Executive Summary I. Introduction II. Definitional Questions III. Causes of Recurrent Cost Problems IV. Recurrent Costs Analysis V. Solutions to Recurrent Cost Problems VI. Conclusions and Recommendations Appendices Wednesday, 11-Jul-2001 16:54:59 EDT

When a government sets out to promote economic development it has two resources at its disposal -- its capital budget and its recurrent budget. In many ways these two resources are substitutes for one another since government resources are, in principle, fungible. There are, however, two restrictions on this general fungibility. There is, in the first place, a technical constraint related to the unidirectional nature of time. A capital expenditure once made can never be converted into a recurrent expenditure ("sunk costs are sunk CostS").¹ On the other hand a proposed recurrent expenditure can always be converted into a proposed capital expenditure. In other words, one can always choose to underfund existing projects in order to make new capital starts; one cannot convert existing capital assets into recurrent financing. Secondly, donor policy, which prohibits, in general, recurrent financing, limits fungibility, especially in the poorest countries which are most dependent on foreign aid for their capital budgets. Thus over any planning horizon a government can choose, within the limits imposed by donors, the degree to which it will invest in new projects, and the degree to which it will finance the variable inputs for existing projects.

As discussed in the previous section a recurrent cost problem exists when the stream of returns to recurrent expenditures is greater than the stream of return to capital expenditures, or is high relative to the opportunity cost of capital. There are two basic causes for such a problem -- donor policy and LDC government policy. We will take each of these up in turn.

A. Donor Policy

The resistance of donors to funding recurrent inputs is a major casual factor of the recurrent cost problem. For a large number of LDC's, foreign assistance and, therefore, government investemnt funds, are a much more abundant resource than are government recurrent resources. For low income USAID recipients, (i.e., those countries with per capita incomes below $400 in 1978) the average excess of investment over domestic savings is 16.5% of GDP; the level of government expenditures on goods and services for these same countries is 15.5% of GDP. Thus, total net foreign inflows available for capital accumulation are greater than total domestic government resources available to service the entire recurrent budget. An estimate of the ratios of development assistance to major government expenditures for low income countries is presented in Table II.

Why do donors limit funding to capital costs? The main arguments seem to be related to the following set of premises:

1. It is better to provide assistance which is directed toward investment, and thus has growth potential, than toward consumption;
2. Funding of recurrent costs is an open-ended commitment which leads to dependence rather than self-reliance on the part of the recipient and which commits the donor to certain activities over a longer run than is politically acceptable.
3. The larger the share of total project costs financed by donors the less interest and commitment does the host government, have in the projects success.

Let us examine each of these arguments.

TABLE II Ratios of Net ODA Receipts to Government Expenditures, Selected Low Income Countries,a 1977 (Figures are in percentages)

Source: OECD, Geographical Distribution of Flows to Developing Countries (1980), and Tait, Alan and Peter Heller, "International Comparisons of Government Expenditure." IMF, DM/81/53 a) Countries with per capital incomes below $400 in 1978 b) Countries for which ODA receipts are greater than current expenditures.

1. Investment vs. Consumption in Recurrent Cost Funding

For the accountant the question as to which expenditures are investments and thus belong in the capital account, and which are consumption and thus belong in the current account is a simple one to answer. Any expenditure which purchases something which is not used up in one year is an investment; any expenditure which is used up within the year purchased is consumption. The former expenditures are put in the capital budget, and the latter in the current account. Thus wages which are paid to road builders enter under capital accounts, while wages paid to street cleaners come under current account. However as one probes carefully behind the accounting constructs, a number of anomalies arise.

First, many expenditures classifed as "current" are actually new investments. Education is widely interpreted by economists as a process through which "human capital" is developed. Both the student and the state are employing resources to produce an investment which has a very long gestation period. Producing a skilled worker or producing a road are conceptually the same thing, and it may be that the worker's skills are the more productive. Therefore, should not expenditures on education, and to a large extent health, be considered as capital expenditures rather than consumption. The same argument can be raised for investments in agricultural research institutions. Here the end product, knowledge, has an extremely long life, as well as a long gestation period, and, therefore, costs involved in producing new knowledge should be considered as investments. To state the case even more strongly, the services of extension workers, who are involved in transmitting knowledge, are similar to those of teachers. Although with extension work the gestation period may be short, the new skills and techniques learned by farmers, continue to pay off over a long period of time and thus the costs of extension needed to spread these skills and techniques are investments.

Secondly, maintenance of capital goods, either physical or human, is surely a gross investment if not a net investment. Thus road maintenance, or health care (including labor costs and drugs) represent investment expenditures designed to keep the quantity and quality of the capital stock from deteriorating. While they do not produce increments to capital, they operate to maintain the integrity of the capital stock. Replenishment of the working capital of credit institutions is another important example of a gross, though not a net, investment.

Lastly, the one year distinction between current expenditures and capital expenditures is largely a matter of convention and has no conceptual legitimacy. Any expenditure which provides a stream of services over time is by nature a capital expenditure, and every expenditure which provides services only instantaneously is a current expenditure. The operational question is the length of the instant. The same expenditure can be viewed as a recurrent or a capital cost depending on the point at view. For example, a bowl of rice provides satisfaction for the length of time, a day perhaps, needed to burn off the calories it contains. However, lack of sufficient nutrition may lead to the stunting of physical and mental abilities for a lifetime. If the length of time invoived is one year then a vehicle with a three-year life is a capital good; if the standard time period is one month, then fertilizer is a capital good. There is, then, a continuum, with expenditures producing outputs over time, some over long periods of time; others over shorter periods. Any expenditure which produces increases in productivity partakes to some degree in the nature of a capital good. But that "capital-good nature" varies in intensity depending on the length of time that the expenditure continues to provide productivity increases.

What this means is that donors who are willing to provide capital costs of projects, but are reluctant to provide recurrent costs, are suffering from conceptual tunnel vision. The "welfare"/"investment" dichotomy is much less clear-cut than is often perceived. For example, the adage, "Give me a fish and I'll eat today; give me a net and I'll never be hungry," could be converted to: "Give me a fish for the first 15 years of my life and I'll be strong enough and smart enough to build my own nets".

2. Self Reliance vs. Dependency

The capital costs of all projects have a fixed termination date. There is no termination date involved in recurrent costs. If the reason donors shun recurrent costs is that they wish to hold to the idea that development means increasing self-reliance, then the main consideration ought to be the prospects for the LDC to take on all the costs associated with each capital expenditure at some date in the future. Under such a rationale it would make sense to finance recurrent costs if there were a firm expectation that such funding was of a reasonable duration.

Thus, funding the recurrent costs of a research project for a number of years would not be inappropriate if there existed a plan for phasing out this type of assistance. Similarly, macro-level budgetary support of various types- PL 480, cash grants, or C.I.P.'s, also make sense as long as there is a realistic plan to move from this temporary situation to permanent self-reliance. This is, indeed, the ultimate objective of the entire aid program -- development allowed by graduation.

We must, however, be realistic about the time it will take for current low income countries to reach the level of development where they can be graduated. For the poorest LDC's, and it is in these countries where the recurrent cost problem is most severe, there is no prospect for becoming less dependent on aid flows in the near future. If average per capita income is $200 and a per capita income level of $1000 is a reasonable threshold for transition to "self-sustained growth," it will take 54 years at 3% per capita growth rates, and 81 years at a more moderate 2% per capital growth rate to reach that level of income. Only ten of thirty-eight low-income countries have exceeded sustained 2% per capita growth in the period 1960-78, and only four have exceeded 3%. While USAID, other donors and the countries themselves hope to improve this performance, and we can undoubtedly expect sectoral successes (such as Indian agriculture) in the medium term, nevertheless, if our objective is to graduate countries from dependence on aid we must be prepared for a long-term donor commitment to low income countries.

3. Donor Financing and Host Country Commitment

There is a danger that as donors pick up a larger proportion of the costs, both capital and recurrent, of a given project that LDC governments will be less involved with, and less committed to the success of that project. It would therefore, be useful to plan a phasing out of donor activities and a phasing in of LDC responsibility, both financial and managerial. Ifthe project is successful, then its support by the LDC government is assured. If it is unsuccessful, that support will not be forthcoming.

The critical question is how success is to be defined. Often donors and recipients have very different objectives, and thus the achievement of one set of objectives may mean the failure to achieve a different set. Only when objectives are congruent and projects are perceived by host governments as successful, will the continuation of project activities be assured. This will tend to be the case whether or not LDC governments have committed their resources to the project. In fact, the central objective in requiring LDC's to be involved financially in any project is to cause them to reveal their preferences, since the long run viability of any activity depends on the host government's placing sufficient value on project outputs.

Since LDC resources are scarce, host country agreement to fund a portion of any project suggests host country interest in that project.

However, if LDC's believe that the level of total donor financing is unrelated to a particular project portfolio, they will have no reason to conceal their preferences. In other words if a country believes that the U.S. aid level is to be $20 million, regardless of whether the money is put into the health or agricultural sectors, then there will be no reason to hide the fact that they would prefer investment in health. There should, therefore, be little difference in donor and host country objectives, and a financial commitment would not increase the degree of host country interest.

However, if LDC's perceive the total level of assistance as being related to the project portfolio, then they would be more likely to accept donor financing of, for example, a population program, even if they have no interest in it, because they expect that it is the population program or nothing. In this case there is little commitment to the project's success.

It would seem, then, that there is some cogency in the requirement that LDC's be obligated to fund some portion of development activities. (As appropriate, local currency generations of PL 480 programs can be used to fulfill the counterpart requirements of USAID projects.) This need not mean that all, or a major portion of, recurrent costs should be the responsibility of recipient countries. In fact, where the objectives of the donor and recipient are disparate, and donors wish to influence LDC governments to move their objectives closer to those of the donor it may be well to finance some portion of project recurrent costs beyond the development phase of the project in order to develop a coinsistency for project continuation.

B. Recipient Country Policies

Just as recurrent cost problems can be caused by donor policies that limit fungibility, recurrent cost problems are often caused by LDC government policies which create budgetary stringencies or lead to inappropriate projects. We can divide these recipient policy problems into three main areas -- government revenues, government expenditures, and project design problems. We will examine each of these in turn.

1. Government Revenues

It is possible that recurrent cost problems are due to the government's inability to raise revenues necessary to fund recurrent expenditures. Domestic revenues can be derived from either general tax receipts or from fees for specific services (i.e., user charges).

1. Tax Revenues

   Most, if not all, LDC's have difficulty mobilizing an economy's resources for the use of the government exchequer. However, there is a wide variety of experience in this respect, and Tait et al found the Tax/GNP ratio for 1972-76 ranging from 5.37% in Nepal to 37.6% in Iraq.² Much of this difference was due to the structure of the economy (mineral-rich countries finding it much easier to raise revenues). They then calculated International Comparison of Taxation (ICT) indices for a sample of LDC's. These ICT's were arrived at by regressing the tax to GDP ratio against a set of exogenous variables, determining what a normal tax effort at a given per capita income would be (given export earnings and mineral production), and calculating a given country's performance as a percentage of the norm. They then identified those countries whose ICT indices were above normal, as well as those below normal. While the underlying implication which follows from the analysis is that countries below normal should be able to improve their tax performance, one must be very careful about such Assertions. In the first place, since we are dealing with norms, it is statistically impossible for every country to be performing well (or badly), even though it is conceptually possible. Thus good performance means only that performance which is better than average. Secondly, the regression equations only explain half of the variance, at a maximum; there may be many other factors, beyond the capacity of the country to affect, which determine its performance. Thirdly, certain variables, such as the export ratio, may be determined by government policy rather than independent of policy as assumed by the model. Lastly, superior tax performance, does not seem to be correlated with GDP growth. Indeed, among those countries with low ICT's are Egypt, Ecuador, Malawi, and Togo, all of which have grown rapidly while among those with high ICT's are Algeria, Benin, Congo, Guyana, Sudan, Upper Volta, Zaire, and Zambia, all of which have stagnated. Indeed the growth rate for the low ICT countries is, on average, one-half percent higher than the medium and high ICT countries. This suggests that how governments use their resources and how their policies affect the private sector are more important than total tax effort in generating economic growth.

   Their results, for USAID recipients, are presented in Table III.

   TABLE III Summary measures of ICT Indices (47 countries)
   Low ICT Index (less than 0.89)	Medium ICT Index (0.89-1.09)	High ICT Index (greater than 1.09)
   Bangladesh	Burma	Benin
   C.A.R.	Burundi	Cameroon
   Ecuador	Costa Rica	Congo
   Egypt	Dominican Republic	Guinea
   El Salvador	Ghana	Guyana
   Gambia, The	Jamaica	India
   Guatemala	Jordan	Kenya
   Honduras	Mali	Morocco
   Indonesia	Peru	Pakistan
   Liberia	Senegal	Sudan
   Malawi	Sierra Leone	Tanzania
   Nepal	Sri Lanka	Tunisia
   Nicaragua	Swaziland	Upper Volta
   Panama	Thailand	Zaire
   Philippines	Yemen	Zambia
   Rwanda
   Togo
   Source: Tait, et al (1979), op. cit.

   If the country with a low ICT seems to be confronting a recurrent cost problem, then one recourse might be to increase the ability of the government to raise and collect taxes. In such cases, USAID assistance to improve tax administration is an appropriate response.

2. User Charges

   The general theory of public finance suggests that even in market-oriented economies there is a legitimate role for government provision of certain goods and service. More importantly, it makes sense under certain conditions for these goods to be paid for from general tax revenues rather than from fees paid by users.

   The basic question is who is to pay for any particular good or service produced by the government, the taxpayer or the user. If it is to be the taxpayer then that implies a degree of subsidization which has substantial import for recurrent cost considerations. Three arguments have been advanced in the public finance literature for government subsidies or specific taxation (taxes are negative subsidies) -- a public goods argument, a merit goods argument, and an equity argument.³

   1. Public Goods Rationale

      Development projects can be divided conceptually along a continuum measuring the degree to which the outputs partake of the nature of a public good. In economic terms, a public good is a good which cannot efficiently be provided by the private economy, because either (1) there is a great difficulty in excluding "free riders" or (2) there are externalities in consumption of the good. For example, fire protection can be justified on externality grounds as a public good because of the danger that a fire in one dwelling can spread to another (on the other hand, smoke alarms which tend to save lives more readily than property are private goods). City streets tend to be public goods while parking places are not, because of the excessive cost of collection as opposed to the value of the good. In other words, toll booths at every intersection would require a payment by the driver, both in time and money, many times the value of driving that city block.

      While there is a continuum from "private" to "public" goods, it might be useful to put goods into three discrete categories -- public, private, and mixed. An illustrative grouping is provided in Table IV.

      TABLE IV Public vs Private Goods: Who Should Pay?
      Public (totally subsidized)	Mixed (partially subsidized)	Private (no subsidy)
      National Defense	Communicable Disease	Agricultural Credit
      Public Health Measures such as mosquito spraying	Health Care	Agricultural Inputs
      	Agricultural Research	Non-Communicable
      	Forestry	Disease
      	Energy Research	Markets
      	Sewerage	Irrigation
      	Family Planning
      	Agricultural Extension
      	Potable Water
      	Education

      The public goods literature argues that on efficiency grounds, public goods should presumably be totally financed out of public funds, private goods should be totally financed out of user charges, and mixed goods by a combination of both. The basic argument underlying this conclusion is as follows.

      In a market system, prices are used as signals to both consumers and producers. For a purely private good, there is no divergence between social benefits and costs and the sum of private benefits and costs. Thus market-determined prices can be used as guides to allocate resources among competing uses. The value to society of an extra bag of maize is measured by the market-determined price of maize. A subsidy (or tax) on maize distorts price signals and would lead to either a greater (or lower) consumption of maize than is economically optimal.

      Markets are not likely to arise, however, for public goods because of their nature. For example, households are not likely to purchase mosquito-spraying services on a voluntary basis if they live in close proximity to other households who might not purchase such services. The non-purchasing households would obtain some external "free-rider" benefit and, moreover, there would be much less assurance that the total mosquito population had been eliminated or effectively controlled. Under these circumstances, the private market would tend to provide much less in the way of these services than is socially optimal; namely, total coverage financed by universal fees or tax revenues. On efficiency grounds, user charges for totally "private" goods, even if produced in the public sector, should be sufficient to cover the costs of production, thus eliminating the need for any recurrent finance out of general revenues. At this level of analysis it makes no difference whether the goods are produced by the government and sold at market prices, or produced and sold on the market by private firms. Since "mixed" goods should be partially subsidized and partially paid for by users, the price of mixed goods would not be sufficient to cover costs and it would therefore be necessary for some measure of recurrent finance. Lastly "public" goods should be totally financed by tax revenues. Again, in each of these instances, it does not matter conceptually whether the goods are produced by the government directly or by private firms which are then reimbursed by the government for the public portion of their costs, much as the way food stamps are provided in the U.S. One final note is in order here. In most cases it is inappropriate to involve the government in the provision of, as opposed to the subsidization of, development services which could and should be handled by the private sector. Government enterprises are often run less efficiently than private enterprises. These parastatals are generally insulated from the strictures of the market, by being able to run losses without being forced to cut costs or lower production. Management is frequently accountable to a different incentive structure than profits and losses. The government would be better advised to concentrate on those activities which are purely public or mixed goods, and create an environment conducive to private production of essentially private goods such a s fertilizer, credit, or health care which is not related to communicable diseases.⁴ LDC's have been pushed into many of these activities by donors who are then surprised when credit funds become decapitalized and physicians not paid.

   2. Merit Goods Rationale

      In the development context it is frequently argued that because of limited information or education, poor people undervalue certain basic commodities including types of health care, potable water, education, and nutritious foods. This argument requires that, to encourage poor people to consume these goods in greater quantities than they otherwise would, the goods must be partially financed by general government revenues. A similar argument is made on the production side where producers, particularly smallholders, are allegedly reluctant to adopt new technologies which require substantial purchases of modern inputs such as fertilizers, hybrid seeds, and insecticides. The reasons for this reluctance are a natural aversion to risk, plus a limited amount of information on the relationship between new technologies and increased productivity.⁵ The argument is then made that subsidizing modem inputs encourages the adoption of new technologies and is therefore justified.

      Both of these merit good arguments maintain that certain goods need to be subsidized for a fixed period of time until experience changes current preferences. It therefore follows that certain subsidies are justifiable if they are imposed for a limited period of time. In other words, subsidies imposed for merit goods should be accompanied by a timetable for their gradual elimination. The political problem is, of course, that it is much easier-to initiate a subsidy program than to eliminate one.

      In general, the merit good argument has little merit. Although there is some anecdotal evidence of "non-rational" consumption patterns among the poor, most careful budget studies show that poor people spend a very large portion of their income on "basic needs" goods.⁶ Table V reproduces an EICEL-Brookings study on consumer behavior in ten South American cities. As can be seen low income consumers in South American cities spend upwards of 85% of their incomes on "basic goods." Without clear evidence that consumer preferences are irrational, subsidy programs could lead to misallocations of resources, producing goods of lower value than non-subsidized programs would.

   TABLE V Basic Needs Expenditures as Percent of Family Budgets for the Lowest Income Quartiles (Ten South American Cities) a
   City	Mean Income of Lowest Quartile(S)	Percent
   Bogota, Columbia	1037	89.7
   Barranguila, Colombia	1100	88.7
   Cali, Colombia	1029	88.6
   Medellin, Colombia	1055	87.8
   Santiago, Chile	942	85.7
   Quito, Ecuador	845	90.5
   Quayaquil, Ecuador	896	91.4
   Lima, Peru	1161	80.8
   Caracas, Venezuela	1602	80.4
   Maracaibo, Venezuela	1452	82.0
   Source: Wheeler and Harris, op. cit. aRelevant definitions are as follows: "Basic" expenditure categories: Food and beverages; housing; clothing; medical; education. "non-Basic" expenditure categories: Furnishings and operations; recreation and culture; vehicle operation; public transportation; communication; other consumption (tobacco, personal care, ceremonies); insurance; gifts and transfers; other nonconsumption.

   3. Equity Rationale

      In many countries there are large groups of people with incomes so low that they face the ever-present danger of starvation. Governments often find it necessary to provide a safety-net, to insure that the minimum requirements for survival are met. The most efficient means of achieving an income-floor goal is through the use of the fiscal system. Few countries, and fewer LDC's, however, have a fiscal system sufficiently sophisticated to make such a program feasible. Consequently, governments have attempted to alleviate poverty by the direct provision of in-kind goods and services thought to be necessary. While this is clearly a "second best" solution, it may represent the only politically and administratively feasible mechanism for providing the necessary resources to the threatened populations.

      The critical problem with subsidies is that they are generally untargetted. The typical policy is universal free primary education, or universal free health care. Food subsidies are often imposed across the board without any clear distinction between those who are able to pay and those who are not.

      For example, consider a country with the income distribution and per capita income of India. Assume that tax revenues are 10% of GDP, and that the recurrent cost of development projects that provide services are $50 per person reached. (India's GNP/capita is $200). Then tax revenues will be $20 per person, and given the cost of services, only 40% of all people can be reached if taxes are the sole source of finance. If we divide the population into three groups: the upper group (top 20%) the poor (middle 40%) and the abysmally poor (bottom 40%), and then assume that all the upper group get the services with the remaining recipients divided equally among the two poorer groups, we obtain the result of Table VI, i.e., that 100% of the upper class get the service, while 25% of each of the next two groups get the services (sixty percent of the total population are excluded).

      Suppose, however we introduce user charges of $25 for the service (it's half a public good and half a private good). Also assume that every member of the two upper groups is able to pay for the service. If you then provided the service free of charge to the bottom group, you would be able to reach 100% of the upper group, 100% of the middle group, and the same 25% of the bottom group, or 70% of the population. These results are summarized in Table VI below:

      TABLE VI An Arithmetic Example of the Effect of User Charges on Service Delivery
      City	Without User Charges	With User Charges
      Cost Per Person	$50	$50
      Tax Revenue per Person	$20	$20
      User Charge Fees per Person	0	$25
      Total Coverage	40%	70%
      Coverage of Upper Group (P.C.I.=$500)	100%	100%
      Coverage of Middle Group (P.C.I.=$175)	25%	100%
      Coverage of Bottom Group (P.C.I.=$75)	25%	25%

      While the general level of poverty and tax collection limits the ability of the government to provide the public good to the entire population (the public good costs $50 per person and tax revenues are only $20 per person) coverage is obviously much broader when user changes are imposed on those groups able to pay.

      The equity question is not as simple as it first seems. Referring back to our India example we see that resources are only available to provide public goods to 25% of the poor population. Which twenty-five percent? In the absence of a purely political response the choice is arbitrary. The result is like a lottery. Without user charges, only 25% of the poor population will receive public services and there is no rational way of allocating these services among the poor.

      By imposing user changes and distributing services to all of the poor community, more people can benefit, though at a lower level of service. While a development project requiring recurrent finance of fifty dollars per year is not affordable for the whole community, one requiring twenty-five dollars might be. The choice becomes one of giving a few poor a high level of service or giving the many poor a lower level of service.

      To be sure there are indivisibilities. It may not be possible to provide everyone with piped water or secondary schools. But improved water supplies, functional literacy, or primary health care may be affordable for most communities, provided the level of services is consonant with the ability of the poor to pay.

      The following general principles apply to the question of subsidies and user charges:

      First, that for goods which are purely public in nature, user charges are inappropriate on efficiency grounds, and these goods should be financed by general revenues;

      Second, for goods which are partially public, e.g., education, partial financing from general revenues is legitimate, but should be carefully monitored. Third, subsidies may be appropriate for a limited period of time to induce producers (particularly small farmers) to adopt new technologies.

      Fourth, subsidies may be used to alleviate poverty through the provision of goods and services to the poorest in order to meet survival needs. Such subsidies should be limited to the segment of the population whose income needs to be raised, and not be provided across the board; ⁷ and Fifth, user charges should be levied on other goods and services provided by government.

2. Government Expenditure Policies

1. Government Budget Allocations

   Even if the government is capable of generating sufficient revenues, there may be recurrent cost problems in development projects, particularly in the health and education sectors, if the government is short-changing these sectors in favor of others. Four principal offenders here are defense expenditures, government enterprises, subsidies of various types, and large prestigious industrial projects. Low income USAID recipient governments are spending on recurrent account an average 2.7% of GDP on education, 1% on health, and another 1% on agriculture; whether these are appropriate depends on particular country circumstances. Table VII summarizes available data on budget allocations for these countries.

   TABLE VII Average Recurrent Government Expenditure Selectors 1%7-1973 (as percent of GDP)
   	Agriculture	Education	Health	Total
   Bolivia	0.4	3.7	0.8	4.9
   Burma	0.9	2.5	1.0	4.4
   Burundi	0.6	2.6	0.7	3.9
   Cameroon	0.6	2.2	n.a	n.a
   Gambia	1.9	2.4	1.7	6.0
   Ghana	0.9	3.7	1.1	5.7
   Guatemala	0.2	1.7	0.9	2.8
   Honduras	0.6	2.8	0.5	3.9
   Kenya	1.3	3.3	1.1	5.9
   Lesotho	2.3	4.9	2.2	9.4
   Liberia	0.4	1.9	1.0	3.4
   Malawi	1.2	2.7	1.1	5.0
   Mali	0.6	2.4	1.2	4.0
   Nepal	0.1	0.3	0.1	0.5
   Philippines	0.5	2.4	0.4	3.3
   Rwanda	0.4	2.5	0.9	3.8
   Senegal	0.9	3.4	1.4	5.7
   Somalia	0.8	1.5	1.7	4.0
   SriLanka	0.6	3.7	1.8	6.1
   Sudan	1.7	2.0	1.1	4.8
   Swaziland	1.9	3.8	1.7	7.4
   Tanzania	1.2	2.7	1.2	5.1
   Thailand	0.7	2.2	0.5	3.4
   Togo	0.6	1.6	0.8	3.0
   Upper Volta	0.5	1.6	0.7	2.8
   Zaire	0.2	4.5	0.6	5.3
   Zambia	2.7	4.0	1.8	8.5
   Mean	0.9	2.7	1.1	4.7
   Source: World Bank Tables, 1976

   Alan Tait and Peter Heller of the IMF have developed an "International Expenditure Comparison Index," similar to the "International Comparison of Taxation Index" discussed above. While these comparisons should be treated with even greater caution than that suggested for interpreting the ICT, the results are interesting and in some cases provocative. Table VIII summarizes the results for USAID recipients for which these Indices were calculated.

   TABLE VIII International Expenditure Comparison Index, 1977: Functional Categories of Expenditures
   	General Public Services	Defense	Education	Health	Agriculture	Transport and Communications
   Bangladesh	Low	Low	Low	Average	Average	High
   Bolivia	Average	Average	Average	Low	Low	Average
   Botswana	Average	Low	High	High	High	n. a.
   Burma	Average	High	Low	Low	High	n. a.
   Burundi	Low	High	Average	n. a.	Average	n. a.
   Cameroona	High	Average	Average	Low	Low	High
   Chad	Average	High	Average	Low	Average	Low
   Costa Rica	Low	Low	Average	Low	Low	High
   Dominican Rep.	Low	Low	Low	Average	Average	Low
   Ecuador	Low	Average	Average	Low	Average	n. a.
   Egypt	Low	Low	High	Average	High	Low
   El Salvador	Average	Low	Average	Average	Low	Average
   Gambia, The	High	n. a.	Average	High	High	n. a.
   Ghana	Average	n. a.	High	Average	Average	Average
   Guatemala	Low	Low	n. a.	Low	Low	n. a.
   Hondurasa	High	Low	Average	High	Low	n. a.
   Jamaica	Low	Low	High	High	High	Average
   Jordanb	Average	High	High	High	High	High
   Kenya	Low	Average	Average	Average	Average	Average
   Lesothoc	High	n. a.	High	Average	High	n. a.
   Liberia	High	Low	Average	High	Average	Average
   Malawi	Average	Low	Average	Low	Average	Low
   Malia	Average	High	High	Average	Low	Low
   Morocco	High	High	High	Average	n. a.	n. a.
   Nepal	Low	Average	Low	n. a.	Average	n. a.
   Nicaraguaa	Low	Low	Low	Low	Low	n. a.
   Niger	High	Low	High	Average	Low	Average
   Pakistan	Low	High	Low	Low	Low	Average
   Panama	High	n. a.	Average	High	Average	Average
   Peru	Average	Low	Average	Low	Average	n. a.
   Philippinesa	Average	Average	Low	Low	Average	High
   Rwanda	Low	High	Low	Low	Low	High
   Senegalb	Average	Low	Average	Low	Low	Low
   Sierra Leonea	Average	Low	Average	Average	Low	Low
   Somalia	High	High	High	High	High	Low
   Sri Lanka	Low	Low	Average	Average	Average	n. a.
   Sudan	Low	Average	Low	Low	High	Average
   Swaziland	Average	Low	High	Average	High	n. a.
   Tanzania	Average	Average	Average	High	High	Average
   Thailand	Low	High	Average	Low	Average	Average
   Upper Voltac	Average	High	Average	Low	Low	Low
   Yemen	Average	High	Low	Low	Low	Average
   Zambia	High	n. a.	High	High	High	High
   Source: Alan A. Tait and Peter S. Heller, "Intemational Comparisons of Government Expenditures: A Starting Point for Discussion IMF, Discussion Memorandum," DM/81/53, July, 1981. Each index runs from zero to four hundred. Any value less than 75 was recorded as "Low", values between 75 and 125 were recorded as "Average", and values greater than 125 were recorded as "High". a1976 b1975 c1974 d1978 e1973

   Frequently, a recurrent cost problem in a given sector is an expression of government priorities. Failure to provide school supplies or petrol for extension workers may reflect the government's decision that these activities are less important than subsidizing agricultural inputs or financing a new steel complex. Where requisite steps to meet recurrent costs are not undertaken, USAID should seriously consider reducing the level of assistance to the affected sector. There is little point in developing projects that call for host government resources to be successful, if those resources are not likely to be forthcoming.

2. Cost of Government Activities

   Sometimes government budgets become tight because the costs of producing government goods and services are excessive. This is frequently due to inflated salary schedules, as is particularly the case in much of the Sahel. For example, let us examine primary education in Upper Volta by comparing its situation with that of Burma and Malawi. Burma, with a nominal GNP per capita level similar to Upper Volta's has enrollment levels of 80%, (five times that of Upper Volta) and Malawi with a per capita income level 12.5% higher has enrollment levels of 60% (almost four times that of Upper Volta). How are these countries able to finance the recurrent expenditures and Upper Volta not?

   TABLE IX Comparative Cost Figures for Upper Volta, Malawi, and Burma
   	Upper Volta	Malawi	Burma
   Per Capita Income (US$)	160	180	150
   Per Cent of Population between 5 and 14 (%)	25	25	25
   Primary Enrollment Rate	16	62	80
   Tax/ GDP Ratio (%)	11.3	10.1	7.6
   Proportion of Govt Exp to Primary Education(%)	8.8	8.5	10.5
   Percent of GDP devoted to Primary Education(%)	1.0	0.9	0.8
   Per Pupil Expenditures (US$)	40	10	6
   Pupil/Teacher Ratio	46	58	7

   The answer is clear. Upper Volta does every bit as well as Malawi and Burma in generating tax revenue and allocating portions of both government budget and GDP to primary education. The key differences is in per pupil expenditures which are four times that of Malawi and almost seven times that of Burma. Often differences in costs are related to particular resource endowments -- either of terrain, population density, or skilled labor. However, in other cases there is no justification for the cost structure of government activities. For example, it has been found that the ratio of public sector wages to the income of a peasant in Mali is approximately twenty to one. In India the ratio is probably closer to six to one. Such a differential may in part be due to the relative shortage of administrative skills, but a larger portion is undoubtedly due to excessive levels of compensation. A clear indicator of the compensation question is whether the private sector or the public sector is exhibiting the greatest shortage of skilled personnel. If the Government sector is having difficulty attracting skilled personnel then public wages are not too high. On the other hand if the private sector is having difficulty attracting trained people then government wages may very well be out of line.

   Salaries are the main cause of high costs, but inappropriate technologies are also important. A health care sector that is made up of high technology curative hospitals is going to have higher costs per beneficiary than a primary care system. This is true for all sectors from education to rural development to road construction. If costs are out of line on a sectoral basis (due to inappropriate technology) donors should tend to stay out of that sector. If costs are out of line in every sector, then clearly, financing of recurrent costs in any sector would be inappropriate, since it would involve subsidizing those inputs which are priced too high.

3. Macroeconomic Policies

Frequently, recurrent cost problems arise through the failure of government fiscal and monetary policies. For example, an overvalued domestic currency coupled with a government marketing system will destroy a country's fiscal base. In Ghana, for example, where the cedi is overvalued by perhaps 10:1, it is impossible for the government to pay cocoa farmers anything like the true market price. Cocoa production has declined precipitously. This has meant a shortfall not only in foreign exchange, but also in government revenues.

As noted in the introduction to this paper, the economic crises threatening most LDC's have led to the acceptance of IMF stabilization packages with an attendant austere government budget. Consequently, many countries are going to face a reduction in their ability to finance the recurrent costs of their development portfolio over the near term.

There are many instances where an apparent recurrent cost problem is due to an over-optimistic attitude on the part of government as to what it can achieve. For example, Egypt and Mali are committed to hire every secondary school leaver. This results in an overblown public service and a shortage of manpower in the private sector. The recurrent burdens of such a policy are enormous.

Similarly, governments may be too ambitious in addressing the needs of the population given the resources they have available. Universal primary eduction may not be appropriate in poor countries with 25% of their population in this age group. The same argument can be advanced for a variety of activities, some worthwhile but expensive, others worthless and expensive.

The recurrent cost problem is, thus, often due to inappropriate government policy, and the rational donor response is to endeavor to persuade governments to change the policy, or to develop activities the success of which is not dependent on recurrent financing which will not be forthcoming.

1. Foreign Exchange Constraints

   It is generally, though not universally, true that the recurrent cost problem is manifested through the shortage of foreign exchange to purchase important intermediate inputs into the production of government services. Thus, there is more typically a scarcity of drugs than medical officers, of school supplies, than teachers, and of petrol than extension workers. One indicator of country performance in this regard is the ratio of expenditures on other goods and services (largely imported) to expenditures on wages and salaries. The IMF expenditure study referred to above has calculated an international expenditure comparison (IEC) index for both "other goods and services" and "wages and salaries" categories of expenditures. The ratio of these two indices is a useful measure of the degree to which there is apparent underspending on materials and replacement of capital equipment, and thus, a likely foreign exchange constraint. These ratios are presented in Table X.

   TABLE X: Ratios of IEC Indices for Other Goods and Services to Wages*
   Ratios below one		Ratios greater than or equal to one
   Bolivia	0.66	Botswana	1.11
   Dominican Republic	0.51	Cameroon	1.06
   Guatemala	0.74	Costa Rica	2.70
   Liberia	0.69	Egypt	1.03
   Mali	0.33	Gambia, The	1.51
   Mauritius	0.59	Honduras	1.16
   Morocco	0.41	Jamaica	1.31
   Rwanda	0.94	Kenya	1.05
   Swaziland	0.90	Malawi	2.21
   Tunisia	0.97	Nicaragua	1.41
   Turkey	0.55	Niger	1.79
   Upper Volta	0.25	Panama	1.16
   		Paraquay	1.35
   		Philippines	1.28
   		Senegal	1.12
   		Sierra Leone	2.12
   		Sri Lanka	1.07
   		Sudan	2.91
   		Tanzania	1.11
   		Thailand	2.41
   Source: Alan A. Tait and Peter S. Heller, "International Comparisons of Government Expenditures: A Starting Point for Discussion," IMF/DM/81/53, July, 1981. *Low ratios imply either (1) the presence of a foreign exchange constraint or (2) high wage structure or (3) both.

   However, even if the recurrent cost problem manifests itself in terms of a shortage of imported inputs rather than locally financed ones, the analysis and the prescriptions presented thus far in this paper remain valid. A recurrent cost problem is defined as a situation in which the stream of returns to the recurrent imported factor of production is greater than the stream of returns to the fixed imported factor. This situation results from either donor policies or recipient policies, and the situation can be resolved by policy reform.

4. Project Design Failures Capital under-utilization which is particularly concentrated in one development sector indicates that the return to the host government of recurrent expenditures in that sector is less at the margin than was anticipated in the project design. A basic economic principle is that fixed costs are fixed costs. Once a health center has been built its capital costs are zero. The government must allocate its recurrent resources among a large number of activities, all of which , presumably have a quantity of capital attached to them. In the project design the returns to the investment were calculated on the expectation that they would be fully utilized through the provision of complementary variable resources -- labor, materials, maintenance, etc. However once in place it makes sense for the government to recalculate the social benefits of its allocation of scarce resources among various activities. The new calculation may differ from the original one due to a number of factors, e.g,:

1. government priorities might change
2. the original design was faulty
3. the values of certain variables such as prices were predicted incorrectly.

If road maintenance is underfinanced, while other sectors are operating with little overt underutilization, then a donor should be very circumspect about getting involved in this sector. At the least it should investigate previous road building projects to determine whether the cause for the underfinancing was a low governmental priority or poor design, and redesign its project accordingly. In general, underutilization confined to a particular sector is prima facie evidence that the expected rates of return to investments in that sector were much too optimistic. In such cases the appropriate response is much more careful design in order to insure that investment decisions anticipate likely responses. In particular, costs of recurrent inputs are often underestimated due to lack of donor coordination. For example, if a number of donors are doing projects that require skilled workers for maintenance, the sum of their demands might raise the price of these workers well beyond the levels anticipated. ⁸

Another problem with project design is the tendency to be overly optimistic about the time necessary for a project to move from the development stage to what might be called the operational stage. During the early stages of a project it is expected that project revenues will be low and expenditures high. It is likely that this situation will continue beyond the typical five year implementation period, thus leading to demands on the public fisc.

In the preceding paragraphs project designs were faulted for failing to anticipate changing environments correctly. Often it is the policy environment that is most at fault. This is particularly the case in income generating projects which should never, if designed correctly, be the source of recurrent cost problems. Any investment is expected to generate a positive rate of return. For an income generation project that return should begin being realized even before the development phase of the project is concluded. Such returns should be more than sufficient to cover the recurrent costs of the project. If a farmer or a herder is increasing his income by a greater amount than the costs of the inputs leading to that increase, he should be willing to pay for it. If a project is not designed so as to capture these variable costs then it is poorly designed.

It is true, however, that many projects will not become financially viable immediately, but require a development phase that allows a time for learning how to use new technologies efficiently. Thus the adoption of new technologies, particularly those with substantial capital costs, such as irrigation or animal traction packages, will have negative cash flows in the start-up or development phase. There is, therefore, a need to develop medium and long-term credit institutions which will provide the resources necessary for investment.

1. Local Participation

   Recurrent cost problems, as we have seen, have many causes. Local communities have substantial resources, both in cash and kind, which can be utilized to finance the recurrent and capital costs of development projects.

   In order to mobilize these resources, local communities need to be shown that it is in their interest, that the benefits of the project exceed the costs. Projects which are designed with the participation of local communities are frequently more responsive to local needs and conditions than centrally designed projects and thus are more likely to elicit local support.

   Moreover, leaving maintenance and finance to local communities is likely to increase the commitment of the beneficiaries to the project and therefore to insure its sustainability. If it is the local community that is to maintain irrigation ditches rather than a government parastatal, the ditches are more likely to be maintained.

   Local communities are also able to mobilize resources in kind and transform them into cash. For example, a water project which saves women's time might be accompanied by a gardening project which converts that time into revenue, part of which could be used to pay for the water.

²Tait et al, "International Comparison of Taxation for Developing Countries, 1972-76" IMF Staff Papers, Vol. 26, No. 1 (March, 1979).
[return to text]

³See for example, Joh G. Head, Public Goods and Public Welfare (Duke University Press, Durham, N.C. 1974).
[return to text]

⁴PPC is currently drafting a policy paper on the role of parastatals in development.
[return to text]

⁵In addition, the weakness of long-term financial markets inhibits the adoption of new technologies because many of these technical packages are associated with negative cash flows during the start-up period.
[return to text]

⁶See David Wheeler and John Harris, "Recurrent Costs and Basic Needs Strategies." (USAID/OTR-G-1733, June 1980).
[return to text]

⁷There are several ways of targeting subsidies. Food subsidies could be limited to those foods preferred by the poor, such as cassava in many countries. Water piped to a house would not be sudsidized while community stand pipes could be. There could be subsidies for certain levels of health care but not for others. These complex issues will be discussed in greater depth in a forthcoming policy paper on subsidies.
[return to text]

⁸See A. Mead Over, "On the Care and Feeding of a Gift Horse: The Recurrent Cost and Problem and Optimal Reduction of Current Inputs," Williams College, Development Studies Program, Research Memorandum, No. 79, Jan. 1981).
[return to text]