Bureaucratic Corruption, Democracy and Judicial Independence

Transcription

1 sian Business Research; ol. 1, No. 1; 16 ISSN Publishe by uly Press Bureaucratic Corrution, emocracy an uicial Ineenence Gang ang 1 1 eartment of Political Science, niversity of Zurich, Switzerlan Corresonence: Gang ang, eartment of Political Science, niversity of Zurich, Switzerlan. wang.ucb@gmail.com Receive: anuary 13, 16 ccete: anuary 18, 16 Online Publishe: February 6, 16 oi: 1.849/abr.v1i1.11 RL: htt://x.oi.org/1.849/abr.v1i1.11 bstract The theoretical literature in economics an olitical science has mae numerous efforts in unerstaning the eterminants of corrution an stresse the imortance of olitical institutions in shaing the atterns of government corrution. Nevertheless, very few researches focus on the role of juicial system. Emloying a formal moel with emirical analyses, I incororate economic factors with olitical constraints to investigate the ifferent roles of emocracy an juicial ineenence in etermining the level of bureaucrats corrution across countries. Emirically, the instrumental variable I aroach is alie to resolve the enogeneity roblems. The evience inicates that ifferent levels of corrution across countries are significantly influence by the egrees of juicial ineenence. To fight corrution successfully, I conten that the juiciary, as a har institutional constraint to resist bureaucratic corrution, has to be ineenent from the government. Keywors: bureaucratic corrution, juicial system, economic factors, constraints 1. Introuction Corrution, as government officials use ublic owers for rivate economic interests, has been the hot toic of ebate among social scientists. Evience of bureaucratic corrution exists in all economies, at various stages of eveloment, an uner ifferent olitical an economic regimes. But why is it more ervasive in some societies than in others? Economists an olitical scientists have stresse the imortance of such variables as economic eveloment, government exeniture, emocracy, checks an balances, etc. Comaratively very few scholars focus on the role of juicial system in shaing the atterns of bureaucratic corrution. Emloying a formal moel with emirical analyses, the author incororate economic factors with olitical constraints to investigate the ifferent effects of emocracy an juicial system on the level of corrution an argue the juiciary, as a har institutional constraint to resist bureaucratic corrution, has to be ineenent from the government. For most economists, corrutions lie in the elegation of ower. Therefore, economic theories ay much attention to the ecision of rational bureaucrats who involve in corrut transactions. ner information asymmetry, the rincial in many cases may not have full control over the agent s misconuct. Thus the agent can circumvent many of the checks an controls that are lace by the rincial Rose-ckerman, It is easier to exlain the nature of corrution accoring to the failure of the rincial-agent relationshi. It is, however, unclear why officials in some countries misuse ublic office for rivate gain more frequently an for larger ayoffs than their counterarts in others. Thus a rincial-agent moel has its limitations in exlaining the level of corrution. Many economists have also stuie the emirical regularities between corrution an a variety of economic variables across the countries. Most of these stuies reort income amania et al., 4; Leerman et al., 5; Treisman,, government exeniture Fishman & Gatti,, an economic freeom Golsmith, 1999; Park, 3; Treisman, have a negative-significant effect of the level of corrution. However, some economic factors like government exeniture which is significant in a articular moel may lose their significance when some other olitical variables are incororate. Hence as Selayo an Haan 6 conten, claims concerning the eterminants of corrution are conitional, an the robustness of the finings is oen to question. Political scientists have been attemting to exlain how olitical institutions influence actual corrution levels in ifferent societies. The logic that olitical institutions lay a key role in fighting bureaucratic corrution is reasonable an obvious. Because government intervention transfers resources from one arty to another, it 7

2 creates room for corrution cemoglu & erier,. Since in a moern society it is unavoiable that the government frequently involves in economic activities, without any olitical constraints, government leaers can make full use of ublic resources for rivate benefits. In other wors, without a strict unishment which can only be offere by olitical institutional arrangements, corrution will be out of control. But various olitical institutions may have very ifferent effects on bureaucrats corrution behavior. For instance, uner a juicial system, a corrutive bureaucrat will be unishe severely base on soli evience of his corrutive activities whereas uner emocratic system, a corrutive bureaucrat can only be unishe via free ress exosure an voting outcomes. Generally seaking, the actual level of corrution in a society shoul be etermine by the strictest of institutional arrangements. Many comarative stuies reort a significant relationshi between emocracy an the level of corrution base on various regression aroaches without controlling juicial ineenence. In my view, these emirical conclusions on the emocratic Clean Theory are theoretically an emirically roblematic. The structure of the rest art is organize as follows: in Section two, I am going to analyze the flawe logic an emirical araox of emocratic clean theory. In Section three, a formal moel will be emloye to investigate why the level of juicial ineenence rather than emocracy has a significant imact to the level of corrution. My emirical evience will be rovie in Section four an the brief iscussion an conclusion in the en.. The Flawe Logic an Emirical Paraox of emocratic Clean Theory Following the logic of the Princile-gent Moel, many olitical scientists argue that a emocratic regime reicts a low level of corrution because electoral accountability enforces an efficient constraint to government officials an olitical cometition an free ress reuce information asymmetries between voters an bureaucrats.. Tella, 1997, 1999; B.. Tella, 4; Chang-Golen, 4; Golsmith, 1999; Gurgur-Shah, 5; ckerman, 5; Leerman et al., 5; Treisman,. Theoretically the corrution level of a country is etermine by the most efficient olitical constraint to bureaucrats misconuct. Thus the emocratic clean theory is reasonable given that the emocratic regime is the strictest of institutional arrangements. nfortunately, thus far there is no literature articulately comaring the efficiency of emocracy an that of juicial ineenence in term of fighting corrution. In aition, without a juicial intervention, whether electoral accountability an free ress imly a high execte cost for the bureaucrat is in oubt. s Rose-ckerman 1999 conclue, the istinctive incentives for corrution in emocracies een on the organization of electoral an legislative rocesses an on the methos of camaign finance Note 1. Hence emocratic elections are not invariably a cure for corrution. The emirical evience of a relationshi between emocracy an corrution is also mixe. hile many regression analyses fin that emocracy has significant imacts on the level of corrution it & utta, 1; Emerson, 6; Ferejohn, 1986; Leerman et al., 1; Sanholtz & Koetzle,. cross-national stuy by Treisman suggests that the current egree of electoral emocracy is not significantly correlate with the level of corrution, but long exosure to emocracy reicts lower corrution. Montinola an ackman confirm that olitical cometition affects levels of corrution, but this effect is nonlinear. Most of these stuies emloy OLS regression analyses without controlling the level of juicial ineenence. Since many measures of olitical institutions are highly associate with each other an enogenous relationshis among aggregate level variables are very oular in social science research, the strategy of research esign is crucial to the creibility of emirical results. Because I am intereste in not only the eterminants of corrution from comarative ersective, but the causal mechanisms between bureaucrats corrution behavior an institutional arrangements, I attemt to emloy a formal moel to further our unerstaning on bureaucratic corrution an then rovie a robust emirical analysis. 3. Corrution an Political Constraints ner certain economic conitions, a bureaucrat s corrutive behavior is a rational ecision base on cost-benefits consieration. Because olitical institutions efine the rules of game in elegation of ower, an anyone breaks the rules may be enforce a enalty, a bureaucrat shoul consier olitical constraints as corrutive action costs. e can observe how the bureaucrat searches for his erfect corrution level by analyzing his utility function. ssume a bureaucrat s execte wage income was in time erio t without corrution. ith corrution his income in the two states, no unishment 1 an unishment, is, I, where 8

3 reresents the benefits from corrution an I the unishment enforce by a olitical institution. Since not all corrutive activities can be monitore, the execte utility theorem suggests that his references for income in these two states are escribe by the following function,,, 1, 1 1 here reresents the utility of money income an the robability of being unishe. Bureaucrats corrution behavior can be seen as a choice between ifferent combinations of the otential corrution levels an unishments, which can be escribe by,,,, I I. From all the corrution-unishment choices the bureaucrat is offere, he chooses the one that maximizes,,. I Since he always has the otion of oing no corrution, a bureaucrat will select corrution only if,, I,,, which imlies that bureaucrats are willing to work in ublic sectors because they can enjoy high rents by using ublic offices or ublic sectors coul offer an efficiency wage. I assume that bureaucrats are ientical in all asects an that they are risk-averse. I + α N O 45 o + α 1 Figure 1. Suose the amount of ublic resources he can corrut is. In Figure 1, let s try to unerstan a bureaucrat s behavior in this way: to gain the corrut benefit on the horizontal axes,, he has to ay the cost of being unishe I with robability, so the execte cost will be on the vertical axes. I 9

4 ssume his corrut behavior will never be tolerate, so the robability of being etecte is equal to the robability of being unishe see Note -1 for roof. If there are no institutional constraints,, I, I, then, 1, an from, which imlies he enjoys every ollar corrute without aying any cost. In figure 1, the rational bureaucrat will get starte from oint N, an go along with the straight line N until oint, is reache. That is, the rational bureaucrat can maximize his utility uner the initial buget line N. t oint,, he makes full use of ublic resources for rivate gains an his utility in this circumstance is,, Bureaucrats Corrution Behavior uner emocracy Now suose an institutional arrangement, emocracy emerges. Consier whether emocratic elections can reuce the amount of corrution significantly. The mechanisms of emocracy over corrution are as follows: 1 Electoral ccountability. Ientical voters elect government officials. hen voters have a signal of a bureaucrat misusing ublic office for rivate benefits, no matter what s the amount of corrution, he will lose his job. Since there is no juicial intervention so far, the most severe an only unishment for him is always to be fire. ssume his wage income in office at the time erio t 1 is 1 ; the robability of winning the election again at the en of t 1 is. So the execte total income after t 1 is t if he wins the election. If he loses the election, he coul fin another osition in the job market an the execte income in the new osition will be, where is a iscount number an [, 1]. Now his execte wage income in t 1 t is 1 t without corrution. ith corrution his income in the two states no unishment an unishment is: Suose,, 1 t, 1 t 1 t, 1 t 1 t t are all exogenous variables an they are uncorrelate with. Let, an 1 1 t t. The bureaucrat corrution choices can be escribe as,, where an are both constants. Generally seaking, in moern societies, a government official s wage income is much less than the ublic resources uner his control. That is,. Monitor System. Because of free ress an olitical cometition, the robability of his corrution behavior being etecte in a emocratic regime is reasonably higher than that in an autocracy. 1

5 can be unerstoo as an information function, an, 1 nature of is that although there is information asymmetry between voters an bureaucrats, voters can get incomlete information through free ress an arty cometition. hen the bureaucrat is oing corrution, the information will accumulate as increases, so. In aition, as. The increases, I assume will be larger, so. The notion of this assumtion is that when a monitoring system fins his corrution behavior, it not only has the information for his current erformance, but will actively collect more information about his revious behavior to fin out whether he i corrution before. Now the bureaucrat s references for corrution can be escribe by the following function: Max, 1 S. t. 1 Max, 1 S. t. Notice the olitical imlication of is the marginal robability of being unishe when the bureaucrat enjoys one more ollar corrution income, thus. From is a very small number even though an see Note - for roof, his corrution behavior now is converte into a single urose nonlinear rogramming roblem an the otimal level of corrution is etermine by the shae of, an. efine the bureaucrat s rrow-pratt coefficient of absolute risk aversion at as r u / u. Given that the bureaucrat is not highly risk averse an the marginal robability is a small number, secifically, when 1 1 t an 1 r u / u [ ] we can rove institutional constraint is a monotone increasing function of an,, in, ] see Note -3 for roof, which imlies as the marginal buget kees going 1 [ u, a bureaucrat will always enjoy every ollar from the corrution income. Therefore there is no Interior 11

6 Solutions but a Corner Solution the equilibrium E in Figure for the bureaucrat uner a emocracy constraint an Max 1 a + α E a E E a α - α E N 45 o O + α 1 Figure. Similarly, if we assume in an autocracy, the corresoning unishment an the robability of being unishe is relatively smaller, we can see that the bureaucrat still fixes his level of corrution at. Max Max 1 1 Max The bureaucrat in a emocracy has to ay much more cost the execte unishment, thus his execte utility might be smaller than his counterarts in an autocracy. Since, his inifference curve 1 1 looks steeer in a emocracy an eole become a little more risk-averse. But the actual level of corrution oes not change from an autocracy to a emocracy. 1

7 In Figure, we can see the ifferent institutional arrangements change the bureaucrat s buget line in ifferent ways. Now the bureaucrat is trying to maximize his utility uner a new buget line NE or NE a rather than N. However, any institutional arrangement which can ecrease the level of corrution significantly has to increase the execte unishment corrution cost at least beyon the bureaucrat s execte corrution benefit at the low level of corrution. In other wors, the new buget line has to be below the horizontal line N since a certain low level of corrution. Theoretically, any corner solutions excet N cannot reuce corrution, so a goo institutional esign shoul lea us to fin the ossible interior solutions an the actual levels of corrution in ifferent countries are etermine by the otential interior solution an thus the roose institution. Basically we have two aroaches to threaten the bureaucrat s corrution behavior. This first one is to increase the unishment. But as is a constant an exogenous variable, without a new institution intervention, electoral accountability has little influence to. Even a emocracy coul raise the enalty, this otential unishment is har to be equivalent to the one that juicial system can offer. On the other han, we may also strengthen the monitoring caability an insect every ollar government exene, as some other scholars have argue, free ress an arty cometition is imortant to limit the scoe of corrution. But the embarrassment is that monitoring bureaucrats behavior is just a rerequisite or necessary conition of enalty. hen emocratic unishment cannot threaten bureaucrats vital interests, how can we exect to see a sufficient better outcome emerging? That s why I will focus on the role of juicial ineenence in the following section. 3. Fighting Corrution: The Role of uicial Ineenence In olitical science literature, juicial ineenence usually refers to the autonomy of juges, which imlies the members of the juiciary ought to have an ineenent relationshi with other arts of the olitical system an they can exect their ecisions to be imlemente free from any outsie ressure Russell, 1. The mathematical form of juicial ineenence can be exresse as follows in the sense of fighting bureaucratic corrution. Suose reresents the mount of ublic resources which have been corrute, an L is a unishment for his behavior efine by law. L is the real unishment mae by juges, where ε is a ranom error. Then juicial ineenence refers to: E L E L E E L L Because a normal form of government intervention in juicial system is to ecrease unishment, we can introuce G as a measure of juicial eenence. hen juges ecision is influence by the government, the execte unishment for a bureaucrat will be: E L G E L E G E L G L Basically, we can use the similar logic to analysis the role of juicial ineenence, an the bureaucrat s strategy is still to select the erfect level of corrution uner unishment constraints. But there are two ifferences between emocratic constraint an juicial constraint. The first one is the romising unishment is not a constant 13

8 in the juicial system. The more the ublic resources corrute, the more severe unishment will be given to him. So is a monotone increasing function of. Because the ineenent juicial system can enforce a strict sanction on government officials, we efine the sanction function an for any [, ] no corrution, no unishment,. In aition, once all the ublic resources are corrute, the bureaucrat will receive a more severe unishment,. To simlify our analysis, let, which imlies the anti-corrution law will not increase the marginal unishment to the bureaucrat suenly because he receive one more ollar illegal income. The secon ifference is that courts have to have substantial eviences to rove he misuse ublic resources for ersonal benefits, so the robability of being monitore, an hence the robability of being unishe will be much smaller in the juicial system, but we still have. Now the bureaucrat s references for corrution uner juicial constraint can be escribe by the following function: Max, 1 S. t. 1 Max, 1 S. t. So his otimal corrution behavior is also converte into a single urose nonlinear rogramming roblem, an there is one an only one interior solution *, see Note iv for roof. * Now suose the levels of juicial ineenence are ifferent in two countries an B. e can substitute L G an L G. Other things being equal, if B B then we reict the bureaucratic corrution will be more ervasive in country B than in country. That is * **. B * Figure 3 resents some intuitionistic notions of why an ineenent juicial system can reuce the level of corrution efficiently an etermine the ifferent levels of corrution across countries. In figure 3, comare to the emocratic constraint, there is an inflection oint in the juicial buget line because the strictest unishment is always enforce by juicial system, so even when the robability of being unishe is relatively small, the bureaucrat is facing a high cost of corrution behavior. To see this, let s comare the features of ifferent buget lines: In a emocracy: B B 14

9 an In a juicial constraint, but when is small an when is very large a + α E a E E a j j α - a α a E j E j E N 45 o R α ** α B ** O + α 1 Figure 3. Basically, the shae of juicial constraint line is influence by two factors, the monitoring technology an the level of juicial ineenence. So my finings rovie a new suort for the arguments that the monitoring technology matters to the level of corrution. The joint effect of the monitoring technology an juicial ineenence etermines the equilibrium when economic variables are hel as constants. ner certain monitoring technology, an efficient way to influence the bureaucrat s behavior is to change the unishment. Let for any. That is, any corrution behavior once monitore will be enforce a horrible enalty., so the rational bureaucrat won t o any corrution. Grahically, this 15

10 buget line will be the vertical line N. In this case, the equilibrium will be N,, which imlies there is no corrution. In most countries, even with L G L, as long as there is an inflection oint in the i i i juicial buget line, the ifferent levels of corrution will be reicte by the levels of juicial ineenence. 4. Emirical nalyses 4.1 Simle Moel of Corrution In revious analyses, I o not consier the effects of economic variables on the amount of corrution. However, both the wage income an the amount of ublic resources uner bureaucrats control can influence the actual level of corrution. Therefore, corrution can be ientifie as a rouct in a certain society with secific economic an olitical structures. Suose E an P resectively reresent the amount of economic an olitical resources controlle by government bureaucrats. Symbol refers to an overall measure of the institutional environment, or to what extent bureaucrats may easily make use of ublic resonsibility for rivate ens. I assume the aggregate level of bureaucratic corrution can be escribe by the following Cobb-ouglas rouction function: Corrution E here enotes the inex that bureaucrats use ublic economic resource for ersonal uroses or interests. E thus refers to how much ublic resources have been corrute ue to economic conitions. Similarly, i P enotes the inex that bureaucrats use ublic olitical resources for ersonal uroses or interests. refers to how much ublic olitical resource has been corrute uner certain olitical environment. P thus Corrution log Corrutio E n log P log E log P Because bureaucrats always have motivations to convert the ublic resources into rivate ens, α > an β >, I exect to see the level of corrution is ositively associate with the economic an olitical resources uner the government s control. 4. Brief Review of Literature on uicial Ineenence I It is well recognize that the rule of law is the cornerstone of a roserous society. mong various asects of law, I is consiere to be a founation for the rule of law nite Nations, Most olitical scientists believe that the searation of owers an checks an balances is essential for a regime to be well-functioning. s one of the most effective checks an balances, the ineenence of the juiciary can revent the abuse of government ower by other olitical branches. The literature has not reache the consensus on the efinition of I. Some scholars have rouce long lists of criteria the juiciary must meet, whereas others focus on more narrow asects of juicial institutions Lanes & Posner, 1975; Larkins, 1996; Shetreet, ccoring to the number of characteristics involve, I can be efine in the following ways. The first tye of efinition eals with two levels of characteristics: juge an juiciary level bbasi, 8. The ineenence of juge refers to the imartiality of juicial ecision. That is, the results of a court shoul not be intervene by the juge s ersonal interest in the outcome of the case. On the contrary, the ineenence of juiciary, autonomy, is efine from the institutional ersective. It refers to the relationshi between the juiciary an olitical branches. In other wors, secific arrangements are create to revent the olitical forces from becoming the main eterminants of juicial behavior. In literature, the ineenence of juge is 16

11 synonymous to ecisional ineenence, whereas the ineenence of juiciary is ientical to institutional ineenence. The secon tye of efinition inclues three characteristics. ccoring to the orl Bank, the juicial ineenence inclues three imensions: imartiality, comliance with the juicial ecisions, an free from interference from other olitical branches. Imartiality refers to the situation in which the results of a court are not influence by the juge s ersonal interest. The comliance with the juicial ecisions imlies that juicial ecisions shoul be well resecte once they are renere. Insulating juges from interference of government officials is often taken to be the most imortant asect of juicial ineenence. Similar efinition is also rovie by Greene 6. Base on interviews of Canaian aellate court juges, Greene 6 also ientifie three asects of juicial ineenence: imartiality, no interference from other juges, an comlete freeom to ecie. nother innovative efinition is ut forwar by Rios-Figueroa. He 5 roose to unack the concet of I into four comonents: autonomy, external ineenence, internal ineenence, an the institutional location of the ublic rosecutor s office. utonomy refers to the relation between the juiciary an the electe branches of government. External ineenence refers to the relation between Sureme Court juges an government branches, whereas internal ineenence is the relation between lower court an sureme. s for the institutional location of the ublic rosecutor s office, the author sets three categories: within the juiciary, the executive, or as an autonomous organ. If juges an rosecutors belong to the same juiciary, they woul be more ineenent of olitical owers from other government organs. In my analysis, I aot the efinition that I has two characteristics: ecisional ineenence an institutional ineenence. In line with these two imensions of I, I am motivate to fin two sources of exogenous variations in olitical institution to instrument I. The first instrument-tenure of Sureme Court juges, inicates the ecisional ineenence, whereas the secon instrument-the year of the constitution significantly revise last time, reresents the institutional ineenence. 4.3 Research esign lot of research has been conucte emirically to uncover the causal effect of secific olitical institutions on corrution. My interest in this aer is to ientify the causal effect of juicial system on the outcome of corrution. However, cross-country analysis often faces an obstacle, the enogeneity roblem if I simly emloy an OLS regression. First, it is confronte by the simultaneity roblem. There are unobservable variables that may etermine both the ecision to form the juicial system an the level of corrution. The secon issue is concerne with reverse causality: a change in corrution level may lea to a change in the ecision to outline the juicial structure. It is quite likely that high-income economies with low level of corrution can affor better juicial system. These enogeneity roblems ut threat to internal valiity of my stuy, an become the biggest concern for the research esign of my stuy. well-known strategy to resolve the enogeneity issue is to use instrumental variables I. The two Is use in my analysis are the year of the constitution significantly revise last time, an tenure of Sureme Court juge. For abbreviation, the first I is calle YCR, an the secon I is calle tenure. The logic is that YCR oes not affect the corrution level of countries irectly, but countries whose constitution were significantly revise more recently ten to have less stable juiciary system an less rigiity of constitution. The volatile juiciary system inicates a weak juicial ineenence. Similarly, tenure of juges oes not influence the bureaucrats corrutive behavior irectly, but longer terms for juges rovie greater job security so that juges concerns about holing their osition or being romote o not influence their ecisions, which may further lea to the ineenence of juiciary. Secifically, the I aroach will rocee in two stes: in the first ste, the juicial ineenence variable is regresse on YCR, tenure an other observable exogenous regressors; in the secon ste, the reicte value of the juicial variable is use in the corrution regression. The valiity of I will be iscusse in etail in the following sections. 4.4 ata an escritive Statistics Table 1 rovies escritive statistics for the key variable of interest. The Corrution Percetions Inex in 6 is my measure of corrution outcome, ranging from 1.8, the most corrut country Haiti, to 9.6, the cleanest country New Zealan. My key exlanatory variable is the level of juicial ineenence in 5, with a range of.3 to 9.,.3 corresoning to the least juicial ineenent country enezuela an 9. referring to the most juicial ineenent country Germany. The I variables-ycr sans almost two centuries, varying from 1814 to 17

12 5; an Tenure ranges from to, with reresenting a tenure less than six years, 1 inicating a tenure more than six years but not life long, an reresenting life long. Since CPI 6 actually refers to the corrution level in the year of 5 in ifferent countries, all the ineenent variables in my emirical stuy are measure in 5. Table 1. escritive statistics ariable Obs. Mean St. ev. Min Max CPI LogGP Government Size Government Consumtion Government Investment uicial Ineenence Political Rights Civil Liberty YCR Tenure Corrution inex in Corrution inex in Political rights Civil liberty Figure 4. Reuce-form relationshi between corrution inex an emocracy olitical rights or civil liberty 18

13 Corrution inex in uicial Ineenence Figure 5. Reuce-form relationshi between corrution inex an juicial ineenence hich institution is more effective in fighting corrution: emocracy or juicial ineenence? Figures 4 & 5 resent simle grahic illustrations of the relative efficiency of these two institutions. Figure 4 lots the corrution inex in 6 against two measures of emocracy: olitical rights an civil liberty. It shows a very weak association between corrution an emocracy regarless of which measure of emocracy is alie. In shar contrast, Figure 5 resents a strong ositive association between corrution an juicial ineenence. Countries with higher level of juicial ineenence resent a lower level of corrution. Naturally, I conclue that emocracy has a weaker imact on reucing corrution, while juicial ineenence has a stronger imact on it. The next section will rovie regression results to substantiate this statement. Figure 6 lots the CPI 6 against the year when the current constitution was significantly revise last time for a samle of 165 countries two countries with YCR in the 18s are roe out of the grah. It shows a strong negative relationshi between these two variables. Countries which revise their constitution more recently are substantially more corrut than countries which revise current constitution many years ago. To valiate this, I will regress current corrution level in 6 on current juicial ineenence level in 5, an instrument the latter by YCR an tenure variables. The I regression result will be resente in the following sections. Corrution inex in The year of the constitution significantly revise last time Figure 6. Reuce-form relationshi between corrution inex an YCR I 19

14 4.5 Orinary Least-Squares Regressions In this art, I will resent the naive result from OLS regression in Table. The basic secification is liste as follows: Corrution uicial_ Ineenence X 1 i i i i here Corrution i is the eenent variable, inicating the corrution level in 6 for country i; uicial_ineenence i is my key exlanatory variable, inicative of the juicial ineenence level of country i in 5. I use the lagge values instea of current values of uicial_ineenence to cature the causal effect. X i is a vector of other covariates, incluing logarithm of GP, government size, government consumtion, government investment, olitical rights, an civil liberty; an i is a ranom error term. The coefficient of interest throughout the aer is, the effect of juicial ineenence on the level of corrution. Table reorts the OLS regression estimates of corrution on emocracy, juicial ineenence, an other controls. First, the effect of emocracy on controlling corrution is not effective. Political rights variable has insignificant effect against corrution no matter whether the level of juiciary ineenence is controlle see Moel an Moel 4. Yet civil liberty, another measure of emocracy, is statistically significant in Moel, which exclues juiciary level. Secon, the coefficient of juicial ineenence variable, an estimate of.464, shows its strong ositive association with the eenent variable CPI 6. Further, once the juiciary level is controlle, two measures of emocracy become insignificant. The overall OLS results inicate the correct reication of my formal moel. Comare to emocracy inex, juicial ineenence can exlain the ifferences of corrution across countries much better. Table. OLS regression analyses eenent variable: CPI 6 Moel 1 Moel Moel 3 Moel 4 Ineenent ariables LogGP 1.5*** 1.118***.84***.794*** Government Size Government Consumtion Government Investment.148*.1.1*.87 uicial Ineenence.475***.464*** Political Rights.8.1 Civil Liberty -.445* j R Significant level: *** <.1, ** <.1, * < uicial Ineenence an Corrution: I Results s I iscusse in the Research esign section, the enogeneity roblems will bias the OLS result an make the OLS estimates inconsistent. Therefore, in orer to estimate the imact of olitical institution on corrution level, I nee fin a source of exogenous variation in olitical institution, an instrumental variable I, to remove the surious correlation between the exlanatory variable an unobserve characteristics. The two instruments I am using are YCR, the year of a country s constitution significantly revise last time, an tenure, term of juges. Concetually, the I aroach imly that, in the first stage, a variable which is unrelate to the outcome variable is use as a reictor of the key exlanatory variable; in the secon stage, the outcome variable is regresse on

15 the reicte measure from the first stage. But, the valiity of inferences from an I analysis eens on the aroriateness of the exclusion restriction assumtion, which imly that, conitional on the controls inclue in the regression, YCR an tenure have no effect on corrution level toay in 6, other than their effect through the institutional eveloment of juicial ineenence. The two-stage least-squares estimates are resente in Table 3. uicialineenence i is treate as enogenous an moele as following: uicialineenence i 1 * YCRi * Tenurei X i i here YCR i is the year when a country s constitution was significantly revise last time, an Tenure i is the tenure of Sureme Court juges. The exclusion restriction is that these two variables o not aear in equation 1 Table 3. I SLS Regression of corrution eenent variable: CPI 6 Instrumente: uicial Ineenence 5 I: YCR an tenure Moel 1 Moel Moel 3 Moel 4 Ineenent ariables LogGP.579*.393**.553** Government Size Government Consumtion Government Investment uicial Ineenence 1.13***.849**.919***.853** Political Rights Civil Liberty j R Significant level: *** <.1, ** <.1, * <.5 The corresoning SLS estimate of the imact of juicial ineenence on corrution in moel 4 is.853 stanar error =.38, which is larger than the OLS estimate.464 reorte in Table moel 4. This suggests the ownwar bias of OLS estimates. In aition, the existence of attenuation bias, an the measurement error in the juicial ineenence variable is likely to be more imortant than reverse causality an omitte variable biases. It is lausible that one single measure of olitical institution can harly cature the whole set of olitical institutions that matter for corrution level. The I aroach resents us a more creible estimate. I erform two tyes of tests to consoliate my I results. First, I test the null hyothesis of exogeneity of juiciary ineenence variable. If this null is not rejecte, the analysis will rocee uner the assumtion of exogeneity an run OLS regressions. Otherwise, the analysis will rocee to o I. Tyically, Hausman test will accomlish this task. Secon, the over-ientification test is erforme. Since I have only one otential enogenous variable but have two caniate instruments, I nee test the exogeneity of the extra instruments. That is, whether all of the instruments are vali. The Hausman test resents a chi-square value of 9.63 with a P-value of.1, inicating rejection of the consistency of OLS an suort for using I regression. The over-ientification test has a P-value of.895, 1

16 which is highly insignificant. So at the tyical 5% significance level, I woul fail to reject the hyothesis that the instrumental variables are all exogenous. These two instruments turn out to be vali for my analysis. 5. iscussion an Conclusion It is wiely iscusse that ifferences in institutions are at the root of large ifferences in economic outcome across countries. However, lague with enogeneity roblems, it is ifficult to isolate exogenous sources of variation in institutions to estimate their effect on economic erformance. Emloying a simle formal moel as micro founation, this aer articularly focuses on the causal effect of juicial ineenence status on the corrution level. Two instrumental variables: the year when a country s current constitution was significantly revise last time, an tenure of juges, serve as a source of exogenous ifferences in juicial ineenence level. However, these two Is o not irectly influence corrution level. Thus, the exclusion restriction assumtion is met. The estimates are robust across various secifications. The research finings inicate that a severe unishment enforce by ineenent juicial system is necessary to eter bureaucrats corrution behavior, an ifferent levels of corrution across countries can be exlaine much better than emocracy oes. There are many questions left unanswere in this aer. In the formal moel, it is assume that government officials are ientical in all asects an they are risk-averse. In reality, ifferent bureaucrats in similar ositions may resent ifferent attitues towars the otential sanction of risk. In aition, informal institutions such as culture an customs were not consiere to influence bureaucrats corrution behavior in the formal moel an emirical analyses. These questions leave room for the future stuy. References es,., & Rafael, T The Causes an Consequences of Corrution: Review of Recent Emirical Contributions. In B. Harris-hite & G. hite Es. Liberalization an the New Corrution. Brighton: Institute of eveloment Stuies Bulletin. es,., & Rafael, T Rents, Cometition, an Corrution. The merican Economic Review, 894, htt://x.oi.org/1.157/aer it, T. S. 3. Economic nalysis of Corrution: Survey. The Economic ournal, 113, 3-5. zfar, O. et al. 1. The Causes an Consequences of Corrution. nnals of the merican caemy of Political an Social Science, 573, htt://x.oi.org/1.1177/ Barhan, P Corrution an eveloment: Review of Issues. ournal of Economic Literature, 35, Colaner, Neoclassical Political Economy: The nalysis of Rent-seeking an P ctivities. Mass: Ballinger Publishing Co. amania, R., Freriksson. P., & Mani, M. 4. The Persistence of Corrution an Regulatory Comliance Failures: Theory an Evience. Public Choice, 11, htt://x.oi.org/1.17/s e Mello, L., & Barenstein, M.. Fiscal ecentralization an Governance: Cross-Country nalysis. In G. T. be, & S. Guta Es., Governance, Corrution an Economic Performance. ashington C: International Monetary Fun. Emerson, P. 6. Corrution, Cometition an emocracy. ournal of eveloment Economics, 81, htt://x.oi.org/1.116/j.jeveco Ervin, S Searation of Powers: uicial Ineenence. Law an Contemorary Problems, 351, Ferejohn, Incumbent erformance an electoral control. Public Choice, 5, 5-5. htt://x.oi.org/1.17/bf1494. Ferejohn,., & Kramer, L. 6. uicial Ineenence in a emocracy: Institutionalizing uicial Restraint in Norms an the Law. New York: Cambrige niversity Press. Fishman, R., & Gatti, R.. ecentralization an Corrution: Evience across Countries. ournal of Public Economics, 83, htt://x.oi.org/1.116/s Gerring,., & Thacker,. 4. Political Institutions an Corrution: The Role of nitarism an Parliamentarism British ournal of Political Science, 34, htt://x.oi.org/1.117/s Golsmith, Slaing the Grasing Han: Correlates of Political Corrution in Emerging Market.

17 merican ournal of Economics an Sociology, 584, htt://x.oi.org/1.1111/j tb3398.x Hanssen,. 4. Is There a Politically Otimal Level of uicial Ineenence? The merican Economic Review, 943, Hanssen, The Effect of uicial Institutions on ncertainty an the Rate of Litigation: The Election versus ointment of State uges. The ournal of Legal Stuies, 81, 5-3. htt://x.oi.org/1.186/4685 ain, Economics of Corrution. Boston: Kluwer caemic Publishers. ain,. 1. Corrution: Review. ournal of Economic Surveys, 151, htt://x.oi.org/1.1111/ Kaufman, I. R The Essence of uicial Ineenence. Columbia Law Review, 84, htt://x.oi.org/1.37/11136 LaPalombara, Structural an Institutional sects of Corrution. Social Research, 61, Larkins, C uicial Ineenence an emocratization: Theoretical an Concetual nalysis. The merican ournal of Comarative Law, 444, htt://x.oi.org/1.37/8463 Leerman,., Loayza, N., & Soares, R. 5. ccountability an Corrution: Political Institutions Matter. Economics an Politics, 17, htt://x.oi.org/1.1111/j x Porta, R. L., Lóez, F. C., Po-Eleches, & Shleifer,. 4. uicial Checks an Balances. ournal of Political Economy, Mariscal, E., & Sokoloff, K.. Schooling, Suffrage, an the Persistence of Inequality in the mericas, Political Institutions an Economic Growth in Latin merica. Stanfor: Hoover Institution. Mauro, P Corrution an Growth. The Quarterly ournal of Economics, 113, htt://x.oi.org/1.37/ Mauro, P Corrution an the Comosition of Government Exeniture. ournal of Public Economics, 69, Montinola, G., & Robert,.. Sources of Corrution: Cross-Country Stuy. British ournal of Political Science, 3, htt://x.oi.org/1.117/s Park, H. 3. eterminants of Corrution: Cross-National nalysis. The Multinational Business Review, 11, htt://x.oi.org/1.118/155383x31 Persson, T., Rolan, G., & Tabellini, G Searation of Powers an Political ccountability. Quarterly ournal of Economics, 11, htt://x.oi.org/1.116/ Przeworski,. 7. Is the Science of Comarative Politics Possible? The Oxfor Hanbook of Comarative Politics. Oxfor niversity Press. Rios-Figueroa,. 6. uicial Ineenence: efinition, Measurement, an Its Effects on Corrution. issertation submitte in artial fulfillment of the requirements for the egree of octor of Philosohy. eartment of Politics, New York niversity. Rosenberg, G uicial Ineenence an the Reality of Political Power. The Review of Politics, 543, Rose-ckerman, S Corrution: Stuy in Political Economy. New York: caemic Press. Rose-ckerman, S Corrution an Government: Causes, Consequences, an Reform. New York: Cambrige niversity Press. Rose-ckerman, S. 6. International Hanbook on the Economics of Corrution. Northamton, M: Ewar Elgar. Rothschil, M., & Stiglitz, Euqilibrium in Cometitive Insurance Market: n Essay on the Economics of Imerfect Information. The Quarter ournal of Economics, 94, Russell, P., & O Brien,. 1. uicial Ineenence in the ge of emocracy. Charlottesville an Lonon: niversity Press of irginia. Sanholtz,., & Koetzle,.. ccounting for Corrution: Economic Structure, emocracy, an Trae. 3

18 International Stuies Quarterly, 44, Shi, S., & Temzelies, T. 4. Moel of Bureaucracy an Corrution. International Economic Review, 453, 4. Shleifer,., & ishny, R Corrution. The Quarterly ournal of Economics, 183, Svensson,. 5. Eight Questions about Corrution. ournal of Economic Persectives, 193, Treisman, ecentralization an Corrution: hy are Feeral States Perceive to be More Corrut. Paer reare for the resentation at the nnual Meeting of the merican Political Science ssociation, tlanta. Treisman,.. The Causes of Corrution: Cross-National Stuy. ournal of Public Economics, 76, arren, M. 6. emocracy an eceit: Regulating earances of Corrution. merican ournal of Political Science, 51, Notes Note 1. By relating corrution to ifferent features of the electoral system in a samle from the late nineties encomassing more than 8 emocracies, Persson et al. 1 fin that larger voting istricts are associate with less corrution, holing constant a variety of economic an social variables. Note. Below are a few mathematical equations. -1, unishme nt unishme nt monitore unishme nt Not monitore unishme nt/monitor e monitore unishme nt/not monitore Not monitore 1 monitore monitore -, Let g g ; g g 1 g g g -3, 1 From ii g 1 1, if 1 t g is a monotone increasing function of., then institutional constraint From, 1, 1 4

19 5 1 1 ] [ ] [ Therefore, when ] [ 1 / u u r e have an min, which imlies there is no Interior Solutions but a Corner Solution. -4, From.. 1, t S Max Let g

20 6 g g So there exits at least one * * * to let, * g. 1 1 g g g g Therefore, So there exits only one * * * to let, * g an for any, *, we have P, which means the bureaucrat might have an interior solution to maximize his utility function. Since * e must have one an only one *, * * an **. t **, he can maximize his utility function. Coyrights Coyright for this article is retaine by the authors, with first ublication rights grante to the journal. This is an oen-access article istribute uner the terms an conitions of the Creative Commons ttribution license htt://creativecommons.org/licenses/by/3./.