Faculty patent assignment in the Chinese mainland: evidence from the top 35 patent application universities

Abstract

Recent research has explored the issue of university faculty patent assignment in the US, Europe and Japan from the individual or organization perspectives. However, there is limited empirical research that examines the real picture of faculty patent assignment in China’s universities. This paper aimed to fill this gap by creating a special dataset including 18,435 faculty/patent pairs. The investigation indicated that 13.16 % of pairs were not solely assigned to universities in 35 top patent application Chinese universities from 2002 to 2012. The empirical study correlates types of patent assignment to invention characteristics, university intellectual eminence and licensing policies, and illustrates that patent assignment changes depending on the research field, that university assignment is positively related to patent claims but negatively related to patent validity, patent maintenance time, and number of co-inventors, and that university intellectual eminence has a weak impact. Through controlling the influence of inventor characteristics, university royalty and equality policies play different roles in faculty patent assignment. This paper provides new insights as well as operational policy implications for China’s university policy makers.

Appendix: Data collection and screening method

Figure 3 gives details about data collection in this paper. First, we collected all professors’ individual information from their CVs in websites on three research fields in the 35 Chinese universities in our sample. The individual information includes date of birth, gender, research field, work experience, the date when they acquired the title of professor, and their experience of administrative position. Note that in this paper we classify research field according to the titles of the university departments (e.g. schools of mechanical engineering, schools of telecommunications, or schools of life science).

Fig. 3

Process of data collection

Second, we collected patent documents from the CNKI database by faculty names, and then matched faculty names with first-inventor names. To ensure that the sample included only patents with faculty who were truly employed by universities on the dates of their patent applications, a screening method was adopted that aimed at eliminating inventors who were not true faculty inventors: (1) in our dataset, faculty with no patents were eliminated. In order to avoid name repetition in two or more universities, we excluded patent/faculty pairs if the faculty names appeared in different universities but in the same city; (2) we excluded patent/faculty pairs if name repetition occurred in two or more departments, regardless of whether the faculty had a part-time job in another department; (3) when a faculty name appeared in both the university and an external enterprise, and the first four figures of the zip code in the patent document was not the same as that of the university, the patent/faculty pair was excluded. Finally, (4) for a given faculty inventor, we checked every one of his/her patents according to the patent classification code. We excluded those patents that were significantly different from others. In the end, 18,435 faculty/patent pairs were created in our tailor-made dataset.

It is worth noting that our dataset is at the patent/inventor level. Because a single patent often has more than one inventor, we define a specific patent as the “faculty patent” only where the university faculty is the first inventor, and thus collect faculty patents according to the faculty name when faculty inventor is the first inventor in the patent document. This means that each patent appears only once in our dataset. In this way, patents in which the university faculty inventor takes part in R&D activities and is the second or third inventor will be excluded. Thus, we admit that our study has neglected some of the R&D collaboration activities between faculty and private firms. Perhaps the real proportion of university-firm assignment is higher than we claimed in our paper. In addition, due to the limited information on faculty patents, we do not have information about co-inventors. An obvious limitation of this data presentation method is that we cannot calculate how many industrial firm employees are listed in faculty patent documents.