The effect of electoral competition on state
level corruption in American states.
A study on the influence of inter and intra-party competition in gubernatorial elections on state level corruption in the United States
Bachelor thesis
Sam Akihary
11060891
Seiki Tanaka
Varieties of democracy and democratization
25 June 2018
Introduction
Corruption is one of the largest problems societies all around the world face, problems resulting from corruption are severe and putting an end to this phenomenon would improve the lives of millions of people. Corrupt politicians and public officials can be found everywhere but in some countries a lot more than in others. While corruption is relatively rare in some other countries, mainly in the developed world, others have to have to deal with it on a daily basis. Differences between countries are thus huge between countries. Because of its highly negative effects and the large differences between countries corruption has been studied a lot by scholars. Past research on this topic has so far discovered two important causal mechanisms. The first and most powerful of these is that the Gross Domestic Product (GDP) has a strong effect, richer countries are far less corrupt than poor ones (Treisman 2000:415). The second finding is that democracies are on average less corrupt than authoritarian countries (ibid). These findings however only partially explain the large differences between countries (ibid). Differences occur even between rich and democratic states (Transparency International: 2018). In Western Europe for example, a region with high levels of GDP and democracy, a country like Italy is far more corrupt than Northern European countries like Sweden and Norway who are one of the least corrupt countries in the world (ibid). It is thus obvious that there is still a lot to discover on the causes of corruption.
This bachelor thesis will contribute to earlier research by focusing on the effect of electoral competition on public corruption. Earlier research on corruption in democracies has primarily focused on the effect of different electoral rules. According to one study majoritarian democracies are less corrupt than democracies with a proportional system (Persson et al 2000), others however refute this (Faller et al 2013). Electoral competition was a mediating variable in these studies, electoral rules are thought to influence competition and thereby corruption (Persson et al 2000). This bachelor thesis will take a closer look at this mechanism by studying the effect of competition on corruption, both competition between parties as competition between politicians from the same party.
To analyze this the case of the United States is used. The advantage of using the US is that we can control for many national-level factors by comparing different states. State level corruption, corruption by public officials in these states, and electoral competition in gubernatorial elections will be analyzed. The research question of this bachelor thesis is: Why are some states in the US more corrupt than others? The fifty different states within the US vary a lot when it comes to both electoral competition and state level corruption. Although these states differ in GDP and how democratic they are, all states are when looked at the global level (very) rich and democratic. Analyzing differences in corruption between states in the US is therefore a good way to explain
why some rich and democratic countries are more corrupt than others.
The biggest limitation of this thesis is that corruption is very hard to measure, finding correct data on corruption levels in the fifty different states is thus hard if not impossible to do. Transparency International is an organization that gives reliable data on corruption but focuses on countries as a whole and not on differences within countries. This thesis uses data on the number of convictions of public officials in corruption cases. Although it is the best data on corruption around, it remains imperfect because not every corrupt officials gets convicted in a court. Because of this the available data on convicted public officials does not fully represent the actual corruption levels within the states. Some states may be more corrupt than the data us while others may be less corrupt than the data makes them look like. Since corruption is the dependent variable in this thesis the effect of electoral competition on corruption can thus not be measured fully correct.
This thesis is structured as follows: the first section is a literature review in which earlier studies on corruption and the relation between democracy and corruption will be discussed. At the end of this section three gaps within the literature are pointed. After the literature review this thesis gives its main argument. The argument is that inter-party electoral competition leads to less
corruption, while intra-party competition increases corruption. The third section discusses how this thesis has done its research. In this section the case of gubernatorial elections will be discussed, an operationalization of competition and corruption is given and the statistical method of using a linear regression is explained. In the analysis this thesis discusses the results of the linear regression.
Literature review
The structure of this literature review is as follows: the first part will look at the most important concept in this thesis,corruption. Earlier studies which have tried to explain why some democracies are more or less corrupt than others will be discussed in the second. The third and final part will look at existing gaps in the literature on corruption and democracy. Electoral competition has only rarely been used to explain corruption and no study has taken both intra-party as inter-party corruption in account.
Corruption
Transparency International defines corruption as ‘’ the abuse of entrusted power for private gain’’ (Transparency international 2018). Corruption can be both public, when done by government officials, and private when done by private individuals (Tanzi 1998:564). This thesis will focus on public corruption meaning the abuse of power for private gain by public officials. They can abuse
their power and thereby be corrupt in a large number of ways. Examples are accepting briberies, embezzlement, nepotism, patronage and electoral fraud (Rose-Ackerman 2016:10). These examples have many differences but are similar in that someone who was entrusted power from the citizens uses his power for his own benefit (ibid). The benefit can be economical like with bribery or embezzlement but it is not necessarily so. Nepotism and electoral fraud benefit corrupt officials by benefiting friends and relatives or by making them more likely to stay in office. Corruption can furthermore be grand or petty, grand corruption involves political leaders and petty crime involves (low-level) bureaucrats (Tanzi 1998: 564). The corrupt officials does not necessarily have to be the one taking the initiative, if he for instance accepts the bribe the briber (the one bribing the officials) has taken the initiative (ibid). This thesis will look at corruption coming from both elected officials as some bureaucrats.
Corruption is harmful and unjust because it goes against the principal agent relation in politics, in which public officials are agents for the citizens (Rose-Ackerman 2016:9). Corrupt officials harms this relation in two ways: by acting inconsistently with the mandate that the principal (the citizenry) has given them or by acting only in line with their mandate if they get a paid for it (ibid). A bureaucrat for instance who favors everyone that pays him extra money or who demands money in exchange for services breaks the mandate because he does not act in the way the citizens expect him to do, namely being an unselfish servant of the citizens.
Causes of corruption
Given how big of a problem corruption is it is not surprising that scholars have already given
attention to this subject. Much research on the causes of corruption has already been done. Research on cross country differences has discovered two important mechanisms. The first of these is the causal relation between income (measured by GDP) and corruption, richer countries are on average way less corrupt than poor countries (Treisman 2000:415). This is not very surprising since one would expect that people with a very low income, like many public officials in developing countries, are more likely to accept bribes than public officials with a high income. A second important finding is the strong causal relation between democracy and corruption, studies have consistently revealed that democratic countries are less corrupt than authoritarian countries (ibid). Moreover, the level of democracy is important. This means that the more democratic a country is the less corrupt it is, the most democratic countries in the world as measured by Freedom House are in general also the least corrupt countries(ibid). To explain the effect of democracy on corruption, scholars have pointed out three mechanisms which link democracy to lower levels of public corruption. The main reason is that democracy raises the costs of being corrupt for public officials and most of all politicians (Nur-tegin and Czap 2012:53). It does this in three ways, the first is that
freedom of speech and free press both it more likely that corrupt public officials will be caught (ibid). Secondly a strong and independent legal system makes it furthermore more likely that corrupt officials will be punished for their crimes, further raising the costs of being corrupt (ibid). Thirdly, elections raise the cost of corruption because it enables voters to punish corrupt politicians by voting them out of office (ibid). For politicians this means that a democracy being corrupt is more risky in democracies since it can decrease your chances of being reelected (ibid).
Democracy and corruption
Research has however also shown that not every rich and democratic country is the same. Some of them like Italy, South Korea and Japan are significantly more corrupt than others like Norway and Sweden, which are the least corrupt countries in the world (Transparency International 2018). Recent studies that have tried to explain these differences have so far used electoral rules as the main independent variable. Persson et al (2003:960) argue that democracies with a proportional voting system are more corrupt than democracies with voting districts. They argue that in majoritarian systems accountability of politicians by voters is more direct and that competition between these politicians is therefore stiffer in majoritarian democracies (ibidem: 962). This competition disciplines politicians and thereby makes them less likely to be corrupt (ibid). They also argue that large voting district lead to less corruption (ibid). In small voting districts honest candidates may not always be available and voters thus may elect a dishonest (and therefore
corrupt) politician because of their party preference (ibid). Districts with a clear party preference are therefore less likely to oust a corrupt politician (ibid). Other scholars however take the opposite standpoint. Faller, Glynn and Ichino (2013) argue that majoritarian systems are actually on average more corrupt than PR systems. In a longitudinal study of ten (developing) countries they found out that a change in electoral system did not result in much more or less corruption (ibidem:25). Chang (2005) even argues that in open list PR systems intra-party competition can drive politicians to become more corrupt. He looked at local elections with an open list proportional system in Italy and his findings shows that increased inter-party electoral competition led the politicians to become more corrupt (Chang 2005:716). His argument is that when “personal votes are expensive yet critical for politicians to win elections,uncertainty drives incumbent politicians to seek illegal rents in order to finance campaigns’’ (ibid).
Gaps in the literature
Although much research has already been done on this topic a number of things that can explain corruption in democracies have not been studied (enough). The first of these is that past studies
have primarily focused on electoral systems as the explaining factor while using electoral competition as a mediating variable. Chang has so far been the only who focused on electoral competition and not on electoral systems. The lack research on the effect of electoral competition is surprising given that the effect of competition (non electoral) on non political corruption has been studied. Ades and Tella (2003:991) for instance have studied the effect of competition on corruption for bureaucrats and found that competition reduces corruption. Because electoral competition is different than competition between bureaucrats (bureaucrats do not have to win elections) the effect of electoral competition on corruption is still largely unknown. Persson et al (2003:960) assume that electoral rules affect public corruption by influencing competition, but since it is unknown whether competition actually influences corruption it may be that the mechanism between electoral rules nd competition is different.
The second gap in the literature is that past studies have not focused on the possible difference of intra-party and inter-party competition. Chang (2005) who was the only one to focus on competition did this solely on inter-party competition. No study to date has studied the possible effects of both inter-part and intra-party competition on amount of public corruption. It may be that inter and intra-party competition have different effects on corruption since personal based voting is lower inter-party elections because the candidates differ more on policy and political ideology than politicians from the same party. Studies on the effect of electoral systems on public corruption often have competition as a mediating variable but only discuss competition in general without
distinguishing between inter and intra-party competition.
The third and final gap is that all studies focused on cross national differences or on
relatively corrupt countries. Variation within one country that is not particularly known to have high levels of corruption has not been researched yet. Chang (2005) for instance focused on Italy and Faller et al (2013) focus on African countries. Countries with relatively low levels of corruption in the rankings of Transparency International are seen as non corrupt but it may be that there is some variety within the country, certainly if the country in question is very large. It is interesting to know whether such variety exists and if so what causes these differences.
Argument
The argument that is made in reaction to the literature review is that inter-party electoral
competition lowers public corruption while intra-party competition increases public corruption. Competition between parties makes politicians less likely to become corrupt while competition between candidates from the same party makes them more likely to become corrupt. Inter and intra-part competition are different in character because the latter is based more on personal votes (Chang
2005: 719) and less on party preference. Two mechanisms cause the negative effect that inter-party competition has. The first one is the disciplining effect that competition has on elected officials because it causes uncertainty for these officials to be reelected. The second mechanism is that higher competition increases the monitoring of politicians, this makes it more likely that corrupt officials will be voted out of office. Intra-party competition on the other hand increases corruption because these elections are based more on personal votes.
Disciplining effect
A number of studies has already used the disciplining effect to explain why competition would lead to less corruption. The theory comes from rational choice theory and assumes that actors (in our case the elected officials) are rational beings who would like to stay in office. To do so they should win the next election and elected officials thus keep the next election in mind when they make decisions. The degree of competition these officials have is important because it increases their uncertainty of being reelected. This uncertainty influences these officials and makes them less likely to make decisions that would harm their chances of being reelected. Being corrupt is one of these things, when an officials turns out to be corrupt some voters are likely to vote for someone else in the next election. Accepting bribes or committing another corruption crime has thus more risks when competition is high. The amount of competition is thus of great importance of the costs that corruption brings. Higher levels of competition make corruption more costly because it would severely hurt the chances of being reelected if the voters would find out about the corruption. This means that because actors are rational high levels of electoral competition prevent politicians from becoming corrupt. When competition is very low politicians do not have much uncertainty of being reelected and being corrupt is less costly. A possible corruption scandal would with low levels of competition not result in losing the election, in contrast to when competition is high. The level of competition thus often decides whether it is rational, from a selfish point of view, for a politician to become corrupt. High levels of competition makes corruption to costly and thereby ‘’disciplines’’ politicians.
Monitoring effect
Besides disciplining politicians, high levels of competition also lead to better monitoring of these politicians by the voters. When competition levels are high, it becomes more likely that voters will vote a corrupt incumbent politician out of office. Voters who always choose the same party are not likely to change their party preference based on a corruption scandal. Those who do not have a preferred party look for things that can decide their vote, a corruption scandal with a politician or a
party involved will in their case often be decisive. In the American case states in which the same party usually wins have lots of voters who always vote for the same party, for instance many evangelists who always vote for a republican candidate. These voters are not likely to vote for a candidate of the democratic party in reaction to a corruption scandal with the republican candidate involved. In the so called swing states with high level of inter-party competition many voters often switch between the republicans and democrats, these voters are way more likely to base their vote on a corruption scandal. Corrupt politicians in states with high levels of electoral competition are therefore more likely to be punished by voters than those in states with low levels of competition.
The monitor and disciplining effects of electoral competition also explain why democracies in general are less corrupt than authoritarian states. In non democratic countries electoral
competition is non existent or very low. The elections that authoritarian countries have, sometimes like in the case of China there are not even any, are rigged or lack any serious competition from the regime (like in Russia). Because of this politicians in these states, in particular political leaders, have much certainty that they will stay in power. The disciplining effect is thus non existent in these countries, being corrupt has much less risks than in democracies since it is highly unlikely that it could harm the chances of being reelected. Also non existent is the monitoring effect, the absence of electoral competition makes it impossible for voters to punish corrupt politicians. Elected officials who engage in corruption will thus never be voted out of office by voters in authoritarian countries, simply because there are no serious elections. Electoral competition in democracies is always higher than in authoritarian countries, so even countries/regions in democracies with low levels of
competition still have much more competition than authoritarian states.
Intra-party competition
In contrast to inter-party competition intra-party competition is likely to cause more corruption and not less. Chang (2005: 719) has made the argument before that elections with much intra-party competition rely more on personal votes. Candidates in these elections cannot rely on the reputation of the party, in our case the Democratic and Republican party, but have to stand out in other ways (ibid). The candidate must distinguish himself from other candidates from the same party and thus relies on his or her personal reputation among the voters (ibid). Since other candidates are from the same party it is harder to distinguish yourself on ideological standpoints when compared to
elections between candidates from different parties. Because of this candidates in elections dominated by intra-party competition (like the primaries in our case) seek personal based votes, some of them will do this by giving their supporters personal favors (ibid). When elected those candidates will give their supporters personal favors and thereby be corrupt. Not all candidates will
seek personal based votes and give their supporters personal favors but some of them do and this causes intra-party competition to increase public corruption. The more uncertain a politician is of being elected or reelected the more desperate he becomes to seek a personal based campaign
(ibidem:720). So the more intra-party competition there is, the more uncertain candidates become of being elected and the more likely it is that they will become corrupt (ibid). It is thus expected that states with more intra-party competition, competition within the parties, have more public
corruption.
This effect/mechanism does not occur with inter-party competition because the need for personal votes is less high. When competing with a candidate from another party it is much easier to distinguish yourself based on ideology or party reputation. Some voters have a clear preference for one party while others focus on the standpoints of the candidates. For candidates in elections dominated by inter-party competition it is thus much easier to convince certain voters. Candidates can emphasize their differences on certain important issues, like abortion in the US, to sway voters to their side. Personal votes are thus not needed to the extent they are in elections dominated by intra-party competition. Candidates furthermore have the party behind them which means that they have the access to the campaign funds of the party. They are thus way less dependent on donors from supporters which further decreases their chances of becoming corrupt.
Research methods
To analyze the effect of electoral competition on corruption of politicians this thesis uses a quantitative, more specifically a statistical regression analysis. Gubernatorial elections and their effect on state level corruption within the USA will be used as a case study, all fifty different states of this country will be analyzed. In this section the case will be explained, an operationalization of corruption and electoral competition will be given and the cross sectional method will be explained.
The case
Competition in gubernatorial elections and its effect of state level corruption within states in the United States is used as a case study for the effect of competition on corruption. Gubernatorial election are held every four years (in some states once every two years) in every state to elect the most influential political figure in state politics, the governor. The candidate with the most votes wins, so in contrast to the presidential elections no districts and delegates. Both the republicans and democrats have one candidate in the general election with in some cases a third party or
elections are held with an unlimited amount of candidates, if there is only one candidate the primary elections are usually not held. Incumbent governors also have to participate in these primary
elections first if they face a rival candidate from their own party. Most states have a imposed a term limit for their governor, often two terms of four years like the president in the US has. Governors have much power on state level politics. He or she is the chief executive officer of the state and is responsible for both the implementation of state laws and the operation of the states executive branch. He also forms the government of the state and often has considerable power over the states budget. Moreover the governor has veto power which means that he can block any bill (in some states this veto power can be overruled by the state legislature). A corrupt governor can thus have considerable impact on the states corruption level.
There are number of reasons for why gubernatorial elections in the US are a good case study for this thesis. The first is that the fifty different states vary when it comes to electoral competition. Some of them like Ohio are so called swing states in which competition between democrats and republicans is tough, others like many states in the Midwest have a clear party preference and thus usually the same party as the winner. It must be said that this only counts for inter-party competition and that might be a different story for intra-party competition. States in which a candidate from the Democratic party usually wins may face more intra-party competition during the democratic primary. In these states however one of the two rounds in the election has already been decided on forehand and winning candidates thus have less competition in total.
The second reason why this is a good case study is that it creates the opportunity to analyze and explain differences in corruption level within one rich and democratic country. As said earlier in the literature review, studies on the relation between democracy and corruption have primarily looked in to cross national differences and if they focused on a single country it usually was a relatively corrupt country. An analysis that focuses on differences within a country has the
advantage that the cultural or historical differences within the same country are much smaller than those between different countries. States in the US vary in their culture and history but this variation is much smaller than those between countries from different continents or countries in Europe. Because other factors like culture, history and political institutions are smaller differences in corruption level are more likely to be caused by competition.
A third and final advantage is that the effects of both inter-party as intra-party competition can be measured. As mentioned earlier gubernatorial elections have primary elections in which candidates from the same party compete with each other. The elections thus have two stages, one with intra-party (the primary election) competition and one with inter-party competition (the main election). Because of this a case study of the gubernatorial elections can take both inter as intra-party competition in to account and measure the influence of both types of competition on
corruption separately and together as a whole. Inter and intra-party competition can be measured separately by looking at primary elections for intra-party competition and by looking at the general elections for inter-party competition. The total amount of electoral competition can be measured by looking at both elections.
Operationalization
How to measure corruption and electoral competition, the two most important variables in this thesis? Measuring electoral competition can be done relatively easily by counting the voter share margin of the winner in the election. The higher the percentage of the voter margin is the lower electoral competition is. When a politician for instance wins by only a very small margin, like for example one percent, he faced much competition from his competitor(s). If he however won with a margin higher than fifty percent, an example would be receiving 75 percent of the votes and your only opponent 25 percent, he faced a small amount of competition. This method of measuring electoral competition makes it possible to quantify the amount of competition. Measuring the amount of competition over a long period of time can be done by calculating the average voter share margin in elections during this period (adding up all the margins would have the same result). Inter and intra-party competition can be measured separately by looking at respectively the general and the primary election.
How much intra-party competition there is can be measured by counting the voter share margin in the primary election. In cases where there was no primary election because of a lack of competitors the voter share margin was counted as 100 percent, the candidate in the general election did not face any competition from candidates within his party in these cases. Examples of this are governors who had no competitor from their party and who could thus directly take part in the general elections . Because the candidate did not face any competition, these cases should be measured as having a 100 percent victory margin. If there would have been an election in such a case the margin of the winner would be 100 percent since he is the only candidate. This thesis will only look at the margin voter share margin in primaries of the winner in the general election. If for instance the democratic candidate wins the general election and becomes the governor, the voter share margin he had in the democratic primary will be counted to measure intra-party competition. The winner in the other primary (who lost the general election) does not become the governor and can thus not influence public corruption.
Corruption is harder to measure. This thesis uses data from the federal bureau of justice on the number of convicted public officials in corruption cases. This data contains information on the number of state level public officials who have been convicted on corruption charges. Examples of corruption crimes are accepting bribes, embezzlement, voter fraud and tax fraud. The convictions
were all in federal courts, the numbers thus do not depend on the quality of courts in the states. This dataset includes politicians but also other public officials, like higher ranked bureaucrats. Although gubernatorial elections do not influence these people directly, governors still have much political power on state politics and thus have a large influence on corruption levels. They can for instance appoint supporters who become corrupt or block anti corruption laws which would make the state less corrupt. Thus although many convicted public officials did not have to participate in elections, gubernatorial elections still have much influence on the number of convicted public officials in corruption cases. To take the size of the state into account, the amount of conviction per 100.000 inhabitants is used. A large state like California has many times more public officials and thus convicted public officials than states like Alaska and Rhode Island with few inhabitants. The total amount of convicted public officials thus does not tell us how corrupt these states are if it is not aggregated to the number of convictions per capita. The data used covers the period from 1976 to 2010, because of this gubernatorial elections in this period are used to measure competition.
It should however be noted that this data is not a perfect measure for public corruption. The main reason for this is that not all corrupt public officials are caught and some corrupt officials are thus not represented in this data which only contains convicted public officials. It could thus be that some states are more corrupt than the data suggests because corrupt officials in some states have a smaller chance of being caught. This is however very unlikely since the convictions were all in federal courts and the number of convicted officials thus does not depend of the quality of state level juridical institutions. The data is therefore still useful for our research. Data by the federal bureau of justice on corruption convictions has moreover been used before in other studies on corruption in the United States (Glaeser 2006:1053).
Regression analysis
A cross sectional study using a regression analysis will be used to research the effect of electoral competition on public corruption. In the cross sectional study the period from 1976 to 2010 will be analyzed to find correlations between electoral competition and corruption. Whether states with much competition (intra and or inter-party competition) in this period have been more or less
corrupt will be studied in this cross sectional study. A rather long period (34 years) has been chosen so a lot of elections could be taken into account, the long term effect of electoral competition can thereby be studied. Taking a long period also has the advantage that the scores for competition can not be biased by elections with very high winning margins. All fifty states of the USA will be analyzed.
The cross sectional study will be done with the use of a regression analysis. The dependent variable in this thesis is public corruption, measured by corruption convictions of public officers per
100.000 inhabitants. This is an interval variable and it is thus possible to make a regression analysis. Electoral competition is the main independent variable, inter-party competition and intra-party competition as well as total competition (average of inter and intra-party competition) are the three main independent variables. For each of the fifty states the average voter share margin in
gubernatorial elections during the period 1976-2010 has been calculated. This has been done for both the general elections in which each party has one candidate and the primary elections in which candidates from the same party compete for a sport in the general election. With the average voter share margin in general elections the level of inter-party competition can be measured, the level of intra-party competition can be measured by the average voter share margin in primary elections. Statistics on gubernatorial elections and primary elections have been retrieved from the Atlas of U.S Presidential Elections, a often used source which other than its name suggests also contains data on gubernatorial elections. The regression analysis contains six different models. Each of the three forms of electoral competition (inter-party competition, intra-party competition and total competition) has two models, one in which it is the only independent variable and one in which control variables are used to control the effect competition has only corruption levels. Because the results of a few primary elections could not be found the intra-party competition could not be calculated of four states (Alaska, Connecticut, North Dakota and Utah). This means that model three to six miss four states.
Median household income is the first control variable. Earlier cross national research has revealed the causal relation between GDP (Treisman 2000:400) and corruption so it is likely that variation in corruption levels between the states is partially caused by different levels of income in the states. Higher incomes are likely to lower public corruption because it lowers the incentive to become corrupt (ibid). People with a low income are more likely to be susceptible to bribes and other acts of corruption since it would raise their wealth significantly (ibid). Those with a higher income are more likely to be satisfied with how much money they have and therefore are less susceptible to corruption (ibid). The median household income of the states in 2010 is used because the variables of corruption and competition are based on a period ending in 2010. Median income is a better variable than mean income or GDP per capita since wealth and income are often very unevenly divided, particularly in the United States which has high levels of income inequality compared to other (developed) countries. The average income is thus distorted by a small number people having extremely high incomes. This is not the case with the median income per household and median income per household is thus used as a variable.
The two other control variables in the regression analysis are education and urbanization level. This thesis uses these control variables because they have been used before in research to explain the variety of corruption levels in the United States (Glaeser and Saks 2005:1059).
Especially education has proven to be having a significant influence on corruption (ibid). The reason for this is that higher educated citizens are better at monitoring their own politicians and are more willing to do something about it (ibidem: 1054). Education is furthermore shown to be a contributor of political participation with higher educated citizens being more politically involved than other citizens (ibid). Higher political participation may cause these citizens to pay more attention to corruption scandals, being better educated may also lead these people to better understand the negative consequences of corruption (ibid).
Analysis
Descriptive statistics
A look at the descriptive data shows that the fifty states vary much when it comes to both corruption and electoral competition. The three most corrupt states are in descending order Louisiana,Alaska and Mississippi with Louisiana having twenty convictions per 100.000 inhabitants. This is almost eight times higher than the number of convictions in the least corrupt state, Oregon has only 2,4 convictions per 100.000 inhabitants. Washington and Utah are the second and third least corrupt states with respectively 3 and 3,1 convictions per 100.00 inhabitants.
Virginia is the state in which the inter-party competition in gubernatorial elections was the greatest, the average voter share margin was only 9,1 percent in general elections. Iowa and New Mexico come in second and third place with average margins of 10,35 and 10,2 percent. Least competitive were general elections in Delaware, Louisiana and Nevada with margins of 26.25, 24.32 and 24 percent. The average score of all states was 16.1 percent with a standard deviation of 4.5.
Differences in competition during primary elections are bigger, the most competitive states in this aspect are Washington, West-Virginia and Georgia with scores of 21.31 22.91 and 25.62. The states with the least intra-party competition were Indiana, Kansas and Colorado with margins of 86.42, 85 and 83.68 percent. On average the states had a score of 50.44 percent with a standard deviation 17.9. Finally when looked at the total competition, both intra and inter-party competition, Delaware becomes the least and Washington the most competitive state with average margins of 51.43 and 15.98. The most and least competitive states can be seen in table 1 below.
Table 1: Most and least competitive states
Most and least competitive states (total competition)
1 Washington 15.98 48 Kansas 50.05
2 Virginia 18.24 49 Colorado 51.00
3 Georgia 22.24 50 Delaware 51.43
Regression analysis
Table 2: Results for competition regressions in the period 1976-2010
1 2 3 4 5 6 inter-party competition .112 (0.112) .072 (0.071) intra-party competition -.078** (-0.332) -.060* (-0.253) total competition -.123* (-0.272) -.092 (-0.203 median income .221* (0.441) 0.056 (0.106) .067 (0.127) education -.556*** (-0.607 -.336 (-0.395) -.364* (-0.428) urban -,063 (-0.209) -.023 (-0.80) -.023 (-0.083) R² 12 215 223 155 N 50 50 46 46 46 46 *:p<0.1 **:p<0.05 ***:p<0.001
Table two above shows the results of the regression analysis. It includes six, each of the two types of competition and the competition in total have two models. Models one and two that the amount of inter-party competition has had no significant impact on corruption in the states. In both models inter-party competition has had no significant effect on corruption level. Two of the control
variables, median household and education do have a significant effect in model two. The effect of education is negative which means that states with more highly educated citizens are on average
less corrupt. The effect of income is only significant at the ten percent level, it is furthermore positive which means that higher incomes led to more corruption. Because the of the high
confidence interval we should however be careful to make strong conclusions about this effect. We can conclude with confidence that inter-party competition has not influenced corruption levels. The argument made in this thesis, that electoral competition causes less corruption, does not hold when it comes to inter-party competition.
The second and third model show that intra-party competition has, in contrast to inter-part competition, a significant relation with corruption. The effect is furthermore negative, which means that states with high levels of intra-party competition during gubernatorial elections are more corrupt. When control variables are included in the fourth model intra-party competition becomes only significant with a ten percent confidence interval. The control variables themselves are not significant in the fourth model in contrast to what we saw earlier in the second model. This thesis argument thus also does not hold true for intra-party competition. Increased intra-party does not cause less corruption, our findings show that it in contrast might lead to increased corruption levels. This corresponds with the research of Chang (2006) earlier who found that intra-party competition in local elections in Italy led to more public corruption. However, it should be noted that the data for corruption can be biased by under reporting. As earlier discussed the number of convicted officials in corruption cases is not a perfect measurement of corruption since not every corrupt officials gets convicted.
When looked at the total amount of electoral competition in the fifth and sixth model, it becomes clear that the effect is only significant when control variables are not taken into account. In the sixth model only education has a significant effect on corruption, competition has not and income and urbanization rate also do not. It can thus be concluded that there is no significant relation between the total amount of electoral competition and corruption levels in states. Income has again no significant effect on corruption, it only had such in the second model. When intra-party competition is taken into account, income no longer has a significant effect on corruption.
To control the surprising results in the third and fourth model on the effect of intra-party competition on state level corruption this thesis will take a closer look at both in the last decade. Data on the number of convicted public officials from 2000 to 2010 has been used to measure corruption while primary elections from the same period have been used to calculate new scores of intra-party competition. In the table below we see the results of this new linear regression analysis using the newly gathered data. The results show that in the 2000s intra-party competition has not had a significant impact state level public corruption. It may be that it had in earlier decades but not in the previous one. Governors facing much competition in primary elections in this decade ave thus not caused more corruption than others.
Education is the only variable in this model that has a significant effect on corruption. This effect is as expected negative like the earlier regression analysis, it furthermore has a fairly strong effect with standardized coefficient of -0.507. The results of this thesis thus confirm earlier research which has pointed out the effect of education on corruption.
Table 3: Corruption and competition from 2000-2010
1 2 Intra-party competition 2000s Income -.007 (-0.072) -.009 (-0.097) .074 (0.365) Education -.174** (-0.507) Urban -.019 (0.162) R² 0.170 N 50 50
The findings of the two regression analyses thus show that the argument of this thesis that electoral competition reduces corruption does not hold. Both inter-party competition and competition in total have no significant impact on state level corruption. Intra-party competition might have an
influence but its effect is the opposite, more intra-party competition enhances rather than reduces corruption. This effect however does not hold when looking at the last decade, intra-party
competition has in the previous decade not had a significant effect on state level corruption. The results furthermore show that income does not explain differences in corruption between states in the US, this in contrast to findings on research on cross national differences.
Conclusion
This thesis has tried to answer whether electoral competition has an effect on the level of public corruption. It did this by looking competition in gubernatorial elections and state level corruption in the fifty states of the US. Both the competition between parties as competition within parties had been taken into account. The results first of all show that the states differ when it comes to both corruption and electoral competition. Louisiana, the most corrupt state, is almost eight times more corrupt than the least corrupt state which is Oregon. Gubernatorial elections are in some state much more contested than in others with having much smaller voter share margins in both the primary
and the general elections. With the use of a regression analysis this thesis has also clearly shown that electoral has not led to less corrupt in the states. Inter-party competition and the total amount of competition have had no significant impact on corruption levels. Intra-party competition may have had an impact but even if this it has the effect has been that it has enhanced corruption rather than stopped.
The results of this thesis make a stark contrast with theories of earlier studies. Rational choice theories used in other studies assume that competition leads to corruption being more costly and officials will therefore be less corrupt. This thesis has however shown that competition does not reduce corruption and serious doubts can thus be raised about the theories in which the officials are seen as rational individuals. Differences in corruption between electoral systems (majoritarian and proportional) may thus not be caused by their effect on competition but by something else. The one study whom this thesis may correspond with is the one by Chang (2005) which showed that intra-party competition makes politicians more likely to be corrupt.
Lastly the results of this thesis also show that the variation in corruption levels between US states cannot be explained by differences in income levels. Poorer states in the US are not more corrupt than richer ones. This is surprising given that income levels explain some of the global variation in corruption levels. It may be that income stops to play a role when it gets above a certain threshold, Officials in some states have a lower income than those in others it is still enough to prevent them from being susceptible to corruption.
Literature
Ades, A., & Di Tella, R. (1999). Rents, competition, and corruption. American economic review, 89(4), 982-993.
Carey, J. M., & Shugart, M. S. (1995). Incentives to cultivate a personal vote: A rank ordering of electoral formulas. Electoral studies, 14(4), 417-439.
Chang, E. C. (2005). Electoral incentives for political corruption under open-list proportional representation.
The Journal of Politics, 67(3), 716-730.
Cordis, A. S., & Milyo, J. (2016). Measuring public corruption in the United States: Evidence from administrative records of federal prosecutions. Public Integrity, 18(2), 127-148.
Ichino, N., Faller, J. K., & Glynn, A. (2012). Electoral Systems and Corruption Leip,D. (2018) Dave Leip's Atlas of U.S. Presidential Elections.
https://uselectionatlas.org/Retrieved on 5-10 June 2018.
Montinola, G. R., & Jackman, R. W. (2002). Sources of corruption: A cross-country study. British Journal of
Political Science, 32(1), 147-170.
Nur-Tegin, K., & Czap, H. J. (2012). Corruption: Democracy, autocracy, and political stability. Economic
Analysis and Policy, 42(1), 51-66.
Persson, T., Tabellini, G., & Trebbi, F. (2003). Electoral rules and corruption. journal of the European
Economic Association, 1(4), 958-989.
Rose-Ackerman, S., & Palifka, B. J. (2016). Corruption and government: Causes, consequences, and
reform. Cambridge university press.
Simpson et al. (2012). Chicago and Illinois , Leading the pack in corruption.
Tanzi, V. (1998). Corruption around the world: Causes, consequences, scope, and cures. Staff
Papers, 45(4),
559-594
Treisman, D. (2000). The causes of corruption: a cross-national study. Journal of public economics, 76(3), 399-457.
Appendix
State Inter-party
competition
(average voter share margin general election)
Intra-party competition
(average voter share margin primary election) Corruption convictions per 100.000 Alabama 19.11 27.68 13.7
Alaska 11.18 Not found 18.3
Arizona 11.52 53.62 5.1
Arkansas 18.2 59.85 6.9
California 13.3 58.15 6.3
Colorado 18.31 83.68 3.8
Connecticut 13.23 Not found 8.7
Delaware 26.65 76.2 8.9 Florida 11.86 51.07 9.4 Georgia 21.97 22.91 8.3 Hawaii 11.83 38.61 8.4 Idaho 20.35 57.28 5 Illinois 11.54 54.08 14.2 Indiana 11.45 86.42 6.5 Iowa 10.6 46.4 4.9 Kansas 15.09 85 5.3 Kentucky 20.38 26.45 13.3 Lousiana 24 28.78 20 Maine 10.92 74.2 7.9 Maryland 21.14 71.8 8.6 Massachusetts 15.7 33.11 8.6 Michigan 15.53 69.88 6.6 Minnesota 11.74 35.92 3.6 Mississippi 11.24 35.9 18.9
Missouri 12.82 34.25 8.5 Montana 19.05 44.1 13.7 Nebraska 23.76 40.2 4.5 Nevada 24.32 55.14 7.9 New Hampshire 20.4 66 3.5 New Jersey 14.35 47.8 10.3 New Mexico 10.34 39.5 6.8 New York 18.62 56.95 13 North Carolina 14.19 61.75 4.8
North Dakota 18.79 Not found 17.5
Ohio 15.29 82.87 12.2 Oklahoma 17.29 35.4 12.6 Oregon 11.58 43.88 2.4 Pennsylvania 12.56 50.3 12.3 Rhode Island 21.2 59.63 7.9 South Carolina 12.95 51.8 8.7 South Dakota 19.3 47 17.7 Tennessee 19.72 32.7 13.3 Texas 13.2 58.3 6.1
Utah 21.05 Not found 3.1
Vermont 19.5 71.45 4.8 Virginia 9.1 27.37 11.2 Washington 10.65 21.31 3 West Virginia 17.65 25.62 11.2 Wisconsin 13.24 53.22 5.2 Wyoming 19.71 37.25 8
