October 30, 2017 | Author: Anonymous | Category: N/A
Yue, & Zhao, 2011; Zheng, Ghoul, Guedhami, & Kwok, 2012), and dividend .. According to Hofstede ......
Collectivism and Corruption in Bank Lending*
Xiaolan Zheng Moore School of Business University of South Carolina Columbia, SC 29208 USA Tel: (803) 800-7342 Fax: (803) 777-3609 E-mail:
[email protected] Sadok El Ghoul Campus Saint-Jean University of Alberta Edmonton, AB T6C 4G9 Canada Tel: (780) 465-8725 Fax: (780) 465-8760 Email:
[email protected] Omrane Guedhami Moore School of Business University of South Carolina Columbia, SC 29208 USA Tel: (803) 777-2175 Fax: (803) 777-3609 E-mail:
[email protected] Chuck C.Y. Kwok Moore School of Business University of South Carolina Columbia, SC 29208 USA Tel: (803) 777-3606 Fax: (803) 777-3609 E-mail:
[email protected]
* We would like to thank Najah Attig, Allen Berger and Narjess Boubakri for their constructive comments. We acknowledge financial support from Canada’s Social Sciences and Humanities Research Council (SSHRC) and the Center for International Business Education and Research (CIBER) at the University of South Carolina for this research project.
Collectivism and Corruption in Bank Lending
ABSTRACT This paper examines how national culture influences corruption in bank lending. Using a sample covering 3,835 firms across 38 countries, we find strong evidence that firms domiciled in collectivist countries perceive a higher level of lending corruption than firms domiciled in individualist countries. This positive link between collectivism and bank corruption is stronger in small and medium firms, privately owned firms, and non-export firms, while it is weaker in countries with stronger uncertainty avoidance, a higher (lower) fraction of foreign-owned (government-owned) banks, a higher number of large banks, and stronger legal institutions.
Keywords: Banking and Finance; National Culture; Corruption; Fraud
1
INTRODUCTION Substantial prior research emphasizes the importance of the banking system in promoting economic growth (e.g., Demirgüç-Kunt & Levine, 2008; Levine, 1997; 2005). A well-functioning banking system facilitates efficient information production about investment opportunities and directs credit to the most productive uses, resulting in a better allocation of capital and faster growth.1 Moreover, it contributes to future economic growth by closely monitoring the use of capital and improving corporate governance (Levine, 1997). However, a major impediment to an efficient banking system is corruption in bank lending to firms. Adapting Macrae’s (1982) definition of corruption to the context of bank lending, we consider lending corruption as an arrangement between two parties (a loan officer and a loan applicant) whereby the loan officer has influence over the final terms of a loan and abuses his or her responsibility for private ends (monetary and/or non-monetary compensation either immediately or in the future). Prior research shows that fraudulent lending disproportionately reduces unconnected small and medium firms’ access to bank loans, which forces them to forgo profitable investment opportunities and hence reduces economic growth (Beck, Demirgüç-Kunt, & Maksimovic, 2005).2 Corrupt lending can also lead to a fragile financial system that is vulnerable to financial crises as it increases banks’ exposure to high-risk lending while not increasing collateral requirements or interest rates to levels sufficient to compensate for the associated increase in risk (Charumilind, Kali, & Wiwattanakantang, 2006; La Porta, Lopez-De-Silanes, & Zamarripa, 2003). Bank corruption further weakens the role of banks in borrowing firms’ corporate governance.3 Despite its serious consequences, corruption in bank lending has been largely overlooked by the literature with the exception of three recent studies that use firm-level data from the World Business Environment Survey (WBES) to investigate the determinants of corruption in bank lending (Barth, Lin, Lin, & Song, 2009; Beck, Demirgüç-Kunt, & Levine, 2006; Houston, Lin, & Ma, 2011). These studies focus on the role of bank competition and various monitoring institutions at the country level. In this study, we extend 2
this emerging literature by examining the influence of national culture on cross-country differences in banks’ lending corruption. Culture can play an important role in influencing corruption in bank lending. Hofstede (2001) defines culture as “the collective programming of the mind which distinguishes the members of one group or category of people from another”, where programming of the mind refers to prescribed ways of thinking, feeling, and acting. Culture influences the way the mind encodes and interprets information as well as the behaviors of human actors (Lonner & Adamopoulos, 1997), placing informal constraints on human interactions (North, 1990). In the context of market activities, culture is viewed as the most fundamental force shaping the incentives of human actors (Williamson, 2000). Indeed, a substantial literature highlights the importance of culture to a country’s saving rates and fiscal redistribution policies (Guiso, Sapienza, & Zingales, 2006), financial system (Kwok & Tadesse, 2006), legal institutions (Licht, Goldschmidt, & Schwartz, 2005; Stulz & Williamson, 2003), life insurance consumption (Chui & Kwok, 2008), momentum profits (Chui, Titman, & Wei, 2010), merger outcomes (Ahern, Daminelli, & Fracassi, forthcoming), cross-border investment activity (Siegel, Licht, & Schwartz, 2011), corporate capital structure and debt maturity choices (Chui, Lloyd, & Kwok, 2002; Li, Griffin, Yue, & Zhao, 2011; Zheng, Ghoul, Guedhami, & Kwok, 2012), and dividend policy (Shao, Kwok, & Guedhami, 2010). Motivated by these studies, and the insight that problems in banks and bank regulation are “surrounded by the entire apparatus of political, legal, cultural and technological forces influencing the operation of banks” (Barth, Caprio, & Levine, 2006, p.7), we argue that culture influences individuals’ attitudes and perceptions towards ethical decision-making (Husted & Allen, 2008; Vitell, Nwachukwu, & Barnes, 1993) and hence the extent to which bank officers engage in corrupt behaviors. To relate national culture to corruption in bank lending, we rely on Hofstede’s (2001) four cultural dimensions, with a focus on the individualism/collectivism syndrome, which prior work argues is the most significant driver of cultural differences across societies (Markus & Kitayama, 1991; Triandis, 2001). Following prior studies (Barth et al., 2009; Beck et al., 2006; Houston et al., 2011), we employ survey data from WBES (World Bank, 2000) to evaluate corruption in bank lending for a sample of 3,835
3
firms across 38 countries. Our main results suggest that firms in more collectivist countries perceive a higher level of corruption in bank lending. This finding is robust to accounting for the role of the government in the economy, political connectedness, alternative dependent variables, endogeneity and potential omitted variables, alternative estimation methods, and alternative explanations. Moreover, this finding persists when, following prior studies, we control for the effects of bank supervision, bank competition, information sharing, and state ownership of the media on the integrity of bank lending: importantly, in comparison to these factors, the cultural dimension (collectivism) is economically the most important. Finally, in additional tests we find that small/medium, private, and non-exporting firms are more vulnerable to lending corruption associated with collectivism, while high uncertainty avoidance, a large proportion of private and foreign banks, more large banks, and better legal institutions help mitigate the adverse effect of collectivism on the integrity of bank lending. Taken together, our results suggest that culture exerts a first-order effect on corruption in bank lending, and provide insights into the firm- and country-level characteristics that can mitigate corrupt lending practices. Given the stickiness of national culture, our results therefore have important policy implications. Our study contributes to the existing literature in four ways. First, our research adds to the extant literature on the determinants of corruption in bank lending (Barth et al., 2009; Beck et al., 2006; Houston et al., 2011) by identifying the importance of national culture in influencing the integrity of bank lending. Second, while numerous studies examine economy-wide corruption, few examine corruption in the banking industry. In this paper, we focus on the context of bank lending and provide novel insights into how culture shapes the incentives for corruption in the loan screening and monitoring processes. Third, while several studies recognize the role of culture in influencing corruption perceived at the country level (e.g., Getz & Volkema, 2001; Husted, 1999; Husted, 2002; Jing & Graham, 2008; Seleim & Bontis, 2009), we focus on firm-level measures of corruption, which allows us to explore how culture interacts with firm characteristics in shaping the integrity of bank lending. Finally, we contribute to the recent literature on culture and finance (e.g., Ahern et al., forthcoming; Chui et al., 2010; Li et al., 2011; Li,
4
Griffin, Yue, & Zhao, 2012; Shao et al., 2010; Siegel et al., 2011) by showing that culture is an important determinant of the integrity of bank lending. The remainder of the paper is organized as follows. The next section reviews the existing literature and develops our main hypothesis. The third section discusses the sample and empirical methodology. The fourth to seventh sections present the empirical results. The last section concludes and discusses policy implications.
HYPOTHESIS DEVELOPMENT Literature Review A growing body of research suggests that the banking sector exerts a first-order effect on economic growth (Demirgüç-Kunt & Levine, 2008; Levine, 1997; 2005). However, in the presence of corruption in bank lending, risk-return analysis may not be the primary criterion in credit allocation decisions, in which case the efficiency of the banking system, and in turn economic growth, can be adversely affected. We adapt Macrae’s (1982) definition of corruption to the context of bank lending. In particular, we view lending corruption as an arrangement between two parties (a loan officer and a loan applicant) whereby the loan officer has influence over the final loan terms and abuses his or her responsibility for private ends (monetary and/or non-monetary compensation immediately or in the future). According to Barth et al. (2009), a lending officer weighs the benefits from engaging in lending corruption against the costs of being caught and punished, while a loan applicant weighs the benefits from better loan terms (e.g., approval of application, lower interest rate, longer maturity, less collateral) against the costs of bribes. Although the same type of cost-benefit analysis determines the likelihood and amount of lending corruption in every country, we argue that differences in institutional environment could shape loan officers’ and loan applicants’ incentives differently and thus contribute to variation in lending corruption across countries. A new but growing literature explores the country-level factors that influence the integrity of bank
5
lending. Using firm-level data from WBES, Beck et al. (2006) examine the relationship between bank supervisory policies and the extent to which corruption in lending constrains the ability of firms to raise external finance. They find no evidence that powerful supervisory agencies improve banks’ corporate governance and reduce corruption in bank lending, suggesting that the traditional approach of strengthening official supervision falls short of its objective of improving the integrity of bank lending. In contrast, they find strong evidence that strengthening private monitoring of banks by requiring banks to disclose accurate information to the public reduces corruption in bank lending, improving firms’ access to external capital. In an extension of Beck et al. (2006), Barth et al. (2009) show that greater bank competition and information sharing help curb lending corruption. More recently, Houston et al. (2011) show that monitoring by the media, especially by private and competitive media, increases the likelihood of fraudulent activity being caught and punished and hence reduces bank officers’ incentives to engage in such activity. In this paper we examine whether national culture influences corruption in bank lending over and above the determinants related to the regulatory environment, market structure, information sharing, and media characteristics. Prior literature suggests that culture is one of the most important determinants of human behavior. Hofstede (2001) argues that culture includes patterned ways of thinking, feeling, and reacting, the essential core of which are value systems that are ingrained in one’s mind during early socialization and that exert a life-long influence on one’s behavior. In addition to constraining certain behaviors, culture may condition human actors’ motivations to be compatible with social norms and value systems (Licht et al., 2005). Culture may therefore affect one’s perceptions towards unethical decision-making (Husted & Allen, 2008; Vitell et al., 1993). Based on this literature, we expect cross-country differences in culture to contribute to different attitudes and behaviors related to corruption in bank lending. This conjecture is consistent with Barth et al.’s (2006) argument that culture is one of the key factors influencing problems associated with bank operations and regulations.
6
Insights into Bank Lending To help develop our hypothesis on the link between culture and corruption in bank lending, we begin by reviewing the existing literature on bank lending decisions. According to this literature, commercial loan production has two important components, namely, loan screening and loan monitoring. Loan screening refers to the process whereby a loan officer collects information on a firm, conducts credit analysis, negotiates the terms of a loan (e.g., interest rate and maturity) with the firm, and seeks approval from the bank to extend the loan. After a loan application is received, the loan officer collects detailed information on the applicant. The finance literature typically classifies such information into two categories: soft (subjective) information and hard (objective) information (e.g., Demski, 1994; Liberti & Mian, 2009; Stein, 2002). In bank lending, soft information is based on the loan officer’s personal assessment of firm quality – this information, which includes the officer’s evaluation of the firm’s competitive position and managerial team, cannot be easily verified by anyone other than the loan officer who produces it. In contrast, hard information, which includes the value of collateral, taxable income, and other accounting data from financial statements, is relatively easy to verify. Between the two types of information collected, soft information tends to account for a larger proportion for small firms than for large and better-established firms (Berger & Udell, 2002). After a loan officer has collected all the necessary information and filed a credit application, the application is submitted to those in the bank vested with approval authority. Such authority could rest with the loan officer if the requested credit is below some limit set by the bank, with high-level bank managers if the requested credit is above the given threshold, or with the staff of a credit analysis department that serves as a check and balance mechanism. Given the information asymmetry between a loan applicant and the bank about the applicant’s default risk, the soft information used in the lending process gives loan officers a significant amount of discretion over the final terms of the loan. This creates ample room for lending corruption. In addition to distorting the soft information used in loan screening, loan officers might falsify the hard information in order to extract rents in the form of bribes. 7
Turning to the monitoring process, Udell (1989) explains that after extending a loan, banks continue to conduct credit analysis and monitor loan performance to detect deterioration in loan quality on a timely basis. Internal risk-mitigating procedures such as loan reviews and loan officer rotation (Hertzberg, Liberti, & Paravisini, 2010; Udell, 1989) help moderate potential agency problems between a bank and its loan officers. Under corrupt lending, a loan officer takes bribes from low quality firms in exchange for loans with preferential terms. A corrupt loan officer is thus more likely to have a poor quality loan portfolio with high default rates and low recovery rates. A poor quality loan portfolio in and of itself may damage a loan officer’s career prospects, and indeed may lead to dismissal, even if a corrupt deal is not exposed. However, loan portfolios of poor quality are likely to draw close scrutiny from the parties that monitor the loans, and thus face a higher likelihood of corrupt deals being exposed. If bribery is detected and proven, the loan officer may face demotion, dismissal, a fine, or even imprisonment as noted by Barth et al. (2009). A loan officer must therefore trade off the benefits of bribery against the risk of being caught and punished by the bank and/or regulators. The banking literature shows that a commercial bank’s loan screening and monitoring decisions help produce valuable private information about borrowers and are important to fulfilling its role in optimizing capital allocation. 4 In this paper, we extend this literature by arguing that, to the extent that culture conditions the incentives of bank officers to engage in corruption in the loan screening and monitoring processes, it is likely to exert an influence on the integrity of bank lending. Collectivism and Corruption in Bank Lending In this paper our main analysis focuses on the link between corruption in bank lending and a specific dimension of culture, namely, collectivism. The importance of collectivism is emphasized by the crosscultural psychology literature. Triandis (2001) argues that individualism/collectivism is the most significant driver of cultural differences across countries. This cultural dimension deals with selfconstrual and the relationship between self and group (Hofstede, 2001; Triandis, 1995). According to social identity theory (Tajfel & Turner, 1985), an individual’s self-identity consists of a personal identity 8
and a social identity. The definition of the self depends on the inclusiveness of some particular domains, which could be represented by concentric circles (Brewer, 1991). At the center of the concentric circles is one’s personal identity based on personal attributes that distinguish one individual from another. From the center outwards, an individual also belongs to different social groups, contributing to the individual’s social identity. Chen, Peng, and Saparito (2002) argue that collectivists and individualists differ significantly in self-identities. An individualist views the self as an self-contained and autonomous entity comprising a unique configuration of personal attributes, whereas a collectivist sees the self as part of a social relationship (Markus & Kitayama, 1991). Chen et al. (2002) suggest that the boundary between the individual self and other individuals is more important for individualists, while the boundary between the one’s in-groups and other groups is more salient for collectivists. According to cross-cultural psychology research (Hofstede, 2001; Triandis, 2001), such a distinction in self-construal between individualists and collectivists leads to two different social patterns. In an individualist society, the ties between individuals are loose – everyone tends to take care of his or her own self and their immediate family only. An individualist emphasizes personal goals over those of a group or clan and tends to decide whether to maintain a relationship based on rational cost-benefit analysis. Such self-construal is exemplified in American and many Western European cultures. In contrast, a collectivist is integrated into different in-groups (for example, nuclear family, extended family, friends/associates, organization, city, and state) from birth. Within such a society, people align their personal goals with those of their in-groups, giving high priority to group interests and working hard to maintain harmony within one’s social environment. This interdependent view is more often found in Asian cultures, but can also be seen in African cultures, Latin American cultures, and many southern European cultures (Markus & Kitayama, 1991). In this study, we propose that collectivism could increase bank corruption through four potential (independent) channels, namely, through its effects on 1) bank officers’ incentives to engage in bribes, 2) the effectiveness of bribes, 3) the probability of being caught, and 4) the severity of punishment if caught.
9
First, we posit that in collectivist societies, bank officers have greater incentives to engage in bribes. In such societies, individuals are embedded in a network of relatives, friends, and associates and are expected to maintain allegiances with their social groups. This value system shapes loan officers’ incentives and constraints when facing a bribe. Prior research on individualism/collectivism finds that collectivists tend to be particularistic, applying rules and standards differently depending on whether their counterparts are in-groups or out-groups (Husted, 2002; Leung & Bond, 1984), while individualists tend to be universalists, applying the same standards and rules to all counterparts (Chen et al., 2002). When ingroup members apply for bank credit and offer bribes in exchange for a better loan terms, the moral legitimacy of preferential treatment towards in-group members provides collectivist loan officers greater incentives than their individualist counterparts to distort soft, or even hard, information to improve loan terms. In turn, these loan officers may think that it is appropriate to be rewarded with kickbacks. Even though the bank is also part of the social identity of a collectivist loan officer, it is more distant away from the center of concentric circles of self-identity than kinship and friends/associates. Following similar arguments as in Rose-Acherman (1999), we argue that in societies based on strong interpersonal relations, a loan officer’s loyalty to his or her closer in-groups (family, friends, associates) may be of much higher priority than his or her obligations as a bank employee. Moreover, the morally legitimate differential treatment of different in-groups in collectivist societies may encourage loan applicants to first turn to banks that their social network can reach. Such self-selection would further strengthen the influence of collectivism on corruption in bank lending. Second, we argue that in collectivist societies, corruption in bank lending is more effective. Due to the illegality of bribes, a loan officer faces punishment from the bank or the state if a corrupt transaction is exposed, making it important to the loan officer that the corrupt aspects of the deal remain concealed. Additionally, given that a bribe is a quid pro quo agreement that cannot be enforced by a court (RoseAcherman, 1999), after a bribe is paid there is no guarantee that a loan officer will meet his or her reciprocal obligations and deliver satisfactory loan terms. However, in a collectivist culture, these two
10
obstacles to corruption are less problematic. On the one hand, pre-existing interpersonal relationships based on family ties, friendship, or other bonds common in collectivist societies foster moral obligation (Chen et al., 2002) and mutual trust between the two parties of a bribe, reducing the loan officer’s concerns about the briber’s threat of exposure while also reducing the briber’s concerns about the loan officer’s failure to deliver. On the other hand, collectivist societies may impose informal enforcement of such quid pro quo agreements. As we mention above, persons embedded in interpersonal networks may care less about market efficiency if they expect relationships with people to help get things done in exchange for “gifts” or “tips” (Rose-Acherman, 1999). In such a society, if a loan officer takes the benefits of a bribe but does not deliver reasonable favors according to an implicit ex-ante agreement, the briber may spread the word about such anti-norm behavior within his or her social group, reducing the briber’s ability to accrue future benefits from the group. Third, we conjecture that in collectivist countries, corruption in lending is less likely to be detected. Loans extended under corrupt exchanges tend to underperform loans extended based on market criteria. During the loan review process, these low quality loans are more likely to surface. Since the poor performance of a loan officer’s loan portfolio is likely to draw close scrutiny from supervisory agents, corrupt loan officers face a higher risk of detection. However, this risk is likely to be lower for loan officers in collectivist cultures. In collectivist societies, avoiding confrontation with one’s colleagues is highly valued (Hofstede, 2001). In our context, collectivist associates in the credit analysis or loan review departments are likely to loosen standards within their discretions in the credit approval and review processes. For example, they may be more lenient when monitoring loan quality to avoid internal conflicts with loan officers who originate the loans. This prediction is consistent with the finding of Gómez, Kirkman, and Shapiro (2000) that collectivists are significantly more generous than individualists when evaluating in-group members. In the worst-case scenario, bank officers in different departments may even cooperate (collude) to facilitate preferential lending. Thus, in collectivist cultures, an emphasis on maintaining harmony with coworkers is likely to reduce the effectiveness of check and balance
11
mechanisms within the bank, increasing the bank’s risk of exposure to corrupt transactions. In addition, in collectivist societies the exchange of (valuable) gifts is a common way to maintain their relationships, especially with non-kin (Triandis, 1995). However, the line between gift-giving and bribes can be hard to distinguish (Rose-Acherman, 1999). A loan officer who accepts bribes in exchange for preferential loan terms could always argue that she merely received a gift from a friend, making corrupt transactions harder to detect. Finally, we argue that even if loan quality is found to be poor during the loan review process, a loan officer associated with poor performance may face less punishment in a collectivist country. As mentioned earlier, corrupt bank lending is more likely to result in a low quality loan portfolio than bank lending based on market criteria. In addition to increasing the chances of a corrupt deal being detected as we discuss above, a low quality loan portfolio damages a loan officer’s career prospects, and indeed may lead to dismissal, even if lending corruption is not exposed. This consequence is likely to be less severe in collectivist societies where the workplace itself is an in-group society. In particular, in collectivist societies the employer-employee relationship is viewed in moral terms, resembling a family relationship with mutual obligations (Hofstede, 2001). Hofstede (2001) argues that in such a society, poor employee performance is less likely to be a reason for dismissal since parents do not abandon their children. Instead, poor performance is more likely to be taken as a signal of low competence and hence lead to a reassignment of tasks. The discussion above might lead one to ask how a collectivist loan officer involved in many in-groups (extended family, friends, associates, and the bank) reconciles sometimes conflicting interests among different in-groups. To address this question, we build on social identity theory and argue that the boundaries of in-groups are not fixed but rather are relative concepts. Chen et al. (2002) claim that personal and social identities are hierarchically structured. Furthermore, according to Triandis, Bontempo, Villareal, Asai, and Lucca (1988), the control imposed by in-groups follows a series of concentric circles from family, friends/associates, organizations, to the nation. When facing a conflict of interest between
12
two groups, collectivists tend to identify in-group members following a pecking order from the closest ingroups (e.g., family) to the furthest (e.g., nation). Therefore, when members from a closer in-group apply for loans and offer bribes, collectivist loan officers prioritize these individuals over the bank and thus have greater incentive to accept bribes and offer more favorable loan terms in return than their individualist counterparts. Similarly, a collectivist loan reviewer’s attachment to associates dominates his or her loyalty to the bank, resulting in more generous performance evaluations. Even if corrupt behavior is detected through the check and balance mechanism, collectivist loan reviewers who are socialized in the same culture may think such behavior is appropriate and thus show greater leniency toward a deviant loan officer during the evaluation process. Similarly, bank loan officers are considered in-group members relative to unknown others outside the bank and more likely to be protected from poor performance in collectivist countries. Taken together, the arguments above suggest that in collectivist countries, loan screening and monitoring mechanisms are likely to be less effective. The cultural values of collectivist societies are associated with incentives, perceptions, and behaviors that lead loan officers and loan applicants to be more likely to engage in corrupt bank lending. This leads to the following hypothesis: Hypothesis 1a (H1a): A country’s level of collectivism is positively related to the level of corruption in bank lending to firms. On the other hand, it is possible that individualism promotes more opportunistic behavior and thus is associated with a higher level of lending corruption. Opportunism (“self-interest seeking with guile”) is one of the fundamental assumptions in transaction cost economics theory (Williamson, 1988). In economics, opportunism often involves manipulating and distorting information to create information asymmetries that facilitate expropriation. Prior studies (Chen et al., 2002; Sakalaki, Richardson, & Thepaut, 2007) suggest that individualists–who view themselves as autonomous entities, prioritize selfinterest ahead of a group’s interest, and are less morally obligated to others–tend to behave more opportunistically than collectivists. 5 In the context of bank lending, individualist loan officers, whose 13
involvement in relationships is calculative rather than moral (Hofstede, 1980), may be more likely to manipulate information and provide better loan terms to applicants willing to pay substantial bribes. This leads to the following alternative hypothesis: Hypothesis 1b (H1b): A country’s level of collectivism is negatively related to the level of corruption in bank lending to firms. Other Cultural Dimensions and Corruption in Bank Lending Following Chui and Kwok (2008), while we focus on the Hofstede (2001) cultural dimension that is most significant in explaining cultural differences, namely, individualism/collectivism, we also extend our analysis
to
the
other
three
dimensions
proposed
by
Hofstede:
uncertainty
avoidance,
masculinity/femininity, and power distance. Uncertainty avoidance. The uncertainty avoidance dimension captures the extent to which members of a culture feel threatened by uncertainty and ambiguity (Hofstede, 2001). An individual in societies with high uncertainty avoidance feels more anxiety when facing unpredictable situations. Taking bribes from loan applicants can bring substantial risk into a loan officer’s life, significantly damaging his or her future career if caught. Thus, when weighing the benefits from taking bribes against the probability of detection and punishment, loan officers in more uncertainty-averse societies are likely to feel more anxious about the related uncertainty and hence be less willing to take bribes. Accordingly, we hypothesize that: Hypothesis 2a (H2a): A country’s level of uncertainty avoidance is negatively related to the level of corruption in bank lending to firms. On the other hand, a lower tolerance for uncertainty may promote lending corruption in two ways. First, as argued by Hofstede (2001), people in societies with high uncertainty avoidance tend to create bureaucratic structures with well-established rules and procedures that delimit socially acceptable behaviors. However, more rules and procedures may encourage people to work through informal channels, such as bribery, to achieve personal goals (e.g., Getz & Volkema, 2001). Second, corruption can serve as a mechanism to reduce uncertainty (Husted, 1999). Following Husted’s (1999) argument, in uncertainty14
averse societies loan applicants may be more willing to offer bribes to secure a more certain result. Taken together, these arguments suggest that loan officers in high uncertainty-avoidance countries may accept more bribes than their counterparts in low uncertainty-avoidance countries. Our alternative hypothesis is thus: Hypothesis 2b (H2b): A country’s level of uncertainty avoidance is positively related to the level of corruption in bank lending to firms. Masculinity/femininity. According to Hofstede (2001), the masculinity/femininity dimension assesses the degree to which “male assertiveness” (e.g., to be ambitious, to strive for material success, to attach importance to making money) is a dominant value as opposed to “female nurturance”. As noted by Husted (1999) and Jing and Graham (2008), people in masculine countries focus more on material success and hence are less likely to perceive unethical practices as problematic and more likely to think that the ends justify the means, creating more room for corruption. Loan officers are thus more likely to engage in corruption in more masculine societies than in less masculine societies. Our third hypothesis is thus: Hypothesis 3 (H3): A country’s level of masculinity is positively related to the level of corruption in bank lending to firms. Power distance. Hofstede (2001) defines power distance as the extent to which the less powerful members of a country’s organizations expect power to be distributed unequally. This dimension captures how different societies deal with hierarchies and the corresponding inequality in their social lives. In high power-distance countries, people who have certain powers and privileges are expected to use their power to increase their wealth. As pointed out by Getz and Volkema (2001), large disparities in power and wealth, although more tolerated in high power-distance societies, are unlikely to be preferred by the underclass, creating latent conflict between the different classes. This could promote corruption in bank lending among both high-level bank managers and low-level loan officers. First, high-level bank managers may simply perceive acquired personal benefits through bribery as a perquisite of their position. 15
Second, low-level loan officers who suffer inequity in power and wealth while observing the behavior of high-level bank managers may seek opportunities to abuse their discretion over loan terms to improve their own living standard and social status. This leads to the following hypothesis: Hypothesis 4 (H4): A country’s level of power distance is positively related to the level of corruption in bank lending to firms.
DATA AND METHODOLOGY Sample To examine the impact of national culture on corruption in bank lending, we compile data from four sources: (1) WBES (World Bank, 2000), which we use to construct proxies for bank corruption and other firm-level characteristics; (2) Hofstede’s (2001) cultural indices, which we use to capture the extent of a country’s collectivism/individualism; (3) the World Development Indicators (World Bank, 2010), which we use to obtain proxies for a country’s level of economic development; and (4) Beck, Demirgüç-Kunt, and Levine (2000), which we use to form country-level financial development variables. The Appendix contains detailed definitions and sources of all variables used in this paper. WBES, which was conducted by the World Bank, is a cross-sectional survey of the managers of more than 10,000 firms from over 80 developed and developing countries. This database includes managers’ perceptions of the investment climate and obstacles to firm growth that are driven by economic policy, governance, regulations, infrastructure, finance, and public service quality. As noted in Beck et al. (2006), using WBES to study corruption in bank lending has several advantages. First, managers are directly asked to rate the degree to which corruption in lending represents a financing obstacle. These direct responses on bank corruption are likely to lead to less measurement error than proxies for financing obstacles inferred from other information. Second, the firms in the survey exhibit a great deal of heterogeneity in terms of size, ownership, industry membership, and country of origin. For instance, WBES contains a large proportion (around 80%) of small and medium-sized enterprises, which alleviates problems associated with unrepresentative samples encountered in many cross-country finance studies. 16
Third, WBES offers the opportunity to investigate the interaction effects between collectivism and firm and country characteristics on the integrity of bank lending. To test our hypothesis, we start with the full sample of 10,032 firm observations included in WBES. We exclude observations with missing values for the measure of corruption in bank lending, resulting in a loss of 1,957 firms. We also omit observations with missing values for the collectivism index (3,667 firms excluded) and with insufficient data for firm-level characteristics (334 firms excluded). Finally, since the survey was conducted between 1999 and 2000, we employ economic and financial development indicators measured as of 1999. This results in the exclusion of 239 observations with missing values for these country-level indicators in 1999. After applying the above filters, we obtain a sample of 3,835 firms from 38 countries. Variables Bank corruption. The dependent variable in our study is corruption in bank lending. We follow prior studies (Barth et al., 2009; Beck et al., 2006; Houston et al., 2011) by constructing our proxy for bank corruption (Bank Corruption) using responses of firm managers to the WBES survey question: “Is corruption of bank officials an obstacle for the operation and growth of your business?” This indicator takes the value of 1 if the answer is “no obstacle”; 2 if “minor obstacle”; 3 if “moderate obstacle”; and 4 if “major obstacle”. A higher value of this variable implies a greater perception of corruption in bank lending. Like Barth et al. (2009) and Beck et al. (2006), we believe that the WBES survey data are not biased in favor of finding results consistent with our hypotheses. First, even if managers from different cultural backgrounds perceive the same obstacles differently, such a bias will only make it harder for us to find, for example, a positive relationship between collectivism and corruption in lending. To illustrate, if giving valuable “gifts” or taking care of in-groups is viewed as appropriate in a collectivist society, managers from collectivist countries may perceive a lower level of lending corruption than managers from individualist countries. To the extent this is the case, we are less likely to find a positive relationship 17
between collectivism and corruption in bank lending. Second, to alleviate the concern that managers from different institutional environments view financing obstacles differently, we control for a number of country-specific characteristics in our regression models. To further mitigate this concern, in robustness tests we incorporate additional controls for managers’ perceptions of general financial obstacles and the overall level of corruption, which should capture firms’ idiosyncratic responses to the WBES survey questionnaire. Third, Beck et al. (2006) and a substantial body of work reviewed in Barth et al. (2009) show that firms’ responses to the survey questions on financing obstacles are robustly associated with various measurable outcomes including the efficiency of investment flows, industrial expansion, property rights, and access to banking services. Thus, the WBES data likely capture more than idiosyncratic differences in perceptions of lending corruption. In our sample, we find that while 62% of the survey firms indicate that bank corruption is not an obstacle to firm growth, 10% indicate that bank corruption is a major obstacle, 9% indicate that it is a moderate obstacle, and 19% indicate that it is a minor obstacle. Culture variables. Following prior research (e.g., Chui & Kwok, 2008; Chui et al., 2010; Li et al., 2012), we employ the four cultural dimensions constructed by Hofstede (2001) and discussed in Section 2: individualism/collectivism (IDV), uncertainty avoidance (UAI), masculinity/femininity (MAS), and power distance (PDI). Hofstede’s framework continues to be widely used and has arguably had the greatest influence among various cultural classifications in cross-cultural research (Kirkman, Lowe, & Gibson, 2006; Schwartz, 1994; Sivakumar & Nakata, 2001). Hofstede’s individualism index, our main cultural dimension, captures the extent to which individuals emphasize their own goals over those of their group, with a higher value indicating a higher degree of individualism. For ease of interpretation, we construct an index of collectivism (CLT) equal to 100 minus Hofstede’s individualism index (IDV). Higher values of this index imply greater collectivism. We later examine whether our results are sensitive to using alternative proxies for collectivism. The other three cultural variables are constructed as in Hofstede (2001).
18
Firm and country characteristics. To isolate the role of culture in influencing corruption in bank lending, we closely follow prior research (e.g., Barth et al., 2009; Beck et al., 2006) in controlling for a comprehensive set of firm- and country-level determinants of bank corruption. As Barth et al. (2009) point out, government- and foreign-owned firms may perceive less corruption in lending due to their bargaining power vis-à-vis banks. We therefore include two dummy variables to capture a firm’s ownership type. Government (Foreign) equals one if any government agency or state body (foreign company or individual) has an ownership stake in the firm, and zero otherwise. We expect to find a negative relationship between each ownership variable and bank corruption. We also include the natural logarithm of firm sales (Sales) to control for firm size and the natural logarithm of the number of a firm’s competitors (N_competitors) to control for the competitive environment. We expect smaller firms and firms operating in more competitive environments to face more bank corruption (Barth et al., 2009; Beck et al., 2006). In addition, larger firms may perceive a lower level of lending corruption because they have better access to bank credit or because they view the bribes paid as a symbolic amount. We next include a dummy variable for whether the firm is an exporter (Export). An exporter may face fewer obstacles caused by bank corruption in its home country if it has access to external finance abroad, as such financing would strengthen its bargaining position with domestic banks (Barth et al., 2009). Since the obstacles that bank corruption pose may vary across industries, we further control for whether the firm is in the service industry (Service) or the manufacturing sector (Manufacture). All firm-level variables are obtained from WBES. 6 Finally, following Beck et al. (2006), we incorporate three macroeconomic controls measured as of 1999, namely, a proxy for overall bank development, which we capture using the ratio of private credit by deposit money banks to GDP obtained from Beck et al. (2000) (Priv), a proxy for monetary instability, which we capture using a country’s inflation rate (Inflation), and GDP growth (GDP_growth), which we use to capture firms’ growth opportunities. The last two variables come from World Development Indicators (World Bank, 2010). We expect that more credit provided by the banking sector and faster economic growth reduce the financing obstacles that firms face due to bank corruption, while higher inflation rates are associated with less efficient capital allocation and reduced banking 19
activities (Boyd, Levine, & Smith, 2001), exacerbating the obstacles induced by bank corruption to firm growth. Summary Statistics Table 1 provides descriptive statistics for our main explanatory variables by country (Panel A) and for the full sample (Panel B). Panel A shows a fair distribution of the 3,835 sample firms across the 38 countries. In particular, corruption in bank lending is perceived differently across countries, ranging from an average response of 1.03 in Canada to an average of 3.0 in Thailand. Moreover, all firm- and country-specific traits exhibit a fair amount of variation across countries. Panel B shows that, on average, the level of bank corruption perceived by the firms in our sample is 1.67 (out of 4), with a standard deviation of 1. ***Insert Table 1 about here*** Preliminary Evidence In preliminary tests, we divide our sample into above- and below-median CLT subsamples. The histogram in Figure 1 depicts the perceptions of bank corruption across the two subsamples, with the dark (light) columns indicating the percentage of responses to each of the four categories of answers for the abovemedian (below-median) CLT subsample. As can be seen from the figure, in below-median CLT countries 71.66% of the firms report no obstacle caused by bank corruption while the same figure for firms in above-median CLT countries is only 51.55%. However, the percentage of firms in above-median CLT countries is higher than that in below-median CLT countries for every level of perceived corruption in bank lending. These preliminary findings provide initial support for our conjecture of a positive relationship between collectivism and corruption in bank lending. ***Insert Figure 1 about here*** Regression Model To investigate the impact of collectivism on corruption in bank lending, we assume that the latent level of corruption in bank lending,
,
, is determined by: 20
, , ,
,
_
,
,
,
,
_
,
,
(1)
where j and k index firms and countries, respectively. Unlike the continuous latent variable, the observed dependent variable, Bank Corruptionj,k, is a polychotomous variable with a natural order whose value depends on the value of
,
,
. More specifically, the dependent variable,
, in our study is a response to a survey question coded as an ordinal variable,
ranging from 1 (perceive no obstacle due to bank corruption) to 4 (perceive a major obstacle due to bank corruption). A firm manager classifies corruption in bank lending into four categories after comparing their perceived level of bank corruption,
,
with three threshold points, µ , µ , and µ
as follows:
,
1 if 2 if 3 if 4 if
,
, ,
.
(2)
,
Following Barth et al. (2009), Beck et al. (2006), and Houston et al. (2011), we estimate equation (1) as an ordered probit model using maximum likelihood, which estimates the threshold ( ) and coefficient ( ) parameters simultaneously. We report t-statistics based on heteroskedasticity-robust standard errors. Although the coefficient estimates give the correct sign and statistical significance for each independent variable, their magnitudes cannot be directly interpreted as marginal effects because of the nonlinear nature of the ordered probit model (Beck et al., 2006, p. 2142). Consequently, we follow the approach of Barth et al. (2009) and Beck et al. (2006) and discuss the magnitude of the marginal effects in an average firm in the empirical results below.
21
REGRESSION RESULTS Main Evidence In this section, we empirically analyze the influence of collectivism on corruption in bank lending. Table 2 presents the coefficients from ordered probit models estimated on our full sample of 3,835 firms from 38 countries.7 In Model 1, we regress bank corruption on collectivism (CLT) and both the firm-specific and macroeconomic controls discussed above. We find that collectivism (CLT) enters the regression with a positive and statistically significant coefficient at the 1% level. 8 This finding lends support to the prediction in H1a that firms in more collectivist countries are subject to more lending corruption, possibly due to increased incentives to engage in bribes, increased effectiveness of bribes, as well as lower probability of detection and lax punishment if caught.9 In the next set of regressions in Table 2 we consider whether our main evidence on the link between collectivism and bank corruption in Model 1 is robust to using alternative proxies for collectivism. First, in Model 2 we employ Schwartz’s (1994) conservatism dimension (CONS), which comprises values important to societies based on close-knit harmonious relations (Schwartz, 1994), and therefore is similar to the collectivism/individualism dimension of Hofstede.10 Second, in Model 3 we use the collectivism index of Tang and Koveos (2008) (CLT_TK), who update Hofstede’s index based on the changing economic environment within each country (averaged between 1990 and 1994).11 Third, in Model 4 we replace CLT with the in-group collectivism measure (In-group CLT) from the GLOBE study of 62 societies (House, Hanges, Javidan, Dorfman, & Gupta, 2004).12 Reinforcing the evidence in Model 1, Models 2 through 4 indicate that the three alternative proxies for collectivism load positively at the 1% level. ***Insert Table 2 about here***
22
Influence of Other Cultural Dimensions Although our primary focus in this study is on the role of collectivism, in Models 5 through 8 of Table 2 we examine whether Hofstede’s (2001) other cultural dimensions, namely, uncertainty avoidance (UAI), masculinity (MAS), and power distance (PDI), influence lending corruption. In Model 5, we find that uncertainty avoidance enters the regression with a negative and statistically significant coefficient at the 1% level. This finding, which is consistent with Hypothesis 2a, suggests that the fear of being caught and suffering damage to their career discourages loan officers in high uncertainty-avoidance countries from accepting bribes.13 Second, the intuition behind the prediction in Hypothesis 3 suggests that loan officers in masculine societies tend to have a higher tolerance for questionable transactions in the pursuit of material success. This conjecture receives statistically significant support in Model 6, which reports a positive and significant relationship between masculinity (MAS) and corruption in bank lending. Finally, we find that coefficient on PDI loads positively at the 1% level in Model 7, implying that firms in high power-distance countries perceive bank corruption as a bigger obstacle to firm growth. This evidence supports our fourth hypothesis, which posits that in high power-distance countries, where transferring power into wealth is expected, bank officers are more likely to engage in unethical practices. In Model 8, we run a horserace regression that incorporates all four cultural dimensions. Except for PDI, our previous findings for CLT, UAI, and MAS continue to hold.14 Importantly, we continue to find evidence at the 1% level supporting the prediction in H1a that bank lending corruption is increasing in collectivism even after we control for other dimensions of national culture. Influence of Institutional Factors We next examine whether our findings are sensitive to including controls for several institutional factors shown by prior literature to influence firm-level perceptions of lending corruption. In Model 9 of Table 2, we follow Beck et al. (2006) and add to the baseline regression in Model 1 the variables Official Supervisory Power and Private Monitoring Index to control for the authority of bank supervisors and the incentives of creditors to monitor bank activities, respectively. In line with the findings of Beck et al. 23
(2006), we find that Official Supervisory Power is positively associated with lending corruption while Private Monitoring Index loads with a negative and significant coefficient. In Model 10, we follow Barth et al. (2009) and add to our baseline regression proxies for bank concentration (Bank Concentration), which we set equal to the assets of the three largest banks as a share of the assets of all commercial banks obtained from Beck et al. (2000), and information sharing (Private Bureau Age), which we set to the age of the oldest private credit bureau in the country by the end of 1999 from Djankov, McLiesh, and Shleifer (2007). Consistent with Barth et al. (2009), we find that the coefficient on Bank Concentration is positive and statistically significant, suggesting that bank concentration increases corruption in bank lending, and the coefficient on Private Bureau Age is negative and significant, indicating that information sharing helps reduce bank corruption. In Model 11, we follow Houston et al. (2011) and include a measure of media ownership (State Ownership, Press), which is defined as the ratio of the circulation of state-owned newspapers to the circulation of the five largest daily newspapers in the country from Djankov, McLiesh, Nenova, and Shleifer (2003). Consistent with Houston et al. (2011), we find that state ownership of the media leads to more bank corruption. More importantly for our purposes, Models 9 through 11 continue to find a strong positive relationship between collectivism and bank lending corruption. This evidence is reinforced by the horserace regression in Model 12 in which we include all the institutional variables. In this regression, the coefficient on collectivism remains positive and significant at the 1% level. Economic Significance In Table 3 we follow Barth et al. (2009) and evaluate the economic magnitude of the effects of the key determinants by computing the change in the probability that a firm reports corruption in bank lending to be no obstacle, a minor obstacle, a moderate obstacle, and a major obstacle due to a change in the variable of interest for an average firm. The results reported in Panels A, B, and C are based on Models 9, 10, and 11 of Table 2, respectively. Panel A of Table 3 presents the results for the variables CLT, Official Supervisory Power, and Private Monitoring Index. We find that the economic effects of CLT are highly significant and larger than those of Official Supervisory Power and Private Monitoring Index. For
24
example, the estimates indicate that an increase in CLT from the minimum to maximum (from the level of collectivism in the U.S. to that in Guatemala) would lead to a 13% increase in the probability that a firm rates lending corruption as a major obstacle and a 32% decrease in the probability that a firm rates lending corruption as no obstacle to firm operation and growth. Moreover, if the level of collectivism increases by one standard deviation (from half a standard deviation below the mean to half a standard deviation above the mean, which is approximately from the level of collectivism in Poland to that in Peru), the probability that a firm rates lending corruption as a major obstacle to firm operation and growth would increase by 4% while the probability that a firm rates lending corruption as no obstacle would decrease by 9%. By comparison, a one standard deviation increase in Official Supervisory Power (Private Monitoring Index) would only lead to a 2.1% increase (1.5% decrease) in the probability that a firm rates bank corruption as a major obstacle and a 5.3% decrease (3.8% increase) in the probability that a firm rates bank corruption as not an obstacle to firm operation and growth. The marginal effect of CLT is thus almost double the marginal effects of Official Supervisory Power and Private Monitoring Index. Similarly, in Panel B we quantify the marginal effects of CLT, Bank Concentration, and Private Bureau Age on lending corruption, and in Panel C we quantify the marginal effects of CLT and State Ownership, Press. We find that in each case, CLT continues to exert the greatest effect on lending corruption compared to the other variables used in previous studies. For example, Panel B suggests that a one standard deviation increase in CLT would lead to a 4% increase in the likelihood that a firm rates bank corruption as a major obstacle to business. In comparison, if Bank Concentration (Private Bureau Age) increases by one stand deviation, the probability that a firm considers bank corruption a major obstacle would increase by 1% (decrease by 3%). Panel C shows that while a one standard deviation increase in CLT is associated with a 6% increase in the probability that a firm rates lending corruption as a major obstacle, a one standard deviation increase in State Ownership, Press is associated with only a 1% increase in such probability. ***Insert Table 3 about here*** 25
ROBUSTNESS CHECKS Does the Link between Collectivism and Bank Corruption Reflect the Role of Government in the Economy and Political Connections? A skeptical reader might argue that our findings merely reflect the role of the government in the economy or the effect of political connections on bank lending. Indeed, the “political/regulatory capture view” holds that politicians do not maximize social welfare but rather their own private benefits (Shleifer & Vishny, 1998). When a government exerts great power over a country’s economy, it creates more room for politicians to engage in rent-seeking activities. In particular, Beck et al. (2006) suggest that if supervisory agencies have the power to monitor and discipline banks, politicians and supervisors may exploit this power to force banks to channel capital to politically connected firms. Faccio (2010) and Khwaja and Mian (2005) provide direct evidence that politically connected firms underperform nonconnected firms but enjoy better access to external capital. ***Insert Table 4 about here*** To reduce concerns that our earlier evidence on the role of culture in bank corruption reflects the government’s involvement in the economy or political connections, in Table 4 we sequentially introduce four additional controls: Economic Freedom in Model 1, Size of Government in Model 2, State Control in Model 3, and Political Connection in Model 4. Economic Freedom is the Heritage Foundation’s Index of Economic Freedom for 1999, with higher values implying less state intervention in the economy. Size of Government, which also comes from Heritage Foundation for 1999, captures overall government expenditures, including consumption and transfers, with higher values indicating a smaller government. State Control, obtained from the Economist Intelligence Unit, assesses the extent to which an economy is owned/controlled by the government, with higher values indicating less control. Political Connection captures the prevalence of political connections within a country using Faccio’s (2006) data on political connections.15
26
In general, the results reported in Models 1 through 3 support the “political/regulatory capture” view. More specifically, we find that firms perceive more lending corruption in countries with more state intervention (i.e., less economic freedom), a larger government, and a higher level of government control. Moreover, we find that Political Connection enters Model 4 positively and significantly, suggesting that firms from countries in which political connections are widespread face greater financing obstacles caused by bank corruption. Most relevant to our study, we find that the coefficient on CLT is positive and significant at the 1% level, reducing the concern that our earlier findings reflect the role of the government in the economy or political connections. Alternative Dependent Variables Although we follow prior research in specifying our main dependent variable, we are concerned that it does not have a balanced distribution across the four categories of answers, which may invalidate the estimated coefficients. Further, a few outliers in one category could drive the results in an unexpected way. To help dispel these concerns, we follow Beck et al. (2006) and replace Bank Corruption with Bank Corruption Dummy, which equals one if the firm manager rates corruption in bank lending as a minor obstacle, moderate obstacle, or major obstacle, and zero otherwise. This creates a relatively balanced sample of responses for probit estimation. As shown in Models 1 and 2 of Table 5, our results for collectivism continue to hold in these firm- and country-level regressions, respectively. ***Insert Table 5 about here*** To alleviate the concern that these self-reported data reflect pure measurement error and not bank corruption, we employ an alternative measure to check the validity of our previous results. We replace Bank Corruption with Need for Special Connections, which captures responses to the question “Is the need for special connections with banks and financial institutions an obstacle for the operation and growth of your business?” Answers are coded 1 (no obstacle), 2 (minor obstacle), 3 (moderate obstacle), and 4 (major obstacle). The results, presented in Model 3 of Table 5, show that CLT enters the regression positively at the 1% significance level. This finding suggests a positive relationship between collectivism 27
and the need for special connections with banks and financial institutions to ensure access to financing, in line with our previous conclusions. Endogeneity and Potential Omitted Variables Bias We examine whether endogeneity arising from sample selection bias, reverse causality, and omitted variables bias may explain the evidence in Table 2. Our study may suffer from endogeneity caused by a non-random sample (Reeb, Sakakibara, & Mahmood, 2012). To examine the effect of collectivism on lending corruption, ideally we would randomly assign banks/firms to countries with different levels of collectivism. We would then observe and compare levels of lending corruption between collectivist and individualist countries to investigate the influence of collectivism on the integrity of bank lending. Unfortunately, conducting such an experiment is not feasible in our setting. Another source of concern is that banks operating in countries with distinct characteristics may also play an active role in shaping their institutional environments. However, Hofstede (2001) and Williamson (2000) argue that national culture is established over a long period and changes very slowly–on the order of centuries or millennia. Consequently, we consider reverse causality of less a concern in our study than in a pure country-level analysis as it is unlikely that individual firms’ perceptions of bank corruption influence the level of collectivism in a given country (Barth et al., 2009). Nevertheless, to address these concerns we use the instrumental variable (IV) probit framework. Based on theoretical arguments in cross-cultural research, we select the following instruments: GDP per capita in 1970, GDP_pc70, collected from the World Development Indicators (World Bank, 2010); Latitude, the absolute value of a country’s latitude scaled to take a value between zero and one, from La Porta, LopezDe-Silanes, Shleifer, and Vishny (1999); and Diseases, an overall index of the historical prevalence of nine diseases within different geopolitical regions around the world, from Murray and Schaller (2010). Hofstede
(2001)
identifies
wealth
and
geographic
latitude
as
important
predictors
for
individualism/collectivism. He explains that while poverty increases the need to rely on in-groups, as a country’s wealth grows people obtain more resources and become more individualistic. He further 28
suggests that in colder climates, people need to show more individual initiative in order to survive and thus tend to be more individualistic. In addition, Fincher, Thornhill, Murrray, and Schaller (2008) argue that collectivists are more wary of contact with out-group members (and/or strangers), and are less likely to eat unusual foods. Thus, collectivism serves an antipathogen defense function and is more likely to emerge in societies that historically suffered a greater prevalence of pathogen. The regression results are presented in Model 4 of Table 5. We perform two tests to assess the suitability of the selected instruments. First, we conduct tests of overidentifying restrictions, in which the null hypothesis is that the excluded instruments are uncorrelated with the error term. We fail to reject the null hypothesis (p-value = 0.397), indicating that these instruments are valid and influence bank corruption through their effect on collectivism. Second, we conduct F-tests of the excluded exogenous variables in the first-stage regression, in which the null hypothesis is that the coefficient estimates of these variables are jointly equal to zero. We reject this null hypothesis at the 1% level. All the instruments enter the first stage with statistical significance at the 1% level and jointly explain around 70% of the cross-country variation in collectivism. In the second stage shown in Model 4, we find that CLT continues to load positively at the 1% level, suggesting that endogeneity is not responsible for our previous evidence. Finally, we are concerned that we might have omitted a variable from our model that is correlated with both Bank Corruption and CLT, leading to a biased coefficient on CLT. We address the concern about potential correlated omitted variables by sequentially adding the following control variables: 1) General Financing Obstacle, which measures the degree to which a firm manager views financing as an obstacle to the firm’s operation and growth (from WBES), because responses to this question may capture managers’ idiosyncratic biases (e.g., pessimism) in the perception of their firms’ environment; 2) Overall Corruption, which assesses the level of general corruption a firm manager perceives (from WBES), because managers’ perception of corruption in bank lending may reflect their perception of the overall level of corruption in their economies; 3) Stock Market Cap, defined as the total market value of listed shares to GDP in 1999, Stock Market Turnover, defined as the ratio of the value of total shares traded to
29
average real market capitalization in 1999 (both from Beck et al., 2000), and Market Based System, defined as a dummy variable that takes the value of one if the country has a market-based financial system and zero otherwise (from Demirgüç-Kunt
& Levine, 2004), because better stock market
development may reduce a firm’s dependence on bank loans and thus influence the integrity of bank lending; 4) Government Bank Ownership and Foreign Bank Ownership, defined as the fraction of the banking system’s assets in banks in which government (foreign) ownership is 50% or greater (from Barth et al., 2004), since private and foreign ownership in the banking industry tend to stimulate competition and increase banks’ incentives to maintain a good reputation, which can help reduce corruption in bank lending; 5) Bank Crisis, a dummy variable taking the value of one if a country suffered a systemic baking crisis during the 1997-1999 period, and zero otherwise (Laeven & Valencia, 2010), and Sovereign Default, a dummy variable that equals one if a country defaulted on its sovereign debt during the 1990-1999 period, and zero otherwise (Benjamin & Wright, 2009), because bank crises and sovereign debt default may intensify the financing obstacles that firms face and thus lead firms in such countries to perceive corruption in bank lending to be more problematic; 6) Creditor Rights, which measures the rights of secured creditors in the bankruptcy process (Djankov et al., 2007), because strong creditor rights allow creditors to monitor banks more effectively and thus may help reduce corruption in bank lending; and 7) Rule of Law and Government Effectiveness (Kaufmann, Kraay, & Zoido-Lobatón, 1999), because the general institutional environment may also play a role in shaping the integrity of bank lending.16 The results, which are not tabulated but are available from the authors, indicate that these additional control variables are statistically significant and generally load with the expected sign as suggested by the prior literature. More importantly for our purposes, we find that across all regressions, CLT is positively and significantly associated with corruption in bank lending at the 1% level. Alternative Estimation Methods For comparison purposes with previous studies (Barth et al., 2009; Beck et al., 2006; Houston et al., 2011), in our main analysis we choose to use an ordered probit model. Here we evaluate whether our 30
results continue to hold when we employ alternative estimation methods. First, we consider multilevel modeling. Our data include 3,835 firms (the base-level observations) nested in 38 countries (the higherlevel observations). If the perceived levels of bank corruption are nested in the country, ignoring the multilevel nature of the data will lead to underestimated standard errors, which is particularly severe for coefficients on country-level predictors. To address this concern, we follow Li, Griffin, Yue, and Zhao (2012) and center every independent variable by its grand mean, which is the mean averaged across firms within each country and then averaged across countries. Next, for each grand mean-centered firm-level predictor, we decompose it into two components: country-level mean value (averaged within each country) and firm-level deviation from the country-level mean value (subtracting the within-country-level mean from the grand mean-centered firm-level variable). We then regress Bank Corruption on the grand meancentered country-level predictors, country mean-centered firm-level predictors (deviations between firmlevel predictors and their country means), and within-country means of the firm-level predictors. As shown in Model 5 of Table 5, the coefficient on CLT is positive and statistically significant at the 1% level, reinforcing our previous findings. In the rest of Table 5 we continue to find evidence supporting the positive link between collectivism and corruption in bank lending when we employ an ordinary least squares estimation in Model 6, a weighted probit regression in Model 7 in which the weights are equal to the inverse of the number of firm observations in each country, an ordered probit with standard errors clustered at the country level in Model 8, and an ordered logit estimation in Model 9.
ALTERNATIVE EXPLANATIONS Biased Response from Disgruntled Borrowers It is possible that disgruntled borrowers may unduly blame corruption in bank lending for the difficulties they face in obtaining bank loans. Such borrowers may therefore have biased perceptions of lending corruption. We conduct two tests to reduce concerns that our results are driven by disgruntled borrowers. First, we include three additional controls for the proportions of a firm’s financing (one year before the 31
survey) coming from local commercial banks, investment funds/special development finance (development banks), and foreign banks, which are obtained from WBES. Second, we remove firms that do not obtain financing from any of the above three sources. In results that are available from the authors, we continue to find positive and statistically significant coefficients on collectivism, suggesting that our main findings are not driven by disgruntled borrowers. Relationship Banking It is possible that loan officers in collectivist countries extend credit to borrowers in their social networks because they know these borrowers better and have private information about their unique business risks, management quality, and growth potential. This may lead firms outside such networks to incorrectly perceive relationship banking as an indication of lending corruption (Cull et al., 2011). If banks are more informed about borrowers in collectivist countries, we would expect bank loans originating in collectivist countries to be of better quality. We examine this implication using actual loan quality from BANKSCOPE. We use two proxies for loan quality: Non-performing Loans and Loan Loss Provisions, both deflated by lagged total loans. We obtain a sample of 68,810 (78,069) observations of commercial banks and bank-holding companies across 58 (62) countries over the 1996 to 2005 period. In addition to CLT, we control for bank- and country-level determinants of loan quality following Laeven and Majnoni (2003): 1) Bank_size, the natural logarithm of bank total assets in U.S. dollars; 2) Loan_growth, the growth rate of total loans; 3) Profitability, earnings before taxes and loan loss provisions, all divided by lagged assets; 4) the same set of country-level controls as in our main regression, including Priv, Inflation, and GDP_growth; and 5) year dummies. We winsorize all firm-level variables at the 1% level in each tail. We estimate random effect models and weighted least squares (WLS) models with weights equal to the inverse of the number of bank observations in each country. For the WLS models, we report robust standard errors adjusted for clustering at the bank level. The unreported results, which are available from the authors upon request, show that both non-performing loans and loan loss provisions are positively related to CLT at the 1% 32
level. This evidence is inconsistent with banks having better quality loans in high CLT countries. Rather, the results lend support to our intuition, which posits that the pervasiveness of corruption in bank lending is associated with poorer quality loans in collectivist countries.
INTERACTION WITH FIRM- AND COUNTRY-LEVEL CHARACTERISTICS We perform a series of split-sample tests to examine whether the impact of collectivism on corruption in bank lending varies according to firm- and country-level characteristics.17 The results help identify both the type of firms that are more vulnerable to the negative impact of collectivism on access to external finance and growth, and the country-level factors that can alleviate this problem. It should be noted that the magnitudes of the coefficient estimates from the ordered probit models cannot be directly compared across the subsamples. Indeed, the economic impact of CLT on corruption in bank lending depends on the coefficient of CLT and the average values of the control variables in each subsample. Hence, to facilitate comparison and interpretation, we report the magnitude of the effects of a one standard deviation increase in CLT on the perceived level of bank corruption for the average subsample firm in Panel A of Tables 6 and 7. For brevity, instead of reporting results for all four response categories, we only report the changes in the probability that a firm rates corruption in bank lending to be no obstacle (1) and a major obstacle (4) due to a one standard deviation increase in CLT. We report the coefficient estimates in Panel B of Tables 6 and 7. Table 6 presents the split-sample results using a set of firm characteristics. This set includes firm size, firm ownership, and whether a firm exports. ***Insert Table 6 about here*** Interaction with Firm-Level Characteristics Firm size. In Models 1 through 3, we explore how the effect of collectivism on perceived bank corruption differs across three firm size categories available from WBES: Small, Medium, and Large. In each subsample, the coefficient on CLT is positive and statistically significant at the 1% level. However, in Panel A we can see that the magnitude of the economic effects of CLT on perceived bank corruption
33
varies with firm size. A one standard deviation increase in CLT leads to a 7% (6%) increase in the probability that a small (medium) firm rates corruption in bank lending as a major obstacle. This percentage increase in probability is only 2% for large firms, which is more than three times smaller than that for small or medium firms. These findings suggest that holding other things equal, the impact of collectivism on perceived bank corruption is more severe for small and medium firms than for large firms. One potential explanation for this result is that small and medium firms are often informationally opaque in which case a loan officer’s screening and monitoring decisions are more likely to rely on soft information that is more difficult to verify (Berger & Udell, 2002). With more discretion left to loan officers, it is not surprising that small and medium firms are more vulnerable to lending corruption associated with collectivism. This evidence is consistent with recent studies showing that small and medium enterprises (SMEs) are more financially constrained and have less access to external finance than large firms (e.g., Beck & Demirgüç-Kunt, 2006; Beck et al., 2005; Beck, Demirgüç-Kunt, & Maksimovic, 2008). SMEs play a significant role in economic development since they account for a majority of firms in the private sector and a large proportion of employment. 18 The disproportionate adverse effect of national culture on SMEs underscores the importance of concerted efforts by international organizations and governments to foster SME financing, especially in more collectivist countries. Ownership structure. We explore how the effect of CLT on perceived corruption in lending differs across firm ownership types in Models 4 through 6. CLT consistently loads positive and significant at the 1% level in subsamples of government-owned (Government), foreign-owned (Foreign), and privateowned (Private) firms. However, private firms perceive more bank corruption resulting from collectivism than foreign-owned and government-owned firms. More specifically, the estimates in Panel A imply that a one standard deviation increase in CLT leads to a 6% increase in the probability that a private firm perceives bank corruption to be a major obstacle and a 4% increase in the probability that a governmentor foreign-owned firm perceives bank corruption to be a major obstacle. These results suggest that private firms are more vulnerable to bank corruption associated with collectivism, while government- and
34
foreign-owned firms experience less adverse effects of collectivism on bank corruption. These findings are also consistent with prior research. For example, stated-owned enterprises, which benefit from soft budget constraints and easy access to capital from state-owned banks and government subsidies (Megginson, 2005), are less likely to experience bank corruption associated with collectivism. Foreignowned firms are also likely to be less constrained since they enjoy alternative financing sources from parent companies and foreign capital markets. Exporting activities. We next separate the sample according to whether firms export and present the results in Models 7 and 8. Although CLT shows positive and statistically significant relations with bank corruption in both subsamples, export firms perceive less lending corruption resulting from collectivism than purely domestic firms. In particular, if CLT increases from half a standard deviation below the sample mean to half a standard deviation above the mean, the probability that an exporting firm rates bank corruption as a major obstacle to firm growth increases by 4% and the probability that an export firm rates bank corruption as no obstacle drops by 12%. In contrast, the corresponding probability changes for non-export firms are 6% and 14%, respectively. Thus, exporters are less affected by the impact of collectivism on lending corruption, possibly because they have alternative financing sources (e.g., foreign banks) or because they benefit from favorable government export-stimulation policies (e.g., access to export finance). Interaction with Country-Level Characteristics In Table 7, we turn to country characteristics and examine whether the impact of collectivism on lending corruption hinges on the level of uncertainty avoidance, the ownership structure of a country’s banking industry, the prevalence of large banks, and the soundness of the country’s legal institutions. We conduct the split using the median value of the conditioning variable for each subsample test.19 ***Insert Table 7 about here***
35
Uncertainty avoidance. In Table 2, we find that the level of lending corruption is lower in high uncertainty-avoidance countries, suggesting that greater uncertainty induced by corrupt lending is a more effective deterrent for loan officers in countries with strong uncertainty aversion. Motivated by this evidence, we split the sample by the median level of uncertainty avoidance of countries in our sample and present results in Models 1 and 2 of Table 7. We find that the coefficient on CLT exhibits a positive sign at the 1% level in both subsamples but has a smaller influence in countries with a high level of uncertainty avoidance. More specifically, a one standard deviation increase in CLT leads to a 3% increase in the probability that a firm rates bank corruption as a major obstacle to its operation in high UAI countries, compared with a 6% probability increase in low UAI countries. For example, among countries with above-median CLT, the average Bank Corruption is 2.47 for Indonesia, with UAI scored at 48 (approximately the 25th percentile of UAI), in comparison to 1.20 for Chile, with UAI scored at 86 (approximately the 75th percentile). These results suggest that a country’s uncertainty-avoidance sentiment could mitigate the adverse effect of collectivism on the integrity of bank lending. State- and foreign-bank participation in the banking sector. Next, we divide the sample by the level of government and foreign ownership of the banking industry and present the results in Models 3 to 4 and 5 to 6, respectively. Although the coefficients on CLT are positive and highly significant in all subsamples, we find that firms in countries with low government bank ownership and high foreign bank ownership tend to face a less adverse effect of collectivism on bank corruption. Specifically, in Models 3 and 4 our estimates imply that the increase in the probability that a firm rates bank corruption as a major obstacle due to a one standard deviation increase in CLT is 7% in countries with high government bank ownership but only 4% in countries with low government bank ownership. Models 5 and 6 show that if CLT increases by one standard deviation, the probability that a firm rates bank corruption as a major obstacle increases only by 3% in countries with high foreign bank ownership but by 6% in countries with low foreign bank ownership.
36
These findings suggest that private and foreign ownership of banks tends to reduce the undesirable influence of collectivism on lending corruption. This conclusion is related to the literature on bank ownership structure in several ways. First, government ownership of banks is inherently associated with weaker managerial incentives to maximize bank value, less intense monitoring, and more pronounced moral hazard problems due to the higher likelihood of a government bailout (Megginson, 2005). These features of state-owned banks could induce loan reviewers to value more harmony in the workplace and leave lending officers with little monitoring, reducing the chance of getting caught in the case of corrupt quid pro quo. Second, La Porta, Lopez-de-Silanes, and Shleifer (2002) provide evidence in support of the political view of government ownership of banks, which holds that politicians use government-owned banks to maximize employment and transfer favors to supporters in exchange for votes, political patronage, and bribes. Using bank loans in Italy, Sapienza (2004) further reports that identical firms borrow at lower interest rates from state-owned banks than from private banks and that the interest rates charged by state-owned banks are lower for firms located in areas where political patronage is more widespread. In more collectivist countries, where lending corruption tends to be more prevalent, politicians may be more inclined to seek bribes from firms in return for financing from governmentowned banks. Both lines of argument—misaligned managerial incentives and political motivations behind government bank ownership—provide potential explanations for why state ownership of banks exacerbates the positive relationship between collectivism and lending corruption. Levine (1996) argues that foreign banks could help enhance the availability and quality of financial services by increasing bank competition, pressuring domestic banks to adopt more modern banking technology, and spurring the government to improve legal, regulatory, and supervisory frameworks related to financial activities. High survival pressure on domestic banks and better financial institutions tend to increase the chance that corrupt transactions will be detected and punished if caught, reducing the positive effect of collectivism on lending corruption.
37
Taken together, our evidence stresses the importance of bank privatization and openness to foreign banks in reducing the adverse impact of collectivism on bank corruption and promoting a healthier banking sector. Bank size. Although information about banks from which our sample firms receive loans is not available, we construct a country-level measure of bank size to broadly capture the presence of large banks in a given country. Using a sample of commercial banks and bank holding companies collected from Bankscope, we identify the number of large banks in a country (Prevalence of Large Banks), where a bank is classified as a large bank if its total assets in 1999 exceed 1 billion USD (Berger & Black, 2011). We then split the sample according to the median level of Prevalence of Large Banks. The results are reported in Models 7 and 8. We find that the coefficient estimate on CLT is positive in both subsamples, but is less significant and smaller in magnitude in the subsample of countries with more large banks. The magnitude of the economic impact is quite different between the two subsamples: if CLT increases by one standard deviation in the subsample, the probability that a firm rates bank corruption as a major obstacle to firm growth increases by 3% for countries with more large banks compared to 8% for countries with fewer large banks. Our results are consistent with the argument that small banks have a comparative advantage in using soft information while large banks rely more on hard information since it is easier to verify and pass along the hierarchy (e.g., Berger, Miller, Petersen, Rajan, & Stein, 2005; Cole, Goldberg, & White, 2004). Given that loan officers have more discretion and opportunities to receive bribes in the presence of soft information, the influence of collectivism on bank corruption is likely to be more pronounced in small banks. Put differently, our results imply that hard information, which is used more by large banks, may help curtail bank corruption resulting from collectivism. From a policy perspective, our findings suggest that in collectivist countries, underweighting soft information and promoting the use of hard information in the lending process may help limit the discretion of loan officers and in turn lending corruption.
38
Legal institutions. Following Beck et al. (2006), we use two proxies for the soundness of legal institutions: Rule of Law and Government Effectiveness. We separate the sample into countries with strong versus weak institutions and find that CLT continues to load positively at the 1% level throughout all subsamples, although this impact is smaller in countries with well-functioning legal systems and more effective governments. Comparing the economic effects in Models 9 and 11 with those in Models 10 and 12, we find that a one standard deviation increase in CLT results in a more pronounced increase in the probability that a firm perceives bank corruption to be a major obstacle in countries with a weak Rule of Law and a low level of Government Effectiveness (6% and 5%, respectively) than in countries with a strong Rule of Law and better Government Effectiveness (2% and 1%, respectively). These results suggest that well-functioning legal institutions allow private investors to exert more effective corporate governance, thereby alleviating bank corruption stemming from collectivism. In sum, our split-sample tests indicate that the effect of collectivism on bank corruption tends to be more pronounced among small and medium firms, privately owned firms, and non-export firms. A high level of uncertainty avoidance, a high fraction of private- and foreign-owned banks, more large banks in the banking sector, as well as strong legal institutions help alleviate the adverse impact of collectivism. Given that culture is difficult to alter due to its stability over time (Hofstede, 2001), these findings have practical implications for policymakers of collectivist countries. In particular, they underscore the need to facilitate bank financing to small, private, and domestic firms as well as the importance of banking sector privatization and openness, hard information, and strong legal institutions in fighting lending corruption.
CONCLUSION AND POLICY IMPLICATIONS A large body of the finance literature argues that a country’s level of banking development has a significant impact on the country’s economic growth. However, corruption in bank lending as a major impediment to the efficient allocation of bank credit has been overlooked by the literature until recent studies by Barth et al. (2009), Beck et al. (2006), and Houston et al. (2011). We contribute to this new but growing thread of the literature by investigating how national culture affects cross-country differences in 39
bank corruption. Bank officers and loan applicants weigh the benefits of bribes against the associated costs. We argue that culture influences their perceptions towards unethical practices and in turn their costbenefit analysis, thereby contributing to variation in lending corruption across countries. More specifically, we argue that culture, particularly the collectivism/individualism dimension, influences lending corruption through its effect on the incentives to engage in bribes, the effectiveness of bribes, the probability of detection, and the severity of punishment if a corrupt deal surfaces. Our paper also adds to recent research on national culture and finance (e.g., Ahern et al., forthcoming; Chui et al., 2010; Li et al., 2011) by linking national culture to the integrity of bank lending, and complements studies of national culture and overall corruption at the country level (e.g., Husted, 1999; Jing & Graham, 2008) by focusing on the banking industry and employing firm-level data. Using a sample of 3,835 firm observations from 38 countries, we find that collectivism is consistently positively and significantly associated with lending corruption. This positive relationship persists when confronted with various robustness tests. In terms of economic magnitude, the effect of collectivism on lending corruption is greater than that of other factors identified in prior studies (including bank supervision, bank competition, information sharing, and media monitoring). In additional subsample analyses, we find that the negative effect of collectivism on lending integrity is more pronounced in small and medium-sized firms, private firms, and non-export firms. Moreover, we find that high uncertainty avoidance, private and foreign ownership of the banking sector, large banks in the economy, and a sound legal environment mitigate the undesirable effects of collectivism. Our findings have implications for policy makers who seek to improve their country’s financial development and economic growth. Our study demonstrates that corruption in bank lending is driven in part by cultural influences. Since culture remains stable, alleviating the adverse effects of corruption on a firm’s access to external financing and in turn on firm growth and overall financial development requires a policy response. Policy makers in collectivist countries should help alleviate financing constraints for small, private, and non-export firms – firms that are more sensitive to bank lending corruption resulting
40
from collectivism. Furthermore, privatizing the banking sector, liberalizing foreign bank entry, increasing the use of hard information, and improving legal institutions can help mitigate the aggravating effect of collectivism on the integrity of bank lending.
41
NOTES 1
Bank lending, as a primary source of external finance for firms, has received increased attention in recent literature
(e.g., Cull, Haber, & Imai, 2011; Li, Qiu, & Wan, 2011). 2
For example, Beck et al. (2005) show that corruption of bank officials constrains firm growth, with this effect more
pronounced among small firms. 3
For example, using data on Mexico’s bank loans, La Porta et al. (2003) show that, compared to unrelated
borrowers, related borrowers are 33 to 35% more likely to default, tend to borrow at a lower interest rate, and are less likely to post collateral. Further, Charumilind et al. (2006) find that connections are the most important factor explaining access to long-term credit in Thailand, dominating other firm characteristics identified in the firm financing literature. Several other studies suggest that cronyism is one of the most important factors contributing to the Asian financial crisis (e.g., Khatri, Tsang, & Begley, 2006). 4
The existence and quality of a bank relationship increase the market value of the borrower. For example, empirical
studies report that a borrower’s abnormal return associated with a bank loan announcement is positively related to the lender’s reputation and screening and monitoring abilities (Billett, Flannery, & Garfinkel, 1995; Johnson, 1997; Lee & Sharpe, 2009). 5
Chen et al. (2002) also point out that the relationship between individualism/collectivism and opportunistic
propensity relies on whether the counterparts are in-groups or out-groups. 6
We additionally incorporate Leverage, which is defined as 100 times the difference between one and the
percentage of a firm’s financing (one year before the survey) from retained earnings and equity. The fraction of financing from retained earnings and the fraction of financing from equity are from WBES. The results presented in Table 2 are not sensitive to the inclusion of Leverage, and the coefficient on Leverage is economically and statistically insignificant in general. Given the smaller sample size due to missing data for the leverage proxy and the reduced comparability with previous studies such as Beck et al. (2006), we exclude Leverage from our main models. The unreported results are available upon request. 7
We also employ the ordered logistic model to estimate the regressions in Table 2 and calculate proportional odds
ratios for each predictor. This method leads to the same conclusions as Table 2 regarding the relations between lending corruption and the measures of collectivism, uncertainty avoidance, masculinity, as well as power distance.
42
In particular, we find that the odds ratios for collectivism in Models 1 to 4 and Models 8 to 12 are greater than one, suggesting a positive relationship between bank corruption and collectivism that is consistent with our findings reported in Table 2. 8
In this model, the value of the goodness-of-fit measure, Pseudo R2, increases from 2% before the inclusion of CLT
(unreported regression) to 5% after its inclusion. Despite our efforts to include every known determinant of corruption in bank lending in our regressions, the pseudo R-squares of our models may appear relatively low. However, for the sake of comparison, we appraise our R-squares in light of the three related papers: Barth et al. (2009), Beck et al. (2006), and Houston et al. (2011), which use the same data and estimation method, and predict the same outcome. Except for Barth et al. (2009), who do not report pseudo R-squares, Beck et al. (2006) and Houston et al. (2011) report pseudo R-squares around 0.05-0.06 in their main regressions. Moreover, according to Beck et al. (2006), the pseudo R-square ranging from 5% to 9% is high for these types of cross-firm empirical studies. 9
The arguments in Chen et al. (2002) may shed light on why the prediction in the alternative hypothesis (H1b) is
rejected. They argue that the relationship between individualism/collectivism and opportunistic behaviors depends on whether the counterparts are in-groups or out-groups. In individualist societies, justifications for morally questionable means to achieve desirable ends primarily come from self-interests, while in collectivist societies, such justifications derive from self-interests, group interests, or both. The legitimacy of taking care of in-groups in collectivist societies makes opportunistic behaviors against out-groups morally less repugnant. Therefore, while an individualist loan officer may behave opportunistically towards loan applicants in general, a collectivist loan officer behaves more opportunistically towards loan applicants from his or her social network. 10
Schwartz (1994) identifies seven cultural value types–conservatism, hierarchy, mastery, affective autonomy,
intellectual autonomy, egalitarian commitment, and harmony–that can be condensed into two widely used dimensions (e.g., Shao et al., 2010), namely, conservatism and mastery. 11
Tang and Koveos (2008) regress the four cultural dimensions on the set of determinants they identified, including
average GDP per capita between 1970 and 1974. They then use the regression coefficients and insert the average GDP per capita between 1990 and 1994 to update Hofstede’s cultural scores.
43
12
House et al. (2004) developed two measures of collectivism. In addition to in-group collectivism, the other
measure is institutional collectivism, which reflects the degree to which institutional practices at the societal level encourage and reward collective distribution of resources and collective action. According to House, Javidan, Hanges, and Dorfman (2002), institutional collectivism measures societal emphasis on collectivism, with high scores indicating collectivistic emphasis by means of laws, social programs, or institutional practices. House et al.’s in-group collectivism is therefore closer to Hofstede’s collectivism index. Nevertheless, in a robustness check we find that our evidence on CLT is not sensitive to including as an additional control House et al.’s proxy for institutional collectivism. 13
Later in the paper, we extend our analysis by examining whether the level of uncertainty avoidance mitigates the
adverse effect of collectivism on bank corruption. 14
One potential explanation for the unexpected sign on PDI could be the high correlation between PDI and CLT
(0.5). Indeed, when we exclude CLT from the regression, the coefficient on PDI becomes positive and significant. 15
According to Faccio (2006, p. 369), a firm is considered politically connected if “at least one of its large
shareholders (anyone controlling at least 10 percent of voting shares) or one of its top officers (CEO, president, vicepresident, chairman, or secretary) is a member of parliament, a minister, or is closely related to a top politician or party.” There are some limitations of using Faccio’s (2006) data in our study. First, as recognized by Faccio (2006), these data underestimate the prevalence of political connections. Second, we consider political connections that existed at the time when the data were collected, but information on the dates when these connections were established is not available. Third, while Faccio’s sample includes only publicly traded firms, 80% of firms surveyed by WBES are small or medium firms. 16
A clawback provision to executive compensation could change the incentives of bank loan officers. For example,
Allen and Li (2011) examine the effect on bank lending practices of a lifetime responsibility program implemented in China in 1998 that instates clawbacks on bank lending agents if the loans they originated defaulted or performed poorly. They find that the implementation of this clawback provision reduces bank lending based on political connections but increases lending relying on economic fundamentals. Moreover, in the United States, a clawback provision that applies to publicly-traded companies is included by the Sarbanes-Oxley Act in 2002, but only become implementable and influential after the Dodd-Frank Act in 2010. According to Fried and Shilon (2011), on the eve
44
of Dodd-Frank, fewer than 2% of S&P 500 firms had robust clawback policies. However, to our best knowledge, information about the clawback provisions for executive compensation across countries is not available. While China is not a concern to our study since it is not in our sample, we find that our core evidence on the effect of collectivism holds when we exclude firms from the United States. 17
In a linear model, adding an interaction term to our original regression is an alternative approach. However, in our
nonlinear ordered probit model, “the marginal effect of a change in both interacted variables is not equal to the marginal effect of changing just the interaction term. More surprisingly, the sign may be different for different observations” (Norton, Wang, & Ai, 2004, p. 154). Therefore, we choose the split-sample technique to conduct our analysis. 18
For instance, data compiled by Ayyagari, Beck, and Demirgüç-Kunt (2007) for 76 countries indicate that SMEs
account, on average, for 54% of manufacturing employment. 19
One concern with the country-level split-sample tests is that the partitioning variables used in the subsample tests
could be endogenous. To mitigate this concern, we adopt a two-stage procedure akin to the switching regression framework: we first regress each partitioning variable, including UAI, Government Bank Ownership, Foreign Bank Ownership, Prevalence of Large Banks, Rule of Law, and Government Effectiveness, on CLT, and then use the median level of each resulting residual to split the full sample into two subsamples (Chen, Chen, & Wei, 2011). By construction, the residual partitioning variables are orthogonal to CLT. In unreported results, which are available upon request, we find that our findings in Table 7 do not materially change.
45
REFERENCES Ahern, K. R., Daminelli, D., & Fracassi, C. forthcoming. Lost in Translation? The effect of cultural values on mergers around the world. Journal of Financial Economics. Allen, L. & Li, G. 2011. Clawbacks and cronyism: Evidence from China. Financial Management, 40(3): 733-756. Ayyagari, M., Beck, T., & Demirgüç-Kunt, A. 2007. Small and medium enterprises across the globe. Small Business Economics, 29(4): 415-434. Barth, J. R., Caprio, G., & Levine, R. 2004. Bank regulation and supervision: What works best? Journal of Financial Intermediation, 13(2): 205-248. Barth, J. R., Caprio, G., & Levine, R. 2006. Rethinking bank regulation: Till angels govern. New York, NY: Cambridge Univ Press. Barth, J. R., Lin, C., Lin, P., & Song, F. M. 2009. Corruption in bank lending to firms: Cross-country micro evidence on the beneficial role of competition and information sharing. Journal of Financial Economics, 91(3): 361-388. Beck, T. & Demirgüç-Kunt, A. 2006. Small and medium-size enterprises: Access to finance as a growth constraint. Journal of Banking & Finance, 30(11): 2931-2943. Beck, T., Demirgüç-Kunt, A., & Levine, R. 2000. A new database on financial development and structure. World Bank Economic Review, 14(3): 597-605. Beck, T., Demirgüç-Kunt, A., & Levine, R. 2006. Bank supervision and corruption in lending. Journal of Monetary Economics, 53(8): 2131-2163. Beck, T., Demirgüç-Kunt, A., & Maksimovic, V. 2005. Financial and legal constraints to growth: Does firm size matter? Journal of Finance, 60(1): 137-177. Beck, T., Demirgüç-Kunt, A., & Maksimovic, V. 2008. Financing patterns around the world: Are small firms different? Journal of Financial Economics, 89(3): 467-487. Benjamin, D. & Wright, M. L. J. 2009. Recovery before redemption: A theory of delays in sovereign debt renegotiations. Working Paper, University of California at Los Angeles. Berger, A. N. & Black, L. K. 2011. Bank size, lending technologies, and small business finance. Journal of Banking & Finance, 35(3): 724-735. Berger, A. N., Miller, N. H., Petersen, M. A., Rajan, R. G., & Stein, J. C. 2005. Does function follow organizational form? Evidence from the lending practices of large and small banks. Journal of Financial Economics, 76(2): 237-269. 46
Berger, A. N. & Udell, G. F. 2002. Small business credit availability and relationship lending: The importance of bank organisational structure. Economic Journal, 112(477): F32-F53. Billett, M. T., Flannery, M. J., & Garfinkel, J. A. 1995. The effect of lender identity on a borrowing firm's equity return. Journal of Finance, 50(2): 699-718. Boyd, J. H., Levine, R., & Smith, B. D. 2001. The impact of inflation on financial sector performance. Journal of Monetary Economics, 47(2): 221-248. Brewer, M. B. 1991. The social self: On being the same and different at the same time. Personality and Social Psychology Bulletin, 17(5): 475-482. Charumilind, C., Kali, R., & Wiwattanakantang, Y. 2006. Connected lending: Thailand before the financial crisis. Journal of Business, 79(1): 181-217. Chen, C. C., Peng, M. W., & Saparito, P. A. 2002. Individualism, collectivism, and opportunism: A cultural perspective on transaction cost economics. Journal of Management, 28(4): 567-583. Chen, K. C. W., Chen, Z., & Wei, K. J. 2011. Agency costs of free cash flow and the effect of shareholder rights on the implied cost of equity capital. Journal of Financial and Quantitative Analysis, 46(1): 171207. Chui, A. C. W. & Kwok, C. C. Y. 2008. National culture and life insurance consumption. Journal of International Business Studies, 39(1): 88-101. Chui, A. C. W., Lloyd, A. E., & Kwok, C. C. Y. 2002. The determination of capital structure: Is national culture a missing piece to the puzzle? Journal of International Business Studies, 33(1): 99-127. Chui, A. C. W., Titman, S., & Wei, K. C. J. 2010. Individualism and momentum around the world. Journal of Finance, 65(1): 361-392. Cole, R. A., Goldberg, L. G., & White, L. J. 2004. Cookie cutter vs. character: The micro structure of small business lending by large and small banks. Journal of Financial and Quantitative Analysis, 39(2): 227-251. Cull, R., Haber, S., & Imai, M. 2011. Related lending and banking development. Journal of International Business Studies, 42(3): 406-426. Demirgüç-Kunt , A. & Levine, R. 2004. Financial structure and economic growth: A cross-country comparison of banks, markets, and development. Cambridge, MA: MIT press. Demirgüç-Kunt, A. & Levine, R. 2008. Finance, financial sector policies, and long-run growth. Commission on Growth and Development Working Paper No. 11, The World Bank
47
Demski, J. S. 1994. Managerial uses of accounting information. Norwell, MA: Kluwer Academic Publishers. Djankov, S., McLiesh, C., Nenova, T., & Shleifer, A. 2003. Who owns the media? Journal of Law and Economics, 46(2): 341-381. Djankov, S., McLiesh, C., & Shleifer, A. 2007. Private credit in 129 countries. Journal of Financial Economics, 84(2): 299-329. Faccio, M. 2006. Politically connected firms. American Economic Review, 96(1): 369-386. Faccio, M. 2010. Differences between politically connected and nonconnected firms: A cross country analysis. Financial Management, 39(3): 905-928. Fincher, C. L., Thornhill, R., Murray, D. R., & Schaller, M. 2008. Pathogen prevalence predicts human cross-cultural variability in individualism/collectivism. Proceedings of the Royal Society B: Biological Sciences, 275(1640): 1279-1285. Fried, J. & Shilon, N. 2011. Excess pay and the Dodd-Frank clawback. The Conference Board: Director notes, No. DN-V3N20: 1-8. Getz, K. A. & Volkema, R. J. 2001. Culture, perceived corruption, and economics: A model of predictors and outcomes. Business & Society, 40(1): 7-30. Gómez, C., Kirkman, B. L., & Shapiro, D. L. 2000. The impact of collectivism and in-group/out-group membership on the evaluation generosity of team members. Academy of Management Journal, 43(6): 1097-1106. Guiso, L., Sapienza, P., & Zingales, L. 2006. Does culture affect economic outcomes? Journal of Economic Perspectives, 20(2): 23-48. Hertzberg, A., Liberti, J. M., & Paravisini, D. 2010. Information and incentives inside the firm: Evidence from loan officer rotation. Journal of Finance, 65(3): 795-828. Hofstede, G. H. 1980. Culture's consequences: International differences in work-related values. . Newbury Park, CA: SAGE Publications. Hofstede, G. H. 2001. Culture's consequences: Comparing values, behaviors, institutions, and organizations across nations. Thousand Oaks, CA: SAGE Publications. House, R., Hanges, P., Javidan, M., Dorfman, P. W., & Gupta, V. 2004. Culture, leadership, and organizations: The GLOBE study of 62 societies. Thousand Oaks, CA: Sage Publications. House, R., Javidan, M., Hanges, P., & Dorfman, P. 2002. Understanding cultures and implicit leadership theories across the globe: An introduction to project GLOBE. Journal of World Business, 37(1): 3-10. 48
Houston, J. F., Lin, C., & Ma, Y. 2011. Media ownership, concentration and corruption in bank lending. Journal of Financial Economics, 100(2): 326-350. Husted, B. W. 1999. Wealth, culture, and corruption. Journal of International Business Studies, 30(2): 339-341. Husted, B. W. 2002. Culture and international anti-corruption agreements in Latin America. Journal of Business Ethics, 37(4): 413-422. Husted, B. W. & Allen, D. B. 2008. Toward a model of cross-cultural business ethics: The impact of individualism and collectivism on the ethical decision-making process. Journal of Business Ethics, 82(2): 293-305. Jing, R. & Graham, J. L. 2008. Values versus regulations: How culture plays its role. Journal of Business Ethics, 80(4): 791-806. Johnson, S. A. 1997. The effect of bank reputation on the value of bank loan agreements. Journal of Accounting, Auditing and Finance, 12(1): 83-100. Kaufmann, D., Kraay, A., & Zoido-Lobatón, P. 1999. Aggregating governance indicators. Policy Research Working Paper Series 2195, The World Bank Khatri, N., Tsang, E. W. K., & Begley, T. M. 2006. Cronyism: A cross-cultural analysis. Journal of International Business Studies, 37(1): 61-75. Khwaja, A. I. & Mian, A. 2005. Do lenders favor politically connected firms? Rent provision in an emerging financial market. Quarterly Journal of Economics, 120(4): 1371-1411. Kirkman, B. L., Lowe, K. B., & Gibson, C. B. 2006. A quarter century of cultures consequences: A review of empirical research incorporating Hofstede's cultural values framework. Journal of International Business Studies, 37(3): 285-320. Kwok, C. C. Y. & Tadesse, S. 2006. National culture and financial systems. Journal of International Business Studies, 37(2): 227-247. La Porta, R., Lopez-De-Silanes, F., & Shleifer, A. 2002. Government ownership of banks. Journal of Finance, 57(1): 265-301. La Porta, R., Lopez-De-Silanes, F., Shleifer, A., & Vishny, R. 1999. The quality of government. Journal of Law, Economics, and Organization, 15(1): 222-279. La Porta, R., Lopez-De-Silanes, F., & Zamarripa, G. 2003. Related lending. Quarterly Journal of Economics, 118(1): 231-268.
49
Laeven, L. & Majnoni, G. 2003. Loan loss provisioning and economic slowdowns: too much, too late? Journal of Financial Intermediation, 12(2): 178-197. Laeven, L. & Valencia, F. 2010. Resolution of banking crises: The good, the bad, and the ugly. Working paper 10/146, IMF Lee, K.-W. & Sharpe, I. G. 2009. Does a bank's loan screening and monitoring matter? Journal of Financial Services Research, 35(1): 33-52. Leung, K. & Bond, M. H. 1984. The impact of cultural collectivism on reward allocation. Journal of Personality and Social psychology, 47(4): 793-804. Levine, R. 1996. Foreign banks, financial development, and economic growth. In C. E. Barfield (Ed.), International financial markets: Harmonization versus competition: 224-254. Washington, D.C.: The AEI Press. Levine, R. 1997. Financial development and economic growth: Views and agenda. Journal of Economic Literature, 35(2): 688-726. Levine, R. 2005. Finance and growth: Theory and evidence. In P. Aghion & S. N. Durlauf (Ed.), Handbook of economic growth: 865-934. San Diego, CA: Elsevier. Li, K., Griffin, D., Yue, H., & Zhao, L. 2011. National culture and capital structure decisions: Evidence from foreign joint ventures in China. Journal of International Business Studies, 42(4): 477-503. Li, K., Griffin, D., Yue, H., & Zhao, L. 2012. How does culture influence corporate risk-taking? Working Paper, University of British Columbia Li, S., Qiu, J., & Wan, C. 2011. Corporate globalization and bank lending. Journal of International Business Studies, 42(8): 1016-1042. Liberti, J. M. & Mian, A. R. 2009. Estimating the effect of hierarchies on information use. Review of Financial Studies, 22(10): 4057-4090. Licht, A. N., Goldschmidt, C., & Schwartz, S. H. 2005. Culture, law, and corporate governance. International Review of Law and Economics, 25(2): 229-255. Lonner, W. J. & Adamopoulos, J. 1997. Culture as antecedent of behavior. In J. Berry, Y. Poortinga, & J. Pandey (Ed.), Handbook of cross-cultural psychology, Vol. 1: Theory and method: 43-83. Needham Heights, MA: Allyn & Bacon. Macrae, J. 1982. Underdevelopment and the economics of corruption: A game theory approach. World Development, 10(8): 677-687.
50
Markus, H. R. & Kitayama, S. 1991. Culture and the self: Implications for cognition, emotion, and motivation. Psychological Review, 98(2): 224-253. Megginson, W. L. 2005. The economics of bank privatization. Journal of Banking and Finance, 29(8-9): 1931-1980. Megginson, W. L. 2005. The financial economics of privatization. New York, NY: Oxford University Press. Murray, D. R. & Schaller, M. 2010. Historical prevalence of infectious diseases within 230 geopolitical regions: A tool for investigating origins of culture. Journal of Cross-Cultural Psychology, 41(1): 99-108. North, D. C. 1990. Institutions, institutional change, and economic performance. Cambridge, UK: Cambridge University Press. Norton, E. C., Wang, H., & Ai, C. 2004. Computing interaction effects and standard errors in logit and probit models. Stata Journal, 4(2): 154-167. Reeb, D., Sakakibara, M., & Mahmood, I. P. 2012. From the Editors: Endogeneity in international business research. Journal of International Business Studies, 43(3): 211-218. Rose-Acherman, S. 1999. Corruption and government: Causes, consequences and reform. New York, NY: Cambridge University Press. Sakalaki, M., Richardson, C., & Thepaut, Y. 2007. Machiavellianism and economic opportunism. Journal of Applied Social Psychology, 37(6): 1181-1190. Sapienza, P. 2004. The effects of government ownership on bank lending. Journal of Financial Economics, 72(2): 357-384. Schwartz, S. H. 1994. Beyond individualism/collectivism: New cultural dimensions of values. In U. Kim, H. C. Triandis, C. Kagitcibasi, S.-C. Chol, & G. Yoon (Ed.), Individualism and collectivism: Theory, method, and applications: 85-119. Thousand Oaks, CA: SAGE Publications. Seleim, A. & Bontis, N. 2009. The relationship between culture and corruption: A cross-national study. Journal of Intellectual Capital, 10(1): 165-184. Shao, L., Kwok, C. C. Y., & Guedhami, O. 2010. National culture and dividend policy. Journal of International Business Studies, 41(8): 1391-1414. Shleifer, A. & Vishny, R. W. 1998. The grabbing hand: Government pathologies and their cures. Cambridge, MA: Harvard University Press. Siegel, J. I., Licht, A. N., & Schwartz, S. H. 2011. Egalitarianism and international investment. Journal of Financial Economics, 102(3): 621-642. 51
Sivakumar, K. & Nakata, C. 2001. The stampede toward Hofstede's framework: Avoiding the sample design pit in cross-cultural research. Journal of International Business Studies, 32(3): 555-574. Stein, J. C. 2002. Information production and capital allocation: Decentralized versus hierarchical firms. Journal of Finance, 57(5): 1891-1921. Stulz, R. M. & Williamson, R. 2003. Culture, openness, and finance. Journal of Financial Economics, 70(3): 313-349. Tajfel, H. & Turner, J. C. 1985. The social identity theory of intergroup behavior. In S. Worchel & W. G. Austin (Ed.), Psychology of intergroup relations: 7-24. Chicago: Nelson-Hall. Tang, L. & Koveos, P. E. 2008. A framework to update Hofstede's cultural value indices: Economic dynamics and institutional stability. Journal of International Business Studies, 39 (6): 1045-1063. Triandis, H. C. 1995. Individualism & collectivism. Boulder, Oxford: Westview Press. Triandis, H. C. 2001. Individualism collectivism and personality. Journal of Personality, 69(6): 907-924. Triandis, H. C., Bontempo, R., Villareal, M. J., Asai, M., & Lucca, N. 1988. Individualism and collectivism: Cross-cultural perspectives on self-ingroup relationships. Journal of Personality and Social psychology, 54(2): 323-338. Udell, G. F. 1989. Loan quality, commercial loan review and loan officer contracting. Journal of Banking and Finance, 13(3): 367-382. Vitell, S. J., Nwachukwu, S. L., & Barnes, J. H. 1993. The effects of culture on ethical decision-making: An application of Hofstede's typology. Journal of Business Ethics, 12(10): 753-760. Williamson, O. E. 1988. Corporate finance and corporate governance. Journal of Finance, 43(3): 567-591. Williamson, O. E. 2000. The new institutional economics: Taking stock, looking ahead. Journal of Economic Literature, 38(3): 595-613. World Bank. 2000. World Business environment survey. The World Bank, Washington, DC. World Bank. 2010. World development indicators. The World Bank, Washington, DC. Zheng, X., Ghoul, S. E., Guedhami, O., & Kwok, C. C. Y. 2012. National culture and corporate debt maturity. Journal of Banking & Finance, 36(2): 468-488.
52
Appendix Variable definitions and sources Variable
Definition
Source
Bank Corruption
The degree to which a firm manager views corruption in bank lending as an obstacle to a firm’s operation and growth (1-no obstacle, 2-minor obstacle, 3moderate obstacle, 4- major obstacle).
World Business Environment Survey, or WBES (World Bank, 2000)
Bank Corruption Dummy
Dummy variable that equals zero if a firm manager rates corruption in bank lending as no obstacle to a firm’s operation and growth , and one if the firm manager rates corruption in bank lending as a minor obstacle, moderate obstacle, or major obstacle to the firm’s operation and growth.
WBES (World Bank, 2000)
IDV
Hofstede’s cultural index of Individualism.
Hofstede (2001)
CLT
A cultural index of collectivism, which is equal to 100 minus Hofstede’s cultural index of Individualism. Higher values of this index imply greater collectivism.
Hofstede (2001)
UAI
Hofstede’s cultural index of Uncertainty Avoidance.
Hofstede (2001)
MAS
Hofstede’s cultural index of Masculinity.
Hofstede (2001)
PDI
Hofstede’s cultural index of Power Distance.
Hofstede (2001)
CONS
Schwartz’s cultural index of Conservatism.
Schwartz (1994)
CLT_TK
100 minus Tang and Koveos’ updated cultural index of Individualism.
Tang and Koveos(2008)
In-group CLT
GLOBE’s cultural index on Should-be In-group Collectivism, which measures the degree to which individuals express pride, loyalty, and cohesiveness in their organization or families.
House et al. (2004)
Government
Dummy variable that equals one if any government agency or state body has a financial stake in the ownership of the firm, and zero otherwise.
WBES (World Bank, 2000)
Foreign
Dummy variable that equals one if any foreign company or individual has a financial stake in the ownership of the firm, and zero otherwise.
WBES (World Bank, 2000)
Export
Dummy variable that equals one if the firm exports, and zero otherwise.
N_competitors
Natural logarithm of the number of competitors, which is from the survey question “Regarding your firm’s major product line, how many competitors do you face in your market?”
WBES (World Bank, 2000)
Sales
Natural logarithm of firm sales.
WBES (World Bank, 2000)
Manufacture
Dummy variable that equals one if the firm is in the manufacturing industry, and zero otherwise.
WBES (World Bank, 2000)
Service
Dummy variable that equals one if the firm is in the services industry, and zero otherwise.
WBES (World Bank, 2000)
Priv
Ratio of private credit by deposit money banks to GDP in 1999.
Beck et al.(2000), World Bank (Data updated April 2010)
Inflation
Consumer price inflation (annual %) in 1999.
World Development Indicator, or WDI (World Bank, 2010)
GDP_growth
GDP growth rate (annual %) in 1999.
WDI (World Bank, 2010)
Official Supervisory Power
Principal component indicator of 14 dummy variables: (1) Does the supervisory agency have the right to meet with external auditors to discuss their report without the approval of the bank? (2) Are auditors required by law to communicate directly to the supervisory agency any presumed involvement of bank directors or senior managers in elicit activities, fraud, or insider abuse? (3) Can supervisors take legal action against external auditors for negligence? (4) Can the supervisory authority force a bank to change its internal organizational structure? (5) Are off-balance sheet items disclosed to supervisors? (6) Can the supervisory agency order the bank’s directors or management to constitute provisions to cover actual or potential losses? (7) Can the supervisory agency suspend directors’ decision to distribute: (a) Dividends? (b) Bonuses? (c) Management fees? (8) Can the supervisory agency legally declare-such that this declaration supersedes the rights of bank shareholders-that a bank is insolvent? (9) Does the Banking Law give authority to the supervisory agency to intervene,
Barth et al.(2004)
53
that is, suspend some or all ownership rights-a problem bank? And (10) Regarding bank restructuring and reorganization, can the supervisory agency or any other government agency do the following: (a) Supersede shareholder rights? (b) Remove and replace management? (c) Remove and replace directors? A higher value indicates wider and stronger authority for bank supervisors. Private Monitoring Index
Principal component indicator of nine dummy variables that measure whether: (1) bank directors and officials are legally liable for the accuracy of information disclosed to the public; (2) whether banks must publish consolidated accounts; (3) whether banks must be audited by certified international auditors; (4) whether 100% of the 10 largest banks are rated by international rating agencies; (5) whether off-balance sheet items are disclosed to the public; (6) whether banks must disclose their risk management procedures to the public; (7) whether accrued, though unpaid interest/ principal enter the income statement while the loan is still non-performing; (8) whether subordinated debt is allowable as part of capital; and (9) whether there is no explicit deposit insurance system and no insurance was paid the last time a bank failed. A higher value indicates more information and higher incentives for creditors to monitor bank activities.
Barth et al. (2004)
Bank Concentration
Assets of the three largest banks as a share of the assets of all commercial banks in 1999.
Beck et al.(2000), World Bank (Data updated April 2010)
Private Bureau Age
By the end of 1999, years since the starting data of the oldest private credit bureau in the country.
Djankov, McLiesh, and Shleifer(2007)
State Ownership, Press
Percentage of state-owned newspapers out of the five largest daily newspapers (by circulation) in 1999.
Djankov, Mcliesh, Nenova, and Shleifer(2003)
Economic Freedom
A country’s overall economic freedom score, given as an average of its 10 subcomponents, including business freedom, trade freedom, fiscal freedom, government size, monetary freedom, investment freedom, financial freedom, property rights, freedom from corruption and labor freedom. This index ranges from 0 to 100, with a higher score indicating that a country is more economically free. Individuals in an economically free society would be free and entitled to work, produce, consume, and invest in any way they please under a rule of law, with their freedom at once both protected and respected by the state. The value of this index in 1999 is used.
Heritage Foundation (2010)
Size of Government
This index is a subcomponent of economic freedom, considering the level of government expenditures (including consumption and transfers) as a percentage of GDP. Government expenditure is rescaled through a special formula to construct this index. This index ranges from 0 to 100, with lower value reflecting a higher level of government expenditure. The value of this index in 1999 is used.
Heritage Foundation (2010)
State Control
An indicator that shows the extent to which an economy is owned/controlled by government. Higher values indicate less state control. The value of this indicator in 1999 is used.
Economist Intelligence Unit
Political Connection
Percentage of firms in each country connected with a minister or member of parliament, or with a close relationship with a top politician or party.
Faccio (2006)
Need for Special Connections
The degree to which a firm manager views the need for special connections with banks and financial institutions as an obstacle to the firm’s operation and growth (1-no obstacle, 2-minor obstacle, 3-moderate obstacle, 4-major obstacle).
WBES (World Bank, 2000)
GDP_pc70
GDP per capita (current US$) in 1970.
WDI (World Bank, 2010)
Latitude
Absolute value of the latitude of a country, scaled to take a value between zero and one.
La Porta, Lopez-DeSilanes, and Shleifer(1999)
Diseases
An overall index of the historical prevalence of nine diseases within different geopolitical regions worldwide. The nine diseases coded include leishmanias, schistosomes, trypanosomes, leprosy, malaria, typhus, filariae, dengue, and tuberculosis. A 4-point coding scheme was employed: 0 = completely absent or never reported, 1 = rarely reported, 2 = sporadically or moderately reported, 3 = present at severe levels or epidemic levels at least once. All nine disease prevalence ratings were standardized by converting them to z scores. The overall index was computed as the mean of z scores for nine diseases. The mean of the overall index is approximately 0; positive scores indicate disease prevalence that is higher than the mean, and negative scores indicate disease
Murray and Schaller (2010)
54
prevalence that is lower than the mean. General Financing Obstacle
The degree to which a firm manager indicates that financing is problematic for the operation and growth of the firm (1-no obstacle, 2-minor obstacle, 3moderate obstacle, 4- major obstacle).
WBES (World Bank, 2000)
Overall Corruption
The degree to which a firm manager indicates that general corruption is problematic for the operation and growth of the firm (1-no obstacle, 2-minor obstacle, 3-moderate obstacle, 4- major obstacle).
WBES (World Bank, 2000)
Stock Market Cap
Total market value of listed shares to GDP in 1999.
Beck et al.(2000), World Bank (Data updated April 2010)
Stock Market Turnover
Ratio of the value of total shares traded to average real market capitalization in 1999.
Beck et al.(2000), World Bank (Data updated April 2010)
Market Based System
Dummy variable that equals one if a country has a market-based financial system, and zero if it has a bank-based financial system.
Demirgüç-Kunt and Levine (2004)
Government Bank Ownership
The fraction of the banking system’s assets in banks in which government ownership is 50% or greater. The data are compiled based on a survey of bank regulation and supervisory practices (received from 1998 to early 2000) for 117 countries.
Barth et al.(2004)
Foreign Bank Ownership
The fraction of the banking system’s assets in banks in which foreign ownership is 50% or greater. The data are compiled based on a survey of bank regulation and supervisory practices (received from 1998 to early 2000) for 117 countries.
Barth et al.(2004)
Bank Crisis
Dummy variable that equals one if a country suffered a systemic banking crisis during 1997 to 1999, and zero otherwise.
Based on Laeven and Valencia (2010)
Sovereign Default
Dummy variable that equals one if a country defaulted on sovereign debts owed to private sector creditors, such as banks and bondholders, during 1990 to 1999, and zero otherwise.
Based on Benjamin and Wright (2009)
Creditor Rights
Index of creditor rights. A score of one is assigned for each of the following rights of secured lenders defined in a county’s laws and regulations: (1) debtors are subject to restrictions, such as creditor consent or minimum dividends, in filing for reorganization; (2) secured creditors are able to seize their collateral after a reorganization petition is approved, i.e., there is no "automatic stay" or "asset freeze"; (3) secured creditors are paid first out of the proceeds of liquidation, as opposed to other creditors such as the government or workers; and (4) management does not retain administration of its property pending the resolution of the reorganization. The index ranges from 0 (weak creditor rights) to 4 (strong creditor rights), and is constructed in January for every year from 1978 to 2003. For the purpose of this study, the value of the creditor rights index in 1999 is used.
Djankov et al. (2007)
Rule of Law
Principal component indicator of survey indicators in 1998 measuring perceptions of the extent to which agents have confidence in and abide by the rules of society, and in particular the quality of contract enforcement, property rights, the police, and the courts, as well as the likelihood of crime and violence.
Kaufmann et al.(1999)
Government Effectiveness
Principal component indicator of survey indicators in 1998 measuring perceptions of the quality of public services, the quality of the civil service and the degree of its independence from political pressures, the quality of policy formulation and implementation, and the credibility of the government's commitment to such policies.
Kaufmann et al.(1999)
Private
Dummy variable that equals one if no government agency, state body, foreign company, or foreign individual has a financial stake in the ownership of the firm, and zero otherwise.
WBES (World Bank, 2000)
Prevalence of Large Banks
Number of large banks in a country, based on a sample including only commercial banks or bank holdings & holding companies. A bank is classified as a large bank if its total assets in 1999 are larger than $1 billion USD. Non-performing loans deflated by lagged total loans.
Constructed from Bankscope BankScope
Loan loss provisions deflated by lagged total loans. The natural logarithm of bank total assets in USD Growth rate of total loans in USD. Earnings before taxes and loan loss provisions scaled by lagged assets.
BankScope BankScope BankScope BankScope
Non-performing loans Loan loss provisions Bank-size Loan_growth Profitability
55
Figure 1 Comparison of corruption in bank lending between high collectivist and low collectivist countries 80 Percentage 71.66 70 60 51.55 50 40 30 22.98 20
14.94
13.91
11.57 6.97
10
6.42
0 1
2
Low CLT Countries
3
4
Level of lending corruption
High CLT Countries
Following the extant literature, we construct our proxy for bank corruption (Bank Corruption) using responses of firm managers to the WBES survey question: “Is corruption of bank officials an obstacle for the operation and growth of your business?” This indicator takes the value of 1 if the answer is “no obstacle”, 2 if “minor obstacle”, 3 if “moderate obstacle”, and 4 if “major obstacle”. We divide our sample into above- and below-median CLT subsamples and compare the percentage of responses to each of the four categories of answers across the two subsamples.
56
Table 1 Descriptive statistics Panel A. Summary Statistics by Country Country Bank Name N Corruption CLT Argentina 85 1.53 54 Brazil 180 1.28 62 Bulgaria 77 2.08 70 Canada 86 1.03 20 Chile 88 1.2 77 Colombia 90 1.54 87 Costa Rica 85 1.74 85 Czech Rep. 94 1.85 42 Ecuador 91 2.68 92 El Salvador 90 1.71 81 Estonia 96 1.35 40 Germany 92 1.52 33 Guatemala 93 1.53 94 Hungary 97 1.48 20 India 158 1.56 52 Indonesia 91 2.47 86 Italy 74 1.16 24 Malaysia 61 1.87 74 Mexico 87 2.01 70 Pakistan 86 2.49 86 Panama 85 1.42 89 Peru 89 2.19 84 Philippines 93 2.18 68 Poland 179 1.38 40 Portugal 87 1.48 73 Russian Fed 347 1.89 61 Singapore 91 1.26 80 Slovak Rep 96 2.02 48 South Africa 81 1.11 35 Spain 87 1.26 49 Sweden 84 1.06 29 Thailand 51 3 80 Trinidad & Tobago 98 1.78 84 Turkey 141 2.33 63 UK 72 1.04 11 US 86 1.44 9
UAI 86 76 85 48 86 80 86 74 67 94 60 65 101 82 40 48 75 36 82 70 86 87 44 93 104 95 8 51 49 86 29 64
MAS 56 49 40 52 28 64 21 57 63 40 30 66 37 88 56 46 70 50 69 50 44 42 64 64 31 36 48 110 63 42 5 34
PDI 49 69 70 39 63 67 35 57 78 66 40 35 95 46 77 78 50 104 81 55 95 64 94 68 63 93 74 104 49 57 31 64
Government 0.04 0.02 0.3 0.02 0.02 0.01 0.04 0.17 0.05 0.01 0.22 0.09 0 0.18 0.14 0.07 0.12 0.05 0 0.06 0.04 0.01 0.02 0.2 0.06 0.15 0.04 0.18 0.06 0.06 0.06 0.04
Foreign 0.32 0.24 0.08 0.28 0.32 0.38 0.33 0.18 0.13 0.17 0.2 0.3 0.19 0.07 0.3 0.16 0.34 0.15 0.15 0.16 0.19 0.27 0.23 0.09 0.29 0.02 0.36 0.05 0.32 0.25 0.24 0.31
Export 0.29 0.29 0.31 0.51 0.44 0.39 0.44 0.34 0.25 0.28 0.63 0.35 0.31 0.32 0.62 0.26 0.32 0.28 0.4 0.47 0.42 0.24 0.3 0.44 0.23 0.07 0.49 0.41 0.85 0.43 0.58 0.41
N_ competitors 0.6 0.59 0.95 0.63 0.65 0.59 0.64 0.93 0.66 0.64 0.93 0.58 0.65 1 1.03 1.02 0.62 0.59 0.65 0.71 0.59 0.59 0.66 1.05 0.76 0.98 0.62 0.9 1 0.6 0.64 1.01
Sales 16.89 16.61 0.91 16.95 16.9 15.65 15.42 0.84 17.07 15.46 1.27 17.48 16.47 1.17 5.92 18.13 20.72 17.84 15.52 15.33 16.77 16.1 15.42 1.28 19.67 0.6 17.47 0.82 18.27 15.63 15.92 18.96
Manufacture 0.36 0.21 0 0.31 0.26 0.26 0.31 0 0.38 0.28 0 0.27 0.26 0 0 0.15 0.28 0.36 0.28 0.21 0.31 0.26 0.16 0 0.2 0 0.32 0 0.04 0.32 0.26 0
Service 0.19 0.22 0 0.52 0.17 0.16 0.38 0 0.51 0.38 0 0.42 0.57 0 0 0.46 0.34 0.23 0.51 0.16 0.45 0.43 0.46 0 0.36 0 0.32 0 0.47 0.15 0.24 0
Priv 0.25 0.29 0.11 0.78 0.58 0.32 0.17 0.55 0.35 0.42 0.31 1.15 0.2 0.24 0.23 0.34 0.63 1.43 0.19 0.24 0.85 0.27 0.41 0.23 0.99 0.11 1.09 0.53 0.65 0.84 0.38 1.43
Inflation -1.17 4.86 2.57 1.73 3.34 10.87 10.05 2.14 52.24 0.51 3.3 0.57 4.86 10 4.67 20.49 1.66 2.74 16.59 4.14 1.25 3.47 5.95 7.28 2.3 85.74 0.02 10.57 5.18 2.31 0.45 0.28
GDP_growth -3.39 0.25 2.3 5.53 -0.76 -4.2 8.22 1.34 -6.3 3.45 -0.3 2.01 3.85 4.17 7.39 0.79 1.46 6.14 3.87 3.66 3.92 0.91 3.4 4.52 3.84 6.4 7.2 0.03 2.36 4.75 4.6 4.45
55 85 35 46
58 45 66 62
47 66 35 40
0.06 0.18 0.01 0.05
0.18 0.09 0.15 0.1
0.43 0.4 0.31 0.3
0.59 0.95 0.69 0.65
14.7 1.58 16.39 17.02
0.28 0 0.25 0.26
0.45 0 0.33 0.33
0.32 0.14 1.13 0.48
3.44 64.87 1.56 2.19
4.39 -3.37 3.47 4.49
57
Uruguay 87 1.13 64 Venezuela 80 1.51 88 Total 3,835 1.67 60.11 Panel B. Summary Statistics for Full Sample Bank STATS Corruption CLT N 3,835 38 MEAN 1.67 60.63 SD 1 25.01 MIN 1 9 P25 1 40 P50 1 66 P75 2 84 MAX 4 94
100 76 71.74
38 73 50.98
61 81 66.58
0 0.01 0.09
0.17 0.23 0.19
0.52 0.38 0.37
0.66 0.55 0.77
15.92 17.11 11.67
0.49 0.2 0.17
0.41 0.45 0.23
0.48 0.11 0.44
5.66 23.57 15.86
-2.85 -5.97 2.52
UAI 38 69.32 22.94 8 49 75.5 86 104
MAS 38 51.5 18.96 5 40 50 64 110
PDI 38 64.21 20.16 31 49 64 78 104
Government 3,835 0.09 0.28 0 0 0 0 1
Foreign 3,835 0.19 0.4 0 0 0 0 1
Export 3,835 0.37 0.48 0 0 0 1 1
N_ competitors 3,835 0.77 0.32 0 0.69 0.69 1.1 2.2
Sales 3,835 11.67 7.71 -2.12 2.08 14.22 17.8 25.33
Manufacture 3,835 0.17 0.38 0 0 0 0 1
Service 3,835 0.23 0.42 0 0 0 0 1
Priv 38 0.51 0.37 0.11 0.24 0.37 0.65 1.43
Inflation 38 10.06 18.34 -1.17 1.73 3.45 10 85.74
GDP_growth 38 2.26 3.61 -6.3 0.25 3.46 4.49 8.22
The table reports the sample distribution by country and summary statistics for all variables in the main regressions. Definitions and sources for all variables are provided in the Appendix. The full sample includes 3,835 firm observations from 38 countries. Panel A presents the average by country of the dependent variable (Bank Corruption) and the key independent variables. Panel B summarizes detailed descriptive statistics for the full sample of 38 countries.
58
Table 2 Collectivism and corruption in bank lending VARIABLES CLT
(1) 0.016*** (13.960)
CONS
(2)
(3)
(4)
(5)
(6)
(7)
(8) 0.023*** (16.076)
(9) 0.011*** (7.592)
-0.003*** (-2.769)
MAS
0.003*** (3.010)
PDI
0.008*** (7.351)
-0.003*** (-3.144) 0.012*** (8.251) -0.008*** (-5.452)
Official Supervisory Power
0.137*** (5.816) -0.134*** (-2.628)
Private Monitoring Index Bank Concentration
0.304** (2.569) -0.010*** (-7.844)
Private Bureau Age State Ownership, Press
N_competitors Sales Manufacture Service Priv Inflation
0.816*** (2.710) -0.266*** (-3.294) -0.242*** (-4.022) -0.066 (-1.386) 0.152* (1.872) -0.024*** (-5.254) -0.082 (-1.155) -0.085 (-1.314) 0.206** (2.541) -0.000 (-0.352)
0.126*** (4.006) -0.162*** (-2.692) -0.065 (-0.341) -0.005*** (-3.702) 0.337 (0.937) -0.268*** (-2.873) -0.185*** (-2.807) -0.023 (-0.415) 0.112 (1.205) -0.011* (-1.945) -0.020 (-0.265) -0.062 (-0.883) 0.476*** (5.248) 0.003* (1.726)
0.259*** (3.489)
UAI
Export
(12) 0.011*** (6.737)
0.015*** (11.693)
In-group CLT
Foreign
(11) 0.018*** (13.671)
0.999*** (11.317)
CLT_TK
Government
(10) 0.012*** (10.123)
-0.240*** (-3.145) -0.183*** (-3.432) -0.108** (-2.498) 0.139* (1.876) -0.018*** (-4.259) -0.130** (-2.139) -0.119** (-2.162) 0.181** (2.445) 0.003*** (3.511)
-0.323*** (-4.020) -0.182*** (-3.027) -0.135*** (-2.820) -0.011 (-0.135) -0.008* (-1.899) -0.006 (-0.078) -0.063 (-0.968) 0.195** (2.369) 0.005*** (5.187)
-0.122 (-1.086) -0.200*** (-3.392) -0.164*** (-3.232) 0.070 (0.787) 0.016*** (2.764) -0.064 (-1.000) -0.088 (-1.502) 0.238*** (2.854) 0.023*** (10.957)
-0.236** (-2.568) -0.115* (-1.865) -0.091* (-1.795) 0.184** (2.026) 0.007 (1.314) -0.118* (-1.726) -0.057 (-0.924) 0.200** (2.401) 0.009*** (8.622)
-0.275*** (-3.667) -0.170*** (-3.200) -0.149*** (-3.479) 0.092 (1.237) 0.003 (0.844) -0.083 (-1.397) -0.071 (-1.302) -0.108 (-1.392) 0.007*** (7.388)
-0.276*** (-3.672) -0.168*** (-3.168) -0.137*** (-3.235) 0.101 (1.355) 0.006 (1.591) -0.086 (-1.458) -0.074 (-1.366) -0.081 (-1.060) 0.007*** (7.452)
-0.269*** (-3.590) -0.166*** (-3.131) -0.123*** (-2.882) 0.102 (1.369) 0.004 (1.104) -0.088 (-1.471) -0.079 (-1.448) -0.081 (-1.087) 0.004*** (4.030)
-0.250*** (-3.310) -0.182*** (-3.370) -0.128*** (-2.941) 0.147** (1.975) -0.019*** (-4.244) -0.137** (-2.242) -0.151*** (-2.735) 0.198** (2.504) 0.008*** (7.033)
-0.268*** (-3.076) -0.140** (-2.376) -0.056 (-1.137) 0.125 (1.475) -0.009* (-1.876) -0.084 (-1.227) -0.078 (-1.267) 0.333*** (3.858) 0.005*** (4.023)
-0.251*** (-3.141) -0.182*** (-3.270) -0.101** (-2.242) 0.097 (1.259) -0.009** (-2.121) -0.124** (-1.979) -0.130** (-2.258) 0.239*** (3.191) 0.003*** (3.226)
59
GDP_growth
-0.016*** -0.030*** 0.043*** -0.030*** -0.024*** -0.017*** -0.024*** 0.006 -0.025** -0.029*** -0.012* -0.040*** (-2.875) (-4.605) (5.350) (-4.768) (-4.357) (-2.996) (-4.390) (0.947) (-2.532) (-4.979) (-1.661) (-3.610) Observations 3835 3205 2751 2939 3835 3835 3835 3835 3223 3641 3206 2761 Pseudo R2 0.05 0.04 0.08 0.03 0.02 0.02 0.03 0.06 0.05 0.06 0.06 0.07 The table presents results of ordered probit regressions of Bank Corruption on country- and firm-level predictors, where Bank Corruption is the response to the question “Is the corruption of bank officials an obstacle for the operation and growth of your business?”; a value of 1 indicates “no obstacle”, 2 indicates “minor obstacle”, 3 indicates “moderate obstacle”, and 4 indicates “major obstacle”. Definitions and sources of all variables are provided in the Appendix. t-statistics based on robust standard errors are reported beneath each coefficient estimate. ***, **, and * denote significance at the 1%, 5%, and 10% level, respectively.
60
Table 3 Magnitude of the economic effects of collectivism and selected country-level variables on corruption in bank lending Panel A. For Model 9 CLT
Official Supervisory Power
Private Monitoring Index
1
2
3
4
change from min to max
-0.320
0.116
0.078
0.125
1 standard deviation increase
-0.091
0.033
0.023
0.036
change from min to max
-0.197
0.078
0.049
0.070
1 standard deviation increase
-0.053
0.019
0.013
0.021
change from min to max
0.133
-0.047
-0.033
-0.053
1 standard deviation increase
0.038
-0.014
-0.010
-0.015
1
2
3
4
Panel B. For Model 10 CLT
Bank Concentration
Private Bureau Age
change from min to max
-0.346
0.116
0.095
0.135
1 standard deviation increase
-0.100
0.032
0.028
0.040
change from min to max
-0.087
0.027
0.024
0.035
1 standard deviation increase
-0.021
0.007
0.006
0.008
change from min to max
0.296
-0.123
-0.081
-0.091
1 standard deviation increase
0.084
-0.026
-0.024
-0.034
Panel C. For Model 11 CLT
State Ownership, Press
1
2
3
4
change from min to max
-0.466
0.152
0.125
0.189
1 standard deviation increase
-0.143
0.046
0.040
0.057
change from min to max
-0.126
0.034
0.035
0.058
1 standard deviation increase -0.027 0.009 0.008 0.011 This table reports the magnitude of the effects of collectivism and selected country-level variables on the predicted probability that lending corruption is rated an obstacle to firm growth. The estimation of the effects reported in Panels A, B, and C is based on Models 9, 10, and 11 of Table 2, respectively. The numbers in columns 1, 2, 3, and 4 indicate the change in the predicted probability that an average firm rates the corruption of bank officials as no obstacle, a minor obstacle, a moderate obstacle, and a major obstacle, respectively, due to a change in the independent variable of interest from its minimum to its maximum and from its mean minus 0.5 standard deviations to its mean plus 0.5 standard deviations.
61
Table 4 Robustness tests set # 1: The role of the government and political connections VARIABLES CTL Economic Freedom Size of Government
(1)
(2)
(3)
(4)
0.016*** (13.491) -0.021*** (-6.405)
0.015*** (10.101)
0.016*** (11.761)
0.013*** (9.615)
State Control
0.001 (0.753)
Political Connection Government Foreign Export N_competitors Sales Manufacture Service Priv Inflation GDP_growth Observations Pseudo R2
-0.273*** (-3.580) -0.178*** (-3.329) -0.111** (-2.559) 0.080 (1.082) -0.016*** (-3.758) -0.078 (-1.266) -0.074 (-1.323) 0.374*** (4.598) 0.002** (2.455) -0.018*** (-3.312) 3835 0.05
-0.237*** (-3.090) -0.185*** (-3.454) -0.109** (-2.513) 0.137* (1.855) -0.019*** (-4.420) -0.130** (-2.140) -0.119** (-2.161) 0.193** (2.565) 0.003*** (3.483) -0.016*** (-2.999) 3835 0.05
-0.102*** (-2.943) -0.257*** (-3.182) -0.222*** (-3.714) -0.114** (-2.365) 0.097 (1.197) -0.024*** (-5.251) -0.036 (-0.519) -0.040 (-0.642) 0.289*** (3.374) 0.002* (1.697) -0.019*** (-2.988) 3201 0.06
0.028*** (5.367) -0.254*** (-2.725) -0.182*** (-2.879) -0.055 (-1.054) 0.085 (0.961) -0.031*** (-5.482) -0.081 (-1.108) -0.058 (-0.876) 0.134 (1.541) -0.003** (-2.217) -0.039*** (-4.365) 2851 0.06
This table presents robustness tests controlling for the government’s involvement in the economy and political connections, in addition to collectivism (CLT) and the other controls included in Table 2. The dependent variable is Bank Corruption, which is the response to the question “Is the corruption of bank officials an obstacle for the operation and growth of your business?”; a value of 1 indicates “no obstacle”, 2 indicates “minor obstacle”, 3 indicates “moderate obstacle”, and 4 indicates “major obstacle”. Ordered probit estimation is performed. Definitions and sources of all variables are provided in the Appendix. t-statistics based on robust standard errors are reported beneath each coefficient estimate. ***, **, and * denote significance at the 1%, 5%, and 10% level, respectively.
62
Table 5 Robustness Tests Set # 2: Alternative dependent variable, instrumental variables (IV) regression, and alternative estimation methods (1)
VARIABLES
Bank corruptio n dummy
(2) Bank corrupti on dummy -country level
(3)
Need for special connecti ons
(4)
IV probit
(5)
HLM firmdev
CLT Government Foreign Export N_competitors Sales Manufacture Service Priv Inflation GDP_growth Constant
0.016*** (13.369) -0.224*** (-2.756) -0.189*** (-3.259) -0.109** (-2.314) 0.156** (2.050) -0.017*** (-3.790) -0.114* (-1.720) -0.099* (-1.649) 0.161** (1.993) 0.002** (2.103) -0.016*** (-2.713) -1.128*** (-9.789) 3835 0.07
0.005*** (4.301) 0.332 (0.422) -0.687 (-1.301) -0.328 (-1.274) 0.236 (0.684) 0.016 (1.223) -0.643 (-1.294) -0.070 (-0.219) 0.034 (0.315) 0.000 (0.113) -0.004 (-0.424) 0.047 (0.129) 38
0.006*** (6.817) -0.322*** (-4.575) -0.179*** (-3.836) -0.137*** (-3.550) 0.028 (0.418) 0.003 (0.921) 0.090* (1.686) 0.101** (2.066) -0.213*** (-3.087) 0.001 (1.076) -0.007 (-1.441)
0.016*** (11.521) -0.054 (-0.448) -0.195*** (-3.058) -0.161*** (-2.933) 0.176* (1.906) -0.000 (-0.036) -0.086 (-1.273) -0.126** (-2.013) 0.041 (0.481) 0.019*** (7.936) 0.038*** (4.386) -1.611*** (-9.979) 2762
(6)
-0.357*** (-2.970) -0.189*** (-3.282) -0.119* (-1.838) -0.127 (-1.587) -0.008 (-1.047) -0.001 (-0.014) -0.057 (-0.094)
ctrymean 0.020*** (6.380) 3.557* (1.792) -1.438 (-1.316) -1.539*** (-2.631) 0.738 (0.890) 0.030 (1.020) -1.301 (-1.157) 0.113 (0.186) 0.076 (0.325) -0.003 (-0.859) -0.039* (-1.822)
OLS
0.010*** (15.407) -0.183*** (-3.419) -0.135*** (-3.624) -0.079** (-2.411) 0.119** (2.039) -0.010*** (-3.297) -0.098** (-2.185) -0.082* (-1.939) 0.153** (2.550) 0.004*** (4.452) -0.014*** (-2.947) 1.103*** (13.578) 3835
(7)
(8)
(9)
Weighte d ordered probit
Ordered probit, clusterin g at the country level
Ordered logit
0.015*** (13.359) -0.234** (-2.372) -0.141** (-2.426) -0.085* (-1.820) 0.001 (0.015) -0.023*** (-5.268) -0.104* (-1.710) -0.055 (-1.013) 0.160** (2.174) 0.002** (2.425) -0.022*** (-3.898)
0.016*** (3.916) -0.240** (-2.124) -0.183*** (-3.370) -0.108* (-1.654) 0.139 (1.034) -0.018* (-1.811) -0.130 (-1.362) -0.119 (-1.202) 0.181 (0.666) 0.003 (1.291) -0.016 (-0.889)
0.027*** (13.591) -0.407*** (-3.094) -0.312*** (-3.404) -0.189** (-2.570) 0.263** (2.017) -0.032*** (-4.366) -0.217** (-2.091) -0.199** (-2.144) 0.314** (2.445) 0.005*** (2.844) -0.028*** (-3.051)
Observations 3715 3835 3835 3835 3835 Pseudo R2 0.02 0.04 0.05 0.05 Adj. R2 0.09 0.41 R2 between 0.60 countries F-test 0.000*** The table presents the results of additional robustness regressions of corruption in bank lending on collectivism (CLT) and other controls. Model 1 is estimated using a probit model where the dependent variable is Bank Corruption Dummy, which equals one if the firm rates corruption of bank officials as a minor obstacle, moderate obstacle, or major obstacle, and zero otherwise. Model 2 presents country-level results on the basis of Model 1 using 38 country-mean observations. Model 3 is estimated using an ordered probit model where the dependent variable is Need for Special Connections, which is the response to the question “Is the need for special connections with banks and financial institutions an obstacle for the operation and growth of your business?”; a value of 1 indicates “no obstacle”, 2 indicates “minor obstacle”, 3 indicates “moderate obstacle”, and 4 indicates “major obstacle”. IV probit estimation is used for Model 4 where the dependent variable is Bank Corruption Dummy. The instrumental variables in the first stage regression are GDP_pc70 (GDP per capita in 1970), Latitude (a country’s absolute value of latitude), and Diseases (an overall index of historical prevalence of nine diseases). In Models 5 through 9, the dependent variable is Bank Corruption. Model 5 fits Hierarchical Linear Modeling. In Model 5, we center every independent variable by its grand mean (mean of within-country means). We then decompose each grand mean centered firm-level independent variable into two components: country-level mean value (averaged within each country) and firm-level deviation from the country-level mean value (subtracting the within-country-level mean from the grand mean-centered firm-level variable). The alternative methodologies used in Models 6 through 9 are OLS regression, weighted ordered probit regression, ordered probit regression with clustering at the country level, and ordered logit regression. Definitions and sources of all variables are outlined in the Appendix. t-statistics, based on robust standard errors, are reported beneath each coefficient estimate. ***, **, and * denote significance at the 1%, 5%, and 10% level, respectively.
63
Table 6 Collectivism and corruption in bank lending: Split-sample tests by firm characteristics Size Firm Ownership Export Small Medium Large Government Foreign Private Yes No (1) (2) (3) (4) (5) (6) (7) (8) Panel A. Magnitude of the Effects: One Standard Deviation Increase in CLT 1 -0.1467 -0.1523 -0.0758 -0.1137 -0.1249 -0.1382 -0.1214 -0.1414 4 0.0728 0.0587 0.0223 0.0363 0.0370 0.0613 0.0407 0.0637 Panel B. Regression Estimates CLT 0.016*** 0.019*** 0.010*** 0.018*** 0.016*** 0.016*** 0.015*** 0.016*** (9.159) (10.389) (4.030) (3.813) (6.631) (11.981) (7.952) (11.564) Additional Controls Yes Yes Yes Yes Yes Yes Yes Yes Observations 1,360 1,670 804 326 745 2,817 1,417 2,418 Pseudo R2 0.04 0.06 0.04 0.07 0.06 0.05 0.06 0.05 This table presents results of ordered probit regressions of corruption in bank lending (Bank Corruption) on collectivism (CLT) and other firm- and country-level controls in various subsamples. All regressions include the controls in Table 2 (Government, Foreign, Export, N_competitiors, Sales, Manufacture, Service, Priv, Inflation, and GDP_growth) with the exception that Models 4-6 do not control for Government and Foreign and Models 7-8 do not control for Export. The coefficients on these additional control variables are not reported for brevity. All unreported results are available upon request. Panel A reports the magnitude of the effect of a one standard deviation increase in CLT on the level of bank corruption perceived by firms. To save space, Panel A only reports the changes in the probability that a firm rates corruption in bank lending to be no obstacle (1) and a major obstacle (4) due to a one standard deviation increase in CLT. Panel B presents the coefficient estimates from ordered probit regressions. Definitions and sources of all variables are provided in the Appendix. t-statistics based on robust standard errors are reported beneath each coefficient estimate. ***, **, and * denote significance at the 1%, 5%, and 10% level, respectively. VARIABLES
64
Table 7 Collectivism and corruption in bank lending: Split-sample tests by country characteristics Government Bank Foreign Bank Prevalence of Large Government Ownership Ownership Banks Rule of Law Effectiveness VARIABLES LOW HIGH LOW HIGH LOW HIGH LOW HIGH LOW HIGH LOW HIGH (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) Panel A. Magnitude of the Effects: One Standard Deviation Increase in CLT 1 -0.1752 -0.0744 -0.1093 -0.1381 -0.1370 -0.0765 -0.1526 -0.1006 -0.0994 -0.0961 -0.0896 -0.0496 4 0.0574 0.0320 0.0354 0.0651 0.0604 0.0295 0.0771 0.0305 0.0554 0.0241 0.0509 0.0110 Panel B. Regression Estimates CLT 0.018*** 0.011*** 0.012*** 0.027*** 0.016*** 0.011*** 0.018*** 0.014*** 0.017*** 0.012*** 0.017*** 0.006*** (12.026) (5.437) (8.892) (9.659) (9.737) (4.104) (9.364) (8.572) (8.300) (8.047) (6.635) (4.440) Additional Controls Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Observations 1941 2154 1,631 1,765 1,660 1,578 1956 1879 1,886 1,949 1,940 1,895 Pseudo R2 0.10 0.03 0.06 0.05 0.06 0.02 0.03 0.06 0.03 0.05 0.03 0.06 This table presents the results of ordered probit regressions of corruption in bank lending (Bank Corruption) on collectivism (CLT) and other firm- and country-level controls in various subsamples. All regressions include the controls in Table 2 (Government, Foreign, Export, N_competitiors, Sales, Manufacture, Service, Priv, Inflation, and GDP_growth). The coefficients on these additional control variables are not reported for brevity. All unreported results are available upon request. Panel A reports the magnitude of the effect of a one standard deviation increase in CLT on the level of bank corruption perceived by firms. To save space, Panel A only reports the changes in the probability that a firm rates corruption in bank lending to be no obstacle (1) and a major obstacle (4) due to a one standard deviation increase in CLT. Panel B presents the coefficient estimates from ordered probit regressions. Definitions and sources of all variables are provided in the Appendix. tstatistics based on robust standard errors are reported beneath each coefficient estimate. ***, **, and * denote significance at the 1%, 5%, and 10% level, respectively. UAI
65