The rise of alternative work arrangements – from temporary and part-time work to self-employment and online gig-economy jobs – has increased the take-up of part-time unemployment benefits in several countries. This column presents the results of a large experiment among recipients of unemployment-benefits insurance in France. It shows that increasing part-time unemployment benefits raises the propensity to work in non-regular jobs, but also extends the duration of compensated unemployment and unemployment insurance expenditure, suggesting that the lock-in effects of compensated unemployment associated with part-time benefits require precise evaluation.
Unemployment insurance plays an increasingly large role in the labour market (Boone et al. 2017, Marinescu 2016, Midões and Serék 2021). At the same time, part-time unemployment benefits provided to persons working non-regular jobs while seeking a regular job play an increasingly important role in unemployment insurance systems. The rise in the incidence of such alternative work arrangements as temporary work, part-time work, and self-employment – as well as the new kinds of work relationships emerging in the online gig economy – has increased part-time unemployment take-up in several countries. In France, almost half of all unemployment-benefit recipients work while on claim during their unemployment spell. Part-time unemployment benefits are also widespread in Belgium, Finland, Austria, and Germany (Ek Spector 2015, Cahuc 2018).
In principle, part-time unemployment benefits supply incentives to job seekers looking for regular jobs to accept non-regular jobs in the meantime. This may increase overall employment and shorten unemployment spells if non-regular jobs act as stepping stones towards regular jobs. However, such benefits may also induce lock-in effects by discouraging unemployed workers from searching for regular jobs. Knowing the relative importance of the stepping-stone and lock-in effects that condition access to regular employment is essential to evaluating the impact of part-time unemployment benefits on labour supply and unemployment insurance expenditure.
Unfortunately, little is known about these issues because the potential selection into part-time unemployment of individuals with non-observable characteristics correlated with their exit rate from unemployment makes the evaluation of part-time unemployment insurance very difficult. Most studies look at the effects of working non-regular jobs on the access to regular employment (see Auray and Lepage-Saucier 2021 for a recent study and references). They find mixed results, showing generally significant lock-in effects while individuals work non-regular jobs and more positive effects on the access to regular jobs after non-regular jobs end. But the effects of part-time unemployment benefits are less explored. Recent contributions relying on natural experiments (Ait Bihi Ouali et al. 2020, Le Barbanchon 2020) find that part-time unemployment benefits have significant incentive effects on the propensity to work while on claim. Another recent contribution of Lee et al. (2019) – which relied on a randomised experiment in the State of Washington in the north-western US during the 1990s – finds that more generous part-time unemployment benefits did not raise labour supply but did increase the unemployment insurance expenditure because they encouraged the propensity to claim benefits.
To evaluate the impact of part-time unemployment insurance in a different and more recent context, we ran a large randomised controlled experiment among about 150,000 recipients of unemployment insurance benefits in France to whom we provided information about the existence of part-time unemployment benefits (Benghalem et al. 2021). We deduce the impact of part-time unemployment insurance on the behaviour of unemployed workers from the change in their behaviour induced by the provision of information. The choice of this strategy is justified by the lack of knowledge about part-time unemployment insurance among job seekers. A survey conducted by the employment agency has shown that 41.2% of job seekers do not know of the existence of the programme, and that 33.6% are aware of its existence but do not know the rules.
In our experiment, initiated in January 2017, individuals who had recently entered unemployment were randomly allocated either to a treated group or to a control group. Individuals assigned to the treated group were sent emails that contained a description of the part-time unemployment insurance scheme. Individuals in the control group did not receive any message, while otherwise facing identical conditions in terms of employment services.
We find that the information provision has a significant positive impact on the propensity to work on while on claim, as shown by Figure 1. The probability that treated individuals will take work while on claim increases by about 6% three months after receiving the information compared with non-treated individuals. The positive effect of the information provision on part-time unemployment benefits take-up means that the treatment has improved the gains expected from work while on claim for the treated group as a whole.
Figure 1 Effects of information provision on propensity to work while on claim
Notes: Each red dot denotes the point estimate for intention to treat effects at a given time horizon based on OLS regressions on the indicator variable equal to one from the first month in which the individuals start working while on claim. The green lines denote a 95% confidence interval for the corresponding point estimate where standard errors are clustered at the agency level.
The hike in the propensity to work while on claim is associated with a drop in the exit rate from unemployment. The information provision raises the probability that individuals will remain unemployed until the initial exhaustion date of their unemployment benefits. The effect is significant: a 6% increase in the probability that job seekers will take work while on claim three months after the start of the treatment is associated with 1.5% hike in the probability that they will remain unemployed the last month before the initial benefits exhaustion date. Therefore, it is clear that increasing part-time unemployment benefits raises the compensated unemployment duration.
Looking at unemployment insurance expenditure, we find that the savings induced by the increase in work while on claim are more than counterbalanced by the lock-in effects into compensated unemployment. We show that the importance of lock-in effects induced by part-time unemployment benefits implies that increasing the marginal tax rate on earnings from work while on claim in the neighbourhood of its current level would decrease the expenditure net of taxes of the unemployment insurance without decreasing labour supply.
These results, together with those of Lee et al. (2019), indicate that the lock-in effects into compensated unemployment associated with part-time unemployment benefits, which play an increasing role in many countries, need to be precisely evaluated to design efficient unemployment insurance benefits.
Ait Bihi Ouali, L, O Bargain and X Joutard (2020), “Partial unemployment insurance and hour decisions: Evidence from administrative data”, Technical report, Aix Marseille University.
Auray, S and N Lepage-Saucier (2021), “Stepping-stone effect of atypical jobs: Could the least employable reap the most benefits?”, Labour Economics 68: 101945.
Benghalem, H, P Cahuc and P Villedieu (2021), “The Lock-in Effects of Part-time Unemployment Benefits”, CEPR Discussion Paper 15921.
Boone, C, A Dube, L Goodman and E Kaplan (2017), “The impact of unemployment insurance expansion on aggregate employment during the Great Recession”, VoxEU.org, 8 January.
Cahuc, P (2018), “Wage insurance, part-time unemployment insurance and short-time work in the xxi century”, IZA Discussion Paper 12045.
Ek Spector, S (2015), “Should unemployment insurance cover partial unemployment?”, IZA World of Labor 199.
Le Barbanchon, T (2020), “Taxes today, benefits tomorrow”, Working Paper.
Lee, D S, P Leung, C J O’Leary, Z Pei and S Quach (2019), “Are sufficient statistics necessary? nonparametric measurement of deadweight loss from unemployment insurance”, NBER Working Paper 25574, also forthcoming in Journal of Labor Economics.
Marinescu, I (2016), “Why unemployment benefits should be extended in recessions: US evidence”, VoxEU.org, 1 March.
Midões, C and M Serék (2021), “Millions of Europeans would fall into vulnerability if it were not for COVID-19 unemployment benefits”, VoxEU.org, 6 February.