An analysis of corruption, through
political stability and firm size
Master Thesis International Business and Management
Nicholas Brightwell S2223880 n.m.brightwell@student.rug.nl
Supervisor: Dr. G. de Jong March 17, 2014
Faculty of Economics and Business
2 Abstract
Today, corruption is seen as an important issue when considering the growth of many major nations. This research project adds to the previous written literature on the causes of corruption by exploring both country-level and firm-level variables, political stability and firm size, and each variables effect on a segment of corruption, bribery. With a total of 3155 observations, 4 different levels of firm size are explored, from 9 different countries, with 3 different levels of political stability. The correlation between these variables is analyzed through a Logit regression analysis, by viewing the following countries: Venezuela, Vietnam, Romania, Botswana, Panama, Lithuania, Turkey, Czech Republic and Uzbekistan. The results provide support for a negative relationship between political stability and bribery. We find an insignificant relationship between firm size and bribery. An interaction effect is also tested for between political stability and firm size and the study shows that political stability acts as a moderator variable towards firm size, in terms of corruption. Moreover, it is also proven that specific countries and factors from within each specific country affect bribery results significantly.
3
Table of Contents
1. Introduction……… 4
2. Literature review………. 6
2.1 What is corruption and how can we measure it?... 6
2.2 Corruption as bribery………...……….……….9
2.3 Hypotheses……….……… 9
2.3.1 Firm Size……….………...9
2.3.2 Political Stability……….……… 11
2.3.3 Political Stability and Firm Size………...……… 13
3. Methodology………. 14 3.1 Countries……… 15 3.2 Dependent variable……… 15 3.3 Independent variables………. 16 3.4 Control variables……….. 18 3.5 Method assumptions………. 19 4. Empirical results……….. 22 4.1 Regression results……… 23 4.2 Robustness tests……… 26 4.3 Interaction effect………. 27 5. Discussion……… 29 6. Conclusion……… 30
6.1 Limitations and further research……….. 33
Appendix 1 – BEEPS questions and WGI specifics...……….. 34
Appendix 2 – Figures and tables……… 36
Appendix 3 – Model fit and Robustness test……… 38
4
1. Introduction
International firms, regulators and governments are constantly intrigued by the effects of corruption on economic development, with anti-corruption strategies being promoted worldwide through various organizations (Organization for Economic Cooperation and Development European, United Nations, World Bank, European Bank for Reconstruction and Development, Transparency International). Today, corruption is seen as an important issue when considering the growth of many major nations. The Greek economy can be seen as an example of a current situation which may lead to the utilization of possible structural reforms (Christodoulakis et al., 2011). From both a business and ethical perspective it is important the corruption is understood and dealt with. Transparency international (transparency.org) believe that corruption to this day is a global issue. Even though there are very different levels of corruption from country to country, bribery still takes place in even the largest economies today.
5
(Jensen, Liand Rahman, 2010). What about linking country-level data with firm–level data? This has rarely been done before and could create an intriguing and revealing set of results. This project is therefore setup to review the link and relationship between bribery, political stability (country-level) and firm size (firm-level).
The aim of this thesis is to get a better understanding of how political stability and firm size effect the bribery of firms in various different economies. Firm size does not come up much as an independent variable in previous research as it is seen more as a control variable. Are smaller firms really more likely to bribe due to less bargaining power? How does political stability effect these firms that struggle with bribery in various different nations? Political stability is another variable of interest and in many previous studies is said to have a negative impact on corruption. However, this is not always necessarily the case and therefore this is also a variable worthy of research. All areas of firm size will be researched from small to large and the same goes for political stability, from the lower levels of stability to the higher levels through various nations and the focus will be the differences in whether bribes were paid by firms when these variables are considered. This study will therefore add to the comprehension of what pushes firms to make bribery payments. How are these 3 factors interrelated? We will view each of the factors and compare different countries with different political stability levels to each other, in order to come to an overall conclusion.
The main research question of this study is: What are the determinants of firm bribery? The following research questions will be answered in this project in order to explore the main research question:
What is corruption and how can we measure it?
What is the relationship between bribery and firm size?
What is the relationship between bribery and political stability? Does a relationship exist between firm size and political stability?
6
and political stability in order to see if they have an effect on bribery together and if one moderates the other. Data will be taken from the Business Environment and Enterprise Survey (BEEPS) and the World Government Index (WGI) to test each hypothesis. The results will be tested afterwards by a robustness test. During this project we must understand that in politically aggressive and unstable areas, firms use not answering and untruthful claims as ways to protect themselves. As we deal with unstable areas, the level of corruption noted could very much be untruthful. In this study, the World Bank enterprise survey database will be used, which contains surveyed information from 44,000 firms in 72 various countries. Firms used in this study in areas that are unstable with lower development and less freedom are more likely not to answer and create untruthful claims in terms of corruption. This could lead to vital misjudgments of the level of corruption. However, the Worldbank are aware of how vital this information is if it used correctly, and have used firm managers to explore levels of corruption with these circumstances in mind, in order to minimize the risk of misjudgment.
2. Literature review
2.1 What is corruption and how can we measure it?
The concept of corruption is a common concept in today’s business world and is being reviewed constantly more and more. A large number of studies are being taken up about this topic as many firms are starting to take the idea of corruption more seriously and have the intention of fighting corruption (Jain, 2001). It is difficult to define corruption. Huntington states that corruption is the “behavior of public officials which deviates from accepted norms in order to serve private ends” (Huntington 1989). Jain states however that “corruption refers to acts in which the power of public office is used for personal gain in a manner that
7
investment is deterred, with start-up businesses inside specific countries, on many occasions, also finding it difficult to overcome the extra start-up costs needed due to corruption
(UNODC). However, what are the effects on corruption and what are the causes? After a review of previous literature the following 3 major categories effect corruption the most:
Organization and individuals
Historical roots
Contemporary causes
However, many published causes have there blind spots as the topic of corruption is rather complex (Pellegrini and Gerlagh, 2008).
Organizations and individuals
Corruption exists in all countries, both developed and emerging, in both the public and private sectors, and in all the various forms of organizations and firms. Therefore, the type, structure and form of organization will have a large effect on corruption as well as specific individuals (UNESCAP). The organizational construction of a firm in relation to size, age, sector, and productivity are frequently cited as factors effecting a firm’s durability and therefore are seen as determinant factors of corruption as well (Hallward-Driemeier, 2009). This is also why firm size seemed intriguing. Corruption is sometimes embedded within an organization, meaning, it is stuck within the organization. In this situation corruption is taken for granted and maintained within. Therefore, corruption, such as bribery, can be ingrained in
organizational structures and systems, embodied by individual members as allowed and correct conduct. This is then transferred to the following generations of individuals as well (Ashford and Anand, 2003).
Historical roots
8 Contemporary causes- institutions, economic structures, the level of development and
political stability
Income levels are seen as a major cause and can effect corruption in many ways. Wealthier countries can build and develop better institutions. Several factors are linked with income, for example, education levels, urbanization and mass media access, are linked with better development levels and they lower the lenience of the state in terms of corruption. Rent-seeking theories publicize the link between corruption and the likelihood of economic agents getting to origins of higher than average rents. Being open to trade increases domestic competition, and therefore lowers rents and corruption. In contrast to this, trade-barriers raise the likelihood of acquiring extra rents by getting trade allowances, which in turn increases corruption (Pellegrini and Gerlagh, 2008). Natural resources are a potential origin of high rents, for those that are legally allowed to extract them. These rents provoke policy makers who have the last say on who can have exploitation rights, which pushes away resources from alternative important and productive activities (La Porta et al. 1999). Therefore an affluent amount of natural resources can be linked with an increase in corruption. Another area to touch on is social institutions and democracies. Many previous studies that concentrate on democracies, using a low amount of control variables, state that democracy lowers corruption levels. However, many recent papers do not agree with this conclusive result (Treisman, 2000).This point then brings us to political stability. In politically stable governmental systems, there is a lower chance of being laid off, and the increasing possibility of a long career, which leads to the motivation to create an honest reputation for the evolution of a career. However, a steady career in a position of power may lead to contributing to already corrupt reputations and networks of corruption. From this it is clear that there are two very different hypotheses on how corruption is effected by political stability (Treisman, 2000). This is a reason why political stability was chosen as a variable in this study. Last but not least, a modern determinant can be seen, in terms of the media. This theory states that corruption occurrences often in the public eye can be used to lower levels of corruption. (Brunetti and Weder, 2003). Another debated theory is that Protestant religion, due to it being less hierarchical in comparison to other religions, is not as likely to suffer as much from corruption. Therefore, protestant countries may not be as effected by corruption.
9
policies” (Jain, 2001). Politicians are meant to allocate resource acquisitions based on the interest of the population. Political elite involved in corruption can change national policies to serve its own needs, at a specific cost however to the population. Bureaucratic Corruption involves acts of corruption by “bureaucrats when dealing with the political elite or the public” (Jain, 2001). Legislative Corruption involves the manner and amount that the “voting behavior of legislators can be influenced”. Legislators can be bribed by groups to “enforce legislation that will change the economic rents associated with assets” (Jain, 2001).
2.2 Corruption as bribery
A large part of corruption is bribery. It is also how we will recognize and measure corruption within this project. Corruption and bribery are seen as similar in many cases. A bribe can be defined as a payment that pushes a person to behave in a fashion that is contrary to his responsibilities, whereas, corruption is the abuse of power for private gain, which can incorporate other corrupt activities, for example, fraud, money laundering, fraud, misuse of private knowledge, gifts and so forth (Weber & Getz, 2004). We will use the BEEPS database to measure bribery. We will look into this deeper in the methodology section. Svensson (2005) describes corruption as the misuse of public office for private gain and he goes on to state that it is the outcome and reflection “of a country’s legal, economic, cultural and political
institutions. Corruption can be a response to either beneficial or harmful rules”. In addition, corruption can also occur due to bad policies or inefficient institutions being set-up to collect bribes from individuals looking to avoid them (Djankov, LaPorta, Lopez-de-Silanes and Shleifer, 2003). Levels of Bribery also change depending on various institutional conditions and tensions dealt with by specific firms (Martin, Cullen, Johnson & Parbooteah, 2007). In stable European nations bribery is seen as a large and important issue, where as in many developing nations bribery payments have become part of the everyday business norm. Therefore, on some level it is difficult to say that bribery is incorrect and the question then becomes should bribery be seen as an individual act or a behavioral method (Steidlmeier, 1999).
2.3 Hypotheses 2.3.1 Firm size
10
11
the matter by stating that “firms that are larger, older, exporting and government owned are widely held and/or have fewer competitors, have more political influence and perceive corruption to be less of a problem and pay bribes less often”. In general, it can be seen that strong firms (assumed to be large) use their networks, knowledge and power to bend and overcome legal regulations, however, weak firms (assumed to be smaller) make bribery payments to remove the effects of government interference (Bennedsen, Feldmann and Lassen, 2009). A purely negative trend is not always seen however. Rose-Ackerman and Stone, (1996) found that “bribes as a percentage of gross-sales were highest for
micro-enterprises- at 2.45% lower than small and medium size firms and higher again for the largest enterprises.” Reasoning for this was that “large firms can avoid involvement in corruption in some of the areas explored. This in turn may limit small-sized firm’s capability to compete in this market, as they are most likely only able to make contacts and connections with small-sized firms” (Kousnetzov and Dass, 2010).
As we can see though, corruption is overall negatively associated with firm size and growth at the firm level (Athanasouli, Goujard, Sklias, 2012). There is definitely a negative relationship and therefore as firm size increases corruption will decrease. The following hypothesis can also be made:
H1: Firm size will have a negative effect on bribery
2.3.2 Political stability
Political instability or (low political stability) depending on which view you take, is regarded by economists as a seriously harmful to economic performance (Aisen and Veiga, 2013). However, what is the effect on corruption? There are various definitions for political stability or instability. Awokuse and Gempesaw (2005) define “political instability as the propensity for a change in the governance of a country, which may include any type of insurgency, revolution, regime change, and military-led coups or the frequency of events that increase the likelihood of social and political unrests”. This can involve politically persuaded
12
13
Do 2008). However, in general, from the research in this study it seems political instability raises corruption. As bribery is a form of corruption and the two are heavily linked, this should be seen through bribery and with this we can produce the following hypothesis:
H2: Political stability will have a negative effect on bribery
2.3.3 Political stability and firm size
Both hypothesis 1 and hypothesis 2 have negative relationships, which means enough to associate the interaction effect between firm size and political stability with a negative correlation in terms of bribery. With firm size being negatively correlated with bribery and political stability being negatively correlated with bribery, it provides enough reasoning to believe that hypothesis 3 will also show a negative combination. Based on the research for hypothesis 2 it is more likely that it will be political stability that may have a larger effect on firm size rather than vice versa, in terms of corruption. This is due to the fact that political stability is a country-level variable and will therefore include many more factors than firm size. Political stability may therefore be seen as a moderator variable. On the other hand however, firm size can be linked heavily to firm growth. According to Kalonda-Kanyama (2013) there is a link seen on occasion between firm growth and political stability, however “political instability does not have a significant impact on firm growth (linked to firm size) when the other obstacles of the business environment are controlled for”. This could be because within political stability, there are many sub-variables which can influence stability, such as an election year. Anaman and Agyei-Sasu (2012) state that return increases in an election year but declines in a transition year after an election” and this obviously has an effect on firm growth. However, once again corruption is overall negatively associated with firm growth at the firm level (Athanasouli, Goujard, Sklias, 2012). Even though firm growth (linked with firm size) shows an insignificant link with political stability, once again we should take note and reconsider hypothesis 1 and hypothesis 2 in terms of corruption. This can be backed up by the following statement “corruption is overall negatively associated with firm growth at the firm level” (Athanasouli, Goujard, Sklias, 2012). As both hypothesis 1 and 2 point to a negative correlation, it is believed the interaction effect will also be negative in terms of corruption. From this we can produce the following hypothesis:
14 Concept Model:
Negative
Firm Size (interaction effect) Political Stability Negative
Negative Corruption
3. Methodology
Data from BEEPs (The Business Environment and Enterprise Performance Surveys also known as Enterprise Surveys) and WGI (Worldwide Governance Indicator project) will be used to test the hypotheses and will be the main resource for information. WGI will be
15
must contain a proportional relationship to the size of the population from which it comes from (Blumberg et al, 2011), the survey does this by taking more surveys in large economies than in smaller ones. The sample size for each country is sorted by industry, firm size, and geographic region, with firm survey amounts ranging from 150 – 1320 per industry. A
disadvantage of the Enterprise Surveys is that the methodology used is not the same across all countries and this can be seen if some of the older surveys are looked at (BEEPs).
In terms of the variable political stability, data will be viewed by using a political stability index called the Worldwide Governance Indicators. The Worldwide Governance Indicators (WGI) project reports aggregate and individual governance indicators for 215 economies over the period 1996–2012, for six dimensions of governance. This is explained in more detail in the variables section. When comparing sets of data between variables, dummy numerals will be used in order to be able to complete a Logit regression analysis between each of the variables described below. Each set of data will be compared in country groups and along the same time period. This will also allow us to set up firm age as one of the control variables. Data from BEEPs and the WGI index will be compared to produce results.
3.1 Countries
Below in Fig.1 is a list of the countries that will be researched. The political stability ranking system runs from 0 being the lowest to 100 being the highest level. These countries influence will be looked at overall. Each variable will be looked at and cross-compared.
Fig. 1
Country: PS-2002% PS-2007% PS-2012% PS-Mean% PS grouping
Botswana 73,08 82,21 88,63 81,30666667 low pol.ins=70+ stability Czech Republic 80,29 81,73 84,36 82,12666667 low pol.ins=70+stability Lithuania 74 74 70,1 72,7 low pol.ins=70+ stability
Romania 58,65 49,52 48,82 52,33 medium =40-70 stability Panama 52,9 41,8 40,3 45 medium =40-70 stability Vietnam 54,8 52,9 55,9 54,53333333 medium =40-70 stability
Venezuela 12,5 12,5 17,5 14,16666667 high pol.ins= 0-40 stability Uzbekistan 16,3 9,6 29,4 18,43333333 high pol.ins= 0-40 stability Turkey 20,19 19,71 13,27 17,72333333 high pol.ins= 0-40 stability
3.2 Dependent variable
16
question to measure corruption is the following question: “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?” Respondents have the
opportunity to refuse to answer the question or might not know the answer. Therefore, a lack of response is removed from the estimation and outliers as well. Data produced from this question is turned into a binary scale format that shows whether a firm bribes or does not. Assuming that not all firms are open to responding on the actual percentage of annual sales paid in informal payments or gifts, this method results in the most answers possible in terms of the bribery variable.
3.3 Independent variables Political stability
Political stability can be seen as one of two independent variables. This will be viewed by using a political stability index called the Worldwide Governance Indicators project. The Worldwide Governance Indicators (WGI) project reports aggregate and individual governance indicators for 215 economies over the period 1996–2012, for six dimensions of governance:
Voice and Accountability
Political Stability and Absence of Violence
Government Effectiveness
Regulatory Quality
Rule of Law
Control of Corruption
17
the World Bank Development Research Group and the World Bank Institute. “The
Worldwide Governance Indicators (WGI) is a research dataset summarizing the views on the quality of governance provided by a large number of enterprise, citizen and expert survey respondents in industrial and developing countries. These data are gathered from a number of survey institutes, think tanks, non-governmental organizations, international organizations, and private sector firms” (WGI). Each of six aggregate WGI measures is constructed by averaging together data from the underlying sources that correspond to the concept of governance being measured. This is done in the three steps described below. (WGI)
Assigning data from individual sources to the six aggregate indicators. Individual questions from the underlying data sources are assigned to each of the six aggregate indicators
Preliminary rescaling of the individual source data to run from 0 to 1. The questions from the individual data sources are first rescaled to range from 0 to 1, with higher values corresponding to better outcomes.
Using an Unobserved Components Model (UCM) to construct a weighted average of the individual indicators for each source. A statistical tool called the Unobserved Components Model (UCM) is used to allow 0-1 scaled data comparable across sources. It is then used to construct a weighted average of the data from each source for each country. Please see appendix 1 for WGI methodological sourcing and areas questioned for political stability. (WGI)
Firm size
18 3.4 Control variables
Firm age: Firm age will be monitored as it seen to be a major determinant of corruption. It will be monitored in a continuous fashion and the correlation will be checked to see if firm age is indeed a major determinant at least in terms of bribery. Business sector: The countries analyzed have their data based in the manufacturing sector and service sector. This might lead to differences in firm performance across sectors as well as different reactions to bribe
requests. Therefore, a control variable is included for the business sector. “There is reason to believe that being either a manufacturing or service firm might have a bearing on how vulnerable the firm is to bribery demands, but a priori it is not clear which sector should be more exposed. Valid arguments can be made for either sector being more susceptible to corruption” (Herrera, Lijane and Rodriguez, 2007). The % of time spent by senior
19
Fig. 2
Type Variable Definition Measure
Control Firm age When firms were established Continuous
Control Business sector The firms major business sector Service = 1, non-service = 0
Control % of time, government regulations
% of time spent by senior management dealing with
government regulations Continuous
Control Obstacle- corruption Whether the biggest obstacle is corruption Non- corruption = 0 Corruption= 1
Control Obstacle- p. stability Whether the biggest obstacle is political stability
Non- political stability = 0 Political stability= 1
Control Country Individual country Individual country
dummy
Dependent Bribery Did the firm bribe? 0=No
1=Yes
Independent Firm size Size of firm
0= below 5 employees 1=Small 2=Medium 3=Large
Independent Political stability Level of stability
3=low stability 2=medium stability 1=high stability
3.5 Method assumptions
Linear regression usually includes assumption tests that are based on ordinary least square algorithms which are needed, however, Logistic regression does not need to make many of these assumptions and therefore testing for normality and homoscedasticity is not necessary (statisticsolutions.com). Normality assumes that the values of the error term are normally distributed about their mean (Hill et al. 2009). However, Logit regression analysis takes this into account and therefore normality does not need to be analyzed separately.
Homoscedasticity assumes that the variance of the error term is consistent and is the same for all measurements. If this assumption is not fulfilled and the error variance for all
measurements is not the same, than we call this phenomenon heteroscedasticity.
20
Heteroscedasticity results in unrepresentative estimates of the standard errors which then leads to unrepresentative results (Hill, Griffiths & Lim, 2009). The fact that it is not needed is supported by Healy (2006) that states “Logistic regression is less restrictive than ordinary sum of squares regression. It does not require normally distributed dependent data or homogeneity of variance (homoscedasticity). Predictions made by ordinary sum of squares regression are based on the observed changes in the independent data itself”. However, endogeneity and multicollinearity do need to be explored in this study though.
Multicollinearity
Multicollinearity refers to a situation when independent variables do not exactly correspond and correlate leading to the values not being precise linear functions of the other variables within the tested model (Hill et al., 2009). If multicollinearity is not fullfilled, then these variables are collinear. This then creates an issue as it becomes hard to see the exact correlations between various variables. When this happens, the true representative values become hard to foresee due to related unclear information. If there is little variation within explanatory variables a bigger issue will be created. In order to test for the presence of multicollinearity, the VIF is determined (Variance Inflation Factor ) (Fig.3). The VIF evaluates the amount of variance an estimated regression coefficient has been increased because of collinearity. Neter, Wasserman & Nachtsheim (1985) state that for
multicollinearity not to be an issue the VIF results must be well below the value of 10. All the values are well below 10 and therefore there are no multicollinearity issues seen here. (See also appendix 2 which provides the correlation values, showing that there is no problematic correlation among the independent variables).
Fig. 3 Model Collinearity Statistics Tolerance VIF Business sector ,954 1,048 Obstacle-corruption ,989 1,011
% time govt regulations ,851 1,175
Obstacle- p.stability ,919 1,088
Country ,816 1,225
Political stability ,871 1,148
Firm size ,920 1,087
21 Endogeneity
An Endogeneiety issue refers to when the error term is uncorrelated with the independent variables. If this is not fullfilled and there are correlations with unobserved variables from outside the model, this can be seen as exogeneity. However, according to Hill et al (2009) there are no scientific test procedures credible. Looking at the variables in this study, there is no evidence to say that there will be issues with endogeneity and unmeasured variables as all data can be judged and looked at by firms and management. If an endogeneity issue does occur, it may be due to the data responses given and judgments made. These judgements can be influenced by unmeasurable variables. However, since there are no tests for endogeneity that fit the current sample in this study, one has to conclude that there will be no issue when it comes to endogeneity.
The method assumptions that can be satisfied have been satisfied showing that there should be no severe problems during this study. In terms of goodness of fit tests (Appendix 3) taken the Chi-squared result shows a Chi-Square value of 588.087 and has a significance level < 0.0001, so we conclude that there is a significant relationship between the dependent
22
4. Empirical results
The table Fig.4 gives an overview of the descriptive statistics of the study. As can be seen from Fig.4, the sample consists of several dummy variables which cannot be seen in depth within this table. To get a greater comprehension of these dummy variables a more in depth frequency table can be found in the Appendix 2. From the total sample of 3155 firms, 715 firms make bribery payments to government officials and 2440 do not. In terms of business sector, there are 2392 non-service firms and 763 service firms. 2997 countries state that corruption is not the biggest obstacle with 158 stating that it is. 2885 state that political stability is not the biggest obstacle and 2700 firms state that it is. Out of the 9 countries researched Vietnam and Turkey contained the largest number of firms with 24,7% and 28%, respectively. Political stability involves 3 categories, high stability includes 555 firms, medium stability includes 1225 firms with low stability including 1375 firms. Independent variable firm size involves 4 categories, with 28 firms having below 5 employees, 1096 firms in the small category, 1185 in the medium category and 846 in the large category. In terms of the non-categorical variables, firm age ranges from under a year old to 140 years old. The percentage of senior management time that was spent in dealing with government officials includes firms ranging from no time spent to 100%. 100% is very high and this means that some firms work with the government constantly. If we look at the correlation table in
Appendix 2 (table 2) we can take an early look at some of the correlations that occur between variables. Our two independent variables are not significantly correlated. Political stability shows significant correlations with all the other variables used except firm size. Firm size correlates with 3 out of the other variables. Country has a high significant correlation rate with the other variables as it correlates significantly with the other variables except industry sector. This is logical and will be taken into account when proceeding with the Logit
23
Fig. 4
Descriptive Statistics
N Minimum Maximum Mean Std. Deviation
Dependent Variable
Bribes Yes/No 3155 0,00 1,00 ,2266 ,41871
Control Variables
Business sector 3155 0,00 1,00 ,2418 ,42826
Obstacle-corruption 3155 0,00 1,00 ,0501 ,21814
% time govt regulations? 3155 0 100 17,45 24,625
Obstacle- p.stability 3155 0,00 1,00 ,0856 ,27978 Country 3155 1,00 9,00 4,6853 2,58210 Firm age 3155 0,00 140,00 15,3876 12,63819 Independent Variables Political stability 3155 1,00 3,00 2,2599 ,73780 Firm size 3155 0,00 3,00 1,9030 ,80114 Valid N (listwise) 3155 4.1 Regression results
The next table (Fig.5) provides the Logit regression results. The results comprise of 3
different models, the first model contains the control variables only, the independent variables are added in the second model and the third model removes the county variables from the equation (to overcome redundancy issues seen in model 2). After this there will be last model which includes the interaction effect between the two independent variables.
Model 1
The first model includes only control variables in order to be able to estimate the effect of each specific control variable on the probability that bribery will occur at firm level. The first control variable to consider is business sector. The dummy variable identifying the service sector produces a negative result however, this is shown to be insignificant (β= -0,148; p>0,05). Therefore this means there is no difference between service and non-service sectors in terms of corruption. The second control variable is whether the biggest obstacle for a firm was seen as corruption. The firms that specified that corruption was the biggest obstacle while doing business also have a significantly higher probability of bribing (β= 1,388; p<0,05). The question involving the % of senior management time that was spent in dealing with
government regulations, also shows a positive and significantly result meaning that firms with a higher percentage of time dealing with government regulations also have a higher
24
firm level (Micro) had a lower probability of bribery. However this was also shown to be insignificant. Firm age as a control variable shows a negative result even though it is seen as a week negative result. However this is also shown as insignificant (β= -0,004; p>0,05). In terms of each country as a control variable, we start with Venezuela as the reference country as it has the lowest mean political stability reading. In reference to Venezuela, each country shows a negative result with each being significant. In terms of the control model (Model 1), the results show that all the countries are negatively related to corruption when using
Venezuela (least politically stable) as a reference point.
Model 2
The second Logit regression model in this study includes the independent variables and the control variables. The two independent variables will be started with and the control variables will be explored after. First, in terms of political stability, we can see we have negative results. This means the higher political stability the less corruption. With low stability as the reference point, medium shows β= -3,672; p<0,05 and high shows β= -2,460; p<0,05.
25
Fig. 5
Model 1 Model 2 Model 2a
B Exp(B) Sig. B Exp(B) Sig. B Exp(B) Sig.
Business sector -,148 (0,122) ,862 ,224 -,153 (0,124) ,858 ,215 -,014 (0,107) ,986 ,898 Obstacle-corruption 1,388 (0,217) 4,010 ,000 1,392 (0,217) 4,022 ,000 ,857 (0,176) 2,357 ,000 % time Govt ,011 (0,002) 1,011 ,000 ,011 (0,002) 1,011 ,000 -,001 (0,002) ,999 ,460 Obstacle-p.stability -,079 (0,199) ,924 ,691 -,084 (0,200) ,920 ,675 -,540 (0,184) ,583 ,003 Firm age -,004 (0,004) ,996 ,270 -,004 (0,004) ,996 ,326 -,007 (0,004) ,993 ,071 Venezuela (ref) ,000 ,000 Vietnam -,623 (0,157) ,537 ,000 3,052 (0,380) 21,150 ,000 Romania -2,593 (0,237) ,075 ,000 1,043 (0,411) 2,837 ,011 Botswana -3,217 (0,319) ,040 ,000 -,758 (0,397) ,469 ,057 Panama -3,670 (0,384) ,025 ,000 -,278 (0,382) ,758 ,467 Lithuania -2,725 (0,303) ,066 ,000 Turkey -2,603 (0,181) ,074 ,000 -2,604 (0,181) ,074 ,000 Czech Rep -2,447 (0,296) ,087 ,000 Uzbekistan -1,163 (0,203) ,313 ,000 -1,201 (0,206) ,301 ,000 P.stability (Low) ,000 ,000 P.stability (Medium) -3,672 (0,384) ,025 ,000 -1,401 (0,174) ,246 ,000 P.stability (High) -2,460 (0,296) ,085 ,000 ,086 (0,099) 1,089 ,389 Firm size (5 or below) ,498 ,627 Firm size (Small) -,699 (0,460) ,497 ,129 -,451 (0,439) ,637 ,304 Firm size (Medium) -,686 (0,462) ,504 ,138 -,378 (0,439) ,685 ,389 Firm size (Large) -,713 (0,467) ,490 ,126 -,345 (0,442) ,708 ,435 Observations 3155 3155,000 3155,000 Pseudo R2 0,258 ,259 ,068 log likelihood (-2) 2791,003 2788,806 3232,399 Chi2 **p<0,05 585,890** 588,087** 144,494**
Czech Rep and Lithuania are seen as redundant and dropped in Model 2. Model 2a has been added to remove country variable interference
26 Model 2a
With the country variables removed, the effect of high political stability, even though it becomes positive also becomes an insignificant result, β= 0,086; p>0,05. Medium stability still remains negative with low as the reference point. Firm size still produces insignificant results. % time dealing with government regulations has changed to a negative result, however this is now also insignificant. We can see that the model is a good fit, if we understand that model 2a is the model with the least variables, it should have the lowest significant Chi-square reading 144,494. This is correct and as more variables are seen by model 1 and model 2, the chi-squared value will also increase.
To summarize the findings we can say the following: In terms of hypothesis 1, there is not enough evidence that the correlation between firm size and bribery is positive or
negative. This is due to firm size producing insignificant results. This is shown throughout the first 2 models. Hypothesis 2 can be supported to a certain extent as the Logit regression results show that political stability has a negative relationship with bribery. However, the results also show that bribery is less likely to occur in an environment of medium political stability rather than in an environment of high political stability. A low political stability environment is still most likely to be corrupt in terms of bribery. The link between the two independent variables will be tested further by an interaction effect which you can view below. By viewing both models 1 and 2, it can be seen that some of the coefficients representing a specific country change from negative to positive as the 2 independent variables firm size and political stability are added to the second model. This shows that the relevance of specific countries change when firm size and political stability are introduced due to unknown factors and variables within each specific country. Model 2a also shows different results when the country variable is removed showing that each specific country has a large effect on the other variables in the model.
4.2 Robustness test
27
removed. However, it is shown as insignificant and therefore we can disregard this result. Implementing the statistical model into a different analytical method that gives similar regressions results, and with the results being only slightly different this implies that the results are robust. Due to this slight difference seen in the Probit analysis, a Hosmer and Lemeshaw goodness of fit test was added, which can be seen in appendix 3 as well. With a significance of 0,003 as a result, this shows a poor fit. However, this could be down to having a large sample size (Archer and Lemeshow, 2006).
4.3 Interaction effect
In this study, an additional analysis is done to test whether an interaction exists between political stability and firm size. “An interaction effect is said to exist when the effect of an independent variable on a dependent variable differs depending on the value of a third variable, usually known as a “moderator variable”” (Jaccard, 2001). In this study, the
28
payments. This proves hypothesis 3 to be true. However, when the separate country variables are added, even though there is still a negative relationship with bribery, it becomes
insignificant. The interaction results described above mean that each country includes other factors which make the interaction effect less significant. Also even though firm size is seen as insignificant in the first regression models, both the independent variables clearly interact with each other once other country factors are not involved. This proved that political stability acts as a moderator variable in terms of firm size and corruption. This could be down to it being a country-level variable and also down to the fact the political stability is effected by institutions, economic development, law, regulations and policies, which in turn effect firm growth, firm performance and firm size. Political stability will therefore act as a moderator in terms of corruption.
Fig. 6
Model 3a (with country) Model 3b (without country)
B Exp(B) Sig. B Exp(B) Sig.
Business sector -,149 (0,124) ,861 ,227 -,010 (0,107) ,990 ,927 Obstacle-corruption 1,391 (0,217) 4,020 ,000 ,863 (0,176) 2,370 ,000 % time govt ,011 (0,002) 1,011 ,000 -,001 (0,002) ,999 ,568 Obstacle- p.stability -,087 (0,200) ,917 ,665 -,550 (0,184) ,577 ,003 Firm age -,004 (0,004) ,996 ,361 -,006 (0,004) ,994 ,091
Political stability (Low) ,000 ,000
Political stability (Medium) -3,664 (0,384)
,026 ,000 ,088 (0,100)
1,092 ,378
Political stability (High) -2,454 (0,296)
,086 ,000 -1,373 (0,174)
,253 ,000
Firm size (5 or below) ,522 ,434
Firm size(Small) -,694 (0,467)
,499 ,137 -,444 (0,444)
,641 ,317 Firm size (Medium) -,670
(0,470)
,512 ,154 -,332 (0,444)
,717 ,454 Firm size (Large) -,695
(0,474)
,499 ,142 -,282 (0,447)
,754 ,528 ZFirm size by ZPolitical
stability -,058 (0,056) ,944 ,299 -,140 (0,052) ,870 ,007 Venezuela (ref) ,000 Vietnam 3,045 (0,380) 21,009 ,000 Romania 1,043 (0,411) 2,838 ,011 Botswana -,734 (0,398) ,480 ,065 Panama -,264 (0,382) ,768 ,490 Turkey -2,598 (0,181) ,074 ,000 Uzbekistan -1,224 (0,208) ,294 ,000
29
5. Discussion
From the different analyses it can be deciphered that the higher the political stability level the lower the probabilityof bribery, however medium political stability countries are more likely to be less corrupt than high. Firm size is seen as insignificant in terms of corruption, however, there is an interaction effect between firm size and political stability when placed together. We also see differences in results when country control variables are added. But what are the possible explanations for these results? The results for hypothesis 2 do partially look a bit U-shaped. This was mentioned by Campante, Chor and Do (2008) who believe that when dealing with political stability and corruption there is a behavior model involving two affects a horizon effect and a demand effect. This behavior model is seen as U-shaped model and this can also be depicted in the results in model 2. However, in model 2, the U-shaped is not fully finished by high political stability and falls just short. Therefore model 2, can be related to the horizon effect and demand effect but only partially. In terms of firm size, the reasoning behind the clear insignificance levels could be down to different reasons. The main reason to bring forward is the fact that there are many factors that could have an effect on the levels of corruption seen by a firm. Placed in the current model, firm size was seen as insignificant. This could be down to the model itself, insignificant data or interfering factors not listed as variables. As we can see by the interaction effect, firm size becomes significant once political stability is also involved, without the involvement of the interference of other country variable factors. The insignificance could mean that on its own and with all the other factors involved, firm size’s effect on corruption is just not potent enough. This phenomenon of other factors may also explain the country variables findings. We see different results when the country control variable is added. For example, the interaction effect between firm size and political stability becomes insignificant when country variables are added to the analysis. This shows that each individual country involves factors which are different and that need to be taken in to consideration and this is shown by the variable causing insignificance to the interaction effect. Some of these factors have been explained in the corruption section of the literature review. They involve, the organizational setup in each country, the historical roots of a country, and the contemporary causes within a country such as institutions,
30
adhere to traditional methods of doing business whether they involve corruption or not (Zaheer, 1995).
It is not the firm characteristics alone that determine whether to bribe or not. The firm’s business atmosphere and institutional surroundings are a vital part as well in the decision to bribe or not (Martin et al, 2007). This could be the reason for different results in terms of country and for insignificant firm size results. Another reason for different firm size results is, if we assume that large firms have a greater capacity to pay, it will mean that the firms with a great capacity to pay are more likely to pay more in bribes than small firms. Therefore, the end result could be either positive or negative and the outcome depends on the effect that influences the most (Svensson, 2003).
6. Conclusion
The purpose of this research project was to add more insight into the effect of firm level determinants in terms of corruption and in this case bribery. The effect of political stability and firm size on the probability of bribery was tested, with a database consisting of firms of various sizes and from 9 different countries around the globe representing different levels of political stability: Venezuela, Vietnam, Romania, Botswana, Panama, Lithuania, Turkey, Czech Republic and Uzbekistan. Corruption has been shown to have positive and negative effects by previous literature however, today corruption is a major hurdle for many firms and bribery has a major part to play in many countries. Recent figures have shown that there is still a big problem in terms of corruption. According to Enterprise Survey (2013), 24.8% of all firms in the world surveyed by the World Bank are expected to give gifts to public officials to get things done (Enterprise Survey, 2013). Overall, from this study it can be concluded that the higher the political stability level the lower the probability of bribery, however medium political stability countries are more likely to be less corrupt than high. Firm size is seen as insignificant in terms of corruption however, there is an interaction effect between the firm size and political stability when placed together. We also see differences in results when country control variables are added and this leads to the fact that each specific country includes a large number of factors which need be accounted for and that take effect in terms of corruption. When dealing with political stability and corruption we see a negative
correlation. A U-shaped model is represented partially as well and this can be linked to the theory described by Campante, Chor and Do (2008) which depicts a behavior model
31
the U-shaped is not fully finished by High political stability and falls just short. The
explanation for the insignificance of firm size is based on there being many unknown factors that were not involved in the test. The model used for the test may also have needed
improving in terms of variables related to firm size. This project looks thoroughly at the effect of firm size and political stability on corruption. The biggest gain of this study is that it is one of the few studies that looks at variables from both the firm-level and country level and tried to determine why some specific firms have higher levels of bribery than other firms.
Fortunately, support for the political stability hypothesis (H2) was found, even though a partial U-shape was also represented. Firm size proved insignificant however, along with the added interaction effect and the removal of the country variables, interesting results were seen. These interesting results show that each country contains many influencing factors that need to be considered when testing corruption. The results of this study have many positive signals and strategic ramifications for organizations, managers and members of government. When companies want to invest across borders in countries of various political stability levels, with an eye on firm size and the interaction effect in mind, they will have an idea of what to expect and what to look out for in terms of bribe payment and corruption. This study gives insight to the management of companies into the methods at which to foresee corrupt acts in their firms and possibly reduce bribe requests. These outcomes also give managers an idea of the environment and situation in which firms will receive bribe requests from
government officials and what circumstances they will be pushed to pay bribes to get things done. For government members, the results show that political stability clearly plays a role in corruption with or without interacting with firm size and in order to reduce corruption, political stability will need to be increased first.
Bribery
As stated earlier, the corruption format in this study is bribery and in order to explore this dependent variable, a unique dataset at the firm level was put together using the World Bank Enterprise Survey. 9 national examples of bribery were explored by viewing 9 different countries, each separated in to their 3 groups based on political stability. By studying bribery, this research project also creates an opportunity to add to previous literature dealing with the causes of bribery and the reasons why it occurs.
32
economically (Perry, 2005). Bribery can lead to an increase in market capitalization for a firm, and according to Cheung et al, “firms bribing high-ranking politicians are awarded projects of larger size”, however, “they have to pay larger bribes to secure these contracts”. Firm performance, the rank of the politicians bribed, as well as bribe-paying and bribe-taking country characteristics affect the magnitude of the bribes and the benefits that firms derive from them (Cheung, Rau , Stouraitis, 2011).
However, when the choice of hiring a firm is based on what will be received and not on a firm’s ability to do the best work for a good value of money, problems are created and the economy suffers as a whole due to the inevitable poor distribution of firm skills, jobs and resources. This can then also lead to reduced industry related quality levels which may have both social and economic consequences on those firms involved (Perry, 2005).
At firm level, bribery can be heavily related to the impact on a firm's reputation, industry links, regulatory relations, and worker morale (Serafeim, 2013). Engaging in bribery also creates a troublesome and damaged business environment by promoting unequal, advantageous and unethical practices. Not only is bribery an act of corruption which plays a help in hand in allowing organised crime to grow, it also adds to a format of crime which is one of the main barriers to the development of a country. It ignores the most basic of laws and regulations, lessens institutional trust and questions democratic standards (anticorruption.ie).
In terms of overall firm performance does bribery really benefit firms though? According to Anna Kochanova (2012), “bureaucratic corruption (bribery) negatively affects both the sales and labor productivity growth of firms.” However, in specific uneven local environments, bribery aids firms which are preferred or are more efficient in terms of bribery, to overcome bureaucratic boundaries and also facilitates firm performance. This shows that the advantages and disadvantages are based on the many surrounding factors which effect bribery and corruption. This is why bribery needs to be looked at closely by both managers and politicians and why this study has shed some light on the factors that effect bribery.
33
acclimatized to corruption and therefore have less stringent standards for judging corrupt practices including bribery (Asiedu and Freeman, 2008). Therefore, it cannot be seen how each firm specifically judges a bribery payment and how it is reported. It is however assumed that the management of firms are capable. As shown by this project, bribery has many
determining factors and definitely needs to be researched further thoroughly, to be able to fully understand the concept.
6.1 Limitations and further research
Throughout the various tests it can be seen that this study has its limits. First of all, in terms of bribery the use of a Logit model may have hindered the study somewhat, as precise
measurement of the amount or percentage of bribery for each firm could not be made. This is due to having to the data used being limited to whether each firm had given bribery payments or not. In terms of enterprise survey data there were many sensitive questions that firms did not want to answer or did not answer correctly. Even though these bits of data were removed, it lowered the final amount of data used in the final data set, which could have hindered the Logit regression analysis somewhat. In order to gain a better understanding of political stability, a larger selection of countries could be used and not only from each stability category. There is also an issue of time the data was taken between 2008 and 2010 from BEEPs with the political stability mean calculated from data coming from 2002, 2007 and 2012. Situations for each variable may have changed and may be different at this precise moment in time. Therefore the data used may not be representative and future research may need a different model. This model produced 2 correct hypotheses and one clearly
34
Appendix
Appendix 1 – BEEPS questions and WGI specifics
Bribery question:
1) 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?
2) 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 = labour 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
3) In what year did this establishment begin operations? Year: 4) Size of firm? Size Less than 5 0 Small =5 - 19 1 Medium =20 -99 2 Large =100 + 3 5) Business Sector
1 = Manufacturing 3 = Other services
2 = Service (retail)
6) What percentage of total senior management’s time was spent on dealing with requirements imposed by government regulations?
Actual % of senior management’s time spent on dealing with regulations……or
0 = No time was spent -9 = Don’t know
35 WGI information
Political Stability and Absence of Violence/Terrorism
Political Stability and Absence of Violence/Terrorism measures perceptions of the likelihood that the government will be destabilized or overthrown by unconstitutional or violent means, including politically-motivated violence and terrorism. This table lists the individual variables from each data sources used to construct this measure in the Worldwide Governance Indicators
Representative Sources EIU Orderly transfers Armed conflict Violent demonstrations Social Unrest
International tensions / terrorist threat
GCS Cost of Terrorism
HUM Frequency of political killings (CIRI)
Frequency of disappearances (CIRI) Frequency of tortures (CIRI)
Political terror scale (PTS)
IJT Security Risk Rating
IPD Intensity of internal conflicts: ethnic, religious
or regional
Intensity of violent activities…of underground political organizations Intensity of social conflicts (excluding conflicts relating to land)
PRS Government stability
Internal conflict External conflict Ethnic tensions
WMO
Civil unrest How widespread political unrest is,
and how great a threat it poses to investors.
Demonstrations in themselves may not be cause for concern, but they will cause major disruption if they escalate into severe violence. At the extreme, this factor would amount to civil war.
Terrorism Whether the country suffers from a sustained terrorist threat,
and from how many sources. The degree of localisation of the threat is assessed, and whether the active groups are likely to target or affect businesses.
Non-representative Sources
WCY The risk of political instability is very high WJP Factor 3.2: Civil conflict is effectively
36 Appendix 2 – Figures and tables
Table 1
Frequency of Categorical Variables
Frequency Percent Cumulative Percent
Bribery No bribe 2440 77,3 77,3 Yes bribe 715 22,7 100,0 Business sector Non-service 2392 75,8 75,8 Service 763 24,2 100,0 Obstacle- corruption
37 Table 2 Correlations BribesYesNo Business sector biggest obst. corr % time govt reg biggest obst. PS Country Political stability Firm size Firm age Bribes YesNo Pearson
Correlation
1
Business sector Pearson Correlation -,026 1 Obstacle- Corruption Pearson Correlation ,087** ,003 1 % Of time with govt Pearson Correlation -,013 -,031 ,003 1 Obstacle- p.stability Pearson Correlation -,060** -,001 -,070** ,098** 1 Country Pearson Correlation -,260** -,006 -,048** ,322** ,261** 1 Political stability Pearson Correlation ,099** -,099** -,043* ,271** ,152** ,221** 1
Firm size Pearson Correlation
,020 -,190** -,036* ,021 -,017 -,084** ,027 1
Firm age Pearson Correlation
-,025 -,033 ,026 ,111** ,020 ,113** ,174** ,178** 1
**. Correlation is significant at the 0.01 level (2-tailed). *. Correlation is significant at the 0.05 level (2-tailed).
38 Appendix 3 – Model fit and robustness test
Goodness-of –fit tests:
Log likelihood Iteration -2 Log likelihood 1 2893,403 2 2795,191 3 2788,886 4 2788,806 5 2788,806 6 2788,806 Chi- Square Chi-square Sig. 588,087 ,000
Hosmer and Lemeshow test added after robustness test below. Pseudo R2 -2 Log likelihood Cox & Snell R Square Nagelkerke R Square 2788,806a ,170 ,259
Hosmer and Lemeshow Test
39
Robustness test:
Probit Model 1 Probit Model 2 Probit Model 2a Estimate Sig. Estimate Sig. Estimate Sig. Business sector -,087 (0,069) ,208 -,090 (0,070) ,199 -,004 (0,062) ,948 Obstacle-corruption ,769 (0,122) ,000 ,770 (0,123) ,000 ,524 (0,107) ,000 % time Govt ,006 (0,001) ,000 ,006 (0,001) ,000 -,001 (0,001) ,512 Obstacle- p.stability -,082 (0,109) ,452 -,084 (0,110) ,445 -,304 (0,101) ,003 Firm age -,002 (0,002) ,271 -,002 (0,002) ,341 -,004 (0,002) ,080 Venezuela (ref) 0a 0a Vietnam -,400 (0,096) ,000 1,664 (0,186) ,000 Romania -1,512 (0,129) ,000 ,529 (0,201) ,009 Botswana -1,843 (0,161) ,000 -,398 (0,198) ,045 Panama -2,059 (0,189) ,000 0a Lithuania -1,580 (0,158) ,000 -,138 (0,194) ,476 Turkey -1,517 (0,103) ,000 -1,517 (0,103) ,000 Czech Rep -1,441 (0,159) ,000 0a Uzbekistan -,703 (0,123) ,000 -,726 (0,124) ,000 P.stability (Low) 0a 0a P.stability (Medium) (0,190) -2,061 ,000 ,055 (0,059) ,346 P.stability (High) (0,159) -1,446 ,000 -,769 (0,090) ,000 Firm size (5 or below) 0a 0a
Firm size (Small)
(0,269) -,395
,142 -,257 (0,262)
,326 Firm size (Medium)
(0,270) -,391
,149 -,220 (0,262)
,401 Firm size (Large)
(0,273) -,409
,134 -,195 (0,264)
40 References
Acemoglu, D. Johnson, S. Robinson, J. and Thaicharoen, Y. (2003)- Institutional causes,
macroeconomic symptoms: volatility, crises and growth, Journal of Monetary Economics, 50: 49 – 123
Aisen, A. and Veiga, F.J. (2013)- How does political instability affect economic growth? European Journal of Political Economy, Volume 29, Pages 151–167
Anaman, K.A and Agyei-Sasu, F. (2012)- Impact of Democratic Political Transition on the Performance of Business Firms in Ghana. Economic Papers, A journal of applied economics and policy ,Volume 31, Issue 3, pages 391–400
Ashforth, B.E. and Anand, V. (2003)- The normalization of corruption within organizations, Research in organizational behavior
Asiedu, E., & Freeman, J. (2009)- The Effect of Corruption on Investment Growth: Evidence from Firms in Latin America, Sub‐Saharan Africa, and Transition Countries. Review of Development Economics, 13(2), 200-214.
Athanasouli, D. Goujard, A. Sklias, P. (2012)- Corruption and firm performance: Evidence from Greek firms, International Journal of Economic Sciences and Applied Research 5 (2): 43-67 Archer, K.J. and Lemeshow, S. (2006)- Goodness-of-fit test for a logistic regression model fitted using survey sample data, The Stata Journal 6, Number 1, pp. 97–105
Arvis, J.-F. Berenbeim, R. (2003)- Fighting Corruption in East Asia: Solutions From the Private Sector, The World Bank, Washington DC
Awokuse, T.O and Gempesaw, C.M. (2005)- Foreign political instability and US agricultural exports: evidence from panel data, Economics Bulletin, 6 (15), pp. 1–12
Bennedsen, M. Feldmann, S.E.Lassen, D.D (2009)-Strong firms lobby, weak firms bribe: A
survey-based analysis of the demand for influence and corruption, EPRU working paper series, EPRU, University of Copenhagen, Department of Economics
Breuning, M and Ishiyama, J. (2011)- Orphans and Political Instability, Social Science Quarterly (Wiley-Blackwell), Vol. 92 Issue 4, p1002-1020. 19p. 4 Charts
Brunetti A, Weder, B. (2003)- A free press is bad news for corruption, J Public Econ 87(7–8):1801– 1824
Campante, F. R. Chor, D. & Do. Q,C. (2008)- Instability and the Incentives for Corruption, Economics and Politics, 21(1): 42-92
Cheung, S. Rau, R. & Stouraitis, A. (2011)- Which firms benefit from bribes, and by how much? Evidence from corruption cases worldwide, 24th Australasian Finance and Banking Conference. Damania, R. Fredriksson, P.G. and Mani, M. (2003)- The Persistence of Corruption and Regulatory Compliance Failures: Theory and Evidence, International Monetary Fund
Dieleman, M. & Sachs, W. M. (2008)- Economies of connectedness: Concept and application, Journal of International Management, 14(3): 270–285
41 Hallward-Driemeier, M. (2009)- Who Survives ? The Impact Of Corruption, Competition And
Property Rights Across Firms, TheWorldbank Elibrary
Healy, L.M. (2006)- Running Head: logistic regression: an overview, Eastern Michigan University, College of Technology
Hellman, J. & Schankerman, M. (2000)- Intervention, corruption and capture: The nexus between enterprises and state, Economics of Transition, 8(2): 325–368.
Hellman, J. Jones, G. & Kaufmann, D. (2003)- Seize the state, seize the day: State capture, corruption, and influence in transition, Journal of Comparative Economics, 31(4): 751–774.
Herrera, A. M. and P. Rodriguez. (2003)- Bribery and the Nature of Corruption, Working Paper, Department of Economics, Michigan State University, MI.
Herrera, A. M. Lijane, L and Peter Rodriguez, P. (2007)- Bribery and the Nature of Corruption, Department of Economics, Michigan State University
Hill, R. C., Griffiths, W. E. & Lim, G. C. (2009)- Principles of Econometrics, 4th Edition, John Wiley & Sons Ltd.
Huntington, S. (1989)- Modernization and Corruption in Political Corruption. A Handbook, Heidenheimer, A.; Levine, V., Eds., New Brunswick: Transaction Publishers, 1989, 377-388 Jaccard, J. (2001)- Interaction effects in logistic regression, Sage Publications, Inc.
Jain, A.K. (2001)- Corruption: a review, Journal of economic surveys, 15(1): 71 – 121
Jensen, N.M. Li, Q. and Rahman, A. (2010)- Understanding corruption and firm responses in cross-national firm-level surveys, Journal of Intercross-national Business Studies (2010) 41, 1481–1504. doi:10.1057/jibs.2010.8
Kalonda-Kanyama, I. (2013)- An analysis of the economic impact of the business environment factors on firm growth in Sub-Saharan Africa, Department of Economics and Econometrics, University of Johannesburg
Kochanova, A. (2012)- The Impact of Bribery on Firm Performance: Evidence from Central and Eastern European Countries. Economics Institute, Academy of Sciences of the Czech Republic. Kouznetsov, A. Dass ,M. (2010) - Does size matter?: A qualitative study into areas of corruption where a firm's size influences prospects for distributors of foreign-made goods in Russia, Baltic Journal of Management, Vol. 5 Iss: 1, pp.51 – 67
La Porta, R. Lopez-De-Silanes, F. Shleifer, A. (1999)- The quality of government., J Law Econ Organ 15(1):222–279
Leff, N.H. (1964)- Economic Development Through Bureaucratic Corruption. American Behavioral Scientist, 8: 8
Martin, K. & Cullen, J. & Johnson, J. & Parboteeh, K.P. (2007)- Deciding to bribe: a cross-level analysis of firm and home country influences on bribery activity, Academy of Management Journal, 50(6): 1401 – 1422
