THE WORLD BANK Discussion Paper EDUCATION AND TRAINING SERIES Report No. EDT14 Recovering the Cost of Public Higher Education in LDCs: To- What Extent are Loan Schemes an Eficient Instrument? Alain Mingat Jee-Peng Tan Monzurul Hoque December 1985 Education and Training Department Operations Policy Staff The views presented here are those of the author(s), and they should not be interpreted as reflecting those of the World Bank. Other Available EDT Discussion Papers EDT1 General Operational Review B. Searle July 1985 of Textbooks EDT2 Does SENA Matter?--Some E. Jimenez August 1985 Preliminary Results on the B. Kugler Impact of Colombia's National Training System on Earnings EDT3 Higher Education K. Hinchliffe August 1985 in Sub-Saharan Africa EDT4 Planning of Education: G. Psacharopoulos August 1985 Where Do We Stand? EDT5 Evaluation in World Bank B. Searle, ed. August 1985 Education Projects: Lessons from Three Case Studies EDT6 Financing Technical Education C. Lee September 1985 in LDCs: Economic Implications from a Survey of Training Modes in the Republic of Korea EDT8 The Structure of Educational E. Jimenez August 1985 Costs: Multiproduct Cost Functions for Primary and Secondary Schools in Latin America EDT 9 Participation in Schooling: D. Jamison December 1985 Determinants and Learning M. Lockheed Outcomes in Nepal EDTIO Classroom Uses of the J. Friend November 1985 Computer: A Retrospective View with Implications for Developing Countries EDT13 Education Costs and Financing A. Mingat In Africa: Some Facts and G. Psacharopoulos December 1985 Possible Lines of Action EDT15 Improving the Quantity-Quality A. Mingat December 1985 Mix in Education: A Simulation J.-P. Tan of Policy Tradeoffs (Continued on back cover) Discussion Paper Education and Training Series Report No. EDT14 RECOVERING THE COST OF PUBLIC HIGHER EDUCATION IN LDCs: TO WHAT EXTENT ARE LOAN SCHEMES AN EFFICIENT INSTRUMENT? Alain Mingat, Jee-Peng Tan and Monzurul Hoque Training Division Education and Training Department December 1985 The World Bank does not accept responsibility for the views expressed herein, which are those of the author(s) and should not be attributed to the World Bank or to its affiliated organizations. The findings, interpretations, and conclusions are the results of research or analysis supported by the Bank; they do not necessarily represent official policy of the Bank. The designations employed, the presentation of material, and any maps used in this document are solely for the convenience of the reader and do not imply the expression of any opinion whatsoever on the part of the World Bank or its affiliates concerning the legal status of any country, territory, city, area, or of its authorities, or concerning the delimitation of its bou-ndaries, or national affiliation. Abstract In many LDCs today, the education sector is still characterized by relatively low levels of enrollment. On the other hand, however, expansion in the public provision of education is limited by constraints in the public budget. To overcome this problem, various strategies could be considered among which is the implementation of cost recovery policies. In this paper, we have argued that higher education would be a suitable candidate for such policies. Cost recovery at this level of education would be helpful, not only because of its potential effect on resource mobilization, but also because it would boost efficiency and equity in the overall distribution of public resources for education. For (other) equity and political reasons, loan schemes have been favored by many authors as an appropriate instrument for implementing cost recovery at this level of education. In this paper, we have examined their likely effectiveness in this role. Our analysis indicates that the outcome depends primarily on what is judged, in each country, to be the maximum acceptable debt burden, measured, say, in terms of duration of repayment, and proportion of a borrower's current income allocated to loan repayment. Our simulations indicate, however, that the terms of repayment required for full cost recovery appear to be easily bearable in Asia; similarly, in Latin America, although probably less so than in Asia. In Francophone Africa, the terms of repayment appear to be more difficult to bear, although they are probably still manageable. In Anglophone Africa, however, the conditions appear to be more problematic. Contents Introduction 1 The Case for Cost Recovery in Higher Education 3 The Equity Argument 4 The Efficiency Argument 12 Loans as a Mechanism for Cost Recovery 15 Some Features of Returns to Education in Malawi 18 The Potential Usefulness of a Loan Scheme in Malawi 22 Upper Bound Estimates 24 The Effect of Repetition and Dropout on the Performance of the Loan Scheme 28 The Overall Effectiveness of Loans as a Cost Recovery Mechanism in Malawi 30 The Likely Usefulness of Loan Schemes in Other LDCs 31 Some Objections to the Implementation of Loan Schemes in LDCs 37 Conclusion 41 References 43 I. Introduction In many developing countries, the government plays an important role in the provision of education. The rapid expansion in school enrollments in the last two decades has largely been a result of strong public investment programs in the sector. Indeed, in many countries today, public spending on education typically represents the single largest share in the national budget. In the 60s and 70s, the growth of educational expenditures came from two sources. First, the public budget as a whole had been growing. Second the share of education in this enlarged pool of public resources had also been increasing. By the late seventies, however, the past pattern of growth in educational expenditures, in both relative and absolute terms, had begun to level off. This slowdown is not surprising since the share of education in the public budget had already become very large in most countries; as a result it became increasingly difficult to compete with alternative collective activities for scarce public resources. At the same time, poor macroeconomic conditions have also. restrained the growth of the overall government budget itself. Over the forseeable future, these unfavorable financial circumstances are unlikely to improve substantially; as a result LDC governments would probably face difficulties in expanding the provision of public education. The effect of this constraint on educational development is often to be exacerbated by demographic pressures (World Bank, 1984). In view of the rapidly growing school-age population, it is even conceivable that in some countries, the outcome would be an absolute, if not relative, increase in the number of unenrolled school-age children. In this respect, Mingat and Psacharopoulos (1984) have shown that the situation is especially difficult in Sub-Saharan Africa. -2- To change the bleak outlook described above, various strategies could be considered. They relate to changes within the education sector itself, since the possibilities for influencing the external conditions are, at best, limited. Broadly, these strategies can be classified under two headings: (i) reducing the unit cost of education; and (ii) increasing private participation in the finanding of education. It would probably be useful to pursue both lines of action since most of the available policy options are often complements rather than alternatives. Clearly, however, the package of instruments most appropriate in particular country settings is likely to be uniquely dictated by a country's economic, financial and political conditions and constraints. For this reason, it would be important to base the choice of a policy package on information about the potential effectiveness of each policy instrument, given initial country conditions. In this paper, we shall document the potential role of cost recovery in education as a policy option. Such a policy would obviously imply a reduction in the per studenc level of public subsidies, and correspondingly, an increase in the private cost of education. We shall consider its application mainly in higher education, since the policy appears to be particularly appropriate at this level of education. Our focus does not, however, imply that cost recovery should not be considered Lor other levels of education. Mingat and Tan (1984a), for example, have shown that a policy which combines user charges for secondary education with a bursary scheme might also be useful in some situations. In particular, such a policy could lead to a sizable expansion of secondary education without reducing equity; in fact equity might even be enhanced. Our discussion in this paper is organized as follows. In Section II, we examine the arguments for cost recovery in higher education. The -3- implementation of this policy via loan schemes is considered in Section III. We illustrate in Section VI and V, their potential effectiveness as a cost recovery instrument, using data from Malawi. In Section VI, we draw from these Malawi-specific results, some general for LDCs. In Section VII, we provide a complement to the analytical discussions in the foregoing sections by considering a commonly-cited set of objections to the implementation of loan schemes in LDCs. Finally, Section VIII concludes the paper. II. The Case for Cost Recovery in Higher Education A policy of cost recovery in higher education and its potential usefulness can be assessed according to three considerations. First, what is the likely impact of such a policy on resource mobilization; second, would this policy be equitable and efficient; and third, would it be feasible to implement? As to the first consideration, let us first recall some of the current characteristics of the education sector in developing countries. Table 1 shows, for example, that the gross enrollment ratio in primary and Table 1: Public Expenditure on Education and Enrollment Ratios, By Region, Circa 1980 Public Expenditure on Education as % of Gross Enrollment Ratio Total Public Region Budget GNP Primary Secondary Higher Francophone Africa 22.6 4.9 46 14 2.4 Anglophone Africa 17.0 5.2 77 17 1.2 South East Asia & Pacific 15.0 4.2 87 43 9.1 South Asia 8.8 2.1 71 19 4.4 Latin America 16.9 4.5 90 44 12.0 Developing Countries 16.1 4.5 75 23 6.9 Developed Countries 9.0 4.9 107 80 21.0 Source: World Bank (1983) and Unesco (1983) -4- secondary education is still relatively low in some parts of the world. It is clear that additional resources are needed to expand the provision of basic education. However, since public expenditure on education is already quite high, these resources might have to be mobilized from within the education sector itself. In this text, we will consider the shifting of public resource currently allocated to higher education. In their study of ten African countries, Mingat and Tan (1984b) have shown that, potentially, a substantial amount of resources could in fact be mobilized. Although the situation differed from country to country, their results show that the public resources released through a reduction in higher education subsidies would permit an increase of between 20 and 40 percent in the provision of primary school places at the current level of unit costs. Alternatively, if these resources are channelled into secondary education, the result would be an expansion of between 35 and 75 percent in this level of education. The magnitude of these effects underline the potential importance of cost recovery in higher education as a policy option. Its importance is strengthened when we consider the equity and efficiency arguments for reducing the level of subsidies at this level of education. II.1 The Ecuity Argument In general, equity is aggravated when more is given to those who are already in advantaged situations. In this regard, it is useful to distinguish two issues. The first concerns the distribution of public resources for education among persons belonging to the same generation. The second concerns the personal and socio-economic characteristics of the individuals who succeed in appropriating a share in these resources. We shall examine separately these two facets of equity in education. -5- (i) The Distribution-of Public Resources for Education A basic fact is that the higher the level of education, the higher the (absolute and relative) level of public subsidization. To document this fact, let us begin by decomposing into two parts, the unit cost of education borne by the government: (a) subsidies to cover the operating cost of teaching; and (b) scholarships, stipends and free or subsidized student welfare services (residence halls and cafeterias). In all countries, the unit operating cost increases with the level of education. Commonly, only a small proportion, if any, of this cost is recouped via tuition charges. As a result, public subsidies to cover operating costs tend to increase with the level of education. In the case of the second type of subsidies, the picture is less clear, since the amount and availability of such subsidies vary across the countries. Table 2 shows the magnitude of these subsidies in Africa and Asia relative to the public budgets for secondary and higher education. The data indicate clearly that the subsidization of students' living expenses is substantial Table 2: Share of Scholarships and Welfare Expenditures in Secondary and Higher Education Budgets Secondary Higher Region Education Education (%) (%) Francophone Africa 23.0 43.0 Anglophone Africa 14.0 14.0 Asia 4.0 4.0 Source: Acharya (1982) -6- in Africa, especially in the Francophone countries. _ Taken together, the two components - operating cost subsidies and welfare subsidies - give the overall public unit cost of education. The data in Table 3 show the regional averages in the level of this cost Table 3: Public Unit Cost of Education, by Major Regions of the World, circa 1980 . Unit Cost in Terms of GNP/Capita Region Primary Secondary Higher Africa Anglophone 0.18 0,50 9.20 Francophone 0.29 1.43 8.04 Asia South Asia 0.08 0.18 1.19 East Asia & 0.11 0.20 1.18 Pacific Latin America 0.09 0.26 0.88 'Middle East & 0.12 0.28 1.50 North Africa Developing 'J.14 0.41 3.70 Countries Developed 0.22 0.24 0.49 Countries Source: World Bank (1984) and Unesco (1983) 1/ There are of course historical reasons for the heavy subsidization of higher education in Africa. To the extent that the colonial administrations neglected higher education, there was in general an acute shortage of suitably qualified nationals when countries in this region became independent. The provision of scholarships and tuition- free higher education was seen as a way to encourage a rapid increase in the supply of graduates to replace expatriates in all sectors of the economy, including especially public administration. This pattern of financing has had an important impact on the development of higher education in the former colonies. In this respect, it is interesting to compare the Anglophone and Francophone countries. The subsidization of students' living expenses has been more important in the Francophone countries (see Table 2); correspondingly, higher education in this region is today twice as developed as in the Anglophone countries. (See Table 1) The shortage-of-qualified-nationals argument is, however, much weaker today than it was one or two decades ago. Indeed, in the changed circumstances in many African countries today, it is no longer a valid justification for the continued heavy subsidization of higher education. One reason is that the scarcity of graduates is not as acute as before. In some countries, it may even be argued that there is in fact an oversuppLy of graduates in some fields of specialization. -~~7- for tb three levels of education. In developed and developing countries, the pattern is clearly one in which the unit public cost increses with the level of study. The rate of increase is, however, much more rapid in the developing countries. For the reasons cited earlier, these characteristics are most prominent in the African countries. Mingat and Tan (1984c) have shown that this cost structure, combined with the enrollment patterns described in Table 1, leads to a very unequal distribution of public resources for education among individuals belonging to the same generation. Table 4 presents some of their results. Table 4: Distribution of Public Resources by Terminal Level of Schooling, Major Regions, circa 1980 No Schooling and Primary Higher Education % Popu- % Popu- GNP/Capita Gini- Region lation % Resource Lation % Resources USS Coefficient Africa Anglophone 83 39.0 1.2 26.4 450 0.567 Francophone 86 15.7 2.4 39.5 410 0.823 Asia South Asia 81 23.2 4.4 39.0 180 0.646 East Asia 57 19.3 9.1 39.6 1260 0.499 & Pacific Latin America Middle East & 64 18.8 9.4 44.9 1370 0.571 N. Africa Developing 77 19.6 6.9 56.2 - 0.697 Countries Developed 20 8.3 21.0 36.7 10000 0.218 Countries Source: Mingat and Tan (1984c) In the developing countries as a group, 77 percent of a given cohort (those with primary or no schooling) share only 19.6 percent of the overall cohort resources, whereas 6.9 percent (those with higher education) get 56. percent of those resources. In some regions of the world, the pattern of distribution is even more skewed. In Francophone Africa, for example, 86 percent of the cohort receive only'15.7 percent of the resources, compared the 39.0 percent share going to 2.4 percent of the cohort who leave the education system with higher education. These distributional patterns are summarized by the Gini-coefficient. According to it, the most equitable distribution of resources among the developing countries occurs in the Latin American and East Asian and Pacific countries. The outcome is slightly less favorable in Anglophone Africa and the Middle East and North Africa. Lower down in the ranking is South Asia. The least equitable distribution of public education resources prevails in the Francophone countries in Africa. In the same study, it has also been shown that cost recovery activities in higher education are likely to lower substantially the inequity described above. Table 5 and Figure 1 document the impact on equity if full cost recovery is implemented in higher education and if the public resources thus freed are used to expand the provision of education at the lower levels. -9- Table 5: Gini-Coefficient Before and After a Shift of Resources from Higher Education Towards Lower Levels of Education, Major Regions of the World, circa 1980 Gini-Coefficient Region Before Shift After Shift Africa Anglophone 0.567 0.270 Francophone 0.823 0.447 Asia South Asia 0.646 0.320 EAst Asia & Pacific 0.499 0.153 La in America 0.498 0.205 Mi lie East & N. Africa 0.571 0.225 Developing Countries 0.697 0.278 Developed Countries 0.218 0.039 Source: Mingat and Tan (1984c) Gial 9.9 - -_ __I_ 0.9 __ _ -- -- a *ranwphow Afrran 0.7 _os. South~ 0 .6 As Uia h +a A.glo6ph ne m Nidd e East N. Africa 0.5 Africa Lati America East Asa 0.4 0.3 _ - - _ O _ ^ ^ ~~~~~Adv nc 0.2 Counte 0.1- -- 0 * After 100 200 300 500 700 1000 1500 2000 10000 Per Capita GN? tW5) (log scale) Fig.l; Gini-cofficient bY Per Cafitat _P. Sefor. and After a Shift of Public Resouri o rL o uOn, Ma~r ~ori-d RtagioiCa, -cica 1-980 - 10 - To summarize, when the distribution of resources component of equity is considered, i. is clear that cost recovery policies in higher education are a potentially powerful instrument for improving overall equity in the education system. Let us now examine the second component of equity, namely, the issue of who benefits from the public expenditure on education. (ii) Who Benefits from Public Expenditure on Education? Or The Equity of Access to Education A classical criticism of cost recovery policies is that they will reduce access to education at the level at which they are implemented. The implied increase in the private cost of education may cause some students to give up their studies. Since students from poorer families are more likely to drop out, it follows that the equity of access to education would deteriorate. Applied to higher education, this argument against cost recovery activities is, however, much weaker that it appears at first sight. There are two reasons why. First, if equity in the access to education is important, it is clear that the potential "loss of equity" within higher education has to be weighed against the potential "gains in equity" resulting from the expansion of elementary education when scarce resources are shifted accordinglv. Second, if we focus on the equity of access within higher education itself, some precautions can be taken to reduce, if not eliminate altogether, the potential adverse equity effects. We shall examine these two points in greater detail below. - II - a) A Trade-off Exists Between Losses and Gains in the Equity of Access to Education In general, but more particularly in LDCs, students in higher education tend to come from the upper socioeconomic strata of society. A large proportion of them are the children of civil servants and salaried employees in the modern sector, and, in general, the urban sector (Anderson, 1983). Whatever the truth of this statement, it is still possible that some students in higher education would not be able to pay the full cost of their education due to their current family circumstances. On the other hand, it is very likely that those who would gain from a shift of public resources from higher to lower levels of educationl/ are even more economically deprived. In this situation, it is clear that we might be confronted with a trade-off between less equity of access to higher education and more equity of access to primary or secondary education. b) The Loss of Equity of Access to Higher Education can be Mitigated In the previous paragraph, we have assumed that the increase in the private cost of higher education would inevitably hurt students from modest socio-econoaic backgrounds. This outcome could in fact be circumvented. First, a selective bursary fee exemption scheme could be used to mitigate the financial difficulties faced by needy students. Second, loans could be provided to help them transfer the burden of financing their education to a later time when their earnings would allow them to repay the loan. The argument for loans is in fact quite strong. This is because education ac this level represents an important 1/ That is, those who are not currently enrolled. - 12 - accumulation of personally-embodied capital which will produce good dividents in the form of high earnings during a graduate's working life. From this rather lengthy discussion, it follows that cost recovery in higher education, coupled with the use of the public resources thus freed to expand lower levels of education, would probably result in an unequivocal enhancement of equity in the education sector as a whole. The outcome would clearly be even better if a loan scheme or a selective bursary scheme were organized in conjunction with the fee increase in higher education, since these schemes would reduce the financial hardship potentially faced by (a small number of) higher education students from needy family backgrounds. Let us now turn to examine the efficiency arguments for reallocating the current public subsidies for higher education towards lower levels of education. II.2 The Efficiency Argument The data relevant to this argument appear in Table 6. They show .he pattern of private and social returns to the different levels of education. For the developing countries as a group, the social rate of Table 6: Private and Social Rates of Return, by Region and Type of Country Primary Secondary Higher Region Private Social Private Social Private Social Africa 29 29 22 17 32 12 Asia 32 16 17 12 19 11 Latin America 29 44 20 17 23 18 All Developing Countries 29 27 19 16 24 13 Intermediate Countries 20 16 17 14 17 10 Advanced Countries - - 14 10 12 9 Source: Psacharopoulos (1981) - 13 - return is 27 percent for primary education, but only 13 percent for higher education. In this situation, an increase in investment in primary education would probably lead to efficiency gains. Under the current financial constraints in the public budget, cost recovery in higher education is one way of mobilizing resources for this purpose. The data in Table 6 also illustrate the growing discrepancy between the private and social rates of return as the level of education rises. This result reflects the common tendency for the public subsidization of education to increase with the level of study. In the developing countries, the private rate of return is higher than the social rate by 7 percent at the primary level; 16 percent at the secondary level; and 38 percent in higher education. The increasing divergence between the two rates is especially dramatic in the African countries; the private rate of return to higher education, for example, exceeds the corresponding social rate by 167 percent. To the extent that individuals are motivated by the private returns to education in making their educational choices, the current pattern of subsidization encourages socially inefficient choices. In other words, it is partly responsible for the prevailing imbalance in the enrollments at the different levels of education, and for the relative underinvestment in primary schooling. Within higher education itself, students' educational choices are also influenced by the relative returns to different fields of study. By increasing cost-consciousness among students cost recovery policies are likely to encourage students to choose more responsibly among their educational options. In other words, students would have greater incentive to behave more like "investors" than "consumers and to pay greater attention to labor market signals in choosing their field of specialization. In general, these - 14 - general, these tendencies would help to improve the linkage between education and development. Furthermore, the implied increase in private costs could be expected to improve internal efficiency as students of inferior academic ability would then be less likely to enroll. The foregoing efficiency arguments are based largely on the rate of return calculations. It is sometimes pointed out that the rate-of- return calculations neglect the existence of externalities and that this neglect results in a downward bias in the estimated returns. However, this problem would invalidate a comparison based on the unadjusted estimated returns only when we have reason to believe that the externalities are greater at higher than lower levels of education. Even in this case, the difference would have to be exceptionally large. The reason is linked to variations in the unit cost of education at the different levels of education. In the LDCs as a group, for example, the unit cost of higher education is 26 times that for primary education. Given this statistic, the problem discussed above becomes important only if it can be shown that che externalities generated by one person with higher education is greater than those generated by 26 persons with primary education. In summary, the arguments for cost recovery in higher education can be stated as follows. First, whatever the historical reasons for the heavy subsidization of higher education, they are no longer valid in the changed economic and social circumstances in most LDCs today. Given constraints in the public budget, the continued subsidization of higher education at existing levels implies a significant loss in the development of lower levels of education. Second, the abolishment of subsidies to higher education would improve equity in two ways: (i) it would avoid the reinforcement of privileges that students in higher education already have - 15 - in terms of family background and future earnings potential; and (ii) it would enable more students to share in the limited pool of public resources for education. Third, the case of cost recovery could also be supported on grounds of efficiency, as argued above. Having stated the arguments for requiring students in higher education to share more fully in financing their education, we shall now examine how this policy could be implemented via student loans. III. Loans as a Mechanism for Cost Recovery As indicated earlier, concern over the equity of access to higher education leads us to consider two alternative approaches for implementing cost recovery in higher education. In one approach, the increase in the private cost of education is combined with the provision of scholarships and/or fee exemptions for a limited number of needy and deserving students. In the second approach, the potential adverse effect on equity is mitigated via the provision of loans; in this case, the recipients need not necessarily be limited to selected students. In the scholarship and fee-exemption scheme, the maximum proportion of the social cost that can be recouped clearly depends on the ability and willingness of individuals to pay for their higher education. However, even if we can expect the wealthiest students to bear the full social cost of their education, this scheme would permit only a partial recovery of costs. In the loan scheme, on the other hand, full cost recovery is theoretically attainable. In reality, however, the outcome may also be limited, since it depends on the practical feasibility of implementing the terms of repayment required for full cost recovery. The scholarship strategy for mitigating the possible adverse equity effect of cost recovery policies has been documented by Mingat and Tan (1984a). In this paper, we shall concentrate on the feasibility of implementing these policies via loans. - 16 - In any given situation, loan scheme can be characterized by several inter-related features. However, once the initial cost of higher education to be financed is known, and once the market rate of interest is given, the two key features are the duration of repayment, and the amount repaid at each installment. Between these two key variables, there is a trade-off relationship: the larger the amount repaid each time, the more quickly the initial loan would be repaid. To the government, a short duration of repayment is desirable since it means that the loan scheme would become self-financing more rapidly. To the student, however, a short duration of repayment may imply an insupportable debt burden. This difficulty could be exacerbated by the possibility of temporary unemployment or extended job search activity in the period immediately following graduation. Some loan schemes accommodate this problem through the provision of a grace period. This feature is important because it allows the borrower to settle into a job before starting to repay the loan. Clearly, however, the longer the grace period, the more difficult it 4ould be to achieve a high rate of cost recovery. As such, the impact of the grace period could be quantified in terms of (i) the increase in the amount repaid at each installment; or (ii) the increase in the duration of repayment; or (iii) the loss in the proportion of costs recovered. Regarding the repayment burden, there are two ways of examining the likely feasibility of alternative scenarios. The first would be to take the classical position of a banker who makes a loan. Given the repayment period and the market rate of interest, a schedule of fixed repayments can be determined so that their total net present value equals the initial amount of the loan. Following this calculation, the banker then evaluates the repayment burden against the borrower's earnings to decide whether or not the borrower can be expected to repay. The second -17- way of examining the problem is to start from the borrower's earnings. Assuming that a given proportion of this income is devoted to repaying the loan, it would be possible to derive the net present value of all the installments paid and to assess the rate of cost recovery. For our purposes here, the second approach is probably better since we are interested more in the extent to which constraints in the repayment burden lower the cost recovery potential of loans, than in whether or not students are low-risk borrowers. To summarize the foregoing discussion, we can use the following expressions: n t (1) PVL C 2 Ct/(l+r) t=O n+l (2) PVR = Z: a.Et/(l+r)t where PVL and PVR are the present values of the loan and repayments respectively; Ct is the cost of study in year t 1/, n is the number of years in the course of study; r is the rate of interest; Et is the borrower's earnings at year t ; a is the proportion of earnings devoted to loan repayment; and a is the duration of loan repayment. The question that interest us can then be stated as follows: (i) given the cost of education (Ct), earnings (Et) and the rate of interest (r), what are the different sets of (a , 3) - proportion of earnings allocated to repayment and duration of repayment - for which there is full cost recovery (that is PVL = PVR); and (ii) for any given set of (a ,a ), what is the proportion of PVL that will actually be recouped? Further, what would this proportion be if "reasonable" values of a and 3 are assumed? I/ In the calculations below, we assume that the annual cost of study is the same for all years in course of study. - 18 - To answer these questions, we need data on costs and earnings. We shall use data from Malawi to illustrate the analysis. We shall first consider some features of returns to education in this country; thereafter we shall consider the role of loan schemes in financing higher education, given these country-specific conditions. Following this analysis, we shall consider the likely usefulness of loan schemes in other LDC contexts; in paricular, we shall assess the extent to which differences in the cost of education (Ct) and the level of graduate earnings (Et) are likely to alter the conclusions drawn for Malawi. IV. Some Features of Returns to Education in Malawi To calculate the returns to education, we used the earnings data from the 1981 Household Income and Expenditure Survey. These have been supplemented by data on the earnings of civil servants (Malawi government, 1983). On the cost side, our data on unit private costs were derived from the 1983 Malawi Education Finance Study while those on unit government- expenditures were obtained from the official source cited above. Our calculation of the rates of return is based on the "elaborate method" (Psacharopoulos, 1981). Since the available earnings data are in tabulated form, they were first smoothed to obtain the age-earnings profiles. For ages above 25 years, these profiles were obtained by fitting a cuadratic function to the data. For the earlier ages, we have used the moving average method. Thie reason for this preference is purely practical, and is based on the noticeably systematic discrepancy between the regression predicted profile and the actual earnings at the earlier ages. Table 7 shows the main features of the earnings and cost data used in our calculations. The estimated rates of return appear in Table 8. These results are contrasted in Figure 2 with those for the African countries in general. - 19 - Table 7: The Private and Social Direct Costs of Education and Earnings by Level of Education, Malawi, Circa 1982 (Kwacha p.a.) Educational Direct Costs 2/ Earnings 4/ Level 1/ Private Social Starting After 10 years No Schooling a0 110 200 Primary(8) 41 53 220 440 Lower Secondary(2) 230 405 560 960 Upper Secondary(2) 230 405 650 1510 University(4) 0 3/ 3936 3260 4800 1/ The figures in parentheses represent the number of years in the corresponding cycle of education. 2/ For details, see Tan, Lee and Mingat (1984). 3/ In Malawi, university students receive grants to cover almost all expenditures for materials, books, room and board. Thus in theory, there are no private direct costs. 4/ See text for data sources. Table 8: Estimated Returns to Education, Malawi, Circa 1982) Educational Beginning Rate of Returns (%) Qualification Working Age 1/ Private Social Primary 7 11.6 11.1 8 12.1 11.6 9 12.8 12.2 10 13.6 12.9 11 14.6 13.7 12 15.7 14.7 Lower Secondary - 26.3 21.2 Upper Secondary 16.8 15.2 University 46.6 11.5 1/ Beginning working age of those without schooling. This assumption affects the calculation of the rate of return to primary schooling only. - 20 - Rate of Return 50 Malavi ,PRR 40 Africa,PRR 30 20 -o' r=a,^; > _^x_ Africa,SRR 10 _Zalavi,SRR Primary Lover Upper Higher Level of Secondary Secondary Education Education FRig. 2: The Social (SRR) and Private (?RR) Rates of Return to tducation, by Level, in Malavi and Africa. The results for Malawi reflect some features of the general pattern of returns to education in developing countries, particulary those in Africa; The most important feature is that the discrepancy between the social and the private rate of return tends to increase with the level of education. As we have highlighted in Section II of this paper, this results arises from the "perverse" pattern of subsidization. It suggests that there are efficiency and equity reasons for advocating greater cost recovery in higher education in Malawi. In terms of the social returns by level of education, the results for Malawi differ sightly from other LDCs. In particular, the social rate of return is highest for secondary rather than primary education, unlike the opposite patterns commcnly found. This feature of the Malawian situation is probably explained by the particular enrollment structure of the educational system in this country. - 21 - Table 9 shows, for example, the enrollment ratio at the different levels of schooling in Malawi compared to the regional averages. Table 9: Enrollment ratio by level of education in Malawi and in Sub-Sahara Africa, Circa 1980 Gross Enrollment Ratio (%) Relative Ratio Country/Region Primary Secondary Higher (2)/(1) (3)/(2) (1) (2) (3) Malawi 62 4 0.4 0.06 0.10 Anglophone Africa 77 17 1.2 0.22 0.07 Francophone Africa 46 14 2.4 0.30 0.17 Source: See Table 1. It is clear that in Malawi the enrollment in secondary education is particulary low. Mingat and Tan (1984a) have shown that this feature corresponds in reality to a situation in which the supply of public secondary school places is relatively restricted, while there exists a large unsatisfied demand for this level of education. As a result, there is a scarcity of secondary school graduates on the labor market; this scarcity is reflected in their relatively high earnings, and hence the relatively high rates of return to their education. These circumstances suggest that some priority should probably be given to the expansion of secondary education. The resources freed from greater cost recovery in higher education could, for example, be directed, at least partly, to finance this expansion. Whatever the use of these resources, however, it appears clear that there should be cost recovery for higher education in Malawi. As we now have at our disposal the relevant factual information, we shall examine next the extent to which a loan scheme could serve as an instrument towards this objective. - 22 - V. The Potential Usefulness of a Loan Scheme in Malawi Given that we have the data on the unit cost of education (operating cost and living expenses) and the average age earnings profile of graduates, it appears relatively straightforward to determine values in the set ( a, 3) that would permit full cost recovery. The calculus is, however, slightly complicated by two considerations: (i) the rate of interest charged for the loans; and (ii) the length of study and the student flow characteristics within higher education institutions. We shall discuss these points in turn. As we are interested in evaluating loans as an instrument for cost recovery, it is clear that we should use the market rate of interest to discount the flow of repayments.1/ Here, we note that the real rate of interest should be used, since the cost and earnings data are expressed in real terms. Although this interest rate is likely to vary from country to country, it is probably less variable than its nominal counterpart. During the 70s, it was very low, if at all different from zero percent. By 1983, however, had risen to around 3 to 5 percent in Europe, and around 4 to 6 in the U.S. (IMF, 1984). It would thus appear reasonable to assume a real 1/ This does not mean that the markec interest rate would necessarily be the one actually charged on the loans. In fact, the government might be willing to subsidize the interest rate so as to reduce the debt burden on students. In this case it is implicitly assumed that only part of the loans' initial value will be recovered. The subsidization of the interest rate can therefore be viewed as an implementation strategy; if only partial recovery of costs is envisaged, the same result could equally be achieved by requiring students to pay back part of the loan at the market interest rate. Analytically, we need only consider one of these two methods of implementation. We shall in fact focus on the first since the results can be more easily interpreted for our purposes. - 23 - market interest rate of 5 percent in our calculations below. We shall, however, assess the sensitivity of the results by varying this rate in the O to 10 percent range. Regarding the student-flow characteristics, two features are pertinent in our discussion. First, it is likely that because of repetition some of the students would take more thani the "normal" (that is, minimum) number of years to complete their program of study, for example, five or six years instead of four. Second, some students may drop out after one or two years of unsuccessful study. If these features are not explicitly allowed for in assessing the cost recovery potential of loans, the results are likely to be overly optimistic. Unfortunately, we do not have reliable data on the proportion of students belonging to the various "survival groups". Even if these data were available they would probably be useful only marginally. The reason is that the existing student-flow characteristics are likely to change in the context of a loan scheme. The corresponding increase in the private cost of higher education is likely to give greater incentive to students to succeed in their studies. As a result, repetition and dropout rates would probably be lower than before. Whatever the effect of cost recovery on repetition and dropout, it is clear that our analysis would be incomplete if these two feature of the student-flow are neglected. In view of the lack of'data, however, we shall address this problem through sensitivity analysis. First, we shall present the results based on the assumption that students neither repeat nor drop-out. The results are clearly upper bound estimates of the potential effectiveness of loans in recovering costs. We shall then test the sensitivity of these optimistic results to less favorable student-flow characteristics. - 24 - V.1. Upper Bound Estimates Figure 3 depicts the locus of different combinations of a (proportion of income allocated to repaying the loan) and 9 (duration of repayment) which imply full cost recovery. The calculations assume that the duration of study is four years for all students. Three levels in the real rate of interest are considered, namely zero, five, and ten percent. The solid curves represent the outcome in the absence of any grace period, that is students begin to repay their loans upon graduation; the dotted line corresponds to the case where a grace period of three years is allowed, assuming a real interest rate of 5 percent. 1 of Income Allocated to Repay Loan (ct) 60 _ \ _ \ __i - No Grace Period 50 -9 ' - 3 Years Grace Period 40 ' \ X t ~ e _ r .10X 0 5 10 15 20 25 30 Duration of Repayment,years (a) Fig. 3: Sets of (na,) for Full Cost Reccverv, given the Real Rate of Interest (r), With and Without Grace Period. Malawi. - 25 - The figure shows, for example, that to achieve full cost recovery at a five percent real interest rate, graduates would have to allocate 50 percent of their current income for 11 years of repayment; or 30 percent for 18 years; and so on, if there is no grace period. It can be seen that the three years grace period exerts a relatively minor impact on the results. The reason is because with the postponment of repayment, the absolute size of the installments repaid also increases, given the usual characteristics of the age-earnings profile. Despite this result, however, it is useful to bear in mind that loan schemes take longer to become self-financing when a grace period is permitted. To evaluate the potential effectiveness of a loan scheme, some judgement needs to be made about the maximum acceptable values of two parameters: the duration of repayment and the size of installments repaid as a proportion of the average graduate's current earnings over this repayment period. For practical reasons, it appears that a 10 year repayment period might be considered as a maximum. As to the second parameter, however, it is less clear where the line should be drawn. In some countries, banks set an upper limit of one-third of income for mortgage amortization purposes. It would thus appear that a 25 percent limit for student loan repayment would be quite bearable. The acceptability of this upper limit is reinforced when we consider the earnings of students by level of education. In Malawi, the average earnings of higher education graduates are around K4000 p.a. over the first ten years of working life. The corresponding figure for a secondary graduate is about K1500 p.a. Thus, if 25 percent of the earnings of a higher education graduate is devoted to loan repayment, the remaining income, K3000 p.a., would still be much higher than that of a secondary school graduate. - 26 - Quite obviously, however, it is not our role as economists to produce here an assessment of what might be politically or socially acceptable in a given situation. Such an assessment requires more than a technical analysis. What is clear, however, is that even if the maximum size of repayment is set at 25 percent of current income, full cost recovery would not materialize in a 10 year repayment period, given the conditions in Malawi. If this maximum is lowered, say to 20 or 15 percent, the performance of the loan scheme would inevitably worsen. Figure 4 summarizes the range of outcomes regarding the proportion of the initial loan amount that would be recovered corresponding to a variety of repayment schedules; a five percent real interest rate has been used in plotting the points in the graph. Rate of Without With 3 Years Cost Grace Period Grace Period Recovery a.602 ct-50% 40% o-30% -25% 190 0t/X 70 60 50 40 ___ __ ___ __ 30 20 _ 10 0 5 :0 ,5 20 25 30 Duration of Raaient, yeass Fi. A4 Rate of Cost Recovery, According to the Procortion of Income Allocated to Repay Loan Ca), and Our&tion of Rzovament, Assuming a 5S Real interest Rate. Malawi. 27 The figure shows clearly that in Malawi, the proportion of costs recovered would be negligible if, in a 10 year repayment period, the proportion of income allocated to loan repayment (c ) is 10 percent or less. If the repayment amount is 15 percent of current income, the rate of cost recovery would be 28 percent. This rate would rise to 36 percent if 20 percent of the current income is allocated to loan repayment; and 45 percent if 25 percent of current income is allocated. Obviously, the outcome would improve if the duration of repayment is extended; for example, if it is 15 years, 26 percent of the initial loan amount can be recouped when 10 percent of income is allocated to loan repayment. The proportion recouped would increase to 42 percent when the repayment amount is 15 percent of income; and 52 percent if this amount is 20 percent of income. Here again, it is an open question whether or not a 15 year repayment period would in fact be acceptable. The calculations indicate, however, that in Malawi, a repayment period shorter than ten years would not lead to a high rate of cost recovery, unless, of course, the repayment amount is a very high proportion of current income. On the whole, the results show that it would be quite difficult to achieve a substantial rate of cost recovery in Malawi via loan schemes. In other words, only a relatively limited range in the loan repayment conditions would assure the recovery of a reasonably high proportion of the initial loan amount. This pessimistic conclusion is reinforced when we recall that the foregoing calculations have been based on the assumption that students obtain their degrees in the minimum time, without repetition and dropout problems. Let us now examine the impact of these student-flow characteristics on the effectiveness of the loan instrument. - 28 - V.2. The Effect of Repetition and Dropout on the Performance of the Loan Scheme We shall analyze separately the effect of dropout and repetition. For dropouts, the problem is that they will borrow to finance one or two years of study without getting any certification. To the extent that they have made some effort to'acquire knowledge during this time, they might have some advantages over "ordinary secondary school graduates". If the labor market rewards these advantages, their earnings are likely to exceed those of secondary graduates. We shall, however, assume that the earnings in these two groups are the same. Under this conservative assumption, we have simulated the proportion of the initial loan recouped corresponding to various combinations of duration of repayment and proportion of income allocated to repayment. The results, which appear in Table 10, show that the level of cost recovery achieved under various scenarios is comparable to the earlier results for students who complete their studies in four years. Table 10: Proportion of the Initial Loan Recouped, Given Various Combinations of ( a , a ) --- The Case of Dropouts 1/ % of Initial Loan Recouped Repayment % Current Income Devoted Dropout Dropout Period (Year) to Loan Repavment After 1 Year After 2 Years 5 25 40 20 10 25 78 38 15 25 94 57 1/ a is the repayment period; and a , the proportion of current income devoted to loan repayment. - 29 - For repeaters, the problem is that they will take five or six years to graduate instead of the normal four. Correspondingly, they would need a larger loan to finance their studies. In order to achieve the same level of cost recovery as in the case of non-repeaters, these students would thus have to allocate a larger amount of their current income for loan repayment. Alternatively stated, if they allocate the same percent of income as the non-repeater graduates, the rate of cost recovery would be lower. This effect is illustrated in Figure 5. For example, if the loan repayment period is 10 years, and the proportion of income allocated is 25 percent, the cost recovery rate would be 36 percent in the case of students Rate of Cost Recovery 4 years to (S) graduate 5 years 100 90 _ -7_ ,;6 yeaTs 80 _ / 70 60 50 __________ l . 30 20 1;9 10 _ _ _ _ __ _ _ _ _ 0 5 10 15 20 25 30 Duraticn of Repaymlent, years Fig. 5: Rate of Cost Recovery to the case of Reveaters. Assuming the Prooortion of Income Allocated to Loan Reoavment is 25t. Real interest Rate is 5X, According to the Duration of Revayment, Mala.t. - 30 - who repeat one year, and 29 percent in the case of those who repeat two years, compared to 45 percent in the absence of repetition. Alternatively, the comparison can be given as follows: to recoup the same percentage of 45 percent within 10 years, it would be necessary for one-year repeaters to allocate 30 percent instead of 25 percent of their current income for repayment. If the percent of incomne allocated for repayment remains at 25 percent, the duration of repayment would have to increase from 10 to 13 years. In the case of two-year repeaters, the corresponding figures are 40 percent over 10 years, or 16 years if the percent of income allocated for repayment remains at 25 percent. To summarize, our calculations show that the incidence of dropout and repetition exert a significant effect on the performance of loans as a cost recovery instrument. The overall effect is, however, likely to be less dramatic, since it is weighted by the frequency of the various types of students. This outcome can be demonstrated as follows. Let us assume, for example, that the repayment period is 10 years, and that the amount repaid each year is 25 percent of current annual income. Further, let us also assume that, ex-post, 50 percent of the graduates complete their studies after four years; 30 percent, after five years; and the remaining 20 percent, after six years. Given these characteristics, the overall rate of cost recovery would be 40 percent, a figure which is not too much below the estimated 45 percent in the optimistic case where students neither drop out nor repeat. V.3. The Overall Effectiveness of loans as a Cost Recovery Mechanism in Malawi As we have emphasized, the cost recovery potential of loans depends to a large extent on policy makers' judgement about what might -31- constitute bearable terms of repayment. In the case of Malawi, we have attempted to show that even if the amount repaid at each installment were as high as 25 percent of graduates' current income, the debt burden would probably remain bearable. Over a 10 year repayment period, the corresponding rate of recovery, after allowing for the incidence of repetition and dropout, would be 40 percent. With lower proportions of income allocated to loan repayment, the percent of the initial loan amount recovered would drop as follows: with 20 percent of income allocated, the recovery rate would be 32 percent; with 15 percent, it would be 24 percent, and with 10 percent, it would be only 16 percent. On the whole, it is unlikely that student loans can be expected to recover fully the social cost of higher education in Malawi. If the terms of repayment are confined within what is likely to be judged as acceptable, the rate of cost recovery would, at the most, be around 40 percent. The conclusions drawn so far relate specifically to Malawi. We shall now examine the extent to which these conclusion are relevant in other LDC contexts. VI. The Likely Usefulness of Loan Schemes in Other LDCs The Malawi-specific computations can be generalized quite easily. The outcome for other country conditions depends primarily on two key factors: (i) the loan amount, that is, the unit social cost of higher education in the country; and (ii) the level of graduates' earnings. The latter is important because it affects the financial capacity of graduates to repay their loans. Both of these factors jointly determine the scale by which the Malawi-specific results should be extrapolated in order to derive estimates for other countries. For example, if the unit cost in a particular country is double that in Malawi, and the earnings of graduates the same, then a loan scheme would achieve only half the cost recovery rate -32 - in Malawi, ceterus paribus. If both the unit costs and earnings of graduates are double that in Malawi, the loan scheme would recoup the same proportion. of the initial loan amount in both countries, assuming again that the terms of repayment (a and a ) are unchanged. The scale factor is, therefore, the ratio between the earnings of graduates, and the unit social cost of higher education; this relationship holds exactly when the duration of study is as in Malawi, four years. Information on the unit cost of higher education is generally available. Although these data are known to be flawed (Eicher, 1984), they are reasonably adequate for the purposes here. Data on the earnings of graduates are more difficult to come by. One proxy that we might be able to use is the salary of secondary school teachers; these are available from calculations based on Tait and Heller (1984) and Wolff (1984). The earnings of higher education graduates is, however, probably higher than those of secondary school teachers. There are two main reasons why. First, salaries for university graduates tend, on average, to be lower in teaching than in other professions. Second, the average secondary school teacher's salary is a statistic weighted by the earnings of all teachers in secondary education; since in the context of LDCs, some proportion of these teachers are likely to have less than the full four years of higher education, this statistic is an underestimate of the earnings of higher education graduates. In the absence of better data, we shall nevertheless use this proxy. In fact, since we are mainly interested in the relative levels of graduate earnings in other countries compared to Malawi, the use of this proxy is not likelv to be damaging. For this purpose, our only assumption is that, across countries, there is a reasonably stable relationship between the earnings of higher education graduates and those of secondary school teachers. - 33 - Table 11 displays the data on unit costs and secondary school teacher salaries in Malawi and in various regions of the world. The data for both variables are expressed in relation to per capita GNP. With regard to the potential cost recovery impact of loan schemes, it is instructive to look at the figures in the last column of the table. They indicate that the outcome in Malawi is less favorable than in the average African country, and very much less so compared to Asian and Latin American countries. This result is due mainly to the exceptionally high unit cost of higher education in Malawi. In Francophone Africa, the favorable outcome is a reflection of the high level of graduates' salaries which enhances their debt repayment capacity. In Latin America and Asia, the earnings in terms of per capita GNP are lower than in Malawi, but the unit cost of higher education is even smaller; these conditions increase the possibilities for implementing a loan scheme to recover the public cost of higher education. Table 11: Unit Costs in Higher Education and Secondary School Teacher Salaries in Malawi and Major Regions of the World, circa 1984 I/ Unit Cost Secondary Salary-Cost Ratio (2)/(1) in Higher Teacher Relative to Country/Region Education Salaries Absolute Malawi (1) (2) Malawi 19.8 12.0 0.61 1.00 Anglophone Africa 9.2 10.0 1.09 1.79 Francophone Africa 8.0 19.4 2.43 3.98 Latin America 1.2 3.4 2.83 4.64 Asia 0.9 4.0 4.-44 7.28 1/ The data in columns (1) and (2) are measured in terms of per capita GNP. Source: Based on calculations from data in Unesco (1983), Tait and Heller (1984) and Wolff (1984). - 34 - Let us now examine what can in fact be expected of loan schemes as an instrument for cost recovery in the various regional contexts. In each region, we can compute the proportion of the initial loan amount that is likely to be recouped by multiplying the results for Malawi by the corresponding salary-cost ratio, as explained earlier. Table 12: Percent of the Initial Loan Amount Recovered Under Various Various Scenarios, Malawi and Major Regions of the World 1/ (Percent) % of Current Income Allocated to Repay Loan Country/Region 5% 10% 15% 20% 25% Malawi 8 16 24 32 39 Anglophone Africa 15 29 43 57 70 Francophone Africa 32 64 96 126 155 Latin America 37 74 112 146 181 Asia 58 11 175 230 284 1/ As explained in the text, we assume a 10 year repayment period and a five percent real rate of interest for all these calculations. The various scenarios thus correspond to different values of a, the proportion of income allocated to loan repayment. Our calculations assume that the regional student-flow characteristics are comparable to those we have assumed for Malawi 1/. Also, they relate to loan schemes with the following features: ten-year repayment period, and a five percent real rate of interest on the loan. The results appear in Table 12. They represent the overall picture across large geographic regions. 4lthough variation across countries is likely, the figures nevertheless provide a rough idea as to the potential effectiveness of loan schemes in the various regions. - 35 - As we do not know what the maximum bearable debt burden is in each regional context, our computations simulate a range of outcomes corresponding to various values of a (proportion of income allocated to discharge debt). For example, if it is judged that the maximum bearable repayment per year is only five percent of the borrower's current annual income, the implied rate of cost recovery would range from 15 percent in Anglophone Africa to 58 percent in Asia. Since this maximum is likely to depend on social and political considerations, we have no analytical basis to indicate what would be appropriate in particular country or regional contexts. What is clear, however, is that in some regions a high rate of cost recovery can be attained with a relatively low repayment burden. In Asia, for example, a ten percent allocation of income to loan repayment assures full cost recovery. The corresponding figure of 117 percent in the table can be interpreted in one or all of the following ways: (i) full cost recovery requires a smaller allocation from income for loan repayment than the ten percent assumed; or (ii) full cost recovery can be achieved with a repayment duration shorter than ten years; or (iii) graduates would have the financial capacity to repay a larger loan. The loan amount assumed in these simulations is actually the current public unit cost of education. For Asia and Latin America, these data do not include the cost of student living expenses, whereas they do for Francophone, and, to a smaller extent, for Anglophone Africa. Thus, for the former two regions, where the simulations indicate a recovery rate exceeding 100 percent, the loan amount could in fact be increased to enable students to meet some of 1/ This assumption is unlikely to bias our results seriously, since our earlier results suggest that the incidence of dropout and repetition exert a relatively minor effect on the potential effective performance of loan schemes. - 36 - their living expenses. The larger loan amount, if made available selectively to poorer students, might be important as an additional feature to widen access to higher education. Overall, the data indicate that loans would be effective as an instrument of cost recovery in Asia, Latin America and, to a lesser extent in Francophone Africa. The assessment for these regions to achieve full cost recovery do not appear to be too stringent. We note, however, that the favorable assessment for Francophone Africa hinges on the relatively high level of graduates' earnings. Although these high earnings improve the financial capacity of students to repay their loans, and hence the potential effectiveness of loan schemes, they nevertheless create severe problems for the education sector as a whole. In particular, through their link with the salaries of school teachers, the high earnings of graduates increase the unit cost of education at the lower levels of study 1/. Correspondingly, the expansion of primary and secondary schooling is, from a fiscal point of view, less easily accomplished. Here, it is useful to recall that the ultimate purpose of cost recovery in higher education is precisely to mobilize resources for expanding these levels of education. As such, we might view the potential effectiveness of loan schemes as the positive side of a costly feature in the educational systems of countries in Francophone Africa. 1/ Mingat and Psacharopoulos (1984) show, for example, that the unit cost of primary and secondary education are 29 and 143 times, respectively, of the average per capita GNP in Francophone Africa; the corresponding figures for Anglophone Africa are 18 and 50; and for other developing countries, they are 14 and 41. - 37 - Among the four regions, the prospects for cost recovery via loans are least promissing in Anglophone Africa. Nevertheless, they would still permit the government to shift onto students a non-regligible part of the current public cost of higher education. Their relatively low potential effectiveness suggests, however, that a selective scholarship and fee exemption scheme might be a more appropriate cost recovery instrument. The choice becween these instruments clearly deserves to be analyzed, but since it is beyond the scope of this paper, we shall leave it as an open question here. VII. Some Objections to the Implementation of Loan Schemes in LDCs In the foregoing discussion, we have shown analytically that loan schemes could be an effective instrument for recovering the social cost of higher education. It is clear, however, that the experience with loan schemes in LDCs has led critics to question their desirability. We shall examine here two broad categories of the commonly cited objections to loan schemes: (i) such schemes do not generate adequate resources for financing higher education (Kausal, 1978; Brodersohn, 1978); and (ii) they do not promote equity (Herrick et. al 1974; Velloso, 1984). With respect to the low cost recovery rate of loan schemes in practice, the disappointing results can often be traced, at least partly, to inappropriate fee policies. Jallade (1974), for example, points out that "loans cannot be considered as a source of finance for public universities because ....... little ..... was done to [increase] the role of fees". Similarly, Rogers (1971 and 1972) stressed that the cost recovery impact of loans would be significant only if they are accompanied by fees. In fact, since loans are only an instrument towards cost recovery, it would not be very meaningful to discuss their performance outside the context of a fee policy. - 38 - Assuming that the "right" level of fees is charged, it is true that the performance of loan schemes could nevertheless be impaired by the incidence of dropout and repetition. Moreover, in the case of graduates who might be temporarily unemployed, the loan authorities might be obliged to accept postponment of repayment by introducing or lengthening the "grace period". Although these problems are likely to be important, our analysis above show that their effect on the rate of cost recovery is in theory, relatively small. The incidence of defaults is another problem that will reduce the performance of loan schemes. Woodhall's (1984) study show that although it could be a significant problem, some countries, such as Costa Rica, Honduras, Jamaica and Mexico, have been able to keep the rate of default below 10 percent. The experience in these countries indicate that mechanisms for improving the efficiency of loan collection do exist. In Latin America and Asia, the organization of social and economic life probably makes it easier to minimize default on loans. In Africa, this problem appears, at first sight, to be more intractable, given the relative lack of the requisite social institutions. However, since almost all graduates of higher education work in the formal sector, the incidence of default could also be minimized by arranging loan collection through source deductions using the existing social security institutions. The fact that loan repayment might impose too great a burden on students has also been cited as a reason why the rate of cost recovery would in practice be low (Brodersohn, 1978). As indicated earlier, there are obviously social and political considerations for cushioning borrowers against what are judged to be unbearably heavy debt burdens. However, in some loan schemes, the terms of repayment appear to be overly generous, - 3,> - thus reducing further the proportion of the initial loan that can be recovered. In Lesotho, for example, borrowers who join the civil service are automatically granted a 50 percent reduction in their debt. Whether or not such a generous offer is important to ensure adequate staffing for the civil service is debatable, but the adverse effect on the rate of cost recovery should be obvious. Critics of loan schemes have also pointed out that inflation and/or subsidization of the interest rate frequently erode the real value of repayments on student loans. Velloso (1984), for example, argues that the 200% inflation rate in Brazil over the last four years has converted loans into virtual full scholarships. In loan schemes with fixed repayment schedules, this outcome is hardly surprising. In our calculations above, we have shown that this problem can be circumvented by taking for loan repayment a fixed proportion of graduates' income over the repayment period. With respect to equity considerations, the arguments against loan schemes are of course another facet of the objections based on their alleged low rate of cost recovery. The reason is because loans would then contain a substantial "hidden" subsidy. In this case, it is clear that for equity reasons, access to loans should be restricted to financially disadvantaged students. On the other hand, if full cost recovery is possible, there is no obvious equity reason for imposing this restriction. To minimize the size of the "hidden" subsidy, we have indicated that an income-contingent loan scheme would be helpful. At first sight, this plan might appear inequitable: Since the repayment amount is a uniform proportion of income, graduates with larger incomes would be repaying more than those with smaller incomes, even though the initial loan is of the same value. According to Rawl's (1971) theory of justice, however, the income contingent-loan scheme would be equitable precisely because of this effect. The higher incomes of some graduates could be viewed as windfall gains, which should be taxed accordingly; and since these are expected rather than confirmed gains at the time the loan is committed, the case for a proportional tax is further strengthened. Fields (1974) has added that this arrangement also permits "a pooling of risks, so that the individual is not liable for a fixed amount in the event of a personal disaster". Finally, with respect to the impact of loans on equity, we reiterate Woodhall's (1984) observation that "The introduction of loans alone is not enough, but should be accompanied by fundamental changes in fee levels". In some instancas, loan schemes have been implemented in the absence of these changes. In Brazil, for example, a loan scheme exists in the context of tuition-free public higher education. Most students in the public universities come from higher socioeconomic backgrounds, whereas their counterparts in the fee-charging private institutions are usually from poorer homes. Although the provision of loans may help to relieve some of the financial difficulties faced by the poorer students, they do not correct for the most important source of inequity, which is structural in nature, and concerns the access to the two types of institutions. (Velloso, 1984) In summary, the effectiveness of loan schemes depends, first and foremost, on the existence of an appropriate fee policy. They also require an efficient mechanism for debt collection; in this respect, the existence of record-keeping, data processing and managerial capacities is important. Overall, the objections to the implementation of loan schemes have been founded on weakness in either or both of these premises. When neither condition is adequate, it is not surprising that some observers have been critical of the performance of loans as a cost recovery instrument. - 41 - VIII. Conclusion In this paper, we have argued that a policy of cost recovery in higher education would be beneficial in many LDCs; it would improve econotic efficiency as well as promote equity at the country level. Within the higher education sector itself, however, such a policy could introduce some equity problems. This is because some students might be unable to bear the implied cost increase, given their current financial circumstances. As a result, access to higher education could become socially more selective. For this and other reasons, a fee policy is likely to be unpopular with the current and prospective recipients of higher education. However, once a decision has been taken to increase the private cost of education, its social and political acceptability could possibly be improved by implementing it through a loan scheme. As Woodhall (1984) has put it, such a scheme would give "students financial support today, when they need it, in return for a promise that they, in turn, will contribute directly ... to support [future] students". Although the provision of loans could reduce the unpopularity of a fee policy, it nevertheless raises, for the policy-maker, new and difficult issues concerning the terms of repayment. Would those required for full cost recovery be feasible to implement; in particular, would the implied debt burden, measured say, in terms of the duration of repayment, and proportion of a borrower's current income allocated to repayment, be "bearable?" The answer clearly requires a political judgement. Although it is outside our role as economists to offer this judgement, our analysis nevertheless provides some information on the potential effectiveness of loan schemes as a cost recovery instrument in LDCs. We have shown that the terms of repayment required for full cost recovery appear to be easily - 42 - bearable in Asia; similarly, in Latin America, although probably less so than in Asia. In Africa, our assessment is less optimistic; the problem is, h6wever, probably still manageable in Francophone Africa. In Anglophone Africa, cost recovery in higher education through student loans is likely to be more problematic. - 43 - References Acharya, M. (1982), "Issues in Recurrent Costs in Social Sectors", (Mimeo) Employment and Income Distribution Division, Development Research Department, The World Bank, Washington, D.C. Anderson, A. (1983), "Social Selection in Education and Economic Development". (Mimeo) Education Department, The World Bank, Washington, D.C. Brodersohn, M. (1978), "Public and Private Financing of Education in Latin America: A Review of Its Principal Sources", in Inter-American Development Bank, The Financing of Education in Latin America, Seminar held in Mexico City, November, Washington, D. C.: IDB. Eicher, J.C. (1984), Educational Costing and Financing in Developing Countries with Special Reference to Sub-Saharan Africa. Staff Working Paper No. 655. The World Bank, Washington, D.C. Fields, G.S. (1974), "Private Returns and Social Equity in the Financing of Higher Education." in Court, D. and Ghai, D.P. (eds), Education, Society and Development: New Perspectives from Kenya. Nairobi: Oxford University Press. Herrick, A., Sharlach, H., and Seville, L. (1974), Intercountry Evaluation of Educational Credit Institutions in Latin America, Washington, D.C.: U.S. Agency for International Development. IMF (1984), International Financial Statistics, Washington, D.C.: International Monetary Fund. Jallade, J.P. (1974), Student Loans in Developing Countries. An Evaluation of the Colombian Performance, Staff Working Paper No. 182, The World Bank, Washington, D.C. Kausel, E.B. (1978), "Alternative Policies for Financing Higher Education" in IDB, The Financing of Education in Latin America, Washington, D.C.: IDB. Malawi Government (1983), Approved Estimates of Expenditure on Revenue Account for the Financial Year 1982/83, Zomba, Malawi: Government Printer. Mingat, A. and Psacharopoulos, G. (1984), "Education Costs and Financing in Africa: Some Facts and Possible Lines of Action", (Mimeo), Education Department, The World Bank, Washington, D.C. Mingat, A. and Tan, J.P. (1984a), "Expanding Education Through User Charges in LDCs: What can be Achieved?", (Mimeo). Education Department, The lWorld Bank, Washington, D.C. - 44 - Mingat, A. and Tan, J.P. (1984b), "Subsidization of Higher Education Versus Expansion of Primary Enrollments: What Can a Shift of Resources Achieve in Sub-Saharan Africa?", Education Department, The World Bank, Washington, D.C. (1984c), "On Equity in Education Again: An International Comparison". Education Department, The World Bank, Washington, D.C. Psacharopoulos, G. (1981), "Returns to Education: An Updated International Comparison", Comparative Education, Vol. 17, No. 3. Rawls, J. (1971), A Theory of Justice, Cambridge, Mass.: Belknap Press of Harvard University Press. Rogers, D. (1971), "Financing Higher Education in Less Developed Countries", Comparative Education Review, Vol. 15, No. 1. (1972), "Student Loan Programs and Returns to Investment in Higher Levels of Education in Kenya", Economic Development and Cultural Change, January. Tait, A. and Heller, P. (1983), Government Employment and Pay: Some International Comparison, Occasional Paper No. 24, International Monetary Fund, Washington, D.C. Tan, J.P., Lee, K.H,, and Mingat, A. (1984), User Charges for Education: The Ability and Willingness to Pay in Malawi, Staff Working Paper No. 662, The World Bank, Washington, D.C. Unesco (1983), Statistical Yearbook. Paris: Unesco. Velloso, J.R. (1984), "Educational Finance Policies, Dependency and Social Inequalities: The Case of Student Loan Programs in Brazil", (Mimeo), paper presented at the Vth World Congress of Comparative Education, Paris, July 2-6. Wolff, L. (1984), "Financing of Education in Eastern Africa", (Mimeo), Education and Manpower Development Division, Eastern Africa Regional Office, The World Bank, Washington, D.C. Woodhall, M. (1983), Student Loans as a Means of Financing Higher Education: Lessons from International Experience, Staff Working Paper No. 599, The World Bank, Washington, D.C. World Bank (1983), Comparative Education Indicators. Education Department, The World Bank, Washington, D.C. World Bank (1984), World Development Report. New York: Oxford University Press. Other Available EDT Discussion Papers EDT17 Childhood Malnutrition P. Moock December.1985 and Schooling in the Terai J. Leslie; Region of Nepal; Child D. Jamison Malnutrition and School Performance in China EDT20 Lending in Primary R. Romain November 1985 Education: Bank Performance Review, 1962-1983 EDT21 Teacher Training: A Review W. Haddad November 1985 of World Bank Experience EDT22 Institutional Development J. Auerhan, et al. November 1985 in Education and Training in Sub-Saharan African Countries