The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
Home / Publications / Research / Economic Brief / 2025 Hiring Puzzle: Why Do Firms Decrease Hiring So Much in Recessions? By Katarína Borovičková Economic Brief March 2025, No. 25-10 Key T akeaways Unemployment rate uctuations observed in the data are much larger than what can be explained by traditional search models. Inspired by explanations that rationalize high volatility of the stock market through uctuations in risk premia, we hypothesize that uctuations in the unemployment rate might be driven by the same factor. While we nd that uctuations in risk premia indeed help, we also conclude that it cannot be the only driving force. T he unemployment rate uctuates with the business cycle, rising and falling as economic activity changes: Businesses increase hiring during expansions (causing unemployment to decline), and they reduce their hiring e orts during recessions (leading to persistently high unemployment rates). Productivity shocks are the main driver of the business cycle, but are these productivity uctuations also the sole driver of such employment cycles? In this article, I discuss my recent work showing that productivity uctuations alone are insu cient to explain employment uctuations. Unemployment Volatility Puzzle From a macroeconomic perspective, economists seek to understand business cycle uctuations in unemployment rates for several compelling reasons. A primary concern is that cyclical unemployment represents wasted productive capacity. Unemployed workers / tend to lose their human capital over time, diminishing the economy's future productive potential. Understanding these uctuations is therefore crucial for designing policies that minimize such ine ciencies. T he primary framework for analyzing unemployment dynamics is the search model developed by Peter Diamond, Dale Mortensen and Christopher Pissarides.1 T his model envisions an economy populated by workers and rms, where rms employ workers to produce output and pay wages. T o hire new workers, rms must post vacancies. Workers exist in one of two states — employed or unemployed — with unemployed workers searching for jobs by pursuing posted vacancies. T heir job search e orts may or may not result in employment. T he business cycle in the model is generated through productivity shocks, as is usual in the business cycle literature. T he model's key insight is that an unemployed worker's probability of nding employment depends on the vacancy-unemployment ratio, known as the labor market tightness. Job- nding probabilities are high during expansions, when vacancies are plentiful and unemployment is low. Conversely, during recessions, fewer vacancies and higher unemployment lead to lower job- nding probabilities. While this model successfully captures the directional relationship between economic conditions and unemployment — low unemployment during expansions and high unemployment during recessions — it falls short in explaining the magnitude of unemployment uctuations. T he model predicts much smaller variation in unemployment rates than what's observed in empirical data. T his discrepancy between the model's predictions and real-world observations has become known as "the unemployment volatility puzzle." T his is also referred to as the Shimer puzzle, after Robert Shimer's in uential 2005 paper "T he Cyclical Behavior of Equilibrium Unemployment and Vacancies," which rst highlighted this limitation. Solutions to the Unemployment Volatility Puzzle A critical limitation of the standard model is its prediction that wages absorb most productivity shocks. During recessions, rms' initial inclination to reduce hiring when productivity falls is largely o set by rapid wage adjustments. T his wage exibility dampens the decline in hiring incentives, resulting in only modest reductions in job postings and, consequently, small uctuations in unemployment rates. T his disconnect between theoretical predictions and empirical evidence has spurred extensive research in labor economics. Scholars have proposed various solutions, including alternative wage-setting mechanisms and model recalibrations. While some of these modi cations successfully generate the observed unemployment volatility, they introduce a new problem: T hey predict implausibly large uctuations in expected pro ts from hiring an additional worker. / The Business Cycle Link to the Stock Market In the nancial literature, economists have documented that stock market valuation also exhibits high volatility over the business cycle.2 T his volatility and its underlying causes have been extensively studied. By de nition, a stock's value equals the present discounted value of its future dividend stream. Research has shown that dividends themselves remain relatively stable over business cycles. T his contrast between highly volatile stock prices and stable dividends has become known as the excess volatility puzzle.3 Consequently, economists have concluded that valuation volatility must stem from changes in how investors value future dividends, a phenomenon captured by uctuations in discount rates. Two Puzzles, One Explanation? Could changes in how rms value future payments explain the unemployment volatility puzzle? A rm's hiring decisions depend on how it values potential workers — speci cally, the present discounted value of a worker's future contributions to the rm pro ts. If this valuation uctuates signi cantly, rms might dramatically reduce job postings during recessions as the perceived value of new hires declines. T he nancial literature has demonstrated that variation in discount rates successfully explains stock market uctuations.4 T his raises an intriguing possibility: Could the same valuation mechanism provide a uni ed explanation for both stock market volatility and the unemployment volatility puzzle? T hat is the question Jaroslav Borovička and I ask in our 2025 working paper "Risk Premia and Unemployment Fluctuations (PDF)." Our Approach Our analysis builds on extensive nance literature that guides the construction and estimation of a model for discount rates disciplined by nancial data. Using this model, we investigate the required properties of rms' pro ts from new hires that would enable the search model to generate unemployment rate uctuations matching empirical observations. T his novel approach (rather than directly modeling the rm's pro t stream) o ers two key advantages. First, it characterizes properties that any labor search model must satisfy to resolve the unemployment volatility puzzle successfully. Second, it serves as a diagnostic tool for evaluating existing models, illuminating the sources of their successes and failures. New Channel Analysis of the labor market search model enriched with uctuations in discount rates reveals two potential resolutions to the unemployment volatility puzzle, each operating through distinct properties of the pro t ow: / High variance of expected pro ts High mean of the conditional variance of pro ts T he rst property — high volatility of expected pro t ow — implies that a worker's contribution to rm pro ts varies substantially over the business cycle: T his pro t is high during expansions but drops signi cantly during recessions. T his represents the traditional channel that models without time-varying discount rates must rely on to explain the unemployment volatility puzzle. T he second property — high expected conditional volatility of pro t ow — focuses on a mechanism that interacts with time-varying discount rates. Under this mechanism, workers generate stable pro ts during expansions. However, during recessions, the marginal pro t becomes highly uncertain while potentially maintaining its average level. T he pro t could either exceed normal levels or fall far below them, but rms cannot predict which outcome will occur at the onset of the recession. T his uncertainty leads risk-averse rms to signi cantly discount the value of new hires during recessions compared to expansions. In a model without time variation in discount rates, rms are not concerned about this uncertainty and hence it does not a ect their hiring decisions. T he increased risk in pro t ow during recessions may stem from several sources. First, sales may become highly unpredictable as customers abruptly reduce spending or postpone purchases, in contrast to the stable purchasing patterns seen during expansions. Second, rms' signi cant xed costs (including facilities and core personnel) cannot be quickly reduced when revenue falls, forcing rms to absorb revenue volatility. T hird, rms often resort to discounting their products during recessions to maintain sales volume, further destabilizing pro t margins. T hese mechanisms contribute to generating substantial pro t volatility during recessions. Figure 1 illustrates these distinct forces through hypothetical pro t patterns. / Enlarge / Enlarge Figure 1a illustrates an example of pro t ow driven primarily by the rst channel: Pro ts remain consistently high during normal periods but systematically plummet during recessions (shown as the shaded areas). Figure 1b demonstrates the impact of the alternative channel: Pro ts maintain the same average level across both expansions and recessions, but their volatility increases dramatically during recessions, uctuating between extreme highs and lows while remaining stable in normal times. In a search model incorporating time variation in discount rates, either channel (or a combination of both) can resolve the unemployment volatility puzzle if su ciently strong. Most existing explanations in the macroeconomics literature rely exclusively on the rst channel, as these models typically feature only very weak concerns about risk on the rm side. However, this approach requires implausibly high pro t volatility — approximately 25 times the volatility of GDP — to rationalize the observed volatility of unemployment. T he introduction of the second channel — which emerges only in models with an appropriately speci ed time variation in discount rates — reduces the burden imposed on the rst channel. T his raises a crucial question: How much can the second channel contribute to resolving the puzzle? / Our Findings Our paper quanti es the trade-o between these two channels. Using an empirically plausible model of time variation in discount rates disciplined by nancial data, we characterize a bound on all possible combinations of the two channels necessary for the model to generate observed uctuation in rms' willingness to hire workers. If we discipline mechanism 1 (volatility of expected pro t ow) by the data, then mechanism 2 (high conditional volatility of pro ts) must also be present to generate unemployment uctuations. T he bound we derived tells us how strong mechanism 2 needs to be. Our main nding is that it must be stronger than what we can support in the data. In particular, the conditional volatility of rm pro t would need to be 8.5 times higher than the riskiness of pro ts measured in the data. T his implies that mechanism 2 alone will not resolve the unemployment volatility puzzle. At the same time, this mechanism turns out to be important. If we discipline both mechanism 1 and mechanism 2 by the data, our model generates 50 percent of the unemployment uctuations observed in the data, while it can generate only less than 5 percent if mechanism 2 is not present. Measurement Challenge: Average Versus Marginal Pro t An important caveat to the calculations above deserves emphasis. When making hiring decisions, rms consider the so-called marginal pro t, or the additional pro t generated by a newly hired worker. T his quantity cannot be directly measured in the data, as individual worker productivity and individual contributions to rm's total pro ts remain unobservable. Instead, we measure the pro t ow generated by the average worker, which can be calculated from observed data on rm pro ts and employment. While marginal pro ts likely exhibit greater volatility than average pro ts, the magnitude of volatility required by our analysis appears to be very high. Conclusion Returning to our central question: Can changes in the valuation of future cash ows explain both stock market volatility and unemployment uctuations? Our analysis suggests a nuanced answer. While valuation uctuations successfully explain stock market dynamics, they appear insu cient as the sole driver of variation in rms' hiring decisions. Although this new mechanism contributes to resolving the unemployment volatility puzzle, a complete explanation likely requires additional factors. Based on our empirical analysis detailed in the paper, we suggest that combining this valuation channel with nancial frictions may o er a promising direction for future research. / Katarína Borovičková is an economist in the Research Department at the Federal Reserve Bank of Richmond. 1 See, for example, Pissarides' 2011 paper "Equilibrium in the Labor Market With Search Frictions." 2 See, for example, the 1989 paper "Business Conditions and Expected Returns on Stocks and Bonds" by Eugene Fama and Kenneth French. 3 See Shimer's 1981 paper "Do Stock Prices Move Too Much to Be Justi ed by Subsequent Changes in Dividends? (PDF)." 4 See, for example, the 1988 paper "The Dividend-Price Ratio and Expectations of Future Dividends and Discount Factors" by John Campbell and Robert Shiller. To cite this Economic Brief, please use the following format: Borovičková, Katarína. (March 2025) "Hiring Puzzle: Why Do Firms Decrease Hiring So Much in Recessions?" Federal Reserve Bank of Richmond Economic Brief, No. 25-10. T his article may be photocopied or reprinted in its entirety. Please credit the author, source, and the Federal Reserve Bank of Richmond and include the italicized statement below. Views expressed in this article are those of the author and not necessarily those of the Federal Reserve Bank of Richmond or the Federal Reserve System. Topics Business Cycles Employment and Labor Markets Subscribe to Economic Brief Receive a noti cation when Economic Brief is posted online. By submitting this form you agree to the Bank's Terms & Conditions and Privacy Notice. / Email Address Subscribe Contact Us RC Balaban © 1997-2025 Federal Reserve Bank of Richmond /