Page 31 - IDEA Study 8 2017 Direct subsidies and R&D output in firms

Heterogeneity checks In this section, we examine the heterogeneity of the output additionality effects with regards to i) size categories of firms; ii) industry versus services; and iii) instruments of IP protection. The analysis is focused on Czech IP protection, because this is where the strongest effects are concentrated and because the data is insufficient for deriving reliable estimates for international IP protection in the subsamples. The presented results are based on the combined propensity score matching and conditional difference-in-differences model, which is preferable from the methodological point of view. Nevertheless, as has been shown above, the results are robust to the model specification. The magnitudes of the estimated effects and respective standard deviations are reported in figures, while tables with the full results are available in the appendix. For the sake of saving space, only the estimated differences between untreated and treated firms are reported, not the underlying propensities, are reported (this information is available from the authors upon request). It should be reiterated that the results for TIP in the third (t+2) year and for ALFA in the second (t+1) and third (t+2) years need to be interpreted with great caution, especially when the dataset is divided into subsamples, as the data on applications for IP protection are incomplete due to the publishing delay. Firm size Following the classification by Eurostat (2017), we divide firms into three broad size categories depending on their number of employees: 1) Small enterprises: fewer than 50 persons employed;13 2) Medium-sizedenterprises:50-249personsemployed; 3) Large enterprises: 250 or more persons employed. 13 Given the aforementioned limitation of the Amadeus dataset, the effective sample predominantly contains firms with 25 or more employees; hence in practice this category refers to enterprises with 25 to 49 employees. 29