Early signs of a recession can lead to a negative feedback loop, with workers’ concerns about unemployment dampening demand and thus deepening the recession. This column uses a heterogeneous agent model to quantify the importance of the ‘unemployment-risk’ channel for business cycle fluctuations in the US economy. It shows that the channel accounts for around one-third of observed unemployment fluctuations. As the demand amplification through precautionary savings is inefficient, this finding provides an additional rationale for stabilisation policies by policymakers.
“Fear of unemployment could well lead to further increases in the saving rate that would dampen consumption growth in the near term.”1 This statement, from the minutes of a Federal Open Market Committee (FOMC) meeting in the wake of the Great Recession, embodies a common view that early signs of a recession might be amplified because workers who start worrying about unemployment increase their savings. The resulting reduction in demand may then set in motion a feedback loop that further increases unemployment and deepens the recession. Although often alluded to in policymaking and the public press, we have little idea about the quantitative importance of this ‘unemployment-risk channel’ of macroeconomic transmission. This is problematic because the design of macroeconomic stabilisation policies hinges on the size of such inefficient amplification.
Quantifying the unemployment-risk channel is difficult, as it requires a general equilibrium analysis where the time-varying risks faced by individual workers play a role for aggregate demand through individual savings decisions. Traditional macroeconomic models, whose representative household implicitly insures its members against unemployment, are thus not suitable. Recent advances in heterogeneous agent macroeconomics, however, have allowed us to study the role of uninsured idiosyncratic risks, such as from unemployment, for macroeconomic dynamics and the effects of policies.2 In a recent working paper (Broer et al. 2021), we build on this literature to quantify the importance of the unemployment-risk channel for business cycle fluctuations in the US economy.
Figure 1 Estimated response of unemployment to monetary policy and total factor productivity shocks
Notes: This figure presents estimated responses of unemployment, and of transitions from employment to unemployment (EU) and vice versa (UE) in response to identified shocks to the Federal Funds Rate (left-hand side) and total factor productivity (TFP) (right-hand side), taken from Broer et al. (2021). Shares of unemployment response explained by EU/ UE in brackets.
Studying the implications of time-varying unemployment risk for aggregate demand first requires a gauge of how that risk fluctuates over the business cycle. In particular, because worker savings react to fluctuations in the prospect of job loss as well as to changes in the likely duration of unemployment, we need a good idea of how both job separation and job finding rates contribute to overall unemployment fluctuations. Figure 1 shows how changes in monthly inflows and outflows contribute to overall unemployment movements in the US in response to two common macroeconomic shocks (monetary policy and total factor productivity, respectively).3 Two stylised facts emerge. First, fluctuations in the separation rate account for a sizeable share (between one-third and two-thirds depending on the shock) of unemployment fluctuations. And second, the peak response in job finding lags that in separations, by between six and 16 months.
To study the implications of these stylised facts, and of fluctuations in unemployment risk in general, for the dynamics of aggregate demand, we build on a recent workhorse model (Ravn and Sterck 2021) from the heterogeneous agent literature. To account for the stylised facts, two additional ingredients are crucial: first, job separations (that are often taken to be constant in standard models) must be responsive to changes in economic conditions. And second, firm vacancies must respond sluggishly, and thus with a lag relative to separations, to changes in the expected profitability of jobs (as in Coles and Kelishomi 2018).
The resulting framework allows a quantification of the unemployment-risk channel by comparing its predictions to those that result when households are insured against idiosyncratic unemployment risk (and only the average unemployment rate thus matters for their savings decisions). Such comparison shows that the unemployment-risk channel accounts for about one-third of observed unemployment fluctuations. Importantly, accounting for the two stylised facts is crucial for this result: when separations are exogenous and firm vacancies adjust freely to changes in economic conditions (and other parameters are adjusted to match the observed volatility of US unemployment), the size of the unemployment-risk channel is only half as large as in the benchmark economy.
These results carry two important messages for policymakers. First, because the demand amplification through precautionary savings is inefficient, they provide an extra rationale for stabilisation policies. And second, they illustrate, again, a common theme of recent research studying the interplay between inequality and macro dynamics: that by affecting unemployment risk and its consequences, structural labour market policies may have an important aggregate demand component.
Broer, T, J Druedahl, K Harmenberg and E Öberg (2021), “The Unemployment-Risk Channel in Business-Cycle Fluctuations”, CEPR Discussion Paper 16639.
Coles, M G and A Moghaddasi Kelishomi (2018), “Do job destruction shocks matter in the theory of unemployment?”, American Economic Journal: Macroeconomics 10(3): 118–36.
Fernald, J G (2014), “Productivity and Potential Output before, during, and after the Great Recession”, NBER Macroeconomics Annual 29(1): 1-51.
Jordà, Ò (2005), “Estimation and Inference of Impulse Responses by Local Projections”, American Economic Review 95(1): 161–182.
Kaplan, G and G L Violante (2018), “Microeconomic heterogeneity and macroeconomic shocks”, Journal of Economic Perspectives 32(3): 167-94.
Krueger, D, K Mitman and F Perri (2016), “Macroeconomics and household heterogeneity”, Handbook of Macroeconomics Vol. 2, Elsevier, 843-921.
Miranda-Agrippino, S and G Ricco (2021), “The Transmission of Monetary Policy Shocks”, American Economic Journal: Macroeconomics 13(3): 74–107.
Ravn, M O and V Sterk (2021), “Macroeconomic Fluctuations with HANK & SAM: an Analytical Approach”, Journal of the European Economic Association 19(2): 1162–1202.
2 See Kaplan and Violante (2018) and Krueger et al. (2016) for recent surveys.
3 The impulse responses for the labour market transitions are computed using a smoothened version of the local projection method from Jordà (2005). We take monetary policy shocks from Miranda-Agrippino and Ricco (2021), and the first difference of the quarterly total factor productivity (TFP) series in Fernald (2014) for technology shocks.