Firms’ loose and tight interactions with university scientific codified knowledge and the economic cycle
The relation between economic growth and economic cycles with innovation has deserved much attention, but not with a particular driver of innovation, academic knowledge spillovers. We do not know yet the role of the economic cycle and, particularly, the effect of economic crises, on academic knowledge spillovers and their localization. In this paper, we hypothesise that economic crises diminish academic knowledge spillovers, and that local academic knowledge spillovers diminish to a larger extent than non-local academic knowledge spillovers. We find evidence in favour of these hypotheses using patent data on Spanish firm patent citations to national and international university patents. The data covers 2000-2014, which includes a crisis period, 2008-2014, the beginning of the Great Recession.
An enduring stream of literature suggests a link between academic and industrial codified knowledge, namely ‘academic knowledge spillovers’, which also constitute an engine of innovation (Jaffe et al., 1992; Mowery and Ziedonis, 2015). An academic knowledge spillover (or flow) is the use of published codified information from universities by firms as a source of inspiration to build new technologies. Academic knowledge spillovers from university to industry can take place through a variety of channels between academics and firms (when reading scientific papers or via direct conversation or informal meetings with the inventors, etc.). In this paper, we focus on one of them: citations contained in firm patents, as an indicator that the university patent cited by the firm has been useful to generate that piece of technological knowledge.
Little is known about the effect of economic cycles on academic knowledge spillovers and their localization, which becomes a particularly relevant topic given the recent Great Recession started in 2008. In expansions and recessions of the economy, framework conditions and incentive systems for university-industry interaction change, and this is likely to affect university-industry knowledge flows. In this paper we focus on the effects of macrolevel economic cycles on the intensity and localization of academic knowledge flows.
2. HYPOTHESES BUILDING
Hypothesis 1. GDP growth decreases academic knowledge spillovers, up to a threshold from which GDP growth increases academic knowledge spillovers
Hypothesis 2. GDP growth increases academic knowledge spillovers during expansions, and decreases academic knowledge spillovers during crises
Hypothesis 3a. GDP growth increases delocalization of academic knowledge spillovers during expansions, and increases the localization of academic knowledge spillovers during crises
Hypothesis 3b. GDP growth increases localization of academic knowledge spillovers during expansions, and decreases the localization of academic knowledge spillovers during crises
3. METHODOLOGY AND DATA
The primary source of patent information is the European Patent Office (EPO) Worldwide Patent Statistical Database (PATSTAT 2017, Spring Edition), produced by the EPO, OECD and Eurostat. Patstat includes patents from more than 90 national and international patent offices and provides details such as technology field, priority date, backward/forward citations and family links.
The methodology to analyze academic knowledge spillovers is based essentially on the analysis of the information retrieved from a large-scale database (Patstat), and involves the following steps:
1. Retrieve patents applications by companies to the EPO with priority date between 2000-2014.
2. Gather citations to previous patent documents (backward citations).
Given the binary nature of our dependent variable, in order to test Hypothesis 1 we put forward a logit model where the dependent variable KS, captures knowledge spillovers. The empirical panel model takes the form:
Where i represents application; j represents family, k refer to citation, l refer to applicant and t indicates time.
AKS represents knowledge spillovers: takes value 1 if the cited patent belongs to a university, 0 otherwise. As explanatory varibles we include GDP growth rate, lagged n years (ΔGDP) and θ includes a set of control variables: Application authority (appln_authi), applicaation kind (appln_kindi), earliest_filing_yeari, number of patent applications in the family (docdbfamily_sizej) and citation origin, that includes information about when the citation was added(by the applicant, during search, opposition, etc).
In order to test Hypothesis 2, related to the localisation of academic knowledge spillovers, we use Local(AKSijklt) as our dependent variable. This variable equals 1 if the citing and the cited applicant share the same country code; 0 otherwise.
Our econometric models show evidence in favour of hypotheses 1, 2 and 3b, not 3a.
This works reveals a novel aspect of academic knowledge spillovers, namely their reaction to economic growth and the economic cycle. Academic knowledge spillovers increase with the economy, but only once the macroecomic context has reached a minimum level of growth. In a crisis, economic growth does not reach this threshold, and the virtuous circle with academic knowledge spillovers becomes vicious. This is a new aspect that adds to the negative effects of crisis.
Economic crises also change the relation between economic growth and localization of academic knowledge spillovers. In expansions, GDP increases translate into the localization of those spillovers, due to increased university research capacity, than makes local universities flourish and become attractive. On the contrary, in crisis, GDP increases lead to delocalization of academic knowledge spillovers, due to the higher capacity to resist the crisis of universities with long-standing international reputation. This gives additional arguments for the protection during crises of small local universities with good spillover record during expansions.