Does it pay to be a multinational? A large‐sample, cross‐national replication assessing the multinationality–performance relationship

• Published date: 01 January 2020
• Date: 01 January 2020
• DOI: http://doi.org/10.1002/smj.3087
• Author: Niccolò Pisani,Eelke Heemskerk,Javier Garcia‐Bernardo

RESEARCH ARTICLE

Does it pay to be a multinational? A large-sample,

cross-national replication assessing the

multinationality–performance relationship

Niccolò Pisani

1

| Javier Garcia-Bernardo

2

| Eelke Heemskerk

2

1

Amsterdam Business School, University of Amsterdam, Amsterdam, The Netherlands

2

CORPNET, Amsterdam Institute for Social Science Research, University of Amsterdam, Amsterdam, The Netherlands

Correspondence

Niccolò Pisani, Amsterdam Business

School, University of Amsterdam,

Amsterdam, The Netherlands.

Email: n.pisani@uva.nl

Funding information

European Union's Horizon 2020 Research

and Innovation Programme, Grant/Award

Number: 638946

Abstract

Research Summary:Does it pay to be a multinational?

Despite decades of empirical research, we still do not know.

We undertake a large-sample, cross-national replication of Lu

and Beamish (2004) and Berry and Kaul's (2016) works to

examine whether the multinationality–performance relation-

ship is S-shaped in a 2009–2016 panel of 889,865 firm-year

observations. Using a two-stage least squares fixed-effects

model that accounts for endogeneity on a subsample of

32,835 multinationals from 64 countries, we find no evidence

of an S-shaped relationship; nor do we see it in any of the

single-country contexts. Our results show no evidence of

any within-firm effect of multinationality on performance,

highlighting the need for more contextually-grounded

research focused on explaining between-firm effects to

advance our theoretical and empirical understanding of the

multinationality-performance relationship.

Managerial Summary:We replicate two studies that

examined the relationship between a firm's multinationality

and its performance. Lu and Beamish (2004) found evi-

dence of an S-shaped relationship—with performance first

decreasing, then increasing, then decreasing again as firms

expanded abroad—in a sample of Japanese firms; Ber ry and

Received: 18 March 2018 Revised: 28 July 2019 Accepted: 3 August 2019 Published on: 6 November 2019

DOI: 10.1002/smj.3087

This is an open access article under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs License, which permits use and

distribution in any medium, provided the original work is properly cited, the use is non-commercial and no modifications or adaptations

are made.

© 2019 The Authors. Strategic Management Journal published by Wiley & Sons, Ltd. on behalf of Strategic Management Society.

152 Strat. Mgmt. J. 2020;41:152–172.wileyonlinelibrary.com/journal/smj

Kaul (2016) found no evidence of an S-shaped pattern in a

sample of U.S. multinationals. We test for the same relation-

ship using data from nearly 250,000 firms based in over

100 countries from 2009 to 2016 and find no evidence of an

S-shaped pattern or of any effect of multinationality. Our

study thus adds substantial evidence to the one shown by

Berry and Kaul (2016), emphasizing the need to focus on

how contextual differences influence the multinationality–

performance relationship.

KEYWORDS

firm performance, multinationality, panel data, replication, S-curve

1|INTRODUCTION

Whether and how multinationality affects firm performance represents a crucial question in the strate-

gic management and international business fields. A substantial body of scholarly work has

attempted to empirically verify whether and, if so, how a firm's multinationality

1

(M)—that is, the

extent to which it undertakes value-adding activities outside its home country—impacts performance

(P). Despite these ongoing efforts, the literature has to date failed to offer a consistent set of empirical

findings (Cardinal, Miller, & Palich, 2011; Hitt, Tihanyi, Miller, & Connelly, 2006; Kirca et al.,

2011). The M–P relationship has been found to be insignificant or very weak (Tallman & Li, 1996),

negative (Denis, Denis, & Yost, 2002), U-shaped (Lu & Beamish, 2001), inverted U-shaped

(Gomes & Ramaswamy, 1999), and S-curved (Contractor, Kumar, & Kundu, 2007). Thus, we still

do not know whether it pays to be a multinational.

A particularly influential contribution to the debate is by Lu and Beamish (2004, henceforth

L&B).

2

Studying a sample of 1,489 Japanese listed firms over 12 years, they found an S-curved rela-

tionship between multinationality and performance. Even though L&B explicitly discussed the speci-

ficities of the (single-country) context used in their empirical analysis, their convincing theoretical

explanation and rigorous methodological approach has made their S-curve hypothesis widely

accepted and assumed to apply in a variety of (country) contexts. In view of the influence of L&B's

contribution and the inconsistent results in the literature, Berry and Kaul (2016, p. 2,276; henceforth

B&K) recently argued that “careful, large sample replication of L&B's study in other samples and

settings is therefore critical.”Thus, they conducted a replication of L&B's study using a panel of

2,023 U.S. multinational firms from 1989 to 2007. However, they were unable to replicate L&B's

results; the little evidence they obtained in support of an S-shaped relationship in a subsample of

manufacturing firms vanished once they accounted for the endogeneity of multinationality.

1

As noted by Kirca et al. (2011) in their meta-analytic review and echoed by Berry and Kaul (2016), “multinationality,”

“degree of internationalization,”“internationaldiversification,”“geographicdiversification,”and “international expansion”

refer to the same construct. In our work, as also clarified in the main text, we use the term multinationality to refer to the extent

of a firm's operations outside its home country.

2

Verbeke and Forootan (2012) listed L&B's contribution as one of the 10 most influential studies in this literature, having

accumulated at the time of their analysis 346 Google scholar citations. By February 2019 L&B's study had reached 1,206

google scholar citations, thus further corroborating its influence in the literature.

PISANI ET AL.153