Designing Matrix Structures to Fit MNC Strategy

• Published date: 01 August 2013
• Author: William G. Egelhoff,Joachim Wolf,Mihael Adzic
• Date: 01 August 2013
• DOI: http://doi.org/10.1111/j.2042-5805.2013.01059.x

DESIGNING MATRIX STRUCTURES TO FIT

MNC STRATEGY

WILLIAM G. EGELHOFF,1JOACHIM WOLF,2* and MIHAEL ADZIC3

1Graduate School of Business, Fordham University, New York, New York, U.S.A.

2Institute of Business Administration, University of Kiel, Kiel, Germany

3Adzic & Partner, Berlin, Germany

Matrix structures and network designs are the primary organizationalalternatives available to

MNCs to implement increasingly complex multidimensional strategies. Despite a growing

interest in matrix structures among MNC managers, theory about how to design and fit the

different types of matrix structure to MNC strategy is largely absent from the scholarly

literature. To address this gap, the present study develops an information-processing model of

strategy-structure fit for MNC matrix structures. It does this by modifying and extending logic

already established for fitting elementary structures to MNC strategy.Hypotheses based on the

model are empirically supported, using a sample of 57 German MNCs with matrix structures.

Multivariate discriminant analysis further reveals that an important element of strategy—the

technological similarity of products in an MNC—is still missing from the literature and

proposed model. Copyright © 2013 Strategic Management Society.

INTRODUCTION

Matrix structures were once embraced as the best way

to organize multinational corporations (MNCs) to

deal with increasingly complex international strate-

gies (Davis and Lawrence, 1977; Galbraith and

Nathanson, 1978; Mintzberg, 1979). The underly-

ing logic linking structure to strategy was that a firm’s

organizational structure should facilitate implement-

ing its strategy, often referred to as the strategy-

structure paradigm (Chandler, 1962). Since

international strategies were becoming more multidi-

mensional, simultaneously focusing on various local

and global advantages (Prahalad and Doz, 1987;

Bartlett and Ghoshal, 1989), MNCs required designs

like matrix structures that could also be multidimen-

sional. Since a matrix structure is an overlaying of

two or more elementary or single-dimension struc-

tures, it is the only type of hierarchical structure that

can provide multidimensional coordination. Despite

this apparent fit between matrix structures and multi-

dimensional strategies, many U.S. MNCs abandoned

their matrix structures during the 1980s when they

experienced difficulties managing them (Peters and

Waterman, 1982; Pitts and Daniels, 1984).

As a result of matrix structures falling out of favor

in the U.S., scholarly research on them largely

ceased before a specified theory about how to design

and use them was developed. The strategy-structure

paradigm, which attempts to conceptually define fit

between elements of a firm’s strategy and the type of

organizational structure required to successfully

implement the strategy, was already well established

for the elementary types of MNC structure when

matrix structures fell out of favor. Following the

work of Chandler (1962), international scholars

played a major role in developing the strategy-

structure paradigm (Stopford and Wells, 1972;

Keywords: strategy; matrix structures; multinational

corporations

*Correspondence to: Joachim Wolf, University of Kiel,

Westring 425, 24098 Kiel, Germany. E-mail: wolf@bwl.uni-

kiel.de

Global Strategy Journal

Global Strat. J., 3: 205–226 (2013)

Published online in Wiley Online Library (wileyonlinelibrary.com). DOI: 10.1111/j.2042-5805.2013.01059.x

Copyright © 2013 Strategic Management Society

Franko, 1976; Egelhoff, 1982; Daniels, Pitts, and

Tretter, 1984). But, strategy-structure fit for matrix

structures was still underdeveloped and not a defined

part of the paradigm when research on matrix struc-

tures ceased. This is the gap or research issue the

present study seeks to address. We want to concep-

tualize fit between elements of MNC strategy and the

various types of MNC matrix structure in use today,

and then empirically test hypotheses that are based

on this conceptualization.

One might wonder why we want to study and

further develop theory for MNC matrix structures if

they have been out of favor for two decades. One

reason is that many MNCs currently require the type

of coordination that matrix structures provide. They

are uniquely appropriate for many of the multidi-

mensional strategies MNCs are attempting to imple-

ment. If an MNC avoids matrix structures, the only

way to implement today’s multidimensional strate-

gies is with an elementary structure and an extremely

heavy use of nonhierarchical network structure. In

some circumstances, this may be appropriate. But an

MNC that has the option to add a second dimension

of hierarchy to its organization design (creating a

matrix structure) clearly has more flexibility for suc-

cessfully fitting and implementing a challenging

strategy than an MNC that lacks this option. So

making matrix structures more understandable and

useable expands the organization design options of

MNCs.

There is also significant evidence that managers in

today’s MNCs are increasingly interested in matrix

structures. Jay Galbraith, a well-known consultant to

MNCs, recently reported: (1) a growing interest on

the part of managers in matrix structures; and (2) the

already widespread use of matrix structures by firms,

often without referring to them as matrix structures

(Galbraith, 2009). Our own recent exploratory

research in German MNCs with matrix structures

found a general belief on the part of most managers

that some form of a matrix structure is inevitable in

firms that seek to implement complex international

strategies. Many managers told us that even if matrix

structures are difficult to work with, there is no way

to completely avoid them. If matrix structures are so

inevitable for MNC managers, it seems appropriate

that scholarly research begin to actively reengage the

subject. Our study seeks to do this by developing a

better conceptual model for fitting the different types

of matrix structure to MNC strategies, an important

step toward extending the existing strategy-structure

paradigm to include matrix structures.

The structural dimensions most commonly used

in MNC matrix structures are functional divisions

(FD), product divisions (PD), and geographical

regions (GR). In a typical matrix MNC, the head of

a subunit in a foreign subsidiary simultaneously

reports up to the parent headquarters (HQ) along two

structural dimensions (Davis and Lawrence, 1977).

For example, the subunit head might report to a

country manager and to the lower-level manager of a

worldwide PD HQ. Using Davis and Lawrence’s

(1977) terminology, this is a ‘multiple command’ or

‘two-boss’ structure. Figure 1 shows a simplified

portrayal of such a product division x geographical

region (PDxGR) matrix structure. The subunit is the

German plastics business of a global chemical

company headquartered in the U.S. Here, the general

manager (GM) of the German plastics business

reports in via two hierarchies. The GM’s immediate

boss in the PD hierarchy is a vice president (VP) in

the plastics division. The GM’s immediate boss in

the GR hierarchy is the German country manager.

These two are called ‘matrix managers’ (Davis and

Lawrence, 1977), and there can be more levels of

management between them and the CEO than are

shown in Figure 1. GMs in other business units in the

German subsidiary and in other subsidiaries will also

report in through a PD hierarchy and a GR hierarchy.

Thus, a two-dimensional matrix can be seen as an

overlaying of two elementary structures, in this case

a worldwide PD structure and a GR structure. When

combined in a matrix, they are typically referred to

as dimensions of the matrix.

Since most matrix structures overlay just two

dimensions, there is a design problem deciding

which two dimensions to include in an MNC matrix.

This design problem is analogous to the design

problem that a nonmatrix structure MNC faces when

it has to decide which type of elementary structure to

adopt. This problem was initially studied by

Stopford and Wells (1972) and then by others

(Franko, 1976; Egelhoff, 1982; Daniels, Pitts, and

Tretter, 1984; Habib and Victor, 1991). It is impor-

tant to distinguish between a matrix and a mixed

structure. A mixed structure has some foreign sub-

sidiaries reporting to the parent along one dimension

(e.g., GR) and other subsidiaries along another

dimension (e.g., PD). It is not a matrix structure

because the two hierarchies do not overlap. The

present study focuses only on matrix structures. It

attempts to specify which structural dimensions

should be included in a matrix structure, given the

strategy the structure is attempting to implement.

206 W. G. Egelhoff et al.

Copyright © 2013 Strategic Management Society Global Strat. J., 3: 205–226 (2013)

DOI: 10.1111/j.2042-5805.2013.01059.x