Elsevier

Research Policy

Volume 37, Issue 10, December 2008, Pages 1892-1908
Research Policy

Does it matter where patent citations come from? Inventor vs. examiner citations in European patents

https://doi.org/10.1016/j.respol.2008.07.011Get rights and content

Abstract

This paper addresses the question of whether patent citations are useful indicators of technology flows. We exploit the distinction between citations added by inventors and patent examiners. We use information from the search reports of European Patent Office patent examiners to construct our dataset of patenting activity in Europe and the US, and apply various econometric models to investigate what determines the probability that a citation is added by the inventor rather than the examiner. Contrary to previous work which uses US Patent and Trademark Office data, we find that geographical distance is a factor that strongly diminishes the probability of knowledge flows. We find other significant effects of such factors as cognitive distance, time and strategic factors on citing behaviour.

Introduction

Patent citations have been used extensively as indicators of technology spillovers, and technology flows more generally. However, this is a very indirect use of patent citation data; citations are not intended to be an indication of technology flows or spillovers. They are instead a response to the legal requirement to supply a complete description of the state of the art in the field of the invention. Thus, citations limit the scope of an inventor's claim to novelty and represent a link to the pre-existing knowledge upon which the invention is built. The latter fact has been used to justify their use as indicators of knowledge spillovers. An inventor's citing of a patent or scientific article may indicate that the knowledge contained in the cited document has been useful in the development of the citing patent, and therefore that the citation might be a proxy for knowledge flow.

A criticism that has been levelled at the use of patent citations as an indicator of spillovers is that citations are a very ‘noisy’ indicator (Jaffe et al., 1998), i.e., they can be interpreted in several different ways and do not always point to the actual flow of knowledge from cited to citing inventor. A crucial factor here is that patent citations can be included by the applicant (or his/her patent lawyer) and also can be added by the patent examiner responsible for judging the degree of novelty of the patent. Where citations are added by the patent examiner, we cannot judge whether or not the applicants were aware of the cited patent. Jaffe et al. (1998) show that in many instances they were not and, hence, citation data are a ‘noisy’ indicator of spillovers or knowledge flows.

Alcacer and Gittelman (2006) joined the debate by proposing two scenarios for examiner citing behaviour: that the patent examiner might add citations that differ in nature from the inventor/applicant citations (‘gap-filling’), or that the examiner might add similar citations (‘tracking’). When patent citations are used in econometric analyses as indicators of inventor behaviour, gap-filling implies that failure to acknowledge the source of the citation may produce statistical results that are biased. Tracking does not lead to any bias but it may cause standard errors in statistical estimations to be inflated. Moreover, tracking raises doubts about patent citations as indicators of knowledge flows (Alcacer and Gittelman, 2006, p. 775). A priori, examiner citations may be taken as a valid reflection of technological and legal relatedness. But, since much of the literature argues that knowledge flows are a limited subset of potential technological relations (e.g., Jaffe et al., 1993 argue that knowledge flows are more likely where short geographical distances are involved), one would not expect inventor citations to be similar to examiner citations. If they are similar (tracking), this may indicate that these citations reflect expectations in examiners’ opinions rather than knowledge flows that played a role during the invention process.

We build on the tracking vs. gap-filling distinction by formulating a research question in terms of potential and actual spillovers. Our research question is aimed at identifying the factors that influence whether an observed patent-to-patent citation was added by the applicant/inventor. We assume a citation to be an indicator of a potential spillover or knowledge flow, and whether or not the inventor/applicant added the citation as a property indicating whether a knowledge flow actually occurred. If the Alcacer and Gittelman (2006) gap-filling model holds, we should be able to explain the occurrence of actual spillovers (i.e., inventor/applicant citation) by the existing theoretical models on spillovers. For example, if we find that geographical distance impacts negatively on the likelihood of an inventor (vs. examiner) citation, this will indicate that inventors tend to choose their citations from within a narrower geographical space than do examiners. In that case, we can conclude that inventors make a selection from the set of technologically and legally relevant citations, which is consistent with the idea of patent citations being indicators of knowledge flows.

The explanatory variables, i.e., those variables expected to have an influence on whether or not a potential spillover or technology flow occurs, are geographical distance, technological or cognitive distance (between the ‘sender’ and ‘receiver’ of the flow), time lapse between the citing and cited patents, strategic behaviour of the applicant, self-citations and international patent application. Geographical distance is the variable that has been the focus of most econometric work in the area. A large body of empirical studies has exploited the use of patent citations to assess the spatial nature of technological spillovers (Jaffe et al., 1993, Jaffe et al., 1998, Jaffe and Trajtenberg, 1996, Jaffe and Trajtenberg, 1999, Maurseth and Verspagen, 2002). These authors looked at whether or not knowledge spillovers between firms, or from (semi-) public knowledge institutes to firms, depend on geographical distance, i.e., whether citing occurs, ceteris paribus, more frequently between inventors located close to one another. The results show that in both the US and Europe, such a relationship exists. Here, we test the hypothesis that proximity between parties increases the probability of a knowledge flow.

With the exception of the European Patent Office (EPO) dataset that we used in this study and recent US Patent and Trademark Office (USPOT) data, it is not possible to identify precisely those citations chosen by the inventor. Moreover, the role of examiner vs. inventor2 citations differs among patent systems. And, ultimately, the final decision about which documents are cited in the published patent is made by the patent examiner. The patent examiner might decide to retain the citations proposed by the applicant and/or add new references, which will lead to the bias identified above that patent citations might not reflect an actual source of knowledge spillovers.

Two recent studies investigated the citations in patents granted by the USPTO. The studies by Alcacer and Gittelman (2006) and Thompson (2006) exploit the fact that, since 2001, the USPTO provides information on the source of patent citations. Thompson (2006) is aimed primarily at investigating whether or not knowledge spillovers are geographically concentrated. Alcacer and Gittelman's (2006) investigation is closer to the present study, and looks at how inventor and examiner citations differ.

In this study we explore the inventor/examiner origin of patent citations in EPO data, where it has been possible to identify the source of the citations since the EPO was established in 1979. This allows us to test whether the results obtained by Alcacer and Gittelman (2006), based on US patents, are typical of a different patent system. We would expect to find some differences for two reasons. First, the nature of the geographical space, which is different to that in the US due to cultural factors, language, the existence of national borders, etc. In order to identify more clearly the effect of the patent system, we implement estimations related to US-based inventors applying for EPO patents. Second, as Alcacer and Gittelman (2006) stress, USPTO and EPO patent examiner practices differ substantially, and particularly with regard to disclosure of prior art. This has a strong effect on the relative number of citations included by the inventor (see below), and therefore could have strong implications for the use of patent citations as a proxy for knowledge spillovers.

Section snippets

Patent citations

Patents contain references to prior patents and the scientific literature.3

Data

Our primary data sources are the EPO database on patent applications (Bulletin CD) and patent citations to other patents within the EPO over the period 1985–2000 (all citations are taken from the EPO REFI database). We complement these data with information from the OECD citations database on patent applications filed under the Patent Cooperation Treaty (PCT) and on equivalent patents (Webb et al., 2004).7

Econometric approach

As discussed in the introduction, we want to investigate the factors that explain whether a potential knowledge flow (spillover) actually occurs. The dependent variable is citation type (examiner or inventor). This is a binary variable that is equal to 1 if the citation was added by the inventor. A zero value indicates that the potential knowledge flow did not occur because the examiner and not the inventor provided the citation, a value of 1 indicates that the potential knowledge flow did

Results

Table 3 reports the descriptive statistics for the variables in the regression for the sample of within-EPO citations and the sample of within-US citations. In both these samples inventor citations are more co-localised (Distance (km) is smaller) than examiner citations: on average, inventors are more likely to cite local patents and the difference in means is quite large – especially for the within-US sample – and statistically significant. Inventors are also more likely to include citations

A closer look at the effect of distance

In Table 4, Table 5, Table 6, we (implicitly) assumed that the effect of distance is linear, but it might be the case that the relation between knowledge flows and distance is non-linear. In particular, and in line with some of the results in Table 7, we would expect that at small distances, an increase in distance by one unit (1 km) would lead to a stronger effect of the likelihood of an inventor citation, than the same increase over a longer distance. In order to test for this, we employ a

Conclusions

The European patent database allows identification of whether citations are added by the applicant/inventor (inventor citations) or the patent examiner for all of its patents. Moreover, since the legal requirements for citing prior art differ between the European and US systems, we expect the indication of knowledge spillovers based on the patent citations in our database to be different from the USPTO citation studies that dominate the literature. On the basis of the EPO database, we asked

Acknowledgements

The authors would like to thank three anonymous referees, Gustavo Crespi, Aldo Geuna, Colin Webb and Joe Hilbe for helpful comments and Grid Thoma and Salvatore Torrisi for letting us use their dataset. Paola Criscuolo acknowledges financial support from the Engineering and Physical Sciences Research Council's IMRC at Imperial College London. The usual disclaimer applies.

References (33)

  • E. Duguet et al.

    How well do patent citations measure flows of technology? Evidence from French innovation surveys

    Economics of Innovation and New Technologies

    (2005)
  • Z. Griliches

    The search for R&D spillovers

    Scandinavian Journal of Economics

    (1992)
  • J.A. Hausman et al.

    Nonparametric estimation of exact consumers surplus and deadweight loss

    Econometrica

    (1995)
  • Hedge, D., Sampat, B.N., 2005. Examiner citations, applicant citations and the private value of patents. Mimeo Columbia...
  • Jaffe, A.B., Trajtenberg, M., 1996. Flows of knowledge from universities and federal labs: modelling the flow of patent...
  • A.B. Jaffe et al.

    International knowledge flows: evidence from patent citations

    Economics of Innovation and New Technologies

    (1999)
  • Cited by (234)

    • Big Pharma and monopoly capitalism: A long-term view

      2023, Structural Change and Economic Dynamics
    • Women in top management teams and their impact on innovation

      2022, Technological Forecasting and Social Change
    View all citing articles on Scopus
    View full text