ABSTRACT

The first step in our data collection was aimed at identifying the object of our analysis, i.e. new firms that innovate (i.e. apply for a patent) within a relatively short time after their foundation. Using the EP-Cespri database, which contains data on patent applications to the European Patent Office (EPO) from 1978, we collected data on all firms that patented for the first time in either the laser technology field or in the field of data-switching networks between 1990 and 2005. The two technology fields correspond to well-defined International Patent

Classification (IPC) codes, respectively H01S and H04L12. From the sample thus generated, we then extracted all companies located either in the United States or in one of the following European countries: Germany, the UK, France, Italy, Sweden and Denmark. Starting from this sample, we collected further information regarding the history of each firm and its legal status. In fact, many of the companies that start patenting each year are not truly new ventures, but most often they are divisions/ subsidiaries of long established companies, joint ventures, firms resulting from mergers of already existing companies, and so on. To sort out innovative entrants from other types of patenting organizations, we went through all the first-time patenting firms and searched for their establishment date and their year of exit from the market (if any) using both international company databases (e.g. AMADEUS), national company databases (AIDA, DIANE, FAME, etc.) and Internetbased searches. Since our focus is on new and innovative firms, we restricted our sample according to the following two criteria. First, we used all the abovementioned sources to identify firms belonging to categories that were not relevant to the object of the research, namely categories that did not respond to the definition of new firms (i.e. firms that are truly new ventures). All firms that turned out to be subsidiaries/divisions of established firms or clearly owned by other institutions (e.g. universities) have been excluded from the analysis. Second, we assumed that truly innovative new entrants are more likely to have applied for their first patent shortly after their foundation; hence we restricted our sample to firms established no earlier than five years before their first patent application at the EPO. The resultant sample includes 49 laser and 184 data-switching networks firms, all of which turn out to be established between 1990 and 2003.1 A large share of these firms is located in the United States (31 laser and 129 dataswitching networks firms, respectively). This raises a major concern as US firms might first apply for a patent at the USPTO and apply at the EPO only at a later stage. This implies that for all the US firms in our sample we could not be sure that their first patent at the EPO was their first patent overall. This could generate two sources of bias. First, we might treat as new entrants in either the laser or data-switching networks technologies firms already active in other technologies, which diversify in the two technologies of interest. Second, since we aim at studying the impact of the technological position of the innovative firm at the time of entry in the market as proxied by characteristics of the first patent (as we shall explain later), we would then mistakenly proxy a firm’s initial technological position using a patent farther in time from the beginning of activity. This could be misleading also if the firm’s true first patent were in one of the two relevant technologies. For these reasons we searched the USPTO database for all the US firms in our sample and excluded all the firms that were found to have applied for a patent at the USPTO before their first patent application at the EPO (based on this patent priority date). Firms with no previous patent applications at the USPTO or with a first patent application at the USPTO equivalent to the first EPO application are instead included in our final sample. We also keep US firms

for which we find at least one patent application at the USPTO in the same year as the priority year of their first patent application at the EPO, even if this other application is not in the two technologies of interest and keep track of this information, which might signal that the new firm enters the market with a wider knowledge base. After excluding from the sample all US firms that did not satisfy our selection criteria, we are left with our final sample, which includes 39 laser firms (21 located in the United States) and 152 data-switching networks firms (97 located in the United States). For each of these firms, using again Internet-based searches, we finally identified at least one founder.