% Q uI I Q7 rubiic Expenditures I 974 on Education - and Income DP, trihitin j* **B29.C234* - 4 - - - E LB *9 C 62*LR2899.C62 134 c.2 * Public expenditures on education and income distribution in SLCI38 I 111189IIIIIIIi _ SLCO38 1 89 Ak.w k L LWI _________ World Bank Staff Occasional Papers No. 1. Herman G. van der Tak, The Economic Choice be- tween Hydroelectric and Thermal Power Develop- ments, with a Note on Joint Costs in Multipurpose Projects, $3.00 No. 2. Jan de Weille, Quantification of Road User Savings, $3.00 No. 3. Barend A. de Vries, The Export Experience of De- vQloping Countries, $3.00 No. 4. Hans A. Adler, Sector and Project Planning in Trans- portation, $3.00 No. 5. A. A. Walters, The Economics of Road User Charges, $5.00 No. 6. Benjamin B. King, Notes on the Mechanics of Growth and Debt, $3.00 No. 7. Herman G. van der Tak and Jan de Weille, Reap- praisal of a Road Project in Iran, $3.00 No. 8. Jack Baranson, Automotive Industries in Developing Countries, $3.00 No. 9. Ayhan ,7ilingiroglu, Manufacture of Heavy Electrical Equipment in Developing Countries, $3.00 No. 10. Shlomo Reutlinger, Techniques for Project Appraisal under Uncertainty, $3.00 No. I l. Louis Y. Pouliquen, Risk Analysis in Project Ap- praisal, $3.00 No. 12. George C. Zaidan, The Costs and Benefits of Family Planning Programs, $3.00 No. 13. Herman G. van der Tak and Anandarup Ray, The Economic Benefits of Road Transport Projects, $3.00 No. 14. Hans Heinrich Thias and Martin Carnoy, Cost-Benefit Analysis in Education: A Casc Study of Kenya, $6.00 No. 15. Anthony Churchill, Road User Charges in Central America, $5.00 No. 16. Deepak Lal, Methods of Project Analysis: A Review, $3.00 No. 17. Kenji Takeuchi, Tropical Hardwood Trade in the Asia- Pacific Region, $4.00 No. 18. Jallade, Jean-Pierre, Public Expenditures on Educa- tion and Income Distribution in Colombia, $4.00 WORLD BANK STAFF OCCASIONAL PAPERS NUMBER EIGHTEEN This paper may not be quoted as representing the view of the Bank and its affiliated organizations. They do not accept responsibility for its accuracy or completeness. WORLD BANK STAFF OCCASIONAL PAPERS NUMBER EIGHTEEN Edited by Ian Bowen Jean-Pierre Jallade Public Expenditures on Education and Income Distribution in Colombia S,CTOROAL LiBRARY l11lRNATIONAL UAI4 RECO4SSTM'ON AND DEVeLO UMORA11OWAL MONETAR UND WASIIU4ojroN71).C Distributed by The Johns Hopkins University Press Baltimore and London Copyright © 1974 by the International Bank for Reconstruction and Development All rights reserved Manufactured in the United States of America Library of Congress Catalog Card Number 74-4216 ISBN 0-8018-1628-9 Library of Congress Cataloging in Publication data will be found on the last printed page of this book. 1A /C&) Foreword I would like to explain why the World Bank Group does research work, and why it publishes it. We feel an obligation to look beyond the projects we help to finance toward the whole resource allocation of an economy, and the effectiveness of the use of those resources. Our major concern, in dealings with member countries, is that all scarce resources, including capital, skilled labor, enterprise and know-how, should be used to their best advantage. We want to see policies that encourage appropriate increases in the supply of savings, whether domestic or international. Finally, we are re- quired by our Articles, as well as by inclination, to use objective economic criteria in all our judgments. These are our preoccupations, and these, one way or another, are the subjects of most of our research work. Clearly, they are also the proper con- cern of anyone who is interested in promoting development, and so we seek to make our research papers widely available. In doing so, we have to take the risk of being misunderstood. Although these studies are published by the Bank, the views expressed and the methods explored should not neces- sarily be considered to represent the Bank's views or policies. Rather they are offered as a modest contribution to the great discussion on how to ad- vance the economic development of the underdeveloped world. ROBERT S. McNAMARA President International Bank for Reconstruction and Development Table of Contents FOREWORD . ............................................... v LIST OF TABLES AND FIGURES ................ .I x GLOSSARY ................................................. xiii INTRODUCTION ........................................... xv I. DESCRIPTION OF THE PROBLEM ....... ............... 3 II. THE FINANCING OF EDUCATION IN COLOMBIA: MAIN CHARACTERISTICS .............................. 8 Basic Facts on Colombian Educational Finance ............ 10 Implications for the Income Redistributive Aspects of Public Expenditures on Education ................... 17 111. PUBLIC EXPENDITURES ON EDUCATION IN COLOMBIA: THE REDISTRIBUTIVE IMPACT AMONG INCOME GROUPS ...................................... 20 The Distribution of Income among Colombian Households in 1970 ........... ....................... 20 The Distribution of the Tax Burden among Income Groups ...................... ...................... 21 The Distribution of Public Subsidies for Education among Income Groups ......... I ...........I......... 28 Comparing the Distribution of Tax Burdens and Public Subsidies for Education among Income Groups .......... 37 IV. CENTRAL GOVERNMENT TRANSFERS FOR EDUCATION: THE DISTRIBUTIVE IMPACT AMONG REGIONS ............ ......................... 42 The Distribution of National Taxes among Departments .... 42 The Distribution of Central Government Transfers among Departments ........... ....................... 46 Comparing the Distribution of Taxes and the Distribution of Transfers ........ ..................... 48 Allocating the Transfers for Education among Departments ......... .............................. 51 Conclusions ......... 68 Vil V. A SUMMARY OF FINDINGS .. ......................... 69 ANNEXES 1. Table of Educational Expenditures and Total Expenditures by Level of Government ...... .......... 71 2. Table of Public Expenditures per Pupil in Colombia ....... 72 SELECTED BIOGRAPHY .................................... 73 Vill List of Tables and Figures TABLES 2.1 Colombia: Expenditures on Education as a Proportion of Total Expenditures by Government Level ..... ..... 12 2.2 Colombia: Public Expenditures on Education by Levels of Government Finance ...................... 13 2.3 Colombia: Enrollments by Level and Type of Education in the Public and Private Sectors ..... ...... 15 2.4 Colombia: Revenue Structure by Levels of Government .. 18 3.1 Distribution of Income among Urban Households ........ 22 3.2 Distribution of Income among Rural Households ..... ... 22 3.3 Distribution of Income among Colombian Households .... 23 3.4 Urban Areas: Allocation of Taxes according to Alternative I ...................................... 26 3.5 Urban Areas: Allocation of Taxes according to Alternative 2 ...................................... 26 3.6 Rural Areas: Allocation of Taxes according to Alternative 1 ...................................... 27 3.7 Rural Areas: Allocation of Taxes according to Alternative 2 ...................................... 27 3.8 Total Colombia: Allocation of Taxes according to Alternative I ...................................... 27 3.9 Total Colombia: Allocation of Taxes according to Alternative 2 ...................................... 28 3.10 Urban Colombia: Enrollments by Level of Family Income ........................................... 30 3.11 Urban Colombia: Enrollments by Level of Family Income in Public and Private Schools ................. 32 3.12 Rural Colombia: Enrollments by Level of Family Income .......................................... 32 3.13 Total Public Expenditures per Pupil ....... ............. 33 3.14 Primary Education (Public): Teachers' Costs in Urban and Rural Areas ............................. 35 3.15 Urban Colombia: Public Subsidies for Education by Level of Income ................. .................. 36 ix 3.16 Rural Colombia: Public Subsidies for Education by Level of Income .......... ....................... 36 3.17 Urban Colombia: Allocation of Taxes and Public Subsidies for Education among Income Groups ........ 38 3.18 Rural Colombia: Allocation of Taxes and Public Subsidies for Education among Income Groups ........ 39 3.19 Total Colombia: Allocation of Taxes and Public Subsidies for Education among Income Groups ........ 40 3.20 Total Colombia: Public Subsidies for Three Levels of Education as a Proportion of Taxes Distributed among Income Groups .......... ................... 40 4.1 Distribution of Central Government Taxes among Departments .45 4.2 Transfers from the Central Government to the Departments .47 4.3 Structure of Transfers and Total Taxes Collected by Departments .49 4.4 Transfers for Education and Departmental Expenditures on Primary Education .54 4.5 Transfers for Education and Departmental and Municipal Expenditures on Primary Education .58 4.6 Transfers per Primary School Teacher .61 4.7 Transfers per Child Enrolled and per Teacher as Compared to Departmental "Effort" on Education . 64 4.8 Transfers per Teacher and per Child Compared with Enrollment Ratios .... ........ ....... 65 FIGURES 4.1 Total Transfers as a Proportion of Taxes Collected; Taxes Paid per Capita ............................... 52 4.2 Transfers for Education as a Proportion of Taxes Collected; Taxes Paid per Capita ....... .............. 53 4.3 Transfers as a Proportion of Departmental Expenditures on Primary Education; Departmental Expenditures on Primary Education per Capita ........ 56 4.4 Transfers as a Proportion of Departmental and Municipal Expenditures on Primary Education; Departmental and Municipal Expenditures on Primary Education per Capita ........................ 57 4.5 Transfers per Primary School Teacher (Public Sector Only); Expenditures on Primary Education as a Proportion of Total Expenditures ....... ............. 62 x 4.6 Transfers per Child Enrolled in Public Primary Education; Expenditures on Primary Education as a Proportion of Total Expenditures ...... .......... 63 4.7 Transfers per Primary School Teacher (Public Sector Only); Enrollment Ratios ......... .................. 66 4.8 Transfers per Child Enrolled in Public Primary Education; Enrollment Ratios ....... ................ 67 xi Glossary DANE Departamento Administrativo Nacional de Estadistica (Co- lombian Statistical Office) ICCE Instituto Colombiano de Construcciones Escolares (Colom- bian Institute for School Construction) ICETEX Instituto Colombiano de Credito Educativo y Estudios Tec- nicos en el Exterior (Colombian Institute for Educational Credit and Technical Studies Abroad) ICFES Instituto Colombiano para el Fomento de la Educaci6n Superior (Colombian Institute for the Development of Higher Education) INANDES Instituci6n Andina de Estudios Sociales (Andean Institution for Social Studies) SENA Servicio Nacional de Aprendizaje (National Apprenticeship Service) xiii Introduction In the field of educational finance, a feature common to many countries is that the costs of education are usually shared by the government and the students or their families. The cost burden accruing to each party does differ very widely among countries, over time within a single country, or according to the type or level of education concerned. Any kind of situation can be found, ranging from, say, private universities wholly financed by the fees paid by the students, to public vocational training programs which are not only tuition-free but also provide compensation for the income foregone by the trainees However, these extreme situations are not to be found very frequently, at least on a large scale. The bulk of educational activities is, in fact, jointly financed by the government and the families. There is little point in review- ing here the various arguments for providing public finance for education, simply because the debate is no longer about the existence but about the desired extent of public subsidies to education which, in turn, will deter- mine the charges accruing to private persons. Economists once believed they were able to provide an adequate solu- tion to this problem. During the early sixties, they started to develop analytical tools such as rates of return to education, hoping that standard cost-benefit analyses could provide an appropriate answer to the in- creasingly heard question: How much "should" a given country spend on education? The existing literature on this subject is abundant and no attempt to review it will be made here. Rates of return calculations consider only the economic benefits of education and governments and educational authorities must pay attention to the psychological, sociological and politi- cal benefits as well. This has engendered a large gap between the univer- sities and research institutions where these calculations are made, and the decision-making bodies where the future of educational systems is actually decided upon. xv Another shortcoming of current economic thinking in this area has been the lack of attention paid to the closely related question of how to share education expenditures between the government, i e., to a large extent the taxpayers, the families and other potential contributors. The economic ap- proach to the problem has been far too global, limiting itself to such broad alternatives as investing more or less in a given type of education, while the decision-makers are governed largely by the more down-to-earth issue of who is going to bear the burden of additional expenditures. As a result, it is only fair to say that despite a decade of rates of return calculations, both the amount of money that is spent on education and the arrangements for sharing the burden of educational expenditures are still the result of the free play of social and political forces in the country under consideration. In any case, the contribution of the economists to the deci- sion-making process has been marginal. It is believed that this situation has developed largely because economists have paid too much attention to the problem of the amount of educational expenditures-which they cannot pretend to solve alone anyhow-while showing too little concern for the distributional impact of these expenditures. The impact will be defined here as the combined effect on the distribution of income resulting from the fi- nancing of educational expenditures and the allocation of educational sub- sidies. In other words, the discussion will be centered around the two basic questions: Who pays for education? Who receives the educational subsidies? As a result, the present study will focus not on the desired extent of public education expenditures but on their impact on income distribution. Although the empirical question of how a given government expenditures program affects different population groups is an interesting one per se, this study will not restrict itself to the allocation of educational benefits among various population groups, but will also consider the distributional effects of the taxes that pay for public subsidies in general. Because public expenditures on education are, in most cases, financed out of general revenue, the ultimate purpose of the present analysis is to spell out the joint incidence of both the taxation system and public educa- tion expenditures on the distribution of income. The country under study is Colombia at the end of the sixties. A brief description of the problem and a reminder of the existing literature will be presented in Chapter 1. Chapter 1I will deal with some basic facts on the fi- nancing of education in Colombia, and Chapter III will analyze the impact of public educational expenditures on the householddistribution of income. A similar analysis will be carried out in Chapter IV in order to pinpoint the impact of central government finance for primary education on the regional distribution of income. xv' Public Expenditures on Education and Income Distribution in Colombia I. Description of the Problem The question of how government policies affect the distribution of in- come in a given country has always aroused interest. The redistributive im- pact of, say, taxation policies has long been a matter of concern for both academics and politicians. More recently, attention has shifted towards the implications of government spending policies for the distribution of in- come. Who benefits from government expenditures on housing, health care, recreational facilities, education and so on? These services, provided to the public at a price sometimes far below their cost, have an obvious im- pact on incomes. Were they unequally distributed among population groups, the distribution of income would be altered. Comprehensive studies identifying the redistributive impact on income of both aspects of government activities, i.e., taxation and spending policies, are still scarce. A recent attempt to allocate both taxes and benefits by income level has been carried out by Gillespie with U.S. data for the year 1960.' The difficulties, both theoretical and empirical, encountered in carry- ing out such a study should not be underestimated. While many fiscal econ- omists still express doubts about the feasibility of developing meaningful tax incidence studies, Gillespie concedes that "the empirical estimation of expenditure incidence is much less precise than the tax incidence, primarily because any hypotheses about the distribution by income class of the bene- fits from government expenditures are less well developed."2 He then pro- ceeds to make a clear distinction between "those government expenditures the costs of which are incurred on behalf of clearly delineated beneficiary groups" and the "general" expenditures3 which are distributed according to 1W Irwin Gillespie, "Effect of Public Expenditures on the Distribution of Income," in Essays in Fiscal Federalism, ed Richard A. Musgrave (Washington, D.C the Brookings In- stitution, 1965). See also Tax Foundation, Tax Burdens and Benefits ofGovernment Expenditures by Income Class, 1961 and 1965, Research Publication no 9 (New York, 1967) 2Gillespie, "Effect of Public Expenditures," p 130 3Expenditures on national defense, international affairs, general safety and so forth 3 a set of alternative assumptions: equally per family, similarly as income, capital income or disposable income. According to Gillespie, "all four alter- natives give rise to extremely similar conclusions concerning the general pattern of expenditures incidence" Oddly enough, very few studies of this kind have ever been made in the developing world. Despite the fact that income and wealth are very un- equally distributed in developing countries-disparities of all kinds exist both between regions and between rural and urban areas-despite also the presence of a "strong" and rapidly growing public sector often advocated as the only remedy at hand to alleviate these disparities, it is astonishing that the impact of taxation and public spending policies on income distribution has been worked out only in a very few instances.4 To be sure, tax econo- mists have lately been issuing warnings about the difficulties and the limita- tions of such studies.5 These warnings, which are certainly justified in view of the often very crude approximations adopted by researchers, have however failed to stop the growing interest of agencies and individuals in this area 6 The literature of the redistributive impact of one single type of govern- ment expenditure is more abundant. In particular, the incidence of govern- ment spending on education alone seems to have awakened the interest of many researchers.' In all these studies, the analysis consists of the follow- ing steps. 4A rather rough attempt at assessing the impact of taxation and public spending policies on income distribution in Colombia is presented in Miguel Urrutia Montoya and Clara Elsa de Sandoval, "Politica fiscal y distribucion del ingreso en Colombia," Revista del Banco de la Re- piiblica XLIV (July 1971) 1072-87 See also Gian Sahota, "Public Expenditures and Income Distribution in Panama," mimeographed (Panama City USAID, 1972) 5For an excellent and skeptical appraisal of the research works carried out in Latin Amer- ica, see R M Bird and L de Wulf, "Taxation and Income Distribution in Latin America A Critical Review of Empirical Studies," IMF Fiscal Affairs Department, DM/73/8 (Wash- ington, D C , 1973) 6An examination of the three-year, USAID-financed research project initiated in 1972 by the Program of Development Studies of Rice University helps to assess, among other things, the redistributive impact of the government taxation and expenditure policies in three developing countries See "A Proposal for Research on Distribution of Gains. Wealth and In- come from Economic and Political Development," mimeographed (Houston, Texas, 1971) A less ambitious project has recently been launched by the United Nations, concerning which see K J Charles, "The Distribution and Utilization of Public Services," mimeographed (New York UN Income Distribution Project, n d ) 7See W Norton Grubb, "The Distribution of Public Costs and Private Benefits in an Ur- ban Public School System," National Tax Journal XXIV (March 1971) 1-12, Douglas M Windham, The Redistriburional Effects of Public Higher Education in Florida, University of North Carolina, Department of Economics and Business Administration, Monograph Series F, no I (Greensboro, 1969), Richard W Judy, "The Income Redistributive Effects of Public Aid to Education in Canada," in Canadian Economic Problems and Policies, ed Lawrence H Officer and Lawrence B Smith (New York McGraw-Hill, 1970), pp 302-17, and W Lee Hansen and Burion A Weisbrod, Benefits, Costs and Finance of Public Higher Education (Chicago Markham, 1969) 4 (i) define population groups: income groups, sometimes broken down by race, are generally used; (ii) allocate total tax payments to each income group; (iii) compute the benefits arising from government expenditures accru- ing to each income group; (iv) compare the distribution of tax burdens and benefits to arrive at the "net" incidence of government taxation and spending policies. The present analysis will use the same methodology. However, the follow- ing additional points will be considered. In the first place, dealing with Co- lombia, which is a country with wide geographic disparities and a strong regional life, any analysis of the incidence of government expenditures on the various income groups should be complemented by a similar analysis, using regions as units of observation. Consequently the tax burdens and the benefits will be distributed successively among income groups (Chapter III) and regions (Chapter IV). Secondly, the study is restricted to the distribution of benefits arising from government expenditures on education alone, excluding all other ex- penditures. It considers only the benefits accruing to students or their families as expenditures that need not be made because of the public sub- sidies to education. They include that portion of educational costs which is borne by the public. Other benefits such as those arising from the consump- tion value of education or an enhanced earning power are excluded. In other words, the analysis will be restricted to the immediate impact on family income arising from subsidized education, thus excluding the long- term changes in personal income distribution resulting from public invest- ment in education.8 Thirdly, the decision to restrict the study to the redistributive impact of public expenditures on education alone has important consequences on the fourth step of the analysis, namely the comparison of the distribution of tax burdens with that of benefits. These consequences are shown below. It is sometimes argued that the distributional aspects of government ac- tivities can be dealt with only if all areas of government expenditures are covered by the analysis, the argument being that the "net" incidence can be computed only if the distribution of all taxes is compared with the distribu- 8These two types of benefits from education are sometimes referred to as short-term and long-term benefits To assess the distribution of the latter type of benefits presents important empirical difficulties One needs to know, for instance, the exact distribution of the present generation of students among future income classes If both types of benefits are to be con- sidered together, the analysis becomes extremely intricate because the recipients of the bene- fits are not the same in both cases short-term benefits are received by the parents who also pay the taxes, while long-term benefits accrue to the educated child who does not now pres- ently pay taxes See Judy, "The Income Redistributive Effects," for an attempt along these lines 5 tion of all expenditures. If only one given type of government expenditure is considered, the distribution of the resulting benefits cannot be compared meaningfully with the distribution of taxes, simply because there is no way of knowing which taxes are used to finance these particular expenditures. This problem can be dealt with in a variety of ways. First, the problem is less acute in the case of a government program financed exclusively by the product of a special tax. A well-known example in the field of educational finance is the frequent use of a tax on payrolls earmarked for financing vocational training programs. However, the bulk of public expenditures on education is financed from the government general revenue and, here, the problem remains. A customary approach is known professionally as the "proportionality" assumption; it assumes that the proportion of total taxes paid by each in- come group or geographical unit for a particular program is the same as the proportion of the government's total expenditures on this program. Thus, if expenditures on education amount to 20 percent of government's total ex- penditures, it is assumed that 20 percent of the total taxes paid by each pop- ulation group is spent on education. The arbitrariness of this assumption has given rise to a long, probably unfinished, controversy9 between those who hold that the equiproportional allocation assumption has no logical or theoretical foundation and, therefore, should not be used,10 and those who maintain that this assump- tion has to be made, given our ignorance of distributional impacts of many public programs." The "proportionality" assumption will not be made in this study. As a consequence, it will not be possible to compute the exact amount of transfers from one population group to another arising from government expenditures on education However, the distribution of educational bene- fits can be usefully compared to the distribution of total taxes paid, by pre- senting education subsidies as a proportion of tax payments in each income 9W Lee Hansen and Burton A. Weisbrod, "The Distribution of Costs and Direct Benefits of Public Higher Education The Case of Cahifornia," Journal of Human Resources IV (Spring 1969) 176-91 This article is taken from Section IV of the same authors' Benefits, Costs and Fi- nance See also Elchanan Cohn, Adam Gifford and Ira Sharkansky, "Benefits and Costs of Higher Education and Income Redistribution Three Comments," Journal of Human Resources V (Spring 1970) 222-36; Joseph A Pechman, "The Distributional Effects of Pubiic Higher Education in California," Journal of Human Resources V (Summer 1970) 361-70, Robert W Hartman, "A Comment on the Pechman-Hansen-Weisbrod Controversy," Journal of Human Resources V (Fall 1970) 519-23, and W Lee Hansen and Burton A Weisbrod, "On the Distribution of Costs and Benefits of Public Higher Education Reply," Journal of Human Resources VI (Summer 1971) 363-74 10W Lee Hansen and Burton A Weisbrod, "Who Pays for a Public Expenditure Program9" National Tax Journal XXIV (December 1971) 515-17 fljames T Bonnen, "The Absence of Knowledge of Distributional Impacts an Obstacle to Effective Policy Analysis and Decisions," in Public Expenditures and Policy Analysis, ed Robert H Haveman and Julius Margolis (Chicago Markham, 1970) 6 group or geographical unit. The impact of public educational expenditures on income distribution will appear when comparing these proportions with the average for the country as a whole or across income groups or geographical units. Obviously, a group which receives an above-average proportion of its taxes back in the form of education subsidies has its in- come positively affected by the public financing of education, and vice- versa. 7 II. The Financing of Education in Colombia: Main Characteristics The most important feature of public education finance in Colombia can be summed up in one single word: decentralization. To start with, each of the three levels of government, namely the central government, the depart- ments and the municipalities, have traditionally shared the financial burden of educational activities provided free-of-charge to the public,' by maintaining their own schools or transferring funds to each other for the same purpose. Moreover, a fourth "government level," namely the de- centralized agencies of the education sector, has emerged during the sixties as a significant contributor to higher education and vocational training fi- nance. Further, Colombia is also referred to as the country with the largest private education sector in Latin America. As a consequence, the amount of private finance that goes towards education is probably bigger than elsewhere, although it is unknown. Clearly, the existing arrangements made to share the burden of educa- tional costs are the result of a host of historical and political facts that need not be reviewed here. It will suffice to say that the financial responsibilities of each of these financing bodies are not always in line with their adminis- trative responsibilities. This situation has engendered a complicated array of transfers of funds between the various financing bodies, with the result that double, if not triple, counting is current practice 2 The divorce between administrative and financial responsibilities is probably more dramatic at the central government level than anywhere else. In 1967, 48 percent of total public expenditures on education in the country originated from the budget of the Ministry of Education, while the schools administered by the Ministry accounted for less than 10 percent of enrollments in primary and secondary education respectively.3 The funds going to the universities, ISome public universities do charge very small fees 2For instance, the same sum of money may be counted once in the budget of the Ministry of Education, twice when it is transferred to a decentralized institution, and a third time when it reaches a university While there exist data on transfers from the central government to the departments, the exact amount oi funds for education transferred to the municipalities is not recorded by official statistics. 3Departmental and municipal schools enroll 80 percent of pupils in primary education and 40 percent of pupils in secondary education The remaining proportions are enrolled in private schools 8 though recorded in the budget of the Ministry, largely escape its control as soon as they are transferred to the decentralized agency responsible for higher education finance, or to the autonomous universities themselves. As a result, while the central government is, by far, the most important source of educational finance, its actual administrative and coordination powers are minimal. Needless to say, this situation is bound to be an important obstacle on the road to educational reforms on a countrywide basis. Three decentralized agencies contribute significantly to the financing of education in Colombia. At the higher education level, the Colombian In- stitute for the Development of Higher Education (ICFESP is the adminis- trative body responsible for the coordination and distribution of central government finance to public and private universities. While its functions are potentially very wide, its actual powers have so far been greatly reduced by the political turmoil aroused in the universities. The Colombian Institute of School Construction (ICCE)5 is responsible for carrying out Ministry of Education plans for school construction and for providing school equipment and teaching materials to primary and sec- ondary schools. A significant part of the capital costs at these levels is now financed through this institute. Lastly, the National Apprenticeship Service (SENA)6 is responsible for providing vocational training programs servicing the industrial and services sector. This institution, which does not receive funds from the government, is nearly wholly financed by a 2 percent tax on the payrolls of all firms employing more than fifty workers. While these three decentralized agencies operate at the central govern- ment level, the departments play a key role in the financing of primary and secondary education. They are responsible for building and maintaining schools and paying the teachers. They also pay the salaries of teachers at departmental universities. The extent of the departments' autonomy in conducting their own educational policies is quite real, despite recent at- tempts by the Ministry of Education to introduce some guidelines for the use of funds transferred to them. The municipalities' financial contributions to education are restricted to the provision of land for school building and to the maintenance of schools. These contributions are, in absolute value, very limited in the overwhelm- ing majority of municipalities which are simply too small and too poor to afford even minor maintenance expenditures. The situation is of course different in big cities: the three biggest municipalities of Bogota, Medellin and Cali are responsible for 70 percent of the total amount of expenditures 4lnstituto Colombiano para el Fomento de la Educaci6n Superior. 5Instituto Colombiano de Construcciones Escolares 6Servic,o Nacional de Aprendizaje. 9 on education made by the one thousand municipalities of the country in 1967. The four levels of government-central government, decentralized agen- cies, departments and municipalities-also contribute to the financing of private education. Although the exact amount of transfers of public funds to the private sector is unknown, it is generally accepted that about 10 percent alone of the budget of the Ministry of Education is spent on helping private schools. The above considerations should be kept in mind when dealing with Co- lombian educational finance statistics. Evidently, the lack of clear and pre- cise definition of responsibilities at each government level is one of the main causes of the poor quality of these statistics, which can provide but a very approximate image of the reality. Basic Facts on Colombian Educational Finance Public Expenditures on Education Detailed statistics on total public expenditures and public expenditures on education broken down by government level are shown in Annex 1. Although these data refer to the whole decade 1960-70, the last year for which public finance statistics for the departments and municipalities are available is 1969; the figures shown for these government levels in 1970 are based on our own estimates. Transfers from one government level to another have been, as far as possible, excluded. However, some double counting resulting, for instance, from minor transfers between the depart- ments to the municipalities and vice-versa, is still possible. In Colombia as in most countries, budget figures based on initial appropriations are often poor approximations of actual expenditures during the fiscal year. All figures presented in Annex I and below are based on the actual expen- ditures statistics collected by the controller's office of the Finance Ministry. In 1969, total spending of the four government levels amounted to 3.7 billion pesos, which represents 10.6 percent of their total expenditures and 3.4 percent of the gross domestic product. In absolute terms, the amount of public expenditures on education has been multiplied by eight (in current pesos) between 1960 and 1969.7 However, because total public expenditures were 9.5 times higher in 1969 than in 1960, there has been a slight decrease in the proportion of total public expenditures devoted to education during this period.8 Progress over the period looks more decisive when one con- 71n constant terms, it has been multiplied by three over the same period 8Ot is fair to say that total public expenditures have increased sharply in 1968, especially at the decentralized-agencies level, and in 1%9 at the departmental level, thus causing a drop in the proportion of total public expenditures devoted to education In both cases, educational ex- penditures did not "tfollow" the sharply rising trend of total public expenditures. 10 siders that educational expenditures as a proportion of the gross domestic product have doubled over the period. However, according to the interna- tional standards set by Unesco for Latin American countries, in 1970 Co- lombia was still below the target of 4 percent of its gross domestic product to be spent on education. Expenditures on education are obviously not given the same considera- tion at all levels of government, as shown in Table 2.1. Their share in the central government total expenditures increased from 9.4 percent to 13.6 percent over the period 1960-70. However, most of this increase took place early in the decade. Between 1962 and 1970 the proportion of education ex- penditures in the budget remained fairly stable. Within the central govern- ment's contribution, the bulk of this increase could be attributed to the funds transferred to the departments, whose share in the total central government budget more than tripled between 1960 and 1967.9 The de- centralized agencies, whose educational expenditures were negligible in the early sixties, devoted close to 5 percent of their budget to education in 1970. The departments make by far the biggest budgetary "effort" in education, by devoting to it between 30 and 40 percent of their total expenditures. In spite of two peaks in 1966 and 1967, it can be said that roughly one-third of departmental budgets has gone towards education during the decade. Education has obviously been a small and decreasing item on the munici- palities' spending list: its share in total municipal expenditures has now stabilized at a little more than 5 percent during the 1967-69 period. To sum up, by the end of the sixties, both the municipalities and the de- centralized agencies spent about 5 percent of their budget on education, with a downward tendency for the former and an upward tendency for the latter. While the proportion of departmental budgets devoted to education has remained roughly stable, it has increased at the central government level. This is largely due to the development of a transfers policy in favor of the departments inaugurated at the beginning of the decade. The concern for education shown by the various government levels has resulted in a changing pattern of their respective contributions to total educational expenditures during the period. Table 2.2 confirms the decreas- ing role of the municipalities as a source of education finance-6 percent of the total in 1969 as against 10.4 percent in 1960-and the increasing role of the decentralized agencies: 13.5 percent in 1969 as against 6.9 percent in 1964. By 1969 the central government had become the biggest single source of finance, accounting for about half of the total. However, its importance as a spending body has remained roughly the same, with slightly more than 40 percent of the total, the funds originating from the central government 91n absolute value, these transfers have been multiplied by twelve between 1960 and 1967. They represent 37 percent of central governrment finance in 1967, against 12 percent in 1960. 11 Table 2.1: Colombia: Expenditures on Education as a Proportion of Total Expenditures by Government Level (Transfers are Netted out) (%) Central Transfers to Total Central Decentralized All Levels of Government Departmentsl Government Agencies Departments Municipalities Government 1960 8.4 1.0 9.4 - 29.2 6.7 12.6 1961 8.4 1.2 9.6 - 34.0 6.9 12.6 1962 11.1 2 8 13.9 - 33.5 5.5 15.0 1963 8.9 4.2 13.0 - 32.3 5.1 14 2 1964 10.6 4.9 15.5 2.8 30.5 5.2 11.9 1965 9.5 4.0 13.5 4.5 34.4 5.1 11.9 1966 9.0 3.4 12.4 5.4 39.4 6.3 12.0 1967 9.5 3.5 13.0 3.8 39.2 5.3 11.0 1968 9.7 2.8 12.5 3.4 39.8 5.3 10.2 1969 10.4 2.3 12.7 3.9 31.5 5.3 10.6 1970 n.a. n.a. 13 6 4.4 33.0 5.3 10.9 lYvon Lebot, "Organizaci6n y funcionamiento del sistema educativo colombiano," in Boletin Mensual de Estadistica no. 243 (October, 1971). Tables 721 and 724. Source See Annex 1. Table 2.2: Colombia: Public Expenditures on Education by Levels of Government Finance (in %) Central Transfers to Total Central Decentralized Government Departments Government Agencies Departments Municipalities Total 1 2 3 4 51 62 7 8 1960 41.4 5.2 46.6 - 43.0 (48.2) 10.4 100.0 1961 46.3 6.4 52.7 - 38.5 (44.9) 8.8 100.0 1962 45.4 11.6 57.0 - 34.5 (46.1) 8.5 100.0 1963 38.5 18.1 56.6 - 35.2 (53.3) 8.2 100.0 1964 38.5 17.6 56.1 6.9 30.0 (47.6) 7.0 100.0 1965 36.3 15.1 51.4 10.5 31.5 (46.6) 6.6 100.0 1966 34.4 13.1 47.5 13.6 31.7 (44.8) 7.2 100.0 1967 35.0 13.0 48.0 12.3 32.4 (45.4) 7.3 100.0 1968 38.7 11.3 50.0 12.7 30.3 (41.6) 7.0 100.0 1969 41.0 9.2 50.2 13.5 30.3 (38.0) 6.0 100.0 1970 n.a. n.a. 51.0 15.8 27.8 (n.a.) 5.4 100.0 'Excluding the transfers from the central government. 2lncluding the transfers from the central government. Source. See Annex I and Table 1. and transferred to the departments accounting for the difference. The departments, whose role as a source of finance has decreased from 43 per- cent to 30 3 percent over the 1960-69 period, managed to maintain their preeminent role as spending bodies until 1968, thanks to the funds received from the central government. In spite of the drop recorded in 1969, the transfers for education from the central government to the departments thus appear to have been a rapidly growing source of educational finance during the sixties.'0 While the development of this transfer policy illustrates the divorce between the administrative and financial responsibilities mentioned earlier, it is not clear whether it reflects more the reluctance of the central government to take over further administrative chores, or the willingness of the depart- ments to maintain their traditional responsibilities in this area in spite of their financial inability to meet increasing expenditures. In any case, the transfers policy provides the central government with a clear opportunity to intervene more actively in the shaping of an education policy valid for the country as a whole. The Importance of the Private Sector As mentioned earlier, there is no way of knowing exactly the amount of private finance going into education in Colombia. However, it is possible to get a clear picture of the extent of the private education sector in Colombia by looking at the enrollment data shown in Table 2.3. Private education accounted for more than half of total enrollments in preprimary (85.2 percent) and secondary education (53.6 percent) in 1968. It can also be said that preprimary education has developed through private initiative, since enrollments in public schools were still negligible in 1960 As far as secondary education is concerned, the proportion of pupils enrolled in private schools decreased from 60.5 percent to 53.6 percent be- tween 1960 and 1968. However, in 1968, public schools enrolled a clear ma- jority of pupils (71.2 percent) only at the primary school teacher training level (escuelas normales). Before 1950, primary education was virtually a state monopoly, with close to 94 percent of the pupils enrolled in public schools. The last ten years have witnessed an increase in private participation at this level, which accounted for 19 percent of total enrollments in 1968 The same evolution took place at the higher education level. In 1950, more than 70 percent of the students were enrolled in public universities. This proportion has decreased steadily over the last twenty years It may be said that, at the beginning of the seventies, the public and private sector are now on the same footing, with 50 percent of the student body. '0Unpubiished data show that this trend is continuing the amount of transfers practically doubled in current value between 1966 and 1971 14 An important feature of the private education sector is its distribution: broadly speaking, private schools are to be found mostly in urban areas. Thus, in 1968, only 11 percent of total private enrollments were in rural schools, as against 42.4 percent of public enrollments. The bias of private education towards big cities is even stronger in the case of bachillerato education, which accounts for 70 percent of total secondary education: in 1968, private enrollments in the provinces of Antioquia, Atlantico, Cun- dinamarca and Valle-which include the four biggest cities, namely Mede- llin, Barranquilla, Bogota and Cali-represented 73 percent of total private Table 2.3: Colombia: Enrollments by Level and Type of Education in the Public and Private Sectors 1960 1968 1940 1950 1960 1968 Levels of Education (Thousands) (Thousands) (%) (%) (%) (%) PrePrimary Public 1 9 16.3 - - 4.5 14.8 Private 40.5 94.2 - - 95 5 85.2 Primary Public 1,432.2 2,213.4 92 9 93.8 84.7 81 0 Private 258.1 520.0 7 1 6.2 15.3 19.0 Secondary: Total Public 100.3 272.8 - - 39.5 46.4 Private 153.5 313.9 - - 60.5 53 6 Bachilleralo Public 50.7 177 6 33 0 39.3 36.1 43.8 Private 89.6 228.2 67 0 60 7 63.9 56.2 Teacher Training Public 18.2 38.6 80.1 83.0 64 9 71.2 Private 9 8 15.6 19 9 17.0 35.1 28.8 Othersi Public 31.3 56.7 - - 36.7 44.7 Private 54.12 70.1 - - 63.3 55.3 Higher Education Public 13 2 35.6 68.0 71.8 58.4 52.7 Private 9.4 31.9 32.0 28.2 41.6 47.3 llncludes all kinds of secondary education but baclullerato and teacher training, i.e., vocational, commercial, agricultLiral education, schools for nurses and arts 21ncluding religious education Sources Boleti, Mensuat de Estadistica 243 (October, 1971) 144-46. The data for 1940 and 1950 are drawn from Miguel Urrutia and Clara Elsa de Sandoval, "Reforma de la educaci6n publica y sus implicaciones fiscales," mimeographed, (Universidad Nacional de Colombia, 1970). "La educaci6n en Colombia, 1960-1968," Anexo Estadist,co, (Bogota. Mmnisterio de Educaci6n Nacional, 1970). Ceoso de establecitie,itos educanivos (Bogota: DANE, 1968). 15 enrollments at this level. The corresponding data for the public education sector is 48 percent. By far the main source of private education finance Is the families' con- tribution in the form of fees and gifts. No exhaustive data exist on the former which, in any case, are believed to vary widely. The importance of the latter may increase periodically whenever a temporary freeze on the tui- tion level is ordered by the government. Many private schools then attempt to maintain their real income by asking parents to increase their voluntary contributions. The private education sector also receives financial help from the four levels of government, although available data do not presently permit the amount of public subsidies going toward private education to be computed with any reasonable degree of accuracy. While the bulk of the income of private schools and universities comes in the form of fees paid by the pupils and gifts, these institutions do receive financial help from the central government, the departments and the mu- nicipalities. At least one decentralized agency, namely ICFES, contributes by channeling public funds into private institutions of higher education. Broadly speaking, private education receives public subsidies under the following forms: (i) direct "cash" grants to schools or universities; (ii) teachers en comisi6n, which means teachers working in private schools but paid by the government; (hi) scholarships to pupils enrolled in private schools; (iv) legal tax-relief arrangements. While there is absolutely no way of computing the amount of public money that goes into private education at the departmental and municipal level, it is theoretically possible to present a crude evaluation of the amount of cash grants originating from the central government. The ICFES series on university finance also permit some reasonable speculations about the amount of public finance supporting private universities. The absence of any statistics on the number of teachers en comisi6n is particularly troublesome, given the importance of teachers' pay in the total costs of education at the primary and the secondary level. As to the volume of scholarships to pupils enrolled in private schools, it is probable that ICETEX," to whom the responsibility for managing scholarship funds and selecting suitable candidates was transferred from the Ministry of Educa- tion in 1971, will be able to provide some reliable data in the near future. At the central government level, the bulk of direct financial help to educational institutions is spelled out in "Program 327" of the budget of the lnstituto Colombiano de Credito Educativo y Estudios Tkcnicos en el Exterior 16 Ministry of Education."2 Though the data are broken down by department, level of education and individual school, there is no immediate way of deciding which schools are private or public.'3 In addition, the statistics shown in the budget are of little help because they refer to initial appropria- tions. Execution budget data broken down in the same way and suffering from similar shortcomings are provided by the Contraloria General de la Republica.'4 Keeping these shortcomings in mind, it was decided to adopt the already mentioned rule-of-thumb that 10 percent of the Ministry of Education bud- get goes into private education. The calculations have been made for the year 1970 and are presented in Annex 2. While the margin of error affecting these rather crude data is likely to be very large, it is safe to say that they probably underestimate the actual amount of public help to private educa- tion, since they refer to some central government subsidies only, thus ex- cluding subsidies under the form of tax relief and scholarships and the sub- sidies from the departments and the municipalities. At the higher education level the situation is more satisfactory, thanks to the unpublished ICFES series on the financing of public and private universities. Implications for the Income Redistributive Aspects of Public Expenditures on Education The foregoing considerations have important implications for the study of the impact of public educational expenditures on income distribution. First, on the financing side of these expenditures, the fact that the various government levels contribute significantly to the financing of education makes it necessary to study the various taxation systems of the country in order to determine the overall incidence of all tax payments. Secondly, considering the benefits side of public educational expen- ditures, the existence of an important private education sector, partially supported by public funds, may introduce a strong and biased corrective factor in the allocation of public education subsidies, insofar as private education is unequally distributed among income groups. Thirdly, the vigorous policy of transferring funds from the central government to regional units, namely the departments, may have an impor- tant distributional impact on the allocation of educational expenditures be- tween departments. Clearly, the expenditures incurred by the departments themselves and the municipalities do not have any geographical redistribu- tive impact since they are made within their own respective administrative limits. '2Presupuesto de Colombia (Bogotai: Direcci6n General del Presupuesto, 1970), p. 245 13ICFES has published a code of educational institutions which may be used for this purpose '41nforme Financiero, vol III (Bogotb: Ejecuci6n Presupuestal, n.d.). 17 Table 2.4: Colombia: Revenue Structure by Levels of Government (1967) Government Levels/ Central Revenue Sources Government Departments Municipalities Bogota Total Revenuc 100.0 1000 100.0 100.0 Total Capital Revenue 16 7 6 7 15.2 19.3 Total Current Revenue 83 3 93.3 84 8 80 7 Total Fiscal Revenue 81 0 46 2 23 5 18 2 Direct Taxes 40 1 2.7 17.3 14 9 (of which Income Tax) 38 3 - - - Indirect Taxes 40.9 43 5 6 2 3.3 Other Revenue - 27.7 53 1 51.3 Transfers 2 3 19.4 8 2 11.2 Source Anuario de Fiscales y, F,,ialicieras clel DANE, 1966-67 (Bogota: DANE). The three preceding points will now be reviewed in turn. The general structure of taxation and its role as a source of revenue for the central government, the departments and the municipalities are shown in Table 2.4. First, it should be noted that fiscal revenue represents 81 percent of the central government's total revenue as against 46.2 percent in the depart- ments and 23.5 percent in the municipalities.is Second, about half of the central government's total fiscal revenue is attributed to direct taxes-which are, presumably, more progressive than indirect taxes-as against 6 percent in the departments As a consequence, while indirect taxes account for more than 40 percent of both the central government's and departments' total revenue, they account for 50 percent and 94 percent of their respective fiscal revenue As to the municipalities, less than a fourth of their total revenue comes from taxation, while more than 50 percent originates in the provision of services such as electricity, water, public transportation and the like At first glance, it can be said it is very unlikely that these three fiscal systems have the same degree of progressivity. The level of government fi- nancing education will, therefore, have implications for equity. Moreover, any switch in their relative contributions to the country's total fiscal revenue is bound to have an impact on the overall incidence of the three taxation systems and, therefore, on the distributional effects of public sub- sidies to education. A practical consequence of this state of affairs is that it will be necessary to carry out a different tax incidence study for each of the three levels of government. As mentioned earlier, one of our hypotheses is that the presence of a large private education sector, partially financed by public funds, is bound to introduce an important bias in the distribution of public subsidies among income groups. In Colombia, as in many other countries, children of higher i5This last proportion is even smaller in the case of big cities, as evidenced by the data shown for the municipality of BogotA 18 income groups have a higher propensity to enroll in private fee-paying schools than children from lower income groups. This will, of course, be truer for the children living in the larger urban areas where private educa- tion facilities are concentrated than in other areas, and for those enrolled in secondary education where private initiative is important than for those enrolled in primary education. In the case of the rural families in the lowest income brackets, the choice does not even exist, since the overwhelming majority of private and secondary schools are concentrated in urban areas. The amount of educational subsidies received by each income group is a function of its enrollments in public and private education, and of the amount of public subsidies received by both the public and private educa- tion sectors. Higher income groups in Colombia will, therefore, all else being equal, receive relatively smaller subsidies than in countries where there is no private education. The two above points will have important consequences on the actual impact of public educational expenditures on the distribution of income: see Chapter III. Turning now to the question of the transfers from the central govern- ment to the departments, it has been argued that these transfers merely reflect the inability of the departments' fiscal systems to raise enough revenue to finance their educational expenditures. More specifically, the lack of income elasticity of many departmental taxes has been emphasized. Whether and how the departments' fiscal systems should be overhauled in the sense of both more elasticity and more equity, to enable them to finance their expenditures in an appropriate way, is surely an important question. However, it is hard to argue that all transfers will or should disappear in the process. Given the already very limited powers of the central government in educational matters, this would be tantamount to giving up the idea of any countrywide educational policies or reforms. In any case, the transfers to the departments are presently an important source of geographical redistribution of educational opportunities. These transfers, which clearly represent educational subsidies from the central government to the departments, are financed out of central govern- ment revenue. The distribution of central government taxes according to their departments of origin will be considered here as equivalent to the regional incidence of national taxation, on the grounds that these taxes, if not levied by the government, would be levied by the departments them- selves. In other words, they can be considered as transfers of revenue from the departments to the central government. By combining the distribution of central government transfers for education among departments, and the departmental distribution of central government taxes, the overall geographical impact of the transfers policy carried out by the government will be assessed (see Chapter IV). 19 III. Public Expenditures on Education in Colombia: The Redistributive Impact among Income Groups This chapter attempts to assess the "net" incidence of public educational expenditures on the distribution of income for the year 1970. The analysis consists of three stages: first the distribution of the tax-burden among in- come-groups will be determined; second, education subsidies will be allo- cated among the same income-groups; and third the ratios of subsidies re- ceived over taxes paid will be computed for each income-group, in order to arrive at an overall picture of the impact of public educational expenditures on the distribution of income. The unit of analysis is the family which is considered as a more appropriate tax-paying and subsidies-receiving unit than individuals in the present study. The Distribution of Income among Colombian Households in 1970 The data on the distribution of income used in this study are those col- lected by DANE, the Colombian Statistical Office, in two recent household surveys,' and adjusted by McLure2 at the request of this study's author. In urban areas the original data provided by the DANE household bud- get surveys refer to the distribution of household incomes among fourteen income classes in the seven largest Colombian cities. These data were rather crudely blown up so as to give a rough approximation of the distribu- tion of household incomes in urban Colombia. According to this distribu- tion which is reported in Table 3.1, the lowest 30 percent of urban house- holds receive 8 percent of income accruing to all urban households, and the 'The various DANE surveys are described in P Cordoba, La encuesta nacional de hogares de Colombia (Bogota DANE, 1971) Data on the distribution of income in rural areas come from Project EH-1 and are reported in P Cordoba, Distribucion de ingresos en Colombia y su cont rbuct6n al proyecto de Naciones Unidas sobre el sistema complementario de estadisticas de la distribicion del ingreso (Bogota DANE, 1971) Data on the distribution of income in urban areas come from Project EH-2 and were provided in unpublished form by DANE 2See Charles McLure, "The Incidence of Colombian Taxes 1970," Discussion Paper no 41, mimeographed Program of Development Studies, Rice University Houston, Texas, 1973, pp 3-12 20 top 10 percent of households about 40 percent of total urban income. Because this income distribution refers to households, there is no way of knowing whether it is more or less equitable than past income distributions, which all refer to the economically active population.3 However, it is, as ex- pected, slightly less unequal than the 1970 income distribution among eco- nomically active persons.4 In rural areas, data for income distribution among economically active persons come from an earlier DANE survey. The problem was thus to con- vert this distribution among individuals into a distribution among house- holds. Although there is no fully satisfactory way of doing this, it was decided to use an approach based upon data on incomes of heads of house- holds, reported by income brackets for each educational level in rural areas.5 The results of the calculations are reported in Table 3.2. According to this income distribution, which includes only seven income classes, the lowest 40 percent of rural households receives approximately 10 percent of rural income and the top 10 percent about 35 percent to 40 percent of in- come. As pointed out by McLure, the distribution of income in rural areas is roughly similar to that of urban areas, but the average urban income is more than four times higher than the average rural income.6 The income distribution for all Colombian households was obtained by combining the distributions of urban and rural areas. The results are re- ported in Table 3.3. According to this estimate, the lowest 40 percent of all Colombian families receive about 8 percent of all income, and the top 13 percent about 50 percent of all income. Average income per household was roughly 30,000 pesos in 1970. The Distribution of the Tax Burden among Income-Groups The next step consists in allocating the major direct and indirect taxes to the income groups likely to pay them. We thereby arrive at an overall pic- ture of the incidence of the taxation system as given by the effective rates of taxation, i.e., the ratio of taxes paid over total income in each income group. This process of assessing the incidence of taxation, which was carried out separately in urban and rural areas, is rather delicate. The final results depend largely on the availability of reliable tax data and on a number of 3See Miguel Urrutia Montoya, "Resetia de los estudios de distribuci6n de ingreso en Co- lombia," Rewista del Banco de la Repilbihca XLIII (February 1970). 180-91. 4According to C6rdoba, D,strrbuciin de ingresos en Colombia, the lowest 30 percent of active persons in urban areas receive about 6 percent of urban income, and the top 10 percent about 44 percent of income. 5lbid , pp 80, 84, and 95, see also McLure, "The Incidence of Colombian Taxes," Table 3, Estimate 3 643,800 pesos against 10,300 pesos per year. 21 Table 3.1: Distribution of Income among Urban Households (1970) (Number of Households in Thousanids, Income in Billions of Pesos) Income Bracket Number of Total Cumulative % of (Pesos/Year) Households Income Households Income 0- 6,000 43.3 0 20 2.1 0.2 6,000- 12,000 257.8 2.38 14.7 2.9 12,000- 18,000 308.9 4 56 29.9 8.0 18,000- 24,000 309.2 6 39 45.0 15.1 24,000- 30,000 204 9 5 44 55.0 21 2 30,000- 36,000 153.6 5.03 62.5 26.8 36,000- 48,000 193.1 7.91 72.0 35 6 48,000- 60,000 148.3 7.87 79.3 44.4 60,000- 72,000 89.7 5 87 83 6 51.0 72,000- 84,000 74.9 5 76 87.2 57.4 84,000-120,000 121.0 12.10 93 1 70.9 120,000-180,000 75:2 10 65 96 8 82.8 180,000-240,000 38.1 6.51 98.7 90.1 Over 240,000 24 6 8 81 100.0 100.0 Total 2,042 6 89.52 Sources. DANE, Special Tabulations of Household Survey (EH2) and McLure, "The Incidence of Colombian Taxes," Table 2. Table 3.2: Distribution of Income among Rural Households (1970) (Number of Households in Thousands; Income in Billions of Pesos) Income Bracket' Number of Total Cumulative % of (Pesos/Year) Households' Incomel Householdsl Income' 0- 6,000 595.7 1 79 45.1 13.2 6,000- 12,000 421.3 3.80 77.0 41.2 12,000- 18,000 217 9 3.92 93.5 70.1 18,000- 24,000 24,000- 30,000 67.3 2.37 98.6 87.6 30,000- 36,000 36,000- 48,000 48,000- 60,000 60,000- 72,000 9.3 0 92 99.3 94.4 72,000- 84,000 84,000-120,000 120,000-180,000 1.3 0 31 99.4 96.7 180,000-240,000 Over 240,000 1 3 0.37 99 52 99.42 Total 1,320.8 13.48 - - 'Wider income bracket for rural households than for urban households were dictated by the format of the original data. Figures are given in the top Iine for the several narrower brackets included in the wider ones. 2Small numbers of persons and amounts of income were not allocated among income classes Sources C6rdoba, Disiribuct6n de ingresos en Colombia and McLure, "The Incidence of Colombian Taxes," Table 3, Estimate 3. Table 3.3: Distribution of Income among Colombian Households (1970) (Number of Households in Thousands; Income in Billions of Pesos) Income Bracket Number of Total Cumulative % of (Pesos/Year) Households Income Households Income 0- 6,000 6390 1 99 19 0 1.9 6,000- 12,000 679.1 6.18 39.2 7 9 12,000- 24,000 8360 1487 64.1 22.4 24,000- 60,000 767.2 28.63 87 0 50.2 60,000-120,000 294.9 24.66 95.8 74 1 120,000-240,000 114.6 17.48 99.2 91 1 Over 240,000 25.9 9.18 100.0 100.0 Total 3,356 7 102.99 - Soioces See Tables 3.1 and 3.2. hypotheses concerning tax-shifting, that is to say, the fact that the persons who pay the taxes are not necessarily those who bear the actual burden. In the case of Colombia, this field has already been pioneered by various researchers and the data, although still far from excellent, are adequate. The thorny question of tax-shifting was dealt with by presenting alternative and extreme assumptions whenever necessary in order to bracket the "true" in- cidence of the tax system.7 The various taxes will now be treated separately: Direct Taxes. The allocation of the personal income tax among income groups was made on the basis of unpublished tax-return data for 1967, sup- plied by the Ministry of Finance and adjusted to the year 1970. It was assumed that the personal income tax was actually borne by those who paid it. The data, which were supplied on an individual tax-payer basis, had to be adjusted to bring them on a household basis. All these calculations led to effective tax rates under the personal income tax, ranging from zero for the lowest income group, to 15 percent for the top income group.8 It is impor- tant to mention that these results refer to liabilities and not to actual tax payments. Taxation rates may thus be overestimated to the extent that ac- tual tax collections do, in fact, fall short of tax liabilities. Three assumptions concerning the incidence of the corporation income tax were made, namely: that it is borne entirely by upper income Colom- bian shareholders, that it is shifted by Colombian consumers, and that it is borne by foreign shareholders. Because of the arbitrariness of a choice 7This section draws heavily on McLure, "The Incidence of Colombian Taxes" This paper- which was originally specially commissioned to Professor McLure by the author of the present study-represents a very thorough updating of McLure's earlier work on the subject, "The Incidence of Taxation in Colombia," in Fiscal Reformfor Colombia: The Final Report and Sraff Papers of rhe Colombian Commission on Tax Reform, ed Richard A Musgrave and Malcolm Gillis (Cambridge, Mass. Harvard Law School International Tax Program, 1971) 8See McLure, "The Incidence of Colombian Taxes," Table 12 23 among these alternatives and of the sensitivity of actual tax incidence to such a choice, it was decided to present two alternatives: alternative I based on the least progressive assumption about the incidence of the corporation tax, i.e , the assumption according to which the tax is shifted to consumers roughly in proportion to their non-food expenditures; and alternative 2 based on the most progressive assumption about the incidence of the cor- poration tax, i.e., the assumption according to which the tax is borne by Co- lombian shareholders who are assumed to be concentrated in the top three income brackets. Alternative I leads to a taxation rate of roughly 3 percent for all income groups and alternative 2 to taxation rates of 6.7 percent, 10 percent and 13.3 percent for the top three income groups (from bottom to top).9 Coffee export duties were responsible for about 22 percent of central government tax revenue in 1970. It is not clear, however, whether the revenue raised went to general revenue or to finance the activities of the coffee sector. In the former case, coffee duties should be treated as any other tax, whereas in the latter they could reasonably be ignored. This is a serious problem from the point of view of tax incidence because coffee duties constitute a strongly progressive element of taxation within the rural sector. If they are excluded, the taxation system will appear to be much less progressive than it is when they are included. Here too it was decided to use two polar alternatives-alternative I (minimum progressivity) according to which coffee duties are excluded, and alternative 2 (maximum progressivity) according to which they are included-in order to bracket the actual incidence of taxation in Colombia in an appropriate manner. The property taxes levied at the municipal level have been allocated among income groups according to actual household payments for this tax, as reported in the DANE household budget survey (Project EH-2). For most urban households, property taxes amount to less than 0.5 percent of income. Only in income groups above 120,000 pesos per year does the burden of these taxes rise to I percent to 2.5 percent of income. In rural areas, the rate of taxation is insignificant (less than 0.1 percent of income). To summarize, the effective rates of direct taxes range from 4.6 percent of income in the poorest income group to 18.8 percent of income in the richest, according to the least progressive hypothesis,'0 and from 5.7 per- cent to 29.3 percent in the most progressive hypothesis." Indirect Taxes. Indirect taxes were allocated among income groups in ac- cordance with the consumption patterns reported in the DANE household 91bid ,Table 13 0Corporation income tax borne by shareholders and coffee export duties excluded ilCorporation income tax borne by shareholders and coffee export duties included See McLure, "The Incidence of Colombian Taxes," Table 18 24 budget survey. The various national,'2 departmental'3 and munici- pal'4 taxes were allocated separately, according to consumption of taxed items. The overall rate of indirect taxation is roughly 6 percent to 7 percent of income and varies very little among income groups.'5 In other words, in- direct taxation is roughly proportionate to income in Colombia. Further- more, its burden is roughly the same in urban and rural areas. These global conclusions are not the result of strongly offsetting patterns between taxes'6 and deviations from the pattern of proportionality are small. When both direct and indirect taxes are combined, effective rates of tax- ation imposed by all Colombian taxes can be computed. These rates are re- ported separately for urban and rural areas, using two alternatives in each case: alternative 1 of minimum progressivity which assumes that the cor- porate income tax is borne by consumers and which excludes coffee export duties, and alternative 2 of maximum progressivity which assumes that the corporate income tax is borne by shareholders and which includes coffee export duties. The results are shown in Tables 3.4 through 3.9. In urban areas, according to alternative 1, the tax system is roughly pro- portionate to income for low and middle income households, and mildly progressive for the top two income groups, which comprise 3.2 percent of all urban households. In the case of alternative 2, the tax system is still pro- portional for the lower ten income groups, but it becomes more strongly progressive for the top four income groups, which comprise 6.9 percent of all urban households. In the rural sector, the tax structure is proportionate to income if coffee export duties are excluded (alternative 1), but shows considerable progressivity if they are included (alternative 2). Total taxes paid by rural households amount to 9.5 percent of their income in the first case, as against 22.6 percent in the second case. The proportion for all urban house- holds is 13.3 percent. For the country as a whole, according to alternative I which yields minimum progressivity, the incidence of taxation is roughly the same-around 1I percent of income-for all incomes under 120,000 pesos, that is to say for 95.8 percent of all Colombian households. For incomes over 120,000 pesos, that is for less than 5 percent of all households, the average rate of taxation is 18 percent, while the average for the country as a whole is 12.8 percent (see Table 3.8). '2Tax on gasoline, customs, stamps and sales tax 3Tobacco, alcohol, beverages and departmental sales tax 14Automobile registration, tax on industry and commerce '5See McLure, "The Incidence of Colombian Taxes," Table 22 16There is no strongly progressive tax the influence on total indirect taxation of which is offset by another very regressive tax 25 Table 3.4: Urban Areas: Allocation of Taxes according to Alternative I (Minimum Progressivity) Income Bracket Income Effective Rates of Allocation of Taxes (Pesos/Year) (Billions of Pesos) Taxation (%) (Mtllions of Pesos) (1) (2) (1) x (2) 0- 6,000 0.20 11 5 23 6,000- 12,000 2 38 7 7 183 12,000- 18,000 4 56 9.6 438 18,000- 24,000 6 39 10 3 658 24,000- 30,000 5 44 10.3 560 30,000- 36,000 5 03 10.9 548 36,000- 48,000 791 120 949 48,000- 60,000 787 105 826 60,000- 72,000 5 87 11 5 675 72,000- 84,000 5.76 10.7 616 84,000-120,000 12 10 12.4 1,500 120,000-180,000 11.50 12.2 1,405 180,000-240,000 7 36 14 2 1,048 Over 240,000 10 52 27.5 2,894 Total 92.89 13 3 12,323 Soiotce McLure, "The Incidence of Colombian Taxes," Tables 7 and 23 Table 3.5: Urban Areas: Allocation of Taxes according to Alternative 2 (Maximum Progressivity) Income Bracket Income Effective Rates of Allocation of Taxes (Pesos/Year) (Billtons of Pesos) Taxation (%) (Millions of Pesos) (1) (2) (1) x (2) 0- 6,000 0.20 7 1 14 6,000- 12,000 2 38 5 0 119 12,000- 18,000 4.56 6.7 306 18,000- 24,000 6.39 7.1 454 24,000- 30,000 5 44 7.2 392 30,000- 36,000 5.03 7.6 382 36,000- 48,000 791 8.7 688 48,000- 60,000 7 87 7.5 590 60,000- 72,000 5.87 8 8 517 72,000- 84,000 5.76 7.9 455 84,000-120,000 12.10 9.4 1,137 120,000-180,000 12.32 15 4 1,897 180,000-240,000 8 18 19.9 1,628 Over 240,000 12.14 34 8 4,225 Total 96.15 13.3 12,804 Somice McLure, "The Incidence of Colonmbian Taxes," Tables 7 and 23. 26 Table 3.6: Rural Areas: Allocation of Taxes according to Alternative 1 (Minimum Progressivity) Income Bracket Income Effective Rates of Allocation of Taxes (Pesos/Year) (Billions of Pesos) Taxation (%) (Millions of Pesos) (1) (2) (1) x (2) 0- 6,000 1.79 11.2 200 6,000- 12,000 3.80 8.6 327 12,000- 24,000 3.92 9.5 372 24,000- 42,000 1.94 95 5 225 42,000- 60,000 0.43 1 1 60,000- 90,000 0.53 9 8 90,000-120,000 0.39 120,000-240,000 0.31 9.9 31 Over 240,000 0.37 10.3 38 Total 13.48 9 5 1,280 Soutice See McLure, "The Incidence of Colombian Taxes," Tables 9 and 24 Table 3.7: Rural Areas: Allocation of Taxes according to Alternative 2 (Maximum Progressivity) Income Bracket Income Effective Rates of Allocation of Taxes (Pesos/Year) (Billions of Pesos) Taxation (%) (Millions of Pesos) (1) (2) (1) x (2) 0- 6,000 1.91 13.5 258 6,000- 12,000 4.03 11.2 451 12,000- 24,000 4.31 14.8 638 24,000- 42,000 2 24 2 8 42,000- 60,000 0 85 1 l 60,000- 90,000 0 89 ( 90,000-120,000 0.55 1 4 120,000-240,000 0.70 58.6 410 Over 240,000 0.81 57.4 465 Total 16.29 22.5 3,670 Source See McLure, "The Incidence of Colombian Taxes," Tables 9 and 24. Table 3.8: Total Colombia: Allocation of Taxes according to Alternative 1 (Minimum Progressivity) Income Bracket Income Effective Rates of Allocation of Taxes (Pesos/Year) (Billions of Pesos) Taxation (%) (Millions of Pesos) 0- 6,000 1 99 11.2 223 6,000- 12,000 6.18 8.3 510 12,000- 24,000 14.87 9.9 1,468 24,000- 60,000 28.62 10.9 3,108 60,000-120,000 24.65 11.7 2,878 120,000-240,000 19.17 12.9 2,484 Over 240,000 10.89 26.9 2,932 Total 106.37 12.8 13,603 Soutrce See Tables 3.4 and 3.6. Table 3.9: Total Colombia: Allocation of Taxes according to Alternative 2 (Maximum Progressivity) Income Bracket Income Effective Rates of Allocation of Taxes (Pesos/Year) (Billions of Pesos) Taxation (%) (Millions of Pesos) 0- 6,000 2 11 12.9 272 6,000- 12,000 6.41 8.9 570 12,000- 24,000 1526 9.2 1,398 24,000- 60,000 29.34 10.0 2,920 60,000-120,000 25.17 10.7 2,689 120,000-240,000 21 20 18.6 3,935 Over 240,000 12.95 36.2 4,690 Total 112.44 14.7 16,474 The discrepancies between the income distributions shown in Tables 3.1-3.3 and those recorded in Tables 3.4-3.9 are due to: -an adjustment for corporate retained earnings which have been allocated to the three top income brackets in Table 3.4 in which, it is recalled, corporation taxes are shifted entirely to Colombian consumers; -an adjustment for both corporate retained earnings and income tax which have been allocated to the three top income brackets in Table 3.5 in which corporation tax is borne entirely by Colombian shareholders; -the inclusion of coffee export duties in Table 3.7. Source. See Tables 3.5 and 3.7. According to alternative 2, the average taxation rate for the country as a whole is 14 7 percent, slightly higher than recorded under alternative I because of the inclusion of coffee export duties. Here too, the incidence of the tax system is proportionate to income-around 10 percent-for all in- comes under 120,000 pesos. Over 120,000 pesos, the average rate of taxation reaches 25 percent of income The taxes that contribute most to progressivity are, at the national level, the personal income tax and the corporate income tax if it is attributed to shareholders. Alcoholic beverage taxes at the departmental level and prop- erty taxes at the municipal level are also progressive at the upper end of the tncome scale. The Distribution of Public Subsidies for Education among Income Groups In order to allocate the public subsidies for education to the various in- come groups, it is necessary, first, to determine the structure of enrollments in each level of education by level of income, then to distribute enrollments in the public and private sector and, lastly, to compute the amount of public expenditures per pupil and per level in each sector. The total subsidies re- ceived by each income bracket will be obtained by multiplying the number 28 of pupils enrolled in each level and sector by the corresponding public ex- penditure per pupil. Data on enrollments by income levels in urban areas were provided in unpublished form by the Colombian Office of Statistics. The source of the data is the same household survey already used previously to extract data on the distribution of income.'7 The questionnaire used in this survey in- cluded data on levels of educational attainment and on the present school situation of each member of more than twelve years of age in each house- hold. The level of educational attainment of members of households who declare at the same time that they are still attending school was adopted as a proxy for the structure of enrollments by income levels. For children be- tween five and twelve years of age, only the level of educational attainment assessed by the number of grades passed in primary education was avail- able. In order to determine whether a person declaring, say, three years of primary education was still attending school or not, the number of grades passed was compared with age and the following rules-of-thumb were adopted for all income groups: (i) whenever a child was four years or more behind in his school career, he was considered as having dropped out; (ii) 50 percent of children who were three years behind were considered as having dropped out; (iii) all other children were considered as still attending pri- mary school except (iv) children who declared they had passed the fifth and last grade of primary education; those children were considered as enrolled in lower secondary education. These rules clearly imply an assumption about the age at which children are supposed to start primary school. In order to take into account the well- known problem of late entry into Colombian education, it was assumed that the age corresponding to the first year of schooling was seven. Thus, a ten-year-old child who should be in the fourth grade but declares that he has not yet passed the first grade was considered a drop-out. The assumption of seven years of age for children enrolled in the first grade is not equally true for all income groups. According to the survey, 38 percent of all seven-year-old children in urban areas are already enrolled in grade 2 or 3; however, this proportion drops to 15 percent to 25 percent for the low income groups, and rises to 50 percent to 60 percent for the high in- come groups. In other words, high incomes are associated with early entry and a low probability of a large age/grade discrepancy, that is with a low drop-out rate; on the contrary, low incomes are associated with late entry and a high probability of a large age/grade discrepancy, that is with a high drop-out rate. All in all, though these assumptions remain arbitrary, they seem reasonable enough in a country like Colombia. A final adjustment was '7Project EH-2, in DANE terminology. 29 Table 3.10: Urban Colombia: Enrollments by Level of Family Income Number of Enrollment in all Income Bracket Children Levels Primary Secondary Higher (Pesos/Year) Age 5-25 (N) (%) (%) (Thousands) % Thousands 0- 6,000 III 48 5 54 88.0 11.5 - 6,000- 12,000 443 43 1 191 96.2 3.6 0.2 12,000- 18,000 670 45.4 304 90.3 9.4 0.3 18,000- 24,000 745 48 2 359 80.6 18.7 0.7 24,000- 30,000 600 48.7 292 78 4 21 4 0.2 30,000- 36,000 423 47 6 201 76.3 21.9 1.8 36,000- 48,000 549 52.0 285 66.3 31.1 2.6 48,000- 60,000 358 544 195 52.7 42.0 5.3 60,000- 72,000 283 52.9 149 54.2 38.1 7.7 72,000- 84,000 217 56.8 123 55.3 40.2 4.5 84,000-120,000 278 58.3 162 47 5 45.1 7.5 120,000-180,000 213 64.9 138 45 9 43.3 10.8 180,000-240,000 107 66.2 71 39.1 52.2 8.7 Over 240,000 114 61 9 70 40.7 48.3 11.0 (%) 50.8 100.0 10.0 26.8 3.2 Total (Thousands) 5,1101 2,596 1,8162 6963 843 'Approximately 50 percent of all Colombians between five and twenity-five are assumed to live in urban areas. 21t is assumed that 60 percent of pupils enrolled in primary education are enrolled in urban areas No adjustment was made for the fact some children from rural families may attcnd schools located in urban areas. 31t is assumed that all enrollments in secondary and higher education are in urban areas. Souirce See Text Unpublished tabulations based on the DANE Household Survey (Project EH2). made to "blow up" the results yielded by the survey tn order to make them representative of urban Colombia as a whole. The expansion factors used for primary, secondary and higher education were 1.37, 1.78 and 2.07 respectively; it was 2 19 for the total number of children. The results of these calculations are shown in Table 3 10. About half of the five to twenty-five age group is enrolled in some level of education. While enrollment ratios are below 50 percent for the lowest six income groups, they are above 60 percent for the top three. The struc- ture of enrollments by level of education also differs according to income. In the lowest five income groups, more than 80 percent of the students are enrolled in primary education, and close to 20 percent in secondary educa- tion. Enrollments in higher education represent less than 1 percent of the total. In the top five income groups, enrollments in primary and secondary education account for 90 percent of the total (half in primary and half in secondary); the share of higher education is close to 8 percent to 10 percent. It is thus clear that the children coming from families with higher in- 30 comes are enrolled in greater proportions than their less privileged counter- parts. Furthermore, these children are more likely to be enrolled in second- ary and higher education than the children from low income groups. The next step is to determine the relative shares of the public and private sector in enrollments at each level of education and for each income group. In this connection, the only information available in Colombia is that of the National Testing Service. It refers to the bachilleres, the students enrolled in the last year of secondary education. Their application forms for the tests include data on family income and the type of educational institu- tions-public or private-in which they received their secondary education. This information, once processed, shows very clearly that the share of the public sector in enrollments, preponderant in low income groups, declines steadily as income rises with a corresponding increase of the share of the private sector.'" This information was "blown up" to total enrollment in secondary, pri- mary and higher education successively, subject to the constraint of the overall shares of the public and private sectors in each level. The results of these calculations are recorded in Table 3.11.'9 The emerging pattern of enrollment is clear: public educational institutions in Colombia are mainly attended by the "poor" and an overwhelming proportion of children com- ing from upper income families are educated in the private sector. Moreover, the propensity of Colombian families to send their children to private schools rises steadily with income. All in all, the private sector is nearly as important as the public sector for secondary and higher education, and accounts for close to 28 percent of urban primary education. Needless to say, this is going to have an important impact on the pattern of subsidies received by the various income groups, since private education is far less subsidized than public education. The DANE household survey mentioned above provides very little in- formation on enrollments by level of family income in rural Colombia. A few families living in the rural "belt" around the big cities were included in the sample, but they are clearly not representative of the rural world as a whole. There is an obvious data problem here which the "guesstimates" recorded in Table 3.12 do not pretend to solve. The distribution of children between five and twenty-five years of age among income groups is assumed to be the same as the distribution of households shown in Table 3.2.20 Less 18See Jean-Pierre Jallade, "The Financing of Education an Examination of Basic Issues," IBRD Staff Working Paper no 157 (Washington, D C. World Bank, 1973), p 34 19There is obviously some arbitrariness in this procedure since it impiies that the relative propensities of the various income groups to enroll their twelfth-grade children in pubiic or private schools are repeated throughout the education system. 20This assumption obviously implies that family size is the same for all income levels It relies on the belief that high- or middle-income families living in rural areas may not have the same incentives to have fewer children than low-income families as they have in urban areas 31 Table 3.11: Urban Colombia: Enrollments by Level of Family Income in Public and Private Schools (%) Income Brackct Primary Secondary Higher (Pesos/Year) Public Private Public Private Public Private 0- 6,000 95 0 5.0 95.0 5.0 - - 6,000- 12,000 90 0 10.0 90 0 10 0 100 0 - 12,000- 18,000 85 0 15.0 85.0 15.0 100.0 - 18,000- 24,000 89 1 109 79 9 20.1 100.0 - 24,000- 30,000 89.1 10.9 81.0 19.0 95.0 5.0 30,000- 36,000 895 105 81.0 190 900 10.0 36,000- 48,000 64.8 35.2 589 41.1 85.7 143 48,000- 60,000 54.6 45.4 49 7 50.3 86.3 13 7 60,000- 72,000 42.5 57.5 38.7 61.3 67.1 32.9 72,000- 84.000 34.5 65 5 31.3 68.7 54.3 45.7 84,000-120,000 22.4 77.6 20.3 79 7 35.2 64.8 120,000-180,000 20.3 79.7 18.4 81.6 32 0 68.0 180,000-240,000 11 9 88.1 11 1 88.9 19.2 80.8 Over 240,000 12.0 88 0 11.0 89.0 19.3 80.7 Total (%) 72 5 27.5 49.0 51 0 54.8 45 2 (Thousands) 1,316 500 341 355 46 38 For the lowest three income groups the "blowing up" of the information available for the twelfth graders lcd to a share of public enrollment in primary and secondary educa- tion above 100 percent. This share had to be adjusted downwards arbitrarily and the "surplus" students were allocated proportionately to the remaining income groups The same adjustment was made with the share of public enrollment in higher education for the lowest five income groups Source: Derived from unpublished data of the National Testing Service (Servicio Nacional de Pruebas), ancd ICFES. Table 3.12: Rural Colombia: Enrollments by Level of Family Income Enrolled in Primary Enrolled in the Number of Children Enrolled on Public Private Income Bracket Age 5-25 Eucation Sector Sector (Pesos/Year) (Thousands) % Thousands (%) (%) 0- 6,000 2,305 15 346 100 - 6,000- 12,000 1,630 23 375 100 - 12,000- 24,000 843 33 278 100 - 24,000- 60,000 261 50 130 100 - 60,000-120,000 36 1001 36 50 50 120,000-240,000 5 1001 5 - 100 Over 240,000 30 1001 30 - 100 Total 5,100 23 1,200 96 4 'The 100 percent enrollment ratios for the three upper income groups imply that the children beyond primary school age in those groups havc migrated to cities to get secondary or higher education, and are no longer included in their original consump- tion-tax-paying unit Souirce See Text Table 3.13: Total Public Expenditures per Pupil Direct1 Indirect2 Total Urban Arcas3 Rural Areas3 Primary Education Public Sector 731 115 846 1,105 553 Private Sector 154 59 213 222 III Secondary Education Public Sector 2,830 450 3,820 - - Private Sector 167 48 215 - - Higher Education Public Sector 19,000 2,848 21,848 - - Private Sector 921 395 1,216 - - tAnnex 2. 2See Text. 31t is assumed that the ratio of public expenditures per pupil enrolled in primary educa- tion in urban areas over rtiral arcas is similar to the ratio of the corresponldilig tcachers wage bills, 2.1, as shown in Table 3.14 than a quarter of this age group, that is 1.2 million persons, is likely to be enrolled in rural primary education-the only education available in rural areas. It is assumed that enrollment ratios rise from 15 percent for the lowest income group, which includes 45 percent of the eligible population, to 100 percent for the top three income groups, which account for 14 per- cent of the eligible population. About 50,000 children-that is, 4 percent of the total enrolled in rural areas-are enrolled in rural private schools, and family income was granted the same importance as in urban areas in deter- mining the propensity to use private schools instead of public schools. As a result, children enrolled in public education are concentrated at the bottom of the income scale, while children enrolled in private education are ex- pected to come from the top three income groups only (see Table 3.12). The next step is to compute the amount of public expenditures per pupil enrolled in each level of education, in the private and public sectors, in ur- ban and rural areas. The total expenditures on education incurred by the four levels of government between 1960 and 1970 are recorded in Annex 1. A breakdown of these expenditures by level of education for the year 1970 is presented in Annex 2, together with rough approximations of the amount of public expenditures allocated to private education. The results of these calculations are shown in the first column of Table 3.13. In addition to these "direct" expenditures per pupil, the four levels of government incurred about 635 million pesos worth of "indirect" expenditures2' related to pri- mary, secondary or higher education. These expenditures, which are mostly 2iThat is, 1,045 million pesos (see column "Other" in Annex 2) minus 410 million pesos which are allocated to the National Apprenticeship Scheme (SENA) not included in the analysis 33 administrative, were allocated to each level of education according to their share in total "direct" expenditures Furthermore, it was assumed that 90 percent of these "indirect" expenditures went into the public sector, as against 10 percent into the private sector The results of these calculations are shown in the second column of Table 3.13. A final adjustment was made in the amount of expenditures per pupil in primary education, in order to include the urban-rural differences in costs. This adjustment rests on the assumption that the ratio of total expenditures per pupil in urban areas over rural areas is similar to the observed ratio of the corresponding teacher costs per pupil. The calculations are shown in Ta- ble 3.14, where the total wage bill corresponding to each of the five catego- ries of the teaching body is shown Each total is then allocated to urban and rural areas according to the proportion of teachers. As the well paid, well qualified teachers (first and second category) are much more numerous in urban areas than in rural areas and vice-versa, since the underqualified teachers are predominantly rural, urban areas account for a much bigger proportion of the wage bill than rural areas, even when the differences in enrollment are brought into the picture To be sure, while urban enroll- ments account for 53 percent of total enrollments, urban teachers represent 66 percent of the teaching force, and they get close to 70 percent of total wages All these differences result in a wage bill per pupil in urban areas twice as high as in rural areas (see Table 3.14) As already mentioned, the differences in teacher costs per pupil between urban and rural areas pro- vided a basis for the estimates shown in Table 3.13, where public expen- ditures per pupil in urban primary schools are twice as high as those in rural primary schools. The amount of public subsidies for education received by each income group was then computed by multiplying the expenditures per pupil for each level of education and in each sector by the corresponding enroll- ments The results of the calculations are shown in Tables 3.15 and 3.16 for urban and rural Colombia respectively. In 1970 urban households received about 3.8 billion pesos of public subsidies for the education of their children: 41 percent of this amount goes to primary education, 31 percent to secondary and 28 percent to higher education The bulk of the subsidies to primary education are predictably concentrated toward the lower part of the income scale. 73 percent of the total goes to families earning less than 48,000 pesos a year. Conversely, subsidies to higher education tend to go to high income groups: 70 percent of the total goes to families earning more than 48,000 pesos a year. Lastly, 66 percent of all the subsidies to secondary education are received by middle income families earning between 18,000 and 60,000 pesos. As a result, the average subsidy per child enrolled is around 1,500 pesos. It shows a tendency to be slightly lower than the 34 Table 3.14: Primary Education (Public): Teachers' Costs in Urban and Rural Areas Total Ist 2nd 3rd 4th 5th Abs % Total Wage Billl (Millions of Pesos/Monthi) 24.5 24.1 8.3 4.5 10.0 71 4 Proportion of tcacliers i1n2 -Urban Areas (%) 89 1 79 0 52 4 41.3 27 4 66.1 -Rural Areas (%) 10.9 21.0 47.6 58.7 72.6 33.9 Wage Bill in: -Urban Areas (Millions of Pesos/Month) 21 8 19 0 4 3 1.9 2 7 49.7 69.6 -Rural Areas (Millions of Pesos/Month) 2 7 5.1 4.0 2 6 7.3 21.7 30.3 Enrollments3 in. -Urban Areas 1,316,000 53.0 -Rural Areas 1,165,000 47.0 Wage Bill per Pupil/Month -Urban Areas (Pesos) 38 -Rural Areas (Pesos) 19 'Derived from Table 54, "La educaci6n en Colombia, 1960-68" 2See, Lebot, "Organizaci6n y funciolaniiento del sistema Cducativo Colomblano," Tables 7, 4, and 3. 3Annex 2 (fn. 8), and estimatcs based on previous years. average for the lower income groups (under 30,000 pesos), higher than the average for the middle income groups (from 30,000 to 72,000 pesos) and lower than the average for the upper income groups (over 72,000 pesos). In the same way, the aver.age subsidy per child-enrolled or not-which amounts to 750 pesos, is higher in the middle income groups than in either the bottom or the top of the income scale. In other words, the urban middle class appears to benefit more from the public subsidies to education than the other social groups. About 640 million pesos of public funds go into primary education in rural Colombia. The subsidy per child enrolled in rural areas is roughly the same-about 550 pesos-for all households under 60,000 pesos, that is, for 98.6 percent of all rural households. Because of their low enrollment ratios, low income groups receive a smaller subsidy per child than high income groups. On the whole it can be said that the distribution of public subsidies for education among income groups in rural areas reinforces existing in- come inequalities. 35 Table 3.15: Urban Colombia: Public Subsidies for Education by Level of lncome Subsidies Subsidies per Child Primary Secondary Higher Total per Child in each Income Bracket (Millions (Millions (Millions (Millions Enrolled Income Group (Pesos/Year) of Pesos) of Pesos) of Pesos) of Pesos) (Pesos) (Pesos) 0- 6,000 51 20 - 71 1,315 640 6,000- 12,000 187 20 10 217 1,136 490 12,000- 18,000 267 80 23 370 1,217 552 18,000- 24,000 292 181 57 530 1,476 711 24,000- 30,000 232 169 12 413 1,414 688 30,000- 36,000 155 119 71 345 1,716 816 36,000- 48,000 150 179 148 477 1,674 869 48,000- 60,000 72 142 196 410 2,103 1,145 60,000- 72,000 48 80 173 301 2,020 1,064 72,000- 84,000 36 58 69 163 1,325 751 84,000-120,000 32 61 103 196 1,210 705 120,000-180,000 25 47 117 189 1,370 887 180,000-240,000 9 21 32 62 873 579 Over 240,000 10 19 40 69 986 605 Total 1,566 1,196 1,051 3,813 1,469 746 Soinices See Tables 3 10, 3 11, and 3 13. Table 3.16: Rural Colombia: Public Subsidies for Education by Level of Income Public Subsidies for Education: Per Child Per Child in Total Enrolled each Income Income Bracket (Millions (Millions Group (Pesos/Year) of Pesos) of Pesos) (Pesos) 0- 6,000 191 552 83 6,000- 12,000 207 552 127 12,000- 24,000 154 554 183 24,000- 60,000 72 554 276 60,000-120,000 12 333 333 120,000-240,000 1 200 200 Over 240.000 3 100 100 Total 640 533 125 Soitnces See Tables 3 12 and 3.13. 36 Comparing the Distribution of Tax Burdens and Public Subsidies for Education among Income Groups The allocation of public subsidies for education to the various income groups can now be compared with the corresponding allocation of taxes. For each income group in urban Colombia, the ratio of the subsidies re- ceived over the taxes paid was computed, using successively the least progressive(alternative 1) and the most progressive(alternative 2) allocation of taxes shown in Tables 3.4 and 3.5. As the fourth and fifth columns of Table 3.17 show, Colombian families living in urban areas receive about 30 percent of their tax payments back in the form of education subsidies. This average proportion varies widely ac- cording to the level of income. Under alternative 1, the lowest two income groups receive more in educational subsidies than they have paid in taxes, three times as much in the case of the poorest income group. At the top of the income scale, the subsidies received by the two richest income groups amount to far less than 10 percent of their taxes. Between these two ex- tremes, the ratio of subsidies received over taxes paid declines steadily as income rises. Only families earning more than 72,000 pesos receive a ratio of subsidies over taxes below the national average. On the whole, one can thus conclude that the public financing of education in urban Colombia results in a number of income transfers from the upper to the lower income groups. This conclusion holds a fortiori if alternative 2 (maximum progressivity of the tax system) is selected for the allocation of taxes to the various income groups. One may be tempted to explain the positive effect of the public financing of education on income distribution in urban Colombia by the importance of the private education sector. As already seen, upper and middle income families show a strong tendency to enroll their children in the private sec- tor, which receives a far smaller public subsidy per pupil than the public sector. In other words, the private sector contributes to diminish the share of education subsidies received by these families, thus contributing to the positive income redistribution observed. In order to check this hypothesis, it was assumed that all students were enrolled in the public sector or, which is the same for our purposes, that pri- vate education was subsidized to the same extent as public education. A model of the distribution of public subsidies was built, by multiplying both public and private enrollments by the subsidy per pupil in the public sector, and this distribution was compared with the allocation of taxes in the last three columns of Table 3.17. As expected, the ratios of subsidies over taxes increase at all income levels, but much more for middle and upper income groups (an extra 20 to 33 percentage points are recorded for income groups between 30,000 and 240,000 pesos) than for the lower income groups. In 37 Table 3.17: Urban Colombia: Allocation of Taxes and Public Subsidies for Education among Income Groups Subsidies as Propor- Public Subsidies Subsidies as Propor- Allocation of Taxes Public Subsidies for tion of Taxes for Education all tion of Taxes Income Bracket (Millions of Pesos) Education % % Education Public' % % (Pesos/Year) Alt. I Alt. 2 (Millions of Pesos) Alt I Alt 2 (Millions of Pesos) Alt I Alt. 2 0- 6,000 23 14 71 309 507 73 317 521 6,000- 12,000 183 119 217 119 182 235 128 197 12,000- 18,000 438 306 370 84 121 421 96 138 18,000- 24,000 658 454 530 81 117 596 91 131 24,000- 30,000 560 392 413 74 105 471 84 120 30,000- 36,000 548 382 345 63 90 393 72 103 36,000- 48,000 949 688 477 50 69 661 70 96 48,000- 60,000 826 590 410 50 69 607 73 103 60,000- 72,000 675 517 301 45 58 528 78 102 72,000- 84,000 616 455 163 26 36 358 58 79 84,000-120,000 1,500 1,137 196 13 17 589 39 52 120,000-180,000 1,405 1,897 189 13 10 593 42 31 180,000-240,000 1,048 1,628 62 6 4 287 27 18 Over 240,000 2,894 4,225 69 2 2 313 11 7 Total 12,323 12,804 3,813 31 30 6,125 50 48 tThe data in this column have been obtained by multiplying total enrollments (public and private) by the subsidy per pupil in the public sector. Sources Ist Column: see Table 3.4 2nd Column see Table 3 5. 3rd Column see Table 3.15 Table 3.18: Rural Colombia: Allocation of Taxes and Public Subsidies for Education among Income Groups Public Subsidies for Subsidies as Proportion Income Brackct Allocation of Taxes Education of Taxes (Pesos/Year) (Millions of Pesos) (Millions of Pesos) (%) 0- 6,000 200 191 96 6,000- 12,000 327 207 63 12,000- 24,000 372 154 41 24,000- 60,000 225 72 32 60,000-120,000 87 12 14 120,000-240,000 31 1 3 Over 240,000 38 3 8 Total 1,280 640 50 In rural Colombia only the allocation of taxes corresponding to the alternative of mini- mum progressivity, that is, the alternative whicil excludes coffee export duties, was used. The reason for doing so is that only a negligible fraction of these duties flow into general revenue, the bulk of them go towards coffee or coffee-related activities in coffee growing areas. Sources See Tables 3.6 and 3.16 spite of this, the ratio of subsidies over taxes still declines as income rises,22 thus causing a redistribution of income from poorer to richer families. The conclusion is that while the positive redistribution of income that takes place through the public financing of education is accentuated by the presence of the private education sector, it is not entirely attributable to it. Table 3.18 compares the distribution of the tax burden among income groups with the distribution of public subsidies for education in rural Co- lombia. On an average rural families receive 50 percent of their tax pay- ments back in the form of education subsidies-as against 30 percent for ur- ban families. This proportion is distinctly higher for low income groups and lower for high income groups, but the resulting income transfers from the rich to the poor are much less than in the case of urban areas. The poorest income group in rural Colombia does not receive more subsidies than it pays in taxes, it just breaks even. Over 60,000 pesos, education subsidies represent about 10 percent of taxes, that is, the same proportion as in urban areas. The overall impact of the public financing of education on the income distribution of total Colombia is summarized in Table 3.19. On average, Co- lombian families receive one peso's worth of education subsidy for each three pesos paid in taxes. For all families under the 60,000 pesos a year 22There is an exception, however The ratio remains approximately the same-around 75 percent-for all families earning between 30,000 and 72,000 pesos 39 Table 3.19: Total Colombia: Allocation of Taxes and Public Subsidies for Education among Income Groups Allocation of Public Subsidies Subsidies as Number of Taxes for Education Proportion Income Bracket Households (Millions of (Millions of of Taxes (Pesos/Year) (%) Pesos) Pesos) (%) 0- 6,000 19.0 223 262 117 6,000- 12,000 20.2 510 424 83 12,000- 24,000 24.9 1,468 1,054 72 24,000- 60,000 22.9 3,108 1,717 55 60,000-120,000 8.8 2,878 672 23 120,000-240,000 3.4 2,484 252 10 Over 240,000 0 8 2,932 72 2 Total 100.0 13,603 4,453 33 Sources See Tables 3.8, 3.17 and 3.18. Table 3.20: Total Colombia: Public Subsidies for Three Levels of Education as a Proportion of Taxes Distributed among Income Groups (%) Public Subsidies Public Subsidies Public Subsidies for Primary for Secondary for Higher Education as Education as Education as Income Bracket Numberof Proportion of Proportion of Proportion of (Pesos/Year) Households Taxes Taxes Taxes 0- 6,000 19 0 109 9 0 6,000- 12,000 20.2 77 4 2 12,000- 24,000 24.9 49 18 5 24,000- 60,000 22.9 22 20 14 60,000-120,000 8.8 4 7 12 120,000-240,000 3.4 1 3 6 Over 240,000 0.8 - I I Total 100.0 16 9 8 Sources See Tables 3.15, 3 16 and 3.19. level, that ts, for 87 percent of all Colombian families, education subsidies amont to close to two-thirds of their taxes (65.1 percent) For the remaining 13 percent of Colombian families whose earnings are 60,000 pesos or more, education subsidies amount to 12 percent of their taxes. While the educa- tion subsidies received by the poorest income group, close to 20 percent of families, are distinctly higher than the taxes paid by this group, the sub- sidies received by the richest income group-less than 1 percent of all families-amount to only 2 percent of their taxes. 40 All in all, the public financing of education appears to have a strong and positive influence on the distribution of income in Colombia. The remain- ing question is now to determine whether this general conclusion holds true for each of the three levels of education, or whether it is the result of dis- tinct and offsetting effects according to levels. In order to do so, the ratios of public subsidies for, successively, primary, secondary and higher educa- tion over taxes have been computed for each income group (see Table 3.20). The table shows a very different situation for each of the three levels of education. The public financing of primary education contributes to redistribute income from the 13 percent richer families to the 87 percent poorer families, the cut-off point being 60,000 pesos a year. The redistribu- tive effect is strongly beneficial to the 40 percent poorer families (under 12,000 pesos a year) more than three-fourths of whom live in rural areas. These families receive 87 percent of their taxes in the form of public sub- sidies for primary education. The picture that emerges for secondary educa- tion is quite different: here, the main beneficiaries are two middle income groups, including about 48 percent of all families. In other words, the public financing of secondary education redistributes income from both the 40 percent poorest and the 13 percent richest families to a sort of lower mid- dle-class, 80 percent of whom are living in urban areas and whose incomes are in the 12,000-60,000 pesos range. The situation for higher education is very similar, with the difference that the two income groups subsidized by both the poor and the rich are higher up in the income scale. Families in these groups represent close to one-third of all families, they are almost ex- clusively urban and earn between 24,000 and 120,000 pesos. Thus a redistribution of income from the poor and the very rich to the upper mid- dle class takes place through the public financing of higher education. All in all, it is clear that the positive effect of the public financing of education on the distribution of income in Colombia is only due to the fi- nancing of primary education which strongly benefits the poor. This posi- tive effect is partly but not wholly offset by the negative, income distribu- tive effects of the public financing of secondary and higher education which benefit most the lower and upper middle class respectively. 41 IV. Central Government Transfers for Education: The Distributive Impact among Regions The purpose of this chapter is to assess the overall distributive impact of the central government's taxation and educational spending policies among regional units In the present analysis, income groups are no longer con- sidered; they are replaced by the twenty Colombian departments as units of analysis. Each year the residents of each department contribute to finance the central government's budget through national taxes,' part of which come back to the departments in the form of cash transfers for education. Obviously the combined incidence of taxes and transfers may have a redistributive impact among departments, as some may pay little and re- ceive much and vice-versa. Chapter 11 demonstrated that an increasingly large proportion of the central government's funds for education is transferred to and spent by the departments. While the central government's subsidies for education may have a substantial impact on the size and structure of departmental spend- ing on education, the criteria used for allocating these subsidies to the departments are not clear. This chapter attempts, in the first place, to check whether the distribu- tion of central government transfers among departments bears any rela- tionship to the distribution of central government taxes paid by the resi- dents of each department. Second, since the combined "tax-transfer" cri- terion is obviously not the only possible one, other redistributive criteria are proposed in order to allocate the central government's subsidies for education among departments. The Distribution of National Taxes among Departments In 1967, the central government tax revenue originated from three types of taxes: (i) direct taxes, which accounted for 50 percent of total tax revenue, 95 percent of which was represented by the income tax; (di) indirect taxes 'Departmental and municipal taxes are not considered in this analysis, since the revenue raised by a department or municipality is usually spent in the same department and munici- pality and does not, therefore, allow for any geographical distributive impact 42 such as gas consumption tax, sales tax and official paper sales which ac- counted for 23 percent of revenue; and (iii) import duties whose share, although declining, amounted to 27 percent. The problem of assessing the regional incidence of these taxes is an ex- tremely difficult one because it involves not only hypotheses about the shifting of the burden of taxes-as in all studies of tax incidence by income groups2-but also hypotheses about the place of residence of those who ac- tually bear the taxes. To undertake a study of the actual regional incidence of the central government taxation system in Colombia would clearly be a major task far beyond the scope of this paper. This problem of the regional incidence of taxation has been solved here in a rather simplified way, the starting point being the distribution of central government tax revenue by departmental collecting offices for 1967, which is the last year for which complete data are available. The reasons why such a distribution may differ from the actual regional incidence of central government taxation vary between taxes as explained below. (i) Personal income tax raises the problem of how to decide whether revenues should be allocated to the region in which taxable income is cre- ated (i.e., at origin) or to the region in which the recipient resides (i.e., at resi- dence). The first alternative, which is clearly the better one, could not be implemented here because of the lack of statistics on value added originat- ing in various regions. Instead, an allocation based on departmental collect- ing offices was adopted. Such an allocation disregards the fact that, even if it is assumed that personal income tax is entirely borne by the residents of each department, part of the income may originate from assets or hold- ings-real estate holdings for instance-located outside the department. It may also differ slightly from an allocation based on residence to the extent that taxpayers file returns with offices outside their regions of residence. This, however, seems to be a minor problem. (ii) Corporate income tax is even more complex to allocate among regions.3 As was shown in Chapter 111, this tax can be allocated according to various alternative incidence assumptions, which all have different implica- tions for actual incidence among regions. If it is, for instance, decided that the tax is shifted to labor, it should be attributed to regions on the basis of corporate payrolls by regions. If the tax is borne by shareholders, then it should be allocated to their regions of residence on the basis of the regional distribution of corporate ownership. It can also be assumed that much of the corporate tax is borne by nonresidents and, therefore, should not be allocated domestically. Lastly, if the tax is shifted to consumers, it could be 2See Chapter III 3The corporate income tax accounted for about half of the total income tax revenue in i967 43 allocated among regions in proportion to nonfood expenditures of resi- dents. Given the lack of data and the arbitrariness of any choice favoring one alternative at the expense of others, it was decided to stick to an alloca- tion by regions of collection of taxes. It must be emphasized that this is likely to be a second-best approach, which overestimates the contributions of departments or cities where headquarters are located, whereas the con- sumers of the products manufactured by the large corporations may be scat- tered throughout the country. However, such an allocation might be an ac- ceptable proxy for the actual regional incidence, to the extent that the tax is borne by shareholders and that most shareholders are to be found among high-income families who are probably also concentrated in big cities, hopefully in rough proportion to corporate tax collections. (in) Indirect taxes are mainly borne by consumers and should be allo- cated to regions in proportion to taxed consumption. Tax collection figures will truly reflect regional incidence patterns to the extent that taxes are col- lected where consumption occurs. Because some indirect taxes are in fact collected centrally, it has been decided to allocate all indirect taxes in pro- portion to sales tax collections. (iv) Customs duties, as well as all other indirect taxes linked with external trade,4 are mostly collected in ports, border towns or big cities where customs offices are located. Such a geographical distribution is of course perfectly artificial and has little to do with the actual incidence of these taxes. If it is assumed that customs duties are borne by consumers-which is a fairly general assumption-a geographical distribution of the con- sumers of imported goods is required in order to assess their regional inci- dence. Since such a distribution is presently not available in Colombia, customs duties have been allocated among departments according to the distribution found for the sales tax returns, on the grounds that the geographical consumption pattern of imported goods follows closely the corresponding pattern for homemade goods The results of these calculations are shown in Table 4.1. It should be stressed that this allocation of national taxes among departments repre- sents only a very rough approximation of the actual regional incidence of national taxes and the data are to be used with great care. As can be seen from this table, the distribution of central government taxes among departments is heavily concentrated on a few, while the others make relatively minor contributions to central government tax revenue. The department of Cundinamarca which includes the capital city of Bogoti is, by far, the largest contributor to the central government tax revenue: 42 7 percent of all taxes are raised in this department. The departments of Antioquia and Valle which include the second and third largest cities in the 4For example, coffee export duties 44 country, namely Medellin and Cali, come in the next position with 19.4 per- cent and 16.4 percent of all taxes respectively. At the other extreme, the proportion of taxes collected in twelve departments out of twenty does not exceed 1 percent of the total. The distribution of indirect taxes is slightly more concentrated on a few urbanized departments than that of direct taxes, and the weight of the capital city is even more overwhelming in this case. About 83 percent of all indirect taxes are raised in the departments of Antioquia, Valle and Cun- dinamarca, as against 72 percent of direct taxes; the share of the department of Cundinamarca alone is close to 45 percent of indirect taxes, as against 40 percent of direct taxes. Table 4.1: Distribution of Central Government Taxes among Departments (1967) Indirect Taxes Dtrect Taxesl Sales Tax and Customs Total Duties Millions Miliions Millions Millions of Pesos % of Pesos % of Pesos % of Pesos %S0 Aniioquia 351.4 11.1 167 9 25.6 919.4 25.6 1,438.7 19.4 Atlantico 281.9 8.9 41.3 6 3 226.3 6 3 549 5 7 4 Bolivar 47.3 1.5 26.2 4.0 143.7 4.0 217 2 2.9 Boyaca 16.6 0.5 0.7 0.1 3.6 0 1 20.9 0.3 Caldas 92.7 2 9 8.5 1.3 46.7 1 3 147.9 2 0 Cauca 18.2 0.6 0.7 0.1 3.6 0.1 22.5 0.3 C6rdoba 27.3 09 0.7 0.1 3.6 0.1 31.6 0.4 Cundinamarca 1,274.7 40.1 292.5 44.6 1,601 9 44 6 3,169.1 42.7 Choc6 3.7 0.1 - - - - 3.7 - Guajira 0.2 - - - - - 0 2 - Huila 26 7 0.8 0 7 0 i 3.6 0.1 31.0 0.4 Magdalena 38.1 1 2 07 0.1 3 6 0.1 42.4 0.6 Meta 13.9 0.4 - - - - 13.9 0.2 Narihlo 11.6 0 4 1.3 0.2 7.2 0.2 20.1 0.3 Norte de Santander 42.1 1 3 0.6 0.1 3.6 0.1 46.3 06 Quindio 22.7 0.7 0.6 0.1 3.6 0 1 26.9 04 Risaralda 111.5 3 5 11.2 1.7 61 0 1.7 183.7 2.5 Santander 81.6 2.6 15.7 2.4 86.2 2.4 183.5 2.5 Tolima 45.5 1.4 2.0 0.3 10.7 0.3 58 2 0.8 Valle 669.9 21.1 84 6 12.9 463.3 12.9 1,217.8 16 4 Total 3,177.6 100.0 655.9 1000 3,591 6 100.0 7,425.1 100.0 'Excluding income tax collected by consulates, lesoteria qeneral, and other miscellaneous sources. Soarce Aniutario(e Fiscalesy Ftiaticieias, 1966-67, pp. 14to 39 and text. The Department of Sucre which has very incomplete fiscal data has been left out of the analysis. 45 If these rather crude tax data can be considered as indicators of eco- nomic activity and wealth throughout Colombia, they confirm what is al- ready known about this country, namely the concentration of economic life in three, possibly four areas when including the department of Atlantico. Four other departments, namely Bolivar, Caldas, Risaralda and Santander, play a reduced, although not negligible, role in the financing of the national budget, while all the remaining departments are of marginal importance. The Distribution of Central Government Transfers among Departments In addition to the revenue raised through departmental taxation, every Colombian department receives, each year, a certain amount of central government money earmarked for specific purposes. The purpose of these transfers, which have been increasing rapidly during the sixties, is to help the departments finance their rising expenditures, particularly in the field of education Another way of doing this would have been to reform drastically the fiscal systems of the departments in order to enable them to meet their responsibilities. It is not the purpose of this paper to argue about the pros and cons of each alternative. Although the inadequacy of the departmental and municipal fiscal systems is a well-known issue in Colom- bian local finance, all recent proposals to remedy this situation have so far remained on paper I In any case, the transfers system, which provides the departments with loosely controlled central government funds, is backed by strong political pressures and may well become a permanent feature of intergovernmental finance in Colombia. Total transfers from the central government to each department are shown in Table 4.2. In absolute value, the three more urbanized depart- ments of Antioquia, Cundinamarca and Valle received 110 million pesos in 1967 out of a total 303 million pesos, i e, 36.3 percent. However, these transfers accounted for only 11.2 percent of these departments' total cur- rent revenue, while the average proportion was 14.3 percent for the country 5The most far-reaching and publicized of these proposals consists of increasing the rates of the property tax or its base or both through a general property reevaluation program. The ad- ditional revenue thus raised would be earmarked for education to make the reform more ac- ceptable politically This proposal was, it seems, formulated for the first time in Richard Bird, Taxation and Development, Lessons from Colombian Experience (Cambridge, Mass Harvard University Press, 1970) The proposal was elaborated in Richard A Musgrave and Malcolm Giliis, eds, Fiscal Reform for Colombia The Final Report and Staff Papers of the Colombian Commission on Tax Reform (Cambridge, Mass Harvard Law School International Tax Program, 1971) A detailed quantitative evaluation of both the costs of educational expansion and the local fiscal reforms required to meet these costs are presented in Miguel Urrutia Mon- toya and Clara Elsa de Sandoval, "Reforma de la educacion pubbica y sus implicaciones fiscales," mimeographed (Bogota Universidad Nacional de Colombia, 1970) 46 Table 4.2: Transfers from the Central Government to the Departments (1967) Total Transfers Transfers as a Transfers for Education from the Total Departmental Proportion of Transfers for as a Proportion of Central Governmentl Current Revenue Revenue Education Revenue (Millions of Pesos) (Millions of Pesos) (%) (Millions of Pesos) (%) 1 2 3=2/1 4 5=4/2 Antioquia 1 35.2 415.7 8.5 32.1 7.7 At1antico 2 9.9 70 1 14 1 8.2 11.7 Bolivar 3 10.9 119.2 9.1 10.9 9.1 Boyaca 4 19.6 99.6 19 7 18.6 18.7 Caldas 5 11.9 69 5 17.1 11.9 17.1 Cauca 6 15 8 61.0 25.9 12.6 20 7 C6rdoba 7 5.7 60.9 9.4 5.6 9.2 Cundinamarca 8 48.4 308.4 15 7 48 3 15.7 Choc6 9 7.8 15.6 50 0 5.4 34.6 Guajira 10 8.5 14 6 58.2 2.5 17.1 Huila 11 9.3 47.3 19.7 6.6 14 0 Magdalena 12 10.4 49.0 21 2 9.5 19.4 Meta 13 5.6 37.0 15.1 5.6 15.1 Narnio 14 11.4 61.2 18.6 11.4 18.6 Norte de Santander 15 13.0 83.0 15.7 12.4 14.9 Qumdio 16 5.8 33.1 17.5 50 15.1 Risaralda 17 5 4 43.9 12 3 5.4 12.3 Santander 18 24.9 159.3 15.6 17.1 10 7 Tolima 19 17.1 104.9 16.3 17.1 16.3 Valle 20 26 3 260.8 10.1 26.3 10.1 Total 302.9 2,114.1 14.3 272 5 12.9 'Transfers represented by oil royalties-about 100 million pesos in 1967-and other minor transfers, have been excluded from the analysis because they are too unevenly distributed. For instance, the two departments of Santander and Norte dc Santander received 78 percent of the total of oil royalties transferred by the central government in 1967 Source: Anuario de Fiscales y Fmancieras 1966-67, pp. 90-I 11. as a whole At the other extreme, poor departments such as Choc6 and Guajira received only 8 million pesos each in transfers which, nevertheless, accounted for 50 percent and 58 2 percent of their current revenue respectively. The bulk of these transfers-90 percent in 1967-is earmarked for education. This situation is the result of a 1960 law according to which all the primary school teachers in the country should be paid progressively from central government funds, while their appointments would remain in the hands of the departments' executive bodies. The transfers for education as well as their contributions to each department's current revenue are shown in Table 4.2. While the national average is 12.9 percent, it is obvious that some departments are helped more than others. It is noteworthy that the department of Cundinamarca, which includes the capital city of Bogota, ranks above this average with 15.7 percent of its revenue accounted for by transfers. On the contrary, the transfers for education to the department of Valle which includes Cali, and Antioquia which includes Medellin, repre- sent only 10.1 percent and 7.7 percent of their revenue, the latter proportion being the lowest in the country. In one department only, Choc6, is the con- tribution of central government money to current revenue significantly higher than 20 percent. Comparing the Distribution of Taxes and the Distribution of Transfers The distribution of central government taxes among departments and the distribution of transfers for education from the central government to the departments are shown in Table 4.3. It is clear that (i) the former is much more concentrated on a few departments than the latter, and that (ii) a certain amount of redistribution is going on, since the four departments already mentioned as the richest, namely Antioquia, AtlTntico, Cun- dinamarca and Valle, receive only 42.2 percent of all transfers, although 78.5 percent of all taxes have been levied within their boundaries. There is only one more department-Risaralda-whose share in transfers is lower than its contribution in taxes. For all other departments, it can be said that educational transfers financed by central government taxes have a positive redistributive impact, since their shares in transfers are higher than their contribution in taxes. It is also interesting to compare, in absolute terms, the amount of taxes flowing out of the departments with the amount of transfers flowing back, in order to arrive at a measure of the combined incidence of central govern- ment taxation and transfers. In order to do so, the total transfers and the transfers for education have been expressed as a proportion of the taxes col- lected within each department. These ratios are shown in Table 4.3, col- umns 3 and 4. 48 Table 4.3: Structure of Transfers and Total Taxes Collected by Departments (1967) ( Total Transfers Transfers for Transfers as a Proportion Education as a Total Taxes for of Total Taxes Proportion of Total Paid Total Taxes Education Collected Taxes Collected per Capita 1 2 3 4 5 Antioquia 1 19.4 11.8 2.4 2.2 514 Atlantico 2 7.4 3.0 1.8 1.5 684 Bolivar 3 2.9 4.0 5.0 5.0 260 Boyaca 4 0.3 6.8 93.8 89.0 18 Caldas 5 2.0 4.4 8.0 8.0 172 Cauca 6 0.3 4.6 70.2 56.0 34 C6rdoba 7 0.4 2.0 18.0 17.7 47 Cundinamarca 8 42.7 17.7 1.5 1.5 980 Choc6 9 - 2.0 210.8 145.9 18 Guajira 10 - 0.9 4,250.0 1,250.0 1 Huila 11 0.4 2.4 30.0 21.3 68 Magdalena 12 0.6 3.5 24.5 22.4 48 Meta 13 0.2 2.0 40.3 40.3 69 Narifio 14 0.3 4.2 56.7 56.7 26 Norte de Santander 15 0.6 4.6 28.1 26.8 78 Quindio 16 0.4 1.8 21.6 18.6 86 Risaralda 17 2.5 2.0 2.9 2.9 423 Santander 18 2.5 6.3 13.6 9.3 166 Tolima 19 0.8 6.3 29.4 29.4 65 Valle 20 16.4 9.7 2.2 2.2 628 Total 100.0 1000 4.1 3.7 391 (Millions of Pesos) 7,425.1 272.5 Sources. Tables 4.1 and 4.2. The population data have been drawn from Enrique Perez Sanin, Parametros demnograficos coloinbilnos, 1951-64, proyecciones de poblacitn 1965-85, (Bogota: Ediciones de la Universidad de los Andes, 1970). At the national level, out of every 100 pesos collected from the depart- ments, the central government returns 4.1-3.7 pesos for education. The situation varies enormously from one department to another. Considering the transfers for education only, two departments, namely Choc6 and Gua- jira, receive more in the form of transfers than they have paid in taxes. Three others, namely Boyaca, Cauca and Narinio, receive between 50 and 90 pesos for education whenever 100 pesos have been collected by the central government. At the other extreme, the ratio of transfers over taxes is, pre- dictably, below the average for the four richest departments, already men- tioned, plus Risaralda. It is the least favorable in the department of Cun- dinamarca (Bogota) which receives only 1.5 pesos for each 100 pesos paid in taxes. These ratios of transfers received over taxes paid have also been com- pared with the total taxes paid to the central government on a per capita basis, which is considered here as an indicator of the general wealth of the departments. As Figures 4.1 and 4.2 show, there is a clear and negative cor- relation between the distributional impact of both taxation and transfers on the one hand and the economic situation of the department on the other. the poorer the department, the more redistributive the central government policies. This conclusion holds whether total transfers or transfers for education are considered. The relationship between the two variables seems to be roughly log-lineal.6 To sum up, the combined impact of the central government's taxation and transfer policies is clearly redistributive among departments, and redistributive in the right direction, that is from the richer to the poorer departments. More surprising perhaps is the rather rigid relationship ob- served between this redistributive impact and the economic situation of the departments. No attempt will be made here to propose interpretations of the steep slopes of the curves shown in Figures 4.1 and 4.2. On the contrary, one should recall the warnings at the beginning of this chapter concerning a possible overestimation of the taxes actually borne by the rich departments and a corresponding underestimation in other departments. Different hy- potheses about regional tax incidence might have led to higher transfers/tax ratios for the former, and lower transfers/tax ratios for the latter, thus resulting in different curve slopes. The next question could be formulated as follows: is the redistributive criteria used so far-namely the combined incidence of taxation and transfers-the "best" possible one or, in a more practical way, is it compati- ble with other redistributive criteria that might be used in the allocation of central government transfers for education? After all, the prime purpose of 60n Figures 4 1 and 4 2, the departments are given the same identiflcation numbers as those in Tables 4.2 and 4 3 50 educational transfers is to foster education where and when it is most needed. This is why it is important to check if the existing pattern of transfers resulting from the correct use-whether intentional or not-of the "taxation-transfers" criterion, as described in this section, is also redistributive in terms of educational needs of the departments. Allocating the Transfers for Education among Departments The purpose of this section is to determine how the existing pattern of education transfers fares when assessed with redistributive criteria different from the "tax-transfer" criterion used in the preceding section. Three criteria, forming an adequate picture of the educational policies often advocated by the Colombian government for primary education,7 have been drawn up. They are: . the existing disparities in the provision of education facilities; . the "effort" already made for education by the departments; and . the potential demand for education in each department. They will be reviewed in turn below. Disparities in the Provision of Educational Facilities The departments and the municipalities are responsible for providing most primary educational facilities. In 1967, they devoted 382 and 160 million pesos respectively to primary education, close to 60 percent of total expenditures at this level and one-fourth of all education expenditures made by the four government levels. A structure by department of total departmental expenditures on pri- mary education is shown in Table 4.4, together with the structure of transfers. It can be said that, although both distributions are largely similar, the latter is slightly less concentrated than the former. Thus, the three richest departments of Antioquia (1),8 Cundinamarca (8) and Valle (20) are responsible for 47.7 percent of all departmental expenditures on primary education, while they receive only 39.2 percent of the central government subsidies. Conversely, the six departments which contribute individually less than 2 percent of expenditures on primary education, and together 6.4 percent of these expenditures, receive together 15 percent of all central government subsidies. At first glance, the transfers are thus allocated to the departments in such a way that the provision of education facilities fi- nanced by the departments' own resources is slightly redistributed from the poorer to the richer departments. 7The emphasis on primary education is due to the fact that the prime purpose of the transfers policy inaugurated by the Colombian government at the beginning of the 1960s was to help the departments finance their expenditures at this level 8Parenthetical figures are identification numbers referring to reievant tables and figures in this chapter; see aiso footnote 6 above 51 Figure 4.1 1000 I I l I I I I I l 8 Total Transfers as a Proportion of Taxes Collected (%) _ °86-=4 =-= =_ -_ __ _ 391 6 _=_=_ =.9== _ 82_--_- lo 6 '14~- 4---.!l3--- -_ 120 2 2 .6 .7 10~~~~~1 10 __5------ 84 @20 2~~~~~~~~~~~. .8 Taxes Paid per Capita (Pesos) 0 200 400 600 800 1000 52 Figure 4.2 1000 I I I I l 8 Transfers for Education as a Proportion of Taxes Collected (%) 6 - - __ - = _ 4 = = -= - -= _ _ _ 391 .9 100 --- --_ 66 015 2~~~~12 * 016 7 10 - iiiiii - - 9 17 lei -2- 2~~~~~~~~. 2 .2 *8 Taxes Paid per Capita (Pesos) 0 200 400 600 800 1000 53 Table 4.4: Transfers for Education and Departmental Expenditures on Primary Education (1967) Transfers as a Departmental Expenditures Departmental Expenditures Proportion of Departmental on Primary Education, Transfers for on Primary Education, Expenditures on Primary excl. Transfers, per Education Excluding Transfers Education, excl. Transfers Capita (%) (%) (%) (Pesos) Antioquia 1 11.8 17.4 47.1 24 Atlantico 2 3.0 3.6 58.2 18 Bolivar 3 4.0 4.3 65.3 20 Boyaca 4 6.8 5.4 88 6 18 Caldas 5 4.4 4.1 74.4 1'9 Cauca 6 4.6 1.7 190.9 10 C6rdoba 7 2.0 1.7 84.8 10 Cundinamarca 8 17.7 19.4 63.7 23 Choc6 9 2.0 1.2 112.5 24 Guajira 10 0.9 0.5 138.9 11 Huila 11 2.4 3.0 55.9 26 Magdalena 12 3.5 0 5 527.8 2 Meta 13 2.0 0.8 169.7 16 Narifio 14 4.2 3.2 89.8 17 Norte de Santander 15 4.6 5.1 62.6 34 Quindio 16 1.8 2.1 61.7 26 Risaralda 17 2.0 2.6 54.0 23 Santander 18 6.3 7.8 55.7 28 Tolima 19 6.3 4.7 93.4 20 Valle 20 9.7 10.9 61.7 22 Total 100.0 100.0 69.8 21 (Millions of Pesos) 272.5 390.6 Sources. Table 4.3 and AnuarLo de Fiscales y Financieras, pp. 90-111. How large are central government subsidies with respect to departmental expenditures on primary education? The answer to this question will be given by looking at the transfers as a proportion of departmental expen- ditures on primary education, both in absolute values. As can be seen from Table 4.4, for each 100 pesos spent by the departments, the central govern- ment adds, on average, 70 additional pesos. But this help varies enormously from one department to another: the most helped department is Magdalena (12) which receives 528 pesos for each 100 spent on its own resources. Five departments receive more funds in the form of transfers for primary educa- tion than they spend at this level on their own resources. If the relative disparities in the provision of educational facilities are to be eradicated, central government help to the departments should be in- versely related to the amount of money spent per capita on primary educa- tion. Figure 4.3 shows that, although such a relationship exists, it is tenuous and exceptions abound. While most of the departments which are not able to spend the national average of 21 pesos per capita on primary education receive more help than the others, the departments spending more than this average receive approximately the same help-between 50 and 70 pesos for every 100 spent-whatever their level of per capita spending. The foregoing analysis has also been made taking both departmental and municipal expenditures into account as a proxy for the distribution of educational facilities throughout the country. It should be noted, that while the funds spent by all the municipalities in a department are very small in most cases, this is not so for big municipalities. In 1967 the three biggest cities of Bogota, Medellin and Cali spent 112.4 million pesos on education, that is, about 70 percent of total expenditures on education by the 1,600 mu- nicipalities of the country, the share of Bogota alone amounting to 55 per- cent. As a consequence, while the contribution of the department of Cun- dinamarca is 19.4 percent of total expenditures on primary education, it amounts to 30.6 percent when expenditures made by the municipalities are included (see Table 4.5). On an average, central government transfers amount to almost 50 pesos for each 100 pesos spent by the departments and the municipalities. Only four departments, namely Antioquia, Cundinamarca, Risaralda and Valle, receive significantly less than this average, while five, namely Cauca, Choc6, Guajira, Magdalena and Meta, receive more money for primary education than they and their municipalities spend at this level from their own resources. Figure 4.4 shows the same conclusion as before when the ratios of transfers over departmental and municipal expenditures are com- pared with per capita spending: it can be observed that the departments and municipalities who spend little on education receive more help from the central government, although exceptions are numerous, while some who can afford to spend more receive also substantial subsidies. 55 Figure 4.3 I l l I l l I Transfers as a Proportion of Departmental = _ -Expenditures on Primary Education (%) _ 1000 _ 9- 87 6 -12 5 *1 2 - --21 === 4 - = = ==- 3 2 _ _ _ _ _ 113 0 20 100 - -14 --s9- - 9 89 07 4--_ 8 -~~~~~~~~~~~~- - 6~~~~~~~~~~~~9.8- 6 2 3 @8 1o6a 5 ~ ~ ~ ~ ~ ~ U 20.017 * *18 __ 15 2 Departmental Expenditures on Primary Education per Capita (Pesos) 10 20 30 56 Figure 4.4 I ,I I I I I Transfers as a Proportion of Departmental and Municipal Expenditures on Primary Education ( g ) 1000 9 8 2- -- …-- - - - 6~~~. 60 __ ___ _ 2______ 8 = _ *7 _-14 4a19_ -;1 - -_-95 43= =07b9 X--- - - - 5~~~~~~~~5 6 - - 2 i.1,29- - 3 9 - _ _ _ _ _ _ _ _ _ _ _ _ _ 05- ~ 5 Table 4.5: Transfers for Education and Departmental and Municipal Expenditures on Primary Education (1967) Departmental and Transfers as a Departmental and Municipal Proportion of Departmental Municipal Expenditures on Transfers for Expenditures on and Municipal Expenditures Primary Education per Education Primary Education on Primary Education Capita (%) (%) (%) (Pesos) Antioquia 1 11.8 17.1 34.1 34 Atlantico 2 3.0 2 8 52.9 19 BolivAr 3 4.0 3.3 60.9 21 Boyaca 4 6.8 4.1 81 6 20 Caldas 5 4 4 3.3 66.1 21 Cauca 6 4.6 1.4 159 5 12 C6rdoba 7 20 1 3 75.7 11 Cundinamarca 8 17 7 30.6 28.7 52 Choc6 9 2.0 09 108.0 25 Guajira 10 09 04 1250 13 Huila 11 2.4 2 3 512 28 Magdalena 12 3.5 0.7 263.9 4 Meta 13 20 0.7 140.0 20 Narifio 14 4 2 2 6 80.3 19 Norte de Santander 15 4.6 4.1 55 1 38 Quindio 16 1.8 1.7 54.3 30 Risaralda 17 2.0 2.2 43 9 28 Santander 18 6.3 6.3 49.1 32 Tolima 19 6.3 3.8 82.2 23 Valle 20 9.7 10.4 45.9 30 Total 100 0 100.0 49.5 29 (Millions of Pesos) 272.5 550.6 Sources See Table 4.3 and Aniario de Fiscales y Financieras. Disparities in "Efforts" Made for Primary Education by the Departments The twenty Colombian departments do not-in the absence of any transfers-spend the same proportion of their budget on primary educa- tion, and these differences in spending are obviously responsible for all kinds of disparities in educational development. To what extent are central government transfers used to wipe out these disparities? Poor departments, like poor countries, are expected to spend lit- tle on education and should therefore receive more help and vice-versa. One could of course turn the argument the other way around and say that only those who already help themselves should receive central government subsidies. However, this may prove to be an untenable position in the case of many departments whose resources are too limited. In any case, it could be possible to design a transfers policy according to the "effort" already made by the departments from their own resources, measured by the pro- portion of their total expenditures spent on primary education, excluding all transfers. As mentioned earlier, the purpose of the transfers policy initiated by the central government in the early sixties was to help the departments to foot the bill of primary school teachers' pay. It seems logical therefore to relate the amount of funds transferred to each department to some index of the amount of teachers' pay in the same department. The Colombian teaching body is traditionally divided into five catego- ries, reflecting both qualifications and seniority. Each department has im- portant responsibilities in the recruitment, promotion and salary policies of the departmental school teachers, while the central government practices a policy of its own only in those schools run by the Ministry of Education. As a result, there are wide disparities among departments in teachers' qualifica- tions. The richest and more urbanized departments do attract the better qualified teachers while the impoverished, largely rural departments tend to be left with an underqualified teaching body. In 1971, two primary school teachers out of three belonged to the first and second category, which is the best qualified. However, this proportion reached 71 percent in the depart- ment of Cundinamarca, 91 percent and 81 percent respectively in the departments of Antioquia and Atlantico (which include the cities of Medellin and Barranquilla) and 94 percent in Bogota alone, as against less than 40 percent in six departments. In the same way, the proportion of pri- mary school teachers in the fifth category, that is, whose level of education is little more than primary education, is 18 percent for the country as a whole but above 40 percent in six departments at least. These disparities in teachers' qualifications, added to the above-men- tioned disparities in wage policies recorded among departments, are 59 reflected in the index of monthly wage rates per teacher shown in Table 4.6, column 3. While the average monthly salary per teacher for the country as a whole was 1,200 pesos (index average = 100) in 1969, it was about 10 percent or more higher in five departments-among them the four departments in- cluding the four biggest cities-and 20 percent lower in three departments. These wage rates have been used to weight the number of primary school teachers in the public education sector of each department so that the amount of transfers per teacher can be computed on a comparable basis in all the departments (see Table 4.6). As can be seen, the departments receive, on average, 5,160 pesos per teacher on payroll from the central government. While some departments receive much more-81 percent, 53 percent and 41 percent more in Meta, Cauca and Bolivar respectively-others, such as Risaralda and Cun- dinamarca, receive less than 85 percent of this amount. This policy, which consists in determining the amount of central govern- ment help according to the size of the teachers' bill, has sometimes been criticized on the grounds that it may drive the departments toward a systematic sub-utilization of their teaching body (low pupil-teacher ratios), their interest being to have as many teachers on the payroll as possible in order to receive central government funds. It is then necessary to relate the transfers to the size of the student body that they are supposed to serve, namely the enrollments in public primary schools. Both types of transfers, per pupil and per teacher, have been successively related to the proportion of total departmental expenditures on primary education in each department (see Table 4.7). On an average, the depart- ments spend 20 percent of their own budget resources9 on primary educa- tion, while they receive 5,160 pesos of central government help per teacher on payroll, or 127 pesos per pupil enrolled. The disparities in the "efforts" made by the departments for primary education are wide: while two depart- ments spend less than 10 percent, there are also two departments beyond the 30 percent mark. Figures 4.5 and 4.6 show that the transfers per teacher or per child re- ceived by the departments bear no relationship to the proportion of their budget that is spent on primary education. In other words, the efforts al- ready made by the departments for education are not taken into considera- tion when allocating central government help to the departments. As a con- sequence, existing disparities in levels of educational development among departments will be maintained in the future. Coping with Potential Demand for Education in the Departments Another redistributive criterion that might be used to allocate central government transfers among the departments is the potential demand for 9Transfers are excluded 60 Table 4.6: Transfers per Primary School Teacher (Public, 1967) Number of Number of Primary Teachers Transfers for School Teachers Monthly Wage Weighed by Transfers per Education (Public) Rate Index Wage Rate Index Teacher (Millions of Pesos) (Thousands) (Average: 100) (Thousands) (Pesos) 1 2 3 4=3x2 5=1/4 Antioquia 32.1 6.5 109 7.1 4,520 Atlantico 8.2 1.5 109 1.6 5,130 Bolivar 10.9 1.9 77 1.5 7,270 Boyaca 18.6 3.2 112 3.6 5,170 Caldas 11.9 2.4 97 2.3 5,170 Cauca 12.6 2.0 80 1.6 7,880 C6rdoba 5.6 1.3 90 1.2 4,670 Cundinamarca 48.3 9.8 115 11.3 4,270 Choc6 5.4 1.0 96 1.0 5,400 Guajira 2.5 0.6 91 0.5 5,000 Huila 6.6 1.6 81 1.3 5,080 Magdalena 9.5 1.7 95 1.6 5,940 Meta 5.6 0.7 91 0.6 9,330 Narfio 11.4 2.1 86 1.8 6,330 Norte de Santander 12.4 2.2 96 2.1 5,910 Quindio 5.0 0.9 93 0.8 6,250 Risaralda 5.4 1.5 93 1.4 3,860 Santander 17.1 3.8 84 3.2 5,340 Tolima 17.1 3.3 83 2.7 6,330 Valle 26.3 4.8 114 5.5 4,780 Total 272.5 52.8 100 52.8 5,160 (Pesos) 1,200 Sources: Column 1: Table 4.2. Column 2: "La educaci6n en Colombia 1960-1968". Column 3: The pattern of wage rates has been constructed by dividing, in each department, the total teachers' pay by the actual number of teachers for the year 1969. The resulting monthly wage rates have been put on an index scale with 100 as the wage rate for the country as a whole. The statistics for teachers' salaries by department have been drawn from "La educaci6n en Colombia", Table 54. Figure 4.5 (See Table 4 7 for Identification Numbers) I I Transfers per Primary School Teacher, Public Sector Only (Pesos) 10000 @3 20 1 9000 8000 @6 .3 7000 19 1 4 6000~~~~~1 6000 12 015 1 80 - ~~~14-g -go 5 160 5000 2e~l I 7 .020 4000 017 Expenditures on Primary Education as a Proportion of Total Expenditures (% 1 0 20 30 40 50 60 62 Figure 4.6 (See Table 4 7 for Identification Numbers) I I I Transfers per Child Enrolled in Public Primary Education (Pesos) 20 1 1 3 220 _I I 200 180 .12 .6 .15 .91 160 .14 190 140 .4 ~~~~3~~~~~ - _______ -~~~127 120 -.20- 18 '10 16 oil 100I 07 017 80 Expenditures on Primary Education as a Proportion of Total Expenditures (%) 10 20 30 40 50 60 63 Table 4.7: Transfers per Child Enrolled and per Teacher as Compared to Departmental "Effort" on Education (1967) Departmental Transfers per Expenditures on Child Primary Education Transfers per Enrolled as a Proportion Primary School in Primary of Total Departmental Teacher, Education, Expenditures, Public Sector Public Sector excluding Transfers only only (%) (Pesos) (Pesos) 1 2 3 Antioquia 1 18 4 4,520 96 Atlantico 2 17.7 5,130 116 Bolivar 3 15.6 7,270 131 Boyaca 4 21.4 5,170 137 Caldas 5 24.1 5,170 116 Cauca 6 14 4 7,880 169 C6rdoba 7 15.0 4,670 95 Cundinamarca 8 26.2 4,270 133 Choc6 9 58.5 5,400 167 Guajira 10 24.7 5,000 126 Huila 11 29.1 5,080 112 Magdalena 12 6.0 5,940 169 Meta 13 7.6 9,330 222 Nariflo 14 21.3 6,330 151 Norte de Santander 15 28 7 5,910 167 Quindio 16 23.3 6,250 115 Risaralda 17 30.9 3,860 87 Santander 18 20.0 5,340 123 Tolima 19 18.9 6,330 146 Valle 20 16.1 4,780 119 Total 20.1 5,160 127 Sources: Column 1: Anuario de Fiscales y Financieras. Column 2 Table 4.6. Column 3: Table 4.2 and "La Educaci6n en Colombia". primary education in each department, that is to say the proportion of school-age children not enrolled in primary education. Enrollment ratios defined as the number of children enrolled in primary schools, both public and private, as a proportion of the total number of five- to fourteen-year-old children, have been computed for each department (see Table 4.8). According to our hypothesis, the transfers per teacher on payroll or per child enrolled should be inversely correlated with enrollment ratios. Figures 4.7 and 4.8 show that this is not the case. There are, for in- stance, five departments in the quadrant 1 of Figure 4.7, that is to say, 64 Table 4.8: Transfers per Teacher and per Child Compared with Enrollment Ratios, 1967 Transfers per Transfers per Child Teacher Enrollment Ratios, Enrolled (Pesos) (%) (Pesos) 1 2 3 Antioquta 1 4,520 47.6 96 Atlantico 2 5,130 42.8 116 Bolivdr2 3 4,830 37.6 108 Boyaca 4 5,170 43.2 137 Caldas3 5 4,960 47.4 107 Cauca 6 7,880 40.7 169 Cundinamarca 8 4,270 55.7 133 Chocb 9 5,400 53.0 167 Huila 11 5,080 47.7 112 Magdalena4 12 3,960 39.3 123 Nari?)o 14 6,330 39.0 151 Norte de Santander 15 5,910 47.6 167 Santander 18 5,340 49.1 123 Tolima 19 6,330 48.1 146 Valle 20 4,780 48.9 119 Total 5,0005 46.7 1245 'Defined as the number of children enrolled in primary schools, both public and private, as a proportion of the total number of five to fourteen year-old children 2Including Sucre and Cordoba. 3Including Quindlo and Risaralda. 41ncluding Cesar. sThe difference with the results shown on Table 3.12 is due to the fact that the depart- ments of Guajira and Meta have been left out of the analysis, due to the lack of demo- graphic data. The department of Sucre is included. departments which display enrollment ratios above the national average of 46.7 percent and systematically receive above the average subsidies. An op- posite situation is that of the departments of Bolivar (3) and Magdalena (12). A very similar picture emerges from Figure 4.8. The department of Cun- dinamarca (8), which includes the capital city of Bogota, has the highest enrollment ratio (55.7 percent) and receives above-average transfers per child but below-average transfers per teacher, which means that primary education in this department enjoys a below-average pupil/teacher ratio, thanks to central government subsidies. To sum up, the present allocation of transfers for primary education in Colombia is not designed to cope with potential demand. It bears no rela- tionship to the geographical structure of this demand. 65 Figure 4.7 (See Table 4 8 for Identification Numbers) Transfers per Primary School Teacher, Public Sector Only (Pesos) 146.7 8000 ____e6I @6 ~ ~ 6 IV 1 7000 _ _ 6000- 5000 -II.I .18 e-5000- .3 15, .20o *I 4000 012 III I Entrollment Ratios (%) 40 50 60 66 Figure 4.8 (See Table 4.8 for Identification Numbers) I I I I Transfers per Child Enrolled in Public Primary Education (Pesos) 200 46.7 . 6 .15 0 160 1 4 .19 140 .4 - - - - - - ~~~~~~~~~~~124 120 - 12__ 92 20 .1l 03 05 Enrollmient Ratios (% 40 50 60 67 Conclusions Two basic findings have emerged from this chapter: First, the combined impact of both central government taxation and transfers policies has a clear and positive redistributive impact on the bud- getary resources of the 20 Colombian departments. By "positive" it is meant that the net effect of these policies is to transfer public funds from the poorest departments to the richest. Whether this situation is intentional or the result of mere chance is still open to question. However, the most probable answer is chance, since the use of the "tax-transfers criterion" described in this chapter is not presently contemplated by the Colombian government, neither explicitly nor implicitly. Secondly, we have seen that the transfers for education to the depart- ments are not presently allocated according to any of the three criteria more specifically designed to assess the redistribution of education opportunities throughout the country. Neither is the actual pattern of transfers clearly re- lated to the often stated government objective of helping the departments to pay the teachers. A note of caution is, therefore, necessary. A transfers policy may be judged as redistributive from the point of view of its net effect on depart- mental resources when combined with the fiscal policy and may, at the same time, have adverse or neutral effects on the distribution of education throughout the country. It would probably be very difficult to design a transfers policy which would have a positive distributional impact on both budget resources and educational opportunities, whatever the criteria used to measure them. The choice of the redistributive criterion for transfers allocation therefore remains a considerable obstacle on the road toward a better understanding of the distributional aspects of public aid to education, and a wider consensus on the subject. 68 V. A Summary of Findings The main findings of the study may be summarized as follows. The public financing of education in Colombia contributes to redistributing in- come from the rich to the poor. This positive distributional impact on in- come is more decisive in urban than in rural areas and takes place because of the strong propensity of higher income groups in urban areas to enroll their children in fee-paying private schools. When each level of education is examined separately, it appears that only the public financing of primary education has a strong and positive redistributive effect on income. This effect is partially but not wholly offset by the impact of the public financing of secondary and higher education, according to which the lower and upper middle class respectively are subsidized by both the poor and the very rich. On the regional level, the combined effect of the taxation and transfers policies of the central government is to redistribute public funds for pri- mary education from the richer to the poorer departments. This conclusion holds even when municipal expenditures are considered. However, the ex- isting distribution of transfers to the departments does little to alter the wide existing disparities of educational opportunities among departments. Great care should be taken in drawing broad conclusions as to whether or not the public provision of education can be used as a policy tool to im- prove the distribution of income in developing countries. The reason is that the approach adopted here has been a static one; it refers to one point in time'-when education is provided-while education has also far-reaching future income redistributive effects that only a dynamic analysis can en- compass. Two observations will be made to illustrate this point. First, as the present analysis has shown, the net incidence of the public financing of education on income results from the comparison between the degree of progressivity (or regressivity) of the tax system and that of educa- tion subsidies. It follows that any policy action aimed at making a system of finance more equitable may act on either the structure of subsidies or the structure of taxation. One way to make sure that education subsidies bene- fit the poor more than the rich is to foster a private, little subsidized educa- tion sector in which the rich can enroll their children. Thus, the rich will re- ceive limited subsidies while they are still paying taxes. This is, to a certain 'For the analysis by income groups, 1970, for the regional analysis, 1967. 69 extent, the course followed by Colombia. Needless to say, if the presence of a strong private education sector may help to achieve equity in the provi- sion of education, it may also preserve and, possibly, exaggerate inequalities of economic opportunities in the future. It is very likely that if private schools are attended by students with favorable socio-economic back- grounds, they will provide them with a "better" education than public schools or at least an education with a "better" label, with the resulting effect that these students' subsequent economic opportunities may be systematically brighter than those of students coming from public schools. In other words, while private education may be the vehicle to achieve pres- ent equity in the provision of education, it may also turn out to be socially and economically divisive in the future. The second way to introduce more equity in the financing of education is to increase the progressivity of the tax system. In a country like Colombia, a special tax earmarked for the fi- nancing of secondary and higher education would alleviate adverse effects which the public subsidization of those levels of education would have on the distribution of income, provided that the progressivity of such a tax be stronger than that of the existing tax system. There is a second reason to believe that the search for equity in the pro- vision of education may not be enough to prevent the future deepening of income inequalities. It has to do with the long-run economic utility of transfers occurring through the public provision of education. Coming back to the case of Colombia, the question arises as to whether or not each peso transferred from the rich to the poor through the financing of primary education is worth as much in terms of the latter's future incomes as the peso transferred to the middle class through the financing of secondary and higher education. To give an appropriate answer to this problem, one should have an idea about the relative investment values of each type of subsidy, as assessed by the rates of return to investments in the various levels of education by income class. In the same way, a transfer from the rich to the poor through the public provision of education may be offset by a transfer from the poor to the rich arising from the public provision of another public good. However, if the relative investment values of these two goods are different, the respective economic positions of the two groups will be differently altered. A dynamic analysis of the incidence of the public financing of education on income distribution should comple- ment the static analysis described here. 70 ANNEX 1 Educational Expenditures and Total Expenditures by Level of Government: 1961-70 (Transfers are Netted out. Millions of Current Pesos) Central Decentralized Departments Municipalities All Levels of Government' Agencies Government Total Ed Total Ed Total Ed Total Ed Total Ed Year GDPI Exp Exp % Exp.l Exp I % Exp.s Exp 6 % Exp 7 Exp 5 % Exp Exp. % % GDP 1960 26,746 2,294 216 94 - - - 681 199 29 2 714 48 67 3,689 463 12 6 1 7 1961 30,421 3,495 335 9 6 - - - 721 245 34 0 814 56 6 9 5,030 636 12 6 2 1 1962 34,199 3,386 472 13 9 - - - 855 286 33 5 1,265 70 5 5 5,506 828 15 0 2 4 1963 43,526 4,176 544 13 0 - - - 1,048 338 32 3 1,539 79 51 6,763 961 14 2 2 2 1964 53,760 4,706 728 155 3,137 89 28 1,278 390 305 1,752 91 52 10,873 1,298 119 24 1965 60,797 5,808 782 135 3,590 160 45 1,391 479 344 1,972 101 51 12,761 1,522 119 25 1966 73,612 7,720 956 124 5,088 274 54 1,618 637 394 2,299 144 63 16,725 2,011 120 27 1967 83,083 8,132 1,057 130 7,049 271 38 1,817 713 392 3,016 161 53 20,014 2,202 11 0 27 1968 96,422 11,011 1,379 125 10,242 352 34 2,104 837 398 3,624 193 53 26,981 2,761 102 29 1969 110,953 14,664 1,866 127 12,781 503 39 3,580 1,126 315 4,167 222 53 35,192 3,717 106 34 1970 129,779 18,238 2,486 136 17,620 774 44 4,1139 1,3579 330 4,94610 26210 53 44,917 4,879 10.9 28 Notes All data refer to expenditures, plus reserves in the case of the central government The figures for the central government and the decentralized agencies are based on actual expenditures There are no data available on expenditures of decentralized agencies net of transfers between 1960 and 1964 As a consequence, the proportion of educational expenditures in the GDP is slightly underestimated during ihis period The figures for the departments and municipalities are based on actual expenditures between 1960 and 1969 For 1970, see our estimation m9 and 10 below Sources I Cuentas Nacionales, 1960-70 Banco de la Republica, Departamento de Investigactones Econ6micas 21961-69 Bo/eiin de ia Direccidn General del Presupuesta 57 (1970) Total Expenditures, p 147, Expenditures on Education, p 154 1970 Infarme Financiera 1970 Controlaria General de la Republica 31964-69 Excluding transfers from ihe central government Boletn de la Direccion General del Presupuesio 1970 Boletin de la Dlireccion General del Presupuesto 59 (August, 1971) 41964-69 Excluding transfers from the central government Boleln de la Direccion General del Presupuesto 57 (1970) 1970 Boleain de la Direccion General del Presupuesta 59 (August, 1971) 51960-67 Excluding the transfers received from the central government DANE Fiscal Statistics for the corresponding years 1968-69 Anuario de Ftscales y Financieras 1968-69 (BogotA DANE, forthcoming) 61960-67 Excluding the transfers for education received from the central government See Lebot, "Organizaci6n y funcionamiento," p 128-29 1968-69 Anuario de Fiscales y Financieras, 1968-69 71960-67 Excluding all transfers received by municipal entities DANE Fiscal Stalistics 1968-69 Anuario de Fisca/es y Financreras, 1968-69 81960-67 DANE Fiscal Statistics See Lebot, "Organizaci6n y funcionamiento," p 129 1968-69 Anuario de Fiscales y Financieras, 1968-69 gTotal expenditures are extrapolated from the 1969 figure on the basis of the 14 9 percent annual growth rate recorded during 1963-68 The pro- portion of total expenditures devoted to education is estimated at 33 percent t0Total expenditures are extrapolated from the 1969 figures on the basis of the 18 7 percent annual growth rate recorded during 1963-68 The proportion of total expenditures devoted to education is estimated at 5 3 percent ANNEX 2 Public Expenditures per Pupil in Colombia-1970 (Millions of Pesos unless Specified Otherwise) Primary Secondaryl Higher Other2 Total Central Government3 838 585 619 444 2,486 Decentralized Agencies4 51 51 154 518 774 Departments5 747 391 136 83 1,357 Municipalities6 262 - - - 262 Total 1,898 1,027 909 1,045 4,879 Subsidies to Private Education7 84 59 35 Enrollments mn Private Schools (Thousands)8 545 354 38 Subsidies per Pupil Enrolled in Private Schools (Pesos) 154 167 921 Total Public Expenditures on Public Schools 1,814 968 874 Enrollment in Public Schools (Thousands)8 2,481 342 46 Public Expenditures per Pupil Enrolled in Public Schools (Pesos) 731 2,830 19,000 lIncludes bachillerato, industrial, commercial and agricultural education, teacher training, schools for nurses and fine arts 2Includes extension services, administration, preprimary education and all types of education not elsewhere classified 3/nforme Financiero de 1970 (Bogota Controlaria General de la Repblica, Folio July 1971) p. 146; includes expenditures plus reserves. 4Decentrahzed agencies Total expenditures minus transfers from the central government: SENA 410 - 0 = 410 ICCE 268 - 167 = 101 Universities 515 - 361 = 154 Others 200 - 92 = 108 Total 1,393 - 620 = 773 Boletin de la D,reccion General del Presupuesto 59 (August, 1971) pp 168-76 and pp. 113-21 (ICCE expenditures on primary education are assumed to be of total expenditures) 5Assuming the same allocation of total expenditures among levels as in 1969. 6Assuming that all municipal expenditures on education are devoted to primary education. 7Primary and secondary education 10 percent of central govemment expenditures at these levels. Higher education Reforma de la educaci6n, superior, documentos de eseudio (Bogota ICFES, 1971) Table 15. 7Enrollment statistics in thousands are the results of forecasts based on the 1958-68 period carried out by INANDES (Instituci6n Andina de Estudios Sociales) Primary Education 3,026 of which 18 percent Bachillerato and are enrolled Commercial Education 572 55 percent in private schools Teacher Training 62 25 percent according to our own estimation Industrial Education 27 18 percent Others 35 55 percent our estimations Total Secondary 696 51 percent Higher Education see Reforma de la educactin superior, documentos de estudio Table 8. Selected Bibliography Bird, Richard. Taxation and Development, Lessons from Colombian Experi- ence. Cambridge, Mass.: Harvard University Press, 1970. Bird, Richard, and de Wulf, L. Taxation and Income Distribution in Latin America: a Critical Review of Empirical Studies. IMF Fiscal Affairs Department, DM/7318, Washington, D.C.: 1973. Bonnen, James T. "The Absence of Knowledge of Distributional Impacts: an Obstacle to Effective Policy Analysis and Decisions." In Public Ex- penditures and Policy Analysis, edited by Robert H. Haveman and Julius Margolis. Chicago: Markham, 1970. Charles, K.J. "The Distribution and Utilization of Public Services." Mimeographed. New York: UN Income Distribution Project. Cohn, Elchanan; Gifford, Adam, and Sharkansky, Ira. "Benefits and Costs of Higher Education and Income Redistribution: Three Comments." Journal of Human Resources 5 (Spring 1970): 222-36. C6rdoba, P. Distribuci6n de ingresos en Colombia y su contribucion alproyecto de las Naciones Unidas sobre el sistema complementario de estadisticas de la distribuci6n del ingreso. Bogota: DANE, 1971. C6rdoba, P. La encuesta nacional de hogares de Colombia. Bogota: DANE, 1971. Gillespie, W. Irwin. "Effect of Public Expenditures on the Distribution of Income." In Essays in Fiscal Federalism, edited by Richard A. Musgrave. Washington, D.C.: Brookings Institution, 1965. Grubb, W. Norton. "The Distribution of Public Costs and Private Benefits in an Urban Public School System." National Tax Journal 24 (March 1971): 1-12. Hansen, W. Lee, and Weisbrod, Burton A. Benefits, Costs and Finance of Public Higher Education. Chicago: Markham, 1969. /Hansen, W. Lee, and Weisbrod, Burton A. "The Distribution and Direct Benefits of Public Higher Education: the Case of California." Journal of Human Resources 4 (Spring 1969): 176-91. /Hansen, W. Lee, and Weisbrod, Burton A. "On the Distribution of Costs and Benefits of Public Higher Education: Reply." Journal of Human Resources 6 (Summer 1971): 363-74. Hansen, W. Lee, and Weisbrod, Burton A. "Who Pays for a Public Expen- diture Program?" National Tax Journal 24 (December 1971): 515-17. Hartman, Robert W. "A Comment on the Pechman-Hansen-Weisbrod Controversy." Journal of Human Resources 5 (Fall 1970): 519-23. 73 Jallade, Jean-Pierre. The Financing of Education: an Examination of Basic Issues. IBRD Staff Working Paper No. 157. Washington, D.C.: World Bank, 1973. Judy, Richard W. "The Income Redistributive Effects of Public Aid to 4 Education in Canada." In Canadian Economic Problems and Policies, edited by Lawrence H. Officer and Lawrence B. Smith, pp. 302-17. New York: McGraw Hill, 1970. McLure, Charles. "The Incidence of Taxation in Colombia." In Fiscal Reform for Colombia: the Final Report and Staff Papers of the Colombian Commission on Tax Reform, edited by Richard A. Musgrave and Malcolm Gillis. Cambridge, Mass.: Harvard Law School International Tax Program, 1971. McLure, Charles. "The Incidence of Colombian Taxes: 1970." Mimeographed. Houston, Texas: Discussion Paper No. 41, Program of Development Studies, Rice University, 1973. Montoya, Miguel Urrutia. "Resenia de los estudios de distribuci6n de in- greso en Colombia." Revista del Banco de la Republica 43 (February 1970) 180-91 Montoya, Miguel Urrutia, and de Sandoval, Clara Elsa. "Reforma de la educaci6n publica y sus implicaciones fiscales." Mimeographed. Bogotai National University of Colombia, 1970. Montoya, Miguel Urrutia, and de Sandoval, Clara Elsa. "Politica fiscal y dis- tribucion del ingreso en Colombia." Revista del Banco de la Republica 44 (July 1971): 1072-87. Musgrave, Richard A., ed. Essays in Fiscal Federalism. Washington, D.C.: Brookings Institution, 1965. Musgrave, Richard A., and Gillis, Malcolm, eds. Fiscal Reform for Colom- bia: the Final Report and Staff Papers of the Colombian Commission on Tax Reform. Cambridge, Mass.: Harvard Law School International Tax Program, 1971. Pechman, Joseph A. "The Distributional Effects of Public Higher Educa- tion in California." Journal of Human Resources 5 (Summer 1970): 361-70. Rice University, Program of Development Studies. "A Proposal for Research on Distribution of Gains, Wealth and Income from Economic and Political Development." Mimeographed. Houston, Texas: 1971. Sahota, Gian. "Public Expenditures and Income Distribution in Panama." Mimeographed. Panama City: USAID, 1972. Tax Foundation. Tax Burdens and Benefits of Government Expenditures by In- come Class, 1961 and 1965. Research Publication No. 9. New York: 1967. Windham, Douglas M. The Redistributional Effects of Public Higher Educa- tion in Florida. Greensboro: University of North Carolina, Department of Economics and Business Administration, Monograph Series F, No. 1, 1969. 74 Library of Congress Cataloging in Publication Data Jallade, Jean-Pierre. Public expenditures on education and income distribution in Colombia. (World Bank staff occasional papers, no. 18) Includes bibliographical references. 1. Education-Colombia-Finance. I. Title. II. Series. LB2899.C62J34 379'.121'09861 74-4216 ISBN 0-8018-1628-9 EXPORI' SALES REPRESENTATIVES THE UNITED KINGDOM, CONTINENTAL EUROPE, THE NEAR EAST AND MIDDLE EAST, AFRICA IBEG Ltd., 2-4 Brook Street, London WIY-IAA, England CANADA Thc Copp-Clark Publishing Co., 517 Wellington Street, West, Toronto MSV-IGI, Ontario, Canada CENTRAL AMERICA, SOUTHI AMERICA, CARIRIBEAN (EXCLUSIVE OF MEXICO, IN- CLUDING l'UERT'O RICO), INDIA, I'AKISTAN Kainian & Polon, Inc., 456 Sylvan Avcnue. Englewood Cliffs, New Jersey, 07632 MEXICO Centro lnterameri is, Mexico 4. D. F. THIROUGHIOUT RIS1T OF *** LB 2899 .C62 J34 c.2 ns Hopkins Universit Jallade, Jean Pierre. placed with the f( Public expenditures on education AUSTRALIA, NEW and income distribution in Australia an Colombia. vale, N.S JAPAN United Pu Chiyoda-ku A Distrib WORLD BANK 1818 H Street, N.W. Washington, D.C. 20433, U.S A. Telephone: EXecutive 3-6360 Cable Address: INTBAFRAD