77410 THE WORLD BANK ECONOMIC REVIEW, VOL. 17, NO. 2 297–309 Child Labor: A Normative Perspective Debra Satz Examining child labor through the lenses of weak agency, distributive inequality, and harm suggests that not all work performed by children is equally morally objection- able. Some work, especially work that does not interfere with or undermine their health or education, may allow children to develop skills they need to become well-functioning adults and broaden their future opportunities. Other work, including child prostitution and bonded labor, is unambiguously detrimental to children. Eliminating these forms of child labor should be the highest priority. Blanket bans on all child labor may drive families to choose even worse options for their children, however. Moreover, child labor is often a symptom of other problems—poverty, inadequate education systems, discrimination within families, ethnic conflicts, inadequately protected human rights, weak democratic institutions—that will not be eliminated by banning child labor. The International Labour Organization (ILO 2002) estimates that more than 246 million children are engaged in labor. Although the incidence of child labor has been falling globally, it is doing so unevenly, and in some areas it appears to be on the rise (Fallon and Tzannatos 1998). The widespread existence of child labor has provoked both popular outrage and legislative initiatives aimed at banning the sale of all products made by children. But developing economies—and many development economists—have cautioned against universally proscribing child labor. They argue that such bans will be inefficient and will hurt poor families and their children. Some economists have voiced concern about paternalistic interference with family strategies that may have evolved rationally in the context of poverty and inadequate education systems. Others point out that because child labor is itself heterogeneous, ranging from light work delivering newspapers after school to child prostitution, uniform policies may undermine the ability to target its worst forms. There is considerable debate, then, as to whether establishing and enforcing a uniform worldwide set of standards for dealing with child labor is desirable. Debra Satz is associate professor of philosophy at Stanford University; e-mail: dsatz@csli.stanford.edu. This article was originally prepared for the conference on ‘‘The Economics of Child Labor,’’ organized by Zafiris Tzannatos and Kaushik Basu and held in Oslo, 28–29 May 2002. Joe Shapiro provided research assistance on this article. The author thanks Elizabeth Anderson, Susan Okin, John Ferejohn, Lori Gruen, Ben Crow, Rob Reich, and two anonymous reviewers for helpful comments. DOI: 10.1093/wber/lhg015 Ó 2003 The International Bank for Reconstruction and Development / THE WORLD BANK 297 298 THE WORLD BANK ECONOMIC REVIEW, VOL. 17, NO. 2 Against the background of this debate, this article explores the normative issues posed by child labor. It identifies several considerations that make child labor morally problematic, considerations that turn on issues of weak agency, distributive inequality, and harm.1 It concludes that the worst forms of child labor, including child prostitution and the use of children in wars or as bonded laborers, should be unconditionally prohibited. Other types of child labor may need to be tolerated under certain circum- stances, at least in the near future, even as efforts are made to eradicate them. Legal toleration, however, does not imply indifference, and states and non- governmental organizations (NGOs) can protect and promote the interests of children in many ways. In particular, they can take broad social measures to improve outcomes for children, especially by ensuring that all working children are educated. Whatever policies are adopted will involve tradeoffs between different values. Policymakers need to make explicit the values they want to promote and the tradeoffs they are willing to accept. Normative judgments cannot be escaped: they are implicated in the selection of research questions, in the data sought, and in policy design. I. WHAT IS A CHILD? Many countries define childhood in terms of chronological age; others take into account social factors. In some African countries, for example, 10-year-old apprentices or brides are no longer assumed to possess all the characteristics that industrial countries bundle together into the status of ‘‘child.’’ They may be eligible for marriage but not entitled to make decisions independently of their parents. Different countries invoke different age thresholds of adulthood; even within countries such thresholds can diverge—one age for voting, another for employment, another for military service. What is the normative basis of modern society’s view of childhood?2 The concept of a child, implicit in moral and legal practices, is that a child is a person who is in some fundamental way not developed but rather developing (Schapiro 1999). Because of this undeveloped condition, adult parents or surro- gates are needed to act on children’s behalf. Parents or surrogates are thus given special obligations, including the obligations to protect, nurture, and educate children. These obligations are paternalistic, because adults feel bound to fulfill them, whether the children in question consent to be protected, nurtured, or educated. 1. For a discussion of the dimensions of ‘‘noxious markets,’’ see Kanbur (forthcoming) and Satz (forthcoming). I have been very influenced here by Kanbur’s approach. 2. Despite the different age thresholds and bundlings they employ, almost all societies share a common view of childhood. If this seems overreaching, it is certainly true that a common notion is shared by the United Nations, liberal democracies, and most international aid agencies. Satz 299 Adults feel justified in treating children paternalistically because children have not yet developed the cognitive, moral, and affective capacities to deliberate and act competently in their own interests.3 At the same time, children have legitimate claims to have their interests considered: they are not simply tools. Children are not yet full persons, but they are persons. II. NORMATIVE DIMENSIONS OF CHILD LABOR Consider the normative dimensions of child labor. Child labor raises moral concerns because of the weak agency of children (and sometimes their parents), its connections to underlying inequalities, and especially its potential for harm. Weak Agency Children cannot be assumed to have full agency. They lack the cognitive, moral, and affective capacities of adults, and they seldom have the power in the family to make decisions about how to allocate their time.4 As Humphries (1999) has pointed out, there is no infans economicus responding to market signals; most children are put to work by their parents. Parents are the primary decision- makers for children, especially very young children, exercising authority and control over most aspects of their children’s lives. Consider the contrast with ideal labor markets, in which workers and employers are fully rational agents who transact on their own behalves with perfect information. In child labor, as noted, parents make the market decisions concerning their children’s time. This gap between chooser and chosen for in the market for child labor opens up the possibility that children’s interests will be discounted. Surrogate decisionmaking is a morally fraught arena, especially in the case of young children, who often cannot even articulate their own interests. Moreover, such surrogate agency often breaks down, as in the case of parents who lose custody of children they have abused, exploited, or neglected. Child labor also differs from ideal labor markets in that the decisionmaker may lack relevant information regarding the consequences of his or her choice. The costs of child labor can extend far into the future, having, for example, long-term adverse effects on health. It is not clear that these costs are taken into account, even by well-meaning parents. Lack of information may be especially 3. Children should not be seen as merely passive ‘‘patients’’ whose opinions never need be consulted. Clearly, the extent of children’s agency increases over time, so that 3-year-olds differ dramatically from 16-year-olds in terms of their level of effective rational agency. The fact that children’s agency is lower than that of adults does not denigrate the contributions children make to their own well-being or the well-being of others. 4. Children orphaned by AIDS or civil wars and older children who have fled abusive homes do make decisions on their own behalf. But even in these cases, to the extent that their powers of decision remain undeveloped, they cannot generally be seen as full agents. 300 THE WORLD BANK ECONOMIC REVIEW, VOL. 17, NO. 2 important if the parents are themselves from very poor or despised castes. As Dre` ze and Gazdar (1996:86) point out, ‘‘the ability of parents to assess the personal and social value of education depends, among other things, on the information they have at their disposal. If their entire reference group is largely untouched by the experience of being educated, that information might be quite limited.’’ It is noteworthy that children in bonded labor tend to have parents who were also bonded laborers (Burra 1995). Agency problems (surrogate decisionmaking, ignorance, uncertainty about the future) may be associated with child labor. But even if child labor were fully informed and voluntary, it would not necessarily be morally justified. If all the options poor children and their parents face are unjust, the option chosen does not by some mysterious process become just. A key input for the moral assess- ment of an action depends on one’s views about the moral legitimacy of the socially available choices an agent faces. Whether a voluntary choice confers legitimacy depends on other conditions besides its being voluntary. Distributive Inequality Child labor may appear particularly objectionable because of the inequalities that underlie it. These inequalities can occur between societies (poor children working in rich multinational firms) or between families within a society (domestic elites whose children receive excellent education versus poor families whose children work as bonded laborers). Child labor appears as a symptom of an objectionable level of inequality. In many countries, undemocratic institutions and caste and ethnic divisions compound these inequalities. Child labor can also manifest and perpetuate inequality within families. Some families may sacrifice a working child for the sake of other children or family members. They may, for example, keep girls out of school to care for younger children while the mother works outside the home.5 The bias in favor of some children within a family over others is troublesome (see Jejeebhoy 1992).6 Harmful Outcomes The nature of the damage generated by child labor depends on the form of child labor. Many international protocols (including the ILO’s Worst Forms of Child Labor Convention 182 and the Sanders Amendment considered by the U.S. Senate in 1997) view forced labor as one of the worst forms of child labor. But 5. Indeed, girls may be systematically undervalued by their families. Such discounting helps explain why, as Amartya Sen has dramatically phrased it, ‘‘more than 100 million women are missing,’’ mainly in South Asia and China (see Sen, 1999). 6. Child labor may also reflect power inequalities between mothers and fathers. A growing body of evidence suggests that mothers have a stronger preference than fathers for investing in their children’s welfare, including education (Haddad, Hoddinott, and Alderman 1977). See Agarwal (1995) for evidence that land allocation to women rather than men results not only in higher productivity in agriculture but also in better outcomes for children. Satz 301 forced labor is not a useful category for distinguishing the most harmful forms of child labor from others. Parents make paternalistic decisions on behalf of their children that can include forcing children to go to school. It follows that almost all child labor (and child education) is forced. It is not possible to identify what is harmful about child labor without a fuller theory of children’s interests. Children have two kinds of interests, what Sen (1985) calls welfare interests and agency interests. Welfare interests concern a person’s overall good; agency interests concern the ability to participate in deciding matters that bear on that good. Both children and adults have these interests but in different ways and to different degrees. Consider welfare interests first. A child’s present welfare interests include shelter, food, health, education, bodily integrity, and a stable, loving relationship with his or her parents (or other caregivers). Children need parents to protect and provide for these interests because they cannot yet provide for them themselves.7 The state needs to play a crucial role in protecting children against parental abuse and neglect. An adult’s welfare interests are different. First, adults are not dependent on others in the same way children are. Given appropriate background conditions, adults are assumed to have the capacity to provide for their own welfare: to obtain nourishment, health, and shelter; avoid escapable mortality and premature morbidity; and exercise a range of capabilities. Second, adults’ welfare is shaped by their own values and concerns—values and concerns that they have the capacity to endorse or change. Very young children have few immediate agency interests. But unlike other dependent and vulnerable people (for example, people with severe cognitive disabilities), given appropriate background conditions children will develop the capabilities to set goals for themselves and act in accordance with their own values. As they develop, children’s interest in exercising their agency grows, although given their lack of competency and experience, societies still reason- ably set legal bounds on it. Adults, by contrast, have a significant interest in exercising their agency, in being educated participants in decisions that affect their lives. They find it offensive to be treated as children. They willingly allow others, such as political leaders, to make decisions on their behalf only with their consent. Ignorance and undemocratic institutions, which prevail in many of the world’s poorest states, are serious obstacles to the achievement and exercise of adult agency. Although the interests of children and adults thus differ, children are also developing into adults. Any theory of children’s interests must look at those interests dynamically, as contributing to the development of their interests as 7. Of course, parents cannot provide all of the things children need, such as a clean environment. In this sense both children and parents depend on larger social institutions. 302 THE WORLD BANK ECONOMIC REVIEW, VOL. 17, NO. 2 adults. No society can be indifferent to how children are raised and educated, because these factors affect the nature of its future citizens. Uneducated, illiter- ate, and passive adults will not be able to contribute much to social develop- ment or play a role in responding to social problems (Sen 1999). Harms can be defined in terms of negative effects on a child’s present or future (adult) agency and well-being. In particular, one can define a level of basic agency and well-being interests, the failure to satisfy which would be abusive to children, stunt the development of crucial adult capabilities, or be subversive to the state’s interests in producing an informed citizenry. Child labor that violates children’s basic interests would constitute extreme harm. It is important to distinguish this standard from the ‘‘best interests of the child’’ standard that some children’s advocates have proposed for judging child labor. That standard suffers from two major problems. First, because there is no widely shared view of what constitutes a child’s best interests, parents can interpret the standard in radically different ways. Broad consensus is much more likely to be reached on a basic interests standard.8 Second, the best interests standard assumes that parents (which in practice usually means mothers) are mere instru- ments for optimizing their children’s interests and do not count independently. From a moral point of view, this is just wrong. There is no inherent injustice in family structures that assume that children must make some contribution to the well-being of their families as a whole or to other family members. Tradeoffs among interests within the family are acceptable and to some extent inevitable. Work performed by children might thus be acceptable under certain conditions and given certain restrictions.9 III. POLICY IMPLICATIONS What should the response be to child labor that scores poorly along these normative dimensions, manifesting weak agency on the part of children or their parents, inequality within and between families, or very bad outcomes for children? One approach, taken by some activists and NGOs, is to perceive all child labor as a violation of the rights of the child and to call for its immediate abolition. Within this framework, drawing distinctions between kinds of child labor—hazardous versus nonhazardous, bonded versus nonbonded, part-time versus full-time—is considered pointless, because anything short of full-time 8. What if a state rejects the existence of these core interests? The welfare and agency interests identified here are those whose satisfaction describes what might be called a minimally decent human life. There are a variety of ways of supporting the claim that such interests are in fact universal features of an adequate human life. One could, for example, appeal to certain physiological and psychological needs that people have regardless of their cultural circumstances or draw on the choices people make for themselves when they are in a position to make meaningful choices (see Barry 2001). 9. Given that in the family parents’ interests are likely to prevail over children’s interests, there are practical reasons to adopt an approach to child labor that focuses on what happens to the children in a family. Satz 303 formal education for children is seen as a threat to children’s basic interests. (Kabeer 2001:4) Although this approach offers little guidance on how it could be implemented—a serious concern in the context of weak states—it nevertheless has an important policy function. Rights, especially legal rights, create, legitimate, and reinforce social understandings about what people deserve (Kahneman, Knetch, and Thaler 1987). Articulating rights for children may thus have positive effects on children’s welfare by reinforcing the idea that children have a claim on the state, society, and the international community for their protection. Assessing the practicality of abolishing child labor by strictly enforcing legal sanctions is difficult, because we do not really know whether child labor is an unavoidable reality for poor countries. Debate continues over the extent to which child labor is caused by poverty and underdevelopment or by policy failures, including failures arising from social and political inequality. Weiner (1991), for example, argues that Indian elites fail to enforce compulsory universal education because they believe that educating the poor will lead to the overthrow of their rule. Indeed, children’s education, rather than child labor, has been linked to economic development. China, the Republic of Korea, and Taiwan (China) all made rapid economic progress while promoting basic education. Banning child labor and thus restricting the labor market may raise the wages of adult workers enough to make children’s work unnecessary (Basu 1999). We do not yet know the limits of the possible within poor countries themselves or what the industrial countries might do to eradicate child labor if they really had the will. Given resource constraints and the likely need for tradeoffs between values, blanket prohibitions on child labor face two important challenges. First, in some contexts, bans on all child labor may drive families to choose even worse options for their children. Children are better off attending school part-time than not at all; they are presumably better off working in factories than as prostitutes or soldiers. Policymakers must thus take care to combine legislation or efforts to ban all child labor markets with policies designed to protect children from worse outcomes on the black market. The second objection to immediate bans on all child labor stems from recognition that child labor is often a symptom of other problems—poverty, inadequate education systems, discrimination within families, ethnic conflicts, inadequately protected human rights, weak democratic institutions—that will not be eliminated by banning child labor. Blanket legislation against child labor may do nothing to address the underlying problems. Many children who do not work do not attend school. Many of these ‘‘nowhere’’ children are likely to be girls (Bhatty 1998). A focus on enforcing legislative solutions may not solve the problems and may direct scarce resources away from other methods of improving children’s lives. The framework adopted here provides the basis for a somewhat different approach. Child labor can be examined through the lenses of weak agency (espe- cially in the form of parental ignorance and adaptive preferences), distributive inequality, and harm. Within this framework not all work performed by children 304 THE WORLD BANK ECONOMIC REVIEW, VOL. 17, NO. 2 is equally morally objectionable. Some work, especially work that does not interfere with or undermine their health or education, may allow children to develop skills they need to become well-functioning adults and broaden their future opportunities. Indeed, in some countries, given the deficiencies of the public education system, some children work to earn the tuition for private education (Brown, Deardorff, and Stern 2003). Child labor is most objectionable where it clearly violates children’s basic interests. The miserable conditions of abuse that children suffer in some kinds of work cannot be seen as in a child’s basic interests, present or future. According to the most recent study by the ILO (2002), 171 million working children—two-thirds of all working children—are routinely exposed to health risks, abuse, and probable injury.10 An estimated 8.4 million children are caught in what the ILO refers to as ‘‘unconditional worst’’ forms of labor, including slavery, trafficking, debt bondage, participation in armed conflict, prostitution, and pornography. Eliminating these forms of child labor should be the highest priority. Even if under some circum- stances children have to work, at least in the short term, there is no reason that they should suffer the kind of maltreatment that underlies such practices. No state, NGO, family, lending agency, or consumer can justify participating in activities in which the basic rights of children are completely disregarded, in which children are treated with contempt, their lives disposed of as carelessly as the contents of a trashcan. Two other considerations should also be used to determine how harmful a child labor practice is. First, children who work and do not go to school will likely lack the capacities that they need—literacy, numeracy, broad knowledge of personal and social alternatives, communication skills—to effectively exercise their agency as adults. One central benefit of education is the ability of an educated person to choose in a more informed way. Education thus deeply influences the quality of a person’s life. For example, the ability to read docu- ments and newspapers can help oppressed people demand their rights; it can be especially important to women. Empirical investigations by Murthi, Guio, and Dre` ze (1995) indicate that female literacy is a crucial variable in empowering women in the family and lowering birth rates. Thus even child labor that is not immediately harmful can be very harmful in terms of a child’s future well-being and agency interests as an adult. Second, significant third-party harms can result from child labor, even in those cases in which it is not directly abusive to the child. Child labor can lead to an illiterate and minimally productive workforce, reduce adult wages, under- mine health, and lead to a passive and ignorant citizenry. These two types of harm—to the child’s future interests as an adult and to society as a whole—are costs that parents may not take into account in making 10. Millions of children are beaten, raped, harassed, and abused, suggesting that more than economic motivations are driving employers (often the children’s parents). Indeed, children’s lives might be much better if only the bloodless impersonal economic motives of an ideal market were at issue. Satz 305 their decisions about how to allocate their children’s time. The discrepancy between parents and children’s short-term interests and children’s and society’s long-term interests suggests two main routes for intervention. First, where child labor reflects the weak agency of children or their parents, action could be to taken to try to increase both parties’ agency. This could be accomplished by providing more information to parents about the costs of child labor and the benefits of education, strengthening the intrafamily decisionmaking process to bolster the mother–child axis, or requiring that parents sign enforceable agree- ments with their children’s employers about the terms of work. Second, interventions could aim at changing the external context of family decisionmaking. A widely cited example of a promising intervention is Mexico’s Programa de Educacio ´n, Salud y Alimentacio´n, which provides cash transfers to mothers whose children attend school. Other strategies include strengthening the education system, restricting children’s work days to a limited number of hours so that they can attend school at least part-time, encouraging measures (training, organizing) to raise adult wages, and providing credit to poor families (see Grootaert and Kanbur 1995 for additional suggestions). It is worth reflecting on the environment in which much child labor thrives: poverty, weak states, poor education systems, ethnic conflicts, massive inequal- ities, lack of democratic institutions. How much of South Asia, which has the highest absolute numbers of working children, has functioning labor markets? How much of the economy is characterized by bonded labor, serfdom, debt peonage, and slavery? Even if one grants that in some circumstances children must work, there is no doubt that children are vastly worse off than they would be if laws created and enforced genuinely free markets, including the right to exit from employment and restrictions on monopoly and monopsony, with perhaps the state stepping in as a source of credit to poor families. Developing and strengthening democratic political and economic institutions is likely to be an essential component in the process of ending child labor. In the absence of broad changes in policy and commitment, different interven- tions will lead to different tradeoffs between values. For example, imposing a uniform and egalitarian educational system in a country may discriminate against children who are at greatest social and economic disadvantage. Some families may simply not be able to afford to send their children to school full-time. But allowing some children to attend school part-time undermines a commitment to educational equity and perhaps perpetuates caste and geographic inequalities. Tolerating child labor in some countries will give rise to worries about unfair competition in the international context. In considering various policy tools, it is thus extremely important to be explicit about which values are being favored. IV. CONCLUSIONS This article endorses a position between the absolutist universalists, who want to immediately abolish all child labor, and the contextualists, who seek to temporarily 306 THE WORLD BANK ECONOMIC REVIEW, VOL. 17, NO. 2 accommodate it.11 Tradeoffs among different values are inevitable, but there is a need to draw some bottom lines. Child labor that is abusive to children— prostitution, bondage, slavery, and the use of children as soldiers—threatens the core of their lives and should not be tolerated under any circumstances. But tradeoffs between different values above this line need to be weighed in working to eliminate other forms of child labor that score high along one or more of the normative dimensions. Although different people, organizations, families, and states will draw those tradeoffs in different ways, it is important to keep the focus on what different policies do to individual children, not to aggregates. Limits should be placed on the costs that policies impose on children in the name of future familial or societal benefits. Contextualism should be guided and regulated by the universalist standards we are trying to realize. In this sense, the normative perspective proposed here is broadly humanitar- ian, which gives priority to the securing of a decent minimum level for all children. Insofar as liberal democratic institutions are instrumental to that humanitarian goal, however, promoting them must be part of overall strategies for addressing child labor. Indeed, gradualist approaches to ending child labor are much more likely to succeed in the context of accountable political entities. The poor are undoubtedly better off where governments do not devote them- selves to theft or ethnically based spoils systems but to providing health clinics, primary schools, roads, and communications. Diminishing certain kinds of social inequality may itself lead to better outcomes for the least advantaged. Although the state of the world may justify the use of gradualist measures, we need to be attentive to the trajectory of societies using child labor. It makes a great deal of difference whether child labor is a transitional strategy that can deliver future benefits to the child or a strategy of exploitation, propping up the profits of multinational corporations, selfish parents, corrupt governments or satisfying the whims of sadistic employers. It is thus crucial to establish benchmarks for progress in educating children. These benchmarks can foster accountability and allow tracking of what is actually happening over time to children’s interests. If children’s interests are to be realized, it is essential that obligations fall where power is exercised. NGOs and lending institutions need to hold the parties they work with—parents, local villages, corporations, national governments—accountable for what happens to children.12 More data and empirical research are needed to identify which gradualist policies should be favored in which contexts. For example, although the claim is sometimes made that children benefit from child labor under some 11. Contextualists should be distinguished from relativists, who deny universal standards as such. Contextualists recognize the pull of such standards but also recognize that it may not be possible to implement them given current conditions. 12. The international lending institutions should not repeat the policies of the past, in which corrupt dictators like Mobutu Sese Seko were repeatedly given new loans for development that did nothing to improve the lives of Zaire’s people (see Easterly 2001). Satz 307 circumstances, insufficient attention has been paid in the empirical literature to the question of whether the child who is working is the child who benefits. More data are needed on intrahousehold tradeoffs between children and between adults and children. It makes a great deal of difference whether all the children in a family work a little but all go to school or whether daughters are pulled out of school so that sons need not work. It is therefore important to continue to gather data on lower levels of analysis to assess the relevance of gender and other factors. Collecting these data could help policymakers formulate effective interven- tions. They could reveal, for example, that the focus should be on informing parents and teachers about the importance of educating girls or that lending agencies should make some of their loans conditional on achieving gender equity in education. Too much of the data are underinclusive. Very few studies provide data on girls working at home who do not attend school. Indeed, the ILO does not include such girls in its statistics on child labor. This limitation on who counts as a working child may be behind the category of nowhere children, children who are neither at work nor at school. Although it may be extremely difficult to obtain survey data on girls working at home, those data are important for assessing the effectiveness and the normative adequacy of different policies. Attention also needs to be paid to children who combine work and school. Subsidy programs may draw children into school without reducing the family’s need for the child’s labor. Kabeer (2001) has noted the implications of this ‘‘double burden’’ for children’s achievements and well-being. Studying this group of children is especially important insofar as gradualist strategies for combating child labor are adopted. Good empirical projects are needed to investigate how and why some states and governments have made substantial progress in educating their children. Poor countries do differ in what they provide to their children. Within India, for example, states with similar levels of poverty have dramatically different levels of educational performance. In Uttar Pradesh only 32 percent of rural 12- to 14- year-old girls have ever attended school—about a third as many as in Kerala, where 98 percent of girls this age have attended school (Kabeer 2001). What factors explain this difference in outcomes?13 Child labor was once prevalent in what is now the industrial world. Eliminating it in poor societies may not be feasible on the basis of the resources and institutions of those societies. But a key difference between historical and 13. Economic analyses of child labor tend to treat the marginal productivity of a child as a property of the child, given a fixed technology of household production. There are two problems with this assumption. First, household technology is not fixed: Parents affect children’s marginal productivity. Parents could, for example, assign household duties in different ways—by challenging gender norms and giving more productive jobs to girls, for example. Second, children affect their own price: They make norm-supported choices concerning their economic commitments to aging parents. Children are not analogous to other economic products; their benefits and costs are not only exogenously determined. Zelizer (1995) argues that a variety of cultural forces rather than changes in the structure of the labor market changed the view of children in the United States during the 19th century. 308 THE WORLD BANK ECONOMIC REVIEW, VOL. 17, NO. 2 contemporary cases of child labor is that today the industrial world exists. Increasing development aid, ending protectionist policies that close off markets to poor countries, encouraging multinationals to pay higher wages to adult workers, facilitating partnerships in the research and development of products needed by the poor (vaccines, drugs), empowering democratic institutions around the world, and transferring technology may all make a difference. The need for a well-funded global initiative on basic schooling, recently stressed by the United Nations, is also clear. Child labor may be understandable in parts of the world as a response to poverty. But different distributions of wealth and power would undercut the need for child labor. Much depends on whether these alternative distributions can be realized. REFERENCES Agarwal, Bina. 1995. A Field of One’s Own: Gender and Land Rights in South Asia. Cambridge: Cambridge University Press. Barry, Brian. 2001. Culture and Equality. Cambridge, Mass.: Harvard University Press. Basu, Kaushik. 1999. ‘‘Child Labour: Cause, Consequence and Cure, with Remarks on International Labor Standards.’’ Journal of Economic Literature 37(September):1083–119. Bhatty, K. 1998. ‘‘Educational Deprivation in India: A Survey of Field Investigations.’’ Economic and Political Weekly 4 and 18 July:1731–40. Brown, D., A. Deardorff, and R Stern. 2003. ‘‘Child Labor: Theory, Evidence, and Policy.’’ In K. Basu, H. Horn, L. Roman, and J. Shapiro, eds., International Labour Standards—Issues, Theories and Policy. Oxford: Basil Blackwell. Burra, N. 1995. Born to Work: Child Labor in India. New Delhi: Oxford University Press. ` ze, J. and H. Gazdar. 1996. ‘‘Uttar Pradesh: The Burden of Inertia.’’ In J. Dre Dre ` ze and A. Sen, eds., Indian Development: Selected Regional Perspectives. New Delhi: Oxford University Press. Easterly, William. 2001. The Elusive Quest for Growth. Cambridge, Mass.: MIT Press. Fallon, P. and Z. Tzannatos. 1998. Child Labor: Issues and Directions for the World Bank. Washington D.C.: World Bank. Grootaert, C. and Ravi Kanbur. 1995. ‘‘Child Labor: An Economic Perspective.’’ International Labour Review 134(2):187–203. Hadadd, L., J. Hoddinott, and H. Alderman, eds. 1997. Intrahousehold Resource Allocation in Developing Countries: Models, Methods and Policy. Baltimore, Md.: Johns Hopkins University Press. Humphries, J. 1999. ‘‘Cliometrics, Child Labor and the Industrial Revolution.’’ Critical Review 13(3–4):269–83. ILO (International Labour Organization). 2002. Every Child Counts: New Global Estimates on Child Labour. Geneva. Jejeebhoy, S. J. 1992. ‘‘Family Size, Outcomes for Children and Gender Disparities: The Case of Rural Maharashtra.’’ In C. B. Lloyd, ed., Fertility, Family Size, and Structure: Consequences for Families and Children. New York: Population Council. Kabeer, Naila. 2001. ‘‘Deprivation, Discrimination and Delivery: Competing Explanations for Child Labour and Educational Failure in South Asia.’’ Working Paper 135. Institute of Development Studies, University of Sussex. Kahneman, D., J. Knetch, and R. Thaler. 1987. ‘‘Fairness and the Assumptions of Economics.’’ In Robin M. Hograth and Melvin Reder, eds., Rational Choice: The Contrasts between Economics and Psychology. Chicago: University of Chicago Press. Satz 309 Kanbur, Ravi. Forthcoming. ‘‘On Noxious Markets.’’ In P. Pattanaik, and S. Cullenberg, ed., Globalization, Culture and the Limits of the Market: New Essays in Economics and Philosophy. New Delhi: Oxford University Press. ` ze. 1995. ‘‘Mortality, Fertility and Gender Bias in India: A District Level Murthi, M., A. Guio, and J. Dre Analysis.’’ Population and Development Review 21(4):745–82. Satz, Debra. Forthcoming. ‘‘Noxious Markets: Why Some Things Should Not Be for Sale.’’ In P. Pattanaik and S. Cullenberg, eds., Globalization, Culture and the Limits of the Market: New Essays in Economics and Philosophy. New Delhi: Oxford University Press. Schapiro, Tamar. 1999. ‘‘What Is a Child?’’ Ethics 109(4):715–38. Sen, Amartya. 1985. ‘‘Well-Being, Agency and Freedom: The Dewey Lectures 1984.’’ Journal of Philosophy 82(4):169–221. ———. 1999. Development as Freedom. New York: Knopf. Weiner, Myron. 1991. The Child and the State in India. Princeton, N.J.: Princeton University Press. Zelizer, Vivianna. 1985. Pricing the Priceless Child: The Changing Social Value of Children. New York: Basic Books.