The aging population and the size of the welfare state: Is there a puzzle?

Razin, Sadka, and Swagel [Razin, Asaf, Efraim Sadka, and Phillip Swagel (2002) “The Aging Population and the Size of the Welfare State” The Journal of Political Economy, vol. 110, no. 4, pp. 900–918.] unveil a puzzling fact: the welfare state appears to be shrinking even as the dependency ratio rises. While they formulate an elegant political economy model to explain the coexistence of an aging population and declining transfers, the resolution of the puzzle turns out to be much simpler. Labor tax rates and per capita transfers are negatively correlated with the dependency ratio in advanced economies only because this measure includes children as well as retirees. Both labor tax rates and per capita transfers in advanced economies are, in fact, historically positively correlated with the ratio of retirees to the working-age population and negatively correlated with the ratio of children to the working-age population. Increasing the number of retirees shifts preferences toward higher taxes and transfers by increasing the fraction of the population that receives transfers. In contrast, workers with more children prefer to spend more of their lifetime income while raising dependents, so they prefer smaller public pension systems. These results suggest that fiscal leakage from workers to retirees is not required to explain the broad trends in the transfer policies of advanced economies. © 2007 Elsevier B.V. All rights reserved. Razin, Sadka, and Swagel (2002) present the surprising result that per capita social transfers and the dependency ratio are negatively correlated in a sample of 13 advanced economies. In their words, “this occurs despite the increased political clout of the dependent population implied by the aging of the population.” They take this as evidence of a negative correlation between a group's size and its per capita transfers and proceed to construct a political economy model to explain this curious finding. However, they mis-measure the retired fraction of the population by using the dependency ratio, which includes children as well as retirees. I show that when children and retirees enter the regression separately, both per capita transfers and labor tax rates are positively correlated with the retired fraction of the population. ⁎ Department of Economics, Wesleyan University, 238 Church Street, Middletown, CT 06459-0007, United States. Tel.: +1 860 685 2944; fax: +1 860 685 2301. E-mail address: [email protected]. 1 I am indebted to Justin Wolfers for the numerous excellent comments and suggestions and to Phillip Swagel for providing the data from Razin, Sadka, and Swagel (2002). 0047-2727/$ see front matter © 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.jpubeco.2007.10.002 648 C.A. Shelton / Journal of Public Economics 92 (2008) 647–651 Their model is a clever derivative of the Meltzer and Richard (1981) model to which they have added a new channel: fiscal leakage from the working to the retired. As in Meltzer–Richard, voters with varying levels of income choose a flat income tax rate the proceeds of which are then redistributed as a lump sum grant to each voter. However, in Razin, Sadka, and Swagel's model, voters live and are politically active for two periods but work only in the first period. As a result, there is a population of retirees who receive benefits but pay no taxes. The net effect on taxes from increasing the fraction of retirees in the population is the sum of two competing effects. On the one hand, increasing the fraction of the population which is retired decreases the income of the median voter relative to the mean thereby increasing the median voter's preferred tax rate. On the other hand, if a greater fraction of the population is not working, then any given tax rate yields a smaller per capita benefit which decreases the median voter's preferred tax rate. The first effect is the classic Meltzer–Richard effect. The authors describe the second effect as “fiscal leakage” from the working population to the retirees. For some parameter values, fiscal leakage dominates the Meltzer–Richard effect and therefore increasing the fraction of retirees in the population actually leads to smaller per capita transfers. They interpret their empirical work as evidence that fiscal leakage has played a major role in the tax and transfer policies of advanced economies between 1965 and 1992. This is an elegant model but, ultimately, it solves an empirical puzzle which does not exist. The negative correlation between per capita social transfers and the fraction of retirees in the population is due entirely to mis-measurement of the latter. The parameter which governs the extent of fiscal leakage is the ratio of retirees to the working age population. In their empirical work, Razin, Sadka, and Swagel measure this with the dependency ratio, defined as one minus the fraction of the population in the labor force. But the labor force excludes children as well as retirees. Were the youthful fraction of the population uncorrelated with both the retired fraction and the welfare state, this would simply be whitenoise measurement error biasing the slope coefficient toward zero. However, this is not the case: in both demographic models and practical experience, population shares of youth and retirees are highly negatively correlated. Moreover, children and retirees likely have different effects on social transfers (the former do not vote and have a lifetime of work ahead of them, the latter do vote, but no longer work), hence restricting their coefficients to be equal imposes a misspecification.