As measured by flows of jobs and workers across employers, U.S. labor markets became much less fluid in recent decades. We document a large, broad‐based decline in these labor market flows, drawing on multiple data sources. An aging workforce and a secular shift away from younger and smaller employers can partly account for the long‐term decline in labor market fluidity, but these forces are not the main story. Instead, we find large declines in worker reallocation across employers within groups defined by gender and age or gender and education. Likewise, we document large declines in job reallocation across employers within groups defined by industry, employer size, and employer age.
International comparisons suggest that the large, secular decline in U.S. job reallocation is somewhat unusual. In light of these facts, we consider whether reduced labor market fluidity is cause for concern about the U.S. economy. There are beneficial and benign aspects of reduced labor market fluidity, but also strong reasons for concern about harmful consequences of reduced fluidity for productivity growth, real wages, and employment.
A first concern is that slower job and worker reallocation might reflect slower arrival of new job opportunities. For the unemployed, this increases the risk of long jobless spells. For the employed, it hampers the ability to switch employers so as to move up a job ladder, change careers, or satisfy location constraints. In line with this observation, previous studies find that job mobility facilitates wage growth and career advancement (Topel and Ward, 1992; Hagedorn and Manovskii, 2013; and Akerlof, Rose, and Yellen, 1988).
A second worry is that available evidence cuts against the view that reallocation slowed because firms now face a more quiescent economic environment. Bloom et al. (2012) find rising volatility of plant‐level productivity shocks in U.S. manufacturing after 1990. Decker et al. (2014b) find that the intra‐industry dispersion of plant‐level productivity rose, not fell, in the past quarter century; they also find a declining trend in the responsiveness of plant‐level growth rates to plant‐level productivity shocks from 1980 to 2010. Although limited to manufacturing, this evidence indicates that job and worker reallocation rates trended down because U.S. employers became less responsive to shocks, not because employer‐level shocks became less variable.
A third concern about reduced fluidity is that it might reflect expansion of government regulations and policies that hamper reallocation. According to Kleiner and Krueger (2013), the fraction of workers required to hold a government‐issued license rose from less than 5 percent in the 1950s to 29 percent in 2008. A large literature finds that employment protection laws suppress labor market flows, sometimes to a powerful extent (Blanchard and Portugal , 2001; Gómez‐Salvador et al., 2004; Boeri and Jimeno, 2005; OECD, 2010; and Haltiwanger et al., 2014).
State court decisions from 1972 to 1999, and especially in the 1980s, seriously eroded the presumption that employees could be fired at will. Autor et al. (2006) find that the introduction of exceptions to employment‐at‐will reduced state‐level employment rates, while Autor et al. (2007) find that these exceptions reduced the volatility of annual employment growth rates in state‐industry cells and, in the manufacturing sector, depressed productivity. In addition to the erosion of the employment‐at‐will doctrine, many federal and state laws enacted in recent decades establish protected classes of workers defined by race, religion, gender, age, disability, national origin, and other worker characteristics. These laws, however well intentioned, likely contribute to the trend declines in job and worker reallocation rates in recent decades, with negative effects on labor market fluidity and perhaps on employment, wages, and productivity as well.
Other policy interventions suppress labor market flows as a by‐product or unintended consequence. The preferential tax treatment of employer‐provided health insurance, which encourages most Americans to obtain health insurance through their employers, may generate “job lock” for many workers, suppressing job‐to‐job mobility (Currie and Madrian, 1999; and Gruber, 2000). Some evidence also suggests that minimum wage laws reduce reallocation (Dube et al., 2013; and Brochu and Green, 2013). Since the real federal minimum wage is lower now than in the 1970s, however, it seems unlikely that changes in the prevalence and bite of minimum wage provisions have contributed to the secular declines in worker reallocation.
Finally, we think the information revolution has played a significant role in the trend declines in worker reallocation. Information about criminal records, credit histories, unfavorable media coverage, and even ill‐advised web postings has become more abundant and cheaper to access and process. The likely result is a shift to stricter selection on the hiring margin and less use of trial employment arrangements that contribute to hires and separations.
Thus, many factors may have contributed to reduced labor market fluidity in the United States. We think restrictions on occupational labor supply, wrongful discharge and anti‐discrimination laws, and the preferential tax treatment of employer‐provided health insurance are among the policy factors that played a role in reducing labor market fluidity. Regardless of other benefits (and costs) associated with these policy factors, their role in suppressing labor market fluidity can lead to negative effects on productivity, welfare, and employment.
Our empirical examination of the relation between fluidity and employment exploits data on state‐level rates of employment, job reallocation, and worker reallocation. We estimate the effects of the reallocation measures on state‐level employment rates for groups defined by gender, education, and age, while controlling for other relevant factors that might have influenced employment rates. We find large effects of worker reallocation rates on the employment rates of the young and the less educated.
The effects are uniformly larger for men. For example, a 100‐basis‐point decline in the worker reallocation rate yields an estimated 77 basis point decline in the employment rate for men who did not finish high school. For men under 25 who did not finish high school, the corresponding estimate is 143 basis points. The larger estimated effects for the young, the less educated, and men comport well with the actual pattern of larger employment rate declines for these groups. When we use the job reallocation rate as our fluidity measure, doubling our sample period, we find positive and statistically significant effects of fluidity in all education groups for men and women. For both fluidity measures, the cross‐state patterns of declines in actual employment rates are captured reasonably well by the predictions of our empirical models.
If our assessment of how labor market fluidity affects employment is approximately correct, then the U.S. economy faced serious impediments to high employment rates well before the Great Recession. Moreover, if our assessment is correct, the United States is unlikely to return to sustained high employment rates without restoring labor market fluidity.
This Research Brief is based on a paper prepared for the Federal Reserve Bank of Kansas City’s economic policy symposium on “Re‐Evaluating Labor Market Dynamics,” held August 21–23, 2014, in Jackson Hole, Wyoming. The paper is available at http://www.nber.org/papers/w20479. All works cited are provided therein.