Stimulating part-time work by legal entitlements? Evidence from a German policy experiment

Abstract

Difference-in-difference estimates indicate that the new law on part-time work in Germany has raised the share of part-time workers in those plants that already used part-time employment whereas it has not stimulated the introduction of part-time work in other plants.

Notes

¹ Statistical tests and the fact that plant size is positively correlated with the existence of part-time employment in a plant, but negatively correlated with the share of part-time workers in such establishments clearly indicate that a Tobit model would be too restrictive in our context.

² Previous microeconometric studies investigating the determinants of part-time work with establishment data include Montgomery (1988), Houseman (2001) and Düll–Ellguth (1999). The latter study is also based on the IAB Establishment Panel.

³ We do not include the share of female employees since this is clearly co-determined with the share of part-time workers, which would result in endogeneity problems.

⁴ Ai–Norton (2003, p. 123) also point out that ‘most applied researchers misinterpret the coefficient on the interaction term in nonlinear models’ as the true interaction effect.

⁵ All computations were performed with Stata 8.2. The marginal effects (and their significance) for the probit model were calculated using the ado-file inteff.ado supplied by Ai–Norton. Our corresponding programme for the truncreg specification is available upon request from the first author of this article.

⁶ In Model 2, the marginal effects for the different size categories in the probit specification are now negative and significant in all but the lowest group. Since most of the larger plants (e.g. 96% of those with more than 500 employees) did already have part-time employment before 2001, the slight reduction in the probability of part-time work indicated by these estimates does not imply that plants have given up part-time employment.