BENGHALEM Helene

We ran a large randomized controlled experiment among about 150,000 recipients of unemployment benefits insurance in France in order to evaluate the impact of part-time unemployment benefits. We took advantage of the lack of knowledge of job seekers regarding this program and sent emails presenting the program. The information provision had a significant positive impact on the propensity to work while on claim, but reduced the unemployment exit rate, showing important lock-in effects into unemployment associated with part-time unemployment benefits. The importance of these lock-in effects implies that increasing the marginal tax rate on earnings from work while on claim in the neighborhood of its current level would not decrease labor supply and would decrease the expenditure net of taxes of the unemployment insurance agency.

No summary available.

My thesis studies how the design of unemployment insurance can affect the behavior of job seekers and employers. In my first thesis chapter, we evaluate the impact of taxing very short-term contracts. Unédic introduced an increase in employer contributions to unemployment insurance in 2013. In order to limit insecurity and segmentation in the labor market, several European countries have also decided to tax fixed-term contracts. In order to analyze this problem, we have developed a model explaining the employer's choice between fixed-term and open-ended contracts, as well as the duration of fixed-term contracts. The structural estimation of the model then allows us to study the effects of taxing contracts shorter than 30 days. We find two opposing effects. On the one hand, employers prefer to create 30-day contracts rather than contracts just under 30 days. On the other hand, the tax reduces the duration of all contracts further away from the 30-day threshold. This latter phenomenon has a significant impact on the average duration of contracts, as very short contracts are created in large numbers. In particular, our results show that taxing short contracts increases unemployment and reduces the welfare of the unemployed. In the second chapter, I study the design of the regime for intermittent workers in the entertainment industry. In particular, I measure the effect of the eligibility threshold on the labor supply of intermittents. Between 2003 and 2016, intermittents had to work 507 hours over the last 10 months to be eligible for unemployment benefits. This 507-hour threshold induces strategic behavior that can be costly for unemployment insurance. For intermittent workers just below 507 hours, a slight increase in the number of hours worked implies a 60% increase in their income from unemployment benefits. I show that individuals optimize by positioning themselves just above the threshold creating an observable jump in the distribution studied and a hole below the threshold. I develop a frictionless model a la Kleven and Waseem (2013). I find a structural elasticity between 0.26 and 0.28. Finally, I determine the impact of this strategic behavior on the unemployment insurance deficit. The induced cost is non-negligible since it represents 27% of the annual deficit. In the third chapter, I examine whether individuals understand the link between the social contributions they pay and their future benefits. If individuals understand this link, social contributions will be less of a tax, thus limiting their disincentive effects on labor supply. To analyze this problem, I explore the discontinuity created by the reference period in the eligibility rules for unemployment insurance. Intuitively, when an individual registers for unemployment insurance, only the hours included in the reference period will be used to determine his or her right to compensation. If individuals are aware of the contribution-allowance link, they have a strong incentive to work more hours during the reference period. To illustrate this point, I study the behavior of intermittent workers in the entertainment industry. Since 2016, intermittents have had to work 507 hours over a reference period of the last 12 months to be eligible for unemployment benefits. This rule implies that one year after they enter the labor market, the reference period will shift excluding some hours of work in the calculation of eligibility. I show that in order to avoid this, a large number of intermittent workers reach the 507 hours required for eligibility one year after entering the labor market, creating an observable jump in the distribution studied. I develop a dynamic frictionless model explaining the choice of the date of eligibility for unemployment insurance and find a structural elasticity of 0.52.