The process of economic development is an important instance of the interanimation of policy, contingency, and justice. Governments construct development strategies based on a variety of political and social objectives; this process may be characterized in terms of a decision-making process based on instrumental rationality. The outcomes of these strategies depend heavily on a variety of factors, only some of which are within the control of policy makers-world commodity markets, the behavior of powerful outsiders, weather and climate variability, and so forth. And the outcomes and implementations of development strategies unavoidably raise problems of justice. This is so because they have deeply significant distributive effects that may endure over many generations. Different groups do well or badly under different policy alternatives. So policy makers ought to be responsive to the demands of justice in their adoption of priorities and the instruments they choose in pursuit of the goods of development.
This essay raises problems about the role of normative considerations in several aspects of this story. These problems arise in two general areas. First, there are problems defining the instrumental rationality of policy makers. Even if we bracket the problem of distributive justice there are perplexing problems that arise in attempting to characterize the goods of development and how these should be treated by policy makers. (For example, how should environmental quality and average longevity be compared, given that a policy that results in lengthening life expectancy may also require worsening environmental quality?) And how should we think about the role of moral considerations within the policy process? The second broad topic is substantive: what are the overriding goals of economic development? I will argue that the highest priority goal ought to be the alleviation of poverty, a stance I describe as "putting the poor first". Policy makers ought to give highest priority to improving the well-being of the poorest strata. But this position is faced with the task of explaining in substance why this is so: what is the bad of poverty? What is involved in placing highest priority on poverty alleviation? And what compelling moral reasons may be offered to support this position? These questions are important because the answers we provide for them will in large part determine how much importance we place on poverty alleviation in comparison with other compelling development goals.
Second, there are problems defining the relationship among the diverse goals of economic development. Even if we bracket the problem of distributive justice there are perplexing problems that arise in attempting to characterize the goods of development and how these should be treated by policy makers. (For example, how should environmental quality and average longevity be compared, given that a policy that results in lengthening life expectancy may also require worsening environmental quality?) And how should we think about the role of moral considerations within the policy process? The second broad topic is substantive: what are the overriding goals of economic development? I will argue that the highest priority goal ought to be the alleviation of poverty, a stance I describe as "putting the poor first". Policy makers ought to give highest priority to improving the well-being of the poorest strata. But this position is faced with the task of explaining in substance why this is so: what is the bad of poverty? What is involved in placing highest priority on poverty alleviation? And what compelling moral reasons may be offered to support this position? These questions are important because the answers we provide for them will in large part determine how much importance we place on poverty alleviation in comparison with other development goals.
Seen from one point of view, this complex of issues is a relatively small part of applied moral philosophy, of interest only to a rather specialized audience. From another point of view, however, this area ultimately invokes virtually all of the most difficult and important topics within normative social philosophy: the problem of planning for a process of social change, the justice of the allocation of the benefits of social cooperation across individuals and groups, the problem of intergenerational justice, the legitimate claims that different groups and individuals have on each other and on outsiders, and the definition of the good of human life and social arrangements. So close study of the problems of economic development may have a considerable contribution to make for social philosophers more generally.
Let us begin with a consideration of the normative position of the development policy maker. Policy making obviously involves an analysis of means and ends. Policies are sets of instruments designed to bring about certain outcomes, and the desired outcomes are selected because they embody one or more goods that are favored by the policy maker. But this formulation glosses over two issues: What is the nature of the goods toward which development policy is oriented? And what role, if any, do normative commitments play within the policy-making process?
In designing a development plan there is always a range of choices that must be made: to encourage export production, to promote cash crops or food crops, to favor heavy industry or consumer goods, and so forth. And the choices that are made among these alternatives will depend on the criteria of evaluation of consequences that are in use. If the goal is to increase GNP at the fastest possible rate, then one family of choices will be made; if the goal is a combination of growth and military security, another set of choices will be made; and so forth. Putting the poor first involves making these choices on the basis of consideration of how various alternatives will affect the welfare of the poorest strata in society. So, for example, a government may be deliberating between investment credits for a television assembly plant and for a sugar-processing plant. The television plant, let us suppose, will produce a greater amount of value added, generating a resultingly higher amount of foreign currency; while the sugar-processing plant involves a substantially higher level of employment at a wage higher than the current average for unskilled labor. Prima facie these circumstances, conjoined with the "poverty-first" principle, entail that the government should select the sugar-processing plant, since this alternative creates the greater amount of additional income for the poor.
Let us reflect briefly on the normative situation of economic development actors-planning agencies within LDC governments, international organizations like the United Nations or the World Bank, or donor agencies like the Agency for International Development. These actors attempt to influence the course of economic development in a country or region-that is, they undertake actions or policies today that lead to a change in particular features of the economy in the future. For example, the Minister of Finance may judge that existing financial conditions are leading to capital flight and disinvestment, and may seek to reverse this process by increasing interest rates at the national bank.
It will be useful to fix the terms of discussion a bit by providing a simple model of the development planning process so as to have a representative decision-problem in mind. Policy de bates can arise only in a context where agents involved have actions among which they can choose. What are the action-opportunities that confront governments concerned to facilitate economic development? There are three broad types of actions available to governments. A government can expend resources in a variety of ways; it can affect the institutional context of economic activity through new legislation or regulations; and it can directly influence the economic environment for private activity through control of a number of macroeconomic variables (exchange rates, tariffs and subsidies, national bank interest rates, and the like).
The selection of an economic policy depends on several things. First, a diagnosis of existing circumstances is needed. Second, a theory of the economic or social mechanisms through which these circumstances might be changed is required. Third, it is necessary to have a coherent conception of the goals of possible intervention-the goods that intervention is designed to secure. Fourth, the minister requires an inventory of the options and resources available to him. And finally, the minister must arrive at a plan-an orderly conception of a series of actions within his power that are well-conceived to bring about the goods selected. Finally the plan must be put into place.
It is useful to consider an abstract analysis of the process of economic policy intervention from the point of view of a government minister in a less-developed country. At a given point of time the policy environment is defined by the assets available to the national economy: currency reserves, natural resources, agricultural technology and resources, industrial equipment, human resources, and infrastructure. These assets are organized within a set of institutional arrangements: property rights, land tenure systems, and the like. And they are currently employed and controlled by a large number of economic agents. These features define the economic system as it exists at a given time; and political economists from Marx to Marglin have shown that the economic system has a dynamic of development of its own that continues without regard to explicit government policy. This ensemble is thus a causally ordered social system.
A second large factor defining the policy situation is the assets available to government through which it can exercise influence on the character and pace of economic change: taxation, subsidies, influence on interest and exchange rates, provision of credit, and establishment of economic regulations of various kinds. These assets represent the instruments through which a government minister can hope to modulate the economic processes that are ongoing within the national economic system.
Third, economic policy requires plans. A plan involves an orchestrated set of choices of such actions designed to bring about some end. Two questions arise: How is a particular policy choice (action) to be evaluated? And how is a plan to be evaluated? In general the answer is, through the probable consequences of the plan.
Finally, we need to be able to identify the range of consequences of economic policies. Several categories of outcomes are particularly important for assessment of policies. First are in come consequences, which may be represented as a vector of incomes to different income groups. (This formulation allows us to characterize both average income to various groups and the pattern of distribution of income.) Second are structural or institutional consequences: changes (or persistence) in features of the economic environment. Likewise, we need to know the impulse that the plan will impose on future patterns of development (agricultural stagnation, rapid industrialization, etc.). Third are environmental consequences, including resource stocks-the consequences for environment and health of the plan. Fourth, we may want to be able to evaluate the demographic characteristics of the plan: what shifts in population size, density, and location may be expected to arise? And finally, we need to take account of political consequences: the effects on regime stability and popular politics that various strategies may have.
Consider first expenditure of resources. State expenditures may be directed toward infrastructure, productive investment, or consumption. Should the state widen the road connecting a secondary market town with the central market town? Should the state provide cheap credit to entrepreneurs proposing a shrimp-processing plant? Should the state widen or narrow the ex tent of basic food subsidies?
Turn now to institutional reform. It is possible for the developing state to alter the institutional arrangements through which economic activity takes place in such a way as to improve the situation of the poor. For example, programs to make tenancy rights more secure shift the balance between landlord and tenant; the tenant can claim a larger share of the harvest if he is more confident that he cannot be expelled from his land. Likewise, establishment of rights of labor-free labor organization, the right to strike-enhance the ability of workers to demand higher wages and a larger share of the product of their enterprises.
This account perhaps gives the impression of an orderly, fully intentional process. How ever, this impression is misleading; economic development is a complex process that involves both policy choice and historical contingency. Governments, agencies, associations, and corporations undertake various actions designed to influence the process of economic change; but the outcome of economic development depends both on these actions (and their unintended interactive effects) and on various non-strategic circumstances-climate, the availability of resources, the discovery of new technologies, the shift of patterns of consumption in the world market, and so forth; and on the strategies chosen by other actors with different and often conflicting goals. And historical experience makes it rather clear that the course and character of a given country's economic development experience is substantially less subject to policy determination than is the construction of a complex building. An architect can produce a set of plans for the building and be confident that the natural obstacles that arise in the process of construction will be resolved. There is a close correspondence between blueprint and the eventual structure of the building, whereas there is often wide discrepancy between the designs and goals of economic planners, and the eventual course of economic development in the event. The upshot of this observation is the need for humility on the part of the economic policy profession; it must be recognized at the start that events are not securely within the control of the agents charged with planning for economic development.
Given the importance of contingency and unanticipated consequences in economic planning-reflecting both the complexity of modern economies and the competition of actors with contending interests within such economies-it is necessary to pay particular attention to the range of risk and uncertainty associated with various policy candidates, as well as the goodness and badness of undesired outcomes. It might be, for example, that it is judged likely that P1 will lead to an unemployment rate of only 6 percent, whereas P2 will lead to an unemployment rate of 8 percent; but that P1 runs the slight risk of causing a prolonged depression, whereas P2 does not. Under these circumstances it might be rational to choose P2, in spite of its being likely to lead to second-best consequences.
Before proceeding with this discussion, we need to consider whether a normative argument is relevant in this context. For it might be held that development theory is a purely positive discipline, concerned only with the technical relations between such variables as rates of saving, technological change, and GNP. Why should we consider the question of justice in this context at all? The short answer to this question is convincing: development processes are the result of policy choices by various agents-national governments, political parties, organized interest groups, donor agencies, and the like; and policy choices unavoidably express normative goals. Whether justice requires putting the poor first is therefore a critical question. And the normative case for the priority of the poor is not difficult to construct. The most immediate answer is a concern for the welfare of all members of a society and a collateral conviction that it both is important and feasible to focus attention first on the strata in the most immediate need.
When we ask the question, what does justice require of economic development policy, we are forced to consider whether the answer we arrive at has any relevance for the behavior of actual policy-makers. To what extent, and in what ways, do policy-makers take account of moral ideas, ideas about justice, and other normative commitments?
There is a line of analysis of policy choice in political science today that works on the assumption that policymakers are motivated solely by their particular private interests-not an estimate of what is best for society at large, but what is best for the policymaker's own interests. This approach is described as "realism" or public choice theory. Margaret Levi represents a recent instance of this approach in her analysis of state revenue policies. "My hypothesis follows from two paired assumptions. The first is that all the actors who compose the polity, including the policymakers, are rational and self-interested. By this I mean that they calculate the costs and benefits of proposed actions and choose the course of action most consistent with their fixed preferences. The second is that actors who compose the state have interests of their own, derived from and supported by institutional power" [Levi, 1988, Revenue]. Likewise, Robert Bates's analysis of African government agricultural policies aims to explain the policies selected as the result of the strategic efforts of groups in charge of government to maximize their power and income [Bates, 1981, Markets]. On this account, state policy formation is similar to any other kind of rational action: the agent has certain interests and chooses instruments that best serve those interests. Political behavior is understood as the result of a straightforward calculation based on self-interest. And the interests in question are private interests; either personal (the wealth, power, and privilege of the particular policymaker) or corporate (the wealth, power, and privileges of the policymaker's group). (On such an ac count, group-oriented policies emerge through the logic of coalition politics: in order to achieve his ends Minister A must gain the support of group B; and this requires A's putting forward policies that serve the interests of B as well as himself.) This model appears to leave little or no room for an influence to be exerted by normative principles or constraints. What, then, is the relevance of a normative theory of justice in development? To what extent and through what mechanisms do moral ideas about the justice of development affect the choices of powerful agents in the development process?
It would be foolish to deny the operation of this sort of calculation of self-interest within the policy process. Here I will only suggest that the approach is flawed in its claimed comprehensiveness. The public choice approach implicitly denies that policymakers are motivated or constrained by moral considerations. But much recent work in rational choice theory converges on the conclusion that this strictly instrumental conception is inadequate for all but the most artificial examples of rational deliberation-e.g. consumer choice among commodities based on price and quality. In more ordinary instances decision-makers have to take into account both their particular interests and a range of normative constraints and values [Margolis, 1982; Sen, 1982, Fools; Scott, 1985; Elster, 1989, Cement; Little, 1991, Rational; Taylor, 1982, #931]. Practical rationality has more structure than the narrow model of economic rationality allows, and deliberation is a complex process in which the agent attempts to balance interests, commitments, and concerns for the public good. Debate arises over whether to regard these factors as extra-rational; my own position is that they are part of "extended practical rationality" which reduces to narrow economic rationality in special circumstances [Little, 1991, Rational].
In the area of economic development policy these points can be extended in several important ways. First, at any given time policymakers are operating within a normative framework that sets broad constraints on the nature and application of policy options that may be considered. (A politician who proposes abolishing state subsidies for public education in the United States would plainly transgress such limits.) Such constraints are often internalized, so that the decision-maker will not even consider adopting policies that flatly contradict the framework; but they are also externally embodied, so that there are substantial political costs that accrue to the regime that flagrantly violates popular assumptions about justice. A normative framework has the effect of helping define and limit the opportunity set for the decisionmaker; options that involve reintroducing chattel slavery are simply ruled out of consideration, whatever their economic effects. Norms affect policy choices by altering the options available to agents, giving priority to some goals over others, and making certain choices entirely unavailable. And they may have the effect of aiding in the mobilization of affected groups (e.g. the landless poor) whose new political dispositions impose new constraints and imperatives on policy makers. So even policy makers who are themselves indifferent to the new moral assumptions may be compelled to act otherwise.
Finally, it is unreasonable to assume that decision-makers' interests are defined narrowly in terms of self-interest. As A. K. Sen argues [Sen, 1987, Ethics], it is better to conceive of agents as having a broad network of interests ranging from narrowly private interests, to family interests, to class or national interests. And interests at any of these levels may be motivationally efficacious at various times. In the development process this leads to a conclusion: it is at least possible to motivate decision-makers to choose policies because they are best for society at large. Whether this concern plays a prominent or minor role is not a matter of logic or economic rationality, but one of moral education. And it may well be that what distinguishes between more and less predatory states is the character and quality of the process through which young policymakers are trained, selected, and promoted.
I conclude, then, that discussion of the justice of development is not a purely philosophical exercise. It is rather an attempt to contribute to a debate that may come to change the framework of our assumptions about the priorities of development, and eventually to change the motives and constraints that govern the behavior of the agents whose choices in fact determine the course of development. Normative theory is behaviorally relevant in explaining the course and character of an ensemble of policies (military, economic, political, or social).
The reason I have dwelt on this topic is that it makes a difference to how we think about problems of economic development policy and economic reform. If as potential reformers we are best advised to look at policymakers as self-serving adversaries to be out-maneuvered, then one line of strategy is recommended. If on the other hand it is true that decisionmakers are to some extent swayed by moral arguments and considerations of the public good, then a strategy based on moral argument is feasible. In the former case our only hope is to create institutional arrangements that assure that the private interest of the policymaker is best served by adopting policies that we think best for the common good. In the latter case it will be causally efficacious to advance reasonable arguments about fairness, the welfare of the poor, or the demands of future generations; considerations of these sorts may be decisive in affecting policy choice.
I propose, then, that we think of the normative position of the policymaker along these lines. The agent is assumed to have a range of interests: personal security, income, and privilege; the continued ability of his or her party to retain power; the ability of the regime to reward its supporters with jobs, state investment and subsidies, and the like; and concern for the present and future welfare of some or all of the rest of society. And the agent is situated within the context of a set of moral assumptions about justice, entitlement, concern for disadvantaged groups, and the like, that are current in the society at large and that affect behavior by constraining and impelling choice. When considering a policy problem, whether a grand development theme (shall we liberalize our agricultural system through price reform and removal of subsidies?) or a more particular policy initiative (shall we replace current rice subsidies with a needs-based food stamp program?) I assume that the agent initiates a complex process of deliberation. What effect will the proposed policy have on me personally? (Perhaps liberalization will mean the ministry of development will have less importance within government and corresponding loss of funds and prestige.) What effect will the policy have on the electoral prospects of my party? (Perhaps the destitution caused to the landless poor will lead to massive loss of support for our party in the next elections.) What effect will the policy have on the welfare of rural society, in the short, medium, and long term? (Perhaps liberalization will lead to much unemployment in the short term, followed by strengthening demand for agricultural goods in the medium term, followed by rising demand for labor in agriculture in the long term.) What effect will the policy have on inter-sectoral equity, in which urban living standards are currently substantially higher than rural? (Perhaps liberalization will improve the terms of trade between urban and rural sectors, leading to rising rural incomes and a slight drop in urban incomes.) Finally, I will assume (with a proviso) that all these factors are potentially efficacious as a source of motivation. The policy chosen may in the end be one that is judged to be consistent with the official's current interests, consistent with the electoral future of the party in power, and relatively progressive in its effects on social welfare and equity; and it may be the case that a different policy would have been chosen if welfare and equity concerns had not entered the calculation.
The proviso is this: there appears to be an almost unavoidable ranking of these considerations in any actual political decision-making context, such that the interests of the policy maker and his or her party must be satisfied as a necessary condition to the selection of a strategy. It is one thing to say that policymakers are concerned to further social welfare and equity; it is another to assume that they will do so even at the cost of loss of position, power, and privilege. We may read this proviso as a stipulation of political feasibility; policy initiatives that require the self-extinction of those who implement them are non-starters. Fortunately, the requirement of political feasibility does not narrow the opportunity set too drastically; there remain a range of options, some of which have better effects on the public welfare than others. What is excluded is the adoption of policies that would be best overall but that require that wielders of power relinquish substantial portions of their power, wealth, and privilege. (This is perhaps the central obstacle to meaningful land reform in the Philippines.)
If there is any force to this conception as a quasi-empirical theory of the reasoning of agents within the policy-making process, then debate and discussion of the goods and justice of development from the point of view of society as a whole have a practical role to play within development policy realm itself. In this case normative theories of the goods of development, the bad of poverty, or the requirements of justice in development may have the effect of clarifying the normative issues involved and thereby inducing policy agents to choose options they would not otherwise have chosen. (This approach would be to put the philosophical examination of the ethics of economic development into the category of other fields of applied ethics-e.g. medical ethics-in which one of the goals of the field is to affect the behavior of practitioners. One might hope that the Kennedy School's program for officials from developing countries might have this effect.)
So far we have considered the planning process from the point of view of the goals of development. But central to rational planning is the availability of reliable expectations about the effects of various possible strategies. Much of the analytic effort expended in development economics is directed toward this goal. Economists have assembled a variety of modeling techniques that are designed to work out the probable consequences of various possible policy interventions. Economic models are used to simulate the effects on a complex national economy, situated within an equally complex international economic system, of various possible strategies. Computable general equilibrium models (CGE) are among the most sophisticated of these systems. [1]
Economic development planning (like most other areas of policy formation) is aimed at multiple goods, superficially at least. (It may be possible to consolidate these goods to some ex tent.) Consider these goals: increasing the national wealth (GNP); modernizing industry and agriculture; improving food security; improving the well-being of all members of society; improving the provision of health and education services; reducing poverty; securing the conditions for stable democracy; ameliorating inequalities of income, power, status, and wealth; preserving and enhancing environmental quality; improving the well-being of future generations; effecting a transition to "sustainable" economic activity; and so on. There are a variety of social welfare goals-for example, economic growth, industrial modernization, improvement of social welfare across the board, and improvement of welfare of the poor. There are environment goals-for ex ample, environmental protection and sustainable development. There are equity goals-distributive justice among different groups in society and intergenerational justice. There are political goals-securing the conditions of democracy, establishing a stable and responsive government, securing electoral support for the regime in power, avoiding popular unrest. These goods are diverse and non-subsumable; so we must consider how multiple goods can be incorporated into the planning process. And we must consider how the priority of the poor is to be accommodated to the undoubted importance of other development goals as well.
The goal of economic development can be described rather briefly, it might appear: to promote a process of social and economic change leading to sustained improvement in the average welfare of society. And the central characteristics of this process are easy to describe as well: modernization of industry and agriculture, technical and organizational innovations leading to rising labor productivity, and establishment of efficient economic institutions. Development economists are inclined to summarize these features of development in this way: economic growth is aimed at increasing per capita national income, and this occurs as a result of the introduction of modern technologies and efficient market institutions within a stable political context. "Economic growth refers to a rise in national or per-capita income and product. . . . Economic development, in addition to a rise in per-capita income, implies fundamental changes in the structure of the economy" [Gillis, 1987]. To this "growth and modernization" stance at least some development economists will add an equity goal as well: income inequalities should be managed in such a way that the benefits of growth have a positive effect on all participants.
This simple description needs amplification, however. To start with, how are we to characterize the good of society? Social welfare theory brackets this problem by assuming something like a classical theory of utility; the goal of social policy is to maximize social welfare. This approach is probably too single-dimensional even for steady-state policy problems. But economic development involves massive social, technological, and economic change over an extended period of time, and it involves a set of heterogeneous goods. So a single dimension of "welfare" is inadequate as a basis for reasoning about development policy.
Consider the following list of goods: average and total national income; income shares disaggregated across social classes; life expectancy and morbidity (again, average and disaggregated); rising literacy and educational levels; broadening the scope of individual liberties; extending the scope of democratic decision-making; the welfare, health, and liberty of future generations; improving the equity of distribution of resources and incomes across regions, groups, and classes; preserving and enhancing environmental quality; and conserving natural resources. Each of these is a good thing; it would be desirable to improve each variable as a result of a given development policy choice. Unfortunately, two facts are relatively clear. First, these goods are largely heterogeneous; there is no common measure to which they could be reduced (e.g. utility). So we do not have the simple option of choosing that policy that maximizes utility across all these goods. And second, there are substantial conflicts among them: policies that favor some of these goods harm others. (There is debate within the development world, for example, over whether democratic institutions may not make the structural adjustments required by economic development difficult or impossible; so we are forced to choose between democracy and efficient growth. For a review of this debate see [Haggard, 1990, Periphery; Kohli, 1986, Democracy].) If some strategies strictly dominated others with regards to these goods, then the problem of choice is relatively simple: the former are strictly preferred to the latter. Most commonly, however, it will be the case that policy options are mixed, with policy A doing better than B with regard to good Gm but worse with regard to Gn.
If we are to make calculating choices in this circumstance, our options are limited. We might try to find a common measure for all the goods-which looks impossible. We might try to arrive at a set of priority rules ranking the goods of development: always favor an improvement in Gn over an improvement in Gm. This would be to give lexical priority to Gn over Gm (as Rawls assigns lexical priority to liberty over welfare [Rawls, 1971]). In general, however, it is difficult to justify a lexical ranking of goods, since this implies that any miniscule improvement in the higher-ranked good will justify massive losses in the inferior good.
Third, we might try to assign a rough and ready scheme of weights to the various goods (perhaps changing in different stages of development and at different levels of provision of the various goods). This option allows the policymaker to consider tradeoffs among goods: is a substantial savings of rain forest sufficient to justify a modest reduction in rights of political association? Is a modest increase in individual liberties sufficient to justify delaying poverty alleviation by a generation? This suggestion gives rise to a microeconomic approach. We might conceive of each of these goods-income to each group and environmental bads-as individual commodities and then construct a series of indifference curves between each pair of goods. These curves establish the "price ratios" of each good against every other, for every level of provision of the goods: the ratio of marginal quantities of each we are willing to exchange.
Figure 1. Indifference curves for income and environment
Panel A represents the idea that we will accept large environmental costs in order to achieve relatively small increases in incomes to the poor-when those incomes are very low; whereas we will require greater improvements in the income of the poor in order to accept a unit of environmental cost at higher levels of income. But the elasticity of environment with respect to the poor is high throughout the range; a small percentage of change in incomes to the poor is sufficient to justify a larger percentage of change in environmental quality. By contrast, the indifference curve in panel C embodies a low elasticity of environmental quality with respect to incomes to the non-poor.
This approach raises a difficult problem of interpretation, however: whose indifference curves are thought to be represented here? It is one thing to postulate individual indifference curves among commodities, based on observed price behavior. It is quite another to postulate a "social indifference curve" analogous to a social preference ranking. Presumably this problem is subject to the same sorts of impossibility results as arise within classical social choice theory. We might attempt to finesse the problem by regarding the policy-maker's indifference curve as the social ordering as well (this is the "dictator" strategy; [Arrow, 1963]). To the extent that we want economic development policy to derive from a democratic process, we need to have some way of aggregating individual preferences into a social preference function.
Finally, we might offer a mix of the last two approaches: rank some goods lexically and the rest according to a scheme of weights. Thus, for example, we might set a minimum level of poverty alleviation and basic democracy that must be satisfied before we begin sacrificing further advances in these goods for the sake of resource conservation or environmental protection.
The point of this discussion is messy but important: the policy-formation process cannot be adequately characterized in terms of a simple utility-maximizing framework of decision-making. There are heterogeneous goods that must be accommodated within the process; there are problems aggregating individual preferences into a social preference function; and the policy process is extended in time within a stochastic world, with the result that the planning process requires constant readjustment.
The book from which this paper derives is an extended argument for putting the poor first in economic development policy. This amounts to something like the following:
Economic development policies, both domestic and international, should be structured in such a way as to give highest priority to improving the welfare of the poor in developing countries.
Let us now attempt to formulate the poverty-first dictum more carefully. In terms of the discussion of heterogeneous goods in the previous section, I will defend this principle as involving a segmented hierarchy of the goods of development. Improving the welfare of the poor is to have lexical priority over all aggregate income improvements-in particular, over growth in per capita GNP. Poverty alleviation is to have a heavy weight in comparison to important non-welfare goods of development (resource conservation, environmental quality), particularly at low levels of income to the poor. And the interests of today's poor are to be weighted more heavily than tomorrow's poor, through a time discount function. Let us examine each relationship more closely.
The lexical priority of poverty alleviation over GNP growth reflects the idea that the needs of the poor are more urgent than those of higher income groups. Rising GNP per capita entails that average income is rising as well. But we can infer nothing from this fact about the behavior of incomes of the various income groups. It is entirely possible for increases in income to be concentrated in higher-income groups. The distribution of the fruits of growth depend entirely on the character of the distributive institutions in place in the developing economy. And it is not only possible but historically common that a widening income stratification will accompany growth, with the result that the poorest 20 to 40 percent of income earners show little or no improvement; their incomes may even fall absolutely as a result of worsening income inequalities [Fields, 1980].
It is important to mark the implications of this ranking. An increase in per capita GNP means an increase in overall social welfare: some groups, at least, derive higher incomes as a result of growth in national income, and this increase will have positive effects on their welfare. Moreover, there is no reason to suppose that the beneficiaries of growth are limited to the idle privileged classes; it may be, for example, that a growth-first strategy has a greater impact on the welfare of the near-poor than either the poor or the rich. (This would be true if the chief result of growth was a rising demand for modern-sector labor, leading to an increase in the size of this class and the real wage; [Fields, 1980, Poverty].) And on any account these improvements are important and desirable. Increase in income to the near-poor may mean that families are able to keep their children in school longer, to provide better nutrition and health-care, or to avoid the imperatives of child labor. Putting the poor first may entail that we select a policy option that provides a small improvement in the welfare of the poorest, at the cost of a much larger improvement in the welfare of the near-poor, and this is a substantial cost. [2] Nonetheless, I will defend this ranking of the priority of the poor over the rest of society, though it may be that others would prefer only to give greater weight to improvements to lower income categories [Chenery, 1974, #119]. [3]
Whatever justification there is for giving lexical priority to poverty alleviation over average growth, it does not extend to non-welfare goods such as environmental quality or resource conservation. It is utterly implausible to maintain that any small improvement in the condition of the poor would justify even massive environmental harm (e.g. burning down the Brazilian rainforest in order to create more cultivable acreage for the landless poor). What does seem true, however, is that increments in the welfare of the poor should count for more than increments in average welfare or the welfare of the non-poor in assessing environmental costs. The urgency of the needs of the poor would seem to justify environmental costs that improvements of the welfare of the non-poor could not justify. At the same time, this position requires that we make a calculating decision about possible tradeoffs between benefits to the poor and costs to non-welfare goods: the higher the non-welfare cost, the greater must be the benefit for the poor.
The reason why environmental quality and resource conservation appear more weighty than growth or average welfare seems to derive from a cross-temporal or cross-generational consideration. Environmental quality is a public good for persons of the present generation, but it is also a public good for persons of future generations. And the value of that future good depends on the development and consumption choices made by the present generation. Whatever else intergenerational justice requires, it demands at least that we should give some weight to important future interests in all our choices. Therefore it cannot be that the welfare of today's poor trumps any consideration of environmental cost. But here the asymmetry of time creates a new problem. Tomorrow's environmental quality is not a public good or a private good for today's consumers. So we must postulate that consumers and policy makers take some altruistic interest in the welfare of later generations; otherwise environmental quality may be expected to decline rapidly.
This brings us to the final priority identified above: placing relatively greater weight on improving the welfare of today's poor over tomorrow's poor. Here I am making several assumptions. Most important is the assumption that poverty alleviation is occurring through a carefully structured process of economic growth. This guarantees that the condition of tomorrow's poor will be better than today's poor. Consider figure 2 once again describing hypothetical trajectories of three strategies in terms of the income flowing to the poorest 40%. Strategy WF leads to an immediate improvement in income to the poor. Strategy LF produces a slower rate of improvement in the first several decades. After 40 years the curves cross; after this point the real income of the poor is higher on strategy LF. By that time, however, the deprivation of the poor is no longer as severe as it is today; real incomes have doubled by this point. On both strategies, we may suppose, the poorest strata have access to health care, adequate nutrition, and secondary education. A temporal discount of future benefits gives us a way of giving immediate improvements in welfare greater weight than future improvements; at the same time, however, it rules out the situation in which extremely small improvements in the present trump massive improvements in the future.
Consider one final difficult comparison of heterogeneous goods: this time the tradeoff between welfare of the poor and overall social well-being indicators such as infant mortality or longevity. Suppose we are faced with a pair of policies, one of which confers a bundle of additional resources on the poor, whereas the other provides for an across-the-board increase in social health subsidies. And suppose that the predictable consequences are these: Plan A leads to higher income and welfare for the poor, which in turn produces an improvement in health indicators. This improvement amounts only to a small percentage increase in average well-being, however. Plan B leaves incomes as they were but produces a uniform improvement in health status, with a larger percentage increase in health indicators for society as a whole. Both plans lead to socially desirable outcomes; so how should we choose between them?
Here too I take the position that improving the welfare of the poor should receive greater weight than improving average social well-being. But this position does not at all amount to diminishing the importance of the latter. It is certainly a good thing to improve the health status of the near-poor, the non-poor, and the rich. Given, though, that there are insufficient social resources to achieve all good things, we need to have a way of comparing the relative urgency of the choices before us. And the following is a strongly supportable generalization: one's position within the scheme of income inequalities within a typical developing society is a very good indicator of one's performance on non-income indicators such as health status or longevity. Low-income groups have sharply poorer access to food, health care, and literacy, with the result that their health status is dramatically lower than the social average. Thus longevity, infant mortality, and morbidity are strongly associated with income, and raising the incomes of the poor may be expected to have a substantial effect on their health status. (K. Nagaraj documents this sort of pattern in his analysis of infant mortality in Tamil Nadu [Nagaraj, 1986].) [4] This circumstance has two implications. First, the poverty-first approach is likely to have the greatest overall effect on health status for a particular group. And second, the improvements that occur do so at the level where they affect the most urgent human needs. It is always a good thing to add five years of life expectancy to a group average; but such an improvement is more significant when it takes the form of increasing the average from 45 to 50, than from 60 to 65.
A second topic in the normative framework of development policy concerns the problem of taking present and future interests of society into account-the problem of intergenerational justice. Suppose we are confronted with two development choices. The first policy package leads to an immediate and gradual improvement in the welfare of the poor, followed by slow but steady growth over the next 50 years. The second policy package leads to some decline in the position of the poorest 40%, then gradual improvement and medium steady growth for the next 50 years. Finally, suppose that the welfare of the poor on the second plan passes that of the first plan after 30 years. (This can best be visualized in the form of a graph of incomes flowing the poor over time.) Considered with strict temporal neutrality the second plan is preferable to the first; for an historically brief time the first dominates the second, whereas for the rest of time the second dominates the first. The gains achieved by the future poor vastly (perhaps infinitely) outweigh the sacrifices of today's poor.
On the other hand, we might reason along these lines. The needs of today's poor are extremely urgent. Infant mortality is high, morbidity is high, literacy is low, nutritional status is compromised, and so on. Today's poor need immediate attention. Tomorrow's poor deserves consideration as well; but their condition will be better than that of the poor today. Therefore we should give first priority to poverty alleviation for today's poor, and then turn to improving the prospects of later generations.
Both these arguments have a certain amount of force. (Ideally, of course, it would be preferable to adopt the poverty alleviation strategy for the first 30 years and then switch effortlessly to the growth strategy. But this may not be feasible; the technological and institutional innovations needed for the first may not support the second.) So how are we to resolve the issue; shall we maximize utility over time-thus favoring the growth strategy? Or shall we maximize the welfare of today's poor-thus favoring the poverty-first strategy?
One possible line of analysis involves introducing a time discount function. If we discount future utilities by even a low rate, then large gains in the distant future will not outweigh small gains in the near future. Moreover, if we do not discount future utilities, then we should always favor investment over current consumption-at every point in time. So a time discount is mandatory. But setting the discount rate high or low is a substantive moral issue, and we need to have a justification for our choice.
The way we treat this issue makes a great deal of practical difference in the kinds of development plans that we favor. If the present generation gets strong priority, then there will be less investment in productive assets and more in current consumption-food grain subsidies, free health care and education, and the like. On the other hand, if we favor future generations very heavily over the present generation (as might be justified by a temporally neutral social choice approach considering a large number of generations), then we will choose a mix of strategies heavily biased toward longterm investments in productive assets.
We now have a reasonably clear idea of what is meant by the principle of poverty-first development. But why should we think that this principle is acceptable? Why should we put such a high priority on the alleviation of poverty? Why is poverty such a singularly important problem, demanding our immediate and highest attention for remediation? We think of poverty as synonymous with low income, implying material deprivation; and none of us would choose to live in the material circumstances of the poor. But we can say more than this about the bad of poverty: there is a fundamental relationship between poverty and obstacles to human development. It is not merely that the poor are able to consume less than the non-poor; it is that they are less able to develop their full human potential. Poor or absent medical services lead to illness and premature mortality; poor education and illiteracy lead to stunted intellectual achievement; long and tedious work hours (when available) give rise to demoralized everyday experience; and the social stigma of shabby dress, poor housing, and low status employment lead to a failure of self-respect. So poverty is a unique bad in its concentrated destructive effects on the realization of the full human capacities of the poor: it is not just that the poor have a lower standard of living, but that they live less than fully human lives.
A. K. Sen's writings have contributed much to clarification of the bad of poverty. In his lectures on the standard of living Sen [Sen, 1987, Living] distinguishes between a commodity-based definition of the standard of living and a "human functioning" view of well-being. His central insight is that we are centrally concerned with human beings in possession of a bundle of human capabilities which can be either realized or impeded through the economic and social environment in which the person is located.
What reasons can be put forward to defend the poverty-first position? It might be argued, against this position, that all members of society should be treated equally; and this means that no group should receive priority. So privileging the poor is unfair and contrary to a principle of equality of treatment. Instead, it might be insisted that improving average welfare should be the goal, or that equal proportional increases in welfare for all income classes should be the goal. Or again, one might hold that poverty alleviation is a good, to be sure; but it is only one out of many important social goals, and should receive no more weight than other important goals-e.g. environmental goals.
In this section I will consider a number of arguments for giving priority to poverty alleviation over other social goods in development planning. None of these arguments is wholly compelling. But taken together they give strong reasons for giving high priority to improving the welfare of the poorest strata of society.
There is a straightforward social-welfare rationale for putting the poor first. Sup pose that we postulate a utility-based social welfare theory and we maintain that policies should be constructed so as to maximize average social welfare. The difference between the poor and the non-poor is that the former have less income. By almost any measure there is a declining marginal utility of income: the utility gained by expending an additional dollar in a high-income situation is less than that gained by expending a dollar in a low-income situation. Therefore we can maximize social welfare by biasing policies in the direction of increasing incomes flowing to the poor. [5]
This approach has very strong implications for distributive justice. Consider the simplest case in which we have a set of n income groups and an amount of incremental income M to distribute. The assumption of diminishing marginal utility of income entails that utility is maximized when incomes are equal across groups. Utility is maximized, then, if income is redistributed to achieve income equality across groups. If we impose the constraint that no existing incomes can be redistributed (reflecting individual entitlements that cannot be changed), then utility is maximized by directing new sources of income toward raising incomes of the poorest groups. It can be shown that the increased utility produced by the expenditure of M is maximized by raising incomes from the bottom up: raising the income of the last group to the level of the second-lowest group; then raise this group up to the level of the third-lowest group; and so on until the resources are exhausted. [6]
It might be noted that this finding is sharply at odds with a common criticism of utilitarian ism: that the principle of utility is indifferent to distribution [Rawls, 1971]. This is true in one obvious sense: the principle is indifferent between cases in which the same amount of utility is distributed over a small minority or diffused over a large majority. However, when we turn to consideration of the distribution of resources, along with the law of the diminishing marginal utility of resources, we get a strongly egalitarian conclusion: utility is maximized by distributing resources more equally.
Social welfare theory is generally premised on a theory of utility as subjective preference. The good to be achieved through social choice is the satisfaction of individual preferences-what ever they are. If we adopt this theory of the good that is the ultimate object of social choice, then it is difficult to see why resources should be biased toward the poor. The poor will enjoy greater utility through expenditure of these resources, to be sure; but this gain will be accompanied by a comparable loss in preference-satisfaction by the non-poor, and the net sum will be zero. To put it as simply as possible, the poor will be able to purchase a more nutritious bundle of foods, but at the cost of the non-poor being less able to purchase textbooks for their children's schooling and of the rich being less able to furnish their homes as they desire. The gains to the poor are balanced by the losses to the non-poor. Therefore it is impermissible to privilege the poor; instead, the equality of persons requires that development should be neutral between all persons' preferences.
This argument turns on a subjective theory of welfare. Well-being is simply the level of preference-satisfaction an individual achieves-whatever the content of his or her preferences. My preference for truffles should receive as much weight as your preference for clean water. However, a number of philosophers have argued that an objective theory of welfare is superior. (See, for example, [Scanlon, 1975], [Griffin, 1986, #297], [Braybrooke, 1987, Needs], and several authors in [Sen, 1982, Utilitarianism].) On this approach, we ought not be neutral among subjective preferences. Instead, we should distinguish between basic and non-basic preferences on the basis of a theory of how the objects of preference relate to human needs at a variety of levels. It is a natural fact about human beings that they need clothing, food, and shelter. Somewhat less fundamental are needs for education, health services, and retirement security. Likewise, humans need meaningful work, opportunities for satisfying social relationships, the grounds of self-respect, and free political activity. Finally, it might be said that persons need varied opportunities for entertainment and sport, access to libraries and parks, access to means of communication (telephone, television, personal computer), and access to the marks of status and public acceptance-attractive clothing, fashionable running shoes, or a mountain bike.
It is fairly clear that there are important differences in the needs listed here. If it is a choice between providing malaria pills to a cohort of Bangladeshi toddlers or the latest Air Jordan sneakers to a cohort of Boston teenagers, it is not difficult to argue that the former should take priority over the latter. And this remains true even if we imagine that the net preference-satisfaction produced by the sneakers is equal to that produced by the malaria pills. (Indeed, it might well be that the Boston teenagers feel intensely deprived, whereas the Bangladeshi toddlers are so demoralized that they would experience no subjective deprivation if the pills were not available.) The harder question is whether this distinction can be generalized. Is there a basis for holding that certain needs are more important than others?
Thomas Scanlon offers the distinction between urgent and non-urgent needs as an effort to make sense of this sort of example. The general idea is that certain needs are more fundamental than others because they are closer to the preconditions of continued existence. Scanlon's own analysis is somewhat undetailed. But A. K. Sen's discussion of capabilities provides a more developed basis for generalizing this notion. On Sen's account, the most fundamental notion involved in human welfare is the idea of normal human functioning and the capabilities that must be realized to permit this level of functioning [Sen, 1987, Living].
Let us attempt to generalize this approach. We can analyze urgency of needs in terms of the centrality of the capability that a certain good serves in the attainment of full human functioning. I will define a need in terms of the goods, resources, or institutions that are necessary for realizing some aspect of a fully developed human being. And one need is more urgent than an other if it is a prerequisite to the achievement of a feature of the full human personality that is prior to the feature satisfied by the second need. For example, food is more urgent than education because a poorly fed child may die, whereas a poorly educated but well-fed child will survive. Likewise, primary education is more urgent than access to libraries, because children who lack primary school will never be able to make use of libraries. (See [Braybrooke, 1987, Needs] and [Griffin, 1986] for extensive development of the conception of need and objective interests.)
If this distinction between urgent and non-urgent needs is viable, then we have a distinctive basis for justifying the poverty-first principle. Both poor and non-poor have a range of needs that are not yet satisfied. But the poor have a much lower level of satisfaction of urgent needs than the non-poor. And development planning ought to be biased toward satisfying urgent needs prior to non-urgent needs.
This line of thought looks most plausible when we consider the extreme cases-e.g. malaria pills versus sneakers. But it should be noted that other cases are more difficult. For most of the non-poor in developing societies are still possessed with a range of relatively urgent needs: higher quality food, more comfortable housing, access to secondary school, and so on. These are not trivial human needs; to the contrary, their satisfaction is mandatory in pursuit of full human functioning. And so the tradeoff to be considered is between satisfying the very urgent needs of the poor versus the moderately urgent needs of the non-poor. And it is not self-evident that justice requires that we give absolute priority to the former; on the contrary, it is persuasive to suppose that large improvements in the latter might justly be chosen over small improvements in the for mer.
We might motivate a priority for the poor along these lines: anyone would favor this principle in advance if uncertain whether or not he or she will be among the poor. So fairness dictates that we should embody this principle in economic development policy. This is a line of thought that follows John Rawls's construction of justice as fairness.
This approach can be developed further through discussion of Rawls's own principle of distributive justice, the difference principle [Rawls, 1971]. Rawls's difference principle establishes a baseline in terms of which to evaluate social and economic inequalities, and has the effect of concentrating our attention on the welfare of the least-well-off position in society. This principle has implications for economic policies in the less-developed world today; it implies that development policies ought to be structured so as to have the largest possible effect on raising the welfare of the poor in those societies. Rawls has not applied this framework to the problem of economic development. (Indeed, Rawls suggests that the "circumstances of justice" do not emerge until rather late in the process of economic development [Rawls, 1971]. ([Pogge, 1989] and [Beitz, 1981] have each borrowed some elements of Rawls's theory in their own theories of international justice.) However, Rawls's theory of justice has the merit of providing an extensively developed discussion of the justice of inequalities, along with a broad literature identifying some of the strengths and weaknesses of this framework.
Let us look briefly, then, at Rawls's theory of distributive justice. The concept of justice as fairness depends on the idea that just social institutions are those that would be acceptable to all members of society, given appropriate restrictions on knowledge about personal circumstances. Rawls makes this idea more specific by introducing the idea of the original position, in which hypothetical members of society deliberate about a basic charter for society (a theory of justice) behind a veil of ignorance. Participants are postulated to have extensive knowledge about economic arrangements but no knowledge about their own particular circumstances; and they are assumed to choose a theory of justice on the basis of their assessment of their own best future interests.
Rawls's theory of justice rests upon his argument that two principles of justice would be chosen in these circumstances of choice to regulate the basic institutions of society. "The first statement of the two principles reads as follows. First: each person is to have an equal right to the most extensive basic liberty compatible with a similar liberty for others. Second: social and economic inequalities are to be arranged so that they are both (a) reasonably expected to be to everyone's advantage, and (b) attached to positions and offices open to all" [Rawls, 1971,, : 60-61]. The first principle is referred to as the "liberty principle," while the second is the "difference principle." It is the difference principle which is of greatest interest in development studies. Rawls amplifies the difference principle in the following terms: inequalities in initial life prospects are justifiable "only if the difference in expectation is to the advantage of the representative man who is worse off, in this case the representative unskilled worker" (78).
Rawls applies the difference principle in terms of what he refers to as "primary goods"-goods that are needed for the furtherance of virtually any human life plan. He includes income, status, and the prerequisites of self-respect. This conception of primary goods converges nicely with the idea of "basic-needs" satisfaction and the fruition of human capabilities that A. K. Sen has offered (discussed above).
The difference principle requires that
inequalities are justifiable only insofar as they work to the advantage of the least-well-off.
Justice requires, then, that the basic structure of society be arranged so as to create the least sys tem of inequalities consistent with maximizing the position of the least-well-off. In the context of development, then, we may formulate the following principle:
development strategies should be chosen so as to produce the most rapid improvement for the least-well-off.
This principle implies that we should not be concerned about average GNP or total GNP directly, but rather on the share of GNP flowing to the poorest 20%. It also implies that if two outcomes assign the same absolute amount of income to the poorest segment, we should choose that out come that contains the least inequality-thereby reducing the total of GNP. We should not "stretch" the ladder of income. This principle, however, has seriously counterintuitive implications in the context of development theory. For one of the central goals of development is to expand the welfare of all members of society-to create a material basis for more fulfilling human life. Growth, in other words, is of inherent value in development. The difference principle, however, appears to imply that we should choose to minimize growth, by selecting against policies that benefit groups other than the least-well-off. This implication violates a reasonable Pareto condition, and would not be chosen by rational agents in the situation of the original position. In stead, agents would be interested in establishing the highest welfare floor possible, and then maximizing the welfare of each group above the floor. Consider, then, the following principle:
First maximize the income to the poorest 10%; then maximize the income to the next decile; and so on through the income ladder.
This principle involves maximizing GNP sequentially from the ground up; it does not permit tradeoffs between gains for higher deciles at the expense of lesser losses for lower deciles. This principle puts the poor first, but also places positive weight on income gains to other groups. It thus stretches out the income ladder and maximizing GNP consistent with maximizing the position of the least-well-off.
Finally, we must consider whether there is a human rights basis for favoring the poor over the non-poor in economic development. (Henry Shue adopts such a position; [Shue, 1980, Rights].) This line of thought flows rather naturally from considerations raised concerning urgent human needs. Consider the following argument. All human beings have a right to access to the minimal goods needed to satisfy basic human needs. This right entails that if we are confronted with a group whose basic needs are not currently satisfied and if there are alternative plans that lead either to increased satisfaction of basic needs or to satisfaction of other persons' non-basic rights, we should choose the former. The poor do not have a high level of basic-needs satisfaction. They therefore have a right to priority in development planning until their basic needs have been satisfied. Therefore development planning should give priority to improving the incomes flowing to the poor.
There is thus a family of converging arguments that support giving priority to poverty alleviation. At the same time, we have seen that there are other important goals within the development process as well. The program of "putting the poor first" must therefore be formulated rather carefully, and the doctrine cannot reasonably be construed as requiring lexical priority over all other social goods.
I have now argued that justice requires that development strategy place top priority on poverty alleviation. More abstractly, development policies ought to conform to the difference principle. We now need to ask to whom this normative advice is directed? There are two agents in the development process who have the ability to adopt this policy goal, and they are agents of quite unequal influence. First, there are the governments of the developing countries themselves. Each government invests substantial resources in development planning and implementation, and these processes are guided by national economic priorities. Many national governments already give rhetorical allegiance to a "poverty first" approach-for example, Indian planning has fairly consistently taken this stance since independence. However, in many instances this rhetoric is not accompanied by corresponding policy commitments. Each government has a set of domestic constituencies who are affected in different ways by the development process. So the first audience of this normative argument is the government planning agencies and domestic constituencies-the poor-in the less developed countries.
The second category of player in the development game is the international development agency that offers advice and aid to developing countries. The World Bank, US-AID, United Nations agencies, EEC development agencies, and the like, offer advice and program aid to poor countries; and these interventions are based on a mix of normative and analytical assumptions. We must not overestimate the ability of international agencies to affect the development process in various poor countries; in general it is domestic governments rather than international agencies that call the tune. However, it is also clearly true that international agencies have some leverage with domestic governments. The sort of argument offered above is intended to reaffirm and deepen the "poverty first" normative orientation that was current in development thinking in the 1960s and 1970s, but that is now regarded as infeasible or undesirable. There are good normative reasons for concluding that the "poverty first" approach is superior to the "growth first" approach on welfare grounds. And there are good analytical reasons to believe that there are economically feasible development strategies available that effectively pursue the poverty first approach. So a second goal of this argument is to make the argument to development policy makers in the industrialized world that development advice and aid should be structured around this goal.
Let us regroup. Economic development is an instance of goal-directed public policy, in which certain goods are selected as goals, and plans and strategies are adopted so as to bring about those goods. Here I have focused on several normative characteristics of this process.
First, I have drawn attention to the complex nature of the goals of development. There are heterogeneous goods to be advanced, these goods have shifting characteristics over time, and the goods themselves may be construed as reflecting either the preferences of specific individuals (policy makers) or society at large.
Second is the problem of the role of justice, and normative considerations more generally, within the policy process. Is it reasonable to assume that assumptions about justice and fairness have some motivational effect on policy makers? Or are we forced to adopt the somewhat cynical perspective of narrow public choice theory, according to which officials are assumed to pursue their only own private interests? I have offered reasons to believe that normative commitments can be motivationally effective, while recognizing the importance and frequency of policies driven by the latter sorts of motives. This argument has a substantial practical import, since it means that philosophical discussions of the justice of development may in fact affect the policy process.
The paper then turns to the substantive content of development policy. I hold that justice requires that development should put a high priority on improving the welfare of the poor. While recognizing that there are many legitimate goods to be furthered in development, I argue that poverty alleviation should have the greatest weight. More fully, improving the welfare of the poor should have lexical priority over average growth, it should have a substantial (but not lexical) weight in comparison to intertemporal goods (environment, resources), and today's poor should receive greater weight than tomorrow's poor (through a time discount function).
It is one thing to advance such a view and quite another to defend it. In the final portion of the paper I argue that poverty is the most important consideration within the development process because of the intimate connection between poverty and the fulfillment of human capacities. The poor are not merely relatively deprived; they are blocked from achieving full realization of their human potential. This basic perspective is then further articulated through a series of arguments that converge on the conclusion that the poor should receive highest priority: a social welfare argument, an urgency argument, a fairness argument, and a human rights argument.
[1] See Taylor, ed. 1990 for an extensive description of CGE models, as well as a number of useful applied models.
[2] The situation is not quite as desperate as this scenario might suggest, however, in that poverty-first growth strategies may be expected to have positive effects on the near-poor as well. This is probably an instance of the situation that Rawls describes in his presentation of the difference principle: focusing on the welfare of the least-well-off leads to reforms that have the effect of pushing the whole income ladder upward.
[3] Hollis Chenery suggests an even simpler scheme: a linear utility function in which income is disaggregated over deciles, with lower deciles having larger coefficients than higher deciles. This has the effect of weighting improvements in lower-income groups more heavily than improvements in higher-income groups; Chenery et al. 1974.
[4] "The average value of infant mortality rate in Tamil Nadu-which in itself is quite high-conceals a great deal of variation across different social classes with the socially disadvantaged sections having an above average value of IMR" (Nagaraj 1986, p.33).
[5] R. M. Hare makes an argument along these lines in "Ethical Theory and Utilitarianism" (Sen and Williams, eds., 1982, p. 27).
[6] Strictly speaking this finding only applies on the assumption that we are dealing with a purely allocative problem, and development is not solely concerned with allocation of consumption resources. The goal of much development spending is not to directly increase welfare through expanded consumption, but rather to create the possibility of future expansions in welfare through investment in the pre sent. However, we saw above that there are distributive choices that planners must make; and this finding shows that social welfare is increased by favoring distributive outcomes that narrow inequalities by raising the welfare of the poor.
