Professional Documents
Culture Documents
Multinationality Performance:
The Moderating of R&D Role
and
Marketing Capabilities
MasaakiKotabe*
TEMPLE UNIVERSITY
SriniS. Srinivasan**
DREXEL UNIVERSITY
PreetS. Aulakh***
TEMPLE
UNIVERSITY
and Li, 1996; Hitt, Hoskisson and Kim, ment of R&Dand marketing on the other
1997; Mishra and Gobeli, 1998; Gomez hand? The existing literature offers con-
and Ramaswamy, 1999; Geringer, Tall- flicting empirical findings on the perfor-
man and Olsen, 2000). That is, increas- mance implications of multinationality.
ing market liberalization around the We develop an argument that coherently
globe, especially in erstwhile-protected piece together seemingly conflicting
economies, has made it easier and some- findings, empirically test and explain the
times necessary for firms to expand into reasons for those findings, and provide
foreign markets (Aulakh, Kotabe and managerial implications.
Teegen, 2000). This liberalization has co-
incided with economic integration, suc- PERFORMANCE IMPLICATIONS OF
cess of international organizations such MULTINATIONALITY
as GATT/WTO and UNCTAD, and ad- The literature on multinationality gen-
vances in information and communica- erally points to the thesis that multina-
tion technologies. These environmental tional expansion allows firms to transfer
trends and the popular buzzwords, such "rent yielding" resources into foreign
as "globalization of markets," "global markets to achieve both economies of
economy," and "think global, act local," scale and scope (Tallman and Li, 1996),
found in both academic literature and exploit market imperfections across
popular press, point toward the growing countries (Dunning, 1988), expand mar-
necessity for firms to find international ket opportunities (Buhner, 1987), and
markets for their products and services maximize location economies by config-
as well as configure their value chain uring value-chain activities (Kogut,
activities around the globe in order to 1985), among others. However, expan-
achieve scale, learning and location sion into diverse foreign markets in-
economies-in essence, to increase their creases the costs (transaction, manage-
multinationality. rial, coordination) of managing far-flung
Multinationality generally refers to the operations, especially for those firms
extent to which firms operate beyond that are located in different cultural en-
their national borders and benefit from vironments (Gomez and Ramaswamy,
product and geographical diversifica- 1999). Thus, performance advantages of
tions through economies of scale and multinationality will reach their limit
scope (e.g., Hitt, Hoskisson, and Kim, when "internal governance costs exceed
1997). For firms that are becoming in- the benefits provided by the economies
creasingly multinational by taking ad- achieved and thus, the range of resources
vantages of liberal trade regimes, some of used and scope of governance exceeds
the relevant normative issues that con- managerial capabilities" (Hitt, Hoskis-
tinue to be asked are: Does the increase son, and Kim, 1997, p. 773).1
in multinationality enhance perfor- Existing studies examining the perfor-
mance? Do firms need to have some mance implications of multinationality
threshold of R&D or marketing intensity have used different theoretical argu-
to benefit sufficiently from multination- ments as well as diverse data sources,
ality? How should they make the tradeoff resulting in mixed and sometimes con-
in resource allocations between geo- tradictory results (see Ramaswamy
graphical expansion into several over- (1995) for an extensive review of this
seas markets on one hand and develop- literature). The findings range from a
80 BUSINESS STUDIES
JOURNALOF INTERNATIONAL
MASAAKIKOTABE,SRINIS. SRINIVASAN,
PREETS. AULAKH
82 BUSINESS STUDIES
JOURNALOF INTERNATIONAL
PREETS. AULAKH
MASAAKIKOTABE,SRINIS. SRINIVASAN,
The data used for this research were as a ratio of sales to operating costs
obtained from the COMPUSTAT data- (OPSALINV). Many researchers have in-
base, which contains firm level data on dicated that variance in firm perfor-
different industries (at the 4-digit SIC mance is partly explained by firm size
classification level). We chose SICs (DeCarolis and Deeds, 1999). Hence, in
based upon the following two criteria: 1) the analysis of the data, we need to con-
data should be available for at least three trol for firm size, or else the parameters
companies in each SIC, and 2) for each estimated might be biased. To avoid the
company, data should be available for at confounding effect of firm size on firm
least 7 years. Our study used data on 49 performance, we used firm size (SIZE),
US companies in 12 different industries measured as a logarithmic function of
(over a 7-year time period ending in sales, as a covariate.
1993) for which COMPUSTAT had com- Multinationality (MULTI) has been
plete data on all the variables of interest. operationalized in a number of ways by
The details of the industries and the different researchers. Some researchers
number of time series and cross-sec- have used sales/profit based measures,
tional observations used are provided in such as, ratio of foreign sales to total
Table 1. sales (Grant, 1987; Tallman and Li,
The variables used in the analysis 1996), foreign income to total income,
were operationalized as follows. Perfor- ratio of foreign sales to total assets
mance was measured in both financial (Daniels and Bracker, 1989; Ramaswamy,
and operational terms in a manner sim- 1995), number of foreign countries in
ilar to that of Gomes and Ramaswamy which a firm has subsidiaries (Tallman
(1999). Financial performance was mea- and Li, 1996). Sullivan (1994) examined
sured in terms of return on assets (ROA). the different ways to measure multina-
Operational outcomes were assessed tionality their associated problems.
TABLE 1
DETAILSOF INDUSTRIESINCLUDEDIN THEANALYSIS
R&D Marketing
Number of Number of Intensity Intensity
SIC Description of the Industry Companies Years (%) (%)
2621 Paper Mills 5 7 1.2 0.4
2670 PackagingPaper, Plastic Film 4 7 3.1 1.8
2800 Chemicals and Allied Products 5 7 5.9 3.7
2851 Paints, Varnishes, Lacquers, 3 7 2.4 3.1
Enamels and Allied products
3270 Concrete, Gypsum, Plaster 3 7 0.7 0.0
3570 Computerand Office Equipment 5 7 9.5 1.3
3571 Electronic Computers 3 7 10.5 1.3
3640 Electric Lighting,Wiring 3 7 4.5 0.5
3674 Semiconductor, Related Devices 9 7 9.1 0.7
3822 Automatic Regulating Controls 3 7 2.4 0.0
3944 Games, Toys 3 7 3.5 16.8
3950 Pens, Pencils, Artistic Materials 3 7 0.7 6.6
84 JOURNALOF INTERNATIONAL
BUSINESS STUDIES
PREETS. AULAK
MASAAKIKOTABE,SRINIS. SRINIVASAN,
Based on his suggestion, and the avail- variables X and Y across four industries
ability of appropriate data, we measured as shown in Figure 1.
multinationality as a ratio of foreign in- In the above figure, the broken-line el-
come to total income. Consistent with lipses represent the point scatter for in-
earlier studies (e.g., Hufbauer, 1970; dividual industries over time, and the
Mansfield, 1981; Kotabe, 1990b), R&Din- broken straight lines represent the indi-
tensity and marketing intensity are de- vidual regressions for the different in-
fined as the annual expenditure on R&D dustries. The solid line represents the
divided by sales and the advertising ex- least-square regression using the data
penditure divided by sales, respectively. points for all the industries. As is illus-
trated by the solid line, even if the two
METHODOLOGY variables are positively related, aggregat-
To extend the generalizability of the ing the data (without accounting for dif-
findings of this research to a number of ferences in the intercepts across indus-
industries, we use data from different tries) and estimating an aggregate model
industries. Cross-sectional studies not might lead us to wrongly conclude that
taking into account variations across in- the two variables are negatively related.3
dustries might lead us to wrong conclu- Therefore, when we pool data from dif-
sions. For the sake of simplicity, let us ferent industries, we need to control for
consider the relationship between two the industry to avoid biased inferences
FIGURE1
MISTAKEN INTERPRETATION OF THE RELATIONSHIP WHEN THE IHETEROGENEITY
IN INTERClfPTSIS NOT ACCOUNTD FOR
Y*
F
I - /
R
M
-7
P
E
R /7~~~~~~~r~I-
F ---
0 '-777~~~
R 7
M 7' --
A /7--
N
c
E
MULTINATIONALITY(x)
about the impact of multinationality on X7it = X3it * X4it) of firm i in time pe-
firm performance. riod t (MULTIRD)
Examining the cross relationship be- X8it = (X3it * X5it) of firm i in time pe-
tween multinationality and firm perfor- riod t (MULTIAD)
mance over a single period of time does Xgit = (X3it * Xit * X5it of firm i in time
not allow us to generalize aboutthe find- period t (MULADRD)
= parameters to be esti-
ings over time. Therefore, we used a /1 through (19
+ I18X8it + Uit
+ 319X9it (1) ANALYSISAND RESULTS
We estimated equation 1 using the
Where:
Fuller and Battesemethod (implemented
Yi = performance of firm i in time
by the TSCS procedure in SAS) and the
results are given in Table 2.4
period t (ROA or OPSALINV) To ensure that the interaction effects
=
Di dummy variable for ith SIC (11 indeed significantly add to the model fit,
dummies for 12 industries)
we ran the following two regression
X2it = size of firm i in time period t models: 1) Model with the main effects
(SIZE)
X3it = multinationality of firm i in only, given by equation (3), and 2) Model
with the main effect and two way inter-
time period t (MULTI)
action effects, given by equation (4).5
X4it = R&D intensity of firm i in time
period t (RDINT) 11
X5it = marketing intensity of firm i in
time period t (ADINT) Yit = Po + + 2X
P12X2Di + 13X3it
i=1
X6it = (X4it * X5it) of firm i in time pe-
riod t (RDAD) + 914X4it + 315X5it + Uit (3)
TABLE 2
(TSCS) REGRESSION
TIMESERIES-CROSSSECTIONAL ANALYSIS
OFFIRMPERFORMANCE*
ROA OPSALINV
Prob >
Variables Parameter Prob > IT| Parameter |TI
Intercept -0.1949 0.0014 1.2971 0.0001
SIZE (X2) 0.0332 0.0001 0.0114 0.6183
MULTI(X3) -0.0097 0.6619 -0.0726 0.0284
RDINT(X4) -0.3191 0.2396 1.4775 0.0514
ADINT (Xs) 0.2196 0.7322 -0.1150 0.9365
RDAD (X6) -6.3330 0.4904 6.9012 0.7448
MULTIRD(X7) 0.1013 0.6609 0.7601 0.0273
MULTIAD(X8) -3.0723 0.0405 -1.6805 0.4757
MULADRD(X9) 100.0757 0.0030 132.3350 0.0131
R2 22.5% 21.3%
*Industrydummy variables are included in the models, but regression coefficients are not
shown in this table.
TABLE 3
PREDICIIVEPOWEROF THETHREE-WAY INTERACTON MODELOVER THE
MAIN EFFECTSAND TWO-WAY INTERACTIONMODELS
ROA* OPSALINV**
Model R2 R2
*Model 2 does not provide significantly incrementalfit over Model 1 (F3323)= 0.99; Model
3 provides significantly higher fit over Model 1 (F4,322) = 2.37 (p < .05)
*Model 2 provides significantly higher fit over Model 1 (F3,323)= 3.009 (p < .05). Model 3
provides significantly higher fit over Model 2 (F1322)= 8.92 (p < .05)
88 BUSINESS STUDIES
JOURNALOF INTERNATIONAL
MASAAKI KOTABE, SRINI S. SRINIVASAN, PREET S. AULAKH
FIGURE2
ROA AS A FUNCTION OF MULTINATIONALITY AND R&D INTENSITY
ROA
1.5
- 0. 2~~~~~~~~~~~~0.5
_ _
0.01
0.03 _
0.07
R&DIntensity 0.09
can be evaluated from equation 7 and RDINT and MULTI on ROA (OPSALINV)
equation 8 by substituting the average after holding the ADINT at the average
RDINT and ADINT values for all the level.
firms in the data set. The averageADINT As can be seen from the above figures,
for all the firms in the data set is 0.025 at very low levels of R&Dintensity, in-
and the average RDINTis 0.051. Substi- creasing MULTIdoes not have a positive
tuting these values in equation 7 and impact on firm performance.However, at
aROAit higher levels of R&D intensity, higher
equation 8, results in aM Tit 0.05
level of MULTIleads to higher firm per-
and aOPSALINVitI
O LI = 0.135. This confirms formance.
aMULTI3it
our expectations that, ceteris paribus, In Figure 4 (Figure 5), we illustrate the
multinationality leads to higher firm per- impact of ADINT and MULTIon ROA
formance. (OPSALINV)after holding the RDINTat
The impact of multinationality on the average level. This graph visually
ROA depends both on the R&Dintensity capturesequation 1, when RDINTis held
and advertisement intensity. As it is not constant at the average value (0.051). As
possible to visually capture all the four can be seen from the above figures, the
dimensions in a single figure, we illus- impact of MULTIis higher at higher lev-
tratethe same in two figures. In Figure 2 els of ADINT than at lower levels of
(Figure 3), we illustrate the impact of ADINT.
FIGURE 3
OPSALINV AS FUNCTION OF MULTINATIONALITY AND R&D INTENSITY
: ?i
19.! ~i-';:.. :i:i
OPSALINV:1.::45-'
1.45-
~" '
'""~"'
'"" ~'~'~~z""1
~'t
0.95
05 0.07
intensity
0.R&D
R&Dintensity
FIGURE 4
ROA AS A FUNCTION OF MULTINATIONALITY AND ADVERTISING INTENSITY
'~~~~~~~~-"
:;'':'.,:":,,e ~:: "
0
1.5
0.5
-0.5 MULTI
-::.:0:.2
' -
0.03 0 '
0.05 0.07
ADINT
90 BUSINESS STUDIES
JOURNALOF INTERNATIONAL
MASAAKI KOTABE, SRINI S. SRINIVASAN, PREET S. AULAKH
FIGURE 5
OPSALINV AS A FUNCTIONOF MULTINATIONALITY
AND ADVERTISING INTENSITY
X*t- D?I;i
.{*i;>.t...
i*. ~ ;rr
OPSALINV <'w;<;~~~~~~,
wi:.- :?I.t;
>? -. , i, t
2 ;
s,2m.
OPSALINV iSS,,cia v _ S t
r- 1.5
1
0.03
ADINT 0.07
FIGURE 6
MARGINAL IMPACT OF MULTINATIONALITY ON ROA
0.4 i~~
0.35 ~ ~
~~~
~~ (id~~~~~~~~~~~~~~~~~~~~~!-]
0.3-,~ ~~~~~~~~~~~~~~~~~~~,
:~'- ,,.
03
?' ~r?-
025
r-0.05~??
,.;:'
0.15 ']
-r0.11.. 0.04?
ti0.1 .0
'C1
0.06
0.04
---
.0.15
0. 003 ADINT
01 002
RDINT 0.09
it equals .0178 (or 1.78%). In other knowledge that these two key factors
words, firms are required to spend at moderate the impact of multinationality
least 1.78% of their sales in R&Dactivi- on firm performance. It sensitizes man-
ties so that the marginal impact of agers on the need to focus not just on
MULTIon OPSALEINVwill be positive. overseas expansion activities, but also to
focus on their R&Dand marketing activ-
CONCLUSIONS AND DISCUSSION ities in order for their overseas expan-
The findings of this research suggest sion to be successful. However, care
that the impact of multinationality on must be taken in interepreting the abso-
firm performance is not unequivocal. lute values of threshold R&D intensity
Rather,the impact of multinationality on calculated from equation 7 and equation
firm performance depends on a number 8. These threshold values are calculated
of firm-specific factors.Two such factors based upon the averagemarketinginten-
that moderate this relationship are the sity for all the firms in the sample. The
R&Dintensity and the marketing inten- threshold R&Dintensity for any particu-
sity. lar industry can be calculated by using
While the existing literature amply the average marketing intensity for that
provides evidence that R&Dand market- particularindustry.
ing intensities positively influence firm At the aggregatelevel (assuming away
performance, our study furtheradvances industry differences), the threshold R&D
92 BUSINESS STUDIES
JOURNALOF INTERNATIONAL
MASAAKIKOTABE,SRINIS. SRINIVASAN,
PREETS. AULAKH
intensity level is 1.78% for the positive and Renforth (1996) discuss the limita-
effect of multinationality on firms' oper- tions of these various measures. In this
ational performance (OPSALINV),but it researchwe used percentage of overseas
takes a threshold R&Dintensity level of income to total income as an indicatorof
3.06% for multinationality to exhibit a multinationaltiy. As data on other mea-
positive effect on firms' financial perfor- sures of measuringmultinationalitywere
mance (ROA). This finding implies that unavailable in the COMPUSTATdata-
the increase in R&Dintensity begins to base, we were unable to replicate this
affect the operational performance of study using these other possible measures
multinationality much earlier than the of multinationality. However, future re-
financial performance of multinational- searcherscould replicate this researchus-
ity. In other words, companies tend to ing othermeasuresof multinationality.
enjoy operational improvement (i.e., po- The R&D intensity measure that we
sitional strengths) from foreign expan- used measures the fraction of sales that
sion before financial improvement as are spent on the research and develop-
they increase their R&Dintensity. This mental activities of the firm.As disaggre-
implication is consistent with the litera- gate data on primary research expendi-
ture in strategic marketing and manage- tures and applied developmental expen-
ment that strategy (multinational expan- ditures are unavailable, we did not
sion in our case) builds firms' positional isolate the impact of research activities
strengths and then subsequently leads to and developmental activities in this
improved financial performance (e.g., study.
Porter, 1986; Day and Montgomery, The marketing efforts of a multina-
1999). Our study has added empirical tional firm are routinely operational-
credence to the existing literatureby ex- ized by their advertising intensity (Ca-
amining the performanceimplications of pon, Farley, and Hoenig, 1990), as
multinationality with the time series firms are reluctant to disclose their to-
data across industries. However, there tal marketing expenditures. In this
are also limitations that would beg for study we used advertising intensity as
furtherresearch inquiries on this issue. a surrogate for the marketing efforts of
The conceptualization of multination- a multinational firm. As data on total
ality of a firmhas created enough contro- marketing activities are unavailable,
versy in the academic literature. Despite we could not use the ratio of marketing
the wide body of researchin the interna- expenditures to total sales as an indi-
tional business area, there is no single cator of marketing efforts. Due to com-
accepted method of measuring multina- petitive reasons, organizations are re-
tionality of firms. The range of measures luctant to disclose finer accounting de-
includes percentage of sales that are tails of their operations and hence
from overseas operations, percentage of researchers have to contend with such
profit measures from overseas opera- limitations.
tions, number of countries in which The R&Dintensity and advertising in-
firms operated, foreign assets as a per- tensity measures used in this research
centage of total assets or a summated are only limited proxies for the rent
measure of these above indicators. Both yielding capabilities of the firms. One
Sullivan (1994) and Ramaswamy,Kroeck possible extension of this research is to
model how the R&D expenditures and impact of R&D intensity on firm perfor-
marketing expenditures improve the rent mance.
yielding capabilities of multinational
firms. NOTES
The entry strategy and mode of opera-
1. Also, as noted by Daniels and
tion of a multinational company also
Bracker (1989), the association between
will have an impact on firm perfor-
foreign operations and foreign perfor-
mance. Companies that aggressively
mance need not be the same across all
source abroad through contractual ar-
industries. As domestic markets are also
rangements, instead of just focusing on an important source of revenues and
domestic suppliers, might show higher
profits for any company, one would nor-
profitability, although the measure of mally not expect a company whose prof-
multinationality used in this research its are entirely from abroad to out-per-
could not capture this dimension. In this form a company with some combination
research we do not control for the entry of foreign and domestic profits. We
strategies of multinational companies or thank an anonymous reviewer for point-
how they operate in foreign countries.
ing this out.
Assuming appropriate data availability, 2. Past studies have, at the most, incor-
researchers could investigate if the im-
porated firm-specific variables as control
pact of R&D and marketing intensities variables in the empirical analyses.
systematically differs across firms using 3. Depending upon the intercept and
different methods of operations in for-
slope of the individual industries, the
eign countries. pooled model, estimated without con-
Another possible extension is to inves-
trolling for industry differences, might
tigate the lagged effect of R&D intensity. show positive, negative or no relation-
As R&D activities might take several
ship between the dependent and inde-
years to yield financial benefits, future pendent variables.
researchers might want to investigate 4. For the sake of brevity, the parame-
how the lagged effects of R&D intensity ter estimates for industry dummies are
interact with multinationality of the firm not reported in the table. Readers inter-
in determining firm performance. Lag- ested in these parameter estimates can
ging the R&D activities by one or two get them from the authors.
periods when estimating the model im- 5. Adding each of the three two-way
plicitly assumes that all the R&D spend- interactions, one at a time, to the main
ing will uniformly have an impact after effects model (equation 3) does not sig-
one or two years. This assumption hides nificantly increase the model fit.
the fact that some projects might take 6. We thank the anonymous reviewer
longer to yield results while some other for suggesting the use of the incremental
projects might yield financial returns in fit statistics to test for the impact of the
a short term. Further lagging the inde- interaction term.
pendent variables by a couple of time 7. As 133and 317 are not significantly
periods will result in a loss of degrees of different from zero, they are excluded
freedom. Future researchers with richer from equation 8.
data set can try to estimate the optimal 8. As p18 is not significantly different
lag that could be used in modeling the from zero, it is excluded from equation 8.
94 BUSINESS STUDIES
JOURNALOF INTERNATIONAL
MASAAKIKOTABE,SRINIS. SRINIVASAN,
PREETS. AULAKH
Helsen, Kristiaan, Kamel Jedidi & Wayne of European and Japanese Multina-
S. DeSarbo. 1993. A New Approach to tionals: An Empirical Investigation.
Country Segmentation Utilizing Mul- Journal of Marketing, 54 (April):
tinational Diffusion Patterns. Journal 19-33.
of Marketing, 57 (October): 60-71. . 1990b. The Relationship Be-
Hitt, Michael, Robert Hoskisson & tween Offshore Sourcing and Innova-
Hicheon Kim. 1997. International tiveness of U.S. Multinational Firms:
Diversification: Effects on Innovation An Empirical Investigation. Journal of
and Firm Performance in Product-Di- International Business Studies, 21(4):
versified Firms. Academy of Manage- 623-638.
ment Journal, 40(4): 767-98. Levitt, Theodore. 1983. The Globaliza-
& R. Duane Ireland. 1994. tion of Markets. Harvard Business Re-
A Mid-Range Theory of the Interactive view, 61 (May-June): 92-102.
Effects of International and Product
Mansfield, Edwin. 1981. Composition of
Diversification on Innovation and Per- R&D Expenditures: Relationship to
formance. Journal of Management, Size of Firm, Concentration, and Inno-
20(2): 297-326. vative Output. Review of Economics
Hufbauer, G.C. 1970. The Impact of Na- and Statistics, 63 (November):
tional Characteristics and Technology
610-615.
on the Commodity Composition of
Mishra, Chandra S. & David H. Gobeli.
Trade in Manufactured Goods. In Ray-
1998. Managerial Incentives, Internal-
mond Vernon, editor, The Technology
ization, and Market Valuation of
Factor in International Trade. New
Multinational Firms. Journal of Inter-
York: Columbia University Press. 145-
national Business Studies, 29(3): 583-
231.
597.
Jaccard, James, Robert Turrisi & Choi K.
Wan. 1990. Interaction Effects in Mul- Peteraf, Margaret A. 1993. The Corner-
stones of Competitive Advantage.
tiple Regression. Newbury Park, Calif:
Strategic Management Journal, 14(3):
Sage Publications
179-192.
Katrishen, Frances A. &Nicos A Scordis.
1998. Economies of Scale in Services: Porter, Michael E., editor. 1986. Compe-
A Study of Multinational Insurers. tition in Global Industries. Boston:
Harvard Business School Press.
Journal of International Business Stud-
ies, 29(2): 305-323. Ramaswamy, Kannan. 1995. Multina-
Kim, W.C., P. Hwang & W.P. Burgers. tionality, Configuration, and Perfor-
1989. Global Diversification Strategy mance: A Study of MNEs in the U.S.
and Corporate Profit Performances. Drug and Pharmaceutical Industry.
Strategic Management Journal, 10: Journal of International Management,
45-57. 1(2): 231-53.
Kogut, Bruce. 1985. Designing Global , K. Galen Kroeck, &William Ren-
Strategies: Comparative and Compet- forth. 1996. Measuring the Degree of
itive Value Added Chains. Sloan Internationalization of Firms: A Com-
Management Review, 27(Summer): ment. Journal of International Busi-
15-28. ness Studies, 27(1): 167-77.
Kotabe, Masaaki. 1990a. Corporate Prod- SAS/ETS User's Guide. Version 6. SAS
uct Policy and Innovative Behavior Institute Inc. Cary, NC.