The importance of firm level multinationality in the country versus industry debate

• DOI: http://doi.org/10.1002/ijfe.1597
• Author: Jenny Berrill,Cormac Mullen
• Published date: 01 October 2017
• Date: 01 October 2017

RESEARCH ARTICLE

The importance of firm level multinationality in the country

versus industry debate

Cormac Mullen | Jenny Berrill

School of Business, Trinity College Dublin,

Dublin 2, Ireland

Correspondence

Cormac Mullen, School of Business,

Trinity College Dublin, Dublin 2, Ireland.

Email: mullenco@tcd.ie

JEL Codes: F23; G11; G15sss

Abstract

We conduct the most comprehensive empirical analysis that exists to date of the

effect multinationality has on the explanatory power of country and industry

factors in international diversification. We investigate the impact the size,

scope, and location of a company's international sales has on country versus

industry factors, analysing 1,276 firms from Belgium, Canada, France,

Germany, Italy, Japan, the Netherlands, Spain, the UK, and the US over the

15‐year period, 1998–2012. We find that the magnitude of the country factor

is greater than the magnitude of the industry factor for the period as a whole

but that a company's level of international sales has a greater impact on the

magnitude of its industry factor than the magnitude of its country factor.

Counter‐intuitively, we find stocks with lower sales exposure to their country

of origin have a higher country factor, and we show the existence of both

a strong local and international industry factor. Our results suggest country‐

of‐origin diversification may no longer be sufficient to exploit country‐specific

risk and the country factor has become a “country classification”factor.

KEYWORDS

Country effects, equities, industry effects, internationalisation, international portfolio diversification,

multinationality

1|INTRODUCTION

The debate on the comparative merits of allocating an

equity portfolio by country or by industry has been ongoing

for over 40 years. On one side, researchers claim country

factors are most important (Baele & Inghelbrecht,

2009; Bai, Green, & Leger, 2012; Heston & Rouwenhorst,

1994), others claim industry factors dominate (Balli &

Balli, 2011; Eiling, Gerard, & De Roon, 2012; Flavin,

2004). A number of research papers, particularly in the

2000s, proposing the benefits of diversifying an interna-

tional equity portfolio across countries is decreasing due

to a rise in sector effects and greater financial market inte-

gration, most notably in Europe (Brooks & Del Negro,

2004; Grisolia & Navone, 2008; Phylaktis & Xia, 2006b;

Steliaros & Thomas, 2006). However, of the 87 papers

1

studied on this debate, just two examine the impact of firm

level multinationalisation on the debate (Brooks & Del

Negro, 2006; Peterson LaBarge, 2008) and both papers

use static measures of multinationality over a dynamic

time frame. Neither considers multinationality as an evolv-

ing process with changing levels of internationalisation

among firms. We investigate the impact the size, scope,

and location of a company's international sales has on

the return variations of equity portfolios diversified across

both country and industry classifications.

Stocks are traditionally classified by country, using the

location of their primary stock market listing, and by

industry sector. Coca‐Cola, for example, is classified as

both a US and a consumer goods stock under the

Received: 11 February 2017 Accepted: 23 September 2017

DOI: 10.1002/ijfe.1597

Int J Fin Econ. 2017;22:403–420. Copyright © 2017 John Wiley & Sons, Ltd.wileyonlinelibrary.com/journal/ijfe 403

Financial Times Stock Exchange (FTSE) International

Industry Classification Benchmark system. Investors use

these classifications to construct portfolios to obtain or

avoid exposure to particular geographies or industry sec-

tors and academics use them to calculate the comparative

benefits of allocating an equity portfolio by country versus

industry. But how efficient is it to classify multinational

firms by their location of primary listing? How American

is Coca‐Cola to a fund manager when Coke is sold in

more than 200 countries around the world?

2

How does

Coca‐Cola impact the calculation of a US country effect

when it is among the most global stocks in the world?

We construct a novel dataset of stock returns and

international sales data for 1,276 companies from 10

countries over a 15‐year period from 1998 to 2012. We com-

pile annual segmental sales data unique to each firm and

allocate it to six geographic regions, North America, South

America, Europe, Asia, Oceania, and Africa & Middle East.

Our dataset is among the most comprehensive longitudi-

nal compilations of international sales data in the interna-

tional business literature to date. It enables us to examine

the degree of internationalisation of the firms in our

sample on an annual basis through three measurements

of multinationality. We use percentage of foreign sales data

to measure a firm's level of multinationality, the number

of regions a firm sells into to measure scope of

multinationality and also examine the regional location

of those sales. We incorporate these measures to classify a

firm's degree of multinationality using a method similar

to Aggarwal, Berrill, Hutson, and Kearney (2011) and one

similar to Rugman and Brain (2003).

We apply the linear factor model approach of Heston

and Rouwenhorst (1994) to analyse the explanatory effect

of country and industry classifications on underlying

stock returns. We examine how the magnitude of country

and industry factors differ as company's international

sales level increases and analyse the results for each coun-

try and industry across the categories of multinational

firm from our classification methodologies. In doing so,

we conduct the most comprehensive analysis that exists

in the international finance literature to date of the effect

multinationality has on the explanatory power of country

and industry factors on stock returns.

The literature to date highlights the dominance of

country factors in explaining stock returns. We find

approximately half of 87 papers studied on this topic

and show country factors are dominant (44 papers) with

just under a quarter concluding that industry factors dom-

inate

3

(23 papers). Our results show that the magnitude of

the country factor is greater than the magnitude of the

industry factor in 10 out of 15 years and for the period

as a whole. However, we show that country factors

outweighing industry factors depends on the firm's degree

of multinationality and provide evidence of both a local

and global industry influence. We find international sales

levels have a larger effect on the explanatory power of

industry effects on returns than they do on country effects

and that the magnitude of the industry factor is higher

than the country factorfor firms with both the greatest geo-

graphical scope and the greatest locational reach. Counter‐

intuitively, we show that stocks with lower sales exposure

to their country of origin have a higher country factor,

higher even than pure domestic stocks. This suggests that

the country effect has become a “country classification”

effect, divorcedfrom exposure to the underlying fundamen-

tals of the country itself. Our subperiod analysis confirms

our main findings. We conclude that fund managers who

choose countries to classify their portfolio should be aware

that stocks with the highest international sales scope may

counter‐intuitively have a higher sensitivity to country

factors than domestic companies. Fund managers who

diversify by industry sector should note that domestic

stocks and companieswith high scope of international sales

show a higher sensitivity to the industry factor than those

with a low–medium scope of international sales.

The remainder of the paper is structured as follows.

Section 2 contains a review of the relevant literature,

Section 3 contains an explanation of the data used and

methodology. Section 4 details the results of the analysis,

Section 5 presents our conclusions.

2|LITERATURE REVIEW

Lessard (1974) was one of the first to investigate the influ-

ences that a company'scountry‐of‐origin and industry sector

had on its stock return and its potential for risk‐reduction

through international diversification. He finds the country

dimension much more important than the industry dimen-

sion in a diversification strategy. Grinold, Rudd, and Stefek

(1989) show that local market factors account for more stock

variance than industry factors in approximately 90% of the

months in their sample. Drummen and Zimmermann

(1992) are more specific. In an analysis of European stock

returns, they calculate that 19% of average stock variance

can be explained by country‐of‐origin factors versus 9% for

industry factors with world and regional factors accounting

for 11% and 8%, respectively.

In the early 1990s, Heston and Rouwenhorst (1994)

introduced a dummy factor model approach to determin-

ing the relative importance of country and industry classi-

fications in stock returns, an approach which has been

used extensively (Bai et al., 2012; Bekaert, Harvey,

Lundblad, & Siegel, 2011; Chong, Hooy, & Lee, 2013;

Miralles‐Marcelo, Miralles‐Quirós, & Martins, 2013). They

suggest that diversification across countries within an

404 MULLEN AND BERRILL