Despite the attraction of lifetime income, it is difficult to measure and whatever measure is employed rests on strong assumptions. An alternative approach is to employ a cohort analysis. Gale, Houser, and Scholz (1997), for example, consider the impact of tax changes on married families in the age range of 40 to 50. By restricting the analysis to households who are likely to be at the same stage of their earnings profile, they avoid mixing people from different stages of the lifecycle. The approach is conceptually appealing and does reduce the measurement problem described above. It does not, however, address the problem of transitory income shocks.
Households with a one time negative income shock may maintain previous consumption levels under the assumption that the poor income realization is a temporary setback that is likely to be offset by a positive income shocks in the future. Hence, consumption to income ratios will be high for this group and any tax that approximates a consumption tax in its effect will look more regressive than it would if transitory income shocks were taken into account. Taking a multi-year window, as for example the Joint Committee on Taxation (1993) has in a number of studies addresses this problem to some extent. Unfortunately, our dataset does not provide multiple observations on the same household’s income and consumption patterns with which we could smooth our income measure. Below, we will report a cohort distribution of taxes as an alternative measure to our lifetime income measure.
I distribute taxes using conventional assumptions about incidence derived from previous economic incidence studies. Individual taxes on wages and factor payments are assumed to be borne by the individual. Corporate taxes are assumed to be borne by owners of capital and are distributed to households in my data set using a methodology developed by Feldstein (1988).