Performance effects of different subsidiary role types on Multinational Corporations: Are there differences in the ROI of multinational corporations using world mandate, specialized
contributor or local implementer subsidiary types?
Matthias Ollinger (11802626) Bachelor Thesis BSc Business Administration
Specialization: Finance
Faculty of Economics and Business, University of Amsterdam Academic year: 2019-2020
Supervised by Florencio López de Silanes
Statement of Originality
This document is written by Student Matthias Ollinger who declares to take full
responsibility for the contents of this document.
I declare that the text and the work presented in this document are original and
that no sources other than those mentioned in the text and its references have
been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision
of completion of the work, not for the contents.
Table of Contents
ABSTRACT ... 4
SUBSIDIARY ROLE TYPES ... 6
LOCAL IMPLEMENTER ... 6
SPECIALIZED CONTRIBUTOR ... 6
WORLD MANDATE ... 7
TYPES OF MULTINATIONAL CORPORATIONS ... 7
GLOBAL STRATEGY ... 7
MULTIDOMESTIC STRATEGY ... 8
TRANSNATIONAL STRATEGY ... 8
LINK BETWEEN MNC TYPES AND SUBSIDIARY STRATEGIES ... 9
GLOBAL STRATEGY AND SPECIALIZED CONTRIBUTOR SUBSIDIARIES ... 9
TRANSNATIONAL STRATEGY AND LOCAL IMPLEMENTER SUBSIDIARIES ... 9
MULTIDOMESTIC STRATEGY AND WORLD MANDATE SUBSIDIARIES ... 10
RETURN ON INVESTMENT ... 12
INDUSTRY SECTOR ... 12
LINK BETWEEN SUBSIDIARY ROLE TYPE AND RETURN ON INVESTMENT ... 12
METHOD ... 14 RESULTS ... 16 DISCUSSION ... 20 CONCLUSION ... 22 BIBLIOGRAPHY ... 23 APPENDIX ... 25
Abstract
It seems there has been much research conducted about the internal structures of Multinational Corporations. Many of them with the goal to analyze the degree of centralization or decentralization. Others examined differences between hierarchy or heterarchy. To some extent many of these studies investigated the degree of autonomous decision making within different levels of an organization.
As local implementer, specialized contributor, and world mandate subsidiary role types (IV), to some extent, reflect such differences within the autonomy of a subsidiary from its parent corporation, this research tries to answer the question whether there are performance differences between German MNCs applying one of these subsidiary role types. To investigate the effect of these subsidiary role types on the corporate performance, hypothesis 1 examines potential performance differences. Subsequently, hypothesis 2 analyzes whether MNCs using specialized contributor subsidiaries might underperform MNCs using the other two types. The potential effect of industry sectors (CV) on the MNCs performance is investigated in hypothesis 3. Hypothesis 4 is examining the effect of the control variable and independent variable on the Multinational Corporation’s performances. Thereafter, hypothesizes 5-7 analyze the robustness of the previous results. As the sample of the research the top 100 German corporations (by revenue) are used. While the data initially supports hypothesizes 1 and 2, the data did not support hypothesis 3 and 4, as either there were to less observations compared to industry sector categories or there really was no relation between industry sectors and performance differences of MNCs. The subsequent analysis of the robustness confirmed potential performance differences between MNCs using different subsidiary role types. Though this research concludes with no significant evidence whether these performance differences are a result of the application of local implementer, specialized contributor, or world mandate subsidiary role types.
Performance effects of different subsidiary role types on Multinational Corporations: Are there differences in the ROI of multinational corporations using world mandate, specialized
contributor or local implementer subsidiary types?
There has been much research conducted on the structures of multinational corporations. Vayanos (2003) investigated the decentralization of information processing through different hierarchical levels to predict the most efficient decision-making level. Birkinshaw and Morrison (1995) further investigated hierarchy and heterarchy models of multinational organizations by examining differences between the subsidiary role types world mandate, specialized contributor, and local implementer.
Interest in differences between hierarchy and heterarchy models, whether in terms of information processing, the interdependence between inter-organizational units or economic return, was shown further by Tolentino (2002) as well as Bartlett & Ghosal (1986). One approach used by Birkinshaw and Morrison (1995) to measure performance differences across subsidiary roles was to compare the subsidiaries’ return on investment relative to its parent MNC. One possible explanation for their finding, that specialized contributors significantly underperform the other two types, was that specialized contributors are often that well integrated that they could serve as cost centers. Although, there might be MNC using multiple different subsidiary types at the same time, the research of Bartlett and Ghosal (1986), Birkinshaw and Morrison (1995) and even to some extent Porter (1986) link the selected subsidiary role types to Global, Multidomestic and Transnational corporate strategies. This indicates that the usage of either world mandate, specialized contributor and local implementer subsidiary types depends on MNCs’ overall strategy.
Since Birkinshaw and Morrison (1995) elaborate how the interrelation of strategy/role and structural context influence the performance of a corporation, the question is now whether the performance differences between subsidiary roles, reported in their research, influence the MNC’s overall performance. Therefore, the purpose of this study is to examine if a corporate’s subsidiary role type influences its return on investment, leaving the research question:
‘Are there performance differences between German multinational corporations using world mandate, specialized contributor or local implementer subsidiary role types?’
As the environment, and thereby indirectly the needs and wants industries face, affect the strategy selection of MNCs, this study does not aim to suggest the optimal subsidiary role type to gain maximum economic efficiency. Rather this study aims to investigate whether there are performance differences between different levels of integration or autonomy and therefore different degrees of product differentiation, as this indirectly reflects the demands of customers. These insights create value
for businesses when selecting and evaluating their strategy and hence optimal internal organizational structure.
Subsidiary role types
Local implementer
According to Birkinshaw and Morrison (1995), subsidiaries of this type have limited geographical scope. As their responsibilities are limited to either one singe country or world region, local implementer subsidiaries have limited functional and value-added scope. (Porter, 1986) As Porter (1986) further describes, common value activities have been integrated globally within the multinational corporation. The subsidiary’s role is to adapt global products to the needs of the local market. Research of Bartlett and Ghoshal (1986) as well as Gupta and Govindarajan (1991), confirm this. Their implementer subsidiaries are described as closely integrated within the international operations of an MNC, but at the same time, autonomous with regards to product adjustments. Local operations of this subsidiary type include R&D, Distribution, Service, Marketing, and Sales activities and, to some extent, Manufacturing, in the case of product adjustments. Although Porter (1986) refers this subsidiary role type to be typically found in a multidomestic strategy, this paper argues that the integration of common value activities, i.e. purchasing, manufacturing, etc., together with some product adjustments, either in terms of Service, Distribution, Marketing, and Sales activities or even sometimes with adaptions of product characteristics to local needs, can be found more commonly in Transnational strategies.
Specialized contributor
This subsidiary type is highly interdependent with affiliated subsidiaries as well as its parent organization. The reason for this is that it serves as a center of expertise within limited functions and activities. (Birkinshaw & Morrison, 1995) Headquarters coordinate the activities of each specialized contributor to coordinate and combine their efforts to archive a combined product. (Roth & Morrison 1992) As Jarillo and Martinez (1990) explain, this subsidiary type arises commonly when environmental pressures are low for local responsiveness and high integration. This fits Bartlett and Ghoshal’s (1986) as well as Porter’s (1986) definition of a Global strategy.
World Mandate
The activities of this subsidiary type are integrated worldwide. In difference to local implementer and specialized contributor subsidiaries, the activities of world mandate subsidiaries are managed from the subsidiary itself rather than the parent MNC. Thereby it attains ‘decentralized centralization’. (White & Poynter, 1984) World Mandate subsidiaries have unconstrained product and value-added scope as they have got either regional or worldwide responsivities for the entire business. Bartlett and Ghoshal’s (1986) ‘strategic leader’ serve as the counterpart of world mandate. As it provides high level of resources and expertise it operates autonomous and only develops and implements its overall strategy together with headquarters. (Roth & Morrison, 1992) This indicates that world mandate subsidiaries arise commonly where environmental pressures are high for local responsiveness and low for integration to headquarters. This does not mean that the internal operations of the subsidiary cannot be highly integrated. With regards to MNC types, world mandate subsidiaries would fit Porter’s (1986) multidomestic strategy.
Degree of integration to HQ (MNC) High
Specialized contributor Local implementer
World mandate
Low High
Degree of international product differentiation / Pressure for local responsiveness
Types of Multinational Corporations
Global Strategy
According to Bartlett and Ghoshal (2002), companies pursuing a global strategy use centralized global-scale operations to build up cost advantages. Porter (1986) describes this type of strategy as mass production pursuing a low-cost strategy through economies of scale. This type of MNCs is organized as a centralized hub. Headquarters exchange information as well as goods resources in the form of a one-way flow to its subsidiaries. (Harzing, 2000) This type of MNC emerges when environmental pressures are low for local responsiveness and high integration. (Bartlett & Ghoshal, 2002) Therefore,
global strategy is defined as centralized. HQ and subsidiaries either take over a specific role within the production or serve as local sales division applying the same coordinated image within all markets.
Multidomestic Strategy
Porter (1986) describes this strategy as one in which companies compete on a country by country base. Its products are adjusted to local preferences of the target audience. Therefore, such a company’s subsidiaries are decentralized, and its important activities are in the area where it conducts business. This follows Porter’s differentiation product strategy, as the MNC’s subsidiary defines its strategy mainly based on country specific circumstances, hence local needs and preferences. Bartlett and Ghoshal (2002) call their counterpart of this strategy ‘Multinational’ and describe it as one strategy which emphasizes responsiveness to national differences through strong local presence. This type of MNC emerges when environmental pressures are high for local responsiveness and low for integration to headquarters.
Transnational Strategy
The Transnational strategy is one which derived out of the environmental pressure for high local responsiveness and high pressure for integration within the MNC. (Porter, 1986) Harzing (2000) furthermore explains that the emergence derives from the need of some product adjustments to regional preferences combined with the pressure for cost efficiency gained by the integration of common value activities. Such activities are usually early in the chain of value-adding activities, which are in that case coordinated by the headquarters. Subsequent value-adding activities can either be the adjustment of products in the local area to meet local product characteristics. Though, mainly such adjustments are within marketing, sales or distribution rather than changes in the product itself. As Bartlett and Ghoshal (2002) elaborate, transnational corporations use economies of scale within upstream value activities. Local subsidiaries than either market products just produced for this market or market products with comparable characteristics differently, according to local demands.
Degree of integration of subsidiary activities to HQ (MNC) / Centralization High
Global Transnational
Multidomestic
Low High
Degree of international product differentiation / Pressure for local responsiveness
Link between MNC types and subsidiary strategies
Global strategy and specialized contributor subsidiaries
As previously explained, MNC’s applying a global strategy aim for a cost-efficient mass production. (Porter, 1986) To serve the global market with the identical generic product, they apply economies of scale and centralize their operations. This fits to the specialized contributor subsidiary type. Such subsidiaries serve as centers of expertise, hence act as centralized centers for specific tasks. (Birkinshaw & Morrison, 1995) As within the global strategy, the activities of specialized contributor subsidiaries are coordinated centralized by its headquarters. Furthermore, the subsidiaries are highly interdependent on each other as they combine their efforts to archive a joint product. (Roth & Morrison, 1992)
Both, global MNC strategy and specialized contributor subsidiaries are commonly used when environmental pressures are low for local responsiveness and high integration.
Transnational strategy and local implementer subsidiaries
According to Harzing (2000) MNCs using a transnational strategy centralize common value activities, which are usually higher in the value chain, to achieve cost saving through economies of scale. In the subsequent step, further elaborated by Bartlett and Ghoshal (1986), transnational corporations adapt their global products to needs of target regional markets or develop products specifically for these regional markets. In difference to multidomestic corporations they do not necessarily adjust the products features. Mainly they adapt the way how they market product features to point out product characteristics of the global product which fit the needs of the market. This fits to the local implementer subsidiary role as such subsidiaries are responsible for a limited geographical
region. It is highly integrated into the MNC operations and adapts some product characteristics to the regional demand.
As previously mentioned, in difference to Birkinshaw and Morrison (1995) this paper argues that local implementer subsidiaries fit the transnational strategy. This is because both, transnational MNC strategy and local implementer subsidiaries, are common for high pressure for local responsiveness and high pressure for integration.
Multidomestic strategy and world mandate subsidiaries
Bartlett and Ghoshal (1986) describe, their counterpart of multidomestic strategy, as one strategy which emphasizes responsiveness to national differences through strong local presence. They emphasize product adaption over a coordinated global product offering. Because of the decentralization of its operations into regions, such MNC’s subsidiaries are decentralized. This fits to the definition of world mandate subsidiaries by White & Poynter (1984), who describe these subsidiaries as ‘decentralized centralization’. Bartlett and Ghoshal’s (1986) definition of their ‘strategic leader’, which acts as the counterpart of world mandate, is defined as highly autonomous operating unit which only develops its strategy with its parent MNC. (Roth & Morrison, 1992)
Both, multidomestic strategy and world mandate subsidiary put large emphasis on decentralized product development and offering. Therefore, both types are common in environments with high pressure for local responsiveness and low for integration to headquarters.
Degree of integration of subsidiary activities to HQ (MNC) / Centralization High Global Specialized contributor Low autonomy Mass production No product adjustments Individual centers of expertise
Economies of scale Transnational Local implementer Medium autonomy Some product/marketing adjustments
Independent R&D, Marketing, Sales & Distribution Economies of scale within value
activities
(higher within value-chain) Multidomestic World mandate
High autonomy
All products adjusted to local needs and preferences
Each local subsidiary totally independent
No coordinated product offerings
Low High
Return on Investment
Return on investment is a performance measure to assess the efficiency of an investment. (Chen, 2020) It aims to value the return resulting from a specific investment relative to its costs. This can be either done with a project or a whole company/corporation. (Phillips & Phillips, 2006)
Industry Sector
As elaborated by Birkinshaw and Morrison’s (1995) model, the environment, and to some extent the industry sector a MNC operates in, influence the strategy/role and structural context of a multinational corporation. Furthermore, Vayanos (2003) explains that different industries are linked to different MNC types.
As industry sectors are, according to Global Industry Classification Standard, (Brain, 2016) a superordinate category to industries, the statement of Vayanos (2003) should be transferable to industry sectors being linked to different MNC types. Thus, industry sectors are used as a control variable to test whether potential performance differences between subsidiary role types might result from different characteristics of industry sectors.
Link between subsidiary role type and Return on Investment
After elaborating the different subsidiary role types (Structural Context), the different multinational corporation types (Strategy/Role) and their interrelation, it is important to address once again the model of Birkinshaw and Morrison (1995):
As mentioned in the introduction, the interrelation of Strategy/Role and Structural Context influence the performance. One result of Birkinshaw and Morrison’s (1995) study was that they found significant evidence for differences within the return on investment of subsidiary role types, local implementer, specialized contributor & world mandate, relative to their parent MNC. Subsequently to the connection of MNC types/strategies and subsidiary types, this research aims to test whether the performance differences between subsidiary roles, reported by Birkinshaw and Morrison (1995), are transmitted to their parent MNC’s.
Therefore, this study derived with the following hypothesizes:
H1: There is a performance difference in return on investment between German MNCs pursuing different subsidiary strategies
H2: MNC’s using the specialized contributor subsidiary strategy report lower return on investment relative to MNC’s with local implementer and world mandate subsidiaries.
H3: Industry sectors affect the return on investment of MNC
H4: Subsidiary types and industry sectors explain deviations within return on investment of MNC’s
As it turned out, during the analysis, that the results of the regression 1 & 2 might not be very reliable and robust, additional hypothesizes and regressions have been added:
H5: The significant results of regression 1 & 2 are not robust
H6: The results of the robust regression do not support those of the second regression and therefore question the robustness of the second regression.
H7: There is no performance difference in return on investment between MNC’s subsidiary strategy using performance data of 2019.
Method Design, Sample and Procedure
In the course of this study, convenience sampling was used to derive with the sample. Research on the top 100 German corporations in terms of revenue was conducted, as they serve as a sample for this study. To determine these top 100 corporations the official ranking from the German stock exchange (Deutsche Börse) was used. Subsequently, those corporations have been investigated whether they conduct business internationally as well as classified into different MNC types and subsidiary role types.
The analysis of the data was conducted with the statistical program STATA. As 8 of 100 corporations did not conduct business internationally, they have been excluded from the sample. Furthermore, four companies within the sample were subsidiaries of other MNC which were included in the sample. As their effect would have been reported multiple times, these companies have been excluded as well. Another 19 MNC did not provide the necessary information to calculate the return on investment, wherefore they have been excluded too. This results in a final sample of 65 observations/MNCs.
Of those MNCs, 10 were specialized contributors, 18 were local implementer and 37 world mandate subsidiaries. The average return on investment in 2018 within the sample was 6.11%, with a range from -11.45% to 24.83% (SD 5.27%).
Measures
For the independent variable, subsidiary role type, the MNC were categorized into three categories, local implementer, specialized contributor and world mandate. The criteria for this classification were developed from the literature, which was previously elaborated, from Birkinshaw and Morrison (1995), Porter (1986), Bartlett and Ghosal (1986) and Harzing (2000). The categorization into these subsidiary role types was conducted using the following questions:
1. Are products adjusted to local demands?
2. Are the subsidiaries operations highly integrated within the parent MNC?
The necessary information to answer these questions was collected from the annual reports of the MNC or from the strategy subpage published on the corporate website.
The dependent variable, return on investment, was calculated manually according to the formula on Investopedia (Chen, 2020) as well as the information provided by Phillips & Phillips (2006).
from Bundesanzeiger, a database of the German government containing official company reports. Those companies for which there was no data provided, either did not have to publish the required information publicly or were part of a larger corporation, wherefore there was often no public information about their total capital balance. The applied formula for the return on investment was:
𝑅𝑂𝐼 = 𝐸𝐵𝐼𝑇
𝑡𝑜𝑡𝑎𝑙 𝑐𝑎𝑝𝑖𝑡𝑎𝑙 𝑏𝑎𝑙𝑎𝑛𝑐𝑒 (𝑡𝑜𝑡𝑎𝑙 𝑜𝑓 𝑎𝑐𝑡𝑖𝑣𝑎 𝑜𝑟 𝑝𝑎𝑠𝑠𝑖𝑣𝑎 𝑜𝑛 𝑡ℎ𝑒 𝑏𝑎𝑙𝑎𝑛𝑐𝑒 𝑠ℎ𝑒𝑒𝑡)
As the control variable for this study the industry sectors in which the MNC are operating are used. Again, this data was hand collected. The industry sectors of the Global Industry Classification Standard by Standard & Poor’s and Morgan Stanley Capital International (MSCI) were applied. (Brain, 2016)
Analytical Plan
To test hypothesis 1, whether there is a performance difference in return on investment between German MNCs pursuing different subsidiary strategies, a linear regression with robust standard errors will be used with the categorical variable subsidiary role type as independent variable and return on investment of the year 2018 as the dependent variable.
Another linear regression with robust standard errors will be conducted to test hypothesis 2, if multinational corporations using the specialized contributor subsidiary strategy report lower return on investment relative to MNC’s with local implementer and world mandate subsidiaries.
Subsequently, to test hypothesis 3, an additional linear regression with robust standard errors will be used to test whether the industry sectors influence the return on investment of MNCs.
To test the potential effect of subsidiary types and industry sectors together (hypothesis 4) a 4th linear regression with robust standard errors will be conducted.
As it turned out, during the analysis, that the results of regressions 1 & 2 might not be very reliable and robust, additional hypothesizes and regressions have been added:
Hypothesis 5, the significant results of regressions 1 & 2 are not robust, will be tested conducting a robust regression and a linear regression with robust SE on performance data of a different year.
Additionally, to test hypothesis 6, whether the results of the robust regression do not support those of the second regression, a robust regression will be performed.
In a final step, hypothesis 7, there is no performance difference in return on investment between MNC’s subsidiary strategy using performance data of 2019, will be tested using a regression with robust standard errors and return on investment data of a different year.
Results
As all observations represent valid performances of multinational corporations, no influential point nor outliers have been deleted. Furthermore, as the test for outliers is going to show in the following section, there have not been any observations with Cook’s Distance above 1.
Because the independent variable and control variable, are categorical variables without any real means and the calculation of correlations requires a means, there will be no correlation matrix to report.
Before one may conduct a linear regression, the data must be tested whether it meets the requirements and assumptions of a linear regression.
The normality of the data was examined using a Qnorm plot of the standardized residuals. (Appendix: Graphic 1) Additionally, the Kernel density estimate was calculated and plotted. (Appendix Graphic 3) According to the plots it was not clear if the data meets the assumption. Therefore, a Shapiro-Wilk test was conducted. All plots and tests were done for the dependent variable RoI2018 as well as the ‘control variable’ for the robustness of the analysis: RoI2019. The Shapiro-Wilk test for RoI2018 is significant. (z =3.931, p-value = .00004) (Appendix: Stata Output 1) The results for RoI2019 is almost significant with z = 1.582 and a p-value = .05685. As the results show that the normality of the residuals is not given, the study was continued based on the central limit theory, which states that as long as n > 30 the data is robust enough for regressions.
The Durban-Whatson test for independence of observations could not be conducted as the sample data does not contains any time series information. Therefore, this research was continued with the assumption of independence of observations.
Succeeding, the test for linear relationship was conducted using a scatter plot of subsidiary role type (IV) and return on investment of the year 2018 (DV). As the plot did not show a clear sign for or against a linear relationship, the research was continued assuming the assumption is met. (Appendix: Graphic 6)
Values for cook’s distance have been predicted and analyzed to test for potential influential outliers. There have been no cook’s distance values above 1 and only 3 observations with cook’s distance values violating the threshold of cook’s distance > 4/n (n = 65). (Stata Output 2) As already stated prior, since all observations represent valid performance data of MNCs, no outliers or influential points were excluded. Nevertheless, the test for the robustness of the first results is going to reweight extreme performance observations, hence influential points to compare the effect of extreme out- or underperforming MNCs on the results of the regressions.
To test for homoscedasticity, first Cameron and Trivedi's decomposition of IM-test and Breusch-Pagan / Cook-Weisberg test were conducted. Both tests show insignificant results, although the second test was close to the significance level with Chi-square = 3.31 and a p-value of .0690. (Appendix: Stata Output 4) In the following step a scatter plot of fitted values and the residuals was used to assess the homoscedasticity. As the plot suggests an increase within the residual variance with increasing predicted values, this might be a sign for an omitted variable affecting the residuals. As a result, although there is no clear sign for heteroscedasticity, regressions with robust standard errors are used for the analysis of this research. (Appendix: Graphic 7)
At last, the mean VIF was calculated for the model. The assumption for multicollinearity is met since the VIF value was below 5 (1.40). (Appendix: Stata Output 6)
As a conclusion, it is not clear if the data does meet all requirements for a linear regression as the independence of observations and linear relationship could not be confirmed.
To test whether there are performance differences in return on investment between MNC following different subsidiary strategies (hypothesis 1), a linear regression was applied. As the test for homoscedasticity was not clear, a linear regression with robust standard errors has been selected for the analysis. To avoid perfect multicollinearity and therefore a dummy variable trap, the analysis excludes one of the three subsidiary type categories. The remaining two subsidiary types are subsequently compared to the omitted subsidiary type to compare their relative effect on the rate on investment in year 2018. The test showed an almost statistically significant result at the a < 5% boundary for specialized contributor. (t = -1.98, p-value = .052, b = -.0313336) (Appendix: Stata Output 8) The results show no significant result for a difference in RoI 2018 between local implementer and world mandate subsidiaries (t = .19, p-value = .852, b = .0030671), The current model reports an R Square of .0538, indicating that the predictor variables explain about 5% of the variance of the outcome variable. There is no adj. R Square for linear regressions with robust SE reported by STATA. The results of this regression support hypothesis 1.
In the following step, an additional regression with robust standard errors with specialized contributor as the omitted category was conducted to test hypothesis 2: MNC’s using the specialized contributor subsidiary strategy report lower return on investment relative to MNC’s with local
implementer and world mandate subsidiaries. The results report, again, the almost statistically significant result at the a < 5% boundary for the difference between MNCs using specialized contributor and local implementer subsidiary types (t = -1.98, p-value = .052, b = -.0313336) (Appendix: Stata Output 9) Moreover, there is a significant result for the difference within the ROI (of the year 2018) of MNCs following the world mandate subsidiary strategy, relative to those using a specialized contributor subsidiary strategy. (t = 2.88, p-value = .005, b = .0344007) (Appendix: Stata Output 9) As, compared to regression 1, only the omitted categorical reference variable was changed from local implementer to specialized contributor, the model stayed the same. Hence, it reports the same R Square of .0538. The results support hypothesis 2.
To test the effect of industry sectors on variations within the return on investment (hypothesis 3) a linear regression with robust standard errors was used. As in the previous two regressions analyses, the different categories of the IV (industry sectors) were compared relative to one omitted category. Significant results were reported for the following industry sectors: Consumer Durables & Apparel (t = 3.70, p-value .001, b = .0836373), Energy (t = -2.05, p-value = 0.046, b = -.0462927) and the Semiconductors and Semiconductor Equipment industry sector (t = 2.94, p-value = 0.005, b = .0663473). (Appendix: Stata Output 10) All other industry sectors reported insignificant results. The analysis was conducted relative to the Automobiles and Components industry sector. The same as for the previous regressions, STATA does not report adj. R Square for the regression type. The R Square of this analysis is .2011. (Appendix: Stata Output 10) As some of the industry sectors show significant results in this analysis, the data supports hypothesis 3.
To test the effect of the different subsidiary role types (IV) and the industry sectors (CV) together on the return on investment of the MNC’s in the year 2018 (DV), another regression with robust standard errors was conducted. Hereby, hypothesis 4, subsidiary types and industry sectors explain deviations within return on investment of MNC’s, was tested. The results of this test show no statistically significant difference between MNC’s using one of the three subsidiary role types. (local implementer: t = .69, p-value = .491, b = .0156049; world mandate: t = 1.34, p-value = .187, b = .0221009; both relative to specialized contributor) (Appendix: Stata Output 11) This is in contrast to the results of regression 1 & 2 which reported statistically significant results for differences within the return on investment of MNCs following different subsidiary role types. The test reports a R Square value of 0.2119. (Appendix: Stata Output 11) This indicates an increase in R Square Change = .1581, relative to regression 2 (R Square .0538). Regarding the industry sector categories, the regression reports statistically significant results for Consumer Durables & Apparel (t = 3.49, p-value .001, b = .0889522 and the Semiconductors and Semiconductor Equipment industry sector (t = 2.69, p-value = 0.010, b = .0651662). (Appendix: Stata Output 11) Compared to regression 3 there has been no significant result for the Energy industry sector anymore. (t = -.91, p-value = 0.370, b = -.0253729)
(Appendix: Stata Output 11) The change in R Square relative to regression 3 (.2011) is = .0108. As there was no significant result for any subsidiary role type, hypothesis 4 is rejected.
The fact that no statistically significant result was derived from regression 4, containing both, IV and CV, raises the question of the robustness of the first and second regression. To test the robustness of the results of regression 2, a robust regression as well as a regression with robust standard errors with ROI data from the year 2019 have been added to the analysis.
To test hypothesis 5, the significant results of regression 1 & 2 are not robust, the subsequent hypothesizes were tested:
The results of the robust regression correct for potential influential points in the sample data, by excluding observations with a cook’s distance above 1. Furthermore, instead of weighting all observations equally, as in regression 1&2, robust regressions reweigh those observations based on absolute residuals. As the robust regression type drops the most influential points and down-weights observations with large absolute residuals it ensures the comparability of the different subsidiary strategy categories. Extreme out- or underperforming MNC’s are therefore not included, or their effect is reduced, in the analysis. Because of that they cannot affect the individual subsidiary type category’s unproportionally, anymore.
Especially in the case of this research, with only a very small sample of n = 65, even just a small number of MNC’s with extreme performance values can manipulate the statistical measures into a direction which does not represent the majority of each specific subsidiary type anymore. As such effects are included in regression 1 & 2, we can compare these against the result of regression 5 & 6.
To test hypothesis 6, whether the results of the robust regression do not support those of the regression 2 and therefore question the robustness of the second regression, a robust regression with subsidiary role type as the independent variable and return on investment of 2018 as the dependent variable has been performed. The results of this robust regression show no statistically significant difference between MNCs using specialized contributor or local implementer subsidiary role types. (t = 1.05, p-value = .296, b = .0150071) Only the results for world mandate are, the same as in regression 2, significant. (t = 2.54, p-value = .014, b = .0326185) (Appendix: Stata Output 12) These results support hypothesis 6 and therefore hypothesis 5, as it questions the robustness of the results of regression 2.
To further test hypothesis 5, hence the robustness of the results of regression 1 & 2, an additional regression with robust standard errors was conducted. This analysis aims to test hypothesis 7, there is no performance difference in return on investment between MNC’s subsidiary strategy using
performance data of 2019. The results of the analysis support hypothesis 7 as there are no significant results for any subsidiary role type. (local implementer: t = 1.765, p-value = .086, b = .0203731; world mandate: t = 1.38, p-value = .173, b = .0132819; both relative to specialized contributor) (Appendix: Stata Output 12)
As these results are in direct contrast to those of regression 2, they further support hypothesis 5, as they question the robustness of the result of regression 1 & 2.
Discussion
This study aims to investigate whether there are performance differences between German multinational corporations using world mandate, specialized contributor or local implementer subsidiary role types.
The results support hypothesis 1, which investigated if there is a performance difference in return on investment between German MNCs pursuing different subsidiary strategies. As this confirms that, compared to companies using the local implementer subsidiary strategy, those companies using a specialized contributor subsidiaries strategy exhibit lower returns on their investment in the year 2018. This serves as a confirmation of Birkinshaw and Morrison’s (1995) results.
The subsequent test of hypothesis 2, whether MNC’s using the specialized contributor subsidiary strategy report lower return on investment relative to MNC’s with local implementer and world mandate subsidiaries, was supported too. The results show that those MNC’s conducting local implementer as well as the world mandate subsidiary strategies exhibit statistically significant higher return on investment for the year 2018 relative to those MNC’s pursuing a specialized contributor subsidiary strategy.
Although the results of regression 3 support hypothesis 3, that industry sectors affect the return on investment of MNC, one must be careful with deriving a conclusion. By looking closer into the underlying data, it appears that due to the small number of observations, there is only one MNC represented in each of these industry sectors. (Consumer Durables & Apparel – Adidas, RoI2018 = 15.23%; Energy – BP Europe, RoI2018 = 2.24%; Semiconductors and Semiconductor Equipment industry sector – Infineon, RoI2018 = 13.5%) (Appendix: Stata Output 7) This indicates, one cannot generalize the result of the regression to the whole industry sector. As each of these industry sectors are solely represented by one MNC, it does not mirror the situation within the industry sectors but rather only the situation within these specific MNC. Hence, the significant result for the industry sectors on the return on investment result from underrepresentation within those industry sectors variables. Furthermore, these underlying MNC either under- or outperformed other industry sector averages, wherefore they have been reported to have significant deviation regarding their return in investment. In contrast, the regression does not show significant statistical results for many other industry sector
variables which contain multiple MNC’s. This could be interpreted in a way that the characteristics of these industry sectors do not have any statistically significant influence on the return on investment of those companies. On the other hand, it is almost impossible to derive any relevant or significant result from a sample that small combined with such many categorical variables. As the size of the sample is the main problem with the reliability of the results of this paper, more on this later in the limitations section.
The same can be observed over the results of regression 4. The results for regression 4, testing whether subsidiary types and industry sectors explain deviations within return on investment of MNC’s, do reject hypothesis 4. Together with the industry sector categories, there is either no significant measurable difference between the return on investment of local implementer, world mandate and specialized contributor subsidiaries or there are too many factors compared to the observations within sample data. As already stated, this is one of the main limitations of this paper.
As the results of regression 4 do not confirm those of regression 1 & 2, an analysis of the robustness of regression 2 was conducted. The results of the robust regression (regression 5) support hypothesis 5 & 6, which question the robustness of regression 1 & 2 and question whether results of the robust regression do support those of the second regression. This means, measuring the sample data, after excluding influential points and reducing the effect of extreme under- or outperformers, indicates there is no statistically evidence of performance differences between MNC’s return on investment applying local implementer or specialized contributor subsidiary strategies. Solely a performance difference between specialized contributor and world mandate was statistically confirmed. Though it is not confirmed whether the performance differences are resulting from the effect of those subsidiary strategies or if there are other reasons for the performance variations in ROI 2018. This stands in direct contradiction to the results of regressions 1 & 2.
This is in line with the results of regression 6 as they support hypothesis 7, that there is no performance difference in return on investment between MNC’s subsidiary strategy using performance data of 2019. This means, since there is no statistical evidence for a difference in the return on investment of 2019 between multinational corporation applying different subsidiary role types, it highly questions the robustness of regressions 1 & 2 and thereby supports hypothesis 5.
Even if this research cannot prove whether there are performance differences between MNCs using different subsidiary role types, it points out how the influence of extreme under- and over performing MNC does affect the analysis in different ways, leading to different result. After extensive elaboration, this research recommends proceeding with the robust regression type to degrease the influence of influential points, hence extreme performing MNC.
Alternative Explanations & Limitations
One possible explanation for the differences in ROI within the subsidiary role types might be the uneven balance of subsidiary types within this sample: specialized contributor 10; local implementer
18 & world mandate 37. As some industries and product types clearly dominate the world industry and corporate structures often share the same subsidiary strategies, this is a possible reason for the dominance of world mandate strategies within this sample. Still, as this cannot be statistically confirmed by this research it is part of the recommendations for future research to investigate further if there are performance differences between MNC’s using different subsidiary role types and if so, where those differences come from.
As previously elaborated, one of the main limitations are the ratio between the number of observations within the sample and the number of categories within the control variable. While STATA might be able to calculate correct statistical results for the effect of the independent variable (three categories) on the dependent variable, the analysis was overpowered after adding the control variable (20 categories). Simply put, the number of categories within the IV and CV compared to the number of observations (DV) was simply too high to receive a reliable statistical result.
Suggestion for future research
As this study is unable to answer its research question, it is recommended to continue the research on this field. For future research, it is highly recommended to use a different sample of a larger size. Furthermore, as the results of the robust regression correct for the extreme effects of influential points by excluding or reducing their weigh, this seems to be the most suitable type of analysis.
Conclusion
As previously mentioned, this research did not find an answer to its question, whether there are differences in the return on investment of Multinational Corporations resulting from a different selection of subsidiary types. Although there was some evidence for performance differences within those MNCs there was not enough evidence to conclude that the source for those differences in the return on investment was a result of the appliance of different subsidiary role types.
Bibliography
Bartlett, C., & Ghoshal, S. (1986). Tap your subsidiaries for global reach. Harvard Business
Review, 64.
Bartlett, C. A., & Ghoshal, S. (2002). Managing across borders: The transnational solution. Harvard Business Press.
Birkinshaw, J. M., & Morrison, A. J. (1995). Configurations of strategy and structure in subsidiaries of multinational corporations. Journal of international business studies, 26(4), 729-753.
Brain, R. B. (2016, August). GICS system – sectors and industries. Retrieved August 12, 2020, from https://www.robertbrain.com/share-market/files/gics-sectors.pdf
Chen, J. (2020, April 27). Return on Investment (ROI). Retrieved August 11, 2020, from https://www.investopedia.com/terms/r/returnoninvestment.asp
Gupta, A. K., & Govindarajan, V. (1991). Knowledge flows and the structure of control within multinational corporations. Academy of management review, 16(4), 768-792.
Harzing, A. W. (2000). An empirical analysis and extension of the Bartlett and Ghoshal typology of multinational companies. Journal of international business studies, 31(1), 101-120.
Jarillo, J. C., & Martíanez, J. I. (1990). Different roles for subsidiaries: The case of multinational corporations in Spain. Strategic management journal, 11(7), 501-512.
Phillips, P. P., & Phillips, J. J. (2006). Return on investment (ROI) basics. American Society for Training and Development.
Porter, M. E. (1986). Changing patterns of international competition. California management review, 28(2), 9-40.
Roth, K., & Morrison, A. J. (1992). Implementing global strategy: Characteristics of global subsidiary mandates. Journal of International Business Studies, 23(4), 715-735.
Tolentino, P. E. (2002). Hierarchical pyramids and heterarchical networks: organisational strategies and structures of multinational corporations and its impact on world
development. Contributions to Political Economy, 21(1), 69-89.
Vayanos, D. (2003). The decentralization of information processing in the presence of interactions. The Review of Economic Studies, 70(3), 667-695.
White, R. E., & Poynter, T. A. (1984). Strategies for foreign-owned subsidiaries in Canada. Business quarterly, 49(2), 59-69.
Appendix
Graphic 1: Qnorm e_ij Residuals ROI2018
Graphic 3: Kernel density estimate ROI2018
Stata Output 1: Shapiro-Wilk test for normality (ROI2018 & ROI2019)
Stata Output 2: List of potential outliers (ROI2018)
Stata Output 3: List of potential outliers (ROI2019)
Graphic 7: Test for Homoscedasticity (ROI 2018)
Stata Output 6: Test for Multicollinearity (IV&CV&ROI2018)
Stata Output 8: Test H1
