Deferred cost recovery for higher education

Public Disclosure Authorized Public Disclosure Authorized Public Disclosure Authorized Public Disclosure Authorized )PA /f3J7 137 3 World Bank Discussion Papers Deferred Cost Recovery for HiPgher Education Student Loan Programs in Developing Countries Douglas Albrecht Adrian Ziderman Recent World Bank Discussion Papers No. 79 Analyzing Taxeson BusinessIncomewith the MarginalEffectiveTax Rate Model. David Dunn and Anthony Pellechio No. 80 EnvironmentalManagementin Development:The Evolutionof Paradigms.Michael E. Colby No. 81 Latin America'sBankingSystemsin the 1980s:A CrossCountryComparison.Felipe Morris, Mark Dorfman, Jose Pedro Ortiz, and others. No. 82 Why EducationalPoliciesCan Fail:An Overviewof SelectedAfricanExperiences.George Psacharopoulos No. 83 ComparativeAfricanExperiencesin ImplementingEducationalPolicies. John Craig No. 84 ImplementingEducationalPoliciesin Ethiopia.FassilR. Kiros No. 85 ImplementingEducationalPoliciesin Kenya. G. S. Eshiwani No. 86 ImplementingEducationalPoliciesin Tanzania.C. J. Galabawa No. 87 ImplementingEducationalPoliciesin Lesotho,T. Sohl Thelejani No. 88 ImplementingEducationalPoliciesin Swaziland.Cisco Magalula No. 89 ImplementingEducationalPoliciesin Uganda.Cooper F. Odaet No. 90 ImplementingEducationalPoliciesin Zambia.Paul P. W. Achola No. 91 ImplementingEducationalPoliciesin Zimbabwe.0. E. Maravanyika No. 92 InstitutionalReformsin SectorAdjustmentOperations:The WorldBank'sExperience.Samuel Paul No. 93 Assessmentof theP'ivateSector:A CaseStudy and Its Methodological Implications.Samuel Paul No. 94 ReachingthePoorthroughRural PublicEmployment:A Surveyof Theoryand Evidence.Martin Ravallion No. 95 Educationand Development:EvidenceforNew Priorities.Wadi D. Haddad and others No. 96 HouseholdFoodSecurityand the Role of Women.J. Price Gittinger and others No. 97 Problemsof DevelopingCountriesin the 1990s. VolumeI: GeneralTopics.F. Desmond McCarthy, editor No. 98 Problemsof DevelopingCountriesin the 1990s. VolumeII: CountryStudies.F. Desmond McCarthy, editor No. 99 PublicSectorManagementIssuesin Structural AdjustmentLending.Barbara Nunberg No. 100 The EuropeanCommunities'SingleMarket:The Challengeof 1992for Sub-SaharanAfrica.Alfred Tovias No. 101 International Migrationand Developmentin Sub-SaharanAfrica. VolumeI: Overview.Sharon Stanton Russell, Karen Jacobsen, and William Deane Stanley No. 102 International Migrationand Developmentin Sub-SaharanAfrica. VolumeII: CountryAnalyses.Sharon Stanton Russell, Karen Jacobsen, and William Deane Stanley No. 103 Agricultural ExtensionforWomenFarmersin Africa.Katrine Saito and C. Jean Weidemann No. 104 EnterpriseReformand Privitizationin SocialistEconomies.Barbara Lee andJohn Nellis No. 105 Redefiningthe Role of Governmentin Agricultureforthe 1990s. Odin Knudsen, John Nash, and others No. 106 SocialSpendingin LatinAmerica:The Storyofthe 1980s. Margaret E. Grosh No. 107 Kenyaat the Demographic TurningPoint?Hypothesesand a ProposedResearchAgenda.Allen C. Kelley and Charles E. Nobbe No. 108 Debt ManagementSystems.Debt and Intemational Finance Division (Continued on the inside back cover.) 13 7 1z1 E World Bank DiscussionPapers Deferred Cost Recovery for Higher Education Student Loan Programs in Developing Countries Douglas Albrecht Adrian Ziderman The World Bank Washington, D.C. Copyright C 1991 The International Bank for Reconstruction and Development/THE WORLD BANK 1818 H Street, N.W. Washington, D.C. 20433, U.S.A. All rights reserved Manufactured in the United States of America First printing November 1991 Second printing December 1993 Discussion Papers present results of country analysisor research that are circulated to encourage discussion and comment within the development community. To present these results with the least possible delay, the typescript of this paper has not been prepared in accordance with the procedures appropriate to formal printed texts, and the World Bank accepts no responsibilityfor errors. The findings, interpretations, and conclusions expressedin this paper are entirely those of the author(s) and should not be attributed in any manner to the World Bank, to its affiliatedorganizations, or to members of its Board of Executive Directors or the countries they represent. The World Bank does not guarantee the accuracy of the data included in this publication and accepts no responsibilitywhatsoever for any consequence of their use. Any maps that accompany the text have been prepared solelyfor the convenience of readers; the designationsand presentation of material in them do not imply the expressionof any opinion whatsoever on the part of the World Bank, its affiliates,or its Board or member countries concerning the legal status of any country, territory, city, or area or of the authorities thereof or concerning tbe delimitation of its boundaries or its national affiliation. The material in this publication is copyrighted. Requests for permission to reproduce portions of it should be sent to the Office of the Publisher at the addressshown in the copyright notice above. The World Bank encourages dissemination of its work and will normally give permissionpromptly and, when the reproduction is for noncommercial purposes, without asking a fee. Permission to copy portions for classroom use is granted through the Copyright ClearanceCenter, 27 CongressStreet, Salem,Massachusetts 01970, U.S.A. The complete backlist of publications from the World Bank is shown in the annual Index ofPublications, which contains an alphabeticaltitle list (with full ordering information) and indexes of subjects, authors, and countries and regions. The latest edition is availablefree of charge from the Distribution Unit, Office of the Publisher, The World Bank, 1818 H Street, N.W., Washington, D.C. 20433, U.S.A., or from Publications, The World Bank, 66, avenue d'lena, 75116 Paris, France. ISSN: 0259-210X Douglas Albrecht is a consultant to the Education and Employment Division of the World Bank's Population and Human Resources Departrnent. Adrian Ziderman, formerly a senior economist in that division, is now a professor of economics at Bar Ilan University, Israel. Library of Congress Cataloging-in-Publication Data Albrecht, Douglas, 1965Deferred cost recovery for higher education: student loan programs in developing countries / Douglas Albrecht, Adrian Zidernan. p. cm. - (World Bank discussionpapers ; 137) Includes bibliographical references. ISBN 0-8213-1952-3 1. Student loan programs-Developing countries-Costs. I. Ziderman, Adrian. II. Title. III. Series. LB2340.4.D44A43 1991 91-35046 378.3'62'091724-dc20 CIP Abstract Given the growing demand for access to higher education, and the declining quality and available resources from governments, many governments, particularly in developing countries, have attempted to increase studentcontributions. In many instances,however,governmentshave encounteredproblems increasing cost recovery without deterring access among lower income students. Despite clear economic and financial rationale, tuition increases can be difficult to implement because of the inabilityof many students (and their parents) to pay fees out of current income.Many economists have advocated student loan programs as a means to increase private contributionswhile also preserving access. This paper analyzes the experience of existing loan programs, particularlyin developingcountries,in order to understand their role in fostering cost recovery. Currently,loan programs existin over 50 developingand industrial countries,and have most commonly been introduced to assist students to pay their livingexpenses. In somewhat fewer cases, they are used to support direct payments of instructional costs, and thus to expand the resource base of institutions. With the exception of four programs, all have taken the form of "mortgage loans", in which students make fixed payments over a fixed time period. In contrast, programs in Australia, Swedenand Ghana, require payments in relation to income; Chile's programs allow for graduated annual payments. Detailed financial analysis of 24 loan programs shows that present value of the repayments collected by loan programs constitutes a small percentage of the loan value disbursed (and the costs of administering the loan). In some instances,loan programs have been more expensivethan continuingwith a policy of outright grants. In general, developingcountry loan programs to date have not reduced significantlythe government's fscal burden for higher education. In order to improve financial effectiveness,programs should be targeted toward the most needy and able students. Hidden subsidies should be limited by charging positive real interest rates, combined with repayment plans that take account of the likely pattern of graduate earnings. Default reductions require that loan programs be managed by institutionswith the capacity and financial incentivesto collect - namely banks, private collection agencies, or taxation departments. Such reforms offer great potential to transform small programs into relatively efficient forms of student support. Larger programs, however, may be more difficult to manage. Some countries have considered alternatives which preserve the basic concept of payingfor education from future income. The most notable is a graduate tax in which a student pays a fixed percentage of income over the entire working life, regardless of how much is repaid. In the presence of an effective tax system, a graduate tax could bring in significantly more revenue than traditional loan programs. Besides improved financial efficiency, income contingent payments may be more equitable since they limit the risk to poorer students. In countries with weak taxation systems, this option may not be feasible. Without reliable financial institutionsor effective taxation mechanisms,loan programs and taxes may not have much potentiaL Rather, national service programs, differential fees, targeted scholarships and community support may be more effective alternatives. In generaL deferred cost recovery can help reduce government burdens, but only where institutional capacity exists. Acknowledgements Special thanks are due to Adriaan Verspoor, Chief of the Education and Employment Division,for general encouragement and comments on earlier drafts. Valuable comments were also received from Jamil Salmi, V. Selvaratnam, Elizabeth King, Moussa Kourouma, WilliamSaint and Christopher Shaw. This paper is the first in a series of papers by the authors on issues related to higher education finance. FOREWORD The World Bank has long acknowledged the important relationship between education and economic development, and in particular, the critical role of higher education institutions in providing leadershipfor education systems as a whole. Ever since the World Bankbegan lending for education in 1963, its aim has beento assist developing countries expand and improve their education systems. But the rapid expansionof higher education systems over the last three decades,compoundedby the more recent global economic crisis, has left many institutions short of funds in relation to the demands imposed on them. The impact has been most severe on institutions solely dependent on governments for funding. The result has been declining quality as well as insufficient funds to help many needy students meet high living costs associatedwith attending universities. It is therefore crucial that nations begin to find alternativeor supplementarysources of revenuefor institutions, as well as to utilize scarce resources more effectively and efficiently in pursuit of their educational objectives. This study is part of a series on issues related to higher education reform and finance currently being conducted by the Educationand Employment Division of the Population and Human ResourcesDepartment of the World Bank. The goal of this study is to help decision makersexplorealternatives to diversify the resource basefor their higher education institutions through cost recovery, while minimizing negative impacts on vulnerable groups. Ann 0. Hamilton Director Population and Human ResourcesDepartment I Tables of Contents Page I Introduction ............................... 2 ExistingStudent Loan Programs ................................................ Coverage ............................. Repayment format ..................... Administering Institutions ............................... Purpose of Support ............................... Loan Value and Students Covered ............. 1 3 3 3 6 9 11 .................. 3 ............... The Financial Impact ....... Loan Recovery Ratio ............................... Loans in Relation to University Costs ...................................... 13 13 19 4 .............................. Improving Performance ............ ....................................... Targeting Loan Support . Reducing Subsidies While Limiting Debt Burden .......... .......................................... MinimizingEvasion . Conclusions.......................................... 23 23 28 30 33 .................. 5 Alternative Scenarios ....................................... Graduate Tax .......................................... ............................. Employer Taxes ............. .......................................... Community Service. .......................................... Some Conclusions. 35 35 42 44 46 6 Moving Forward .......................................... 47 Annex 1 Annex 2 Checklist of Policy Options for Deferred Cost Recovery.50 MethodologicalNote for CalculatingSubsidies on Mortgage Loan Programs.52 Box 1 : Ghana: Using Social Security for Repayment. Box 2: Student Maintenance and Higher Education Budgets .10 Box 3: Equity and Risk Aversion.19 Box 4: Brazil: Establishingthe Costs of a Loan Program .22 Box 5: Means Testing at the Universityof the Philippines .26 Box 6: USA. Quality Restrictions and Efficiency.27 Box 7: Grants in Addition to Loans .28 Box 8: Sweden: Using Financial Efficiencyto Improve Equity .30 Box 9: Honduras: Reducing Default is Expensiveand Deters Low Income Students.................................................... Box 10: Equity Finance at Yale . Box 11: Argentina's Proposed Graduate Income Tax....................................... Box 12: Nepal: National Development Service ........................................... .................... ............ Table 1. ExistingStudent Loan Programs ...... Table 2. Hidden Subsidiesand Costs of Selected Student Loan Programs. Table 3. Non Repayment of student loans as percentage of total loans .17 Table 4. Effective Cost Recovery from Loan Recipients at Public Universities (as a fraction of Unit Instructional costs ) .20 9 ,,.. 31 36 41 45 5 15 Table 5. Student Loans Verses Graduate Taxes: Contrasts and Similarities....................................................... Table 6. Present Value of Net Payments Under Alternative Deferred Cost Recovery Programs (Australan Data) ............ ............................. References ................................................................ 38 40 54 1. Introduction The financial problems of higher education have been well documented in recent years (Psacharopoulosand Woodhall 1985). Decliningquality due to overcrowding,growingdemand for access and constraints on government budgets imply that higher education systems must seek alternative sources of income. In parts of the developing world, slowing growth during the 1980's and rising costs of providing training at internationally competitive standards have led to a further erosion of institutional capacity. In addition, resource constraints for higher education institutionsare compounded by government commitments to subsidize student living expenses. In many instances, government expenditures on student support has equaled or even surpassed educational expenditures (Psacharopouloset al 1986). Many governmentsargue that student support is justified as a means to enable students to attend higher education while they are not earning income. In other instances,student support is part of a general welfare policy that relies on progressive taxation to redistribute income: students are entitled to a minimum social income while they forgo earnings. The combinationof rapid expansion,macro-economicdifficulties,and commitmentsto student support have left governmentsseeking means to relieve budgetary pressures. Additional funds can come from two sources. Institutions can become more efficient, and thus free up resources; or institutionscan diversify their resource base by bringing in more external funds. The most obvious source of additional income is from the direct beneficiaries of higher education - the students. In addition to the budgetary rationale for mobilizingstudent contributions,recent economic analyses have demonstrated efficiencyand equity rationale for recovering at least part of the cost of higher education from students (Psacharopoulosetal 1986;Jimenez 1987; Birdsall and James 1990). In sum, cost recovery is believed to lead to a more efficient use of public and private resources; to increase the equity of educational systems which tend to attract elites or produce future high income earners; and to provide an expanded source of revenue to support more educational opportunity and better quality. In many instances, however, imposing cost recovery -- either for living expenses or for instructional costs - has proven politicallydifficult,and has raised the problem of how to relieve the pressure on students who cannot afford to pay. To resolve this problem, much economicliterature has advocated student -2 loans to enable students to defer payment for the costs of attending higher education until they are earning incomes. We refer more broadly to defrred payment programs to include those policy instruments which secure payment from the future incomesof students, rather than their current resources. Extensivetheoretical and comparative literature on student loans has been developed by Maureen Woodhall. A particular emphasis of her work has been on the potential role of loans in developingcountries (Woodhall 1983,1987(a), 1987(b), 1991). Johnstone (1986) has surveyed student support mechanisms in industrial countries. More theoretical discussionshave been developed by Mingat, Tan and Hoque (1985),and Psacharopoulosand Woodhall (1985). In recent years, alternative formats for loans, particularly loans with income contingent repayments' have received considerable attention (Barnes and Barr 1988; Barr 1989; Woodhall 1990b and 1991). While most of this literature has been extremely optimistic about the efficiencyof student loans, few studies have actuallyexamined their financial impact, particularly in devcloping countries. In this paper, therefore, we examine the financial impact of current and past programs on government and student budgets, highlightkey obstacles, particularly with regard to payment formats and administering institutions. The paper then tuTnsto strategies for improvemenL Overall,we concludethat while it is possible to improve small scale loan programs that have had, until now, only a marginal impact on reducing government expenditures, most student loan programs possess severe limitationsin their present forms. The plan of the paper is as follows. The main characteristics of loan programs in fifty countries are discussed in Section 2. In Section 3, the financial performance of 23 of these programs are examined in detail while in Section 4 we suggest policy reforms that would lead to improved fiancial performance. Alternative cost recovery mechanisms are discussed in Section 5 and some conclusionson the feasibilityof introducinga loan program concludes the paper in Section 6. Loanswith income-contingentrepaymentshave somewhat misleadinglybeen labeled "income-contingent loans"in the remainder of the paper, we use the more common term. -3 - 2. E ft1gStudentL Program Student loans programs have been developed in various forms in over 50 countries throughout the world. Summaryinformation on these programs is listed in Table 1, in terms of geographiccoverage, type of repayment format, administering institution, purpose of loan support, average value of the loan and the proportion of students covered by the loans scheme. In general, developing country student loan programs have been used to assist with student livingexpenses and typicallycover only a few percent of the student population. Coage The present study has identified 20 programs in Latin America and the Canbbean, eight in Asia, four in the Middle East and Northern Africa, seven in Sub-Saharan Africa and 14 in industrialized countries. Noteworthy is the large number of loan programs in Latin America and the Caribbean: first implemented in Colombia in 1953 to assist graduate students to meet the costs of overseas study (Woodhall 1983), loan programs (referred to locallyas student credit programs) are now in place in most countries in the region. This contrasts with the paucity of programs in other developingcountries, especiallyin the Middle East and Africa, where indeed some programs have been abandoned in recent years. Many countries have no single loan program. Federated countries often have locally run systemsof support. Canada, for instance,has national and provincialloan schemes. The US has federal, state and institutionalloan programs. In Latin America, many countries haveseveral loan programs, often sponsored by private non-profit groups, govemment ministries,and large companies. Repayment ftmat With the exception of four schemes, all programs offer students credit in the form of a T mortgage"loan. In this traditional mortgage-tvpeloan,repayment is made over a specifiedperiod, usuallywith fixed monthlypayments; interest rates and the maximum length of repayment are used to calculate the fixed - 4periodic payments.In contrast to this regime of equal nominal payments,most of the universitiesin Chile allow graduated nominal payments: borrowers from Chile's CatholicUniversity,for example, pay equal real (rather than nominal) amounts, thus ensuring that the first payments are not excessivelylarge in real terms in relation to others. A third type of repayment mechanismis an income contingent loan, in whichloans are repaid as a proportion of a graduate's income each year. Income contingentloans are expected to be more favorable to low-incomestudents. The basic problem of borrowing for education, is that the outcome is risky,since the future value of a degree is not immediately apparent. The risk is greatest for students from poorer backgrounds: future job and earnings opportunities may be less favorable for the poor, and fLxedfuture repayments commit the debtor to repay an open ended proportion of his income. In addition, the poor tend to be more risk averse than the well-to-do(Reuterberg and Svennson1990; Barr 1990). Therefore mortgage loans may deter access among the very groups that loans are intended to reach. -5TabLe 1. Existing Country (Loan Organization) Repayment Mechanism LATIN AMERICAANDCARI BBEAN Argentina (INCE) Mortgage Loan Barbados (SRLF) Mortgage Loan Bolivia (CIDEP) Mortgage Loan Brazil (CEP) Mortgage Loan Chile Graduated Cotombia (ICETEX) Mortgage Loan Costa Rica (CONAPE) Mortgage Loan Dominican Republic (FCE)Mortgage Loan Ecuador (IECE) Mortgage Loan El Salvador (Educredito)Mortgage Loan Honduras (Educredito) Mortgage Loan Jamica (SLB) Mortgage Loan Mexico Mortgage Loan Nicaragua (Educredito) Mortgage Loan Panare (IFARHU) Mortgage Loan Peru (INABEC) Mortgage Loan Trinidad (SRLF) Mortgage Loan Venezuela (Educredito) Mortgage Loan (FGMA) Mortgage Loan (BANAP) Mortgage Loan ASIA China India indonesia * Mortgage Loan Mortgage Loan Mortgage Loan Korea Malaysia Philippines Pakistan Sri Lank Mortgage Mortgage Mortgage Mortgage Mortgage Loan Loan Loan Loan Loan MIDDLEEAST. NORTHAFRICA Egypt Mortgage Loan Israel Jordan Morocco SUB-SAHARAN AFRICA Ghana Kenya Nigeria* Riwada Burundi* Malawi Tanzania* INWUSTRIAL COUNTRIES Australia Canada (Quebec) Denmrk Finland France Cermany Hong Kong Netherlands Norway Japn Singapore Mortgage Loan Mortgage Loan Mortgage Loan Student Loan Progr Administering Institution AutonomousBody AutonomousBody Commercial Banks Universities AutonomousBody Commercial Banks AutonomousBody AutonomousBody AutonomousBody AutonomousBody AutonomousBody Commercial Banks Autonomous Body Autonomous Body Autonomous Body Autonomous Body Other Universities Commercial Banks Universities Other Universities and Commerciat Banks Commercial Banks Commercial Banks Commercial Banks Commercial Banks L Purpose of Su,oport Living Tuition Living Tuition Tuition Tuition Tuition Living Living Tuition Tuition and Living and Living and Living and Living and Living and Living and Living and Living Tuition Living Tuition Tuition and Living Commercial Banks Tuition $280 Percent of Year students Begun with loans Data year 1976 12X 1989 1974 1981 1953 1977 25X 1989 1988 1985 1983 oX 3X Tuition Tuition Tuition Tuition $11,000 $400 Tuition and Living Tuition and Living Tuition and Living Tuition and Living AutonomousBody Commercial Banks Commercial Banks Average Loan Value $2,700 $405 $400 $2,200 $700 $85 S550 $1,300 and Living 1976 1970 1X 20X 1966 6a 1972 1967 1975 1991 1985 1X 1X 1X 1991 1991 1991 1987 1963 1982 30K 1X 3K 1989 1989 1986 1975 1985 1976 1974 1964 1X and Living SK 2K 12X 1980 1980 1983 <1X 1990 Income Contingent Government Dept. Mortgage Loan Commercial Bank Mortgage Loan Autonomous Body Living Living Tuition $200 S845 1989 1973 68X 100K 1990 1990 Mortgage Loan Mortgage Loan Living Living $80 1988 50K 1989 $1,750 $2,800 $3,700 $2,200 1989 1963 1975 1986 81K 59K 1990 1990 1985 1987 $1,500 $1,050 S200 $4,000 $2,500 1974 1969 Income Contingent Mortgage Loan Mortgage Loan Mortgage Loan Mortgage Loan Mortgage Loan Mortgage Loan Mortgage Loan Mortgage Loan Mortgage Loan Mortgage Loan Government Dept. Commercial Banks Commercial Banks Government Dept. Government Dept. Tuition Tuition Living Living LIving Living Tuition Living Living Tuition Tuition and living and Living Sweden United Kingdom USA Autonomous Body Autonmous Body Government Dept. Coamercial Banks Income Contingent Autonmous Body Mortgage Loan Autonomous Body Mortgage Loan Commercial Banks Blanks imply informtion Program in Indonesia, was not available. Israel, Nigeria, Tanzania and 8urundi have been abandoned. Living LivIng Tuition and Living and Living and Living $5,828 $750 $2,176 1990 1964 1K 30X 26K B0X 19X 39X 1987 1969 1969 1986 1967 1990 7K 28X 1990 1987 -6 Income contingent loans constitute a mechanism for achievinga balance between effective recovery of costs and minimum risk to the borrower. Here, the size of repayment is linked to the graduate's income. Income contingencythus limits debt burden in a given period, and also targets more subsidiesto ower wage earners. Since high earners have to pay their loans more quickly,they benefit less from any subsidy low earners are able to repay more slowly,and therefore receive greater subsidies. Currently, there are three income contingent loans programs - in Sweden, Australia and Ghana. In Sweden, students are now required to pay four percent of their annual income to the loan fund until their debt is repaid. The schemes in Ghana and Australia respectively,use social security contributionsand income taxation for loan repayment.2 Mmllsterlg Instittons For the most part, credit programs are administered through public institutions. Even where the private sector is responsible for lending (as in the US) the government acts as a guarantor on loans. Public interventionstems from a failure on the part of private markets to supply credit for unsecured human capital loans.3 Public interventions have taken four major forms. In most countries, public intervention has led to the creation of autonomous public lending bodies. These institutionshave often been labeled "revolvingfunds"which, once capitalized, are expected to finance themselves through repaymentsfrom earlier loans. Yet, as wil be argued, this is rarely the case, since loans are generallyheavily subsidized and result in losses. The advantage of this type of lending institution is that it allowsstronger control over targeting policies,and the introductionof non-traditionaltype loans - such as income contingent loans. Such autonomous bodies exist throughout Latin America and Europe, as well as in Hong Kong,Egypt and Nigeria. They manage student selectionbased on merit, need and national priorities. 2 In Australia, former students will repay their debt through a graduated surplus income tax (one, two or three per cent of income). Although the Australian scheme is sometimes referred to as a graduate tax, it is in fact a loan collected through the taxation system.While a true graduate tax has not been implemented in any country, it will be examined as an alternative strategy in section 5. 3 Govemment intervention is also necessitated from a demand perspective. Student demand for credit is likely to be constrainedbecause of uncertaintiesamong poorer students as to the value of a degree in relation to proposed debt. This requires the government to act to minimizerisk both for borrowers as well as creditors. In Latin America, many manage overseas scholarshipand loan programs (such as the FGMA in Venezuela and Educredito in Honduras). Institutionalstrength howevervaries tremendously: some, particularlyin Europe and in Colombia, are quite strong, while others lack basic managerial and physical resources necessary to administer the program. As a result, many of these bodieshave begun to delegate loan administration to third parties such as commercial banks. A second common administrativearrangement is the use of commercialbanks. Participating banks have been both publicly and privately owned. Some manage programs entirely, with or without government guarantees, while others act simply as collection agents. In Brazil, the government owned commercial banking system has managed the portfolio of student loans since 1975; the government sets broad policy regulations for the loan program, while local branches of a commercial banking system execute distribution and recovery. Decentralization can make processing and collection more efficient, while the banking system'sprevious lendingexperience, and control over individual'saccessto future credit, makes them more effective in reducing default. The administrativeefficiencyof these institutions tends to be better than the autonomous bodies. Public commercial banks have been used in Indonesia, Pakistan, Barbados and Venezuela. Private Banks have managedprograms both with and without government support. In the US, private banks disburse and collect money from students,while the government guarantees and subsidizesthe loans. There are three major motivationsfor relyingon the private sector: first, the government does not have to make initial capital outlays;second, the government hopes to harness the efficienciesof the private sector and reduce the costs of a loan program; third, the government does not have to set up a potentially costly administrative apparatus to handle the program. Other countries utilizingprivate banks in a similar fashion are Canada and Denmark. Opting for private banks does not ensure effectiverecovery. Default is a problem in the US, but the source of high default is not reliance on the private sector per se, but rather faulty policies and incentive structures. While private banks may represent the lowest cost approach for governments, the strategy is feasible only if a banking system is in place and even then, banks may not wish to participate in the program - as in the UK - because of potential losses and feared "bad images' with future clientele. - 8In some countries, private banks have begun student loan programs without any government guarantees or subsidies. Such programs, as in Morocco,typicallysupport private institutionsthat offer training in fields that lead to high salaried employment A program exists in Indonesiato help finance high tuition fees for elite business programs. These banks tend to loan money to secure borrowers (not poorer students) and for students studying in fields that guarantee high private returns to the investment As a consequence, they indirectlyprovide incentives for universitiesto expand programs in fields of relevance to the labor market A final administrativeapproach utilizes existinggovernment structures for disbursement and collection. To address some of the administrative problems involvedwith income contingent collection, two countries (Ghana and Australia) utilize the governmentrevenue collectionsystemsto recover loans. In Ghana, the collection is managed by the Social Security department; in Australia, through the national income tax. Transactions are made directly to and from the budgets of different government bodies, without creating new administrativestructures. While there may be little conceptual difference between a loan repaid through the taxation system or to a bank, there may be a considerable difference in the effectivenessin recovering funds and in administrativecosts. In addition, if government structures are used, then the government usuallyneeds to make the initial capital outlays for the program. The mechanics of collecting contributions in Ghana and Australia are quite different. In Ghana, students have been able to borrow money from the government to pay newly imposed fees for living expenses; repayment is made through the existing social security tax on all wage earning employees, by deferring the accumulation of retirement benefits (see Box 1). Social security payments have a particular advantage because in many systems individualshave an incentive to pay this tax, since they derive benefits in proportion to what they pay. Australia, in contrast, has implementeda systemwhere the Ministryof Education disburses funds to students to cover fees; repayment is through a graduated surplus income tax, with outstanding debt assessed at a zero percent real interest rate, and tax rates of one per cent, two percent and three percent depending on individualincome. 9- Box 1: Ghana.Using Social Securityfor Repayment In 1989, the Ghanaian government began to charge universitystudents for housing and meals. At the same time, it offered students an optional loan worth about $200 to help meet these costs. The most innovative aspect of the loan is the collection mechanism, through the social security system. Graduates repay their loans through their standard social security deduction which goes to the education budget rather than to their own benefit accounL Students, therefore, repay their loans not through an increased social security tax rate, but rather by deferring contributing to their own retirement accounts until the loan is repaid. Once a graduate finds employment,the standard 5 percent payroll deduction plus the employer's 12 percent contribution goes to the Ministry of Education rather than the retirement accounts. The program is not without problems. A first concern is a large interest subsidy on the loan. More puzzlinghowever is whether the student actually makes any contribution. The scheme may not actually collect any additional revenues for the government; rather, the social securitysystem may be subsidizinguniversityeducation. This is because workers usuallyaccumulate maximumretirement benefits some years before retirement, but continue to contribute to the social security system. Thus, even if students wait four years before starting to accumulate their retirement benefits, the normal work life may be such that these students anyhowwould have worked an extra four or five years beyond the period that full retirement benefits had been accumulated. In the final analysis, the government may have to find extra funding for the social security system. Several other countries have considered usingthe taxation system rather than specializedloans institutions,but have rejected this on grounds of administrativeinfeasibility. In the UK and New Zealand, the taxation institutions did not want the additional burden they felt a loan scheme would impose on them. Tax collection in developing countries may present more severe obstacles. While many anglophone African countries have effective taxation structures such as social security, this is not the case in much of francophone Africa, the Middle East nor much of Latin America where social securitytaxes are quite fragmented. In these situations,utilization of the taxation system to collect student debt does not constitutea feasible option. Finally, a few other countries rely on the taxation department as a collector of last resort, as in Canada and the US. Purposeof Support An initiallysurprising feature of Table 1 is the large number of programs that offer support for students' livingexpenses. Of the forty programs for which informationis available,33 offered maintenance support (for lodgings, food etc); of these, nearly half supported livingexpenses solely (European countries, - 10 - Kenya, Ghana), the rest in combinationwith support for tuition expenses (US, Colombia,Hong Kong, Korea, Japan). The programs supportingcombined tuitionand livingexpensesoften attempt to promote student choice between public and private institutions. A student can use support to pay tuition at more expensiveprivate schools or for livingcosts by attending a public university(as in Colombia and the US). The purpose the loan program depends on the structure and policies within the universitysystem. In many countries student living allowancesabsorb a very high proportion of the higher education budget; in Africa, for example,where public universityeducation is typicallyfree, generous student support often accounts for more than half of higher education budgets. In theory, loan programs aimed at living expenses can free up budgets to finance educationalinputs; there remains enormous scope for further movesin this direction. As a consequence,many African governmentswith larger student support budgets (Box2), have either recently implemented new loan programs, or are contemplatingnew ones. Ghana and Malawihave just introduced loans to reduce government expenditures on livingexpenses. Bac k Student Malntenanweand Higher Eduation Budgets Government commitments to funding student livingexpenses have represented a growing share of higher education budgets, often at the expense of institutional budgets. While only representing six percent of recurrent expenditure in Asia, and 14 percent in industrial countries, student support represents around twenty percent of spending in the Middle East and Latin America. Allowancesin East Africa for non-tuition expenses, constituted 35.2 percent of the total expenditure and an alarming 65.6 percent in Western Africa. Allowancesare as high as 62 percent of the average public sector salary in Burkina Faso. A more recent survey conducted reveals that, on average, scholarships account for 37 percent of higher education spending in Africa, but 47 percent in the lowest income countries (World Bank 1988). Share of LivingAllowancesto students in Recurrent Higher Education Budget c. 1980 Region East Africa West Africa Asia LAC EMENA OECD % 35.2 65.6 6.5 17.4 19.1 13.7 Source: Psacharopoulos,Tan, Jimenez 1986. - 11- At issue, in many countries, is whether individualsover the age of 18 should be treated as familydependents. Requiringfamilies to maintain students after a certain age, in some instances,may impose unfair hardship on other familymembers, and also lead to discriminatorypractices,particularly against female children. On the other hand, if the state regards all eighteen years olds as independent, then individualsfrom wealthy families income will be likely to receive support - because while family income could be high, the student's income would probably be low. In only five countries are loans stiDlused to fund tuition fees only (Brazil, Chile, the Philippines, Morocco, and Australia). Indeed, tuition loans have often been essential to the development of fee charging private sectors. In Colombia, Brazil, and Morocco, loans to assist students in private institutions have permitted their expansion, and thus increased the overaDlaccess to higher educationwith lower budgetary demands on the government Australia combined new tuition fees with an option to pay the whole amount as a loan through the tax system. In Chile, large tuition increases were combined with student support programs managed by universities. ILan Vaue and StudentsCoered To understand the financial impact of loan programs it is important to examine the amount that students are receiving,and the number of students receiving loans. While average annual loan values in industrial countries typically range between $1,000 and $5,000 per year, loans have been much smaller in developing countries. With the exception of countries that use loans to finance overseas study, programs normally lend under $500 per student. Those programs lending large amounts on average (Venezuela, Honduras and Barbados) have extensive overseas programs. Barbados is exceptionallyhigh since the country does not have its own university,and students rely almost exclusivelyon foreign training. A relatively higher proportion of students receive loans in industrial countries (between 20 percent and 80 percent) than in developingcountries,where coverage is almost alwaysless than 10 percent of the student population. Exceptionsare found in Kenya and Ghana, where aUlpublic universitystudents receive loans for livingexpenses. High coverage usuallyindicatesa situationwhere loans replaced outright grants. As - 12 a rue of thumb, the higher the coverage, the lower is the average loan amount. When institutionscover less than one percent of the student population, they are able to lend larger amounts; when they expand to 10 percent, the average size dwindles. The limitation on loan organizations in developing countries is their overwhelmingdependence on the government for their budgets: when student repayments are relatively insignificant,total support in a given year is determined by government allocations. In general, loan programs have not been used to support cost recovery for higher education. They have served as support mechanisms for the maintenance of students, at somewhat lower costs than outright grants: the next section will evaluate the extent to which different programs have operated at lower costs than a regime of grants. In addition, we shall illustrate that loan programs have had only marginalimpacts on higher education finance,supporting a few students or providingrelativelylittle support on highlysubsidized terms. Before developingcountries can fashion larger scale programs to enable students to meet the costs of higher tuition fees, important lessons from existing programs need to be learned. These issues are discussed in the next two sections. - 13 - 4 3 The Financial Impact The major purpose of a loan scheme is to enable students to share the financialburdens for tuition and/or maintenance expenseswith the government through paymentsfrom their future earned incomes. The financial efficacy of any loan scheme will depend centrally on the "loan recovery ratio" - the extent to which the loan is repaid in full. One can consider the relationship between what governments lend out to students and what is returned in repayments as an indicator of the loan's efficiency. An inefficientprogram, where the government recovers little of what it lends out implies that the government continues to bear the cost burden of higher education and/or student maintenance expenses. A second issue raised in this section regards what costs are being recovered. Even if loan recovery were complete, with loan expenditures fully repaid, the vast majority of loan programs would only reduce governmentburdens for maintenance expenses, and not tackle the problem of diversifyingthe resource base of higher education institutions. Most public higher education institutionsdo not require students to pay tuition fees that cover a significant portion of educational costs. Therefore, the institutional 'cost recovery ratio" (average loan repayment in relation to unit costs) will be low. Institutional cost recovery cannot be substantial unless tuition fees are high and loans are used to support students payingtuition. Loan Rcwery Ratio The efficiency of student loans, and their relationship to institutional cost recovery are examined in the present section. 24 separate deferred cost recovery programs (from 20 different countries) are analyzed in detail to evaluate their financial impact, in terms of both loan recovery and the cost recovery ratio. Both types of existing loan programs have been evaluated -- mortgage loans, and income contingent loans. While these latter programs have been implemented recently in only three countries and it is thus too 4 In order to evaluate programs, the authors have developed simulation models for mortgage-type loans (included tilted payment schemes), income-contingentloans and graduate taxes. These allow flexible inputs for repayment streams and costs, and can project budgetary requirements. For information on their use, contact the authors. - 14 - early to assess their full impact, it is possible to predict their revenue generating potential by projecting future earnings for university graduates. Loan recovery will depend on three major issues: the amount of hidden interest subsidies on loans; repayment losses due to default; and administration costs.Discussionof these three issues relates to Table 2. Interest Subsidies(Hidden Grantsa: Student loans are subsidizedif they charge an interest rate that is less than normal market rates for borrowing; this subsidy can be considered a 'hidden grant' to students.5 To calculate the size of the hidden grant portion of the loan, we examine the loan account of the individualborrower, assuming regular repayments are made in conformitywith the formal conditions of the loan agreement. Thus we examine the amount and timing of repayment in relation to the loan disbursed to the student. Annex 2 contains a methodologicalnote outliningthe method used to measure this hidden grant. The factors that influence the size of the subsidy are the real interest rate charged and the length of repayment these are shown in Table 2. Column two and three, respectively,list the nominal interest rate charged in the data year and the real interest rate (nominalrates adjusted for average inflation). Column four lists the length of repayment6 for loans (excludinggrace periods); for income contingent loans, this is the length of repayment that is implied using an average income profile for universitygraduates. Column five presents the hidden subsidy to the student as a percentage of the original loan: this figure compares the net present value of the student's repayment account to the present value of the loan disbursement. We note that all of the loan programs in the sample are subsidized,some very highlyso, ranging from 13 percent subsidy in Barbados to 93 percent in Venezuela. In half of the programs examined, subsidyexceeds 50 percent of the loan, indicating that less than half of the real loan value would be recovered if all students repaid on time. Even when real interest rates are positive - as in Barbados and Sweden- the loans are still subsidizedbecause the interest charged is below market rates. S More precisely, even if the government were to provide loans at normal market interest rates, the government is still providinga subsidybecause loans for education investment have a higher inherent risk. 6 In some instances, repayment length is a function of borrowing length. The analysisassumes four years of borrowing. - 15 - Table 2. HiddenSubsidiesand Government on SelectedStudentLoanPrograms Country (1) Nominal Real Interest Interest rate rate (2) (3) Maximum or Projected Repayment Period (4) Hidden Grant to students per cent of loan (5) Government loss with default (6) MORTGAGE LOANS Colombia I 11.02 a Colombia II 24.02 Sweden 4.3% Indonesia 6.02 USA (GSL) 8.02 a Hong Kong 0.02 UK 6.02 Norway 11.52 a Denmark 8.0% a Finland 6.5% a Brazil I 15.02 Brazil II 318.02 Jamaica I 6.02 Jamaica II 12.02 Barbados 8.02 Kenya 2.0% Quebec 10.O2 a Chile varies Japan 0.02 Venezuela 4.0% Honduras 12.0% -10.62 3.02 b -3.0% -2.32 b 3.82 b -6.32 b 0.02 5.6X 1.62 -0.6% -35.0% b -14.92 -10.72 -5.6Z 4.12 -6.9% 5.22 1.02 b -1.4% -232 32 8 5 20 10 10 5 7 20 10 10 5 8 9 9 12 10 10 10 20 20 8 732 292 612 572 292 43Z 262 332 522 45Z 912 622 742 562 132 702 312 482 50% 932 51% 76Z c 382 c 622 612 412 43Z 302 332 562 462 94% 652 842 c 62% c 182 94Z c 312 69% c 51% 982 53% INCOME CONTINGENTLOANS Australia varies Sweden varies 0.0% 1.0% 17 10 48% 282 52% 302 es Government loss with default and administr- Year Estimates (7) (8) 872 472 702 712 532 47% 412 482 62% 522 982 712 92% 702 332 103% 372 822 60% 1082 732 1978 1985 1988 1985 1986 1985 1989 1986 1986 1986 1983 1989 1987 1988 1988 1989 1989 1989 1987 1991 1991 572 33% Administrative 22 Administrative21 Administrative12 Default 102, Adm 2% Administrative22 Administrative22 Default 52, Adm 12 Administrative12 Administrative12 Default 22, Adm 12 Default 302, Adm 22 Default 102, Adm 22 Administrative22 Default 202, Adm 22 Default 52, Adm 22 Administrative22 Administrative12 Administrative22 Administrative12 Administrative32 Administrative52 1990 Evasion 32, adm 0.52 1990 Evasion 32, adm 0.52 General notes: All subsidy calculationsuse a real opportunitycost of capital accordingto the governmentrate of borrowingor estimatesused by the World Bank. Loans are assumed to be paid in equal installmentsover a four year period, adjusted in size each year to keep up with inflation.Given the availabilityof relevantdata, Swedish income contingentcalculationis based on Australia'sage earning profile information; (1) Countries with I and II refer to situationswhere the loan program underwent reform. (2) Nominal interestrate refersonly to the rate duringrepayment.'a' refer to loans that use a different rate during the disbursementand grace period. 'b' denotes those programs with interestrates which are indexed. (3) Real interest rates use PurchasingPower Parity formula,where inflationis based on the average of the 1980-1988 period as reported in the World DevelopmentReport, except in instancesnoted where a 5 year average of inflationwas calculatedfrom the data date. (4) The repaymentlength is the maximum prescribedin the loan, exceptfor the two incomecontingentloans where it is the repaymentlengththat is impliedby the average incomeprofileof a graduate.This does not include grace periods. (5) The hidden grant percentage is calculatedas a discountedcash flow of the student's account, and thereforeexcludes default and administrativecosts. (6) The governmentloss due to default subtractsthe 2 of default from each year of the repaymentstream. 'c' denotes where these figures have been estimated. For Colombia, Jamaica,Chile and Kenya the figure used is loans in arrears. (7) The loss with default and administrativecosts subtracts an annual administrativecost related to outstandingdebt each year. (8) Year is date from which loan informationwas collected,and from which inflationcalculationswere made. - 16 - Default and Evasion: The loan subsidymeasures the percentage loss to the governmentfrom each loan that is repaid according to the established loan conditions;however, it fails to reveal the overall loss to the government from the loan program. Lending agencies receive back less than is indicated by the hidden student subsidy because not all students meet their repayment obligations and the administration of the program is not costless. The experience with default has been mixed.7 Table 3 indicates that default and evasion can constitute a more severe problem than subsidies.For example, non-repayment was as high as 81 percent in Kenya. Thus, even with theoretically tight repayment terms, little revenue comes back. In other instances, default is less of a problem (Sweden, Hong Kong, Israel). While default rates are lower among developed countries, particularlywhen they are small and have populations which are easy to track, it has yet to be demonstrated that default can be minimizedeffectivelyin large developingcountries, without extensive administrative costs. When default losses are factored into the return to the government - the method is described in Annex 2 - measured losses from the loan program are enhanced (Table 2, Column six). In the original Brazilian scheme, Venezuela and Kenya, losses increased to over 90 percent of the value of the loan. 7 Measuring default is difficultfor definitionalreasons.Some countries measure arrears rather than default. A more serious problem is whether default is measured as a percent of the number of loans that do not pay (as used in our calculations),or the value of outstanding debt that is not being repaid in relation to the total outstanding debt. West (1988),suggests that if the latter were used as a measurement, default rates would be even higher. The figures reported above (and those used in subsidy calculations)are therefore conservative estimates of default and non-payment. Losses therefore tend to be underestimated. For loans that are collected through the tax system, as in Australia, default rates have not yet been measured, but it can realisticallybe assumed that the default rate will be similar to the evasion rate on taxes generally. - 17 - Table 3. Non Repaymentof student oans as percentageof total kamn Selected Countrie Country Jamaica Sweden Ontario, Canada Colombia Chile USA Denmark Israel Japan Kenya Hong Kong Note: Non Repayment as percent of loans Year 38.8 1.0 0.5 12.0 40.0 17.0 < 10.0 < 2.0 2.3 81.0 < 1.0 1985 1988 1988 1985 1989 1987 1987 c. 1980 1985 1987 1988 Country Definition Arrears Default Default Arrears Arrears Default Default Default Default Arrears Default Each country has different definitions of non-repayment. Default means that countries have written off the loan, while some countries list payments as in arrears, when in reality students have defaulted. 1989was the first year of repayment for the Chilean loan program. Sources: Chile-Schiefelbein 1990; Jamaica-World Bank data; Sweden-Woodhall (1989), Canada-QuebecStudentFinancialAssistanceProgram (1990),Colombia-Woodhall (1987a), USA-Department of Education (1990); Denmark-OECD (1989); IsraelWoodhall (1983); Japan-OECD (1989); Kenya-World Bank data; Hong KongWoodhall (1990). Administrative Costs: To establish the true cost of a deferred cost recovery program, administrative costs, too, must be taken into account These costs generally fall into initial processing costs, overall maintenance costs and collection costs. In developing countries, tracking mobile students can be extremely difficult, making administrative costs higher. The small average size of loans makes them proportionatelymore costly. No detailed comparativestudy of costsof loan programs has been conducted,and data are mostly limited to those from developingcountries. The most efficientlyrun operations -- in Sweden, Hong Kong and Canada -- costs report ranging between a half and one percent of outstanding debt each year. - 18 - (Woodhall 1983,Woodhall 1990(b), Quebec Student Financial Assistance Program 1990). In Latin America, the overall cost of managing a loan has been put at between 12-23percent of the value of the loan (Woodhali 1983). Annual reports from Latin American loan organizationsconfirm these estimates, and suggest that the institutionsinvestingin recovery are spendingeven more, as high as 30 percent in Honduras. In calculatingthe net return of loan programs, when costs are unknown,we assume an annual cost of only two percent of outstanding debt each year; when discounted, this implies an overall cost of approximately 10 percent of total loan value, and thus is likely to understate the full cost of a loan program. Programs that rely on commercial banks or taxation departments have been far less costly to administer. Operating costs for commercialbanks tend to be much smaller than autonomous loan bodies. In Brazil, operating costs for the commercialbanks are approximately 10 percent of the total loan value (World Bank data). Administrativecosts for taxation collection may be even less expensive, due to large economies of scale. The Honduran Ministryof Finance reports payingthe Central Bank a service charge of between one percent and two percent of money recovered. Overall losses on loan programs, taking account of administrationcosts, in addition to interest subsidies and non-repayment, are shown in Column seven of Table 2; given the low assumed value of administration costs, these results should be regarded as conservative estimates of what the true net loss to government is likely to be. The most efficient programs are in Sweden and Quebec, whichboth recover well over 60 percent of the loan's value (ie. losses of 33 and 37 percent respectively), while the programs in Venezuela and Kenya actuaLlycost more than would outright grants to students. - 19 - Box 3: Equity and Risk Aversion The equity considerations of student loans are no less important than financialefficiency. While loans can be an important tool to assist people meet their educational costs, poorer individualsare less likely to borrow than middle class students. The problem of 'risk aversion" has been confirmed by empirical studies. Sweden's former mortgage-type loan was not found to promote access among lower income groups (Reuterberg and Svennson1991). Other studies confirm this findingin industrial countries. Borrowing to finance higher education is unlike borrowing to purchase a house because, when people borrow to finance a degree they are not completelycertain what they are purchasing (especially if their parents did not attend higher education); there is a risk of failing the course; and not all degrees lead to high private returns. That is, while mean incomes may show a high rate of return, in reality, incomes can vary considerably. In addition, while private returns are likely to be high for wealthier students, they are less likely to be high for poorer students who lack familyconnections (Barr 1990). To minimize the risk to low income students, most governments subsidize loans. But large subsidies undermine the purpose of having the loan in the first place. Governments can minimize more effectively the problem of risk aversion by limiting the repayment burden in any given year by linking payments to income, imposingpayment ceilings,or providingexemptions if income falls. Such measures can minimize the risk to low income students and encourage them to borrow to finance their studies. Lo7 in ReRtiDnto University Costs One of the central theoretical and practical rationales for loan programs is to diversify (broaden) sources of fundingfor higher education. As noted, however, most loans are used not for institutional funds, but to limit government burden for student maintenance. Table 4 examines the experience in seven countries where a loan scheme is coupled with fees in public universitiesto help cushion the impact of cost recovery. - 20 - Table 4 Eflccdve Cost Reovery from Loan Recpien1s at Publc Universitis (as a fraction of unit insuctional costs) (1) Country Chile Colombia Indonesia Australia Canada (Quebec) Japan USA -; Unit Instructional Costs 100 100 100 100 100 100 100 (2) Average Tuition From NonLoan Students 35 4 25 18* 22 9 24 (3) Implied Cost Recovery Ratio For Loan Recipients 5 4 7 9 14 4 11 (4) Average Loan Size in Relation to fees Greater Greater Greater Equal Greater Greater Greater Fees in Australia are nominally set at 21 percent of recurrent costs, but students receive a 15 percent discount on fees if they pay them up-front. Sources for unit cost and tuition data: Chile-Brunner (1990); Colombia-Gomez Buendia (1984); IndonesiaWorld Bank data; Australia-Hope and Miller (1989); Canada-Quebec Student Financial Assistance Program (1990); US Department of Education (1990); Others-OECD (1990). Unit instructional expenditures are estimated where only total unit cost is known. Estimates assume 30 percent of expenditures for research. Table 4 compares present value contributions from students paying direct fees from their own funds, and those paying with government sponsored student loans, to higher education instructional costs. Column three shows the proportion of teaching expenditures that governments effectively recover from students who receive a loan. In these seven countries, with some of the highest public sector cost recovery in the world, governments recover only between four percent (Colombia) and 14 percent (Quebec) of instructional costs from loan recipients. Actual cost recovery, however, is even lower because in every instance, except Australia, the average size of loan is larger than tuition costs. So governments are actually spending large amounts of money on student support in addition to institutional subsidies. In Australia, the loan is fixed at the level of fees charged. It is easiest, therefore, in this instance to understand the relationship between immediate and deferred cost payments. Each student has an - 21 option: to pay up front or to pay in the form of a loan. Tuition fees are set at 21 percent cost recovery8 . If the student decides to repay in the form of an income contingent loan, the government recovers, on average, about 43 percent of the loan value. The effectivecost recovery therefore represents only nine percent of unit costs. Overall, effective cost recovery is extremelylow. This is so for two reasons. First, fee levels generaly do not represent significant portions of the costs of higher education. The low initial cost recovery is compounded by loan programs which require further government expenditure just to recover costs in a deferred form. If loans are to be used to foster cost recovery, significant fee levels must be established. To date, loans have been operating only at the margins of cost recovery. 8 There is a 15 percent discount if the student pays fees directly. The Australian scheme allows this discount in recognition of the hidden subsidyon the loan. However,as calculated for average income earners, this discount is well below the loan subsidy. Israel allowedstudents eligible for loans the option of a 35 percent tuition discount if they turned down the loan: this calculation was based on the actual subsidy in the loan (Woodhall 1983). -22 - Box 4: Brazlh Euabllhlng the Cosasof a Loan Progam Brazil has contemplated reform of their student loan program, which primarily is intended to promote access among low income students to private institutions. The question arises how much wil the government have to lay out each year, and how much will be returned from repayments. The loan program will be loss making. even when revenues are fullybuilt up. The loss depends on the subsidy and the default leveL The graph ilustrates the projected relationship between income and expenditures for a program that covers 25 percent of the Brazilian student population, with loans averaging S500 per year. It also assumes that the higher education system (and the loan coverage) is expandingat three percent per year. Thus, expenditures too expand in real terms by three percent per year. ProjectedExpendituresand Revenuesfor ProposedBrazilianLoan Program 400 US $ millions 350 _.....-. 300 -.. ;;; 250 . .......................................... 200 _--- ....................................... 150 I I I 10 1s 20 50-_.-v..-.-..-. 0 5 25 Year - Expenditure -I- Revenues P1 In conducting these forecasts, it is important to consider that average loan value per student must be maintained in real, rather than nominal, terms. Revenues build up slowly,and reach a maximum in relation to expenditures after approxinately nine years. - 23 4 Improv Performance Cost sharing for higher education is important in many countries as universities need to broaden their financial base to improve the efficiency,equity and qualityof education. Many students,however, cannot afford to pay the up-front costs of their education, or even a sizable proportion of them; deferred cost recovery programs fill this void by allowing students to pay by tapping their future earnings. While this principle is well established, we have seen that past experience with loan programs has been disappointing, particularly from the viewpoint of financial efficiency. Yet, it should not be concluded from disappointing results of past experience that loans programs should be abandoned. On the contrary, we argue that reform and improvement in several key elements of program design as necessaryconditionsfor well functioningloan programs. In this section, we outline three major issues that require attention for programs to work well: effective targeting, reducing subsidies while limitingdebt burdens, and minimizingevasion. Targeting Loan Support Many student loan programs are open to all students, regardless of need or ability. In Africa, loan programs in Ghana and Kenya allow all students to borrow money for their maintenance expenses in public institutions. Recently implemented schemes in the United Kingdom and in Australia also provide students with access to credit, regardless of income. But open access can be expensive to governments, particularlyif support is subsidized. The primary advantage of open access to loan support is that no one will be missed. The chief disadvantage is that usually fewer funds are available for needier students, and limited available support may often go to benefit those who can afford to pay. A successfulsupport program needs to be targeted effectively,to those who are deemed most deserving of support. Without effective targeting,growingstudent numbers in the future, as well as less-thanfull loan recovery,will result in increasing,and unsustainable, pressures on limited loan funds. Given that loan funds are subsidized and most likely will continue to be so (though we argue for much lower subsidy levels), - 24 - targeting will facilitate the task of limitingthe extent of loans subsidization. Other reasons, in addition, may underline the need for deliberate loan targeting.The finding(Table 1) that in many countries the percentage of students receivingloans is not high, indicates that rationing mechanismsare at work. But with open access, these will not have been established deliberately by government; self-selectioninto the loan scheme amongst well-to-dostudents, together with a disinclinationof risk-adversepoorer students to enroll, mayhave introduced implicitrationing criteria that may not be in conformitywith overall policy objectives. Several targeting criteria are presented in the discussionthat follows,the most effective - and acceptable - being targeting according to need and to ability. Although discussed separately, many programs employ more than one of the criteria discussed below. Means testine: Access to loans may be limited to those students whose family or personal income falls belowa threshold value. Means tests may take a variety of forms. In the US Stafford/Guaranteed Student Loan Program (GSLP), complete support is available to students below the income threshold.9 Alternatively,loan amounts can vary according to the difference between an individual's available resources and the costs of a given course of study, as in Canada, Barbados, Brazil and Sweden. The calculationof need can be adjusted for the number of familydependents; or parental income can be completelyignored and the student's assets and income assessed independently as in Sweden, the Netherlands and Norway. In these countries, students over 19 years of age are treated as financiallyindependent of their family. Almost all students are eligible for support. This stipulation, it is believed, has been significant in ensuring access for women. But this same requirement in many developingcountries has enabled students from wealthier families to benefit enormouslyfrom student support, simplybecause students at age 19 are unlikelyto have their own sources of income. In developingcountries means testing can be extremely difficult, particularlywhere income 9 In the late 1970's, the SLP,whichhad previouslybeen restricted to needier students,was made available to all students in accredited higher education institutions. In the 1980's, the government felt that the costs of the program had become too high and reintroduced targeting: the loan program was scaled back, with access based on their need. - 25 - tax systems are not in place, where the extended family is important and where the non-market sector is sizeable. Experiences in Colombia and Brazil indicate that programs could be better targeted if stronger restrictions on income ceilings were imposed. In Colombia, funds were allocated to students whose families fell below an income threshold. But the threshold used was relativelyhigh. In Brazil, need is prioritized -- that is students are ranked. The government disburses all the funds it has in a given year according to the ranking and does not try to conserve funds (Vahl 1990). The major problems in assessingfinancial needs are presented by McMahon (1988) together with a suggested method for computing family ability to pay, illustrated for Indonesia. One effective technique for targeting funds is to allow higher education institutionsto manage a pool of loan money, and to allocate funds to neediest students. Being in closer contact with student population may give these institutions an advantage in evaluating needs. Chinese, Chilean and Indonesian universitieshave been fairlysuccessfulin identifyingneedy students. The targeting of student support in Chile illustrates steps that can be taken to circumvent limitations on reported income. In addition to student and family income, students must submit information on their parents' occupations and education levels, family assets and place of residence. This information is verified through spot checks and students are barred access if they provide false information. - 26 - Box 5: Means Testingat the Universityof the Philippines In 1989, the University of the Philippines combined sharp tuition increases with increased financial assistance to needy students. The "Socialized Tuition and Financial Assistance Program" (STFAP) has both increased overall institutional revenues and support for needy and academicallyqualified students. The University grants two types of financial assistance. The first level of support are tuition discountswhich are awarded solely on the basis of need. The second are maintenance grants which are given for both need and academic merit. To assess financial need, the University has had to move beyond income tax returns, which often understate true ability to pay. Around 40 percent of the 15,000 students who apply for financial aid receive less assistance than they would have if means testing were based on income tax returns alone. STFAP applicants must complete a twelve-page questionnaire which are encoded for computer processing. The questionnaire asks about family assets, parental occupation and education levels, and location of residence. The questionnaire in itself does not stop dishonest applicants, but home visits and harsh disciplinary actions are believed to make applicants answer questions more truthfully. Home visits verify the accuracy of most reports. Several students have been expelled from the university for givingfalse information. Abilitycriteria: Access to support can be based on student performance, either at secondary schoolor university.Abilityrestrictionsgivestudents a strong performance incentive,while also rewarding those who are most likely to benefit from higher education. Restricting eligibilityin this way can help to avoid providing subsidies to students that are most likely to repeat or drop out. In Indonesia, students were only eligible for loans as they approached graduation, after they had proven their academic ability. It may also be useful to define publicly the academic standards that must be achieved to attain access to loan funds. In Venezuela and Honduras, a student failing to receive minimum grades, will lose access to loan support and must begin repayment of loans immediately.In Colombia,access to loans is determined partially by results on the national secondary school examination. There is concern, however, that the use of ability criteria could result in the selection of wealthier students with access to better educational facilities. - 27 - Box 6: USA. QualityRestrictionsand Effidency The Stafford Loan Program (formerly the Guaranteed Student Loan Program) has been the principal government mechanism for promoting access to higher education. Under the program's initial terms, all post-secondary students at accredited institutions satisfying need criteria would have access to subsidized loan funds from private commercial banks. Loan funds could be used at public or private institutions, includingvocationally oriented proprietary institutions.The government would act as the loan guarantor and pay a subsidy to the banks. Guaranteeing widespread access has undoubtedlyhelped disadvantagedstudents, but the lack of quality standards (either among institutions or students) has led to an increasingly costly program. During the 1980's, approximately 17 percent of borrowingstudents failed to repay their debts. Default stems from the high risk involved guaranteeing access to all low income students, since the government makes little effort to control the quality of the students receiving support. The subsidized loans have been available to students at institutions, even if they accepted students without a secondary school diploma. Therefore, the subsidy in the loan has encouraged less qualified students to enter poor quality schools. Default is by far the highest in proprietary and two-year institutions. In 1989, the rate was 33 percent among students at proprietary schoolswhile only seven percent among students attending four year institutions. Reform of the program requires the government to reconcile increasingquality standards with the need to preserve access for the neediest students. To resolve this problem, quality control focuses on institutions rather than placing restrictions on student achievement level. Institutions that accept students without a secondary school degree or its equivalent will no longer be eligible for loan funds. While this will not correct all the problems, it will certainly reduce the rate of default and encourage institutionsto raise their entrance standards. Duration: The period for whichstudent support is available can affect student flows,and thus the efficiencyof the education system. In many higher education systems, repetition is fostered by open ended availabilityof support. Limiting loans to the prescribed length of a course can improve student performance, and also conserve funds. This consideration has been important in program reforms in Australia, the Netherlands and Brazil, where support has been limited to the officialduration of study (sometimeswith one year extensions). A potentially negative consequence of this type of restriction is that it can penalize students who work and study at the same time, and are therefore likely to take longer to complete their courses. An interesting innovation to address this problem has been implemented recently in Denmark: the "clip card" approach allowsstudents to draw upon a fixed total loan on a month by month basis as they choose. That is, although aid is limited to the equivalent of four years, students can spread the aid over the expected period of study as they choose. This flexibilityis intended both to improve incentives to finishon time, and to mitigate - 28 - problems for students that need to take longer to complete their studies. Box 7: Grantsin addition to loans The poorest students will not be able to gain access based on loans alone: for them, foregone earnings are too high. Tilak (1985) illustrates the importance of opportunity costs in determining access for India. Fees generally make up only a small percentage of total private costs for higher education, and changes in effective fee levels, given loans, have a relatively minor impact on access for most people. On average, tuition fees represent about 19 percent of total private expenditure (including foregone earnings) for university students and about 13 percent for college students. Access for groups at lower income levels, therefore, is much more a function of opportunity costs than of fees. Therefore, without sufficient support, they will not attend. Many countries provide grants rather than loans for the poorest students. The US has a system of Pell grants in addition to loans. Similarly,in Canada, needy students receive a grant (onlyafter they have received a loan). In Colombia,grants rather than loans are given to a few students. Reducingsubsidieswhile limitingdebt burdens The manner in which interest charges are assessed is central to the balance between efficient cash flows for the government and equitable debt burdens to students. The lower the interest rate, the larger the subsidy on loans. But higher interest rates increase debt burden and the likelihood of default. Therefore, in designingrepayment plans that limit the subsidy element in the loan, it is extremely important to examine the likely range of incomes that graduates will be earning. A successful loan program will not simply raise interest rates, but will redesign the repayment format so that graduates will be in a position to pay. Fixed Real or Floating Interest Rates: A common solution to the open ended subsidyproblem is to tie interest rates to an indicator of inflationor commercial lendingrates. By doing so, the level of subsidy remains fairly constant, and it is easier to project the financial implicationsof a loan program. After the poor experience of charging fixed nominal interest rates, some countries such as the UK and Australia, now adjust outstanding debt for inflation. Alternatively, in Sweden, interest rates move with the government lending rate while in Barbados interest rates are adjusted according to the interest on government bond issues. Controlling the level of subsidy on a loan implies increasing interest charges both during the period of study and the repayment period. In the US, Quebec and Norway, for instance, no interest is assessed during the study and grace period. The resulting subsidy during this short period, however, can be quite significant. - 29 Repayment Period: Equally important for minimizingthe subsidy is ensuring a limited repayment length. Longer repayment periods are an effective guard against default and are less burdensome to the student, but lead to larger hidden subsidies. In Sweden,the old student loan program allowedstudents until their 51st birthday to repay their loans, and the result was minimal defaults. The average interest rate subsidies, however, have been calculated at approximately53 percent (Woodhall 1989). Reducing SubsidiesImplies Linking Repayments to Income: One of the major problems with traditional mortgage loans is that, even when subsidized,they impose heavy repayment burdens in the first years after graduation. Typically,a graduate's earnings are low immediatelyafter graduation and rise quickly. Inflation implies that the real value of equal nominal payments decreases over time. Students therefore have the largest debt burdens when they are earning relativelylittle. In Venezuela, where interest rates for student loans are well below the rate of inflation, the real value of the first monthlypayment is more than 250 times the real value of the last payment So although the loan is heavily subsidized, the student might default because the initial payment represents an unmanageable proportion of income. Unless payment terms are restructured, hon-subsidizedstudent loans are likely to lead to payment plans that require excessive portions of a graduate's income in the first years of repayment. This problem will be particularly acute in inflationaryenvironments, since the real value of the first payment will be so much greater than the last payment To circumvent this problem, graduated or income-contingent payment plans should be designed so that payments are related to income. - 30 - Box & Sweden:Using FlnancialEfficency to ImproveEquily The reforms in Sweden's student support system, implemented in 1989, sought to increase participation rates in higher education among low income groups,while increasing the financial efficiency of the program. The reforms followed two basic strategies: (i) increasing the availability of support funds for poorer students by converting the hidden subsidy in the old loan program into open grants; and (ii) minimizingthe risk of borrowingby linkingrepayments to income. The original student support program, begun in 1964, assisted virtually all students. A grant/loan allowance was calculated at 140 percent of the government's social security subsistence leveL The program was widely regarded as one of the most successful student support schemes, particularly because of its low administrativecost (one percent), and low default rate (one percent). In the 1980's, the program was criticized for poor financial performance and more importantly for its failure to raise the higher education participation rate of students from working class backgrounds. To redress these problems, the new loan/grant mix essentially allows more money to be channeled to student support by cutting back on hidden grants (Morris 1990). The new support package has raised both the total support and the percentage which is awarded as an outright grant. The rest is given as a loan to be repaid on an income contingent basis of four percent of income. The same deferment clauses are still in effect. But the loan carries with it an interest charge that is half the government lending rate. This yields a positive real interest rate of about one percent, and has therefore feduced the hidden subsidy to about 20 percent. Thus, with the savings from eliminating the hidden grant, outright grants have been expanded. This should be important in assistinglow income students in attending higher education, as their effective risk is now minimized,and the overall availability of credit and open support has been increased. Income-Contingent payments circumvent this problem, but they are not always a feasible option (because of the absence of accurate income reporting).An alternative is to designscheduled repayments so that they approximate the growingtrend in expected incomes. This implies tilting repayment schedules so initial payments are smaller than later ones. Such graduated repayment plans could effectivelyminimize burdens on students after they graduate while eliminatingsubsidies. fmizingEvasion The other major problem that has plagued the finances of student loan programs has been the failure of many students to repay their debt. Default can be dividedinto two problems: students whocannot pay, and students who evade payment. Properly defmed repayment plans will help students that do not earn large salaries after graduating.In addition, the followingsteps have been shown to improve performance. - 31 Income contingent deferrals: These are an essential minimum step in avoiding situations where students who cannot pay are unnecessarilyclassifiedas in default. The lowestloan default rates have been in Sweden, Hong Kong and Quebec, in which loans have low income contingent deferment clause. In these loans, when a graduate's income falls below a threshold level, students are exempted from payment while still accruing intcrest charges. The new student loan program in the UK also has an income deferral clause. In all of these programs, students must submit proof that their incomes indeed have fallen below the threshold level before deferral is granted. Box 9: Hondura& Reducing Default can be Expensiv and Deter Low Inowme Students The Honduran loan program, Educredito, has provided approximately 300 students per year with loans to study both within Honduras and abroad. In its twenty year history, the program has encountered severe problems with default. In 1990,the government moved to privatize Educredito. As a consequence, Educredito has taken steps to eliminate these losses. In its earliest years, when the program was small, students were followedcloselyto ensure repayment. However, as the program grew, both in numbers and loan amounts, many students succeeded in avoiding paying their debts. Overall, the non-payment rate was about 90 percent of the loan portfolio. Due to concerted efforts in the last three years, Educredito is recovering loans from almost all students, and of the latest cohorts only about two percent fail to pay. This success has not been costless, and could have important, although undocumented, equity implications. To boost recovery, the loan program now requires either two guarantors or collateral on every student loan. In the event of non-payment,Educredito reserves the right to confscate property or seize assets of the guarantor. Before taking such radical action, the loan organization uses private agencies to locate students that are not paying nor responding to contact. If after locating students and demanding payment the debtor still refuses to pay, Educredito utilizes a private agency for collection. The costs of these operations are high. In 1990, nearly 30 percent of the operating budget went to administrative costs, and a substantial portion of these paid for private agencies. In the future these costs should fall as attitudes change towards non-payment A more worrying problem, however, has been that the steps taken to ensure payment, particularly the requirement of guarantors and/or collateral, have deterred low income students from applyingfor credit These students have great difficultyin securing guarantors given Educredito's determination to secure repayment. This problem has yet to be resolved. Ensuringlncentives for Financial Agent:In many instances,guaranteed publicbudgetsfor loan programs undermine incentives for institutionsto collecL The autonomousloan bodies in Latin America often prefer to rely on public funds to provide new loans rather than stepping up efforts to secure repayments. In - 32 Honduras, moves to privatize the loan institution,Educredito, have led to needed investments in the recovery apparatus, and have successfullyreduced default from 90 percent to under 10 percent. As noted, experience with private and public banking systemshas been such that sometimes it may be cheaper for a bank to collect from the government rather than the debtor. In the US and Indonesia, loans were guaranteed to nearly 100 percent of their value. However, recent steps in the US have sought to minimize the extent of the government guarantee and discount its value -- enough so that the institution has the incentive to collect. Requirement of a Guarantor: A controversial policyto ensure repayment is requiring a wage earning guarantor who agrees to pay the loan if the student does or can not. This type of arrangement has been implemented in most of Latin America. In Ghana, each borrowingstudent must have two guarantors, who are wage eamers (and thus trackable by the government). The result of such an effective guarantee should be that default willbe negligible. Requiring a guarantor, however, can have negative consequencesthat defeat the purpose of a credit scheme. It might very well be the case that precisely those people who most need support will be the least able to provide guarantors (see Box 9). Direct Payroll Deductions: Increasingly,loan schemes authorize companies to deduct wages from the salaries of debtors in arrears. In some instances this may be difficult if legal restrictions prohibit deducting salaries for loan repayments. This strategy also requires the lending agent to know where the debtor is. The approach, has been implemented in Jamaica, Honduras and Colombia and seems to be effective. ChoosingAppropriate CollectionInstitutions: While autonomousbodiesand universitiesmay possess comparative advantages in selecting students and targeting support, it is less clear that they have the capacity to collect repayments. But banks and tax systemsoften have the necessaryinfrastructure that they lack. The former can operate efficientlyat collectingmoney so long as policy does not undermine their incentives to do so. In Venezuela, while exact figures are not available, the student loan program operated by the national savingsbank (BANAP) does not suffer from problems of default, while the public collection agencies - 33 have much greater problems. Besides relying on banks, some institutionsare utilizingthird party collection agencies. Recovery in Honduras has been dramaticallyraised by usingprivate agencies to locate students, and also agencies to collect money. Default, as a consequence, has fallen below five percent (Box 9). Colombia passes on additional charges for such services to the student and students now are reluctant to default on their loans. Insurance fees: Studentscan be required to contribute an up-front insurance fee on their loan. Currently, insurance for disability or death is required in BraziL but it may be possible to extend the idea further to a general default insurance fund as is being discussedin France. Bar Access to Further Credit: When borrowers in Brazil realized that they were ineligiblefor car loans as a result of failure to repay student loans, they quicklybegan repaying. Simple measures such as barring access to further credit can ensure that those who are able to pay (those who can afford a car or a house) will do so. Maintain ContinuousContact: It may be helpfulto maintain contact with students at periodic intervals while they are borrowing to remind them of their loan obligation. The French government is proposing a student loan program that willrequire students to make small payments each year even while they are borrowing; if the student fails to make any payment, the loan will be cut off. Conclusions The goal of most deferred cost recovery programs is to secure student contributionsto higher education costs. If subsidies and losses are too high, the program is essentially meaningless and should be abandoned. Policymakers should insure that a loan program is for the most part self financing.Three basic strategies can significantlyimprove the performance of current loan programs. To utilize resources efficiently,a deferred cost recovery program must be properly targeted - 34 to students who need and can benefit from support. Efficient targeting of subsidies to needy and qualified students will improve the financialviabilityof a program and increase the availabilityof support for those who need it. Second, hidden subsidies can be eliminated by charging positive real interest rates, but this will have to be combined with repayment plans that make sense in relation to graduate incomes. Where income information is accurate an income contingent repayment plan would be most appropriate. A similar option is to allow deferrals of paymnenton the basis of low income,and therefore place a ceilingon repayment burden. This can be achieved through (i) deferral clauses, (ii) maximumrepayments as a percentage of income (say, no repayment should exceed 10 percent of monthly income), or (iii) payments as a fixed percentage of income. If these collection methods are administrativelydifficult,then adjusting scheduled payments to the likely pattern of graduate incomes (i.e. graduating the payments) would improve collectionas well. Third, a strong strategy to deal with default must be in place, beginningwith the removal of institutional disincentivesto collect On the one hand, repayment terms should allow graduates whose income has fallen to defer payments, and therefore limit their payment obligation in any year. On the other hand, recovery agents should take strong measures against borrowers who are evading payment. Insuring that the most suitable type of institutionsare collectingis an important step. Barring access to other credit, deducting from payrolls, usingtax informationand strengtheningcollectingagents are important steps availableto reduce defaulL - 35 - 5 Aola Scenarlo' Our discussionin the preceding section suggested a range of reforms, based on "best-practice" measures currently in place, to improve the fnancial performance of existingloans schemes. The range of deferred payment options, however,extends beyond the formal loan schemesdiscussedthus far. A more radical strategy than the reform of the traditional loan schemewould be to implementalternative (or additional) forms of deferred cost recovery, a consideration particularly relevant to countries that are weighing the merits of introducing for the first time some form of delayed cost recovery. In this section we discuss the efficacyand advantages and disadvantagesof three such schemes: equity finance (the "graduate tax"), employer taxes, and national service. GiraduateTax The idea behind a graduate tax is straightforward.In subsidizing higher education, the state assumes a share in financing the creation of human capital. This produces a future stream of benefits that accrue mainly to the graduates in the form of enhanced earnings. By participating in the finance of higher education, the government essentiallyacquires an equity share in the human capital created and is thus entitled to a dividend from the ensuing income benefits. In the case of a graduate tax this dividend can take the form of a percentage tax (say, one to thTee percent) on graduates' income over their working lives. The term graduate tax is somewhat misleadingsince it legitimatelyapplies to individualswho attend higher education but fail to graduate. The tax is a form of user charge, and therefore could accumulate for each year that the student attends university. Percentage tax rates could also be made to vary with income level, while graduates with low income (low incomes being defined perhaps in relation to median incomes) would be exempt from the tax. Thus the government assumes the risks of human capital investment (depending on the size of the subsidy),which are spread over the student cohort; high-earninggraduates willprove to have been good risks, while those with low incomes or high unemployment,poor ones. - 36 - BDu10: Equity Fianr at Yale In 1972,Yale University attempted to implement a novel equity finance scheme. The University offered students the option of deferring a fixed portion of their annual tuition payments in exchange for payments of 0.4 per cent of their annual income, for each $1000 deferred. Graduates who opted for this program were to repay as a cohort, not as individuals.Payments would terminate when the cohort's repayment was complete. Thus some individualswould repay less than tuition deferral, others more; there was an exemption for individualswhose payments had reached 1.5 times their original debt However, the program failed to attract a sufficient number of students and was abandoned after the first year. A central problem was that existingstudent loan programs offered more generous (Le. highlysubsidized) terms ( Hope and Miller 1988). Indeed, the failure of such a program might be expected in the presence of a student loan scheme. A potentially high wage earner would shun such equity finance arrangements. He would always pay less under a loan scheme (whether subsidized or not) than in an equity finance program: in the latter case, his total payments would exceed the average, whereas in the former total repayments are equal for all participants. The absence of potentially high wage earners from the equity finance scheme would necessarilyraise payments for those who remained. This, in turn, would discourage their participation in a scheme that had become financiallyless attractive. First suggested by Milton Friedman this equity finance approach has been urged frequently 1 0 . It has not as yet been implemented anywhere, although by other educationeconomists and policy advocates there was an interesting, but unsuccessful,attempt to introduce an equity finance scheme at Yale University in the early 1970's (see Box 10). The feasibilityof a graduate tax for the UK is discussed in Glennerster et al (1968). A graduate tax of the type discussedhere in many ways resembles the income-contingentloan scheme recently introducedin Australia (which has been labeledas a graduate tax). However,the two schemes are quite different. While in the Australian scheme, income related loan repayments are made through the income tax system, this is done for administrativeconvenienceonly. In principle,repayments could be effected through other collection institutions, though there are clear advantages in using the taxation system for collection. The major differences between the two schemes are outlined in Table 5, which also offers comparable informationfor the traditional mortgage loan scheme. The motivationbehind both loan and equity 10Friedman 1962, Blaug 1973,Barnes and Barr 1989. - 37 - finance schemes is, ultimately, cost recovery,with the beneficiaries of higher education forgoingpart of the return on education that they would otherwise capture. However, they are conceptuallydistinct. In the case of loans, there is a creditor-borrower relationship between the government and graduate, which terminates when the original loan has been repaid, as defined in the loan agreement In the case of the graduate tax, the government's involvement takes the form of an equity holding, enttling the government to a share in the benefits of higher education, in the form of a percentage of the graduate's income over his working life. Thus payments made by graduates are defined as loan repayments in the case of loans, but are to be seen as dividend payments accruing to the government in the case of a graduate tax. - 38 - Table 5 Student Lam Verses Graduate Taze Contrasts and Simliries Mortgage Loan Income Contingent Loan Graduate Tax Government Provides Student Loans to Pay Fees or Living costs Government Provides Student Loans to Pay Fees or Living Costs Government Acquires Share in Human Capital Equity Government Recovery of Costs Government Recovery of Costs Government Share in Benefits Loan Pays Fees (Tuition or Living) Loan Pays Fees (Tuition or Living) Tax Applies to SubsidizedEducation Payments Accrue to Loan Fund Payments Accrue to Loan Fund Taxes Accrue to the Treasury Level of Annual Payments Fixed Level of Payment Contingenton Annual Income Level of Tax Payments Contingent on Annual Income Annual Payments a Declining Proportion of Income Annual Payments a Fixed Proportion of Income Tax Payments a Fixed Proportion of Income Fixed Term Payment Obligation Payment Obligation Until Loan Repaid Tax Obligation While in Employment Loan Disbursement Institutions Loan Disbursement Institutions No Disbursement Need to Maintain Individual Accounts Need to Maintain IndividualAccounts No Individual Accounts - 39 - Another important distinctionregards the likelybudgetary arrangements of a loan or a tax. The revenues generated on account of the loan schemes, either through direct fee payments, or the fee payments made with loan money, accrue to the education budget. Cost recovery is implemented to expand overall resources. Graduate taxes, however, would be applicable only to graduates that had benefitted from subsidized higher education institutions,and are not related to fee charges. A graduate tax is a mechanismfor the government as a whole to recover its expenditure to the higher education sector, and the revenues would be part of general treasury accounts. There is no prima facie for earmarking graduate tax payments to higher education. How effective are graduate taxes as a cost recovery device? In order to illustrate the impact of a graduate tax, the Australian loan program has been simulated as a graduate tax in which students contribute two percent of their income per year"1 , and compared with an income contingent loan scheme with repayments set also at two percent of income. We assume that a graduate tax is collected for thirty years, rather than over the whole working life: this compares with income contingent loan repayments of 17 years (Table 2). While the present values of net benefits of a mortgage and income contingent loan scheme are roughly similar, they are only about half of the value of graduate tax (Table 6). Whereas an income contingent loan scheme achieves only nine percent cost recovery (Table 4), a graduate tax would result in roughly ful recovery of the equivalent loan for 20 percent of teaching costs, though this may not accrue to higher education. Within twenty years (assuming student cohort growth of three percent a year), a two percent tax would generate about 15 percent of the total universitycosts in Australiat2 . t"Technically a graduate tax should be charged only on the income enhanced by human capital investment in university education (ie. on income earned over that received on average by those with university entry qualifications). For administrativeefficiencya lower average rate, leviedon all income is assumed, rather than a higher marginal rate only on the graduate earnings differentiaL 12 Details of these calculations are available on request. - 40 Table 6 Present Value of Net Payments for Alternative Deferred Cost Recovery Programs13 (Australian Data) 10 Year Mortgage loan $A 3,602 Current Income Two percent Graduate Contingent Loan Tax $A 3,126 $A 6,877 The chief justification for the equity finance approach is that it generates more revenue than a loan scheme. Since there is no formal connection with costs of education, tax payments can continue long after a loan would have been paid off; moreover, taxes are levied on higher salaries, given the upward movement of graduate salaries with age. Yet this gives rise to the criticism of graduate taxes, particularlyin comparison to mortgage loans, that they are "front-loaded':the government has to pay out money immediately, but receives much of the return only in the more distant future when the stock of tax paying graduates accumulates. This argument may be overstated since in principle, the government may borrow against these outlays - just as it would if it ran a student loan program and all calculations have discounted reserves to their present value. In practice, however, there maybe some obstacles to a graduate tax (whichapply also to loan schemes with repayment effected through the tax system). In many countries there are constitutionalor legal barriers to creating a graduate tax; the tax may not be administrativelyfeasible in some developingcountries where collection mechanisms are weak; and it may be difficultto track down the self-employed. Where tax systemsare weak, administrativecapacity to identifygraduates may be absent. 13 Assumptions:All calculationsuse a discount rate of 5 percent. Loans assume three years A$ 2,500.Loan charges interest rate equal to inflation; has one year grace period; default rate of 10 percent; administrative costs equal to two percent of outstanding debt each year. Collectionsthrough tax system assume evasion of 3 percent and administrativecosts of 0.5 percent per year. - 41 - Box 11 Argentina's ProposedGraduatelnwm Tax In 1986, in the wake of severe fiscal shortages, the government of Argentina drafted a proposal for a graduate income tax on higher education. The draft contained three essential elements. First, a three percent tax on all income from professionals after the third year of graduation, to be deposited in a special account for each university. Second, a one percent tax on all transactions for professional services involvinguniversity graduates, to be paid by the contractor of the service. Third, parents of the students would also be required to pay and additional one percent of their income, beginningat the same time as student payments. The total income from the three components was expected to equal 15 percent of the entire higher education budget Only half of this revenue was to accrue to the universities,the rest would return to general treasury funds. The tax never received parliamentary approval. Source: Gertel 1991. Despite these obstacles, in many instances there could be practical advantages to a tax as opposed to a loan. First, a graduate tax obviates the need for the government to discuss the sensitive issue of payment of interest. Chargingnear market rates of interest (central to ensuring that loans do not lose too much money) can be politically difficult. A graduate tax allows the government to avoid this controversy because payments extend sufficientlyso that present value returns are greater than would have been a loan with market interest rates. A second advantage may lie with the simplicityof calculatingwho must pay. Rather than determining who has completed payments, the tax simplyassumes that all graduates must pay. In terms of generating extra resources for higher education, there may be some dangers associated with true graduate taxes. With an income contingent loan, it is clear that the treasury acts as a loan collection agency and that legitimately the proceeds should return to the higher education sector (or at least to the loan fund). Graduate taxes howeverresult from the earlier acquisitionby the government of an equity share in the graduates' human capital: although the proceeds of a graduate tax could be earmarked for higher education, there is no overwhelmingjustificationfor doing so and it is unlikelythat the treasury would accede to this readily. Finally, it is arguable that income contingent loans and graduate taxes may be more -42 - complementary than competing.Equity finance maybe appropriate to recover costs from students in subsidized (mainly public) institutionsonly, while a loan program seems more justified for students attending fee-paying (mainly private) institutions. That is, loans may be seen as a tool to help students to meet existing fee payments, while a graduate tax serves as a means of inplementing cost recovery, obviating the need to introduce fees. EmpkWz M Taxationof firms,the users of educated manpower, is an alternative that has begun to receive attention; it is suggested that in certain country settings, notablyskill-shortagestates in Sub-Saharan Africa, a payroll tax on the employment of graduates would result not only in the generation of revenues that offset the cwstsof higher education14 , but would also lead to a more economicaluse of graduates in the labor market (Colclough 1989)13.Graduate payroll taxation is unlikely to be feasible in situations of excess supplies of graduate manpower and high graduate unemployment,because of the disincentiveeffects on the employment of graduates; it is more appropriate in economiessufferingfrom general shortages of higher educated workers or of particular high level skills.In this case, taxes on employersrelated to the use of graduates in short supply can be regarded as a scarcity tax,whichwould not only result in revenues, but also encourage parsimony in the use of graduate manpower by firns. Employers tend to pass the cost of general payroll taxes onto the employees in the form of 14 Using data for Botswana, Colclough(1989) shows payroll taxation levied on graduate earnings would be effective in terms of cost recovery,and compares well with an alternative income contingent loan scenario. Whfle some of the assumptions employed in the simulationsdo not appear to be realistic, a reworkingof the results by the authors using alternative assumptions give results that are even more favorable for payrol taxation. These results are available on request uS See also Tilak and Varghese (1991); although referring to this as a "graduate tax", they essentially advance a similar idea. - 43 - lower wages16; a sharing of the incidence of payroll taxation between the employer and workers is to be expected (the proportions depending on the elasticities of supply and demand of labor). Thus Colcloughsees additional merit in a payroll tax on graduates in loweringgraduate salaries, thereby reducingtheir scarcityrents and the unnecessarily high private rates of return they derive from higher education. However, even with backwardshifting of payrolltaxes, it is not clear that firmswill pass the costs of a tax on graduate employment onto the graduates alone. It is possible that firms will be tempted to shift such a tax onto workers in general (again depending on the elasticitiesof supplyand demand for different categoriesand levels of skiled workers). If this were so, a tax on graduate payrolls wouldbe inequitable,in effectivelyrequiringworkersof all skilllevels to contribute to the costs of the education of the highlyeducated. Not all suggested variants of the payroll tax idea seem feasible,however. Tilak and Varghese (1991) unrealisticallycall for full cost recovery of higher education, coupled with a regime of differential tax rates, related to costs of major disciplines (engineering verses arts for example). An alternative scheme is suggested in a recent comprehensive review of financing options for post-secondary education, by the International Academyof Education (1990).This calls for a payrolltax coveringall workers (not just graduates) to be earmarked for education, on the lines of the French apprenticeship tax. There seems to be little theoretical justification for such a tax (unless it could be shown that there are external benefits such that aUl workers benefit from graduates) and, given tax shifting, the equity implicationsare unlikelyto be acceptable. In some countries, firms contribute to the cost recovery through the repayment of student loans. In Ghana, we have noted, employersof graduates who have taken student loans, contribute 12 percent of wages to the national social securityfund, which is redirected to the education budget until the student loan is repaid. Although this is, formally, a payroll tax on graduate employment,the Ghanaian scheme may exact no real contribution from the employer; these payments might have been made to the pension fund even in the absence of the loan (Box 1). In China, a de facto policy of employer loan repayment exists.Students who 16 Forward shifting onto the consumer is also possible. The classic study of payroll tax shifting is by Brittain (1972) and relates to the US. See Whalleyand Ziderman (1991) and Middleton, Ziderman and Adams (forthcoming) for further references and for an application of payroll taxation to the fnance of training in developing countries. - 44 receive loans often have them repaid by their employer, the compression of wage differentials existingin the Chinese labor market necessitates (and perhaps justifies) such employer contributions. CommunityServic Governmentscan move beyondexp1icitlyfinancialinstruments to exact paymentsfrom students or graduates, who would perform work or provide service in areas of high societal value, as a means of partially 'paying off" the costs of their higher education. This approach has much in common with the compulsoryor voluntary study-serviceschemes found in many developingcountries, in which students (or recent graduates) perform community service outside the university.However, the objectives are very different though not necessarilyin conflict. Most current programs may be regarded as "awareness"schemes, directed primarily towardsthe students themselves,with the aim of inculcatingsocietal values and countering tendencies towards student elitism and isolationfrom the life of the general community.Cost recovery schemes concern socially productive activitiesthat are in short supply. They may relate to student activitiesconcurrent with study or, more generally,to labor market activitiesbefore or after graduation. We consider each briefly7 . Could students not meet part of the costs of their eduction by "workingtheir way through college",either by performing tasks within the university(assisting as library clerks, for example) or part-time work in the general labor market? In many job-shortage developingcountries, this approach is not feasible, particularlygiven the resulting displacementof universityjunior personnel by students that would result. An alternative is to utilize students for tasks of high societal value, against which tuition fees could be waived, whollyor in part. The 'Perach" program, covering all the universitiesin IsraeL provides an example of a well functioningscheme,along these lines. Israeli students maywork as tutors to disadvantagedteenagers, for which they receive payment equivalent to a half of universitytuition fees (in turn covering about 20 percent of universitycosts). Some twentypercent of Israeli students are enrolled in the program, whichoffers a valuable service that the free market seems unable to provide. 17 In a concurrent paper, the authors survey current schemes and discuss ways in which they could be enhanced to provide for cost recovery, in addition. See Albrecht and Ziderman (forthcoming). - 45 In every society, shortages persist (at least in particular locations) in the supply of certain occupations that are deemed to be sociallyof some importance. The persistence of excess social demands for these activities is to be explained by such factors as the presence of externality effects or inflexiblewages differentials. Thus, in many countries there are acute manpower shortages in such areas as rural health care and secondary school teaching; graduates are normally unwilling to perform these tasks, at least in the numbers that society deems necessary. A partial solution is to require recent graduates to perform a period of national service in one of these sociallyproductiveemployments,for perhaps two years followinggraduation, as a form of partial repayment for their education. Boz 12: Nepak NationalDevelopment Service Between 1974and 1980, the Nepalese government implemented a program of required rural service for all higher degree university students. The primary emphasis of the National Development Service Program (NDS) was twofold: to supply educated manpower for rural development and to improve the higher education system. Each participating student worked for one year under both universityand local supervision, partly as a teacher in a rural secondary school and partly as a general community development worker in the surrounding community. Participants were responsible for mobilizing local resources and manpower for community projects, including health and nutrition education, reforestation campaigns, adult literacy teaching, improved sanitation, water supplies, bridges and schools, familyplanning promotion, agriculturaland horticultural demonstrations. The societal benefits associated with the program were high. Rural school enrolments rose sharply, particularly for girls. Literacy campaigns proved successful,and clean drinking water and public health campaigns improved livingconditions. Students were able to transport materials to remote villages, as well as providinga feedback mechanism for the government,of information on rural needs. In addition to its manpower function, the NDS was planned as a tool to make higher education itself more relevant to resolving the most pressing needs of the society. University relevance had come into question as it retained much of the colonial legacy. The NDS was seen as a means to adapt the universitycurriculum to national concerns facing people outside Kathmandu (where 95 percent of the population lives). The NDS therefore served as an important feedback mechanism for university planners and teachers. Unfortunately the program was abandoned due to politcal unrest, but the current government is considering reinstating it In the US, the federal government pays the tuitionof a limitedgroup of medicalstudents each year who agree to serve in areas of acute medical manpower shortage; during the 1960's, a similar program - 46 provided a supplyof rural teachers. The Nepal scheme (Box 12) providesa relevant example from a developing country. Other programs operate in Indonesia, Yemen and Mexico. Such national service schemes provide indirect cost recovery through generating positiveexternalides,such as those related to increased education or health care. Using national service as a form of payment for higher education, however, does not represent a financial addition to the sector. In addition, it is possiblefor the government to effect further cost recovery through payingthe graduates on these programs a wage that is lower than market rates. In what was formerly the Yemen Arab Republic, graduates were used as a lower cost replacement of overseas primary and secondaryschool teachers. However, this is really equivalent to a graduate tax, as discussedabove. Tle opportunities for increasing student contributions to the costs of higher education are many. Student loans have received much attention both in the literature and in practice. While they have not alwaysworked well, we have argued that suitablyreformed, they can constitute a productive,though limited, mechanism for cost recovery. In certain countries, however, other mechanisms may be more appropriate. Indeed, the policy maker is presented with a wide menu of policy choices, though some creativity may be required in their application to particular local settings. Some of these have been outlined in the present section. -47 - While this paper has primarilyfocussed on the financial implicationsof loan programs, equity considerationsare of considerable importance. Despite the lack of empirical work on the equity impactof loans on access in developingcountries, it is clear that increases in cost recoverywill,on the margin,discourage some individualswho would otherwise have attended. This may be seen as a negative equity impact As noted earlier, however, most developing country higher education systems are not very equitable to begin with Access tends to be skewed towards higher income groups,where children attend better primary and secondary schools and families can afford to have their children out of work for longer periods. Thus, a large group of talented individualsoften lacks de facto access to educational opportunities, while large subsidies accrue to groups that are well-off. Increases in cost recovery will make it harder for these groups to have access, but it willalso allow the governmentto invest in better access to primary and secondary education and provide grants to the least well off. The central equity concern of a deferred payment program should be how to design it so that any tendency to deter access is minimized. Loan programs can be expensive enterprises which do not easily satisfy the needs for cost recovery. Without careful consideration,it is unwiseto start a loan or tax scheme. With that said, the following list of issues can serve as a guide in consideringwhether a deferred payment scheme should be implemented. Annex 1 provides a more complete check list of options, summarizingthe major issues discussed in the body of the paper. First, a deferred paymentprogram requires the participation of a debleaoflmdoah.dkkc nk which in most instancesrequires the direct or indirect participation of the taxation department or social security agency - either for direct collection or indirect support for the collection agent. The current evasion rate among graduates on taxes, the number of graduates that work in the public and private sectors, and the current rate of graduate unemploymentare relevant variables to be taken into account If default or evasion is likly to be greater than say 25 percent, it would be inadvisable to implement a program of refundable support; in - 48 such cases, a carefully targeted grants program is likelyto be more cost-effective. Second, with loans, there must be a willingnessto charge interest rates equal to or above inflationin order to minimize subsidies. With tax or income contingent collection,the rate assessed must be sufficient to ensure significantcash flows.Careful financialcalculations must be conducted,which account for the likely impact of inflation - particularly on the size of annual disbursements -- and growth of the higher education system. From this information,one can assesswhether the program will generate significantincome for the higher education system. Third, the relationshipbetween necessaryrepaymentsand the likelyincome of graduates must be examined to ensure that repayment burdens never pose an excessive burden on graduates. Excessive burdens only result in higher default. Average income profiles of graduates are not sufficient to understand likely problems. The income range according to profession and sector will be equally important in program design. Fourth, developinga means of targetingsupport to needier and more academicaly deserving students will be crucial to a program's efficiency.The larger the expected participation rate, the greater is the need for tight repayment terms and strict enforcement of collection. In developingcountries, good targeting means that an institution with access to information beyond income tax information. Institutions closer to students, such as universities, are often able to make the best judgements regarding need. Fifth, loan losses can only be justified if there are potential social Pins that would not be reflected in a graduate's income. Subsidiescan promote, indirectly, private institutional development and/or manpower direction (graduates as teachers, rural development workers, private sector entrepreneurs), by forgivingloans. If these are desirable options, one can consider whether a student loan program is an efficient way of transferring subsidies to these areas. A critical issue is whether a ministryof finance or a taxation department will support the program. In most developing countries, such support will be required for accurate targeting and efficient collection. In some instances, these ministries are too weak to handle the load, and alternative solutions to financialproblems in higher education may have to be explored. These includeoutright feeswith limited grants, - 49 - or restriction of numbers that participate in public higher education. In the instanceswhere there is a credible possibilityfor the programs, support amongnecessaTyorganizationsand the proper setting of interest rates and recovery terms increase the likelihood of a program's success. Cost recovery for student living expenses and institutional costs continues to be a pressing concern in many developingcountries. Resources to promote access to quality higher education systems have frequently been eroded because of rapid universityexpansionwithout sufficientgovernment resources. Student loans and alternative forms of deferred paymentspresent and important policyoption to assist in cost recovery, without deterring access to qualified students. In order to achieve these twin goals, however, programs require careful planning, particularly to ensure recovery. - 50 ANNEC LIChecklist of Policy Options for Deferred Cost Recovery Structure/ Policy Options Lendina Institution a. Autonomous Public Body The most comon institutionalstructureis to create a publicly administered and financed loan to distribute and collect loans. organization b. Another comon institutional to administer loans. Public Banks c. Private Coemercial Banks Descrintion structure utilizes publicly owned comlercial banks In countries with more developed banking systems private banks may be used to allocate loans. (US, Indonesia,Denmark). d. Higher Governmentsmay transferfunds to higher education institutionsfor the purpose of administeringloans. Education Institution (China,Chile). e. Directly Money is disburseddirectly from government from Govern- ministriesor trust fund, and collectedby treasury. ment Accounts (Australia,Ghana). RePewment Mechaniam a. Mortgage type loan The most common approachby which the capitalizedloan is broken into equal monthly payments. b. Income Contingent Loan Payments are a fixed portion of monthly or annual income, thus putting a limit on the debt burden to a graduate (Sweden). c. Graduated payments Payments fixed in advance,but increasewith time. d. Income Contingent Loan (Tax) Same as 'b' except payment may be collected through the taxation system (Australia). a. Deferral of Social Benefits Repaymentis through an already existingpayroll tax in which pension benefits do not begin to accrue the loan is repaid (Ghana). f. Students contributethrough a lifetime increase in their tax contribution.(Offered briefly at Yale University,proposed in US and UK). Graduate tax/equity finance until g. Employer In countrieswhere graduatesare scarce, employers Contribution contributeto loan or tax repayments Through Tax as a form of "scarcity"tax. Loan repaymentsare shared or Loan between employersand employeesin Ghana and China. Targeting h. National Service Repaymentthrough labor that is sociallyvaluable to and in demand by the society. a. Means Testing Selectionof credit recipientson the basis of family or individual(Sweden,Norway) income. Or more complex socio-economicstatus indicators(Chile). b. Ability Criteria Selectionof studentson the basis of performance at secondaryschool, on national exams or within universities(Indonesia). - 51 - Interest Rates and Subsidies c. Priority Areas Priority support for studentswho study in fields of nationalmanpower priority -- e.g. engineering,teacher training, health. (Colombia,Barbados) d. Restricted Length Limitationon availabilityof funds to a fixed period of study -- as the officialduration of a given course. (Brazil, Denmark) a. Fixed Real or floating Interest rates can fixed in relation to inflation at either negative, zero percent or positive real rates, can float with an index of commercial rates. or they b. Differential Students charged differentrates of interestbased Interest on their economic situation,thus targetingmore rates subsidizedsupport to needy. (US, Japan). Default Minimization c. Repayment Length The length of the repaymentperiod can be varied to achieve a balance between debt burden and financialefficiency. d. Graduated Annuities Payments can be calculatedso they are smaller in the first years and larger later on. e. Up-front Discount on Tuition Allow studentswho are eligible for a subsidized loan to have their fees reduced by a fixed percentageif they forgo the loan. (Australia,Israel). a. Grace Period Allow students a specifiedtime after graduation before repaymentbegins, with the assumption that they need time to find employment. b. Income Threshold ALLow graduatesto defer payment during any time in which their income falls below a specified level (Sweden,Kenya, UK). c. Incentives for Financial Agent Where the governmentis the guarantoron the loans, the governmentdiscounts the value of that guarantee sufficientlyso that institutions prefer to collect from the student. d. Require Guarantor Requiring an income earning co-signer on a loan who agrees to pay in the event that the graduate does not. (Ghana,Barbados,Brazil) e. Payroll Deductions Requiring employeesto withhold a portion of salary of graduatesfor the purpose of paying the loan. (Jamaica) f. Income tax to locate defaulters Governmentsto locate individualsthat might be in default, through taxation institutionsCanada g. Moral Pressure Publish lists of defaulters(Jamaica) h. Required Insurance Require student to pay an up-front fee to insure against losses that result from death or debilitatingillness or accidents. (Brazil). i. Bar Further Bar access to further credit if default. (Brazil) credit j Collection Agencies Utilize private collectionagenciesto locate studentsand secure payment. (Honduras,Colombia). - 52 - Annex 2. MethodologicalNote for CalculatingSubsidieson Mortgage Loan Programs Aumptioni =ad In clulationg for Tabk 2 1. Students receive equal real value loans over a four year disbursement period in lump sums at the beginningof each year. 2. Administrativecosts are spread out evenly during the life of the loan. 3. Default is the frequency of loans that fail to repay. It is expressed as a probability for each year of repayment 4. Grace periods have been rounded to the nearest year. 5. Repayments are in equal nominal amounts in yearly installments,at the beginningof each payment period. 6. Inflation is constant throughout the life of the loan. 7. Defaulted loans carry an administrativecost equal to good loans. (1) Calculatingthe Student subsidy PV= present value D = disbursement value i = initial interest rate (during lending period) I = Interest rate during repayment period g = grace period in years n = repayment length r = Opportunity cost of capital, from time of lending onwards. L = disbursement length IL A = DE (1 + ji.U(l-) Amortization value = i-t p = A*I 1 - (1 + The annual payment = The cash flow is as follows: 4 years of loan disbursementsof equal real values (adjusted for inflationeachyear), O during the period of the grace, and P during the repayment length (n) Di IaL PV disbursement = 1.1 (1 + - - 53 - 1 a- PV repayments = -1 (1 + ry L-1+ $ Subsidyto student = PVdjsb- PVm % Subsidyto student = (PVdib,- PV,,)/PVdi, (2) Calculatingloss with default The calculations are the same, except that payment amounts are reduced to include the probability that they are not made. Thus, the cash stream uses the followingrepayments: Pd,r = P*(1-d) where d is the probabilityof default. Thus the cash stream is only adjusted during the years of repayment (3) Calculatingthe total loss to the government Each year of the cash stream is adjusted to reflect the cost of administering the loans. This is calculated by using the annual percent cost of servicingoutstanding debt od = outstanding debt on loan ac = administrativecost of servicingloan, as percent of outstanding t = year in the loan life cf = previous cash flow, includingdeductions for likelihood of default CF = adjusted cash flow, includingdeductions for both default and administrative costs. Thus in each year, the cash flow is adjusted: CF. = cf, - (od, * ac) and the PV and subsidies are calculated as in section 1. debt each year - 54- REFERENCES Albrecht, Douglas and Adrian Ziderman (forthcoming). National Service as Cost Recovery for Higher Education. Washington,D.C.: The World Bank. Australia (1988).Report on the Committeeon Higher Education Funding (the Wran Report), Canberra. Barnes, John and Nicholas Barr, 1988. Strategies for Higher Education: The AlternativeWhite Paner. Edinburgh:AberdeenUniversity Press. Barr, Nicholas. 1989. Student Loans: The Next SteRs, Edinburgh:Aberdeen University Press. c. Forthcoming. "IncomeContingentLoans: An Idea whose Time has come". Essays in Honor of Mark Blaug. Blaug, Mark. 1973. An Introductionto the Economics of Education. London: Penguin. . 1987. The Economicsof Educationand the Educationof an Economist. Aldershot: Edward Elgar. Birdsall, Nancy and Estelle James. 1990. Efficiency and Eguity in Social SRending: How and Why GovernmentsMisbehave. World Bank Working Paper No. 27. Washington,D.C.: The World Bank. Brittain, J.A. 1972. The Payroll Tax for Social Security. Washington, D.C.: Brookings Institution. Brunner, J. 1990. La Educacion Superior en Chile: Politicas. (Processed). 1960-1990 Evolucion y Colclough,Christopher."RaisingAdditionalResourcesfor Educationin Developing Countries: Are Graduate Payroll Taxes Preferable to Student Loans?". Universityof Sussex. (processed). Departmentof Education & Science (UK). 1988. ToR-UR Loans for Students. London. Friedman,Milton. 1962. CaRitalismand Freedom. Chicago University of Chicago Press. Gertel,Hector, 1991. "Issuesand Perspectivesfor Higher Educationin Argentina in the 1990's".Higher Education.Vol. 21, No. 1. Pp. 63-81. Glennerster,Howard, Stephen Merrett, and Gail Wilson, 1968. "A Graduate Tax", Higher Education Review. Vol. 1, No. 1. Gomez-Buendia,H. 1984. FinanzasUniversitarias. Bogota: Fedesarrollo. - 55 - Hope, J. and P. Miller, 1988. "FinancingTertiary Education:An Examinationof the Issues".AustralianEconomicReview. Vol. 4. InternationalAcademy of Education. 1990. Rethinking the Finance of PostCompulsoryEducation. (Processed). Jimenez, Emmanuel, 1987. Pricing Policy in the Social Sectors: Cost Recovery for Educationand Health in DeveloRingCountries.Baltimore:Johns Hopkins University Press. Johnstone,Bruce. 1986. Sharing the Costs of Higher Education:Student Financial Assistancein the United Kingdom. the Federal ReRublicof Germany.France Sweden and the United States.New York College EntranceExaminationBoard. McMahon,Walter. 1988. "PotentialResourceRecoveryin Higher Education in the DevelopingCountriesand the Parents'ExpectedContribution'. Economics of Education Review, Vol.1, No.1, pp 135-152. Middleton,John, Adrian Zidermanand Arvil Van Adams. Forthcoming. Skills for Productivity: Policies for Vocational Education and Training in DevelopingCountries. New York: Oxford UniversityPress. Mingat, A., Jee-Peng, Tan, and M. Hoque, 1985. "Recoveringthe Cost of Public Higher Education in LDC's: To What Extent are Loan Schemes an Efficient Instrument?".World Bank Education and Training Department Discussion Paper, Washington,D.C.: The World Bank. Morris, Martin. 1989. "StudentAid in Sweden: Recent Experienceand Reforms', in Maureen Woodhall, ed., Financial SuRRort for Students: Grants. Loans or Graduate Tax? London Kogan Page. OECD, 1990. FinancingHigher Education: Current Patterns. Paris: OECD. Psacharopoulos,George, Jee-Peng, Tan, and Emmanuel Jimenez, 1986. Financing Education in Developing Countries: An Exploration of Policy Options. Washington, D.C.: The World Bank Psacharopoulos,George and Maureen Woodhall, 1985. Education for DeveloRment: An Analysis of InvestmentChoices. New York: Oxford UniversityPress. Quebec Student FinancialAssistance Program. 1990. Annual ReRort. Reuterberg,Sven-Ericand Allan Svensson.1990. "StudentFinancialAid in Sweden" Higher EducationPolicy. Vol. 3, No 3. Schiefelbein,Ernesto. 1990. "Chile: Economic Incentivesin Higher Education", Higher Education Policy. Vol. 3, No 3. Tilak, Jandhyala. 1985. 'Analysis of Costs of Education in India". National Institute of EducationalPlanning and Administration. - 56 - Tilak, Jandhyala and N.V. Varghese. 1991. "FinancingHigher Education in India". Higher Education,Vol. 21, No. 1, pp 83-102. U.S. Departmentof Education.1990. Reducing Student Loan Defaults: A Plan for Action. Washington,D.C. Vahl, Teodoro. 1990. Report to the World Bank on Student Program in Brazil. Washington,D.C.: The World Bank (processed). Loan West, Edwin. 1988. Higher Educationin Canada:An Analysis.Vancouver:The Fraser Institute. Whalley, John and Adrian Ziderman. 1990. "FinancingTraining in Developing Countries: The Role of Payroll Taxes". Economics of Education Review, Vol.9, No.4, pp 377-387. Woodhall,Maureen.1983. StudentLoans as a Means of FinancingHigher Education: Lessons from InternationalExperience.World Bank Staff Working Paper No. 599. Washington,D.C.: The World Bank. 1986. (processed)FinancialAid to Studentsin Indonesia.Wa4diirn D.C.: The World Bank. _________. 1987(a). "Establishing Student Loans in Developing Countries: Some Guidelines".World Bank Educationand TrainingDepartmentDiscussion Paper. Washington,D.C.: The World Bank. _ . 1987(b). Lending for Learning: Designing a Student Loan Programme for DevelopingCountries.London CommonwealthSecretariat. 1989. ed. Financial Support for Students: Grants. Loans or Graduate Tax. London: Kogan Page. _. , ed. 1990. Financial Support for Students:Grants. Loans or Graduate Tax? London, Kogan Page. . 1990a.Ken=a: World Bank Assistance to Education Sector ReRort on UniversityStudents Loan Scheme. Washington,D.C. (processed) .1990b.Student Loans in Higher Education:1. Western Europe and the USA. Paris: InternationalInstitute for EducationalPlanning. .1991.StudentLoans in Higher Education:2. Asia. Paris International Institute for EducationalPlanning. World Bank. 1988. Education in Subsaharan Africa: Policies for Adjustment. Revitalization,and Expansion.Washington,D.C.: The World Bank. Distributors of WorldBank Publications ARGENTINA Cdo},kd~SRL Gd Gacim PAidd 1, 4thP 1Xw-O 453/465 133Bu3_oAire FINLAND Ak-einiIea IOjkw P.O.9E-125 SF40101 HdsldlO10P.O.a11,41 AUSTRAIA, APUANEW GUN. FULSOLOMONISLANDS, VANUATU,ANDWESTERNSAMOA DA.BRob&j@aU 618WItidihrmeRtod Mhtdiag 3132 Vidala PRANCE Waid BankPu,ltaliaoi 446 miw dleta 75116PD AUSTUIA Cagd Vd CeA Graboo1 A-2I0UWCEME RAHRAIN Bairaie Ruerch andCnAtamy Ai.od" Led. PO.EBxm 22103 M aaTown 317 SANGtADES Miaoh du.htaDeedepAa Aad USotdIyO ASI Ilase, 5Rcd1AUa Dhmondt R/A. Dhaka 129 Bjnu m MA Road Mdee Cant NoMkha -311w HONGXONC.MACAO MMDlad. 414 Wydham SReet WteeeCeetre 2ed Ploor Co"rHl Kong 42, NORWAY NasarhnnmMeen Cabr ok Dpartead P.O.&e 4125Efttaitd N4120.1Oi6 Pratm Rl, 71dFAeer NearThakwesg, awee Ahadabad t-3100I PAJaSTAN Mite. RediAgaoy 65,.Sahrah..Qumld.1-Az P.O.B-Nt72 Lmh= 3 ELSALVADOR Fuada Averda Maet.4 SnArm* WadoSUAler. a Sm Savadw EGYPT.A RURLKCOF Al Alm AlGaleeSked Cai SWEDEN Prslgi.dle FdtOaF _ POJUGAL UwTat Sa44 RwDCwmo70-74 1200Liba SAUDIARADA QATAR la*wkSW P.O 13e3196 Elydh 11471 SOt2SPIoma JAPAN Enm BoSar 37A4HNp 34Em% Smhkyo.u113 Tokyo rDma XENYA ARmSeakSSer(BA IM. P.O. Bt 45245 Nowj FSt lf AIAla SheeT AD aC.nt Flew P.O. B 7PN Rip" HeBAdAllA AkES. Vuiblift beg l sted sod P.O.BRe3a9 33U1MdmainaedflaemaAwedSh P.O. BOmwS 5.edd1 S30b KOREA,REPUBLIC OF PaKneeamoeCCaporatio PDO. 101, ar Kmwgwhaa S SDM KUWAIT M5IE WB Yauga it, P.O. BO5XSW MAAYSIA ha i4_dled P.O. Rot II2Zda Pamd FM" B SINGAPOR.TAIWAN, MYANMAARRUNBI IbIamaMmD P- di m Pdvet LtdL 61J lto FtP&durid SMde 24Newidudd Road fw*lg63 s Feraw*am W _wpat.W IMAB 8=3D0 S-1042SSUKIhao, SWrITBELAND F5wdLuur POLAND ORPAN Pr K y1Natld 01- Wana41 _ B S103 n rlEe.at MEDRI kdauaauusvMm DOMINICAN JRI:C bIdmasTallmCparA. BerIblC D Apdtad Pai 2190 SmteDatgo SAJ SRI ANKAANDTIEMADIVES Iiere p PAXam244 1C0L SkrOittpt A. Gwdbtwh meaa CAt;eb 2 PILEPrDNES 1 ' CInteralemkmt. Fiamh R PD e Bad" Apla AvewnJMak MtroMnal a ITALY LMmuCA _imlda SAmd SPA Via DBme1deoPatle120L0 Ccs PbQeSl DENMAtX Smiardellmeahr RaU,teemAilI DCd19 FrederisbegC SPAIN Msni-%m UIh COJ.37 01 MndlHd UfralatdoradoeADABOS Came1Mde Ca 351 UWEmalata PEWU EdikldoDameelloSA Apurtmlad 3524 Umm DIDONESIA Pt 1dra UIad JL So Rairdae37 PDO.me 11 Jkata Pula NlEemi F.rmtha_4immrws Subma"ofSuyim PO. E 409 G u224 RBw1456 Paimela Home 16-AAblMalg L.,kw.2260Dt RBuasvMm Svc OMAN MEMIRIInfoea 5ieervim PO.Bex1613.SedbArprt MIucat 3-C-1152 t<digda CrnmReod Hyduebad-S6027 COTEiVOIRE C-metEftomekdeafludom AMcdnm(CSDA) 04SF. 541 AlrdinOl4 P1. 41.aetShe NIGERIA Unively Pi Uaited TheeetaL PdlvateMdlSags Ibadan ieyadeyaHteetdltdldtn SkiMadtRoadGAdtagar Baidore-So5t COLMWBIA _NfER161*ala. Apar4a, Ae 3M0 BqeisBD. CYPRUS MMSM P.O Sac9 GUATEMAIA Ukain Peda SWa Ss.CoIne7-SS Zhl Guataba City 17 ChiffemAvowe Calmttb-70t2 CHINA Chta IReamdal & Econttc PubmhldigHmae S4 DeoSi Dke BdNp CqeTtaW0 GREECE X1 24,Ipdau st5 Plnt. mant_ Atha,1163S 13/14Ad Al Rad NewD0d-110012 CANADA LeDlte. CFP.8ktSBnwAeApe Somidv9il Q:Od )a 855 MOROCCO SoaA iEt,dammkShigM arece 12seM rteBd..dAnf CsbLca NEWZEALAND H2le Ul *d IeloiE PrivateBs NewMarke Audaid Brmab.,ffr. 15j N.Heredta arg Bdle Retae Boombey-4C01 RUISIUM je DeL_ey A,. du Rd202 I0ODBnom SOUTHAFRIC SOTSWANA FadqlM OdUa r S A a NETIHERLANDS 1b0-Pub1MlK.bv. P.O.Ear 14 724DBA T u GERMANY UNO-Va1ag PpdpeladfA1IAeSS D040 aut I NDIA Allied PubhPh.PdvateLd. 731MomtRoad MeM-650012 76,f4KDO-Avaen jale. MEXICO INPOIEC Ap.tadoPeeBi 22146 lW1ATL MedDP. & r Ca_apoeltal5t CH 1211G_wa It Formaima*morim lfdepayt Uavk-edmAboammemdi Chp 33ti2a CH0102 1uam TANZANIA Odtd UIvatty Prm rim Ramm 1aiSdemi iTAlAND CmtD = tseun S3D691kO Rad oow TRINIDADAaTOSAG0,ANTUA SARSDWA. DARADOSL DoMN.A, GC NADA.GUYAKA, JAMAJCA.MONTERRAT, Sr. snTTE &ANEVR,ST.CLA ST.VINCET&GAENADNES *_m stUdestinm fwm tr w. _edl UNITm ATARESMIAXB3 MS4RGulfCc PAXlUme Sh UNmIKINGDOM M-LaieieUd. P.O.BD 3 Abm ampa d VNEUELA Ladde Rae Apbo.*0.337 C.tntI6O.A EUGOLAVIA jqaiEaM*a PJOE1mSI3 TrBR11tl30 IG wteGtSi Recent World Bank Discussion Papers (continued) No. 109 Indian Women:Their Healthand EconomicProductivity.Meera Chatterjee No. 110 Sodal Securityin Latin America:Issuesand Optionsforthe WorldBank. William McGreevey No. 111 HouseholdConsequences of HighFertilityin Pakistan.Susan Cochrane, Valerie Kozel, and Harold Alderman No. 112 StrengtheningProtection of Intellectual Propertyin DevelopingCountries:A Surveyof theLiterature.Wolf;ang Siebeck, editor, with Robert E. Evenson,WilliamLesser,and CarlosA. Primo Braga No. 113 WorldBank LendingforSmalland MediumEnterprises.Leila Webster No. 114 UsingKnowledgefromSocialSciencein DevelopmentProjects.Michael M. Cernea No. 115 DesigningMajor PolicyReform: Lessonsfromthe TransportSector.Ian G. Heggie No. 116 Women's Work, Education,and Family Welfarein Peru.Barbara K. Herz and Shahidur R. Khandker, editors No. 117 DevelopingFinancialInstitutionsforthePoorand ReducingBarriersto AccessforWomen. Sharon L. Holt and Helena Ribe No. 118 ImprovingthePqformanceof SovietEnterprises. John Nellis No. 119 PublicEnterpriseReform:Lessonsfromthe Pastand Issuesforthe Future.Ahmed Galal No. 120 The InformationTechnologyRevolutionand Economic Development.Nagy K. Hanna No. 121 PromotingRural Cooperatives in DevelopingCountries:The Caseof Sub-SaharanAfrica.AvishayBraverman,J. Luis Guasch, Monika Huppi, and Lorenz Pohlneier No. 122 Performance EvaluationforPublicEnterprises. Leroy P. Jones No. 123 UrbanHousingReformin China:An EconomicAnalysis.George S. Tolley No. 124 The New FiscalFederalism in Brazil. Anwar Shah No. 125 HousingReformin SocialistEconomies.Bertrand Renaud No. 126 AgriculturalTechnology in Sub-SaharanAfrica: A Workshopon ResearchIssues.Suzanne Gnaegy andJock R. Anderson, editors No. 127 UsingIndigenousKnowledgein Agricultural Development.D. Michael Warren No. 128 Researchon Irrigation and DrainageTechnologies:FifteenYearsof WorldBank Experience.Raed Safadiand Herve Plusquellec No. 129 Rent Controlin DevelopingCountries.Stephen Malpezzi and Gwendolyn Ball No. 130 Pattemsof DirectForeignInvestmentin China. Zafar Shah Khan No. 131 A New Viewof EconomicGrowth:FourLectures.Maurice FG. Scott No. 132 AdjustingEducationalPoliciesto ConserveResourcesWhileRaisingSchoolQuality. Bruce Fuler and Aklilu Habte, editors No. 133 LettingGirlsLearn:Promising Approachesin Primaryand SecondaryEducation.Barbara Herz, K. Subbarao, Masooma Habib, and Laura Raney No. 134 ForestEconomicsand PolicyAnalysis:An Overview.William F. Hyde and David H. Newman, with a contribution by Roger A. Sedjo No. 135 A StrategyforFisheriesDevelopment.Eduardo Loayza No. 136 IntermediateMeansof Transportin Sub-SaharanAfrica:Its PotentialforImprovingRural Traveland Transport.John D.N. Riverson and Steve Carapetis The World Bank Headquarters 1818 H Street, N.W. Washington, D.C. 20433, U.S.A. Telephone: (202) 477-1234 Facsimile: (202) 477-6391 Telex: wuL 64145 WORLDBANK RCA 248423 WORLDBK Cable Address: European Office 66, avenue d'lena 75116 Paris, France Telephone: (1) 40.69.30.00 Facsimile: (1) 40.69.30.66 Telex: 640651 Tokyo Office Kokusai Building 1-1 Marunouchi 3-chome Chiyoda-ku, Tokyo 100,Japan Telephone: (3) 3214-5001 Facsirile: (3) 3214-3657 Telex: 26838 INTBAFRAD WASHINGTONDC 0 0 0 0 0 0 19- 0 9 5~~~~ ISBNOL8213-1952-3i