Skip to main content

On the misery of losing self-employment


Using German panel data, this study compares responses of self-employed workers’ well-being and paid employees’ well-being to the loss of work. The empirical analyses show that life satisfaction decreases substantially more in the probability of losing work when self-employed than when paid-employed. It also turns out that becoming unemployed yields a much stronger decline in self-employed workers’ life satisfaction than in paid employees’ life satisfaction. Although these results do not necessarily represent causal effects, they indicate that losing self-employment is an even more harmful life event than losing dependent employment. Further analyses substantiate this conclusion. It is also shown that non-monetary reasons may explain much more than monetary reasons why the self-employed seem to suffer in particular from unemployment. In addition, the difference in the responses of well-being to unemployment between self-employed workers and paid-employed workers originates from varying levels of life satisfaction after terminating work, but not from divergent levels of life satisfaction before terminating work. One implication of these findings is that the potential psychological cost of unemployment might constitute a risk that prevents workers from going into business by themselves.

This is a preview of subscription content, access via your institution.

Fig. 1


  1. See Parker (2009) for an extensive account of theoretical and empirical work that has been done on this issue.

  2. Krause (2013) also focuses on reemployment and documents that overall well-being during unemployment is more positively related to being self-employed than to being paid-employed in the future.

  3. In consequence, the second approach does not rely on information about the current likelihood of losing work and, thus, obviates assumptions of the first identification strategy in this respect.

  4. The first identification strategy can be seen as a robustness check in this respect. Subgroup analyses will also address this issue.

  5. With respect to some of the reasons for entering unemployment, the data contain only very few observations of workers who stay continuously unemployed for more than two SOEP interviews in a row. Therefore, these observations are not considered at all.

  6. Reported differences in Sects. 5 and 6 are statistically significant at least at the 5 % level.

  7. F tests confirm for all of the specifications that the interaction effect of self-employed × q and the level effects (self-employed, q) are jointly significant. This also applies to all of the subgroup analyses (Sect. 5.3).

  8. According to Knabe and Rätzel (2011b), this calculation might overestimate the CIV as it does not account for the effect such a hypothetical compensation would have on permanent income. The underlying test is not applicable to individual fixed effects estimations and thus not feasible here. However, the CIVs would continue to point to high economic relevance even if they were overestimated by 55 % as in the example of the authors.


Download references


The author is very grateful to Ronnie Schöb, Andreas Knabe, C. Katharina Spieß, Adrian Chadi, Frank Fossen, Katja Görlitz, Laszlo Goerke, Malte Preuß, Silva Lea Haselon, two anonymous referees and the editor Rui Baptista. He also thanks participants of the SOLE-EALE World Meeting (Montréal, 2015), the Quality of Life Conference (Berlin, 2014), the HEIRs Conference on Public Happiness (Rome, 2013) and the Workshop in Economics (Trier, 2013).

Author information

Authors and Affiliations


Corresponding author

Correspondence to Clemens Hetschko.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Supplementary material 1 (DOCX 71 kb)



See Tables 3, 4 and Fig. 2.

Table 3 Descriptive statistics—first analysis
Table 4 Descriptive statistics—second analysis
Fig. 2
figure 2

Predicted and actual probability of losing work. Source: SOEP 1999–2011. Note: the actual probability of losing work is the share of losses of work during the next 2 years, dependent on the predicted probability. Salaried workers lose their jobs because of any kind of dismissal (including plant closures), the end of a fixed-term contract or the end of an apprenticeship. Self-employed workers close their businesses. The figure illustrates classes of the predicted probability of losing work because of small numbers of observations. Thin lines illustrate 95 % confidence intervals

Rights and permissions

Reprints and Permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Hetschko, C. On the misery of losing self-employment. Small Bus Econ 47, 461–478 (2016).

Download citation

  • Accepted:

  • Published:

  • Issue Date:

  • DOI:


JEL Classifications