2012 an improvement to kogut and singh

8
Research in International Business and Finance 26 (2012) 196–203 Contents lists available at SciVerse ScienceDirect Research in International Business and Finance journal homepage: www.elsevier.com/locate/ribaf Short communication An improvement to Kogut and Singh measure of cultural distance considering the relationship among different dimensions of culture Yener Kandogan School of Management, University of Michigan-Flint, 303 E. Kearsley, Flint, MI 48503, United States a r t i c l e i n f o Article history: Received 8 July 2011 Received in revised form 24 October 2011 Accepted 4 November 2011 Available online 17 November 2011 Keywords: Cross-cultural research/measurement issues Internationalization theories and foreign market entry Cultural distance a b s t r a c t National cultural distance construct has wide-spread use in the international business literature, with many applications. Despite its limitations as summarized by Shenkar (2001), the method in Kogut and Singh (1988) is commonly adopted by researchers to measure cultural distance. This article demonstrates that this method is a special case of the distance measure in Mahalanobis (1936) under the assumption of zero covariances between dif- ferent dimensions of culture. Further, it demonstrates that this assumption is not valid for several cultural dimensions of countries measured by Hofstede (1980), and suggests a simple modification to the method that corrects for this invalid assumption, and hence produces more accurate measures of cultural distance. The article concludes with a comparison of cultural distances as measured by the original and the modified version of the method. © 2011 Elsevier B.V. All rights reserved. 1. Introduction Various international business activities and cross-border managerial decisions are affected by differences in national cultures. These have been first demonstrated by Hofstede (1980) and the sub- sequent work of others. Despite its limitations later demonstrated by several researchers, most of the relevant work in the literature used a measure of cultural distance developed by Kogut and Singh (1988). Tel.: +1 810 2376675; fax: +1 810 237 6685. E-mail address: [email protected] 0275-5319/$ see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.ribaf.2011.11.001

Upload: mcbayon

Post on 06-Sep-2015

4 views

Category:

Documents


2 download

DESCRIPTION

Calculating cultural distance

TRANSCRIPT

  • Research in International Business and Finance 26 (2012) 196 203

    Contents lists available at SciVerse ScienceDirect

    Research in International Businessand Finance

    journal homepage: www.elsevier.com/locate/r ibaf

    Short communication

    An improvement to Kogut and Singh measure of culturaldistance considering the relationship among differentdimensions of culture

    Yener Kandogan

    School of Management, University of Michigan-Flint, 303 E. Kearsley, Flint, MI 48503, United States

    a r t i c l e i n f o

    Article history:Received 8 July 2011Received in revised form 24 October 2011Accepted 4 November 2011

    Available online 17 November 2011

    Keywords:Cross-cultural research/measurementissuesInternationalization theories and foreignmarket entryCultural distance

    a b s t r a c t

    National cultural distance construct has wide-spread use in theinternational business literature, with many applications. Despiteits limitations as summarized by Shenkar (2001), the methodin Kogut and Singh (1988) is commonly adopted by researchersto measure cultural distance. This article demonstrates that thismethod is a special case of the distance measure in Mahalanobis(1936) under the assumption of zero covariances between dif-ferent dimensions of culture. Further, it demonstrates that thisassumption is not valid for several cultural dimensions of countriesmeasured by Hofstede (1980), and suggests a simple modificationto the method that corrects for this invalid assumption, and henceproduces more accurate measures of cultural distance. The articleconcludes with a comparison of cultural distances as measured bythe original and the modified version of the method.

    2011 Elsevier B.V. All rights reserved.

    1. Introduction

    Various international business activities and cross-border managerial decisions are affected bydifferences in national cultures. These have been first demonstrated by Hofstede (1980) and the sub-sequent work of others. Despite its limitations later demonstrated by several researchers, most of therelevant work in the literature used a measure of cultural distance developed by Kogut and Singh(1988).

    Tel.: +1 810 2376675; fax: +1 810 237 6685.E-mail address: [email protected]

    0275-5319/$ see front matter 2011 Elsevier B.V. All rights reserved.doi:10.1016/j.ribaf.2011.11.001

    dx.doi.org/10.1016/j.ribaf.2011.11.001http://www.sciencedirect.com/science/journal/02755319http://www.elsevier.com/locate/ribafmailto:[email protected]/10.1016/j.ribaf.2011.11.001
  • Y. Kandogan / Research in International Business and Finance 26 (2012) 196 203 197

    Management scholars used this cultural distance measure to assess team performance (Gibson,1999), effectiveness of training (Tung, 1982), and conflict management in teams (Von Glinow et al.,2004). The same construct was also used by researchers in the strategy field to measure the efficiency ofthe global value chain (Govindarajan & Gupta, 2001), that of knowledge transfer and diffusion (Ghoshal& Nohria, 1989), institutional strength (La Porta et al., 1998), corporate governance (Bushman et al.,2004), as well as the impact of cultural differences on market entry (Brannen, 2004) and on subsidiaryrelations (Gupta, 1987). In finance, it was used to determine the capital structure (Chui et al., 2002),financial systems (Kwok & Tadesse, 2006), as well as foreign portfolio investment (Aggarwal et al.,2012). Last but not least, in accounting, national cultural differences have been shown to impinge onaccounting practices (Salter & Niswander, 1995).

    In several occasions, the results have been inconsistent. For example, in determining the foreignmarket investment location, while Davidson (1980) suggests that cultural similarity encourages moreinvestment, Dunning (1988) argues that larger cultural distance helps overcoming transaction andmarket failures and thus promotes foreign direct investment. In another application of this constructto predict the choice of entry mode into a foreign market, Kim & Hwang (1992) argue that multina-tional enterprises want less control over their operations in culturally distant foreign countries. Incontrast, some empirical evidence suggests a positive correlation between control and cultural dis-tance (Boyacigiller, 1990). In one last example of its application, Gomez-Mejia and Palich (1997) andothers use this construct to assess the performance of multinational firms with inconsistent results.

    The measurement and application of cultural distance to these variables in management and otherfields is clearly an important part of the process in understanding their effects. Given its widespreaduse as illustrated above, it is no doubt very important to measure cultural distances between nationscorrectly. Also to discover meaningful relationships that are useful for practitioners, accurate measure-ment of cultural difference is critical. However, partly driven by the above-mentioned inconsistencyin the results, Shenkar (2001) challenged the methodology used in measuring the cultural distanceconstruct, as well as its conceptual properties, pointing out to its hidden assumptions. He suggestedthat these methodological properties, the assumptions of corporate homogeneity within a nation, lackof intra-cultural variation, and reliance on single company data, result in measurement biases. Thisarticle points out to another source of measurement bias and suggests a simple correction. Such cor-rections are especially important, especially after the seminal work of Tung & Verbeke (2010) on thefuture of study of distance in international business. They studied various important issues that hadrisen as a result of recent studies, and urged the international business scholars to develop an impec-cable command of the full. . . arsenal of instruments for measuring distance dimensions and providingdistance scores. Several researchers are already working towards this goal (Brewer & Venaik, 2011).The suggestions made in this article will hopefully contribute in this direction as well.

    The rest of the article is organized as follows: after briefly reviewing various measures of distancefrom the mathematics realm, it is demonstrated that the Kogut and Singh (1988) method is a specialcase of a more general method with the assumption of zero covariances between the different dimen-sions of culture. The validity of this assumption is questioned using the actual data on various culturaldimensions of countries as measured by Hofstede (1980) and demonstrated not to be the case. In par-ticular, statistically significant positive covariance exists between individualism and power distancedimensions, as well as between individualism and uncertainty avoidance dimensions. A significantnegative correlation is also observed between power distance and uncertainty avoidance dimensions.A simple modification to the method is suggested that corrects for this invalid assumption and takesinto account these significant correlations. The paper concludes with comparison of cultural distancesas measured by the original and the modified version of the method. It is found that cultural distancemeasured using the original Kogut and Singh method could under- or over-estimate the distance asmuch as 60%.

    2. Measures of distance

    There are numerous methods of measuring distance, which is basically a numerical descriptionof the space between objects. Mathematics generalizes the concept of distance and the metrics ofdescribing the elements of some space to be close to or far from each other. The space can have

  • 198 Y. Kandogan / Research in International Business and Finance 26 (2012) 196 203

    multiples of dimensions and mathematicians have developed various multi-dimensional measures ofdistance. The most ordinary measure of distance is the Euclidean distance. This is what one wouldmeasure with a ruler and can be proved by the Pythagorean Theorem between two elements with ndimensions:

    Eij =n

    d=1(Id

    i Id

    j)2

    (1)

    The Euclidean distance is a special case of a much more general method of measuring distances,introduced by Mahalanobis (1936). It is a useful method of determining the similarity between twoelements:

    Mij =

    (Ii Ij)T S1(Ii Ij) (2)

    Ix is the vector for element x with n dimensions as follows:

    Ix =

    I1x

    I2x

    ...

    Inx

    (3)

    S is the covariance matrix for n dimensions of the space:

    S =

    V1 CV12 . . . CV1n

    CV21 V2 CV2n

    .... . .

    ...

    CVn1 CVn2 . . . Vn

    (4)

    where V are CV the variance and the covariance of the dimension(s) in the superscript, respectively.Note that if the covariance matrix is an identity, the Mahalanobis distance is reduced to the

    Euclidean distance. In other words, this special case would require that there is no covariance amongthe dimensions of the space, and the unity variance for each dimension.

    Kogut and Singh (1988) offer a simple standardized quantitative measure of cultural distance to beused alongside other data from the cultural dimensions of Hofstede (1980):

    KSij =1n

    nd=1

    (Idi

    Idj)2

    Vd(5)

    where KSij is the cultural distance between countries i and j, idx is the index of a country x in thedimension d, Vd is the variance of the index for the dimension d, and n is the number of culturaldimensions. Note that the above distance measure is a special case of the Mahalanobis distance. As

  • Y. Kandogan / Research in International Business and Finance 26 (2012) 196 203 199

    Table 1Covariances across Hofstedes cultural dimensions.

    Dimensions Mean Variance 1 2 3 4

    1. Power distance (PDI) 61.2 422.02. Individualism (IDV) 41.1 505.6 0.636**

    (0.00)3. Masculinity (MAS) 49.4 260.4 0.039 0.073

    (36.9) (26.4)4. Uncertainty avoidance (UAI) 65.9 490.2 0.210** 0.172* 0.013

    (3.67) (7.01) (45.5)5. Long term orientation (LTO) 44.3 754.8 0.347** 0.395** 0.180 0.024

    (4.21) (2.58) (17.9) (45.1)

    90% and 95% significance are denoted by * and **, respectively. P-values are given within parentheses.

    illustrated below in the Kogut and Singh measure, the covariances across all dimensions are assumedto be zero:

    Mij =

    [ I1i I1j I2i I2j . . . Ini Inj ]

    V1 0 . . . 0

    0 V2 0

    .... . .

    ...

    0 0 . . . Vn

    1

    I1i

    I1j

    I2i

    I2j

    ...

    Ini

    Inj

    =

    nd=1

    (Idi

    Idj)2

    Vd

    =

    n KSij

    (6)

    3. Suggested method

    As Table 1 demonstrates, the assumption of zero covariances across dimensions does not hold.The table shows the summary the mean, the variance for power distance, individualism masculin-ity, uncertainty avoidance and long term orientation cultural dimensions in Hofstede (1980) and thecovariances between these dimensions using all 77 countries or areas in his dataset. Accordingly,power distance and individualism are negatively correlated with individualism and positively corre-lated with uncertainty avoidance and long term orientation, which are all statistically significant at95% level of confidence. Furthermore, individualism is negatively corrected with uncertainty avoidanceand long term orientation.

    Since the assumptions that reduce the Mahalanobis distance to the one used by Kogut and Singh(1988) do not hold, this paper suggests using a modified version recognizing the non-zero covariancesamong the dimensions. The modified version would basically square the original Mahalanobis distanceand divide it by the number of dimensions so that it is comparable to the Kogut and Singh measuresin magnitudes as follows:

    MMij =14

    IPDIi

    IPDIj

    IIDVi

    IIDVj

    IMASi

    IMASj

    IUAIi

    IUAIj

    T

    c11 c12 c13 c14c21 c22 c23 c24c31 c32 c33 c34c41 c42 c43 c44

    IPDIi

    IPDIj

    IIDVi

    IIDVj

    IMASi

    IMASj

    IUAIi

    IUAIj

    (7)

    The above distance measure has a similar functional form to the Kogut and Singh method, yetrecognizes the non-zero covariances among various dimensions. Only four primary dimensions aretaken into account and long term orientation dimension is omitted due to the lack of measurement

  • 200 Y. Kandogan / Research in International Business and Finance 26 (2012) 196 203

    Fig. 1. Distances according to modified Mahalanobis and Kogut and Singh. Modified Mahalanobis, Kogut and Singh.

    of this dimension for many countries. The covariance matrices are given in Table 2 for the Kogut andSingh and the suggested modified method for comparison purposes.

    4. Comparisons

    The next set of figures and tables provides a comparison of the cultural distances using the originalKogut and Singh and the suggested modified Mahalanobis methods. Fig. 1 gives the frequency distri-bution of distances between all possible 2896 pairs of countries in Hoftstedes analysis. Accordingly,the Kogut and Singh method tends to overestimate the distance between countries that are culturallyvery similar and very distant, and underestimate the distance between countries with some but notvery big differences in culture. This can be seen with fewer pairs of countries with distances smallerthan 1 and larger than 4, and larger number of country pairs with distances between 1 and 4 usingthe modified method. Table 3 provides descriptive statistics of cultural distances using both meth-ods. While the means under each are the same, the smallest and largest distances are both smallerunder the modified method with a smaller standard deviation, which implies that the Kogut and Singhmethod tends to overestimate the cultural distance with greater variability. In fact, computation ofthe percentage difference in measurements under these two methods show that the Kogut and Singhmeasurements are an average of 1.42% higher than the modified method. The largest overestimationby the Kogut and Singh method is as big as 63% and the lowest underestimation is as big as 73%.

    Table 2Covariance matrices.

    Kogut and Singh (103) Modified Mahalanobis (103)2.369 0 0 0 4.4049 2.287 0 0.3880 1.978 0 0 2.287 3.330 0 0.1370 0 3.840 0 0 0 3.840 00 0 0 2.040 0.388 0.137 0 2.139

    Table 3Descriptive statistics of distance measures.

    Modified Mahalanobis Kogut and Singh Deviation

    Minimum 0.0193 0.0221 73.02%Maximum 9.9087 10.4233 63.50%Average 2.0473 2.0473 1.42%Standard deviation 1.4148 1.5621 29.71%

  • Y. Kandogan / Research in International Business and Finance 26 (2012) 196 203 201

    Table 4Over- and under-estimation by Kogut and Singh.

    PDI IND MAS UAI estimate1 0 0 0 41%0 1 0 0 41%0 0 1 0 0%0 0 0 1 5%1 1 0 0 64%1 0 1 0 21%1 0 0 1 19%0 1 1 0 19%0 1 0 1 30%0 0 1 1 2%1 1 0 0 55%1 0 1 0 21%1 0 0 1 37%0 1 1 0 19%0 1 0 1 23%0 0 1 1 2%1 1 1 0 48%1 1 0 1 53%1 0 1 1 11%0 1 1 1 18%

    1 1 1 0 23%1 1 1 0 23%1 1 1 0 48%

    1 1 0 1 7%1 1 0 1 64%1 1 0 1 56%

    1 0 1 1 24%1 0 1 1 11%1 0 1 1 24%0 1 1 1 13%0 1 1 1 18%0 1 1 1 13%1 1 1 1 41%

    1 1 1 1 4%1 1 1 1 32%1 1 1 1 41%1 1 1 1 45%

    1 1 1 1 45%1 1 1 1 4%1 1 1 1 32%

    Table 4 lists the deviations of Kogut and Singh method from the modified Mahalanobis methodwhen there are differences by 1 unit in only one, two, three and all four cultural dimensions betweentwo countries compared. The outcomes of only distinct situations are presented. In particular, perthe nature of the distance, the order of countries in finding distances does not matter. Hence, if all ofthe differences in various dimensions between two countries are reversed, the amount of under- orover-estimation remains the same. In the first set of situations listed in the table, there are differencesin only one of the dimensions. Accordingly, when the difference is only in the masculinity dimen-sion, the two measures of distance produce the same result. Similarly, when the difference is only inuncertainty avoidance, the deviation in the measures is only 5%. However, when the difference is onlyin power distance dimension or only in the individualism dimension, the Kogut and Singh methodunderstimates the distance by 41%.

    The next set of situations illustrates the deviations between the two measures when there aredifferences in two cultural dimensions. Differences in power distance and individualism cause thelargest deviation. When a country is higher (or lower) in both dimensions, the Kogut and Singh measure

  • 202 Y. Kandogan / Research in International Business and Finance 26 (2012) 196 203

    Table 5Measures of cultural distances among select countries.

    Country i Country j PDI IDV MAS UAI MM KS estimateFrance Spain 11 20 1 0 0.71 0.27 62%East Africa India 13 21 15 12 1.16 0.61 47%Arab countries Thailand 16 18 18 4 1.17 0.63 46%France Thailand 4 51 9 22 2.81 1.62 42%Arab countries East Africa 16 11 11 16 0.78 0.46 41%Germany United States 5 24 4 19 0.84 0.50 40%Brazil Germany 34 29 17 11 0.99 1.44 45%India United States 37 43 6 6 1.22 1.78 46%Russia United States 53 52 26 49 3.20 4.87 52%Mexico United States 41 61 7 36 2.24 3.54 58%Germany Mexico 46 37 3 17 1.30 2.09 60%Arab countries United States 40 53 10 22 1.64 2.68 64%

    In computing, the distances in each dimension, country js index is subtracted from country is index.

    seriously underestimate the distance by as much as 64%. When one country is higher in one of thesedimensions and lower in the other, it overestimates the distance by 55%. When there is a differencein power distance and a difference in masculinity or uncertainty avoidance, the underestimation isreduced from 41% down to 21% or 19%, respectively. A similar situation exists when the differencein individualism is accompanied by a difference in these two dimensions. Differences in masculinityand uncertainty avoidance dimensions continue to result in small deviations between the distancemeasures compared. Similar observations can be made when there are differences in three of the four,or in all four cultural dimensions. Same direction differences in power distance and individualism causelargest underestimation by Kogut and Singh, and opposite direction differences in these cause largestoverestimation. These deviations get smaller when there are additional differences in masculinity anduncertainty dimensions. For example, the size of the underestimation when the differences in powerdistance and individualism are in the same direction (64%) is reduced first to 48% when additionaldifference exists in masculinity or to 53% when the additional difference is in uncertainty. Further,when one more additional difference exists in both masculinity and uncertainty avoidance, the sizeof the underestimation is further reduced to 41%. The same is true when the differences in powerdistance and individualism are in opposite directions. In this case, the overestimation is reduced from55%, down to 23% or 7%, when masculinity or uncertainty differences also exist, respectively, and thenfurther down to 5%, when differences in both also exist.

    Lastly, Table 5 gives the absolute value of differences in four Hosftedes cultural dimensions andmeasurement of cultural distances using both methods for select countries representing variousregions of the world. The table contains instances of the largest percentage underestimations andlargest percentage overestimations by the Kogut and Singh method for these countries. Consistentwith previous discussions in Table 4, it appears that differences in masculinity do not cause any majordifferences in the measurements of these methods. However, large differences between countries inthe other three dimensions that are correlated with each other seem to cause underestimation oroverestimation of the cultural distance. When there are large differences between the countries inthe power distance coupled with large differences in either or both individualism and uncertaintyavoidance dimensions, the Kogut and Singh method overestimates the cultural distance. However,when differences between countries in individualism are small, the Kogut and Singh method tends tounderestimate the cultural distance. These are consistent with the frequency distribution of distancesdiscussed earlier.

    References

    Aggarwal, R., Kearney, C., & Lucey, B., 2012 Gravity and culture in foreign portfolio investment. J. Bank. Finan. 36(2), 525538.Brannen, M.Y., 2004. When Mickey loses face: recontextualization, semantic fit, and the semiotics of foreignness. Acad. Manage.

    Rev. 29 (4), 593616.Brewer, P., Venaik, S., 2011. Individualism-collectivism in Hofstede and GLOBE. J. Int. Bus. Stud. 42 (3), 436445.

  • Y. Kandogan / Research in International Business and Finance 26 (2012) 196 203 203

    Boyacigiller, N., 1990. The role of expatriates in the management of interdependence. J. Int. Bus. Stud. 21 (3), 357381.Bushman, R.M., Piotroski, J.D., Smith, A.J., 2004. What determines corporate transparency? J. Accounting Res. 42 (2), 207252.Chui, A.C.W., Lloyd, A.E., Kwok, C.C.Y., 2002. The determination of capital structure: is national culture a missing piece of the

    puzzle? J. Int. Bus. Stud. 33 (1), 99127.Davidson, W.H., 1980. The location of foreign direct investment activity: country characteristics and experience effects. J. Int.

    Bus. Stud. 11 (2), 922.Dunning, J.H., 1988. The eclectic paradigm of international production: a restatement and some possible extensions. J. Int. Bus.

    Stud. 19 (1), 131.Gibson, C., 1999. Do they do what they believe they can? Group efficacy beliefs and group performance across tasks and cultures?

    Acad. Manage. J. 42 (2), 138152.Gomez-Mejia, L.R., Palich, L., 1997. Cultural diversity and the performance of multinational firms. J. Int. Bus. Stud. 28 (2), 309335.Ghoshal, S., Nohria, N., 1989. Internal differentiation within multinational corporations. Strategic Manage. J. 10 (4), 323337.Govindarajan, V., Gupta, A., 2001. The Quest for Global Dominance: Transforming Global Presence into Global Competitive

    Advantage. Jossey-Bass, San Francisco.Gupta, A., 1987. SBU strategies, corporate-SBU relations, and SBU effectiveness in strategy implementation. Acad. Manage. J. 30

    (3), 477500.Hofstede, G., 1980. Cultures Consequences. Sage, New York.Kim, W.C., Hwang, P., 1992. Global strategy and multinational entry mode choice. J. Int. Bus. Stud. 23 (1), 2953.Kogut, B., Singh, H., 1988. The effect of national culture on the choice of entry mode. J. Int. Bus. Stud. 19 (3), 411432.Kwok, C.C.Y., Tadesse, S., 2006. National culture and financial systems. J. Int. Bus. Stud. 37, 227247.La Porta, R., Lopez de Silanes, F., Shleifer, A., Vishny, R., 1998. Law and finance. J. Polit. Econ. 106 (6), 11131155.Mahalanobis, P.C., 1936. On the generalised distance in statistics. Proc. Natl. Inst. Sci. India 2 (1), 4955.Salter, S.B., Niswander, F., 1995. Cultural influence on the development of accounting systems internationally: a test of Grays

    (1988) theory. J. Int. Bus. Stud. 26 (2), 379397.Shenkar, O., 2001. Cultural distance revisited: towards a more rigorous conceptualization and measurement of cultural differ-

    ences. J. Int. Bus. Stud. 32 (3), 519535.Tung, R., 1982. Selecting and training procedures of US, European, and Japanese multinational. Calif. Manage. Rev. 25 (1), 5771.Tung, R.L., Verbeke, A., 2010. Beyond Hofstede and GLOBE: improving the quality of cross-cultural research. J. Int. Bus. Stud. 41

    (8), 12591274.Von Glinow, M., Shapiro, D., Brett, J., 2004. Can we talk, and should we? Managing emotional conflict in multicultural teams.

    Acad. Manage. Rev. 29 (4), 578592.

    An improvement to Kogut and Singh measure of cultural distance considering the relationship among different dimensions of ...1 Introduction2 Measures of distance3 Suggested method4 ComparisonsReferences