Get started today!

Applying Is As Easy As 1-2-3!



While these results are encouraging for proponents of equity investment by the Social Security system, we note two caveats. First, lower Social Security tax rates reduce welfare by 0.6% of consumption in a model in which investors are extremely impatient, with a low time discount factor of 0.8, but the government judges their welfare using a higher time discount factor of 0.96. This calculation is a crude way to capture factors that might lead to inadequate private saving and justify a mandatory retirement saving system. In this model, however, there is a substantial gain of 2.0% from Social Security equity investment with fixed tax rates. These findings suggest that the appropriate adjustment of tax rates will depend on detailed assumptions about the behavior of households, but that under a wide range of assumptions there are welfare gains to be had by investing some retirement wealth into equities.
Second, a system with partial investment of retirement wealth in the equity market has greater variability of outcomes across cohorts. Particularly negative outcomes for some cohorts might provoke pressure for political bailouts, and the anticipation of bailouts might change the consumption behavior modelled here. This is an important issue for research on Social Security reform.
Using data from the PSID, we study heterogeneity in labor income processes across households. We find that some households—particularly self-employed college graduates—are exposed to much greater volatility in their labor income than are typical households. The labor income of these households also tends to be more highly correlated with returns on stocks and long-term bonds. This heterogeneity affects optimal investment strategies and may help to justify Social Security reform that includes an element of personal choice.
We also consider the possibility that investment of retirement wealth in equities might reduce the equity premium. We do not build a general equilibrium model to study this issue, but we do compare results under alternative assumptions about the equity premium. We find that it matters greatly whether the equity premium falls through a decline in the expected return on stocks or through a rise in the riskless interest rate. In the former case the welfare gains from investing retirement wealth in equities are reduced, while in the latter case they are actually greater than in our benchmark model.
In evaluating our results, it is important to be aware of several respects in which our model is oversimplified. First, we consider only self-financing retirement systems in which there is no net payment from any household to any other. Thus we ignore the redistributive features of the present Social Security system, and we have nothing to say about the overhang of liabilities to previous generations implied by the present system. Second, we assume that asset returns are independently and identically distributed. Thus we ignore the variation in real interest rates and equity premia that is the subject of much recent research. Third, we abstract from the existence of owner-occupied housing. This is an important omission since housing is the main component of wealth for many people. Cocco (1998) takes a first step towards the realistic incorporation of housing into a life-cycle model. Fourth, we assume that labor income shocks have constant variances. Some researchers have argued that the variance of idiosyncratic shocks to labor income is higher when the economy is weak and risky asset returns are low; this can have important effects on the demand for risky assets as shown by Mankiw (1986), Constantinides and Duffie (1996), and Storesletten, Telmer, and Yaron (1998a). Finally, we assume that labor income and the retirement age are exogenous to the household. Bodie, Merton, and Samuelson (1991) point out that households with flexible labor supply can afford to hold riskier portfolios because they can adjust to negative asset returns both by changing their consumption and by changing their labor supply. An important task for future research is to incorporate these and other realistic complications into the basic life-cycle model of portfolio choice.