Earnings Risks, Savings and Wealth Concentration

A criticism of earnings risk models of wealth inequality is that they do not accurately capture individual's earnings risks. I construct a stochastic process that directly determines workers earnings. I use a set of new empirical moments to match moments of earnings changes in the universe of workers in the U.S. economy. Despite its computational challenges, a stochastic process that determines earnings instead of labor productivity improves the modeling of risks as it allows me to (1) reproduce the distribution of earnings, (2) match several moments of the distribution of earnings changes, and (3) skill-dependence of earnings profiles. To study the implications of such data-guided earnings risks for wealth concentration, I develop a general equilibrium stochastic life cycle production economy with skilled and unskilled workers. I show that data-guided earnings risks, contrary to what Castaneda et al. (2003) implies, are not large enough to explain high saving rates of top earners. Therefore, wealth in the model is less concentrated than the data. I also study how changes in earnings affect the distribution of wealth over time. Consistent with the data, I allow for earnings profiles, skill premium, share of skilled workers in the population, and tax schedules to vary with time. My findings show that changes in Wealth inequality between 1989 and 2013 can almost entirely be attributed to changes in earnings dispersion. Although in both periods, wealth concentration is lower than the data, changes in the wealth share of top one percent over time matches the data.