The household production model provides a framework for the theory of the household, and most applications have focused on its implications for market and nonmarket behavior. In this paper I examine the consequences of the new home economics for welfare analysis, and in particular for the cost-of-living index. In the household production framework market are combined with time to produce These commodities, rather than the market goods, are the arguments of the household's preference ordering; the demand for and time is a derived demand, since are not desired for their own sake, but only as inputs into the production of commodities.' This paper is an analysis of the implications of the household production model for welfare evaluation, not a critique of the model. Hence, it accepts the fundamental distinction between and commodities, and assumes that commodities as well as are observable and measurable.2 The distinction between technology and tastes follows unambiguously from that between and commodities. In orthodox demand theory the household's preference ordering is defined over the goods and welfare analysis is based on those preferences. The cost-ofliving index is defined as the ratio of the minimum expenditures required to attain a particular indifference curve of this preference ordering under two price regimes. In the household production model the preference ordering over the commodity space provides a corresponding basis for welfare evaluation. One way to extend the notion of the cost-of-living index to the framework is to define it as the ratio of the minimum expenditures required to attain a par

When households have different consumption patterns, whose cost of living should actual price represent? This issue was first raised by J. L. Nicholson and S. J. Prais in 1950's. Both made essentially same point: official price indexes give each household's consumption pattern an implicit weight proportional to its total (see Nicholson, p. 540). Prais calls such plutocratic, and both Nicholson and Prais suggest alternative democratic price index which gives all households equal weight. A cost-of-living index is that measures impact of price changes on welfare of a group or population of households. To define such requires explicit or implicit concept of the welfare of a group, and hence requires interpersonal comparison and distributional judgments. Since group indexes such as Consumer Price Index play important role in our perception of inflation and formation of macro-economic policy and are used to escalate wages and Social Security benefits, they have significant effects on government decisions and economic welfare. Despite their intellectual interest and practical importance, however, until recently they have been virtually ignored by number theorists. The theory of cost-of-living (CLI) provides a generally accepted framework for measuring impact of price changes on welfare of a particular household. This paper extends CLI concept to groups and discusses which questions require group indexes and which do not. I begin by introducing some notation and terminology in context of household CLIs. A household's CLI is ratio of expenditures required to attain a particular base indifference curve in two price situations. Suppose there are n goods and S households, and denote preference ordering of rth household by R r. The base indifference curve can be identified by a goods collection, Xro, which lies on it. The function, Er(P, xr, Rr), shows minimum expenditure required to attain base indifference curve at prices P. The CLI of rth household, Ir(Pa,pb,XroRr) is ratio of minimum expenditure required to attain base indifference curve at prices pa (comparison prices) to that required at prices pb (reference prices). Except in very special cases, value of CLI depends on base indifference curve at which it is evaluated; as successively higher base indifference curves are specified, one would expect prices of luxuries to become more important relative to prices of necessities.' Hence, it is convenient to regard CLI as a function of base indifference curve rather than as a single number corresponding to a particular base. Thus, instead of offering guidance in choosing appropriate base indifference curve, theory suggests that there is no need to choose. To construct exact CLI, investigator needs to know household's preferences. Lacking this knowledge, he rnust fall back on indexes which require less information and which are upper bounds on exact index. The Laspeyres index, Jr(papbXrb) is ratio of cost of purchasing reference period consumption basket at comparison prices to its cost

Everyone recognizes substantial discrepancies between the public's rankings of hazards and those of the experts. For example, experts at the Environmental Protection Agency think that hazardous-waste sites pose mediumto-low risks to the public, while indoor air pollution poses a high risk; yet public perceptions have driven policy to focus on hazardous-waste sites rather than on indoor air quality (Stephen Breyer, 1993 pp. 19-20). beliefs should determine government policy when the public's beliefs differ from those of the experts? The problem evaporates if the public, perhaps recognizing its inability to deal with complex technical issues, entrusts risk assessment to the government and its experts. But what if the public, perhaps distrusting government and experts, is unwilling to leave risk assessment to the experts?' Paul Portney (1992 p. 131) posed a version of this Whose beliefs? question succinctly in his fable, Trouble in Happyville:

Writers from diverse intellectual traditions inside and outside the social sciences criticize neoclassical economics for neglecting Their criticisms focus on economists' analyses of labor markets and of distribution within families; Marxists and feminists are among the leading critics. The classic definitions of power come from sociology and political science and, not surprisingly, resonate more for sociologists and political scientists than for economists. Instead of discussing lofty abstract definitions, I stay near the ground and focus on a concrete application: specifically, distribution within marriage and, more generally, distribution between women and men. Economists have three alternative models of distribution within marriage: Gary Becker's altruist model, cooperative bargaining models, and noncooperative bargaining models. The altruist model remains the leader. Becker's model implies that the equilibrium distribution maximizes the utility of the altruist (the husband, father, dictator, patriarch) subject to the family's resource constraint. Becker does not describe the altruist model in game-theoretic terms, but I have argued elsewhere (Pollak, 1985) that his model can be interpreted as a twostage bargaining game in which the altruist moves first and confronts other family members with take-it-or-leave-it choices. The game-theoretic interpretation makes it clear that the crucial postulate of the model is not the altruism of Becker's altruist, but his position in the game-one is tempted to say, his power. Altruism does play a role in Becker's model: the altruism of the altruist (the assumption that his utility is an increasing function of his wife's utility or consumption) allows the model to have an equilibrium in which the wife receives more than her reservation level of utility. Within the past 15 years, Becker's altruist model has been challenged by models that explicitly view distribution within marriage as the solution to a cooperative or a noncooperative game. Cooperative bargaining models are exemplified by the divorcethreat models of Marilyn Manser and Murray Brown (1980) and of Marjorie B. McElroy and Mary J. Horney (1981) and by the model of Shelly Lundberg and myself (1993). All of these models use the Nash bargaining solution or a similar axiomatic solution concept to obtain a unique equilibrium corresponding to a point, which specifies the payoffs the players receive if they fail to reach an agreement. In divorce-threat models the threat point is the utility each spouse would receive in the event of divorce; thus, the threat point is external to the marriage. In the separate-spheres model, the threat point is internal to the marriage and, more specifically, is the equilibrium of a noncooperative game in which the quantities of household public goods are determined by voluntary contributions by the spouses. Noncooperative models of distribution within marriage, aside from Becker's altruist model, are less common than cooperative models. Ravi Kanbur and L awrence Haddad (1994) analyze intrahousehold allocation using a Rubinstein alternating-offer game. Lundberg and I (1994) discuss a repeated game in which the voluntary-contri* Department of Economics, University of Washington, Seattle, WA, 98195. This paper is based on a longer manuscript entitled Taking Power Seriously. I am grateful to Jere R. Behrman, Douglas H. Blair, Paula England, Nancy Folbre, Margaret Levi, Shelly Lundberg, Jane J. Mansbridge, Julie A. Nelson, Mark Rosenzweig, Dick Startz, and Diana Strassmann for helpful conversations and to Judith Goff for editorial assistance.

Why do parents make inter vivos transfers to their children and leave them postmortem bequests?1 Gary Becker's notions of altruism (1981, ch. 8)-by which he means that children's utilities are arguments of their parents' utility function-provides one explanation. Denoting the children's utility functions by U'(ci), where ci denotes consumption by child i, the preferences of parents with two children can be represented by a utility function of the form W[cp, U1(c1), U2(c2)] where cp is the parents' own consumption.2 In the altruistic model, parents' sole motive for intergenerational transfers is to increase their children's utility. In models, however, parents may have nonaltruistic as well as altruistic motives for transferring resources to their children. The literature on economic development emphasizes old-age support as a motive for fertility and, to a lesser degree, as a motive for providing children with capital as part of an explicit or implicit intergenerational contract. When capital is the primary focus of the analysis, as in discussions of education and earnings, it is useful to decompose inter vivos transfers into human capital formation and other inter vivos Such a decomposition can mislead in discussing intergenerational transfers, however, because it obscures the fact that the provision of capital by parents constitutes an intergenerational transfer. Laurence Kotlikoff and Avia Spivak (1981) analyze another old-age support model, one in which the family operates as an incomplete annuities market. In their model, children make regular transfers to their aging parents, and the share that each child contributes to the parents determines his or her share of the parental estate. Although the prospect of old-age support may be an important motive for intergenerational transfers (i.e., from parents to in some societies, in the United States today upstream transfers appear too small and too uncertain to make this motive credible. To explain downstream transfers in the United States today, economists have investigated models in which parents have selfish as well as selfless motives. For example, B. Douglas Bernheim, Andrei Shleifer, and Lawrence Summers propose a model in which parents use the prospect of bequests to exact from their children: we envision a testator who, though altruistic, is also affected by actions taken individually by a number of potential beneficiaries (he may, e.g., enjoy receiving attention from his children) (1985, p. 1046). Bernheim et al. assume that such actions increase parents' utility and decrease children's utilities. In their model the children's utility functions become U1(ai, ci), where ai denotes the ith child provides the parents, and the parents' utility function becomes W[cp, al, a2, Ul(al, c1),U2(a2, cA]. To measure these services, Bernheim et al. use frequency of contact (i.e., visits plus telephone calls) between parents and children. This approach expands the concept of child to include those provided by adult children who live outside the parents' household in an attempt to explain bequests and inter vivos transfers. Child services originally appeared in discussions of fertility and the allocation of resources to young children living with their parents, and thus tends to evoke the joys of young parenthood *University of Pennsylvania, Philadelphia, PA 19104, and University of Washington. I am grateful to the National Science Foundation and the National Institutes of Health for financial support, to Gary Becker, Samuel Preston, David Stapleton, and Paul Taubman for helpful comments, and to Judith Farnbach for editonal assistance. 1 Even if bequests are unplanned, as some versions of the life cycle savings model assume, inter vivos transfers must be intentional. 21 ignore the possible dependence of the children's utility on their own children's utility, etc., because it is not relevant to the issues discussed in this paper.