The story of income growth over a lifetime

Income growth, as we usually think about it, over the course of one year is an incomplete picture. Income growth, it turns out, varies quite a bit over the course of workers’ lives. Pay boosts that might be just adequate at one point in life could be a massive improvement at another point. So how exactly does income growth vary over the lives of workers? A new working paper sheds light on this issue in a variety of ways.

The new paper, by economists Fatih Guvenen of the University of Minnesota, Fatih Karahan of the Federal Reserve Bank of New York, Serdar Ozkan of the University of Toronto, and Jae Song of the U.S. Social Security Administration, looks at the income growth of male workers over the course of their working years. More specifically, they look at that growth from the age of 25 to 60. And the data they use is quite large. Using data from the Social Security Administration, their data covers10 percent of all male workers in the United States in every year from 1978 to 2010.

Guvenen, Karahan, Ozkan, and Song document a large number of facts about income growth over the bulk of these men’sworkers’ working lives. They find that a large chunk of income growth—pretty much all of it—happens in the first decade of workers’ careers. But this income growth varies substantially up and down the income ladder, as Danielle Paquette of The Washington Post points out in her write-up of the paper. A worker at the median of the income spectrum will have a growth rate of about 38 percent, while a worker at the 99^th percentile will see his income grow by 1,450 percent.

The three economists also look at shocks to income growth in the first decade of workers’ careers. Take, for example, the variance (or range) of growth. The authors find that variance decreases as a worker moves up the income ladder. Income growth moves within a smaller band as a worker earns more. But once a worker starts earning over the 90^th percentile, this relationship changes: Variance increases as incomes get close to the very top of the ladder.

Then the authors focus on the so-called “skewness” of income growth. Skewness is a measure of the symmetry of income growth—how often it is growth positive or negative. Guvenen, Karahan, Ozkan, and Song find that income growth is negatively skewed. This means negative shocks to income growth are more likely than positive shocks. And according to their results, the likelihood of negative shocks to income increases as workers earn more and as they become older. This result fits well with the idea that income gains come earlier in workers’ careers. As workers age, it appears income growth mostly has one place to go: down.

As a consequence, the economists explain that the distribution of income growth is quite different from what is usually suspected. Instead of an anticipated smooth bell curve, the distribution has a very high peak and long tails in addition to being negatively skewed. This means most income shocks are small and very close to zero. But there are a few workers who see very large positive and negative shocks to their incomes.

What’s the implication of this work? By documenting the dynamics of income growth, Guvenen, Karahan, Ozkan, and Song demonstrate that the common assumption that income growth is distributed normally, like a bell curve, is misplaced. And if we want accurate economic models of the U.S. economy then our models should match the facts, regardless of how abnormal and asymmetric they are. Today, alas, most economic models do not account for the actual state of income distribution among workers over their prime working years.