Master thesis International Financial Management, August 15, 2008 Department of Economics and Business, University of Groningen
Thesis supervisor: Dr. G. de Jong
The Effect of Cultural, Economic and Institutional Distance on
Deal Characteristics of Cross-Border
Mergers and Acquisitions
Abstract
This paper contributes to the cross-border M&A literature by examining the effect of cultural distance on the deal characteristics of cross-border M&As (mergers and acquisitions). With the ratio of cross-border acquisitions relative to acquisition within the same country rising steadily, the wave of cross-border M&As expanded steadily after 1996 and reached a peak of $2.7 trillionin the first half of 2007. These recent developments stimulated empirical research in the domain of cross-border M&A activities, the majority of which with a focus on explaining success of cross-border M&A in terms of post-acquisition performance. This study deviates from this research tradition, and its significance lies in its focus on the specific mechanisms through which cultural distance might influence post-acquisition performance. This different focus is made explicit in the central research question of this study: what is the relationship between cultural dimensions on M&A deal characteristics? Moreover, this study intends to empirically test some model assumptions that in most studies are taken for granted and not subjected to any form of diagnostic checking. These model assumptions are implicit in the operationalization of cultural dimensions as cultural ‘distance’ constructs. The use of distance measures implies that country effects satisfy symmetry and similarity assumptions, two strong and testable model assumptions. Making use of a large data set of M&A data, this study derives models explaining five M&A deal characteristics of cross-border acquisitions by cultural, institutional and economic dimensions. Both in the derivation of these models, as in the investigation of M&A deal characteristics of domestic acquisitions, it is demonstrated that symmetry and similarity model assumptions are not supported by empirical data: neither culture, nor institution or economic conditions manifest themselves primarily in the format of distance constructs.
Table of contents
1. Introduction 1
2. Theoretical background 4
2.1 Cultural distance and cross-border M&As 4
2.1.1 Limitations of Hofstede and the cultural distance measure 6
2.1.2 Culture index by GLOBE 9
2.2 Economic distance and cross-border M&As 11
2.3 Institutional distance and cross-border M&As 12
2.4 Relatedness 13
2.5 Deal size 15
3. Theoretical model 16
3.1 Concepts and definitions 16
3.2 Hypotheses 19
4. Methods 21
4.1 Data sources 21
4.2 Sample 22
4.3 Measures and descriptive statistics 24
4.4 Econometric methods 31
5. Empirical results 33
5.1 Domestic M&As as the natural context to test the role of ‘distances’ 33 5.2 Cultural distance and M&A deal characteristics in cross-border M&As 36
5.2.1 The similarity and symmetry assumption 36
5.2.2 The model for cultural distance 41
5.3 Hofstede vs. Globe cultural dimensions 42
5.4 The final model 46
6. Conclusion and discussion 49
7. References 53
8. Appendices 57
Appendix A: Target Nation * Acquirer Nation Cross-Tabulation 57
Appendix B: Diagnoses of normality of model variables and model residuals 59
1. Introduction
Since past two decades the practice of mergers and acquisitions has gained popularity as the mode of international expansion for firms participating in the race of globalization. The growth of cross-border mergers and acquisitions (M&As hereafter) is fuelled by several factors, among them industry consolidation and privatization, liberalization of economies, and the opportunity to learn new knowledge and acquire new capabilities (Shimizu et al., 2004). Furthermore, the ratio of cross-border acquisitions relative to acquisitions within the same country is rising steadily. The wave of cross border M&A expanded steadily after 1996 and reached a peak of $2.7 trillionin the first half of 2007. Cross-border activity accounted for a record of 48.5% of total worldwide M&A activity for the first half of 2007 (Thomson Financial, 2007 Mergers & Acquisitions Review). These recent developments stimulated empirical research in the domain of cross-border M&A activities, the majority of which with a focus on explaining success of cross-border M&A in terms of post-acquisition performance. Earlier studies in this domain of cross-border acquisition activities have primarily focused on the explanatory role of legal and accounting systems in the performance of acquisitions (Kupiers et al., 2003; Rossi and Volpin, 2004). Since last decade research into the concept of cultural differences in relation to post acquisition performance of cross-border acquisitions has substantially increased.
an indirect effect through acquisition mode. Little research has focused on these specific mechanisms through which cultural distance influence post-acquisition performance, and one potential mechanism, culture influencing deal characteristics, is the main topic of my thesis.
Statistical studies examining cultural distance have used the Hofstede cultural distances in order to quantify cultural distances (see for example Morosini et al., 1998; Chakrabarti et al., 2004). In doing so, the majority of these studies use an index constructed by Kogut and Singh (1988) to measure cultural distance. The index arithmetically averages differences in scores for four Hofstede dimensions and produces one single cultural distance (CD) measure for each pair of countries. Using the simple Kogut and Singh index implicitly comes down to making two rather strong assumptions: that of cultural symmetry, and cultural similarity. The symmetry assumption is the result of including cultural differences in one’s theory, instead of adopting both cultural dimensions of target and acquirer. Hence, the assumption implies that the acquirer and target country effects are equal in size but opposite in sign. However, empirical research found no prove of symmetry in cross-border M&As (Shenkar, 2001). The similarity assumption refers to the use of one index composed of four cultural dimensions. The assumption implies that in all conditions all four cultural dimensions have an equal impact. Again, empirical proof of this assumption is scarce, so it is our challenge to investigate if these assumptions hold when empirically tested. In part of this testing procedure, we will make use of a data set that might seem not obvious: that of domestic acquisitions. However, models that explain M&A characteristics purely on the basis of distance concepts imply that the effect of culture is absent when that distance is small. The perfect case of small distances is that of domestic M&As: by definition, distance equals zero in domestic acquisitions. Hence, the dataset of domestic M&As will function as the most pure context for testing our model assumptions, and as benchmark model for the impact of cultural dimensions in case these model assumptions are to be rejected.
measuring cultural distances: the index of cultural distance by Hofstede and, subsequently, project GLOBE; a multi-phase, multi-method research program.
To achieve the research objective, this study attempts to investigate the following main research questions:
- Do cultural factors impact cross-border M&A deal characteristics?
- If an impact is present: is it appropriate to operationalize the mechanism of it in terms of cultural distances, or are crucial assumptions to do so, such as the symmetry and similarity assumptions, not satisfied, so that we better base the mechanism on individual cultural dimensions?
- Do institutional and economic factors play a role beyond that of cultural factors, and if so, what is the mechanism along which they operate?
To the best of our knowledge, the concept of cross-border M&A transactions in relation to cultural, economic and institutional distances has not been addressed before, at least not explicitly. Although studies have analyzed the effects of both cultural and institutional distance on cross-border M&A (Dikova et al., 2006), or both cultural and economic distance (Chakrabarti et al., 2004), we are not aware of any research probing the effect of all three variables in one empirical model.
2. Theoretical background
In this chapter the core concepts of cultural, economic and institutional distances as well as the control variables will each be examined and reviewed. Existing literature will be presented for all variables separately and focus will be placed at the cultural distance variable.
In the literature reviewed in this chapter, the majority of studies examine the relationship between cultural, institutional and economic factors and cross-border M&A post-acquisition performance, instead of the dependent concept deal characteristics we have chosen to examine. However, empirical literature on the latter topic is very limited, so that we can not do better than report on post-acquisition performance studies and hypothesize that factors being influential in those studies are strong candidates for being influential in M&A deal characteristics studies too.
2.1 Cultural distance and cross-border M&As
Cultural distance is a useful proxy for country risk as it indicates the differences between the acquirer’s and target country (Shimizu et al., 2004). Herewith, cultural distance represents the distance in the norms, routines and repertoires of the target and acquirer nations. The issue of cultural distance in relation to M&A performance has been studied to great extent in the international business literature; the major part of the studies being descriptive as the intangible concept of culture is complex to translate into quantifiable variables. However, although being a much researched topic, no conclusive results have emerged. Teerikangas and Very (2006) state that, with current knowledge, it is challenging to predict the nature and direction of the impact of cultural distances on M&A. In other words, there exists a lack of understanding of the effect of cultural distances in cross-border M&A (Stahl and Voigt, 2003; Teerikangas and Very, 2006).
that a combination with a culturally distant company could provide access to various routines and practices, which would enhance performance over time. Hence, cultural differences between merging firms are explained as being a source of value creation and learning. Subsequently, the authors propose that when firms acquire companies with different cultural backgrounds, they might apply diverse routines and repertoires in their acquisition process (Morosini et al., 1998).
Research by Morosini et al. (1998) is the single study in this field that measures the effect of a specific cultural dimension. The authors argue that the Hofstede dimension Uncertainty Avoidance can influence post-acquisition performance as this dimension has been associated with a preference for organizational rules and procedures favoring monitoring, planning and control (Hofstede 1980). Subsequently, it was found that Uncertainty Avoidance had a slightly positive but significant effect on post-acquisition performance (Morosini et al., 1998).
In their study Chakrabarti et al. (2004) use the index of culture drafted by Hofstede. Hofstede defined culture as following: “Culture is the collective programming of the human mind that distinguishes the members of one human group from those of another. Culture in this sense is a system of collectively held values. This definition applies to national as well as to corporate cultures.” Hofstede collected data from a large sample of employees of IBM from 40 countries in the time period 1967 to 1973 (Williamson and Dermot, 2002). He found that cultures differed mainly along four dimensions: Power Distance, reflecting the extent to which less powerful members of organizations and institutions accept and expect power to be distributed unequally (Hofstede & Peterson, 2000). Individualism versus Collectivism; the degree to which individuals are integrated into groups (Hofstede & Peterson, 2000). The third dimension comprises Masculinity versus Femininity, which represents the distribution of roles between the sexes (Hofstede & Peterson, 2000). A fourth dimension refers to man’s search for Truth, or, Uncertainty Avoidance. This dimension reflects to what extent members of a particular culture feel either comfortable or not in unstructured situations (Hofstede & Peterson, 2000). The fifth dimension, Confucian Dynamism, or the long vs. short term orientation, was later added. Subsequently, scores were developed for several countries on these different dimensions. The “cultural distance” between countries can be measured with the use of these scores.
2.1.1 Limitations of Hofstede and the cultural distance measure
The vast majority of researchers relating culture to cross-border M&A base their measures of cultural distance on Hofstede’s national culture scores. Arguments in favor of the Hofstede model focus on the assumed validity and reliability present in the Hofstede measures (Morosini et al., 1998; Kogut and Singh, 1988). However, as mentioned by Teerikangas and Very (2006) “one can question the extent to which Hofstede’s dimensions capture all relevant categories of cross-national differences.” In this section several limitations and drawbacks of the cultural distance measure and subsequently, the Hofstede scores, will be discussed.
single cultural distance (CD) measure for each pair of countries. However, as Shenkar (2001) makes explicit in his article, severe limitations to such a measure exist: “The appeal of the CD construct is, unfortunately, illusory. It masks serious problems in conceptualization and measurement, from unsupported hidden assumptions to questionable methodological properties, undermining the validity of the construct and challenging its theoretical role and application” (Shenkar, 2001).
The index constructed by Kogut and Singh (1988) assumes symmetry to exist in the measurement of cultural distances. Symmetry suggests that both the target and acquirer nation face the same cultural distance when involved in cross-border M&As. However, empirical results suggest otherwise: some studies found the acquirer nation to be of especially importance (Kogut and Singh, 1988) while other research found significant influence of the target nation to be present. None of the studies found symmetry to exist, which implies that the acquirer and target country effects are different in nature (Shenkar, 2001).
Furthermore, when calculating an average of the Hofstede dimensions one loses the opportunity to determine which of the four dimensions is detrimental in explaining possible cultural distance effects. One cannot assume all dimensions to be of equal weight and direction (positive or negative) in explaining cultural distance effects. Hence, although applying a single cultural distance index provides statistical straightforwardness, one looses valuable information in explaining the specific ways in which culture influences the practice of cross-border M&A.
Next paragraph will examine the limitations present in the construction of the Hofstede cultural dimension scores.
partly determined by their cultural background. This limitation stems from the ecological fallacy of attempting to predict individuals’ values or behavior from data about their culture (Williamson, 2002).McSweeney’s argument opposes the view of seeing human nature as wholly determined by external forces.
Furthermore, the cultural dimensions constructed by Hofstede, like many models in the field of cross-cultural research, examine culture in terms of bipolarized cultural dimensions: the either/or lists (Williamson, 2002). To illustrate, Hofstede classified cultures as clustering around two poles dimensions: individualistic or collectivistic, low-context or high-context, long-term oriented or short-term oriented etc. Although such a paradigm provides simplicity in separating cultural variables and facilitates the comparison among cultures, the paradigm offers limitations regarding “the intricacy, diversity, richness and dynamism of culture” (Williamson, 2002).
A subsequent drawback refers to the accurateness of the data used. The data collection chosen by Hofstede, through a single multinational corporation sample, has received substantial critique (Kirkman et al., 2004). Also given expression to by McSweeney: “The scale problem of Hofstede’s research is radically compounded by the narrowness of the population surveyed”. It is argued that the IBM employees are not a representative of their respective national cultures as of the company’s selective recruitment (McSweeney, 2002). As McSweeney points out: “In countries such as Taiwan, with cultures very different from that of IBM’s home country of the USA, IBM staff may be expected to be generally more unrepresentative of their national culture than are US staff of IBM.” (McSweeney, 2002). Furthermore, it can be questioned to what extent the data is still up to date. Cultural values by Hofstede are seen as static variables; however, one can argue that culture changes as a result of internal and external influences which are progressively present over the last decades due to globalization pressures.
Hofstedeian hegemony” (Javidan et al., 2006). In next section project Globe will be examined in more-depth.
2.1.2 Culture index by Globe
Globe is a research project that consists of over 150 researchers who conceptualized and developed measures of nine cultural dimensions. The nine dimensions are aspects of a country’s culture and can be used as an instrument to distinguish societies from another, subsequently, they possess important managerial implications. Globe studied cultures in terms of cultural practices and cultural values (Javidan et al., 2006). The nine dimensions studied by Globe include dimensions formulated by Hofstede, hence, the explanations of the dimensions Power Distance, Uncertainty Avoidance, Future Orientation, and Gender Egalitarianism can be considered similar in both models. Regarding the dimension of Collectivism, the Globe model makes a distinction between Institutional and In-Group Collectivism: Institutional Collectivism can be defined as the level to which societal and organizational institutional practices reward and encourage the collective distribution of resources and collective action. In-Group Collectivism can be described as the level of individuals expressing concepts such as pride, loyalty, and cohesiveness in groups of their society (Javidan et al., 2006). Subsequently, Globe formulated three new cultural dimensions: Assertiveness, which refers to the level to which individuals are assertive, aggressive, and confrontational in their relationships with others. Orientation is the level to which a society rewards and encourages her members for excellence and performance improvement. Finally, Humane Orientation is the level to which a society rewards and encourages members for being altruistic, generous, fair, and kind to others (Javidan et al., 2006).
2.2 Economic distance and cross-border M&As
In subsequent section we will focus on the relationship between economic distance and crossborder M&A. Although a substantial amount of research has focused on cultural -and to a lesser extent institutional- distance, very few research has incorporated economic distance as a factor influencing cross-border M&A. Moreover, the studies that have incorporated economic distance measures predominantly focus on cross-border bank M&A (Buch and DeLong, 2004; Schmautzer, 2006). The argument of increased transaction costs due to country differences in cross-border M&A applies to the concept of economic distance as well. This is supported by research of Schmautzer (2006) who examined the announcement effects of 96 international cross-border bank mergers between 1985 and 2005. He finds that similarity in the degree of economic development between bidder and target country positively influences target wealth creation.
Chakrabarti et al. (2004) investigate the effect of economic distance between the acquirer’s country and that of the target in terms of the difference in per capita income. According to the authors, differences in per capita income are often associated with major socio-economic differences between countries. Chakrabarti et al. (2004) measure economic distance with the use of the relative difference in per capita income in order to capture economic disparity. The authors found that the coefficient of variable economic distance between the target and acquiring country is statistically significant (Chakrabarti et al., 2004). According to the authors: “The positive coefficient suggests that the market reacts more favorably if the acquiring firm’s country is expected to grow at a higher rate than the target firm’s country. Markets appear to have a more negative view of cross-border acquisitions involving firms from countries with larger economic disparities. In fact, this indicates that the short-term returns of the acquirers are decreasing with the relative superiority of the acquirer nation’s economy as compared to the target’s economy”. Hence, markets have a negative view of mergers involving firms from countries with very different income levels.
shareholder wealth effects for developed-market acquirers are positive only when an acquisition is made in an emerging market (lower GDP) and not in developed markets (higher GDP). Furthermore, economic development (as measured by GDP) and the subsequently explained variable of institutional distance are found to be highly correlated: it was found that institutional features affect the property rights setting and the incomplete contracting problem in emerging markets (Chari et al., 2006). In next section the mentioned institutional distance will be analyzed in more-depth
2.3 Institutional distance and cross-border M&As
Countries differ in their institutional (governance) context and, as a result, M&A activity of firms is affected by the institutional environment of the home country as well as the other country in which the firms operate (Oliver, 1996). According to Hitt et al. (2006) national culture exists in a larger setting of the institutional environment. “National culture reflects the values of society which establish the norms for behavior. In turn the norms of behavior represent a dimension of the institutional environment” (Hitt et al., 2006). The distances in institutional environments may be based on different regulations, accounting standards, and value systems (Shimizu et al., 2004).
Hitt et al. (2006) examine the interrelationship between culture and institutions and conclude that international strategy, culture and institutions are inextricably interwoven. Buch and DeLong (2004) also emphasize the importance of legal aspects in cross-border M&A. They state that one could assume that the presence of a common legal system, and hence small institutional distance, has a positive impact on cross-border M&A. This can be explained by the fact that when the institutional distance between two countries is significant, conflict between managers and employees is likely to increase which in turn leads to decreased performance rates (Shimizu et al., 2004).
indicator of institutional quality while the acquirer firm belongs to a country that scores above the median1.
Dikova et al. (2006) examined the influence of formal institutional constraints on the likelihood of a merger completion and duration. It is argued that the legitimacy of an organization is reflected in its acceptance and approval by the legitimating environment (Kostova and Zaheer, 1999). Hence, if institutional environments of the countries in which the multinational corporation operates are more similar, the better able the corporation will be in responding appropriately to the legitimacy requirements (Kostova and Zaheer, 1999). Or, in other words, through their institutions, nations influence the norms according to which buying firms manage the acquisition process (Majidi, 2007). Dikova et al. (2006) argue that in M&As with partners from institutionally different environments it may take more time to address all aspects of the merger deal and meet regulatory requirements. Their found results indicate that the greater the institutional distance, the greater the likelihood of M&A deal abandonment. However, found results were rather conservative as the used sample composed solely developed economies with relatively similar institutional environments. As is stated: “maybe a sample with representatives at the extremes of the institutional spectrum is needed to find support for all our theoretical predictions” (Dikova et al., 2006).
In next sections we will focus on two of the control variables in our research: industry relatedness and deal size. Previous found effects of the control variables in relation to cross-border M&A are presented.
2.4 Relatedness
Relatedness is commonly defined as the common link in market, product or technology between the acquirer and the target (Singh and Montgomery, 1987). Studies on industry relatedness in cross-border M&A and post-acquisition performance have provided mixed
1
results. Results of some studies indicated that relatedness between the acquirer and the target improves post M&A performance (Healey et al., 1992; Slangen, 2006). To illustrate, Singh and Montgomery (1987) examined a sample of M&As completed in the 1970s and found that related acquisitions performed better compared to unrelated acquisitions. Higher degrees of relatedness can explain better performance as a result of operating synergies in overlapping businesses. Such synergies may be the result of economies of scale and scope (Muehlfeld et al., 2007) or market power. Furthermore, Stahl and Voigt (2008) performed a meta-analysis of 46 studies; one of their findings was that the effects of cultural differences vary depending on the degree of relatedness. As was stated: “the degree of relatedness is a potential moderator of the relationship between cultural differences and M&A outcomes because of its impact on the level of integration”(Stahl and Voigt, 2008).
Other studies advocate that industry relatedness is not an advantage, or find no relation at all. For example, Morosini et al. (1998) found that their regression coefficient associated with the relatedness variable was positive and non-significant. This indicated that “the degree of relatedness of the acquirer and the target firms did not influence sales growth of the combined firm during the two-year period following the acquisition” (Morosini et al., 1998). Subsequently, in a comparative research by Chakrabarti et al. (2004) the relatedness and cash flow measures of cross-border M&A were added as explanatory variables in long-term post-acquisition performance. The authors found both variables to be statistically insignificant. Hence, they do not add any explanatory power in their regressions for long-term performance of cross-border acquirers. (Chakrabarti et al., 2004)
Hagedoorn and Schakenraad (1994) already acknowledged the fact that strategic partnering is dependent of sectoral features. It was found that strong patterns exist in the preference for acquisitions across services and manufacturing industries (Kogut and Singh, 1988). Subsequently, in their empirical research Morosini et al. (1998) show that the type of industry can have a significant and positive effect on post-acquisition sales growth. This indicates the importance of controlling for industries in studies on cross-border M&As (Morosini et al., 1998).
2.5 Deal size
The control variable deal size has been incorporated in several studies and opposing results emerged, most likely as a result of different constructs of the variable. In their study of national cultural distance and cross-border acquisition performance, Morosini et al. (1998) defined size of an acquisition as the dollar value of the target’s net sales in the year of the acquisition. The authors found that the size variable was negative and significant, although very small. Hence, they concluded that larger size transactions resulted in a slightly poorer post-acquisition performance compared to smaller transactions.
However, opposing results were found by Slangen (2006). In his study of cultural distance and initial foreign acquisition performance Slangen assumes large acquisitions to show higher post-acquisition performance as they generally receive more attention and support (Slangen, 2006). He controls for the relative size of the acquisition in terms of the number of employees. Slangen found the relative size to possess a significantly positive effect on foreign acquisition performance.
acquired firm, relative to the size of the buying firm, appeared to be invariant to post-merger performance.
3. Theoretical model
In this chapter the theoretical model will be presented. First, the dependent, explanatory and control variables will be explained and defined. Subsequently, hypotheses will be presented for three research areas.
3.1 Concepts and definitions Dependent variable
The dependent variable in this research is defined as the concept deal characteristics. As this term constitutes a rather vague wording, a subsequent explanation is in place. The concept deal characteristics is composed of five measures: both a measure of shares: Percentage Shares Sought (%Sought) as well as four financial measures: Enterprise Value (LogEntValue), Equity Value (LogEqValue), Deal Size (LogDealSize), and the Price per Share (LogPrice/Share).
Lastly, Price per Share is the offered price per share of the target firm by the acquiring firm.
Important to note is that there exists an extensive assortment of variables on which M&A transactions can, and are, explored. The selection of the five measures of the dependent variable is based on two arguments. First, the selected measures in this paper are purely concerned with the acquisition activity itself, compared to most measures constituting broader firm-specific information. Second, the availability of transaction data varies considerably; the selected measures are known for a large part of the sample unlike other measures. We will elaborate on the selection process in section 4.3.
Explanatory variables
Cultural distance is defined as the distance in the norms, routines and repertoires of the target and acquirer nation. As mentioned, in order to quantify the cultural distance between acquirer and target country scores of both Hofstede’s model as well as the extension of project GLOBE are used. One should be aware we do not imply the GLOBE model to be guarded from limitations; however, it is expected that the enclosure of project GLOBE will enrich the examination of cultural distance among countries.
Economic distance is defined as the distance in per capita income of the target and acquirer nation.
Institutional distance is defined as the distance in basic requirements, efficiency enhancers and innovation and sophistication factors of the target and acquirer nation. Three potential sets of indicators have been evaluated: World Bank’s Doing Business, World Economic Forum’s Global Competiveness Index, and IMD World Competitiveness’ Overall Competitiveness index. The datasets and subsequent selection will be shortly described in next section.
178 for each specific regulations and the total Ease of Doing Business indicator for all countries, and no specific indices, or scores, on which these rankings are based, problems emerge in the adoption of these indicators in any model. Statistical models assume underlying variables to be normally distributed, which is not the case with rank data.
The IMD World Competitiveness indices look at the relationship between a country’s
national environment and the wealth creation process of enterprises and individuals (IMD
World Competitiveness Yearbook 2006). When incorporating the indices it was found that the dataset had 7 missing country values: Costa Rica, Ecuador, Egypt, El Salvador, Guatemala, Iran, Kuwait, and Morocco. Hence, as a substantial amount of countries (7 of 48) is absent in the IMD Competitiveness dataset, the indices were not deemed useful for the construct of institutional distance in this study.
The third, and selected, set of institutional measures was derived from the Global Competitiveness Report by the World Economic Forum (http://www.gcr.weforum.org/). The Competitiveness Report "assesses the ability of countries to provide high levels of prosperity to their citizens. This in turn depends on how productively a country uses available resources. Therefore, the Global Competitiveness Index measures the set of institutions, policies, and factors that set the sustainable current and medium-term levels of economic prosperity." The Global Competitiveness Report separates countries into three specific stages: factor-driven, efficiency-driven, and innovation-driven. Next description of the development stages is subtracted from the 2007-2008 Global Competitiveness Report.
In the factor-driven stage countries compete based on their factor endowments, primarily unskilled labor and natural resources. To maintain competitiveness at this stage of development, competitiveness is achieved mainly on well-functioning public and private institutions (pillar 1), appropriate infrastructure (pillar 2), a stable macroeconomic framework (pillar 3), and good health and primary education (pillar 4). These four pillars are together defined as Basic Requirements.
productive uses (pillar 8), the ability to adopt existing technologies to enhance productivity (pillar 9), and market size (pillar 10). Subsequent 6 pillars are defined as Efficiency Enhancers.
Finally, countries move into the innovation-driven stage; companies must compete by higher efficiency in the production of goods and services to enhance business sophistication (pillar 11) and through innovation (pillar 12). Last 2 pillars are defined as Innovation and Sophistication factors.
Control variables
Relatedness is defined as the common link in market or product between the acquirer and the target. The level of relatedness will be based on the industry of the acquirer and the target firms.
Deal size is defined as the value of the M&A transaction in US dollars. As scholars have shown that the sizes of an M&A will influence post-acquisition performance, size of the transaction will be controlled for in the study as well.
Industry is defined as the specific type of sector in which the acquisition took place. In order to test for industry effects, acquisitions classified by three industry sectors were incorporated: manufacturing, service, and finance. The selection of the three industries will be explained further in section 4.2.
3.2 Hypotheses
From the literature review presented previously, three specific areas are identified in which this study will enhance our knowledge regarding cross-border M&A and the effect of cultural dimensions. In next section for each research area a hypothesis will be formulated.
M&A. The same is true for institutional distance. Hence, our first research area concentrates on the role of cultural and institutional dimensions in case distances are zero.
Hypothesis 1: When distances are zero and no cultural distances have to be bridged, as is the case in domestic M&As, national cultural dimensions will still impact the M&A deal characteristics.
Subsequently, numerous studies have focused on the link between cultural distance and post-acquisition performance of cross-border acquisitions, of these, most studies assumed the existence of symmetry as a result of adopting the Kogut and Sing (1988) cultural distance index. As stated, in this study neither the existence of symmetry in cultural distance effects, nor the similarity of effects of the several cultural dimensions, will be assumed; hence, no simplified index will be applied. The assumption of similarity will be examined by incorporating all cultural dimensions as separate predictors, instead of aggregating their values into one index. The assumption of symmetry will be examined by incorporating cultural dimensions of both acquirer and target countries as separate predictors, instead of adopting a difference value as predictor. We expect to find evidence for the described critique by Shenkar (2001) of the Kogut and Singh index (1988). More specifically, two hypotheses can be formulated in this research area:
Hypothesis 2a: The effects of the cultural dimensions of any nation, both as target and acquirer, are not similar in size and sign. Hence, the assumption of similarity can be rejected.
Hypothesis 2b: The cultural dimensions of the target and acquirer nations cannot be modeled as distance effects, implying that the effects do not have equal size and opposite signs. Hence, the assumption of symmetry can be rejected.
distance and post-acquisition performance in cross-border M&As suggest a link between the two to exist. Our expectation is that this link can be explained by specific mechanisms, for example the specific deal characteristics under examination in this study. The same argument applies to the economic and institutional distance variable. Moreover, as was stated earlier, the mere finding in the literature that post-acquisition performance is related to cultural distance, is in itself no prove that cultural distance has an effect on the success of the acquisition. It might be the case that cultural distance has an impact on the acquisition deal characteristics. If that is true, one will find a relationship between culture and performance, but that relationship is the outcome of the existence of a third variable: acquisition mode. In addition, the impact of cultural, economic and institutional distances can best be determined in a relative way, rather than an absolute way. Therefore, the prime hypothesis of this study reads:
Hypothesis 3: There exists a relationship between the cultural dimensions of both target and acquirer nations and the deal characteristic of cross-border M&A under examination.
4. Methods
In this methods chapter, we will discuss the topics of data sources, sample used, measures of theoretical concepts and methods of statistical analyses. Those topics are highly interwoven: statistical methods are based on data assumptions, such as normality, that might require the transformation or raw data, or need samples of some size that might imply restrictions on the choice of measures. For that reason, it is inevitable to run ahead later discussions in some parts of this chapter.
4.1 Data sources
related companies as target. Subsequently, the data regarding the five measures that comprise the dependent variable deal characteristics as well as data of all three control variables were obtained from Thomson Financial Securities as well.
Regarding the explanatory variable cultural distance, the Hofstede indexes for the four cultural dimensions were derived from the researcher’s website http://www.geert-hofstede.com. Furthermore, the GLOBE cultural measures were subtracted from The GLOBE study of 62 societies, by House et al. (2004).
Data for the institutional distance were derived from the Global Competitiveness Report 2007-2008 by the World Economic Forum (http://www.gcr.weforum.org/). Lastly, the gross domestic product per capita figures for measurement of the economic distance were obtained from the World Economic Outlook (WEO) database.
4.2 Sample
The large majority of empirical studies (if not all) in the field of cultural and institutional distances focus on a small range of cultures in the sample under examination. Several studies use samples that consist of cross-border acquisitions in which one of the partners is from one singular Western nationality (Kogut and Singh, 1988; Weber et al., 1996; Morosini et al., 1998, among many). Other research use samples originating from solely developed economies (Dikova et al., 2006; Teerikangas and Very, 2006). Some empirical studies restrict their sample to deals of a specific size (Very et al., 1997), often choosing $100 million as the minimum deal size value (Chakrabarti et al., 2004). As a result, empirical findings derived from these studies might lose value in their applicability beyond their specific sample contexts. In order to get a sample that is as representative as possible, we decided to use a very broad sample of M&As.
geographical distribution is not proportional among the 48 countries. From the sample of 25,005 cross-border M&As, most of the merger activities took place in the United States (6828), followed by the United Kingdom (4362), Canada (1481) and Germany with 1241 cross-border M&As. Appendix A shows the Target nation * Acquirer nation Cross Tabulation: the number of domestic and cross-border M&A for each of the 48 countries.
Subsequently, two datasets were constructed: a cross-border M&A dataset and a domestic M&A dataset. The sample of worldwide M&A transactions is based on data for the manufacturing, service and finance related industries. These are three main categories out of six provided by Thomson Financial Securities (also: trade, natural resources and other). For all three sectors seven industries were selected, with the selection being mainly based on the size of provided datasets. For the manufacturing sector this resulted in the selection of the following industries: Aero, Communication Equipment, Computer, Electronic Equipment, Measuring Equipment, Software, and Transport Equipment. Subsequently, industries composing the service sector are: Amusement, Air, Hotels, Motion Pictures, Radio, Telecommunication, and Transport. Lastly, the industries representing the sample for the financial sector are: Banks, Credit Institutions, Insurance, Investment, Other, Real Estate, and Saving Banks. Furthermore, data encompasses the time period 01/01/1979-today and the transactions must be completed.
below 1%. In addition, availability is not equally spread over target countries; the dataset exhibits high missing rates for non-western targets. Excluding too many M&As because of missing data would make the resulting sample rather unrepresentative, especially since it would exclude mainly M&As with high distance scores for cultural, institutional and economic dimensions. Therefore, a compromise was sought between the number of and the diversity in M&As, versus the number and their diversity of M&A characteristics available for the research. Selecting the above mentioned variables (also incorporated in Table 2) as dependent measures in our study, is the outcome of this compromise.
Table 2: M&A characteristics made available by Thomson SDC
Data entries % data entries
Total M&A’s 341937
% of Shares Acq. 227009 66
% Owned After Transaction 228635 67
% sought 311068 91
Enterprise Value at Announcement ($mil) 48949 14
Equity Value at Announcement ($mil) 51601 15
Price Per Share 61510 18
Rank Value of Deal ($mil) 190676 56
Value of Transaction ($mil) 180540 53
Valid N (listwise) 23285 7
4.3 Measures and descriptive statistics
In this section the measures used in our study and their descriptive statistics will be presented and explained.
Dependent variable
Acquired. For that reason, only one was selected in the final set of dependent measures: % of Shares Sought, the variable with highest availability. Similarly: Rank Value of Deal ($mil) and Value of Transaction ($mil) correlate 0.993. As data availability is not very different, the non-rank measure was opted for.
Four out of five measures from this final selection of deal characteristics represent economic values. Like the majority of economic value variables, these variables are not normally distributed, but are skewed to the right. As will be discussed in the next section, the modeling approach used in this study requires normality. Therefore, data transformations were sought that would result in approximately normally distributed dependent measures. For right-skewed distributions the log-transform was applied to all four value-based dependent measures. The upper part of Table 3 presents descriptive statistics of all dependent measures, both with and without log-transform. Statistics for skewness, kurtosis and the Kolmogorov-Smirnov test statistic make clear that the log-transform highly improves normality in all four cases. In a formal test normality of transformed data is still rejected, but that rejection is mainly due to the sensitivity of the test statistic to the large sample size of our study. Graphical inspection of normality plots, exhibited in Appendix A for the case of Value of Transaction (or Deal Size), demonstrates that log-transformation is appropriate for producing normality. This also applies to the other value measures.
Explanatory variables
In order to quantify the cultural distance we make use of two cultural measurement models: the Hofstede indexes for the four cultural dimensions which range from 1 to 100 (no decimals), and subsequently, the GLOBE cultural measures for nine cultural dimensions which range from 1 to 6 (including 2 decimals). Table 5 presents the range of the scores of the specific cultural dimensions for both Hofstede and Globe. Table 3 presents descriptive statistics of a selection of the explanatory variables: those included in the final model presented in Chapter 5. The statistics are based on individual M&A entries in the cross-border data set, and can thus be regarded as weighted averages of country data, with the weights being the number of acquisitions per target country. Two different weighting schemes will be distinguished: weights based on the number of occurrences of a country as target, and as acquirer. For that reason, all explanatory variables are included twice in Table 3. However, differences between the two weighting schemes are small, as can be seen from the average values of the Hofstede measures for the several cultural dimensions: PDI (48.30 and 44.30), IDV (67.81 and 73.78), MAS (56.89 and 58.27) and UAI (53.70 and 49.54).
The economic distance is measured with the use of the gross domestic product per capita; figures are presented in U.S. dollars. Every single M&A transaction was matched with that country’s particular GDP per capita measure of the year in which the transaction was completed. From Table 3 one can see that the average value of the target GDP per capita is 24979 US dollar; the average value for the acquirer GDP per capita is 29510 US dollar.
can be seen from Table 3, the average values of the target and respectively the acquirer weighting for Basic Requirements (5.39 vs 5.56), Efficiency Enhancers (5.18 vs 5.39), and Innovation and Sophistication factors (4.99 vs 5.23) are rather similar.
Table 4 presents correlation coefficients of all pairs of independent variables, which enable us to detect whether there exists any risk of (multi)collinearity among the independent variables. The correlations indicate that especially institutional distance measures, with a .892 value for the target-country weighted correlation coefficient between the Efficiency Enhancers and Innovation and Sophistication factors, are a potential source of collinearity. We will continue the discussion of collinearity in the next section.
Control Variables
Downloads from the Thomson M&A database are organized by target industry sector. As explained in the first section of this chapter, we selected M&As from three different industry sectors: manufacturing, service and finance related industries. The first obvious choice for a control variable is to distinguish between these three industrial sectors of target.
Based on industry data, a further control variable can be defined: the relatedness of target and acquirer. Relatedness measures are operationalized using both the Macro Code (for example the broader category of Finance or Media) and the Mid Code (for example the more specific classification Bank or Publish) of both target and acquirer. This results in three categories of industrial relatedness: no relatedness (both Macro and Mid Code of target and acquirer are different), relatedness of 1 (target and acquirer have the same Macro Code, but different Mid Code), and a relatedness of 2 (target and acquirer match both on Macro Code and on Mid Code).
Table 5: variables, measures and data sources
Constructs Measure Data sources
1. Dependent variable: M&A deal characteristics - Percentage Shares Sought
- Enterprise Value - Equity Value - Deal Size - Price per Share
Score of 0 or 1 Log(EntValue) Log(EqValue) Log(DealSize) Log(Price/Share) Thomson Financial Securities 2. Explanatory variables: a. Cultural distance:
+ Hofstede cultural dimensions index: - Power Distance (PDI)
- Individualism (IDV) - Masculinity (MAS)
- Uncertainty Avoidance (UAI) + Globe cultural dimensions index:
- Uncertainty Avoidance (UA) - Future Orientation (FO) - Power Distance (PD)
- Institutional Collectivism (IC) - Humane Orientation (HO) - Performance Orientation (PO) - Family Collectivism (FC) - Gender Egalitarianism (GE) - Assertiveness (AS) Score 1-104 Score 1-91 Score 1-110 Score 1-104 Score 2.85 - 5.36 Score 4.73 - 2.80 Score 5.68 - 4.51 Score 3.25 - 5.20 Score 3.11 - 4.96 Score 3.20 - 4.94 Score 3.46 - 6.36 Score 2.50 - 4.12 Score 2.79 - 4.67 Hofstede’s Cultural Dimensions Index Globe Cultural Dimensions Index b. Economic distance: - GDP per capita In US $ World Economic Outlook database c. Institutional distance: - Basic requirements - Efficiency enhancers
- Innovation and sophistication factors
Score 2.71 - 6.14 Score 2.59 - 5.77 Score 2.47 - 5.77 Global Competitiveness Report 2007-2008 3. Control variables: - Industry - Relatedness - Deal size
Manuf., Fin., Serv. Score of 0, 1 or 2 Small: <100 mill. US$ and Large: ≥100 mill. US$
4.4 Econometric methods
Although data in our data set cover a period of 30 years, our hypothesis with regard to the relationships between M&A deal characteristics and cultural, institutional and economic distances are non-dynamic in nature. Our hypotheses assume both cultural and institutional distances to be relatively stable, so that country measures can be approximated by estimates that are constant over the time period, leaving only economic distances as a source of variation over time. As explained in the previous section, the time variability of economic distances was solved by substituting target and acquirer country GDP of the year of the transaction into the data set. As a result, models estimated on this data set possess a cross-section dimension, but no time-series dimension.
Models based on cross-section data can be estimated by Ordinary Least Squares if several assumptions satisfy. Normality is an important assumption that refers both to the dependent measure, and the explanatory variables. Normality can be checked in a univariate context, checking all variables individually, and in a multivariate context, checking the model residuals. In the previous section, reference was made to the outcomes of univariate diagnosis of model variables; both Table 3, and Appendix A present the outcomes of such checks. The Kolmogorov-Smirnov test statistic reported in Table 3 provides a formal test on normality; however, the test is of limited use in this study due to the large sample size. Normality plots as exhibited in Appendix A for the case of one dependent measure, Deal Size, provides more useful information on the degree of normality.
All potential model variables have been checked in the univariate context, both using the K-S test and normality plots. As indicated in the previous section, normality of all five dependent measures is rejected, in contrast to normality of the explanatory variables. However, for four out of five dependent measures we applied the log-transform: Enterprise Value, Equity Value, Deal Size, and Price per Share are significantly non-normal, but their log-transforms are, approximately, normal.
lack of heteroscedasticity, or the presence of non-linear relations. On all models estimated in this study, such residual diagnosis has been applied. Appendix A, Figure 9 presents the outcome of the residual diagnosis for the final model estimated on aggregate data with the log-transform of Deal Size as dependent measure. All three panels indicate that all model assumptions are satisfied. This figure is representative of all estimated linear regression models.
A subsequent assumption focuses on the independence of explanatory variables in a multivariate regression models: the absence of (multi)collinearity. Collinearity can be investigated before and after the estimation of the regression model: before, by checking bivariate correlations of explanatory variables, and after, by checking the variance inflation factors (VIFs) produced by the multivariate regression. Table 4 reports the bivariate correlations of our final model; as discussed in the previous section, the high correlation among institutional distance factors present a potential case of collinearity. Table A1 in Appendix B provides variance inflation factors of the final model including all institutional distances estimated on aggregate data. As can be seen from the Table, the VIF value for T_EfficiencyEnhancers is larger than the benchmark value of 10. Of the other VIF values, A_EfficiencyEnhancers is high with a score just under 10. As a result, evidence from VIF is in line with the checks based on bivariate correlations, and it was decided to exclude Efficiency Enhancers scores from the model.
5. Empirical Results
In this chapter the hypotheses from section 3.2 will be answered. Section 5.1 will focus on our first hypothesis regarding the effect of national cultural dimensions and institutional variables on domestic M&A. Next, section 5.2 will test the similarity and symmetry assumptions implied in the Kogut and Singh (1988) index. Furthermore, in this section the model for disaggregate data will be presented. Section 5.3 will compare the Hofstede and Globe model on their findings. In section 5.4 we will present the final model examining the impact of cultural, institutional and economic distances on deal characteristics of cross-border M&As. Furthermore, an advanced analyses section is presented in appendix C. In this section we present additional sub-models for the explanatory variables institutional and economic distances. Moreover, in this section sub-models analyzing the effect of the three control variables under examination will be presented as well.
5.1 Domestic M&As as the natural context to test the role of ‘distances’
In this section we will check if the ‘distance’ assumption is correct to the extent that it can also be applied for zero distances: the case of domestic M&As. Collecting all industries in an aggregate dataset, Table 6 shows the beta coefficients and their statistical significance levels of linear regressions for the four value related variables, and the logistic regression for seeking full ownership (both its beta’s and the Nagelkerke R2 have a different interpretation than the beta’s and the R2 in the linear regressions).
Table 6: Beta’s of regression equations explaining M&A deal characteristics by Hofstede’s cultural dimensions using aggregate data
PDI IDV MAS UAI R2
LogEnterpriseValue - 0.171*** -0.135*** 0.181*** 0.040 LogEquityValue -0.050*** 0.125*** -0.120*** 0.174*** 0.039 LogPrice/Share 0.047*** 0.449*** -0.080*** 0.305*** 0.175 LogDealSize - 0.207*** -0.083*** 0.100*** 0.037 Sougth100% - 0.036*** -0.02* -0.014*** 0.252 *: p<0.05; **: p<0.01; ***: p<0.001
that the Hofstede dimensions individualism versus collectivism (IDV) and uncertainty avoidance (UAI) consistently have a positive impact on the four different M&A deal characteristics, whereas masculinity versus femininity (MAS) consistently has a negative impact. The signs of the beta coefficients coincide the signs of bivariate correlations (not reported here), with one exception: power distance (PDI) has all negative correlations with the four M&A deal characteristics, but in a multivariate regression, collinearity with other Hofstede dimensions, especially IDV (see also the correlation matrix presented in Table 4 of Chapter 4) make the beta’s disappear. The strong positive role of IDV can be explained by its relation to general economic conditions. Confronting countries with lowest IDV (Guatemala, Ecuador, Venezuela, Colombia, & Indonesia) with those with highest IDV (Netherlands, United Kingdom, Australia, Canada, & United States) makes clear that in countries with high IDV one can in general expect M&A deal characteristics to have higher levels than in countries with low IDV. The explanation of the beta’s of the two remaining scales is less easy. Both MAS and UAI are only weakly correlated with other variables investigated. With Uncertainty Avoidance, one might assume that high levels of this variable contribute to the aim of seeking control in the acquisition, which would have an upward impact on all M&A characteristics. High levels of Masculinity would then have a reverse role: it prevents achieving high levels of control.
Summarizing: deal characteristics of domestic M&As are dependent on national culture; for some variables they explain up to 25% of variation. If the ‘distance’ assumption was to be true, we would have expected no cultural effect in domestic M&As. Therefore, the assumption that the effect of culture expresses itself solely through the distance between cultural dimensions of two countries should be rejected.
(Inn&SophF) has the strongest and positive impact, the other two variables, Basic requirements (BasicRequirm) and Efficiency enhancers (EfficiencyEnh) have a weaker and negative impact on the value variables. Different from the cultural distances, signs of the beta’s are not identical to the signs of the bivariate correlations (not reported here): these are all positive. Apparently, the direct effect of all Global Competitiveness scales is that they push deal characteristics in upward direction. Furthermore, due to relatively high collinearity in institutional distances, the best predictor, Inn&SophF gets the chance to collect all these positive influences, whereas the other two are left over with minor, negative side effects. However, different from the cross-border dataset, in the domestic dataset the collinearity is not so strong as to produce VIF’s larger than 10. Hence, in the domestic models all three institutional variables are included, although the reversal in signs of beta’s signals the presence of modest collinearity.
Confronting the two Tables 6 and 7, one can see that the extent to which deal characteristics can be explained by the two models is rather comparable, with the LogPrice/Share variable being the best predictable characteristic in both sets. Cultural dimensions somewhat outperform institutional dimensions in their ability to predict M&A deal characteristics.
Table 7: Beta’s of regression equations explaining M&A deal characteristics by Global Competitiveness Report institutional dimensions using aggregate data
BasicRequirm EfficiencyEnh Inn&SophF R2 LogEnterpriseValue -0.059*** -0.038** 0.160*** 0.016 LogEquityValue - -0.063*** 0.169*** 0.014 LogPrice/Share -0.084*** -0.122*** 0.518*** 0.158 LogDealSize -0.111 0.080*** 0.128*** 0.031 Sougth100% -1.122 3.657*** -1.153* 0.229 VIF 1.386 4.449 3.823 *: p<0.05; **: p<0.01; ***: p<0.001
5.2 Cultural distance and M&A deal characteristics in cross-border M&As
5.2.1 The similarity and symmetry assumption
In line with described research on cultural influences on M&A characteristics, we have run regressions of the five M&A deal characteristics of aggregate data on the Kogut and Singh (1988) index of national culture. The index consists of the weighted sum of the four scores for the Hofstede dimensions of the target and acquirer, with the weights being equal to the variance of the specific Hofstede dimension. First, to distinguish between the similarity and symmetry assumptions, we created two culture indexes: one for the target and one for the acquirer, beyond the distance index. Starting with the investigation of the similarity assumption, Table 8 shows the beta coefficients and their statistical significance levels of linear regressions for the four value related variables, and the logistic regression for seeking full ownership, when using both target and acquirer cultural index scores as explanatory variables.
Table 8: Beta’s of regression equations explaining M&A deal characteristics by target and acquirer cultural indices (assuming similarity)
CultDimIndexTarget CultDimIndexAcq R2 LogEnterpriseValue 0.045** 0.048** 0.004 LogEquityValue - 0.041** 0.002 LogPrice/Share 0.075*** - 0.006 LogDealSize - 0.032*** 0.001 Sougth100% -0.022*** -0.009*** 0.003 *: p<0.05; **: p<0.01; ***: p<0.001; -: non-significant at 5%
Table 9: Beta’s of regression equations explaining M&A deal characteristics by
individual Hofstede dimension scores of target and acquirer (no similarity assumption)
T_PDI T_IDV T_MAS T_UAI A_PDI A_IDV A_MAS A_UAI R2
LogEnter priseValue -.127*** -.132*** -.053** .176*** .091*** .100*** -.094*** .164*** 0.087 LogEquity Value -.108*** -.072** - .106*** .046** .109*** -.083*** .155*** 0.057 LogPrice /Share - .248*** -.070*** .246*** -.043* - -.044** .059*** 0.142 LogDeal Size -.048*** .023* -.016** .066*** - .110*** -.085*** .166*** 0.052 Sought100% -.013*** .020*** -.007*** -.003*** -.004** .012*** - -.010*** 0.209 *: p<0.05; **: p<0.01; ***: p<0.001; -: non-significant at 5%
When one compares the two tables, it is clear that the individual Hofstede dimensions lead to a regression model that is superior to the one based on the two indexes of cultural dimensions, both in terms of explained variation, as in terms of consistency of regression coefficients over the several acquisition mode characteristics. That finding leads a strong rejection of the similarity assumption. Focusing solely on Table 9, it is not difficult to conclude what contributes to this rejection. As in the domestic case, there is a consistent pattern amongst signs of beta’s (and signs of underlying bivariate correlations, not reported here) and the acquisition mode characteristics. UAI, of both target and acquirer, has a positive impact, whereas MAS has a negative. The pattern in PDI and IDV is slightly less strong compared to the domestic case, but again: PDI tends to have a negative impact, IDV a positive impact. It is obvious that the creation of a simple index of four variables, when two have a positive and two have a negative impact, is no good modeling practice: the several effects will cancel each other out.
assumption satisfies, only the regression coefficients of the two index variables will turn out to be statistically significant. If however any of the individual cultural dimensions enters the model with a significant coefficient, the similarity assumptions should be rejected. From Table 10 one can see that not only some, but most individual dimensions enter the model significantly, while the index variables explain the acquisition deal variables only weakly.
Table 10: Regression with both cultural indices and individual Hofstede dimension scores
CultIndT CulIndA T_PDI T_IDV T_UAI A_PDI A_IDV A_UAI R2 LogEnter priseValue -.078** -.129*** -.081** -.103*** .216*** .158*** .141*** .226*** 0.087 LogEquity Value - -.113*** -.081** -.055* .129*** .105** .146*** .209*** 0.057 LogPrice /Share -.102*** -.060** - .286*** .299*** - .049* .089*** 0.142 LogDeal Size -.024** -.119*** -.033** .031** .078*** .067*** .148*** .223*** 0.052 Sought 100% -.013*** - -.008*** .022*** - -.003** .012*** -.009*** 0.209 *: p<0.05; **: p<0.01; ***: p<0.001; -: non-significant at 5%
Moreover, all regression coefficients of the individual cultural dimensions have their sign in the same direction as in Table 9, whereas all the regression coefficients of the cultural indices switch sign, as compared to Table 8. This significant instability caused by extending the model with variables that according to the assumption should not matter, provides clear evidence that the assumption of similarity should be rejected, as was expected according our hypothesis.
Table 11: Regressions with cultural distance index (assuming symmetry) CultDistIndex R2 LogEnterpriseValue - 0.000 LogEquityValue - 0.000 LogPrice/Share 0.052*** 0.003 LogDealSize -0.018** 0.000 Sougth100% -0.006*** 0.003 *: p<0.05; **: p<0.01; ***: p<0.001; -: non-significant at 5%
In order to isolate the symmetry aspect, we compare Table 11 with Table 8, where similar regression equations were estimated with indices of national dimensions, in stead of single distances. Comparison shows that the model based on cultural dimensions outperforms the model based on cultural distances for all five acquisition deal characteristics. A clarifying case is provided by the first variable, logged Enterprise Value. In the version with two national indexes, the regression coefficients of both indexes are approximately equal in size (beta’s of .045 and .048, respectively, for target and acquirer). When transforming it into a distance variable, as one can see in Table 8, the two different effects cancel out, which results in no predictive power left. Hence, the symmetry assumption is rejected.
Table 12: Regressions with difference and sum scores of Hofstede target and acquirer dimensions to test the symmetry assumption
PDIsum PDIdiff IDVsm IDVdiff MASsum MASdiff UAIsum UAIdiff R2 LogEnter priseValue - -.134*** -. -.136*** -.104*** - .254*** - 0.087 LogEquity Value - -.093*** - -.108*** -.079*** - .196*** -.034* 0.057 LogPrice /Share -.052** - .222*** .119*** -.084*** - .232*** .121*** 0.142 LogDeal Size -.029** -.031*** .111*** -.056*** -.073*** .045*** .175*** -.069* 0.052 Sought 100% -.008*** -.005*** .016*** .004*** -.004*** -.003*** -.007*** .003*** 0.209 *: p<0.05; **: p<0.01; ***: p<0.001; -: non-significant at 5%
The last column of the table, containing the explained variation, is exactly equal to the last column of Table 10, which does not surprise, since the transformed variables cannot do better or worse than the original variables. More interesting is the spread of the significant regression coefficients over the ‘sum’ and ‘difference’ columns. More ‘sum’ regression coefficients are significant than ‘difference’ regression coefficients, and on top of that, all regression equations are dominated by ‘sum’ variables having the highest beta coefficient, and in this way contributing strongest to the explained variation. The clearest manifestation of the important role of the ‘sum’ variable is the uncertainty avoidance variable, in all linear equations the most powerful predictor. Inspection of Table 9 would have predicted the strong role of UAIsum: both T_UAI and A_UAI have relative strong beta’s with always the same, positive sign, making the sum variable much more appropriate to catch most of the effect, than the difference variable. The sign of the beta coefficients of the difference variable UAIdiff changes over the several equations, in contrast to the sign of UAIsum, again a further indication that the symmetry assumption is strongly rejected.
