Introduction

Despite all societal efforts to fight corruption, this phenomenon is a fact of life around the world. Corruption negatively affects economic development as a major impediment for investments and economic growth (Mauro 1995; Dreher et al. 2007) or for subjective well-being (Tay et al. 2014). Various research has attempted to find the real drivers for corruption, so as to determine the best channels to address it. However, the problem of fighting against corruption still persists. In order to better explain the economic phenomena, numerous recent studies have gone beyond economic explanations and analyzed the impact of values, social norms, and attitudes on the economic behavior of people (Trompenaars 1993; Elenkov and Fileva 2006; Torgler and Schneider 2007; Stiglitz et al. 2010; Frey and Stutzer 2012).

Following these researchers, this study fills a gap in the literature by investigating the subjective factors to see if they are determinants for corruption around the world. Shadabi (2013) noted that corruption is an important variable which results from social and cultural conditions. Various studies have investigated the relationship between corruption and culture, but their results are more or less inconclusive. This study was conducted to fill this gap in the literature. For this purpose, a cross-country survey of 98 countries was used.

Literature Review

According to Transparency International (2015, p. 1) “Corruption is the abuse of entrusted power for private gain. It can be classified as grand, petty, and political, depending on the amounts of money lost and the sector where it occurs.” The World Bank (1997, p. 1) provides a short definition of corruption in which corruption is “the abuse of public office for private gain.”

The most common factor for showing the moral dimension of economic behavior is culture. According to Hofstede (2011), culture is defined as “the collective mental programming of the human mind which distinguishes one group of people from another.” Issues such as honesty, trust in authorities, trust in people, pride, relationship with nature and the world, relationship with others, the ability to do things in one’s own way, avoiding uncertainty, orientation in time and space, and long-term and short-term orientation are elements that characterize the culture of a nation, which are likely to determine the behavior of individuals acting in economic activity.

Husted (1999) finds that corruption is significantly associated with culture. A cultural profile of a corrupt country is one in which there is high power distance, high masculinity, and high uncertainty avoidance. Fisman and Miguel (2007) investigated the relationship between culture and corruption by conducting a survey of New York’s parking violations by diplomats from over 149 countries. They found out that diplomats from highly corrupt countries are more likely to violate parking law than diplomats from less corrupt countries. They concluded that corruption is partially a cultural phenomenon. The same findings were reached by Barr and Serra (2010). In 2005 and 2007, they conducted two experiments on bribery, with Oxford University students participating as subjects who belong to some of the most and least corrupt countries in the world, 33 and 22 countries total, respectively. Both experiments revealed that among undergraduates, culture significantly influences corruption, but the reverse is the case for graduates. Thus, the values and beliefs towards corruption may be strongly related to the country of origin. For the immigrants, these values and benefits could be changed following the change context. Based on the literature review, we test the following working hypothesis:

  • Hypothesis 1. Cultural factors affect the level of corruption.

The studies conducted by Hofstede are extremely useful for understanding the cultural dynamics of nations (Javidan et al. 2006). Hofstede’s cultural model has six dimensions: 1) attitude towards social inequality or power distance (PD); 2) attitude towards the community or individualism versus collectivism (IDV); 3) attitude towards success or masculinity versus femininity (MAS); 4) attitude towards the unknown or uncertainty avoidance (UAI); 5) attitude towards the passage of time or long-term orientation (LTO); and 6) attitude towards control of one's own desires or indulgence and restraint (IND). Each dimension places the culture of a nation on a scale from 0 to 100. Hofstede’s model was applied to 100 countries (Hofstede Centre 2015).

According to the Hofstede Centre (2015), PD refers to “the degree to which the less powerful members of a society accept and expect that power is distributed unequally.” A large degree of power distance means a hierarchical order in which everyone has a place and there is no need for any justification. In a high power distance culture, superiors provide favors to subordinates in return for their loyalty and corruption may occur as a result of nepotism and favoritism (Husted 1999). Various studies document a positive relationship between power distance and corruption (Husted 1999; Davis and Ruhe 2003; Murdoch 2009; Halkos and Tzeremes 2011; McLaughlin 2013; Tong 2014). For our sample, we test the following secondary hypothesis:

  • Hypothesis 1.1. The higher the power distance, the higher the level of corruption.

The IDV dimension of culture refers to the extent to which the decision regarding a person’s life is viewed by an individual or by a group (family or relatives). It considers “whether people’s self-image is defined in terms of “I” or “we” (The Hofstede Centre 2015). A high score on this dimension indicates a high individualist society in which the law is respected. In countries such as the United States, Great Britain, and Australia, individual initiative, competition, and democracy are highly valued (Davis and Ruhe 2003). It is expected that, in a collectivistic society, people are inclined to violate the law in order to support their own group based on unquestioning loyalty. Therefore, corruption may increase. Various studies document that the less individualistic (more collectivistic) a society is, the higher the level of corruption (Davis and Ruhe 2003; Murdoch 2009; Halkos and Tzeremes 2011; Tong 2014). Thus, another secondary hypothesis can be stated:

  • Hypothesis 1.2. The less individualistic (more collectivistic) a society is, the higher the level of corruption.

The MAS dimension of culture refers to the concern of a society for achievement, heroism, assertiveness and material rewards for success (masculinity) or for cooperation, modesty, caring for the weak, and quality of life (femininity) (Hofstede 1980; The Hofstede Centre 2015). In a survey conducted on 42 countries, Davis and Ruhe (2003) empirically found a significantly positive relationship between masculinity and corruption. They concluded that in countries with the highest score in masculinity, people prefer to receive money, titles, or other material or status rewards, thus enhancing the level of corruption. In Venezuela, Gonzales-Fabres (1996, p.60) found that the high corruption is motivated by “personal accumulation of riches.” Husted (1999) empirically found a significant relationship between corruption and masculinity and associated high corruption with high earnings, recognition, advancement and challenging work. After investigating the explanations for different levels of corruption in countries from Scandinavia and Africa, McLaughlin (2013) shows that some cultural variables like masculinity play a role in determining the level of corruption. We might conclude that a preference for material rewards creates the framework for extending corruption practices. The third secondary hypothesis can be stated as:

  • Hypothesis 1.3. The greater the masculinity of a society, the higher the level of corruption.

The UAI dimension of culture expresses “the degree to which the members of a society feel uncomfortable with uncertainty and ambiguity” (The Hofstede Centre 2015). In a society with a high level of uncertainty avoidance, corruption can be viewed as a mechanism to reduce uncertainty in order to obtain more certain results (Husted 1999). By performing a large meta-analysis of the literature. Tong (2014) conclude that “in low UAI countries such as China, ambiguity and adaptability regarding laws and rules to fit situations make corruption more likely.” Therefore, for our purpose, the following fourth secondary hypothesis is stated:

  • Hypothesis 1.4. The greater the level of uncertainty avoidance, the higher the level of corruption.

The LTO dimension of culture refers to “how every society has to maintain some links with its own past while dealing with the challenges of the present and future” (The Hofstede Centre 2015). This dimension is also called by Hofstede as “Confucian dynamism” and deals with a choice between two contrasting poles: long-term orientation versus short term orientation. The first pole is a positive one and reflects a dynamic and future-oriented mentality whereas the second pole is a negative one and shows a more static and tradition-orientated mentality (Fang 2003). Hofstede (1997) characterized the long-term orientation by “persistence, ordering relationships by status, thrift and having a sense of shame” and the short-term orientation by “personal steadiness, protecting face, respect for tradition, reciprocation of greetings, favors and gifts.” Under these assumptions we may expect that a short-term orientation may enhance the need to ask for gifts and favors in order to get immediate benefits. Therefore, the fifth hypothesis can be introduced as follows:

  • Hypothesis 1.5. The shorter the term of orientation, the higher the level of corruption.

The IND dimension of culture refers to the preference of a society to allow relatively free gratification of basic and natural human drives to enjoy life, in contrast to suppressing the gratification of needs and regulating it by means of strict social norms (The Hofstede Centre 2015). A high score of indulgence would mean a society that exhibits a willingness to realize their impulses and desires with regard to enjoying life and having fun. This society highly values leisure time and spending a lot of money. Individuals in a restrained society are restricted by social norms, breaking away towards incentives for both demanding and offering, even some illicit private payments. In this context, we test the following sixth hypothesis:

  • Hypothesis 1.6. The more indulgent the society is, the lower the level of corruption.

Methodology and Data

The first step is to test the aforementioned working hypotheses in order to identify the impact of culture on the level of corruption. Correlation coefficients are used along with ordinary least square (OLS) multiple regression analysis and analysis of variance (ANOVA). The level of corruption was determined using the Corruption Perceptions Index report (2014), provided by Transparency International (2015). This index was used to measure the perceived levels of public sector corruption in 175 countries. The scores range from 0 (highly corrupt) to 100 (corruption free, or very clean). The ranked countries range was from 1 (lowest level of corruption) to 175 (highest level of corruption). Perceptions of corruption were dealt with when the corruption term was used.

For culture, Hofstede’s cultural model was used based on six dimensions. Each dimension places the culture of a nation on a scale from 0 to 100. At the time of our research, Hofstede’s model was applied to 100 countries (The Hofstede Centre 2015).

The hypotheses were tested using a sample of countries for which both data for Hofstede’s culture dimensions and the level of corruption were available. This resulted in an initial sample of 98 countries.

Results and Discussion

Table 1 shows the pairwise Pearson correlations between the variables of corruption and dimensions of culture. Table 1 and Figs. 1, 2 and 3 reveal that the highest correlations (which were significant at the 1 % level of significance) between culture and corruption were found in relation to three dimensions of the culture, namely PD, IDV and LTO. Also, a medium and negative correlation was observed between PD and IDV (r = −0.656). Such moderate levels do not affect the assumptions of OLS regression.

Table 1 Pearson correlation
Full size table
Fig. 1
figure 1

Correlation between the PD dimension of culture and corruption

Full size image
Fig. 2
figure 2

Correlation between the IDV dimension of culture and corruption

Full size image
Fig. 3
figure 3

Correlation between the LTO dimension of culture and corruption

Full size image

Hypothesis 1.1 examines the relationship between power distance and the level of corruption. The higher the power distance, the higher the level of corruption expected. The average power distance of the sample countries is 64. Countries with the highest power distances are Malaysia (100), Saudi Arabia, Iraq and Guatemala (95), Russia (93), Albania and Kuwait (90). Countries with the lowest power distance are Austria (11), Denmark (18), New Zealand (22), and Norway (31). Figure 1 provides in graphical form the correlation between the PD dimension of culture and corruption. From Table 1, it can be seen that the correlation coefficient reflects a positive and medium correlation (r = 0.585), statistically significant at the 1 % level of significance. The result remains significant for the initial regression (Table 2) and controlling for another cultural variable in the multiple regression analysis (Table 3). A positive relationship was found between power distance and level of corruption, which was highly significant (p < 0.000). Variation in corruption may be explained by power distance (33.5 %). Therefore, hypothesis 1.1 is accepted. Our results are also supported by the findings of Husted (1999); Davis and Ruhe (2003); Halkos and Tzeremes (2011); McLaughlin (2013) and Tong (2014) who also found a positive correlation between power distance and corruption.

Table 2 Descriptive statistics and results of the simple regression analysis
Full size table
Table 3 Models of corruption as a function of culture
Full size table

Regarding hypothesis 1.2, a highly score of the IDV dimension indicates a highly individualist society in which law is respected. The average score for IDV is 39. Countries with the highest score on this dimension are USA (91), Australia (90), UK (89), Netherlands (80), New Zealand (79), and Denmark (74), while countries with the lowest score are Guatemala (6), Ecuador (8), Panama (11), Venezuela (12), and Colombia (13). Figure 2 reveals the correlation between the IDV dimension and corruption. From Table 1, a negative and medium Pearson correlation (r = −0.613) was found significant at the 1 % level of significance. Therefore, the more individualistic a society is, the lower the level of corruption. Tables 2 and 3 also show a negative relationship between individualism versus collectivism and the level of corruption. In both cases, the results are statistically significant (p < 0.001 and p < 0.05). For individualism-collectivism, 37 % of the variation in corruption is explained. The results revealed that hypothesis 1.2 is accepted, which indicates that the less individualistic (more collectivistic) a society is, the higher the level of corruption. These results are in line with those of Davis and Ruhe (2003); Halkos and Tzeremes (2011) and Tong (2014), who also highlighted the main role of the collectivistic society and social network in fostering corrupt acts.

Hypothesis 1.3 examines whether the MAS dimension of culture would influence the level of corruption. A high score means a high masculine and a low feminine society. The average of the masculinity-femininity index is 47. The most masculine societies are Slovakia (100), Japan (95), Hungary (88), and Austria (79) while the most feminine societies are Sweden (5), Norway (8), and Latvia (9). Table 1 points out a weak and positive correlation between these two variables (r = 0.162) and they are not statistically significant. By running simple and multiple regressions (Tables 2 and 3), the positive influence of masculinity was found at the corruption level and is not statistically significant. All the results reject hypothesis 1.3, which means that the more masculine societies are not perceived to have a higher level of corruption than the more feminine societies. The variation in corruption due to masculinity versus femininity is also very low (only at 1.6 %) resulting in the same conclusion. These results are in line with those of Tong (2014) for China, but disagree with those of Husted (1999) and Davis and Ruhe (2003) conducted on 50 and 42 countries, respectively.

Hypothesis 1.4 investigates whether the level of UAI explains the level of corruption. A high level of UAI means high concern on UAI, and in this context, a high level of corruption in getting more certain results. The average score for UAI is 63. The highest score of UAI was recorded in Greece (100), Portugal (99), Guatemala (99), Uruguay (99), and Belgium (94) reflecting a high preference for avoiding uncertainty and the emotional needs of rigid rules and codes of faith and conduct. Innovation may be rejected in business and safety is an important element in individual motivation (Hofstede 1997). The lowest scores for this dimension are registered in Singapore (8), Jamaica (13), Denmark (23), Sweden (29), and Hong Kong (29). Also, countries that show a low risk of aversions and innovative ideas of diversity are being promoted. Table 1 shows a positive and very low correlation coefficient (r = 0.048), which is not statistically significant. Furthermore, by running a simple regression analysis (Table 2), and a multivariate regression analysis (Table 3), the results are also not statistically significant. These results do not support hypothesis 1.4, so the UAI in a given society failed to explain the level of corruption for that society. All the same, the results are in line with those of Davis and Ruhe (2003), but contradict the findings of Husted (1999) or Tong (2014).

Hypothesis 1.5 analyzes whether the LTO dimension of culture may explain the level of corruption. A high score on LTO indicates a high-long term orientation of this society. The average score of LTO for our sample is 41. A high score means that societies value tradition and long-term commitments, while a low score reflects the desire to embrace change, which is not hampered by tradition (Hofstede 1997). According to Hofstede’s cultural model, countries with the highest LTOs are South Korea (100), China (87), Japan (88), Germany (83), Belgium, Lithuania, and Estonia (82), while Honduras (8), Nigeria (13), and Ghana (4) have more short-term orientations. Figure 3 graphically depicts the correlation between LTO and corruption. Table 1 shows a negative correlation between LTO and corruption (r = −0.344), which is statistically significant (p < 0.01). The linear and multiple regressions in Tables 2 and 3 reveal a negative influence of LTO and corruption, which is statistically significant (p < 0.001). Approximately 10.8 % of the variation in corruption can be explained by LTO. The lower long-term orientation (meaning the higher short-term orientation) is, the higher the level of corruption. Thus, corruption increases under a short-term orientation, confirming our hypothesis. However, for China, Tong (2014) found opposite results. He found the long-term outlook fosters the importance of maintaining relationships, which may involve corrupt activities. But, various studies conducted in China contradict this economic theory. For instance, Teixeira et al. (2016, p. 71) and Jiang and Nie (2014) talked about “China’s miracle” of continuing high GDP growth by the prevalence of government corruption.

Hypothesis 1.6 examines whether a more indulgent society may lead to a lower level of corruption. A high IND score indicates a high indulgence and low restraint society. The average IND score of our sample is 48.60. Countries with the highest IND score are Venezuela (100), Malaysia (97), El Salvador (89), and Angola (83), reflecting very low impulse control compared to countries like Egypt (4), Latvia (13), Lithuania (16), Albania (15), and Hong Kong (17), nations with more restrained actions and manifestations of pessimism and cynicism. In Table 1, the correlation coefficient was found to be low and not statistically significant (r = −0.165). The level of significance of the coefficient for IND from linear regression (Table 2) is also not statistically significant. However, when controlling for the other cultural variables, the influence became statistically significant (p < 0.05) (Table 3, model 1 and model 2). Given mixed results and inconclusive findings, 1.6 cannot be supported. A strong relationship could not be identified between indulgence-restraint and corruption. Our results are in line with those of Tong (2014), who also found that indulgence-restraint is not associated with corruption.

Generally, hypothesis 1 investigates whether cultural dimensions explain the level of corruption. Table 3 shows that culture expressed by all dimensions of Hofstede’s model explains a statistically significant level (p < 0.001) of corruption around the world. In Model 1, corruption is a function of culture with 52% of the variation in corruption explained by culture (in all six dimensions) of a nation (adjusted R-square = 0.52). If the non-significant dimensions of MAS and UAI are removed, the remaining four cultural dimensions explain 51 %, of the variation in corruption (Model 2). Model 2 appears to have better fit than Model 1, as shown by the higher Fisher's test value. Model 3 is obtained by also removing IND which has some shortcomings given the lack of statistically significant results (Tables 1 and 2) and a low adjusted R-square value. The proportion of variation in corruption explained by the three cultural factors (PD, IDV and LTO) also remains high (49 %), but in this case, the Fisher's test (F = 28.085), indicates that this model has the best fit. Also, looking at the adjusted R-square values of the variables (Table 2), PD, IDV and LTO play major role in determining the corruption levels of countries. All three models of corruption as a function of culture are statistically significant (p < 0.001), hence hypothesis 1 is supported.

Conclusions

Various studies have documented that many economic behavioral acts are better explained by investigating subjective factors, as well as the role played in national cultures. This study investigated whether culture plays a major role in determining the levels of corruption in countries. The cultural dimensions of Hofstede’s model are used. The worldwide level of corruption is calculated as a perceived level of corruption by the Corruption Perceptions Index.

First, three of the six main components of the Hofstede’s model were found to have a significant influence on corruption. They are: PD, IDV, and LTO. The statistical influence of MAS, UAI, and IND were rejected by the statistical tests.

Second, the results of this study reveal that 35 % of the variations in corruption is explained by PD. A higher PD culture involves a higher level of corruption. It is a hierarchical society, in which employees acknowledge the legitimate power of the boss. To maintain the loyalty of subordinates, superiors can ask for a bribe as a prerequisite for their position.

Third, IDV explains 37 % of the variation in corruption. The more collectivistic the society is, a higher the level of corruption. In a collectivistic society, a network of friends and family creates lasting relationships which could stimulate corrupt behaviors.

Finally, LTO, as a pattern of national culture, may also explain the level of corruption, but the proportion explained is somewhat lower, 10.8 %. Corruption significantly increases under a short-term orientation supporting our hypothesis. This can be explained by the fact that, a short-term orientation culture is concerned with the need to ask for favors and gifts in order to get immediate benefits.

In total, our main hypothesis is accepted, which implies that cultural factors affect the level of corruption. About half of the level of corruption in countries can be explained based on national culture. These findings suggest that national culture may help to explain the level of corruption from each of the countries. Therefore, this research may have significant social implications for policy-makers who are seeking ways to reduce the level of corruption, in order to safeguard the national economy. Governments should acknowledge the role of culture and adopt the most appropriate decisions in policy reforms undertaken to fight corruption.

A limitation of this research is the use of only one measure of culture as provided by Hofstede’s model. Future research needs to focus on the culture-corruption nexus and use other data sources for culture (for example, World Value Survey data, www.worldvaluessurvey.org/wvs.jsp). Also, this study suggests further investigation of factors which create incentives for corruption, such as trust, religion or subjective well-being.