Optimal unemployment insurance with monitoring

Abstract

I model job-search monitoring in the optimal unemployment insurance framework, in which job-search effort is the worker's private information. In the model, monitoring provides costly information upon which the government conditions unemployment benefits. Using a simple one-period model with two effort levels, I show analytically that the monitoring precision increases and the utility spread decreases if and only if the inverse of the worker's utility in consumption has a convex derivative. The quantitative analysis that follows extends the model by allowing a continuous effort and separations from employment. That analysis highlights two conflicting economic forces affecting the optimal precision of monitoring with respect to the generosity of the welfare system: higher promised utility is associated not only with a higher cost of moral hazard, but also with lower effort and lower value of employment. The result is an inverse U-shaped precision profile with respect to promised utility.

Optimal unemployment policy

“…Several studies have found that the cost of monitoring the job search of unemployed individuals is quite small, typically around 1% of per capita output per individual monitored; this, for example, is similar to findings in Boone et al. (2007) for Sweden, van den Berg and van der Klaauw (2013) for the Netherlands, and Setty (2019) for the US 28 . Therefore, I assume that a similar program is to be implemented here, with q = 0.01; additional results available upon request show that my results are not very sensitive to the value of q , except that monitoring intensity ϕ increases rapidly as q approaches zero.…”

“…See, for example,Gruber (1997) on consumption smoothing, andMeyer (1990),Lalive et al (2006) andChetty (2008) on the moral hazard costs of UI.2 These papers find that traditional UI should be followed by monitored or assisted search and then social assistance. Other related papers include a simple study of monitoring with linear precision in van der Linden(2003), and more recent principal-agent papers such asSetty (2019), which considers monitoring of binary search,Pavoni et al (2016), who study 'soft' programs including JSA, andJung and Kuester (2015), who consider the optimal combination of UI, vacancy subsidies, and layoff taxes in recessions.3 See also, for example,Hansen and İmrohoroğlu (1992) for optimal UI andMichelacci and Ruffo (2015) for life-cycle UI.4 van der Linden (2003) includes a payroll tax paid by firms, whileWunsch (2013) andSetty (2019) calibrate their models to tax rates of 36% and 29% respectively. There is no inherent distinction between welfare-maximization and principal-agent approaches on this dimension: even though principal-agent models assume that the planner 'owns' the agent's consumption, such models will not incorporate fiscal externalities from income taxes unless the model features a revenue requirement for spending by the planner.…”

“…2 2 These papers find that traditional UI should be followed by monitored or assisted search and then social assistance. Other related papers include a simple study of monitoring with linear precision in van der Linden (2003), and more recent principal‐agent papers such as Setty (2019), which considers monitoring of binary search, Pavoni et al. (2016), who study ‘soft’ programs including JSA, and Jung and Kuester (2015), who consider the optimal combination of UI, vacancy subsidies, and layoff taxes in recessions. …”

Imperfect Monitoring of Job Search: Structural Estimation and Policy Design

“…To our knowledge, only two studies have considered such imperfect monitoring technology. Setty (2015) builds a dynamic principal-agent model with two levels of search effort to address a question of optimal UI with a probabilistic monitoring technology in which the probability of a negative evaluation decreases with the effort level. Boone et al (2007) consider a stationary equilibrium search model with sanctions.…”

Unemployment Insurance and Unemployment Accounts: The Best of Both Worlds

“…Although some monitoring of such rejections takes place in the United States,Setty (2014) shows that the average monthly monitoring probability in that country is 0.20. This is an upper bound for the probability of observing rejections because some rejections are undetected.4 The key result of this paper depends on some source of moral hazard that would allow voluntarily unemployed workers to receive government benefits in either UI2 or UIA.…”