Report No 19066-PE Peru Education at a Crossroads Challenges and Opportunities for the 21 st Century (In Two Volumes) Volume I Main Report December 30, 1999 Human Development Department Bolivia, Paraguay and Peru Country Management Unit Latin America and the Caribbean Region uetoteWm Doument o beWrld Bank ACRONYMS AND ABBREVIATIONS ADE Areas de Desarrollo Educativo (Education Development Areas) AE Area de Ejecuci6n (Area of Execution) AFP Administraci6n de Fondo de Pensiones APP Authorized Pensionable Position CIAS Comite Interministerial de Asuntos Sociales (Inter-ministerial Committee of Social Affairs) CORDELICA Corporaci6n de Desarrollo de Lima y Callao CTAR Consejo Transitorio de Administraci6n Regional (Transitional Council of Regional Administration) DRE Director Regional de Educaci6n FONAVI Fondo Nacional de Vivienda (National Housing Fund) FONCODES Fondo Nacional de Compensaci6n y Desarrollo Social (Social Fund) GRADE Grupo de Analisis para el Desarrollo INEI Instituto Nacional de Estadistica e Informatica INFES Infraestructura Nacional para Educaci6n y Salud (National Infrastructure for Education and Health) IPSS Instituto Peruano de Seguro Social (Peruvian Institute of Social Security) IST Institutos Superiores Tecnicos (Higher Technical Institutes) ISP Institutos Superiores Pedag6gicos (Higher Institutes of Pedagogy) MECEP Proyecto para Mejoramiento de la Calidad de la Educaci6n Primaria MED Ministerio de Educaci6n (Ministry of Education) MEF Ministerio de Economia y Finanzas (Ministry of Economy and Finance) MINSA Ministerio de Salud (Ministry of Health) OECD Organization for Economic Cooperation and Development ONP Oficina de Normalizaci6n Previsional (Pension Office) PLANMED Planning Unit in MED PROMUDEH Ministerio de Promoci6n de la Mujer y del Desarollo Humano (Ministry for the Promotion of Women and Human Development) PRES Ministerio de la Presidencia (Ministry of the Presidency) USE Unidades de Servicios Educativos (Educational Service Units) UNESCO United Nations Educational, Scientific and Cultural Organization Exchange Rates (1997): Soles 2.66 = US$1 Fiscal Year: January 1 to December 31 School Year: April I to December 31 (180 days/year) Vice Presidents Shahid Javed Burki (through June 30, 1999) David de Ferranti (from July 1, 1999) Country Director Isabel Guerrero Lead Economist Emesto May Sector Director Xavier Coll Education Sector Manager Jamil Salmi Lead Specialist in Human Donald Winkler Development Country Sector Leader Evangeline Javier Task Team Leader Kin Bing Wu Peru Education at a Crossroads: Challenges and Opportunities for the 21st Century Volume 1: Main Report PREFACE Peruvian government policy in the 1990s attained macroeconomic and political stability. These accomplishments were particularly noteworthy given the problems of macroeconomic imbalances, hyper-inflation, and civil unrest of the late 1980s. These achievements created the preconditions for the success of long-term policies that can both sustain economic growth and address problems of persistent poverty. The Government of Peru (GOP) has made poverty alleviation one of the cornerstones of its policy. In 1997, about 49 percent of Peru's population of 25 million lived in poverty and 15 percent in extreme poverty. The World Bank's assistance program for Peru emphasizes support for the GOP's poverty reduction strategies and projects. In order to guide its own work, to deepen its dialogue with government, and to inform public discourse, the Bank initiated a coordinated set of studies bearing on poverty reduction. The studies covered four topics: poverty itself, health, indigenous peoples and education. This report conveys the findings of the education study. Almost all Peruvian children enroll in primary education and opportunities for secondary and higher education well exceed what would be expected for a country of Peru's income level. The GOP views investment in education as essential for social cohesion, for personal and moral development, and for improving individual economic productivity and employment prospects. Educational improvements thus underpin both poverty reduction and long-term economic and social development. Despite achievements to date, significant gaps remain -- between the rich and poor, between the rural and urban areas, and between the indigenous and nonindigenous populations -- in school completion rates and learning outcomes. Overall challenges also remain for upgrading quality at all levels and for extending coverage of early childhood, secondary and higher education. Peruvian education has thus reached a crossroads. The current situation reflects important accomplishments. One path into the 21" Century would pursue steady incremental improvements of the current situation. Another path for policy would aim for a marked improvement in intellectual and technical capacity of the population through a focussed commitment to closing gaps and meeting remaining challenges. This report seeks to inform discussion of these options by examining the impact of public and private finance, and policies for their use, on education and labor market outcomes. As such, it is one contribution to a larger discussion of human resource development and poverty reduction. This World Bank document reports the work not only of its own staff but, also, to an unusual extent, that of Peruvian academics, policy analysts, and public officials. The document reflects a collective effort. IT TABLE OF CONTENTS VOLUME 1: MAIN REPORT Task Team and Reviewers ....................................... vii Ackinowledgments ....................................... lx Executive Summary ....................................... xi 1. Sector Overview.1 1.1. Achievements.1 1.2. The Evolving Education System ........................................3 2. Education Finance..11 2.1. The Budget Process.11 2.2. Public Expenditure on Education .13 2.3. Household Expenditure on Education ....................................... 24 3. System Performance Indicators ........................................ 29 3.1. Access, Repetition, and Retention ....................................... 29 3.2. Labor Market Outcomes ........................................ 34 3.3. Learning Outcomes ....................................... 37 4. The Teaching Profession ....................................... . 47 4.1. Teacher Qualifications and Employment Status ....................................... 47 4.2. Conditions of Service and Compensation ....................................... 51 4.3. Incentives and Accountability ....................................... 56 5. Second-Generation Reform ....................................... 61 5.1. Improve Equity ....................................... 61 5.2. Enhance Quality ............................... 64 5.3. Improve Efficiency of Resource Use ............................... 65 5.4. Conclusion ............................... 68 References ............................... 75 111 VOLUME II: BACKGROUND NOTES AND APPENDICES Background Notes ........................................l. 1. The Structure of Education .1 2. Income Elasticity of Demand for Education and Engel's Curve .3 3. Private and Social Returns to Public Education in Urban Peru .7 4. Determinants of Achievement .13 5. Teacher Education and Professional Development .41 Appendices ....... 51 1. Student Enrollment Statistics .51 2. Teacher Statistics .59 3. School Statistics .75 4. Indicators of Equity and Efficiency .83 5. International Comparison of Between-School Variation in Achievement . 105 6. Public Expenditure on Education.1. II 7. Household Expenditure on Education .157 8. Population Projection .173 9. External Support for Education Since 1990 .179 10. Selected Indicators for International Comparison ..................................... 189 iv LIST OF TABLES IN THE MAIN REPORT 1. A Comparison of the Existing and New Education Structures ........................................6 2. Urban Peru: Rates of Return to Public Education, 1997 ............................................... 35 3. Index of Fourth Grade Mathematics Outcomes, 1996 ................................................... 38 4. Summary of Effects Crossing between Departments, Schools and Students ............ 43 5. First Generation Reform and Its Implications for Second Generation Reform ......................................................... 70 6. Summary of Policy Options ......................................................... 72 LIST OF FIGURES IN THE MAIN REPORT 1. International Comparison of Enrollment of Students between the Ages of 3 and 23 ..........................................................2 2. International Comparison of Public Expenditure on Education as a Percentage of Gross Domestic Product ..........................................................2 3. Trend of Enrollment in Public Institutions by Level, 1990 to 1997 ................................7 4. Trend of Enrollment in Private Institutions by Level, 1990 to 1997 ...............................7 5. Ministry of Education Operational Structure ......................................................... 10 6. Public Expenditure on Education and Central Government Expenditure as a Percentage of Gross Domestic Product Expenditure, 1970 to 1997 ................ 14 7. Percentage Change of Gross Domestic Product, Central Government Expenditure, and Public Education Expenditure, 1970 to 1997 ............................. 14 8. Public Spending on Education and GNP Per Capita in Lower Middle Income Countries ......................................................... 15 9. Total Recurrent and Capital Expenditures on Education 1970 to 1997 (in constant 1997 soles) ......................................................... 17 10. Composition of Public Expenditures on Education, 1990 to 1997 (Percentage, Re-grouped According to the Latest Classification) .......................... 17 11. Inter-Governmental Transfer of Resources (Percentage of Total), 1990 to 1997 ......................................................... 17 12. Public Expenditure on Education by Level, 1990 to 1997 (Percentage) ..................... 19 13. Per Student Recurrent Public Expenditure on Education by Level, 1990 to 1997 (Constant 1997 Soles) ......................................................... 19 14. Lorenz Curve for Incidence of Public Expenditure ..................................................... 22 15. Lorenz Curves for Incidence with 5 Simulations ........................................................ 22 16. Lorenz Curves by Education Level ......................................................... 22 17. Lorenz Curve for Incidence of Private Expenditures - All Levels .............................. 27 18. Lorenz Curve for Incidence of Private Expenditures - Only Public Schools .............. 27 19. Lorenz Curve for Incidence of Private Expenditures - Only Primary Schools ........... 27 20. School Survival Rates by Gender, 1997 ......................................................... 32 21. School Survival Rates by Urban and Rural Areas, 1997 ............................................. 32 22. School Survival Rates by the Mother Tongue, 1997 ................................................... 33 23. School Survival Rates by Poorest and Richest Consumption Quintiles, 1997 ............ 33 24. Evolution of Estimated Premia by Educational Level, 1985 to 1997 .......................... 34 v 25. Mathematics Outcomes and Recurrent Public Expenditure on Basic Education Per Student ....................................................... 39 26. Mathematics Outcomes and Household Expenditure on Basic Education Per Student ........................................................ 39 27. Poverty and Recurrent Public Expenditure on Basic Education Per Student .............. 40 28. Poverty and Household Expenditure on Basic Education Per Student .................. 40 29. Determninants of Effective Learning in Primary Education: Findings from Literature Review ....................................................... 44 30. Remuneration of Teachers in Real Terms, 1990 to 1997 ............................................... 54 31. Index of Remuneration of Government Employees 1970-1997 .................................. 55 32. Index of Private Sector Salaries in Metropolitan Lima, 1970-1997 (Based August 1990=100) ....................................................... 55 33. Index of Purchasing Power of Private Sector Salaries 1972, 1990, 1997 .................... 55 34. Index of Purchasing Power of Public Sector Salaries 1972, 1990, 1997 ..................... 55 35. Estimates and Projections of School Age Population, 1995-2020 .............................. 68 vi TASK TEAM AND REVIEWERS This report was prepared by: Kin Bing Wu, Team Leader With contributions from: Juan Pablo Silva and Arturo Review of Public Expenditure on Education Miranda Suhas Parandekar and Juvenal Diaz Analysis of Household Expenditure on Education Pete Goldschmidt Analysis of Academic Achievement Patricia Arregui and Sandro Analysis of School Survey in Lima/Callao and Cusco Marcone Jaime Saavedra and Eduardo Analysis of Social and Private Rates of Retum to Maruyama Education and Labor Market Outcomes Kye Woo Lee Teacher Supply and Demand and Compensation Maria Amelia Palacios Teacher Training Richard Webb Budget Process & Higher Education Livia Benavides External Support for Education Eduard Bos Projection of School-age Population Supported by: Anna Maria Mayda, Roberto De Research Assistance Vogli, Olympia Icochea, Aude Damon, Yukiko Arai and Atsuko Toi Aracelly Woodall, Douglas Task Assistance Flandro and Pierre Sandoval Leslie Evans English Editing Luisa Maria Rojas, Berta Van Spanish Translation Zuiden, Ivonne Treneman Elizabeth Mestanz Word processing in Spanish Gladys Eopez, Gladys Del Valle, Support of various kinds from the staff of the World Ana Maria Arteaga and Nancy Bank Resident Mission in Lima Escalante World Bank Reviewers: Peter Moock Marlaine Lockheed Ernesto Cuadra Fernando Reimers Jaime Saavedra vii Ministry of Education Reviewers: Juan Fernando Vega Cesar Guadalupe Jose Rodriguez Juan Pablo Silva viii ACKNOWLEDGMENTS The study is made possible by the support of the Ministry of Education (MED), Ministry of Economy and Finance (MEF), Ministry of the Presidency (PRES), and Regional Education Directorates. Special thanks are due to current and former officials of the following agencies: Ministry of Education Sr. Felipe Garcia Escudero (Ministro), Srta. Susana Seto (Vice Ministra), Srta. Esther Gago (Ex-Vice Ministra), Sr. Juan Fernando Vega, Sr. Cesar Guadalupe, Sr. Jose Rodriguez (PLANMED), Sr. Walter Twanama (Ex-Oficina de Planificaci6n Estrategica y de Medicion de la Calidad), Sr. Jorge Ferradas (MECEP), Sra. Virginia Cachay (Ex- Unidad de Seguimiento y Monitoreo), Sr. Cesar Serpa (Ex-Unidad de Presupuesto), Dra. Blanca Encinas (Ex-Direcci6n Nacional de Educaci6n Inicial y Primaria), Sr. Rolando Andrade Talledo (Direcci6n Nacional de Capacitaci6n y Formaci6n Docente), Sra. Carmen Trelles (PLANCAD), Sra. Tula Luna (Ex-Direcci6n Nacional de Educaci6n Secundaria y Superior Tecnologia), Sra. Luz Chung (Ex-Unidad de Desarrollo Curricular y Recursos Educativos de Educaci6n Secundaria), Sra. Marimoly Vela Tellez (Direcci6n de Educaci6n de Adultos), Sra. Carmen Cecilia L6pez Diaz (Oficina de Personal), Dra. Maria Belen Suazo (Oficina de Personal), Sra. Nery Escobar (Unidad de Formaci6n Docente), Sr. Juan Carlos Godenzzi (IJnidad de Educaci6n Bilingue Intercultural), Sr. Paul Roeders (GTZ-Proyecto Reforma de Formaci6n Magisterial), Sra. Carmen Salardi (Directora General de Administraci6n). Ministry of the Presidency Sr. Alejandro Arrieta (Ex-Secretario Interministerial de Asuntos Sociales), Dr. Nelson Vidal La Torre (Ex-Asesor del Despacho Ministerial), Sr. Adrian Fajardo Christen (Asesor Principal del Despacho Ministerial). Ministry of Economy and Finance Sr. Javier Abugattas (SIAF), Dra. Hedy Huarcaya (Direcci6n de Sectores Sociales, DNNPP-MEF), Sr. Carlos Pichilingue (Direcci6n de Regiones, DNPP-MEF), Sra. Neride Sotomarino (Ex-Oficina de Inversion), Sr. Rafael Capristan Miranda (Oficina de Inversiones). Direcci6n Regional de Educaci6n La Libertad: Sra. Rosa Neyra (Direcci6n Regional de Educaci6n), Sr. Estuardo Loyola Rabines (Oficina Asesoramiento Tecnico), Sra. Rene Vereau Orbegozo (Oficina de Administraci6n), Sr. Walter Rebaza Vasquez (Direcci6n Tecnica Pedag6gica), Sra. ix Regina Pacheco Ponce (Direcci6n, Aldea Infantil Santa Rosa de la Fundaci6n por los Nifios del Peru). Arequipa: Prof. Carlos Urbiola L6pez (Direcci6n Tecnica Pedag6gica), Sr. Oscar Lazo Rodriguez (Relaciones Pfblicas), Sr. Hector RodrIguez Alvarez, Sra. Luz Perez, Postigo, y Sr. Sauil Begozo Jacobo. Cusco: Sr. Juan C. Galvin (Director de Asesoramiento Tecnico). Puno: Prof. Elias Peralta Hinostroza, (Direcci6n Regional de Educaci6n), Arq. Jose Manuel Pineda Barreda, (Infraestructura Educativa), Sr. N6stor Marca Limachi, (Top6grafo), Prof. Florencio Madarriaga (Especialista en Educaci6n), Lic. Leticia Ramos Cuba (Relaciones Publicas). Loreto y Lima: Prof. Luis Veintemilla Soria (Direcci6n Regional de Educaci6n), Sra. Herman Rodriguez Flores (Area Estadistica), Prof. Bernardo Vasquez Cahuaza (Jefe de Presupuesto). Banco Central de la Reserva del Peru Sra. Maritza Guabloche Colunge (Apoderado General), Sr. Victor Hugo Diaz (Dpto. Entidades Gubernamentales), Sra. Judith Guabloche, Sra. Augusta Alfageme. Instituto Nacional de Estadistica e Informatica (INEI) Sra. Genara Rivera (Directora de Estudios Sociales), Sr. Renan Quispe (Sub Jefe de Estadistica). The World Bank team is grateful to educators and researchers in universities, research centers, and non-governmental organizations for sharing their work and perspectives. Universidades/Entidades de Investigaci6n/Organizaciones No Gubernamentales Sr. Juan Ansi6n (Pontificia Universidad Cat6lica del Peru), Sra. Carmen Montero (Instituto de Estudios Peruanos), Sr. Hugo DIaz (Instituto de Jnvestigaci6n para el Desarrollo y la Defensa Nacional), Sr. Moises Ventocilla (Instituto Cuanto), Sr. Luis Carlos Gorritti (Foro Educativo), Sra. Cecilia Thorne (Pontificia Universidad Cat6lica del Perui), Sr. Jose Martin Vegas (Foro Educativo), Sr. Santiago Cueto (GRADE - Grupo de Analisis para el Desarrollo), Sr. Julio Dagnino (Instituto de Pedagogia Popular), Sr. Eduardo Palomino (Colegio de Los Andes). x EXECUTIVE SUMMARY Peruvian education has achieved notable successes in the second half of the 20th Century. Primary education now reaches almost all children. Secondary and tertiary education institutions enroll about 80 percent of the 12- to 16-year olds and 30 percent of the 17- to 25-year olds, respectively. Few countries in Peru's income range achieve comparable coverage. These accomplishments are particularly impressive given Peru's geographical and ethnic diversity and its recently overcome problems of macroeconomic instability and civil unrest. The current government has steadily increased public expenditure on education, despite constraints imposed by disciplined fiscal policies. Nonetheless, at 3 percent1 of Gross Domestic Product (GDP) in 1997, Peru's public spending on education remained significantly less than the Latin American average of 4.5 percent. That Peru was able to attain high enrollment with a very low level of public spending results from several factors: (i) relatively equitable distribution of public resources to benefit the poorer quintiles of the consumption distribution; (ii) the ability to contain the growth of personnel expenditure, channeling the additional public resources to build up the infrastructure and capacity; (iii) mobilizing high household expenditure on education (total household spending on public and private education accounted for about 2 percent of GDP-much higher than the OECD's 1.3 percent); and (iv) having achieved near universal primary education before qualitative improvement and expansion of other levels. These factors operated in the context of what might be labeled a first generation of reform occurring in the early 1990s. This reforrn was characterized by rationalization of the public sector, regionalization of administration, deconcentration of social services, encouragement of private education, and extension of free and compulsory education. Despite these accomplishments, important challenges remain. Disparity between the rich and poor -- and between rural and urban areas -- remains pronounced in access to preschool, secondary education, and tertiary education, as well as in school completion rates. Disparity is also manifested in levels of achievement between indigenous and non- indigenous populations, and between public and private schools. Meanwhile, the earnings differentials between workers with different levels of education are growing. In urban areas, the largest increase in earnings differentials is between university educated and secondary educated workers. This signals a growing demand for higher skill levels in an open economy that is facing increasing international competition and technological change. These trends have serious implications for the employment prospects and future life-time earnings of disadvantaged groups. Peru has thus reached a crossroads for education policy. The status quo reflects major accomplishments. One direction for policy would, therefore, involve useful but In Peru, pensions are paid out of the recurrent budget of each ministry, not out of a separate pension fund as in many other countries. Pensions accounted for about 21 percent of the total public expenditure on education. Net of pension, public spending on education accounted for only 2.4 percent of the GDP in 1997. xi relatively limited improvements in the current situation. This direction, however, may prove inadequate to fulfill the government's goal of having a highly educated citizenry - fully competitive by international standards - to underpin poverty reduction and to facilitate economic and social development. A second generation of reform would be required if this direction were taken. This study explores these possibilities. It reviews evidence to guide reforn in the education process itself. In this context, it lays out options for a second wave of reforn. Either of the broad directions for policy will need to identify specific ways of improving educational quality. Experiences of other countries provide important insights. Of more direct relevance is analysis of the determinants of educational outcomes in Peru itself. An innovative assessment of fourth graders' achievement levels in 1996 generates an empirical basis for such an analysis.2 The 1996 assessment has limitations - as do those in other countries, particularly in initial efforts - and hence the results obtained should be viewed as suggestive rather than definitive. That said, the findings are encouraging in indicating potential directions where change could make a difference. This report draws preliminary conclusions based on analyses of determinants of mathematics achievement. After controlling for a number of explanatory variables, the perfornance of poor and extremely poor departments was better than non-poor departments. Some departments were doing a better job in educating over-aged students. Aymara students performed as well as Spanish-speaking students. Quechua students could achieved as much as others if they were not studying in predominantly Quechua schools, thereby indicating the potential for policy to reduce the disparity. Teachers who graduated from universities and from Institutos Superiores Pedag6gicos, teachers who have had longer years of service, and teachers who have had more in-service training courses, were positively associated with higher student achievement, relative to those teachers who have not had these qualifications, years of service and training opportunities. Non-availability of textbooks was associated with lower achievement. Parental expectations - potentially modifiable through publication of assessment results and learning determinants - helped shape outcomes. Even within the limitations of this first assessment effort, the findings are sufficiently important to warrant attention for the policy possibilities to equalize educational outcomes and improve quality more generally. Because the burden for financing education is disproportionately heavier for poorer households than richer households, public policy cannot rely on general increases in income to bring improved educational quality for disadvantaged populations. The public sector has assumed the mandate to ensure the equality of educational opportunity for all and improved economic opportunity for all - by directing more public resources to the poor. Past constraints on public expenditure allow room for new allocations to meet this mandate without exceeding reasonable overall public allocations to the sector. But 2 The 1996 assessment is not yet in the public domain. This report, therefore, provides no data on specific test results or on averages of scores of departments. To protect the underlying data, statistical analyses report only the effects of various independent variables on outcomes, rather than more comprehensive fimdings. xii increased public commitment to education will prove to be of little value unless the resources are committed not only to the right groups but also to the right interventions. To level the playing field for all Peruvian children, then, it is necessary to improve equity, quality, and efficiency. All countries face these challenges. In Peru, the policy options to meet them include the following: * Improve equity: Government support can come in the form of proven supply-side interventions. These include extending the current provision of each class with a set of instructional materials and supplies from primary education to preschool and secondary education; expanding provision of bilingual education programs and texts; stepping up recruitment and strengthening training of indigenous teachers; training rural teachers in multi-grade teaching; providing incentives to rural teachers; extending cost-effective health and nutrition programs for school-age children; and expanding access to secondary education (in part by establishing distance learning programs). Also important are demand-side financing measures (such as grants and scholarships targeted to rural areas and indigenous students, particularly girls). * Improving quality: Many of the interventions just listed to improve equity will also enhance quality. Additional options include: changing the system of incentives -- such as adjusting the salary scale to reward higher levels of skills and competency, opportunity for professional development, and open-ended tenure to be determined by performance. Setting standards for learning and teaching, strengthening teacher pre- service and in-service training, teacher performance evaluation, and rewarding schools for improved performance (perhaps through formula-based funding) are among the options to sustain the efforts to enhance quality. These measures require complementary improvement in the frequency, quality, and transparency of student assessment (with results available to administrators, principals, teachers, parents, students, and the general public); the recently introduced program of national testing provides an excellent start in this direction. Building consensus with all stakeholders on the direction and means for change will be critical to build a culture of accountability. * Improving efficiency in planning and policy: Efficiency objectives can be advanced by (i) strengthening coordination of educational policy and financial matters between the budgetary entities that have responsibility for education (Ministry of Education, the Regions, universities, decentralized institutions, and Ministry of the Presidency, as well as the Ministry of Women and Human Development); (ii) proactive gathering of information on teachers and school-level finance (in both public and private schools) to guide policy; and (iii) using student achievement data systematically to target additional resources for schools falling below certain performance levels, while recognizing schools which have made above average progress over time. Some measures are easier to implement (such as provision of educational materials) than others (such as setting standards to drive development of teacher professionalism) xiii because the latter involve institutional and cultural change. Therefore the timeframe of implementation will vary. Many countries have committed far more public resources to education than has Peru, but without achieving universal coverage for basic education. For these countries, increasingly binding fiscal constraints and continued needs to expand coverage of basic education sharply constrain the policy agenda. Peru, in contrast, has positioned itself at a crossroads. One direction to take involves continuation and improvement of the system much as it currently is. On the other hand, Peru has achieved the pre-conditions to initiate a major drive to consolidate equity gains and expand access while improving quality. Choosing this path would require, over time, substantially increased public expenditures on education. A gradual increase from 3 percent to 4.5 percent of GDP net of pensions (that is, to the Latin American average) is, for Peru, feasible given its fiscal discipline and generally improving macroeconomic performance. By increasing public expenditure levels to only the Latin American average, Peru has the opportunity to enhance markedly the intellectual ability and competitiveness of its labor force within a generation. No policy challenge is more significant. xiv Chapter 1. Sector Overview As a lower middle-income country with a Gross National Product (GNP) per capita of $2,460 in 19971, Peru has made impressive progress in extending education opportunities over the last five decades. Between 1950 and 1997, enrollment expanded 6.6 times, more than double the three-fold increase of the population.2 Total enrollment grew from a mere 14 percent of the population in 1950 to 36 percent in 1997. As a consequence, over the period, the average education level of the population of age 15 and over increased from 1.9 years to 8.6 years3, and the illiteracy rate was reduced from 58 percent to 11 percent. Female illiteracy was reduced from 70 to 18 percent, and rural illiteracy from over 60 to 29 percent. These are impressive accomplishments in a country as physically and ethnically diverse as Peru, particularly in view of its recent history of macroeconomic instability and civil unrest. Chapter I summarizes achievements to date and describes recent developments and issues in the sector. 1.1. Achievements In 1997, school enrollment was practically universal for the 6- to 11-year olds, about 80 percent of the 12- to 16-year olds, and over 30 percent of the 17- to 25-year olds.4 The education system now encompasses about 8 million children and adults, of whom 6.7 million are in the public, formal system. (See Background Note 1 for the structure of the education system, Appendices 1 to 3 for education statistics, and Appendix 4 for gross and net enrollment ratios.) International comparison shows Peruvian enrollment ratio in a remarkably favorable light, as is presented in Figure 1. Each bar in the figure above stands for a particular country in the World Bank's education database, and Peru is high up on the list, where the neighbors on the graph are mostly developed nations. When international comparison is made of the level of public spending on education as a percentage of Gross Domestic Product (GDP), however, the position of Peru changes, as is shown in Figure 2, which is also drawn from the same World Bank database. In the second graph, the neighboring countries include a number of low-income countries. I See World Bank, 1998c, World Development Report, p. 191. 2 See Diaz, Huayte, Farro, and Tavara (1995, p. 22), which cites Instutito Nacional de Estadistica e Informatica (INEI) and Ministry of Education (MED) statistics. 3 The average of 1.9 years of education corresponds to the national census of 1940. The 8.6 years of education is based on the National Survey of Living Standards by INEI (Encuesta Nacional de Niveles de Vida, or ENNIV), 1997. 4 The findings were based on analysis of a household survey conducted by Instituto Cuanto in 1997. The Cuanto dataset was more extensive in its questionnaire about education expenditure than the government's INEI dataset. Therefore, it was used for this study. Figure 1. International Comparison of Enrollment of Students Between the Ages of 3 and 23 100 Peru (Rank 15) *390 x/ X 80- ¢1 70 - c 0 = 60 11 50- -_ 00 40 -_ 30 - X 20- 102 - 0 Figure 2. International Comparison of Public Expenditure on Education as a Percentage of Gross Domestic Product 12 10 1- C - M0 m 6 - D- Peru (Rank 118) 2 4 - 0 0) Source Edstats Database of The World Bank 2 These figures provide a compressed account of achievements to date and also pose a question: what can explain the puzzle that Peru has been able to achieve an unusually high participation rate with a relatively low level of public spending on education? How has this been accomplished? Answering this question raises the main policy issues that this report addresses. Several hypotlheses for explaining the puzzle are investigated anid their policy implications are discussed: (a) Is it because public resources have been well used and well targeted? (b) Is it because Peruvian households value education highly and spend heavily on education? (c) Has expansion of access to basic education come at the expense of qualitative improvement? (d) Is the low level of public spending attributable to the ability to contain the salary cost of teachers? How does this and other policy on teachers affect the profession? The chapters in the report correspond roughly with the above questions. Chapter 1 sets the context for discussion by summarizing achievements to date and by describing recent developments in the sector. Chapter 2 reviews public expenditure on education in order to address the question of whether public resources have been used efficiently and equitably; it also examines private spending on education to assess the impact on equity. Chapter 3 analyzes indicators on access, school survival rates, quality, and labor market outcomes to evaluate the trade-off between quantitative expansion and qualitative improvement within the constraints of public and private finance. Chapter 4 reviews the policy towards teacher employment and deployment, conditions of services and compensation, incentives and accountability. Chapter 5 explores the options for improving equity, quality, and efficiency, and discusses the resource implications. 1.2. The Evolving Education System Many of the policy changes that have taken in Peru in the 1990s represent, to a considerable extent, a break with the past. The evolution of this policy environment must be viewed against the economic and political crises of the late 1980s. Fiscal deficit (which included debt servicing) was equivalent to 10 percent of GDP in 1988. Hyperinflation cumulated to a rate of over 7,000 per cent in 1990. Insurgency was rampant in the countryside and at times in the cities. The year 1990 marked a turning point. Structural adjustment under the Fujimori Administration has restored fiscal discipline. Sound macroeconomic management, in combination with the ending of the Shining Path insurgency in 1994, have gradually set the economy on a growth path. By 1997, a balanced budget (which included debt servicing) was achieved; savings in the current account progressively increased to 4.4 percent of GDP; and both GDP growth and inflation were estimated at around 7 percent each. Before the effects of the East Asian financial crisis was spread to Latin America in 1998, Peru's economic growth rate was second only to that of Chile in the region. In the 1990s, along with major reform of macroeconomic policy, a series of measures have been implemented to contain public expenditure, to mobilize private 3 resources and to delegate social services to the regions. These measures and the changes they set into motion are as follows: * Rationalization of the public sector and introduction of a private personalized pension plan Retrenclhment of government services between 1991 and 1993 resulted in a reduction in employment in the central administration of the Ministry of Education (MED) by 72 percent.5 Vacant or new positions in the Ministry are filled by consultants on contract. To contain personnel expenditure, authorized pensionable positions (APP) for the entire sector were frozen since 1995. In 1994, a n-ew personalized pension plan1 (AFP) (Law 25897), modeled after the Chilean private pension plan, was introduced.6 People on public payroll, including teachers, are given the option of choosing whether they want their pension to be covered by the previous laws, or have the individualized account which they can invest with a private company for capitalization. Since it is a personalized plan, they can take the 5 World Bank, 1994b: Peru Public Expenditure Review, Report No. 13190-PE, p. 50. 6 Before the reform, two laws governed the pensions for teachers: (I) Law 20530, which affects those teachers who joined the service before 1980, has no minimum retirement age. It allows female retirees to receive 7/25 and male retirees to receive 7/30 of their basic salaries after they have contributed 6 percent of their basic salaries for 7 years. Female retirees who have worked for 25 years and male retirees who have worked for 30 years are entitled to 100 percent of their basic salaries. Pension benefits are not only filly adjusted for inflation, but will be linked to the salary increments of current serving employees. When pensioners were still in active service, they also contributed 9 percent of their salaries to social security (Instituto Peruano de Seguro Social or IPSS) and 5 percent to a housing fund (Fondo Nacional de Vivienda or FONAVI), while the employer (that is, MED) contributed 6 percent. Pensioners have to continue to contribute 4 percent of their pension to the Treasury in order to enjoy the benefits. (2) Law 19990, which affects those teachers who joined the service after 1980, imposes a minimum retirement age of 55 for women who had completed 25 years of service, and 60 for men who had completed 30 years of service. Pension benefits were fixed and not adjusted for inflation. Active teachers have to contribute 13 percent of thleir basic salaries, 9 percent to IPSS and 5 percent to FONAVI. There is no doubt that the pensioin provided under Law 20530 was veiy generous, and provided coverage to those who might be as young as their late thirties and early forties (which are the age groups of the last cohort of ISP graduates who joined the teaching profession before 1980). In fact, Law 20530 provides a strong incentive for teachers to retire with pension and start a second career, such as teaching in or founding private schools. It was not surprising that the retrenchmlent of the early 1990s resulted in launching the second career of many enterprising persons in this sector. The incomplete system of data collection left the Government unable to calculate the total number of teaclhers and their age structure governed under these three retirement laws. This has made it difficult to project the total pension obligations of the education sector. The pension issue has been a dominant one in public expenditure on education. The reform has stopped future drain on public expenditure. 4 pension wherever they are employed.7 Salaries paid to teachers who choose the personalized pension plan are higher than those under previous laws in order to provide incentives for conversion (Appendix 6.32). Meanwhile, the Government has created a Pension Office (Oficina de Normalizacion Previsional, or ONP for short) which will eventually handle all pension matters. The new personalized pension plan is expected to lessen the burden of the state treasury anid make employment more flexible by de-linking it with specific employers. These measures to colntain public expenditure, along with other policies, helped restore fiscal balance. These, however, came at the cost of low morale in the public sector, which could ultimately undermilne high performance. Cost containment alone could not result in efficiency gain withouit concomitant use of transparent criteria for personnel recruitment alnd resource allocating, setting up of incentives to reward performance, and introduction of accountability. Towards the second half of the 1990s, there has been increased attention to these complementary needs. Competitive examinations was introduced in 1997 to select new principals and teachers to fill vacant authorized pensionable positions (APP). A national student assessment program was set up in 1996 to monitoring achievement. These measures, although remaining in an early stage of implementation, provide the building blocks towards the establishment of a merit-based system. Further policy guidelines on standards for teachers and students, strengthened pre-service and in-service training, and incentives and accountability, backed up by resources to fwmd them, would counteract the morale issue, professionalize the teaching force, and improve quality of education. * Extension of free and compulsory education in the 1993 Constitution The Constitution of 1993 extends compulsory and free education from primary to secondary education. In 1997, a proposed structural change of the education system redefines basic education by extending downward by one-year to include initial education for 5-year-old children, and by shortening secondary education from five to four years. This will make the overall duration of basic education 11 years. The plan is to progressively extend universal initial education to cover children of four years of age, and then, those of three years of age. Since secondamy education was shortened, it is compensated by the introduction of two years of preparatory course work (bachillerato) which is not compulsory but free and will provide the transition to tertiary education or to the world of work. Certificate examinations will be held at the end of basic education, bachillerato, and tertiary education (Table 1). This structural change has far reaching implications on the resource requirements to implement the policy, the supply and demand for teachers at different levels and in different subject specialties, and the content and methods of teacher in-service and pre- service training. I-low various components of the proposal can be synchronized remains to be worked out. 7 Employees wlho choose private pension lhave to contribute 11 to 13 percent of their basic salaries to future pension, and 5 percent to FONAVI, while the Government contributes 9 percent to tPSS to cover health insurance. 5 Table 1: A Comparison of the Existing and New Education Structures Age Existing Structure Age New Structure Objective of Change 3-5 Nonuniversal 5 1 year ofuniversal initial Facilitate the articulation initial education education at the age of 5 between initial and prrmary education to improve efficiency of the system 6-11 6 years of 6-1 I Duration of primary Develop the capacity of universal primary education remains learning education of unchanged 'but er-phasis is uneven quality on improving the quality 12-16 5 years of 12-15 4 years of universal Guarantee free access and tle secondary secondary education use of distance education for education that has rural areas to extend uneven access coverage. I year of preschool. between rural and 6 years of primary, plus 4 urban areas years of secondary education will form Li years of universal basic education. Certification of study at the end of basic education. 16-17 2 years of bachillerato is a Preparation for work and for new introduction It is not tertiary education. compulsory but free in public Certification of study at the schools. end of bachillerato. 17- Tertiary education 18 and Tertiary education Remains unchanged. over Souirce MED, Nueva Estructutra del Sistema Educativo Per uzano Fundamento de ki Propuesta, 1997 Note The shaded area is basic education, which is compulsory and free E Encouragement of private schools Complementary to the need to contain public spending and to the constitutional mandate for expanded compulsory edtucation is a new law that encourages establishment of private schools. This legal framework, combined with retrenchment of education administrators and teachers, has led to a rapid growth of private schools8. This supply has met the demand of parenits who have grown weary of frequent closing of public schools due to teachers' strikes in the early 1990s, who consider the quality of public education unsatisfactory, and who can afford to pay for private schools. 8 There are a number of private schools: secular schools, cooperative schools, Catholic and other religious schools, all of which are privately financed and privately run. In addition, there is Fe y Alegria, whicLi is operated by the Cathiolic Church, but financed by the State. 6 Figure 3. Trend of Enrollment in Public Institutions by Level, 1990 to 1997 4,000,000 3,500,000 3,000,000 2,500,000 ' 2,500,000 % 2,000,000 E 1,500,000 it z 1,000,000 500,000 1990 1991 1992 1993 1994 1995 1996 1997 !4-Initial *-Prima Secondary c Sup No Univ _Universiity Source Ministry of Education Figure 4. Trends of Enrollment in Private Institutions by Level, 1990 to 1997 600,000 500,000 - 400,000- 4) , 300,000 E Z 200,000 100,000i C* - 9 0 1990 1991 1992 1993 1994 1995 1996 1997 1nitiaa UPrimary *'Secon up No niv University Source: Ministry of Education 7 Between 1990 and 1997, enrollment in private education grew by 62 percent in initial education, 9 percent in primaly education, 28 percent in secondary education, and 37 percent in tertiary nonuniversity education (MED statistics). This outpaced the rate of increase at these levels in the public sector, wlich grew only by 34 percent in initial education, 8 percent in primary education, 10 percent in secondary education, 25 percent in tertiary nonuniversity education.9 Within the rapidly expanding tertiary nonuniversity sector, private teacher training institutions accounted for a significant share, (See Figures 3 and 4 for increase in enrollment in public and private schools). The overwhelming majority of private school students are from the richest consumption quintile of the country (Appendices 4.3b and 4.3c). For example, in the urban areas, these students accounted for as much as 39 percent of net enrollment in the 6 to 11 age group, 27 percenit of the 12 to 18 age group, 21 percent of the 17 to 25 age group. A significant percentage of the fourth quintile also chose private schools. By contrast, there were less than 1.5 percent of students from the poorest quintile of all age groups in private schools. In the rural areas, overall, there was not even 1 percent of children of all age groups in private schools. In the public school system in urban areas, net enrollment of the richest quintiles (53 percent) was much lower than that of the poorest quintile (85 percent). The implications for policy will be discussed in Chapter 2. * Setting up of regional administration, deconcentration of education services, and creation of new ministries The 1993 Constitution also restructures the political system. It divides the country into regions, departments, provinces, and districts.'0 At each region, the Transitory Council of Regional Administration (Consejos Transitorios de Administraci6n Regional or CTAR in short) coordinate all regional affairs and finances. In the case of Lima and Callao, the Development Corperation of Lima and Callao (CORDELICA) serves a similar functioni as the CTAR. The CTARs and CORDELICA are directly under the Ministry of the Presidency (PRES). PRES was created at the same time. Not only does it coordinate the region's education budget but also has responsibility for most of the capital expenditure on education through National Infrastructure for Education and Health (Infraestructura Nacional para Educaci6n y Salud, or INFES in short). 9 The decline in public and private university enrollment in official statistics is inconsistent with reality. The most probable reasoni is due to non-reporting in the case of private universities. As for public universities, the dramatic decline in enrollment between 1996 and 1997 is most probably because only the registration in the first semester is taken into account. Normally, registration in both semesters would be averaged out for the entire academic year. The figure for 1997 was probably not updated yet. A survey by the World Bank found that enrollment in public universities has remained stable in the 1990s, while that in private universities has grown rapidly. 10 The political levels that have elected offices are the central government and the municipal government (provincial and district municipal governments). The central government has an elected president and congress. The provincial and district municipal governments have their respective elected mayors and councils. The CTAR each appoints a president and a regional coordination council. 8 While the central government and the municipal governments have their own revenue sources, the regional administrations do not, and depend on the central government for transfer of revenue. Regionalization of administration has affected the budget process, intra-governmental allocation of resources, and the balance of power between various ministries in the center and the regions. In education, MED retains the overall responsibility for setting education policy on pre-primaiy, primary, secolndary, vocational, and tertiary nonuniversity education, but not on university education. Public universities remain autonomous and outside the jurisdiction of the MED or regional administrations. They have their own coordinating body, the National Assembly of Rectors (Asamblea Nacional de Rectores). A new ministry, Ministry for the Promotion of Women and Human Development (Ministerio de Promoci6n de la Mujer y del Desarollo HLumano or PROMUDEH in short), which was created in 1996, is in charge of early childhood care for children from birth to four or five, and the literacy program. MED is charged with the missions of developing the character of the individual, improving the quality of life, and facilitating social development in Peru through promotion of culture, science and technology, physical education, and pursuit of excellence. The responsibility for provision of educational services from pre-school to tertiary non-university education has been delegated to 23 Regional Education Directorates (Direcciones Regionales de Educaci6n or DREs in short) which sit within the regional administration, and to the Directorate of Education in Lima and that of Callao. MED makes educational policy for the entire nation, and gives technical and normative directions to the DREs, whieh implement policies. The Education Director of Lima is appointed by the Minister of Education, and the other Regional Education Directors are also appointed by the Minister with the approval of the regional administration. UJnder the DREs are Areas of Execution (AEs), Educational Service Units (Unidades de Servicios Educativos or USEs for short), and Education Development Areas (Areas de Desarrollo Educativo or ADEs in short)."1 The USEs manage some 58,000 schools and about 18,000 nonformal educational programs, both public and private (Figure 5). Both AEs and USEs are administrative units, but ADEs are educational supervision and support units. USEs are line units executing the functions and budgets of the DREs or Sub-DREs. Each DRE is headed by a director, who is assisted by two committees: one composed of the heads of subordinate units (Sub-DREs), another composed of the heads of internal line units, such as internal control, administration, legal and technical advisors' office, and technical pedagogic and technical cultural departments. Sub-DREs and USEs are organized essentially along the same lines. I Decree 26011 provides the framework of decentralization of educational administratioll and manageiment and transfers the private right to Communal Councils of Education (Consejos Comuunales de Educaci6n or COMUNED) for the administration of public schools. But months after the approval of the law, it was decided that it would not be implemented. 9 Figure 5: MINISTRY OF EDUCATION OPERATIONAL STRUCTURE 05-29-98 MINISTRY OF EDUCATION (1)t EDUCATION REGIONAL DIRECTORATE EDUCATION SUBREGIONS AREAS OF UNIT OF EDUCATION ,AREA OF EDUCATION EXECUTION SERVICES DEVELOPMENT TOTAL EDUCATION TOTAL NONFORMAL7 CENTERS (SCHOOLS) PROGRAMS PUBLIC EDUCATION PRIVATE EDUCATION NONFORMAL PUBLIC NONFORMAL PRIVATE CENTERS CENTERS EDUCATON PROGRAMS ! EDUCATION PROGRAMS t Init Educ 9,064 :nit Educ 4,936 1 nit Educ 17.029 ' lnit Educ 69 iPrim Educ 26, 963 ' Prim Educ 4,976 Prim Educ 309 Prim Educ 67 Sec Educ 6231 1 Sec Educ 2,060 Sec Educ 94 Sec Educ 169 Other levels 3,022 1 Other levels 1,764 Other levels 130 Other levels 4 Source Ministry of Education Summary. Policies implemented since 1990 have irrevocably changed the education system. Containment of public spending and mobilization of private resources may be loosely considered as the first generation of reform. Important achievements ensued. Yet, in the course of implementation, many issues have arisen that must be addressed in order for the sector to move forward. These issues have largely defined the scope for a second wave of reform which must deal with remaining inequities, quality improvement, further expansion, and institutional issues. Table 5 at the end of Chapter 5 provides an overview of how these new measures cross cut into with the issues of public and private finance, quality, efficiency and equity, and the teaching profession. 10 Chapter 2. Education Finance Two key questions in the puzzle posed at the beginning of the report are whether Peru's ability to finance high level of enrollment is because public resources have been used efficiently and equitably and whether households spending on education is high. This chapter evaluates these questions by first reviewing public spending on education and then by looking at the magnitude and impact of household financing. It then discusses the policy implications. 2.1. The Budget Process To understand public finance of education in Peru, it is important to first understand the budget process. The fiscal year in Peru coincides with the calendar year as well as the school year. The budget process begins in May every year when the lowest units submit their budget requests for the following year. The Ministry of the Economy and Finance (MEF) consolidates all requests in August and presents to the Congress in September. The Congress approves the budget in November/December for funds to be allocated in January. There are five budgetary entities in education: (a) MED, which covers the greater Lima/Callao area, (b) regions, (c) public universities, (d) other decentralized institutions, and (e) PRES. The first four are entitled primarily to handle recurrent expenditure under their jurisdiction, while the last one is responsible for investment and, hence, controls most of the capital expenditure through INFES."2 Since the establishment of regional administration in 1991, each of these entities prepares their own budget and negotiates directly with MEF.13 The process begins when schools present their requests for recurrent budget to the USEs, which submit a consolidated request to the DREs which, in turn, forward the aggregated budget to the CTARs. These budgets are consolidated in PRES, and then presented to MEF. MED's budget which also covers greater Lima and Callao and some national programs are submitted directly to MEF. The universities and decentralized institutions submit their own respective budgets directly to MEF. PRES also presents its budget for capital expenditure in education directly to MEF. Coordination is weak between MED and the regions, between MED and PRES, and between the regions and PRES.14 12 The budget for PROMUDEH, which has responsibility for literacy program and early childhood care, is not consolidated with education. The budget for these activities is very small. Since PROMUDEH was created only in 1996, this report does not cover this new ministry. 13 In principle, this practice would change after April 1, 1998 to have MED coordinate all recurrent budgetary matters for the regions. The new process was supposed to enable MED to have an overview of the nation's education budget and spending pattern. However, this was implemented for only three months and then was reverted to the old process. 14 DREs have little control over their own capital investment. For example, since all the education allocation to them covers only recurrent spending, if DREs want to buy a computer for use in the office, they have to submit a separate request to the CTARs for incorporation into the regions' capital budget request. 11 Under this process, MED does not have complete information about what the regions have requested and the regions do not report to MED about their actual expenditure. Therefore, much of the policy by MED which has national application has been made without clear information onl the requirements and availability of resources in the country. MEF has the ultimate say over allocation of education resources but without having an overall view of priorities and strategies in the education sector. There is no coordinating body in education that can assure the coherence of policies and provide the necessary resources to support them. MEF's decisions for allocating resources are based on the availability of public funds to match with what have been requested by MED and the DREs to deliver services. The first obligation is to pay salaries and pensions, and then to meet the needs for providing basic services of each of the entities. There are no funding formula to allocate resources other than the aforementioned priorities. Funds are allocated on a quarterly basis, but spent on a monthly basis. Funds not used as planned every month have to be returned to MEF at the end of the month, to be deposited back to the public treasury. There is no incentive to award savings. The main ground for allocating recurrent budgets to each DRE is the number of authorized pensionable positions (APPs) for teachers and administrative staff. These positions, in turn, are based on the ratio of allocating, on average, one teacher for every 35 students in the urban areas, and one teacher for every 20 students in rural areas, with some variations by level and for remote areas. On the surface, this method of allocation has given special consideration to the rural areas. However, thie 20 students in the rural areas may be of different ages and grade levels, and the teacher does not have a full range of skills to meet all of their educational needs. Moreover, since the freeze of pensionable positions in 1995, the departments that have high birth rates (usually poorer and with a larger indigenous population) have been more adversely affected than others. Teachers and administrators in the urban areas are paid monthly through deposit to their bank accounts although some are paid by checks; most of those in rural areas are paid by check. Textbooks, library books, and other educational materials and supplies are generally purchased by MED at the central level and are delivered to DREs, which distribute them to all schools under their jurisdiction. Electricity and water for evening schools are paid by the USEs. For many schools, parents' contribution pay for water and electricity. Many rural schools have neither water nor electricity. Given the fragmentation of the budgetary process, there is a strong case for improving coordination among various budgetary entities to improve the coherence of educational policy and efficiency in resource allocation to the sector as a whole. Equally strong is the case for improving the consistency, flexibility, and transparency of funding decisions through formula that reward efficiency and allow for adjustment to local needs. 12 2.2. Public Expenditure on Education'5 Historical trend. Government allocation constitutes the most important source of funding for education in Peru. Enrollment growth, however, has far exceeded the growth rates of either the GDP or public expenditure on education. Between 1970 and 1990, GDP increased by 85 percent in real terms, total government expenditure by 84 percent, public expenditure on education by 72 percent, while enrollment in public institutions by 130 percent (Appendix 6.2). Public spending on education fluctuated widely throughout the last 27 years. Between 1970 and 1997, public expenditure on education16 peaked in 1972 at 3.7 percent of GDP, falling to 2.2 percent in 1988 at the lowest point, and recovering to 3 percent in 1997.17 The steep decline in public spending on education in the late 1980s reflected the extremely volatile macroeconomic environment. In 1988, when GDP contracted by 8.4 percent in real terms and total government expenditure by 29 percent, total public spending on education declined disproportionately by 40 percent (Figures 6 and 7, Appendices 6.1 and 6.2). The recovery of public spending on education in the 1990s started from this extremely low base in the late 1980s. After declining in real terms by 40 percent in 1988, 10 percent in 1989, and 7 percent in 1990, allocation to education increased annually by some 3.6 percent in 1991 and 1992, respectively, by 18 percent in 1993, by 23 percent in 1994, and by 20 percent in 1995. It was reduced by 7 percent in real terms in 1996 but rose by 18 percent in 1997. The overall trend in the 1990s is a reversal of that in the 1980s: education expenditure has increased at a higher rate than that of GDP or total government expenditure (except for two years) (Figure 7). This trend indicates the government's commitment to education. Consideration for fiscal balance, however, has led to a gradual approach to increasing public spending on education. The enormous fluctuation of public expenditure on education over time, nonetheless, reflected deep- seated instability and unpredictability in resource allocation, which made it difficult for strategic planning, and undermined continuity of projects. 15 This review of public expenditure on education by Juan Pablo Silva of the Ministry of Education and Arturo Miranda of Universidad de San Marcos has updated and deepened the analysis by Jaime Saavedra and Roberto Melzi (1998). Jaime Saavedra reviewed the work to ensure consistency in methodology. 16 This review focuses on direct public expenditure for educational institutions, which coincides with the Government's official account of public education spending. It does not examine public subsidies to households such as school health and school meals, which is funded under the Ministry of Health and PRES, or early childhood care and the literacy program under PROMUDEH. The reason for doing so is to ensure that the scope of discussion remains focused. It is also consistent with OECD's classification, which divides public expenditure into three groups: (a) direct public expenditure for educational institutions, (b) total public subsidies to households and other private entities, and (c) financial aid to students not attributable to households (see Appendix 10.1). When the scope. of the review is clearly defined, it would be possible to compare across countries. This includes external finance, but not interest payment from borrowing. 13 Figure 6. Public Expenditure on Education and Central Government Expenditure as a Percentage of Gross Domestic Product, 1970 to 1997 ~-*---TotaIpubic_Expendte -0-e-TotaI E d-u ca thcn petue Source- Ministry of Econom-y and Finance (MvEF) Figure 7. Percentage Change of Gross Domestic Product, Central Government Expenditure, and Public Education Expenditure; 1970 to 1997 40 30 20 10 I iCJ~~~~~~~~~~~~~~~~~~~~i 0 -~~~~~~~~ ~~1099 1994 1905 1097 | -G DP -. -Total Publi. Eop.nditure -...To tal Educatio tr Source Ministry of Economy and Finance (MEF) 14 The level of public spending on education in Peru is low in comparison with other non-socialist lower-middle-income countries (Figure 8). It is substantially lower than the Latin American regional averafe of 4.5 percent of GDP (UNESCO, 1998), which is also the similar as 4.6 of OCED' (OECD, 1998). Because the school-age population of OCED is much smaller than that of Latin America, even if the level of public spending on education as a percentage is similar, the need for educational services is proportionally higher in the latter. In Peru, about one-third of the population are attending schools, in contrast to 16 percent in France and the Ulnited Kingdom, respectively, 14 percent in Japan, 28 percent in Mexico, 26 percent in Colombia, and 23 percent in Chile. This comparison males Peru's level of public spending even lower in both relative and absolute terms. It should be noted that pensions of retired teachers and administrators are paid out of the reculTent expenditure on education. This accounted for 22 percent of the total education expenditure in 1997. Net of pensions, public expenditure on education was about 2 4 percent in 1997 Many countries19 pay pensions out of a separate ftmd, such as a provident fund which may be inivested in the capital or financial markets to increase the fund, not from the recurrent allocation to the sector. Although public education expenditure that includes pensions reflects the true cost of education, when comparison is made with other countries' spending levels, the proper way is to compare public expenditure net of pensionis. This will make Peruvian public spending on education as a percentage of GDP less than half of the region's average for most of the years in 1 990s. While it is a tribute to MED and the teaching profession to be able to sustain such high enrollmeiit ratios at all levels with so little resources, the situation highlights the predicament of the e-ducation sector, with adverse implications for quality. Figure 8: Public spending on education and GNP per capita in lower middle income countries 10 oU 9. o 8** (DZ4 gLO - . r 6 0O 5 ;a8- 4 3~~~* PER$* .2 a.~ 0 2,000 4,000 6,000 8,000 10,000 GNP per capita PPP (Purchase Price Parity) in current international $ Source. Edstats database of the World Bank The OECD's average cited here refers to educational institution related expenses, but excludes educational subsidies to households, and student financial assistance. '9 Paying pensions out of the sector's recurrent expenditure is uncommon, except in socialist countries such as China. The United States, Singapore, Hong Kong, South Korea, Jamaica, Trinidad and Tobago, Colombia, Chile, Mexico and Argentina pay pensions from a separate fund. 15 Changes in composition of education spending. What did the additional public expenditure in the 1990s finance? Unlike many countries where most of the increase in public expenditure on education has been absorbed in personnel cost, Peru put the additional resources in educational infrastructure, in quality enhancing inputs (such as textbooks), in teacher training and in capacity building. For example, the World Bank Project for Improvement of Quality of Basic Education (Mejoramiento de la Calidad de la Educacion Primaria or MECEP in short) finances a class-set of free textbooks for all grades in primary education throughout the country. Capital investment increased from 1.4 to 15 percent of total public expenditure on education between 1990 and 1994, and then gradually fell back to 8 percent in 1997. Spending on other capital goods also increased from 1.4 to 2.1 percent. Spending on goods and services as a percentage of total education expenditure more than doubled from 4 to 10 percent, and other recurrent costs also more than doubled from 0.7 to 1.8 percent. It should be noted, however, that the fluctuation in non-personnel education expenditure still bore the mark of unpredictability which undermines planning and implementation (Figures 9 and 10 and Appendix 6.7). By contrast, total personnel cost (remuneration and pensions) increased by 64 percent, substantially below the 94 percent increase of total public expenditure (Appendix 6.7). As a result, the percentage share of personnel cost was reduced from 92 to 78 percent of total public spending during the period. Net of pensions, compensation for teachers and administrators (which includes salaries, allowances, and contribution to future pensions) accounted for under 60 percent of total public expenditure. This is substantially below the personnel expenditure of most lower-middle-income countries. Intra-governmental transfer of resources is the area where the most far-reaching change in education finance has occurred. The Government initiated in 1991 a policy to transfer public funds directly to the regions. In 1990, the MED managed 71 percent of the public education expenditure, the regions 17 percent, the universities 10 percent, other public institutions 1.6 percent, and the PRES 0.3 percent. By 1997, only 25 percent of public expenditure was managed by MED, as 56 percent was transferred directly to regions, 16 percent to the universities, 2 percent to other decentralized public institutions, and nearly 4 percent to the PRES (Figure 11 and Appendix 6.5). Given that such a large share of public expenditure on education is transferred to the regions, the universities, decentralized institutions, and PRES, the case for strengthening coordination among these bodies for policy and resource allocation is even stronger. The departments can generate own resources to invest in education, most of which are used to purchase goods and services, for administrative purposes, or for post- secondary education. However, the department's capacity is limited, and they depend heavily on transfer from the central government (Appendices 6.11 to 6.17). In 1997, central transfer accounted for 100 percent of pensions, almost 100 percent of all salaries in administration, planning, initial education, primary education, secondary education, tertiary education, and special education. PRES provided most of capital expenditure in primary and secondary education in the country. 16 Figure 9. Total Recurent and Capital Expenditures on Education 1970 to 1997 (In constant 1997 soles) 6000, 5000 4000 3000 E 1000 0c C) ON 00 00 Nco co co Co ) c n m ° Total Expenditure °3 Recurrent Expenditure -^- Capital Expenditure Figure 10. Composition of Public Expenditures on Source: Ministry of Economy and Finance (MEF) Education, 1990 to 1997 (Percentage, Regrouped According to the Lastest Classification) 40 20 1990 1991 1992 1993 1994 1995 1996 1997 1- - 0*larI.s.,d0on .fsl.,ln ,P.fl.fl,Ca It.IOx dd00,. .0' Sod,and 0008 **0IOh,,xpnkus Figure 1 1. Inter-Governmental Transfer of Resources (Percentage Source: Ministry of Economy and Finance (MEF) of Total), 1990 to 1997 80 70 60 50 40 - % * 30 _ 1990 1991 1992 1993 1994 1995 1996 1997 - Ministry of Education - Regional govemments -Public Institutions I t Public Universities '-MInIstry of the Presidency Source: Ministry of Economy and Finance (MEF) 17 It should be noted that pensions are paid through the department where the retiree resides, not where he/she used to teach. That is why the share of pensions as a proportion of total public expenditure varies from one department to the next. In 1997, pensions accounted for 32 percent of MED's total expenditure, about 21 percent of the region's expenditure, but 14 percent of total university expenditure. Since MED has jurisdiction over LimalCallao, 43 percent of the total pensions of the education sector were paid out of Lima/Callao, and the rest through other departments. (Appendixes 6.8) Intra-sectoral allocation. Among various subsectors, public universities are the only one which has benefited from uninterrupted increase in public expenditure (from about 10 to 16 percent) between 1990 and 1997. In 1997, about 6 percent of total public expenditure was spent on initial education, 27 percent on primary education, 19 percent on secondary education, 2 percent on non-university tertiary education, 16 percent on university education, and 21 percent on administrations. (Figure 12 and Appendix 6.9). It should be noted that administration expenditure includes compensation to all principals, school administrators and inspectors at all levels of education. Disaggregated information on administration is not available to pro-rate it to various educational levels. That makes spending by each level low and administration rather high. For comparison, OECD countries classify the salaries of all administrative personnel as personnel cost, not as administrative cost (Appendices 10.1-10.5). Per student spending. Between 1990 and 1997, per student recurrent public spending steadily increased at all levels. It grew by 70 percent in initial education, 87 percent in primary education, 71 percent in secondary education, 79 percent in tertiary non-university education, and 335 percent in university education. While the percentage increase was impressive, it started from a very low base (Figure 13 and Appendix 6.10). For university education, the very low per student spending in 1990 signaled poor quality. The rapid increase in per student spending throughout the 1990s, however, reflected not only additional public allocation to this subsector, but also reduced enrollment in public universities. Whether increased resources to improve quality of university education should come from the public or private sector will be discussed in Chapter 3. Converted to US dollars, per student public spending (inclusive of expenditure on pensions) in 1997 was US$175 in initial education, US$201 in primary education, US$260 in secondary education, US$324 in nonuniversity tertiary education, and US$1,255 in university education (Appendix 6.10). The difference in public spending per student between higher education and primary education in Peru was 6 times. Since the distribution of pensions differs by level of education, expenditure net of pension that goes to operating the university system is much higher than the gross figure, whereas expenditure net of pension that goes to basic education is lower than the gross figure. Net of pensions, per student spending on university education was 7 times higher than that of primary education in 1997. Nevertheless, this differential is still lower than that in many countries of Latin America (which may be as high as 20 times). In many countries in the region, public expenditure on higher education per student is often above $2,000. 18 Figure 12. Public Expenditure on Education by Level, 1990 to 1997 (Percentage) 1. I0 1990 ___i91 1992 1993 1994 1995 1996 1997 BIijal Education -*-ac duaio 'Secondary rducation ~Tertar_on-University - A*iisrtO -Universities Source. M inistry of Econoimy and Finance (MEF) Figure 13. Per Student Recurrent Public Expenditure on Education by Level, 1990 to 1997 (Constant 1997 Soles) 3500 3000 - 2500 - 2000 1500 1000 500) 0 _- - r-- r 1990 1991 1992 1993 1994 1995 1996 1997 -Initial Education Primary Education -*-Secondary Education 4Tertiarv Non-University Adinivrstro --_ _ ___ _ Source. Ministry of Economy and Finance (MEF) 19 Equity of distribution of public expenditure. How equitable has public expenditure been distributed? A standard method to measure the incidence of public expenditure is to construct a Lorenz curve20 to show the proportion of education expenditure which accrues to each consumption or income quintile. (This report uses consumption quintiles at all places).21 Since capital expenditure varies from year to year, only recurrent expenditure of 1997 was used for the incidence analysis. Figure 16 shows a number of Lorenz curves with recurrent public expenditure disaggregated by level of education.22 This analysis included expenditure on pensions. Recurrent public expenditure oni pre-primary and primary education was skewed toward the lowest consumption quintile (29 percent) and that on higher education was skewed toward the highest consumption quintile because the vast majority of students (47 percent) in higher education were from the top quintile and only 4 percent were from the bottom quintile. (Appendix 4.4a.) Public expenditure on primary education is equity enhancing not only because of the universal enrollment in primary education, but also because many families in the top two quintiles have opted out of sending their children to public pre-primary, primary, and secondary schools, leaving the public system mainly to the less well-off (Appendix 4.2a- 4.2d, and 4.3a-4.3d). However, the top quintile is the major user of public universities because children from that group have been better prepared for it and could afford to forgo the income to pursue further education. That is why public spending on pre- university level is more equitable than that oni university education. 20 The Lorenz curve is read as follows: the heavy straight black line joining the two corners as showni in Figures 14-16 is the line of "perfect equality" or the line which would obtain if each consumption quintile received an equal amount of educational expenditure-for instance, if 20 percent of expenditures accrued to the poorest quintile just as to the richest quintile. The curved line(s), the Lorenz curve(s) shown in these figures, represent the actual distribution of expenditures. The closer the curves are to the diagonal, the more equitable is the distribution of expenditures-in Figure 14 the curved line is very close to the diagonal, and the claim can be made, subject to certain assum1ptions, that public education expenditures in Peru are equitable. 21 The methodology for undertaking this analysis is simple. A table is constructed whichl shows the enrollments from each quintile, separately for each level of education. The number of students in each of the cells in the table is then multiplied by public expenditure per student on that level. This method is to get around the lack of data on actual expenditure per student by quintile. It assumes that the same amount of public expenditure is spent on a child from a poor family as for a child from a rich family. It does not adjust for the difference in teacher-studeiit ratios in rural and uirban areas. 22 The heavy straight black line in Figure 16 shows the line of equality. This figure indicates that public expenditure for the pre-primary and primary levels is not only equitable, it is actually biased towards the poor, so that more public expenditures accrue to the poor than to the rich. The diagram also shows that higher education expenditures are very inequitable, especially for ulliversity education. Interestingly, equity does not appear to be too much of a problem for secondary educationi-the broad dashed linie for the Loreniz curve for secondary education falls close to the diagonal. 20 Figure 14 shows that when recurrent public expenditure per student at all levels were combined, overall recurrent public spending was distributed quite equitably. The three lowest quintiles each received over 21 percent of the recurrent public expenditure on education, while the top quintile received 17 percent (Appendix 4.4a). This curve, which includes pension expenditure, is also referred to as Simulation 1 in Figure 15. Since the distribution of pensions differs by level of education, Simulation 2 in Figure 15 tested what the Lorenz curve might look like without pension expenditure. The simulation took an average of 26.5 percent of pension expenditure out of pre-primary, primary, secondary, and nonuniversity spending per student, and 13.5 percent out of university spending per student (Appendix 4.4b). The curve of Simulation 2 looks less equitable, but is not significantly different from Simulation 1 (Appendix 4.4b). This simulation is closer to the true picture (assuming that similar proportion of teachers are retired from the various groups). Still, both Simulations 1 and 2 were built on the assumption that the public spending per student in each level of education was uniform across all quintiles.23 However, the variation of public expenditure per student by department indicates that this assumption is unlikely to hold. In 1997, for example, the average public recurrent expenditure on primary and secondary education per student in the poor Department of Huancavelica was only 40 percent that of the national average, in contrast to richer Moquegua and Tumbes which had a level of per student spending that was about 160 percent of the national average (Appendix 6.18). This variation may be attributable to three reasons: (i) the pension burden (which is included in the expenditure) is much smaller in the poorer, inlterior departmenit but much heavier in the richer, coastal departments; (ii) the ability to generate own resources varies between departments; and (iii) since resources are based on student-to-teacher ratios, the freeze of pensionable positions in 1995 put departments with a high birth rates at a disadvantage. Whatever the causes might be, the assumption of uniform per student spending is questionable. Sinmulation 3, therefore, varied the per student spending by quintile. The simulation held the public spending per student of the middle quintile constant for all levels of education, but reduced that of the second quintile to 15 percent below that of the middle quintile, and that of the first quintile 30 percent below that of the middle quintile. By the same token, per student expenditure of the fourth quintile was raised 15 percent higher than that of the third quintile and the top quintile was 30 percent higher (Appendix 4.4c). Although the choice of these percentages for the simulation already reduced the variation in per student spending in the Departments by more than half, the Lorenz curve of Simulation 3 is still dramatically more unequal. 23 An account of the criticisms and the relevance of studies about the incidence of benefits can be found in "Assessing the Welfare Impacts of Public Spending," by Dominique van de Walle, World Development, 1998. 21 Figure 14. Lree- C.- fe,- fcid nueofPublic Eupenditu-e Figure 15. Loren Curves for lno.dence with 5 Simulatios t1 -| UolPodectEwd17 / /7 1 L |n. tPu "It Eq>l1;- 01 0 o,2 ,3 0,4 '0.5 2, 07 0s o.s 1 r,npor$oeno tPnrpu22n Figuere 1E. Lorenz Curves Py Educrtron Level I ___ _____ h j I 0 2 I ,I 0.2 0,2 0,4 o,5 0, 0.7 0.2 0,4. Propontlon ofFopulatlun P-p-- - -p...i. 5--02u2l -Pomury - - Oucuolryt -Sup NC UI0v-uelt Source: World Bank Analysis of Household Survey by Instituto Cuanto, 1997 22 Simulation 4 combined the principles of Simulations 2 and 3 and repeated the same experiment after taking out the pensions. Predictably, the distribution is tlhe worst among all simulations (Figure 15 and Appendix 4.4d). Simulation 5 asked: if per student public expenditure on higher education was much higher than the culTent level, while per student spending at all other levels remained the same, how inequitably would public resources be distributed? The Lorenz curve of this simulationl is almost as unequal as those in Simulations 3 and 4. This shows that the relatively low public expenditure on university education per student is an important reasoln for the overall Lorenz curve to look equitable in Simulations 1 and 2 (Figure 15 and Appendix 4.4e). However, if sclhool resources are distributed inequitably across quintiles, no matter how high enrollment ratios are in basic education, the Lorenz curve will look worse. This exercise shows that the Lorenz curve looks equitable overall (Simulations 1 and 2) can be attributable to three reasons: (i) universal primary education benefits the poor more than the better-off; (ii) the better-off has opted out of public schools, thereby consuming less of public subsidies; and (iii) public spending on higher education per student is relatively low. The system could be extremely unequal, if per student allocation in any given educational level is less for lower quintiles than upper quintiles, and if public expenditure allocated to universities continues to escalate. If capital investment is taken into account, the distribution of public expenditure is likely to be even more inequitable. The negligible historical capital investment has resulted in a highly inadequate learning environment. Historical investment in school infrastructure and equipment tended to favor large urban schools, the argument being to bring the greatest benefit to the largest number of pupils. As a result, urban schools are better endowed than rural schools. In the mid-1990s onwards, the National Social Fund (Fondo Nacional de Compensaci6n y DesalTollo Social or FONCODES in short) which was administered under PRES has responded to the demands of rural communities. INFES is also building schools in rural areas. Summary. The exercises in this chapter found that public resources have been relatively well-targeted and well-used. The pattern emerged in the 1990s of a modest but steady increase public spending on education reflects both commitment to education and fiscal restraint. The Lorenz curve constructed from using uniform per student expenditure for all quintiles suggests equitable distribution of public expenditure, although the absence of actual cost data by quintile leaves lingering questions on the methodology. Future investigation should collect data on the actual amount of public resources spent on students from different quintiles to shed light on this question. Furthermore, the largest increase in per student spending in the 1990s is that on public universities, and yet the majority of students in this level are from upper quintiles. This trend should be monitored closely to ensure that public resources are targeted to the truly disadvantaged. Given that the level of public spending on education is still low in absolute and relative terms, to expand access and improve quality for the poor, additional and targeted investment needs to be sustained for a long time in order to equalize educational opportunity. 23 2.3. Household Expenditure on Education With respect to the second question of whether Peruvians value education and have invested heavily in the education of their children, the answer is unequivocally affirmative. Historically, communities built schools and organized learning long before the Government began to play a key r ole in financing or provision of education. This was borne out by the high level of household spending on education, totaling to about 2 percent of GDP24 , according to analysis of household surveys of 1994 and 1997. This level of household expenditure is higher than the 1.3 percent of GDP spent by OECD countries, also higher than Argentina's 0.75 percent and Mexico's 1.1 percent, but lower than Chile's 2.6 percent, Colombia's 3.6 percent, and Jamaica's 6 percent.25 It should be noted that composition of household expenditure on education may vary from country to coun-try. Since household expenditure estimates are obtained from household surveys, how the questionnaires are phrased affect the information obtained. The cross- country information provided above is intended to show the range of household expenditure on education. It should be not be taken as comparable. The key question is not whether households in Peru spend too much or too little b'ut what this level of spending by households implies for educational policy in Peru. Understanding the breakdown of expenditures across consumption quintiles would provide an answer to the question of whether certain groups of Peruvians are deprived of educational benefits because they are too poor to afford the necessary expenditure. Secondly, it is important to address the question of what variables determine the variation in expenditures across households-this would aid understanding the implications for educational development in the future. Disparity in household spending on education by quintile. Figure 17 shows the Lorenz curve for total private expenditure, which looks very dissimilar to the equitable Lorenz curve for total public expenditures in Figure 14, but looks very similar to the very inequitable distribution of expenditure on public universities in Figure 16. The expenditures are indeed inequitable, with the lowest quintile accounting for only about 4 percent of the total household expenditures on education, and the upper quintile as much as 57 percent. This Lorenz curve for all private expenditures does include spending on private schools. Figure 18 shows the Lorenz cu-rve of household expenditure which is spent only on public schools. However, the curve is only slightly better. Peruvian households spent 24 In 1997, household spending on education amounted to about US$1,300 million (Appendix 7.14) Given Peru's GDP in that year of US$65,221 million, household expenditure on education was about 2 percent of GDP. This was consistent with the findings by Saavedra et. at. in the analysis of 1994 household survey. 25 OCED, Education at a Glance, 1997; Colomiibia - Departamento Nacional de Planacion, 1996; Chile, Ministerio de Hacienda and Banco Central, 1998; World Bank, 1999d, Janaicae Secondary Education: Improving Quality acnd Extending Access, Report No. 19069. 24 approximately $781 million (41 percent) for the education of children who were enrolled in public schools to complement the public spending on education that was about $1,932 million. These household expenditures include registration fees and contributions to parents' associations (Asociacion de Padres de la Familia or APAFA in short), uniforms, school lunches, and transportation.26 Figure 19 shows the Lorenz curve of household expenditure on public primary schools; the situation is only worse in regard to secondary schooling. The level of household expenditure on education varies tremendously by income level-the total amount spent on education by the richest quintile in Peru was 13 times the total amount spent on education by the poorest quintile. (Appendix 7.14). Even this figure is likely to be underestimated because the household survey questionnaire which provided the data for analysis did not include spending on extra tutoring and other school activities such as field trips. It is far more difficult for poor families to provide sufficient educational inputs for their children. For example, spending on books accounted for 35 percent of total household expenditure on public primary education for the poorest quintile, but only II percent for the richest quintile whose children attend private schools (Appendix 7.2). In the case of primary education, the average out-of-pocket cost for parents of the top quintile to send their children to a public primary school was 194 soles (US$73), or 2.2 times the amount spent by the poorest quintile, which was 88 soles (US$33). The average cost per child in the richest quintile in a private school was 1,645 soles (US$618), amounting to 19 times the average cost spent by households in the poorest quintile on public primary schools (Appendix 7.2). Since public expenditure covers mostly salaries, household contributions in school fees and to the APAFA are often used by schools for repair and maintenance, educational materials and supplies, and water and electricity. The disparity in the ability of parents to pay, therefore, has contributed to the disparity in school resources. Appendix 4.6 presents the result of a survey of some 400 rural and urban public schools in Lima and Cusco by MED in 1994. It shows that the annual APAFA contribution to very large urban schools (with an average of over 1,600 students) amounted to 11,735 soles, in contrast to only 279 soles of contribution to small rural schools (with an average of 96 students). This survey confirmed the disparity in school resources among very large, large, medium, and small urban and rural public schools27 (Appendix 4.6). To cite a few examples, 81 percent of very large urban schools have a library, compared to 26 percent of small rural schools; 82 percent of very large urban schools have brick or cement walls, 26 Unfortunately, the data was recorded on the questionnaire instrument of the household survey in such a way that it is not possible to separate out items such as bus tickets from the direct transfer of resources to schools through items such as APAFA fees. 27 This was based on the MED's survey in Cusco and Lima conducted in 1994. Although the situation might be different now, it is unlikely to have changed so dramatically that it alters the picture. 25 compared to only 21 percent of small rural schools; 79 percent of very large urban schools have electricity, compared to 37 percent of small rural schools; and 76 percent of very large urban schools have latrines that work, compared to 32 percent of small rural schools. Appendix 4.5 also illustrates how the disparity in the ability of parents to pay is translated into inequity in resources of schools attended by children of different quintiles (proxied by water and drainage). Note the high percentage of both public and private schools attended by poor children in the first and second consumption quintiles that lacked either water or drainage or both, in contrast to those schools attended by children in the top quintiles. Poor children attend schools with little resources, be that public or private. Elasticity of demand for education. Notwithstanding the aforementioned fact, Peruvians value education highly and would go to great lengths to make sure that their children have an education. Analysis of the behavioral aspect of household education expenditures (Engel's curves) found that the income elasticity of demand is a low 27 percent (See Background Note 2).28 This means that education expenditure is considered to be a necessity by Peruvian households and that there is a strong underlying demand for education, by both rich and poor.29 To make sure that the conclusion was not based just on one pooled set of regressions, the regressions were run separately for subsamples by indigenous and nondigenous, rural and urban, and poor and rich. Consistently, the pattern is that the income elasticities are lower for the more disadvantaged groups. It was 12 percent for the poorest quintile, 14 percent for rural populations, and 10 percent for indigeneous people. The finding resonated with that of Rodriguez and Abler (1998) for a sample of Peruvian children 6 to 16 years old. They found that even if there is a positive relationship between income of the family and the probability of school attendance, the estimated marginal effects are small. Moreover, the magnitude of the negative effect of family income over participating in the labor force is also small. That is why overall enrollment did not decline, and child labor did not increase, during the time of economic crisis. The study by Gertler and Glewwe (1989) had similar findings: that rural Peruvian households were willing to pay fees high enough to more than cover the operating costs of new secondary schools in their villages. This is even true of the poorest quarter of the income distribution. 28 See Deaton and Case (1988), Analysis of Household Expenditures, LSMS Working Paper No. 28, World Bank, Washington, D.C. and Sadoulet and de Janvry (1995), Quantitative Development Policy Analysis, Johns Hopkins University Press, Baltimore. The specification used here also borrows heavily from a Yale University Working Paper by Mwabu (1994), Household Composition and Expenditures on Human Capital Inputs in Kenya. 29 Mwabu's work on Kenya indicated a much higher income elasticity of Education expenditures of 73 percent. 26 Figure 18. Lorenz C.rve for Incidence of PriVate Expenditures Figure 17. Lnrenz Curve for Ine.dence of Private E.penditures All Levels Only Public Schools o ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ 00 O - I 0.6~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~06 Q20 0 0.2 0,4 0,6 0,8 1 ° 0,2 0.4 0 6 0. 1 Proporion of Populatin Prportio f Populatio,n Figure 19. Lorenz Curve for Incidence of Private Expenditures Only Primary Schools 1- . ... ... ......, - 0 0,2 0,4 0 _ _ 0,8 1 Propo.Mon of PopultCon Source: World Bank Analysis of Household Survey by Insituto Cuanto, 1997 27 From the point of view of educational policy, however, the Government cannot rely on general increases in income to bring about greater expenditures on education to improve educational quality for disadvanitaged groups. For every doubling of household income, the budget share spent on education would go up only by a quarter on average; for the poor, the rural people, and the indigenous people, their budget share in education would only go tup by 10 percent or so. Given that levels of household expenditure on education vary vastly by income level, there is a great need for specific policy instruments that will address the inability of poorer households to incur additional expenditures. Conclusion. Because the burden for financing education is disproportionately heavier for poorer households than richer households, the public sector has a special mandate to ensure the equality of educational opportunity for all by directing more public resources to the poor. Past constraints on public expenditure allow exercising this mandate without exceeding reasonable overall public allocations to the sector. 28 3. System Performance Indicators This chapter reviews indicators on educational access, internal efficiency, quality, anid labor market outcomiies in order to address the question of whether expansion of education in the past has come at the expense of quality, and to assess the implications for policy in the future. 3.1. Access, Repetition, and Retention Household surveys have repeatedly found evidence of near universal enrollment in primary education for children between the ages of 6 to 1 1.30 This is reflected in very high gross enrollment ratios. Access to early childhood, secondary, and tertiary education, however, varied tremendously by socioeconomic status, gender, and urban or rural location. In general, both males and females between the ages of 12 and 17 in rural areas, irrespective of consumption quintile, are less likely to be in school than their cotnterparts in urban areas. For the 17- to 25-age group, girls in rural areas definitely have fewer opportunities than boys of the same consumption quintiles in the rural areas or girls in the urban areas. (Appendices 4.1 a- 4. lb). It should be noted that these gross enrollment ratios indicate how many students of all ages are studying at a given level; it encompasses late entrants, under- and over-aged students, repeaters, and adult learners. Gross enrollment ratios, therefore, often exceed 100 percent. Net enrollment ratios, however, indicate what percentage of children of a particular age group are studying in the level designated for that age group. It never exceeds 100 percent and is a more accurate measure of the amount of schooling acquired by the age group. Where large differences exist hetween net cnd gross enrollment ratios, they signal that a large proportion of students are late entrants and repeaters It is, therefore, very telling that gross enrollment rates in the rural areas among lower quintiles are higher (over I 1O percent) thani even their counterparts in the same quintile in urban areas. The gross enrollmenit of the top quintile in urban areas is under 99 percent, showing that they move through the system rapidly without wasting time in it. (Appendices 4. l a and 4. lb). 30 There are two major institutions in Peru which have conducted regular household surveys and whlich provide databases for analyzing education indicators: INEI, which is under the GOP and Instituto Cuanto, which is a privately run organization. INEI has conducted a school census in 1993 and another one in 1998. Most of the official education statistics including enrollment ratios and average years of schooling are drawn from INEI's data. This report, however, draws from the household survey of 1 997 conducted by Instituto Cuanto because its questionnaire is richer and also because the World Bank's other studies (on poverty and labor market) also drew from this dataset. It should be noted that Instituto Cuanto's sample size is much smaller than INEI's, and its sample frame is different. For this reason, indicators on access, repetition and retention reported in this report are not the same as those reported in government statistics. Nonetheless, the broad picture revealed by data from Instituto Cuanto is similar to those by INEI. 29 Peru 's net enrollment ratios are much lower than the gross ratios, averaging in the rural areas only 10 percent in initial education, 65 percent in primary education, 28 percent in secondary education, and 4 percent in tertiary education. These are lower than the urban areas' 12 percent in initial education, 71 percent in primary education, 57 percent in secondary education, and 22 percent in. tertiary education. Net primary education enrollment differs substantially by income quintile and gender within and between urban and rural areas. The difference was most pronounced at the tertiary level, where, in the rural areas, only 2.3 percent of girls and 6 percent of boys of the first quintile enrolled, in contrast to 16 percent of girls and 11 percent of boys in the same quintile in the urban areas. (Appendix 4.1 e and 4.1 d). Rural children tend to enter late inito the school system because they often have to walk to school. As schools are usually established in population centers, allowing for 1.5 to two hours' walking distance from other settlements, only older children who can endure the journey can go to school. As a result of their late entrants and high absenteeism (due to the need to help their families and due to vulnerability to climatic factors), they also tend to have high repetition rates. Research has found a clear association between late ently and high repetition rates on the one hand, and poverty, indigenous language speakers, and uneducated mothers on the other. About 63 percent of Quechua speaking children are over-aged. For children who work in the countryside, 68 percent are over-aged, and the dropout rate among them is as high as 55 percent (Montero et. al, 1998). Official statistics on repetition31 and dropout are highly aggregated, without gender or urban and rural breakdown. In the aggregate, 80 percent of a cohort reached Grade 6, while 60 percent of the same cohort reached Grade 1 1. However, only 40 percent of Grade 6 students and 20 percent of Grade 11 students had not repeated during their course of study. Dropouts spent an average of 6.7 years in the system (Appendix 4.7). To obtain a picture of the differential repetition and dropout rates by the rich and poor, household survey data are used. The analysis by Saavedra and Felices (1996) of the 1994 Cuanto household survey confirmed the relationship between repetition and income-the percentage of repeaters went from 17 in Lima, to 24 in other urban areas, and rose further to 35 in rural areas. Repetition is also much higher in public schools than in private schools. The study also revealed the relationship between income and status dropout (defined as the proportion of individuals in a cohort that have not finished an educational level and are not enrolled in any educational institution). For those aged 17 to 24, the status dropout rates were 13 percent in Lima, 20 percent in other urban areas, and 54 percent in rural areas. 3 1 The MED has adopted a policy of automatic promotioni in lower grades in primary education since 1998. Therefore, repetition rates are not good indicators of whether students have mastered the requisite skills for a given grade level. 30 An anialysis of Instituto Cuanto's household survey of 1997 found large disparity in school survival rates32 between urban and rural areas. Figure 21 shows that although urban and rural children started out the same in the first year of schooling, they rapidly diverged after the Fourth Grade. Figure 23 displays a similar pattern of school survival rates between children from the top and bottom quintiles. In both cases, disadvantaged children (that is, rural and poor) dropped out much earlier. School survival rates did not differ much by gender (Figure 20) but disparity was significant between children whose mother tongues are indigenous languages, and those who are Spanish speakers (Figure 22). In comparison with the school survival gap betweeni rural and urban areas, the indigenous gap appears to be smaller. However, this may simply reflect the reluctance of people to identify their own mother tongue (which is the variable often used to construct ilndigenous and non-indigenous populations). That the school survival rates of Spanish-speakers were much lower tlhan those of urban dwellers seems to lend support to the above-mentioned point. Conclusion. High repetition and low retention rates indicate low internal efficiency of the system. This means that children spend time in the system without attaining the education level and mastering the skills commensurate with the numnber of years of enrollmelnt. This is nlot only a waste of public anld household resources but also has grave consequences for the future employment prospects and life-time earnings of children. The solution is not to adopt a policy of automatic promotion but to ensure that children do not learn the skills relevant to the grade-level every year by a range of policy options to be discussed in Chapter 5. 32 The survivor function plots in Figures 20 to 22 are known as "Kaplan-Meier Survivor Functions". They are non-paramnetric maximum likelihood estimates of the survivor function. (See The Statistical Analysis of Failure Timie Data, by J.D. Kalbfleisch and R.L. Prentice, Jolh Wiley and Sons, 1980). The survivor function uses information about the years of schooling completed and the current status of enrol lment at schooL. It is the complement of the empirical cumulative distribution ftnction. The Kaplan-Meier plot is not based on any regression model - the figure simply represents a coun1t of people as they leave the educational system compared to the number of people who stay on1. The numerator is the number of people who stay on, and the denominator is the number of people who have been in school up to the time. Hence, the plot always begins at a value of 1, since every one stays oni at the first instant, and subsequent "steps" on the figure show people leaving. Thiere is one underlying assumption, that if you have left school, you will not enroll again at a later date. This assumption is more valid in some cases than others, but it is a fairly standard one in the literature on educational attainment. 31 Figure 20: School Survival Rates by Gender, 1997 = ,/ ~~Female Male (0 D- 0 o 0 0 5 10 15 GTrade Source: World Bank Analysis of Household Survey by Instituto Cuanto, 1997 Figure 21. School Survival Rates by Urban and Rural Areas,1997 X : -12 J~~~~~~~~~~rban C3 ~ ~ ~ ~ ~ ~~3 Rural 0 5 1 0 1 5 Grade Source, World Bank Analysis of Household Survey by Instituto Cuanto, l1997 32 Figure 22. School Survival Rates by Mother Tongue, 1997 0 Castellano Indigenous n .~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~j.~~~~~~~~~~~~~~ 0 5 1 0 1 5 Grade Source: World Bank Analysis of Household Survey by Instituto Cuanto, l1997 Figure 23. School Survival Rates by Poorest and Richest Consumption Quintiles, 1997 1 0 0 8 . . . ~~~~~~~~~~~Richest Quintile 0 0 6 Poorest Quintile 0 4 0 0 2 00 0 5 10 15 Grade Source: World Bank Analysis of Household Survey by Instituto Cuanto, 1997 33 3.2. Labor Market Outcomes The consequence of low internal efficiency will impact on external efficiency (labor market outcomes). This will become even graver in the 21St century as the trend of the 1990s has already shown rapidly increasing wage differentials among workers with various education levels. After falling between 1985 and 1991 during the economic crisis, the preinia of all education levels bounced back in recenlt years after the implementation of structural reform and opening of the Peruvian economy (Saavedra, 1998b). The magnitude of decline between 1985 and 1991 was different among workers of different education levels-it was minimal for university educated workers but was almost halved among workers witlh other levels of education. When the premia bounced back in the 1990s, the increase was also steepest for university graduates. Figure 24 shows that in urban areas, in 1997, the earnings differentials between workers who had no education and those who had primary education barely recovered to the level of 1985. Improving quality of primary education will provide some real benefit to those who would only have this opportunity to acquire the requisite skills that distinguish them from workers with no education. The premia of secondary education (over that of primary education), has not grown as fast as that of the primary education premia and still has not reached the level of 1985. This might be related to the relatively slow growth of formal sector work which normally employ secondary school graduates. Figure 24. Evolution of Estimated Premia by Educational Level, 1985 to 1997 160 . 140 0 Univetaity Higher! 11 120 >< Secondary 120 00~ 100 Primary/ No education ,- Secondary/ | 80 . x Primnary t 60 Nonl-univeisity highet/ 5 s Secondary a 40^ 20 ---- _,__ _ __ . _ 1985 1991 1994 1997 Year Source World Batik Analysis of Household Survey by Instituto Cuanto, 1985, 199 l, 1994, 1997 Although the wage differential between non-university tertiary education graduates anid secondary school graduates also recovered by 1997, it was only returning to the level of 1985. Only the university premium exceeded tlhe level of 1985. This signals an increasing demnand for a higher level of skills in an open economy that is facing growing 34 international competition and technological change. This trend is similar to those in many Latin American economies such as Colombia (Cardenas and Gutierrez, 1997), Costa Rica (Gindling and Robbins, 1995), Chile (Robbins, 1996) and Argentina (Pessino, 1995) where returns to education have also increased after structural reforms. Rising education premia has implications for policy. If education premia continue to increase, many people will have incentives to pursue further education, even in adult or evening classes. As they attain more education they are likely to climb in the earnings scale. This will probably increase their likelihood of not being poor. If repetition and dropout rates fall, people will have higher levels of education attainment, even if their years of schooling remain unchanged. Given the positive relationship between literacy of parents and school attendance of children, there is an intergenerational positive effect on education. This is the start of a virtuous cycle. In the case of workers that do not improve their skills through education and training, they are less likely to improve their income due to their lower productivity. In rural areas, Escobal, Saavedra, and Torero (1998), using a sample of rural families in household surveys from 1985 to 1996, found a positive relationship between years of educationi of the household head and of the other family members and per capita household expenditure. Taking the trend to its logical conclusion, an improvement in the quality of basic education will certainly have effects over the skill structure of the labor force and income distribution. Table 2 shows private and social rates of return to education in 1997 (See Background Note 3 for methodology and explanation). Returns to primary education for men are very high. This is not inconceivable for this educational level. A paper by Psacharopoulous shows the rates of 24 percent and 31 percenit for Venezuela and Guatemala, respectively. On the other hand, the large difference between males and females miglht be attributable to the small number of non educated males actually working in urban areas. . That they comprised only 2.3 percent of the sample of males might bias the estimate. Table 2: Urban Peru: Rates of Return to Public Education, 1997 Female Male Private Social Private Social Primary education 5.9% 5.1% 37.8% 26.3% Secondary education 10.4% 7.4% 7.2% 6.1% Non-university higher education 12.1% 10.4% 9.4% 8.2% University higher education 13.9% 12.4% 12.1% 11.1% Source World Bank analysis of Cuanto's 1997 household survey 35 Private returns to education increased with the level of education for both men and women (except for primary educated males). This is completely consistent with the trends observed in the labor market all over the world. What is peculiar to Peru is the veTy small difference between private and social returns to various levels of education (Table 2). This corroborates with the point made in Chapter 2 about the low level of public spending on education per student at all levels because the calculation of social returns used public expenditure to estimate the cost. When the costs are low, the returns would be high. Another feature special to Peru is that social returns to university education were higher than those to other levels of education. Again, this supports the point made in Chapter 2 about public spending on university education per student being much lower than those of other countries. Both the social and private rates of return to all levels of education, except primary education, were higher for women than for men. (Table 2). This indicates the profitability for public investment in girls and women's education. Given the very low enrollment ratio of females in higher education in the rural area, and among the lower quintiles, the beneficiaries of these high private returns are urban women from upper quintiles. Since rural women are not likely to have access to such opportunities without specific government intervention, provision of scholarship rural girls who have a good academic record would yield high benefits to society. Conclusion. The policy implications of these findings are that investment in basic education, both in terms of qualitative improvement and quantitative expansion of secondary education, will have a positive effect on alleviation of poverty, although returns to this level are probably lower because the initial general human capital is low. The rising private returns to higher education means that children from the upper quintiles who are the dominant consumers of university education, are the main beneficiaries of public investment in this subsector. (See Appendix 4.2a to 4.2b for enrollment in higher education by quinitile and Appendix 4.4a, b, c, d, and e for incidence analysis). Although high social rates of return to education justify continuous government investment in this level, the social returns will not remain high if public spending per student in this level keeps on rising. Additional resources needed to improve quality of higher education can come from cost sharing at the university level. Given that students will benefit from a high level of life-time earnings, it is justifiable to ask them to contribute their fair share to finance their own study. However, increased cost sharing also needs to be supported by student financial assistance such as student loans to ensure the academically deserved will not be disqualified due to financial constraints. To ensure gender equity, proactive policy by the Government is needed to support women, particularly those in the rural areas, with good academic standing to access higher education. 36 3.3. Learning Outcomes While repetition, retention, and dropout rates are indirect indicators of quality, the most direct measure of learning outcomes is student achievement. Like other Latin American countries, Peru has only set up measures of student achievement in recent years.33 Notwithstanding start-up problems -- which are common to many countries, and which limit interpretation of the initial tests, --information on the relative performance of students in the country provides the opportunity for a preliminary diagnosis of determinants of achievement. In addition, it provides a basis for improving the assessment instrument and other technical aspects for policy research in future. (See Background Note 4 for description of the analytical procedures and preliminary findings.) Table 3 presents an index of relative outcomes in mathematics. It should be cautioned that such indices are not meaningful by themselves. Furthermore, validity (measuring what it should measure for students at that grade level) and comparability of the instrumenit across years have yet to be established. An assessment instrument that is set to be too difficult, even if it is based on the curriculum, can have the effect of making students perform badly; on the contrary, if it is too easy, it can make all students perform well.34 Also, because the test was not equated with other internationally known studies, -- such as the Third Interniational Math and Science Studies, whose Population A of 9- and I 0-year-olds were quite comparable with the Peruvian Fourth Graders, -- the test results cannot be interpreted as an indicator of how well students performed in comparison with students of other countries. What is informative for policy analysis purposes is the relative performance of students compared among themselves. In Table 3, an index is reported in which the national average is set to 1. The average of other subgroups can be compared against it. The index shows that the achievement gap between private and public schools is large. Among public schools, disparity exists between urban and rural areas. The urban and rural outcome differential is likely to be much bigger than the data show, because children in very small rural schools with only a single teacher were not included, and yet these schools represented about 29 percenit of all schools and about 6 percent of the population of Fourth Graders 35 Students on the coast performed on average better than 33 The first national standardized test of achievement in mathematics and language was conducted in 1996 among nearly 50,000 Fourtht Graders in a national sample. The results of the 1996 assessment remain to be publicly released. Thus, this document reports no statistical tabulations of test scores directly, nor does it attribute average scores to schools or departments. It is nonetheless still possible to undertake linited statistical analyses without reporting scores. These more limited analyses are used in this section as well as in Background Note 4. If the assessment results become publicly available, a more thorough and explicit analysis can be reported. 34 A good test should have both easy and difficult items for botlh the average and the exceptional students to score It slhouild be consistenit in its difficulty level from year to year in order to measure progress 35 According to the 1993 Census, 29 percent of all schools were single-teacher schools, 33 percent had at least some multigrade classrooms, and 38 percent were complete schools. 37 those in the mountains (sierra), who in turn, fared better than those in the jungle (selva). There is also a gender difference in outcomes, as girls did less well than boys. The lowest index is among Quechua-speaking students. To the extent that the coefficient of variability is large on country average, and much larger among certain subgroups (public rural schools, Quecha-speaking students, and in the jungle areas), variability of learning outcomes is a key issue in education. Table 3: Index of Fourth Grade Mathematics Outcomes, 1996 Index of Mathenmatics Coefficient Outcomes Of Variability Countrywide Average 1.00 0.47 Male 1.04 0.46 Female 0 96 0.48 Public rural 0.85 0.54 Public urban 0.97 0.46 Private 1.37 0.34 Spanish-speakers 1.04 0.46 Quechua-speakers 0.73 0.54 Ayinara-speakers 0.99 0.44 Coast 1.10 0.43 Mountain 1.00 0.47 Jungle 0.83 0.50 Note: The coefficient of variability is computed by dividing the value of the standard deviation by the corresponding mean of the group While it is a very standard measure, the coefficients of variability in this table do not have direct relationship with the index, except in the case of the countrywide average To assess the impact of public finance on school achievement, mathematics outcomes by department were regressed on per student public expenditure on primary education by department of 1994. The reason for using 1994 expenditure data to predict outcomes in 1996 is because there is a time lag between public spending and demonstration of effects on learning. No relationship was found between public spending per studelnt by department and outcomes by department, as the R-square was zero (Figure 25). A weak relationship was also found between poverty by department and per student public expenditure on primary education by department (R-square = 0.06) (Figure 27). This might be attributable to the formula of allocating teachers: in the rural areas, one teacher is allocated for every 20 students, whereas in the urban areas, one teacher is allocated for every 35 students. Given the high correlation between poverty and rural population as a percentage of total population in the department (Appendix 5.2), it could well be that no relationship was found between poverty and public expenditure per pupil. One should also taken into account that massive increase in public spending on education began only in 1993 through 1995. It is not surprising to see little effects of public spending on outcomes during this short time period because of the natural time lag between delivery of infrastructure, goods and services (including textbooks) to the schools, and when these facilities and goods are used for teaching and learning. 38 Figure 25: Mathematics Outcomes and Recurrent Public Expenditure on Basic Education Per Student High 4, : : E 0 E *04, y = 0 0209x 4+ 42029 RI = 0 0069 Low - r 0 50 100 150 200 250 Recurrent Public Expenditure on Primary Education per student of 1994 (in US dollars) Source World Bank Analysis of data from MEF and MED. Figure 26: Mathematics Outcomes and Household Expenditure on Basic Education Per Student High 0 y 0 11 55x +436 153 R' 0 3797 0 20 40 so 50 100 120 140 160 Household expenditure on Basic Education per student In US dollars Source: World Bank Analysis of data from- Cuanto's H4ousehold Survey and MED. 39 Rgure 27: Poverty and FRcurrent Public Expencditure on Basic Education Per Student 250 _ . . . o 200 * 0 Ul ~.0 a. ui 100 C 5D 'fl 50 y -7.2356x + 163 28 a. R =0 0663 0 00 05 10 15 20 25 30 35 40 45 50 Poverty Index Source: World Banik Analysis of Cuanto's Household Survey and MEF ;Rgure 28. Poverty and H-busehold Ekpenditure on Basic Biucation Per Student io 140 1R2=0 54 0~- '.C80 :3 W 60 40 20 0 00 05 10 15 20 25 30 35 40 45 50 Poverty Index Source: World Bank Analysis of data from Cuanto's Household Survey and FONCODES. 40 To examine the effects of household expenditure on achievement, family spending on basic education per capita in 1994, by department, was used to predict mathematics outcomes in 1996. It should be noted that the level of household spending on education reflects a very long tradition of family support for education, and hence, the level of private spending is likely to represen-t a continuity of this tradition, rather than an abrupt change as in public spending. When mathematics outcomes by department were regressed on household spending on basic education per capita by department, a positive relationship was found and the R- square was a strong 0.38 (Figure 26). This indicates that the higher the level of household spendinig per capita by department, the higher the learning outcomes by department. This raises the question of whether poorer departments were particularly disadvantaged. To answer this question, poverty index by department was regressed on household spending on basic education per capita by department. A negative relationship was found, meaning that the poorer the department, the lower the departmental average of household spending per capita. The R-square was a strong 0.54 (Figure 28). These reinforce the poinits inade earlier about the inherent inequality in relying on households to finance basic education because it would only replicate the socioeconomic inequality in society. Nonetheless, some departments which had low levels of household spending on basic education had departmental outcomes well above the predicted line (Figure 26). This raises the issue of whether household expenditure by department captures the effects of other variables. To disentangle these issues at the departmental level, the technique of hierarchical linear modeling (HLM) was used. (See Background Notes 4 for details.) It was found that, although the combination of public and private expenditure per capita by department explained nearly half of the between-department variance in outcomes, public and private expenditure did not substitute for each other. Poverty alone had a negative effect on outcomes, but it did not have a linear relationship with them. Dividing departments into non-poor, average, poor and extremely poor categories provided a more precise measure of the effects of poverty on achievement. All of the above mentioned variables, in combination with departmental percentage of female students, students in private schools, Quechua speaking students, and over-aged students, and the proportion of teachers graduated from universities and teachers trained in ISP, explained 94 percent of the variance in test scores between departments. Student-level variables, (namely, age, gender, mother tongue, the availability and usage of textbooks, student attendance and study habits, and parental roles) cumulatively explained 5 percent of the within-school variance in achievement. School-level variables (namely, geographic factors, availability and usage of textbooks and homework assignments, teachers' characteristics, teachers' role, principals' characteristics, and parental role) cumulatively explained 35 percent of the variance in achievement between schools. (See discussion in Background Note 4 for details.) 41 When department-level variables were taken into consideration, in addition to student- and school-level variables, about 12 percent of the variance in math achievement was attributable to differences in characteristics between departments. Within departments, 43 percent of the variance in test scores was attributable to characteristics between schools. Within schools, 45 percent of the variance was due to characteristics among students. The between-school variance in achievement over 30 percent is normally considered as indicators of inequity in learning outcomes. (See Appendices 5.1 anid 5.2 for international comparison). It should be noted that the relative weights of the above-mentioned variables reflect more the imprecision in constructing the explanatory variables in the questionnaires than the lack of predictive power of the above mentioned variables. The relatively low percentage of variance explained at the student-level also reflects the absence of some crucial predictors in the dataset, such as parental educational level, hours of study at home and home resources (which could be proxied by measurable materials such as type of dwelling and sanitary facilities, ownership of refrigerator or telephone). Even at the school-level, where explanatory power was higher, information was not collected on school level resources (which could be proxied by material the school was constructed of, type of sanitary facilities, availability of water, electricity, and library, etc.); government allocation per student in the school and household contribution per student; whether the school is on shift; and student and teacher absenteeism. To inform policy for more precise intervention, it is desirable to collect these variables in the future, as well as to revise the questionnaire. Another limitation to be overcome in futture is the appropriate sampling of rural schools, including single-teacher schools in the sample. Due to these limitation, the results obtained from analysis of determinants of learning outcomes should be viewed as suggestive rather than definitive. That said, the findings are encouraging in indicating potential directions where change could make a difference. Although there are gaps in mathematics outcomes between gender, region, and private and public schools, after controlling for a number of explanatory variables the picture has changed. Students in poor and extremely poor departments performed better than those in non-poor and average departments, holding other variables constant. Some departments were doing a better job in educating over-aged students. Amyara students performed as well as Spanish speakers. Quechua students could perform as well as others if they were not educated in predominantly Quechua schools, thereby indicating that the problem is not with the students themselves. Teachers who have had longer years of service, and teachers who have had more in-service training, were positively associated with higher student achievemenit nationwide. Non-availability of textbooks has been negatively associated withi mathematics achievemelnt nationwide. Parental expectation for better performance in the relevant subject has been translated into higher student performance nationwide. Table 4 summarizes the findings of cross-level HLM analysis to indicate which explanatory variables have positive or negative effects on math achievement and whether these effects varied across schools and departments. 42 Table 4: Summary of Effects Crossing between Departments, Schools, and Students Statistically Effects Effects did Effects Effects did Significant Varied across not vary Varied across not vary Effects Department across Schools across (controlling Department schools for other variables) Between Department Household expenditure per capita + Poor departments + Extremely poor departments + Department % of: Teachers from ISP + Cross-level between Schools And Departments Female students x Over-aged students - x Quechua students - x Teachers' years of service + x Number of training courses + x attended Cross-level between Students, Schools, and Department Female students - x x Over-aged students - x x Quechua students - x x Parental expectation for the + x x subject being tested No text materials x x Note: The coefficients of variables that had no statistical significance were not reported. Source: Background Note 4. Policy implications. These findings point to the opportunity for public policy to make a real difference for disadvantaged students. The policy interventions should be universal where the effects have nationwide impact (that is little variation at the school and departmental levels). These include textbook provision (by extending beyond the primary level which was provided by MECEP to pre-school and secondary level); strengthening teacher pre-service and in-service training; providing incentives for experienced teacher to remain in the profession, deploying qualified and experienced teachers to the rural areas; specific training to teach more effectively to over-aged students; and using the mass media and parents associations to enhance parents' role in supporting their children's education. Where the effects vary across departments or schools, targeted interventions are desirable. These include specific support for schools where Quechua (and other indigenous people) are predominant. This might require strengthening bilingual education and text materials. In better schools, special attention might need to be paid to bring girls and over-aged students up to the standards of other students. Finally, it should be remembered that there was no statistically significant difference in performance between Spanish speakers and Aymara speakers. Future 43 research should find out what factors have enabled them to overcome the barriers facing speakers of this indigenous language. These findings were consistent with those from the literature about the effectiveness of several commonly used policy instruments to improve learning outcomes. These instruments can be grouped into five categories: (i) provision of instructional materials and facilities (such as textbooks, libraries, and laboratories); (ii) increasing the opportunity to learn through increasing instructional time and homework; (iii) provision of teacher training; (iv) increasing teachers' salaries; (v) and reducing class size. Figure 29 summarizes the findings in existing literature. Figure 29. Determinants of Effective Learning in Primary Education: Findings from Literature Review 100T 90 80 C 70i .S60 ;540 U)30 20 10 Libraiies Instructional Homework Textbooks Teachl Teacher LaboTatories Teaciser Class size time knowledge experieice salary School Input Source: Reproduced from World Bank, 1996, based on literature review by Fuller and Clarke It should be noted that the effectiveness of these instruments depends on specific country context and that some measures are administratively simpler to implement (such as provision of a library, or textbooks) than others (such as improving teachers' knowledge). Therefore the graph should not be interpreted as a hierarchy of effective interventions. Nonetheless, it is useful to see the frequency of effectiveness in many settings as an indicator of degree of easiness for implementation. Providing school libraries is found to be cost-effective in nearly 90 percent of the studies, and increasing instructional time (such as lengthening the school day or providing additional instruction in a certain subject) almost equally effective. Asking students to do homework comes third, and provision of textbooks to children who do not have them comes fourth. Improving teachers' knowledge through training is found to be effective in the majority of cases, and having experienced teachers also comes close. Provision of laboratories, 44 increasing teacher salaries, and reducing class size are also effective, but in fewer cases, most probably because they are more difficult to implement.36 Policymakers are well advised to evaluate which interventions are relevant for their country's conditions and to assess the feasibility of implementation and the recurrent cost implications. Conclusion. This chapter aims to address the question of whether expansion of education in the past has come at the expense of quality. A review of indicators on educational access, internal efficiency and quality has found that there has been a trade- off between access and quality as resources have been spread thinly to meet multiple demands, In light of rising premia for education, which signal a growing demand for higher skill levels in an increasingly competitive environment, the serious implications for the employment prospects and future life-time earnings of disadvantaged groups cannot be ignored. This calls for a focus on quality improvement particularly for disadvantaged groups as a centerpiece for education policy. 36 World Bank, 1995; Lockheed and Verspoor, 1991; Harbison and Hanushek, 1992; Postlethwaite and Ross, 1992; Warwick, Reimers, and McGinn, 1989; Tatoo et al. 1990; Fuller and Clarke, 1994. 45 46 Chapter 4. The Teaching Profession The fourth question asks whetlher the ability to contain personnel cost in public expenditure on education has contributed to Peru's ability to extend educational access and what might this and other policies towards teachers affect the sector. Since education is a labor intensive enterprise because interaction between students and teachers is critical to lealning, sound policy towards teachers that can enhance student learning will improve efficiency of resource use. The findings in the analysis of determinants of math achievement in Grade 4 in Peru affirmed the positive impact of teacher qualification, experience and professional development on achievement nationwide. Research evidence from the United States also found that skilled teachers are the most critical of all schooling inputs (in an environment where the needs for textbooks, instructional materials, and facilities have been met).37 These findings underscore that focusing on teachers and teaching is the only way in which an education reform can make an impact in the classroom and help improve student achievement. This chapter reviews the issues related to the teaching profession with the aim of identifying options for improvement. 4.1. Teacher Qualifications and Employment Status Qualifications and deployment. In the Peruvian public education system, there are some 248,000 primary and secondary school teachers (Appendix 2). About 62 percent of these are qualified with titles (con titulos), that is, they have a diploma in pedagogy from one of the 318 tertiary-level teacher training institutes (Instituto Superior Pedag6gico or ISP for short) or from one of the 38 education faculties of a university. The rest (38 percent) are unqualified (sin ti/ulos), that is, they do not have pedagogical titles. They either did not fully satisfy all requirements for the pedagogic diplomas, have other tertiary-level diplomas, or have only completed secondary education. Teachers without titles have lower pay and lower status than those with titles in the profession (Appendix 6.32). There are no statistical breakdowns of the academic qualifications of teachers without titles by level of education. The only information comes from the sample survey of teachers that accompanied the 1996 national assessment. Among teachers of Grade 4 students in the sample, 15 percent graduated from university pedagogical programs, 51 percent graduated from ISPs; 1 percent graduated from Institutos Superiores Tecnicos (IST); 6 percent held university bachelor's degrees, 17 percent obtained their teaching qualifications through part-time professional studies; 1 percent graduated from other programs; 6 percent had only secondary education plus teacher training; and 3 percent had only secondary education without any training (see Table 3 in Background Note 4). 37 According to Ferguson (1991), in the United States, 49 percent of learning outcomes is attributable to homne and family factors (suclh as parental education, income, language background, ethnicity, and location), whereas 43 percent is attributable to teacher qualifications. and experience, and only 8 percent to class size. 47 Having the required pedagogical qualifications is not synonymous with being a good teacher. In fact, ISPs are alleged to be academically weak and also attract poorly prepared students. However, since teachers without titles have lower pay and lower status in the profession, those who remain in the profession signal that they have even lower opportunity cost than the trained teachers. Furthermore, the analysis of the 1996 test confirmed that teachers who were graduates from university and ISPs were associated with higher student achievement than those who were not graduates from these programs. Although the number of teachers with titles seems low, this already represents a dramatic improvement even from five years ago. According to the 1993 census, only 52 percent of teachers in service had titles while 48 percent did not. By 1995, teachers without titles declined to 42 percent. By 1997, they were reduced to 38 percent.38 At present, to be appointed to an authorized pensionable position, one has to have a title and have passed a nationally competitive examination. The rapid growth in the supply of teachers with titles is due to a dramatic increase in teacher training institutions, particularly the private ones. Backgrouid Note 5 discusses the implications of the rapid growth of teacher training inistitutions and the issues of teacher training. While progress has been made in the supply of teachers with titles, the key question is whether there is a particular pattern in the deployment of teachers without titles and its potential impact on quality of education. Accordinig to the 1993 census, the vast majority (68 percent) of the unqualified teachers taught in the rural areas. The MED's 1994 survey of rural and urban public schools in Lima and Cusco found a positive relationship between school size and the percentage of teachers with titles. In large and very large urban schools, as well as large rural schools, over 81 percent of teachers had titles. In medium-sized urbani schools, teachers with titles dropped to 73 percent, while in small rural schools, they declined furtlher to 50 percent. The percentage of teachers trained in regular programs (that is, in ISPs and the education faculties of universities) varied from a high of 85 percent in large and very large urban schools, to 64 percent in small urban schools, to 53 percent in small rural schools (Appendix 4.6). The same pattern is also observed among principals of schools. For example, over 92 percent of principals in very large and large urban schools, as well as in large rural schools, had titles, compared with only 74 percent in small rural schools. Over 86 percent of principals in these large urban and rural schools were trained in regular programs, in contrast to only 72 percenit in small ruLral schools. (Appendix 4.6.) These patterns show that teachers who have had the required qualifications are reluctant to take up hardship positions in remote rural areas. They still go to large rural schools which are located within reasonable distance from an urban area, and which tend to have better school resources, and easier access to health and other social services for 38 In the Inka Region, for example, unlqualified teacheis accounted for almost 60 percent of the teaching force in t 990, but now account for only 32 percent. 48 the teacher and his or her family.39 The same is not true for small rural communities, where the poor working conditions, the lack of opportunities for additional part-time employment or work for the teacher's spouse, and inaccessibility to various social services make it hard to attract teachers with titles. The findings of the 1994 survey of rural and urban schools illustrate just how large a disparity in working conditions exists between urban and rural schools. For example, electricity was available to over 97 percent of large urban schools, 89 percent of medium- sized urban schools, and 55 percent of small urban schools, in contrast to only 29 percent of small rural schools. All teachers of Second Grade who were surveyed in very large urban schools had textbooks, compared to only 91 percent of teachers in small ruLral schools. The nunmber of visits by an inspector from a USE, no more than 1.5 times per year at best, was twice as frequent in urbani schools than in rural schools, reflecting both the difficulty of access as well as marginalization of rural schools. The average number of training courses attended by teachers was much higher for those in urban schools than rural schools (see Appendix 4.6). And yet, there was insignificant earnings differential between teachers who worked in difficult rural conditions and those who taught in urban areas. (See Appendix 6.32.) That is why the turnover rate of rural teachers is much higher than that of urban teachers. For example, the average year of teaching in the schools surveyed was much shorter (2.3 to 3.6) for teachers in small and medium-sized rural schools than for teachers in very large urban schools (5.9), although their average ages were within a narrow range of 32 to 36. For school principals, taking tip postings in smaller schools appeared to be a chanlel for career advancement. The average age of principals in large urban and rural schools was between 45 and 48, but that of principals in small and medium sized rural schools was between 36 and 39. (Appendix 4.6.) Policy to address the inequity in learning outcomes canlot avoid tackling the issue of deployment oj trained teachers to rural areas. This entails providing a larger rural allowance tied to positions in rural schools, anid using job rotation every three years to attract teachers with titles to take up positing in remote communities because it will not be pernanelnt assignmnent. Meanwhile, public investments should be made in rural schools to improve school resources and working conditions, as well as to enable rural teachers to break their isolation and maintain professional contact with other teachers withiin a cluster of schools. Teachers who are going to teach in rural schools should be given additional training on how to handle multigrade teaching. Concomitantly, since many remote areas are populated by indigenous people, it makes sense both in terms of promoting multiculturalism and bilingualism to recruit indigenous teachers for schools in their communities. Given the association of indigenous teachers (except the Aymara) with low achievement of students, this indicates indigenous teachers should be given additional pre-service and in-service training in order to prepare them better for the task. 39 See the cotmpanioni Health Sector Study by the World Bank for the inadequate and low qLuality rural healtlh service (World Bank, 1 999a). 49 Employment status. Teachers are classified either as permanent staff who are appointed to authorized pensionable positionis (APP) (knows as nombrados), or on contract (known as contratados). Contracted teachers are not eligible for pensions, and can be dismissed without severanice pay. There is no relationship between teachers' qualification and emnployment status. APPs are the main means of allocating public education budgets to lower levels of educational authorities. Theoretically, APPs were distributed to each Regional Directorate of Education on the basis of enrollments in each school (one APP for 35 students in the trban areas and 20 students in the rural areas), and incremental APPs are supposed to be allocated annually on the basis of the changes in enrollment. But the organic APPs have not been reviewed for a long time, and the number of annually adjustable APPs has beeni fixed since 1995. Therefore, no systematic relationships seemn to exist between the number of APPs and enrollment, and the budget allocation system has lost its rationale. Since one APP can be used to hire more than one teacher by converting them into conitract or part-time positions,40 there are no precise statistics as to hiow many teachers are in the system or how many are permanenit or on contract. When a Regional Directorate of Education asks for additional APPs on the basis of incremental enrollments in one school, the MED has no information at hand on the possibility of redistributing the APPs in the sanite region or in its USEs, taking into account the APPs in learby schools. The organic APPs are supposed to be filled by appointed teachers through the competitive selection examination, and the annually adjustable APPs by contracted teachers. But when fewer candidates than positionls are selected, the organic positions are filled by contracted teachers as well.4l More than half of the contract teachers are estimated to have titles because most of them are recent graduates from ISPs. Of the 93,000 unqualified teachers nationwide, some 80 percent are estimated to be appointed teachers. In other words, it is possible that there is a higher percenitage of unqualified teachers in permanent positions than those on contract, althouglh in absolute numbers, they are minorities among permanent staff. The nonexistence of a relationship between qualification and employment status is largely due to historical school expansion and demand for teachers. This is probably a major reason why the determinant of achievement study using the 1996 test data did not find any association between teachers' employment status and student achievement, but it did find a statistically significant positive relationship between teacher qualification and studenlt achievement. 40 Nationally selected teachelrs are appointed by the DRE for a specific school on the basis of teachers' preference, scores, and prior teaching experience. Therefore, there is no room for a school principal or a USE director to select the candidates. (However, a principal can hire contracted teachers with the DRE's approval.) 41 In the case of the Cusco Subregion in 1997, a total of 13,674 organizational APPs was budgeted. However, at the end of the year, only 12,320 teachers were appointed and the balance was contracted. In addition, 158 teachers were contracted as provided by the budget for annual incremental APPs, whichi has been fixed since 1995. 50 The deployment of appointed and contracted teachers does not display a clear pattern because political considerations also may enter into the decision. For example, in the frontier areas, teachers in remote rural schools are often made permanent staff irrespective of their qualifications. Their job security, status, and satisfaction are considered vital to national security. In remote internal areas, the pattern appears to hold as well. For example, the analysis of the 1994 survey of rural and urban schools in Lima and Cusco founid that among teachers in small rural schools, 90 percent are permanent staff, even though only 50 percent of them have titles and their average age is about the same as those in large urban and rural schools. By contrast, only 84 percent of teachers in very large and large urban schools are permanent, reflecting largely the fact that 81 percenlt of them have titles. (Appendix 4.6.) In the old system, appointed teachers enjoyed generous benefits and life-long tenure (see Footnote 6 in Chlapter 1 for the various laws that have regulated pensionis). However, the pension benefits are rapidly eroded for new entrants. The new individualized private pension system would enable even contracted teachers to contribute to their retirement benefits. At the same time, the institution of performance review in the public sector may soon erode the job security of appointed teachers as well. Nonetheless, appointed teachers would be entitled to severance pay while contracted teachers will not. Given the rapid erosion of job security even for appointed teachers, the major difference between these two types of teacher may be the status conferred. This difference, however, could have a very negative impact on the morale and commitment of contracted teachers. Therefore, it is important for policy to address the issue of incentives for higher performnance. An optionl is to convert everyone into an open-ended contract with personalized pension plan. This could eliminate the two-tier system and allow performance to determine duration of tenure.42 This may require a Congressional decision to change the law. 4.2. Conditions of Service and Compensation The differences are miniimal in the conditions of service and compensation of appointed and contracted teachlers; they are also minimal between primary and secondary school teachers, and between teachers with titles and teachers who have academic degrees in other professions or post-graduate degrees. This indicates a lack of incentives in the system to encourage commitment, professional development, and higher performance, which could translate into better student achievement. Conditions of service. Both appointed and contracted teachers are obligated to work for 40, 30, or 24 hours per week, but that distinction is largely artificial. Most teachers work about the same number of hours whether they were appointed to do it or not. The difference in pay between 40-, 30-, and 24-hours of weekly work is also 42 This is the option choseln by the World Bank for its staff in 1998 in order to address the inequity of the two-tier system of having pensioniable staff and consultants doing the same job but with different compensation. 51 minimal. There is also no difference between appointed and contracted teachers in working hours. The majority of initial and primary education teachers are appointed for 30 hours, and secondary teaclhers for either 30 or 40 hours. There is no difference in pay between primary and secondary school teachers even though the latter are expected to have higher knowledge of subject content. Principals, deputy principals of all schools, and senior teachers at secondary schools are appointed for 40 hours a week. Teachers have the right to ask for reassignment to a different school in a different location after three years of service, but in reality, it is hard to get reassignment, leading to a lot of dissatisfaction among teachers. Teachers' working conditions have not changed much over time. They usually have two months of vacation in the summer (January and February) and one week vacation in the winter. In remote rural areas, teachers do come to the town to collect their pay checks if they do not have a bank account and to take in-service training.43 On those occasions, they ofteni take off a few days informally. Absenteeism is therefore often higher in rural and public schools than in urban and private schools. The academic year starts on April I and ends on December 15, but the net instructional period is often less than 180 days, as in many countries in Latin America, which is on the low side of the international range. There is also no difference in remuneration between appointed and contracted teachers. Remuneration. There is only one national salary scale for all regions and departments (Appendix 6.19). Teachers under different pension schemes (governed by Law 20530, 19990, and individualized accounts) have different salary structures. Those under Law 20530, who enjoy the most generous pension benefits, have lower salaries, whereas those under the individualized pension scheme have the highest salaries (Appendix 6.32). The basic salaries are adjusted for marital status (unmarried or married with tup to 5 family members), rural allowance (s/45 per month, which has not changed since 1993), alnd three fixed bonuses (s/300 each in March, July and December). The average salary for those under Law 20530 is 627 soles per month, those under Law 19990 receive 646 soles, and those under the individualized pension plan receive 689 soles. Rural allowance, however, is portable, that is even after teachers leave the rural position and teach in the city, they will still be able to keep the monthly rural allowance permanently. This practice should be discontinued. Refornm in salary scale should raise the rural allowance, but link it to the position and not to the person. There are five scales of salaries (I to V) for promotion. However, the differelnce between scales is insignificant (onily some s/.12 per month). There is only a 10 percent difference in remuneration between the top grade for 40 hours work and the lowest grade 43 To give an example, in Loreto Department, a teachier who has to work in a remote community near the frontier withi Ecuador has to travel 20 days upstream by boat from Iquitos to his/her posting and about the same number of days downstream to Iquitos. He/she usually has a few monthis' advance salary, and picks up his/her pay check in lquitos three times a year. This means 120 travel days every year. During the teachers' travel to Iquitos, students would not have classes. Even though some of the travel time coincides with vacation days, a significant amount of time is lost. This again accounts for why the learning outcomes of rIural childreni are lower than that of urbani students 52 for the same amount of work (Appendix 6.19). Promotion between scales has been frozen since 1991. Consequently, the majority of teachers are at levels I and II, and do not have strong motivation for better performance, professional development, and promotion. Appointed teachers receive their salaries through a bank account, and contracted teachers by checks issued to them. The salaries for teachers without titles have also five scales (A to E) depending on their qualifications, but the difference among scales is more insignificant than among teachers with titles (Appendix 6.19). A is for those who completed pedagogic studies but have not earned the title yet. B is for those who have completed higher education studies with professional titles in other than pedagogy. C is for those who have not completed pedagogic studies at higher education level. D is for those who have studied at higher education level without any university titles. Finally, E is for those who have only secondary level education. The salary differential between a teacher without title who works for 40 hours and is in A scale and his/her counterpart in E scale is only 39 soles per month. The salary differential between teachers with titles and those without is more substantial, ranging from 80 soles to 170 soles per month. However, the emphasis on pedagogical titles at the expense of downgrading those who have other higher education degrees (scale B) discourages talent from other fields from entering teaching. There are no statistics on the qualifications, conditions of service, and remuneration of private school teachers. Given the uneven conditions and quality of private schools, it is likely that teachers' salaries also vary a lot. Those private schools that serve the poor are most likely to pay teachers on an hourly basis (about 5 to 7 soles per hour), and teachers probably have to have two jobs at the same time in order to make ends meet. However, according to the preliminary results of a 1997 survey of 1,000 private and public school teachers, salaries in the top private schools may be as high as 4,000 soles per month (Saavedra, 1999). Teachers' average monthly salary of 646-689 soles is more than twice the minimum wage. This amounts to an average annual salary of about US$2,903-3097, or 1.5-1.6 times the GDP per capita, which is lower than the 2 to 4 times prevalent in other countries at a similar level of development. This is one of the reasons why Peru has been able to provide such broad education coverage at such a low level ofpublic expenditure. Teachers' salaries took a hard hit in the 1980s, but their remuneration in real terms has steadily recovered in the l990s (Figure 30). The key policy questions are not whether salaries were high or low relative to other countries, or whether the purchasing power has recovered, but (i) how teachers' salaries fare relative to other professionals who have similar years of tertiary education, which would impact on the ability of the sector to attract and retain academically capable individuals; (ii) whether the salary structure provides incentives for teachers to take up hardship posts in rural areas, to continue to develop their professional skills, and to improve student learning, and (iii) what the recurrent cost implications for restructuring the salary scale are. 53 Figure 30. Remuneration of Teachers in IReal Terms, 1990 to 1997 7u 0 12 500 4 0 V, 300 - -- CO 20 0- 20 _ __ _ __ __ _ __ - _ __ __ _ __ __ _ __ __ _ _ _ .__ _ _ 10 - _ _ _ __ _ _ _ , .__ _ _ _ ,__ _ _ _ __ _ _ _ _ _ --- -j- -,- _ _,__ _ _ _ _ _ _ _ _ .. I-- 19 9 0 1 9 9 1 1 992 1 9 9 3 1 994 1 9 9 5 1 9 9 6 1 9 9 7 1 9 9 8 1 'W ith data up to AprilI9a8 Source: M nistry of Educa0 n Saavedra (1999)44 found that teachers' relative position eroded by 30 percent between 1986 and 1992, but other professionals' relative earnings declined by 16 percent between 1992 and 1996, so that for the whole decade 1986 to 1996 teachers' earnings deteriorated by 10 percent in comparison with other professionals. This salary differential between teaching and other professions could induce the best and most adaptable teachers to leave the profession to take up jobs in other sectors. The data compiled by INEI show that real salaries of all sectors declined sharply between 1970 and 1990, but made some recovery in the 1990s. Average teacher salaries lost rmore than the private sector as a whole, but fared much better than the administrative staff of the public sector in general and nurses.45 This has not even taken into consideration the two months' vacation enjoyed by teachers, which is not available to other employees in public or private sectors. In summary, although teachers salaries lost markedly their real purchasing power in the 1980s in comparison with the private sector, they gradually recovered after 1990 and fared better than the public sector as a whole. Nonetheless, better salaries in the 44 Saavedra compared the earnings of teachers and other professionals with university-level education, using panel data from the household surveys done in 1986, 1992, and 1996. The work is in draft. 45 The current mission's findings partly modify the conclusion of Psacharopoulos's 1996 study using household survey data during the 1980s; that is, that teachers are not underpaid in comparison with all other occupations. In fact, teachers are underpaid in comparison with the private sector workers, but not with the public sector workers. The current mission findings also modify the conclusion of the November 1994 Staff Appraisal Report on the Primary Education Quality Project, which states that teachers have suffered the worst fall in income among all categories of government employees. In fact, teachers were better off among all government employees in terms of falls in remuneration. 54 sector could induce migration of the more competent teachers to other sectors. (Figures 3 1-34.) Figure 31. Index of Remuneration of Government Figure 32. Index of Private Sector Salaries in Metropolitan Employees, 1970-1997. Base August 1990=1 00 Lima, 1970-1997. Base August 1990=100 1 800 1200 1500 900 900 600 _ ____ 600 -- ----------- _ _ __0 300 O -1- --1-- FI- I I I -I- I-1-I-~-! -I-I-I-1-I I i-1- p t o t 8 g g g t 340 ,4 40> , 40 , 0 ,° , 0 40 et eq , Figure 33. Index of Purchasing Power Figure 34. Index of Purchasing Power of Private Sector Salaries 1972, 1990, of Public Sector Salaries 1972, 1990, 1997 1997 120 - 100 ~_ __ _120 - -- - _ 60 J --20_ 83 ~~~~~~~~~~~~0l 1970 1980 1990 2000 1970 1980 1990 2000 Figure 31-34 Source: INEI The unattractiveness of teaching is particularly serious for secondary education, where growth is expected to occur. The lack of salary differelntials between primary and secondary school teachers does not reward knowledge of subject matter, which is more intense in secondary education than primary education. Furthermore, the current salary scale that rewards teachers with pedagogical titles more than those with other degrees (such as in arts and science or with postgraduate training) would not attract talent from other fields into the profession. In the short run, reforming teachers' salary scales to make remuneration to teachers with bachelor or masters degrees in nonpedagogical fields on a par with those with pedagogical titles will be the first step to widen the pool of talent in teaching. This will, of course, have implications for the type of training provided by ISPs, and the basic requirements for teaching. In the long run, the optioln should be considered of requiring all secondary school teachers teaching above Grade 7 to have the first degree in a subject area, and have additional pedagogical training. The salaries of these teachers would reflect this better academic preparation. 55 4.3. Incentives and Accountability Many countries in the world, including OECD countries and in Latin America,46 are pursuing education reform in order to meet the challenges of global economic initegration and technological clhanges of the 21St century. What distinguishes reform of the 1990s from those in previous decades is the focus on learning outcomes. In varying degrees and at different paces, reforming countries are embracing a set of principles to overhaul their education systems. These are the needs to set standards for student learning, to set standards for teaching, to improve teacher education and professional development, to provide incentives for teacher knowledge and skill upgrading, and to encourage schools to organize for learning. The broad scope of the changes entailed makes this wave of reform one of the most ambitious in the history of education. In Peru, MED is in the process of building Up a meritocratic-based education sector with some of the elements for accountability being set in place. For example, the first national assessment test was implemented in 1996 (Background Note 4); a plan to modernize teacher pre-service and in-service training in primary education was also piloted in 1996 (Background Note 5); and a national competitive examination was introduced to select teachers for appointed teacher positions in 1997. The examination has the dual purposes of selection and quality assurance. Its setup was timely, given the rapid growth of private ISPs, and the diverse curricula offered by ISPs and the education faculty of universities. In the first administration of the exam, some 96,000 qualified teachers applied for the 34,000 appointed teacher positions, but only 12,000 candidates or 12.5 percent passed the exam. The results of the second examination are very similar to the first one. Since the examination is not a criterion reference test and the validity of the test has yet to be established, it is premature to judge whether the low pass rate reflects poor preparation of teachers or an admi-nistrative decision. To truly test tiLe skills of teachers, there is room for modification. While Peru is headed in the right direction to pursue a systemic reformn, it is necessary to ensure that a few key building blocks are in place.47 Specifically, what needs to be done is as follows: 46 In the 1990s, Latin Americain countries have devoted considerable financial and human resources to reform the education sector. However, the recent experience of imiplementationi of educational reforms throughout the region suggests that these programiis have not been able to transform substantially and extensively the behaviors of teachers, the main actors, in the school level where it matters most The Second Summit of the Americas held in Santiago, Chile, in 1998 outlined a Plan of Action committing all countries of the hemisphere to new reform efforts, including an increase in the level of professionalism amnong teachers that combines pre-service anid in-ser-vice traininig, and the development of incentive mechanisms tied to updating their skills. 47 In the United States, 5 years of work has gone into improving the instrument that are making it possible today to pay teachers for what they know and do. 56 * Overall goals and standards for students, and assessment of outcomes. As a first step, the education system has to have a clear set of goals spelled out in detail about what it expects students to know and be able to do. Standards are not the same as curriculum because the latter is usually very content- and topic-specific. Standards are more about skills and competency that could be developed in the course of learning various subjects. Without standards for students, it is impossible to set standards for teachers or to measuLre whether the goals have been achieved. To be able to measure outcomes, an assessmenit system must be in place, and test validity and reliability must be secured. The results of the assessment should be fed back not only to policymakers, but also to various levels of administrators, teachers, and parents so that they have a clear sense about their school's relative performance, and can gauge their value-added efforts. Peru miglht want to start with standard setting and provide the results of the 1996 assessment at least to Regional Education Directorates. * Standards for teachers. It is important to spell out the content teachers need to kniow, and the specific kinds of skills and behaviors that constitute good teaching need to be spelled out in meaningful detail. This means providing written documents on what excellence in teaching in a given subject and in a given level (such as science teaching in primary education) would be. Teachers would be assessed against these standards. This might be the next step for MED. - Aligning pre-service and in-service training programs with these standards. Virtually no country in the world has been able to do this yet. However, steady progress has been made. The development of meaningful teacher standards (such as PRAXIS and INTASC in the United States) is beginning to make an impact. Peru can shorten its development time by building on these materials, and by adding what is refevant for their conditions. 3 Teacher assessment. Often, this begins with the use of paper and pencil examinations to test teacher's knowledge of subject content, or for certification of teachers. This process has beguni in Peru. However, even with the most carefully designed test items, this form of examination alone is inadequate to assess a broad range of skills (such as classroom management, pedagogical repertoire, and team work with other teachers) needed in order to be an effective teacher. New innovative instruments being experimenited with in OECD countries include peer examinations of a portfolio of the teacher's work, videotapes of his/her teaching, interviews, competency tests, and other means to ensure a truly comprehensive assessment of a teacher's demonstrated competence as well as knowledge base. Such assessment is predictably expensive, but probably will be cost-effective in the long run when it can positively affect student learning. Assessment techniques are the vehicle to measure progress and inform any corrections which should be made by the teacher and/or system. In future, Peru may want to start by modifying the methods of recruiting teachers to fill APPs by requiring candidates who perform well in the competency test to demonstrate classroom teaching to determine suitability for teaching. Similarly, the 57 recruitment of principals should be made beyond a paper and pencil test by having candidates visit schools anid recommend plans for school improvement.48 * Reward teachers' knowledge and skills individually and schools collectively. How the incentive in the systein is structured affects behaviors of teachers. Teachers should be rewarded for what they kniow and do-as measured by objective and multifaceted performance assessments-rather than for how long they have been in service. Promotion from one level to the next should be based on demonstrated higher level of competency, not seniority. This could provide incentives for teachers to invest in their professional development, which is not necessarily restricted to in- service training, but could include doing more reading at home, more reflection on their practice, or networking with other teachers to keep abreast of the latest developments in the profession. The results to be measured nleed not be restricted to student achievement, either, but could be extended to broader student inltellectual, artistic, athletic, and moral development, and parent involvement. Whatever it is, it must be measurable, preferably in a value-added way. This is not easy, and that is why, despite the desirability of including more diverse measures, most school reward systemiis in countries that implemenit them are basically driven by student achievement tests. At the same time, there should be rewards to schools collectively in order to encourage collaboration among teachers. Rewarding schools may not require more than one to two percent of the total education budget. This should be combined with public recognition of excellence. The amount should motivate teachers at the margin, but not be so central that teachers will focus their work exclusively on "the test" or whatever else is being measured.49 Building professionalism among teachers could be the center piece of reform, and standards should drive the change in each stage of professional development. The best way to build teacher professionalism is through the work of groups of teachers reflecting on what constitutes excellence in the practice of teaching. Since some countries have pionieered the work, Peru cani benefit from their experience without having to reinvent the wheel. It should be acknowledged, however, that this requires changing institutions and culture and will be the most difficult task to accomplish. 48 In New York City's District Two, which has attained nationial fame for being able to improve overall studenit achievement and reduce the variability of learning outcomes of an entire cohort, candidates for principal positions are asked to visit a school or watch a video tape of a school and tlhen tell the hiring commnittee what problems they have identified and what solution they would propose. This would ensule that the principals selected have practical experience of runninig a schIool and of providing instructional leadership. 49 Colombia has intiroduced an inn1ovative incenitive system to reward and recognize good teachinlg. Students and parents in every school are asked to elect their best teacher based on a set of criteria provided by the Ministry of Education, while every municipality will elect the best schools, again based on1 provided criteria. Then the departments will elect the best schools from the list, based on the relative positions of the schools selected by the lower levels. The criteria focuses on1 efforts, rather thani existing conditions. Then there are awards for the best schools in the nation, the best schools in the departments and municipalities, and the best teachers. Both teachers and schools would receive public recognition and a cash award. 58 Summary. The answer to the fourth research question is a affirmative. However, this introduces a policy dilemma towards teachers. Because of the enormous recurrent cost financing implications, education ministries in the world are often constrained by what they can do about salaries across the board, particularly in view of the consensus on working to extend access to basic education. Yet it is important to provide incentives to attract and retain competent people in the profession because teachers are critical to qualitative improvement, as well as quantitative expansion of the system. The issues that need to be addressed in Peru are (i) the disparity in qualifications between rural and urban teachers; (ii) the lack of reward for subject matter knowledge, particularly in secondary school teaching, and the disincentive in the salary scale and promotion criteria for those who have had university and postgraduate education in fields other than pedagogy to enter and remain in the teaching profession; (iii) the morale problem related to different statuses of employment between appointed and contracted teachers; and (iv) the lack of incentives for higher performance. The first issue calls for increasing the salary differential between rural and urban teachers to compensate the rural teachers for hardship positions, combined with a job rotation system so that they have an opportunity to return after a few years, with investment in rural school inputs to make teaching there more attractive, and with recruitment of indigenous teachers into teaching in their communities. The second issue calls for (a) in the short run making the salary scale of teachers who teach in secondary education higher than those who teach in primary schools, and also raising the salaries of teachers who have university degrees in other subject areas at least to a par with teachers with pedagogical titles so that they do not have a second-class status as teachers without titles; and (b) in the long run changing the academic requirement for secondary school teachers with stronger emphasis on subject matter knowledge. The third issue calls for unifying a two-tier employment status. The fourtlh issue really requires the support of a systemic reform geared to establishing an accountability system that rewards group and individual performance, preferably based on value-added efforts. This last will be a long- term endeavor anid can only be done when tlhe assessment tools are perfected and a culture of evaluation is accepted. 59 60 Chapter 5. Second-Generation Reform This report begins by posing the puzzle as a guide to its inquiry of why Peru has been able to have high education participation rates with a relatively low level of public spending on education. After reviewing the issues in the sector, the report found that, progress made in Peruvian education is attributable to relatively equitable public expenditure that focuses on basic education; the ability to contain salary cost in the sector; and high value attached to education by Peruvian households and their high level of spending on education. However, expanding access under extreme resource constraints has come at the expense of quality. The large gap in school survival rates between the rich and poor, and the rural and urban population; the large between-school variance in student achievemenlt; and the rising returns to higher levels of education signal that further policy changes can make a real difference, particularly to disadvantaged students. These changes may be loosely considered to be the second wave of reform because they will build on the foundation laid by the first-generation reform that rationalized the public sector, balanced the budget, and mobilized private resources. These changes would help realize the country's aspiration of having a highly educated citizenry to meet the challenges of the 2 1st century. (Table 5 discusses the implications of the first generation of reform. Table 6 summarizes the objectives, issues and suggested measures for the second wave of reform). Peru thus find itself at a crossroads with respect to education policy. The current situation, which partially reflects the success of the first generation of reform, represents important accomplishmenits. That path could be continued. A second generation of reform. however, would take a path of focussed commitment to improving education as a more central goal in itself and means for accelerating growth and reducing poverty. The following paragraphs summiarize tlhis report's conclusions concerning the potential, the content, and the cost of second generation of reform. 5.1. Improve Equity Although public expenditure on education on the whole has been distributed relatively equitably, poor households still have disproportionately heavier financial burden than rich households for their children's education. The disparity in household spending has led to inequitable learning outcomes. The indigenous people are among the poorest in the country. To equalize opportunity, resources have to target the disadvanitaged groups. The options include a range of supply- and demand-side measures. On the supply side, the Ministry should extend the provision of a class set of textbooks to all levels of secondary education and supply anl additional class set of enrichment reading materials tlhroughout basic education. Given the major difference made by textbooks (as revealed in the 1996 test), this should be the first priority. Since the Ministry is already experienced in the development and provision of textbooks to primary education through the MECEP project, it is both easy and logical for it to extend them downward to cover 61 pre-school and upward to cover secondary education. Since this intervention does not involve changing the administrative or finance structure, it is feasible and achievable. This might raise achievement and help retain niore children in the system. Meanwhile, the Ministry should also explore the feasibility of providing cost-effective educational teclnology such as interactive radio to supplement classroom teaching, or audio materials to assist second-language acquisition to support multigrade teaching. A number of countries in the world have experience with using interactive radio to reach rural communities. Some of experiences have been properly evaluated and found to be highly cost-effective when children are tested for what they learn in comparison with those who are in traditional classrooms. Given Peru's difficult terrain, radio sets can be given to families who have school-age children but live in remote communities. An estimated 6 percent of total studelnts are in single-teacher schools in remote communities. Although radio broadcasts can be used for school instruction, the best use of this medium is during adverse conditions, such as rain, snow, and flood, when children who cannot travel to school can stay home and still learn the lesson. When this is done in combination with programmed text, this will help them overcome the problem of their own and their teacher's absenteeism. Parenting education can also be broadcast by radio so that parents get more involved with their children's education. Specifically helpful to indigenous children is the expansion of bilingual and multicultural education through strengthening teacher training programs in these areas, as well as in multigrade teaching, and recruitment of indigenous teachers to these programs through affirmative action and scholarship. Arregui et. Al (1994) found that the vast majority of teachers are employed in the department they were trained in. Therefore, ISPs in departments heavily populated by indigenous people should offer bilingual and multicultural programs that are tailored to the groups within their jurisdiction. At present, offerin-gs in these programs in ISPs are extremely limited. To seriously improve the learning outcomes of indigenous children, the central government should support such programs through its allocation of resources. Also, given the findings from the 1996 test that showed a relationship between indigenous teachers, except Aymara, and lower student test scores, it signals that indigenous teachers need better preparation during their training. Therefore, not only should there be bilingual education teacher training programs, but there should also be some form of compensatory education in ISPs to assist indigenous trainees wlhose performance might be weaker than average. The training of indigenous teachers needs to be complemented by the provision of bilingual textbooks and educational materials including interactive radio or audio and video tapes throughout primnary education. Currently, the resources that have gone into these programs are negligible, and that is why the impact has yet to be demonstrated. Other countries that have adopted bilingual education, such as Guatemala, have positive results. It is worth examining the approaches in these countries in order to inprove on them. Given that Peru has many indigenous groups who speak different languages, and that indigenous communlities in the Amazon regions are small and dispersed, such intervention is predictably expensive. Nonetheless, given the cost of marginalizing indigenous people, the benefit of poverty alleviation and social cohesion is high. 62 Although it is unrealistic to develop bilingual texts for all groups within a few years, this should be a long-term project, with targets to be met within a medium term time frame of, say, a decade. Training of teachers for rural schools should also emphasize multigrade teaching. Again, this should be complemented by the provision of program materials. The Govermnent is already planning to provide distance learning through educational television to enable children in remote areas or out-of-school children access to secondary education. While this is a commendable move, it should be noted that a range of support measures, such as face-to-face tutorials, needs to be put in place in order for distance education to work. Otherwise, dropout rates could be a high. To ensure that all children learn the requisite skills, formative and summative assessment could be used more systematically and frequently. Those students who are falling behind should be provided remedial education during the school year as well as during holidays. The Ministry has adopted an automatic promotion policy from Grade 1 to Grade 2 in order to improve the promotion rates and reduce dropout rates. The impact of this measure on learning outcomes should be evaluated. Given that automatic promotion policy has been associated with failure for students to acquire the requisite skills in manly countries, timely comnpensatory education might be a more cost-effective intervention in the long run. This might benefit particularly disadvantaged children. Another important issue is poor nutrition and health of children in poverty which has contributed to under achievement in many countries. Although this study on education has not devoted much discussion on the topic, the companion World Bank study on health has examined the issue (World Bank, 1999b). Addressing the health of school-age children either through services provided directly through the schools, or tlhrough targeted publicly financed insurance such as the on-going Seguro Escolar, would improve attendance and learning. At the same time, it is desirable to consider targeted expanlsion of the school feeding program (such as Desayunos Escolares), given initial positive findings of favorable outcomes with respect to nutrition, health and attendance (Pollitt, Jacoby, and Cueto, 1996). Expansion of the school feeding program would need to be weighed against typically high costs of such programs, as well as the feasibility of reaching schools in remote areas. Obviously, such multisectoral interventions require a much more coordinated approach between ministries such as MINSA, PROMUDEH, and PRES. Demand-side financing measures such as scholarships and grants for indigenous or poor children should be explored to enable them to defray the direct costs of education. The mechan-ism for distributing such scholarships or granits needs to be worked out carefully to prevent abuse. An indirect mechaniism is to give schools that have an indigenous enrollmelnt exceeding a certain percentage an additional per student instructional grant based on attendance. The money could be used by the school for purchase of instructional materials, to provide compensatory education, or to subsidize students' clothing or transportation costs. 63 5.2. Enhance Quality A key measure of qualitative improvement is reduction in variance in student achievement. The 1996 test shows that textbook availability and usage, homework assignments, the characteristics and roles of teachers anid principals, and parental role and expectations can reduce between-school differences in learning outcomes. Besides the interventions through textbook provision and teacher training, educating parents about good child-rearing practices and the positive effects on achievement of school attendance, after school studies, and parental involvement in their children's learning by means of the mass media could also help to improve learning outcomes. The policy of universalizing early childhood education to enhance students' school readiness might help reduce late entry and repetition and reduce the between-student differences. Grants for compensatory educationi or to defray the direct private cost of education to facilitate attendance of indigenous students and girls, who tended to have lower average math scores, may also make a difference. These intervenitions could only have maximum impact if they are accompanied by systemic reform that focuses on standards. The necessary components include (i) setting standards for student learning, (ii) settinig standards for teaching, (iii) strengthening teacher education and professional development, (iv) providing incentives for teacher knowledge and skill upgrading, and (v) encouraging schools to organize for learning. In other words, making the career of the teaching force more like that of other professions should be the center piece of reform, and stan-dards should drive the change in each stage of professional development. Part and parcel of this reform is to make available indicators ol performance by school (test scores; repetition, promotion, and dropout rates by grade) available not only to the DREs, USEs, schools and teachers, but also to parents, students, and the public. This will allow the families and public to benchmark the schools' performance against other schools. This will generate pressure for improvement, and will also build the groundwork to set up an accountability system. At the same time, studies on1 determinants of student achievement at all levels of education could be used to decide what qualifications should be required of teachers and at what level of education. This kind of study would be very similar to the one reported in Background Note 4. If teachers graduated from universities and from ISPs are found to be associated with higher student achievement even at Grade 4 levels, relative to teachers who entered the profession through a different ladder, then, it is very important to ensure that teachers are appropriately qualified. To ensure that qualification matches all the requisite competencies to be a teacher of a given grade and subject, teacher pre- service anid in-service training should be examined closely and reformed to ensure high standards. The broader issues that need to be addressed are: (i) the disparity in qualifications between rural and urban teachers; (ii) weakness in subject matter knowledge of teachers, particularly in secondary education; (iii) the morale problem related to different status of employment between appointed and contracted teachers; and (iv) the lack of incentives for higher performance. The first issue justifies increasing the salary differential between rural and urban teachers to compensate them for hardship positions, conmbined with a job 64 rotation system so that they have an opportunity to return after a few years, with investment in rural school iniputs to make teaching there more attractive, and with stepped up efforts to recruit indigenous teachers into teaching in their communities. The second issue calls in the short run for raising the salary scale of teachers without title but who have university degrees in other disciplines to the same level as teachers with titles in order to enlarge the pool of teachers with stronger subject matter knowledge. In the long run, it requires changing the academic requirements of secondary school teachers and reforming ISP cuiricula. The third issue calls for unifying a two-tier employment status to make open-ended contracts, with the tenure determined by evaluated performance. The fourth issue really requires the support of a systemic reform geared to establishing an accountability system that rewards group and individual performance, preferably based on value-added efforts. Given that data on teachers are not available, it would be useful to conduct a census of teachers, as part of the school census to be mentioned below, to review their age, qualifications and specialized areas, experience, subjects taught, and various types of comnpensation received. This would provide data that could serve as the basis of a review of the supply and demand for teachers and the financial implications of increasing the salary differentials between urban and rural teachers, tying promotions to demonstrated competency, and providing moonetary incentives to reward schools. Colncomitantly, it is necessary to review the legal framework anid incentives for teachers in private and public schools so as to provide a benchmark for improvement. Inevitably, the policy towards teachers and their pre-service and in-service training will make a lasting difference in quality. These measures require institutional and cultural change, which take time. Consensus with stakeholders (namely, teachers) needs to be reached in order for the reform to take root. Therefore, as technical measures need to be put in place (suclh as spelling out teaching standards for each subject and grade, evaluationi tools, and assessment of fiscal impact of changinig salary structures), consensus building process should be set into motion by reaching out to teachers, NGOs, and the business comm unity. 5.3. Improve Efficiency of Resource Use The review of public expenditure in this report has only been able to access data down to the departmenltal level, but not to the USE level. Hence, the report was unable to evaluate the actual unit cost by level and by urban and rural areas, across departments. Even the MEF does not have the expenditure data; it only has the budget figures. To piece together a complete pictLre of the nation's public spending pattern, particularly to evaluate the equity and efficiency of resource use, in order to adjust policy, it is of utmost importance that monitoring continues to talce place to cover the following areas. * The trend of public spending on education over lime gross and net of pension. It would be desirable to update and extend the analysis of public expenditure on education to cover not only school-related expenditure, but also education- related subsidies to households (such as school health programs financed by the 65 Ministry of Health, school feeding programs under the PRES), as well as other education-related expenditures unider PROMUDEH (such as early child care and literacy programs). • It is equally important to monitor differential spending by urban and rural areas at the USE level so that interventions can be designed to equalize the resources allocated to different schools. This should take into account department's owIn resources and capital expenditure allocated to the USE level by PRES (through FONCODES). The growth of inequity in allocation of public resources should be watched closely. * Indicators of learning outcomes (e g test scores) would be logical areas for moniforing so that input measures can be tied to outcomes. The existing questionnaire does not contain questions on school-level resources and public allocation per student. Expansion of the questionnaires is desirable to cover these and other variables on family baclcground (parental education, income, family resources, amount of household expenditure on education), study habits, and teacher characteristics, to assess deteTminants of achievement. It would help analysis if more questions are constructed for obtaining continuous variables rather than categorical variables. Studies of determinants of achievement would be helpful to identify effective policy intervenltions to help disadvantaged schools. - Monitoring of household expenditure on education would be another key areas to follow up Given that INEI conducts such a survey annually, the data source for such an undertaking is available; only analysis needs to be undertakenl. Closer cooperation between INEI and MED would help improve the questionmaire for data collection so that the education portion contains the relevant questions that can address issues for household finance. For example the Cuanto dataset used by this report merges tuition fees with transportation and lunch. This does not allow assessment of the impact of fees alone on a household's decision to send their children to one type of school versus another. The question of extra tutoring should also be included because this area is where the rich tend to spend much more than the poor and it provides a proxy as to how much additional money is needed to ensure desirable learning outcomes. Analysis of the household survey would enable the Ministry of Education to monitor distribution of public expenditure by consumption or income quintile (Lorenz curve), changes in the elasticity of demand (Engel's curve) for education, and private and social rates of return to education. Currently, responsibility for all aspects of education is fragmented across ministries and institutions-MED, universities, regions, decentralized institutions, PRES, as well as PROMUDEH and MINSA. There is an urgent need to improve the coordination of educationial policy and financial matters between these budgetary entities. It is recommended that a sectorwide coordinating body be established that meets at least quarterly to review overall education policy, performance indicators of each subsector, 66 and intrasectoral allocation of resources (gross and net of pension), in order to ensure consistency of overall education policy and finance. The minutes could be made available to all ministries concerned to keep them informed, as well as to CIAS to improve intersectoral coordination. Given that there is no accurate information on how many teachers (both appointed and contracted) are in the system, it is very difficult to assess whether resources have been used efficiently. At the samne time, the lack of information in school-level finance in both public and private schools impedes the formulation of policy towards regulating private schools to ensure standards and safety, and towards expanding secondary education through better use of private-school capacity and resources (such as the use of vouchers to buy places in private schools). The combined need to have accurate information on teachers anid school-level finance in public and private schools, therefore, calls for conducting an education census. This needs to be taken at the school to obtain information on schools, teachers, and students. Information could provide a useful database to map out strategy to expand secondary education and to improve quality of primary education, as well as to explore options for formula-based funding. Currently inflexible criteria for allocating resources based on the number of teachers on perrnanent payroll in 1995 without responding to the reality of changing student populations is not an efficient mechanism in the long run. This inflexible criteria has not taken into account the rural to urban migration antd would also disadvantage the populations with very high birth rates, namely, the indigenous and low income populations. It is desirable to evaluate the option of using capitation grants in large public schools (but using a different formula for small rural schools) as a basis of allocations to reflect the reality of chatnging student populations and to make allocation decisions transparenlt. Schools should be given the discretion of how best to use the capitation grants in terms of purchasing a mix of inputs, such as hiring teachers, buying instructional materials, or inistalling lab facilities. In addition, certain categorical grants should be provided to regional education directorates for them to earmark assistance for disadvantaged areas/schools/persons. Increased international competition and technological change have led to growing demand for a higher level of skills in the labor force. This has translated into rising private returns to higher education, which will fuel further demand for higher education. To properly support higher education and ensure quality, it is important to improve the transparency of the funding mechanism, with incentives to reward efficiency and quality. as well as to share costs with students, who are the chief beneficiaries of their own education. Increased cost sharing also needs to be supported by student financial assistance such as student loans to ensure the academically deserved will not be disqualified due to finanicial constrainits. Although this report does not cover higher education, given that policy and expenditure on higher education impacts on lower levels of education, it will also point out areas for further investigation in higher education. In-depth review of funding of higher education is desirable to assess options for introducing funding formulas for 67 higher education (such as full-time equivalency based allocation) with incentives for improving efficiencies (such as that a certain allocation is based on graduation rates within a certain time frame), and examine the adequacy and impact of cost sharing (such as estimating the elasticity of demand and a survey of students' financial situation and expected earnings after graduation). This exercise should involve MED, MEF, and higher education institutions and inform overall education policy of the country. 5.4. Conclusion The 1993 Constitution enshrines the principle of compulsory and free pre-school, primary and secondary education. To the extent that there is strong evidence in many countries of the positive impact of pre-school on subsequent student behavior and achievement, it is educationally very soulnd for the Ministry to include pre-school in basic education. As for secondary education, the projected growth of the cohort between 12 and 15 years of age in the first 15 years of the 21"t century highlights the need to address their educational needs (Figure 35). Given the association of high crime rates and poor, young males in many countries, the education of adolescents will provide many unmeasurable social benefits. Population projections by the World Bank estimate that the cohort of ages 6 to 1 1 (the primary education age group) would rise modestly between the present and 2010 and then decline afterwards, while the 12-to-15, 16-to-17, and 18-to-20 age-groups would increase throughout the first two decades of the next century. There is likely to be increased pressure for education resources, including demand for qualified teachers in secondary and tertiary education. The reprieve provided by the declining primary school age population would come only after 2010. If nothing is done immediately, many cohorts of students would miss their educational opportunity and sink into poverty. Fig, 35. Estimates and Projection of School-Age Population, 1995-2020 4000 . -..."-- 3500 - - 3000 ° 2500 - - Age=5 X __ __Age = 6-11 v 2000 Ai - _ - - -Age= 12-15 o 3Age = 16-17 Xo1500 - - - Age = 18-20 0 1000 1 _ 500 -- *** - -- - _____ ---+**** ++* 1995 2000 2005 2010 2015 2020 68 What miglht the resource requirement be to meet the Constitutional Mandate of universal basic education? Based on a rough estimate, it would probably cost the country an additional 2 percent of GDP, net of pensions, if the minimum standards are to be achieved for all students ot all grade levels. This would still be within the range of the regional average of the Latin American Region. This would enable the government to simultaneously improve internal efficiency and extend access to the, out-of-school population in order to ensure every child the opportunity to acquire the skills of a complete secondary education cycle. This would also raise per student spending. Although the exact costing could only be done after knowing the Government's input mix in providing universal coverage, simply relying on efficiency gains or shifting resources in the margin could not provide the resources necessary to meet the Constitutional Mandate. Many countries have committed far more public resources to education than has Peru, but without achieving universal coverage for basic education. For these countries, increasingly binding fiscal constraints and continiued needs to expand coverage sharply constrain the policy agenda. Peru, in contrast, has positioned itself to initiate a major drive to consolidate equity gains while improving quality. The country has indeed reached a crossroads concerning education. The status quo reflects already substantial progress; one direction continues that path. The other direction would reflect commitment to a re-oriented human resource strategy for poverty reduction and economic growth. This path of second-generation reform will require, over time, substantially increased public expenditures on education. The increase from 3 percent to 4.5 percent of GDP net of pension expenditures (to the Latin American average) is, for Peru, feasible given its fiscal discipline and generally improving international relations. The recently signed peace accord with Ecuador provides an unprecedented opportunity to shift resources from defense to investment in human capital. By increasing public expenditure levels to only the Latin American average, Peru has the opportunity to markedly enhance the intellectual ability and competitiveness of its labor force within a generation. No policy challenge is more significant. 69 Table 5: First Generation Reform and Its Implications for Second Generation Reform First Generation Public Finance Household Equity, Quality, The Teaching Reform Finance and Efficiency Profession Rationalization Measures to There is no Could be negative Negative effect of public service, contain cost immediate impact in the short run, on morale in the increased use of provides the on household but should short run. Fairer contract staff in starting point to finance of improve quality in to offer an open- the central improve efficiency education. But if the long run, if an ended contract ministries, and in the use of these measures accountability for all and use privatization of resources. release more system is properly performance to pensions. Potentially, public public resources to set up. determine resources can be improve quality or duration of used to improve extend access, employment. quality or expand households will Need to set up access. It is benefit from it and incentives to important to match public reward establish investment. performance and transparent introduce funding criteria to accountability. improve efficiency Lack of timely and equity. statistics of appointed and contracted teachers on the payroll hinders development of policy. Increasing Need to establish If the quality of Monitoring and In the short run, attention to transparent education has evaluation should teachers will feel meritocracy and evaluation criteria markedly underpin an pressured. Needs quality. and need policy improved, accountability to support and resources to households are system that is improvement in reward likely to be willing central to a merit- pre-service performance. This to invest even based system. training and in- will ultimately more and start the service improve the viltuous cycle. professional efficiency of development. resource use. Meritocracy should raise status of teachers in the long run. Needs to change salary scale to attract and retain competent people ________________ in the profession. Expansion of Expanded mandate Increased public Need targeted Affects the constitutional needs to be investment will support for the supply and mandate for supported by have a matching poor, indigenous, demand for basic education, increased public effect to solicit and rural students teachers. Need to investment to additional private to ensure equity of pay special realize the goal. spending. learning outcomes. attention to the There must not be quality of teacher 70 Table 5: First Generation Reform and Its Implications for Second Generation Reform First Generation Public Finance Household Equity, Quality, The Teaching Reform Finance and Efficiency Profession a trade-off between pre-service and quality and in-service quantity. training. Legal Opting out of the Led to Need to improve Need to improve encouragement public system by mobilization of quality of public incentives in the of private the urban rich and more private education to avoid public system to education and middle class may resources for further deepening attract good the growth of free up resources education. Demand the socioeconomic teachers. Non- private for the working for private divide. Need to monetary education. class and rural education will accredit private incentives students. Lack of induce supply in schools for quality include making supply of private cities, and greater assurance. work schools in rural availability of environment areas calls for good quality of attractive and increased public education may breaking teacher investment there. attract more isolation. middle class to go for it. Regionalization Affects the budget Neutral at the Need to set Need to provides of process, inter- beginning but can national standards, quality assurance administration governmental be positive if the monitor through and deconcen- transfer of regions become performance, and accreditation of tration of resources, and more effective in undertake teacher training services. balance of power. meeting local measures to reduce programs, and Need to strengthen needs and in regional variation certification of the ability of the assisting the poor. in learning teachers. The regions to deliver outcomes. regions should quality services by have more giving them more flexibility in the discretionary recruitment and power and deployment of resources. teachers. 71 Table 6: Summary of Policy Options Objectives Issues Suggested Measures Improve Inequity in learning Provide teachers' guides, textbooks, workbooks, equity outcomes of poor, rural, supplemental reading, and audiovisual materials and indigenous students and media (for example, interactive radio) due to inadequate public throughout 1 years of basic education to all resources, unwillingness schools. Provide radio sets to families with of qualified teachers to school-age children in remote communities to serve in remote enable them to access lessons. communities, and perhaps poor nutrition and health Expand bilingual education programs; provide and lack of preparedness bilingual instructional materials, recruit of students themselves. indigenous teachers to serve their own communities, train rural teachers in multigrade teaching. Increase compensation to rural teachers and assure job rotation to induce teachers to teach in rural communities. Address the health of school children either through services provided directly through the schools, or through targeted publicly financed insurance such as the on-going Seguro Escolar, to improve attendance and learning. Consider targeted expansion of the school feeding program given initial positive findings of favorable outcomes with respect to nutrition, health and attendance. Weigh costs against scope of expansion of the school feeding program. Provide information (through the mass media) to parents about how good child-rearing practices and involvement in their children's learning to enhance children's interest in learning and improve their school achievement. Extend pre-school education to disadvantaged communities in order to improve school readiness. Provide compensatory education to ensure all children learn all requisite skills relevant to their grade level. Inequity in access to Establish distance learning programs (by a secondary and tertiary combination of programmed text and education of poor, rural, communication media) to provide secondary and indigenous students education to rural communities. due to lack of supply and 72 Table 6: Summary of Policy Options Objectives Issues Suggested Measures inability of households to Provide scholarships and grants to rural children pay. to enable them to attend secondary schools. Enhance Lack of-incentives to Change salary scale to better reward secondary quality attract and retain school teachers, as well as those who have competent people in university and postgraduate degrees in disciplines teaching and to improve other than education. Unify employment status performance. and offer open-ended tenure to be determined by performance. Reward schools that show improvement. Set standards for student learning; set standards for teaching, strengthening teacher pre-service and in-service training (possibly changing the academic requirement for teaching in secondary education); provide incentives for teacher knowledge and skill upgrading, and encourage schools to organize for learning. This entails improving the tools for student assessment and teacher evaluation, as well as building consensus with all stakeholders. Lack of information for Publish information on performance by school benchmarking own and distribute to relevant actors (DRES, USE, performance to know schools and parents). how much to do in order to improve Improve Lack of sectoral overview Monitor the trend of public spending on education efficiency of of policy to assess its over time (gross and net of pension), particularly resource use. coherence and differential spending by urban and rural areas at consistency. the USE level; track indictors of learning outcomes; survey household expenditures on education and follow changes in the elasticity of demand as well as the private and social rates of return. Insufficient coordination Establish a sectorwide coordinating body that of educational policy and meets at least quarterly to review sectoral policy, financial matters between performance indicators of each subsector, and the budgetary entities that intrasectoral allocations to ensure consistency of have responsibility for overall education policy and expenditure. education (MED, universities, decentralized institutions, Regions, PRES, PROMUDEH.) Insufficient information Conduct an education census to obtain on teachers and school- information on teachers' profiles and school-level level finance in public finance. and private schools to guide policy to improve 73 Table 6: Summary of Policy Options Objectives Issues Suggested Measures quality and expand access. Inflexible criteria for Clean up payroll list and set up a centralized allocating resources database. based on the number of teachers on permanent Consider using funding formula based on average payroll without daily attendance with grants to compensate for responding to changing various categories of disadvantage. Give schools school-age population. greater discretion to use resources. Potential issue of intra- Given that policy and expenditure on higher sectoral allocation education impacts on lower levels of education, between universities and in-depth review of funding for universities is lower levels of education. desirable to assess options for introducing funding formulas with incentives for improving efficiency, .____________ and for cost-sharing. 74 REFERENCES Aitkin, M. and N. Longford. 1986. "Statistical Modeling Issues in School Effectiveness Studies." Journal of the Royal Statistical Society 149, pt. 1, pp.1-43. Akin, John S. and Irwin Garfinkel. 1977. "School Expenditures and the Economic Returns to Schooling." Journal of Human Resources 12 (4), pp. 460-48 1. Arregui, Patricia y Ernesto Melgar Salm6n. 1992. "Financiamiento de las Universidades Puiblicas en el Perni: Respuestas a la Crisis y al Ajuste Econ6mico." GRADE y Consorcio de Investigaci6n Econ6mica. Arregui, Patricia; Barbara Hunt y Hugo Diaz. 1996. "Problemas, Perspectivas y Requerimientos de la Formaci6n Magisterial en el Peri." Informe Final del diagn6stico elaborado a solicitud del Ministerio de Educaci6n y la GTZ, Lima, Perui, Octubre. Asamblea Nacional de Rectores. 1996. Universidades del Perui. Facultades y Carreras Profesionales. 1996. Lima, Perui. Asociaci6n Peruana de Administraci6n para el Desarrollo. 1991. "Regimen de Pensiones a Cargo del Estado D.L. 20530 y Conexas." Lima, Peru, Agosto. Banco Central de Reserva del Peru. 1994. "Politicas Macroecon6micas y Aspectos Sociales en Alemania y PeNr." Moneda, Debate. Junio Bosker, Roel J. and Jaap Scheerens. 1990. "Issues in the Interpretation of the Results of School Effectiveness Research." International Journal of Educational Research 13 p. 745. Bricefio, A.; A. Pasc6-Font, J. Escobar y J. Rodriguez. 1992. "Gesti6n Puiblica y Distribuci6n de Ingresos: Tres Estudios de Caso para la Economia Peruana." Serie de Documentos de Trabajo 115, Banco Interamericano de Desarrollo, Washington, D.C. Bryk, Anthony and Yeow Meng Thum. 1989. "The Effects of High School Organization on Dropping Out: An Exploratory Investigation." American Educational Research Journal 26 (3), pp. 353-383. Bryk, Anthony and Stephen Raudenbush. 1992. Hierarchical Linear Models: Applications and Data Analysis Methods. Newbury Park, CA: Sage Publications. Burstein, Leigh. 1980. "The Analysis of Multi-Level Data in Educational Research and Evaluation." Review of Research in Education 4(8), pp.158-223. Cardenas, Mauricio y Catalina Gutierrez.1997. "Impacto de Las Reformas Estructurales sobre la Eficiencia y la Equidad" in Cardenas M. (eds.) Empleo y Distribucion del Ingreso en America Latina: Hemos Avanzado? TM Editores, BogotA, Colombia. Cueto, Santiago; Enrique Jacoby y Ernesto Pollitt. "Factores Predictivos del Rendimiento Escolar en un Grupo de Alumnos de Escuelas Rurales." Revista Educaci6n, Universidad Cat6lica. Cueto, Santiago; Enrique Jacoby y Ernesto Pollitt. "Tiempo en La Tarea y Actividades Educativas en Escuelas Rurales Del Peru." (Borrador). 75 Deaton, Angus and Anne Case. 1988. "Analysis of Household Expenditures'. Living Standard Measurement Survey," Working Paper No.28, The World Bank: Washington, D.C. May. Diaz, Hugo; Vicente Ramirez Huayte, Francisco Custodio Farro y Jose Alvarado Tavara. 1995. La Educaci6n Privada en Lima Metropolitana. Instituto Internacional de Planeamiento de la Educaci6n, UNESCO, Informe de Investigacion No. 104. Diaz, Hugo. 1996. "Oferta y Demanda de Docentes para la Educaci6n Inicial, Primaria y Secundaria." Tomo III del Informe diagnostico Problemas y Perspectivas de la Promoci6n Magisterial en el Perui. Lima. DIaz, Hugo y Otto Revoredo FJ6spuru. 1998. "B,ducaci6n Plrivada: Situaciodn y Tnendencias. Encuestas de Opini6n, Estadtsticas Bdsicas". Instituto de Investigaci6n para el Desarrollo y la Defensa Nacional, INIDEN. Febrero. Diaz, Hugo. 1999. Background paper on teacher incentives, school-level management, and decentralization. (Draft). Educaci6n Primaria (Directivos). 1998. "Examen Desarrollado del Concurso Piblico, Ley No. 26815," Abedul. F'scobal. Javier; Jaime Saavedra y Maximo lorero. 1998. "Activos y Recursos de la Poblaci6n Pobre en el Peril". Banco Interamericano de Desarrollo. Washington, D.C. Fergeson, Ronald. "Paying for Public Education: New Evidence of How and Why Money Matters," Harvard Journal on Legislation 28: 465-98. Figueroa, Adolfo. 1995. "Perui: Social Policies and Economic Adjustment in the 1980s." In Coping with Austerity: Poverty and Inequalitv in Latin America, edited by Nora Lustig, Chapter 10. Washington, D.C.: The Brookings Institution. Franke, Pedro. 1994. La Educaci6n Publica, los Pobresy el Ajuste. Septiembre. Fuller, Bruce and Hua Haiyan. 1992. "When Girls Learn More than Boys: Do Teaching Practices Influence Achievement in Botswana?" Unpublished paper presented at the BRIDGES/TEES Conference on Schooling Effectiveness, Harvard Institute for International Development. September. Fuller, Bruce and Prema Clarke. 1994. "Raising School Effects While Ignoring Culture? Local Conditions and the Influence of Classroom Tools, Rules, and Pedagogy." Review of Educational Research 64 (1): 119-57. GarcIa Guadilla, Carmen. 1996. Situaci6n y Principales dinamicas de Transformaci6n de la Educaci6n Superior en America Latina Liima, Peru. Gindling, T.H. and Donald Robbins. 1994. "Earnings Inequality, Structural Adjustment and Trade Liberalization in Costa Rica" Mimeo. Gertler, Paul arid Paul Glewwe. 1989. Willingness to Pay for Education in Developing Country: Evidence from Rural Peru. Living Standards Measurement Study Working Paper No. 52. Washington, D.C.: The World Bank Glewwe, Paul. 1987. The Distribution of Welfare in Peru in 1985-86. Living Standards Measurement Study Working Paper No. 42. Washington, D.C.: The World Bank. and Gillette Hall. 1992. Poverty and Inequality During Unorthodox Adjustment: The Case of Peru, 1985-90. Living Standards Measurement Study Working Paper No. 86. Washington, D.C.: The World Bank. 76 . 1995. Who Is the Most Vulnerable to Macroeconomic Shock: Hypothesis Testing Using Panel Data from Peru. Living Standards Measurement Study Working Paper No. 86. Washington, D.C.: The World Bank. Godenzzi, Juan Carlos. 1996a. "Diversidad Etnolinguistica, Genero y Alfabetizaci6n en el Peru." . 1996b. "Educaci6n Bilingiue e Interculturalidad en los Andes y la Amazonia." Revista Andina 14 (2). Cusco, Peru. Diciembre. Gonzalez, J.E.V. y A.S. G6mez. 1997. "Descentralizaci6n de los Servicios de Educaci6n en Colombia." Marzo. Guerrero, Luis Ortiz; Patricia Pacora Andrade, Roxana Ball6n Del Valle, Silvia Rivera Ochoa y Miriam Vertiz Ponce. 1996 Problemas y Posibilidades de la Educaciin Inicial en el Peru. Lima, Peru. Diciembre. Harbison, Ralph and Eric Hanushek. 1992. Educational Performance of the Poor: Lessons from Rural Northeast Brazil. New York: Oxford University Press. Huaman, M.; S. Nnfiez, L. Salazar, A. Carhuatocto, J. Williams, E. Araujo y V. Vasquez. 1993. "Innovaciones Pedag6gicas en la Subregi6n Jaen." Lima, Peru. Instituto Cuanto. 1997. "Encuesta Nacional de Hogares sobre Medici6n de Niveles de Vida en el Peru: Informaci6n Basica." Lima, Peru. Instituto de Investigaci6n para el Desarrollo y la Defensa Nacional (INIDEN). 1998. "Educaci6n Privada: Situaci6n y Tendencias, Encuesta de Opini6n, Estadisticas Basicas." Lima, Peru. Instituto Nacional de Estadistica e Informatica. 1997a. "Censo Nacional Universitario 1996." Sintesis Estadistica. .1997b. "Peru: Caracteristicas Educativas de los Hogares. Encuesta Nacional de Hogares 1996." Direcci6n Tecnica de DemografIa y Estudios Sociales. Jacoby, Hanan G. 1994. "Borrowing Constraints and Progress through School: Evidence from Peru." In The Review of Economics and Statistics, pp.15 1-160. Jimenez, Emmanuel; Vicente Paqueo and Ma. Lourdes de Vera. 1988. "Student Performance and School Costs in the Philippines' High Schools." Policy, Planning, and Research Working Papers WPS 61. Washington, D.C.: The World Bank. Kalbfleisch, J.D. and R.L. Prentice. 1980. The Statistical Analysis of Failure Time Data. John Wiley and Sons. King, Elizabeth M. 1990. "Does Education Pay in the Labor Market?: The Labor Force Participation, Occupation, and Earnings of Peruvian Women." Living Standards Measurement Study Working Paper no. 67. Washington, D.C.: The World Bank. Lockheed, Marlaine E. and Nicholas T. Longford. 1989. "A Multi-Level Model of School Effectiveness in a Developing Country," Policy, Planning, and Research Working Papers WPS 242. Washington, D.C.: The World Bank. Lockheed, Marlaine and Barbara Bruns. 1990. "School Effects on Achievement in Secondary Mathematics and Portuguese in Brazil." Policy, Research, and External Affairs Working Papers WPS 525. Washington, D.C.: The World Bank. Lockheed, Marlaine and Adriaan M. Verspoor. 1991. Improving Primary Education in Developing Countries. London: Oxford University Press for the World Bank. Loera, Armando y Noel F. McGinn. 1992. "La Repitencia de Grado en la Escuela Primaria Colombiana: Resultados de una Exploraci6n sobre los Factores 77 Asociados a la Repitencia y las Politicas de Promoci6n." Cambridge, MA: Harvard Institute for International Development, Education Development Discussion Paper. L6pez, Renan and Carla della Maggiora (with collaboration of Gustavo Anriquez). 1998. "Rural Poverty in Peru: Facts and Analysis for Policy." (Draft.) University of Maryland at College Park. L6pez de Castilla; Marta, Lucy Trapnell y Moises Rengifo. 1990. Ser Maestro en la Amazonia. Serie: Autoeducaci6n Docente. Lima, Peru: Instituto de Pedagogia Popular. Ministerio de Educaci6n del Peru. 1995. Censo Escolar 1993. . 1996. Estadisticas B6sicas 1995 1 1997a. "Nueva Estructura del Sistema Educativo Peruano: Fundamentos de la Propuesta." . 1997b. "Politica y Estrategias para el Desarrollo de la Educaci6n Bilingule Intercultural." Documento de Trabajo. _ 1997c. "Manual para Docentes del Nivel de Educaci6n Primaria." Plan Nacional de Capacitaci6n Docente. . 1997d. "Guia de la Docente Coordinadora de Programas no Escolarizados de Educaci6n Inicial, PRONOEI." . 1997e. "Educaci6n en Acci6n: Educaci6n en Loreto, Rumbo a la Excelencia." . 1997f. "Sistema Privado de Pensiones." (Curso de Capacitacion Dirigido a: Funcionarios y Trabajadores del Ministerio de Educaci6n). . 1998. El Plan de Mediano y Largo Plazo de la Educacion Peruana: 1998-2021. Ministerio de la Presidencia del Peru. 1996. Elements of the Focalized Strategy to Combat Extreme Poverty 1996-20. 00 Ministerio de Promoci6n de la Mujer y del Desarrollo Humano del Peru. La Nihez Primero. _____._ Las Mujeres en el Peru de Hoy. 1996 Ministry of Education of Egypt. Education Planning and Information Division, Research and Analysis Directorate. 1993. "General Report on National Survey of Teaching Practices and School Effectiveness." Cairo. Montero, Carmen. 1996. "La exclusi6n Educativa de las Nifias del Campo: Dimensiones, Causas y Posibilidades de Atenci6n." Instituto de Estudios Peruanos. , Patricia Oliart, Patricia Ames, Zoila Cabrera y Francesca Uccelli. 1998. La Escuela Rural: Estudio para Identificar Modalidades y Prioridades de Intervenci6n. Instituto de Estudios Peruanos. Mwabu, Germano. 1994. Household Composition and Expenditures on Human Capital Inputs in Kenya. Department of Economics, Yale University. Nyagura, Levi M. and Abby R. Riddell. 1992. "Primary School Achievement in English and Mathematics in Zimbabwe: A Multilevel Analysis." Unpublished paper presented at the BRIDGES/TEES Conference on Schooling Effectiveness, Harvard Institute for International Development. September. OECD. 1998. Education at a Glance. Paris, France. Palacios, Maria Amelia y Manuel Paiba Cossios. L[997. Consideraciones para una Politica de Desarrollo Magisterial. Lima, Peru: Foro Educativo. 78 Pessino. 1995. "Labor Market Consequences of the Economic Reform in Argentina." in Social Tensions, Job Creation and Economic Policy, D. Turmham, C. Foy and G. Larrain (eds.) OECD, Paris. Planning Institute of Jamaica. 1997. Survey ofLiving Conditions. Kingston, Jamaica. Pollitt, Ernesto; Enrique Jacoby y Santiago Cueto. 1996 "Desayuno Escolar y Rendimiento; A proposito del Programa de Desayunos Escolares de Foncodes en el Peru." APOYO, Lima, Peru. Postlethwaite, T. N. and Kenneth Ross. 1992. Effective Schools in Reading: Implications for Educational Planners. The Hague: International Association for the Evaluation of Educational Achievement. Psacharopoulos, George. 1996. Economics of Education Review 15 (4). Quineche Meza, Daniel. 1992. El Trabajo del Maestro con la Comunidad. Serie: Autoeducaci6n Docente. Lima, Perui: Instituto de Pedagogia Popular. y Luis Velez. 1994. Experiencias de Educaci6n y Trabajo Productivo: Regi6n Nororiental del Maran6n. Serie: Autoeducaci6n Docente. Lima, Peru: Instituto de Pedagogia Popular. Ramirez de Sanchez-Moreno, Eliana; Luis Sime, Santiago Cueto, Emilio Morillo y Arturo Miranda. 1997. Hacia una Propuesta de Educacion Primaria para el Perui. Alternativas Pedag6gicas y de Gesti6n. Lima, Peru: Foro Educativo. Raudenbush, Stephen; Somsri Kidchanapanish and Sang Jin Kang. 1991. "The Effects of Preprimary Access and Quality on Educational Achievement in Thailand." Comparative Education Review 35 (May): 255-273. Reglamento Interno No. 001-97-DIRELL: Reglamento para el Programa Semestral de Evaluaci6n del Rendimiento Laboral de los Trabajadores Administrativos del Ambito de la DIRELL y USES, TI semestre-97. Reimers, Fernando. 1992a. "Influences on Student Achievement in Pakistan." Unpublished paper presented at the BRIDGES/lEES Conference on Schooling Effectiveness, Harvard Institute for International Development. September. _ . 1992b. "The Role of Multigrade Education in Honduras." Unpublished paper presented at the BRIDGES/IEES Conference on Schooling Effectiveness, Harvard Institute for International Development. September. Riddell, Abby Rubin. 1988. "School Effectiveness in Secondary Education in Zimbabwe: A Multilevel Analysis," unpublished Ph.D. thesis, University of London, Institute of Education. and Levi Martin Nyagura. 1991. "What Causes Differences in Achievement in Zimbabwe's Secondary Schools?" Policy, Planning and Research Working Papers WPS 705. Washington, DC: The World Bank, Population and Human Resources Department, June. Robbins. 1996 "HOS Hits Facts: Facts Win. Evidence on Trade and Wages in the Developing World." Harvard University. Robles, Hector; Josd Rodriguez and David Abler. 1997. "What Can We Learn about Rates of Return to Schooling before and after Structural Adjustment? The Mexican and Peruvian cases." Rodriguez, Jose. 1994. "Gasto Piiblico en Educaci6n y Distribuci6n del Ingreso en el Peri." El Trimestre Econ6mico 61 (4): 679-706. Octubre-Diciembre. 79 y David, Alber. 1998. "Asistencia a la Escuela y Participaci6n en el Mercado Laboral de los Menores en el Perfi" (borrador). Penn State University. Roeders, Paul. 1996. "Los Principios de la Formaci6n Docente Peruana y su Contexto: Un Primer Analisis de los Problemas y el Plan de Trabajo." Proforma-Proyecto Reforma Formaci6n Magisterial, Informe No. 1. Octubre. Saavedra, Jaime and G. Felices. 1997a. Inversion la Calidad de la Educacion Puiblica en el Peru y Su Efecto sobre la Fuerza de Trabajo y la Pobreza. Documento No. SOC97 - 104. Departamento de Porgramas Sociales y Desarrollo Sostenible, Banco Interamericano de Desarrollo, Washington D.C. Saavedra, Jaime. 1997b. "Quienes Ganan y Quienes Pierden con una Reforma Estructural: Cambios en la Dispersi6n de Ingresos seguin Educaci6n, Experiencia y Genero en el Perui Urbano". Notes para el Debate/14. 1 Edici6n. Lima, Perui: GRADE. I Roberto Melzi Rios y Arturo Miranda Blanco. 1998. Financiamiento de la Educacion en el Peru. Documento de Trabajo 24. Lima, Peru: GRADE. . 1998. "What Do We Know About Poverty and Income Distribution in Peru, with Emphasis in its links with Education and the Labor Market." Lima, Perul: GRADE. and Hugo Diaz. 1999. Preliminary Analysis of a Study of Public and Private School Teachers. (Draft). Sadoulet, Elisabeth and Alain de Janvry. 1995. Quantitative Development Policy Analysis. The Johns Hopkins University Press. Baltimore. Sanchez Moreno Izaguirre, G. and Hugo Diaz. 1997. Gesti6n Educativa. Lima, Perui. Schady, Norbert. 1999a. Seeking Votes. The Political Economy of Expenditures by the Peruvian Social Fund (FONCODES). World Bank Policy Research Working Paper Series, No. 2229, Washington, D.C. and Christina Paxson. 1999b. Do School Facilities Matter? The Case of the Peruvian Social Fund (FONCODES), 1991-1995. World Bank Policy Research Working Paper Series, No. 2166, Washington, D.C. Sifuentes, Victor; Manthoc Walter Alarc6n, Sigfredo Chiroque, Chris Van Dam, Manuel Valdivia, Nicole Bernex, Marta Dolorier, Daniel Quineche, Wilfredo Kapsoli, Keneth Delgado y Carmen Mendoza. 1991. Escuela de Trabajo, Escuela Productiva. Serie: Autoeducaci6n Docente. Lima, Peru: Instituto de Pedagogia Popular. Subregion Jean, Cads. 1994. Experiencias de Educaci6n y Trabajo Productivo: Regi6n Nororiental del Aiarafi6n. Serie: Autoeducaci6n Docente. Lima, Peru: Instituto de Pedagogia Popular. Tatoo, Maria T.; K. H. Dharmadasa, K. G. Kularatna and D. Nielsen. 1990. "Effectiveness and Costs of Three Approaches to Train Elementary School Teachers in Sri Lanka." Research report for the BRIDGES Project, Harvard Institute for International Development and the United States Agency for International Development. Thorne, Cecilia; Monserrat Moreno and Evangelina L6pez. 1996. "El Nifno Eje del Cambio Social para una Educaci6n de Calidad." Cuadernos de la Facultad de Letras y Ciencias Humanas, Pontificia Universidad Cat6lica del Peru, Lima, Peruf. UNESCO. 1998. Statistical Yearbook. Paris. 80 Valdivia, Marin; and Edmundo Murugarra. 1997. "Wages and Health Determinants for Peruvian Adults: Searching for Differentials across Genders, the Life-Cycle, and the Wage Distribution." GRADE and University of Califoinia, Los Angeles (Draft). Van de Walle, Dominique. 1998. "Assessing the Welfare Impacts of Public Spending." World Development. Vega, Juan Fernando; C6sar Guadalupe, Jose Rodriguez y Juan Pablo Silva. "Estudio Sectoral Sobre Financiamiento Educativo y Educaci6n Rural en el Peru." Vexler, Idel; Ricardo Morales, Arturo Miranda, Hugo Trelles y Rauil Haya de la Torre. 1997. La Educaci6n Secundaria de Adolescentes en el Peru, Realidady Propuesta de Desarrollo Pedag6gico. Lima: Foro Educativo. Waiser, Myriam and Donald Winkler. Children in Poverty in Latin America. (in progress). Warwick, Donald P.; Fernando Reimers and Noel McGinn. 1989. "The Implementation of Reforms in the Primary Schools of Pakistan." Harvard Institute for International Development, Cambridge, Mass. Winkler, Donald. 1997. "Descentralizaci6n de la Educaci6n: Participacion en el Manejo de las Escuelas a Nivel Local." Grupo de Desarrollo Humano, Regi6n de Am6rica Latina y el Caribe. Informe no. 8. Mayo. World Bank. 1994. "Peru: Calidad, Eficiencia, Equidad. Los Desafios de la Educaci6n Primaria." Washington, D. C. (Draft yellow cover.) . 1994. Peru: Primary Education Quality Project. Staff Appraisal Report No. 13426-PE. Washington, D.C. . 1995. Priorities and Strategies for Education: A World Bank Review. Washington, D.C. . 1996. "Peru: Did the Ministry of the Presidency Reach the Poor in 1995?" Revised Draft, Washington, D.C. . 1997. Country Assistance Strategy, Perui. Latin America and the Caribbean Region. Washington, D.C. . 1998b. "Peru: Indigenous Peoples' Development: Background, Policies and Program Strategy." Draft, Washington, D.C. . 1 998c World Development Report. Washington, D.C. 1 1999a. Peru: Improving Health Care for the Poor. A World Bank Country Study. Washington, D.C. . 1999b Poverty and Social Developments in Peru, 1994 - 1997. A World Bank Country Study. Washington, D.C. . 1 999c Education Sector Strategy. Education Advisory Service. Washington, D.C. . 1999d. Jamaica: Secondary Education: Improving Quality and Extending Access. Washington, D.C. Report No. 19069. 81