The Labor Demand and Labor Supply Channels of Monetary Policy (with Christopher Huckfeldt and Eric T. Swanson)
Monetary policy is conventionally understood to influence labor demand, with little effect on labor supply. Estimating the response of labor market flows to high-frequency changes in interest rates around FOMC announcements and Fed Chair speeches, we find that a contractionary monetary policy shock leads to a significant increase in labor supply, by reducing the rate at which workers quit jobs to non-employment and stimulating job-seeking behavior among the non-employed. Holding the response of supply-driven labor market flows constant, the overall decline in employment from a contractionary monetary policy shock becomes nearly twice as large.
We incorporate time-averaging into the canonical model of Heckman, Lochner, and Taber (1998) (HLT) to study retirement decisions, government policies, and their interaction with the aggregate labor supply elasticity. The HLT model forced all agents to retire at age 65, while our model allows them to choose career lengths. A benchmark social security system puts all of our workers at corner solutions of their career-length choice problems and lets our model reproduce HLT model outcomes. But alternative tax and social security arrangements dislodge some agents from those corners, bringing associated changes in equilibrium prices and human capital accumulation decisions. A reform that links social security benefits to age but not to employment status eliminates the implicit tax on working beyond 65. High taxes with revenues returned lump-sum keep agents off corner solutions, raising the aggregate labor supply elasticity and threatening to bring about a ``dual labor market'' in which many people decide not to supply labor.
[NEW DRAFT: May 2023]
[AEJ: Macroeconomics, Conditionally Accepted] (Federal Reserve Working Paper)
In this paper, I show that the decline in consumption during unemployment depends on both liquid and illiquid wealth; that unemployment predicts illiquid asset withdrawal, primarily when households have few liquid assets; and that increased idiosyncratic unemployment risk leads to a rise in saving overall, but also to a decline in investment in illiquid assets. Motivated by these new findings, I embed endogenous unemployment risk in a two-asset heterogeneous-agent New Keynesian model. The model is consistent with the new evidence and suggests that aggregate shocks are amplified by a flight-to-liquidity when unemployment risk rises, particularly when monetary policy is constrained.
[AEJ: Macroeconomics, Vol. 15, No. 2, April 2023] (Federal Reserve Working Paper)
The responsiveness of job creation to shocks is procyclical, while the responsiveness of job destruction is countercyclical. This new finding can be explained by a heterogeneous-firm model in which hiring costs lead to lumpy employment adjustment. The model predicts that policies that aim to stimulate employment by targeting the job creation margin, such as hiring subsidies, are significantly less effective in recessions: These are times when few firms are near their hiring threshold and many firms are near their firing threshold. Policies that target the job destruction margin, such as employment protection subsidies, are particularly effective at such times.
Unemployment Insurance Financing as a Uniform Payroll Tax (with Jonathon Hazell, Walker Lewis and Christina Patterson)
[AEA Papers and Proceedings, Vol. 112, May 2022]
In the United States, unemployment insurance is financed by taxes levied on employers. We develop a model to decompose UI taxes into a firing tax component, levied on firms that layoff workers, and a uniform payroll tax component, levied on all firms regardless of their layoffs. We develop a novel methodology to measure the two components and document a number of facts about the uniform payroll tax component: it is large, accounting for just under half of UI taxes, it rises significantly after recessions, and it is more cyclical in states with poorly funded UI system.
This paper outlines a new method for solving the agent’s problem in models where agents trade two assets. This method involves splitting the two-dimensional maximization problem into two sequential one-dimensional problems. I show that this approach is fast, can be used in settings when an endogenous grid method cannot be applied, and is simple to implement.