Principal investigator: Bruno Benevit
Original title: Unemployment Insurance and Job Search Behavior Get access Arrow
Authors: Iona Marinescu and Daphné Skandalis
Location of the Intervention: France
Sample Size: 457.000 worker-months
Primary Variable of Interest: Reserve salary
Type of Intervention: Unemployment insurance
Methodology: OLS, Poisson Counting Model
Summary
Unemployment insurance is a tool designed to reduce the impact of job loss shocks on workers' income. However, the way the job search process unfolds can have distinct consequences for this type of policy in terms of the preferences of its beneficiaries. To examine how unemployment insurance affects this dynamic, this article used data from French job application databases to assess how unemployment insurance impacts job search effort and the reserve wage of beneficiaries. The results indicate that workers increase their job search efforts in the year before and after the benefit ends to a similar extent, and that the wage desired by workers falls continuously throughout the unemployment insurance period.
- Policy Problem
Unemployment insurance aims to mitigate the effects of income shocks caused by job loss, providing greater income predictability for workers. Economic literature highlights the existence of a stable consumption pattern for individuals throughout their lives, supporting the objective of this type of policy.
The process of searching for a new job, however, involves an investment of time and effort. Furthermore, the period of absence from work leads to the depreciation of human capital and/or its signaling to the labor market. In this sense, unemployment insurance can have repercussions on salary expectations and the search for a new job. The absence of an immediate financial need to re-enter the labor market can reduce the effort in the job search. Added to the depreciation of human capital, this can also negatively affect workers' reservation wages (MARINESCU; SKANDALIS, 2021). On the other hand, job search models predict that greater availability of time to re-enter the labor market can imply an increase in the desired salary (CHETTY, 2008).
- Policy Implementation Context
Unemployment benefits in France can be claimed after job loss by registering on the Public Employment Service (Pôle Emploi) platform. This platform allows employers to create standardized job advertisements, which candidates can apply to by filling out an online form. Approximately one-fifth of unemployed French workers have submitted an application on the search platform, and the number of applicants is similar to the unemployed worker population. Behavior on this platform is largely representative of the French labor market, and the proportion of searches conducted through this channel can be considered exogenous (MARINESCU; SKANDALIS, 2021).
Eligibility for the program occurs if the beneficiary has worked for more than 4 months in the 28 months prior to the involuntary loss of employment – for workers over 50 years of age, 36 months prior to job loss. The potential duration of benefits (DPB) depends on the number of days worked during this reference period, and cannot exceed two years for workers under 50 years of age (three years for unemployed individuals aged 50 or older).
The value of benefits depends on salaries received in the last 12 months, with a minimum and maximum limit. The net replacement rate is around 70% on average, a lower value than in most European countries. Disability benefits are longer compared to other European countries, and the most frequent discontinuation periods observed in France are two and three years, the maximum limits for workers under and over 50 years of age, respectively. After the benefits end, and provided that individuals are actively seeking employment, they can apply for welfare benefits.
- Evaluation and Policy Details
The data considered in this study comprise observations of workers who began a period of unemployment between 2013 and 2017 and submitted at least one application on the platform during that period (regardless of eligibility for benefits). Data from the Public Employment Service platform were cross-referenced with three other administrative datasets, allowing for the identification of the unemployment period, the value of unemployment benefits, the DPB (Detailed Percentage of Unemployment) of each unemployed individual, periods of benefit receipt, replacement rate, salary from previous employment, and period of job placement.
The data were organized as a panel at the levels of the unemployment period (monthly). The observations consider online search activity from the date of unemployment registration on the platform until the end of the unemployment period, for up to three years. This study evaluated two dimensions of job search behavior: (i) the effort in the job search process – measured by the number of applications per month during the unemployment period – and (ii) the reservation wage – measured by the monthly average of wages offered in the job advertisements targeted by the workers.
- Method
To estimate the impact of unemployment insurance on job search behavior, this study adopted several models to isolate the individual behavior of the unemployed. The analyses considered unemployed workers ineligible for unemployment insurance and eligible unemployed workers with a period of unemployment between 12 and 24 months. Two samples were used: (i) the complete sample – considering all observations of unemployment episodes – and (ii) the balanced sample – considering only observations of unemployment episodes occurring more than 3 years prior to the worker's death.
To estimate the impact on the number of applications, the Poisson counting method was used, and for the effect on the reservation wage, an ordinary least squares (OLS) model was estimated. The base model considered the variables related to the period in relation to the unemployment benefit period (assuming a value of 0 when ineligible) and the fixed effects of unemployment benefit to estimate the impact of unemployment insurance on job search behavior in the two years before and after the end of the unemployment benefit period. In order to correctly identify the impact of unemployment insurance on job search behavior, the authors added the fixed effects of the total unemployment period (to control for effort levels among workers) and the period (month) of unemployment (to isolate the effect of unemployment insurance from the effect of individual worker reactions).
The authors also investigated how the time elapsed since the end of unemployment benefits and the length of unemployment (in years) impact the preferences of unemployed workers in the job search process. To this end, the effects on wage measurement variables (log) were estimated for different periods: monthly, weekly, and hourly wages. Furthermore, possible changes in preferences for different occupations with lower pay and the same occupations with lower wages were examined. Finally, it was also verified whether unemployment insurance induced the search for jobs with lower qualification requirements than those of unemployed workers and what the impact was on the number of applications for positions with open-ended contracts. Two models were adopted to verify the impact on these outcome variables: one considering the interaction of the period after the end of benefits with a binary variable identifying wages below the distribution average, and another disregarding this interaction.
- Main results
Estimates regarding the impact of unemployment insurance on job search effort indicated that workers gradually and significantly increase the number of job applications until the benefit ends, at which point they reach their peak. The benefit end period was followed by a gradual decrease in the magnitude of the impact, but it remained significantly positive for all models that considered the fixed effects of the unemployment period. The magnitude of the effect during the end period ranged from 72% to 121%, and the effect under the model with both fixed unemployment effects (best specification) indicates a 100% increase in effort. The coefficients maintain significance and behavior across periods when considering the balanced sample, but with smaller magnitudes (approximately half the value).
Regarding the impacts on the target salary in job applications, the results demonstrated that the desired salary is significantly and gradually reduced until the end of the benefit period, with the magnitude of the effects remaining relatively stable for subsequent periods when considering the models with the best fit. For these models, the coefficients for the period after the end of the disability benefit ranged from -4,9% to -2,6% for the complete sample and from -5,5% to 4,2% for the balanced sample.
The results of the analyses on the impact on different salary variables indicated a considerably smaller reduction in weekly and hourly wages compared to monthly wages. In other words, the observed reduction in desired monthly wages can be explained by a preference for shorter working hours after the policy intervention. Regarding the impacts on the quality of desired jobs in terms of required qualifications and type of contract, the estimates indicated that workers began to accept more jobs with qualifications below their level and decreased their applications for positions with open-ended contracts. The results were robust for both proposed models.
- Lessons in Public Policy
This article analyzed how unemployment insurance affects the behavior of unemployed workers in their search for new employment, based on data from a French job application platform. Specifically, the authors examined how this type of policy affects the preferences of the unemployed in this process in terms of search effort, salary expectations, and the quality of desired positions around the time the benefit ends.
Through the adoption of several robust identification strategies, the results indicated a convergence of efforts by unemployed individuals around the period of benefit termination, at which point the greatest effort was observed – a result consistent with that observed in the literature. However, unemployment insurance induced a reduction in the desired job aspirations of workers in terms of salary and job quality. The evidence from this study highlights the importance of understanding the unexpected negative effects of public policies on their target audience, enabling the improvement of current and future programs.
References
CHETTY, R. Moral Hazard versus Liquidity and Optimal Unemployment Insurance. Journal of Political Economy, v. 116, no. 2, p. 173–234, apr. 2008.
MARINESCU, I.; SKANDALIS, D. Unemployment Insurance and Job Search Behavior. The Quarterly Journal of Economics, v. 136, n. 2, p. 887–931, 24 Mar. 2021.
