Cover Image
close this bookEducational Cost-benefit Analysis - Education Research Paper No. 02 (DFID, 1993, 27 p.)
View the document(introduction...)
View the documentExecutive summary
View the document1. Introduction
View the document2. Definition
View the document3. Development
View the document4. Methodology
View the document5. An alternative approach to rates-of-return
View the document6. Other techniques in educational planning
View the document7. Some cost-benefit results
View the document8. CBA in third world countries: Earlier findings
View the document9. CBA in third world countries: More recent studies
View the document10. Criticisms of CBA in third world countries
View the document11. The educational effectiveness literature
View the document12. The comparative education literature
View the document13. Towards a new approach to cost-benefit analysis
View the documentAppendix 1: Project proposal
View the documentAppendix 2: Returns to investment in education by level and country
View the documentAppendix 3: Bibliography

9. CBA in third world countries: More recent studies

The previous section referred to the evidence available to 1985. Whatever the validity of doubts about such attempts at international comparisons, these do seem to have stimulated considerable interest in cost-benefit studies in Third World countries, and a significant number of new studies have appeared subsequently. These more recent studies will be discussed in greater detail.

Home studies appeared at about the same time as the Psacharopoulos (1985) review but too late for inclusion in it. Heyneman (1984) used the traditional method to estimate the returns to investment in Malawi's Certificate of Education, taken at the end of upper secondary schooling, and found these to be high, of the order of 20% for the social rate-of-return and 50% for the private rate-of-return; the calculations allowed for income foregone, assumed an alpha-coefficient as high as 90%, and assumed unemployment at a constant level. The results were for males only; returns could not be calculated for females because from the available sample in the base year (1976) no females chose to enter the labour market.

Guisinger et al (1984) using a Mincerian function and data for males only found low rates-of-return to schooling in Pakistan, for all schooling 3.4% for employees and 7.6% for the self-employed. Returns were particularly low in the Rawalpindi area, due apparently to many of the sample working in the government sector whilst the government had a policy of compressing pay-scale differentials. In contrast to many findings elsewhere, returns were higher at higher levels of education.

The analysis by Marar and Fraser (1986) of the Harijan Education Programme operated at the pre-degree and degree levels by the Kerala State Government in India found the net present value of the programme and its redistribution benefits in favour of the Harijans (ex-untouchables or ex-outcasts) to be negative. These results followed from the facts that nearly 90% of Harijans were unable to complete their courses successfully and that their preferential inclusion in the limited number of places available entailed restrictions on the admission of Christian and higher-caste Hindu students. The researchers suggest, however, that other less quantifiable benefits may follow in the longer run and may help to reduce the effects of caste origins and untouchability and reduce discrimination and illiteracy.

For college education in Mali, Hough (1987), in a simplified calculation as part of a World Bank consultancy report, found a low social rate-of-return of 2%, high private rate-of-return of 59%; this extreme disparity followed from a combination of the high student grants and the high subsequent rate of graduate unemployment (90%). The social rate-of-return to primary education was very low, around 3 %, due to the high cost of examination failures and repetitions.

Psacharopoulos and Steier (1988) used a Mincerian function in their study of returns to education in Venezuela and found an overall return of 11.2% for 1984 data, down from the previous finding of 13.7% for 1975, lending support to the view that returns to education decline over time. Separate calculations relating solely to those workers in the competitive sector of the economy gave results that were of the same order of magnitude.

Al-Qudsi (1989) also used the Mincer approach in connection with education in Kuwait and found returns to education to be relatively low but to be significantly higher for those in the private sector (overall, 8.15% against 4.52%). A complicating factor was ethnic origin since the majority of workers were non-Kuwait) nationals and these were paid significantly less than Kuwaitis, especially in the public sector. 88% of public sector workers were nationals but returns were highest, at 9.36%, for those Kuwaiti nationals who were in the private sector.

Tan and Paqueo (1989), using a Mincerian function, found returns to education in the Philippines which were described as lower than the average for developing countries. Social rates-of-return averaged around 12.7% and were comparable among the three levels of education but private returns were significantly higher for primary than for higher levels of education: the former was calculated at 18.2% but dropped to 12.2% when primary pupils' income foregone was assumed to equal one-tenth of the average earnings of 19-year-olds. Where pupils failed to complete a cycle (e.g., primary, secondary), the returns were much lower. A Mincerian function approach gave a private return (average over all education) of only 8.1 %.

Gomez-Castellanos and Psacharopoulos (1990), using a Mincerian approach, found social returns to education in Ecuador to average around 12% for primary and university education and 9% for secondary education; the former was more equity-enhancing on account of pronounced sex discrimination in the case of forms of employment associated with higher education. Returns were higher for private sector workers than for those in the public sector.

Grootaert (1990) applied Mincer-type functions to data for the Ivory Coast and found that secondary vocational and technical education (VTE) yielded a high private return of 15.84% but a social return of only 3.86%, with a similar contrast for VTE at the post-secondary level (private 21.2% against social 4.4%). Since all the social returns were below the social opportunity cost of capital,

"to justify the investment in VTE thus requires the invocation of non-quantifiable benefits, such as general externalities from having a pool of vocational and technically trained manpower available".

Alongside formal VTE, which predominantly led on to becoming an employee, informal apprenticeships led on to informal labour markets and yielded broadly similar returns.

Hinchliffe (1990) found high social returns to education in Botswana: 20% for the three years of junior secondary schooling, 35% for the two years of senior secondary, and very large earnings increments following vocational training. For those with no schooling, those who had completed primary, those with junior secondary, and those with senior secondary, respectively, returns were calculated at 51%, 82%, 52%, and 30%. There were problems in calculating returns for education separately from training and in arriving at a single operational definition of training; nevertheless, it was clear that vocational training was socially very profitable.

Knight and Sabot (1990) found average social rates of return in Kenya and Tanzania to be around 13% but, since educational expansion over time compressed the educational structure of wages, marginal rates of return could be significantly less than average.

Riveros (1990) calculated internal rates-of-return to education in Chile both via a standard approach and via a Mincerian function. The former gave returns for 1985 of:












whereas the Mincer function gave an average to all schooling of 11.2%. Returns showed a relatively clear trend of declining over time. The author referred to the ability-adjustment problem but does not seem to have allowed for any corresponding alpha-coefficient in his calculations. He viewed the Mincer approach as unsatisfactory since it failed to correct for the fact that his income-related data excluded the unemployed, who were predominantly those with less education: an overestimation of the returns to education was therefore probable.

McGavin (1991) presented updated rates-of-return to education in Papua New Guinea, ranging from (private) 37% primary to 23% university and (social) 13% primary to 8% university. Important local factors were that, for both males and females, wage employment did not begin much before 19 years and that the average life span did not much exceed 50. Many pupils did not complete secondary schooling; where they could be brought to do so, the returns were high. For unskilled and lesser-skilled workers, the reporting of earnings data was probably incomplete, leading to some overestimation of returns to lower level education. All findings were significantly higher than those quoted in an earlier report which had found some social returns approaching zero.

For Brazil, Tannen (1991) used a Mincerian-function approach and data for working males only and found average private rates-of-return averaging around 12%-13%. These were substantially lower than previous estimates. Correction for the probability of unemployment might reduce the findings by one or two percentage points. Regional data enabled the calculation of geographical variations but these were not substantial. The incorporation in the Mincer framework of estimates of subsidies enabled social returns to be calculated but these involved some "guesstimates" relating to public expenditure figures. Vocational training in industrial skills was observed to yield sizeable private and social returns over an academic curriculum at the primary school level. The only notable difference in the returns to education between private and federal government employees occurred for persons who had attended high school; they fared substantially less well in the federal sector.

Psacharopoulos and Alam (1991) found the return to education in Venezuela, from a Mincerian function approach, to average 10.7% (10.0% for males, 13.1 % for females). Returns were higher for workers in urban areas; returns had not fallen significantly over time, despite the education explosion in Venezuela. Calculations via the "elaborate method" found somewhat higher figures, up to 16.2% in the case of the private primary return, with some evidence of rates falling over the previous decade.

Two separate studies reviewed recent evidence relating to rates-of-return. Psacharopoulos (1989) assembled data relating to 23 countries studied by the "elaborate method" and 16 by the Mincer function, to examine whether returns to education were falling over time: overall, they were in the majority of countries but the trend was quite mild. Returns to education continued to be quite high in developing countries, usually above a reasonable measure of the opportunity cost of capital such as 10%. Jain (1991) also found only weak support for the declining rate-of-return hypothesis, especially when temporary, cyclical, variations in local economies were taken into account; also, over time it would be necessary to drop a number of assumptions such as constant technology. The author concluded by emphasising the diversity of cross-country experience.

The latter point may perhaps serve as a useful concluding comment for this section. As the dates of the above publications show, there has been recently and there is currently much interest in studies of educational rates-of-return in Third World countries. A summary reading of their findings, however, shows:

(i) the variety of approaches used by the various researchers,
(ii) the varying data bases with which they had to work, and
(iii) the wide variety in their results and in the conclusions that they were able to draw from these.

It can not be doubted that there are currently in progress many other rate-of-return studies, the results of which will be published in due course.