Panel Paper: Unemployment Insurance as a Worker Indiscipline Device? Evidence from Scanner Data

Saturday, March 10, 2018
Burkle 12 (Burkle Family Building at Claremont Graduate University)

*Names in bold indicate Presenter

Lester Lusher, University of Hawaii at Manoa, Geoffrey C. Schnorr, University of California, Davis and Rebecca Taylor, University of Sydney

Economic theory predicts that an increase in the generosity of unemployment insurance (UI) benefits will lead to a decrease in on-the-job effort among eligible workers. In a simple model, if effort is costly, and a worker’s probability of job loss is partially affected by their effort at work, workers will respond to an increase in UI benefit level or duration by exerting less effort. UI programs are ubiquitous in developed countries and they account for substantial portions of government spending. Even a small effect of UI benefit changes on worker productivity would have substantial implications for public policy. However, empirical evidence for this ex-ante moral hazard effect is extremely limited, and it is unclear whether or not it exists in practice.

We test for this effect by exploiting state-time variation in the duration of UI benefits in the United States during the Great Recession, which occurred due to the Extended Benefits (EB) and Emergency Unemployment Compensation (EUC) programs. High-frequency transaction level data from 80 locations (in three states) of a large national supermarket chain allow us to directly observe the productivity of individual cashiers, measured as the length of a given transaction conditional on a range of transaction, store, register, and cashier characteristics. The data allow us to control for unobserved confounders which are constant within cashiers, stores, and registers.

In preliminary results from a generalized difference-in-differences specification with day and cashier-register fixed effects, we find evidence of a causal, negative relationship between expanded UI benefits and worker productivity. This effect is strongest for longer-employed cashiers, for whom UI expansions are especially likely to be pertinent, and in larger transactions, during which cashiers have more of an opportunity to shirk. We are able to provide suggestive evidence against several observable confounding responses by both cashiers and employers. These results suggest that cashiers in our sample reduced their on-the-job effort in response to increases in the potential duration of UI benefits.