188Fig. 1 Post-treatment effect of grant on labor demand over ﬁrm size

imaginary ﬁrm with zero employees, whereas the interaction term describes how the effect changes with ﬁrm size. To interpret the interaction effect in the extended model with greater ease, we display the marginal effect and how it varies with ﬁrm size in Fig. 1. Firm size is measured as the natural logarithm of value added, ln(va). Here, we choose to only present the post-treatment effect of the grants from the interaction models in our preferred model, i.e., the matched DiD-model (i.e., column 5 in Table 4), over the observed range of ﬁrm sizes. The plots show that the marginal post-treatment effect of the grants on the number of employees in general is not statistically signiﬁcant regardless of ﬁrm size. There is a tendency that the effect of the grants increases with ﬁrm size and there is a negative and signiﬁcant post-treatment effect for the smallest ﬁrms. These results deviate to some extent from the literature, where the most positive effects of ﬁrm support programs in general are found for small ﬁrms.4

5.2 Effects of Targeted R&D Grants on Employees

with Higher Education

One aim of the programs under study is to help ﬁrms manage R&D projects. It is therefore expected that these grants should encourage ﬁrms to invest more in skilled labor, and thus increase their relative demand for workers with higher human capital. Hence, even without any impact on the total number of workers, it is possible that ﬁrms that are supported by the government programs would substitute less qualiﬁed workers for workers with higher human capital.

4The marginal effect of the grants during the duration of the program is not statistically signiﬁcant anywhere and therefore not depicted in a ﬁgure.

189Do Targeted R&D Grants toward SMEs Increase Employment and Demand for High. . .

Following Hansson (2000), we estimate how the demand for highly educated labor has been affected by the support programs. The results are presented in Table 5, showing no positive effect of the support programs on the relative demand for workers with post-secondary education in our main model speciﬁcations (columns 1–3). In fact, there are no estimates that are statistically signiﬁcantly different from zero during the program period or after the program has ended. This implies that there is no effect in the short- or longer-term for these grants.

In the extended models (columns 4–6), we observe a negative post-treatment effect when comparing treated ﬁrms with their own growth pattern and with all other non-treated ﬁrms. The interaction term, however, loses its signiﬁcance when we compare treated ﬁrms with the matched control group (column 5).

To capture the marginal effect of the grants on demand for skilled labor over the ﬁrm size distribution (as we did with labor demand in Fig. 1), we compare the treated ﬁrms with a matched control group. However, the results are not statistically signiﬁcant, and we can therefore not conclude that the effects of the targeted R&D grant programs on the demand for employees with higher education are dependent on ﬁrm size.5

5.3 Post-Treatment Effects

So far, we have presented the results when estimating the averaged post-treatment effects of the support programs. The reported estimated effects thus display the average effect of the programs after the grants are no longer paid out to the companies. One concern is that there might be a non-linear response from the