IN DAC SUMMER Monthly average consumption (kWh) 110.6 Number of Households 8,486,435 HISTOGRAM WITH SOURCE DATA SUBSIDY CONSUMPTION CUMULATE DAC ESTIMATED BASED ON ESTIMATED INTERVAL midpoint HOUSEHOLDS HOUSEHOLDS TARIFF COST PAYMENT DAC SUBSIDY kWh/month % % $/hh-month $/hh-month $/hh-month $/hh-month $/hh-month 0 5 2.5 6.97 6.97 39.19 38.99 12.38 26.80 26.61 6 25 15.5 7.67 14.64 58.72 57.49 12.37 46.35 45.11 26 50 38 10.63 25.27 92.53 89.50 18.81 73.72 70.70 51 75 63 12.65 37.92 130.09 125.08 31.17 98.93 93.91 76 100 88 13.91 51.83 167.66 160.65 44.77 122.89 115.88 101 125 113 12.87 64.70 205.22 196.23 59.58 145.64 136.65 126 140 133 6.71 71.41 235.28 224.69 71.51 163.77 153.18 141 150 145.5 3.91 75.31 254.06 242.48 110.96 143.10 131.52 151 175 163 7.53 82.85 280.36 267.38 141.32 139.03 126.06 176 200 188 5.11 87.96 317.92 302.95 185.63 132.29 117.33 201 225 213 3.41 91.37 355.49 338.53 233.35 122.14 105.18 226 250 238 2.31 93.68 393.05 374.10 293.44 99.61 80.66 301 400 350.5 2.18 98.64 562.09 534.19 545.09 17.01 -10.90 401 450 425.5 0.45 99.09 674.79 640.92 665.90 8.89 -24.98 451 500 475.5 0.28 99.37 749.92 712.07 742.40 7.52 -30.33 501 600 550.5 0.29 99.65 877.45 818.79 872.62 4.83 -53.83 601 750 675.5 0.17 99.83 1,101.99 996.67 1,097.38 4.60 -100.72 751 850 800.5 0.05 99.88 1,326.53 1,174.54 1,326.53 0.00 -151.99 851 900 875.5 0.02 99.89 1,461.25 1,281.27 1,461.25 0.00 -179.98 901 1,000 950.5 0.02 99.92 1,595.97 1,387.99 1,595.97 0.00 -207.98 1001 1,200 1100.5 0.03 99.95 1,865.42 1,601.44 1,865.42 0.00 -263.98 1201 2,000 1600.5 0.00 99.95 2,763.58 2,312.94 2,763.58 0.00 -450.64 2001 2500 2250.5 0.00 99.95 3,931.18 3,237.89 3,931.18 0.00 -693.29 2501 4500 3500.5 0.00 99.96 6,176.58 5,016.64 6,176.58 0.00 -1,159.94 TOTAL (million pesos) 1,732 1,654 852 881 802 Table 8 CFE TARIFF 1 HOUSEHOLDS IN 1 AND IN DAC WINTER Monthly average consumption (kWh) 122.1 Number of Households 8,486,435 HISTOGRAM WITH SOURCE DATA SUBSIDY CONSUMPTION CUMULATE DAC ESTIMATED BASED ON ESTIMATED INTERVAL midpoint HOUSEHOLDS HOUSEHOLDS TARIFF COST PAYMENT DAC SUBSIDY kWh/month % % $/hh-month $/hh-month $/hh-month $/hh-month $/hh-month 0 5 2.5 6.62 6.62 34.94 34.76 11.96 22.98 22.80 6 25 15.5 7.17 13.79 52.23 51.14 11.95 40.28 39.19 26 50 38 10.67 24.46 82.17 79.49 18.16 64.01 61.33 51 75 63 12.86 37.32 115.43 110.99 30.07 85.37 80.92 76 100 88 13.77 51.09 148.70 142.49 43.16 105.53 99.33 101 125 113 12.56 63.66 181.96 173.99 57.41 124.55 116.58 126 140 133 6.54 70.19 208.57 199.20 68.93 139.64 130.27 141 150 145.5 3.83 74.02 225.20 214.95 103.50 121.70 111.44 151 175 163 7.63 81.66 248.49 237.00 132.73 115.76 104.27 176 200 188 5.37 87.03 281.75 268.50 175.30 106.45 93.20 201 225 213 3.67 90.70 315.01 300.00 220.69 94.32 79.31 226 250 238 2.48 93.18 348.28 331.50 272.64 75.63 58.85 251 300 275.5 2.97 96.15 398.17 378.75 354.89 43.28 23.86 301 400 350.5 2.33 98.48 497.96 473.25 482.70 15.26 -9.44 401 450 425.5 0.48 98.96 597.75 567.76 592.30 5.45 -24.55 451 500 475.5 0.31 99.26 664.28 630.76 661.39 2.88 -30.63 501 600 550.5 0.32 99.58 777.63 725.26 776.61 1.01 -51.35 601 750 675.5 0.20 99.78 977.50 882.77 977.23 0.27 -94.46 751 850 800.5 0.06 99.84 1,177.38 1,040.27 1,177.38 0.00 -137.10 851 900 875.5 0.02 99.86 1,297.30 1,134.78 1,297.30 0.00 -162.52 901 1,000 950.5 0.03 99.89 1,417.23 1,229.28 1,417.23 0.00 -187.95 1001 1,200 1100.5 0.03 99.92 1,657.08 1,418.28 1,657.08 0.00 -238.79 1201 2,000 1600.5 0.03 99.95 2,456.58 2,048.30 2,456.58 0.00 -408.27 2001 2500 2250.5 0.00 99.96 3,495.93 2,867.33 3,495.93 0.00 -628.59 2501 4500 3500.5 0.00 99.96 5,494.68 4,442.38 5,494.68 0.00 -1,052.29 TOTAL (million pesos) 1,573 1,501 829 745 672 - 168 - Table 9 CFE TARIFF 1 HOUSEHOLDS IN 1 AND IN DAC SUMMER Monthly average consumption (kWh) 110.6 Number of Households 8,486,435 HISTOGRAM NORMALIZED BY DECILES SUBSIDY CONSUMPTION CUMULATE AVERAGE ADJUSTED AVERAGE DAC ESTIMATED BASED ON ESTIMATED INTERVAL HOUSEHOLDS HOUSEHOLDS CONSUMPTION CONSUMPTION TARIFF COST PAYMENT DAC SUBSIDY kWh/month % % kWh/month kWh/month $/hh-month $/hh-month $/hh-month $/hh-month $/hh-month 0 13 10 10 6.5 7 45.20 44.68 12.38 32.82 32.30 14 38 10 20 26 27 74.50 72.43 15.62 58.88 56.81 39 59 10 30 49 51 109.06 105.16 23.43 85.63 81.73 60 79 10 40 69.5 73 139.86 134.33 33.20 106.66 101.13 80 96 10 50 88 92 167.66 160.65 42.67 124.99 117.99 97 114 10 60 105.5 110 193.96 185.56 51.22 142.73 134.34 115 136 10 70 125.5 131 224.01 214.02 57.31 166.70 156.71 137 166 10 80 151.5 158 263.08 251.01 129.84 133.23 121.17 167 210 10 90 188.5 197 318.67 303.67 203.34 115.34 100.33 211 288 10 100 249.5 260 410.33 390.47 353.91 56.42 36.56 AVERAGE (kWh/month) M$/month M$/month M$/month M$/month M$/month TOTAL (million pesos) 106.0 110.6 1,652 1,580 783 869 797 Table 10 CFE TARIFF 1 HOUSEHOLDS IN 1 AND IN DAC WINTER Monthly average consumption (kWh) 122.1 Number of customers 8,486,435 HISTOGRAM NORMALIZED BY DECILES SUBSIDY CONSUMPTION CUMULATE AVERAGE ADJUSTED AVERAGE DAC ESTIMATED BASED ON ESTIMATED INTERVAL HOUSEHOLDS HOUSEHOLDS CONSUMPTION CONSUMPTION TARIFF COST PAYMENT DAC SUBSIDY kWh/month % % kWh/month kWh/month $/hh-month $/hh-month $/month $/hh-month $/hh-month 0 14 10 10 7 8 40.92 40.43 11.95 28.97 28.48 15 40 10 20 27.5 31 68.20 66.26 15.56 52.64 50.70 41 61 10 30 51 58 99.47 95.87 23.29 76.18 72.58 62 80 10 40 71 80 126.08 121.07 32.62 93.46 88.46 81 98 10 50 89.5 101 150.69 144.38 41.93 108.77 102.46 99 117 10 60 108 122 175.31 167.69 50.97 124.33 116.72 118 139 10 70 128.5 145 202.58 193.53 67.18 135.40 126.34 140 170 10 80 155 175 237.84 226.92 126.39 111.45 100.53 171 214 10 90 192.5 217 287.74 274.17 198.87 88.87 75.30 215 288 10 100 251.5 284 366.24 348.51 335.81 30.43 12.70 AVERAGE (kWh/month) M$/month M$/month M$/month M$/month M$/month TOTAL (million pesos) 108.2 122.1 1,489 1,425 768 722 657 - 169 - Table 11 Adjusted Coefficients by decile required to include non-served households ALL CUMULATE coefficients HOUSEHOLDS HOUSEHOLDS true by decile % % weights previous current 5 5 4.76 10 15 9.52 1 0.55 10.00 10 25 9.52 0.45 0.60 10.00 10 35 9.52 0.4 0.65 10.00 10 45 9.52 0.35 0.70 10.00 10 55 9.52 0.3 0.75 10.00 10 65 9.52 0.25 0.80 10.00 10 75 9.52 0.2 0.85 10.00 10 85 9.52 0.15 0.90 10.00 10 95 9.52 0.1 0.95 10.00 10 105 9.52 0.05 1.00 10.00 100.0 Table 12 CFE TARIFF 1 HOUSEHOLDS IN 1 AND IN DAC SUMMER Monthly average consumption (kWh) 105.3 Number of Households 8,910,757 HISTOGRAM NORMALIZED BY DECILES, INCLUDING HOUSEHOLDS WITHOUT ELECTRICITY ADJUSTED AVERAGE AVERAGE UNITARY SUBSIDIES BY DECILE DECIL CONSUMPTION CONSUMPTION SUBSIDY kWh/month kWh/month $/kWh $/hh-month $/hh-season 1 4 4 5.00 17.77 106.59 2 19 18 2.64 48.62 291.73 3 44 42 1.81 75.85 455.09 4 69 65 1.52 99.40 596.37 5 91 86 1.38 118.83 712.97 6 111 106 1.29 136.96 821.79 7 133 127 1.26 160.07 960.40 8 162 154 0.86 132.56 795.37 9 203 193 0.56 107.43 644.58 10 270 257 0.16 41.58 249.47 kWh/month kWh/month M$/month M$/year 105.3 105.3 110.6 105.3 837 5,021 - 170 - Table 13 CFE TARIFF 1 HOUSEHOLDS IN 1 AND IN DAC WINTER 116.3 Number of customers 8,910,757 HISTOGRAM NORMALIZED BY DECILES, INCLUDING HOUSEHOLDS WITHOUT ELECTRICITY ADJUSTED SUBSIDY BY DECILE AVERAGE AVERAGE UNITARY INCLUDING NON-SERVED DECIL CONSUMPTION CONSUMPTION SUBSIDY HOUSEHOLDS kWh/month kWh/month $/kWh $/hh-month $/hh-season 1 4 4 3.78 15.66 93.98 2 22 21 2.05 43.23 259.40 3 50 47 1.42 67.46 404.74 4 76 73 1.20 87.32 523.93 5 100 95 1.09 103.38 620.28 6 123 117 1.02 118.99 713.95 7 148 141 0.93 130.73 784.40 8 179 171 0.64 109.43 656.56 9 224 213 0.38 81.59 489.52 10 295 281 0.06 16.46 98.78 kWh/month kWh/month M$/month M$/year 116.3 116.3 122.1 116.3 690 4,140 Table 14 CFE TARIFF 1 ANNUAL 2002 Monthly average consumption (kWh) 110.8 Number of households 8,910,757 Total annual consumption (GWh) 11,847 SUBSIDIES BY DECILE AVERAGE UNITARY INCLUDING HOUSEHOLDS WITHOUT ELECTRICITY DECILE CONSUMPTION SUBSIDY Cumulate kWh/month $/kWh $/hh-month $/hh-year M$/year % % 1 4 4.35 16.71 201 179 2.0% 2.0% 2 20 2.32 45.93 551 491 5.4% 7.3% 3 45 1.60 71.65 860 766 8.4% 15.7% 4 69 1.35 93.36 1,120 998 10.9% 26.6% 5 91 1.22 111.10 1,333 1,188 13.0% 39.5% 6 111 1.15 127.98 1,536 1,368 14.9% 54.5% 7 134 1.09 145.40 1,745 1,555 17.0% 71.5% 8 162 0.74 120.99 1,452 1,294 14.1% 85.6% 9 203 0.47 94.51 1,134 1,011 11.0% 96.6% 10 269 0.11 29.02 348 310 3.4% 100.0% kWh/month $/kWh M$/month M$/year 9,159 110.8 0.77 763 9,160 - 171 - ANNEX 2 A1 DATA, METHODOLOGICAL ASSUMPTIONS AND REDISTRIBUTIVE DEMANDS 1. This section describes the data used and methodological assumptions made in the imputation of benefits to households for the specific programs analyzed in chapter 2. It also includes some information on redistrubutive demands relevant to section 2.1.3 in the main text. 1. EDUCATION. 2. Coverage. Prior to 1992 the ENIGH did not distinguish between students in public and private education. Since then, this source reports whether a person is currently enrolled in a public school, and the level of education attained, from which we may infer the level of current enrollment. Before 1996, the level attained was reported in terms of complete/incomplete cycles (primary, lower secondary, higher secondary, tertiary, post-graduate), and since then in individual school years. With the exception of tertiary education, the total enrollment for each level obtained from ENIGH matches closely administrative enrollment records reported by SEP. 3. Public expenditure. The information available on public education expenditure has some limitations associated with the decentralization process. First, there is no consistent source for education spending by state governments from own resources or tagged transfers (aportaciones estatales). We estimate public (federal and state) spending at each educational level by applying the cost per student reported by the federal government to the total public enrollment reported by SEP, which includes federal, "federalized", autonomous, and state schools. This assumes that costs (mainly teacher salaries) are on average the same at each educational level across the public education system, independently of the source of finance and control. Secondly, however, administrative information on the distribution of basic education spending between its three sub- levels (preschool, primary, and lower secondary) is unavailable even for the federally (FAEB) financed schools after 1998-1999. For 1992 to 1999 we use 6º Informe de Gobierno (Zedillo, 2000). For more recent years we use estimates provided by SEP. 2. HEALTH. 4. Coverage. The ENIGH reports use of health services in public institutions, including the services offered to the uninsured population by the health ministry (SSA) and the social security institutions serving formal sector workers (Instituto Mexicano del Seguro Social, IMSS), public sector workers (Instituto de Seguridad y Servicios Sociales de los Trabajadores del Estado, ISSSTE), and workers of the state oil company (PEMEX) and military personnel. The type of services used is identified in 44 (77 in 2002) categories, but we report three broad aggregates here: primary health care, maternal health care (pregnancy and delivery), and hospital services. 5. Public expenditures. We use the National Health Accounts recently compiled by SSA, except in the case of IMSS, where we refer directly to financial reports published by that institutions (Coordinación de Presupuesto, Contabilidad y Evaluación Financiera). - 172 - 3. PENSIONS. 6. Coverage. We use workers affiliated to IMSS to impute government contributions to the pension funds of active IMSS-affiliated workers. In 2002 ENIGH reports for the first time whether a worker has an individualized pension account under the new law (AFORE), and there is some (unexplained) discrepancy with the distribution and total number of workers reporting IMSS affiliation. The imputation of government spending on current pensions is more uncertain, as ENIGH does not report the institutional source of pensions received, or even separate between private and public sources. To estimate the relevant distribution we use pensioners who also report being IMSS/ISSSTE right holders or whose household reports using health services from these institutions, but this captures only a small proportion of pensions received. Given these difficulties (and the impossibility of neting out past contributions), we present the relevant distributions, but do not impute government spending on current pensions when we obtain the distribution of total benefits. 7. Public expenditures. We consider government contributions to the pension funds of active workers affiliated to IMSS, which under the 1995 IMSS Law includes the “Cuota Social” and a contribution to the Seguro de Invalidez y Vida. In the case of pensioners, we consider pensions currently being paid under the previous IMSS law (1973), and the deficit of the ISSSTE pension fund. At present, both are fully financed by the federal government (see table 2.4.1). 4. MONETARY TRANSFERS: OPORTUNIDADES AND PROCAMPO. 8. Coverage. The ENIGH 2002 reports benefits received by households from the two most important programs of direct monetary transfers operated in Mexico at present, Oportunidades (previously PROGRESA) and PROCAMPO. In both cases there are problems of comparability with earlier ENIGHs. Oportunidades benefits where not reported before 2002, though they were captured in a broader question about scholarship programs and donations from institutions, public as well as private. However, the ENIGH 2000 (and 2002) does report whether students receive scholarships from this program (not the amount of benefits received). For this year we have used the change between 2000 and 2002 in the latter distribution to estimate the distribution of benefits, assuming the relationship between the distribution of beneficiaries and benefits in 2000 to be as observed in 2002. Since the program was extended from basic to upper-secondary education in 2001, and the value of the scholarships at this level is substantially higher than at the lower secondary level, this underestimates the progressivity of benefits in 2000 (and thus the reduction in progressivity between 2000 and 2002, section 2.2.5). 9. Public expenditures. The transfers received from Oportunidades and Procampo in 2002 reported in ENIGH fall short of the transfers reported by the federal government (net of administrative costs) by about 80 percent. This may be explained by limits in the representativity of the survey and by underreporting of income. To estimate the incidence of benefits we use the distribution from ENIGH but impute the transfers reported by the government to make the analysis consistent with the other programs. When we estimate the impact on poverty, however, we use transfers as reported in ENIGH, to ensure consistency with poverty measures. 10. In the case of PROCAMPO the problem of comparability is more subtle. In 1996, 1998, and 2000 the number of cases captured by the survey was small and the total amount of transfers - 173 - reported in the latter year represent only 36 percent of transfers reported by the government. This may be explained by a failure to capture adequately transfers received by upper income producers, which are small in number but absorb a large part of the program’s resources (as these are allocated per hectare). This problem of representation is considerably reduced in 2002 (with 1378 cases, in contrast to only 388 in 2000). We assume the distribution of benefits observed in this year to be the more accurate one, and discard the 2000 observation as unrepresentative. 5. ELECTRICITY SUBSIDY 11. Coverage. The ENIGHs from 1992 to 2002 report access to electricity (in 2002 it also identifies consumers with a formal contract), as well as household expenditure on electricity. In contrast to total income and expenditure, which are significantly underreported in ENIGH in comparison to the National Accounts, ENIGH over reports monthly electricity expenditure in relation to administrative records by a magnitude by a factor close of almost 2. This may be explained by the fact that most households pay a bimonthly electricity bill but may forget to divide this when reporting monthly expenditures in ENIGH. 12. Public expenditure. We apply the residential tariff schedule covering the majority of households and close to half of total residential electricity expenditure (tariff 1) to impute electricity consumption (kWh) to households on the basis of their reported expenditure (divided by two, for the reason mentioned in the previous figure). The great majority of the rest of households and expenditure corresponds is to tariffs 1A-1C, which are sufficiently similar to the latter to make this a reasonable estimate for them as well. To estimate costs of provision we assumed the highest tariff rate for residential consumption to reflect true marginal cost, which implies a total subsidy for residential consumers close to the estimate reported by the federal government. For 1992-2000 this implies a single tariff schedule with three progressive marginal rates, but the 2002 tariff reform implies three different schedules, making direct imputation of electricity consumption levels more difficult in this year. For this reason, to estimate the distribution of the subsidy in the latter year we have applied the 2002 tariffs to the consumption structure estimated for 2000 (see figure A1.1), assuming the latter to have remained basically stable between the two years. - 174 - Figure A1.1 Simulated distribution of household expenditures on electricity (2002 tariffs, 2000 electricity consumption) 500 450 Low consumption tariff Medium consumption tariff 400 High consumption tariff 350 Observed expenditure pesos (base=2002) 300 250 200 150 100 50 0 1 7 13 19 25 31 37 43 49 55 61 67 73 79 85 91 97 Household centiles Data: ENIGH 2000, 2002; Fig. A2.9 in this Annex. Table A1.1 Evolution of national population shares by sector and age groups RURAL 0-14 65+ 1992 27.8% 37.9% 4.2% 1994 26.9% 38.2% 4.4% 1996 26.3% 38.4% 4.7% 1998 25.7% 38.3% 5.1% 2000 25.3% 37.9% 5.5% 2002 24.8% 37.4% 5.9% Data: ENIGH 19992-2002. Table A1.2 Distribution of Population by Age Groups: 2002 (age) Deciles 0-5 6-12 13-15 16-18 19-24 25-45 46-64 65 + 1 13.4% 14.2% 13.0% 10.9% 8.4% 7.2% 8.1% 7.9% 2 12.2% 13.8% 12.0% 9.8% 6.9% 8.4% 8.3% 10.4% 3 11.7% 12.6% 12.6% 10.7% 8.3% 9.2% 7.2% 9.4% 4 10.9% 11.1% 10.9% 9.5% 9.5% 9.4% 9.0% 10.0% 5 11.5% 10.5% 9.6% 9.8% 9.9% 10.3% 7.6% 10.3% 6 10.2% 8.7% 9.4% 10.6% 11.4% 10.0% 10.5% 9.2% 7 8.8% 8.5% 9.5% 10.6% 11.1% 10.6% 11.0% 8.9% 8 7.6% 8.3% 8.4% 10.7% 12.2% 10.6% 11.5% 10.3% 9 7.6% 6.7% 8.1% 9.2% 10.5% 11.9% 12.3% 10.6% 10 6.2% 5.7% 6.6% 8.1% 11.6% 12.6% 14.5% 13.0% Data: ENIGH 2002. Population deciles ordered by per capita expenditure. - 175 - A2 DISTRIBUTION OF BENEFITS: 1992-2002 Table A2.1 Distribution of total schooling years of adult population (25-65 years) Deciles 1984 1989 1992 1994 1996 1998 2000 2002 1 3.3% 2.7% 2.8% 2.8% 2.9% 2.6% 2.8% 3.3% 2 3.8% 4.2% 4.2% 4.1% 4.5% 4.3% 4.2% 4.5% 3 5.2% 5.3% 5.3% 5.1% 5.3% 5.8% 5.9% 6.1% 4 5.5% 6.2% 6.3% 6.2% 6.6% 7.0% 7.3% 7.0% 5 7.2% 7.9% 7.3% 7.6% 7.8% 8.2% 8.6% 8.2% 6 8.6% 9.1% 8.8% 8.6% 9.5% 9.1% 9.8% 9.4% 7 10.6% 10.6% 10.7% 10.7% 10.5% 10.4% 10.8% 10.8% 8 14.2% 13.6% 13.1% 12.4% 12.8% 13.3% 13.1% 12.4% 9 17.4% 16.8% 16.6% 17.0% 16.8% 16.8% 15.3% 16.1% 10 24.3% 23.5% 24.8% 25.7% 23.3% 22.4% 22.2% 22.1% CC 0.3453 0.3311 0.3390 0.3473 0.3195 0.3135 0.3001 0.2942 Data: ENIGH 1984, 1989, 1992, 1994, 1996, 1998, 2000, 2002. Population deciles ordered by per capita income. Table A2.2 Gains in schooling of adult population (25-65 years) 1984-1994 1994-2002 1984-2002 Absolute Distribution of Deciles Percentage gain gain gain 1 13.6% 35.5% 1.11 4.7% 2 38.3% 31.4% 1.88 8.0% 3 17.7% 41.4% 1.99 8.4% 4 42.4% 32.9% 2.65 11.2% 5 35.6% 31.4% 2.83 12.0% 6 29.2% 27.7% 2.69 11.4% 7 28.2% 16.8% 2.47 10.5% 8 12.8% 17.1% 1.98 8.4% 9 17.8% 14.7% 2.54 10.8% 10 30.8% 7.5% 3.47 14.7% Data: ENIGH 1984, 1989, 1992, 1994, 1996, 1998, 2000, 2002. Population deciles ordered by per capita income. - 176 - Figure A2.1 Enrollment in Primary Education as a Percentage of 6-12 year population 120% Private Public 100% 80% 60% 40% 20% 0% 1 2 3 4 5 6 7 8 9 10 Enrollment in Lower Secondary Education as a Percentage of 13-15 year population 140% Private Public 120% 100% 80% 60% 40% 20% 0% 1 2 3 4 5 6 7 8 9 10 Data: ENIGH 2002. Population deciles ordered by per capita expenditure. - 177 - Figure A2.2 Enrollment in Upper Secondary Education as a Percentage of 16-18 year population 100% Private 90% Public 80% 70% 60% 50% 40% 30% 20% 10% 0% 1 2 3 4 5 6 7 8 9 10 Enrollment in Tertiary Education as a Percentage of 19-24 year population 60% Private Public 50% 40% 30% 20% 10% 0% 1 2 3 4 5 6 7 8 9 10 Data: ENIGH 2002. Population deciles ordered by per capita expenditure. - 178 - Figure A2.3 Distribution of Students in Public Education (million) 0.9 Pre-school 0.8 1992 0.7 2000 2002 0.6 0.5 0.4 0.3 0.2 0.1 0.0 1 2 3 4 5 6 7 8 9 10 2.5 Primary 2.0 1992 2000 2002 1.5 1.0 0.5 0.0 1 2 3 4 5 6 7 8 9 10 Population Deciles 0.8 Lower Secondary 0.7 1992 0.6 2000 2002 0.5 0.4 0.3 0.2 0.1 0.0 1 2 3 4 5 6 7 8 9 10 Population Deciles Data: ENIGH 1992, 2000, 2002. Population deciles ordered by per capita expenditure. - 179 - Figure A2.4 Distribution of Students in Public Education (million) 0.50 Upper Secondary 0.45 0.40 1992 2000 0.35 2002 0.30 0.25 0.20 0.15 0.10 0.05 0.00 1 2 3 4 5 6 7 8 9 10 Population Deciles 0.40 Tertiary 0.35 1992 0.30 2000 2002 0.25 0.20 0.15 0.10 0.05 0.00 1 2 3 4 5 6 7 8 9 10 Population Deciles Data: ENIGH 1992, 2000, 2002. Population deciles ordered by per capita expenditure. - 180 - Table A2.3 Distribution of Public Expenditure on Education Services: 2002 Lower Upper Deciles Total Preschool Primary Tertiary Secondary Secondary URBAN 1 8.7% 17.5% 17.0% 10.4% 6.1% 2.6% 2 10.4% 13.9% 15.7% 11.0% 8.6% 3.2% 3 9.8% 13.5% 13.3% 12.0% 8.5% 4.0% 4 9.4% 13.1% 10.4% 11.7% 9.3% 12.1% 5 9.6% 11.1% 11.1% 11.6% 8.6% 7.6% 6 11.0% 8.9% 9.3% 11.2% 14.8% 10.2% 7 12.0% 8.3% 8.7% 9.8% 11.3% 11.9% 8 11.5% 6.0% 6.9% 11.1% 12.8% 11.9% 9 9.8% 5.8% 4.7% 7.4% 13.1% 19.3% 10 7.9% 2.0% 2.8% 3.7% 7.1% 17.2% CC 0.007 -0.250 -0.244 -0.096 0.074 0.286 RURAL 1 10.9% 14.0% 12.1% 8.7% 6.7% 0.2% 2 11.7% 15.2% 13.3% 9.8% 4.2% 0.0% 3 9.9% 13.3% 10.5% 9.8% 3.1% 0.0% 4 10.4% 8.7% 10.4% 11.0% 13.5% 0.0% 5 11.0% 10.5% 11.8% 11.6% 7.9% 0.0% 6 10.7% 9.5% 11.3% 10.3% 10.0% 9.9% 7 11.0% 9.5% 10.8% 11.1% 14.4% 12.9% 8 9.5% 8.0% 8.2% 10.2% 17.2% 14.0% 9 7.9% 5.7% 6.6% 9.4% 13.9% 17.7% 10 7.0% 5.6% 5.1% 8.3% 9.0% 45.2% CC -0.061 -0.167 -0.121 -0.006 0.163 0.648 Source: ENIGH 2002, table 2.2.1. Population deciles ordered by per capita expenditure. - 181 - Table A2.4a Distribution of Marginal Benefits from Growth in Public Education Expenditure: 1992-2002 Lower Upper Deciles Total Preschool Primary Tertiary Secondary Secondary Change in Expenditure Share (percentage points gained) 1 2.2% 1.0% 1.8% 6.4% 2.4% -0.5% 2 1.6% -1.1% 1.5% 5.0% 1.8% -0.5% 3 1.6% 0.7% -0.3% 2.7% 4.9% 1.9% 4 0.9% 0.6% -0.1% -0.8% 2.1% 2.8% 5 -0.4% 0.1% 0.3% -3.4% -1.2% 0.9% 6 -1.4% -1.2% -0.9% -3.7% -4.4% 0.6% 7 -0.8% 0.4% -0.5% -2.4% -1.1% -0.8% 8 1.2% -1.1% -0.1% -1.1% 0.1% 7.4% 9 -2.0% -0.2% -0.5% -0.3% 0.0% -6.8% 10 -3.0% 0.6% -1.3% -2.5% -4.6% -5.1% Urban -3.1% 1.9% -1.6% -9.4% -4.1% -0.4% Rural 3.1% -1.9% 1.6% 9.4% 4.1% 0.4% Distribution of Additional Public Expenditure 1 12.4% 18.7% 18.3% 16.7% 5.5% -0.9% 2 11.3% 11.6% 17.0% 16.1% 6.4% -0.2% 3 12.5% 14.3% 13.4% 16.1% 11.6% 5.9% 4 10.9% 13.5% 11.8% 9.7% 10.6% 9.5% 5 9.2% 10.5% 11.1% 6.3% 9.3% 8.4% 6 7.5% 8.2% 7.5% 6.4% 6.5% 9.3% 7 9.6% 9.2% 7.9% 8.8% 13.6% 9.6% 8 13.9% 6.2% 7.4% 9.3% 16.1% 35.3% 9 8.4% 4.8% 5.0% 9.4% 14.0% 9.9% 10 4.1% 2.9% 0.4% 1.3% 6.3% 13.1% Urban 73.7% 67.8% 64.4% 63.4% 84.1% 96.9% Rural 26.3% 32.2% 35.6% 36.6% 15.9% 3.1% Data: ENIGH 1992-2002, table 2.2.3. Population deciles ordered by per capita expenditure. - 182 - Table A2.4b Distribution of Marginal Benefits: Preschool 1992-1996 1996-2000 2000-2002 1992-2002 Thousand Additional 375 93 187 655 Students Coverage % 14.4% 3.1% 6.1% 25.1% Million$ Additional 7,197 5,061 3,028 15,286 2002 Spending % 52.2% 24.1% 11.6% 110.8% Deciles Change in expenditure share (percentage points) 1 1.6% -2.5% 1.8% 1.0% 2 0.7% 1.2% -3.0% -1.1% 3 -0.3% -0.5% 1.5% 0.7% 4 -1.6% 0.8% 1.4% 0.6% 5 -0.1% 2.6% -2.3% 0.1% 6 -1.3% 0.4% -0.2% -1.2% 7 -0.4% 0.2% 0.6% 0.4% 8 -0.1% -0.5% -0.5% -1.1% 9 0.6% -1.3% 0.5% -0.2% 10 0.8% -0.3% 0.1% 0.6% Urban -2.8% 4.2% 0.5% 1.9% Rural 2.8% -4.2% -0.5% -1.9% Deciles Distribution of additional spending 1 21.5% 5.8% 33.7% 18.7% 2 15.8% 20.6% -13.2% 11.6% 3 12.2% 10.0% 26.7% 14.3% 4 7.8% 14.7% 25.1% 13.5% 5 9.9% 23.3% -9.6% 10.5% 6 6.6% 11.0% 7.2% 8.2% 7 7.2% 9.1% 14.0% 9.2% 8 7.9% 5.6% 3.1% 6.2% 9 6.9% -1.0% 9.7% 4.8% 10 4.1% 0.9% 3.4% 2.9% Urban 55.9% 83.1% 70.6% 67.8% Rural 44.1% 16.9% 29.4% 32.2% Data: ENIGH 1992, 1996, 2000, 2002. Population deciles ordered by per capita expenditure. - 183 - Table A2.4c Distribution of Marginal Benefits: Primary 1992-1996 1996-2000 2000-2002 1992-2002 Thousand Additional 221 -76 19 163 Students Coverage % 1.6% -0.6% 0.1% 1.2% Million$ Additional 24,578 13,841 2,307 40,725 2002 Spending % 39.0% 15.8% 2.3% 64.6% Deciles Change in expenditure share (percentage points) 1 0.3% 3.0% -1.5% 1.8% 2 0.6% 1.0% 0.0% 1.5% 3 -1.2% 1.1% -0.1% -0.3% 4 0.8% -1.7% 0.9% -0.1% 5 0.7% -0.8% 0.4% 0.3% 6 0.2% 0.2% -1.3% -0.9% 7 -0.9% 0.5% -0.1% -0.5% 8 0.5% -1.8% 1.3% -0.1% 9 -0.2% -0.7% 0.5% -0.5% 10 -0.7% -0.7% 0.0% -1.3% Urban -1.9% -1.1% 1.4% -1.6% Rural 1.9% 1.1% -1.4% 1.6% Deciles Distribution of additional spending 1 14.7% 36.3% -51.3% 18.3% 2 15.3% 20.9% 12.5% 17.0% 3 9.9% 20.6% 8.4% 13.4% 4 14.7% 0.1% 51.6% 11.8% 5 12.8% 5.2% 28.8% 11.1% 6 10.4% 11.6% -48.1% 7.5% 7 6.1% 11.9% 3.2% 7.9% 8 9.3% -5.4% 64.1% 7.4% 9 5.4% 0.7% 26.7% 5.0% 10 1.4% -1.8% 4.1% 0.4% Urban 61.6% 58.4% 130.5% 64.4% Rural 38.4% 41.6% -30.5% 35.6% Data: ENIGH 1992, 1996, 2000, 2002. Population deciles ordered by per capita expenditure. - 184 - Table A2.4d Distribution of Marginal Benefits: Secondary 1992-1996 1996-2000 2000-2002 1992-2002 Thousand Additional 615 453 283 1,351 Students Coverage % 15.9% 10.1% 5.8% 35.0% Million$ Additional 9,054 14,588 4,433 28,075 2002 Spending % 27.9% 35.1% 7.9% 86.4% Change in expenditure share (percentage Deciles points) 1 1.2% 3.4% 1.8% 6.4% 2 1.8% 4.3% -1.0% 5.0% 3 -0.9% 2.0% 1.6% 2.7% 4 0.3% -1.6% 0.5% -0.8% 5 -2.1% -0.3% -1.0% -3.4% 6 -1.1% -3.0% 0.5% -3.7% 7 -2.0% 0.7% -1.0% -2.4% 8 2.0% -3.8% 0.7% -1.1% 9 1.0% -1.1% -0.2% -0.3% 10 -0.2% -0.5% -1.8% -2.5% Urban -2.3% -6.2% -0.8% -9.4% Rural 2.3% 6.2% 0.8% 9.4% Deciles Distribution of additional spending 1 8.6% 17.2% 31.9% 16.7% 2 13.4% 23.5% -2.7% 16.1% 3 6.0% 17.2% 33.2% 16.1% 4 12.8% 5.6% 16.9% 9.7% 5 4.1% 10.2% -2.3% 6.3% 6 9.1% 1.5% 17.3% 6.4% 7 4.7% 14.5% -1.8% 8.8% 8 20.8% -1.0% 19.5% 9.3% 9 14.7% 6.7% 7.1% 9.4% 10 5.8% 4.6% -19.1% 1.3% Urban 72.8% 57.5% 63.5% 63.4% Rural 27.2% 42.5% 36.5% 36.6% Data: ENIGH 1992, 1996, 2000, 2002. Population deciles ordered by per capita expenditure. - 185 - Table A2.4e Distribution of Marginal Benefits: Upper Secondary 1992- 1996-2000 2000-2002 1992-2002 1996 Thousand Additional 429 218 272 920 Students Coverage % 25.8% 10.4% 11.8% 55.3% Million$ Additional 12,805 4,696 6,997 24,499 2002 Spending % 62.6% 14.1% 18.4% 119.7% Deciles Change in expenditure share (percentage points) 1 -0.2% 0.9% 1.7% 2.4% 2 -0.6% 0.6% 1.7% 1.8% 3 1.0% 2.2% 1.7% 4.9% 4 0.7% 1.1% 0.3% 2.1% 5 -0.8% 0.5% -0.9% -1.2% 6 -3.8% -0.1% -0.6% -4.4% 7 -0.2% -2.9% 2.0% -1.1% 8 -0.5% -1.2% 1.9% 0.1% 9 3.5% 2.4% -5.8% 0.0% 10 1.0% -3.5% -2.1% -4.6% Urban -0.2% -1.5% -2.4% -4.1% Rural 0.2% 1.5% 2.4% 4.1% Deciles Distribution of additional spending 1 0.5% 8.1% 12.8% 5.5% 2 1.6% 7.4% 14.3% 6.4% 3 5.2% 21.6% 16.8% 11.6% 4 8.5% 16.4% 10.6% 10.6% 5 9.5% 14.8% 5.4% 9.3% 6 4.8% 10.4% 7.1% 6.5% 7 15.1% -8.0% 25.5% 13.6% 8 14.5% 5.3% 26.4% 16.1% 9 23.0% 36.5% -17.7% 14.0% 10 17.3% -12.5% -1.1% 6.3% Urban 91.2% 79.1% 74.6% 84.1% Rural 8.8% 20.9% 25.4% 15.9% Data: ENIGH 1992, 1996, 2000, 2002. Population deciles ordered by per capita expenditure. - 186 - Table A2.4f Distribution of Marginal Benefits: Tertiary 1992- 1996-2000 2000-2002 1992-2002 1996 Thousand Additional 1,812 877 862 3,551 Students Coverage % 16.7% 15.7% 7.3% 44.8% Million$ Additional 60,334 47,419 22,305 130,058 2002 Spending % 35.8% 20.7% 8.1% 77.3% Change in expenditure share (percentage Deciles points) 1 0.6% 1.7% -0.1% -0.5% 2 0.6% 1.6% -0.7% -0.5% 3 0.1% 1.2% 0.3% 1.9% 4 0.6% -0.5% 0.8% 2.8% 5 -0.1% 0.3% -0.5% 0.9% 6 -1.0% 0.0% -0.5% 0.6% 0.7 -0.6% 0.3% -0.4% -0.8% 8 0.8% -2.3% 2.7% 7.4% 9 -1.4% 0.1% -0.7% -6.8% 10 0.3% -2.3% -1.0% -5.1% Urban -2.5% -1.1% 0.6% -0.4% Rural 2.5% 1.1% -0.6% 0.4% Deciles Distribution of additional spending 1 9.7% 17.6% 8.9% -0.9% 2 10.1% 17.9% 0.9% -0.2% 3 9.3% 15.8% 14.4% 5.9% 4 11.3% 6.4% 19.8% 9.5% 5 9.5% 11.8% 3.0% 8.4% 6 7.1% 9.9% 3.2% 9.3% 7 9.0% 12.4% 5.4% 9.6% 8 14.2% -1.6% 46.2% 35.3% 9 7.8% 12.0% 2.5% 9.9% 10 12.1% -2.2% -4.2% 13.1% Urban 71.1% 71.6% 85.1% 96.9% Rural 28.9% 28.4% 14.9% 3.1% Data: ENIGH 1992, 1996, 2000, 2002. Population deciles ordered by per capita expenditure. - 187 - Table A2.4g Distribution of Marginal Benefits: Total Education Expenditure 1992- 1996-2000 2000-2002 1992-2002 1996 Thousand Additional 1,812 877 862 3,551 Students Coverage % 8.0% 3.6% 3.4% 15.7% Million$ Additional 60,334 47,419 22,305 130,058 2002 Spending % 35.8% 20.7% 8.1% 77.3% Change in expenditure share (percentage Deciles points) 1 0.6% 1.7% -0.1% 2.2% 2 0.6% 1.6% -0.7% 1.6% 3 0.1% 1.2% 0.3% 1.6% 4 0.6% -0.5% 0.8% 0.9% 5 -0.1% 0.3% -0.5% -0.4% 6 -1.0% 0.0% -0.5% -1.4% 7 -0.6% 0.3% -0.4% -0.8% 8 0.8% -2.3% 2.7% 1.2% 9 -1.4% 0.1% -0.7% -2.0% 10 0.3% -2.3% -1.0% -3.0% Urban -2.5% -1.1% 0.6% -3.1% Rural 2.5% 1.1% -0.6% 3.1% Deciles Distribution of additional spending 1 9.7% 17.6% 8.9% 12.4% 2 10.1% 17.9% 0.9% 11.3% 3 9.3% 15.8% 14.4% 12.5% 4 11.3% 6.4% 19.8% 10.9% 5 9.5% 11.8% 3.0% 9.2% 6 7.1% 9.9% 3.2% 7.5% 7 9.0% 12.4% 5.4% 9.6% 8 14.2% -1.6% 46.2% 13.9% 9 7.8% 12.0% 2.5% 8.4% 10 12.1% -2.2% -4.2% 4.1% Urban 71.1% 71.6% 85.1% 73.7% Rural 28.9% 28.4% 14.9% 26.3% Data: ENIGH 1992, 1996, 2000, 2002, table 2.2.3. Population deciles ordered by per capita expenditure. - 188 - Figure A2.5 Evolution of the Distribution of Public Education Expenditure: Rural Population 45% Share of Rural Sector 1992 40% 1994 1996 35% 1998 2000 30% 2002 25% 20% 15% 10% 5% 0% Total Preschool Primary Lower Sec Edu Upper Sec Edu Tertiary Edu Private expenditure Data: ENIGH 1992, 1996, 2000, 2002, table 2.2.3. Population deciles ordered by per capita expenditure. - 189 - Table A2.6 Male and Female Enrollment in Public Education (million students) 0.45 Preschool 0.40 1992-2002 0.35 1992 0.30 0.25 0.20 0.15 0.10 0.05 0.00 -0.05 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 Boys Girls 1.10 Primary 1992-2002 1992 0.90 0.70 0.50 0.30 0.10 -0.10 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 Boys Girls 0.35 1992-2002 Lower Secondary 1992 0.30 0.25 0.20 0.15 0.10 0.05 0.00 -0.05 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 Boys Girls Data: ENIGH 1992, 2002. Population deciles ordered by per capita expenditure. - 190 - Table A2.7 Male and Female Enrollment in Public Education (million students) 0.25 Upper Secondary 1992-2002 1992 0.20 0.15 0.10 0.05 0.00 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 Boys Girls Tertiary 0.24 1992-2002 1992 0.19 0.14 0.09 0.04 -0.01 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 Boys Girls Data: ENIGH 1992, 2002. Population deciles ordered by per capita expenditure. - 191 - Figure A2.6 Distribution of Public Health Coverage 18% SSA 16% 14% 1996 2000 2002 12% 10% 8% 6% 4% 2% 0% 1 2 3 4 5 6 7 8 9 10 Population Deciles 18% IMSS 16% 1996 14% 2000 2002 12% 10% 8% 6% 4% 2% 0% 1 2 3 4 5 6 7 8 9 Population Deciles - 192 - 35% ISSSTE 30% 1996 2000 25% 2002 20% 15% 10% 5% 0% 1 2 3 4 5 6 7 8 9 10 Population Deciles Data: ENIGH 1996, 2000, 2002. Population deciles ordered by per capita expenditure. - 193 - Table A2.5 Distribution of Public Expenditure on Health Services: 2002 By type of attention By Institution Total (All Public Expenditure) Uninsured Insured Federal All Public Expenditure Contribution Primary Maternal Hospital SSA IMSS ISSSTE PEMEX to IMSS URBAN 1 6.6% 8.2% 7.3% 4.8% 1.1% 17.4% 3.9% 0.6% 2 6.8% 7.8% 6.9% 9.4% 3.1% 13.5% 5.2% 4.2% 3 7.8% 7.7% 8.3% 7.4% 5.8% 10.4% 8.0% 4.8% 4 9.2% 8.8% 9.4% 9.4% 15.7% 10.9% 9.7% 6.4% 5 11.1% 10.8% 11.1% 13.4% 16.0% 12.8% 11.7% 8.9% 6 9.5% 9.2% 9.8% 12.1% 8.4% 10.9% 9.9% 7.9% 7 12.7% 12.0% 12.9% 21.5% 7.5% 7.3% 13.8% 21.4% 8 10.9% 9.5% 11.6% 7.3% 13.9% 6.8% 13.3% 11.3% 9 13.1% 11.6% 14.1% 10.7% 9.3% 7.4% 15.5% 17.0% 10 8.2% 7.7% 8.6% 4.0% 19.2% 2.6% 9.1% 17.7% CC 0.038 -0.033 0.088 0.036 0.213 -0.207 0.156 0.320 RURAL 1 4.7% 4.9% 4.6% 9.3% 0.6% 5.5% 3.9% 0.0% 2 7.5% 8.5% 8.2% 3.2% 1.2% 10.0% 3.7% 0.0% 3 7.0% 7.4% 7.3% 5.9% 5.1% 8.3% 5.3% 0.2% 4 7.8% 8.5% 8.1% 8.7% 0.0% 9.6% 5.1% 3.9% 5 9.0% 9.3% 9.4% 6.7% 5.3% 10.3% 7.5% 1.0% 6 10.9% 11.6% 11.4% 12.2% 4.5% 12.8% 8.5% 4.8% 7 11.0% 11.7% 10.7% 17.2% 21.2% 13.1% 8.2% 0.0% 8 12.9% 12.4% 13.2% 8.8% 21.0% 12.8% 14.9% 0.1% 9 12.3% 10.1% 12.6% 12.1% 15.0% 8.3% 21.1% 15.3% 10 14.5% 12.6% 14.4% 15.9% 26.0% 9.3% 21.9% 74.6% CC 0.136 0.084 0.158 0.167 0.467 0.058 0.342 0.770 Source: ENIGH 2002, table 2.2.1. Population deciles ordered by per capita expenditure. - 194 - Table A2.6 Distribution of Marginal Benefits from Growth in Public Health Expenditure: 1996-2002 Total SSA IMSS ISSSTE* Change in Expenditure Share (percentage points gained) 1 4.1% 8.4% 1.3% -0.4% 2 3.0% 3.8% 1.2% 1.0% 3 2.3% 1.0% 0.7% 1.1% 4 -0.2% -3.7% -0.1% 1.7% 5 1.7% 0.0% 1.7% 2.5% 6 -0.2% -2.3% 0.2% 2.8% 7 -2.0% -2.4% -1.8% 2.6% 8 -1.2% -0.8% -1.1% 5.4% 9 -2.4% -0.8% -0.4% -5.4% 10 -5.1% -3.0% -1.7% -11.3% Urban -8.4% -7.6% -3.8% 1.6% Rural 8.4% 7.6% 3.8% -1.6% Distribution of Additional Public Expenditure 1 17.8% 23.0% 9.3% 2 14.8% 17.4% 9.4% 3 15.0% 17.7% 9.3% 4 8.1% 7.5% 7.5% 5 14.4% 11.1% 17.8% 6 10.4% 7.0% 13.6% 7 6.2% 5.5% 4.5% 8 9.3% 5.7% 10.0% 9 6.5% 5.6% 13.2% 10 -2.4% -0.5% 5.4% Urban 50.9% 41.3% 65.0% Rural 49.1% 58.7% 35.0% Data: ENIGH 1992, 1996, 2000, 2002; table 2.2.5. Population deciles ordered by per capita expenditure. *ISSSTE health expenditure did not grow in real terms - 195 - Figure A2.7 Evolution of the Distribution of Public Health Expenditure: Rural Population 60% Share of Rural Sector 1996 50% 1998 2000 40% 2002 30% 20% 10% 0% Total SSA IMSS ISSSTE Source: ENIGH 1996, 1998, 2000, 2002, table 2.2.5. Population deciles ordered by per capita expenditure. - 196 - Table A2.7 Distribution of beneficiaries of public expenditure on pensions: 2002 IMSS ISSSTE Active Active Workers Pensioners Pensioners Workers Pensioned Pensioned right Decile Total AFORE Right holder Right holder right holder holder* URBAN 1 2.0% 1.5% 2.9% 3.0% 0.5% 0.0% 2 4.7% 4.4% 6.4% 5.1% 2.5% 3.8% 3 6.1% 5.8% 7.6% 8.2% 4.2% 0.8% 4 6.7% 7.9% 10.2% 8.3% 4.7% 1.6% 5 8.6% 8.8% 9.8% 10.0% 7.0% 4.7% 6 9.9% 10.0% 10.3% 10.1% 9.6% 9.5% 7 11.3% 12.7% 12.0% 11.7% 12.5% 9.2% 8 14.8% 13.9% 13.0% 14.9% 11.3% 15.2% 9 17.1% 16.4% 12.8% 12.5% 25.6% 29.4% 10 18.9% 18.6% 15.0% 16.3% 22.2% 25.9% CC 0.298 0.294 0.187 0.216 0.419 0.512 RURAL 1 0.1% 0.3% 0.4% 0.0% 0.0% 2 0.3% 1.1% 1.6% 0.0% 0.0% 3 3.5% 2.1% 2.3% 4.5% 1.8% 4 4.7% 2.2% 5.4% 6.2% 2.4% 5 3.2% 4.8% 7.1% 3.0% 1.1% 6 16.0% 6.5% 11.5% 16.9% 7.0% 7 10.6% 6.5% 10.2% 8.7% 5.0% 8 16.5% 10.5% 16.9% 18.1% 5.2% 9 19.5% 22.0% 21.3% 21.3% 15.1% 10 25.7% 43.8% 23.3% 21.3% 62.5% CC 0.461 0.595 0.436 0.430 0.698 Source: ENIGH 2002, table 2.2.1. Population deciles ordered by per capita expenditure. *Note: Number of observations of ISSSTE pensioners in the rural sector is too few to obtain representative distribution. - 197 - Figure A2.8 Distribution of IMSS and ISSSTE Right holders 1,600 1992 IMSS 1,400 2000 2002 1,200 1,000 800 600 400 200 0 1 2 3 4 5 6 7 8 9 10 700 ISSSTE 1992 600 2000 2002 500 400 300 200 100 0 1 2 3 4 5 6 7 8 9 10 Source: ENIGH 1992, 2000, 2002. Population deciles ordered by per capita expenditure. - 198 - Table A2.8a Distribution of Marginal Benefits from Growth in Public Expenditure on Pensions: 1996/7-2002 IMSS* ISSSTE Change in Expenditure Share (percentage points gained) 1 0.2% -0.1% 2 0.2% -0.2% 3 1.6% 0.5% 4 0.0% 0.4% 5 2.2% -3.9% 6 -0.9% 1.7% 7 -1.2% -0.4% 8 -0.5% -4.5% 9 -0.9% 5.1% 10 -0.8% 1.4% Urban 0.5% -3.9% Rural -0.5% 3.9% Distribution of Additional Public Expenditure 1 1.1% 0.2% 2 3.4% 1.2% 3 7.3% 2.1% 4 7.8% 4.9% 5 12.6% 3.7% 6 10.7% 8.3% 7 11.8% 11.4% 8 13.8% 13.5% 9 14.8% 25.1% 10 16.8% 29.6% Urban 92.4% 90.3% Rural 7.6% 9.7% *1997-2002 Source: ENIGH 1996, 2002, table 2.2.1. Population deciles ordered by per capita expenditure. - 199 - Table A2.8b Distribution of Marginal Benefits: IMSS Pensions 1992- 1996-2000 2000-2002 1996-2002 1996 Additional Thousand 645 1,270 312 1,582 Coverage % 10.6% 18.9% 3.9% 23.6% Additional Million 23,413* 7,561 30,974 Spending % 111.4% 17.0% 147.4% Deciles Change in expenditure share (percentage points) 1 -0.3% 0.1% 0.1% 0.2% 2 0.4% -0.6% 0.7% 0.2% 3 -0.7% 0.9% 0.7% 1.6% 4 0.1% -0.9% 0.9% 0.0% 5 -0.6% 1.0% 1.2% 2.2% 6 0.3% -1.1% 0.2% -0.9% 7 1.2% -0.2% -1.0% -1.2% 8 -0.5% 0.4% -0.9% -0.5% 9 -0.7% 1.0% -1.9% -0.9% 10 0.8% -0.6% -0.1% -0.8% Urban 1.1% 1.0% -0.5% 0.5% Rural -1.1% -1.0% 0.5% -0.5% Deciles Distribution of additional spending* 1 0.9% 1.7% 1.1% 2 2.0% 7.7% 3.4% 3 6.3% 10.3% 7.3% 4 6.1% 13.2% 7.8% 5 10.7% 18.4% 12.6% 6 10.1% 12.5% 10.7% 7 13.5% 6.7% 11.8% 8 15.4% 8.8% 13.8% 9 18.3% 4.2% 14.8% 10 16.8% 16.5% 16.8% Urban 93.4% 89.4% 92.4% Rural 6.6% 10.6% 7.6% *1997-2000 Data: ENIGH 1992, 1996, 2000, 2002; table 2.2.8. Population deciles ordered by per capita expenditure. - 200 - Table A2.8c Distribution of Marginal Benefits: ISSSTE Pensions 1992-1996 1996-2000 2000-2002 1996-2002 Additional Thousand 236 225 68 529 Coverage % 15.3% 12.7% 3.4% 34.4% Additional Million 2,261 4,550 4,540 11,351 Spending % 66.9% 80.7% 44.6% 336.1% 1 0.0% -0.2% 0.1% -0.1% 2 -0.4% -0.6% 0.9% -0.2% 3 0.4% 0.3% -0.2% 0.5% 4 1.0% -2.0% 1.4% 0.4% 5 -4.1% 1.9% -1.8% -3.9% 6 0.5% 2.4% -1.3% 1.7% 7 0.6% -1.1% 0.1% -0.4% 8 -3.8% 1.6% -2.3% -4.5% 9 4.5% -4.1% 4.7% 5.1% 10 1.2% 1.9% -1.7% 1.4% Urban -4.3% 2.1% -1.7% -3.9% Rural 4.3% -2.1% 1.7% 3.9% 1 0.3% -0.2% 0.5% 0.2% 2 0.5% -0.4% 3.2% 1.2% 3 2.5% 2.5% 1.4% 2.1% 4 6.9% 0.8% 8.0% 4.9% 5 -1.3% 9.1% 0.9% 3.7% 6 7.5% 12.1% 5.0% 8.3% 7 13.4% 10.1% 11.8% 11.4% 8 10.0% 19.1% 9.8% 13.5% 9 29.6% 13.7% 34.2% 25.1% 10 30.7% 33.3% 25.3% 29.6% Urban 84.6% 95.8% 87.5% 90.3% Rural 15.4% 4.2% 12.5% 9.7% ENIGH 1992, 1996, 2000, 2002; table 2.2.9. Population deciles ordered by per capita expenditure - 201 - Figure 2.8d: Distribution of social security right holders in public sector -ISSSTE and Armed Forces (2002) 35% ISSSTE 30% PEMEX, SECMAR, SEDENA 25% 20% 15% 10% 5% 0% 1 2 3 4 5 6 7 8 9 10 Source: ENIGH 2002; World Bank calculations. Table A2.9 Distribution of Residential Electricity Users (Households) % of National % of Decile 1992 2002 1992 2002 1 69% 84% 7.5% 8.7% 2 82% 95% 8.9% 9.8% 3 88% 96% 9.6% 10.0% 4 92% 96% 10.0% 9.9% 5 95% 99% 10.4% 10.2% 6 96% 99% 10.5% 10.3% 7 98% 97% 10.7% 10.1% 8 98% 100% 10.7% 10.3% 9 99% 100% 10.7% 10.4% 10 100% 100% 10.9% 10.3% Average 91.6% 96.5% Urban 80.5% 80.2% Rural 19.5% 19.8% CC 0.051 0.021 ENIGH 1992, 2002. Population deciles ordered by per capita expenditure. - 202 - Figure A2.9 Residential electricity Tariff 1 (MxP per kWh; base=2002) 2 1-75 kW/h 76-125 125-140 140-200 200-500 500+ 1.8 1.6 1.4 1.2 1 0.8 0.6 0.4 0.2 0 Low (<140) Middle (140-250) High (>250-2000) 1996 1998 2000 2002 Source: www.cfe.gob.mx. Table A2.10 Monthly electricity expenditure per household (Base=2002): ENIGH 1992 2000 2002 1992-2000 2000-2002 1992-2002 1 84 73 87 -13.7% 19.4% 3.0% 2 110 104 137 -5.5% 32.1% 24.8% 3 124 132 171 6.9% 29.5% 38.4% 4 143 158 189 10.5% 19.8% 32.3% 5 181 186 200 2.6% 7.6% 10.4% 6 195 192 232 -1.7% 20.9% 18.9% 7 196 199 265 1.4% 33.1% 35.0% 8 209 221 278 5.7% 25.8% 33.0% 9 325 322 329 -1.0% 2.3% 1.2% 10 596 389 579 -34.7% 48.6% -2.9% Average 194 179 224 -8% 25% 15% ENIGH 1992-2002. Population deciles ordered by per capita expenditure. - 203 - Table A2.11 Distribution of Household Expenditure on Electricity 1992 2000 2002 1 2.4% 2.8% 2.9% 2 4.1% 4.2% 4.6% 3 5.0% 6.1% 6.4% 4 6.1% 7.4% 7.2% 5 7.9% 8.3% 7.9% 6 8.9% 9.3% 9.6% 7 9.2% 10.1% 10.3% 8 9.7% 11.7% 11.6% 9 15.1% 16.9% 13.9% 10 31.8% 23.3% 25.8% CC 0.376 0.311 0.308 ENIGH 1992, 2000, 2002. Population deciles ordered by per capita expenditure. Table A2.12 Distribution of households with access to piped water in the house 1992 2000 2002 Coverage 10.2 14.2 15.1 (million) 1 3.5% 1.6% 2.0% 2 5.4% 4.1% 4.5% 3 6.4% 6.6% 6.8% 4 7.5% 8.0% 7.6% 5 9.7% 9.9% 9.9% 6 10.4% 11.7% 11.4% 7 11.8% 12.6% 13.0% 8 13.3% 14.0% 13.9% 9 14.9% 15.1% 14.9% 10 16.9% 16.4% 16.0% CC 0.235 0.263 0.252 ENIGH 1992, 2000, 2002. Population deciles ordered by per capita expenditure. - 204 - Table A2.13 Distribution of Household Expenditure on Water 1992 2000 2002 Distribution 1 1.6% 2.0% 2.0% 2 3.7% 4.1% 4.3% 3 4.7% 5.7% 6.1% 4 6.3% 5.5% 6.7% 5 9.4% 7.9% 8.9% 6 8.8% 9.7% 11.2% 7 11.1% 9.9% 12.1% 8 14.5% 10.9% 13.5% 9 14.1% 12.6% 13.4% 10 25.8% 31.6% 21.9% CC 0.353 0.367 0.297 Monthly Expenditure per Household 1 44 27 42 2 55 43 50 3 63 49 59 4 82 55 67 5 100 71 78 6 100 78 102 7 114 81 103 8 140 84 108 9 138 95 110 10 234 204 169 Average 118 86 95 ENIGH 1992, 2000, 2002. Population deciles ordered by per capita expenditure. - 205 - ANNEX 3 Table A3.1 Taxes Imputed to Households: 2002 Million pesos % del GDP % Total 687,380 11.18 100 ISR P. Físicas 152,479 2.48 22.2 Social Security 163,887 2.67 23.9 IVA 218,261 3.55 31.8 IEPS 136,493 2.22 19.9 Tenencia 11,704 0.19 1.7 ISAN 4,557 0.07 0.6 Source: Cuenta de la Hacienda Pública Federal 2000, 2001 y 2002. Social security contributions (from employers and employees): Poder Ejecutivo Federal (2003). Table A3.2 Distribution and Incidence of Taxes: 2002 Total Personal Special Taxes Social VAT Decile Excl. Social Income Tax (IEPS, Tenencia, Security (IVA) Security (ISR) ISAN) Distribution 1 -0.1 0.2 1.3 -2.0 1.0 0.3 2 0.9 1.1 1.8 -1.7 2.3 1.5 3 1.7 1.8 2.4 -1.4 3.2 2.5 4 2.5 2.7 3.4 -0.9 4.4 3.2 5 3.6 3.9 4.8 0.1 5.5 4.4 6 5.0 5.3 6.3 1.1 6.8 6.4 7 7.1 7.6 8.9 3.0 9.0 8.6 8 11.1 11.3 12.1 6.4 12.5 13.8 9 17.7 18.0 19.1 14.4 17.6 21.0 10 50.5 47.9 39.9 81.0 37.7 38.2 CC 0.635 0.611 0.535 0.911 0.499 0.537 Incidence 1 -0.4 1.7 2.02 -2.98 2.13 0.48 2 2.9 4.7 1.87 -1.62 3.10 1.39 3 4.2 6.2 1.93 -1.00 3.41 1.83 4 5.3 7.5 2.23 -0.52 3.81 1.97 5 6.7 9.5 2.76 0.05 4.24 2.40 6 7.6 10.7 3.02 0.47 4.33 2.84 7 8.8 12.2 3.40 1.08 4.60 3.08 8 11.6 15.6 3.96 1.94 5.47 4.22 9 13.9 18.6 4.70 3.29 5.76 4.82 10 19.4 24.2 4.79 9.06 6.03 4.28 Distributions of individual taxes were obtained from the Sub-Secretaría de Ingresos, SHCP. The total distribution is based on the relative weights obtained from tax revenues collected as reported in table A3.1. Source: ENIGH 2002. Population deciles ordered by per capita expenditure. - 206 - Fig. A3.0 Total federal transfers 500,000 450,000 400,000 350,000 300,000 Million 2002 MxP 250,000 200,000 150,000 100,000 50,000 0 1998 1999 2000 2001 2002 FEDERAL REVENUE SHARES RAMO 33 DECENTRALIZATION AGREEMENTS PAFEF Source: SHCP data and World Bank calculations Figure A3.1. Geographic distribution of federal public investment, 1990-2002 16000 14000 12000 10000 2002 MxP per capita 90-95 8000 96-02 6000 4000 2000 0 Chihuahua Guanajuato Aguascalientes Yucatán Baja California Sur Tlaxcala Coahuila Tamaulipas Puebla Durango Quintana Roo Baja California Nuevo León Veracruz Distrito Federal Guerrero Oaxaca Campeche Querétaro National Avg. Chiapas Sinaloa Zacatecas San Luis Potosí Michoacán Nayarit Hidalgo Colima Jalisco México Sonora Tabasco Morelos Source: Annexes of the Presidential State of the Union data and World Bank calculations. - 207 - . 2002 MxP per capita 2002 MxP per capita 0 500 1,000 1,500 2,000 2,500 3,000 3,500 4,000 0 500 1,000 1,500 2,000 2,500 3,000 3,500 4,000 calculations. Chiapas Chiapas Guerrero Guerrero Oaxaca Oaxaca Veracruz Veracruz Hidalgo Hidalgo San Luis Potosí San Luis Potosí Puebla Puebla Campeche Campeche Tabasco Tabasco Michoacán Michoacán Yucatán Yucatán Zacatecas Zacatecas Guanajuato Guanajuato Nayarit Nayarit Sinaloa Sinaloa Querétaro Querétaro - 208 - Durango Durango Tlaxcala Tlaxcala Morelos Morelos Quintana Roo Quintana Roo México México Colima Colima Tamaulipas Tamaulipas Sonora Sonora Jalisco Jalisco Chihuahua Chihuahua Baja California Sur Baja California Sur Aguascalientes Aguascalientes Coahuila Coahuila Baja California Baja California Source: SEP (Dirección General de Planeación, Programación y Presupuesto) data and World Bank Nuevo León Nuevo León Fig. A3.2. Geographic distribution of spending in education --Ramo 11--, 2002 Fig. A3.3 Geographic distribution of spending in education – R33, R25, 2002 Distrito Federal Distrito Federal National Avg. National Avg. 2002 MxP per insured population 2002 MxP per uninsured population 0 500 1000 1500 2000 2500 3000 3500 4000 4500 - Chiapas 500.00 1,000.00 1,500.00 2,000.00 2,500.00 3,000.00 Guerrero Chiapas Oaxaca Guerrero Veracruz Oaxaca Hidalgo Veracruz San Luis Potosí Hidalgo San Luis Potosí Puebla Puebla Campeche Campeche Tabasco R-12 Tabasco Michoacán Michoacán Yucatán PAC Yucatán Zacatecas Zacatecas Guanajuato Guanajuato Nayarit Nayarit Sinaloa IMSS Sinaloa Querétaro OPORTUNIDADES Querétaro Durango - 209 - ISSSTE Durango Tlaxcala Tlaxcala Morelos Morelos PEMEX IMSS-OPORT Quintana Roo Quintana Roo México INI México Colima Colima Tamaulipas FASSA Tamaulipas Sonora Sonora Jalisco Jalisco Chihuahua Chihuahua APORT-EST Baja California Sur Baja California Sur Aguascalientes Aguascalientes Coahuila Coahuila Baja California Baja California Source: SSA (Coordinación General de Planeación Estratégica) data and World Bank calculations. Nuevo León Nuevo León Source: SSA (Coordinación General de Planeación Estratégica) data and World Bank calculations. Distrito Federal Distrito Federal All States and the Federal District All States and the Federal District Fig. A3.4. Geographic distribution of health spending for the insured population, 2002 Fig. A3.5. Geographic distribution of health spending for the uninsured population, 2002 2002 MxP per capita 2002 MxP per rural population 0 50 100 150 200 250 300 350 400 450 APOYOS 0 500 1,000 1,500 2,000 2,500 3,000 3,500 4,000 4,500 5,000 PROCEDE Chiapas Chiapas Guerrero Guerrero Oaxaca Oaxaca Veracruz Veracruz Hidalgo Hidalgo PROCAMPO PRODEPLAN San Luis Potosí San Luis Potosí Puebla Puebla Campeche Campeche Tabasco Tabasco ALIANZA Michoacán PRODEFOR Michoacán SHCP data and World Bank calculations. Yucatán Yucatán Zacatecas Zacatecas Guanajuato Guanajuato Nayarit FAPPA Nayarit Sinaloa INCENDIOS Sinaloa Queretaro SAGARPA and SHCP data and World Bank calculations - 210 - Querétaro Durango Durango Tlaxcala Tlaxcala Morelos Morelos Quintana Roo PROMUSAG Quintana Roo México México Colima INF. HIDROAGRICOLA Colima Tamaulipas for agriculture per rural population, 2002 Tamaulipas Sonora Sonora Jalisco Fig. A3.6. Geographic distribution of public spending Jalisco OBLIGACIONES Chihuahua INF. HIDRAULICA Chihuahua Baja California Sur Baja California Sur Aguascalientes PET Coahuila Aguascalientes Fig. A3.7. Geographic distribution of public spending in roads, 1995-2002 PIASRE Baja California Coahuila Nuevo León Baja California National Avg. Nuevo León 2002 MxP per capita 0 20 40 60 80 100 120 Chiapas Guerrero Oaxaca Veracruz Hidalgo San Luis Potosí Puebla Campeche Tabasco Michoacán CNA data and World Bank calculations. Yucatán Zacatecas Guanajuato Nayarit Sinaloa Querétaro - 211 - Durango Tlaxcala Morelos Quintana Roo México Colima Tamaulipas Sonora Jalisco Chihuahua Baja California Sur Aguascalientes Coahuila Baja California Nuevo León Fig. A3.8. Geographic distribution of CNA spending in water infrastructure, 2002 Distrito Federal National Avg.