Anandarup Ray and Herman G. van der Tak
The World Bank lends for projects that contribute to the development objectives of the borrowing countries—primarily faster economic growth and the alleviation of extreme poverty. The economic analysis of a project assesses its likely impact on the relevant development objectives by comparing the various ways in which the scarce resources required by the project might be used instead. These resources may include different types of labor and skills, land, imported and domestic equipment and materials, and so on. The costs of the project are the forgone benefits which these resources would have produced elsewhere, which must of course be less than the project benefits if the project is to be a sensible one.
Cost-benefit calculations also help to identify the critical parameters of a project. In an agricultural project, for example, the key measures that determine the outcome, and therefore need to be closely examined, might be the yield per hectare, the labor intensity of farm operations, or the expected prices for the project’s output. This identification helps to improve the project design, or at least to indicate the chances of the project having its expected benefits. Trade-offs between different policy objectives are analyzed by testing how a project’s net benefits increase or decrease as, say, the project design is changed to give more benefits to poorer income groups.
The framework for cost-benefit analysis along these lines has been extensively discussed in recent years within and outside the World Bank, resulting basically in two types of improvements. First, some of the old concepts of analysis, such as the shadow exchange rate, have been redefined and in the process made more precise. Second, an attempt has been made to make the framework more relevant to policy objectives in developing countries, stressing the flexibility needed to adapt the analysis to the great diversity of situations to which it is to be applied. This article is concerned especially with this second aspect.
The costs and benefits of a proposed project are always measured against an alternative situation—generally that of not proceeding with the project at all. Thus, the benefits and costs are those expected from the project over and above those expected without it. Net benefits to be realized over future years are given a present value and are expressed in constant prices (adjusted for purely nominal changes due to inflation) to demonstrate whether the total net benefits over the life of the project will be positive or negative.
Another approach, equivalent to the first, compares the return on the investment in the project with the return on investment at the margin in the economy, that is, the “opportunity cost of capital.” When the economic rate of return on the project is above the opportunity cost of capital, the project clearly helps the economy; conversely, if it is below, the project will involve an outright waste of resources.
It is often thought that a project needs to be analyzed carefully only when it appears marginal. But a project with a high rate of return, of, say, 100 per cent, is not necessarily an acceptable investment, since there may be better ways of designing the project. A highway may be designed according to different standards or it may be started later; an irrigation project may be designed to supply water thinly over a large command area or concentrated in a smaller area; there may be several hydroelectric sites or different techniques for generating enough power to meet the growth in demand, and so on. Project analysis attempts to ensure that the chosen option for a project is the best possible—not only in terms of its size, technology, and location, but also in terms of the ultimate beneficiaries and the quality of output. The analysis, in short, must demonstrate that the proposed project will create more net benefits to the economy than any other option. To be sure, the search for a better option may be limited by practical considerations, including its cost; but it is always wasteful to proceed with the project if a better option is known to be feasible. Since it is not sufficient for the calculation to show only that undertaking the project is better than doing nothing, it is necessary to define costs and benefits carefully in most cases.
The definitions of costs and benefits used in the economic analysis of a project depend on the national objectives that are to be included in that analysis. When the only objective is the maximization of the total income of the economy, then the costs are the reductions in income suffered elsewhere due to the project’s use of scarce resources, and the benefits are the additions to the total income brought about by the project. If a second objective were to be included, say, the reduction of income inequality (the “equity” objective), then the project’s effects on equality would have to be taken into account—an increase in income disparity in the country due to the project would be a cost, and a reduction a corresponding benefit. Another objective could be the alleviation of absolute poverty, as distinct from merely reducing the income gap between rich and poor. These last two objectives would involve weighting the income gains flowing to the poor more heavily than the gains flowing to the affluent.
An attempt to calculate the effects of a project on such broad objectives as growth, poverty, or equity, and to assign weights to them according to a country’s socioeconomic preferences, poses difficult problems for economic analysis, since market prices do not necessarily provide a satisfactory basis for measurement. Prices which do reflect the proper weights to be given to the various objectives are called accounting or shadow prices. If, for example, a unit of labor is used in a project, the resulting sacrifice in the economy’s total income would be the shadow price of that labor, if maximizing total income were to be the only selected objective. If equity were also an objective, then a different shadow price would be used which would also take into account the project’s effect on equity. To distinguish between these different types of shadow prices, the shadow prices related to the income objective only are usually called “efficiency” prices; by contrast, the shadow prices reflecting total income measured with differential income weighting are called “social” prices.
Not all objectives need to be, nor indeed can be, reflected in each cost-benefit analysis. Suppose that a country is not particularly concerned about reducing poverty, or that it can do so more effectively through means other than the project. It would then be proper to exclude poverty-alleviating aspects altogether from the design of the project, let alone from its economic analysis. On the other hand, if the alleviation of poverty were a prominent consideration, then it must be included in the analytical framework if systematic decisions are to be made about the relative merits of projects which have different effects on poverty. However, if the analysis tries to incorporate too many objectives—say, more than three—it may become too complex for practical use.
The issues addressed and the precision desired in the analysis tend to vary over the project cycle. The study of an irrigation project might begin with the choice of the areas to be irrigated, move on to the choices regarding the operation of the particular schemes devised within a project, and then proceed to alternative methods of cost recovery. The economic analysis of alternatives is likely to be relevant to all such decisions. Even though the analysis of project designs is bound to be rather crude in the early stages, it should still incorporate the relevant socioeconomic objectives.
The traditional approach
Cost-benefit analysis has traditionally focused only on maximizing incomes (an objective variously referred to as the “economic,” the “efficiency,” or the “social surplus” objective). To be precise, the traditional approach is defined in terms of total real consumption of goods and services in the economy, rather than of incomes, since the economic welfare of individuals is related to their levels of consumption rather than to their incomes per se. A project investment reduces the total goods and services available for current consumption but increases the level of consumption possible in the future. Projects also change the relative consumption levels of various individuals in the economy, both at a point in time and over time. In order to judge the worth of a project from the national point of view, it is necessary to aggregate the various gains and losses accruing to different individuals over different periods into a single gain/loss measure. For this, some rules or conventions need to be chosen to define how the different gains and losses can be compared.
The traditional practice has been to regard all gains and losses at a point in time to be equivalent, regardless of whether they affect the poor or the rich. The practice does, however, treat the gains and losses accruing in different periods differently—future gains and losses being discounted to make them comparable to changes in consumption during the current period. Once aggregate consumption is defined in this way, the cost-benefit analysis can proceed to measure the project’s net impact on total consumption over time.
This traditional framework has been very helpful in organizing thought and focusing attention on the economy-wide changes in total income and consumption that result from a project. However, the choice of a discount rate for making changes in future consumption comparable to changes in current consumption can be a source of major inconsistencies. The lower the discount rate, for example, the more weight is given to future gains in consumption relative to sacrifices in current consumption, and hence the greater the importance given to savings and growth. A low discount rate—of, say, 2–6 per cent—may be appropriate for cost-benefit analysis in a developing country which has a commitment to rapid growth. However, the opportunity cost of capital in such a country, reflecting the yield expected on investment, may in fact be much higher because the level of investment is low in relation to existing opportunities and available funds are invested efficiently.
Premium on savings
If the yield on investments in an economy exceeds the yield necessary to compensate people for lower current consumption, then the level of investment is clearly inadequate—a situation which is presumed to be a key feature of most developing countries. In such a case, simply discounting future costs and benefits by the opportunity cost of capital, as in the traditional approach, gives incorrect results since consumption gains and losses in different periods are not properly compared. If, instead, the rate appropriate for discounting future consumption—the “consumption rate of interest”—is used, this will also lead to errors as it underestimates the productivity of investments and thereby causes additional investments resulting from the project to be undervalued. To reflect properly both the relative value of current and future consumption and the unsatisfactory level of investment, it is necessary to use the “consumption rate of interest” as the discount rate in combination with a special premium for adjusting the value of investment expenditures. Thus, if a 5 per cent return is all that is necessary to compensate for a sacrifice of $1 in current consumption, but if that $1 when invested yields 10 per cent, then investment at the margin should be regarded as twice as valuable as current consumption.
This introduction of a premium on investment, and thus on savings, requires the project analyst to judge how much of the income created by the project would be saved. Since the public sector and the private sector, and the different income groups within the private sector, save at different rates, one needs to estimate how the incremental income derived from the project is going to be distributed among the various beneficiaries. Great precision is not necessary in this estimation. A distinction between, say, three income groups in the private sector—the very rich, the very poor, and a large middle-income group—may be sufficient.
An investment premium makes investments more attractive in those public or private sector enterprises that reinvest a greater share of their profits productively.
In many developing countries… the introduction of equity or poverty objectives into project selection tends to be an important complement to other policy measures.
On the other hand, any gains derived by the poor tend to be penalized, insofar as these classes tend to save less of their income gains than the rich. Investments in heavy industries, such as steel and petrochemicals, and in revenue-generating utilities, such as power and telecommunications, are likely to become relatively more attractive. Large-scale mechanized farming and estate plantations will perhaps also be favored. In other words, the premium will tend to make capital-intensive projects more attractive and reduce the emphasis on employment generation. Higher taxes on consumption goods, on income, and on land will appear more desirable, assuming that the government uses the tax revenues productively. The allocation of investment funds between private and public sectors may also be affected, insofar as these sectors have different propensities to reinvest and different levels of efficiency. If such differences are considered significant, they should be reflected in different investment premiums for the public and private sectors.
A primary purpose of the new cost-benefit analysis is to take proper account of the “scarcity of foreign exchange” faced by many developing countries. It is often thought that this “scarcity” is also allowed for in traditional analysis. However, the scarcity value of foreign exchange depends on the economic objectives it adversely affects. If the benefit of additional foreign exchange is that it permits higher levels of investment in the economy, then the scarcity of foreign exchange is reflected in the premium on investment. It has recently become clear that the so-called “shadow exchange rate,” or “shadow price of foreign exchange,” as used in traditional analysis, does not bear on the scarcity of foreign exchange in this sense. This “shadow exchange rate” is only a device for correcting the distortions in the relative prices of internationally traded and non-traded goods, and for that purpose it is also used in the new method.
There are practical difficulties, of course, with the use of a premium on investment. It is often hard to decide the proper size of the premium, and estimating the increases in income and savings of different groups from a project may be a demanding task. Would it then not be better to rely solely on qualitative judgments in this respect? The World Bank, for example, has always placed strong emphasis on financial viability, high levels of cost recovery, replicability, and other policies which directly or indirectly reflect concern about the scarcity of investable resources. Unfortunately, qualitative adjustments in project decisions rarely work satisfactorily. Suppose the economic rate of return of a project, measured without an investment premium, is marginally above the cutoff rate; can this project still be rejected if all of its gains are expected to be spent on additional consumption? Or, can a project with an unsatisfactory rate of return be accepted if all of its gains are expected to be reinvested? It is clear that answers to questions such as these implicitly involve a quantification of the value of savings and investment. Such implicit, ad hoc quantification can, however, lead to grossly inconsistent project decisions.
Inequality, poverty, and basic needs
Introducing the investment premium does not require any change in the basic economic objective of traditional cost-benefit analysis, which will still treat consumption gains or losses to different individuals equally. The premium focuses on the correct assessment of a project’s impact on total consumption, but does not affect the concept of the costs and benefits that are being aggregated.
The concepts themselves, however, will need to be changed if concerns with issues such as employment, income inequality, and the alleviation of poverty are to enter into the economic analysis. Gains and losses to different income groups will then be weighted differently to reflect these concerns, by giving more value to benefits to the poorer groups. For this purpose it would be necessary to assess which income groups are expected to gain or lose from the project. A broad distinction between only a few income groups is likely to suffice in practice. An even simpler distinction of beneficiaries into only two groups, above or below a threshold level of poverty, would suffice if the reduction of absolute poverty is the desired objective.
It is sometimes thought that even though governments may be concerned with income distribution and poverty alleviation, they need not introduce such concerns into project decisions, but instead should rely on other instruments of policy. Even though most governments have many policy instruments available which could directly or indirectly affect equity and poverty among their populations, the majority of developing countries seem to have found poverty redressal or the alteration of income distribution very difficult. The redistribution of land, for example, is generally crucial to redistributing incomes in most of these countries; but effective land reform has often proven infeasible, and land taxes are notoriously difficult to administer. The imposition of progressive income taxes also has practical limits, especially if serious adverse effects on earning incentives are to be avoided. Moreover, reliance on indirect taxes, or on inflationary finance, would affect the allocation of resources adversely and tend to weigh more heavily on the relatively poorer groups.
In many developing countries, therefore, the introduction of equity or poverty objectives into project selection tends to be an important complement to other policy measures. It is usually easier to locate projects in backward areas or to design them for urban or rural poverty groups than, for example, to change the tax system or to redistribute assets directly through a national land reform. It is much harder to shift the distribution of existing assets than to direct the creation of new assets in favor of the poor—although the allocation of public sector investments also has political constraints.
The current practice in the World Bank treats the alleviation of absolute poverty as a very important aspect of many of the projects it finances. In order to orient projects toward this goal, several informal rules of search are used in the identification stage, such as upper limits for the cost per job created, or for the acceptable cost per beneficiary. However, search rules are not an adequate substitute for a fully integrated analysis of the conflicts between objectives, such as more employment or income for the poor versus more rapid growth in output. In land settlement projects, for example, the question frequently arises whether to allocate small units to each settler and thus spread the benefits widely or to allocate fewer, larger units in the interests of higher productivity of land use. A rule restricting the cost per beneficiary may be counterproductive in such cases unless it is derived from a full analysis of the trade-offs involved.
If poverty or equity objectives are introduced in the analysis then suitable rules must be specified for aggregating the various gains and losses accruing to different individuals into overall benefit and cost figures for each year of the life of the project. The decision maker who rules on a project, or the advisor who recommends a project, must necessarily use a scheme for weighting the gains and losses of different income groups. The question for any particular country is then which type of weighting scheme is most realistic and relevant for this purpose? Should one choose the equal weights used in traditional practice or should one differentiate according to income groups? The answer to this question obviously depends on the specific socioeconomic priorities of the country for which the project is planned, and no single weighting scheme is universally applicable. But these priorities are usually not explicitly formulated, and the analyst is faced with having to deduce their relative importance.
It is, however, possible to test the plausibility of relative weights reflecting different policy objectives by analyzing various national policies. For example, equity is often an important aspect of taxation policy, and there is always an exemption limit for income taxes. Moreover, many governments run large subsidy programs for the poor. Such policies suggest that if a person is poor enough, then an extra dollar to him is valued more highly than an extra dollar of government revenue, and therefore there is a critical or break-even point—the “critical consumption level”—at which marginal private gains are socially worth just about as much as marginal increases in government revenues. A person should not receive subsidies unless the level of consumption he can afford is below this critical level. If the country concerned is deeply committed to growth, as are Brazil, Ivory Coast, and Korea, for instance, then heavy weight is given to generating incomes for investment, and hence this critical consumption level should be very low.
The cutoff point for subsidies might then be at an income level which is only, say, 25 per cent of the national average. Subsidies will thus tend to be. restricted to the very poor groups in such countries. If, on the other hand, the country is more concerned with equity or with alleviating poverty, as are, say, Sri Lanka and Tanzania, the appropriate critical consumption level would be much higher, perhaps as high as 75 per cent of average income. The critical consumption level is a relative income measure in the country and is usually well above levels representing absolute poverty. It is widely used in practice, especially in the context of project-related pricing and cost-recovery policies.
The critical consumption level is one of the benchmarks for assigning distribution weights that reflect a country’s policy priorities. There are many other tests that can be devised to determine the most reasonable weighting scheme for the country concerned. Generally speaking, assigning equal weights to different income groups, as in the traditional economic analysis, would appear to be appropriate only in exceptional cases. Some degree of differentiation between income groups, at least to take account of extreme wealth and extreme poverty, is usually likely to be more realistic.
The introduction of different weights for different income groups would counteract some of the effects of giving special weight to the generation of additional investment. Projects which lead to additional savings and reinvestment will still be favored, other things being equal, unless the benefits accrue to those below the critical consumption level. Labor-intensive operations and employment generation will be favored to the extent that the additional labor income accrues to the poor.
The differences between countries can be easily reflected in the analysis since the emphasis given to employment, equity, or poverty alleviation can be “controlled” by varying the critical consumption level: the lower the level, the less the importance given to such concerns. However, since the same differential income weights and the same critical consumption level are to be used for all projects within a country, it is clear that ad hoc judgments are avoided by this method. The use of poverty or equity considerations on an ad hoc basis tends to give a “free license” to accept any and all projects that help the poorer groups. In contrast, the new approach demands consistency and discipline in project choice.
Another important objective for many developing countries is to meet the “basic needs” of their people. Definitions of basic needs vary, but the principal interpretation treats certain goods and services as basic needs or “merit wants” that should be satisfied as a matter of government policy, rather than being met through charity dependent on private preferences. The planners or policymakers decide therefore which needs are basic, and what quantity and quality of service should be provided. They fix the weights that determine the importance to be given to additional consumption of the goods or services which meet the basic needs of various (usually income) groups, and how soon these needs should be fully satisfied in relation to other objectives of growth and distribution. These specific basic need weights are a straightforward variation on, and complement, the more general distribution weights discussed above.
The economic rate of return of a project defined in terms of prices that incorporate distributional weights (social prices) may be called its “social rate of return.” It will frequently differ from that calculated on the basis of traditional efficiency prices. There is no built-in tendency for social rates to be higher than the traditional rates of return. The new approach is designed in such a way that the social rates will be higher only to the extent that any increases in consumption due to the project accrue to those below the poverty line, and will be lower to the extent that the project increases the consumption of the relatively affluent. Social analysis does not make it easier to justify projects, but it tends to justify different projects, that is, projects that favor the poor and/or increase the level of investments in the economy.
The rigid adherence to one particular set of weights, as in traditional cost-benefit analysis, appears too doctrinaire to be appropriate for all developing countries, or even for the same country at different stages of its development. In countries where the distribution of project benefits is important, the traditional way of analyzing projects is only a partial indicator of the economic impact of a project and is not necessarily a reliable guide to project decisions. The new approach, on the other hand, focuses directly on the hard choices facing developing countries between growth and redistribution and is likely to improve the decision-making process. As the experience with social pricing in cost-benefit analysis accumulates and the methodology is adapted accordingly, it is likely to become a widely employed tool of analysis, not only in the World Bank but also in other international and national institutions with responsibility for selecting projects which best meet the policy objectives of the country concerned.