186Beginning with the basic results (columns 1–3), we find no indications that the support programs have increased employment, neither during the support period nor after the program has ended. When comparing treated firms with the population of all non-treated firms, the post-support effect of the programs is negative and statistically significantly different from zero. Thus, firms that receive support do not grow faster in terms of number of employees compared to the average firm. On the contrary, they grow slower than the average firm when the support program ends.

In the extended models, the marginal effect of the grants is a combined effect of the direct effect and the interaction term that allow the results to vary with respect to firm size. The treatment effect can therefore be seen as the impact of the grant on an Do Targeted R&D Grants toward SMEs Increase Employment and Demand for High. . .

187Note: *, **, ***, corresponds to levels of significance of 10, 5, 1%, with robust standard errors in parenthesis. Estimated with system GMM with endogenous variables: employment, value added, wages. Collapsed IV-matrix, lag depth is set to two, see Hayakawa (2009). Estimation 2 and 5 weighted with 0.809*(0.466)

1.044***(0.394)

0.055(0.078) 0.294(0.239) 2.602(1.866) Treated vs. all 455.1(518) 0.081(0.244) 0.317(0.465) 50.11(55.7)

0.75 0.54 (6) 28

1.148**(0.505)

1.151*(0.633)

0.896***(0.171)

Treated vs. ctrl.

591.3**(293)

0.008(0.036) 0.270(0.379)

49.40**(25.2)

0.076(0.135) 0.035(0.228)

0.27 0.50 (5) 28

Interaction models 0.0004(7.7e-04) 7.6e-05(1.2e-04)

2.055**(1.044)

0.001(0.490) 0.242(0.210) 0.364(1.664) 0.086(0.057) 0.613(0.397) 0.021(0.842) Treated only

0.88 0.61 (4) 28

9.0e-05(5.8e-04)

0.103***(0.033)

0.909***(0.092)

0.173***(0.052)

3.037(7.780) Treated vs. all 0.019(0.013) 0.028(0.111)

CEM-matched weights. Firm- and period fixed effects included in all models 0.05 0.37 (3) 28

0.0001(5.5e-04)

0.215**(0.090)

0.909***(0.092)

Treated vs. ctrl.

35.57**(17.2)

0.021(0.017) 0.050(0.109) 0.494(0.323)

0.04 0.28 (2) 21

2.218**(1.130)

5.8–05(5.9–05) Table 4 Estimation results, labor demand 0.137(0.847) 0.756*(0.435) Basic models 0.089(0.229) 1.179(1.418) 0.122(0.455) Treated only

0.83 0.46 (1) 28

ln(Y)t*(Post-support)t

(Skill intensity)t-1

ln(Y)t*(Grant/Y)t (Post-support)t # instruments Hansen p-val AR2 p-val (Grant)t

ln(VA)t

(Profit)t

ln(L)t-1

ln(w) t

188S.-O. Daunfeldt et al.

95% CI

Marginal effect

500

0

ln(va)

1.3

3.3

5.3

7.3

9.3

11.3

13.3

15.3

-500

-1000

-1500