There is wide consensus among economists that over the long run, skill-biased technical change (SBTC) is the primary force behind income inequality. As technology advances, says the theory, workers with the training and skills to harness those technologies earn far more than those without advanced skills. Biotech engineers, for instance, command high salaries, while farm workers reap little. Over time, the premium for advanced skills grows, so the gap between rich and poor yawns larger.
Skill-biased technical change fuels the long-term trend toward growing inequality, but labor force dynamics during business cycles guide it.
But new research from the Minneapolis Fed suggests that this dominant trend isn’t the whole story. By itself, SBTC can’t account for all dimensions of inequality, nor does it explain the rising share of working-age men who have simply dropped out of the labor force. A comprehensive explanation, say the economists behind this work, must incorporate business cycles—recessions and expansions—as potent factors in their own right.
But even more significant, they say, is the interaction between cycles and trend. SBTC fuels the long-term trend toward growing inequality, but labor force dynamics during business cycles guide it. Neither force tells the full story, but together, cycles and trend create a wedge that separates prosperous from poor, explaining much about American inequality over the past half century.
And at the heart of the interaction: men who stop working because their skills have too little market value.
The cost of recessions
In “The Rise of U.S. Earnings Inequality: Does the Cycle Drive the Trend?” Minneapolis Fed economists Jonathan Heathcote and Fabrizio Perri, with Giovanni Violante of Princeton University, argue that when low-skilled workers lose their jobs in recessions, it often costs them not only current paychecks, but future employment. Because technological change always favors higher skills, whatever skills now-unemployed workers gained on their previous job quickly depreciate in the eyes of prospective employers.
Skills gained on the job quickly depreciate. Losing a job in a recession can thus permanently scar a worker’s ability to make a living.
If the downturn lasts long enough, low-skilled workers may stop looking for work altogether. It’s an entirely rational decision: Why look for a job when chances are slim and potential wages are low? Over time, the value of their job skills falls still further. Losing a job in a recession can thus permanently scar a worker’s ability to make a living.
It’s a “double whammy,” say the economists. Knocked down by job loss in the wave of recession, a worker might get back up and find a new job when the economy rebounds. But because there’s an undertow—technological change that rewards advanced skills—their market value declines, and they’re pulled down and out of the labor market.
“Long-term unemployment coupled with falling low-skill wages causes many low-skilled individuals to exit the labor force during recessions.”
“Long-term unemployment coupled with falling low-skill wages causes many low-skilled individuals to exit the labor force during recessions,” write Heathcote, Perri, and Violante. The interaction helps explain both the rising share of men who’ve stopped looking for work and “an important share of the rise in earnings inequality at the bottom of the income distribution.”
Data drive the theory
Their investigation begins with a close look at trends in the earnings distribution of prime-age men (25-to-55 years old) from 1967 to 2018 and at the components of those earnings: wages per hour worked and hours worked per year. Using data from the Current Population Survey, the economists confirm two key findings.
First, the data show that inequality for the top half of the earnings distribution (comparing the 90th and 50th percentiles) increases at a smooth, steady pace. In contrast, earnings inequality for the bottom half (comparing the 50th and 20th percentiles) moves upward in a jagged path, jumping sharply during recessions and declining slowly during expansions.
Second, wages and weeks worked look entirely different for workers at the top, middle, and bottom of the earnings scale. At the top and middle, almost all men work full time, so inequality is driven purely by trends in relative wages; wages at the top rose 70 percent while in the middle, they were “essentially flat” over the 52-year period.
For the bottom fifth of the earnings scale, wages have relatively little influence, falling by less than 10 percent (in real terms) over 52 years. But weeks worked declined by 60 percent. During recessions, weeks at work plummet, increasing only slowly during expansions. (See figure.)
Loading chart 1...
Over the long run, working hours for low-income workers have declined dramatically, falling from 38 hours a week in 1967 to just 17 by 2018. Further analysis reveals that this decline is explained largely by an increase in the fraction of men not working at all—a fraction that jumps sharply during recessions and never fully recovers. In 1967, nearly 90 percent of prime-age, low-income men were working. By 2018, just 43 percent worked.
A model with two levers
To examine the dynamics behind these empirical findings, the economists create a model that focuses on interaction between skill-biased technology and business cycles. A steady upward trend in the relative importance of higher-skilled workers represents SBTC in the model. Time variation in job-finding probabilities is the proxy for fluctuating unemployment rates during booms and busts.
Workers in the model have one decision: whether or not to join the labor market. It’s a decision made by weighing the potential for higher earnings against the value of leisure. Wages vary by skill level; skills increase on the job, but lose value while unemployed.
Calibrated to replicate inequality patterns at upper earnings levels, this benchmark model is able to generate patterns at lower levels as well—indicating that the two mechanisms and their interaction do explain U.S. inequality trends.
Low-skilled men stop looking for work while earnings for others rise when the economy rebounds.
The economists then run two simulations. The first sets the unemployment rate at a constant level (artificially creating an economy without business cycles); the second experiment shuts down the SBTC mechanism (so the wage gap doesn’t grow over time).
Turning off either mechanism, they find, generates a far smaller increase in the fraction of workers earning nothing. When both forces—trend and cycle—are at work, the total effect is greater than the sum, reflecting the dynamic interaction that amplifies their respective strengths.
Model and reality
What does this suggest about labor markets, earnings, and inequality? To the economists, it means that business cycles play a central role in determining long-run inequality. Interacting with the underlying trend of SBTC, recessions exacerbate inequality by pushing low-skilled workers out of the workforce. Many don’t return during subsequent expansions because their earnings potential erodes while unemployed. Going back to work literally isn’t worth it.
“Recessions,” therefore, “cause persistent increases in inequality.”
“There is an important interaction between trend and cycle,” write Heathcote, Perri, and Violante. SBTC favors high-skilled workers but, while employed, even low-skilled men maintain market value by learning on the job. But if they lose their jobs in a recession, their value declines and job prospects fall. They stop looking for work while earnings for others rise when the economy rebounds.
“Recessions,” therefore, “cause persistent increases in inequality.”