Impact of Behavioral & Socio-Political Culture Characteristics on Prevalence of Corruption

By Konstantin Ravvin E-mail: KonstantinRavvin@gmail.com

For Research Methods in Economics ECO 4451 Professor Richard Hofler Spring Semester, 2010

1

ABSTRACT Studies have show that, aside from economic development, behavioral, normative and socio-political culture has a significant impact on corruption levels in a given country. Using 79 observations, this study will attempt to evaluate the influence of behavioral/normative cultural characteristics such as Power Distance, Individuality, Masculinity and Risk avoidance along with socio-political factors such as democratic indices, political participation, and civil liberties on the outcome of corruption (measured in CPI units).

2

1. INTRODUCTION Corruption has always been an obstacle impeding the progress of democracy, rationale and self-improvement in societies across the globe. There are diverse amounts of scholarly arguments that pin corruption as a result of two varying factors, economic development and culture. The intuitive nature of the former argument accounts for the presence of low-income levels, high-income inequality, low access to resources and stagnant socioeconomic conditions as a strong contribution to corruption levels within a country. The later factor, however, examines a cross-cultural view of a country’s traditional values and analyzes how a nation’s “regional distinctiveness” relates to the corruption level. This study will focus on the later of the two factors and observe its impact on regional corruption levels. The objectives involve obtaining quantitative data on cultural values across a single time period of one year (2009) and finding what impact the presence of those values has on the corruption level within a given region. The research will attempt to determine whether traditional and culturally maintained values present a significant underlying cause to policy formulation and efficacy within a nation’s borders. The importance of this study is omnipresent in varying areas, including forecasting, risk management, and policy formulation. In cases of political analysis, the ability to forecast a political event based on recurring significant factors is essential to understanding how, why and to which extent certain political regimes formulate policy & approach social, economic, and unforeseen issues. Although forecasting plays an important role in understanding the most likely outcome of a sociopolitical scenario within a nation, it is overshadowed by the marginal benefit private corporations obtain in understanding the extent to which culture plays a 3

roll in establishing a lack of ethical standards (such as in corruption). A lack of ethics, in an incentive driven world, is the antecedent to corruption. In a corporation, the number one goal is to mitigate risk in order to minimize costs legal, moral, tangible and intangible. The corporation’s ability to understand the regional attributes of culture and their reflection on corruption could allow for a better assessment of investment risk by reflecting the regions cultural make-up towards its corrupt tendencies, this process could be used in both calculating the risk of a direct foreign investment or perhaps a merger with a foreign corporation, determining the long term orientation of the investment. As a rule of thumb, this study could be efficient for personal use for someone traveling internationally. Understanding how cultural values differ by region is detrimental to understanding reasons, motives and rationale behind each action taken. For the personal safety, sanctity and security of a tourist, it would be beneficial to understand how the cultural make-up of a nation may impact the process of traveling to and interacting with the nation’s culture. As a significant indicator of economic progress, corruption is important to stratify and pinpoint, specifically its cause and the long-term orientation and presence. It is a strong determinant of investment risk evaluation, political stability, as well as means and motives of each nation, all of the preceding which have a strong influence on the outcome of an economies stature in the global arena. 2. LITERATURE REVIEW In an effort to further familiarize this research study with the topic, it is necessary to obtain prior research results in an effort to make a focused, reasonable and stable hypothesis. A key constraint of conducting and analyzing cross cultural studies is the validity of the quantitative values put upon measuring values that are generally more qualitatively descriptive in nature. In an effort to eliminate this error, this study will 4

relate to a previous study, which contains the analysis of the impact cultural values have on the prevalence of corruption (Das, Eargle, Esmail 2009, Journal of Social Change). The study was a cross-national panel design focused on the presence of secular vs. traditional values as well as a societies focus on self-expression vs. survivalist mentality in accordance with the civil structure of the country (whether democratic or authoritarian). Panel analyses using Seemingly Unrelated Regression (SUR) technique are conducted, since the residuals for the panels have high likelihood of being correlated with one another. The explanatory factors were quantified using survey analysis from various agencies such as the World Value Survey, the United Nations, Transparency International, Inter-Consortium of Political and Social Research and measured cultural values based on questions including (but not limited to) the importance of God, the justification of abortion and the acceptance of authority. Religion was also factored into the study as an indicator variable to observe countries that contained a 50%+ presence of a certain Abrahamic religion including Catholicism and Islam. Corruption level was obtained through surveyed CPI (Corruption Perception Indices) scores from Transparency International, measured on a scale of 1-10, with 1 exhibiting a high level of corruption and 10 being the least likely to exhibit characteristics of corruption. Corruption Level specifications will be discussed further in the Methods and Data Section. It should be noted that the study was done based on data in 4 distinct years including 1999, 2001, 2003 and 2005. The hypothesis of the study maintained that “socioeconomic resource attainments, greater self-expression values than survival values, greater secular values than traditional values, and institutional arrangements for democracy reduce corruption.” 5

The results of the study showed that a democratic institution indicator tended to decrease corruption and increase the CPI score by 1%, on average in all 4 panels. Although the self-expression factor bared little significance on determining corruption, a secular political and civil orientation tended to increase CPI score by 7%, on average, in all 4 panels. In regards to the religious aspect of the studies, research found that countries that exhibiting an Islamic indicator variable (Islamic population > 50%) on average tended to increase CPI score by 11% in the 2005 panel. Christian countries where found to decrease CPI score by 13% on 2005 but this was a highly controversial measurement considering much of this significance was attributed to Latin countries, exhibiting Catholic leaning. As stated in the final conclusion, the study found that “based on the above findings, clearly, socio-economic resources, rational over traditional beliefs, selfexpression values and democratic institutional arrangements – when combined together – reduce corruption more than the reform variables, thus the hypothesis was fairly supported.” Reviewing the above findings of the Das, Eargle, Esmail study, we find evidence that both refutes and confirms portions of the study. Aside from the various studies conducted on the effects of economic reform and conditions on corruption level (Coolidge, Rose-Ackerman 1997; World Bank 2000), there have also been studies that review the influence of socio-political-cultural characteristics on corruption levels (Paldam 2001; Welzel, Inglehart, Klingemann 2003; Inglehart, Oyserman 2004; Kaufman 2004; Kaufman, Siegelbaum 2004; Inglehart, Welzel 2005; Kaufman, Kraay, Mastruzzi 2005; Sandholtz, Taagepera 2005). Ulrich von Alemann, a prominent German political scientist defined corruption as “a multi-faceted phenomenon, consisting of social decline, deviant behavior, logical exchange, perceptions, and shadow politics” (Politische Korruption). 6

Inglehard (1997) argues that socioeconomic resources generally reinforce secular values over traditional values in society. Secular values create a switch from survival values to self-expression values, this is perhaps because there is more availability for advancement, reducing the need for unethical behavior, “enhancing democracy, increasing elite integrity, and lowering the level of corruption” (Inglehart, Welzel 2005), this trend is present in the prior study showing a significance of secular values over traditional values in having a positive effect on CPI scores. Although as separate entities, only a couple were able to pass a significance test, the model was able to predict corruption level significantly. Despite its absence of influence in the study, Triesman (2005) mentions that corruption decreases when self-expression values are emphasized over survival values. Lipset and Lenz (2000) agree with this trend, citing the lack of democratization and self-expression values in these societies also restrict access to resources for most people hence increasing corruption by denying the people the means to obtain those goods. In regards to religious tendencies, Inglehard and Welzel (2005) regarded most Islamic societies as exhibiting a lack secular values (associated with self expression), this characteristic is not likely to promote an effective democracy, which in turn promotes corruption. They further concluded that the Islamic culture place an emphasis on family, and despite the fact that Islamic society promotes the equal allocation of resources among people by the state (David and Robinson 2006), the absence of democracy enhances the control of resources by few in society, increase corrupt tendencies. In regards to Christianity, Haller and Shore (2005) believed that Christian traditions do not promote secularism and rational values, hence they fail to promote self-expression, leading to a lack in effective democracy. Lipset and Lenz followed this notion by adding “Christian orientation promotes a communitarian lifestyle as opposed to an individualist 7

and universalist lifestyle, corruption is more likely to prevail in this environment”. Paldam’s views on religion (2001) suggest that it has both direct and indirect effects on corruption ranging from emphasis on conservative values to supporting group mentality, all characteristics of a tradition based approach to culture. Although these views provide justifications for the above findings, there is a lack of adjustment to the data, specifically in stratifying the devotion of certain religions to their specified deities, in other words, the studies do not account for the reforms that have taken place in various cultures to allow for a distinction between conservative religious views and reformed views. In retrospect to the previous findings, the general consensus revolves around a struggle between secular (also described as rational) values vs. and traditional values, which tend to determine a culture’s long-term orientation towards democracy. Using these findings, it must be noted that the explanatory variables must revolve around and encompass statistics that illustrate the secular and/or traditional leanings of a nation. Picking up on previously included data; religion will be a significant factor and hence must be included in the study along with socio-political tendencies such as democratic index ratings, along with political participation as well as the existence of civil liberties. Inquiring about ratings in regards to democracy will be most effective when survey statistics are applied due to the difficulty in quantifying certain idiosyncrasies in cultures. Further specifications, including the measurement and derivation methods of obtaining the previously mentioned data will be discussed in the methods section. Behavioral and normative characteristics of cultures that relate to secular vs. traditional values must encompass a few key qualities: self-expression, acceptance of authority, and uncertainty avoidance. These are important elements of cultural behavior because they determine the willingness of a culture to accept outside influences and adapt to change (as would a secular culture) as opposed to displaying a resistance to the 8

changing times. Hofstede’s 4 cultural dimensions are prominent in the field of socioorganization and cultural studies, and are reliable for the measurement of the above requirements. Using 4 cultural dimensions measure power distance, individuality, masculinity and risk-averse behavior, we will be able to measure the influence of behavioral culture on corruption within a country. Further specifications, including the measurement and derivation methods of obtaining the cultural dimension scores will be discussed in the methods section. The aforementioned corruption level will consist of a CPI (Corruption Perception Index) obtained from transparency international. The CPI score is currently the most valid corruption measurement for the study and will provide for a reliable dependent variable. Further specifications, including the measurement and derivation methods of obtaining CPI will be discussed in the methods section. Based on the analysis of previous research studies conducted for similar reasons along with previous significant evidence supporting the impact of behavioral and sociopolitical culture on corruption levels. The hypotheses of this study are as follow: Ho: Behavioral and socio-political culture (as represented by the previously mentioned explanatory variables) does not have a significant impact on corruption levels (as represented by CPI) Ha: Behavioral and socio-political culture (as represented by the previously mentioned explanatory variables) does have a significant impact on corruption levels (as represented by CPI)

3. METHODS AND DATA The variables used within this study are all gathered for the year 2009. The independent variables are chosen for their relevance to the question at hand and will 9

attempt to determine if there is a significant impact of behavioral and socio-political culture on corruption levels (as measured by CPI scores). The findings will allow for an in-depth analysis of the significance that the following independent cultural variables have on corruption levels among public officials and politicians: o Power distance o Individuality o Masculinity o Risk-averse behavior (uncertainty avoidance) o Majority presence of certain religions   Christianity Islam

o Democratic Index o Political Participation Index o Civil Liberties Index o Democratic Institutions vs. Authoritarian Institutions
3A. Assessment of Corruption

The dependent variable in this study is a quantitative value of corruption level (ranging from 0 to 10 in intervals of .1) also known as the CPI as calculated by Transparency International’s Corruption Index. The CPI score progressively decreases the corruption level, therefore, the higher the CPI score, the lower the corruption level. The sources used in the aforementioned study by Das, Eargle, Esmail utilize the same data sets and organization, differing only in the time value of the data (99,01,03,05 as opposed to this studies 09 data set).

10

The sources used to provide valid scores for the CPI trace their lineage to reputable organizations, which account for 13 different surveys that are used to surmise the total CPI, these 10 organizations include: Africa Development Bank’s Country Policy and Institutional Assessments, Asian Development Bank’s Country Performance Assessment Ratings, Bertelsmann Foundation’s Transformation Index, Economist Intelligence Unit’s Country Risk Service and Country Forecast, Freedom House’s Nations in Transit Survey, Global Insights’ Country Risk Rating, Institute for Management Development’s World Competitive Report, Political and Economic Risk Consultancy Asian Intelligence Survey, World Economic Forum’s Global Competitiveness Report, and the World Bank’s Country Policy and Institutional Assessments Survey. For an insight on the methodology used to determine CPI scores, please refer to the Appendix.
3B. Accounting for Hofstede Scores

The explanatory variables that pertain to behavioral and normative culture include the Hofstede scores for the four cultural dimensions that resulted from Hofstede’s cross cultural surveys obtained from the ITIM culture and management consulting firm. Dr. Geert Hofstede conducted perhaps the most comprehensive study of individual values in relation to culture. From 1967 to 1973, while working at IBM as a psychologist, he collected and analyzed data from over 100,000 individuals from forty countries. His work continued into the next decades to encompass over 70 countries and expand on the outreach of the surveys to be directed to students, airline pilots, office workers, and other demographics, expanding outside of organizational research. From those results, and later additions, Hofstede developed a model that identifies four primary dimensions to differentiate cultures based on the survey results:

11

o Power Distance: the willingness of a society to accept hierarchical superiority or view those that are at a higher status to be above lower level statuses in terms of autocratic power and influence. (Measured from a value of 0 to 100, indicating low power distance to high power distance respectively). “A high power distance ranking indicates that inequalities of power and wealth have been allowed to grow within the society. These societies are more likely to follow a system that does not allow significant upward mobility of its citizens. A low power distance ranking indicates the society de-emphasizes the differences between citizen's power and wealth. In these societies equality and opportunity for everyone is stressed.” (Hofstede, Geert. "Geert Hofstede's Cultural Dimensions"). Power distance is an important factor in determining the willingness of a society to accept an authoritarian regime, and hence procure corruption. A higher power distance is also associated with a more traditional and conservative approach to behavioral culture. o Individuality: the degree to which a culture allows for an individual to create an affinity based on their own preferences as opposed to those of the collective group (measured from a value of 0 to 100, indicating high individuality to low individuality respectively). “A high individualism ranking indicates that individuality and individual rights are paramount within the society. Individuals in these societies may tend to form a larger number of looser relationships. A low individualism ranking typifies societies of a more collectivist nature with close ties between individuals. These cultures reinforce extended families and collectives where everyone takes responsibility for fellow members of their group.” (Hofstede, Geert. "Geert Hofstede's Cultural Dimensions"). Individuality

12

is a necessary factor in determining self-expression, which is a core value of democratic institutions, which contribute to a decrease in corruption. o Masculinity: the value a society places on competitiveness, assertiveness, and ambition, along with the collection of material wealth, (male oriented values) as opposed to valuing positive relationships between individuals and quality of life, known as femininity (measured from a value of 0 to 100, indicating low masculinity to high masculinity respectively). “A High Masculinity ranking indicates the country experiences a high degree of gender differentiation. In these cultures, males dominate a significant portion of the society and power structure, with females being controlled by male domination. A low masculinity ranking indicates the country has a low level of differentiation and discrimination between genders. In these cultures, females are treated equally to males in all aspects of the society.” (Hofstede, Geert. "Geert Hofstede's Cultural Dimensions"). Generally speaking, masculine cultures have an affinity for competition and aggressive behavior; this illustrates a tendency towards a traditional view of behavioral culture, and hence put pressure on creating a democratic society, bolstering corruption. o Uncertainty Avoidance: the risk averse behavior of a culture, based on the willingness of an individual from a given culture to attempt actions outside the norm of their respective community (measured from a value of 0 to 100, indicating low risk avoidance to high risk avoidance respectively). “A high uncertainty avoidance ranking indicates the country has a low tolerance for uncertainty and ambiguity. This creates a rule-oriented society that institutes laws, rules, regulations, and controls in order to reduce the amount of uncertainty. A low uncertainty avoidance ranking indicates the country has less concern about 13

ambiguity and uncertainty and has more tolerance for a variety of opinions. This is reflected in a society that is less rule-oriented, more readily accepts change, and takes more and greater risks.” (Hofstede, Geert. "Geert Hofstede's Cultural Dimensions"). The affinity for risk-averse behavior is present in cultures that rely on strong norms to guide their actions, this places a restraint on self-expression as it limits the ability of an individual to reconstruct their behavioral paradigm towards their liking, this limits democratic expression, and hence may increase corruption.
3C. Accounting for Indicator Variables of Religion

As in the previous study, to account for information gaps and large variations in data, indicator variables will be used (also known as dummy variables). Measuring religion by percentage is attainable, however, the large gaps between countries with majority Islam, Christian and Jewish populations are more clearly defined when measured with a dummy variable so as to understand the impact of the country’s major religion as opposed to the average coefficient impact on the dependent variable of having added a 1 unit increase of a certain religion to a countries population. As stated previously in the assumption, the presence of religion emphasizes conservative values that are characteristics of a society that stresses traditional norms for behavioral culture, thus producing a counterproductive effect on corruption level. The indicator variable, expressed with a (D) will be given values of 1 or 0 such that when the country has a majority (religion>50%) religion, a value of 1 has been as opposed to a value of 0, when the country does not have the specified religion as a majority. The dummy will be applied as a separate regressor in the model in effort to find the interactive affect on the intercept so that: D= {1 or 0} 14

Religion>50% (D=1) Yi=Bo+B1X1+B2X2+Bc*(D) Yi=Bo+B1X1+B2X2+Bc Yi=(Bo+ Bc) +B1X1+B2X2 This indicator variable will be used to calculate the average difference in corruption levels between countries that have a specified religion as a majority as opposed to the countries that do not, and will account for lags in corruption levels on the basis of religious makeup. The data on religious populations will be obtained from the CIA World Factbook 2009 National Statistics.
3D. Accounting for Socio-Political Cultural Indicators

The Das, Eargle, Esmail study emphasized the inclusion of socio-political culture factors along with behavioral and normative cultural values. These socio-political values served as key determinants of the countries tolerance towards self-expression and political/finance autonomy based on explanatory factors such as political participation, civil liberties index, the presence of a democratic institution, ext. To account for the presence of these socio-political factors, the Democratic Index, Political Participation Index, and the Civil Liberties Index have been factored into the model. All such sociopolitical traits are obtained from the Economist Intelligence Unit Index of Democracy and have also been included to account for dummy variables that represent fully democratic institutional presence (CPI score > 8) as opposed to those that do not. D= {1 or 0} Full Democratic (D=1) All others (D=0) Yi=Bo+B1X1+B2X2+Bc*(D) Yi=Bo+B1X1+B2X2+Bc 15

Yi=(Bo+ Bc) +B1X1+B2X2 *”The Economist Intelligence Unit’s index of democracy, on a 0 to 10 scale, is based on the ratings for 60 indicators (survey questions) grouped in five categories: electoral process and pluralism; civil liberties; the functioning of government; political participation; and political culture. Each category has a rating on a 0 to 10 scale, and the overall index of democracy is the simple average of the five category indexes. The category indexes are based on the sum of the indicator scores in the category, converted to a 0 to 10 scale. Adjustments to the category scores are made if countries do not score a 1 in the following critical areas for democracy: 1. Whether national elections are free and fair 2. The security of voters 3. The influence of foreign powers on government 4. The capability of the civil service to implement policies. If the scores for the first three questions are 0 (or 0.5), one point (0.5 point) is deducted from the index in the relevant category (either the electoral process and pluralism or the functioning of government). If the score for 4 is 0, one point is deducted from the functioning of government category index. The index values are used to place countries within one of four types of Regimes: 1. Full democracies —scores of 8-10 2. Flawed democracies—score of 6 to 7.9 3. Hybrid regimes—scores of 4 to 5.9 4 Authoritarian regimes—scores below 4” * Summarized methodology of Economist Intelligence Index of Democracy.

16

The reason these factors are taken into consideration represents a multidimensional view of culture that is can apprehend a behaviorally oriented variation in culture as well as a politically motivated variation in culture and how it impacts corruption level.
3E. Linear Regression

Although the previous study performed a Seemingly Unrelated Regression (SUR) for their panel study, this regression will note be required as the study inquires only on the impact of culture on corruption during a one year period of 2009. Instead a linear regression will be performed to determine the best-fit equation with a minimal amount of error: Y=βo+βkXk There are 14 independent variables (represented by βkXk) in total along with a constant (represented by βo). The linear regression will account for all of the variation in the dependent variable (Y, Corruption) that the 14 independent variables explain linearly. Each coefficient will explain how on average a one unit increase in a given independent variable affects the dependent variable.
3F. Non-Linear Double Log Regression

Although linear modeling accounts for the linear variation in corruption level, there may be an incentive to measure the model in a double log model in order to account for all of the variation in Y that the independent variables could explain non-linearly. To do this we simply take the natural logs of both the dependent variable and the independent variables to obtain the equation.

Ln(Y)=βo+ln(βkXk)

17

There are 14 independent variables (represented by ln(βkXk)) in total along with a constant (represented by βo). The double log regression will account for all of the variation in the dependent variable (ln(Y), ln(Corruption)) that the 14 independent variables explain non-linearly. Each coefficient will show the elasticity between the independent and dependent variable, that is to say, what percentage effect a one percent increase of an independent variable has on the dependent variable.
3G. Testing for Multicollinearity

Collinearity occurs when two independent variables within a study are correlated significantly, as such the effect that this may have on the outcome of the study could result from two seemingly different independent variables having a highly similar effect on the dependent variable, this may skew the data significantly as the coefficient estimates may change in an erratic fashion in response to small changes in the model or the data. To account for multicollinearity, we simply find the correlative characteristics of each independent variable with each of the other independent variables.
3H. Sample

The sample size will consist of 79 countries out of a total available of 180; the size of the sample is relatively large (44%) in comparison to the requirements for statistical significance of total sample equaling to 30 or 5% of entire population, the larger of which is to be chosen.

18

4. RESULTS
4A. Collinearity Measure
Corrup~n Corruption PwrDst Indiv Masc Risk islam Christ democracy civillib polpar fulldem 1.0000 -0.6748 0.6555 -0.1136 -0.2227 -0.3851 0.2656 0.6999 0.5768 0.5923 0.7259 PwrDst Indiv Masc Risk islam Christ democr~y civillib polpar fulldem

1.0000 -0.6220 1.0000 0.1234 0.1147 1.0000 0.1750 -0.1814 -0.0267 1.0000 0.3637 -0.2224 -0.0063 -0.0492 1.0000 -0.2678 0.2700 0.0177 0.1385 -0.5773 -0.6225 0.5567 -0.0700 0.0523 -0.6438 -0.4375 0.4060 0.0144 0.1252 -0.6605 -0.6155 0.5799 -0.1120 -0.0361 -0.5151 -0.6515 0.6311 -0.0958 0.0498 -0.3592

1.0000 0.5483 0.5924 0.4697 0.3913

1.0000 0.9310 0.8509 0.7114

1.0000 0.7072 0.5652

1.0000 0.6449

1.0000

According to the above data, the democracy index is highly correlated with the political participation index (.8509) and the civil liberties index (.9310). We are no able to judge whether each factor should be thrown out because of this collinearity due to our unknown significance of each of the correlated independent variables. We must first run a linear regression.

19

4B. Linear Regression for Multicollinearity
Source Model Residual Total SS 306.938115 116.680327 423.618441 df 10 67 77 MS 30.6938115 1.74149741 5.5015382 Number of obs F( 10, 67) Prob > F R-squared Adj R-squared Root MSE = = = = = = 78 17.62 0.0000 0.7246 0.6835 1.3197

Corruption PwrDst Indiv Masc Risk islam Christ democracy fulldem polpar civillib _cons

Coef. -.014814 .0238158 -.012656 -.021515 -.2747903 -.6315224 .4389434 1.691224 -.2616047 .1074877 4.473742

Std. Err. .0113666 .0102009 .0093108 .0073591 .5394915 .4144939 .3956085 .5396822 .1615441 .2320237 1.465874

t -1.30 2.33 -1.36 -2.92 -0.51 -1.52 1.11 3.13 -1.62 0.46 3.05

P>|t| 0.197 0.023 0.179 0.005 0.612 0.132 0.271 0.003 0.110 0.645 0.003

[95% Conf. Interval] -.0375019 .0034547 -.0312404 -.0362038 -1.35162 -1.458856 -.3506944 .6140134 -.5840482 -.3556335 1.547846 .0078739 .0441768 .0059285 -.0068263 .8020392 .1958109 1.228581 2.768434 .0608388 .570609 7.399638

Because the Political Participation Index lacks a significant p-value (.110) and is highly correlated to the Democracy Index (which has a significant p-value), it is safe to drop the Political Participation Index variable from the regression. We can conclude the same for the Civil Liberties Index (.645). Exempting the problem of multicollinearity, we continue to produce the linear model without the omitted collinear variables.

20

4C. Linear Regression Model

Source Model Residual Total

SS 299.911103 123.707338 423.618441

df 8 69 77

MS 37.4888879 1.79285997 5.5015382

Number of obs F( 8, 69) Prob > F R-squared Adj R-squared Root MSE

= = = = = =

78 20.91 0.0000 0.7080 0.6741 1.339

Corruption PwrDst Indiv Masc Risk islam Christ democracy fulldem _cons

Coef. -.0114135 .0198156 -.0094392 -.0201491 -.3343111 -.6322558 .3832085 1.666126 3.938819

Std. Err. .0109875 .0101239 .0091826 .0074302 .5462679 .4055708 .1434662 .5341911 1.370658

t -1.04 1.96 -1.03 -2.71 -0.61 -1.56 2.67 3.12 2.87

P>|t| 0.303 0.054 0.308 0.008 0.543 0.124 0.009 0.003 0.005

[95% Conf. Interval] -.0333329 -.0003811 -.0277581 -.034972 -1.424086 -1.441347 .0970013 .6004437 1.204431 .0105058 .0400123 .0088796 -.0053263 .7554636 .1768357 .6694157 2.731808 6.673207

Analyzing the linear regression model, 3 of the 9 independent variables show a significant p-value, their coefficient estimates model the following effects: o On average, for every one unit increase in risk averse behavior exhibited by a culture, the CPI score decreases by -.0201491. This statement has a possibility of error of 0.8% within the model. This statements states that for the data observed, on average, risk-averse behavior had a negative impact on CPI scores and thus increased corruption. This relates strongly to the traditionalist view of behavioral culture, which aims to remain conservative in the face of change in an effort to preserve itself, hence displaying a strong tendency towards avoiding risk. Generally, this long-term orientation produces a more corrupt society based on its emphasis on survivalist values rather than Secular values, which create a switch

21

from survival values to self-expression values. Enhancing democracy, increasing elite integrity, and lowering the level of corruption (Inglehart and Welzel 2005) o On average, for every one unit increase in democracy index ranking, a CPI score increases by .3832085. This statement has a possibility of error of 0.9% within the model. As previously mentioned by Lipset and Lenz (2000), a lack of democratization restricts access to resources for most people. o On average, a full democracy (CPI > 8) has a CPI score equivalent to 1.666126 higher than a flawed democracy, a hybrid democracy or an authoritarian regime. As compared to flawed democracies, hybrid regimes and authoritarian regimes, it’s not surprising to find that CPI scores increase as democratic institutions grow. This finding conforms with the findings of the previous study in assessing the critical value of democracy as an impact on corruption.

22

4D. Non-Linear Regression Model

A non-linear regression may account for some of the variations in the dependent variable that wasn’t explained by the independent variables in the linear model, indicator variables are excluded from the double log non-linear regression due to their discrete values of 1 and 0:
Source Model Residual Total SS 7.88138884 72.4236477 80.3050365 df MS Number of obs F( 5, 72) Prob > F R-squared Adj R-squared Root MSE = = = = = = 78 1.57 0.1803 0.0981 0.0355 1.0029

5 1.57627777 72 1.005884 77 1.04292255

Corruption~g PwrDst_log Indiv_log Masc_log Risk_log democracy_~g _cons

Coef. -.3384044 .1170287 .0478952 -.4942042 .0220904 4.336382

Std. Err. .3422271 .2201664 .2504092 .2606253 .3612047 2.32048

t -0.99 0.53 0.19 -1.90 0.06 1.87

P>|t| 0.326 0.597 0.849 0.062 0.951 0.066

[95% Conf. Interval] -1.020622 -.321865 -.4512864 -1.013751 -.6979579 -.2894102 .3438128 .5559225 .5470767 .0253429 .7421387 8.962175

4E. F-Static Interpretation

The significance of the F-value is an overall significance test assessing whether the group of independent variables, when used together reliably, predict the dependent variable. We obtained the F-value by dividing the MSM (mean squared model) by the MSR (mean squared residual), which is simply a ratio of explained variance over unexplained variance , this F-value is now weighed against the critical p-value (.05). The p-value of the F-stat helps us determine if the overall model is a good fit for predicting CPI scores.

23

Observing the linear model, there 20.91 times more explained variance than unexplained variance in the model. This hefty statistic is also backed up by a very significant p-value for the F-statistic of nearly 0 percent (.0000 at 4 decimal places). Observing the non-linear model, there is stark contrast between its F-statistic as well as its p-value for the F-stat in comparison to the linear model. There is only 1.57 times more explained variance than unexplained variance and the F-stat p-value is highly insignificant with a 18.03% chance of being erroneous, the non-linear model is henceforth insignificant and unfit for prediction.
4F. T-Statistic Interpretation

The t-value estimates the value of each variable and subtracts the hypothesized value of the given variable, further dividing that value by the standard error to determine whether there is significance in the data presented from the independent variable. The significance is obtained by measuring the p value of the derived t-test. Generally speaking, for a 95% confidence level, we must restrict the probability of error to 5% or less.
4G. Coefficient of Determination

Interpreting the variation in the CPI levels in relation to the independent variables, we look at R-squared and R-squared adjusted. R-squared is simply a measure of the variance in CPI scores explained by the model divided by the total sum of squares (sum of square residuals plus sum of squares explained by the model), this will give you an idea of how much (percentage) of the variation in CPI score is explained by the model. In this case, 70.80% of the variation in CPI scores is explained by the linear regression model as shown above, as opposed to the 29.2% that is still unexplained by the model and left to error.

24

Adjusted R-squared is simply adjusting for the amount of explanatory independent) terms in the model:

Where p is the total number of explanatory variables and n is the total number of observations. When adjusting for a degree of freedom, generally R-squared adjusted is used to determine the impact on the ability of the model to explain variation with every additional independent variable that is added. 5. CONCLUSIONS Conclusively, the study produced significant results that can be applied to reject the null hypothesis that Behavioral and socio-political culture (as represented by the previously mentioned explanatory variables) does not have a significant impact on corruption levels (as represented by CPI). The evidence produced by this study shows that there exists a significant relationship between behavioral and socio-political culture and corruption level.
5A. Comparative Analysis

In comparison to the Das, Eargle, Esmail study, many similarities can be found. Similarities between the significance of secular values (significant in the 2005 panel of the Das study) and their impact on the reduction of corruption level have also been found within this study. The significance of risk averse behavior in reducing CPI scores and increasing corruption reflects the views of the Das study that secular cultural values, such as low risk avoidance have the potential to decrease corruption.

25

In relation to democratic institutions, this socio-political cultural trait exhibits a strong significance in both the Das and the current study, given credence to the previously presented findings that secular beliefs along with democratic institutions have a reductive effect on corruption. Although religion played a fairly regressive roll in the Das study, it seemed to have no significant impact on predicting CPI scores in the current study, this can be attribute to the lack in the paneling process in the current study. In regards to the Das study, significance of the same explanatory variables varied across panels, this is perhaps a flaw that this study has incurred as opposed to the Das study. The research has only accounted for the year 2009 and may have diverse significance if it were compared amongst panels, as it did in the Das study. Another drawback of this study as well as that of the aforementioned Das study is the validity of survey results and subjective quantification of such results. Cross-cultural studies within the fields of sociology, anthropology, economics and psychology aim at identifying the interaction of given cultural traits with behaviors. Upon the adoption of such studies, the respective specialization of cross-cultural sociology, anthropology, and economics has come under scrutiny since its inception. The criticism is deeply rooted in the qualitative nature of culture and behavior, as a result, cross-cultural studies have relied heavily on quantified surveys to obtain results. Therefore it is very difficult to account for the validity and reliability of surveys. When attempting to replicate the results of this study, it would be highly advisable to find more concrete determinants of cultural values that would yield more reliable results, not necessarily based on subjective interpretation. It would be beneficial to dwell deep into understanding what quantitative factors are strongly influenced by culture, such as income inequality, and perhaps extrapolate on those factors to get more reliable results. 26

In reference to validity of surveys, there is always the existence of omitted variable bias. The study’s linear regression was not able to account for 29.2% of the variation in corruption, this is undoubtedly as a result of a missing and significant variable. This does not necessarily mean that adding independent variables at random will solve the problem. Although R-squared will usually always increase when an independent variable is added, regardless of the significance, it is nonetheless important to obtain intuitively convincing explanatory variables that have empirical precedence. The time and resource constraint of this study could not be ignored. While the Das study was conducted over a period of time which was sufficient enough to include data from 4 separate years and account for a generally more professional regressive analysis, the current study was undertaken within the time of one semester, limited to the knowledge of undergraduate economics students. As such, when replicating results it would be beneficial to budget time wisely in an effort to use resources more effectively and efficiently.

27

REFERENCES "International Journal of Cultural Studies". SAGE. 02/10/2010 <http://ics.sagepub.com/>. Hofstede, Geert. "Geert Hofstede's Cultural Dimensions". ITIM International. 02/01/2010 <http://www.geert-hofstede.com/>. Hofstede, Geert. Culture's Consequences, Comparing Values, Behaviors, Institutions, and Organizations Across Nations Thousand Oaks CA: Sage Publications, 2001 - for every detail on Hofstede's research. Deegan, Heather. 2005. “Culture and Development”. In Shireen T. Hunter and Huma Malik (Eds.). Modernization, Democracy, and Islam. Washington D. C.: Center for Strategic and International Studies, pp. 21-34. Forsythe, Nancy, Roberto Korzeniewicz, and Valerie Durrant. April 2000. “Gender inequalities and economic growth”. Economic Development and Cultural Change 48(3): 573-617. Haller, Dieter, and Cris Shore. 2005. “Introduction”. In D. Haller and C. Shore (Eds.), Corruption: Anthropological Perspectives Ann Arbor: Pluto Press, pp. 1-26 Huntington, Samuel P. 1996. The Clash of Civilizations and the Remaking of World Order. Simon & Schuster Paperbacks: New York. Inglehart, Ronald. 1997. Modernization and Postmodernization: Cultural, Economic and Political Change in 43 Societies. Princeton: Princeton University Press. Inglehart, Ronald, and Wayne Baker. 2000. “Modernization, Cultural Change, And The Persistence of Traditional Values”. American Sociological Review 65 (February:19-51). Inglehart, Ronald, and Daphna Oyserman. 2004. “Individualism, Autonomy and Self Expression: The Human Development Syndrome”. In H. Vinken, J. Soeters, and P. Ester (Eds.), Comparing Cultures, Dimensions of Culture in a Comparative Perspective Inglehart, Ronald, and Christian Welzel. 2005. Modernization, Cultural Change, and Democracy. New York: Cambridge University Press. Kaufman, Daniel. 1997. “Corruption: The Facts”. Foreign Policy Lipset, Seymour Martin, and Gabriel Salman Lenz. 2000. “Corruption, Culture, and Markets”. In Lawrence E. Harrison and Samuel P. Huntington (Eds.), Culture Matters: How Values Shape Human Progress. New York: Basic Books, pp.112-124. Moreno, Alejandro. 2002. “Corruption and Democracy: A Cultural Assessment”. Comparative Sociology 1(3-4): 495-507. The Economist Intelligence Unit’s Index of Democracy. 2008. The Economist. < http://www.eiu.com/index.asp?rf=0> 28

Transparency International Corruption Perception Index. 2009. < http://www.transparency.org/> CIA World Factbook. 2009. < https://www.cia.gov/library/publications/the-world-factbook/> Das, Shyamal, Ashraf Esmail, and Lisa Eargle. 2007. “Are the Economic Reforms Real Devils?: A Cross-National Comparison of the Roles that Economic Reforms, Socioeconomic Resources, Values, and Culture Play in Shaping Corruption levels in Developing Countries”.

29

Appendix

30

Summary Statistics of Applied and Omitted Variables

Variable Corruption PwrDst Indiv Masc democracy civillib polpar

Obs 78 78 78 78 78 78 78

Mean 4.923077 61.44872 41.35897 50.08974 6.67 7.522564 5.514103

Std. Dev. 2.345536 21.24663 22.67041 17.53082 2.015092 2.479667 2.027717

Min 1.5 11 6 5 1.9 1.18 1.11

Max 9.4 104 91 110 9.88 10 10

31

0

2

Frequency Histograms of Applied Variables

20

4

40

Power Distance

6 Corruption

32
PwrDst 60 80 8 100 10 ycntsDrer0 ycntsDrer0 euq w2 euq w2 004 004 02 02 00 00 01 01 08 08 06 06 1 1 8 8 6 6 4 4 P P F F noineuqr00 yc tp r er1 o2 8 6 5 4 3 C F 0 2 Frequency 4 6 8 10 0 1 Frequency 2 3 4 5

Individuality

0 0

2

Frequency 4

6

8

0

2

Frequency 4

6

8

ycneuvier0 q dnI 002 1 8 6 4 F 20 ycneuqer0 c0aM s02 1 5 8 6 4 F 0

40

Indiv

60

80

100

Masculinity

50 Masc

100

33

Risk Averse Behavior
8 0 0 2 Frequency 4 6 ycneuqer0 0si2 k04 1 5 8 6 R F 50 Risk 100

Democracy Index
5 0 2 1 Frequency 2 3 4 ycareuqer0 ycn com01 e2 8 6 5 4 3 d F 4 6 democracy 8 10

34

laudiser sulp tnenopvidnm00 0 o2 2I 1 8 6 4 C

laudiser sulp tnentsDrw0 laudiser sulp tnentsDrw0 opm02 opm02 004 004 04 04 02 02 00 00 o6 o6 22441 1 8 8 C C P P

Best Fit Line of Each Independent Variable in Regards to Residuals

0

0

20

20

Individualism

40

Power Distance

40

35
PwrDst Indiv 60 60 80 80 100 100 -2 Component plus residual 0 2 4 6 -4 Component plus residual -2 0 2 4

laudiser sulp tnenopmsi0 002 k o4 2 41 5 C R

laudiser sulp tnenopmaM laudiser sulp tnenopmaM c000 c000 s02 s02 o4 o4 22441 1 5 5 0 0 C C

0

0

Risk Averse Behavior

Masculinity

36
50 50 Masc Risk 100 100 -4 Component plus residual -2 0 2 4 -4 Component plus residual -2 0 2 4

laudiser sulp tycaropm00 nen co o2 e4 1 8 6 C d

2

4

Democracy Index

37
6 democracy 8 10 0 Component plus residual 2 4 6

38