INTRODUCTION
A variety of conceptual and empirical studies have examined the relationship between culture and corruption (Beets, Reference Beets2005; Chandler & Graham, Reference Chandler and Graham2010; Davis & Ruhe, Reference Davis and Ruhe2003; DiRienzo, Das, Cort, & Burbridge, Reference DiRienzo, Das, Cort and Burbridge2007; Getz & Volkema, Reference Getz and Volkema2001; Husted, Reference Husted1999; Park, Reference Park2003; Martin, Cullen, Johnson, & Parbotheeah, Reference Martin, Cullen, Johnson and Parbotheeah2007; Sanyal, Reference Sanyal2005; Seleim & Bontis, Reference Seleim and Bontis2009; Treisman, Reference Treisman2000). The results of this research support the importance of culture, however, are inconsistent with respect to specific cultural dimensions. For instance, Husted (Reference Husted1999) and Park (Reference Park2003) found that power distance facilitates corruption. Seleim and Bontis’ (Reference Seleim and Bontis2009) results, however, did not confirm this finding. Similarly, DiRienzo et al. (Reference DiRienzo, Das, Cort and Burbridge2007) and Sanyal and Samanta (Reference Sanyal and Samanta2002) discovered a positive relationship between masculinity and corruption, while Getz and Volkema (Reference Getz and Volkema2001) did not find such a relationship. Uncertainty avoidance was found to be positively related to corruption by Getz and Volkema (Reference Getz and Volkema2001) and Seleim and Bontis (Reference Seleim and Bontis2009), while Park (Reference Park2003) did not find a relationship between the two variables. Although individualism has been shown to decrease corruption (Davis & Ruhe, Reference Davis and Ruhe2003; Seleim & Bontis, Reference Seleim and Bontis2009), this result is not confirmed by DiRenzio et al. (Reference DiRienzo, Das, Cort and Burbridge2007), and, in fact, Martin et al. (Reference Martin, Cullen, Johnson and Parbotheeah2007) uncovered that collectivism decreases corruption. Finally, humane orientation has been found to both increase (Seleim & Bontis, Reference Seleim and Bontis2009) and decrease corruption (Martin et al., Reference Martin, Cullen, Johnson and Parbotheeah2007). We suggest that these inconsistencies may in part be due to how culture was conceptualized and measured in a specific study.
Although cultural frameworks exist (e.g., Hofstede, Reference Hofstede2001), the debate on appropriate conceptualization of culture continues (Hofstede, Reference Hofstede2010, Reference Hofstede2006; Maseland & van Hoorn, Reference Maseland and van Hoorn2009; McCrae, Terracciano, Realo, & Allik, Reference McCrae, Terracciano, Realo and Allik2008; Taras, Steel, & Kirkman, Reference Taras, Steel and Kirkman2010; Venaik & Brewer, Reference Venaik and Brewer2010). This debate reflects a longstanding dispute between cultural psychologists (e.g., Geertz, Reference Geertz1973) and anthropologists (e.g., Clifford & Marcus, Reference Clifford and Marcus1986) on whether culture should be conceptualized based on values (i.e., beliefs, norms, cognitions, and other intangible factors) or practices (i.e., habits, symbols, and other observable behavior). At the core of the debate is the often implicit assumption that individuals’ values and behaviors should be consistent with each other and that, therefore, cultural characteristics can be captured by an assessment of either values or behaviors (Triandis, Reference Triandis1995).
Hofstede’s (Reference Hofstede2006) widely recognized conceptualization of culture is reflective of such a value-behavior consistency assumption. His framework does not differentiate between values and practices and is based on questions that ask respondents to indicate the extent to which “In this society, the accepted norm is to …” In contrast, the GLOBE researchers suggest a conceptualization of culture that includes measures of values and practices (House et al., Reference House, Hanges, Ruiz-Quintanilla, Dorfman, Javidan, Dickson and Gupta2004). They argue that values and practices are different constructs—namely “how a society ought to be” and “how a society is”—that relate to distinct phenomena and thus, may vary (Maseland & van Hoorn, Reference Maseland and van Hoorn2009).
This cultural debate is akin to the discussion about the relationship between “ought” and “is” in ethical philosophy. While Pragmatists claim that values (“ought”) and practices (“is”) might be consistent (De Gaynesford, Reference De Gaynesford2006; Putnam, Reference Putnam2002), Logical Positivists suggest the possibility of a value/fact dichotomy and that deriving “ought” from “is” may result in a naturalistic fallacy error (Friedrich, Reference Friedrich2010; Friedrich, Kierniesky, & Cardon, Reference Friedrich, Kierniesky and Cardon1989; Hume, Reference Hume1739). In other words, the debate centers on whether “ought” and “is” might be consistent (Pragmatists) or not (Logical Positivists). Following Logical Positivists, we recognize that cultural values (“ought”) and practices (“is”) might be inconsistent and present cultural discrepancy—the difference between cultural values and practices—as an alternative conceptualization of culture beyond those offered by Hofstede (Reference Hofstede1980) and House et al. (Reference House, Hanges, Ruiz-Quintanilla, Dorfman, Javidan, Dickson and Gupta2004). Our study explores the value of such a conceptualization by applying the construct of cultural discrepancy to explain national corruption. Considering that previous research has identified national wealth as the strongest predictor of corruption (Beets, Reference Beets2005; Husted, Reference Husted1999, Reference Husted1994), we also examine the relationship between cultural discrepancy and corruption at different national wealth levels.
Notwithstanding its exploratory nature, this study makes three theoretical contributions to business ethics. First, we offer a unique approach to the conceptualization of culture that considers both values and practices. Second, we contribute to the “ought” vs. “is” debate in ethical philosophy by demonstrating that there is value in acknowledging possible inconsistencies between values (“ought”) and practices (“is”). Third, we contribute to understanding how cultural conditions promote or distract from corruption, both in general and at different national wealth levels. Hereby, we address Doh, Husted, Matten, & Santoro’s (Reference Doh, Husted, Matten and Santoro2010) challenge to conduct research that explicitly integrates business ethics with international business. Finally, the results of this study have practical implications for ethical behavior in business. Decision makers may better understand why corruption occurs in societies and develop new ideas to more effectively confront this problem.
The remainder of this article is organized as follows. In the theoretical section, we introduce cultural discrepancy, define corruption, differentiating two distinct types, pervasive and arbitrary corruption, and develop hypotheses about the relationship between discrepancies in specific cultural dimensions and both corruption types. In the empirical section, we test these hypotheses with a sample of 54 countries. We also conduct two exploratory analyses to further investigate the hypothesized relationships. First, we explore the relationships between cultural discrepancies and corruption at different national wealth levels; second, we explore whether there is value in acknowledging differences between “ought” and “is” by comparing regression models that use cultural discrepancies vs. cultural values vs. cultural practices. The paper concludes with a discussion of the results, theoretical and practical implications for business ethics, an acknowledgement of limitations, and guidelines for future research.
THEORETICAL FRAMEWORK
Cultural Discrepancy
We derive the construct of cultural discrepancy from the ongoing debate in ethical philosophy between Pragmatists and Logical Positivists on values vs. facts or “ought” vs. “is.” Pragmatists claim that moral prescriptions (“ought”) and empirically observed human behaviors (“is”) might be consistent and argue that knowledge of facts presupposes knowledge of values (Bucholz & Rosenthal, Reference Bucholz and Rosenthal2001; Putnam, Reference Putnam2002). Indeed, Putnam (Reference Putnam2002: 396) states that facts often “…. only come into view through the lenses of an evaluative outlook.” In contrast, Logical Positivists propose a value/fact dichotomy and claim that “ought” and “is” are distinct concepts and might be inconsistent. Deriving “ought” from “is” may result in a naturalistic fallacy error because it might be falsely concluded that observed behavior reflects the desirable (Friedrich, Reference Friedrich2010; Friedrich et al., Reference Friedrich, Kierniesky and Cardon1989; Hume, Reference Hume1739). The business ethics literature provides examples for such inconsistencies: members of organizations might be aware of what their moral obligations are (“ought”), but their behaviors (“is”) may be deviant (Lea, Reference Lea1999).
Since ethical philosophers do not agree upon the value/fact dichotomy, we propose that in studying the relationship between culture and ethical decision-making, an explicit acknowledgement of each, “ought” and “is,” might be beneficial. In particular, we consider “ought” as cultural values, defined as “a conception, explicit or implicit, distinctive of an individual or characteristic of a group, of the desirable” and “is” as cultural practices, defined as “the selection from available modes, means and ends of actions” (Kluckhohn, Reference Kluckhohn, Parsons and Shils1951: 395). Recognizing the debate between Pragmatists and Logical Positivists, we use a comprehensive conceptualization of culture that comprises both values (how a society should be) and practices (how a society is), which may or may not fall apart.
Accordingly, we introduce the concept of cultural discrepancy, defined as the difference between values (“ought”) and practices (“is”) on specific cultural dimensions. It captures the degree to which cultural values and practices diverge and ranges from low to high. Low cultural discrepancy means that practices are consistent with values; something is equally desirable and practiced (values = practices). A high discrepancy occurs if the values reflected in a cultural dimension do not coincide with the respective practices: something is not equally desirable and practiced (values ≠ practices). The presented construct is valence-free. It only relates to the degree rather than to the kind or quality of inconsistencies, that is, whether cultural values exceed (values > practices) or fall short of practices (values < practices). Our conceptualization corresponds to the classic debate on “ought” vs. “is,” which merely relates to whether “ought” and “is” are consistent, but not to whether “ought” is greater or smaller than “is.” We expect that the mere inconsistency in a particular dimension plays a role in explaining unethical behavior such as corruption—independent of whether values exceed or fall short of practices.
Corruption
Corruption is the “misuse of entrusted power for private gains” (Transparency International, 2011: 4). It involves collaboration of at least two parties (DiRienzo et al., Reference DiRienzo, Das, Cort and Burbridge2007), who exchange benefits (Rabl, Reference Rabl2011) where each party engages in behavior that produces desirable results for the other party. Per se, these kinds of arrangement make sense, but they become dysfunctional if they prevent a society’s economic and political systems from functioning effectively. Further, the misuse of power for private benefits violates the hypernorm of “necessary social efficiency” and is unethical (Donaldson & Dunfee, Reference Donaldson and Dunfee1994) and illegal in most countries (Chandler & Graham, Reference Chandler and Graham2010).
Uhlenbruck, Rodriguez, Doh, and Eden (Reference Uhlenbruck, Rodriguez, Doh and Eden2006) and Rodriguez, Uhlenbruck, and Eden (Reference Rodriguez, Uhlenbruck and Eden2005) differentiate between pervasive and arbitrary corruption. Pervasive corruption is also called predictable corruption because it represents an integral part of the society—it is built in (Nielsen, Reference Nielsen2003; Rodriguez et al., Reference Rodriguez, Uhlenbruck and Eden2005), a way business is done. Accordingly, this behavior is a social arrangement, which has been fully institutionalized within commercial activity (Uhlenbruck et al., Reference Uhlenbruck, Rodriguez, Doh and Eden2006). Pervasive corruption comprises two fundamental characteristics. First, it is socially valid, that is, accepted by the members of a society. Second, it is an omnipresent behavior that can be observed in everyday life (Rodriguez et al., Reference Rodriguez, Uhlenbruck and Eden2005). In contrast, arbitrary corruption is unpredictable corruption, resulting from uncertainty and ambiguity (Uhlenbruck et al., Reference Uhlenbruck, Rodriguez, Doh and Eden2006). Arbitrary corruption occurs when rules and laws are subject to capricious and varied interpretation (Rodriguez et al., Reference Rodriguez, Uhlenbruck and Eden2005). On one hand, this is the case when jurisdiction is tenuous (Oldenburg, Reference Oldenburg1987). People seek to fill this institutional void by claiming their own authority and interpreting tenuous rules for their private benefit (Rodriguez et al., Reference Rodriguez, Uhlenbruck and Eden2005). On the other hand, arbitrary corruption may also occur in more democratic and stable environments (Rodriguez et al., Reference Rodriguez, Uhlenbruck and Eden2005), where jurisdiction is so detailed and complex that law enforcement is characterized by continuous appeal processes (Oldenburg, Reference Oldenburg1987). Here, ambiguity and arbitrariness do not occur due to a lack of rules, but due to a complexity and excessiveness thereof.
HYPOTHESES
We propose that cultural discrepancy is related to corruption and that the direction of this relationship depends on the type of corruption, pervasive vs. arbitrary, and on the GLOBE cultural dimensions: in-group collectivism, institutional collectivism, power distance, assertiveness, uncertainty avoidance, future orientation, performance orientation, gender egalitarianism, and humane orientation (House et al., Reference House, Hanges, Ruiz-Quintanilla, Dorfman, Javidan, Dickson and Gupta2004).
Low Cultural Discrepancies and Pervasive Corruption
In-group collectivism, institutional collectivism, power distance, and assertiveness show properties entailed in the nature of corrupt behavior: a relational component in the form of mutual favor doing (Rabl, Reference Rabl2011) or a self-focused component in the form of private benefit seeking (O’Higgins, Reference O’Higgins2006). We propose that low discrepancies in these dimensions are associated with pervasive corruption. Specifically, we argue that a congruence of values and practices in these dimensions triggers corruption as an institutionalized social arrangement: people accept and engage in corruption on a day-to-day basis (Rodriguez et al., Reference Rodriguez, Uhlenbruck and Eden2005)—it is both desirable and practiced.
In-group collectivism is “the degree to which individuals express pride, loyalty, and cohesiveness in their organizations or families” (House et al., Reference House, Hanges, Ruiz-Quintanilla, Dorfman, Javidan, Dickson and Gupta2004: 30). It describes the individual integration into a social entity and the importance of interpersonal relationships (House et al., Reference House, Hanges, Ruiz-Quintanilla, Dorfman, Javidan, Dickson and Gupta2004). Collectivism is linked to the reciprocity norm, which requires us to repay what others have done for us (Whatley, Webster, Smith, & Rhodes, Reference Whatley, Webster, Smith and Rhodes1999). Given that corruption involves such mutual favor doing, it is perceived as a means of relationship building in collectivistic cultures—even though it might be illegal (Chandler & Graham, Reference Chandler and Graham2010). Accordingly, empirical research shows a positive relationship between collectivism and corruption (Davis & Ruhe, Reference Davis and Ruhe2003; Seleim & Bontis, Reference Seleim and Bontis2009).
Regarding in-group collectivism discrepancy, we argue that low discrepancy provides a double legitimacy for pervasive corruption. The organizational literature defines an action as legitimated, if it is appropriate (Suchmann, Reference Suchmann1995), and such legitimacy can occur at the normative (value-based legitimacy) and behavior level (behavioral legitimacy) (Levi, Sacks, & Tyler, Reference Levi, Sacks and Tyler2009). Low in-group collectivism discrepancy represents such a double legitimacy: it renders relationship-orientation as equally desirable and practiced; people conform to the norm of reciprocity—values are aligned with practices. As a result, mutual favor doing is both accepted (value-based legitimacy) and widely used (behavioral legitimacy). This constellation provides ideal conditions for pervasive corruption (Nielsen, Reference Nielsen2003), a socially accepted and omnipresent behavior (Rodriguez et al., Reference Rodriguez, Uhlenbruck and Eden2005).
In contrast, high in-group collectivism discrepancy means that values are not aligned with practices (values ≠ practices); mutual favor doing is not equally desirable and practiced. This inconsistency may come in two forms: mutual favor doing can be more desirable than practiced (values > practices) or more practiced than desirable (values < practices). Yet, the quality of the discrepancy may not matter, as the inhibiting effect on corruption will be the same. This is because both constellations contradict the nature of pervasive corruption as socially accepted and omnipresent (Rodriguez et al., Reference Rodriguez, Uhlenbruck and Eden2005) and at least one of the two legitimations for such unethical behavior is not given.
In sum, we propose that low in-group collectivism discrepancy triggers pervasive corruption. This means that a higher discrepancy in this dimension decreases the level of pervasive corruption, which corresponds to the following hypothesis:
Hypothesis 1: There is a negative relationship between in-group collectivism discrepancy and pervasive corruption.
Institutional collectivism is “the degree to which organizational and societal institutional practices encourage and reward collective distribution of resources and collective action” (House et al., Reference House, Hanges, Ruiz-Quintanilla, Dorfman, Javidan, Dickson and Gupta2004: 30). While in-group collectivism focuses on “in-groups,” institutional collectivism refers to society and its institutions as a whole. However, institutional collectivism, like in-group collectivism, describes the extent of individual integration into an entity (House et al., Reference House, Hanges, Ruiz-Quintanilla, Dorfman, Javidan, Dickson and Gupta2004). Hence, the same reasoning as for in-group collectivism applies, and we suggest that:
Hypothesis 2: There is a negative relationship between institutional collectivism discrepancy and pervasive corruption.
Power distance is the “degree to which members of a collective expect power to be distributed equally” (House et al., Reference House, Hanges, Ruiz-Quintanilla, Dorfman, Javidan, Dickson and Gupta2004: 30). In high power distance countries, people in subordinate positions accept the superiority of others (House et al., Reference House, Hanges, Ruiz-Quintanilla, Dorfman, Javidan, Dickson and Gupta2004). As a result, people in endowed positions are able to follow their own interest, even at the expense of people in lower positions who would not doubt the appropriateness of their actions. Such constellations may induce superiors to abuse power for private benefits, which corresponds to the definition of corruption (O’Higgins, Reference O’Higgins2006). Further, high power distance is associated with paternalism: superiors favor subordinates who remain loyal (Husted, Reference Husted1999). Paternalism reinforces mutual favor doing, which is inherent to corruption (Rabl, Reference Rabl2011). Indeed, prior research shows that power distance increases corruption (Davis & Ruhe, Reference Davis and Ruhe2003; Getz & Volkema, Reference Getz and Volkema2001; Husted, Reference Husted1999; Park, Reference Park2003).
We argue that low power distance discrepancy fosters pervasive corruption because this constellation reflects social inequality to be equally desirable and practiced. Hence, people in powerful positions receive a double legitimacy to misuse entrusted power for private gains. First, such behavior is perceived as socially valid. Second, (mis)using power is simply a way business is done, people are used to it. Corruption then is pervasive, that is, an integral part of society (Rodriguez et al., Reference Rodriguez, Uhlenbruck and Eden2005). In contrast, high power distance discrepancy means that the unequal distribution of power is more desirable than practiced or vice versa. In the first case, misusing power may be accepted, but it is simply not practiced to the aspired degree and the behavioral legitimacy for pervasive corruption—as an institutionalized activity (Uhlenbruck et al., Reference Uhlenbruck, Rodriguez, Doh and Eden2006)—is missing. In the second case, misusing power is practiced, but the value-based legitimacy for pervasive corruption—as being widely accepted (Uhlenbruck et al., Reference Uhlenbruck, Rodriguez, Doh and Eden2006)—is missing. Accordingly, we propose that a low discrepancy in power distance is associated with pervasive corruption, reflected in a negative relationship between the two constructs:
Hypothesis 3: There is a negative relationship between power distance discrepancy and pervasive corruption.
Assertiveness is the “degree to which individuals are assertive, confrontational and aggressive” (House et al., Reference House, Hanges, Ruiz-Quintanilla, Dorfman, Javidan, Dickson and Gupta2004: 30). Members of assertive societies have sympathy for the strong, value tough and aggressive behavior, and seek to be dominant (House et al., Reference House, Hanges, Ruiz-Quintanilla, Dorfman, Javidan, Dickson and Gupta2004), reflecting selfishness. As corruption represents self-interested behavior (Husted, Reference Husted1994) that aims at private gains (O’Higgins, Reference O’Higgins2006), we argue that this cultural dimension is also related to corruption. In particular, low assertiveness discrepancy means that dominance and self-focus are equally desirable and practiced, resulting in pervasive corruption because it provides both value-based and behavioral legitimacy for aggressively seeking private interests. In contrast, a high assertiveness discrepancy lacks one of the two legitimacies. If assertiveness is more desirable than practiced, people may not dare to seek private interests at the expense of the public because this is unusual (no behavioral legitimacy). If assertiveness is more practiced than desirable, forcefully seeking private interests would be frowned upon (no value-based legitimacy). Hence, we propose that a low, but not a high, discrepancy in assertiveness is associated with pervasive corruption, as represented by a negative relationship between the two constructs:
Hypothesis 4: There is a negative relationship between assertiveness discrepancy and pervasive corruption.
High Cultural Discrepancies and Arbitrary Corruption
For uncertainty avoidance, future orientation, and performance orientation, we argue that cultural discrepancy is related to arbitrary corruption because these dimensions describe adherence to explicit or implicit (work-related) rules, which is central to this type of corruption (Uhlenbruck et al., Reference Uhlenbruck, Rodriguez, Doh and Eden2006). We expect that high cultural discrepancies in these dimensions foster arbitrary corruption because such value-practice incongruence leaves room for varied interpretations of rules.
Uncertainty avoidance is the “extent to which a society, organization, or group relies on social norms, rules, and procedures to alleviate unpredictability of future events” (House et al., Reference House, Hanges, Ruiz-Quintanilla, Dorfman, Javidan, Dickson and Gupta2004: 30). It relates to the adherence to law and order and is also referred to as rule orientation (Venaik & Brewer, Reference Venaik and Brewer2010). High uncertainty avoidance is associated with detailed societal requirements, formal procedures, and laws that inform citizens on how to behave (House et al., Reference House, Hanges, Ruiz-Quintanilla, Dorfman, Javidan, Dickson and Gupta2004). A high discrepancy in this dimension occurs when the degree of desired rule orientation is not in line with the practiced degree (values ≠ practices). Although two such constellations are conceivable, we suggest that they have the same impact on corruption because both represent precarious voids that trigger ambiguity—a precondition for arbitrary corruption (Rodriguez et al., Reference Rodriguez, Uhlenbruck and Eden2005).
First, values can exceed practices (values > practices): people aspire to clear-cut rules, but rules are non-existent or ambiguous. In such a constellation, existing regulations are perceived as inappropriate to alleviate the unpredictability of future events (House et al., Reference House, Hanges, Ruiz-Quintanilla, Dorfman, Javidan, Dickson and Gupta2004). This void comprises uncertainty and ambiguity in the application of rules and laws, which fosters an arbitrary interpretation thereof. Such a free interpretation is not only possible (as rules are vague), but also necessary (as people aspire predictability). Since ambiguity and varied rule interpretation are the primary reason for arbitrary corruption (Uhlenbruck et al., Reference Uhlenbruck, Rodriguez, Doh and Eden2006), we propose that the described constellation fosters such behavior.
Second, practices can exceed values (values < practices): people wish clear guidance, but the rules are too detailed. This constellation represents a perceived overregulation, which often occurs in developed democracies (Rodriguez et al., Reference Rodriguez, Uhlenbruck and Eden2005). Given such high specificity, law enforcement is characterized by continuous and long-lasting appeal processes (Oldenburg, Reference Oldenburg1987). This situation again leads to uncertainty and ambiguity—not resulting from the lack of rules, but from their complexity and perceived overregulation. As uncertainty leaves room for a varied interpretation of rules (Rodriguez et al., Reference Rodriguez, Uhlenbruck and Eden2005), we argue that this second constellation also fosters arbitrary corruption.
In contrast, a low uncertainty avoidance discrepancy means that the aspired degree of guidance and orientation corresponds to the practiced degree (values = practices). Hence, there is no precarious void for a capricious interpretation of rules, and thus, no reason for arbitrary corruption. We expect that:
Hypothesis 5: There is a positive relationship between uncertainty avoidance discrepancy and arbitrary corruption.
A similar reasoning applies to future orientation, the “extent to which individuals engage in future-oriented behaviors, such as delaying gratification, planning, and investing in the future” (House et al., Reference House, Hanges, Ruiz-Quintanilla, Dorfman, Javidan, Dickson and Gupta2004: 30). A high discrepancy in this dimension occurs when planning ahead is not equally desirable and practiced. Again, though this discrepancy may come in two forms, both should have the same impact on arbitrary corruption. First, people may consider it desirable to plan ahead, but the members of their society do not comply with this value (values > practices). The lack of clear-cut future plans entails ambiguity, which exacerbates an objective view of rules. Laws and policies are subject to capricious interpretation (Rodriguez et al., Reference Rodriguez, Uhlenbruck and Eden2005), which tempts citizens to engage in arbitrary corruption (Uhlenbruck et al., Reference Uhlenbruck, Rodriguez, Doh and Eden2006). Second, high future orientation discrepancy may occur, when planning ahead is more practiced than desirable (values < practices). Such perceived regulatory overkill may also result in ambiguity and leave room for a capricious interpretation of rules and arbitrary corruption. Hence, we propose:
Hypothesis 6: There is a positive relationship between future orientation discrepancy and arbitrary corruption.
Performance orientation is the “degree to which a collective encourages and rewards group members for performance improvement and excellence” (House et al., Reference House, Hanges, Ruiz-Quintanilla, Dorfman, Javidan, Dickson and Gupta2004: 30). Societies with a high performance orientation give rewards based on performance only and the established appraisal system emphasizes visible results. Low performance orientation is associated with ambiguous appraisal systems: rewards are also based on loyalty and cooperative spirit (House et al., Reference House, Hanges, Ruiz-Quintanilla, Dorfman, Javidan, Dickson and Gupta2004). As such, performance orientation, like uncertainty avoidance and future orientation, relates to rules, but these rules specifically relate to appraisal systems.
Again, high performance orientation discrepancy may come in two constellations with similar impacts on arbitrary corruption. First, a performance-based appraisal system may be more desirable than practiced (values > practices). This condition may foster arbitrary corruption because people want to be evaluated based on their performance, but the visible results of their work do not count to the aspired extent. Hence, work appraisal is subject to interpretation, which facilitates arbitrary corruption (Rodriguez et al., Reference Rodriguez, Uhlenbruck and Eden2005). In the second constellation, a performance-based appraisal system is practiced to an undesirable extent (values < practices). People perceive a regulatory overkill and may not understand the over-complex appraisal systems. The resulting uncertainty leaves room for a capricious interpretation of the appraisal systems, and thus, give rise to arbitrary corruption (Uhlenbruck et al., Reference Uhlenbruck, Rodriguez, Doh and Eden2006). Hence:
Hypothesis 7: There is a positive relationship between performance orientation discrepancy and arbitrary corruption.
Cultural Discrepancies Unrelated to Corruption
We expect that gender egalitarianism and humane orientation discrepancies are unrelated to corruption. Gender egalitarianism is the “degree to which a collective minimizes gender inequality” (House at al., Reference House, Hanges, Ruiz-Quintanilla, Dorfman, Javidan, Dickson and Gupta2004: 30). High gender egalitarianism is associated with equal gender roles; low gender egalitarianism means that gender roles are predetermined, with women having a lower status (House at al., Reference House, Hanges, Ruiz-Quintanilla, Dorfman, Javidan, Dickson and Gupta2004). As such, gender egalitarianism does not seem to be relevant to pervasive corruption as systematic mutual favor-doing and private benefits seeking or arbitrary corruption as a capricious rule interpretation.
Humane orientation is the “degree to which a collective encourages and rewards individuals for being fair, altruistic, friendly, generous, caring, and kind to others” (House et al., Reference House, Hanges, Ruiz-Quintanilla, Dorfman, Javidan, Dickson and Gupta2004: 30). High humane orientation implies care for others; low humane orientation is associated with the expectation that the state provides support for people’s well-being (House et al., Reference House, Hanges, Ruiz-Quintanilla, Dorfman, Javidan, Dickson and Gupta2004). This dimension does not relate to pervasive corruption, which involves mutual benefits or benefits for the corrupt person. Humane orientation speaks to taking care of others without expecting a favor in return. Further, it does not relate to rules or performance standards; thus, it is unrelated to arbitrary corruption.
METHOD
Sample and Measures
The countries were chosen based on data availability for corruption, GLOBE dimensions, and controls. For three countries, the cultural variables were available for sub-populations only (Switzerland: French vs. German, South Africa: black vs. white, Germany: East vs. West). Here, we calculated a mean across sub-populations. For Kuwait and Taiwan, we replaced one missing control variable (i.e., access to information scores) with the respective scores from neighbor countries that display similar development levels (Qatar, Hong Kong, respectively). Ultimately, we retained a sample of 54 countries. Footnote 1
All variables were measured at the national level, using different sources to prevent common method variance (Chang, van Witteloostuijn, & Eden, Reference Chang, van Witteloostuijn and Eden2010). Cultural discrepancy (CDISCjk), was measured based on the response bias corrected scores of the GLOBE values and practices on the nine cultural dimensions (House et al., Reference House, Hanges, Ruiz-Quintanilla, Dorfman, Javidan, Dickson and Gupta2004). CDISCjk was calculated for each country j as the difference between the value and the practice score on the respective GLOBE dimension k. We used the absolute rather than the squared value of the difference (Homburg, Wieseke, & Bornemann, Reference Homburg, Wieseke and Bornemann2009), because squaring a score lays emphasis on outliers, which may distort regression results (Cohen, Cohen, West, & Aiken, Reference Cohen, Cohen, West and Aiken2003). High (low) CDISCjk scores represent a high (low) cultural discrepancy in the respective cultural dimension:
${\rm{CDISC}}_{{\rm{jk}}} = \left| {{\rm{cultural}}\,{\rm{value}}_{{\rm{jk}}} - {\rm{cultural}}\,{\rm{practice}}_{{\rm{jk}}} } \right|$
Such gap scores are widely used in business research (e.g., Homburg et al., Reference Homburg, Wieseke and Bornemann2009), but they comprise two problems. The first problem is the low reliabilities of difference scores, which occur when both component scores have low reliabilities and are positively correlated (Edwards, Reference Edwards2001). Table 1, which displays the correlations between GLOBE values and practices in our sample, indicates eight negative correlations, and the only positive correlation is relatively low (gender egalitarianism: +0.41). Hence, there should be no reliability problem. The second problem refers to high correlations of the difference score with at least one of the components, which may lead to low discriminant validity, high multicollinearity when used conjointly in regression analyses, and spurious correlations with other variables (Edwards, Reference Edwards2001). Therefore, the difference scores may not explain more variance of a criterion variable than the components alone. Table 1 presents the correlations of cultural discrepancies with the respective GLOBE values and practices in our sample. CDISCjk shows correlations beyond 0.7 with three GLOBE value scores (power distance: -0.77, future orientation: +0.82, gender egalitarianism: +0.71) and with five GLOBE practices scores (in-group collectivism: +0.87, power distance: 0.85, future orientation: -0.88, performance orientation: -0.85, humane orientation: -0.92). Therefore, the CDISCjk scores will only add to the explanation of corruption if they perform better than GLOBE values and practices separately. Thus, we will compare models that use CDISCjk, values, and practices as predictor variables.
Correlations between GLOBE values, practices, and cultural discrepancies
N = 54.
† p < 0.10, * p < 0.05; ** p < 0.01; *** p < 0.001 (two-tailed).
IGC = in-group collectivism, INC = institutional collectivism, PD = power distance, ASS = assertiveness, UA = uncertainty avoidance, FO = future orientation, POR = performance orientation, GEG = gender egalitarianism, HO = humane orientation.
In the absence of validated measures for pervasive and arbitrary corruption, we had to approximate their measurement. We used the Transparency International’s (2011) Corruption Perception Index (CPI), a valid, reliable and widely used measure of corruption (DiRienzo et al., Reference DiRienzo, Das, Cort and Burbridge2007; Husted, Reference Husted1999). CPI indicates the degree to which corrupt practices are perceived to exist in a country, based on assessments and business opinion surveys and bundled to a weighted average (0 = highly corrupt to 10 = not corrupt). For the sake of consistent scale labeling, we reversed the CPI (0 = not corrupt; 10 = highly corrupt). Using the CPI as a proxy variable for both corruption types seems acceptable given the exploratory nature of our study. Further, both pervasive and arbitrary corruption will be reflected in the same sort of behavior, namely the misuse of entrusted power for private benefits.
National wealth has been shown to be negatively related to corruption (Husted, Reference Husted1999); it was measured by GDP per capita (in US$, at PPP), taken from the IMD World Competitiveness Online database (WCO, 2011). Access to information indicates the access, usage, and skills of/in information and communication technology and has been found to decrease corruption (DiRienzo et al., Reference DiRienzo, Das, Cort and Burbridge2007). This is because such access provides citizens with means and knowledge regarding public and private business transactions and, hence, generates a transparent society. We measured access to information through the Information and Communication Technology Development Index (IDI), ranging from 0 (low) to 10 (high access to information), taken from International Telecommunication Union (2011).
Analyses
Although multivariate analyses are desirable, our small sample size may inflate error rates and limit generalizability of regression results (Osborne, Reference Osborne2008). The pertinent literature is inconsistent regarding the minimum ratio of cases-per-independent variables required for multiple regression analysis (Brooks & Barcikowski, Reference Brooks and Barcikowski2012; Knofczynski & Mundfrom, Reference Knofczynski and Mundfrom2008; Tabachnick & Fidell, Reference Tabachnick and Fidell2014). We followed Miller and Kunce’s (Reference Miller and Kunce1973) rule of thumb to consider ten observations per independent variable as sufficient, which is well above the five-to-one ratio viewed by others (Tonidandel, Williams, & LeBreton, Reference Tonidandel, Williams, LeBreton, Lance and Vandenberg2015).
Data analysis comprised three steps. First, we tested hypotheses about the relationship between cultural discrepancies and corruption, H1 to H7, on the full cross-sectional data (n = 54 countries). Here, the rule of thumb allowed for a maximum of five predictors in a regression analysis. As there would be nine predictors, seven cultural discrepancies and two controls, we restricted our analysis to bivariate correlations.
In the second step, we retrieved the most salient cultural discrepancies from the correlation results and explored their relationship with national corruption at different national wealth levels. Given the high variation in GDP per capita, from $3,492 (India) to $88,160 (Qatar), we divided the sample into three subgroups with different wealth levels. Using $10,000 and $30,000 as meaningful cutoff values yielded three subgroups, labeled as poor (n = 15), medium (n = 18), and rich countries (n = 21). In these subgroups, we also ran bivariate correlations rather than multiple regression analyses because of the small sample sizes.
The third step of analysis explored the relative predictive power of cultural discrepancies, the corresponding cultural values, and cultural practices. Using the full cross-sectional data (n = 54), but only retaining the most significant cultural dimensions from the correlation analysis, yielded less than five independent variables (i.e., more than ten observations per independent variable), which allowed for multiple regression analyses. We conducted multistep hierarchical regressions to examine whether the cultural variables significantly increased the variance explained in corruption beyond the controls (Cohen et al., Reference Cohen, Cohen, West and Aiken2003). Starting with a baseline Model 1 with the two controls, the cultural variables were added in Model 2. We used ordinary least squares (OLS) regressions since our data were cross-sectional and the dependent variable was continuous (Husted, Reference Husted1999). To account for potentially grouped standard errors in clustered data (Cameron & Trivedi, Reference Cameron and Trivedi2010), we defined clusters by six regions (Europe, Asia, Africa, Oceania, North America, and South America) and estimated the regression coefficients based on cluster-robust standard errors. This procedure represents a more conservative estimation as it prevents a downward bias in the standard errors (Wooldridge, Reference Wooldridge2003). Again, we added subgroup analyses to compare the role of cultural discrepancies, values, and practices at different wealth levels. Given the limited number of observations in the three subgroups, multivariate regressions were inappropriate, but we computed partial correlations between each cultural dimension and corruption, using the two control variables as covariates.
RESULTS
Relationship between Cultural Discrepancies and Corruption
Table 2 presents the bivariate correlations among all variables comprised in H1 to H7. H1 to H4 predict a negative relationship between pervasive corruption and the discrepancies in in-group collectivism (H1), institutional collectivism (H2), power distance (H3), and assertiveness (H4). The correlation is significant and in the expected direction for in-group collectivism (r = -0.63, p < 0.001), significant and in the opposing direction for institutional collectivism (r = 0.32, p < 0.05), and marginally significant and in the opposing direction for power distance (r = 0.26, p < 0.10). The correlation between assertiveness discrepancy and corruption is not significant (r = 0.17, p > 0.10). Hence, H1 is supported, H2 to H4 are not.
Descriptive statistics and correlations
N = 54.
† p < 0.10, * p < 0.05; ** p < 0.01; *** p < 0.001 (two-tailed).
CPI rev. = Corruption perception index (reverse-coded), GDP = Gross Domestic Product per capita, IDI = Information and Communication Technology Development Index, disc = discrepancy, IGC = in-group collectivism, INC = institutional collectivism, PD = power distance, ASS = assertiveness, UA = uncertainty avoidance, FO = future orientation, POR = performance orientation, GEG = gender egalitarianism, HO = humane orientation.
H5 to H7 predict a positive relationship between arbitrary corruption and the discrepancies in uncertainty avoidance (H5), future orientation (H6), and performance orientation (H7). The correlations are significant or marginally significant and in the expected direction for uncertainty avoidance (r = 0.25, p < 0.10), future orientation (r = 0.65, p < 0.001), and performance orientation (r = 0.32, p < 0.05), supporting H5, H6, and H7.
Table 2 also shows the correlations for the two dimensions that we expected to be unrelated to corruption, neither pervasive nor arbitrary corruption. Unexpectedly, we observe a small negative and significant correlation for the discrepancy in gender egalitarianism (r = -0.33, p < 0.05) and a small negative, marginally significant correlation for the discrepancy in humane orientation (r = -0.23, p < 0.10).
Cultural Discrepancies and Corruption at Different Levels of National Wealth
Results of the correlation analyses show correlations between corruption and in-group collectivism and future orientation discrepancy beyond 0.6 (p < 0.001); all other correlations are minor. Therefore, we explore the relationship of cultural discrepancy and corruption at different levels of national wealth for these two salient dimensions only. The results of the correlation analyses presented in Table 3 show that in-group collectivism discrepancy is significantly and negatively related to corruption at all wealth levels, although the relationship is slightly weaker in rich (r = -0.48, p < 0.05) than in poor (r = -0.71, p < 0.01) and medium countries (r = -0.69, p < 0.01). The positive relationship of future orientation discrepancy is significant—and very strong—for rich countries only (r = 0.85, p < 0.001), but not significant for poor (r = 0.39, p > 0.10), and medium countries (r = 0.09, p > 0.10).
Correlations for IGC and FO discrepancies with corruption at different wealth levels
Bivariate correlations with corruption (CPI reverse-coded);
† p < 0.10, * p < 0.05; ** p < 0.01; *** p < 0.001 (two-tailed).
IGC = in-group collectivism, FO = future orientation.
Comparing GLOBE Discrepancies, Values, and Practices
We compared the explanatory contribution of cultural discrepancy, values, and practices by conducting three hierarchical regressions using cluster-robust standard errors. For each model, we first entered the two controls (national wealth and access to information) and then in-group collectivism and future orientation discrepancies, values, and practices, respectively (see Table 4). In addition to ∆R2, we report Akaike’s information criterion (AIC), which is more appropriate for non-nested model comparisons, with lower (higher) values indicating a better (worse) fit (Akaike, Reference Akaike1987).
Model comparison of the relationships between corruption and discrepancies, values, and practices
N = 54.
Dependent variable: corruption (CPI reverse-coded); unstandardized regression coefficients; cluster-robust standard errors (adjusted for the following six regional clusters: Europe, Asia, Africa, Oceania, North America, and South America).
† p < 0.10, * p < 0.05; ** p < 0.01; *** p < 0.001 (two-tailed).
GDP = Gross Domestic Product per capita, IDI = Information and Communication Technology Development Index, IGC = in-group collectivism, FO = future orientation.
R2 = variance in the dependent variable explained by the independent variables.
∆R2 = changes in R2 when adding the respective independent variables.
F-value = overall model test; significant F-values indicate model support.
∆F-value = changes in F-value when adding the respective independent variables.
In line with the correlation analyses presented earlier, in-group collectivism discrepancy decreases (b = -1.03, p < 0.001) and future orientation discrepancy increases corruption (b = 0.78, p < 0.05). For the GLOBE values, in-group collectivism is negative and marginally significant (b = -1.09, p < 0.10) and future orientation is not significant (b = 1.13, p > 0.10). For the GLOBE practices, in-group collectivism exerts a significant positive effect (b = 0.89, p < 0.05) and future orientation has a significant negative effect (b = -1.38, p < 0.05). The significant R-square changes indicate that the two cultural variables significantly contribute to the explanation of corruption, over and above the control variables, although this contribution is marginal in the case of cultural values. In general, the cultural discrepancies perform much better (∆R2 = 0.135, AIC = 147.67) than the cultural values (∆R2 = 0.027, AIC = 179.08) and well as the GLOBE practices (∆R2 = 0.135, AIC = 147.55).
Table 5 displays the results of partial correlations between the cultural dimensions and corruption for poor, medium, and rich countries Footnote 2 . Cultural values do not play a significant role in any subgroup. Further, in-group collectivism discrepancy is significantly related to corruption in poor countries, while cultural practices and values are not. For medium and rich countries, the discrepancies perform as well as the practices.
Partial correlations for IGC and FO discrepancies with corruption at different wealth levels
Partial correlations with corruption (CPI reverse-coded), when controlling for GDP per capita and IDI;
† p < 0.10, * p < 0.05; ** p < 0.01; *** p < 0.001 (two-tailed).
IGC = in-group collectivism, FO = future orientation.
DISCUSSION
The results of this study contribute to the business ethics literature in several ways. We will first discuss the results related to the construct of cultural discrepancy as a new approach to the conceptualization of culture and its derivation from the “ought” vs. “is” debate held in ethical philosophy. Then, we address how cultural discrepancy helps to understand the occurrence of an unethical behavior like corruption. Specifically, we elaborate on the hypothesized relationships between low and high cultural discrepancies and pervasive and arbitrary corruption, respectively. Further, considering the results of our exploratory analyses, we explain the role of national wealth in these relationships and assess the predictive power of cultural values, practices, and discrepancies.
Cultural Discrepancy Derived from the Ought vs. Is Debate
Our research presents cultural discrepancy as a new cultural concept. While extant research conceptualizes and measures culture either as values (e.g., Hofstede, Reference Hofstede2001) or as values and practices (e.g., House et al., Reference House, Hanges, Ruiz-Quintanilla, Dorfman, Javidan, Dickson and Gupta2004), we look at the difference between values (“ought”) and practices (“is”) on specific cultural dimensions. In doing so, we consider the discrepancy between values and practices itself as a distinct cultural characteristic. This idea was derived from the debate in ethical philosophy on whether “ought” (normative statements) and “is” (observed human behavior) are consistent or not. While Pragmatists argue that people may derive normative statements from observable facts (Putnam, Reference Putnam2002), Logical Positivists call for a value/fact dichotomy (Friedrich, Reference Friedrich2010). In considering cultural values and practices as specifications of “ought” and “is,” our research contributes to this debate in two ways. First, we show that there is value in acknowledging a potential dichotomy of values and practice. We provide preliminary empirical evidence that they may not only fall apart, but that this divergence matters for business ethics: it explains an unethical behavior like corruption. Second, our study results suggest that a pejorative view on either position—a convergence or divergence of values and practices—is premature. A high or low ought/is divergence may not necessarily be negative or positive, but it depends on the nature of the cultural dimension and the larger context. It is not simply about “ought” vs. “is”, but in determining the relevance of the consistency between the two, it matters what specific values and practices are addressed. For example, low divergence in values and practices related to in-group collectivism results in “negative” behavior and fosters corruption, whereas low divergence in values and practices related to future orientation results in “positive” behaviors and hinders corruption.
Cultural Discrepancies and Corruption
Low cultural discrepancies and pervasive corruption. We proposed that for four of nine cultural dimensions, a low discrepancy would result in pervasive corruption because mutual favor doing is both accepted and widely used. Concerning H1, results show that low discrepancy in in-group collectivism fosters corruption (r = -0.63). An examination of the GLOBE data reveals that in-group collectivism values are well above the midpoint of the 7-point scale and that practices lag behind (House et al., Reference House, Hanges, Ruiz-Quintanilla, Dorfman, Javidan, Dickson and Gupta2004). The concurrence of values and practices at a high level implies that mutual favor doing receives both value-based and behavioral legitimacy (Levi et al., Reference Levi, Sacks and Tyler2009), and thus, fosters pervasive corruption as socially valid and fully institutionalized (Uhlenbruck et al., Reference Uhlenbruck, Rodriguez, Doh and Eden2006).
Prior research has shown that collectivism fosters corruption (Davis & Ruhe, Reference Davis and Ruhe2003; Seleim & Bontis, Reference Seleim and Bontis2009), but the results were non-significant in some studies (e.g., DiRienzo et al., Reference DiRienzo, Das, Cort and Burbridge2007). Our exploratory model comparisons shed light on these inconsistencies (see Table 4). When independently regressing corruption on cultural discrepancies, values, and practices—and controlling for national wealth and access to information—in-group collectivism values are only marginally significant (b = -1.09, p < 0.10), but the discrepancy is significant and negative (b = -1.03, p < 0.005), and the practices are significant and positive (b = 0.89, p < 0.05). Hence, the inconsistent prior results may stem from using values or practices as a predictor. While in-group collectivism values do not play a salient role, the practices and the discrepancy matter. At different wealth levels, the discrepancy even seems to be a better predictor: it is related to corruption in poor and medium countries; the practices are significant in medium countries only (see Table 5).
Unexpectedly, discrepancies in institutional collectivism (H2), power distance (H3), and assertiveness (H4), foster corruption, although the effect is nonsignificant for assertiveness. This result is particularly surprising for institutional collectivism, which is similar to in-group collectivism. Yet, the nature of these two dimensions seems to differ substantially. In-group collectivism mainly refers to family as a narrow social entity—an “in-group” (House et al., Reference House, Hanges, Ruiz-Quintanilla, Dorfman, Javidan, Dickson and Gupta2004). Blood-related mutual obligations are closely linked with nepotism, as a particular, widespread form of corruption among family members (Beets, Reference Beets2005). Institutional collectivism refers to the society and its institutions (House et al., Reference House, Hanges, Ruiz-Quintanilla, Dorfman, Javidan, Dickson and Gupta2004): it emphasizes the collective distribution of resources. Corruption may harm institutions (DiRienzo et al., Reference DiRienzo, Das, Cort and Burbridge2007) through public resource drain. Therefore, a low discrepancy in this dimension (i.e., collective distribution of resources is equally desirable and practiced) rather tends to prevent pervasive corruption.
Power distance refers to the acceptance of social inequality (House et al., Reference House, Hanges, Ruiz-Quintanilla, Dorfman, Javidan, Dickson and Gupta2004). The GLOBE results for this dimension reveal the lowest mean scores for values of all nine dimensions (2.75), implying that power distance is undesirable. Further, practices exceed the values across all countries (House et al., Reference House, Hanges, Ruiz-Quintanilla, Dorfman, Javidan, Dickson and Gupta2004). Hence, a low power distance discrepancy means that values and practices are both at a low level. Equal power distribution is desirable and practiced, which rather prevents pervasive corruption.
Similarly, assertiveness is the second of two GLOBE dimensions with a mean value score below the midpoint (mean = 3.81), and practices exceeding the values (House et al., Reference House, Hanges, Ruiz-Quintanilla, Dorfman, Javidan, Dickson and Gupta2004). Hence, a low discrepancy means that assertiveness is neither desirable nor practiced, which may not lead to pervasive corruption.
High cultural discrepancies and arbitrary corruption. The results support positive relationships between high cultural discrepancies and arbitrary corruption for uncertainty avoidance (H5), future orientation (H6), and performance orientation (H7). These dimensions describe adherence to explicit or implicit (work-related) rules, and high cultural discrepancies in them foster arbitrary corruption because such value-practice incongruence leaves room for varied interpretations of rules.
The effect is most salient for future orientation (r = +0.65), which reflects the degree of planning ahead. As the value scores are well above the scale midpoint, with practices lagging behind (House et al., Reference House, Hanges, Ruiz-Quintanilla, Dorfman, Javidan, Dickson and Gupta2004), a high discrepancy means that planning ahead is highly desirable, but not practiced (values > practices). Such a constellation reflects ambiguity, which invites people to capriciously interpret regulations, and thus, arbitrary corruption (Uhlenbruck et al., Reference Uhlenbruck, Rodriguez, Doh and Eden2006). In contrast, a “surplus” of future-orientation practices (values < practices) is not observed in the data, so arbitrary corruption that results from over-complex regulations does not play a role. These results enhance prior research that reports future orientation as non-significant in multivariate analyses (Seleim & Bontis, Reference Seleim and Bontis2009).
The positive relationship of uncertainty avoidance discrepancy with corruption is marginally supported (r = +0.25, p < 0.10). Uncertainty avoidance relates to rule-orientation (House et al., Reference House, Hanges, Ruiz-Quintanilla, Dorfman, Javidan, Dickson and Gupta2004), and the results support that an inconsistency in the aspired vs. practiced level of rule-orientation fosters arbitrary corruption, but the proposed link is weak. The correlation between values and practices for UA is negative and the highest for the nine dimensions. It is also the only dimension for which the discrepancy score is neither correlated with practices nor values. Clearly, this dimension warrants further investigation.
The positive relationship of corruption with performance orientation discrepancy is significant (r = +0.32, p < 0.05). Performance orientation describes a focus on visible results (House et al., Reference House, Hanges, Ruiz-Quintanilla, Dorfman, Javidan, Dickson and Gupta2004). Our findings support that a mismatch in the aspired vs. practiced degree of performance orientation increases arbitrary corruption. Again, these results may shed light on the inconsistent findings in prior studies, which report a nonsignificant (Seleim & Bontis, Reference Seleim and Bontis2009) vs. a positive effect on corruption (Martin et al., Reference Martin, Cullen, Johnson and Parbotheeah2007). Yet, the discrepancy’s effect is weak, which may be due to nature of this dimension. In fact, both low and high performance orientation reflect some sort of rule. Highly performance-oriented societies base rewards on visible results; less performance-oriented societies include criteria like loyalty and cooperative spirit (House et al., Reference House, Hanges, Ruiz-Quintanilla, Dorfman, Javidan, Dickson and Gupta2004). Thus, high discrepancies in this dimension may not only indicate a mismatch of explicitness, but of the type of practiced and aspired rules. This dimension warrants further examination.
Cultural discrepancies hypothesized as unrelated to corruption. Surprisingly, the discrepancies in gender egalitarianism (r = -0.33, p < 0.05) and humane orientation (r = -0.23, p < 0.10) are related to corruption. Yet, the negative relationships are weak and only marginally significant in the latter case. Given that prior multivariate research did not find effects for gender egalitarianism and contradictory results for humane orientation (Martin et al., Reference Martin, Cullen, Johnson and Parbotheeah2007; Seleim & Bontis, Reference Seleim and Bontis2009), these dimensions may be of minor importance for corruption.
The Role of National Wealth
Overview. Consistent with prior research, the correlation between national wealth and corruption is high, negative, and significant (r = -.78, p < .001). In poor countries, many people struggle to sustain a living and are motivated to improve their income through any possible means, including corruption (Beets, Reference Beets2005). Further, the relationship between the two salient cultural discrepancies—in-group collectivism and future orientation—and corruption is contingent on national wealth (Table 3). Specifically, there seem to be different explanations for the cultural drivers of corruption in poor vs. rich countries. In-group collectivism discrepancy relates strongly to corruption in poor (r = -0.71, p < 0.01) and in moderately wealthy countries (r = -0.69, p < 0.01), but to a lower extent in rich countries (r = -0.48, p < 0.05). In contrast, the positive relationship of future orientation discrepancy and corruption is negligible in poor (r = 0.39, p > 0.10) and moderately wealthy countries (r = 0.09, p > 0.10), but becomes salient in rich countries (r = 0.85, p < 0.01).
A possible explanation of this finding lies in the value change hypothesis, which posits that rapid economic development and a subsequent increase in national wealth dramatically shifts a society’s value system. As income increases, societies move away from materialistic goals like physical and economic security to post-materialistic goals of autonomy and self-actualization (Inglehart, Reference Inglehart2008, Reference Inglehart1971). Transferred to our context, national wealth tends to affect the relationship between low in-group collectivism discrepancies and pervasive corruption and high future orientation discrepancies and arbitrary corruption.
In-group collectivism discrepancy and national wealth. Results suggest that pervasive corruption is salient in poor to medium countries, where low cultural discrepancies in in-group collectivism have a particularly pronounced relationship with corruption. In poor countries, income hardly suffices to support a family. Hence, people focus on survival and materialistic values (Inglehart & Baker, Reference Inglehart and Baker2000). If mutual favor doing and private benefit seeking are desirable and practiced in these societies, people may be particularly inclined to engage in corruption because it not only receives value-based and behavioral legitimacy, but it also improves their economic situation. Hence, corruption is “justified” from an economic standpoint as well. In rich societies, material values become less important (Inglehart, Reference Inglehart2008). Here, there is a lower economic justification for such a crime: even if mutual favor-doing and private benefit seeking receive value-based and behavioral legitimacy, corruption remains illegal. Accordingly, low in-group collectivism discrepancy is a strong driver of pervasive corruption in poor and medium countries, but the link is weak in rich countries.
Future orientation discrepancy and national wealth. The value change hypothesis may also explain why arbitrary corruption prevails in rich countries, driven by a high future orientation discrepancy. A high discrepancy in this dimension indicates that planning ahead is more desirable than practiced (values > practices) or vice versa (values < practices). Both situations comprise high levels of ambiguity because rules are either too fuzzy or too complex. Thus, people may feel entitled to interpret rules capriciously, which fosters arbitrary corruption (Rodriguez et al., Reference Rodriguez, Uhlenbruck and Eden2005).
People in rich countries may feel particularly inclined to interpret ambiguous rules in their own favor as they emphasize post-materialistic values like self-actualization (Inglehart & Baker, 2008). Further, post-materialism describes the ideal of a free individual (Inglehart, Reference Inglehart2008). Hence, citizens of affluent societies may feel frustrated with the vagueness or complexity of laws and regulations and use this ambiguity as an invitation to take things into their own hands. In poor countries, ambiguous laws may not prompt favorable law interpretations because such behavior requires a high degree of freedom, which is less important when survival is precarious (Inglehart, Reference Inglehart2008). Even in moderately wealthy countries, people’s behavior may not (yet) be driven by post-materialistic values, as indicated by the non-significant relationship of future orientation discrepancy with corruption.
Interestingly, moderately wealthy countries, though often being in transition, seem to be more similar to poor than to rich countries. The negative effect of in-group collectivism is almost as high in moderately wealthy countries as in poor countries; future orientation discrepancy has no effect. Obviously, the economic justification for illegal tit-for-tat remains pervasive at a moderate—and even at a high—wealth level, whereas the reason for arbitrary corruption when future plans are lacking is salient in affluent economies only.
Cultural Values, Practices, and Discrepancies
Comparing the predictive power of cultural values, practices, and discrepancies for in-group collectivism and future orientation shows that each of these three aspects helps to explain corruption beyond the control variables (Table 4). Yet, the additional contribution of cultural values is marginal: the change in R-square is minor, and only in-group collectivism values have an—albeit marginal—effect. Further, the discrepancies and the practices make a comparably strong contribution, as indicated by similar and significant changes in R-square and—importantly for non-nested model comparison—by an equally low AIC value. At the variable level, in-group collectivism has a negative effect on corruption for the discrepancy and a positive effect for the practice, while the opposite holds true for future orientation.
Further insights can be gained from the partial correlation analyses across different wealth levels (Table 5). Overall, the role of cultural values is negligible. However, both, discrepancies and practices explain corruption in moderately wealthy (in-group collectivism) and rich (future orientation) countries. Note that the discrepancy in in-group collectivism is the only cultural variable significantly related to corruption in poor countries, while the practices and values are not relevant. This result sheds light on the findings of prior research that values or practices may not sufficiently explain corruption in poor countries (Husted, Reference Husted1999). We find that cultural discrepancy really makes a contribution here.
In conclusion, cultural discrepancy has a distinct phenomenology that may, better than values (and practices), explain corruption in poor vs. wealthy nations. Further, our results suggest that research on culture and corruption needs not only clearly conceptualize and operationalize culture as discrepancy, values, or practices, but must also interpret culture in the context of a country’s wealth.
Practical Implications for Business Ethics
In understanding the cultural factors that drive poor and rich countries to be corrupt, business managers and other decision makers may develop adequate measures to prevent corruption, depending on a country’s wealth level. In poor countries, where corruption is pervasive and driven by a double legitimacy of mutual favor-doing (low in-group collectivism discrepancy), decision makers may establish mechanisms that allow fulfilling the norm of reciprocity by other than corrupt practices. For instance, the government as well as organizations could restrict the monetary value of gifts that public officials, purchasing managers, and others are allowed to accept, rather than generally forbidding gift giving and taking. The gift then has a symbolic meaning rather than a monetary value; it will fulfill the function of relationship building. In rich countries, where corruption is arbitrary and driven by a perceived lack of future plans (high future orientation discrepancy), decision makers could reduce corruption by clearly communicating their future plans. Rather than making ambiguous statements, politicians should openly indicate their goals, along with a measureable outcome, public officials should ensure planning reliability, and organizations should provide binding agreements to their stakeholders. As a result, they will be perceived as a reliable partner who aids people in planning ahead.
LIMITATIONS AND FURTHER RESEARCH
We recognize three limitations to our study. First, we measured cultural variables at an aggregate level. Given that heterogeneity within a country may sometimes be greater than between countries (Samiee & Jeong, Reference Samiee and Jeong1994), averaging individual scores means to generalize the stereotypical culture of a society to all members of this society. Future research may operationalize culture at the individual level (Sharma, Reference Sharma2010) and aggregate the findings through hierarchical linear modeling (Martin et al., Reference Martin, Cullen, Johnson and Parbotheeah2007).
Second, we used secondary data in the context of an exploratory study and thus, were not able to distinguish between pervasive and arbitrary corruption at the measurement level. Further research may better acknowledge the different phenomenologies comprised in the two corruption types and operationalize both constructs separately.
Third, the sample of 54 countries is small; correlations need to be rather high to reach significance. This is a general problem in country-level research, and our sample size is similar to pertinent studies that examined between 13 and 64 countries (Davis & Ruhe; Reference Davis and Ruhe2003; Dwyer, Mesak, & Hsu, Reference Dwyer, Mesak and Hsu2005; Getz & Volkema, Reference Getz and Volkema2001; Jing & Graham, Reference Jing and Graham2008; Husted, Reference Husted1999; Kanagaretnam, Lim, & Lobo, Reference Kanagaretnam, Lim and Lobo2011; Peng & Lin, Reference Peng and Lin2009; Treismann, Reference Treisman2000). An individual-level study may not only solve this problem, but also avoid a potential reversed causality with regard to culture and corruption. For example, one could manipulate cultural values and practices (e.g., “Imagine, in your society, …”) in a scenario experiment and measure corrupt behavior as the dependent variable. Yet, scenario experiments only measure what subjects report they would do in a particular situation. As corruption is socially undesirable, subjects may refrain from indicating the respective behavior. Although researchers can use specific methods such as a matrix task to detect lying behavior (Mazar, Amir, & Ariely, Reference Mazar, Amir and Ariely2008), this problem limits the ability to examine corruption in scenario experiments.
