World Bank Reprint Series: Number 322 Nicholas Prescott and Jeremy Warford UEconornc Appraiaal in the "rI"ealth Sector Reprinted with permission from Kenneth Lee and Anne Mills, eds., The Economics of Health in Developing Counties (Oxford: Oxford University Press, 1983), pp. 127-45. World Bank Reprints No. 281. Bela Balassa, "Trade Policy in Mexico," Wwrld Dtvhelopnlrt No. 28 Ia. Bela Balassa, "La politica de comercio exterior de Mexico,' Coniercio Exterior No. 282. Clive Bell and Shantayanan Devarajan, "Shadow Prices for Project Evaluation under Altermative Macroeconomic Specifications," The Quarterlty Journal of EconoDmics No. 283. Anne 0. Krueger, "Trade Policies in Developing Countries," Handbook of Itfernational Economnics No. 284. Anne 0. Krueger and Baran Tuncer, "An Empirical Test of the Infant Industry Argument," Amiierican Econoitii Reviewzz No. 285. Bela Balassa, "Economic Policies in Portugal," Economiia No. 286. F. Bourguignon, G. Michel, and D. Miqueu, "Short-run Rigidities and Long-run Adjustments in a Computable General Equilibrium Model of Income Distribution and Development," Journal of Development Economics No. 287. Michael A. Cohen, "The Chllenerge of Replicability: Toward a New Paradigm for Urban Shelter in Developing Countries," Regional Development Dialogue No. 288. Hollis B. Chenery, "Interaction between Theory and Observation in Development," lAVorld Development No. 2'3 J. B. Krnight and R. H. Sabot, "Educational Expansion and the Kuznets Effect," The Amierictan Economnic Review No. 290. Malcolm D. Bale and Ulrich Koester, "Maginot Line of European Farm Policies," The V'Vrld Fconuliii No. .2 C" 3. Danny M. Leipziger, "Lending versus Giving: The Economics of Foreign Assistance," Vbrld Developmnent No. 292. Gregory K. Ingram, "Land in Perspective: Its Role in the Structure of Cities," World Co,l'ress on Land Polici, 1 i8&) No. 293. Rakesh Mohan and Rodrigo Villamizar, "The Evolution of Land Values in the Context of Rapid Urban Growth: A Case Study of Bogota and Cali, Colombia," World Congress on Landl Polict, 1980 No. 294. Barend A. de Vries, "International Ramifications of the Extemal Debt Situation," The AMEX Bank Review Special Papers No. 2 9. Rakesh Mohan, "The Morphology of Urbanisation in India," Economic and Political Weekil' No. 296. Dean T. Jamison and Peter R. Moock, "Farmer Education and Farm Efficiency in Nepal: The Role of Schooling, Extension Services, and Cognitive Skills," World Development No. 297, Sweder van Wijnbergen, "The 'Dutch Disease': A Disease after All?" The Econiomic Journal No. 298. Ame Drud and Wafik M. Grais, "Macroeconomic Adjustment in Thailand: Demand Management and Supply Conditions," Journral of Policy Modeling No. 2c9. Shujiro Urata, "Factor Inputs and Japanese Manufacturing Trade Structure," Thle Review Of' L 1 7Ioomics anid Statistics No. 300. Dipak Mazumdar, "The Rural-Urban Wage Gap Migration and the Working of Urban Labor Market: An Interpretation Based on a Study of the Workers of Bombay City," Indian Econb?nic Reo'iezv No. 3.01. Gershon Feder and Roger Slade, "Contact Farmer Selection and Extension Visits: The Training and Visit Extension System in Haryana, India," Quarterly Jountal of Inrterna- tionial Aric irlfl .c 7 Economic appraisal in the health sector Nicholas Prescott and Jeremy Warford Introduction This paper discusses the role of economic appraisal in the formulation of health sector policy in less developed countries (LDCs). Specifically, it focuses on the application of economic analysis to public investment decisions in the health sector, that is to the design, selection, and financing of projects which constitute the health component of a development plan. The underlying theme is the desirability of integrating the health sector more closely with the overall frame- work of development planning through a systematic process of project appraisal based on clearly specified objectives and resource constraints. Generalizations about LDCs as a group are inevitably heroic when-as defimed by the World Bank-they comprise countries as diverse as India, Lesotho, Fiji, Korea, Brazil, and Chad, However, by definition, all of these countries share the characteristic of a relatively low level of per capita income */orld Bank 1981), typically associated with a lirmited tax base due to weak tax administration and low ratios of taxable surplus to GNP (Tait, Gratz, and Eichengreen 1979; Meerman 1980). The combination of these factors produces an acute shortage of resources to finance public expenditure, thus emphasizing the iimportance of achieving efficient investment plaining in the public sector. This is particularly true today when the prospects for per capita income growth in LDCs have deteriorated markedly (World Bank 1981) and the adoption of structural adjustment policies is enforcing a more rigorous review of public investment programmnes (Balassa 1982). Other characteristics commonly observed in LDCs include overvalued exchange rates, substantial unemployment or underemployment, and a skewed income distribution, together with a variety of constraints which limnt the ability of their governments to increase private savings and redistribute income with the conventional instruments of fiscal and monetary policy. These factors have important implications for project selection in the health sector. Resource allocation decisions have in the past been notoriously in- efficient and inequitable, and are now reflected in an emphasis on expensive urban and hospital-based curative care which is not directed at the major causes of ill-health in the majority of LDC populations. Moreover, health has tended to claim a low and often declining share of public expenditure (IMF 1981), and the prospects for a substantial increase in real budgetary allocations to 128 The economics of health in developingcountries the sector are poor. At the same time, recent interest in basic needs and the efforts by WHO and UNICEF to promote 'Health for all by the year 2000' have created ambitious objectives for health improvement in LDCs. The Director- General of WHO has suggested that an infant mortality rate of less than 50 per thousand live births, and a life expectancy at birth greater than 60 years, should be minimum objectives for health improvement by the year 2000 (Mahler 1977). These contrast with infant mortality rates higher than 200 per thousand, and life expectancy lower than 40 years, observed in some low-income LDCs. The financial implications of these objectives are enormous. A recent WHO estimate implies that their achievement would require up to an eightfold increase over the present level of public expenditure on health in low-income LDCs of approximately $2.5 per capita (WHO 1981). However, the health sector con- tinues to operate, to an extent that is unmatched in any other sector, without the guidance of systematic project appraisal criteria which could help to recon- cile these health objectives with other dev'elopment objectives in the face of increasingly severe resource constraints. It is, therefore, essential to develop ways of improving the efficiency of public expenditure on health in LDCs, and also of mobilizing additional revenue to finance increased levels of expenditure. This chapter examines these issues in the following framework. First, the basic elements of financial analysis which provide the starting point for analysis of efficiency and resource mobilization are discussed. Then there are two sections which review alternative forms of efficiency criteria, namely cost-effectiveness and cost-benefit analysis, with special attention to the treatment of uncertainty and shadow pricing in health projects. The final section concludes with a consideration of cost recovery options to mobilize additional resources from outside the public sector. Financial analysis Health ministries, using conventional public accounting procedures in LDCs, are generally unaware in detail of the allocation of public resources between different interventions and its relationship to stated health objectives, or of the unit costs of different interventions, or of the full potential of available sources of finance. The situation is complicated by the activities of numerous other agencies, both public and private, that affect health, major examples being those agencies concerned with water supply, nutrition, and pharmaceutical production and distribution. This lack of basic financial data severely limits the ability of health ministries to make systematic judgements about the merits of existing and proposed expenditure patterns and financing policies in the public sector, and reform of their financial accounting systems is clearly indicated. The starting point for economic analysis of investment decisions should be a detailed analysis of total public and private health-related expenditures in terms of expenditure patterns and sources of finance. Chapter 3 is devoted to this set of issues, but it is appropriate here to highlight a few of the main features. Economic appraisal in the health sector 129 Expenditure patterns. Analysis of the allocation of public expenditure between different interventions indicates the extent to which the current expenditure pattern responds to a country's major health problems and proclaimed objec- tives. TIus, it helps to signal the desirability of reallocating existing expenditures or undertaking new investments. If, as rhetoric suggests, distributional objectives are also important, analysis of the distribution of expenditure by beneficiaries in different income classes (Meerman 1979; Selowsky 1979; Tan 1975), or sinply by region (Ofosu-Amaah 1975), will be similarly revealing. Estimates of the unit costs of different interventions, making an appropriate -if not scientific-allocation of joint costs, provide additional signals. Ideally, these should be estimated in terms of long-run marginal costs, which relate capital expansion and recurrent (i.e. operating and replacement) costs to the volume of additional services provided (Saunders, Warford, and Mann 1977). Differences in the unit cost of interventions suggest the possibility of efficiency gains by switching expenditure to lower cost interventions if these are equally effective. The same applies to unit cost differences for similar interventions delivered by different facilities. For example, a study in Malaysia found that the unit cost of out-patient services in rural clinics exceeded those provided by hospitals, suggesting the existence of significant excess capacity in the rural clinic system (Meerman 1979). Similar comparisons can be made between the public and private sectors. Data collected in Lesotho (Smith 1980) indicate that the public sector is a relatively inefficient producer of health services. The average cost per in-patient was almost three times higher in public than private hospitals, and similar differentials existed for hospital and clinic out- patients. Although unadjusted for case-mix and labour cost differences, it is improbable that these factors fully accounted for such high unit cost differ- entials. Finally, with the objective of cost containment in mind, trends in the unit costs of similar interventions at different points in time may be used to monitor efficiency in service provision. Sources of finance. In a similar way, analysis of the sources of finance of existing health expenditure patterns-by public revenues, external donors, private consumers, employers, and charity organizations-can provide important signals to more efficient methods of mobilizing resources for the health sector. For example, available data showing that a major proportion of total health expenditure is financed by private consumers (VHO 1977) suggests the existence of a high level of private willingness (and ability) to pay for health services which remains largely untapped by the public sector. This type of analysis relates intimately to the analysis of expenditure patterns since the potential role of various financing mechanisms, ranging from direct user charges at one extreme to full subsidy from public revenues at the other, tends to vary across different interventions. Recurrent cost financing. A critically important aspect of the analysis of health sector financing, and one which emphasizes the importance of developing 130 The economics of health in developingcountries alternative cost recovery mechanisms, concerns the future availability of fiscal resources to subsidize the recurrent costs generated by new investment projects in the health sector. The availability of public funds to subsidize ongoing oper- ating and maintenance costs is typically the binding constraint on health invest- ments. The reason is that external finance from aid donors is often forthcoming to cover capital investment costs, while f, :ure recurrent costs have to be financed from domestic revenues after project implemeaitation. The problem of recurrent cost financing for social sector projects was recognized a long time ago (Stolper 1966), and has recently attracted attention throughout the public sector in LDCs (Heller 1979; Club du Sahel 1980). There are a number of reasons wvhy it is especially important in the health sector. First of all, the neglect of project analysis criteria for health projc-ets has tended to preclude attention even to elementary financial appraisal. Secondly, the amount of recurrent expenditure generated per unit of invest- ment in the health sector is typically higher than in most other sectors, pnrn- cipally because of the labour and phannaceutical requirements of health facilities. Moreover, the amount of recurrent public subsidy paid per unit of recurrent expenditure is also relatively high in the health sector. Most public sector health projects are financed exclusively, or largely so, from public revenues. lealth sector pricing policies are such that user charges are frequently nominal and recover only a low proportion of operating costs. In addition, the revenues generated by this or other cost recovery mechanisms usually accrue directly tr, the central government Treasury, without any earmarked allocation to the health sector or to the project itself. Thus the allocation of investment resources to health tends to have more onerous implications for recurrent expendituTe financing than in other sectors. Public finance constraints have clear implications for the analysis of project choice in the health sector. Whether partiaBy or totally financed by public subsidy, every project requires a cash flow of revenues to cover its financial costs in order to operate at optimal scale. Analysis of expected future financial costs and revenues of the project is critically important to detennine the feas- ibility of a project over its planned operating life. Indeed, analysis of long-run financial feasibility itself provides a partial test of efficiency simply because future underfinancing of recurrent costs will result in the operating of the project at less than optimal scale and, hence, in a reduction in the net benefit of the investment. A useful way of summarizing the recurrent expenditure implications of health projects is to estimate the ratio of annual recurrent costs to total invest- ment costs, the so-called 'r-coefficient'. For example, $1000 invested in the health sector in Malawi is expected to generate, on average, $250 of incremental recurrent costs per year, compared with only about $100 resulting from the same investment in education or agriculture (World Bank 1982). Moreover, the recurrent cost implications of health investment are higher at lower levels of the health sector. Estimated recurrent cost ratios in Malawi average 0.5 for Economic appraisal in the health sector 131 clinics and primary health care compared to 0.2 for hospitals. Similar patterns have been observed in Kenya and Malaysia (Heller 1974, 1975) and the Sahel countries (Over 1979). Consideration of financial feasibility leads directly intct the analysis of efficiency criteria. Clearly, the objective of planners is not just to ensure a financially feasible choice of health projects but the best choice from various possible alternatives. One approach, using cost-effectiveness analysis, is to select the mix of feasible projects which yields the largest expected improvement in health status, relative to cost. Ideally, the use of cost-benefit analysis would be preferable, to select those projects which have the largest net social benefit, defined in terms of a wider set of development objectives. Some of the key problems involved in cost-effectiveness and cost-benefit analysis are considered below. Cost-effectiveness analysis The least controversial efficiency criterion for health project selection, that of cost-effectiveness, is to choose projects which yield the maximum health improvement subject to available resources. Alternatively, the problem can be inverted to choose projects which minimize the cost of meeting a specified objective or objectives.Cost-effectiveness analysis (CEA) is frequently used in sectors where problems of benefit valuation occur. For example, electricity investment planning models are typically specified in cost-minimization form, to find the least-cost solution to meeting an exogenous future demand upon which a value may or may not be placed (Turvey and Anderson 1977). The method lends itself naturally to the health sector where the concept of meeting basic needs suggests that quantifiable minimum standards exist and can serve as planning objectives. Unfortunately, the basic needs literature is still distin- guished by an absence of concrete suggestions as to what these standards should be (see, for example, Streeten, Burki, Ul Haq, Hicks, and Stewart 1981, or Richards and Leonor 1982). A fundamental problem in applying CEA to the health sector lies in the choice of an appropriate unit of effectiveness to measure the output of health projects. It may seem obvious that health improvement is the primary objective of public expenditure on health. Effectiveness measures framed in terms of intermediate or service outputs, such as provision of MCH services, are therefore inadequate tests since they do not measure actual improvements in health status. However, a central. difficulty in measuring health status is that it is not homogeneous, since health project outputs comprise reductions in both morbidity and mortality. These two types of health improvement can be measured in different ways or aggregated into a composite measure whici can itself be constructed in different ways. Examples of alternative approaches are discussed below. Morbidity reduction. The most straightforward, but limited, measure of health impact is the reduction in the prevalence or incidence of a specific disease. An example is provided by Rosenfield, Smith, and Wolman (1977), who used 132 The economics of health in developingcountries a simple model of schistosomiasis transmission to simulate the effectiveness in Iran of different control techniques (mollusciciding, engineering measures, chemotherapy, and a combination of these) subject to a given resource con, straint over seven years. Maximum output, specified in terms of the greatest reduction in the prevalence rate obtained by the end of the seven-year period, was achieved with a combination of chemotherapy and mollusciciding. This intervention reduced the prevalence rate from 64 per cent to 20 per cent, whereas the next best altemative, chemotherapy, achieved a terminal prevalence rate of 60 per cent. This measure of effectiveness did not take account of the prevalence reductions achieved during the period. A more appropriate measure, in terms of the total number of cases of schinosomniasis prevented over the period, changed the ranking of alternatives. Chemotherapy yielded the greatest output at a cost per case prevented of $1.26, followed by the combination of controls with a unit cost of $1.29. Measures of effectiveness specified in terms of cases of a disease prevented tend to limit the application of CEA to the choice between different methods of controlling that disease, and preclude its use to evaluate the choice between interventions directed at different diseases. This is so because different diseases have different effects on the duration and extent of morbidity, and also mortality, which are not captured by measures of cases prevented. A more useful measure which would permX . omparison of the morbidity effects of the prevention of different types of diseases is the number of days of disability prevented. This can be expressed as the number of days on which individuals would have experienced some degree of dysfunction due to the relevant disease, and can be weighted to reflect the degree of impairment. Paqueo (1976) illustrates the use of a weighted measure to show higher morbidity rates below the poverty line in the Philippines. An unweighted measure has been used in Indonesia by Grosse, De Vries, Tilden, Dievler, and Day (1979) and Grosse (1980) in a detailed simulation of the effectiveness of alternative health interventions directed at 31 diseases, and subject to seven alternative resource constraints. The choice of activities comprised 48 possible combinations of curative and preventive interventions based on health centres, health sub-centres and village health workers, and sanitation, immunization, and nutrition programmes. Resource constraints varied from $2.06 to $30.00 per capita per year. Simulations at each level of resource availability identified the mix of interventions which minimized the number of days of incapacitating illness per person per year from all causes. At all resource levels health centres were selected but with varying combinations of other inputs. At $5.00 per capita these included village health workers and an immunization programme; at $15.00 a sanitation programme only; and at $30.00 village health workers together with sanitation, immunization, and nutrition programmes. Mortality reduction. As is the case with morbidity, mortality reduction can be quantified in different ways. The simplest measure, the number of deaths Economic appra4al in the health sector 133 prevented, has the advantage of not being disease-specific and therefore can be used to compare interventions against different diseases. However, in the aggre- gation of deaths prevented it implicitly assigns the same weight to all regardless of the age at death, whereas a social premium may be attached to the prevention of, for instance, infant deaths or deaths occurring in the productive age groups. Social preferences of this kind can be introduced by using methods which weight deaths prevented at different ages by the additional years of life accruing to the survivors. One approach is to assume that all survivors live to some arbitrarily deter- mined age. For example, Romeder and McWhinnie (1977) suggest a measure of potential years of life lost (gained) could be given by taking the difference between age at death and 70 years. Other methods take the terminal year as life expectancy at birth, or the life table age at which fewer than an arbitrary proportion, such as 10 per cent, of the original radix are survivors. These methods are unsatisfactory because they do not take account of the probability of survival from the age at death to the hypothetical terminal year. It is more appropriate to measure the potential years of life gained by the expectation of life at the age at which death is averted, estimated from the relevant life table. Other refinements have also been suggested, particularly to take account of selective biases which cause the survival chances of survivors from some causes of death to differ from the cohort average (Shepard and Zeckhauser 1980), as rnay be the case with measles vaccination in LDCs (The Kasongo Project Team 1981). Weighting for specific age groups can be made even more extreme by simply giving zero weight to certain age groups. Thus planners may be interested exclusively in minimizing infant and child mortality, or adult mortality. The conventional arguments for population control in LDCs could, of course, imply a very low weight for infant and child mortality reductions (depending on one's view of the child survival hypothesis). The optimal choice of interventions by mortality-based effectiveness measures tends to be different from that resulting from morbidity criteria. For instance, the Grosse model also simulated the effectiveness of intervention alternatives in reducing mortality. For five of the seven different resource constraints the optimal combination of inputs differed substantially according to the choice of objective. For example at $15.00 per capita, the niinimum crude death rate was obtained with a health centre combined with village health workers and nutrition. and immunization programmes; this compared to the health centre plus sanitation programme choice which minimized days of illness. Adopting the minimization of infant mortality instead of total mortality as the planning objective apparently altered the choice of interventions by introducing nutrition programmes more frequently into the optimal solutions. A similar analysis using a programming model to determine the optimal mix of activities to minimize infant and child mortality has been carried out by Barnum, Barlow, Fajardo, and Pradilla (1980) with data from Colombia. 134 The economics of health in developing countries Composite measures. Since morbidity and mortality objectives tend to have different implications for the best choice of health interventions, it seems preferable to construct a composite measure of health improvement, that is one which aggregates morbidity and mortality reductions. Attempts to construct a so-called health status index have generated a considerable literature relating to developed countries (Culyer, Lavers, and Williams 1971; Berg 1973), but very little concerning LDCs. The central concept underlying this approach is that at any point in time an individual occupies one of a continuum of possible health states ranging from good health to death. For estimation purposes the con- tinuum is divided into discrete health states, j = 1 . .. n. Each state is assigned a vatue Uj on a utility scale ranging from 0 to 1. Health interventions alter the probability of occupying different health states. The effectiveness of an inter- vention is measured in terms of the increment in health status units, defined as the discounted sum over all years of the number of days changed from state j to k in each year, multiplied by the utility weight of state k net of the utility weight of state j (Torrance 1976). Methods which have been used in attempts to derive utility weights on the basis of individual preferences, including the Von Neumann-Morgenstern standard gamble approach, are reviewed in Culyer (1978). Effectiveness measures of this type are sometimes called quality-adjusted life years. An Mllustration of the application of this approach to measurement of the health benefits of the reduction in schistosomiasis and cholera prevalence attributable to water supply improvement in LDCs is presented in Saunders and Warford (1976). A variant of the health status index approach has been developed and applied in Ghana by the Ghana Health Assessment Project Tean (1981). Prevention of death from a specific cause results in a gain of healthy days of life: this gain is computed as the life expectancy at age at death from that cause, converted to the equivalent in days. Prevention of disability from a given cause yields an increase in healthy days equivalent to the expected duration of disability in days weighted by the degree of disability per day (assigned arbitrarily). The sum of both the mortality and morbidity components then provides a measure of effectiveness in terns of healthy days of life saved. An empirical application of this method of ranking 48 causes of morbidity and mortality in Ghana showed the dominant effect of mortality on the selection of disease priorities. These experiments with different approaches to cost-effectiveness analysis in LDCs suggest two important observations. First, since the most effective choice of health interventions appears to be quite sensitive to changes in resource constraints, however health objectives are foriulated, accurate specification of these constraints is critically important to avoid getting locked into inefficient strategies. As noted previously, an assessment of the future availability of recurrent cost financing is especially critical. Second, because the optimal mix of interventions is also sensitive to the choice of health objectives, it is equally important that these should be clearly articulated. This presents particularly formidable problems in practice. Most health pians formulate health objectives Economic appraisal in the health sector 135 in terms of desired resource inputs or intermediate outputs rather than health improvement targets in part because the health impact of particular inputs and interventions is often not known with any certainty. Moreover, specifica- tion cf the trade-off between morbidity and mortality objectives necessarily raises fundamental questions about the relative value of different types of health improvement which policy-makers are notoriously reluctant to confront explicitly, even though such trade-offs are irnplicit in all resource allocation decisions they make. Yet those value judgements have to be made explicit in order to determine, through CEA, an efficient allocation of resources between the full range of potential interventions. Cost-benefit analysis The usefulness of CEA in project selection is strictly limited by its measurement of project outputs in physical units which cannot be compared directly with costs. The fact that a least-cost solution exists to meeting a given health objec- tive says nothing about the desirability of achieving that objective in terms of the relationship between social benefits and costs. Moreover its necessarily limited focus on health benefits alone implies that other objectives are un- important, whereas policy-makers typically seek to pursue several social objec- tives of which health improvement is only one. The idea that the social benefits of a health improvement project shoald be measured in terms of its contribu- tion to different social objectives such as increased per capita income and equity in addition to health improvement is advocated by Feldstein (1970), Paglin (1974) and Barlow (1976, 1980), though a dissenting note is offered by Loucks (1975) who excludes health improvement from a long list of develop- ment policy objectives, This section examines in the context of LDCs the distinctive problem which arises in applying the CBA approach to health pro- jects, namely the choice of an appropriate measure of the value of health benefits. The conventional approach to benefit valuation in project analysis is to rely on individual valuations as revealed by consumer behaviour. In a perfectly competitive economy, consumers' willingness to pay for a project output is equal to the prevailing market price. Thus in many cases project outputs can be valued directly at market prices. However, health projects generally cannot be handled in this way, First, the prevalence of zero or heavily subsidized market prices in the health sector results in an absence of observable market data from which to deduce willingness-to-pay. Secondly, where such data are available they may underestimate the true social benefits of intervention if, because of externalities, the consumer is not the sole beneficiary of the output. Preventive interventions directed at communicable diseases, such as immunizations, present a classic example of where the benefits to society are greater than those which accrue to their immediate consumers. Thirdly, market price data may also be unsatisfactory if consumers are not well informed about the benefit to themselves of health services. This again may be especially true of preventive 136 The economics of health in developing countries interventions. Finally, even if acceptable measures of private valuations are available from market data, consumer preferences might be rejected in favour of socially deternined consumption patterns, not least because the consumption patterns of individuals will reflect a particular distribution of income and wealth. The notion that health and certain other goods such as nutrition and education merit higher levels of consumption than would normally occur without public intervention underlies the basic needs approach to development planning. Consideration of these conceptual and empirical difficulties has necessitated resort to indirect methods of valuation. At least five major alternatives have been attempted. The most common method is the human capital approach. Health improvement is treated as an investment Ln human capital formation which yields an incremental flow of future income or output (Grossman 1972). Health improv tnent exerts both quantity and quality effects on the effective supply of labo Mortality reduction increases the stock of potential workers and hence the 1.otential flow of labour services. Morbidity reduction increases the potential number or efficiency of the flow of labour services. This increment in labour units over time can be estimated and multiplied by their marginal product in order to provide an estimate of the incremental output attributable to any given health improvement (Mushkin 1962). Classic applications of this method are estimates of the economic cost of disease in the United States by Weisbrod (1961), and Rice (1966). In the context of LDCs, the theme of the output-augmenting effects of disease control has been expanded to include its impact on the effective supply of land (attributable to migration induced by the elimination of disease vectors) and accumulation of the capital stock (resulting from a reallocation of private and public expenditure from consumption to savings and investment). For a general review, see Barlow (1979) and Ram and Schultz (1 979). Application of the human capital approach faces certain empirical difficulties, notably in estimating the increment in effective labour supply resulting from morbidity reduction and the marginal product which should be assigned to it. As it is rather easy to apply compared with other approaches it has been used extensively, although there is widespread dissatisfaction with it on conceptual grounds. The method values health only to the extent that it is aln investment good which increases aggregate output; it assigns no value to health as a con- sumption good. Accordingly, it implies that the value of health is greater for those with higher earnings and therefore discriminates against the young, the aged, females, and the poor. Two alternatives to the human capital approach reflect a desire to base health benefit valuation on consumer preferences. Implicit private valuations have been sought in analyses of revealed market preferences in the United States. Blomrrquist (1979) derived a value of life equal to $387 000 from an analysis of the use of automobile seat belts which yielded a small reduction in the statistical probability of death. In an analysis of risk-compensating wage differentials in the labour market, Thaler and Rosen (1975) estimated the Economic appraisal in the health sector 137 implicit value of life at $176 000, An alternative approach using questionnaire surveys designed to elicit an explicit private valuation of mortality reduction has been applied by Acton (1973) to changes in the risk of cardiovascular mortality. He obtained a value of life ranging between $28 000 and $43 000, For a review of these and other consumer preference models see Linnerooth (1979). This approach has not yet been applied to changes in morbidity, or in an LDC context. Other approaches attempt to replace individual valuations by social prefer- ences. A number of attempts to elicit implicit public values, as revealed by government choices in the United Kingdom, are summarized by Card and Mooney (1977). The range of implicit values placed on human life by various programmes or regulations were as follows: £50 in screening maternal oestriol concentration to prevent stillbirths; £1000 in the provision of child-proof drug containers; £100 000 in legislation on tractor cabs; £20 000 000 in building regulations for high-rise apartment blocks. The extreme Inconsistency revealed in implicit 'alues across these different public policy areas clearly suggests that this approach serves more as an indicator of the undesirable consequences of ad hoc decision-making rather than as a reliable guide to valuing benefits. Feld- stein (1970) has suggested instead that explicit public values could be obtained by direct questioning of public policy-makers. This idea is similar to the general procedure proposed by Dasgupta, Marglin, and Sen (1972) for the derivation of social weights in project appraisal. A variation of this approach, involving social specification of a basic needs standard for calorie consumption, has been developed by Scandizzo and Knudsen (1980). The health economics literature offers many examples of cost-benefit analysis (Culyer, Wiseman, and Walker 1977; Griffiths, Rigoni, Tacier, and Prescott 1980; Drummond 1980a,b), but there have been few applications to LDCs. All of these concern communicable disease control projects, mainly immunization programmes (see Chapter 8), or parasitic disease control. Such analyses, especially of parasitic disease control, exemplify many of the empirical diffi- culties encountered in benefit valuation. For example, various studies have assumed that schistosomiasis entails, on average, an impairment of working efficiency ranging from between 4 per cent to 100 per cent. This upper limit is inconsistent with epidemiological evidence that only a minority of infected cases sustain infections of sufficient intensity to provoke clinically severe morbidity, and corroborative physiological evidence from Sudan that significant reductions in physical performance capacity (up to 20 per cent of maximum aerobic power output) are obtained only- at very high levels of infection intensity (2000 eggs/g faeces in S. Mansoni infections) (Prescott 1979a). Similarly, analyses of malaria control have included assumptions about the duration of disability associated with the acute clinical attack (varying fronm 6 to 44 days) in excess of that which has been observed empirically (2.4 to 5.7 days depending on the species of parasite and the level of host imnmunity). Such studies have also generaly failed to recognize the zero marginal product of additional labour inputs at 138 The economics of health in developingcountries seasonal periods of labour surplus (Prescott 1979b, 1980). Most fundamentally, these analyses have not made realistic predictions of the enidemiological effec- tiveness of disease control interventions. Like. many estimnates of the economic cost of disease in developed countries, they are really abstract measures of the gain that could be attributed to hypothetical elimination of a givern disease for a single year. Thus they have not addressed the policy-relevant question of what benefits could be generated over time with the application of feasible control techniques. It is clear that the issue of benefit valuation in health projects has not been resolved satisfactorily, and no single 'correct' method exists which can be recommended to project analysts in LDCs. All of the approaches reviewed have some merits and the choice of a benefit measure will vary in different circum- stances. At a mninimum, efforts should be made to estimate the unit cost of achieving health improvements with different interventions, so that at least some sensible judgements may be made in the choice between alternative expen- ditures. Even this needs refinement of basic epidemiological data and analysis, and is but a first step to more powerful analysis involving benefit valuations. Uncertainty and shadow pricing The need to take account of uncertainty and to use appropriate 'shadow' prices applies to project analysis, both CEA and CBA, in all sectors but these techniques deserve special mention in the context of health project investment appraisal. Uncertainty. Uncertainty over the true value of basic parameters in CEA or CBA calculations is inherent in project analysis, partly because by definition the future is unknown and partly because of practical limitations in the quality of relevant data. For instance, a special feature of health projects is the degree of uncertainty about the technology of health improvement. Not enough is known about the relevant production functions, that is the relationship between i puts and outputs, to predict with confidence the health outputs that will result from health project inputs. This fact partly stems from the unusual complexity of health sector production functions. With the exception of pre- ventive interventions which have the characteristic of 'public' goods, improve- ments in health are produced jointly by health service providers and consumers. Consumer demand is therefore one of the determinants of improved health but knowledge about its characteristics is limnited by a lack of empirical data on consumer demand functions for health services and health-related activities. Another source of uncertainty relates to the technical efficacy of some health interventions in compliant populations, for example BCG vaccination against tuberculosis (Tuberculosis Prevention Trial 1979), or improved water supply and sanitation facilities (World Bank 1975; Saunders and Warford 1976). In addition, uncertainty exists about the accuracy of much of the basic epidemio- logical data on the incidence or prevalence of morbidity and mortality due to specific causes against which project interventions are directed. For example, Economic appraisal in the health secror 139 estimates of the health benefits of schistosorniasis control are sensitive to assumptions about the morbidity and mortality which it causes, a matter which continues to be the subject of controversy. Apart from such uncertainty about quantifying the benefits of health projects, further uncertainty arises over the value which should be placed on them. In practice, this problem of uncertainty should be handled by basing the project analysis on best estimates of the basic project parameters, and then analysing the sensitivity of the conclusions to plausible variations of the para- meter values around their expected values. The sensitivity analysis will indicate which, if any, of the basic assumptions have a significant effect on the accept- ability of a project. The converse approach is to estimate how much different parameter values would have to change before the project switches from being acceptable to unacceptable (or vice versa), and to check whether these are plausi.ble. Such an approach is especially well suited to handling the pervasive uncertainty in health sector projects. Given that in some projects only costs may be known with any confidence, the minimum health output (such as lives saved) required to make the project acceptable can still be estimated and then checked to see whether it is plausible. These estimates can be done within either a CEA or CBA framework. Shadow pricing. The idea of 'shadow' pricing is to determine the 'real' price of a good or service either where no market price exists or where such a market price is considered to be an incorrect indicator of its 'real' value, In the analysis of health investments, two kinds of shadow prices need to be considered: eco- nomic (or efficiency) prices which correct for distortions in factor and product market prices; and social (or distributional) prices which reflect a government's growth and equity objectives. The two main systems of shadow pricing are those developed by Dasgupta et at. (1972) and Little and Mirrlees (1974). later refined by Squire and Van der Tak (1975). A full account of the justification for, and derivation of, shadow prices is available in these original sources; only the major points are summarized here. The use of economic prices mainly affects the value of foreign exchange and labour inputs used by projects, In most LDCs it is common to encounter exchange rates which tend to underestimate the scarcity value of foreign exchange. Con- versely, market wage rates often tend to overestimate the opportunity cost of labour. In these circumstances, the effect of the use of 'shadow' prices will be to increase the cost of foreign exchange and reduce the cost of labour. These adjust- ments will discourage projects which use scarce foreign exchange and favour those which are labour intensive. Likewise, if faster economic growth or income redistribution are key social objectives, then shadow prices in the form of social prices may be used to bias project selection against investment projects which rely heavily on public sector financing, in favour of those which generate high rates of saving and reinvestment by, for example, generating additional public revenue from user charges, and those which redistribute income to the poor. 140 The economics of he4lth In developingcountries Some features of the health sector make consideration of shadow pricing of special interest, and the use of economic and social prices has important implica- tions for the choice of health projects. The penalty attached to foreign exchange intensive projects, as a result of the use of economic prices, gives particular importance to pharmaceutical policies directed at minimizing the use of imported pharmaceuticals and increasing domestic production, including primary manu- facture which may not rely on imported inputs. The mutually reinforcing incentives to employ labour with a low opportunity cost (for efficiency reasons) and to favour employment of the poor (for distributional reasons) emphasize the desirability of substituting lower for higher grade health workers where technically feasible. The disincentive to using scarce public resources to finance health projects directs attention to exploring methods of self-financing. It is pertinent to observe that all of these concems-pharmaceutical policy, the use of village health workers and community participation in project financing-are currently the subject of intense policy debate and can each be accommodated through a straightforward application of shadow pricing in project selection. Resource mobilization Existing and prospective financial constraints in LDCs indicate & need to re- evaluate the traditional approach to health sector financing which relies heavily on public subsidies from central government revenues. A significant increase in the allocation of public revenues to the health sector is unlikely to be a realistic option, given the traditional strength of competing sectoral demands and the continuing diffIculty of demonstrating that the marginal social rate of return on health expenditure exceeds the returns yielded in other sectors. Expansion of the health sector in many countries may, therefore, depend largely on the implementation of cost recovery mechanisrns. In general, the relevance of alternative financing methods will vary with the type of health intervention. In each case the choice between financing alternatives can be assessed using the same framework already developed for other public sector projects in which efficiency, equity and fiscal cbjectives are traded off for the various services provided (Ray 1975). The benchmark for analysis of financing alternatives is given by the long- run marginal cost of each intervention, though departures from marginal cost pricing can be justified in certain circumstances. In the presence of externalities or poorly informed consumers, both being characteristic of preventive inter- ventions aimed at individuals, marginal cost prices will fail to induce socially optimal levels of consumption, and subsidized prices are appropriate. Public health measures, such as mollusciciding for schistosomiasis control and aerial larviciding for onchocerciasis control, have some of the characteristics of 'public goods', where the benefits of intervention are not consumed by any one individual and the exclusion principle therefore fails to apply, in that access to the benefits of intervention cannot be made conditional on payment of a user Economic appraisal in the health sector 141 charge. Direct charges for such services, levied separately on each beneficiary, may either be too expensive to administer, or simply not feasible. The achievement of equity objectives will also be affected by the choice of financing mechanism. Interventions directed at low income beneficiaries, for example a programme of primary health care, may be ineffective under a system of direct user charges if the price elasticity of demand is high, meaning that potential consumers are deterred from seeking care by the level of prices charged. Evidence on price elasticities for health services is scarce, but data from Malaysia indicate that the total demand is highly inelastic with respect to cash prices, although there are significant cross-price elasticities for the services of different providers (Heiler 1982), implying that consumers do switch between providers according to variations in prices. Similar findings have been reported for the Philippines (Akin, Griffin, Guilkey, and Popkin 1982). If demand is inelastic, the impact of raising charges on the use of health services will by definition be minor and its fiscal impact positive, but the income transfer from the bene- ficiary to the public sector may not accord with distributional objectives. fHowever, even where these difficulties apply, there may be no alternative but to generate revenues from 'oeneficiaries or elsewhere in the private sector in order to finance the provision of health services. It is clear, none the less, that for some interventions reliance on direct user charges will not be appropriate and other financing mechanisms will need to be explored. But whether the very limited resort to user charges in LDCs is appro- priate-less than 10 per cent of public sector costs are recovered by this means in countries such as Malaysia, Malawi, Lesotho, and Ghana (Brooks 1981)- requires detailed examination. In general, the arguments against pricing are stronger for preventive interventions and for lower income beneficiaries. Thus a system of selective charges for curative services, especially for higher income users, may be justified to cross-subsidize preventive activities and utilzation by the poor. Private sector experience, notably of mission facilities and traditional practitioners and birth attendants, demonstrates the feasibility of recovering a high proportion of recurrent costs from user charges without reducing utiliza- tion. In Lesotho, some 70 per cent of recurrent expenditure by private hospitals is recovered from user charges (Kolobe and Pekeche 1980). Alternatives to charges in use in many countries include earmarked taxes, social security and insurance schemes, or indirect cost recovery by self-help (Evans, Hall, and Warford 1981). These methods are explored in Chapters 3 and 4 of this volume, which show that a wide range of financing mechanisms can be employed in the health sector in LDCs. Nevertheless, the approaches generally used are dominated by the tradition of central government subsidy and the introduction of innovative methods encounters very difficult political obstacles. The prevalent view that because health is good it should be free or subsidized may have to be increasingly challenged, however, if the goals set by WHO and UNICEF, referred to in the introduction, are to be achieved. 142 The economics of health in developing countries Conclusion In general, conventional approaches to investment appraisal in the health sector are much less rigorous than in most other sectors. A major reason for this may be that there are greater difficulties of theory and measurement at all stages of project analysis, for example in the prediction of effectiveness of health inter- ventions, the valuation of benefits, and the design of appropriate cost recovery policies. However, these problems differ in degree rather than substance from those found in other sectors such as water supply, energy, and education, where considerable analytical progress has been achieved. They certainly do not justify the prevalent tendency to avoid rigorous analysis of investment decisions altogether simply because it is difficult. It may even be argued that many of the problems which are encountered are the result of past neglect of economic analysis in the health sector. The potential role of economic appraisal is there- fore very large and, indeed, essential to attainment of the dramatic improve- ments in health status which are now sought by LDCs. REFERENCES Acton, J. P. (1 973). Evaluating public programs to save lives: the case of heart attacks, R-950-RC. The Rand Corporation, Santa Monica, California. Akin, J. S., Griffin, C. C., Guilkey, D. K., and Popkin, B. M. (1982). The demand for primary health care services in the Bicol region of the Philip- pines. Paper presented at the National Council for International Health Conference on Financing Health Services in Developing Countries, 14-16 June, Washington DC. Balassa, B. (19F2). Structural adjustment policies in developing economies. World Development 10, 23-38. Barlow, R. (1976). Applications of a health planning model in Morocco. Int. J. Hlth Serv. 6, 103-21. - (1979). Health and economic development: a theoretical and empirical review. Res. hum. Capital Dev. 1, 45-75. - (1980). Economic goals in health planning. Document TDR/SER/SWG(2)/ 80.WP.6. World Health Organization, Geneva. Barnum, H., Barlow, R., Fajardo, L., and Pradila, A. (1980). A resource alloca- tion model for child survival. Oelgeschlager, Gunn, and Hain, Cambridge, Mass. Berg, R. L. (ed.) (1973). Health status indices. Hospital Research and Educa- tional Trust, Chicago. Blomquist, G. (1979). Value of life saving: implications of consumption activity. J. Polit. Econ. 87, 540-58. Brooks, R. G. (1981). Ghana's health expenditures 1966-80: a commentary. Strathclyde Discussion Papers in Economics No. 80/1, University of Strath- clyde, Glasgow, Scotland. Card, W. J. and Mooney, G. H. (1977). What is the monetary value of a human life? Br. med. J, ii, 1627-9. Club du Sahel (1980). Recurrent costs of development programs in the countries of the Sahel: analysis and recommendations. Club du Sahel, Paris. Culyer, A. J. (1978). Need, values and health status measurement. In Economic aspects of health services (ed. A. J. Culyer and K. G. Wright). Martin Robert- son, London. Economic appraisal in the health sector 143 - Lavers, R. J., and Williams, A. (1971). Social indicators: health. Soc. Trends 2, 31-42. Wiseman, J., and Walker, A. (1977). An annotated bibliography of health economics, English language sources. Martin Robertson, London. Dasgupta, P. S., Marglin, S. A., and Sen, A. K. (1972). Guidelines for project evaluation. United Nations Industrial Development Organizatic n, New York. Drummond, M. F. (1980a). Principles of economic appraisal in health care. Oxford University Press. New York. - (1980b). Studies in economic appraisal in health care. Oxforfd University Press, New York. Evans, J, R., Hall, K. L., and Warford, J. J. (1981). Shattuck lecture-Health care in the developing world: problems of scarcity and choice. New Engl. J. Med. 305, 1117-27. Feldstein, M. S. (1970). Health sector planning in developing countries. Eco- nomica 37, 139-63. -Piot, M. A., and Sundaresan, T. K. (1973). Resource allocation model for public health planning: a case study of tuberculosis control. Bull. Wld Hlth Org. 48, Suppl. Ghana Health Assessment Project Team (1981). A quantitative method of assessing the htealth impact of different diseases in less developed countries. Int. J. Epidemiol. 10, 73-80. Griffiths, A., Rigoni, R., Tacier, P., and Prescott, N. (1980). An annotated bibliography of health economics, Western European sources, Martin Robertson, Oxford. Grosse, R. N. (1980). Interrelation between health and population: observations derived from field experiences. Social Sci. Med. 14c, 99-120. - De Vries, J. L., Tilden, R. L., Dievler, A., and Day, S. R. (1979). A health development model: application to rural Java. Department of Health Planning and Administration, School of Public Health, The University of Michigan, Ann Arbor, Michigan. Grossman, M. (1972). On the concept of health capital and the demand for health. J. Polit. Econ. 223-55. Heller, P. S. (1974). Public investment in LDCs with recurrent cost constraints: the Kenyan case. Q. J. Econ. 88, 251-77. - (1975). Issues in the costing of public sector outputs: the 'public medical services of Malaysia. Staff Working Paper No. 207. World Bank, Washington DC. -(I1979). The underfinancing of recurrent development costs. Finance Dev. 16, 38-41. - (1982). A model of the demand for medical and health services in peninsular Malaysia. Social Sci. Med. (forthcoming). IMP (International Monetary Fund) (1981). Government finance statistics year- book. Vol. V, International Monetary Fund, Washington DC. Kasongo Project Team (1981). Influence of measles vaccination on survival pattern of 7-35 month old children in Kasongo, Zafre. Lancet 764-7. Kolobe, P. and Pekeche, T. (1980). A survey of the financial status of the private health association of Lesotho's hospitals. Private Health Association of Lesotho, Maseru. Linnerooth, J. (1979). The value of human life: a review of the models. Econ. Inquiry 17, 52-74. Little, I. M. D. and Mirrlees, I. A. (1974). Project appraisal and planning for developing countries. Heinemann Educational Books, London. 144 The economics of health in developing countries Loucks, D. P. (1 975). Planning for multiple goals. In Economy-wide models and development planning (ed. C. R. Blitzer, P. B. Clark, and L. Taylor), pp. 213-33. Oxford University Press, London. Mahler, H. (1977). Blueprint for health for all. WHO Chron. 31, 491-8. Meerman, J. (1979). Public expenditure in Malaysia: who benefits and why. Oxford University Press, New York. - (1980). Paying for human development. In Implementing programs of human development (ed. P. T. Knight), pp. 111-82. Staff Working Paper No. 403. World Bank, Washington DC. Mushkin, J. (1962). Health as an investment. J. Polit. Econ. 70, 129-57. Ofosu-Amaah, S. (1975). Reflections on the health budget: a preliminary analysis of the 1974/75 Ministry of Health Budget. Ghana med. J. 215-22. Over, A. M. (1979). Five primary care projects in the Sahel and the issue of recurrent costs. Harvard Institute for International Development, Cambridge, Massachusetts. Paglin, M. (1974). Public health and development: a new analytical framework. Economica 41, 432-41. Paqueo, V. (1976). Social indicators for health awid nutrition. In Measuring Philippine development: report of the social indicators project (ed. M. Mangahas), pp. 41-115. Development Academy of the Philippines, Manila. Prescott, N. M. (1979a). Schistosomiasis and development. Wld Dev. 7, 1-14. -(1979b). The economics of malaria, filariasis and human trypanosomiasis. Document TDR/SER(SC-l)/80.4. World Health Organization, Geneva. - (1980). On the benefits of tropical disease control. In Health policies in developing countries (ed, C. Wood and Y. Rue), pp. 41-8. Royal Society of Medicine International Congress and Symposium Series No. 24. Academic Press, London. Ram, R. and Schultz, T. W. (1979). Life span, health, savings and productivity. Econ. Dev. cultural change 27, 399-421. Ray, A. (1975). Cost recovery policies for public sector projects. Staff Working Paper No. 206. World Bank, Washington DC. Rice, D, P. (1966). Estimating the cost of illness. Health Economics Series No. 6, PHS Publication No. 947-6. US Government Printing Office, Washington DC. Richards, P. and Leonor, M. (ed.) (1982). Target setting for basic needs. Inter- national Labour Office, Geneva. Romeder, I. M, and McWhinnie, J. R. (1977). Potential years of life lost between ages I and 70: an indicator of premature mortality for health planning. Int. J. Epidemiol. 6, 143-51. Rosenfield, P. L., Smith, R. A., and Wolman, M. G. (1977). Development and verification of a schistosomiasis transmission model. Am. J. trop. Med. Hyg. 26, 505-16. Saunders, R. J, and Warford, J. J. (1976). Village water supply: economics and policy in the developing world. Johns Hopldns University Press, Bmalti- more. -Warford, J. J., and Mann, P. C. (1977). Alternative concepts of marginal cost for public utility pricing: problems of application in the water supply sector. Staff Working Paper No. 259. World Bank, Washington DC. Scandizzo, P. L. and Knudsen, 0. K. (1980). The evaluation of the benefits of basic need policies. Am. J. agric. Econ. 62, 46-57. Selowsky, M. (1979). Who benefits from government expenditure? A case study of Colombia. Oxford University Press, New York. Economic appraisal in the health sector 145 Shepard, D. S. and Zeckhauser, R. J. (1980). Long-term effects of interventions to improve survival in mixed populations. J. chron. Dis. 33, 413-33. Smith,J, (1980).A review of the resources and service area of the health facilities in the Kingdom of Lesotho. Ministry of Health, Maseru. Squire, L. and Van der Tak, H. G. (1975).Economic analysis of projects. Johns Hopkins University Press, Baltimore. Stolper, W. F. (1966). Planning without facts. Harvard University Press, Cam- bridge, Mass. Streeten, P., Burki, S. J., Ul Haq, M., Hicks, N., and Stewart, F. (1981). First things first: meeting basic human needs in developing countries. Oxford University Press, New York. Tait, A. A., Gratz, W. L. M., and Eichengreen, B. J. (1979). International com- parisons of taxation for selected developing countries. IMF Staff Pap. 26, 123-56. Tan, E. A. (1975). Taxation, government spending and income distribution in the Philippines. Income Distribution and Employment Programme, Working Paper No. 26. International Labour Office, Geneva. Thaler, R. and Rosen, S. (1975). The value of saving a life. In Household pro- duction and consumption (ed. N. E. Terleckyj). National Bureau of Eco- nomic Research, New York. Torrance, G. W. (1976). Health status index models: a unified mathematical view. Mgt Sci. 22, 9 90-1 001. Tuberculosis Prevention Trial (1979). Trial of BCG vaccines in South India for tuberculosis prevention: first report. Bull. Wld Hith Org. 57, 819-27. Turvey, R. and Anderson, D. (1977). Electricity economics: essays and case studies. Johns Hopkins University Press, Baltirnore. Weisbrod, B. A. (1961). Economics of public health: measuring the economic - impact of diseases. University of Pennsylvania Press, Philadelphia. World Bank (1975). Measurement of the health benefits of investments in water supply. Public Utilities Report No. PUN 20. World Bank, Washington DC. - (1981). World development report 1981. Oxford University Press,New York. - (1 982). Malawi: growth and structural change, a basic economic report. Report No. 3082-MAI. World Bank, Washington DC. WHO (World Health Organization) (1977). Financing of health services: report of a WHO study group. Tech. Rep. Ser. Wld Hlth Org. 625. -(1981). Review of health expenditures, financial needs of the strategy for health for all by the year 2000, and the international flow of resources for the strategy. Document EB69/7. World Health Organization, Geneva. Authors 'note The views and interpretations in this chapter are those of the authors and should not be attributed to the World Bank, to its affiliated organizations, or to any individual acting on their behalf. o. 302. James H i.nson and Jaine de Melo, "The Urguay.3n Experience with Liberalization and Stabilization, 1974-198I," joutrnatil of Ilterainoican Sftudies and Wvorld Affairs o. 303. Nancy Birdsall and Deani T. Jamison, "Income and Other Factors Influencing Fertility in China," Ibpulafion and Df.lolj'nwent RcoiL'w \o. 304. Graham Donaldson,"Food Security and the Role of the Grain Trade," American JoLrnial of AC";ri(lIuraI L'OUiWS 4o. 305. William F. Steel and Yasuoki Takagi, "Small Enterprise Development and the Ernployment-Output Trade-Off," Oxford Economic Papers \o. 306. Oli Havrylyshyn and Engin Civan, "Intra-Industry Trade and the Stage of Develop- ment: A Regression Analysis of Industrial and Developing Countries," Intra-Industpn Trade: Empirical and Metlhodological Aspects 'o. 307. Mateen rhobani, "A Nested Logit Model of Travel Mode to Work and Auto Ownership," Journal of LUrbizn Economics o. 308. Johannes Bisschop and Alexander Meeraus, "On the Development of a General Algebraic Modeling System in a Strategic Planning Environment," Mathematical Th'o'4rinm*ZXinl, Studtvl o. 309. Rtrvnaldo Martorell, Joanne Leslie, and Peter R. Moock, "Characteristics and Deter- minants of Child Nutriticonal Status in Nepal," The ,Anericain Journal of Clinical Nutrition o. 310. Robert H. Litzenberger and Jacques Rolfo, "An International Study of Tax Effects on Govemment Bonds," IThe Journal of Finance o. 311. Jere R. Behrman and Nancy Birdsall," The Quality of Schooling: Quantity Alone is Mlisleading." Aiknerictan Ec,ion.mii Reviewv o. 312. Bela Balassa, "Adjustment Policies in Developing Countries: A Reassessment," Worldi iict'lo;'f'llL'Z'7Jt o. 313. Keithl\ 1farsden, "Services for Small Fiirms: The Roles of Government Programmes and Market Networks in T1hailandl," hi item,a ithmaF uil Labour Review o. 3 14. Gunter Schramm, "The Changing World of Natural Gas Utilization," Natural Resoutrces J0ui11'nai o. 315. Bela Balassa and Carol Balassa, "Industrial Protection in the Developed Countries," 'The World Tc.'imi/ o. 3 16. Nancy Birdsall and Jere R. Behiman, "Does Geographical Aggregation Cause Overestinmate:- of the Returns to Schooling?" Oxford Builletiti of EconLomtics and Statistics o. 317. Jan Peter Wogart and lose Silverio Marques, "Trade Liberalization, Tariff Redundancy, arid Inflation: A Methodological Exploration Applied to Argentina," Weltwirt- Ft.t;caf'licb1,l Arclitiz o. 318. Sweder var Wijnbergen, "Credit Policy, Inflation, and Growth in a Financially Repressed Economy," Journal of DeivLlopneif Ecoinomnlics o. 319. David M. Newbery, "Commodity Price Stabilization in Imperfect or Cartelized Markets," eClonomdetri ci o. 320. George Psacharopoulos, "The Contribution of Education to Economic Growth: International Comparisons," with a Comment by Thieodore W. Schultz, Iternlatioiial .n)1I7?ristni;s of' Prodicfi'ifi, and CaeuzsL, of the SlI5L'vdJ'zi o. 321. Bela Balassa, "Trade between Developed and Developing Countries: the Decade Ahead," OECD Ecotnomiiic Studies Issues of the World Bank Reprint Series are available free of charge from the address on the bottom of the back cover.