The switcher-stayer gap
The Federal Reserve Bank of Atlanta maintains a Wage Growth Tracker derived from the Current Population Survey's matched rotation groups. The tracker compares year-over-year wage growth for two categories of employed individuals: those who remained with the same employer over the twelve-month period and those who changed employers. Over most of the series' history, the median wage growth for job switchers has exceeded that of job stayers, often by one to two percentage points. The gap tends to widen during economic expansions and narrow — or occasionally reverse — during recessions, when voluntary switching declines and those who do switch may be moving out of necessity rather than into better-paying positions.
Early evidence on job mobility and earnings
The relationship between employer changes and wage growth has been studied since at least the early 1990s. Robert Topel and Michael Ward, in a 1992 paper published in the Quarterly Journal of Economics, analyzed longitudinal earnings records from Social Security Administration data and estimated that roughly one-third of the total wage growth experienced by young men during their first decade in the labor force was attributable to job changes rather than within-firm raises. Their finding established a baseline that subsequent researchers have revisited with different datasets and methodologies.
Bruce Fallick and Charles Fleischman extended this line of inquiry in a 2004 Federal Reserve Board working paper that used matched Current Population Survey data to construct monthly estimates of employer-to-employer flows. Their estimates showed that direct job-to-job transitions — where a worker moves to a new employer without an intervening spell of unemployment — accounted for a substantial share of total labor reallocation, and that these direct transitions were associated with larger wage gains than transitions that passed through unemployment.
The role of the business cycle
Giuseppe Moscarini and Fabien Postel-Vinay have argued in a series of papers, including a 2017 contribution to the American Economic Review, that the rate of employment-to-employment reallocation serves as a more informative predictor of aggregate wage dynamics than the unemployment exit rate. Their theoretical framework models the labor market as a job ladder, where workers climb toward higher-paying employers during expansions and the ladder collapses during downturns. In this model, the wage gains from switching are not uniform: they depend on the relative productivity of the origin and destination employers.
Henry Haltiwanger, Henry Hyatt, Lisa Kahn, and Erika McEntarfer provided empirical support for a related idea in a 2018 paper in the American Economic Journal: Macroeconomics. Using linked employer-employee administrative data from the Census Bureau's Longitudinal Employer-Household Dynamics program, they documented that job-to-job moves disproportionately flow from smaller, lower-paying firms to larger, higher-paying ones, and that this pattern intensifies when the labor market is tight.
Methodological cautions
The finding that job switchers experience faster wage growth than job stayers is well documented, but it requires careful interpretation. Selection effects are a persistent concern: workers who switch employers voluntarily are not a random subset of the workforce. They may be younger, more educated, employed in higher-demand occupations, or in possession of outside offers that less mobile workers do not receive. The observed wage premium associated with switching therefore reflects both the causal effect of changing employers and the characteristics of the workers who are able and willing to do so. Disentangling these two components remains an active area of research.
Additionally, survey-based wage data — including the CPS data underlying the Atlanta Fed tracker — are subject to measurement error. Respondents may report wages imprecisely, and matched-rotation designs can produce attrition that is not random with respect to mobility. Administrative datasets such as the LEHD reduce some of these problems but introduce others, including the absence of hours-worked data, which makes it difficult to distinguish between changes in hourly wages and changes in hours or employment intensity.