Collection and Processing of the Data

For our analysis we needed two types of data. First, the patent data of the inventions in the fields of e-mobility and fuel cells made in the region of Stuttgart. Second, citations to these patents. Accordingly, two datasets were prepared. Each consisted of two parts: the information about e-mobility and fuel cell technologies since they are retrieved and processed separately. The first dataset is the patent dataset. It

Table 2 Relevant e-mobility patent classes

Source: Karl and Jager (2011)

includes information about the patents, protecting inventions in the field of chosen technologies made by the firms and research institutions in the region of Stuttgart. The dataset represents a table of all the patents registered at the EPO and includes the following information about every patent retrieved: applicant name, regional code, person ID, address, IPC of the invention, year of the priority date and an application ID. The second dataset is referred to as patent citation dataset. It is made on the basis of the patent dataset. It consists of the following information about backward citations of the patents from the first dataset: citing applicant ID, personal ID, the name of the cited company, address, country code, regional code, year of the priority date, applicant name (the citing company). The information for the first dataset is collected on the basis of IPC classes related to e-mobility and fuel cell technology. The IPC classes related to e-mobility technology are presented in Table 2 and those related to fuel cell technology in Table 3.

The retrieved patents have been filtered by regional codes. As described in Sect. 1, the region of Stuttgart includes the city of Stuttgart itself and a number of neighboring counties. Each county has its regional code (NUTS 3 level). The codes from DE111 to DE119 form the Stuttgart region. Since the aim of our research is to study the regional knowledge outflows, citations made by firms in the region of

Table 3 Relevant fuel cell patent classes

Source: Deutsches Patent und Markenamt

Stuttgart are excluded from the analysis. This also automatically excludes selfcitations (an inventor in his new patent cites his previous inventions). Self-citations obviously do not depict spillovers (Jaffe et al. 1993).

In this way, we have obtained the datasets that we were looking for. The first dataset consists of patent information including 15,265 patents related to e-mobility technology registered in the period from 1977 to 2011. Fuel cell technology appears to provide much less inventions, since only 192 patents have been registered from 1990 till 2010. Note, the time domain is different for these two technologies due to the fact that fuel cells is as an applied technology relatively young.

The second dataset includes information about citations of patents, found during the first search. There are 4184 citations that reference the e-mobility technologies made from 1978 to 2009. For fuel cell technologies, from 1990 till 2002 we have retrieved 45 citations all over the world (excluding the region of Stuttgart). The number is much smaller than for e-mobility. The small number of citations is related to the small number of patents in this field. It shows that the innovation input in this technological field, such R&D activities, is not very high. This can be explained by the fact that the technology is very narrow and also new to the industry, and that is why not many firms are involved in the research in this area and it is not their top-preference research activity. Still, patent (citation) analysis constitutes a useful methodology to identify the pioneers in this field. We rely on patent citations that show the geographical localization of knowledge outflows. However, while a comparison of the intensity of knowledge flows between the countries is meaningful, the numbers should not be overstated. In order to assess the impact of each firm in the technological development of the region, we calculate how many times the country has been cited.