Project Evaluation

Many economists if asked to nominate Ian's major contribution to devel­opment economics would select his work on project evaluation. Given that, it is notable that his entry to that field was almost accidental.

The other main product of my two years at the Development Centre was the OECD Manual of Industrial Project Analysis. This was jointly authored by myself and Jim Mirrlees. This was not the outcome of research that I had started. The Development Centre had already commissioned a French consultancy firm to produce such a manual, soon after it heard that the UN was doing so. A draft arrived which I thought terrible. I criticized it fundamentally, and revisions were promised. I considered the revised draft which eventually arrived to be still unacceptable. A small conference was called, most participants of which sided with me. But I had to threaten resignation to get the ball rolling. Baron [the then President of the OECD Development Centre] was convinced that my opposition simply stemmed from an Anglo-Saxon attitude (LbL 2004: 132).

Here, the discussion of the contribution made to project evaluation theory by Little and Mirrlees (henceforth L&M) will concentrate on their 1974 pub­lication (henceforth Project Appraisal) rather than the original 1969 manual.^[148] Two reasons support this choice. First, the 1974 book develops and presents their ideas more thoroughly and richly than the original. Secondly, the later publication responds in detail to the UNIDO Guidelines volume published between the two in 1972 (see Dasgupta et al. 1972). A comparison of the L&M approach and that of UNIDO is made difficult because the two vol­umes have distinct orientations.

Fundamentally, L&M and UNIDO follow similar paths in that they adjust market-based returns by using shadow prices that are designed to better reflect social valuations. A difference between the two methods that received great attention is in itself of limited significance: the two systems use different numeraires (accounting units). The choice of a numeraire cannot of itself make a great difference. However, once a numeraire has been selected, conver­sion factors are required, that is, formulae to convert other values, such as wage rates, into values expressed in the numeraire. Then the details of conver­sion can make a substantial difference. The L&M numeraire is uncommitted social income measured at border prices, which contrasts with UNIDO's aggregate consumption measured at domestic market prices. To cut short what could become a lengthy discussion, it suffices to say that the L&M method is simpler and more reliable in practice. This is because it avoids the complex issue of deciding how far domestic market prices correctly reflect their contribution to consumption. In a highly distorted economy this is a complex exercise. L&M, on the other hand, avoid this tangled maze, either because if the good is traded one goes directly to the border price or, if it is not traded, its value can be measured at its marginal cost of production, broken down into its direct and indirect traded-good content (valued at border prices) and labour costs.

The focus of any project evaluation exercise is on the particular project and the numerical values associated with it. For that reason, the impression is too easily arrived at that the theory is entirely microeconomic, concerned only with the project itself.

SWR = m + (c' - c) + (1 -1 / 5)(c - m)

where c' = value of consumption at market prices, including items that do not directly contribute to welfare such as transport costs; c = welfare-producing consumption; m = marginal productivity of the wage earner; and s = the value of uncommitted government income in terms of consumption.

The first term in the above equation is the marginal product of labour, the second term adds the costs of delivering consumption, such as transport costs, and the third term shows the increase in consumption of the marginal worker minus that part of it which is reckoned to be a benefit. The final total SWR is in domestic local-currency value. That must be converted to the numeraire (foreign exchange) by the application of the shadow exchange rate. This last number is an economy-wide value with which all project evaluators will be provided.

The derivation of the SWR illustrates beautifully some of the basic princi­ples that underlie the L&M analysis. Wages display two contrary aspects. On the one hand, they are a welfare benefit; they provide workers and their fami­lies with consumption, and the higher they are the more consumption they provide. On the other hand, they are a cost to the national budget because each rupee of wage paid out might otherwise be applied to beneficial govern­ment expenditure. In a simple case, let (c' - c) be zero, so no additional resources are devoted to the provision of consumption. Also, let m equal zero because, for example, labour employed on the project comes from agriculture where the marginal worker adds nothing to output.

SWR-(1 -1/5) w where w is the market wage rate that the project will have to pay. Note that the SWR is below the market wage rate. This implies that public sector proj­ects evaluated positively by the L&M method will be more labour intensive than would be a similar project chosen to maximise profit in the private sector.

Another important value for the accurate assessment of projects is the accounting rate of interest (the ARI), the number that measures how future numeraire values are weighted relative to current numeraire values. This rate of interest may vary over time, but the discussion concentrates reasonably on the case where it is nearly constant. The role of the ARI is to act as a gate­keeper for the projects being assessed. It must not accept too many, when taxes would have to rise sharply, and present consumption would be depressed excessively. Equally, it must not accept too few, when welfare-increasing pos­sibilities would be wasted. The questions at issue here are easier to answer in a classroom on a blackboard than in reality. The two fundamental effects that need to be taken into account are the rate at which per capita consumption will rise, and the root discounting of the future that reflects the impatience of the planner (or the population). Growth of per capita consumption argues for weighting future consumption more lightly. Impatience adds an additional effect in the same direction. These two effects together generate an ARI that should be equated to the rate of return on the marginal project—the one that only just gets accepted. L&M discuss an interesting, although special, case in which the return on private investments provides a useful estimate of the ARI.

The OECD Manual was hugely influential.

The OECD Manual was strongly attacked by the development establishment, especially the Oxford branch. The essential principle it promoted was that, in considering the costs and benefits of domestic production of something, both export possibilities and the alternative of satisfying domestic demand by import­ing should be carefully considered. The implied insistence on trying to use inter­national trade optimally was anathema to those who had been taught that free trade was a colonial tyranny designed to ensure that developing countries would forever produce only primary commodities.. Since those days relatively open trading policies have become more widely practised in developing countries, and few would now deny the benefit of such policies. But I myself continue to be reviled as The Great Satan in some development schools (LbL 2004: 138).

The critiques of L&M pursued many arguments, these of variable merit. The February 1972 edition of the Bulletin of the Oxford University Institute for Economics and Statistics was devoted entirely to a symposium concerned with the OECD Manual. Several of these papers, including one by Vijay Joshi, took a favourable view of L&M, and the paper by Nicholas Stern on an appli­cation to tea farming in Kenya provided a valuable example of the L&M method in practice. Partha Dasguptas paper compared the OECD and UNIDO manuals. In contrast, the long paper by Frances Stewart and Paul Streeten is not unlike a prolonged artillery assault on L&M (see Stewart and Streeten 1972).^[151] Elsewhere, a paper by Amartya Sen explored the issue of irrational (or at least immovable) government policies (see Sen 1972), a point also stressed by Stewart and Streeten.

Leading issues raised by the Oxford critics of L&M are the following: irra­tional governments; economic linkages; and non-traded goods. It was claimed that L&M assume that the government of the country to which project evalu­ation is applied is as rational and detached as the authors themselves. Another assertion is that L&M ignore the multiple linkages—forward, backward, and sideways—that are characteristics of underdeveloped countries. The final claim from the prosecution is that L&M give insufficient weight to non­traded goods and fail to price them correctly.

In the final paper in the Bulletin of the Oxford University Institute issue, Little and Mirrlees provide a vigorous and robust reply to their critics. They agree that recommendations may be conditional on a rational government response but note that the implication of an irrational response is often con­tained in the recommendation. Thus, if the project evaluator recommends the adoption of a scheme to manufacture motor vehicles domestically, provided that the engines are imported, this implies, and that could be made explicit, that the scheme should not be adopted if the government insists on all pro­duction being domestic. On linkages, L&M confirm their scepticism con­cerning their universality and measurability, yet point out that if a linkage is evident and important it becomes part of the project, to be assessed with other components of the same. They underline their flexibility concerning the shadow pricing of non-traded goods, such as electricity supply in many countries. Non-traded goods can often be priced by their opportunity costs in terms of traded goods. If that is not possible, the values in domestic prices can be translated to border values using the conversion factor that already figures in their analysis. Notable in the L&M response is how, rather than mounting new arguments, the authors usually point their critics to what is already there in the Manual.

The 1970s were the years when project evaluation based on cost-benefit analysis was at its high point, both in developed and developing countries. Since then, its status has declined, although it is still used (or abused).^[152] A leading problem that emerged when institutions such as the World Bank tried to impose the method is that project evaluation proved to be strongly liable to manipulation. As L&M show clearly in their writing, estimates and guesses have important parts to play. That opens the door to biased estimates designed to achieve a particular result—usually the acceptance of a dubious project. A senior Indian civil servant once told one of the authors of this memoir that, given the book of rules, he and his colleagues could arrange for almost any favoured project to get over the finishing line. In fact, the bias affecting proj­ect evaluation is two-sided. Governments receiving aid favour certain projects and will twist the assessment process to favour those schemes. Lenders also have their own biases. They are not paid for turning down projects; their job is to lend money. So, a rigorous tough approach to project proposals does not suit donors any more than recipients. Little was sharply aware of the problems created when political forces encroach on project evaluation. He writes: ‘The main difficulty facing cost-benefit analysis is that large public, or publicly subsidized, investments are a source of prestige, patronage, and kick-backs for those in power, and their relatives and cronies. They do not want their proj­ects submitted to hard-nosed appraisal by economists’ (LbL 2004: 142).

Aside from the problems of manipulation discussed above, there is another major reason why cost-benefit analysis on L&M lines has declined in impor­tance. A leading motivation for the L&M approach, and the same could be said of the UNIDO method, is to surmount the misleading price signals prev­alent in highly distorted economies, especially those subject to strong and unbalanced trade protection. All this has become far less important as devel­oping countries have become more open, their markets less interfered with, and their tariffs and controls diminished, often to levels below those of rich industrial countries. A great deal of credit for this belongs to Ian and to economists who thought on similar lines. So, perhaps Little, the trade and protection specialist, was the executioner of Little, the project evaluation innovator. If that is the case, he would probably not have minded.

5