Learning remotely: R&D satellites, intra‐firm linkages, and knowledge sourcing

• Author: Joel Blit
• Published date: 01 December 2017
• Date: 01 December 2017
• DOI: http://doi.org/10.1111/jems.12213

Received: 12 June 2014 Revised: 4 May 2017 Accepted: 8 May 2017

DOI: 10.1111/jems.12213

ORIGINAL ARTICLE

Learning remotely: R&D satellites, intra-firm linkages,

and knowledge sourcing

Joel Blit

Department of Economics, University of

Waterloo,200 University Ave. W.,Waterloo,

ON, Canada (Email: jblit@uwaterloo.ca)

Abstract

Firms venture abroad not only to access resources and markets but also to learn.

Yet there remains limited empirical evidence that headquarters can access geographi-

cally remote knowledge byestablishing a presence in the remote location. Using U.S.

patent data, I show that firm headquarters disproportionately source knowledge from

third parties in remote locations where they have an R&D satellite. This “satellite

effect” on knowledge flow is economically significant, representing up to 60% of the

knowledge-flow premium associated with collocation. Furthermore, the effect seems

to be stronger for recent knowledge, as well as in areas of satellite technological spe-

cialization, suggesting that firms can target cutting-edge knowledgein specific sectors.

In addition, the results show that firms with stronger internal linkages between head-

quarters and satellites, and those that staff satellites with inventors that previously

patented while at other local firms, experience a larger satellite effect on knowledge

acquisition.

1INTRODUCTION

A principal way that firms build competitive advantage is through innovation. To be successful, firms must both pursue incre-

mental improvements that build upon their existing stock of knowledge (exploitation) and explore broader avenues in search of

breakthrough advances (March, 1991). Yet, the latter can be particularly problematic for firms. As Singh and Fleming (2010)

show, breakthrough innovation requires access to diverse sources of knowledge. Because these sources of knowledge are often

external and can reside in geographically distant locations, firms may need to put in place mechanisms that facilitate access.1

One such mechanism to access distant external knowledge is the establishment of a satellite R&D unit in the remote location

(knowledge-seeking foreign direct investment [FDI] in the international context). But how large is the impact of the R&D satel-

lite on the headquarters’ acquisition of remote third-party knowledge? Is the knowledge acquired primarily cutting edge, or

mature and, hence, of potentially lower value? How does the satellite facilitate knowledge acquisition? This paper sets out to

empirically address these questions.

The idea that firms can tap into remote knowledge through the strategic deployment of subsidiaries is not new (Alcacer &

Chung, 2002; Cantwell, 1993; Feinberg & Gupta, 2004; Kuemmerle, 1999). Indeed, firms like Hitachi and BMW have well-

established knowledge-seeking R&D units in Cambridge, U.K., and Silicon Valley, respectively (Boutellier, Gassmann, & Von

I am grateful to Ajay Agrawal, Ignatius Horstmann, Joanne Oxley, and Daniel Trefler for their indispensable guidance. The seed idea for this paper originates

from a discussion with Gilles Duranton. Alex Oettl generously provided some of my data. I also would like to thank Wilbur Chung, Steve Klepper, Carlos

Serrano, Peter Thompson, and others who pushed me to produce a better paper through their insightful critiques. This paper also benefited greatly from the

insightful comments of two anonymous referees. I gratefully acknowledge funding from the NBER Productivity, Innovation, and Entrepreneurship Program,

the Social Sciences and Humanities Research Council of Canada, and from the Martin Prosperity Institute’s Program on Innovation and Creative Industries.

Any errors and omissions are my own.

J Econ Manage Strat. 2017;26:757–781. © 2017 WileyPeriodicals, Inc. 757wileyonlinelibrary.com/journal/jems

758 JOURNAL OF ECONOMICS & MANAGEMENTSTRATEGY

Zedtwitz, 2008). Because knowledge diffusion is partially localized (Audretsch & Feldman, 1996; Branstetter, 2001; Jaffe,

Trajtenberg, & Henderson, 1993; Thompson, 2006), collocating a satellite R&D unit with an important source of knowledge

can be an effective way to access that remote knowledge. This is particularly true for knowledge that is tacit in nature and hence

disseminated through worker mobility or social networks (Almeida & Kogut, 1999; Breschi & Lissoni, 2009). Indeed, strong

evidence suggests that remote subsidiaries effectively embed themselves in local networks and make extensive use of local

knowledge (Almeida, 1996; Frost, 2001; Singh, 2007). What is less clear, however, is whether the headquarters also taps into

this remote knowledge through the medium of their subsidiary.

This paper addresses this question using patent citation data to track the flow of knowledge. Unlike the previous literature

that has focused on international knowledge flows, this paper focuses on knowledge flows across different regions within the

U.S., and it does so for two reasons. First, the U.S. offers a geographically large but relatively culturally and institutionally

homogeneous setting that allows us to focus on the satellite effect on knowledge flows in isolation from the potential effects

of cross-country developmental, cultural, and institutional differences. Second, as it is not clear that establishing a subsidiary

in Boston would facilitate access to knowledge from San Francisco, the relevant level of analysis needs to be at the level of the

city/region.2We would expect,however, that any findings of a satellite effect on knowledge diffusion within the U.S. wouldalso

apply to international knowledge flows, though the magnitudes might perhaps depend in part on the characteristics of country

pairs.

The analysis focuses on U.S. firms with R&D centers in two different U.S.metropolitan regions. I quantify the satellite effect

on knowledge acquisition using a matching methodology in the spirit of Jaffe et al. (1993). To partially address endogeneity

in the choice of satellite location and omitted variable bias, I employ a difference-in-differences approach. I show that prior

to the establishment of the satellite, the headquarters do not disproportionately source third-party knowledge from the satellite

location. However,it does dispropor tionatelysource knowledge from the remote location after the satellite has been established.

Moreover, this finding is not driven by firms that establish satellites to shift into new sectors. Nonetheless, it should be noted

that the analysis is limited to exploring the effect of satellites for firms that have chosen to establish a satellite and the results are

therefore only suggestive of the potential benefits for firms that have not.

The principal result is that when a firm has a satellite research center in a remote location, its R&D headquarter patents

cite third-party patents from the remote location 40%–60% more often than would be expected given the geographic dis-

tribution of innovative activity (accounted for by a set of control patents). To offer a benchmark, a separate analysis is

conducted to determine the extent to which patents are more likely to cite patents in their same location (the collocation

effect). Patents cite third-party patents in their same region 66%–187% more often than would be expected given the geo-

graphic distribution of innovative activity. Clearly then, knowledge sourcing through a satellite cannot fully substitute for

being collocated with an important source of knowledge, but it does account for between 32% and 60% of the knowledge-

flow premium associated with collocation and an even higher fraction in those technological areas where the R&D satellite is

active.

Collocation with a source of knowledge also confers the benefit of timely access to knowledge. We would similarly expect

the role of satellites to be most significant in facilitating access to recently developed knowledge. Indeed, this paper presents

some evidence that the knowledge received from the satellite location is disproportionately recent, though these results are

inconclusive.

A final set of results begins to address the mechanism underlying the satellite effect by examining heterogeneous firm

responses. Because effective transmission of knowledge from the satellite to the headquarters is crucial, firms with stronger

linkages between the two locations should experience a larger satellite effect on knowledge acquisition. Similarly,satellites with

stronger links to other firms in their location will gather more knowledge and hence be stronger channels for knowledge dif-

fusion. I confirm these hypotheses by finding that the magnitude of the satellite effect is positively related to the intensity of

cross-location patent coauthorship, the intensity of remote self-citations, and the fraction of satellite inventors that previously

patented at other firms in the satellite location.

Taken together, the findings suggest that we need to temper the view that high-tech firms should locate

within technology clusters to be successful innovators. From the point of view of knowledge sourcing, the

establishment of a satellite is an effective (if imperfect) substitute for collocation with an important source of

knowledge.

The paper is organized as follows. Section 2 reviews the literature and discusses the theoretical underpinnings behind

R&D satellites as channels for knowledge sourcing. Sections 3 and 4 describe the data and methodology. Section 5 presents

the empirical results related to the existence of a satellite effect, studies whether knowledge sourced through the satel-

lite is primarily recent, and examines firm characteristics that influence the magnitude of the satellite effect. Section 6

concludes.