The effect of the managers' characteristics on
corrupt behavior
Sander Oosterwechel
S1555332
s.w.oosterwechel@student.rug.nl
Supervisor: Dr. G. de Jong
January 10, 2014
Master Thesis for IB&M
2
Abstract
This thesis investigates which individual-level determinants characterize whether or not a firm will pay a bribe. The relationship between work experience, education level and the gender of the top manager on bribery is estimated using firm-level data from Indonesia and the Philippines. With a total of 1986 observations, a Logit regression is used to empirically test the relationship. The analysis draws on social capital theory, top management team theory and social role theory. The findings provide significant support for the positive linear relationship between education level and bribery. Furthermore, we find that male managers bribe significantly more than female managers. We find no relationship between work experience and bribery.
3
Table of Contents
1. Introduction ... 4 2. Literature review ... 7 2.1 Corruption ... 7 2.2 Measuring corruption ... 9 2.3 Individual-level bribery ... 10 2.4 Hypotheses ... 122.4.1 Work experience and education level ... 12
2.4.2 Gender ... 15 3. Methodology ... 16 3.1 Data source ... 16 3.2 Countries ... 18 3.3 Sample ... 19 3.4 Dependent variable ... 19 3.5 Independent variables ... 20 3.6 Control variables ... 20 3.7 Statistical model ... 22 3.8 Method assumptions ... 22 4. Empirical results ... 26 4.1 Descriptive statistics ... 26 4.2 Regression results ... 27 5. Robustness tests ... 31
5.1 Additional analysis of bribery ... 33
6. Conclusion ... 34
6.1 Added value ... 35
6.2 Limitations and suggestion for further research ... 37
Appendix 1: Survey questions ... 39
Appendix 2: Homoscedasticity ... 41
Appendix 3: Normality ... 42
Appendix 4: Goodness-of-fit test ... 43
Appendix 5: Frequency statistics ... 44
4
1. Introduction
'It was about keeping the business unit alive and not jeopardizing thousands of jobs overnight'
These are the words of Reinhard Siekaczek, a former midlevel executive at Siemens A.G. According to the New York Times, together with several others, he arranged a rainfall of payments that eventually streamed to well-placed officials around the world. In 2008, Siemens A.G. ended up paying the largest fine for bribery in modern history, a total sum of $1.6 billion. Siemens A.G. isn't the only company that is convicted for bribery. Other examples of companies that are convicted for bribery are; KBR/Halliburton (U.S., $579 million fine), BAE Systems (U.S., $448 million fine), Snamprogetti (the Netherlands, $240 million fine), Technip S.A. (France, $240 million fine)1. As can be concluded from the examples, the firms that receive the highest fines for bribery are from the developed countries. This shows that bribery is still a universal problem and can happen anywhere. However, there is a widespread difference between countries regarding the amount of bribery (Bribe Payers Index, 2011). The sentence of Siemens A.G. give thought about who is bribing and which factors determine the amount of bribes, since the bribing culture seems to be widespread within the Siemens A.G. organization. Current research into corruption has mainly focused on differences in corruption between countries, by using country-level data. However, lately increasing bodies of research focus on the firm-level and individual-level of bribery, using firm-level data.
It is often assumed that corruption has damaging effects on the economic growth of a country; however research on this topic find mixed results. Leff (1964), for instance, argues that corruption might have a positive impact on economic growth if the right circumstances apply in a country. However, other scholars find a negative relationship between corruption and economic growth, arguing that corruption reduces the amount of economic growth, due to a delay in private investment (Mauro, 1995; Shleifer & Vishny, 1993). Political violence, an inefficient government and a weak rule of law further increases the negative influence of corruption on investment (Méon & Sekkat, 2005). The precise impact of corruption on economic growth remains still vague, since some highly corrupt countries show very high foreign direct
1
5
investment rates and economic growth rates. Scholars are not able to find a clarification for this phenomenon.
According to Svensson (2003), country-level research can explain differences in corruption levels between countries, but not the differences in corruption levels within countries. In his research he claims that firms end up paying different amount of bribes, even if firms are facing comparable policies and institutions. In order to understand these differences in bribery between firms, the focus of the research has shifted towards firm-level research. Studies on a firm-level of bribery focus on firms that offer bribes, and try to explain differences by taking into account the institutional, cultural and country differences (e.g. Spencer & Gomez, 2011; Mocan, 2004; Tonoyan, Strohmeyer, Habib & Perlitz, 2010). Recently, more and more research focus on what differentiates bribing firms from non-bribing firms and why some firms are more involved with bribery than other firms (e.g. Swamy, Knack, Lee & Azfar, 2001; Martin et al., 2007; Jeong & Weiner, 2012; de Jong et al. 2012; Lee, Oh & Eden 2010; Svensson, 2003; Wu, 2009;). However, research on the individual level, that of the manager of the firm, didn’t receive much attention in the literature. Researchers studied the influence of personal characteristics of the manager, e.g. gender, income, education, work experience and marital status on the amount of bribes paid by the manager (Swamy et al, 2001; Hunt & Lazio, 2012; Mocan, 2004; de Jong et al. 2012). The manager of the firm is the actual person who is performing the bribe and therefore studying the individual characteristics of the manager is an interesting point of research.
6
Researching corruption in a transition economy on the individual-level would therefore be an interesting starting point of research. This study will focus on the characteristics of the individual manager on the amount of corruption of firms operating in a transition economy. The characteristics that are researched are work experience, education level and gender. Not much research is performed on the influence of these characteristics on corruption and is therefore an underexposed topic and an interesting topic to research. The first characteristic is the work experience of the manager. Literature argues that managers with little work experience are more likely to engage in high-risk activities, such as bribery. However, more work experience might lead to a better recognition and exploitation of opportunities and reduce uncertainty about the value of opportunities, like the opportunity to bribe. However, extensive research on work experience and bribery is lacking. The second characteristic is the education level. Current research is divided on the influence of education level on the amount of corruption. Country-level research claims that a high education Country-level is associated with a smaller amount of corruption. On the individual level however, it appears to be the opposite; high educated managers can better recognize the movements when to bribe and how to bribe and are therefore more corrupt (de Jong et al., 2009). This study will provide a better insight in the relation between education level and bribery on an individual-level. Researchers on gender and corruption often take the perspective of the demand side of bribery i.e. the difference between male and female public officials regarding bribery. Taking this perspective it is argued that females are less corrupt than males (Swamy et al., 2001). Researching corruption and gender from the supply side, is an underexposed topic and would therefore be worth researching. This research will contribute to a better understanding of why some managers bribe more than others. The main research question of this thesis is:
Why are some managers more corrupt than others?
The following research questions will be addressed in order to answer the main research question:
What is corruption and how can it be measured?
What is the relationship between the work experience of the top manager and
7
What is the relationship between the education level of the top manager and corruption? What is the relationship between the gender of the top manager and corruption?
In addition, this study will measure the interaction effect between work experience and education level. The data that is used for this empirical research comes from the Business Environment and Enterprise Survey (BEEPS). The aim of this research is to get a better insight in the influence of the individual characteristics of the managers on the amount of corruption. This can help managers to work in a more ethical matter in transition economies and provide awareness for managers how to deal with corruption. Three existing theories, namely Human Capital theory, Top Management Team theory and Social Role theory will be used to supplement the development of the hypotheses.
2. Literature review
2.1 Corruption
8
legislators. Probably the most common type of corruption is bureaucratic corruption, since it involves the payment of bribes and gifts to receive for example a service or approval. Weber & Getz (2004) make a distinction between corruption and bribery; corruption is the misuse of power for private benefit, and can include a broader range of corrupt activities, such as money laundering, fraud, abuse of insider information, gift giving etc. A bribe is a payment to persuade a person to act in a way that is contrary to his responsibilities. According to Weber & Getz (2004) the concepts of bribery and corruption are closely related phenomena and are used together in this thesis. According to Transparency International (2009), bribing to win public contracts, avoiding regulations or accelerating services is a persistent concern in the world; Two out of five respondents has been asked to pay bribes to get things done. In emerging and developing countries, the occurrence of paying bribes is even higher.
Literature on the consequences of corruption for firms find mixed results. Wu (2009) argues that corruption lead to higher cost for many firms. The financial costs will increase, due to an increased risk premium. Secondly, the costs of goods and services will increase due to bribe payments. De Jong et al. (2012) argue that bribes are subject to diminishing returns. According to Transparence International (2012), corruption distorts markets and creates unfair competition. Furthermore, firms have an unfair unwarranted advantage in securing their business, due to corruption (D’Souza, 2012). Therefore many countries are trying to set up anti-corruption strategies. On the other hand, some scholars find positive consequences of bribery for firms. Leff (1964) and Mauro (1995) argue that government officials that oblige bribes are found to work harder. Furthermore, they argue that corruption avoid bureaucratic delays, because corruption can act as a grease payment. Svensson (2003) argues that, even if firms are facing the same institutional environments, they may end up paying different amount of bribes, meaning that firm characteristics and the characteristics of the individual manager may influence the amount of bribes that are paid and the size of the bribes. The predictability of corruption, (e.g. if a person seeking favors from the government also receives that favor), influence the level of investments. The high growth rates in some Asian countries are contributed by the clear organization of corruption and the high predictability of corruption.
9
countries with more developed institutions. According to North (1994), the institutions are made up of informal constraints, formal constraints, and their enforcement characteristics. The institutions define the incentive structure of societies. Corruption can be transformed into social behavior and social norms and can be seen as an informal institution (Aidis, Estrin & Mickiewicz, 2012; Jain, 2001). Due to the different institutional pressures and conditions that an individual firm is facing, bribery can vary significantly (Martin, Cullen, Johnson & Parbooteah, 2007). Spencer & Gomez (2011) argue that managers are more likely to conform to the social expectations to acquire legitimacy when the managers observe that the corrupt practice has become institutionalized. Contrary, government officials may impose more pressure on firms to engage in bribery, when corruption is taken for granted (Spencer & Gomez, 2011). It could be concluded that it depends on the context whether or not bribery is seen as harmful. In many emerging markets bribery is daily practice, while in the Western parts of the world it is seen as a problem. Steidlmeier (1999) argues that one has to consider whether it is a rule of behavior or an individual act, in order to judge whether bribery is wrong. As an individual act is not accepted as a rule of behavior, one cannot achieve legitimacy of the corrupt practice.
2.2 Measuring corruption
Since corruption happens outside the scope of the public and is hidden, it is difficult to measure it by nature (Jain, 2001). Studies that measure corruption do so on a macro-level, firm-level or individual-level. Furthermore, corruption can be measured from the perspective of the bribe-payer and bribe-taker. Research on the determinants of corruption has three common features: research uses data on corruption derived from perception indices, it explains corruption as a function of countries’ policy-institutional environment and it is based on cross-country analysis. These determinants allow for a measure on corruption on a macro-level. However the within country variation of corruption cannot be explained by macro level research (Svensson, 2003). In order to explain why some managers bribe and other managers do not while operating in the same institutional environment, an individual level measurement of corruption is required. Furthermore, micro-level empirical research helps to understand the likely heterogeneity of bribery within countries (De Jong et al., 2012).
10
individual-level. In order to increase the credibility of the answers and trust between the manager and the interviewer, the questions on corruption are asked at the end of the interview. Furthermore, in order to avoid the suggestion of bad behavior, the questions are formulated in an indirect manner. Finally, questions on corruption are asked in different sections, to improve the reliability of the answers (Reinikka & Svensson, 2006). Regarding the sensitiveness of the topic and the possibility to protect oneself from possible government reprisals, it could be difficult to measure corruption (Jensen, Li & Rahman, 2010). This difficulty holds for measuring corruption on the macro-, firm-, and individual-level; hence the answers may be subject to false response or non-response. Concerning the large amount of respondents that refuses to answer or do not know whether or not the firm bribes, the BEEPS data is subject to this problem as well.
The BEEPS survey measures three different but connected aspects of corruption, that is; state capture, administrative corruption and the influence on the formation of rules, regulation and laws. State capture measures the extent to which firms make payments to public officials in order to influence the formation of laws, rules and regulations. Administrative corruption measures “the extent to which firms make illicit and non-transparent private payments to public officials in order to alter the prescribed implementation of administrative regulations placed by the state on the firm’s activities” (Hellman, Jones, Kaufmann, 2000:5). The last aspect measured by the BEEPS is the extent to which firms have an influence on the formation of rules, regulation and laws, without making private payment to public officials. The advantage of the BEEPS is that it measure corruption on a firm- and individual-level, and therefore allows for research on the individual-level and take intra country differences into account.
2.3 Individual-level bribery
11
requests and the vulnerability of the manager to bribery (Jeong & Weiner, 2012; Lee et al, 2010). Other firm characteristics that are essential determinants of corruption are corporate governance, firm size, firm age, level of internationalization, level of competition, and embeddedness in networks (Wu, 2009).
According to Li & Ouyang (2007), firms cannot operate or behave without individual people acting on its behalf. Consequently, the characteristics of the individual managers influence the different pressures that the firm is facing to engage in corrupt activities, even for firms operating in the same environment. For instance, Hunt & Laszlo (2012) argues that the income level of the individual can affect the incidence and amount of bribery. Rich individuals tend to bribe more, mainly because they bribe more officials to get things done. Nevertheless, individuals with a low level of income use a greater share of that income for bribing activities, meaning that the poor are hit much harder than the rich, since the bribes are unequally distributed. Research on gender and bribery shows that women are less likely to participate in corrupt activities than men (Swamy et al, 2001). Torgler & Valev (2006) claim that women are more willing to comply and are less likely to agree that corruption can be justified. Research on bribery and education level reveal that education increases a person's stock of information and skills, what lead to a better recognition and pursue of an entrepreneurial opportunity, including the opportunity to bribe (Marvel & Lumpkin 2007). De Jong et al. (2009) find empirical support for this argument by researching entrepreneurs’ education level in Vietnam. Moreover, it is argued that managers with little work experience are more likely to engage in high-risk activities, like bribery (de Jong et al., 2009).
12
2.4 Hypotheses
2.4.1 Work experience and education level
A difference may exist between work experience, education level and bribe payment of the top manager. This difference can be explained by using human capital theory. The importance of human capital was raised as in early as Adam Smith’ The Wealth of Nations in 1776. In his work he reasons that an investment in capital enables an increase in future productivity. According to his theory, the role of education and training are important determinants of earnings and individual productivity. Mincer (1958), Schultz (1961), and Becker (1964) emphasized the importance of human capital, and contributed to the development of the human capital theory. However it was Becker who truly established the conceptual framework of the human capital theory. In his framework he formalized educational choices as rational choices of optimizing agents. According to his theory, the present value is the earnings expected from education, compared to its related costs, over a life-cycle period. In his view education and training are investment, similar to the purchase of new equipment.
Bribery involves uncertainty and ambiguity and is therefore a complex strategic tool. In order to offer the appropriate bribe to the appropriate official, the manager must select the appropriate information. For a manager to be able to bribe, he must recognize the opportunity to bribe. Since corruption happens outside the scope of the public and is hidden, it may be difficult for a manager to discover the opportunity to bribe. In this sense, opportunity recognition and bribery are closely related phenomena. Rondstadt (1988) argues that an individual’s unique possession of prior knowledge creates a knowledge corridor that allows him or her to identify certain opportunities but not others. In general, managers will discover only those opportunities directly related to their knowledge (Venkataraman, 1997). Each person’s accumulation of human capital is different, because the knowledge is generated through an individual’s unique life experiences.
13
Work experience
According to Becker (1964), experience includes work experience, practical on the job learning and non-formal education, such as training. Additionally, human capital is theorized to increase by more depth in work experience and more broad experience across markets.
Fiet (1995) observed that previous work experiences set the context in which an entrepreneur decides whether to invest in a particular venture; in effect, gathered experience increases alertness by providing more cues about when to take action. Shane (2003) indicates that business experience, functional experience, and industry experience are all useful for the discovery and exploitation of opportunities. Through work experience, people develop skills and information that facilitate the formulation of entrepreneurial strategy, the acquisition of resources, and the process of organizing. Experience increases a person’s human capital and reduces uncertainty about the value of opportunities. Breadth in work experience provides access to new information that facilitates opportunity recognition. Recognizing opportunities is often like solving puzzles, because a new piece of information is often the missing element necessary to provide meaning or trigger an entrepreneurial assumption (Shane, 2003). The more diverse a person’s experiences, the greater number of puzzle pieces they can draw upon (Casson, 1995). In addition, people often learn about entrepreneurial opportunities through participation in markets (Casson, 1982). Furthermore, variation in market experience provides access to different types of information that may be useful in the discovery process. Consequently, participation in more markets should increase the likelihood that a person will gain access to the information that is needed for opportunity recognition. More depth in work experience and breadth in work experience lead to a higher prior knowledge, hence a higher chance for the manager to recognize opportunities. Furthermore, work experience reduces uncertainty about the value of opportunities. Since corruption happens outside the scope of the public and is hidden, it may be difficult for a manager to discover the opportunity to bribe. Work experience increases the chance to recognize the opportunity to bribe.
14
more risks and pursuing a more innovative growth strategy (Grimm & Smith, 1991). For this reason, it is more likely that managers with little experience are more likely to engage in high-risk activities, like bribery (de Jong et al, 2012). Based on TMT theory, it could be argued that managers with little work experience are showing higher level of corruption. However, this amount will decrease over time, since they are less involved in high risk activities. When managers become more experienced, the level of corruption will increase again, since they can better recognize and pursue corrupt activities. This lead to the following hypothesis:
H1: There is a u-shaped relationship between work experience and the likelihood of bribery
Education level
Another central characteristic of the human capital theory is the education level. Most studies use formal education or years of schooling as a measure of human capital and have accepted the notion that 'more is better' (Marvel & Lumpkin, 2007). On a macro-level, studies have found that countries with higher level of education are positively associated with lower level of corruption (Ades & Di Tella, 1999; Treisman, 2000). Supported by the argument that a more educated society would expected to tolerate less bribes (Rest & Thoma, 1986), the association has been interpreted as proof that education decreases corruption. However, on the individual-level of bribery, higher educated managers are likely to pay more bribes than lower educated managers. Bribery involves uncertainty and ambiguity and is therefore a complex strategic tool. In order for a manager to bribe, he must offer the right bribe to the right official at the right time. Well-educated managers are endowed with superior cognitive skills, awareness levels, and decision-making capabilities. Besides, education increases a person's stock of information and skills, what lead to a better recognition and pursue of an entrepreneurial opportunity, including the opportunity to bribe (Marvel & Lumpkin 2007). De Jong et al. (2009) find empirical support for this argument by researching entrepreneurs’ education level in Vietnam. This lead to the following hypothesis:
15
2.4.2 Gender
This section will use Social Role theory to explain gender differences regarding corruption. According to social role theory, the historical division in labor between men and women has led to sex difference in social behavior (Eagly, 1987). Men are often assumed responsibilities outside the home and women are often expected to have responsibilities at home. The expectancies about different roles are transferred to future generations and, in turn, impinge on the social behavior of each gender (Eagly, 1987, 1997; Eagly, Wood, & Diekman, 2000). The different expectancies lead to sexual stereotypes (Williams & Best, 1982), what in turn lead to different behavior between men and women. Men are showing more masculine traits, while women are showing more feminine traits. According to Eagly & Karau (1991) females develop traits that manifest communal or expressive behavior, while man develop traits that manifest agency. Communal traits entail the tendency to be caring, friendly, emotional and unselfish. In a communal relationship, a benefit is given in response to a need or to demonstrate a general concern for another. Agency relates to traits such as the inclination to be decisive, independent, assertive and competent (Eagly & Karau, 1991). Agency can be defined as the capacity for human beings to make choices and to impose those choices on the world. Empirical research by Eagly & Johnson (1990), Wright (1988), Rubin (1985) and Feingold (1994) support the argument of the development of different traits between male and female. It could be argued that due to the development of different traits, males often occupy leadership roles. Therefore, it is often assumed that leadership demands for an agentic personality. Consequently, the male character is assumed to better align the leadership role (Peters, Kinsey, & Malloy, 2004). Due to the development of different traits between male and females, it is likely that a difference exist in the amount of bribes that are paid by the different sexes. Corruption is a complex strategic tool, since it involves ambiguity and uncertainty. Since male managers develop agentic traits that involve the capacity to make choices and impose those choices on the world, it is expected that male top managers will be more often involved in bribery activities. The communal trait could withdraw a female manager from participating in bribery activities.
16
USA, Canada, Australia, UK and many European countries, women comprise less than fifteen percent of corporate board members. In some Asian countries, the percentage is as low as two percent. The lower presence of females in top management positions lead to a dissimilar exposure to corruption among gender. A higher exposure to corruption leads to a different attitude towards corruption and a transformation of the corrupt practice into social behavior and social norms. Since male managers are more often exposed to corruption, it would be likely that they develop a different attitude towards corruption and develop social norms and behavior that are more acceptable towards corruption. The arguments lead to the following hypothesis:
H3: Female top managers are likely to pay less bribes then male top managers
The social role theory has some limitations. According to Diekman, Goodfriend, & Goodwin (2004) women are gaining more access to positions that are associated with power and therefore the social role seems to be changing. Furthermore, gender differences in values, emotions and personality are more pronounced in European and North American countries, relative to Asian counties (Guimond, 2008). Additionally, social stereotypes do not always affect the behavior or decisions of individuals (Sczesny and Kuhnen, 2004).
3. Methodology
3.1 Data source
17
“It is said that establishments are sometimes required to make gifts or informal payments to
public officials to “get things done” with regard to customs, taxes, licenses, regulations, services etc. On average, what percent of total annual sales, or estimated total annual value, do establishments like this one pay in informal payments or gifts to public officials for this purpose?“ (Enterprise Surveys, Questionnaire Codebook, 2012 ).
In order for a firm to be interviewed the World Bank made some restrictions. First of all, state-owned firms are excluded from the survey. Furthermore, the firm must have five or more employees and must be formally registered. The business owners and managers are the ones who are interviewed. The number of surveys that are conducted depends on the size of the economy, ranging from small, medium-sized and large economies. The Enterprise Survey make a distinction between manufacturing and service firms, by asking questions only applicable for only the service firm or the manufacturing firm. However, for the most part the questions for the different modules are the same. In order to assure more reliable and honest answers of the participants regarding the sensitivity of some questions, private contractors are used instead of government workers.
18
3.2 Countries
This study focus on two large economies in Southeast Asia, namely Indonesia and the Philippines. On the Corruption Perception Index (CPI) 20122 Indonesia and the Philippines are ranked respectively 118 and 105 on a total of 174 countries, with 1 being less corrupt and 174 being very corrupt. The scores given by the CPI are 32 for Indonesia, 34 for the Philippines, meaning that these countries are very corrupt. The countries are located in one of the fastest growing regions of the economy, and studying the effects of corruption in those countries makes it even more interesting. Indonesia and the Philippines are members of the Association of Southeast Asian Nations (ASEAN)3 consisting of ten countries with the goal of fostering economic, cultural and political collaborations. All member countries have high corruption levels, except for Singapore.
Indonesia is an emerging market economy (IMF) and with a GDP per capita of $3,557 in 2012 (World Bank), it has the largest economy in Southeast Asia. Indonesia is classified as a newly industrialized country and is a member of the G-20 major economies. The annual GDP growth in Indonesia is high, with a growth of 9.3 in 2005. The worldwide economic crisis influenced the growth of the GDP in 2009 a little, but was still growing with 4.6%. In 2010, 2011 and 2012 the annual growth is above 6%. The development of the annual economic growth by GDP is given in table 1. The Indonesian government plays a significant role through the ownership of state-owned enterprise and administration of prices of a range of basic goods, such as rice, fuel and electricity. The Philippines is also considered an emerging market (IMF), with a GDP per capita of $2,58 in 2012 (World Bank). The annual GDP is growing from 2003 till 2007 with more than 4.8% per year. In 2009 the economy grows with only 1%, due to the worldwide economic crisis however in 2010 it was growing again with 7.6%. A total of 1444 managers and company owners are surveyed by the BEEPS in Indonesia and a total of 1326 respondents from the Philippines. Both surveys are conducted in 2009.
2
www.transparency.org
19
Table 1: Annual GDP growth (%)
2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 Indonesia 4.8 7.8 9.3 5.5 6.3 6.0 4.6 6.2 6.5 6.2 Philippines 5.0 6.7 4.8 5.2 6.6 4.2 1.1 7.6 3.6 6.8
Source: World Bank
3.3 Sample
Regarding the sensitivity of the questions about bribery, the raw data set contains many refusal to answer and non-responses. This causes missing data in the sample. The missing values are deleted from the questionnaire as no pattern could be observed, since they were missing at random. Furthermore, incorrectly entered data and outliers are deleted as well. Outliers are extreme values for which the standardized residual is > 3 or < -3. These outliers are excluded from the data set, since they can misrepresent the regressions coefficients. In order to specify influential cases in the independent variables, the Cook’s distance is calculated (Siero, Huisman & Kiers, 2009). Nevertheless, no influential cases are identified, meaning that all observations were below 1. After deletion of the non-responses, missing values and outliers, the dataset contain 1986 observation. 1051 Responses are from Indonesia and 935 responses from Philippine firms. The total response rate for Indonesia is 73%, and for the Philippines 71%.
3.4 Dependent variable
Questions on corruption and bribery are asked throughout the interview. However, the most exact question to quantify corruption is:
“It is said that establishments are sometimes required to make gifts or informal payments to
public officials to “get things done” with regard to customs, taxes, licenses, regulations, services etc. On average, what percentage of total annual sales, or estimated total annual value, do establishments like this one pay in informal payments or gifts to public officials for this purpose?”
20
3.5 Independent variables
The independent variables for this research are work experience, education level and the top managers’ gender. In order to measure the influence of these independent variables on the bribes that are paid, the independent variables should be operationalized.
Work experience (H1): Work experienced is measured with the following question 'How many
years of experience working in this sector does the top manager have?' It measures the years of experience within the sector that the manager is active in and is a continuous variable.
Education level (H2): This ordinal variable measures the education level of the top manager by
the following question: 'What is the Top Manager’s highest completed level of education?'
Gender of the top manager (H4): The question 'Is the top manager female?' is used to measure
the amount of female top managers. The question is a dummy variable, with possible 'yes' or 'no' answers.
3.6 Control variables
21
Table 2: Variables, Definitions and Measures
Type Variable Definition Measure Dependent Bribery Whether a firm uses informal payments or
gifts to get things done.
Binary variable, equals 1 if a firm bribes.
Independent Work experience Amount of years working in the same industry Continuous
Independent Education level The level of education of the top manager Nominal: 1 = No Education, 2 = Primary School, 3 = Secondary School, 4 = Vocational Training, 5 = University degree, 6 = Graduate degree
Independent Gender of the top manager
If the gender of the top manager is female Binary variable: 1 if it is a male, 0 if it is a female
Control Export Average % of direct exports Continuous
Control Firm size The number of fulltime employees per firm Small (1 to 19 employees) Medium (20 to 99 employees) Large (more than 100 employees)
Control Business sector The firms major business sector Service = 1, non-service = 0
Control Firm age 2010 or 2009 - year of establishment Continuous
Control Ownership structure
The ownership structure of the establishment Dummy (1 = Sole proprietorship, 0 = other)
Control Contact with officials
Average % of time spent per week on dealing with regulations imposed by public officials
Continuous
Control Obstacle Corruption indicated as the biggest obstacle Dummy (1 = Corruption, 0 = other)
Control Country Indonesia Whether the sample includes Indonesia Dummy ( 1 = Indonesia, 0 = other)
22
3.7 Statistical model
This research will use the following model to test the hypotheses:
Logit Pr(y=1) = β0 + β1X1 + β2X2 + β3X3 +βcXc + ε
Where, y indicates whether a firm bribes; β0 is the intercept; X1 measures work experience; X2 measures the education level; X3 measures gender; XC stands for the control variables including: export (XC1), firm size (XC2), business sector (XC3), firm age (XC4), a dummy for ownership structure (Xc5), contact with public officials (Xc6), a dummy indicating corruption as the biggest obstacle (XC7),a country dummy for Indonesia (XC8); and ε is the error term.
A Logit regression will be used to test the hypothesis, since the dependent variable is non-metric binominal, i.e. a firm bribes or not. The regression will follow a three-step approach; first the control variables will be included in the model. Secondly, the control- and independent variables will be integrated. Finally, in order to test for the non-linearity in the relationship between work experience and bribery, a squared term is added to the equation.
3.8 Method assumptions
In order to have linear unbiased estimates, Ordinary Least Square (OLS) has several assumptions that must be satisfied. However, a Logit regression is performed in this study, the method assumptions for OLS still apply. This thesis will test for the following assumptions: homoskedasticity, endogeneity, multicollinearity and normality.
Homoskedasticity
23
from the true regression line provides the least information about the true regression line. There are methodical ways to correct for the weight of larger error variances.
In the sample, heteroskedasticity could be expected in the variables work experience, firm size and firm age of the manager. Managers with little work experience might less easily resist bribes than more experienced managers. Furthermore, it might be less easy for smaller firms to resist bribes than larger firms. In addition, older firms might have more resources to resist a bribe than younger firms. These variables may lead to differences in error variances. The residuals of the variables are plotted in figure 1, 2 and 3 (see appendix 2). The plots estimate the errors, against manager experience, firm age and firm size. It could be concluded from the graph that heteroskedasticity does not seem to be a problem in the sample. The variance is similar for inexperienced and experienced managers, smaller and larger firms, and for young and older firms.
Table 3: Breusch-Pagan / Cook-Weisberg Table 4: Breusch-Pagan / Cook-Weisberg test for heteroskedasticity test for heteroskedasticity
Ho: Constant variance Ho: Constant variance Variables: experience Variables: firm age chi2(1) = 1.54 chi2(1) = 3.68 Prob > chi2 = 0.2152 Prob > chi2 = 0.0549
Table 5: Breusch-Pagan / Cook-Weisberg test for heteroskedasticity
Ho: Constant variance Variables: firm size chi2(1) = 3.67 Prob > chi2 = 0.0554
24
Endogeneity
An endogeneity problem will occur if the assumption that the error term is uncorrelated with the independent variable is violated such that one dependent variable is correlated with an unmeasured variable. In that case the effect of the independent variables to the dependent variable will be overestimated. No simple statistical or numerical tests are available to test for the endogeneity assumption; therefore it has to be satisfied theoretically4. Generalized least squares (GLS) and two-stage least-squares (2SLS) regression with instrumental variables is two options to overcome the endogeneity problem. Work experience, firm age and firm size can be easily estimated and observed by managers; hence theoretically speaking there is little reason to assume that these variables correlate with unobserved variables in the error term.
In case endogeneity exists, it is more likely that indicators that are based on the judgment of the respondent are correlated with the error term. These indicators could be influenced by unobserved variables. However, no indicators that are based on the judgment of the respondent are present as independent variable in the sample. If the indicators cannot be observed directly, it is common practice to use instrumental variables. However, such instrumental variables are not present in the sample. Since the common test for endogeneity relies on instrumental variables, and these variables are not present in the sample, the assumption will be made that the nature of the independent variable is exogenous.
Multicollinearity
According to Hill et al. (2009) the multicollinearity assumption is that the independent variables are not perfectly correlated such that the values of the independent variables are not exact linear functions of other explanatory variables. If this assumption is violated, variables are collinear. Collinearity makes it difficult to isolate the relationship between the different variables. The problem with multicollinearity is that the estimates of the regression model become unstable and the standard errors of the coefficients get inflated. The variance inflation factor (VIF) is used in order to test for multicollinearity. The VIF represents an index that estimates how much the variance of an estimated regression coefficient is inflated due to collinearity. The values in table
4 The sample residuals will always be uncorrelated to the independent variables, so using those to test for
25
6 indicate that multicollinearity is not a problem, since all values are below the cut-off value of 10 as suggested by Neter et al. (1985).
Table 6: Test for Multicollinearity
Normality
Normality assumes that the values of the error term are normally distributed around its mean. A violation of this assumption might lead to biased p-values, what might affect the test for statistical significance (Hill et al., 2009). The graphical analysis in appendix 3 plots the residuals against the normal distribution, which is a straight line. It can be concluded from this plot that the residuals are approximately normal distributed. In case this assumption has been violated, a possible solution was to transform the dependent variable.
The above tests indicate that the model does not contain problems with heteroskedasticity, endogeneity, multicollinearity and normality. Additional goodness-of-fit test reveal that the Logit model fits the data and is correctly specified, see appendix 4. The following section will provide the empirical results.
Variable VIF 1/VIF
26
4. Empirical results
4.1 Descriptive statistics
This section will provide an overview of the descriptive statistics. Table 7 gives an overview of the dependent, independent and control variables. The table indicates the mean, standard deviation, the minimum value and the maximum value of the variables. Since these variables include dummy variables, appendix 5 provides more information about the frequencies of the dummy variables.
Table 7: Descriptive statistics
From the total number of 1986 firms in the sample, 402 firms indicate that they pay bribes to government officials. The other 80% of the respondents indicate that they don’t pay bribes to civil servants. The average work experience of a manager within the industry is 15 years, with a standard deviation of 11 years. The minimum amount of work experience within the industry is 1 year and the most experienced manager has 70 years of work experience. Most top managers have a university degree (51%) as their highest degree, while 11% has graduated from university. Taking a closer look at the lower educational levels indicates that 1% of the top managers don’t have any education, 9% completed primary school as their highest education and 23% completed
Mean Std. Dev. Min Max
27
secondary education as their highest education. 6% Of the top managers completed vocational training as their highest educational level. A total number of 535 firms indicate that their top manager is female, being 27% of the total sample. The other 1,587 firms have a male top manager. The control variable export, measures the amount of direct export to a foreign country. On average, 10% of the products are directly exported to another country. However, the standard deviation of 27% shows a wide variation of the direct export. Some firms don’t export at all while other firms directly export all their products. Almost half (47%) of the firms in this sample are indicated as small (19 or less employees), 33% are indicated as medium sized (between 20 and 99 employees) and 20% of the firms are large (100 or more employees). 22% Of the firms in this sample are active in the service sector, the other 78% are active in another sector. The age of the firm is a continuous variable, with an average age of 18 years and a standard deviation of 12 years. The youngest firm is 2 years old, the oldest firm 97 years. Taking a closer look at the ownership of the firms, 46% is single owned, the other 54% is owned otherwise. The continuous variable foreign ownership ranges from 0% foreign owned, to 100% foreign owned. The average firm is for 12% owned by foreigners with a standard deviation of 30%. The average senior manager’s time spend with public official requirements per week is also a continuous variable. On average a senior manager spend 8% of their total time on dealing with the requirements, with a standard deviation of 16%. The highest amount of time spends on dealing with requirements imposed by government regulations are 100%, the lowest amount 0%. 6% of the firms see corruption as the biggest obstacle of all the elements in the business environment, while the other 94% see another obstacle as the biggest problem. A total number of 1,051 firms are from Indonesia (53%), the other firms are from the Philippines.
4.2 Regression results
28
controls for the difference between the manufacturing and service sector is positive and significant. These results are as expected, since the service sector is newer and more bribes are expected to be paid in this sector. The age of the firm is positively related to bribery. However, this study doesn't find a significant result. According to the results in Model 1, single owned firms have a significantly higher probability to bribe, than firms with another ownership structure. These findings are in line with the argumentation that single owned firms are more sensitive to bribery. Firms that are spending more time with governmental officials, have a significantly higher probability to bribe on a p < 0.001 level. Furthermore, firms that indicate bribery as the biggest obstacle while doing business, have a higher probability to pay bribes. The country dummy for Indonesia is significant in all three models, showing a significant difference between Indonesia and the Philippines in the probability to bribe. The Indonesian firms tend to be more corrupt.
29
results are in line with the social role theory. Social role theory argues that a higher exposure to corruption lead to an increase in acceptance and tolerance of corruption. Furthermore, males develop traits that manifest agency. Therefore we accept the third hypothesis.
30
Table 8: Logit Regression
Standard errors in parentheses
+
p < 0.10, * p < 0.05, ** p < 0.01, *** p < 0.001
Model 1 Model 2 Model 3
Export 0.00214 0.00216 0.00211 (0.00208) (0.00208) (0.00209) Firm size 0.282** 0.215* 0.215* (0.0901) (0.0940) (0.0940) Sector 0.264+ 0.246+ 0.245+ (0.142) (0.144) (0.144)
31
5. Robustness tests
32
Table 9: Probit regression
Standard errors in parentheses
+ p < 0.10, * p < 0.05, ** p < 0.01, *** p < 0.001
Model 1 Model 2 Model 3
Export 0.00119 0.00119 0.00116 (0.00123) (0.00124) (0.00124) Firm size 0.165** 0.126* 0.126* (0.0518) (0.0541) (0.0541) Sector 0.148+ 0.137+ 0.136+ (0.0814) (0.0823) (0.0823)
33
5.1 Additional analysis of bribery
This section will provide an analysis if an interaction effect exist between work experience and education level. According to Jaccard (2001), an interaction effect exists when the effect of an independent variable on a dependent variable differs, depending on the value of a third variable. This third variable is frequently called a moderator variable and is in this research the education level. In this thesis, work experienced is tested on a u-shaped relationship with bribery. However, no significant evidence is found in this study supporting this argument. The results show a significant relationship on a p < 0.05 level for a positive linear relationship between work experience and bribery payments. These results are in line with human capital theory. Therefore this thesis will test for an interaction effect between work experience as a positive linear variable and the education level of the top manager.
Education and experience are two central characteristics of the human capital theory (Becker, 1964). It was previously argued that managers with more depth and breadth in work experience are more corrupt, since they can better recognize and exploit opportunities and reduce uncertainty about the value of opportunities. Furthermore, it was argued that a higher education level leads to a better recognition and pursue of an entrepreneurial opportunity, including the opportunity to bribe. It is researched that managers with superior human capital are better able to effectively plan and play bribery games to their advantage (Guerrero & Rodriguez-Oreggia, 2008; Olken, 2009). Consequently, managers with both high education level and work experience are expected to be even more corrupt. In other words, managers endowed with superior human capital, are more corrupt. Thus, the interaction effect between work experience and education level is expected to be positively related to bribe payments. Experience and education level are two continuous variables. In order to test the moderating effect successfully, both variables are standardized and mean centered (Jaccard, 2001).
34
This is contradicting the expectation that a higher human capital lead to a higher level of corruption.
Table 10: Interaction effect
Model 3 Model 4 Experience (log) 0.266 (0.272) 0.258 (0.272) Experience2 (log) -0.0258 (0.0631) -0.0226 (0.0631) Education 0.117* 0.159 (0.0588) (0.221) Gender 0.248+ 0.243+ (0.139) (0.139) Experience X Education -0.0714 (0.0550) Constant -3.760*** -2.070*** (0.540) (1.407) N 1986 1986 Pseudo R2 0.051 0.052 Log Likelihood -949.2 -948.3 Chi2 102.5*** 104.3***
Standard errors in parentheses
+ p < 0.10, * p < 0.05, ** p < 0.01, *** p < 0.001 Note: Control variables are included in model 3 and 4.
6. Conclusion
The positive and negative effects of corruption are stressed in previous literature. Corruption remains common in many countries around the world. From all the firms surveyed by the World Bank, 25% of all firms in the world are expected to bribe public officials to get things done. For many firms, corruption increases the costs of doing business. This figure suggests that bribery is still a severe problem around the world.
35
been examined. The hypotheses were tested by using a sample of 1986 observations from two South East Asian countries, namely Indonesia and the Philippines. These countries show a high growth rate and are operating in a highly corrupt environment.
The first hypothesis argues for a u-shaped relationship between work experience and bribery. According to TMT theory managers with little work experience tend to be more corrupt since they are more willing to change, show less market conformity and are less focused on stability, efficiency and reliability. The amount of corruption will decrease when the manager is getting more experience. However, according to human capital theory, the amount of corruption will rise again when the managers are getting experienced, since they can better recognize and exploit opportunities and reduce uncertainty about the value of opportunities, like the opportunity to bribe. The findings in this study don't support the hypothesis and no u-shaped relationship between work experience and bribery has been found. However, this study find significant support for the second hypothesis on a p < 0.005 level. In line with human capital theory, it was argued that managers with a higher level of education are more corrupt, since education increases a person's stock of information and skills, what lead to a better recognition and pursue of an entrepreneurial opportunity, including the opportunity to bribe. The results are thus in line with the theoretical expectations. In addition, significant support on a p < 0.10 level is found for the third hypothesis, indicating that male managers bribe significantly more than female managers. This is in line with the social role theory, arguing that a higher exposure to corruption increase the acceptance and tolerance of corruption and males develop traits that manifest agency.
6.1 Added value
36
support for the hypothesis that managers with a higher education level are likely to pay more bribes, than managers with a lower education level. Furthermore, significant evidence is found for the third hypothesis, indicating that male top managers pay significantly more bribes than female top managers. The World Bank Enterprise Survey provides this research with firm level data. The data is gathered in a unique dataset, providing the input for this research. Indonesia and the Philippines are countries with high corruption levels and have a high economic growth rate. Therefore, these countries are attractive for foreign firms to do business in. Since only two countries in South East Asia are covered in this research, comparable results might be expected in other South East Asian countries.
Several strategic key implications for managers and policymakers can be revealed from the findings of this study. Firstly, due to the significant evidence found in this study, should make managers aware that higher educated managers are more corrupt. This might feel counterintuitive since most country-level studies on corruption reveal that countries with higher levels of education are positively correlated with lower figures of corruption. When dealing with high educated managers from the same firm or another firm, the manager should be aware of more corrupt behavior from these managers. This implies that highly educated managers from the same firm should be monitored more closely on corrupt behavior. Furthermore, when dealing with other firms, highly educated managers should be examined more thoroughly. Secondly, firms could expect more corrupt behavior from male managers than from female managers. Therefore, male managers should be monitored more closely on corrupt behavior than female managers. Additionally, a firm could promote more females in top positions within the firm in order to decrease the probability of bribery.
37
(Huther & Shah, 2000). Contrary, in countries with little corruption and strong governance, priorities might be stronger financial management, increased public and government awareness, explicit anti-corruption agencies and programs, no-bribery pledges and so on. Most countries focus on fighting corruption from the demand side by trying to limit the corrupt behavior of public officials. Since the private sector is involved in most corrupt practices from the supply side, policy makers could try to limit the corrupt behavior of the private sector, including firms, as well. Therefore it would be very fruitful for policy makers to identify the characteristics of firms and managers that show corrupt behavior. This study identifies two characteristics, namely education level and gender that helps policy makers to identify corrupt managers. Secondly, based on the significant evidence that female managers are less corrupt than male top managers, policy makers could reduce the amount of corruption by introducing policies that favor females in higher positions within the company. Since females are less corrupt than males, having females in higher positions within the company could reduce the amount of corruption.
6.2 Limitations and suggestion for further research
This study has several limitations. Firstly, the sample of this study covers only two different countries, which is a rather small part of South East Asia. Therefore, the generalizability of the results are limited. Results could be generalizable for other countries in South East Asia with comparable growth rates and corruption levels. Nevertheless, the results could not be generalized for other countries in the world. Secondly, the business environment in Indonesia and the Philippines might have changed since the sample was taken in 2009. Especially the corruption rates might have changed. Making use of longitudinal data would improve the validity of the results. Thirdly, the measurement of bribery is limited by using a Logit model, i.e. whether a firm bribes or not. The actual size of the bribe is not taken into account. Lastly, regarding the sensitivity of the questions on bribery, the use of the data collected by the World Bank is limited. The answers might be suspect to false responses or non-responses. Many non-responses were detected in the data set.
38
39
Appendix 1: Survey questions
Bribery:It is said that establishments are sometimes required to make gifts or informal payments to public officials to “get things done” with regard to customs, taxes, licenses, regulations, services etc. On average, what percentage of total annual sales do establishments like this one pay in informal payments or gifts to public officials for this purpose?
…….% of bribes paid
Dummy: Does the firm pay bribes?
1 = Yes 0 = No
Work experience:
How many years of experience working in this sector does the Top Manager have? ...years
Less than one year 1
Don't know -9
Education:
What is the Top Manager's highest completed level of education?
No education 1
Primary school 2
Secondary school 3
Vocational training 4
University degree (B.A., B.S., etc.) 5
Graduate degree (M.B.A, M.S., Ph.D. etc.) from a university in this country 6 Graduate degree ((M.B.A, M.S., Ph.D. etc.) from a university abroad 7
Don't know -9
Gender:
Is the Top Manager female?
Yes 1
No 2
Don't know -9
Export:
What percentage of this establishment’s sales was exported directly (without the use of a third party: …………..(%)
40
Firm size:
At the end of fiscal year 2009, how many permanent, full-time employees did this establishment employ?
……….(no. of employees)
Business sector:
1 = Manufacturing 3 = Other services
2 = Service (retail)
Firm age:
In what year did this establishment begin operations?
………. (year)
Ownership structure:
What is the firm’s current legal status?
Contact with public officials:
What percentage of total senior management’s time was spent on dealing with requirements imposed by government regulations?
………. % of senior management’s time spent on dealing with regulations
0 = No time was spent -9 = Don’t know
Corruption as the biggest obstacle:
Which of the following elements of the business environment, if any, currently represents the biggest obstacle faced by this establishment?
1 = access to finance 10 = labor regulations 2 = access to land 11 = political instability
3 = business licensing and permits 12 = practices of competitors in the informal sector
4 = corruption 13 = tax administration
5 = courts 14 = tax rates
6 = crime, theft and disorder 15 = transport
7 = customs and trade regulations 16 = telecommunications
8 = electricity 17 = economic and regulatory policy uncertainty 9 = inadequate educated workforce 18 = macro-economic instability
Country:
1 =Indonesia 2 =Philippines
1 = Shareholder ownership with shares traded in the stock market 6 = Other
2 = Shareholder ownership with shares traded privately 7 = Single member private ltd. Co.
3 = Sole proprietorship 8 = Business permit
4 = Partnership 9 = Unregistered
41 -1 0 1 2 3 Pe a rs o n re si d u a l 0 20 40 60 80
B7. How many years of experience working in this sector does
Appendix 2: Homoscedasticity
Figure 1: Graphical analysis of Figure 2: Graphical analysis of homoscedasticity (work experience) homoscedasticity (firm age)
Figure 3: Graphical analysis of homoscedasticity (firm size) -1 0 1 2 3 Pe a rs o n re si d u a l 0 20 40 60 80 100 2009 - year of establishment -1 0 1 2 3 Pe a rs o n re si d u a l 0 2000 4000 6000 8000
42
Appendix 3: Normality
Figure 1: Test for normality
43
44
Appendix 5: Frequency statistics
Frequency Percent Cumulative
45
References
Aidis, R., Estrin, S. & Mickiewicz, T. 2012. Size matters: entrepreneurial entry and government. Small Business
Economics, 39(1): 119 – 139
Ardichvili, A. & Cardozo, R. 2000. A model of the entrepreneurial opportunity recognition process. Journal of
Enterprising Culture, 8(2): 103 – 120
Ades, A. & Di Tella, R. 1999. Rents, competition, and corruption. American Economic Review, 89: 982–993
Becker, G.S. 1964. Human capital, Chicago: Chicago University Press
Blumberg, B., Schindler, P. & Cooper D. 2011. Business Research Methods, McGraw-Hill Higher Education
Casson, M. 1982. The entrepreneur: An economic theory. Totowa, NJ: Barnes & Noble
Casson, M. 1995. Entrepreneurship and business culture. Brookfield, VT: Edward Elgar
De Jong, G., Ahn Tu, P. & van Ees, H. 2012. Which Entrepreneurs bribe and what do they get from it? Exploratory Evidence from Vietnam. Entrepreneurship: Theory & Practice, 36(2) 323 – 345
Diekman, A., Goodfriend, W. & Goodwin, S. 2004. Dynamic stereotypes of power: Perceived change and stability in gender hierarchies. Sex Roles, 50: 201-215
D’Souza, A. 2012. The OECD Anti-Bribery Convention: Changing the currents of trade. Journal of Development
Economics , 97: 73–87
Eagly, A. 1987. Sex differences in social behavior: A social role interpretation. Hillsdale, NJ: Erlbaum
Eagly, A. 1997. Sex differences in social behavior: Comparing social role theory and evolutionary psychology. American Psychologist, 50, 1380-1383
Eagly, A. & Johnson, B. 1990. Gender and leadership style: A meta-analysis. Psychological Bulletin, 108: 233-256
Eagly, A. & Karau, S. 1991. Gender and the emergence of leaders: A meta-analysis. Journal of Personality and
Social Psychology, 60: 685-710
Eagly, A. H., Wood, W. & Diekman, A. 2000. Social role theory of sex differences and similarities: A current appraisal. In T. Eckes & H. Trautner. The developmental social psychology of gender. Mahwah, NJ: Erlbaum
Economist. 2006. Women and the world economy: A guide to womenomics, April 12
Feingold, A. 1994. Gender differences in personality: A meta-analysis. Psychological Bulletin, 116: 429-456
Fiet, J. 1995. Risk avoidance strategies in venture capital markets. Journal of Management Studies, 32(4): 551–574
Finkelstein, S., Hambrick, D. & Canella, A. 2009. Strategic leadership. Theory and research on
executives, top management teams, and boards. Oxford: Oxford University Press
