Abstract
Transfers for particular client groups such as children are often in-kind rather than cash. However, this may, at least partially, crowd out private expenditures on the goods in question because they reduce the incentive for other individuals, like parents, to make altruistic transfers. They are often made to one household member on behalf of another so there may also be agency concerns. This paper uses three nutrition programs for children in UK households to cast light on altruism and agency effects.
This is a preview of subscription content, log in via an institution to check access.
Similar content being viewed by others
Notes
See Bergstrom (1989) for a review of the “rotten kid” theorem that lies behind such behaviour, and see Jacoby (2000) which refers to the absence of altruism as generating a “flypaper” effect because the transfer sticks where it lands. Gruber and Hungerman (2007) demonstrates crowding out of charitable giving by New Deal spending during the Great Depression, while Hungerman (2009) suggests that the degree of crowding out, also in the context of transfers to communities, is affected by diversity.
Prior to 1988 Family Credit was called Family Income Supplement and Income Support was called Supplementary Benefit. In 1999 Family Credit was replaced by Working Families Tax Credit and in 2003 by Working Tax Credit and Child Tax Credit. Income Support became Job Seeker’s Allowance. While there are important administrative differences, the new benefits are essentially more generous versions of their predecessors. We use the terminology Family Credit and Income Support throughout.
We compute entitlements on the basis of recorded incomes, children, etc. using a very detailed routine based on the Institute for Fiscal Studies' TAXBEN program. See Giles and McCrae (1995).
Data prior to 1981 cannot be used because free school lunch receipt was not recorded, and data post 1992 cannot be used because there is no longer sufficient detail to compute welfare entitlements from the data available because of a local taxation reform.
One difficulty with the data is that once Family Credit entitlement is established it can then last for up to 6 months (12 months prior to the reform). Indeed, it was the practice of some schools to provide Free School Lunches for a whole school year so that those in receipt of Family Credit or Income Support, at the beginning of the school year, may have still been receiving them more than 9 months later, at the end of the year, even though they were no longer eligible on current circumstances.
Indeed, we do not require that there be no measurement error. In the difference-in-difference analysis we are, in any event, estimating an intention-to-treat so the presence of non-compliers are not problematic.
The decrease in total expenditure for the Income Support recipient group across the reform reflects compositional changes: it arises because of the increase in the representation of lone (especially never-married) parents, who have substantially lower household income, in this category.
Eligibility for Welfare Milk Tokens or Free School Lunches requires household eligibility to an associated cash transfer, and children in the relevant age ranges. The importance of the distinction between receipt and eligibility groupings is emphasized in the next section.
In the case of Day Care Milk, the question routing prevents us from observing ineligible recipients. The larger proportion of ineligible participants in the case of Welfare Milk Tokens is likely to be due to our inability to identify expectant mothers currently without pre-school aged children, who would be eligible during pregnancy.
While it might be tempting to rely on a simple difference in differences methodology based on data grouped by FC and IS eligibility pre and post reform, we are reluctant to do so here. In particular we are concerned that, even if we assume that the reform is a clean natural experiment, restrictions on preferences are required for the aggregate data to be consistent with consumer theory. In particular, incomes are changing over time, within the treatment and control groups, and only if changes in the distribution of income did not affect budget shares can we meaningfully aggregate the data into group means. This condition would imply that preferences are quasi-homothetic, a restriction that is typically rejected in micro-data. The requirement that individual behavior satisfy aggregation conditions is usually ignored entirely in difference in difference studies. Despite our reservations, when we did collapse the data into month/group cells and re-estimated equation 3 we found almost identical results to Table 5, but with larger standard errors.
We have also estimated both the difference-in-differences dropping April 1988 through March 1989 which may be considered a phase-in period for the reform. These difference-in-differences results are slightly higher but lose precision. We also estimate a structural model below. Here the structural budget share estimates are unchanged when we drop the phase-in period. In view of this, and the relatively short post-reform observation window, our preferred estimates include data for the phase-in period.
The price is the average price observed in the data for those that buy school lunches within each region. In principle this price is fixed nationally although we do find that there is a small cross section variation, especially after the mid 1980s.
Over the time period considered here the real price of school lunches increased by 10 %, while the real price of food fell by 13 % and the real wages of unskilled workers remained approximately static.
In addition, a further exclusion restriction is that benefit-year (April–March) dummies only enter into the program participation equation. This is to capture the effects of other changes in transfer programs, over and above entitlement value—such as administrative procedures. We would argue that these proxies for administrative changes should not affect the budget shares beyond their effect on participation and hence on the number of in-kind units received. For the budget share, month-of-year dummies capture seasonality and a quadratic time trend is added to capture long run changes in spending patterns.
We would expect the extent to which milk (tokens or liquid) crowds out spending on milk would be comparable with the effects of Food Stamps in US research. The Hoynes and Schanzenbach (2009) estimate is approximately 80 %.
References
Banks, J., Blundell, R. W., & Lewbel, A. (1997). Quadratic engel curves and consumer demand. Review of Economics and Statistics, 79, 527–539.
Bergstrom, T. C. (1989). A fresh look at the Rotten Kid theorem—and Other household mysteries. Journal of Political Economy, 97, 1139–1159.
Blundell, R. W., Duncan, A., & Pendakur, K. (1998). Semiparametric estimation of consumer demand. Journal of Applied Econometrics, 13, 435–461.
Blundell, R. W., Pashardes, P., & Weber, G. (1993). What do we learn about consumer demand patterns from micro data? American Economic Review, 83, 570–597.
Deaton, A. S., & Muellbauer, J. (1980). Consumer Behaviour. Cambridge: Cambridge University Press.
Giles, C., & McCrae, J. (1995). The IFS micro-simulation tax and benefit model. IFS working paper 95/19.
Gruber, J., & Hungerman, D. (2007). Faith-based charity and crowd-out in the Great Depression. Journal of Public Economics, 91, 1043–1069.
Hoynes, H. W., & Schanzenbach, D. W. (2009). Consumption responses to in-kind transfers: Evidence from the introduction of the food stamp program. American Economic Journal: Applied Economics, 1, 109–139.
Hungerman, D. (2009). Crowd-out and diversity. Journal of Public Economics, 93, 729–740.
Jacoby, H. G. (2000). Is there an intrahousehold ‘flypaper effect’? Evidence from a school feeding program. Economic Journal, 112, 196–221.
Heckman, J. J. (1979). Sample selection bias as a specification error. Econometrica, 47, 153–161.
Manan, R., Millimet, D. L., & Tchernis, R. (2012). Federal nutrition programs and childhood obesity: Inside the black box. Review of Economics of the Household, 10, 1–38.
National Food Survey Committee. (1989). Household food consumption and expenditure. London: HMSO.
Acknowledgments
We are grateful to the UK Economic and Social Research Council (grant RES-000-27-0200), the Danish Social Science Research Council (24-04-0240), the Leverhulme Trust, and the Education Research Section at Princeton University for supporting our collaboration. We are indebted to Chris Giles and Yu Zhu for their assistance with the Family Expenditure Survey data used in this analysis, and to participants in workshops at Essex, Keele and Warwick Universities for their comments. The data was provided by the UK Data Archive and is used with the permission of the Controller of Her Majesty’s Stationery Office. The data are available on request, subject to registering with the Data Archive. The usual disclaimer applies. A more detailed version of the paper is also available on request.
Author information
Authors and Affiliations
Corresponding author
Rights and permissions
About this article
Cite this article
Bingley, P., Walker, I. There’s no such thing as a free lunch: evidence of altruism and agency from household expenditure responses to child nutrition programs. Rev Econ Household 11, 371–392 (2013). https://doi.org/10.1007/s11150-013-9196-7
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11150-013-9196-7