Democracy, Religion and Inequality
University of Groningen
Faculty of Economics and Business
Bachelor Thesis International Economics and Business

Name Student: Yitian Jing
Student ID Number: s2012790
Student email: y.jing@student.rug.nl
Date Thesis: Jun. 5th, 2012
Name Supervisor: Dr. Robbert K. J. Maseland

First and foremost, I would like to express my sincere appreciation and gratitude to my advisor, Dr. R. K. J. Maseland, for his academic guidance and encouragement throughout the research. He has been very generous sharing his experiences on institutional and cultural determinants on economy, as well as on academic research methodology and beyond. I would not have finished such a thesis paper without his support. His effort and patience would never be forgotten.

Abstract
The democracy’s inequality decreasing effect has been appealing to researchers for long but lacks concentrated argumentation and empirical evidence, as well as the interaction between democracy and religion. This paper conduct an empirical analysis covering time period of 1978-2010 with 86 countries to test the hypotheses of whether democracy decreases inequality and whether an egalitarian religion decreases the influence of democracy. The result shows the direct effect of democracy is weak, however, the hypothesis of religion’s effect on the democracy’s influence is partially confirmed. Therefore, democracy itself has minor influence on inequality while a large proportion of the effect is religion-related.

Key words
Democracy Religion Inequality

摘要：民主对贫富差距的削弱作用长期以来受到学界的广泛关注，但是迄今尚未有统一的论证和支持的实证研究，同样值得研究而没有定论的还有民主和意识形态之间的关系。本文对1978-2010年间86个国家的民主和贫富差距数据进行了实证分析，试图验证民主对贫富差距是否有削弱作用以及平等主义的意识形态是否会削弱民主的效果。回归结果显示民主对贫富差距的直接效果弱而不显著，但是意识形态在影响民主对贫富差距的效果却很大程度得到印证。因此，民主对贫富差距至多仅是微弱效果，而意识形态的加入可以大幅增强解释效果。
关键词：民主 贫富差距 意识形态

Introduction
The relationship and causality issue between democracy and income inequality has been the significant interest of not only economists but also sociologists and political scientists for long. Some researchers, such as Cutright (1967) and Muller (1989) have long held the proposition that democracy and inequality are negatively related in that democracy redistributes political power in favor of the disadvantaged and majority of the society. Others like Chong (2001) argue such a relationship is non-monotonic in that introduction of democracy would benefit the wealthy more at the early stage of social economic development as they display a greater marginal propensity to save and would benefit the mass at the late stage after labors shifting from agriculture to industrial sector during the urbanization process. On the other hand, quite a few empirical studies have shown that there is no considerable relationship between democracy and income inequality, including Jackman (1974) and Hewitt (1977). These studies either criticized the sample selection process of those with contradictory results, or introduce additional variables to eliminate the effect of democracy measurement. In fact even to this day there is no unanimous or even mainstream point of view and all the researchers suggest that future studies should equip with better measurement, larger sample and different specification.
Due to the complicated theoretical arguments and empirical results, the academic field never stopped looking for additional explanatory factors that may interpret the difference between studies and one of the most attractive conclusions involve religion. It is argued by Lipset (1960) that the effect of democracy over inequality is interfered by the religion of egalitarian such as Protestant by demonstrating that “the emphasis within Protestantism on individual responsibility furthered the emergence of democratic values.” This argument was supported by later studies like Burkhart (1997) and Gradstein et el. (2001).
The most recent global financial and economic crisis has had tremendous impact on all major economies and consequently economic growth is largely hampered. Such stagnancy raised the discussion of economic inequality again, a problem largely ignored or concealed at more prosperous times when low-income families were easier to make ends meet. Besides, the Arab Spring revolutionary waves before long have aimed at fighting against dictatorship or absolute monarchy and build a nation of democracy. This paper would contribute to the literature by re-analyzing the relationship between democracy measurement and income inequality level as well as providing policy implications for well-democratic states, newly established democracies and yet-democratized countries, with the help of numerous theoretical discussions going on for the past half century, different kinds of instrumental and explanatory variables introduced by former researchers and most integrated cross-national economic database ever.
The model between income distribution and democracy should not only adopt cross section techniques but also take time factor into consideration. Therefore, an unbalanced panel regression model is to be built studying a certain range of countries within certain period of time. Also it is reasonably believed that income level, socioeconomic development, demographic structure and industrial structure may be good control variables when carrying out the regression model. They will help to confirm or reject the significant relationship between democracy and income inequality and the effect of religion by the end of the research paper. However, due to incomplete data available, undiscovered causal mechanism or inappropriate regression model, the results could possibly be partly inconclusive.

Research Framework
Democracy – Inequality Argument
The idea that reducing inequality in the distribution of political power helps to reduce inequality of wealth dated back to as early as Aristotle (1905) and there has been a long and vulnerable history of debate over the relationship between political democracy and social equality. There are three main arguments discussing the underlying relationship that would be illustrated below.
The first argument is the effect of democracy on inequality. The argument was made by J. S. Mill (1862) by advocating the political movements that have fought for political rights for the under-privileged and was refined later by Key (1949), Lipset (1960) and Lenski (1966). Key (1949) studied the one-party factionalism in the American South and concluded that such political system undermines debate on issues and leads to the lack of competing political power. Without competition, there would be no room for effective political engagement. Lipset (1960) covered a broader topic of the role of elections in industrial countries in his investigation. He discovered the extension of franchise since the last century gave rise to the political competition for the votes from people with little property. The classic example of this is the British Labour Party grew its appearance at the expense of the Liberals at the first half of the century so that the interests of the working class were increasingly satisfied accordingly. Lenski’s (1966) story was very much similar as he demonstrated the increased political power of the majority and disadvantaged called for a more egalitarian distribution of social welfare.
The second argument is the effect of inequality on democracy in that economic inequality may discourage the formation of democratic political structure. There are two theories behind this as suggested by Dahl (1971). One is that privileged class who already gained economic resources may use the privilege to prevent possible political movement that may be against them. Their suppressive actions can range from military crack-down to cultural censorship. Another way in which inequality can demolish democratization is that political democracies are vulnerable against social disparities. A conflict between social groups which may be repressed under the authoritarian could compromise a democratic system.
The third argument is the non-relationship between democracy and inequality. This argument showed that two factors are subordinated to a third variable. Introduced by Kerr et al. (1964), the “logic of industrialization” thesis advocates that technological advancement influences both social welfare and political power distribution. In Marxism, the element that affects both distributions is the interrelationship between social classes.

Democracy Literature Critique
Due to the long history of research, a variety of divergent arguments and explanations and reality significance, the empirical studies of the relationship between democracy and inequality are of a huge number. However, their results have been contradictory for several reasons explained as follows.
The first problem regards specification. As introduced before, theoretically causal relationship between democracy and income inequality can be mutual. Therefore the regression model should consider this and include simultaneous equations. In spite of this, only a few studies of all have taken this into account, for instance, Rubinson and Quinlan (1977) and Burkhart (1997). Most models limited their scopes on the democracy-inequality effect while not specifying the reverse direction. Besides, though there is evidence to believe that the level of income inequality could be an inverted U-shape curve along with time or income level as promoted by Kuznets (1963) and Lenski (1966), quite many researchers only examine the simple linear effect or the logarithmic effect while ignoring the quadratic one, for example, Cutright (1967), Jackman (1974) and Hewitt (1977). These specification errors can bias the remaining coefficient and jeopardize the explanatory power of the whole model.
A second problem considers the measurement. The two most important variables in this study field are democracy and income inequality, neither of which is straightforward and uncontested. For income inequality, all papers published use either the distribution of individual income or family income. However, the income receiving unit in reality can be both and this causes the issue of comparability because the income distribution of individuals can be less equal than that of families. The difference of measurements in democracy is even larger. There have been debatable choices among social and political democracy, political stability and electoral participation. Take electoral participation as an example. A certain level of participation is necessary to democracy though, it is unclear whether a higher level means higher democracy standard or legal obligation as e.g. in Australia (Huntington and Nelson, 1976).
A third problem concerns the sample size. Early studies in 1960s and 70s were enslaved to insufficient database and thus the validity was compromised by small number samples. Most of the researches then studied samples of 20 to 60 countries (see the table for exact numbers), which were extremely vulnerable against the inclusion or deletion of a few specific cases. In addition, the small number samples make it difficult to compare the results between different empirical studies. After the millennium, thanks to database established and maintained by the United Nations and the World Bank, newly published investigations now have notably larger samples. For example, Chong (2001) included 99 sample cases with the help of the data provided by Inter-American Development Bank, Gradstein et al. (2001) has 126 sample cases and Reuveny and Li (2003) has 142.

Religion – Democracy – Inequality Argument
The chaotic and somewhat contradictory results studying the relationship between democracy and income inequality has led to the exploration of additional factors. A clue for such thinking may find its root in the experience of Eastern Europe countries under the communist regime. Although the democracy records in these countries then were notorious, they managed to maintain the income inequality at a low level. One probable reason is the Religion which treasures a tradition of egalitarian. In brief, if the society fosters prudent material life and condemns money worship, it should enjoy a more egalitarian welfare distribution than another society with everything else the same, e.g. democracy level. This may imply that a cross national study must take religion into consideration as an explanatory variable when examining the relationship between democracy and inequality. Specifically, when the religion creeds include ideas of egalitarian, then equity itself could be viewed as one sort of treasure. As a result, vested interest groups would intend to give up some of their real material treasure to the poor and needed in exchange for some virtual treasure of “equity”. On one hand, when democracy level is high, the transfer of wealth from the rich to the poor is done through the institutional channel constructed by democratic authorization, and thus the vested interest groups would have less further intention to conduct voluntary transfer. On the other hand, when the religion features egalitarian, the “exchange rate” of real material treasure against virtual treasure of “equity” would be lower so that the effect of democracy on inequality is compromised. Some researchers offered theoretical support for this argument. Abramson and Inglehart (1995) suggested a substantial body of political literature exists, which connects the cultural religion and political institution. Bisin and Verdier (2000) investigated a dynamic evolutionary model of cultural transmission.
This paper would examine the egalitarian nature of the three major religions in the world, namely Muslim, Christian and Buddhism. Muslim is believed to be the religions that treasures egalitarian the most. One of the five pillars of Islam, Zakat (alms-giving), which is defined as giving a fixed portion of accumulated wealth by those who can afford it to help the poor or needy, is one crucial system designed to close the gap between the rich and the poor. (Ali, Liu and Humedian 2004) This gives support to the theoretically most egalitarian religion of Islam. On the contrary, Buddhism does not intentionally encourage spontaneous or forced fortune transfer from the rich to the poor. Instead, it explains income difference as part of the result of “the three periods (trayo-dhvana)” and “redistribution for sin (karma)”. (Walters 2012) People who did good works in their previous existence are blessed in this life and similarly people who do evil in this would be punished in their afterlife. Despite the fact that all major religions encourage followers to behave, the Five Precepts of Buddhism do not have much to do with fortune or welfare but rather pursuing truth and purity. The situation of Christianity is much more complicated. On one hand, traditional Christian has much importance on religious hierarchy, emphasizing reign and obedience. These ideas are still widely witnessed in Orthodox and some denominations of Catholic. On the other hand, reformed Christian gives more attention on the equal rights of humans and encourages simplicity and mutual assistance. These notions could be seen mostly in Protestant. Synthetically, the integrity of Christian shows a complex attitude toward income inequality. In this paper, Christianity is studied as one entity regardless of its denominations as well as Islam regardless of its branches of Sunnite and Shiah in order to be parallel with Buddhism. To sum up, a descending order of religions with egalitarian thoughts within is Muslim, Christian and Buddhism.

Religion Literature Critique
One of the earliest empirical studies giving significant importance on political culture and religion within the democracy – income inequality equation is the pioneering work by Lipset (1959). He started by raising the hypothesis that a more equal distribution of political rights in the form of a political democracy should bring about a more equal income distribution. He examined data from 48 countries covering European and English-speaking nations and Latin America nations and classified the countries being studied into a 2×2 category where the stable against the unstable and democracies against dictatorships. More importantly, he acknowledged the effect of Protestants in smoothing the gap between different income groups, saying that Protestantism advocates an individual work ethic along with democratic ideals. In other words, the more a country believes and practices in Protestant, the more likely it is to be democratic. Though innovative and theoretical valid, Lipset (1959) concluded that it is unjustifiable that “an increase in wealth, in the size of the middle class, in education, and other related factors will necessarily mean the spread of democracy or the stabilizing of democracy”. Quite a few other researchers followed his thinking of religion and culture as an explanatory variable between democracy and income inequality however reached divergent results. Burkhart (1997) analyzed with the simple linear theory model and the parabolic inverted U-curve theory with an explanatory of population of Protestants included in both models. He confirmed a salutary yet weak parabolic relationship and rejected any variance caused by Protestant population in his empirical study. The most recent as well as very comprehensive study about the effect of different religion on the democracy – income inequality equation is the one by Gradstein, Milanovic and Ying (2001). It raised and proved with empirical study results the viewpoint that the effect of democracy on inequality is through the prism of religion.
The summary of all the literature reviewed is in Table 1. This list of studies is far from complete, but they should represent most of the others in a sense of procedures and results.
Table 1. Literature review Study | Mutual | Religion | Model1 | Inequality | Democracy | N= | Result2 | Cutright (1967) | N | N | L | Sectoral | Stability | 44 | A | Jackman (1974) | N | Y | LO | Sectoral | Bollen’s index | 60 | C | Hewitt (1977) | N | N | L | Individual | Stability | 25 | C | Rubinson & Quinlan (1977) | Y | N | LO | Individual | Stability, voter participation | 32 | A | Bollen & Grandjean (1981) | N | N | Q | Individual | Bollen’s index | 50 | C | Weede (1982) | N | N | Q | Individual | Bollen’s index | 33 | B | Muller (1989) | N | Y | L | Gini | Rustow’s index | 55 | A | Simpson (1990) | N | N | Q | Gini | Bollen’s index | 62 | B | Burkhart (1997) | Y | Y | L, Q | Sectoral | Freedom House | 53 | B | Gradstein et el. (2001) | N | Y | L | Gini | Polity98D, DPI | 126 | C | Chong (2001) | N | N | Q | Gini | Freedom House, Polity III | 99 | B | Sylwester (2002) | N | N | L | Gini | Freedom House | 49 | A | Reuveny & Li (2003) | N | N | L | Gini | Polity III | 142 | A | Lee (2005) | N | N | L | Gini | Polity IV | 64 | A | Ansell and Samuels (2010) | N | Y | L, Q | Gini | Polity IV | 55 | A | Timmons (2010) | N | N | L | Gini | Freedom House, Polity III | 53 | A |
1 L=Linear, Q=Quadratic, LO=Logarithmic 2 A=Negative linear, B=Inverted U-curve, C=No significant relation

Control Variables
Besides religion, there are several other variables that may influence the degree of democracy and thus it is important to control them in the regression model. One most distinguished factor in prior researchers is socioeconomic development. Lerner (1958), Lipset (1960) and Lenski (1966) all argued an increase of socioeconomic development level would produce a larger population of modern labor force and bourgeoisie, whose demand for economic resources redistribution could be met. The association between democracy and socioeconomic development has received empirical support from Cutright (1963) and Smith (1969). In most recent studies, Jackman (1973) and Bollen (1979), it is more popular received that such relationship is curvilinear, the theoretical background of which is parallel to the Kuznets’s inequality curve.
Along with the level of socioeconomic development, the demographic structure should also be controlled for. Because different income groups appear to have diversified population growth rate, high population growth rate promotes inequality by expanding the proportion of low-income groups of people. Besides, younger population structure would further jeopardize income equality by lowering per capita productivity. As argued by Hargens and Felmlee (1984), the influence of population growth on inequality depends on the way population growth affects the age structure.
A third control variable focuses on the industrial structure, specifically the size of agricultural sector. In most developing countries, the pattern of urban-suburban duality development structure is spotted and the income of the agricultural labor force is substantially lower than that of urban areas. People working in the agricultural sector have less access to educational opportunity and political power. Cutright (1967) argued that the larger the size of this “vulnerable” population, the higher income inequality level in this country would be.
With all the analysis in mind, a set of hypotheses is presented.
One last control variable focuses on the communism legacy from the collapse of former Soviet Union. One known paradoxical effect (Gradstein and Milanovic 2002) would emerge if the communism history is not controlled for. Therefore, a dummy variable is created to indicate whether the country was transited from Communism.

Hypothesis 1 Higher level of democracy results in lower income inequality.
Hypothesis 2 A stronger (weaker) prevailing religion of egalitarian will decrease (increase) the effect of democracy on income inequality. Particularly, for three major religions, the decreasing (increasing) effect of democracy on inequality in a descending (ascending) order is Muslim, Christian and Buddhism.

Research Design and Analysis
Measurement
For the dependent variable inequality, the Gini coefficient (GINI) is used as it has been adopted by almost all major databases including the World Development Indicator 2011. It contains high-quality observations covering 157 countries with a time-span from 1978 to 2010, which is the time period this study concerns basically due to data availability. However, except for only a few cases such as the United States and the United Kingdom, Gini coefficient data are not complete but with time gaps in between. To avoid compromising the validity and persuasiveness, countries with no more than 3 observations throughout the 33 years are eliminated from the dataset.
For the explanatory variable democracy (DEMOC), two different standards from two databases are adopted. In the democracy indicator (POLI) provided by Polity IV, democracy is conceived as three essential and interdependent elements, the presence of institutions and procedures of freedom of speech, the existence of institutionalized constraints on the executive power and the guarantee of political participation, respectively. The advantage of Polity is the completeness of data regarding either country or time. Quite many countries have data for democracy back to 1800. The disadvantage of Polity is the lack of transparency of how the indexes are measured. Although the authors provide explanation for the criterion they comply with, it remains to be subjective and unrepeatable. Hence the Database of Political Institutions (DPI) is called for complement. Two variables are used, respectively Executive Index of Electoral Competitiveness (EIEC) and Legislative Index of Political Competitiveness (LIPC). They are very much similar in that both scales range from 1 to 7 and that higher scores mean more competition within the electoral / legislative procedures, thus higher level of democracy. DPI employs a set of clear and objective rules to clarify between different systems, electoral and political competitiveness, which serves to be the biggest advantage over Polity database. Table 2 illustrates the simple correlation table between POLI, EIEC and LIPC. Relative high correlation seems to show that there the three measurements have a lot in common.
Table 2. Correlation between various democracy measures Corr. | POLI | EIEC | LIPC | POLI | 1.00 | | | EIEC | 0.57 | 1.00 | | LIPC | 0.67 | 0.69 | 1.00 |

For the explanatory variable religion, four categories are created to describe the religion situation, respectively Christian (CHR), Muslim (MUS), Buddhism (BUD) and others (OTH). The rule of deciding the dominant religion is that over 40% of total population has the same religion and no other religion has over 25%. All countries that cannot be classified into one of the three religious groups would be categorized as others. Some examples are illustrated in Table 3 and the complete chart is shown in Appendix I. Most data come from CIA The World Factbook, complemented with Wikipedia for “difficult” countries.
Table 3. Examples for religion classification. Country | Category | Description | Belgium | CHR | Roman Catholic 75% | Albania | MUS | Muslim 82% | India | BUD | Hindu / Buddhist 80% |

Sample
In accordance with the provisions of United Nations, Gini coefficients lower than 0.2 are considered perfect equality, 0.2-0.3 (ex) relative equality, 0.3-0.4 (ex) reasonable equality, 0.4-0.5 (ex) relative disparity and no-lower than 0.5 huge disparity. Besides, 0.4 is primarily regarded as the watershed of equality and inequality. As can be seen from Table 4, less than half of the observations are on the equality side.
Table 4. Variable GINI from World Development Indicator 2011 Gini coefficient category | Number of observations | Percent | <0.2 | 0 | 0% | 0.2-0.3 (ex) | 77 | 10.69% | 0.3-0.4 (ex) | 221 | 30.69% | 0.4-0.5 (ex) | 224 | 31.11% | ≧0.5 | 198 | 27.5% | Total | 720 | 100% |

The democracy indicator from Polity IV is estimated in a 0-10 level with a larger number indicating a higher level of democracy. As shown in Table 5, countries with higher level of democracy tend to stay in the same group. For countries with level of 8, 9 and 10, the probability of their staying in the same group is 77%, 78% and 100% respectively, compared to total average of about 66%. A very similar pattern can be observed in the democracy indicator of LIEC from DPI 2010 in Table 6.
Table 5. Variable POLI from Polity IV Est. level | n obs. | % obs. | n countries at least 1 obs. | % countries at least 1 obs. | % obs. in same group | 0 | 120 | 16.67 | 41 | 47.67 | 60.98 | 1 | 29 | 4.03 | 13 | 15.12 | 69.23 | 2 | 21 | 2.92 | 10 | 11.63 | 50.00 | 3 | 21 | 2.92 | 10 | 11.63 | 10.00 | 4 | 15 | 2.08 | 7 | 8.14 | 71.43 | 5 | 40 | 5.56 | 16 | 18.60 | 62.50 | 6 | 57 | 7.92 | 19 | 22.09 | 57.89 | 7 | 90 | 12.50 | 24 | 27.91 | 66.67 | 8 | 150 | 20.83 | 31 | 36.05 | 77.42 | 9 | 95 | 13.19 | 23 | 26.74 | 78.26 | 10 | 82 | 11.39 | 11 | 12.79 | 100.00 | Total | 720 | 100.00 | 205 | 238.37 | 65.85 |

The summary of religion composition is as illustrated in Table 7.

Table 6. Variable LIEC from DPI 2010 Est. level | n obs. | % obs. | n countries at least 1 obs. | % countries at least 1 obs. | % obs. in same group | 1 | 22 | 3.06 | 12 | 13.95 | 41.67 | 2 | 5 | 0.69 | 5 | 5.81 | 0.00 | 3 | 12 | 1.67 | 4 | 4.65 | 25.00 | 4 | 36 | 5.00 | 18 | 20.93 | 5.56 | 5 | 16 | 2.22 | 7 | 8.14 | 71.43 | 6 | 67 | 9.31 | 30 | 34.88 | 56.67 | 7 | 562 | 78.06 | 74 | 86.05 | 91.89 | Total | 720 | 100.00 | 150 | 174.42 | 64.67 |

Most countries with high levels of democracy are modern Western countries, which are featured mostly with Christianity rather than Buddhism and Muslim. This can be clearly seen from the correlation between democracy level and religious dummy in Table 8.
Table 7. Religion composition of the sample Category | n obs. | % obs. | n countries | % countries | Christian | 467 | 64.86 | 43 | 50.00 | Muslim | 125 | 17.36 | 21 | 24.42 | Buddhism | 35 | 4.86 | 6 | 6.98 | Others | 93 | 12.92 | 16 | 18.60 | Total | 720 | 100.00 | 86.00 | 100.00 |

Table 8. Correlation between democracy level and religions. Corr. | CHR | MUS | BUD | POLI | 0.4823 | -0.4097 | 0.0297 | EIEC | 0.3996 | -0.3121 | 0.0329 | LIEC | 0.2842 | -0.1225 | -0.0177 |

Research method
All regression models are divided into a two dimension matrix. To compare between different methods of regression, both simple OLS regression and panel regression with random effects are adopted. The random effect is used because in a fixed effect model, it is assumed that the effect sizes in meta-analysis differ only because of sampling error and they all share a common mean while in random effects model there are random variation because the effect sizes themselves are sampled from a population of effect sizes. When an unobserved country-specific is uncorrelated with each explanatory variable, random effect model is more efficient than fixed effect model. The sources of country-specific differences are not the main focus of this pap er. To compare between different measurements of democracy, democracy would be measured as POLI, EIEC and LIEC respectively.

Results
First we start from the simplest model where the only independent variable is democracy. All results are demonstrated in Table 8 for Model 1.
Model 1 Inequalityit=f (Democracyit)
Table 8. Model 1. Dependent variable: GINI1 Method | OLS | Random effects | No. | (1) | (2) | (3) | (4) | (5) | (6) | Democracy measurement | POLI | EIEC | LIEC | POLI | EIEC | LIEC | Democracy explanatory variable | DEMOC | 0.01 ***(0.00) | 0.03 ***(0.01) | 0.03 ***(0.01) | 0.01 ***(0.00) | 0.01 ***(0.00) | 0.00(0.00) | CONS | 3.64 ***(0.02) | 3.52 ***(0.03) | 3.51 ***(0.04) | 3.65 ***(0.02) | 3.62 ***(0.03) | 3.67 ***(0.03) | ADJ-R2 | 0.04 | 0.05 | 0.03 | 0.04 | 0.05 | 0.04 | N | 720 | 720 | 720 | 720 | 720 | 720 |
1 *p<0.1 **p<0.05 ***p<0.01

The coefficients of democracy variable are positive and almost all significant across different measurements and regression methods, which is contradictory to the hypothesis. The reason of this may be the high correlation of democracy level and Christianity domination, which will be explained later in the Democracy section. Besides, since there are no religious factors included as well as control variables, the explanatory power of the models are quite low. To control for the regression model, four control variables are adopted. GDP per capita in constant 2000 US$ is used to control for the socioeconomic development level as it is the most widely-accepted variable to be controlled for. To avoid heteroscedasticity, all GDP per capita observations are in natural logarithm forms and their squares, namely LNGDPPC and LNGDPPC2. In addition, population proportion under 15 is used to control for demographic characteristics and agriculture sector value added as percentage of GDP for economic structure. Both variables are also in their natural logarithm forms, POP15 and AGRI respectively. Whether a country is transited from former Soviet Union is observed in the dummy variable TRANS. All results are demonstrated in Table 9 for Model 2.
Model 2 Inequalityit=f (Democracyit; Socioeconomicit; Socioeconomic2it; Demographicit; Structureit; Transitionit)
Table 9. Model 2. Dependent variable: GINI1 Method | OLS | Random effects | No. | (7) | (8) | (9) | (10) | (11) | (12) | Democracy measurement | POLI | EIEC | LIEC | POLI | EIEC | LIEC | Control variables | LNGDPPC | -0.69 ***(0.26) | -0.33 (0.26) | -0.50 *(0.27) | -0.35 **(0.17) | -0.35 **(0.18) | -0.44 **(0.18) | LNGDPPC2 | 0.05 ***(0.02) | 0.02 (0.02) | 0.03 *(0.02) | 0.03 **(0.01) | 0.03 **(0.01) | 0.03 ***(0.01) | AGRI | -0.12 ***(0.01) | -0.13 ***(0.01) | -0.13 ***(0.01) | -0.06 ***(0.02) | -0.06 ***(0.02) | -0.07 ***(0.02) | POP15 | 0.46 ***(0.03) | 0.44 ***(0.03) | 0.41 ***(0.03) | 0.14 ***(0.05) | 0.13 ***(0.05) | 0.12 **(0.05) | TRANS | -0.11 ***(0.02) | -0.11 ***(0.02) | -0.13 ***(0.02) | -0.21 ***(0.00) | -0.21 ***(0.04) | -0.21 ***(0.04) | Democracy explanatory variable | DEMOC | 0.01 ***(0.00) | 0.04 ***(0.00) | 0.03 ***(0.01) | 0.01 ***(0.00) | 0.01 **(0.00) | -0.00 (0.00) | CONS | 4.77 ***(0.88) | 3.47 ***(0.90) | 4.18 ***(0.93) | 4.55 ***(0.55) | 4.57 ***(0.56) | 4.95 ***(0.56) | ADJ-R2 | 0.51 | 0.52 | 0.48 | 0.44 | 0.44 | 0.41 | N | 720 | 720 | 720 | 720 | 720 | 720 |
1 *p<0.1 **p<0.05 ***p<0.01

The coefficients of democracy maintain same characteristics and all control variables are significant and of expected signs. Besides, the explanatory powers of models rise to over 40%, a large improvement over model 1. Finally, religion variables and their interaction terms are added into the regression models. All results are demonstrated in Table 10 for Model 3.
Model 3 Inequalityit=f (Democracyit; Religionit; Democracy*Religionit; Socioeconomicit; Socioeconomic2it; Demographicit; Structureit; Transitionit)
Table 10. Model 3. Dependent variable: GINI1 Method | OLS | Random effects | No. | (19) | (20) | (21) | (22) | (23) | (24) | Democracy measurement | POLI | EIEC | LIEC | POLI | EIEC | LIEC | Control variables | LNGDPPC | -1.11 ***(0.24) | -0.82 ***(0.25) | -0.94 ***(0.25) | -0.44 **(0.18) | -0.45 **(0.18) | -0.53 ***(0.18) | LNGDPPC2 | 0.08 ***(0.02) | 0.06 ***(0.02) | 0.07 ***(0.02) | 0.03 ***(0.01) | 0.03 ***(0.01) | 0.04 ***(0.01) | AGRI | -0.10 ***(0.01) | -0.10 ***(0.01) | -0.10 ***(0.01) | -0.06 ***(0.02) | -0.05 ***(0.02) | -0.06 ***(0.02) | POP15 | 0.49 ***(0.03) | 0.48 ***(0.03) | 0.46 ***(0.03) | 0.17 ***(0.05) | 0.15 ***(0.05) | 0.15 ***(0.05) | TRANS | -0.11 ***(0.02) | -0.11 ***(0.02) | -0.12 ***(0.02) | -0.20 ***(0.04) | -0.21 ***(0.40) | -0.21 ***(0.04) | Democracy explanatory variable | DEMOC | 0.02 ***(0.00) | 0.03 ***(0.01) | 0.03 ***(0.01) | 0.01 (0.00) | 0.00 (0.01) | 0.01 (0.01) | Religion explanatory variables | CHR | 0.12 ***(0.03) | 0.03 (0.07) | 0.08 (0.07) | 0.03 (0.05) | -0.04 (0.06) | 0.10 (0.08) | MUS | -0.08 ***(0.03) | -0.12 **(0.06) | -0.05 (0.08) | -0.06 (0.05) | -0.06 (0.07) | 0.01 (0.07) | BUD | 0.06 (0.07) | 0.15 (0.14) | 0.19 (0.13) | -0.05 (0.09) | 0.07 (0.11) | 0.10 (0.11) | Democracy / Religion interaction terms | DEMOC×CHR | -0.01 ***(0.00) | 0.00 (0.01) | -0.00 (0.01) | -0.00 (0.01) | 0.02 **(0.01) | -0.01 (0.01) | DEMOC×MUS | -0.01 **(0.01) | -0.00 (0.01) | -0.02 (0.01) | -0.00 (0.01) | -0.00 (0.01) | -0.01 (0.01) | DEMOC×BUD | -0.02 **(0.01) | -0.04 (0.02) | -0.04 **(0.02) | -0.01 (0.01) | -0.03 *(0.01) | -0.03 **(0.01) | CONS | 6.07 ***(0.82) | 5.07 ***(0.85) | 5.47 ***(0.84) | 4.77 ***(0.56) | 4.81 ***(0.56) | 5.13 ***(0.56) | ADJ-R2 | 0.59 | 0.59 | 0.59 | 0.52 | 0.53 | 0.52 | N | 720 | 720 | 720 | 720 | 720 | 720 |
1 *p<0.1 **p<0.05 ***p<0.01

Democracy
The first column of the result tables uses democracy measurement from Polity IV, second EIEC and third LIEC. Almost all models report positive and significant signs of democracy level. This seems to reject Hypothesis 1. However, one disadvantage of using OLS regression is that it ignores country effects. As introduced in Table 7, the positive sign of democracy may be by reason of their dominated religions rather than democracy level itself, that Christianity is the primary reason why these high democratized countries suffer from high inequality, which would be explained in the interaction between democracy and religion section.

Religion
Among all the models, countries with a dominant religion of Christianity and Buddhism show higher but rather insignificant income inequality level than the others while countries characterized as Muslim are of significantly lower inequality. In different models, on average and all else equal, the direct effect of Islamic rule country has 5% - 12% lower Gini coefficient. Under same conditions, countries of Christianity and Buddhism has 2%-10% and 5%-19% higher Gini coefficient, respectively.

Interaction between democracy and religion
As suggested in research hypotheses, religion can have an influence on the effect of democracy on inequality and this influence is partially supported by the regression models, mainly within the religion of Buddhism. Specifically the coefficients of interaction terms between democracy and different religions are expected to be a Muslim > Christianity > Buddhism. The regression results demonstrate no significant differences among Muslim, Christianity and other countries, but the coefficients of the interaction between democracy and Buddhism are significantly lower. This partially conforms to Hypothesis 2. Because Buddhism does not encourage the wealthy of the society to convey part of their fortune to the needy but rather explain such an income gap as the result of “the three periods (trayo-dhvana)” and “redistribution for sin (karma)”, democracy has to play a more important role in redistributing social welfare than that of Christian and Islamic countries.
One possible explanation that Islamic countries do not show significant decreasing effect on the influence of democracy on inequality is that these countries involved in civil and regional wars in a large scale of time and geographical range. After the Second World War, quite a few large-scale regional wars involve Islamic countries, including the Iran-Iraq War in the 1980s, the Gulf War in the early 1990s, the War in Afghanistan since 2001, the Iraq War from 2003 to 2011, the Arab-Israeli Conflicts throughout the whole post-World War II period and countless regional collision. These wars consumed huge amount of national wealth and such consumption burdened the bottom level ordinary people to a large extent. As a result, the decreasing effect of Muslim on democracy over inequality is jeopardized.
Another feasible explanation refers to the so-called “oil curse”, that oil rich nations are doomed to autocracy and inequality. It is common sense that there is high coincidence between Islamic countries and oil reserve. According to Samuel (2006), oil dependence can lead to skewing political forces in that the concentrated production causes concentrated power. “It became a fisherman’s market for rent-seeking behavior, where those with money jock for positions and influence to acquire lucrative contracts, the revenues from which are used to further bribe and manipulate those in power”. As a result, powerful persons secure the positions of the beneficiaries, creating a vicious circle of inequality in political power and social welfare.

Conclusion and Report
The main findings of this paper are as follows. First, the direct influence of democracy on inequality is weak and insignificant according to empirical analysis. Apart from socioeconomic development level, demographic and industrial structure, there are still complicated implicit social context and values that are difficult to observe and quantify within which democracy takes effects influencing income inequality. Second, the theoretical egalitarian effect of Communist legacy of former-Soviet Union countries is confirmed by regression results. Third, there are distinctive correlations between democracy level and different dominant religious countries. Christian countries maintain higher level of democracy than others while Islamic countries’ democracy records are poor. However, the direct effect of dominant religion on inequality shows that Islamic countries tend to decrease income inequality not through democratization but through religious activities, which gives rise to the final finding that when a dominant religion is featured with egalitarian, the influence of democracy on inequality is weakened. Such difference margin is somewhat offset by the direct effect of religion itself.
Admittedly, this paper is subject to several flaws in methodology and data sample, which may to some extent jeopardize the validity and persuasiveness. First, according to Muller (1988), the effect of democracy on inequality is not instant but requires a period of time. He argued that the stability of democracy is a better predictor of inequality than any one time level. Second, despite the fact that all three major religions are being studied in this paper, quite a proportion of countries and people with other dominant religion or mixed religions are left in the group of others, which shall be clarified given more detailed data. Besides, the group of countries of Christian composes of three main denominations, namely Catholic, Protestant and Orthodox, each with even more denominations within. The complex structures are also observed in Islam and Buddhism, which both have their own denominations and different rules. It is reasonably to argue that the attitudes toward egalitarian among all these denominations are not largely the same. Third, the limitation of available data restricts this paper from more explanatory power, especially the deficiency of cross national and cross time inequality index which may lead to biased estimation. Future researchers should either construct better Gini coefficient database or find a reasonable way to estimate the unknown data between the known.

Reference
Abramson, P. and Inglehart, R. 1995. Value change in global perspective. Ann Arbor, MI: University of Michigan Press.
Ali, S. Rasheed, W. Ming Liu and M. Humedian. Islam 101: understanding the religion and therapy implications. Professional Psychology: Research and Practice, 35(6): 635-42.
Ansell, B. and Samuels D. 2010. Inequality and democratization: a contractarian approach. Comparative Political Studies, 43(12): 1543-1574.
Aristotle. 1905. Aristotle’s Politics. Benjamin Jowett, trans. Oxford: Clarendon Press.
Bisin, A. and Verdier, T. 2000. A model of cultural transmission, voting and Religion. European Journal of Political Economy, 16: 5-29.
Bollen, Kenneth A. 1979. Political democracy and the timing of development. American Sociological Review, 44: 572-87.
Bollen, Kenneth A. and Grandjean, B. 1981. The dimension(s) of democracy: further issues in the measurement and effects of political democracy. American Sociological Review, 46: 651-59.
Burkhart, Ross E. 1997. Comparative democracy and income distribution: shape and direction of the causal arrow. The Journal of Politics, 59(1): 148-64.
Chong, A. 2001. Inequality, democracy, and persistence: is there a political Kuznets curve? Inter-American Development Bank Research Department Working paper #445, Inter-American Development Bank and Georgetown University, Washington, D.C.
Cutright, P. 1963. National political development: measurement and analysis. American Sociological Review, 28 (April): 253-64.
Cutright, P. 1967. Inequality: a cross-national analysis. American Sociological Review, 32 (August): 562-78.
Dahl, Robert A. 1971. Polyarchy. New Haven: Yale.
Gradstein, M., Milanovic, B. and Ying, Y. 2001. Democracy and income inequality: an empirical analysis. CESifo working paper no. 411, Center for Economic Studies & Ifo Institute for Economic Research, Munich.
Gradstein, M. and Milanovic, B. 2002. Does Liberte=Egalite? A Survey of the Empirical Links between Democracy and Inequality. World Bank Policy Research working paper no. 2875, The World Bank Development Research Group Poverty Team, Washington DC.
Hargens, Lowell L. and Felmlee, Diane H. 1984. Structural determinants of stratification in science. American Sociological Review, 49: 685-97.
Hewitt, C. 1977. The effect of political democracy and social democracy on equality in industrial societies: a cross-national comparison. American Sociological Review, 42: 450-64.
Huntington, S. and Nelson, J. 1976. No easy choice: political participation in developing countries. Cambridge, U.S.: Harvard University Press.
Jackman, Robert W. 1973. On the relation of economic development to democratic performance. American Journal of Political Science, 17 (August): 611-21.
Jackman, Robert W. 1974. Political democracy and social equality: a comparative analysis. American Sociological Review, 39: 29-45.
Kerr, Clark, John T. Dunlop, Frederick H. Harbison and Charles A. Myers. 1964. Industrialism and Industrial Man. New York: Oxford University Press.
Key, V.O., Jr. 1949. Southern Politics in State and Nation. New York: Alfred A. Knopf, Inc.
Kuznets, S. 1963. Quantitative aspects of the economic growth of nations: VIII. Distribution of income by size. Economic Development and Cultural Change, 11: 1-80.
Lee, C. 2005. Income inequality, democracy, and public sector size. American Sociological Review, 70(1): 158-81.
Lenski, G. 1966. Power and privilege: a theory of social stratification. New York: McGraw-Hill.
Lerner, D. 1958. The passing of traditional society. Glencoe, IL: Free Press.
Lijphart, A. 1984. Democracies: patterns of majoritarian and consensus government in twenty-one countries. New Haven: Yale University Press.
Lipset, Seymour M. 1959. Some social requisites of democracy. American Political Science Review, 53: 69-105.
Lipset, Seymour M. 1960. Political man: the social bases of politics. London: Heinemann.
Mill, John S. 1862. Considerations on representative government. New York: Harper and Brothers.
Muller, Edward N. 1988. Democracy, economic development and income inequality. American Sociological Review, 53: 50-68.
Muller, Edward N. 1989. Democracy and inequality. American Sociological Review, 54(5): 868-71.
Reuveny, R. and Li, Q. 2003. Economic openness, democracy, and income inequality, an empirical analysis. Comparative political studies, 36(5): 575-601.
Rubinson, Richard, and Quinlan, D. 1977. Democracy and social inequality: a reanalysis. American Sociological Review, 42: 611-23.
Samuel, R. Schubert. 2006. Revisiting the oil curse: are oil rich nations really doomed to autocracy and inequality? Oil and Gas Business, 2006(1): 1-16.
Simpson, M. 1990. Political rights and income inequality: a cross-national test. American Sociological Review, 55: 682-93.
Smith, Arthur K. 1969. Socioeconomic development and political democracy. Midwest Journal of Political Science, 13: 95-125.
Sylwester, K. 2002. Democracy and changes in income inequality. International Journal of Business and Economics, 1(2002, 2): 167-78.
Timmons, Jeffrey F. 2010. Does democracy reduce economic inequality? British Journal of Political Science, 40: 741-757.
Walters, W. 2012. The centrality of karma in early Buddhism. Philosophy East and West, 62(1): 114-27.
Weede, E. 1982. The effects of democracy and socialist strength on the size of distribution of income. International Journal of Sociology, 23: 151-65.

Appendix I Religious composition of the countries Country | Religious composition | Category | Albania | Muslim 82% | Muslim | Argentina | Catholic 91% | Christian | Armenia | American Orthodox 94% | Christian | Burundi | Catholic 60%, Protestant 15% | Christian | Burkina Faso | Animist belief 65%, Muslim 25-30% | Others | Bangladesh | Muslim 85%, Hindu 16% | Muslim | Bulgaria | Orthodox 85%, Muslim 13% | Christian | Bosnia and Herzegovina | Muslim 40%, Orthodox 31%, Catholic 15% | Others | Belarus | Orthodox 90% | Christian | Bolivia | Catholic 95% | Christian | Brazil | Catholic 90% | Christian | Central African Republic | Christian 33%, African tradition 25%, Muslim 15% | Others | Chile | Catholic 89%, Protestant 11% | Christian | China | Communist state | Others | Côte d'Ivoire | African tradition 63%, Muslim 25%, Catholic 12% | Others | Cameroon | African tradition 51%, Christian 33%, Muslim 16% | Others | Colombia | Catholic 95% | Christian | Costa Rica | Catholic 95% | Christian | Czech Republic | Catholic 40%, Protestant 5% | Christian | Dominican Republic | Catholic 90% | Christian | Ecuador | Catholic 90% | Christian | Egypt, Arab Rep. | Muslim 80-90% | Muslim | Estonia | Evangelical 62%, Orthodox 30% | Christian | Ethiopia | Orthodox 50% | Christian | Georgia | Orthodox 83% | Christian | Ghana | African tradition 38%, Muslim 30%, Christian 24% | Others | Guinea | Muslim 85%, Christian 8% | Muslim | Guatemala | Catholic 80%, Protestant 20% | Christian | Honduras | Catholic 97% | Christian | Croatia | Catholic 88% | Christian | Hungary | Catholic 65%, Calvinist 20% | Christian | Indonesia | Muslim 87%, Christian 10% | Muslim | India | Hindus 80%, Muslim 11% | Buddhism | Iran, Islamic Rep. | Muslim 95% | Muslim | Jamaica | Protestant 56% | Christian | Jordan | Muslim 90%, Christian 8% | Muslim | Kazakhstan | Muslim 47%, Orthodox 15% | Muslim | Kenya | Catholic 28%, Protestant 20%, African tradition 20%, Muslim 8% | Others | Kyrgyz Republic | Muslim 70% | Muslim | Cambodia | Buddhism 96% | Buddhism | Lao PDR | Communist state | Others | Sri Lanka | Buddhism 70%, Hindus 15%, Muslim 8%, Christian 8% | Buddhism | Lesotho | Christian 90% | Christian | Lithuania | Catholic 80%, Orthodox 10% | Christian | Latvia | Catholic 50%, Orthodox 37% | Christian | Morocco | Muslim 99% | Muslim | Moldova | Orthodox 90% | Christian | Madagascar | Animist 50%, Christian 43% | Others | Mexico | Catholic 90% | Christian | Macedonia, FYR | Orthodox 64%, Muslim 33% | Others | Mali | Muslim 80%, Animist 18% | Muslim | Mongolia | Buddhism 53% | Buddhism | Mozambique | Catholic 28%, Zionist Christian 15%, Muslim 18% | Others | Mauritania | Muslim 99% | Muslim | Malaysia | Muslim 60%, Buddhist 19% | Muslim | Niger | Muslim 95% | Muslim | Nicaragua | Catholic 95% | Christian | Nepal | Hindus 90%, Buddhist 5%, Muslim 3% | Buddhism | Pakistan | Muslim 97% | Muslim | Panama | Catholic 85%, Protestants 15% | Christian | Peru | Catholic 90% | Christian | Philippines | Catholic 84%, Muslim 5%, Protestant 4% | Christian | Poland | Catholic 90% | Christian | Paraguay | Catholic 90% | Christian | Romania | Orthodox 83%, Catholic 6% | Christian | Russian Federation | Orthodox 85% | Christian | Rwanda | Christian 74%, African tradition 25% | Christian | Senegal | Muslim 94%, Christian 4% | Muslim | El Salvador | Catholic 88% | Christian | Slovak Republic | Catholic 60%, Protestant 8% | Christian | Slovenia | Catholic 58% | Christian | Swaziland | Protestant 35%, Zionist 30%, Catholic 25% | Others | Thailand | Buddhism 95%, Muslim 4% | Buddhism | Tajikistan | Muslim 98% | Muslim | Turkmenistan | Muslim 87% | Muslim | Tunisia | Muslim 98% | Muslim | Turkey | Muslim 99% | Muslim | Tanzania | Christian 40%, Muslim 33%, African tradition 25% | Others | Uganda | Christian 60%, Muslim 16% | Christian | Ukraine | Orthodox 75%, Catholic 14% | Christian | Uruguay | Catholic 47%, Protestant 11% | Christian | Uzbekistan | Muslim 96% | Muslim | Venezuela, RB | Catholic 92% | Christian | Vietnam | Communist state | Others | South Africa | Christian 73% | Christian | Zambia | Christian 60% | Christian |

--------------------------------------------
[ 1 ]. Abstaining from taking life, taking what is not given, sexual misconduct, false speech and fermented drink that causes heedlessness.
[ 2 ]. For details, see Polity™ IV Project Dataset Users’ Manual.
[ 3 ]. For details, see DPI2010 Database of Political Institutions: Changes and Variable Definitions.