I. Introduction

Understanding whether national culture affects a society’s likelihood to generate risk-seeking firms is important for effective policy-making and for improving corporate governance. It can enrich discussions on government policies that encourage entrepreneurship and innovation. A grasp of the impact of cultural influences on corporate risk-taking would allow policy-makers to better customize their policies for firms with different risk appetites, thus promoting more competitive business environments. Understanding the impact of culture on corporate risk-taking decisions is also important to the internal conduct of multinational firms. Internal decisions in multinational firms, such as the decision to pursue a risky R&D project, require well-orchestrated responses from executives with diverse cultural backgrounds. Even in firms with standardized operating procedures, the interpretation of various financial decisions can vary among executives from different societies as a result of their cultural differences (Tse et al. 1988). Accounting for the impact of cultural influences on decision-making allows the firms themselves to accommodate and adapt to such differences, hence diminishing “noisy” interactions among executives and errors in decision-making.

This study employs four dimensions of national culture identified by Hofstede (2001) and an international sample of 50,000 firms spread across 400 industries in 51 countries to analyze the effects of cultural differences on corporate risk-taking. More specifically, it tries to identify the channels through which cultural values can influence corporate risk-taking. Culture can affect the institutional and economic development at the macro level, the industrial diversification and industry concentration at the market structure level, as well as the corporate and individual decision-making at the micro level, all of which may in turn influence firm risk-taking decisions.

Previous literature has shown that national culture does in fact predict cross-country differences in the degree of institutional and economic development. Culture has been linked with creditor rights and investor protection (Stulz and Williamson 2003), with judicial efficiency (Radenbaugh et al. 2006), with corporate governance (Doidge et al. 2007), with bankruptcy protection and insolvency management (Beraho and Elisu 2010) and with overall levels of transparency and corruption (Husted 1999). Research has further established that national culture has an impact on the composition and leadership structure of boards of directors (Li and Harrison 2008) and also on individual decision-making at the micro level (Hilary and Hui 2009; Halek and Eisenhauer 2001; and Graham et al. 2009). On the other hand, attitudes towards risk are likely to be indirectly affected by culture through many of the factors listed above, as well as directly by national cultural norms, which may encourage or deter risk-taking.

This paper is not the first to study the impact of cultural values on corporate risk-taking. The extant literature has briefly studied the relation between culture and risk-taking, but has mostly focused on firms in the banking and the financial sectors (Houston et al. 2010; Kanagaretnam et al. 2011; Lehnert et al. 2011; Li and Zahra 2012). For example, Kanagaretnam et al. (2011) show that aggressive risk-taking activities by banks are more likely in societies with low uncertainty avoidance and high individualism. They show that cultural differences between societies have a profound influence on the level of bank risk-taking, and the ability to explain bank financial troubles during the recent financial crisis. On the other hand, Griffin et al. (2012) show that uncertainty avoidance is negatively and individualism is positively associated with firm-level riskiness in the non-financial sector (in the manufacturing sector).

This paper innovated in at least four ways. First, this paper takes a more holistic approach to the study of cultural influences on corporate risk-taking by studying not only the banking and the financial sectors, but all industries in a market economy. We take this approach in order to capture cross-industrial differences in risk-taking. The influence of cultural factors, such as national uncertainty aversion, may be of greater importance for firms in more informationally opaque industries such as information technologies, financial services, oil extraction, and chemicals, where information uncertainty is higher relative to manufacturing and industrial firms, because of the greater complexity of operations and the difficulty of assessing and managing risk. Thus, we test whether corporate risk-taking in informationally more opaque industries is more sensitive to a country’s national cultural norms. Second, we differentiate between the direct and indirect effects of national culture on firm risk-taking. We specifically test whether cultural norms remain important in determining corporate risk-taking behaviors even after taking into account their impact on the institutional, economic and industrial environments. Third, unlike previous research which has used standard ordinary least squares analyses, we model both the direct and indirect effects of culture on risk-taking by employing a hierarchical linear mixed model. The hierarchical linear mixed model allows testing multi-level theories, simultaneously modeling variables at the firm, industry and country level without having to recourse to data aggregation or disaggregation as previous cultural economics studies have had to do. Fourth, by using a hierarchical linear model in explaining firm-level risk-taking, we can model not only the firm, industry and country-level influences on risk-taking, but also their cross-level interactions.

This paper finds that:

• Culture impacts corporate risk-taking directly and not merely though indirect channels such as the legal and regulatory frameworks.

• Corporate risk-taking is higher in societies with low uncertainty avoidance, low tolerance for hierarchical relationships and in societies which value individualism over collectivism, with these effects even more accentuated in societies with better formal institutions.

• Additionally, firms in countries ranking high in uncertainty-aversion and low in individualism take significantly less risk in industrial sectors which are more informationally opaque (e.g. finance, IT, oil refinery and mining), compared to firms in countries lower in uncertainty-aversion and higher in individualism.

• Risk-taking by foreign firms is best explained by the cultural norms of their country of origin.

• These cultural dimensions are not proxying for legal constraints, economic development, bankruptcy costs, insurance safety nets, or many other factors.

The results of this study inform both theory and policy in several ways. First, these findings strengthen the argument that the same institutional rules can produce different economic outcomes in culturally-different societies. Second, they imply that policy-makers should take into account cross-cultural values and norms when drafting policies that promote competitive business environments. Third, they enrich governmental discussions on policies that address risk-taking in informationally opaque sectors.

The remainder of the paper is organized as follows. Related literature is reviewed in section II. Section III details the data and its limitations. Section IV describes the hypotheses under analysis. Section V presents the empirical methodology, while Section VI discusses and interprets the results. Sections VII and VIII conduct additional identification analyses on accentuating/mitigating factors and distinguish between foreign and domestic firms. Section IX concludes.

II. Literature Review

Several research studies in the financial, accounting, and management literatures have explored the importance of cultural values in decision-making. These studies find that culture can explain the institutional, legal and economic environments of a country at the macro level which can influence corporate risk-taking decisions, and offer evidence of the impact of culture on financial decision-making by individuals at the micro level beyond traditional economic arguments.

At the micro level, culture has (unsurprisingly) been shown to affect individual risk-taking behaviors. Breuer et al. (2011) find that individualism is linked to overconfidence and overoptimism and has a significantly positive effect on individual financial risk-taking and the decision to own stocks. Tse et al. (1988) show that home culture has predictable, significant effects on the decision-making of executives. Two decades later, Graham et al. (2010), using survey data in the U.S., also show that CEOs are not immune to the effects of culture. They find that CEOs’ decision-making is strongly influenced by cultural values such as uncertainty-aversion.

At the macro level, cultural heritage has been linked to corporate governance, investor protection, creditor rights, bankruptcy protection, judicial efficiency, accounting transparency, and corruption. Doidge et al. (2007) find that cross-cultural differences explain much more of the variance in corporate governance than observable firm characteristics. Hope (2003a) shows evidence that both legal origin and culture (as proxied by Hofstede’s cultural dimensions) are important in explaining firms’ disclosure practices and investor protection. In fact, he finds that although legal origin is a key determinant of disclosure levels, its importance decreases with the richness of a firm’s information environment, while culture still remains a significant determinant. Licht et al. (2005) find that social norms of governance correlate strongly and systematically with high individualism and low power distance. Stulz et al. (2003) find that cultural heritage, proxied by religion and language, predicts the cross-sectional variation in creditor rights better than a country’s trade openness, economic development, legal origin, or language. Other studies find that culture predicts judicial efficiency and the transparency of accounting systems. Radenbaugh et al. (2006) find that countries in the Anglo cluster have an accounting system which is more transparent and less conservative than either the Germanic or the Latin accounting systems. Beraho et al. (2010) show that cross-cultural variables have a direct influence on the propensity to file for bankruptcy and on insolvency laws. Lastly, both Getz and Volkema (2001) and Robertson and Watson (2004) link cultural differences to corruption levels.

Furthermore, recent research has also linked cultural variables to economic and market development, although the evidence is mixed. Guiso et al. (2006) find that national culture impacts economic outcomes, by influencing national savings rates and income redistributions. Kwok and Tadesse (2006) find that culture explains cross-country variations in financial systems, with higher uncertainty-avoidance countries dominated by bank-based financial systems, rather than by stock-markets. Kirca et al. (2009) show that national culture impacts the implementation of market-oriented practices (i.e., generation, dissemination, and utilization of market intelligence) and the internalization of market-oriented values and norms (i.e., innovativeness, flexibility, openness of internal communication, speed, quality emphasis, competence emphasis, inter-functional cooperation, and responsibility). Lee and Peterson (2000) show that only countries with specific cultural tendencies (i.e., countries which emphasize individualism) tend to engender a strong entrepreneurial orientation, hence experiencing more entrepreneurship and global competitiveness. On the other hand, Pryor (2005) argues that cultural variables do not seem related to the level of economic development and are not useful in understanding economic growth or differences in levels of economic performance across countries. Additionally, Herger et al. (2008) also argue that cultural beliefs do not seem to support or impede financial development. This mixed evidence points to the idea that national culture might only indirectly influence economic and market development through its effects on the legal and institutional contexts.

The institutional and economic environments have been shown to affect corporate risk-taking decisions. There is a small strand of literature which has explored corporate risk-taking around the world which reflects countries’ institutional and economic environments. For example, Laeven and Levine (2009) show that risk-taking by banks varies positively with the comparative power of shareholders within each bank. Moreover, they show that the relations between bank risk-taking and capital regulation, deposit insurance mechanisms, and bank activities restrictiveness, depend critically on the bank’s ownership structure. Claessens et al. (2000) show that corporations in common law countries and market-based financial systems have less risky financing patterns, and that the stronger protection of equity and creditor rights is also associated with less financial risk. Overall, while the literature is relatively small, national culture has been indirectly linked with corporate risk-taking decisions in formal studies, although most of them only analyze the banking sector.

Culture has also been directly linked with corporate risk-taking, although again, most studies have focused on either the financial or the manufacturing sectors separately. Kanagaretnam et al. (2011) show that banks in high uncertainty avoidance societies tend to take less risk, whereas banks in high individualism societies take more risk. However, they do not control for institutional variables such as corporate governance, bankruptcy protection, judicial efficiency, transparency, and corruption, which have shown to be affected by national cultural norms and which could at their turn affect corporate risk-taking. Griffin et al. (2012) study the impact of culture on firms in the manufacturing sector in the period 1997-2006. To the best of our knowledge, they are the only ones who use a hierarchical linear mixed model to analyze the impact of culture on corporate risk-taking. They show that individualism has positive and significant direct effects, while uncertainty avoidance has negative and significant direct effects on corporate risk-taking.

This paper contributes to the literature on the impact of culture on firm risk-taking in several ways. While previous studies have studied either the direct or the indirect effects of culture on risk-taking, this paper tries to reconcile the two strands of literature and assess them simultaneously by using a hierarchical linear mixed model. This allow to test whether cultural norms remain important in determining corporate risk-taking behaviors even after taking into account their impact on the institutional, economic and industrial environments. Moreover, this paper extends the analyses of Griffin et al. (2012) and Kanagaretnam et al. (2011) to capture cross-industrial differences in risk-taking. Given the importance to national and global economies of the highly leveraged sector of finance, or the highly innovative sector of IT, or the highly risky commodity industries¹, and given that firms in these industries are markedly different from manufacturing firms and have been more adversely affected by the recent global economic crisis, it is very important to understand the role of culture on cross-industrial variation in corporate risk-taking.

III. Data

This section details the data, describes the cross-cultural variables, the cross-firm risk-taking measures and the country-level, industry-level, and firm-level control instruments. We also discuss limitations of the cultural dimensions and offer guidance about how they could bias the results.

A. Measuring National Culture

The national cultural dimensions employed in the analysis are derived from a psychological survey into national and organizational culture conducted by the Dutch sociologist Geert Hofstede². The analysis makes use of four dimensions of national culture that distinguish 66 countries from one another: (1) tolerance of uncertainty; (2) the degree of group integration; (3) immobility between social classes; and (4) division of roles between genders. The dimensions come from a very large sample of employee values scores³ collected by IBM between 1967 and 1973 in 66 countries: 31 advanced economies and 35 emerging markets. Subsequent studies⁴ over the following decades have validated and extended these results on occupationally different populations: school teachers, airplane pilots, high school students, etc.

The first index explored is national uncertainty aversion (UAI) which deals with a society’s tolerance for uncertain, unknown, or unstructured situations. UAI is defined as “feeling uncomfortable with uncertainty and ambiguity, and therefore valuing beliefs and institutions that provide certainty and conformity (Hofstede 2001).” People in uncertainty avoidant cultures favor an orderly structure in their organizations, institutions and personal relations and prefer well-anticipated events. Emerging economies tend to be more uncertainty-averse than developed economies. A simple mean-comparison t-test at the 5 percent level shows that, on a scale from 1 to 100, the average UAI score for low income countries is 70, six points higher than the average UAI score for high income countries (see Figure A1 in Appendix A for more details). This observation is consistent with the idea that growth has an inherent risk component to it, and suggests that the lower a country’s ambiguity aversion, the more business ventures it will pursue and thus, the higher its economic development. In fact, this hypothesis is verified by the modeling work of Giordani and Zamparelli (2011), who show that the lower the ambiguity aversion, the higher the R&D efforts, and ceteris paribus, the higher the economic performance. Our sense is that firms in countries low is uncertainty-aversion will benefit from the upside risky effects of their risky financial actions in periods of economic upswing and rapid development, whereas in periods of economic downturn they are likely to run into problems when confronted with the downside risky consequences of their actions.

The second dimension used is individualism (IDV). It describes the relationship between the individual and the greater society. Hofstede defines it as “a society in which the ties between individuals are loose. Everyone is expected to look after himself and his immediate family only. Collectivism stands for a society in which people from birth onwards are integrated into strong, cohesive in-groups, which throughout people’s lifetime continue to protect them in exchange for unquestioning loyalty (Hofstede 2001).” Developed countries tend to be, on average, more individualistic than emerging economies. In fact, there is a 27 point mean-difference between the mean IDV score of high income countries (57) and the mean IDV score of low income countries (25), significant at the 5 percent level. Whether wealthier countries are more individualistic because there is less need to be reliant on other people, or whether economic development is facilitated by individualism (the invisible hand and the ardent pursuit of self-interests) are two hypotheses which are worth investigation. Our sense is that collectivism has many advantages at one stage of economic development, after which it becomes a burden on growth. In an underdeveloped economy, collectivism can be advantageous because it provides some kind of informal insurance against hazards and potential risks. However, once a society becomes more economically developed, collectivism can prove disadvantageous because it can interfere with the functioning of the economy by contributing to corruption and by providing a disincentive to investing and accumulating wealth, both because one is expected to share with the larger in-groups and because one can rely on free-riding on others for economic support (Ball 2001). To account for these effects, we will control for overall corruption at the country level, and we will have an interaction variable between individualism and the stage of economic development to account for the hypothesized nonlinear effects of collectivism / individualism on growth and risk-taking.

The third dimension we employ for analysis is power distance (PDI), which measures “the extent to which less powerful members of a society accept and expect that power is distributed unequally (Hofstede 2001).” A high PDI index means that national elites hold relatively authoritarian views, and that authority is based on tradition rather than on secular arguments. It also characterizes highly stratified societies that value conformity more than independence. Low income countries (67) tend to have, on average, higher power distance scores than high income countries (48) and the 19 point mean-difference is significant at the 5 percent level (see Figure A1 of Appendix A for more details). National wealth can itself be interpreted as both a cause and a consequence of small power distance scores. We believe that wealth is negatively correlated with power distance because development goes together with the growth of the middle classes in a society, which forms a bridge between the powerful and the powerless. On the other hand, high power distance societies are also highly authoritarian and authoritarian political systems which tend to not innovate as much, thus growing poorer in the long-run.

The fourth dimension analyzed is masculinity (MAS). It is defined by Hofstede as referring “to the dominant gender role patterns: the patterns of male assertiveness and female nurturance” (Hofstede 2001). ‘Masculinity’ stands for societies where social gender roles are clearly defined: men are supposed to be assertive, tough and focused on advancement and earnings; women are supposed to be more modest, nurturing, and concerned with the quality of life. ‘Femininity’ stands for societies where the gender roles overlap: both genders are supposed to be modest, tender and concerned with the standard of living (Hofstede 2001). While the choice of naming this index “masculinity” is an unfortunate one because it implies gender-role stereotypes which might not be necessarily true, the index still points to important cross-country differences. High income economies tend to have a slightly larger MAS score of 51, compared to low income countries which have a mean score of 42, but the 9 point difference is not statistically significant at the 5 percent level.

In Hofstede’s terms, these four dimensions are independent of each other, operate along a continuum between two extremities, and can be combined in different ways. Table 1 shows the pair-wise correlations between the four cultural dimensions discussed in this paper. While all the other variables are uncorrelated at the 1 percent level, individualism is negatively correlated (-0.66^*) with immobility between social classes (power distance). This is due to the fact that both low power distance and high individualism correlate with economic development. Once economic development is statistically controlled for, the correlation disappears (-0.27). These results point to the idea that the four cultural dimensions capture dissimilar values and societal norms which will be useful in analyzing corporate risk-taking decisions.

Table 1.

Correlation Matrix of Hofstede’s National Cultural Dimensions

Notes: Pair-wise correlations of national cultural variables. Partial correlations controlling for the level of economic development in parentheses. The star (^*) represents significance at the 1 percent level.Source: Hofstede 2001.

Table 1.

Correlation Matrix of Hofstede’s National Cultural Dimensions

	Uncertainty Aversion	Individualism	Power Distance	Masculinity
Uncertainty Aversion	1
Individualism	-0.21 (-0.30)	1
Power Distance	0.15 (0.16)	-0.66* (-0.27)	1
Masculinity	0.05 (0.05)	0.09 (0.14)	0.07 (0.09)	1

Notes: Pair-wise correlations of national cultural variables. Partial correlations controlling for the level of economic development in parentheses. The star (^*) represents significance at the 1 percent level.Source: Hofstede 2001.

View Table

Table 1.

Correlation Matrix of Hofstede’s National Cultural Dimensions

	Uncertainty Aversion	Individualism	Power Distance	Masculinity
Uncertainty Aversion	1
Individualism	-0.21 (-0.30)	1
Power Distance	0.15 (0.16)	-0.66* (-0.27)	1
Masculinity	0.05 (0.05)	0.09 (0.14)	0.07 (0.09)	1

Notes: Pair-wise correlations of national cultural variables. Partial correlations controlling for the level of economic development in parentheses. The star (^*) represents significance at the 1 percent level.Source: Hofstede 2001.

Even if a fair share of differences in cultural values can be attributed to personal, individually-characteristic factors, much more is related to the society in which an individual has been brought up. Societies share a system of societal norms which consist of the value systems shared by major groups in the population. These societal norms have allowed the development and preservation of institutions with fairly persistent structures and ways of functioning such as: the family, the school, the political system and the legislation. These institutions tend to reinforce the societal value system in such a way that a member of a society will get rejected if he does not follow the value system (”the mental software”) of the society he belongs to (Hofstede 2001). By analyzing matched samples of local IBM employees of marketing and customer service positions in 66 countries, a homogenous group of people with similar backgrounds, similar education levels, type of work and personality, Hofstede manages to eliminate idiosyncratic preferences and to obtain relative crosscountry indices which reflect national cultural differences.

It is important to understand that these dimensions of culture are relative - societies are compared to other societies. As Hofstede suggests, they should only be used as a framework for understanding cultural differences between countries and not as laws set in stone. In other words, without making a comparison, a country score is meaningless.

D. Measuring Industry Informational Opacity

The high scale of divergence between bond raters implies that financial, insurance, oil and gas mining, and IT firms are inherently more opaque than other types of firms. Bond rating agencies such as Moody’s and S&P split more often over these types of firms (Morgan 2000). Uncertainty over the financial firms comes from certain assets, loans and trading assets which are particularly difficult to monitor or price. Uncertainty over the oil and gas mining firms stems from the fact that outsiders have a hard time in determining the market value of their primary assets: their oil and gas reserves.

This paper follows Huang (2008) in developing for each industry an index measuring the amount of private information available, or inversely, the informational opacity of each industry. Huang (2008) uses Durnev et al. (2004) and Rajan and Zingales’s (1998) indicator of “relative firm-specific stock return variation”, which measures stock price informativeness at the industry level. The technical details on how to construct such an index for a wide array of industries can be found in the appendix, but the rationale is the following. Using high-frequency firm-level data, stock returns can be decomposed into market-related, industry-related and firm-specific components. When there is little firm-specific information available, stock prices become less informative, and most of the variation in stock price returns can be explained by industry or market related factors.

Following Huang (2008), the stock price informativeness of U.S. firms is calculated and used as a proxy for intrinsically given information availability at the industry level.⁶ This is justifiable because: 1) the U.S. market is the market with the most efficient information environment, where new information is rapidly (and efficiently) incorporated into stock returns, and 2) using the stock price informativeness of U.S. firms avoids any risks of endogeneity. The opacity indicator is constructed based on weekly stock returns of all U.S. public-listed firms, from 2000 to 2005. A five-year window was chosen to measure stock price informativeness because it is small enough to capture cross-industrial differences, yet long enough to represent more than a transient trend in stock price informativeness. Appendix B describes the technical details in creating the indicator and lists both the industry informativeness and its’ inverse, the industry informational opacity.

Table 2 ranks the main industries in terms of informational opacity and the results look quite intuitive. Table B in Appendix B provides a much grainer distinction of the measurements, reporting the informational opacity for more than 140 industries and in the course of the analysis the finer breakdowns are used. The pattern is that stock prices of newer industries, such as finance or the computer industry are more informationally opaque, whereas stock prices of the “old economy” industries, such as manufacturing, construction or public administration reveal more firm-specific information. A casual inspection also suggests that low opacity industries are also the ones using simpler production and organization technologies, compared to the highly opaque industries (chemicals, petroleum refinery, and finance).

Table 2.

Industry Informational Opacity

Notes: R-squared represents the R² of regressing firm specific return variations on industry-wide and market-wide returns. The informativeness index is calculated as one minus average R² of the regressions using each firm in an industry. The more information contained in the stock prices of individual firms, the lower the R²will be. The informativeness index is highly skewed and mildly leptokurtic, therefore to normalize it we calculate the ln(1/R²)-ln(R²). The result of this calculation is the informational opacity index. All variables are mean-collapsed at the industry level. U.S. data only.Sources: Datastream; author’s calculations.

Table 2.

Industry Informational Opacity

Industry	R-squared	Informativeness Index	Opacity Index
Finance, Insurance, And Real Estate	0.0034	0.9966	5.684
Wholesale Trade	0.0039	0.9961	5.550
Mining	0.0042	0.9958	5.477
Transportation, Communications, Electric, Gas	0.0059	0.9941	5.123
Retail Trade	0.0060	0.9940	5.111
Services	0.0066	0.9934	5.019
Agriculture, Forestry, And Fishing	0.0068	0.9932	4.986
Construction	0.0072	0.9928	4.931
Manufacturing	0.0085	0.9915	4.756

Notes: R-squared represents the R² of regressing firm specific return variations on industry-wide and market-wide returns. The informativeness index is calculated as one minus average R² of the regressions using each firm in an industry. The more information contained in the stock prices of individual firms, the lower the R²will be. The informativeness index is highly skewed and mildly leptokurtic, therefore to normalize it we calculate the ln(1/R²)-ln(R²). The result of this calculation is the informational opacity index. All variables are mean-collapsed at the industry level. U.S. data only.Sources: Datastream; author’s calculations.

View Table

Table 2.

Industry Informational Opacity

Industry	R-squared	Informativeness Index	Opacity Index
Finance, Insurance, And Real Estate	0.0034	0.9966	5.684
Wholesale Trade	0.0039	0.9961	5.550
Mining	0.0042	0.9958	5.477
Transportation, Communications, Electric, Gas	0.0059	0.9941	5.123
Retail Trade	0.0060	0.9940	5.111
Services	0.0066	0.9934	5.019
Agriculture, Forestry, And Fishing	0.0068	0.9932	4.986
Construction	0.0072	0.9928	4.931
Manufacturing	0.0085	0.9915	4.756

Notes: R-squared represents the R² of regressing firm specific return variations on industry-wide and market-wide returns. The informativeness index is calculated as one minus average R² of the regressions using each firm in an industry. The more information contained in the stock prices of individual firms, the lower the R²will be. The informativeness index is highly skewed and mildly leptokurtic, therefore to normalize it we calculate the ln(1/R²)-ln(R²). The result of this calculation is the informational opacity index. All variables are mean-collapsed at the industry level. U.S. data only.Sources: Datastream; author’s calculations.

IV. Hypotheses Development

The main concern of this paper is whether national culture still influences corporate risk-taking decisions, even after taking into account its effects on the institutional, economic and industrial environments. The second concern is whether any of these environments can accentuate or curb the influence of national culture on firm risk-taking.

B. Indirect Effects of Culture

As mentioned previously, national cultural differences may also indirectly affect corporate risk-taking decisions, through their effects on the institutional, economic and industrial environments in which a firm finds itself in. Formal institutions and informal national cultural norms are strongly interconnected. More specifically, formal institutions can be the product of cultural norms because the political, contractual and economic rules are all linked to the society’s perception of how things ought to work. These institutions, once established, tend to reinforce the societal values that led to their establishment. Formal institutions such as rule of law, regulatory quality, government effectiveness, control of corruption, investor protection, creditor rights, bankruptcy costs, financial development, market competition, and industry diversification can all be shaped by national cultural values including uncertainty aversion, individualism, power distance or masculinity.

Country Effects

Societies differ in the quality of institutions that regulate or encourage corporate risk-taking, such as regulatory institutions, governmental and administrative institutions, or judiciary institutions. National cultural variables affect the quality of these institutions, which at their turn have been shown to affect corporate risk-taking.

Table 3 explores the Pearson’s correlations between national cultural values and governance indicators and shows that there is indeed a strongly significant positive relationship between individualism and governance, and a strongly significant negative relationship between power distance and good governance. In fact, the higher a country’s level of individualism, the better its rule-of-law, the better its’ government effectiveness, its’ control of corruption and its’ regulatory quality. Similarly, the lower a country’s level of power distance (the more egalitarian the society), the better its formal institutions. There is no strong correlation relationship between governance and uncertainty aversion, or masculinity. Thus, we believe that individualism and power distance are the two most important cultural variables that affect institutional development.

Table 3.

Correlation Matrix between National Culture and National Governance Indicators

Notes: The formal institution index is a measure of institutional development (higher values mean better institutions) and is calculated as an average index of the other six governance indicators. The stars (^*) represent significance at the 1 percent level.Sources: Governance indicators from WGI World Bank; cultural variables from Hofstede 2001.

Table 3.

Correlation Matrix between National Culture and National Governance Indicators

	Rule of Law	Control of Corruption	Regulatory Quality	Govt. Effectiv.	Voice and Account.	Political Stability	Formal Institution Index
Uncertainty Aversion	-0.17*	-0.20*	-0.14*	-0.21*	0.00	-0.05	-0.14
Individualism	0.71*	0.68*	0.60*	0.68*	0.64*	0.53*	0.69*
Power Distance	-0.64*	-0.65*	-0.56*	-0.60*	-0.64*	-0.48*	-0.64*
Masculinity	-0.07	-0.11*	-0.01	-0.04	-0.04	-0.07	-0.06

Notes: The formal institution index is a measure of institutional development (higher values mean better institutions) and is calculated as an average index of the other six governance indicators. The stars (^*) represent significance at the 1 percent level.Sources: Governance indicators from WGI World Bank; cultural variables from Hofstede 2001.

View Table

Table 3.

Correlation Matrix between National Culture and National Governance Indicators

	Rule of Law	Control of Corruption	Regulatory Quality	Govt. Effectiv.	Voice and Account.	Political Stability	Formal Institution Index
Uncertainty Aversion	-0.17*	-0.20*	-0.14*	-0.21*	0.00	-0.05	-0.14
Individualism	0.71*	0.68*	0.60*	0.68*	0.64*	0.53*	0.69*
Power Distance	-0.64*	-0.65*	-0.56*	-0.60*	-0.64*	-0.48*	-0.64*
Masculinity	-0.07	-0.11*	-0.01	-0.04	-0.04	-0.07	-0.06

Notes: The formal institution index is a measure of institutional development (higher values mean better institutions) and is calculated as an average index of the other six governance indicators. The stars (^*) represent significance at the 1 percent level.Sources: Governance indicators from WGI World Bank; cultural variables from Hofstede 2001.

Because individualistic cultures emphasize self-interested competition, freedom and well-established individual rights and obligations, they also require a rule-of-law state that provides people with a comprehensive set of protection rights (e.g. property, contract rights, etc) and that effectively enforces them (Licht et al. 2005). Such a rule-of-law state protects the rights of competing parties against opportunistic behavior and gives everyone a fair chance to pursue their ambitions. On the other hand, more collectivist countries may give less emphasis to enforcing laws intended to shield personal rights and property from violations. At the same time, better institutional governance has been shown to encourage greater firm risk-taking (Houston et al. 2010; Li and Zahra 2012; Laeven et al. 2009). Thus, we hypothesize that individualism will positively and indirectly affect risk-taking through its effects on institutional development.

Furthermore, cultures with a low power distance, that have a low tolerance for hierarchical relationships, also tend to be more egalitarian. Such cultures have better rule-of-law systems which treat citizens equally and which do not disfavor ones against the others, giving them the same rewards or punishments for their actions. Egalitarian societies, by encouraging societal members to treat one another equally, will also promote a rule-of-law norm that entitles all to equal protection under the law. We predict that power distance will be negatively associated with institutional development, which in turn will negatively affect corporate risk-taking. To summarize the fifth hypothesis,

H5: The higher the level of individualism, the stronger the positive relationship between the level of formal institutional development and the level of corporate risk-taking. The higher the level of power distance, the weaker the positive relationship between formal institutional development and corporate risk-taking.

Countries also vary extensively in the protection mechanisms they offer to creditors and shareholders, and in their bankruptcy rules. As previous literature has found, differences in protection rules, contract enforceability and bankruptcy costs all affect risk-taking decisions.

Countries high in individualism tend to have good shareholder and creditor protection mechanisms and a strong enforceability of deals. Such societies tend to clarify the framework and purpose of people’s entitlements, allowing them to better plan their independent (and sometimes selfish) actions (Raz 1979). On the other hand, societies high on collectivism, by assigning priority to the group in the detriment of the individual, tend to give less importance to protecting individuals’ interests or to satisfying their preferences (Licht et al. 2005). Creditor and shareholder rights have been shown to affect risk-taking decisions in opposite ways. Stronger creditor rights in bankruptcy tend to reduce corporate risk-taking (Acharya et al. 2010), while stronger shareholder rights tend to increase firm risk-taking (John et al. 2008; Paligorova 2010). This is consistent with the conflict of interest between shareholders and creditors, with the creditors receiving only fixed payoffs and being rather risk-averse to risky projects, while the shareholders have an interest in pursuing risky projects because they are only subject to the upside consequences of risky decisions. Therefore we expect that individualism will indirectly affect corporate risk-taking through its distinct effects on creditor and shareholder protection, on the one hand encouraging firm risk-taking and on the other hand discouraging it.

Table 4 explores the correlations between national cultural variables and creditor/shareholder protection mechanisms. Uncertainty aversion is negatively correlated with shareholder protection. This implies that highly uncertainty-averse countries tend to be associated with a lower protection of shareholders’ rights. This is not surprising given the fact that usually, highly uncertainty-averse countries tend to be less supportive of market-based financial practices, usually being dominated by bank-based financial systems, rather than by stock-markets (Kirca et al. 2009; Kwok et al. 2006). Since they are not inclined to encourage market-based financial practices, we could also expect that they will give less importance to creditor and shareholder protection, which in turn will discourage corporate risk-taking. Individualism is, as expected, highly positively correlated with strong property rights, and with a good efficiency of debt enforcements and contracts. Power distance is negatively correlated with property rights, efficiency of debt enforcement and enforceability of contracts.

Table 4.

Correlation Matrix between National Culture and Protection Mechanisms

Notes: The indices are: creditor rights, property rights, the anti-self-dealing index, disclosure requirements, case-A efficiency, bankruptcy costs and propensity to file for bankruptcy. The stars (^*) represent significance at the 5 percent level.Sources: Djankov et al. 2003, 2006, 2007; The Heritage Foundation; La Porta et al. 2006; Hofstede 2001.

Table 4.

Correlation Matrix between National Culture and Protection Mechanisms

	Credit rights	Property rights	Legal protect. of shareholders	Efficiency of debt enforce.	Enforce. of contracts	Bankruptcy costs	Propensity tofile forbankruptcy
Uncertainty Aversion	-0.30	-0.20	-0.46*	-0.32	-0.35	0.04	-0.02
Individualism	0.11	0.68*	0.08	0.51*	0.76*	-0.31*	0.32*
Power Distance	-0.05	-0.62*	-0.05	-0.52*	-0.73*	0.37*	-0.26*
Masculinity	-0.05	-0.04	-0.03	-0.06	-0.10	0.20	0.04

Notes: The indices are: creditor rights, property rights, the anti-self-dealing index, disclosure requirements, case-A efficiency, bankruptcy costs and propensity to file for bankruptcy. The stars (^*) represent significance at the 5 percent level.Sources: Djankov et al. 2003, 2006, 2007; The Heritage Foundation; La Porta et al. 2006; Hofstede 2001.

View Table

Table 4.

Correlation Matrix between National Culture and Protection Mechanisms

	Credit rights	Property rights	Legal protect. of shareholders	Efficiency of debt enforce.	Enforce. of contracts	Bankruptcy costs	Propensity tofile forbankruptcy
Uncertainty Aversion	-0.30	-0.20	-0.46*	-0.32	-0.35	0.04	-0.02
Individualism	0.11	0.68*	0.08	0.51*	0.76*	-0.31*	0.32*
Power Distance	-0.05	-0.62*	-0.05	-0.52*	-0.73*	0.37*	-0.26*
Masculinity	-0.05	-0.04	-0.03	-0.06	-0.10	0.20	0.04

Notes: The indices are: creditor rights, property rights, the anti-self-dealing index, disclosure requirements, case-A efficiency, bankruptcy costs and propensity to file for bankruptcy. The stars (^*) represent significance at the 5 percent level.Sources: Djankov et al. 2003, 2006, 2007; The Heritage Foundation; La Porta et al. 2006; Hofstede 2001.

It also appears that the propensity to file for bankruptcy is influenced by national culture. The correlation analysis shows that individualism is positively and power distance is negatively linked to the propensity to file for bankruptcy. Cultures which are more individualistic might be more inclined to offer individuals “a fresh start” providing the financially troubled entity with an opportunity take responsibility for the financial failure and to obtain relief and properly position itself to re-join the economy as a productive member of society (Efrat 2002). Thus, more individualistic societies might render it easier to file for bankruptcy, while at the same time encouraging greater risk-taking because bankruptcy costs are not that high. More egalitarian societies (lower power distance countries) also seem to make it easier to file for bankruptcy. Thus, we hypothesize that both individualism and power distance will indirectly affect corporate risk-taking through their effects on a society’s bankruptcy rules. To summarize, the sixth hypothesis is

H6: The higher the level of individualism, the stronger the positive link between the level of bankruptcy protection and risk-taking. The lower the level of uncertainty aversion, the stronger the relationship between shareholders’ rights and corporate risk-taking. The higher the level of power distance, the lower the link between protection mechanism indicators and corporate risk-taking.

Industry Effects

Culture has an important impact upon economic development and upon the pattern of industrial relations found in a given country. National cultural values can dictate the types of industries economic activity is concentrated in, besides more obvious reasons such as resource availability or financial development (Whyte 1963).

Table 5 explores the correlations between national cultural variables and industry diversification, industry competition and economic development. It shows that more individualistic countries tend to have economic activity spread into a larger palette of industries. Also, the lower the level of uncertainty aversion and the lower the level of power distance, the higher the diversification of the country’s industries. There are also significant correlations between cultural norms and industry competition. Countries ranking high in masculinity and in individualism tend to have, on average, more competitive industries. On the other hand, countries ranking high in uncertainty aversion tend to have more concentrated industries with few players. The results are especially strong for industries which are more informationally opaque (finance, IT, petroleum refining, and mining). There is also a strong relationship between national cultural values and the level of economic development. Countries ranking high in individualism tend to be strongly associated with a higher level of economic development, while countries high in power distance (less egalitarian) and high in uncertainty aversion tend to be strongly negatively correlated with economic development.

Table 5.

Correlation Matrix between National Culture and Industry Indicators

Notes: Industrial diversification is an index which measures the diversity of industrial activities (higher values correspond to a spreading of employment and investment over a wider range of industries). The industry concentration represents the amount of competition among firms within an industry and is calculated as the Herfindahl index (the higher the index, the lower the competition). Opaque industries represent any industry whose stock price informational opacity is higher than 3.5. Economic development is the log of GDP per capita. The stars (^*) represent significance at the 5 percent level.Sources: CVU; Worldscope; Bankscope; Datastream.

Table 5.

Correlation Matrix between National Culture and Industry Indicators

	Industrial diversification	Industry concentration	Concentration in highly opaque industries	Concentration in less opaque industries	Economic development
Uncertainty Aversion	-0.31*	0.06*	0.12*	0.05*	-0.54*
Individualism	0.69*	-0.09*	-0.25*	-0.06*	0.65*
Power Distance	-0.41*	-0.01*	0.08*	-0.02*	-0.51*
Masculinity	0.24*	-0.28*	-0.35*	-0.27*	0.08

Notes: Industrial diversification is an index which measures the diversity of industrial activities (higher values correspond to a spreading of employment and investment over a wider range of industries). The industry concentration represents the amount of competition among firms within an industry and is calculated as the Herfindahl index (the higher the index, the lower the competition). Opaque industries represent any industry whose stock price informational opacity is higher than 3.5. Economic development is the log of GDP per capita. The stars (^*) represent significance at the 5 percent level.Sources: CVU; Worldscope; Bankscope; Datastream.

View Table

Table 5.

Correlation Matrix between National Culture and Industry Indicators

	Industrial diversification	Industry concentration	Concentration in highly opaque industries	Concentration in less opaque industries	Economic development
Uncertainty Aversion	-0.31*	0.06*	0.12*	0.05*	-0.54*
Individualism	0.69*	-0.09*	-0.25*	-0.06*	0.65*
Power Distance	-0.41*	-0.01*	0.08*	-0.02*	-0.51*
Masculinity	0.24*	-0.28*	-0.35*	-0.27*	0.08

Notes: Industrial diversification is an index which measures the diversity of industrial activities (higher values correspond to a spreading of employment and investment over a wider range of industries). The industry concentration represents the amount of competition among firms within an industry and is calculated as the Herfindahl index (the higher the index, the lower the competition). Opaque industries represent any industry whose stock price informational opacity is higher than 3.5. Economic development is the log of GDP per capita. The stars (^*) represent significance at the 5 percent level.Sources: CVU; Worldscope; Bankscope; Datastream.

Because individualism and masculinity are more negatively correlated with concentration in the highly opaque industries, as compared to the more transparent industries, we hypothesize that firms in these industries will take on more risk. Excessive competition among firms within highly opaque industries can encourage them to pursue riskier policies in an attempt to maintain their former profits. To summarize the seventh hypothesis

H7: The higher the level of uncertainty aversion, the less the risk in industrial sectors which are more informationally opaque. The higher the levels of individualism and masculinity, the higher the risk-taking behavior of firms in more informationally opaque industries.

Firm Effects

National cultural variables can influence the way a firm chooses its compensation practices and ownership structure. Equity-based compensation is usually awarded to managers to overcome managerial risk aversion and induce optimal risk-taking behavior (Smith and Stulz 1985; Guay 1999). Nash et al. (2012) have shown that individualism is positively, while uncertainty avoidance is negatively related to firm level equity-based compensation practices. At the same time, several studies have proven that corporate risk-taking behavior depends on compensation practices and ownership structure. Low (2009), among many others, has proved that equity-based compensation affects managers’ risk-taking behavior. Laeven et al. (2009) have shown that risk-taking also varies positively with the comparative power of shareholders within the corporate governance structure.

We postulate that both individualism, which accentuates self-interests and individual autonomy, and uncertainty aversion will indirectly influence corporate risk-taking through their effects on compensation practices and ownership structure. Since firm-level equity-based compensation is more frequent in highly individualistic and low uncertainty-avoidant countries, we predict that firms in such countries will be more risk-seeking. These suggestions only reinforce the previous hypotheses, suggesting that uncertainty aversion will be negatively, and individualism will be positively related to firm risk-taking decisions.

V. Empirical Model

Because the dataset analyzed is a hierarchical model, consisting of 50, 000 firms, nested across 400 industries in 51 countries, all analyses are performed using a linear mixed model procedure in Stata MP/11.2, controlling for firm, industry and country-level characteristics. From an econometric perspective, observations at the firm level are clustered under higher entities (e.g. firms grouped within industries within countries) and analyzing the data through ordinary least squares (OLS) or general linear models (GLS) could lead to a multitude of problems such as obtaining correlated errors, biased estimates of coefficient standard errors, and wrongfully interpreting the results and significance of the predictor variables (Garson 2012).

While previous literature has studied the effects of culture on risk-taking through OLS (Kanagaretnam et al. 2011) or through GLS methods (Li and Zahra 2011), we make full use of a more recently developed modeling technique, hierarchical linear mixed models (HLM), which can process multilevel data where observations are not independent. That is, linear mixed models are not only needed, but required in this case, since the OLS and GLS regression assumption of independent errors is violated: the data is clustered by a higher order grouping variable and by repeated time measures. Compared to OLS and GLS regressions, hierarchical linear mixed models correctly compute coefficient estimates and standard errors even when observations display intraclass correlation - that is, individual-level observations from the same upper level group will not be independent, but rather more similar due to factors such as shared group history and group selection processes.

There are several advantages of using a multilevel hierarchical model. First, we can statistically test multi-level theories, simultaneously modeling variables at the firm, industry and country level without having to recourse to data aggregation or disaggregation as previous cultural economics literature has had to do. Second, the multilevel hierarchical model has the ability to handle unbalanced data where sample size varies across higher levels, as in our case where the number of firms varies widely across industries across countries. Third, in explaining corporate risk-taking, the linear mixed model focuses on differences between groups (e.g. industries, countries) in relation to differences within groups (e.g. among firms within industries within countries), making it possible to model not only individual firm, industry and country-level characteristics, but also their cross-level interactions.

Before specifying the hierarchical linear mixed model, we mean center the independent variables by their grand mean. This is a customary procedure which improves the interpretability of coefficients and reduces multicollinearity (Garson 2012). Then, we create grand-mean adjusted independent variables averaged within an industry and a country which we will denote as ending with the suffix “_indus_ctry_mean”. Thus, for each industry within a country we will have grand-mean adjusted independent variables which do not change across firms, but only across industries within countries. Lastly, we create an additional set of variables of within-country variability, ending with the suffix “_firm_var”, defined as the grand mean-adjusted independent variables in step 1 minus the grand-mean adjusted independent variable averaged within an industry and a country in step 2. These variables are practically deviations of firm-level variables from their corresponding industry and country-level means and inform our analysis by taking into account and separating the covariances within- and between industries in a given country.

The basic model we fit is specified as follows:

where Risk - Taking_ijk can be either the z-score, or the standard deviation of ROA, or the R&D index of firm i in country j in industry k.

Culture_j is a vector composed of four national cultural variables: uncertainty aversion, individualism, power distance and masculinity as defined at the beginning of this section. Country_Controls_j is a vector of country-level variables such as economic development, the policy rate, and aggregate economic volatility. At the country level, we also control for variables such shareholder protection, creditor rights, and bankruptcy costs, besides controlling for a country’s overall quality of formal institutions, which is an index of seven institutional measures calculated through principle component analysis. Industry_Controls_k is a vector of industry-level variables such as industry competition and industry informational opacity. Similar to Griffin et al. (2012), we decompose the firm-level controls into firm-level deviations and industry/country-level means to understand the differential firm and industry/country-level effects. More specifically, Firm_Controls_Firm_Var_i is a vector comprising of firm size, leverage, profitability, sales growth, and dependence on external finance. Firm_Controls_Indus_Ctry_Mean_i is an additional set of controls which will help us understand the differential industry/country-level effects.

VI. Results and Discussion

Clearly, national cultural differences play a very important role in corporate risk-taking decisions. Table 6 on the following page presents the results of the empirical model. In Appendix C, we present more detailed HLM and OLS regression results for each measure of corporate risk-taking.

Table 6.

Effects of National Culture on Corporate Risk-Taking

Notes: Hierarchical linear mixed model estimation, with observations clustered at the industry- and country-levels. All regressions control for country-specific variables (financial development; policy rate; shareholder rights; creditor rights; bankruptcy costs; an index of the quality of formal institutions comprising of rule of law, regulatory quality, government effectiveness, efficiency of debt enforecement, control of corruption, voice and accountability and political stability; and aggreagate economic volatility), industry-specific variables (industry opacity; industry competition), and firm-specific variables (size; leverage; profitability; sales growth; and dependence on external finance). The estimation period is 2000-2012 and includes 50,000 firms, from 50 countries. OLS coefficients presented for comparison purposes. OLS regressions employ robust standard errors clustering at the country level with industry fixed effects not reported in this table. All regressions exclude U.S. firms. Robust standard errors in parentheses. Stars ^***, ^**, and ^* denote significance at the 1, 5, and 10 percent levels respectively.

Table 6.

Effects of National Culture on Corporate Risk-Taking

Dependent variable: Risk-Taking
	Explanatory variables ↓	Inverse of The z-Score		Standard Deviation of Return on Assets		Research & Development Index
		(1)	(2)	(3)	(4)	(5)	(6)
		OLS	HLM	OLS	HLM	OLS	HLM
Culture	Uncertainty Aversion (UAI)	-0.005***	-0.005***	-0.568***	-0.191**	-0.568***	-0.011**
		[0.001]	[0.003]	[0.013]	[0.086]	[0.027]	[0.006]
	Individualism (IDV)	0.003***	0.008***	0.726***	0.226**	0.242*	0.343**
		[0.001]	[0.003]	[0.015]	[0.103]	[0.147]	[0.160]
	Power Distance (PDI)	-0.006***	-0.001**	-0.415***	-0.053	-0.210***	-0.143***
		[0.001]	[0.001]	[0.019]	[0.102]	[0.064]	[0.058]
	Masculinity (MAS)	0.001	0.001	0.194***	0.043	0.077***	0.131
		[0.001]	[0.001]	[0.015]	[0.072]	[0.015]	[0.126]
Country	Financial Development	0.003***	0.004	0.002	-0.411*	0.029	0.116
		[0.001]	[0.006]	[0.022]	[0.236]	[0.156]	[0.348]
	Central Bank Policy Rate	-0.007***	-0.008	-1.076***	0.323	-0.682***	-0.950
		[0.001]	[0.016]	[0.026]	[0.555]	[0.237]	[0.944]
	Shareholders’ Rights Index	0.047	0.085	54.129***	17.687**	8.596	15.744
		[0.068]	[0.260]	[1.627]	[7.840]	[5.623]	[14.059]
	Creditor Rights	0.012	-0.020	-17.687***	-3.525**	0.761	-1.592
		[0.011]	[0.055]	[0.359]	[1.654]	[2.219]	[2.762]
	Cost Of Bankruptcy	-0.006***	-0.002	-0.318***	0.026	-0.029	0.106
		[0.002]	[0.007]	[0.034]	[0.224]	[0.077]	[0.414]
	Formal Institutions Index	1.025***	1.053*	28.067**	30.940**	2.268**	4.524**
		[0.255]	[0.622]	[11.196]	[15.559]	[1.136]	[2.198]
	Aggregate Econ Volatility	3.181**	4.247	575.432***	-117.931	249.115*	310.907
		[1.494]	[5.486]	[32.444]	[181.592]	[136.537]	[337.379]
Industry	Industry Opacity	-0.035***	-0.038***	-0.433*	0.136	-2.081**	-2.852**
		[0.008]	[0.007]	[0.224]	[0.209]	[1.007]	[1.307]
	Industry Competition	0.091***	0.216*	8.005***	9.391***	6.475***	2.697*
		[0.034]	[0.124]	[1.377]	[2.603]	[1.881]	[1.492]
Firm	Firm Size	-0.121***	-0.082***	2.331***	2.216***	-0.733	-1.233
		[0.004]	[0.004]	[0.089]	[0.121]	[0.554]	[0.751]
	Firm Leverage	0.001	-0.003***	-0.010	-0.098***	-0.022	-0.054
		[0.000]	[0.000]	[0.015]	[0.007]	[0.040]	[0.044]
	Firm Profitability	-0.115***	-0.029***	0.016***	0.701***	-0.248**	-0.173***
		[0.001]	[0.000]	[0.002]	[0.008]	[0.109]	[0.046]
	Firm Sales Growth	-0.047***	-0.289***	-0.495***	0.809	-2.939**	-7.465**
		[0.008]	[0.022]	[0.078]	[0.501]	[1.232]	[3.032]
	Firm Depend. Extern Finance	0.000**	-0.003***	0.000	0.018***	-0.016	-0.050
		[0.000]	[0.000]	[0.001]	[0.007]	[0.013]	[0.041]
	Observations	94997		132571		132755
	Number of firms	20055	20055	24197	24197	24328	24328
	R-square overall	0.207		0.504		0.060

Notes: Hierarchical linear mixed model estimation, with observations clustered at the industry- and country-levels. All regressions control for country-specific variables (financial development; policy rate; shareholder rights; creditor rights; bankruptcy costs; an index of the quality of formal institutions comprising of rule of law, regulatory quality, government effectiveness, efficiency of debt enforecement, control of corruption, voice and accountability and political stability; and aggreagate economic volatility), industry-specific variables (industry opacity; industry competition), and firm-specific variables (size; leverage; profitability; sales growth; and dependence on external finance). The estimation period is 2000-2012 and includes 50,000 firms, from 50 countries. OLS coefficients presented for comparison purposes. OLS regressions employ robust standard errors clustering at the country level with industry fixed effects not reported in this table. All regressions exclude U.S. firms. Robust standard errors in parentheses. Stars ^***, ^**, and ^* denote significance at the 1, 5, and 10 percent levels respectively.

View Table

Table 6.

Effects of National Culture on Corporate Risk-Taking

Dependent variable: Risk-Taking
	Explanatory variables ↓	Inverse of The z-Score		Standard Deviation of Return on Assets		Research & Development Index
		(1)	(2)	(3)	(4)	(5)	(6)
		OLS	HLM	OLS	HLM	OLS	HLM
Culture	Uncertainty Aversion (UAI)	-0.005***	-0.005***	-0.568***	-0.191**	-0.568***	-0.011**
		[0.001]	[0.003]	[0.013]	[0.086]	[0.027]	[0.006]
	Individualism (IDV)	0.003***	0.008***	0.726***	0.226**	0.242*	0.343**
		[0.001]	[0.003]	[0.015]	[0.103]	[0.147]	[0.160]
	Power Distance (PDI)	-0.006***	-0.001**	-0.415***	-0.053	-0.210***	-0.143***
		[0.001]	[0.001]	[0.019]	[0.102]	[0.064]	[0.058]
	Masculinity (MAS)	0.001	0.001	0.194***	0.043	0.077***	0.131
		[0.001]	[0.001]	[0.015]	[0.072]	[0.015]	[0.126]
Country	Financial Development	0.003***	0.004	0.002	-0.411*	0.029	0.116
		[0.001]	[0.006]	[0.022]	[0.236]	[0.156]	[0.348]
	Central Bank Policy Rate	-0.007***	-0.008	-1.076***	0.323	-0.682***	-0.950
		[0.001]	[0.016]	[0.026]	[0.555]	[0.237]	[0.944]
	Shareholders’ Rights Index	0.047	0.085	54.129***	17.687**	8.596	15.744
		[0.068]	[0.260]	[1.627]	[7.840]	[5.623]	[14.059]
	Creditor Rights	0.012	-0.020	-17.687***	-3.525**	0.761	-1.592
		[0.011]	[0.055]	[0.359]	[1.654]	[2.219]	[2.762]
	Cost Of Bankruptcy	-0.006***	-0.002	-0.318***	0.026	-0.029	0.106
		[0.002]	[0.007]	[0.034]	[0.224]	[0.077]	[0.414]
	Formal Institutions Index	1.025***	1.053*	28.067**	30.940**	2.268**	4.524**
		[0.255]	[0.622]	[11.196]	[15.559]	[1.136]	[2.198]
	Aggregate Econ Volatility	3.181**	4.247	575.432***	-117.931	249.115*	310.907
		[1.494]	[5.486]	[32.444]	[181.592]	[136.537]	[337.379]
Industry	Industry Opacity	-0.035***	-0.038***	-0.433*	0.136	-2.081**	-2.852**
		[0.008]	[0.007]	[0.224]	[0.209]	[1.007]	[1.307]
	Industry Competition	0.091***	0.216*	8.005***	9.391***	6.475***	2.697*
		[0.034]	[0.124]	[1.377]	[2.603]	[1.881]	[1.492]
Firm	Firm Size	-0.121***	-0.082***	2.331***	2.216***	-0.733	-1.233
		[0.004]	[0.004]	[0.089]	[0.121]	[0.554]	[0.751]
	Firm Leverage	0.001	-0.003***	-0.010	-0.098***	-0.022	-0.054
		[0.000]	[0.000]	[0.015]	[0.007]	[0.040]	[0.044]
	Firm Profitability	-0.115***	-0.029***	0.016***	0.701***	-0.248**	-0.173***
		[0.001]	[0.000]	[0.002]	[0.008]	[0.109]	[0.046]
	Firm Sales Growth	-0.047***	-0.289***	-0.495***	0.809	-2.939**	-7.465**
		[0.008]	[0.022]	[0.078]	[0.501]	[1.232]	[3.032]
	Firm Depend. Extern Finance	0.000**	-0.003***	0.000	0.018***	-0.016	-0.050
		[0.000]	[0.000]	[0.001]	[0.007]	[0.013]	[0.041]
	Observations	94997		132571		132755
	Number of firms	20055	20055	24197	24197	24328	24328
	R-square overall	0.207		0.504		0.060

Notes: Hierarchical linear mixed model estimation, with observations clustered at the industry- and country-levels. All regressions control for country-specific variables (financial development; policy rate; shareholder rights; creditor rights; bankruptcy costs; an index of the quality of formal institutions comprising of rule of law, regulatory quality, government effectiveness, efficiency of debt enforecement, control of corruption, voice and accountability and political stability; and aggreagate economic volatility), industry-specific variables (industry opacity; industry competition), and firm-specific variables (size; leverage; profitability; sales growth; and dependence on external finance). The estimation period is 2000-2012 and includes 50,000 firms, from 50 countries. OLS coefficients presented for comparison purposes. OLS regressions employ robust standard errors clustering at the country level with industry fixed effects not reported in this table. All regressions exclude U.S. firms. Robust standard errors in parentheses. Stars ^***, ^**, and ^* denote significance at the 1, 5, and 10 percent levels respectively.

Models (2), (4) and (6) present the estimated coefficients for the direct and indirect effects of individual cultural variables, uncertainty aversion, individualism, power distance and masculinity respectively, on corporate risk-taking after controlling for country, industry and firm-level characteristics. Models (1), (3) and (5) present the coefficients obtained through OLS estimation, which we include for comparison purposes.

VII. Accentuating/Moderating Factors

Table 7 - and the more detailed Table C4 in Appendix C - present the accentuating/moderating factors that increase/decrease corporate risk-taking. They show that culture interacts with the various institutional, political and economic forces to produce very different and specific outcomes.

Table 7.

Accentuating/Moderating Factors

Notes: Hierarchical linear mixed model estimation, with observations clustered at the industry- and country- levels. All regressions control for country-specific variables (financial development; policy rate; shareholder rights; creditor rights; bankruptcy costs; an index of formal institutions; and aggreagate economic volatility), industry-specific variables (industry opacity; industry competition), and firm-specific variables (size; leverage; profitability; sales growth; and dependence on external finance). The estimation period is 2000-2012 and includes 50,000 firms, from 50 countries. OLS coefficients presented for comparison purposes. OLS regressions employ robust SE clustering at the country level with industry fixed effects not reported. All regressions exclude U.S. firms. Robust standard errors in parentheses. Stars ^***, ^**, and ^* denote significance at the 1, 5, and 10 percent levels respectively.

Table 7.

Accentuating/Moderating Factors

Dependent variable: Risk-Taking
	Explanatory variables ↓	Inverse of the z-score			Stdev(Return on assets)			Research & Development
		(1)	(2)	(3)	(1)	(2)	(3)	(1)	(2)	(3)
Culture	UAI	-0.005***	-0.007**	-0.004*	-0.297***	-0.339**	-0.360***	-0.010**	-0.089**	-0.038**
		[0.003]	[0.003]	[0.003]	[0.096]	[0.170]	[0.097]	[0.004]	[0.49]	[0.184]
	IDV	0.013*	0.004**	0.015***	0.538***	0.381*	0.444***	0.113*	0.267	0.297**
		[0.006]	[0.002]	[0.004]	[0.183]	[0.199]	[0.098]	[0.060]	[0.437]	[0.132]
	PDI	-0.001**	-0.004	-0.008*	-0.097	-0.493**	-0.242**	-0.001**	-0.411*	-0.695**
		[0.001]	[0.006]	[0.004]	[0.139]	[0.199]	[0.103]	[0.001]	[0.242]	[0.339]
	MAS	0.001	0.003	0.004	0.094	0.060	0.098	0.095	0.278	0.151
		[0.001]	[0.003]	[0.003]	[0.080]	[0.096]	[0.090]	[0.088]	[0.180]	[0.161]
H5	IDV × F. Institutions	0.006**			0.155*			0.259*
		[0.003]			[0.083]			[0.141]
	PDI × F. Institutions	1.306			25.691			0.009
		[2.703]			[104.102]			[0.245]
H6	UAI × Creditors’ Rights		-0.002			-0.022			-0.008
			[0.002]			[0.081]			[0.094]
	1/UAI X Shareholders’ Rights		4.566			4.691			-231.375
			[3.876]			[151.232]			[142.655]
	IDV × Creditors’ Rights		-0.009***			-0.031			-0.076
			[0.002]			[0.089]			[0.129]
	IDV × Shareholders’ Rights		0.037***			0.264			0.555
			[0.010]			[0.423]			[0.678]
	IDV × Eff. of Debt Enforcement		0.000			0.004			0.013*
			[0.000]			[0.004]			[0.007]
	PDI × 1/Eff. of Debt Enforcement		-0.118*			-1.121			-2.995
			[0.068]			[2.198]			[4.670]
	IDV × 1/Cost of Bankruptcy		0.005			-0.002			0.237
			[0.004]			[0.156]			[0.272]
	PDI × Cost of Bankruptcy		0.000			-0.001			-0.001
			[0.000]			[0.012]			[0.018]
H7	UAI × Industry Opacity			-0.001*			-0.006**			-0.058*
				[0.000]			[0.003]			[0.034]
	IDV × Industry Opacity			1.185*			27.323**			94.36*
				[0.659]			[12.298]			[50.038]
	PDI × Industry Opacity			0.001			0.019			0.107
				[0.001]			[0.020]			[0.065]
	MAS × Industry Opacity			0.167			-3.189			0.031
				[0.244]			[6.314]			[19.579]
	Number of firms					20055

Notes: Hierarchical linear mixed model estimation, with observations clustered at the industry- and country- levels. All regressions control for country-specific variables (financial development; policy rate; shareholder rights; creditor rights; bankruptcy costs; an index of formal institutions; and aggreagate economic volatility), industry-specific variables (industry opacity; industry competition), and firm-specific variables (size; leverage; profitability; sales growth; and dependence on external finance). The estimation period is 2000-2012 and includes 50,000 firms, from 50 countries. OLS coefficients presented for comparison purposes. OLS regressions employ robust SE clustering at the country level with industry fixed effects not reported. All regressions exclude U.S. firms. Robust standard errors in parentheses. Stars ^***, ^**, and ^* denote significance at the 1, 5, and 10 percent levels respectively.

View Table

Table 7.

Accentuating/Moderating Factors

Dependent variable: Risk-Taking
	Explanatory variables ↓	Inverse of the z-score			Stdev(Return on assets)			Research & Development
		(1)	(2)	(3)	(1)	(2)	(3)	(1)	(2)	(3)
Culture	UAI	-0.005***	-0.007**	-0.004*	-0.297***	-0.339**	-0.360***	-0.010**	-0.089**	-0.038**
		[0.003]	[0.003]	[0.003]	[0.096]	[0.170]	[0.097]	[0.004]	[0.49]	[0.184]
	IDV	0.013*	0.004**	0.015***	0.538***	0.381*	0.444***	0.113*	0.267	0.297**
		[0.006]	[0.002]	[0.004]	[0.183]	[0.199]	[0.098]	[0.060]	[0.437]	[0.132]
	PDI	-0.001**	-0.004	-0.008*	-0.097	-0.493**	-0.242**	-0.001**	-0.411*	-0.695**
		[0.001]	[0.006]	[0.004]	[0.139]	[0.199]	[0.103]	[0.001]	[0.242]	[0.339]
	MAS	0.001	0.003	0.004	0.094	0.060	0.098	0.095	0.278	0.151
		[0.001]	[0.003]	[0.003]	[0.080]	[0.096]	[0.090]	[0.088]	[0.180]	[0.161]
H5	IDV × F. Institutions	0.006**			0.155*			0.259*
		[0.003]			[0.083]			[0.141]
	PDI × F. Institutions	1.306			25.691			0.009
		[2.703]			[104.102]			[0.245]
H6	UAI × Creditors’ Rights		-0.002			-0.022			-0.008
			[0.002]			[0.081]			[0.094]
	1/UAI X Shareholders’ Rights		4.566			4.691			-231.375
			[3.876]			[151.232]			[142.655]
	IDV × Creditors’ Rights		-0.009***			-0.031			-0.076
			[0.002]			[0.089]			[0.129]
	IDV × Shareholders’ Rights		0.037***			0.264			0.555
			[0.010]			[0.423]			[0.678]
	IDV × Eff. of Debt Enforcement		0.000			0.004			0.013*
			[0.000]			[0.004]			[0.007]
	PDI × 1/Eff. of Debt Enforcement		-0.118*			-1.121			-2.995
			[0.068]			[2.198]			[4.670]
	IDV × 1/Cost of Bankruptcy		0.005			-0.002			0.237
			[0.004]			[0.156]			[0.272]
	PDI × Cost of Bankruptcy		0.000			-0.001			-0.001
			[0.000]			[0.012]			[0.018]
H7	UAI × Industry Opacity			-0.001*			-0.006**			-0.058*
				[0.000]			[0.003]			[0.034]
	IDV × Industry Opacity			1.185*			27.323**			94.36*
				[0.659]			[12.298]			[50.038]
	PDI × Industry Opacity			0.001			0.019			0.107
				[0.001]			[0.020]			[0.065]
	MAS × Industry Opacity			0.167			-3.189			0.031
				[0.244]			[6.314]			[19.579]
	Number of firms					20055

Notes: Hierarchical linear mixed model estimation, with observations clustered at the industry- and country- levels. All regressions control for country-specific variables (financial development; policy rate; shareholder rights; creditor rights; bankruptcy costs; an index of formal institutions; and aggreagate economic volatility), industry-specific variables (industry opacity; industry competition), and firm-specific variables (size; leverage; profitability; sales growth; and dependence on external finance). The estimation period is 2000-2012 and includes 50,000 firms, from 50 countries. OLS coefficients presented for comparison purposes. OLS regressions employ robust SE clustering at the country level with industry fixed effects not reported. All regressions exclude U.S. firms. Robust standard errors in parentheses. Stars ^***, ^**, and ^* denote significance at the 1, 5, and 10 percent levels respectively.

The columns numbered (1) in Table 7 show that the effects of national governance indicators depend on the level of informal national cultural values. These columns examine the interaction effects between various cultural variables and an index of the overall quality of institutions within a country (rule of law, control of corruption, regulatory quality, political stability, etc). The results verify hypothesis H5, showing that the higher the level of individualism, the stronger the positive relationship between formal institutions and corporate risk-taking. This implies that the higher a country’s individualism and the better its formal institutions, the more risk firms in those countries will take. On the other hand, the coefficient on the interaction variable between power distance and formal institutions is not significant. Individually, power distance is negatively, while the formal institutions index is positively and significantly directly related to corporate risk-taking. The fact that their interaction is not significant suggests that the higher societies’ level of power distance, the weaker the positive relationship between formal institutional development and corporate risk-taking. The results are robust across all three specifications: for the inverse of the z-score, the σ(ROA) and the R&D index.

The columns numbered (2) in Table 7 present the interactions between national cultural values and bankruptcy rights. The analysis shows no consistent pattern, and therefore we reject our hypothesis H6. We do mention that for the inverse of the z-score, which is a proxy for corporate risk-taking, stronger shareholders rights tend to increase the level of firm risk-taking in highly individualistic societies. Stronger creditors’ rights in highly individualistic societies tend to slightly decrease risk-taking as shown by the negative and significant coefficient (-0.009^***), while worse debt enforcement in high power distance countries tends to also decrease corporate risk-taking (-0.118^*).

The columns numbered (3) in Table 7 show that corporate risk-taking in informationally opaque industries is strongly moderated by national cultural values. The interaction between uncertainty aversion and industry informational opacity is negative and significant across all model specifications (-0.001^* for the inverse of the z-score, -0.006^* for the stdev(ROA) and -0.058^* for the R&D index). This suggests that firms tend to take even less risk in industries which are high in informational opacity if they are also located in a society which is high in uncertainty aversion. Such firms will be even more cautious in pursuing risky projects. Surprisingly, the coefficient on the interaction variable between individualism and industry opacity is highly positive and significant across all models (1.185^* for the inverse of the z-score, 27.323^** for the stdev(ROA) and 94.36^* for the R&D index). This implies that in highly informationally opaque industries, such as finance, IT, mining or oil refinery, firms tend to take even more risk if they are located in a highly individualistic society. This result is very surprising because the general expectation is that a firm in an opaque industry takes less risk. However, this is not the case in a highly individualistic society. This puzzling result might be the consequence of highly opaque industries being more competitive in highly individualistic societies compared to the less opaque industries, and also compared to the more collectivistic societies. Neither masculinity, nor power distance seem to moderate risk-taking in highly opaque industries and the coefficients on these two interaction variables are not significant. Overall, these results partly confirm hypothesis H7.

VIII. Further Identification Test: Foreign Versus Domestic Firms

By this stage, we have shown that on average national cultural values influence firm risk-taking both directly and indirectly, subsequent to the quality of formal institutions and the industry a firm operates in. In this section, we further study whether domestic firms and foreign firms respond to cultural norms in the same way. To test this, we divide our sample into domestic versus foreign firms and re-estimate regression (1) for each sample separately. For the foreign firms, we use as explanatory variables both the host country’s and the origin country’s cultural values and macro-economic indicators separately. Before we present the results, we draw attention to the fact that the sample of foreign firms used in the analysis is very limited (only 500 foreign firms).

Regressing corporate risk-taking on cultural, legal, firm-specific and country-specific economic controls for the large subsample of domestic firms strengthens the initial results and renders some coefficients even more significant. We do not report them here since they are very similar to Table 6, and likely driving the previous results. Table 8 presents the regression results for foreign subsidiaries. It shows that (with the exception of power distance) a host country’s cultural values do not explain the corporate risk-taking behavior of foreign firms, while the parent country’s cultural values do. This is a strong identification test which takes care of the potential omitted variable bias which might influence the previous results. In other words, it is less about the economy in which a firm operates, and more about the culture of its country of origin.

Table 8.

Foreign Firms’ Risk-Taking Behavior and Culture

Notes: Hierarchical linear mixed model estimation, with observations clustered at the industry- and country-levels. Columns (1), (3) and (5) present the results from regressing corporate risk-taking on cultural and macroeconomic variables belonging to the host country a firm operates in. Columns (2), (4) and (5) present the results from regressing risk-taking indicators on cultural and macroeconomic variables belonging to the origin country a firm comes from. All regressions exclude U.S. firms. Robust standard errors in parentheses. Stars ^***, ^**, and ^* denote significance at the 1, 5, and 10 percent levels respectively.

Table 8.

Foreign Firms’ Risk-Taking Behavior and Culture

Dependent variable: Risk-Taking
	Explanatory variables ↓	Inverse of The z-Score		Standard Deviation of Return on Assets		Research & Development Index
		(1)	(2)	(3)	(4)	(5)	(6)
		Host Country	Origin Country	Host Country	Origin Country	Host Country	Origin Country
Culture	Uncertainty Aversion	-0.001	-0.013*	-0.299	-0.314*	0.048	-0.010*
		[0.006]	[0.008]	[0.182]	[0.177]	[0.049]	[0.005]
	Individualism	0.005	0.031***	0.213	0.554***	0.056	0.086*
		[0.006]	[0.010]	[0.211]	[0.203]	[0.040]	[0.048]
	Power Distance	-0.014*	-0.031***	-0.527**	-0.778**	-0.156***	-0.231***
		[0.008]	[0.010]	[0.218]	[0.316]	[0.053]	[0.064]
	Masculinity	0.004	0.002	-0.123	0.043	-0.038	-0.061
		[0.005]	[0.006]	[0.147]	[0.072]	[0.039]	[0.050]
Country	Financial Development	-0.016	-0.048***	-1.453***	-1.340***	-0.123	-0.329***
		[0.013]	[0.016]	[0.345]	[0.335]	[0.094]	[0.089]
	Central Bank Policy Rate	0.065	-0.082	-0.739	0.244	-0.048	-0.479
		[0.043]	[0.068]	[1.379]	[1.342]	[0.324]	[0.455]
	Shareholders’ Rights	1.204**	-1.173*	43.282**	26.477*	9.989***	16.078***
		[0.514]	[0.641]	[16.929]	[15.624]	[3.519]	[4.977]
	Creditor Rights	-0.103	-0.071	-7.054**	-8.128***	-0.497	-0.373
		[0.105]	[0.127]	[2.890]	[2.829]	[0.812]	[0.900]
	Cost Of Bankruptcy	-0.004	-0.029	-1.331**	-1.372**	0.128	-0.033
		[0.022]	[0.029]	[0.636]	[0.613]	[0.161]	[0.222]
	Formal Institutions Index	0.374	-0.358	22.548	52.519***	2.901	3.572
		[0.265]	[0.772]	[16.075]	[20.219]	[1.956]	[4.780]
	Aggregate Econ	42.260**	39.522**	-1,122.111***	-1,059.191***	143.569	-248.645**
		[18.494]	[18.125]	[309.063]	[298.680]	[178.317]	[98.337]
Industry	Industry Opacity	0.006	0.035	0.930	-4.753	-0.425	-0.307
		[0.052]	[0.051]	[1.040]	[3.728]	[0.449]	[0.473]
	Industry Competition	0.831	0.779	22.624	24.305*	4.04	2.436
		[0.607]	[0.546]	[14.011]	[13.591]	[4.948]	[5.247]
Firm	Firm Size	-0.341***	-0.145***	2979***	2.886***	-0.266	-0.231
		[0.084]	[0.035]	[0.683]	[0.681]	[0.297]	[0.302]
	Firm Leverage	-0.040***	-0.382***	-8.158***	-8.202***	-3.826***	-2.934***
		[0.004]	[0.085]	[1.586]	[1.539]	[0.699]	[0.606]
	Firm Profitability	-0.419**	-0.041***	0.707***	0.706***	-0.172***	-0.160***
		[0.177]	[0.004]	[0.057]	[0.056]	[0.026]	[0.026]
	Firm Sales Growth	-0.001	-0.395**	-5.721*	-6.140*	-1.075	-0.271
		[0.002]	[0.177]	[3.185]	[3.163]	[1.448]	[1.471]
	Firm Depend. Ext. Fin.	0.001	0.001	0.142***	0.141***	0.043***	0.042***
		[0.002]	[0.002]	[0.034]	[0.034]	[0.015]	[0.015]
	Number of firms	500	500	491	491	499	499

Notes: Hierarchical linear mixed model estimation, with observations clustered at the industry- and country-levels. Columns (1), (3) and (5) present the results from regressing corporate risk-taking on cultural and macroeconomic variables belonging to the host country a firm operates in. Columns (2), (4) and (5) present the results from regressing risk-taking indicators on cultural and macroeconomic variables belonging to the origin country a firm comes from. All regressions exclude U.S. firms. Robust standard errors in parentheses. Stars ^***, ^**, and ^* denote significance at the 1, 5, and 10 percent levels respectively.

View Table

Table 8.

Foreign Firms’ Risk-Taking Behavior and Culture

Dependent variable: Risk-Taking
	Explanatory variables ↓	Inverse of The z-Score		Standard Deviation of Return on Assets		Research & Development Index
		(1)	(2)	(3)	(4)	(5)	(6)
		Host Country	Origin Country	Host Country	Origin Country	Host Country	Origin Country
Culture	Uncertainty Aversion	-0.001	-0.013*	-0.299	-0.314*	0.048	-0.010*
		[0.006]	[0.008]	[0.182]	[0.177]	[0.049]	[0.005]
	Individualism	0.005	0.031***	0.213	0.554***	0.056	0.086*
		[0.006]	[0.010]	[0.211]	[0.203]	[0.040]	[0.048]
	Power Distance	-0.014*	-0.031***	-0.527**	-0.778**	-0.156***	-0.231***
		[0.008]	[0.010]	[0.218]	[0.316]	[0.053]	[0.064]
	Masculinity	0.004	0.002	-0.123	0.043	-0.038	-0.061
		[0.005]	[0.006]	[0.147]	[0.072]	[0.039]	[0.050]
Country	Financial Development	-0.016	-0.048***	-1.453***	-1.340***	-0.123	-0.329***
		[0.013]	[0.016]	[0.345]	[0.335]	[0.094]	[0.089]
	Central Bank Policy Rate	0.065	-0.082	-0.739	0.244	-0.048	-0.479
		[0.043]	[0.068]	[1.379]	[1.342]	[0.324]	[0.455]
	Shareholders’ Rights	1.204**	-1.173*	43.282**	26.477*	9.989***	16.078***
		[0.514]	[0.641]	[16.929]	[15.624]	[3.519]	[4.977]
	Creditor Rights	-0.103	-0.071	-7.054**	-8.128***	-0.497	-0.373
		[0.105]	[0.127]	[2.890]	[2.829]	[0.812]	[0.900]
	Cost Of Bankruptcy	-0.004	-0.029	-1.331**	-1.372**	0.128	-0.033
		[0.022]	[0.029]	[0.636]	[0.613]	[0.161]	[0.222]
	Formal Institutions Index	0.374	-0.358	22.548	52.519***	2.901	3.572
		[0.265]	[0.772]	[16.075]	[20.219]	[1.956]	[4.780]
	Aggregate Econ	42.260**	39.522**	-1,122.111***	-1,059.191***	143.569	-248.645**
		[18.494]	[18.125]	[309.063]	[298.680]	[178.317]	[98.337]
Industry	Industry Opacity	0.006	0.035	0.930	-4.753	-0.425	-0.307
		[0.052]	[0.051]	[1.040]	[3.728]	[0.449]	[0.473]
	Industry Competition	0.831	0.779	22.624	24.305*	4.04	2.436
		[0.607]	[0.546]	[14.011]	[13.591]	[4.948]	[5.247]
Firm	Firm Size	-0.341***	-0.145***	2979***	2.886***	-0.266	-0.231
		[0.084]	[0.035]	[0.683]	[0.681]	[0.297]	[0.302]
	Firm Leverage	-0.040***	-0.382***	-8.158***	-8.202***	-3.826***	-2.934***
		[0.004]	[0.085]	[1.586]	[1.539]	[0.699]	[0.606]
	Firm Profitability	-0.419**	-0.041***	0.707***	0.706***	-0.172***	-0.160***
		[0.177]	[0.004]	[0.057]	[0.056]	[0.026]	[0.026]
	Firm Sales Growth	-0.001	-0.395**	-5.721*	-6.140*	-1.075	-0.271
		[0.002]	[0.177]	[3.185]	[3.163]	[1.448]	[1.471]
	Firm Depend. Ext. Fin.	0.001	0.001	0.142***	0.141***	0.043***	0.042***
		[0.002]	[0.002]	[0.034]	[0.034]	[0.015]	[0.015]
	Number of firms	500	500	491	491	499	499

Notes: Hierarchical linear mixed model estimation, with observations clustered at the industry- and country-levels. Columns (1), (3) and (5) present the results from regressing corporate risk-taking on cultural and macroeconomic variables belonging to the host country a firm operates in. Columns (2), (4) and (5) present the results from regressing risk-taking indicators on cultural and macroeconomic variables belonging to the origin country a firm comes from. All regressions exclude U.S. firms. Robust standard errors in parentheses. Stars ^***, ^**, and ^* denote significance at the 1, 5, and 10 percent levels respectively.

Thus, for a firm belonging to country A (with headquarters in country A) but operating in country B, most of country B’s cultural norms will not impact the risk-taking appetite of that firm. However, country A’s cultural values will be significant with the predicted sign in explaining corporate risk-taking decisions. This result is somewhat surprising. As an example, it implies that the risk-taking behavior of a Japanese⁷ branch operating in the U.S. is less determined by American cultural values, but rather by Japanese values. As another example, the risk-taking behavior of a German firm operating in highly uncertainty-averse Greece will not be determined by Greek cultural norms as much as by German values.

These results suggest that even if a firm moves its business to another country, it takes a while for it to adopt new cultural norms, and that it will continue to behave according to the cultural norms of the society it originated in. Domestic regulatory laws will have different effects and produce different outcomes for domestic versus foreign firms. Foreign firms will most likely exhibit the same behavior as domestic firms in the country of origin. Thus, a firm coming from a society which values individualism, but operating in a collectivist society will take on more risk than a domestic firm in the collectivist society. More broadly, this suggests that the same regulatory structures can produce very different outcomes for culturally different firms (domestic versus foreign firms).

These surprising results also open up the question as to whether globalization can affect the stability of sovereigns. If the risk-taking behavior of a foreign branch is not determined by the host country’s values in as much as by the parent country’s values, we are left with the question of whether separate regulatory frameworks should be applied to foreign versus domestic firms. While a worthwhile topic, this idea should constitute the subject of further in-depth research. A caveat of the results presented in this section is that the sample of foreign firms is very limited (only 500 foreign firms) compared to the sample of domestic firms (about 20 000 entities) and more research is needed in order to draw reasonable conclusions. A further limitation of these results is that the analysis does not take into account the length of time a foreign firm has operated in a country, nor the cultural diversity of the management board, two additional caveats which might strengthen or weaken certain conclusions.

IX. Concluding Remarks

Because firm practices, institutional/political systems, economic development and social mores are intertwined with the national culture from which they derive, studying corporate risk-taking decisions under a cultural umbrella seems appropriate. This paper has taken a different approach to the empirical study of the impact of culture on corporate risk-taking. Whereas prior literature has focused mostly on the behavior of firms in the financial or in the manufacturing industry separately, this paper has considered firms in each industry in a market-economy simultaneously, trying to infer whether there are differences between the effects of culture on corporate risk-taking behaviors of firms not only across countries, but also across industries. Furthermore, whereas prior work has largely studied the impact of culture on firm risk-taking through ordinary least squares or generalized least squares estimation techniques, this paper has used hierarchical linear mixed models to identify both the direct and indirect effects of national cultural values on corporate risk-taking.

The evidence indicates that culture remains an important determinant of corporate risk-taking decisions, even after taking into account its indirect effects on the institutional, economic and industrial environments in which a firm operates in. Firms in societies which are highly uncertainty-averse, low in individualism and high in power distance will tend to take on less risk. This pattern is especially strong for firms operating in industrial sectors which are more informationally opaque. The evidence also suggests that the effects of formal institutions, such as shareholders’ protection or creditors’ rights, depend on the level of informal national cultural values such as individualism. Culture interacts with the various social, political and economic forces to produce very different and specific outcomes.

The evidence also indicates that these results hold only for domestic firms. The behavior of foreign firms is most likely determined by the cultural norms of the society they originated in.

In designing this study, we had to make some tradeoffs in examining the effects of both formal and informal institutions on corporate risk-taking behaviors. Firstly, the cross-cultural comparison of firm risk-taking is constrained by data availability. In the contemporary globalized world, it is more likely than not that firms are managed by executives with diverse cultural backgrounds. Future research would benefit from a more detailed analysis which also examines the cultural distribution of a firm’s board of directors and individual-level cultural traits. Secondly, the definitions of formal and informal institutions are complex and multi-faceted and require more in-depth analysis. Whether the national cultural dimensions can be considered as de facto practices (”as is”) or as general values (”as should be”) is important for conceptualizing their effects on corporate risk-taking and further research is clearly warranted. Thirdly, corporate risk-taking practices may vary across regions and organizations within the same country. Future research would benefit from analyzing the influence of regional and organizational differences on the heterogeneity of firm-level risk-taking behaviors within countries. And lastly, future research would also benefit from a better decomposition of risk-taking into the active decisions that a firm’s managers take versus those institutional/regulatory constraints on risk-taking. For example, further research could look at cross-industrial differences in the regulatory environments and how regulation of different industries affects corporate risk-taking.