فاصله های فرهنگ ملی به عنوان تعهدات خارجی: به مسئله سطح تجزیه و تحلیل

In this study, we focus on the use of measures of national cultural distance as a proxy for liabilities of foreignness. In particular, we focus on the dominant measure of national cultural distance: a linear combination of the differences between the two countries where the workplaces are located using indices from previously published research. Our question concerns whether measures of distance based on previously published indices at the national level are appropriate measures of cultural distance at the organizational level. Our results suggest that they are not; implications for theory and research are discussed.

As economic activity becomes increasingly global, there has been a notable increase in attention to cultural factors (Erez and Earley, 1993, pp. 4–7), especially as they may impact activities at firms that cross borders. Liabilities of foreignness represent one topic that has recently received considerable attention as part of this larger trend to study cultural factors. Of course, no single paper can adequately address a topic as broad as liabilities of foreignness. In order to navigate this broad topic in a way that allows us to make progress in a single study, we will focus on studies that have used national measures of cultural distance from previously published work as measures of liabilities of foreignness at the level of the organization. This literature has largely focused on decisions about entry mode when a firm established operations in a host country outside its home country. The remainder of this study is addressed specifically to the issue of whether these measures of national cultural distance are adequate proxies for liabilities of foreignness in this context. We proceed as follows. We will begin our study by discussing the concept of national cultural distance and how it has been linked with liabilities of foreignness at the firm level. Having introduced the concept, we will review empirical studies that have measured the construct and suggest some open questions that remain from this empirical work. Then, we will review the question of levels of analysis in measuring cultural distance and suggest some reasons for why the organizational level of analysis might need to be examined in more detail as a source of potential cultural distance. Having established the conceptual framework of cultural distance at the national and organizational levels of analysis, we will turn our attention to some data on cultural beliefs in seven countries measured at the organizational level of analysis. We will compare these measures at the organizational level of analysis with published results for the same measures at the national level of analysis; our interest is in the extent to which measures at these two levels converge. Based on our finding that they do not converge, we urge researchers who wish to have measures of the liability of foreignness to be more sensitive to the risk of relying on indices of cultural differences established by earlier studies.

The message of the analyses conducted comparing the organizational and national levels of analysis is clear: Systematic errors can result when measures derived at the national level are used to characterize phenomena at the organizational level. This is one form of what Robinson (1950) termed the ecological fallacy. We found evidence for this using the formula for computing distance along a cultural dimension derived by Kogut and Singh (1988) and the seven dimensions of culture described by Schwartz (1994). In comparing the difference between distances measured in this way at the country and organizational levels, we found significant main effects from both country and dimension. We included drift toward the US in the measures by Schwartz to control for the fact that the firm we studied was an American firm; the finding of significant main effects on the mean difference in distance between the national and organizational levels of analysis was unchanged. Hofstede et al. (1990) have reported on similarly systematic differences between the organizational and national levels of analysis using his measures of culture, which are the ones actually used by Kogut and Singh. We generalized this finding on cultural distance at different levels of analysis by looking at means and ranks at different levels of analysis. Our finding about means was that there was a significant main effect from dimension. Measures of distance based on the difference in means were likely to be characterized by larger errors for some dimensions than for others. This was true even after controlling for drift toward the culture of the US parent firm. Our finding about ranks was that there was a significant main effect from country. Measures of distance based on the difference in ranks were likely to be characterized by larger errors for some countries than for others. Once again, this was true even after controlling for drift toward the culture of the US parent firm. The principal implications of these results are related to the measurement of culture for the purpose of understanding organizational outcomes. We agree with Kogut and Singh (1988, p. 427) that cultural distance measured at the organizational level would be preferable to predict an organizational outcome like entry mode. Indeed, we would go a step further and suggest that it is no longer justified to assume that measures of cultural distance at the national level will be accurate measures of cultural distance at the organizational level. Hofstede et al. (1990) provided evidence on the general point about how culture would differ at the national and organizational levels using the scales that Hofstede himself had earlier developed. We have provided evidence for a similar effect of the same difference in levels on measures of cultural distance using the dimensions of Schwartz (1994). The robustness of the findings across different dimensionalizations of culture, different countries, and a variety of measures, including cultural distance between countries, mean scores on dimensions by country, and the ranks of dimensions by country, speaks for itself. Like all studies, ours has its limitations. First, the sample size of our current study is relatively small. Yet, we do not see any clear conceptual or empirical rationale for why our results would have been different had our sample size been much bigger. Second, similar to the Hofstede (1980) study, we have only examined a single American multinational. However, our purpose in discussing the structure of culture revealed by these multiple sites of one American company is fairly modest. We simply wanted to assess whether national measures of culture would replicate at the organizational level; one organization is a large enough sample to conclude that the answer is not necessarily. In addition, it is important to note that Hofstede et al. (1990) came to the same conclusion using a variety of units of various companies from a variety of home countries and using different scales to measure culture. We also hope in future work to provide a direct test of the power of measures at different levels of analysis to predict organizational outcomes. It is important to note that we are not suggesting that culture is not an important variable or that the construct of cultural distance is of no utility. Nothing we have found here precludes direct measurement of culture and cultural distance in the course of the study of organizations and organizational behavior. In fact, several useful approaches have recently been adopted in both organizational behavior and strategic management research areas. For example, Very et al. (1996) measured the acculturative stress felt by managers in recent European mergers by having them complete questionnaire items directly, including items from Hofstede (1980). O'Grady and Lane (1996) measured cultural beliefs, again using items from Hofstede, among the managers studied in their research on psychic distance. Brett and Okumura (1998) measured the cultural beliefs among participants in a negotiation exercise using the scales proposed by Schwartz (1994). It is also possible to study how national culture serves as a setting in which organizations are embedded. For example, Newman and Nollen (1996) studied the financial performance of European and Asian work units of one multinational company as a function of the congruence between management practices and national culture. They developed a theory of fit between practices used at the organizational level and national culture, proposing that greater fit would be associated with higher performance. They used Hofstede's (1980) dimensions of culture to characterize the various national cultures where these units operated and measured the practices used at each unit. Their results support the notion that national culture seems to mediate the effectiveness of work practices at the organizational level. While the mechanisms by which this effect took place were not measured directly, the authors did not assign national scores at the organizational level, avoiding the ecological fallacy. Another approach is to control for the effects of culture through the design of the study rather than by direct measurement: For example, studies have posed a contrast between foreign and domestic firms. For example, Shaver (1998) found systematic differences between American and non-American firms in choosing US plant locations. A study using a similar methodology has shown that foreign firms entering the US suffer from a significantly greater number of labor-related lawsuits Mezias, 1999, Mezias, 2000 and Mezias, 2002. Yet another approach is to use dummy variables to capture the differences between home and host countries. For example, Rosenzweig and Nohria (1993) compared the human resource management practices of foreign subsidiaries in the US with the human resource management practices at locations in between the US and each of the eight other home countries included among firms in the sample. Rather than imposing a national average computed from Hofstede's study, this work computes an average effect of country using the very same firms being studied. Thus, even though our finding suggests that using measures of distance in cultural beliefs at the national level of analysis to characterize the organizational level of analysis results in systematic bias, it does not preclude the use of cultural distance either conceptually or empirically. Rather, we view these results as cautionary. First and most importantly, we need to be acutely aware of the issue of levels of analysis in measuring the effects of culture or cultural beliefs. Thinking through the implications of the assumptions imposed by crossing levels of analysis is important for future work. For example, using the measure of cultural distance at the national level to characterize the firm level implies that there will be little variance among firms from the same country; this is illustrated by the fact that the same scores get assigned to all firms from the same country. However, the existing research suggests that there is significant within-country variation. For example, Vandello and Cohen (1999) found a wide variation along the individualism–collectivism dimension across 50 states within the United States. Furthermore, Chen et al., 1998 and Chen et al., 2001 found that the within-culture difference in individualism–collectivism between their US samples was in fact larger than the between-culture difference in individualism–collectivism between their US and Chinese samples. This suggests the possibility that multinational enterprises and local firms investigated might not be typical organizations in their own countries, and it is possible that the distance between them might be smaller than the distance they have with other firms from their own country. Hence, great caution needs to be made when we use the national cultural distance index as a proxy for differences between firms from different countries. As Mezias et al. (1999) have suggested, workplaces themselves are repositories of social information (Salancik and Pfeffer, 1978). The link between this social information and measures of national culture is an issue that needs better understanding and careful study. Merely asserting they are the same and plugging in national values as scores for the organization is no longer warranted. Measures of cultural distance used to predict outcomes at the organizational level need to be linked explicitly with mechanisms that cross levels of analysis. Implications of the assumptions implied by the discussion of these mechanisms need to be examined thoroughly. Our final point is that the time has come to rethink cultural distance both theoretically and empirically. In light of our findings, one of the most troubling aspects of dominant measures of cultural distance has been the failure to develop alternative measures. While we believe that differences in cultural beliefs do matter and are likely related to theoretical constructs like cultural distance and liability of foreignness, we also believe that other differences between cultures and countries matter. We believe that closer attention to industry norms and national systems—economic, legal, and political—that can affect the link between features at the national level and outcomes of the organizational level is warranted Orru et al., 1991, Earley and Singh, 1995 and Meyer et al., 1994. These are issues that need better understanding, as Luo and Mezias have suggested in their call for papers. Thus, despite finding some problems with currently predominant measures such as cultural distance, we remain firm believers that a deeper understanding of the construct of cultural distance and the concept of liability of foreignness is both possible and desirable.