Unemployment insurance (UI) protects workers against the loss of earnings when they lose their job, but more generous benefits can also lengthen unemployment spells and lead more workers to enter unemployment. In France, the generosity of UI is governed mainly by the benefit level, the qualifying period of work needed to open a claim, and the potential benefit duration.
In a policy note for France’s Conseil d’analyse économique (CAE), Fontaine, Rathelot, and Roulet (2026) use the MVPF to compare possible reforms that would reduce the generosity of French UI. One of these is a further cut in the potential benefit duration. Since a 2023 reform, which cut potential duration by 25 percent, a jobseeker’s entitlement has been 75 percent of the time worked over the reference period, or 18 months for a 24-month reference period. The hypothetical reform the paper considers lowers this conversion coefficient from 0.75 to 0.6.
The paper draws upon estimates of behavioral responses from Le Barbanchon (2016) and Jessen et al. (2025) to compute the MVPF of lowering the conversion coefficient from 0.75 to 0.6, per euro of mechanical savings. It pairs them with its own estimates, from Crédit Mutuel Alliance Fédérale bank data, of the drop in consumption six months after job loss among the jobseekers the reform affects.
MVPF = 1.0
The net cost is estimated as €1 of UI benefit plus the fiscal externalities from behavioral responses: a longer potential duration lengthens unemployment spells, which means more benefit payments and fewer payroll contributions, and makes contract endings, layoffs and mutually agreed terminations more frequent, raising entries into unemployment. The paper measures these effects per euro of mechanical savings from the cut.
Selecting the available estimates most relevant to France, the paper assumes that a 10 percent cut in potential duration shortens unemployment by 2.7 percent, following Le Barbanchon (2016). Following Jessen et al. (2025), it assumes that the same cut lowers entries into unemployment by 1.2 percent.
In the paper’s accounting budget model, the cut saves €2,137 million a year in benefits mechanically once fully phased in, and behavioral responses save a further €1,049 million, with each of the two responses contributing about half. The net cost per euro of benefit is therefore (€2,137 million + €1,049 million) / €2,137 million = €1.49.
The willingness to pay for €1 of UI benefits is estimated as
WTP = 1 + \gamma \frac{\Delta c}{c}
to capture that jobseekers are willing to pay more than €1 to the extent that they are risk averse and that their consumption falls when they lose their job. Risk aversion is measured by the coefficient of relative risk aversion, γ, and the consumption drop by Δc/c. The paper assumes a coefficient of relative risk aversion of 4.5, and its estimate of the drop in consumption six months after job loss for the jobseekers this reform affects is 10.42 percent. This leads to a willingness to pay of €1 + 4.5 × €0.1042 = €1.469.
With a willingness to pay of €1.469 and a net cost of €1.491 per euro of benefit, the MVPF is 0.99. Because the reform is a cut, the MVPF can also be read the other way: the affected jobseekers would be willing to pay €0.99 to avoid the cut for each euro of net savings it generates. With a coefficient of relative risk aversion of 2 instead of 4.5, the willingness to pay falls to €1.208 and the MVPF to 0.81.
Fontaine, François, Roland Rathelot, and Alexandra Roulet (2026). “Measuring the Efficiency of Unemployment Insurance.” Les notes du conseil d’analyse économique, No. 90. https://cae-eco.fr/static/pdf/CAE090_Assurance-chomage_260630_EN.pdf
Jessen, Jonas, Robin Jessen, Andrew C. Johnston, and Ewa Gałecka-Burdziak (2025). “Moral Hazard among the Employed: Evidence from Regression Discontinuity.” NBER Working Paper 33450. https://doi.org/10.3386/w33450
Le Barbanchon, Thomas (2016). “The Effect of the Potential Duration of Unemployment Benefits on Unemployment Exits to Work and Match Quality in France.” Labour Economics, 42: 16–29. https://doi.org/10.1016/j.labeco.2016.06.003