Page 21 - IDEA Study 8 2017 Direct subsidies and R&D output in firms

this type of outcome. On the one hand, applying for IP protection locally is generally easier, faster and cheaper, so this might be a preferred strategy to generate eligible project results for reporting purposes. On the other hand, applying for IP protection internationally in any of the four major foreign patent offices, signals a higher quality of invention and a higher value IP application. The motivation for this comparison is to find out whether there are any “unconditional” differences at all, irrespective of whether the difference can be attributed to the impact of the programme. Table 3 shows this comparison in the effective sample within the first three years of funding from the respective programme call, i.e. over the period (t) to (t+2). Clearly, the “treated” programme participants had a substantially higher propensity to apply for both Czech and international IP protection than the “untreated” non-participants. According to t-tests these differences are highly statistically significant in all cases with the exception of just one outlier. Nevertheless, it would be misleading at this point to consider the untreated to be a relevant control group and to associate this difference with the true effect of the subsidy programmes; this distinction between the two groups may result from the two-step selection of firms into applicants, and applicants into subsidy recipients, as explained above. Programmes may increase or reduce this difference depending on the effectiveness of the subsidies. To find out whether this is the case, we compare the characteristics of the treated and untreated firms one year before the funding started, i.e. in the year (t-1), which mostly refers to the period when the call was announced and the project proposals evaluated (see Appendix Table A1 for full results). This reveals that there were indeed strong and highly statistically significant differences before the subsidies started. We can see that the firms that went on to receive subsidies were on average more likely to apply for IP protection even before they received any programme funding. Moreover, there were sizeable ex-ante differences in many other respects, including size, age, subsidy history, labour productivity and solvency.10 As expected, the overall sample of non-participants is not a relevant control group for the subsidy participants and we therefore need to rely on the statistical propensity score matching method to pin down the impact of the programmes. 10 The solvency ratio, i.e. the firms’ ability to meet their long-term financial obligations, is calculated as (Shareholders funds / Total assets) * 100%. 19