5. CAPÍTULO V CONCLUSIONES Y RECOMENDACIONES
5.2. Recomendaciones
A detailed discussion of the different models of various models of
intrahousehold resource allocation is given in Appendix I. The most simplistic and restrictive model of intrahousehold resource allocation is the common preference model. Researchers that measure only aggregate consumption or aggregate income in the household, or those who interview only a single respondent for household
demand, implicitly use this model.
This model assumes that either all household members have homothetic preferences or one of the household members (generally the head of the household) is a dictator responsible for all decision-making within the household. The common
preference models enables us to conduct economic analyses without taking into account the intricacies of intrahousehold resource allocation.
Modern theoretical advances and empirical findings in the field of household economics have questioned the findings of the common preferences model. The “Rotten Kid Theory” put forth by Gary Becker (1974, 1981) redefined the way economists examined household decision-making. He conceptualized an altruistic parent (or husband) who manipulated transfers for his rational but egotistic “rotten” kids (or wife), such that each member in the household behaved to maximize household income. Thus, household choice was essentially a collective choice. Although widely criticized by economists and feminists (Lindbeck and Weibull , 1988, Bergstorm, 1989, Bruce and Waldman, 1990), this model provided a real alternative to the common preference theory and a starting point to the study of household economics. McElroy and Horney (1981), and Manser and Brown (1980) defined the concept of bargaining in which husbands and wives were assumed to be independent utility maximizers indulged in a game of Nash Bargaining. These models assumed that individuals will stay in a marriage only if they have certain benefits to enjoy that outweigh the costs. Whether an individual stays in a relationship depends upon her/his reserve utility, defined as a “threat point,” that she/he would get if she/he were single. If an individual does not derive benefits with respect to this reserve utility that she/he gets from the marriage, the marriage may dissolve and result in a divorce. Further modifications of the Nash Bargaining model investigate a more appropriate threat point than divorce (Bergstorm, 1996) as each argument in a marriage does not result in a divorce. Defining an appropriate threat point is often
problematic, and may require data that is difficult to get. Chippori (1988a), (1991) put forth a collective bargaining model that depends only on a pareto efficient outcome rather than Nash equilibrium to generate testable restrictions on labor supply
functions and recover individual preferences and outcomes, thus reducing the need to define threat points and thus reduces data requirements. Non cooperating bargaining models (Chen and Wooley, 2001) focus on non-cooperative bargaining framework such as Cournot-Nash bargaining, where each person maximizes his or her welfare, given the expected actions of others and his or her own resource constraints.
While the common preference model is the most restrictive, the non-
cooperative bargaining model is the least restrictive model. The key in differentiating between models lies in looking at what people do with their income—whether they pool it into a single “pot” or keep it for themselves. Since the 1990’s “resource pooling has emerged as the crucial empirical issue” (Pollak, 2003). If individual labor income rather than total household income affects expenditure patterns, the common preference model can be rejected.
The collective model of bargaining is a generic model with Becker’s unified model and the co-operative bargaining model as restricted cases of this model. Becker’s unified model also involves pooling of household resources. However, merely testing whether individual labor income affects expenditure patterns is not enough. Among the early works on pooling that rejected Becker’s unified model include those by Thomas (1990). Thomas (1990) showed that children have better mortality and morbidity if their mother controls a larger fraction of the family’s resources. Thomas contended that labor income is endogenous as it reflects labor
supply decisions. Hence, he examined non-labor income, which he assumed to be exogenous. Thus, determining whether expenditure patterns are dependent upon non- labor income could provide us information to reject Becker’s model of a unified household.
However, non-labor income could be the result of previous labor related decisions, and hence not really exogenous. In Thomas’s case, one can also provide a counter-argument of unobserved heterogeneity, that “better mothers” control more resources in the household, and also raise children in a better manner (Pollak, 2003). Situations or shocks involving exogenous changes in income and expenditure are needed to test for income pooling.
Such a shock was provided in the late 1970’s when Margaret Thatcher restructured the child-benefit program in UK from a tax deduction to a direct cash installment to mothers (at the post office). It indirectly transferred income from primary wage earners (typically the men) to their wives. Lundberg et al. (1997)
studied this natural experiment, and found a significant increase in children’s clothing expenditure relative to men’s clothing expenditure and also a significant increase in women’s clothing expenditure relative to men’s clothing expenditure. Lundberg et al., (1997) provide perhaps the most compelling argument that husbands and wives do not pool their income, and hence the unified model of intrahousehold resource allocation is not appropriate. Thus, the unified model can be rejected if expenditure patterns are dependent on non-labor income, and there is no strong argument for any unobserved heterogeneity that affects outcomes.
Measuring non-labor income is often difficult. However, even if it is measured with error and income is pooled, the ratio of their marginal utility of income for the good should be the same for the common preference model and the Beckerian unified model, i.e., equality of the ratios cannot be rejected (Thomas, 1990).
Further tests are based on inspecting for pareto efficiency (Doss, 1996), i.e., whether there exists a constant ratio of income effects across all goods (as shown in the equation 2.15). The tests of intrahousehold resource allocation, summarized by Doss (1996), are reported in table 2.1.
Table 2.1: Tests of intrahousehold resource allocation models
Bargaining models
Null hypothesis Common
preferences model
Unified
model Collective model
Cooperative Non- cooperative Individual labor income does
not affect expenditures R C C C C
Individual exogenous income does not affect
expenditures/ does not affect labor supply
R R C C C
Pareto efficiency: constant ratio of income effects across different goods/ constant ratio of marginal productivity of inputs.
R R R R C
(The table is reproduced from Doss(1996). Cells indicate whether the rejection of null hypothesis is consistent with or implies a rejection of the model).
R: Reject, C: Consistent