When people quit their jobs to take outside offers or to exit the job market, it often sets in motion replacement hiring to fill job vacancies.1 Measuring the number of these hires helps us understand labor market dynamics, shedding light on the overall state of the economy and informing policy.2

In their paper, “Vacancy Chains,” Michael Elsby and Axel Gottfries of the University of Edinburgh, Ryan Michaels of the Philadelphia Fed, and David Ratner of the Board of Governors of the Federal Reserve System examine how quits and replacement hiring create vacancy chains and how these chains in turn impact the aggregate labor market.

As background, the authors explain that at any given time, some firms are expanding and creating new positions, other firms are downsizing by leaving vacant jobs unfilled, and still others are operating in an intermediate replacement region. Those firms operating in the replacement region are neither productive enough to expand their positions nor unproductive enough to need to shed workers. Instead, they hire workers only to replace those that quit. (In general, whether a firm is expanding, downsizing, or in the replacement region depends on firm-specific factors as well as overall macroeconomic conditions.)

Why might many firms lie in the replacement region? One key reason, the authors explain, is that firms face potentially high costs when restarting formerly idle positions after the business environment improves. Specifically, it is costly to recreate the positions (which can involve substantial capital investments) and to integrate them into the workflow of the firm. The authors reason that firms, facing these costs, may find it more profitable to fill their vacancies rather than to leave positions empty — and thus retain the capacity after workers quit.

A critical implication of the replacement region, they argue, is the emergence of vacancy chains, which originate when expanding firms invest in new positions and poach workers from other firms. They note that vacancy chains propagate when “the poaching of one worker triggers a cascade of further hires among the many firms that seek to maintain their employment [in the replacement region].” The vacancy chains terminate, they explain, when positions are filled by workers from contracting firms or from the unemployment pool.

The authors begin their analysis using microdata from the U.S. Bureau of Labor Statistics (BLS), which allows them to study a cross-section of establishments over time to determine the extent of replacement hiring.3 Next, the authors develop a model of vacancy chains that captures replacement hiring at the firm level. To this end, their model includes workers’ on-the-job search, which generates quits, and two kinds of labor market adjustment “frictions”: (i) the cost of hiring a worker when a position already exists; and (ii) the cost of creating or later restarting a position. (The key aspect of replacement hiring, they explain, is that the latter friction is substantially larger than the former, leading more firms to replace quits using currently vacant positions.) Next, they use their model to analyze the implications of vacancy chains for the aggregate labor market.

Their analysis of the microdata reveals that over their sample period, establishments often reported no net change in their employment, often over many years, even though these same firms had extensive gross turnover resulting from workers quitting. These observations suggest a significant role for replacement hiring and, indeed, the authors estimate that over their sample period, replacement hiring accounted for 45 percent of total hiring.

The authors also show that their model, which incorporates replacement hiring, generates a significant amount of labor market volatility. Consider what happens during an economic expansion. Worker replacement costs rise as quits increase, and the higher cost of sustaining a given level of employment limits the amount of hiring. By contrast, the cost of creating positions, while large, does not change. Since the latter cost is by far the larger of the two, the overall cost of expanding rises relatively little (despite the increase in worker replacement costs), which leads firms to increase employment more than otherwise, contributing to labor market volatility.4

The authors also find that firms’ motive for replacement hiring is responsible for the gradual pace at which overall (aggregate) hiring often unfolds. Recall that replacement hiring is grounded in the cost of creating positions. If this cost is large and the replacement region is wide, then firms only gradually move out of this region and commence either expanding (following an improvement in market conditions) or shrinking (following a deterioration in conditions). Crucially, this process unfolds at different speeds for different firms, depending on their own idiosyncratic circumstances. That is, some firms begin expanding early on in a business cycle recovery, whereas others defer hiring until later. When firms have fully adjusted, total hiring will have changed substantially, but it is a gradual process, mimicking the sluggishness we tend to see in the data.

In summary, the authors show that replacement hiring, and the vacancy chains that follow, are pervasive in the U.S. economy and meaningfully shape the dynamics of aggregate employment.

  1. The views expressed here are solely those of the authors and do not necessarily reflect the views of the Federal Reserve Bank of Philadelphia or the Federal Reserve System.
  2. U.S. quit rates have historically averaged 2 to 3 percent per month. For more details, see, for example, Steven Davis, R. Jason Faberman, and John Haltiwanger, “Labor Market Flows in the Cross Section and Over Time,” Journal of Monetary Economics, 59:1 (2012), pp. 1–18.
  3. According to the U.S. Bureau of Labor Statistics (BLS), a vacancy implies that “a specific position exists and there is work available for that position.” The BLS began collecting vacancy data when it created the Job Openings and Labor Turnover Survey (JOLTS) in December 2000. Since then, JOLTS has become the primary source of U.S. vacancy data.
  4. The authors used JOLTS data, which contain a sample of 16,000 establishments per month for the period December 2000 to mid-2016. JOLTS provides data on hires and separations. (The latter consist mainly of quits and layoffs.) The authors also used the BLS’s Quarterly Census of Employment and Wages (QCEW) data. Each month, firms are required to report their employee count and quarterly compensation bill, which is used by the BLS to create the QCEW. The QCEW covers about 98 percent of U.S. nonfarm employees in private-sector establishments. The authors had access to QCEW data on 40 states from the early 1990s to 2014.
  5. What is crucial here is the change in rather than the level of the cost of expanding employment. The level is higher on average because of the cost of starting new positions.