In the standard case in which the interest rate is assumed to be greater than the rate of population growth, implementation of a social security program leads to a reduction in capital formation and a loss of welfare of the representative individual. This dissertation asks whether the parameters of a stylized social security program can be manipulated to reduce this welfare loss. By attaching weights to the earnings used in computing the average monthly earnings, an instrument is created which the social security administrator can use to manipulate the net marginal tax rates and the relative cost of leisure between years. If, as a result, aggregate savings increase, then steady-state welfare may also increase. The effect of changing the weights in the benefit formula is considered first in a simple three-period partial equilibrium model. Individuals work for two periods and are retired in the third. It is shown, under assumptions of separability, that first-period labor supply must go up and second-period labor supply must go down in response to an increase in the earnings weight attached to the first period. Furthermore, although there is an element of ambiguity, a strong case can be made that aggregate savings must increase. It is also shown that, contrary to intuition, a zero net tax is not neutral and in fact must lead to a reduction in capital formation and welfare. These same issues are then considered in a many-period model in which interest rates and wage rates are allowed to respond to changes in aggregate savings. It is found that alternatives to the current program that provide more weight to earnings of younger workers can reduce the welfare loss by a small amount. Because of the intractability of the many-periods case a computer simulation is used to perform the analysis. In addition, the adjustment costs of a public savings program are considered. (Feldstein, among others, has suggested that social security be used as a vehicle for a public savings program to increase private investment in the economy.) It is shown that while such a program would adversely affect that welfare of a number of generations, these welfare losses are quite small: less than 0.05% for all the cases considered. / Ph. D.