Page 19 - IDEA Study 8 2017 Direct subsidies and R&D output in firms

the state budget or to contribute to the proposed project only using its own private resources. In all these cases, we count the firm as a programme participant: the firm represents the so-called “treated” observation in the analysis. In other words, as soon as a firm is listed in the administrative records as participating in a subsidized project, it is considered to have been treated by the subsidy. The longitudinal structure of the dataset is organized by the calls for proposals. All participants in projects funded in a given call are grouped together as a cohort, regardless of the exact start date or duration of their project. As a result, the data for each year refer to a particular programme call, such that the dataset is divided into five cohorts for IMPULS and four cohorts each for TIP and ALFA. The main advantage of delineating the treated status by calls is that the effects of the subsidies during the first (t), second (t+1) and third (t+2) year after the start of funding can be clearly distinguished. The fourth and last call of ALFA is not included in the analysis, because of the limited availability of recent IP protection data due to the publishing delay described above. A firm may be treated in some calls of a programme but not in others. The subsidized projects typically last longer than one year, while the calls are announced annually. So if a firm is treated in one call but not in subsequent calls, this does not necessarily mean that the firm did not benefit from the programme in the latter years. It is quite likely, in other words, that a firm may be recorded as untreated for a year when it was still participating in a multi-year project supported from a previous call. In addition, theeffect of participation may extend beyond the duration of the subsidized project. Admittedly, this potential overlap is problematic for the matching procedure. Hence, to avoid matching the treated firms with themselves just in a different period, the treated firms are eliminated from the sample for the calls in which they were not treated. As a result, the control group of untreated observations only includes firms that have never been treated in the given programme. 17