Mette Ejrnæs får publiceret artikel i Economic Journal

In this article we investigate if self-employed men who has an unemployment insurance are more likely to become unemployed (moral hazard).

The main challenge when analyzing the effect of unemployment insurance on subsequent unemployment is the fact that individuals who have a higher risk also are more likely to take-up an insurance, e.g. if they are having health problems.  We also find that insured self-employed have a higher risk of unemployment than uninsured. However, this will not tell us about the causal impact of insurance but may simply be due to the fact that those who are insured are different from those who are not insured (selection). To disentangle the causal impact we use three particular features of the Danish unemployment scheme: first it is voluntary to take-up unemployment insurance, second unemployment insurance is also available for self employed and third the eligibility to an early retirement program is closely linked to unemployment insurance. The two first features enable us to compare the risk of unemployment for insured and not insured self-employed. The eligibility to early retirement gave many people an additional reason for being enrolled in the unemployment insurance. By exploiting the fact that substantial fraction of self-employed actually enroll in the unemployment insurance to be eligible to early retirement rather than being covered by insurance we are able to disentangle the effect of insurance.

Using detailed longitudinal data from the official registers on Danish self-employed men we can estimate the causal effect. We use data from the period 1980 to 1998. We find that the raw data suggest that self-employed insurees are more likely to subsequently become unemployed than the uninsured self-employed. The difference in transition rates is about 1.8 percentage points, and hence very sizeable compared to a raw unemployment rate of about 2 percent in the sample. However, this observed difference will be due to both selection and moral hazard. When we take care of the non-randomness in the insurance choice, the marginal effect drops to 0.6 percentage points. This suggests that of the original difference only about 30 percent is due to moral hazard, while the remaining 70 percent is due to heterogeneity or sorting. The overall magnitude is not very big, however, such that only a limited proportion of business failure can be attributed to the lack of effort.
Stefan Hochguertel, VU Amsterdam is a co-author on this article.

Article: Ejrnæs and Hochguertel, 2013, “Is business failure due to lack of effort?- Evidence from a large administrative sample”, Economic Journal, vol. 123, pp 791-830.