What Do Data on Millions of U.S. Workers Say About Life Cycle Income Risk?

Abstract

This paper sheds new light on individual labor income risk using a unique and confidential dataset from the Social Security Administration on individuals' earnings histories. The substantial sample size allows us to cut the data in different and novel ways and document how earnings risk varies over the lifecycle and across individuals that differ in their lifetime income. The main conclusion of our research is that earnings risk varies significantly across the population in the following ways. First, the overall size of income risk becomes smaller with age, from age 25 to 50 and then increases again. Second, as individuals age, the likelihood of getting very small and very large shocks increases relative to the likelihood of middling shocks. Third, income shocks become more left skewed with age, meaning that, relative to the average change in income, a large fall becomes more likely than a large rise as individuals get older. Finally, earnings growth rates are dramatically different for individuals ranked by their lifetime income: individuals with lifetime earnings in the top 5% experience a growth rate from age 25 to 55 that is 10 times larger than individuals with average lifetime earnings. To provide useful input to policy relevant research, we estimate an econometric process that captures these salient features of earnings dynamics to provide a reliable “user's guide” for applied economists.