|Educational Cost-benefit Analysis - Education Research Paper No. 02 (DFID, 1993, 27 p.)|
From the foregoing sections in this report, it is clear that:
(i) Educational cost-benefit analysis is currently, and has been for some years, a widely-accepted technique, used to assess the profitability of investment in education. Educational rate-of-return studies have been carried out in most developed countries and in many developing countries. In the great majority of cases, the results are in favour of additional investment in education, which the studies show to be profitable both from the point of view of the national economy and from that of the individual student. In general, the studies particularly favour additional investment at the level of primary education.
(ii) There are serious doubts regarding a number of aspects of the methodology used in rate-of-return studies. It is not difficult to show that the underlying assumptions may be invalid, the data may be faulty, or there may be inherent sources of bias in the results and that some of these may be very large, possibly so large as to invalidate the findings. Some writers see the application of cost-benefit techniques to education as being essentially flawed.
Educational cost-benefit analysis will undoubtedly continue. But the above criticisms point towards the desirability of a revision of this approach, possibly in the direction of incorporating elements of the two principal alternative approaches, namely manpower planning and social demand. Such a possibility was in fact envisaged by Professor Blaug over twenty years ago, when he wrote:
"Faced with the difficulties of manpower forecasting, difficulties that seem to increase at a progressive rate the longer the time period over which we are forecasting, the remedy is to begin modestly with short-term forecasts which are then extrapolated with a compounding margin of error. As we accumulate more experience, we begin to adjust the margin of error, gradually producing more and more reliable medium-term and eventually long-term forecasts. As a check on such forecasts of demand, we ought to make continuous year-by-year projections of the future supply of educated people. Indeed, the forecasts of demand ought to be of the type that provides a range of alternative estimates, given different projections of the projected supply. If the demand for educational inputs depends in any way on their prices, and this will necessarily be so if there is any substitutability between educated people, changes in supply are just as capable of altering prices as changes in demand and, therefore, the quantity demanded of educational inputs is not independent of its supply. It follows that manpower forecasts must always be combined with projections of the demand-for-places. As we combine forecasts of demand for manpower with projections of the supply of manpower, we start thinking quite naturally of earnings associated with education as possible indicators of impending shortages and surpluses; and since the costs of training various types of specialized manpower differ considerably, we shall be led to consider variations in earnings in relation to variations in the costs of education. This is rate-of-return analysis, whether we call it that or not. If earnings are inflexible and fail to reveal shortages and surpluses of manpower, the remedy lies in imputing "shadow prices" to labour of different skills and calculating the critical rates of return that lead to definite investment priorities in education. By making such calculations on a year-to-year basis, we keep a continual check on labour markets for highly qualified manpower and gradually develop insights into the ways in which education interacts with economic growth. Rates of return as such can never provide more than an ex post check on the efficiency of investment already embodied in different kinds of educational facilities and, of course, a signal for a possible direction of change in the pattern of educational investment. By supplementing rates of return with exante estimates of the likely changes in the demand and supply of skills over the planning period, however, we convert them into tests of the validity of predictions of demand and supply. If we get different answers from rate-of-return calculations than from manpower forecasts, it may be that (a) earnings are divorced from the marginal productivity of labour, (b) the costs of education are artificially inflated, (c) future rates of return will diverge from present rates or (d) the manpower forecasts are wrong. Which of these four factors or which combination of them is responsible for the difference in answers cannot be settled on a priori grounds. What we have been trying to do is to build up a framework in which such factors can be systematically considered. The message of this framework is that the manpower requirements approach, the "social demand" approach and rate-of-return analysis are reconcilable and, in fact, complementary techniques of educational planning, but not as these approaches are presently practiced around the world." (Blaug, 1970, underlining added)
Professor Blaug was particularly critical of attempts to make use wholesale of elements of the different approaches in their present form, as happened with higher education in the UK in the post-Robbing era: higher education places were expanded to meet the increasing demand (the social demand approach) but the government attempted to maintain the principle from previous manpower planning exercises that two-thirds of the additional places outside medicine and agriculture should be in science and technology. Given the different assumptions embodied in each approach, Blaug concluded that "this really combines the worst of both worlds".
Subsequently, there has been considerable interest in developing educational planning models which combine elements of all three approaches (cost-benefit analysis, manpower planning and social demand), or at least combine cost-benefit analysis and manpower planning. Such approaches became known as "synthetic" educational planning models. Synthetic models: "purport to offer a compromise between the polarized assumptions of the manpower requirements approach and the cost-benefit model" (Psacharapoulos, 1985b).
Such models may proceed in a number of stages, for example:
1. From: Base year labour structure via: Manpower forecasting
2. to: Target year labour structure, 1st approximation.
3. Using: Linear programming incorporate: Shadow wages of target year labour structure.
4. CBA: To give shadow rates of return, corresponding to above labour structure.
5. This enables: rate of return comparison between each other and the social discount rate.
6. This gives: Optimal target year labour structure
(Source: Psacharopoulos, 1985b).
This approach can either commence from a quantity solution, i.e. estimation of quantities of labour skills required in the target year (via manpower planning), and subsequently add in relative prices, or commence with a set of relative prices of skills, or rates of return (via cost-benefit analysis) and then proceed to find corresponding quantities, thus giving "a cost-benefit evaluation of manpower planning" (Psacharopoulos, 1985).
Such synthetic models are also sometimes referred to, incorrectly, as linear programming models: linear programming is simply a mathematical technique for arriving at a solution to a set problem and has nothing to do with educational planning as such.
In adopting a synthetic planning model, it has to be remembered that the cost-benefit and manpower planning approaches, which are both being used here, embody quite different, indeed opposing, assumptions. The most important of these assumptions may be represented as follows:
elasticity of substitution between different skills
elasticity of demand for skills
Thus, in including both approaches, we must necessarily, in each case, be assuming some elasticity around mid-point between zero and infinity. To do so may well be reasonable and realistic but may also undermine some of the findings. Thus in the manpower planning part of the exercise, if there are some genuine elasticities, both of substitution between skills and of demand for skills, then the forecast quantity figures will be partially invalidated. Similarly, if in the cost-benefit part either or both of these elasticities are "sticky", then price signals will not have the effects anticipated and, again, the expected outcome will be partially invalidated.
This is a fundamental problem with any attempt to combine cost-benefit analysis with manpower planning, since the two approaches are based on quite different assumptions, indeed on contrasting views of the economic world to which they apply. These were outlined in some detail by Blaug (1970) in what he termed "Two Views of the World"; he went on to recommend an "active manpower policy" which would consist partly of attempts to move the real world in the direction of the rate-of-return end of the continuum, e.g. reductions in specialization and greater flexibility.
The above account of the development of synthetic approaches is at the conceptual or model-building level. There seem to have been very few attempts to apply such an approach in practice.
Dougherty (1971) showed that when the standard cost-benefit approach is modified to allow for relative wage levels to change over time, it is possible to incorporate the effects of the growth of the education system on the growth of each category of labour and thence on future wage rates. This clearly incorporates elements of manpower planning. Dougherty tested this approach for data relating to Colombia and calculated rates of return to primary, secondary and higher education for successive years, allowing for the effects of changing wage rates. For 1985, he found returns of 19.8%, 17.8%, 0.9%, respectively.
One of the rare attempts known to attempt to embody the principles of both rate-of-return analysis and manpower planning has taken place in Cyprus. The Director of the Department of Statistics of the Ministry of Finance, Government of Cyprus, has given outline details of the way his government has developed an "eclectic" approach which uses both manpower forecasts and cost-benefit studies and which: "focuses much more on particular forms of education and particular occupational and industrial employment categories, rather than combining all relevant factors into a single model" (Demetriades, 1989).
There has been little other progress in the direction of empirical studies incorporating both the cost-benefit and manpower planning approaches, and it is not difficult to see why. Quite apart from the differing assumptions underlying each and the fact that, as indicated previously, each of the manpower planning and social demand approaches has been the subject of at least as much criticism as has cost-benefit analysis, research studies would become not only more complex but also much more costly.
Also, there have been very few attempts to relate cost-benefit results to questions of school "quality" (as argued by Behrman & Birdsall - see section 10 above) or educational "effectiveness" (see section 11 above), instead of merely taking number of years of schooling as the adequate school variable as is conventionally the case with rate-of-return studies.
In conclusion, three further points, all of which have been referred to earlier in this report, require emphasis here. Firstly, the link with cost-effectiveness: cost-benefit analysis may be used as a means of comparing alternative uses of resources in order to identify the most cost-effective. Psacharopoulos and Woodhall gave examples of this, including general versus vocational education in Colombia and Tanzania, on-the-job training versus formal training in Israel and formal schooling in Brazil, and the effectiveness of a new school building programme compared with a school repair programme in El Salvador.
Secondly, there could be a more extended treatment of the use of sensitivity analysis to compare alternative manpower forecasts and different patterns of manpower utilisation, and of cost-benefit analysis to study the determinants of private demand, which is a way of linking up the three approaches; the determinants of demand would include students, perceptions of costs and benefits, i.e. the private rate of return, and their forecasts of future job prospects, i.e. crude manpower forecasts. Government may then seek to shift costs from taxpayers to students, as the British Government has done recently. Compare with, e.g. the massive expansion of higher education in Kenya and transfers of costs to students.
Thirdly, cost-benefit analysis has been used in several countries to develop or at least justify new policies on financing education, see e.g. the World Bank recommendations on financing education, which draw heavily on rate of return studies. This represents the main way in which governments in both developed and developing countries are currently using cost-benefit analysis to guide and formulate policy. To quote Maureen Woodhall again:
"To sum up, a new approach to cost-benefit analysis is already in evidence in many developing countries, which are changing traditional patterns of financing higher education in the light of evidence of high private rates of return and are switching emphasis to primary education, just as cost-benefit analysis recommended.
The crucial need in the next decade is to monitor the effects of these changes within a cost-benefit framework".
Finally, as we have seen, educational cost-benefit analysis, as currently practiced, has been the subject of much criticism, and yet the principal alternative, the manpower planning approach, has been the subject of even more. Given that both are currently is use in different ways or for different sectors in many different countries, from the point of view of this report it does seem regrettable that there seem to have been so few attempts to combine the two in empirically-based studies.
Further research on these lines would be welcome even though it would be both costly and time-consuming. This might take the form of studying recent cohorts emerging from the education system and charting their subsequent employment progress, including vis the use of such competitive labour market signals as are available. Such research should enhance the validity of educational cost-benefit studies and should lead to increased confidence in the effectiveness of educational planning.
Any such future studies should be carried out in conjunction with local staff from the country in question, perhaps suitably-qualified staff from the Ministry of Education or possibly from a local university, with a view to developing local capacity to undertake independent cost-benefit studies.