Inside the Household

Transcription

1 Inside the Household Spring 2016

2 Inside the Household

3 Outline for Today I model II Evidence on : Lundberg, Pollak and Wales III Evidence on : Duflo IV Cooperative models V Noncooperative models VI Evidence on Cooperative: Udry

4 Decision Making In general, agents will differ in their views on how household income should be spent Three broad classes of decision making processes: i ii Noncooperative iii Cooperative

5 Model The family is a remarkable institution. And a complex one. Indeed, so complex that much of economic theory proceeds as if no such thing exists Sen, A. (1984) Resources, Values and Developments

6 Model Standard assumption = household can be modelled as if an individual Implies that interactions between individuals in the household are not relevant How might this be justified? i Consensus model (Samuelson) - agree to maximise a consensus social welfare function ii Alturist model (Becker) - preferences of a benevolent parent become the preferences of the household; household maximand becomes the parent s utility function

7 Rotten Kid Theorem (Becker, 1974 & 1981)

8 Rotten Kid Theorem (Becker, 1974 & 1981) Family consists of a group of purely selfish but rational kids" and one altruistic parent whose utility function reflects a concern for the wellbeing of the other family members Altruistic parent desires to make gifts/transfers to the kids Under his assumptions, sufficient to induce the kids to act in an unselfish way altruistic parent adjusts transfers so that each rotten kid" finds it in their interest to choose actions to maximise family income Resulting income levels and distribution maximises altruist s utility function subject to the family resource constraint

9 Rotten Kid Theorem (Becker, 1974 & 1981) Note: Depends implicitly on the (strong!) assumption of transferable utility Theoretical arguments against it include: King Lear and the importance of having the last word (Hirshleifer, 1977); the Samaritan s Dilemma (Bernheim and Stark, 1988); the controversial night-light (Bergstrom, 1989)

10 Public Transfers & Income Pooling model implies that household decisions are not affected by the identity of who receives public transfers/additional income > targeting particular household members unnecessary However, policy makers seem to think differently... When the British child allowance system was changed in the mid-1970s to make child benefits payable in cash to the mother, it was widely regarded as a redistribution of family income from men to women and was expected to be popular with women. Lundberg and Pollak

11 The Evidence A theory is vindicable if its consequences are empirically valid to a useful degree of approximation; the empirical unrealism of the theory itself, or its assumptions, is quite irrelevant to its validity and worth... [Yet] if the abstract models contain empirical falsities, we must jettison the models, not gloss over their inadequacies Samuelson (1963)

12 Income Pooling model implies income pooling, i.e. who controls what portion of the income should have no impact on household demands. Only total family income matters. Potential test: does giving women control over a larger portion of the household budget change consumption behaviour? Idea: compare the demand behaviour of families in which women contribute a large share of (earned/unearned) total income to families where they do not

13 Income Pooling: Econometric Issues Potential endogeneity of measures of control over income in the household How much women earn and work is likely determined jointly with household expenditures, and correlated with differences in preferences and wage rates E.g. Phipps and Burton: expenditures on eating out are more elastic with respect to the wife s than husband s earnings. She gains more bargaining power and... women like eating out more (?) Restaurant expenditures depend on the cost of substitutes, and the wife s wage is an important component of the cost of home prepared meals

14 Income Pooling: Lundberg, Pollak and Wales In the late 1970s, the UK changed the form of its universal child benefit scheme, shifting receipt of the transfer from fathers (reduction in the amount withheld for taxes from paycheck) to mothers (cash payment) in two-parent families A natural experiment that provides an exogenous source of variation in the control of resources within the family (and rather large 8% av. male earnings) Compare the ratio of children and women s clothing expenditures to men s before and after the policy change Identification strategy: no time trends in relevant outcome variables.

15 Income Pooling: Lundberg, Pollak & Wales

16 Income Pooling: Duflo Might worry that time trends biasing LPW results Duflo (2003) looks at the impact of the introduction of pensions for Black South Africans in 1993, which represented a substantial unexpected transfer of income for men older than 65 and women older than 60 Many children live with their grandparents, although those that do relatively disadvantaged on average Consider the impact on child weight-for-height (fast reacting) and height-for-age (persistent)

17 Income Pooling: Duflo Step 1: Compare outcomes of children in eligible and ineligible households Compare weight-for-height in households with no member eligible for the pension, those in households with an eligible man, and those with an eligible woman (normalising differences by differences in the probability to receive pension) Also control for the presence of older ineligible household members and for a further set of rich demographic variables to control for differences in background Identification strategy: no systematic differences in nutrition between eligible and ineligible households with an elderly member

18 Income Pooling: Duflo Pensions received by women increase the weight-for-height of girls by 1.19 standard deviations but did not significantly increase that of boys. Pensions received by men not associated with an improvement in the nutritional status of girls nor boys

19 Income Pooling: Duflo Step 2: But what if there are still intrinsic differences between eligible and ineligible households? Would hope to use a difference-in-differences identification strategy but no time dimension to exploit (no representative surveys of African households before the end of apartheid) However, height-for-age reflects past nutritional history effectively, can create a panel dataset from the cross-section

20 Income Pooling: Duflo If households eligible for pensions have worse characteristics than ineligible households, older children would be smaller in eligible households. If the pension program leads to better nutrition, children measured when younger will have been better nourished for a larger fraction of their lives The younger children are, the smaller their relative disadvantage should be in eligible households Strategy: compare the difference in height between children in eligible and those in ineligible households among children exposed to the program for a fraction of their lives to the same difference amoung children exposed all their lives

21 Income Pooling: Duflo

22 Collective models Recognise the separateness of persons within the households how do individual preferences lead to collective choice? Cooperative models: household decisions are always efficient Noncooperative models: cannot enter into binding and enforceable contracts with each other. Instead, actions conditional on the actions of others.

23 Cooperative models

24 Cooperative models Can model household decision making problem as: max q A + µ(z)q B q A,q B,Q s.t. p (q A + q B ) + P Q = y A + y B

25 Cooperative models

26 Cooperative models Note that this doesn t imply the absence of conflict, it imposes that the solution leave no money left on the table Different models supply different threat points or leave the bargaining model implicit Divorce threat point Noncooperative solution Cooperation seems natural - long term relationships with good information. Folk theorem arguments that with sufficient patience, Pareto efficient outcomes can be supported as a subgame perfect Nash equilibrium of a repeated game

27 Noncooperative models - Example: Separate Spheres Lundberg & Pollak (1993): each spouse makes decisions within his or her own sphere" and responds to the other s decisions by altering the level of voluntary contribution to shared, public goods Socially recognised and sanctioned gender roles assign primary responsibility for certain activities to husbands and other activities to wives Wife treats level of public good provided by husband as fixed, and chooses quantities of her private good and public good subject to her own budget constraint. Same reasoning for husband. Decisions lead to a pair of best response functions that determine a Cournot-Nash equilibrium un which public good contributions are inefficiently low

28 Pareto Efficiency: Udry model cannot capture household decision making. What about the evidence in favour of cooperative versus noncooperative models? Key difference: cooperative models lead to efficient decisions Enrich model to introduce household production and the notion that the household needs to maximise the size of the pie Pareto efficiency in this context household first maximises its total income and then shares this according to the set of bargaining weights that characterise the intrahousehold decision making process

29 Pareto Efficiency: Udry Agricultural, often polygamous households in Burkina Faso (1.8 wives per man) Men and women control different plots of land Can test if the allocation of resources across different plots of land is efficient

30 Udry: Model Household production of good k Y k = i P k G(N i F, Ni M, Ai ) Household budget constraint p C = p Y

31 Udry: Model Pareto efficient allocation of resources solves max U F ( ) + λu M ( ) C j,nj i,pk subject to the budget and resource constraints Maximisation implies that the allocation of labour across plots maximises total household output max G(N NF i F i, Ni M, Ai ),Ni M i P k

32 Udry: Model If a crop is planted both on plots controlled by men, and on plots controlled by women then must have that men and women have access to the same technology for producing that crop Should apply labour on each plot until the marginal product of labour is equalised across plots Thus, variations across plots in output and factor inputs should be functions only of variation in plot characteristics, and so yield (production divided by size) of each plot should be independent of the owner of the plot Test: for a given year, household, and crop, is the yield a function of the gender of the person who owns the plot?

33 Udry: Empirics Yield (output per area) Q k (A i ) = G(Ni F, Ni M, Ai ) A A Taylor approximation (way of approximating a function) Q k (A i k Qk (Ā) ) Q (Ā) + A (Ai Ā) Leads to fixed effects estimation - estimate the deivation of plot yield from mean yield as a function of the deviation of plot characteristics from mean plot characteristics within a group of plots planted with the dame crop by members of the same household in a given cropping season

34 Udry: Empirics Generalise to control for other plot characteristics Q htci = X htci β + γg htci + λ htc + e htci X htci = characteristics of plot i planted with crop c at time t by member of household h λ htc is the household-year-crop fixed effect restrict attention to the variation in yields across plots planted with the same crop within a single household in a given year

35 Udry: Results Finds that the household is not efficient could achieve an increase of 5.8% of production by reallocating inputs across plots Marginal product of land controlled by women is less than that of similar land controlled by her husband Could produce more if either reallocated variable factors from men to women or if all the land was given to the husband who then compensated the woman with a promised transfer of income...

36 Summary model, whilst elegant and tractable, cannot be reconciled with evidence on household decision making Collective models, especially cooperative models, have become the norm in recent years However, unclear that the assumption of Pareto efficiency is justified.