The Quality of Government - scholar.harvard.edu · The Quality of Government ... effective...
Transcript of The Quality of Government - scholar.harvard.edu · The Quality of Government ... effective...
222 JLEO, V15 N1
The Quality of Government
Rafael La PortaHarvard University and NBER
Florencio Lopez-de-SilanesHarvard University and NBER
Andrei ShleiferHarvard University and NBER
Robert VishnyUniversity of Chicago and NBER
We investigate empirically the determinants of the quality of governments in a largecross-section of countries. We assess government performance using measuresof government intervention, public sector efficiency, public good provision, sizeof government, and political freedom. We find that countries that are poor, closeto the equator, ethnolinguistically heterogeneous, use French or socialist laws, orhave high proportions of Catholics or Muslims exhibit inferior government perfor-mance. We also find that the larger governments tend to be the better performingones. The importance of (reasonably) exogenous historical factors in explain-ing the variation in government performance across countries sheds light on theeconomic, political, and cultural theories of institutions.
1. Introduction and OverviewIn recent years economists have stressed the idea that good economic insti-tutions, particularly those in the public sector, are instrumental to economicgrowth. Some of such institutions include limited government, a relatively be-nign and uncorrupt bureaucracy, a legal system that protects property rights andenforces contracts, and modest taxation and regulation. Good government hasbeen shown to contribute to the economic development of European countriesover the last millennium (North, 1981; De Long and Shleifer, 1993), to growthacross countries over the last 40 years (Knack and Keefer, 1995; Mauro, 1995;Easterly and Levine, 1997), and to the successful transition from socialism tocapitalism (Weingast, 1995; Johnson, Kaufmann, and Shleifer, 1997). The im-
The authors are extremely grateful to Alberto Alesina, Robert Barro, Olivier Blanchard, ColinCamerer, Brad DeLong, Edward Glaeser, Claudia Goldin, Simon Johnson, Lawrence Katz, DavidLaibson, David Landes, Daniel Treisman, and Lynn Zucker for discussions of this article, and toTatyana Nenova for research assistance.
c© 1999 Oxford University Press
The Quality of Government 223
portance of good government for growth thus appears to be a well-establishedempirical proposition.
This proposition raises an obvious question: How did some countries cometo have good government and others did not? How does history influence gov-ernment performance? In this article we address these questions using dataon government performance and its possible determinants for a large sampleof countries. We present, and evaluate empirically, a number of theories thatfocus on economic, political, and cultural determinants of government perfor-mance.
As a first step, it is important to agree on what constitutes “good govern-ment.” We use the term “good” in this article to stand for good-for-economic-development. One can alternatively consider good government performanceto manifest itself in lower inequality, greater diversity among people, or main-tained traditions, but here we keep the narrow view. Even in this narrow view,however, there are many distinct dimensions of good government performance.Following Montesquieu (1748) and Smith (1776), economists have focusedon the security of property rights—lack of intervention by the government,benign regulation, low taxation—as the crucial metric of good performance.Other signs of well-functioning government include high quality of the bureau-cracy, successful provision of essential public goods, effective spending, anddemocracy—which is both an end in itself and a mechanism for modifyinginstitutions. In this article we examine all these separate dimensions of thequality of government.
The theories of determinants of institutional—and more specificallygovernment—performance fall into three broad categories: economic, polit-ical, and cultural. We treat these groups of theories as sharply distinct, eventhough their advocates would surely accuse us of caricaturing their views. Eco-nomic theories (Demsetz, 1967; North, 1981) hold that institutions are createdwhen it is efficient to create them, that is, when the social benefits of buildinginstitutions exceed the transaction costs of doing so. Thus private propertyrights over land are created when land becomes scarce, and when the costs ofenforcing such rights fall below the benefits. Political theories (Marx, 1872;North, 1990; Olson, 1993), in contrast, focus on redistribution rather than effi-ciency, and hold that policies and institutions are shaped by those in power tostay in power and to transfer resources to themselves. For Marx, the interestsin power are identified with “class,” but they can also be autocrats, autonomousbureaucracies, organized religion, ethnic groups, or even particular organizedeconomic interests (the nobility, state enterprises, or even retirees). Accord-ing to political theories, government policies are inefficient not because theyare expensive to fix, but because their creators want them so. Finally, accord-ing to cultural theories (Banfield, 1958; Weber, 1958; Putnam, 1993; Landes,1998), societies hold beliefs that shape collective action and government. Somesocieties are so intolerant or distrustful that their governments simply cannotfunction effectively.
To evaluate the ability of the different theories to explain the variation in gov-ernment performance across countries we need to find (reasonably) exogenous
224 The Journal of Law, Economics, & Organization, V15 N1
sources of variation in the economic, political, and cultural characteristics ofthese countries. Begin with economic theories. Perhaps the most direct impli-cation of these theories is that as the scale of economic activity expands, betterinstitutions become affordable, and hence government performance should im-prove (North, 1981). Of course, better government in turn improves economicperformance, so the scale of economic activity is endogenous. It turns out to betrue that richer countries have better government, but this is hardly the wholestory.
With political and cultural theories we have better exogenous measures ofcountry characteristics. We use two strategies for evaluating political theories,the essence of which we take to be that political divergence in society—betweensocial, ethnic, class, or other interests—is detrimental to government perfor-mance. One strategy is to look at ethnic heterogeneity in a country, whichseveral recent studies have identified as an important exogenous measure of re-distributive tendencies (Alesina, Baqir, and Easterly, 1999; Easterly and Levine,1997). Another strategy is to look at the legal systems, which can be viewedas indicators of the relative power of the State vis-`a-vis property owners. Inparticular, common law has developed in England to some extent as a defenseof Parliament and property owners against the attempts by the sovereign toregulate and expropriate them. Civil law, in contrast, has developed more asan instrument used by the sovereign for State building and controlling eco-nomic life. Finally, socialist law is the expression of ultimate control of theeconomy by the State. We consider a country’s legal system as a potentialdeterminant of government performance. In this article we do not use standardpolitical variables—parliamentary versus presidential system, fragmentation ofthe parliament, democracy versus dictatorship—as independent determinantsof government performance because, from our perspective, these variables arelargely endogenous.
To examine cultural theories, we follow a variety of authors, from Weber toLandes, who use religion as a proxy for work ethic, tolerance, trust, and othercharacteristics of society that may be instrumental in shaping its government.We look at the religious affiliations of the population as the potential culturaldeterminants of performance, and in particular focus on the Catholic and Mus-lim religions which have been recently singled out by Landes (1998) as hostileto institutional development.
Our data show a strong positive association between per capita income andgovernment performance, but they also suggest that economic theories of insti-tutions are not the whole story. To a significant extent, government performanceis determined by political, and perhaps also cultural, factors. We find that eth-nolinguistic heterogeneity and the use of a more interventionist legal system,such as socialist or French civil law, predict inferior government performance.So do significant shares of Catholics or Muslims in the population. The influ-ence of religion on government performance has a number of political, as wellas cultural, interpretations.
In the next section we spell out the different theories and their potentiallytestable predictions. In Section 3 we describe our data and empirical strategy.
The Quality of Government 225
In section 4 we present the main results bearing on the alternative theories.Section 5 concludes.
2. Theories of Institutional Development2.1 Measuring Government Performance
The government shapes the economic life of a country in a variety of ways:protecting or grabbing property, allowing or suppressing dissent, serving orabusing the public through its agents. To begin understanding what explainsthe variation in government performance across countries, our first step is todescribe some of the dimensions of such performance.
Perhaps the most standard view is that a good government protects propertyrights, and keeps regulations and taxes light; that is, a good government isrelatively noninterventionist (Smith, 1776; North, 1981; Knack and Keefer,1995). Our first group of measures, therefore, focuses on interventionism, andspecifically on the quality of regulation and the security of property rights. Onearea where the interpretation of interventionism is ambiguous is taxation. Inone view, high taxation is a measure of high intervention. In another view, hightax rates are imposed with the consent of the governed to finance sought-afterpublic goods, whereas low taxes are all that a deeply interventionist governmentcan hope to collect. Recent interpretations of higher taxes in Britain than inFrance in the 18th century (Brewer, 1988; Finer, 1997), as well as the evidenceon transition from socialism (Johnson, Kaufmann, and Shleifer, 1997), areconsistent with the notion that higher tax rates may go hand in hand with betterinstitutions.
Noninterventionism is only one aspect of good government. A further im-portant, and separate, dimension can be described as the efficiency of govern-ment, or the quality of the bureaucracy (Rauch, 1995; Rauch and Evans, 1997).When a government intervenes, it can do so reasonably efficiently, or with de-lays, corruption, and other distortions (Mauro, 1995; Treisman, 1997). Whena government taxes, it can do so with relatively high compliance, or with lowcompliance, which often leads to corruption and arbitrary variation of effectivetax rates across similarly situated taxpayers. On average, greater interven-tionism should be associated with lower efficiency, since entrusting officialswith greater regulatory and taxing powers invites corruption and bureaucraticdelay (Shleifer and Vishny, 1993). Nonetheless, some bureaucracies delivera given bundle of interventions more efficiently than others. There are manyexamples indicating that interventionism and efficiency are not just mirror im-ages of each other. The government in 18th-century Britain was efficient bythe standards of the day and (outside of taxation) relatively noninterventionist,whereas the government of Frederick the Great of Prussia was also efficient, buthighly interventionist (Johnson, 1975; Ertman, 1997). The Giolitti governmentin Italy before World War I was highly corrupt and inefficient, but relativelynoninterventionist; under Mussolini, the government became much more inter-ventionist while keeping its low efficiency levels; under Christian Democratsafter World War II, interventionism was reduced while the efficiency levelswere maintained (Mack Smith, 1997). In theory, it may help development for a
226 The Journal of Law, Economics, & Organization, V15 N1
highly interventionist government to be inefficient (Huntington, 1968). In anyevent, interventionism and efficiency are in part distinct aspects of governmentperformance.
In addition to controlling, taxing, regulating, and charging bribes, govern-ments in many countries actually provide public services that are essential foreconomic development. Government performance of a given country shouldbe assessed in part by evaluating the quality of public good provisions such asschooling, infant mortality, literacy, and infrastructure. While some of thesegoods are in part provided privately, governments have come to play a largerole in delivering health, education, and infrastructure. High quality of thesegoods, as opposed to just high expenditures, is a sign of a well-functioninggovernment.
A more problematic, but nonetheless important, indicator of performance isgovernment expenditures on transfers, its own consumption, and public sectoremployment. High government expenditures in these areas may reflect the citi-zens’ willingness to pay taxes because they like what the government does, andas such reflect good government. Alternatively, high expenditures on transfersand subsidies or on government consumption may reflect high levels of distor-tionary taxes and redistribution (Barro, 1991), and hence represent a failure toprotect the public from state intervention. Some measures of the size of gov-ernment, such as the size of the state-owned enterprise sector and the relativesize of public sector employment (Shleifer and Vishny, 1994, 1998), are per-haps more indicative of political and redistributive, rather than public-spirited,intent. Examining the determinants of these types of government spending maythus help us distinguish alternative theories of institutions.
A final dimension of good government which we examine is democracyand political rights. We look at this both because political freedom is a cru-cial element of good government and because economic freedom generallygoes together with political freedom (Hayek, 1944). The relationship betweendemocracy and economic success has been difficult to find in recent data (Barro,1996b), although over the longer span of history more limited governments havepresided over more successful development (DeLong and Shleifer, 1993).
2.2 Theories of InstitutionsEconomists, historians, sociologists, and political scientists have developed alarge number of theories of institutional development. Unfortunately, scholarsoften pursue their own theories without paying too much attention to the al-ternatives. This is true, for example, of Marx (1872), Weber (1958), Demsetz(1967), North (1981, 1990), Putnam (1993), Fukuyama (1995), and Landes(1998). Our goal is to provide some evidence that illuminates, and perhapscan distinguish, alternative theories of institutions and their implications forgovernment performance. To this end we first organize these theories (withoutpresenting any new ones) and then present their possibly testable implications.
We divide the available theories into three broad categories: economic, po-litical, and cultural. These theories focus on different reasons why institutionslook the way they do: social efficiency needs in the case of economic theories,
The Quality of Government 227
redistribution toward powerful groups in the case of political theories, and socialbeliefs in the case of cultural theories. Yet all these theories are consistent witheconomics more broadly in the sense that political actors pursue their selfishobjectives in light of their beliefs. Even so, this division is probably too crude,and the writers we examined—most notably North—typically have economic,political and cultural elements to their stories.
Economic theories of institutions suggest that they are created whenever thesocial benefits of doing so exceed the costs (Demsetz, 1967; North, 1981).1
For example, a government protects private property when the returns to suchprotection exceed the cost of police. This theory generally takes the view thatinstitutions are efficient, and that the trouble is the absence of institutions, ratherthan the existence of bad ones. Even if this theory is an important part of reality,it is at best incomplete—we see too many extractive and inefficient governmentsaround (Bates, 1981; Olson, 1996; Finer, 1997; Shleifer and Vishny, 1998).
A second group of theories of institutions is political; they state, roughly,that institutions and policies are shaped by those in power to stay in power andto amass resources. Government policies are used to control assets, includingpeople, and to convert this control into wealth.2 To Marx, societies are dividedinto social classes, and policies are designed by the ruling class. Moderntheories are more nuanced and allow redistributive policies to be shaped bysovereigns, bureaucracies, ethnic groups, religions, or even particular lobbies.Thus the Mandarin bureaucracy shaped the Chinese government for its own andthe emperor’s benefit for hundreds of years, and the Communist bureaucracyin the USSR did the same during Soviet rule. What political theories havein common is the idea that, when some group in a society becomes powerfulenough, it shapes policies to its own rather than social advantage.3
Perhaps the best illustrations of political theory come from the history ofsovereign state-building, which has been magnificently recounted by Finer(1997). Finer shows, for example, how the Russian czars, Ottoman sultans,and Tokugawa shoguns created polities in which they had absolute uncheckedpower through near complete control over the military, the aristocracy, the re-ligion, and the bureaucracy [see also Pipes (1974), Jones (1981), and Tilly(1990)]. These empires were short on property rights, long on government in-tervention, and (eventually) quite short on government efficiency as well; theyhad few laws or civil rights and relatively small governments. These qualities
1. Following Olson (1965), these theories pay a great deal of attention to free-rider problemsthat need to be overcome to produce any public goods, including institutions. This is where thecultural theory typically enters: ideology or beliefs help overcome free-riding.
2. Analytically we can analyze the political theories of institutions in terms of the control rightsof various agents (Grossman and Hart, 1986; Boycko, Shleifer, and Vishny, 1995, 1996).
3. Governments become massively redistributive when there are relatively few very powerfulgroups with different interests, not when there are many relatively weak groups each pushing inits own direction. This view is consistent with Becker’s (1983) notion that interactions of manyrelatively small lobbies lead to efficient outcomes and is inconsistent with Olson’s (1982) view thatincreased density of such lobbies reduces efficiency through political stalemate. Having lots ofweak articulated interests is better than having a few very strong ones.
228 The Journal of Law, Economics, & Organization, V15 N1
of government were an immediate consequence of the intent of the politicalrulers to maintain complete control over their subjects.
Finer (1997) contrasts this oriental despotism with European absolutism,where the power of the monarch was at least partially checked by law, andwhere the Church (in Catholic countries) and the aristocracy to some extentrestricted autocratic control. These countries, as a consequence, had moresecure property rights, and greater political rights of the subjects, although thesovereigns continually tried to restrict these rights to enhance their revenues.Civil law developed in Western Europe as part of such restrained control bythe sovereigns over their subjects. Consistent with the point that governmentefficiency is a separate matter than intervention, some European countries, suchas Prussia, managed to create relatively efficient bureaucracies, while others,such as Italy and France, have created “patrimonial” bureaucracies that usedtheir powers to pursue the personal interests of the officials rather than those ofthe sovereign (Ertman 1997).
Finally, in contrast to European absolutism, British government was shapedby the victory of aristocracy over the crown, and resulted in a more limited gov-ernment, greater political freedoms, and a more efficient bureaucracy. Commonlaw was developed, in part, as a mechanism for protecting the subjects from thecrown. Perhaps surprisingly, the consent of the governed enabled the BritishParliament to raise considerably higher tax revenues for military spending thanthe French kings could ever raise (Brewer, 1988): the British government wasbigger though less interventionist.4 Whatever the nuances of these histories,the basic point that institutions and policies are shaped not by considerations ofefficiency but rather by the ability of the various players to extract rents seems tobe broadly confirmed by the narrative accounts of world history [see also Jones(1981), DeLong and Shleifer (1993), Ertman (1997), Kamen (1997), etc.].
In some cases, political and economic theories of institutions share similarpredictions. For example, it may serve the interest of tax-extracting sovereignsto impose some constraints on their own powers so as to “fatten” the goosebefore roasting it, or even to provide public goods such as justice and propertyrights protection to enhance tax revenues (North, 1981; Olson, 1993). In thosecases, even though the ultimate goal of a policy is redistribution toward thesovereign through enhanced tax collections, its immediate effect is to increaseefficiency. Frederick the Great’s reforms are a great example of this phenomenon(Johnson, 1975). Although kings are surely different from outright plunderers,the economic and political theories differ in that the former, in the first instance,do not accommodate purely rent-seeking policies. Unlike economic theories,political theories readily predict the existence of inefficient, interventionist, anddistortionary policies that are put in place not because they raise the social prod-uct but because they redistribute it. Such clearly redistributive policies renderpolitical and economic theories distinguishable at least in principle.
4. This contrasts with the very limited aristocratic government of Poland, where the baronsoccasionally could not agree to pay for armies needed to resist aggression, with the result that thegovernment was small and the country got even smaller (Ertman, 1997; Finer, 1997).
The Quality of Government 229
Finally, cultural theories state that some societies form beliefs and ideas thatare conducive to good government, while others do not (see Weber, 1958).Some of these beliefs and ideas are nonverifiable (e.g., beliefs in punishmentsare rewards after death), others verifiable and false (e.g., certain racists andanti-Semitic beliefs), and still others self-fulfilling (e.g., the belief that yourneighbors do not cooperate in any collective action and so you should noteither). When these beliefs are highly pervasive and persistent, they get to becalled “culture.”
Two recent instantly famous versions of such cultural theories are due toRobert Putnam (1993) and David Landes (1998). Putnam’s theory [see alsoColeman (1990) and Fukuyama (1995)] states that trust in strangers facilitatescollective action, which he views as essential for the provision of public goods.Putnam comments that in Italy the Catholic Church had adverse effects ontrust, because “vertical bonds of authority are more characteristic of the ItalianChurch than horizontal bonds of fellowship” (p. 107). Putnam compares avariety of social outcomes in high-trust northern Italy and low-trust southernItaly, and finds confirmation of his views. Knack and Keefer (1997) and LaPorta et al. (1997b) also find empirical support for the view that higher trust instrangers predicts better government performance, even across countries.
More recently, Landes (1998) argued that during the period of economictakeoff of Protestant countries, Catholic and Muslim countries in particularhave acquired cultures of intolerance, xenophobia, and closed-mindedness thatretarded their development. According to Landes, starting in the 15th cen-tury or even earlier, Catholic countries of southern Europe and Latin America,egged on by the Church, burned heretics, forbade learning and travel by theircitizens, censored books, and restricted the inflow of new ideas. In part, theCatholic Church felt threatened by other religions emerging from the ProtestantReformation, and in part it wanted to extend its influence. The consequencesof such intolerance were both the enormous increase in the power of churchand state as they dedicated substantial resources to keeping out new ideas, andthe inability of Catholic countries to learn from others. Landes believes thatthis intolerance was responsible for the decline of Spain, Portugal, and Italy,for poverty of Latin America, and for many other ills. Similarly, the declineof Muslim countries after the 13th century (that followed the golden age ofopenness and prosperity) is also explained by the newly found but long-lastingintolerance as a means of political and religious control.
Many cultural explanations of institutions and policies have a political el-ement to them, as Landes’s emphasis on the use of intolerance for politicalends makes clear. Thus the Spanish Inquisition might have been the supremeexample of Catholic intolerance that Landes takes it to be. But it was, in thefirst place, a political attempt by Ferdinand and Isabella to wrest control overreligion from Rome by organizing religious life around the courts of Inquisitionpaid for by the Crown (Kamen, 1997). Similarly, the Church throughout theCatholic world has fought the State to tax and regulate the citizenry (Barra-clough, 1972; Putnam, 1993; Finer, 1997; Mack Smith, 1997). It ultimatelylost, but not before making a significant imprint on the governments of these
230 The Journal of Law, Economics, & Organization, V15 N1
countries.5 Interestingly, as Huntington (1991) explains, the Catholic Churchin the 1960s switched from being a powerful force against liberalism (and there-fore a source of poor culture, according to Landes) to a powerful force towarddemocratization, presumably again for the political reason of trying to maintainmembership. Perhaps in part as a consequence, most of the countries that havedemocratized in the last 30 years, including Portugal, Spain, Poland, Hungary,Philippines, Nicaragua, Argentina, Brazil, and Mexico, have been Catholic.Finally, one could argue that Tokugawa Japan—perhaps the most intolerantand xenophobic polity of them all—changed its policies in the Meiji Restora-tion because the cultural beliefs of the officials have become more “Western”(Beasley, 1972). But there is an obvious political explanation: the politicalinterests of the Tokugawa shoguns, namely controlling internal threats, werevery different from those of the Meiji reformers, namely addressing the externalthreat. In short, culture appears to be quite often shaped by politics.
2.3 Developing HypothesesOur goal is to look for exogenous variation in economic, political, and culturalcircumstances across countries that may account for the variation in governmentperformance. Finding such exogeneity is rather difficult. Economists looked atsuch variables as government consumption (Barro, 1991), democracy (Barro,1996a), corruption (Mauro, 1995), and institutional quality (Knack and Keefer,1995) as determinants of economic growth. More recently, Hall and Jones(1999) have examined institutional quality, size of government, openness totrade, and private ownership as determinants of productivity across countriesand made the brave argument that productivity is unlikely itself to influenceinstitutions. From our perspective, of course, the quality of government isendogenous; to understand what determines government performance, and todistinguish between theories, we need to look at the more fundamental, or atleast historically predetermined, variables.
From this perspective, economic theories of institutions are the hardest to test.The thrust of these theories is that economic development itself creates a demandfor good government, and hence the appropriate measure of demand is per capitaincome. Based on this logic, our test of economic theories is to consider percapita income as a potential determinant of government performance. But goodinstitutions themselves improve economic conditions. For example, we wouldexpect better protection of property, improved government efficiency, higherquality private goods, perhaps big but good government, and perhaps politicalfreedoms to enhance per capita income. We therefore have a relatively weaktest of economic theories and, in fact, the article focuses on other determinantsof government performance.
We use two strategies to identify the relevance of political theories of insti-tutions. The first measure of the State’s predisposition to major redistribution
5. DeToqueville (1858) discusses the motivation of French revolutionaries by noting that “tooverthrow the institutions of existing social order they must begin by destroying those of theChurch, on which they were modeled and from which, indeed, they derived” (p. 151).
The Quality of Government 231
is ethnic heterogeneity [see, e.g., Mauro (1995); Alesina, Baqir, and Easterly(1999); Easterly and Levine (1997)]. In ethnically heterogeneous societies, ithas been common for the groups that come to power to fashion governmentpolicies that expropriate (or kill) the ethnic losers, restrict their freedom ofopposition, and limit the production of public goods to prevent those outsidethe ruling group from also benefiting and getting stronger. Alesina, Baqir, andEasterly (1999) offer a theory in which disagreement over the nature of pre-ferred public goods leads ethnically heterogeneous societies to demand fewerof those goods, but there are many other—much more hostile—mechanismsthrough which heterogeneity can undermine government performance. Thepolitical theories predict that, as ethnic heterogeneity increases, governmentsbecome more interventionist and less efficient, and the quality of public goodsfalls, as does the size of government and political freedom.6
Our second strategy for evaluating political theories is to consider each coun-try’s legal origin. We divide national commercial legal traditions into commonlaw, French civil law, German civil law, Scandinavian law, and socialist law.These traditions were developed in England, France, Germany, Scandinavia,and the Soviet Union but then spread through the world through conquest,colonization, imitation, and voluntary adoption. The distinctions between theFrench, German, and Scandinavian families are relatively subtle (as we discussbelow), but the distinctions between socialist, civil, and common law traditionsare not.
To begin, socialist law is a clear manifestation of the State’s intent to createinstitutions to maintain its power and extract resources, without much regardfor protecting the economic interests or the liberties of the population. Thegoal of socialist law is to keep the Communist Party in power, not to protectproperty or freedom.
Civil law, particularly since the time of Codification in the 19th century, hasalso been largely an instrument of the State in expanding its power, thoughin a more constrained way than socialist law [see Elster, Offe, and Press(1997:39); Finer (1997:1564)]. It is not surprising that the principal legalcodes of the world were introduced by the two greatest State builders of the19th century: Napoleon and Bismarck. Civil law is largely legislature cre-ated and is focused on discovering a just solution to a dispute (often from thepoint of view of the State) rather than on following a just procedure that pro-tects individuals against the State (David and Brierley, 1978:331). Amusingly,Rene David—whose book can be fairly seen as a treatise on the superiorityof civil law over common law—is surprised at France’s weakness in the areaof public law that deals with restraints on public officials (David and Brierley,1978:76).7 A civil legal tradition, then, can be taken as a proxy for an intent
6. One can alternatively argue that ethnic heterogeneity is related to ethnic hatred, which iscultural rather than political. Often, though not always, ethnic hatred seems to have politicalantecedents.
7. An illustration is worth quoting in full: “Thus a taxpayer, even in the absence of any textuallyestablished right or some error committed in his case, can bargain with the tax authorities or obtain
232 The Journal of Law, Economics, & Organization, V15 N1
to build institutions to further the power of the State, although not to the sameextent as in the socialist tradition.
The English common law tradition is entirely different in that its developmentstarting in the 17th century has been shaped by Parliament and the aristocracy atthe expense of the Crown, and hence it has reflected to a much greater extent theintent to limit the power of the sovereign (David and Brierley, 1978:303; Finer,1997:1347–1348). As a result of this influence, the judges who made commonlaw “put their emphasis on the private rights of individuals and especiallyon their property rights” (Finer, 1997:1348). There is also more emphasison restraining the government and on protecting the individual against thegovernment. A common law tradition, then, can be taken as a proxy for theintent to limit rather than strengthen the State.
In our earlier work (La Porta et al., 1997a, 1998) we considered legal tradi-tions in a narrower context of examining legal protection of investors and theeffect of such protection on capital markets. Here we use legal traditions ascruder proxies for the political orientation of governments. Most clearly, weexpect that countries with socialist laws have the most interventionist govern-ments, next come countries with civil laws, and finally common law countries.We would predict the same ordering for political freedom. In other dimen-sions of the quality of government, however, the relationships might be morecomplicated.
We expect that government efficiency is the lowest in socialist law countries,in part because the extreme power of the State corrupts the bureaucracies. Wealso expect that government efficiency should be high in common law, Scan-dinavian, and German civil law countries, despite the interventionist stance ofthe law in the latter two groups, largely because these countries have managedto build professional rather than patrimonial bureaucracies based on the armiesand professional civil servants rather than aristocrats and clerics (Ertman 1997;Finer 1997). We expect the French civil law countries to be intermediate ingovernment efficiency because the bureaucracy was built to be powerful andlargely unconstrained. Having made these conjectures, we note that our theo-retical and historical priors in this dimension of government performance arerelatively weak.
The predicted effects of the legal systems on public good provision andgovernment spending are not entirely straightforward either. We expect socialistlaw countries, in part because of their commitment to equality and in partbecause of their militarism, to do well on education and infant mortality, thoughnot necessarily on infrastructure. With respect to public good provision and thesize of government, we have no strong priors: civil law countries may have agreater interest in state expansion but a lower ability to tax than the commonlaw countries. A more interventionist state is not necessarily a larger one.
a release in view of his special circumstances. According to the French Ministere de l’economieet des finances, about 200,000 taxpayers each year obtain some reduction in or release from taxesupon an appeal to thebienveillanceof the administration.Le Monde, August 9, 1968” (David andBrierley, 1978:77).
The Quality of Government 233
One possible concern with using legal systems as independent variables isthat they are, to some extent, endogenous—a symptom of the relative powerof the State and its subjects rather than the cause. Thus common law acquiredits rules because the Crown was weak in the first place, and not the other wayaround. But this concern, while legitimate, makes legal systems particularlyappropriate independent variables for testing political theories of institutions.Suppose that legal systems, which were acquired centuries ago as part of thepolitical process (both voluntarily and not), are found to influence governmentperformance today. This would mean that the historically determined politicalallocation of power, as reflected in the law, shapes government performancetoday, precisely as the political theory predicts. Despite their historical endo-geneity, legal variables serve our empirical purpose extremely well.
Finally, we come to the cultural theories. Following Weber (1958), Put-nam (1993), and Landes (1998), our proxy for the dimension of culture thatinfluences government performance is religion—more precisely, the percent-age of population in each country belonging to different religious affiliations.In an earlier article (La Porta et al., 1997b) we provided some evidence thatcountries whose populations belong primarily to what we called “hierarchicalreligions”—Catholicism, Islam, and Greek Orthodox—exhibit inferior gov-ernment performance to that of largely Protestant countries. In this articlewe separate out the three most widespread religions—Muslim, Catholic, andProtestant—from all others.8 We also focus on Muslim and Catholic religionsbecause Landes (1998) singles them out as particularly detrimental to develop-ment.
To begin, we note that cultural theories typically do not focus on government.Weber (1958) is interested in entrepreneurship, Putnam (1993) focuses on publicgood provision, and Landes (1998) is concerned with the flow of people, goods,and ideas between countries. Nonetheless, we try to develop some hypothesesfor the effects of religion on governments.
It is probably fair to say that both Muslim and Catholic countries would beviewed by the adherents of cultural theories as being more interventionist—in part because the doctrines of these religions are more interventionist (theylike to tell people what to do) than Protestantism, and in part because thesereligions grew to support State power. Cultural theories would probably alsopredict that governments in predominantly Catholic and Muslim countries areless efficient partly as a consequence of excessive power, and partly becausebureaucracies in these countries, particularly the Catholic ones, have developedfrom religious ranks (clerk comes from cleric), and hence were not as dependenton the sovereign. Cultural theories would definitely predict inferior provision ofpublic goods in Muslim and Catholic countries, although Putnam’s story here isa bit political: religion competes with, and undermines, the State in the provision
8. We do not focus on the Orthodox religion here because it is not as widespread as the otherthree and because many people in (at least formerly) Orthodox, but later on socialist, countriesreport to be nonreligious.
234 The Journal of Law, Economics, & Organization, V15 N1
of public goods.9 We would not even conjecture what these theories predict forgovernment consumption. Finally, these theories, particularly Landes (1998),would surely predict that Muslim and Catholic countries have fewer civil rights:the State-supported intolerance in these countries requires a curtailment offreedom to be carried out. (Remember, however, the switch by the CatholicChurch after Vatican II in the 1960s.)
3. DataOur analysis is based on a dataset of measures of government performance andtheir potential determinants in (up to) 152 countries. Since we are combininga large number of datasets, we have different numbers of observations fordifferent variables, with some regressions covering as few as 47 countries. Thedata typically also come from different years, although most of the data are fromthe 1990s. The definitions and sources for all the variables used in this articleare summarized in Table 1; Appendices A and B present summary statisticsand the data on the independent variables.
To gain robustness, we try to use as our measures of government perfor-mance both objective and survey measures from different data sources. This isparticularly important for subjective assessments of government performance,sincewithin the same surveyresponses to different questions may simply reflectsome general underlying sentiment toward a country. When different surveysuse different respondents, this risk is reduced.
We measure government intervention by an index of property rights protec-tion, an index of the quality of business regulation, and the top marginal tax rate.We measure government efficiency by survey scores on corruption, bureaucraticdelays, and tax compliance. We also include in this category a measure of rel-ative wages of government officials to see if higher wages are associated withmore efficient government. We measure the output of public goods by infantmortality, school attainment, illiteracy, and an index of infrastructure quality.(With illiteracy and infant mortality, we depart from the convention that highervalues of the measure refer to better performance.) We measure the size of thepublic sector by government transfers and subsidies, government consumption,an index of the size of the state enterprise sector (higher value means smaller),and a measure of the relative size of public sector employment. Finally, we useindices of democracy and of political rights in our democracy category.
Table 2 presents the correlations between the dependent variables, whichreveal a number of fascinating patterns. The correlations between measures ofgovernment performance within the same category are generally high. In theintervention category, there is a positive correlation between property rightsprotection and the business regulation index, but the top tax rate is not signif-icantly related to these two measures of government intervention. This result
9. Another version of the cultural story is that religions that require their followers to read thesacred texts (such as Protestantism) encourage literacy, whereas religions that rely on the auditoryabsorption of information (such as Catholicism) discourage literacy.
The Quality of Government 235
Table 1. Description of the Variables
Variable Name Description and Source Number ofObservations
Interference with the private sector:Property A rating of property rights in each country 149rights (on a scale from 1 to 5). The more protectionindex private property receives, the higher the score.
The score is based, broadly, on the degree of legalprotection of private property, the extent to whichthe government protects and enforces laws thatprotect private property, the probability thatthe government will expropriate private property,and the country’s legal protection to privateproperty. Source: Holmes, Johnson, andKirkpatrick, 1997.
Business A rating of regulation policies related to 149Regulation opening a business and keeping open aindex business (on a scale from 1 to 5). Higher
score means that regulations are straight-forwardand applied uniformly to all businesses and thatregulations are less of a burden to business. Source:Holmes, Johnson, and Kirkpatrick, 1997.
Top tax rate Top marginal tax rate for each country 82in 1994. Source: Gwartney, Lawson, andBlock, 1996.
Efficiency:Corruption Corruption in government index. Low 126
ratings indicate “high government officialsare likely to demand special payments” and“illegal payments are generally expected thoughtlower levels of government” in the form of“bribes connected with import and exportlicenses, exchange controls, tax assessment,policy protection, or loans.” Scale from 0to 10. Average of the months of April andOctober in the monthly index between 1982and 1995. Source: Political RiskServices, various years.
Bureaucratic An indicator of bureaucratic delays 60delays (red tape). Low ratings indicate lower
levels of red tape in the bureaucracy of thecountry. Scale from 0 to 10. The index ispublished three times per year. The data isthe average of the years between 1972 and1995. Source: Business Environmental RiskIntelligence’s (BERI) Operation Risk Index.
Tax compliance Assessment of the level of tax 49compliance. Scale from 0 to 6, where higherscores indicate higher compliance. Data isfor 1995. Source: World Economic Forum, 1996.
Continued
236 The Journal of Law, Economics, & Organization, V15 N1
Table 1. Continued
Avg. government The ratio of average wages of 63wages/GDP per central government to per capitacapita GDP in each country. Certain non-wage
benefits are not included in the estimate ofthe average central government wage. Source:Schiavo-Campo, de Tommaso and Mukherjee, 1997.
Output of public goods:Log of infant Logarithm of the number of 196mortality deaths of infants under one year
of age per one thousand live births forthe years 1970–1995. Source: WorldBank, World Development Indicators1997 (WDI).
Log of school Log of schooling taken over 106attainment five year periods (1960–65,
1970–75, and 1980–85). Each value isobtained as the logarithm of (1 + averageyears of school attainment during therespective period). Source: Barroand Lee, 1994.
Illiteracy rate Average of adult illiteracy 128rate for the years 1990–1995. Adultilliteracy rate is the proportion of adultsaged 15 and above who cannot, withunderstanding, read and write a short,simple statement on their everyday life.1990–1995. Scale 0 to 100. Source: WDI.
Infrastructure Assessment of the “facilities 60quality for and ease of communication between
headquarters and the operation, and within thecountry,” as well as the quality of thetransportation. Average data for the years1972 to 1995. Scale from 0 to 10 with higherscores for superior quality. Source:BERI’s Operation Risk Index.
Size of public sector:Transfers and Total government transfers and 90subsidies/GDP subsidies as a percentage of GDP
(scale from 0 to 100). Average for theyears 1975–1995. Source: Gwartney,Lawson, and Block, 1996 (with datafrom the World Bank and InternationalMonetary Fund).
Continued
The Quality of Government 237
Table 1. Continued
Government Government consumption expenditures 104consumption/ as a percentage of GDP (scale fromGDP 0 to 100). Average for the years 1975–1995.
Government consumption expenditures “includeall spending on goods and services purchasedby the government—things like nationaldefense, road maintenance, wages and salaries,office space, and government-owned vehicles.Since it is obtained from the nationalincome accounts, it includes all levelsof government spending. It does not includedirect transfers and subsidies, since thesedo not enter into the national income accounts.”Source: Gwartney, Lawson, and Block, 1996(with data from the World Bank and InternationalMonetary Fund).
SOEs in the Index of State-Owned Enterprises 104economy as a share of the economy (scale from
0 to 10). Higher scores include countries withless government-owned enterprises which areestimated to produce less of the country’s output.As the estimated size and breadth of the SOE sectorincreases, countries are assigned lower ratings.Average of the score for the years 1975–1995.Source: Gwartney, Lawson, andBlock, 1996.
Public sector Average of the ratio of public sector employment 124employment/total in general government to total populationpopulation for the years 1976–1996. General government
employment includes employment in “all governmentdepartment offices, organizations and other bodieswhich are agencies or instruments of the centralor local authorities whether accounted for orfinanced in, ordinary or extraordinary budgets orextra-budgetary funds. They are not solely engaged inadministration but also in defense and public order, inthe promotion of economic growth and in the provision ofeducation, health and cultural and social services.” Source:Schiavo-Campo, de Tommaso, and Mukherjee, 1997.
Political freedom:Democracy index Average of democracy score for the period 161
1970–1994. Scale from 0 to 10, with lower valuesindicating a less democratic environment. Source:Jaggers and Gurr, 1996.
Political rights Index of political rights. Higher ratings 209index indicate countries that come closer “to the ideals
suggested by the checklist questions of: (1) free and fairelections; (2) those elected rule; (3) there are competitiveparties or other competitive political groupings; (4)the opposition has an important role and power; and(5) the entities have self-determination or an extremelyhigh degree of autonomy.” Source: Freedom House,1996.
Continued
238 The Journal of Law, Economics, & Organization, V15 N1
Table 1. Continued
Determinants:Ethnolinguistic Average value of five different indices 161fractionalization of ethonolinguistic fractionalization.
Its value ranges from 0 to 1. The five component indicesare: (1) index of ethnolinguistic fractionalization in 1960,which measures the probability that two randomly selectedpeople from a given country will not belong to the sameethnolinguistic group (the index is based on the numberand size of population groups as distinguished by theirethnic and linguistic status); (2) probability of tworandomly selected individuals speaking differentlanguages; (3) probability of two randomly selectedindividuals do not speak the same language; (4)percent of the population not speaking the officiallanguage; and (5) percent of the population notspeaking the most widely used language. Sources:Easterly and Levine, 1997. The sources of thecomponents of the average index are (1) AtlasNarodov Mira, 1964; (2) Muller, 1964; (3)Roberts, 1962; (4) and (5) Gunnemark, 1991.
Legal origin Identifies the legal origin of the 212Company Law or Commercial Code of each country.There are five possible origins: (1) English CommonLaw; (2) French Commercial Code; (3) GermanCommercial Code; (4) Scandinavian CommercialCode; and (5) Socialist/Communist laws. Source:La Porta et al., 1998, extended using“Foreign Laws: Current Sources of BasicLegislation in Jurisdictions of theWorld,” 1989; and CIA WorldFactbook 1996.
Religion Identifies the percentage of the population 209of each country that belonged to the three mostwidely spread religions in the world in 1980. Forcountries of recent formation, the data is availablefor 1990–1995. The numbers are in percent (scale from0 to 100). The three religions identified here are:(1) Roman Catholic; (2) Protestant; and (3) Muslim.The residual is called “other religions”. Sources:Barrett, 1982, Worldmark Encyclopediaof Nations 1995, Statistical Abstractof the World 1995, United Nations, 1995,CIA 1996.
Economic Development:Latitude The absolute value of the latitude of the 209
country, scaled to take values between 0 and 1.Source: CIA 1996.
Log GNP per Logarithm of GNP per capita expressed in 186capita current U.S. dollars for the period 1970–1995.
Source: WDI.
The Quality of Government 239
gets back to the earlier point: high taxes are not necessarily a sign of an inferiorgovernment.
In the government efficiency category, less corrupt countries also have fewerbureaucratic delays and higher tax compliance. Tax compliance and the scoreon bureaucratic delays are also positively correlated. These correlations sup-port the view that corruption is another side of bureaucratic discretion, sincedelays create the opportunities to take bribes. Finally, corruption is negativelycorrelated with the relative wages of government officials. Contrary to the viewthat higher pay for bureaucrats is a remedy for corruption, the raw correlationsuggests just the opposite. It is likely instead that in countries where bureaucratshave much power, they collect both higher wages and significant bribes. Con-sistent with this view as well, countries where the government is more efficient(including less corrupt) are on average also less interventionist.
The measures of public good output are highly correlated with each other.Low infant mortality, significant school attainment, low illiteracy, and high-quality infrastructure typically come together. Countries with good publicgoods are also the ones with efficient government, low (!) relative governmentwages, and less intervention (except for the high top tax rate).
With the exception of the SOE index, the various measures of the size ofgovernment are strongly correlated with each other. Countries with more gov-ernment consumption, transfers, and labor are less corrupt; they have fewerbureaucratic delays and better provision of public goods, but also higher taxrates. Figures 1 and 2 present the scatter plots of the corruption index and infantmortality, respectively, against government transfers and subsidies as a shareof GDP. The scatter plots, and the correlations more generally, show clearlythat bigger governments, while taxing more, look better on just about everymeasure of performance. This result, that the larger governments tend to be thehigher quality ones, is one of our key findings. It does not imply, of course,that it is often desirable to expand a government of a given quality.10
The measures of democracy and political rights are highly correlated witheach other. Freer governments are larger, more efficient, intervene less, andprovide better public goods.
The correlations between the independent variables are presented in Table 3,and they too deliver interesting findings. We have three broad factors thatmay be related to government performance: ethnolinguistic fractionalization[measured, following Easterly and Levine (1997), as the average of severalmeasures of ethnic diversity], the origin of commercial laws, and the religiouscomposition of the population (the percentage of the population in the threemost widespread religions). We also include latitude and (the log of) per capitaincome as control variables. We discussed per capita income already. Weinclude (the scaled absolute value of) latitude because temperate zones havemore productive agriculture and healthier climates, which has enabled them
10. Lindbeck (1997), for example, suggests that even a reasonably efficient government, such asthat of postwar Sweden, can become too imposing to sustain economic growth.
240The
JournalofLaw,E
conomics,&
Organization,V15
N1
Table 2. Correlations of Dependent Variables
Interference with private sector Efficiency Output of public goods Size of public sector Politicalfreedom
Property Business Top tax Corrupt. Bureau. Tax Gov. L. infant L. school Illiteracy Infrastr. Transf. Gov. SOEs Gov. Democracy
rights regulation rate delays compl. wages / mortality attainm. rate index & Subs. / consump / index labor / Index
index index GNP pc. GDP GDP population
Business .7382a
regulation
Top tax rate .1190 .0752
Corruption .5035a .4066a .2613
Bureaucratic .7100a .6938a .2789 .8550a
delays
Tax .6628a .5640a .0059 .5353b .6552a
compliance
Gov. wages −.0277 .0022 −.2862 −.5039c −.3598 .0013
/GNP percap.
Log infant −.5860a −.5704a −.2385 −.7192a −.8075a −.5516b .4585b
mortality
Log school .6553a .5684a .1888 .6912a .6783a .5080 −.4874 −.8674a
attainment
Illiteracy rate −.4211a −.4392a .3115 −.2250 −.4342 −.1162 .6113 .7811a −.8694a
Continued
TheQ
ualityofG
overnment
241
Table 2. Continued
Interference with private sector Efficiency Output of public goods Size of public sector Politicalfreedom
Property Business Top tax Corrupt. Bureau. Tax Gov. L. infant L. school Illiteracy Infrastr. Transf. Gov. SOEs Gov. Democracy
rights regulation rate delays compl. wages / mortality attainm. rate index & Subs. / consump / index labor / Index
index index GNP pc. GDP GDP population
Infrastructure .7410a .6836a .4035 .8142a .9298a .6229a −.3429 −.7907a .6793a −.3759
index
Transfers & .4831a .3257 .4183b .6512a .4858c .1207 −.5229 −.7147a .6514a −.3038 .5545a
subs. / GDP
G. consumpt. .1788 .1884 .3324 .4604a .4662 .2402 −.1852 −.3300 .2734 .0012 .4686c .4005b
/GDP
SOEs index .2431 .2950 −.1683 .1550 .5379a .6008a .1657 −.1794 .1446 −.1705 .4735c −.2976 −.1434
Gov. labor / .3811b .3157 .2692 .6077a .4703 .3047 −.4202 .7064a .6833a −.5207a .4969b .7222a .5235a .0362
population
Democracy .5921a .4735a .1209 .6395a .6653a .3120 −.2132 −.7033a .6606a −.4545a .7381a .4727a .3178 .2386 .5141a
index
Political .6022a .4600a .0338 .5388a .6251a .2165 −.1169 −.5673a .6708a −.4171a .6681a .5954a .2465 .1520 .3702a .7394a
rights index
aSignificant at 1% level.bSignificant at 5% level.cSignificant at 10% level.
242 The Journal of Law, Economics, & Organization, V15 N1
Figure 1. The size of government and corruption.
Figure 2. The size of government and infant mortality.
The Quality of Government 243
Tabl
e3.
Cor
rela
tions
ofIn
depe
nden
tVar
iabl
es
Eth
nolin
guis
ticE
nglis
hS
ocia
list
Fren
chG
erm
anS
cand
inav
ian
Pro
test
ant
Cat
holic
Mus
limO
ther
Latit
ude
fract
iona
lizat
ion
lega
lorig
inle
galo
rigin
lega
lorig
inle
galo
rigin
lega
lorig
inre
ligio
ns
Eng
lish
lega
lorig
in.1
568
Soc
ialis
tleg
alor
igin
−.17
41−.
3223
a
Fren
chle
galo
rigin
.053
8−.
6345
a−.
3894
a
Ger
man
lega
lorig
in−.
1491
−.13
39−.
0822
−.16
18
Sca
ndin
avia
nle
galo
rigin
−.15
66−.
1126
−.06
91−.
1361
−.02
87
Pro
test
ant
−.05
76.2
446b
−.20
62−.
2754
a.0
238
.516
5a
Cat
hlol
ic−.
1882
−.23
16−.
1984
.401
3a.0
328
−.13
86−.
1569
Mus
lim.1
912
−.07
25−.
0483
.177
9−.
1152
−.09
75−.
3473
a−.
4885
a
Oth
erre
ligio
ns.0
615
.134
8.4
504a
−.46
01a
.074
5−.
1104
−.16
83−.
4746
a−.
3021
a
Latit
ude
−.44
58a
−.27
58a
.442
6a−.
2429
.174
5.3
382a
.187
2−.
1389−.
0664
.116
0
Log
GN
Ppe
rcap
ita−.
4427
a−.
0096
−.01
93−.
1651
−.26
87b
.281
7b.2
614b
.121
6−.
1875
−.11
02.5
314a
a Sig
nific
anta
t1%
leve
l.b
Sig
nific
anta
t5%
leve
l.c
Sig
nific
anta
t10%
leve
l.
244 The Journal of Law, Economics, & Organization, V15 N1
to develop their economies and possibly their institutions as well (Landes,1998).
Ethnolinguistic fractionalization (EF) is unrelated to any of our religious orlegal origin variables, but is significantly negatively correlated with latitudeand per capita income. It is noteworthy that EF captures a different aspect ofsociety than religion and law.
The correlations between the legal and religious variables are predictable, butimportant. First, there are significant negative correlations between the percent-ages of Protestants and Muslims, Catholics and Muslims, Catholics and otherreligions, and Muslims and other religions across countries. Second, Catholiccountries typically have French legal origin; Protestant countries typically havean English or Scandinavian legal origin, and are less likely to have Frenchlegal origin; other religions are more likely to have a socialist legal origin, pre-sumably because of the significant presence of Orthodox religions in socialistcountries. Third, Scandinavian and German legal origin countries, as well asProtestant countries, are richer. Fourth, ethnolinguistically fractionalized andcommon law countries tend to be closer to the equator, whereas socialist andScandinavian legal origin countries tend to be further away from it. Of course,Scandinavia and Germany have not had many colonies, and hence their legaltraditions are restricted to wealthy European and East Asian countries.
4. Regression ResultsIn this section we present regressions of determinants of government perfor-mance. We present each regression both with and without the logarithm of percapita income and latitude controls.11 The argument for including per capita in-come (or for that matter, latitude) is that, on the economic theory of institutions,development improves government performance.12 On the other hand, goodpolicies themselves enhance per capita income, and so a regression with trulyexogenous independent variables should not include income. Including incomein the regressions together with other determinants of performance would thenspuriously reduce our estimates of the impact of these determinants on the qual-ity of government. For example, if EF undermines government performance,and poor performance reduces per capita income, then including income in theregression of government performance on EF would reduce the estimated effectof the latter. We estimate the regressions both ways, and try to understand howthe results depend on the specification.
We present the results in Tables 4–6. Since religious affiliations and legal
11. The logarithm of per capita income and latitude are strongly correlated with each other. Wealso ran the regressions including them one at a time. Each is nearly always significant whenincluded alone. Countries that are further from the equator, or richer, have better governmentperformance. We include them together because the correlation between them is not perfect.
12. Hall and Jones (1999) make the argument that per capita income (or productivity) shouldnot have a direct effect on institutional quality, based on their impression that, in the United States,institutions were better in 1790 than in 1990. The economic theory of institutions, which Hall andJones appear to endorse, argues that causation goes precisely in the opposite direction.
The Quality of Government 245
origins are correlated with each other, and EF appears to be uncorrelated witheither, we include legal origin and EF together as independent variables inTable 4 and religious affiliation and EF in Table 5. In Table 6 we include EF,legal origin, and religious affiliation.
Looking at Table 4 (or Table 5) we note that the logarithm of per capita income(and latitude) generally has a strong positive effect on government performance.Richer countries are less interventionist in that they protect property rights andregulate better, although they also have higher marginal tax rates. Richer coun-tries also have more efficient governments, better provision of public goods,and—as Wagner law suggests—a larger public sector (except a smaller stateenterprise sector).
The results on per capita income mean either that there is some merit tothe economic theories of institutions, or that good governments promote de-velopment, or both. Consistent with the importance of the correlation betweengovernment performance and per capita income, the explanatory power of theregressions generally rises sharply when per capita income is included in theregressions. Having established the importance of this correlation, we can nowturn to the central question: Do historical factors, as suggested by the politicaland cultural theories, help explain the variation in government performanceacross countries? As we show below, the answer is yes.
4.1 Ethnolinguistic FractionalizationWe discuss EF results based on Table 4, which only includes variables sug-gested by the political theories of institutions. The results on EF in Table 5are similar. In the specifications excluding latitude and per capita income, EFhas a very consistent adverse effect on government performance. Higher frac-tionalization is associated with more interventionism (worse property rightsand regulation), lower government efficiency (more corruption, longer delays,lower tax compliance), inferior provision of public goods (higher infant mortal-ity and illiteracy, lower school attainment and infrastructure quality), smallergovernment (transfers, consumption, and public employment), though morestate enterprises, and finally less political freedom. The adverse effects of EFon government performance are broadly consistent with the argument that EFcaptures the predisposition of ethnic groups in power to redistribute.
The negative effect of EF on government performance generally becomesinsignificant once we control for per capita income and latitude. (Recall thatEF is itself negatively correlated with per capita income.) Controlling for howpoor they are, ethnolinguistically fractionalized countries do not have espe-cially bad governments. One exception to this finding is public good provision,which is inferior in divided countries even controlling for their poverty. An-other exception is the prevalence of state ownership of firms, which is higherin divided countries even controlling for per capita income. The latter result isparticularly noteworthy because state-owned enterprises are such an importantmechanism of political redistribution. The effect of per capita income controlmay mean that EF adversely affects government performance, and bad perfor-mance in turn reduces per capita income, as the standard version of the political
246 The Journal of Law, Economics, & Organization, V15 N1
Tabl
e4.
Gov
ernm
entP
erfo
rman
cean
dLe
galO
rigin
Dep
ende
ntVa
riabl
esIn
depe
nden
tVar
iabl
es
Eth
nolin
guis
ticS
ocia
list
Fren
chG
erm
anS
cand
inav
ian
Latit
ude
Log
GN
PIn
terc
ept
Adj
.R
2
fract
iona
lizat
ion
lega
lorig
inle
galo
rigin
lega
lorig
inle
galo
rigin
perc
apita
[N]
Inte
rfere
nce
with
the
priv
ate
sect
or
Pro
perty
right
sin
dex
−0.8
259b
−1.3
861a
−0.7
037a
1.08
39a
0.87
94a
——
3.98
13a
0.31
64(0
.299
5)(0
.380
0)(0
.200
5)(0
.220
6)(0
.285
3)(0
.234
6)[1
24]
0.33
17−1
.113
9a−0
.559
2a0.
0684
−0.5
310b
1.63
82a
0.41
72a
0.19
070.
6311
(0.2
630)
(0.2
618)
(0.1
455)
(0.2
481)
(0.2
619)
(0.4
957)
(0.0
720)
(0.5
028)
[124
]
Bus
ines
sre
gula
tion
inde
x−1
.099
5a−1
.050
8a−0
.466
8b−0
.090
7−0
.096
6—
—3.
3774
a0.
1974
(0.2
747)
(0.2
988)
(0.1
860)
(0.2
685)
(0.2
680)
(0.2
216)
[124
]
−0.2
452
−0.6
104b
−0.3
457b
−0.8
799a
−0.9
913a
0.17
760.
4382
a−0
.104
90.
5216
(0.2
483)
(0.2
365)
(0.1
329)
(0.2
646)
(0.2
471)
(0.5
372)
(0.0
767)
(0.5
005)
[124
]
Top
mar
gina
ltax
rate
in4.
8702
13.1
457b
2.13
2710
.819
019
.645
6a—
—36
.153
8a0.
1460
1994
(5.4
254)
(5.1
124)
(3.0
166)
(5.6
843)
(5.2
545)
(3.0
844)
[79]
17.0
138b
11.5
953c
4.48
686.
6684
10.4
847c
20.8
175b
2.39
567.
4816
0.27
58(6
.543
9)(6
.188
4)(2
.896
9)(6
.086
7)(5
.897
6)(9
.287
5)(1
.675
4)(1
3.21
7)[7
9]
Effic
ienc
y
Cor
rupt
ion
−2.1
657a
−0.0
716
−0.7
552
1.65
25c
3.55
06a
——
6.60
85a
0.29
83(0
.762
2)(0
.672
9)(0
.512
5)(0
.893
2)(0
.587
9)(0
.630
6)[1
14]
Con
tinue
d
The Quality of Government 247
Tabl
e4.
Con
tinue
d
Dep
ende
ntV
aria
bles
Inde
pend
entV
aria
bles
Eth
nolin
guis
ticS
ocia
list
Fren
chG
erm
anS
cand
inav
ian
Latit
ude
Log
GN
PIn
terc
ept
Adj
.R
2
fract
iona
lizat
ion
lega
lorig
inle
galo
rigin
lega
lorig
inle
galo
rigin
perc
apita
[N]
0.64
660.
2337
−0.4
105
0.00
910.
7869
c4.
0864
a0.
7953
a−1
.248
50.
5764
(0.8
108)
(0.6
066)
(0.3
873)
(0.5
378)
(0.4
840)
(1.3
598)
(0.1
910)
(1.4
530)
[114
]
Bur
eauc
ratic
dela
ys−1
.977
4a−2
.127
2a−1
.494
2a0.
5035
0.36
17—
—6.
1036
a0.
5027
(0.6
289)
(0.6
034)
(0.3
956)
(0.5
862)
(0.4
016)
(0.3
861)
[54]
0.18
33−0
.489
8−0
.847
8a0.
1754
−0.0
665
−0.1
704
0.76
75a−0
.837
60.
7435
(0.6
424)
(0.5
783)
(0.2
774)
(0.3
257)
(0.2
724)
(0.9
509)
(0.1
722)
(1.3
409)
[54]
Tax
com
plia
nce
−1.3
688b
−2.0
663a
−1.6
107a
−0.5
537
−0.9
029a
——
4.44
35a
0.49
95(0
.601
4)(0
.290
7)(0
.256
6)(0
.335
9)(0
.298
7)(0
.235
3)[4
8]
−0.7
822
−1.3
561b
−1.3
226a
−0.6
174b
−0.8
640b
−0.7
798
0.34
69b
1.57
820.
5718
(0.6
782)
(0.5
349)
(0.2
738)
(0.2
744)
(0.3
392)
(0.9
042)
(0.1
708)
(1.4
022)
[48]
Avg
.go
vern
men
twag
es/
0.49
62−0
.550
3c1.
0939
a−0
.359
5−0
.470
7c—
—1.
5625
a0.
4272
GN
Ppe
rcap
ita(0
.900
3)(0
.284
4)(0
.288
1)(0
.226
7)(0
.255
4)(0
.269
7)[4
7]
0.16
54−1
.691
3a0.
5171
0.09
39−0
.452
61.
9247
b−0
.642
5a6.
4064
a0.
6041
(0.6
898)
(0.4
749)
(0.3
128)
(0.3
640)
(0.3
705)
(0.8
489)
(0.1
502)
(1.1
449)
[47]
Out
puto
fpub
licgo
ods
Log
ofin
fant
mor
talit
y1.
4426
a0.
4598
c0.
3816
a−0
.852
1a−1
.247
7a—
—3.
2093
a0.
4299
(0.1
816)
(0.2
509)
(0.1
313)
(0.2
688)
(0.1
399)
(0.1
393)
[151
]
Con
tinue
d
248 The Journal of Law, Economics, & Organization, V15 N1
Tabl
e4.
Con
tinue
d
Dep
ende
ntV
aria
bles
Inde
pend
entV
aria
bles
Eth
nolin
guis
ticS
ocia
list
Fren
chG
erm
anS
cand
inav
ian
Latit
ude
Log
GN
PIn
terc
ept
Adj
.R
2
fract
iona
lizat
ion
lega
lorig
inle
galo
rigin
lega
lorig
inle
galo
rigin
perc
apita
[N]
0.44
52a
−0.1
008
0.29
04a
0.05
06−0
.232
5b−0
.145
1−0
.523
3a7.
3733
a0.
8385
(0.1
247)
(0.1
630)
(0.0
726)
(0.1
336)
(0.1
153)
(0.2
973)
(0.0
401)
(0.2
769)
[151
]
Log
ofsc
hool
atta
inm
ent
−0.9
514a
−0.4
098a−0
.276
3a0.
1962
0.39
62a
——
1.83
09a
0.40
04(0
.159
6)(0
.113
8)(0
.105
7)(0
.146
6)(0
.114
1)(0
.113
6)[1
02]
−0.2
384
0.54
94a−0
.188
3a−0
.231
7a−0
.087
6−0
.305
40.
3479
a−0
.860
4a0.
7556
(0.1
185)
(0.0
831)
(0.0
686)
(0.0
844)
(0.0
977)
(0.2
547)
(0.0
352)
(0.2
514)
[102
]
Illite
racy
rate
32.0
410a
9.91
418.
1702
b−1
1.55
6a—
——
14.4
063
0.21
54(6
.027
4)(8
.298
3)(4
.174
4)(4
.292
1)(4
.292
1)[1
16]
16.4
024a
−12.
947
5.10
94c
−7.1
497b
—28
.496
7b−1
4.28
9a11
3.08
6a0.
5945
(5.0
588)
(8.7
744)
(2.9
942)
(3.2
073)
(11.
410)
(1.7
650)
(12.
882)
[116
]
Infra
stru
ctur
equ
ality
inde
x−2
.472
2a−3
.553
4a−1
.713
9a0.
8913
0.60
43—
—7.
2514
a0.
5314
(0.6
830)
(0.9
442)
(0.4
666)
(0.7
314)
(0.4
535)
(0.4
666)
[54]
0.73
78−1
.534
5c−0
.772
6a0.
4386
−0.3
074
1.34
200.
9478
a2.
0253
0.82
40(0
.558
1)(0
.802
6)(0
.278
7)(0
.388
2)(0
.299
4)(0
.902
2)(0
.177
6)(1
.352
5)[5
4]
Size
ofpu
blic
sect
or
Tran
sfer
san
dsu
bsid
ies
as%−1
1.06
9a9.
5935
a−1
.740
80.
7947
8.43
80a
——
12.3
694a
0.41
35of
GD
P(1
.943
3)(2
.815
0)(1
.408
6)(3
.346
2)(3
.062
6)(1
.584
7)[8
8]
Con
tinue
d
The Quality of Government 249
Tabl
e4.
Con
tinue
d
Dep
ende
ntV
aria
bles
Inde
pend
entV
aria
bles
Eth
nolin
guis
ticS
ocia
list
Fren
chG
erm
anS
cand
inav
ian
Latit
ude
Log
GN
PIn
terc
ept
Adj
.R
2
fract
iona
lizat
ion
lega
lorig
inle
galo
rigin
lega
lorig
inle
galo
rigin
perc
apita
[N]
−1.0
853
6.79
89b
0.52
68−3
.727
1−1
.623
022
.218
3a1.
8792
a−1
1.53
8a0.
7218
(1.7
668)
(2.7
920)
(1.0
485)
(2.3
284)
(3.3
284)
(3.9
537)
(0.5
528)
(4.2
783)
[88]
Gov
ernm
entc
onsu
mpt
ion
as−2
.034
4−5
.735
0a−1
.678
8−1
.549
95.
9173
a—
—16
.340
2a0.
1818
%of
GD
P(1
.462
0)(1
.345
9)(1
.119
8)(1
.893
6)(1
.987
4)(1
.228
5)[1
00]
0.66
17−6
.769
9a−1
.205
2−3
.304
9c2.
6416
5.91
78c
0.65
559.
0833
a0.
2594
(1.5
967)
(1.3
787)
(1.0
411)
(1.9
298)
(2.2
078)
(3.2
642)
(0.5
206)
(3.4
203)
[100
]
SO
Es
inth
eec
onom
yin
dex−1
.765
3b−4
.963
5a−0
.120
90.
5434
−1.1
847
——
5.31
44a
0.22
13(0
.704
2)(0
.474
6)(0
.477
4)(1
.026
2)(0
.760
3)(0
.503
1)[1
00]
−1.4
844b
−4.4
589a
−0.0
683
0.12
12−1
.210
5−2
.180
00.
4492
b2.
5190
c0.
2575
(0.7
489)
(0.5
099)
(0.4
662)
(1.0
293)
(0.8
639)
(1.7
007)
(0.2
102)
(1.3
091)
[100
]
Pub
licse
ctor
empl
oym
ent/
−3.4
957a
−0.6
837
−1.1
089b
−0.5
941
8.86
44a
——
5.23
41a
0.58
77to
talp
opul
atio
n(0
.623
2)(0
.848
1)(0
.476
2)(0
.999
8)(1
.173
3)(0
.546
5)[1
01]
−0.9
478c
0.62
69−0
.672
3a−2
.596
6a6.
6636
a0.
1170
1.24
29a
−4.7
012a
0.76
44(0
.561
8)(0
.743
9)(0
.328
3)(0
.922
6)(1
.211
2)(1
.843
5)(0
.242
1)(1
.485
7)[1
01]
Polit
ical
freed
om
Dem
ocra
cyin
dex
−4.1
753a
−5.4
741a−2
.488
9a0.
8801
3.40
16a
——
6.90
52a
0.32
17(1
.076
9)(0
.794
2)(0
.745
3)(1
.763
9)(0
.801
9)(0
.851
0)[1
25]
Con
tinue
d
250 The Journal of Law, Economics, & Organization, V15 N1
Tabl
e4.
Con
tinue
d
Dep
ende
ntV
aria
bles
Inde
pend
entV
aria
bles
Eth
nolin
guis
ticS
ocia
list
Fren
chG
erm
anS
cand
inav
ian
Latit
ude
Log
GN
PIn
terc
ept
Adj
.R
2
fract
iona
lizat
ion
lega
lorig
inle
galo
rigin
lega
lorig
inle
galo
rigin
perc
apita
[N]
−0.4
720
−3.8
954a
−2.0
485a
−2.5
033b
−0.7
896
1.63
261.
8107
a−7
.619
2c0.
6235
(0.9
398)
(0.7
875)
(0.5
504)
(1.2
611)
(0.6
022)
(2.6
534)
(0.3
016)
(1.7
230)
[125
]
Polit
ical
right
sin
dex
−2.1
852a
−1.4
376c
−0.5
438
1.28
31a
1.47
13a
——
5.68
93a
0.16
96(0
.562
5)(0
.802
8)(0
.363
1)(0
.405
8)(0
.324
3)(0
.346
2)[1
52]
−0.5
865
−1.0
240
−0.3
322
−0.2
927
−0.6
364
2.13
000.
6545
a−0
.044
20.
3409
(0.5
799)
(0.6
698)
(0.3
262)
(0.3
348)
(0.4
028)
(1.3
905)
(0.1
962)
(1.2
936)
[152
]
Ord
inar
yle
asts
quar
esre
gres
sion
sof
acr
oss-
sect
ion
ofco
untri
esar
ound
the
wor
ld.
The
depe
nden
tvar
iabl
esar
ecl
assi
fied
into
five
diffe
rent
grou
ps:
(i)in
terfe
renc
ew
ithth
epr
ivat
ese
ctor
;(ii
)ef
ficie
ncy;
(iii)
outp
utof
publ
icgo
ods;
(iv)s
ize
ofpu
blic
sect
or;a
nd(v
)pol
itica
lfre
edom
.R
obus
tsta
ndar
der
rors
are
show
nin
pare
nthe
ses.
The
inde
pend
entv
aria
bles
are:
(1)a
nav
erag
eof
five
indi
cato
rsof
ethn
olin
guis
ticfra
ctio
naliz
atio
n;(2
)ase
tof“
lega
lorig
in”
dum
mie
s(S
ocia
list,
Fren
ch,G
erm
an,S
cand
inav
ian,
and
the
omitt
eddu
mm
ybe
ing
Eng
lish)
;(3)
latit
ude;
and
(4)l
ogG
NP
per
capi
ta.
a Sig
nific
anta
t1%
leve
l.b
Sig
nific
anta
t5%
leve
l.c
Sig
nific
anta
t10%
leve
l.
The Quality of Government 251
Tabl
e5.
Gov
ernm
entP
erfo
rman
cean
dR
elig
ion
Dep
ende
ntV
aria
bles
Inde
pend
entV
aria
bles
Eth
nolin
guis
ticC
atho
licM
uslim
Oth
erLa
titud
eLo
gG
NP
Inte
rcep
tA
dj.
R2
fract
iona
lizat
ion
deno
min
atio
nspe
rcap
ita[N
]
Inte
rfere
nce
with
the
priv
ate
sect
orP
rope
rtyrig
hts
inde
x−0
.678
2b−0
.013
6a−0
.019
1a−0
.014
6a—
—4.
8511
a0.
1751
(0.2
837)
(0.0
033)
(0.0
036)
(0.0
044)
(0.2
669)
[124
]
0.59
77b
−0.0
018−0
.002
8−0
.000
81.
0599
c0.
5065
a−0
.621
40.
5534
(0.2
715)
(0.0
023)
(0.0
039)
(0.0
031)
(0.5
927)
(0.0
905)
(0.6
470)
[124
]
Bus
ines
sre
gula
tion
inde
x−0
.842
5a−0
.005
1c−0
.008
0b−0
.003
0—
—3.
3945
a0.
1280
(0.2
397)
(0.0
026)
(0.0
032)
(0.0
038)
(0.2
259)
[124
]
−0.0
938
0.00
220.
0044
0.00
69b
−0.5
668
0.50
19a−1
.084
90.
4832
(0.2
499)
(0.0
028)
(0.0
037)
(0.0
031)
(0.5
431)
(0.0
811)
(0.6
024)
[124
]
Top
mar
gina
ltax
rate
in−4
.195
0−0
.174
1a−0
.073
3−0
.121
0—
—54
.613
2a0.
1123
1994
(5.3
031)
(0.0
621)
(0.0
614)
(0.0
740)
(5.4
033)
[79]
10.3
710
−0.0
528
0.04
01−0
.028
721
.608
0b2.
3691
16.0
797
0.25
24(6
.554
5)(0
.065
2)(0
.067
7)(0
.079
3)(9
.827
6)(1
.876
5)(1
5.30
3)[7
9]
Effic
ienc
yC
orru
ptio
n−2
.749
0a−0
.037
2a−0
.043
0a−0
.028
1a—
—9.
7124
a0.
2856
(0.6
294)
(0.0
086)
(0.0
084)
(0.0
078)
(0.7
729)
[114
]
Con
tinue
d
252 The Journal of Law, Economics, & Organization, V15 N1
Tabl
e5.
Con
tinue
d
Dep
ende
ntV
aria
bles
Inde
pend
entV
aria
bles
Eth
nolin
guis
ticC
atho
licM
uslim
Oth
erLa
titud
eLo
gG
NP
Inte
rcep
tA
dj.
R2
fract
iona
lizat
ion
deno
min
atio
nspe
rcap
ita[N
]
0.54
25−0
.008
8−0
.008
9−0
.000
24.
5201
a0.
7685
a−0
.795
30.
5785
(0.8
118)
(0.0
076)
(0.0
077)
(0.0
072)
(1.3
451)
(0.1
656)
(1.6
261)
[114
]
Bur
eauc
ratic
dela
ys−1
.398
3b−0
.027
5a−0
.036
8a−0
.019
5b—
—7.
3743
a0.
4174
(0.6
435)
(0.0
060)
(0.0
067)
(0.0
080)
(0.5
321)
[54]
0.46
64−0
.011
8b−0
.015
1c−0
.002
9−0
.179
10.
8158
a−0
.946
40.
7465
(0.5
602)
(0.0
048)
(0.0
076)
(0.0
054)
(0.9
667)
(0.1
608)
(1.3
895)
[54]
Tax
com
plia
nce
−0.6
224
−0.0
120−0
.008
1−0
.000
5—
—3.
8719
a0.
1934
(0.6
206)
(0.0
042)
(0.0
053)
(0.0
051)
(0.3
225)
[48]
0.07
04−0
.006
00.
0061
0.00
52−1
.366
9c0.
6488
a−1
.684
4c0.
5143
(0.5
118)
(0.0
046)
(0.0
068)
(0.0
049)
(0.7
438)
(0.0
962)
(0.9
590)
[48]
Avg
.go
vern
men
twag
es/
0.64
540.
0125
a0.
0220
a0.
0078
——
0.93
57a
0.30
44G
NP
perc
apita
(0.9
849)
(0.0
041)
(0.0
041)
(0.0
076)
(0.2
620)
[47]
0.56
920.
0039
0.01
02c−0
.001
1−0
.506
8−0
.292
34.
2686
a0.
3990
(0.8
712)
(0.0
042)
(0.0
057)
(0.0
096)
(0.9
240)
(0.1
828)
(1.5
396)
[47]
Out
puto
fpub
licgo
ods
Log
ofin
fant
mor
talit
y1.
4039
a0.
0083
a0.
0166
a0.
0082
a—
—2.
5138
a0.
4292
(0.1
656)
(0.0
027)
(0.0
024)
(0.0
033)
(0.2
244)
[151
]
Con
tinue
d
The Quality of Government 253
Tabl
e5.
Con
tinue
d
Dep
ende
ntV
aria
bles
Inde
pend
entV
aria
bles
Eth
nolin
guis
ticC
atho
licM
uslim
Oth
erLa
titud
eLo
gG
NP
Inte
rcep
tA
dj.
R2
fract
iona
lizat
ion
deno
min
atio
nspe
rcap
ita[N
]
0.40
47a
0.00
030.
0046
a−0
.001
6−0
.494
0b−0
.487
9a7.
2572
a0.
8506
(0.1
284)
(0.0
015)
(0.0
017)
(0.0
017)
(0.2
450)
(0.0
370)
(0.3
063)
[151
]
Log
ofsc
hool
atta
inm
ent
−0.8
640a
−0.0
043b−0
.011
2a−0
.003
1—
—2.
1424
a0.
4830
(0.1
414)
(0.0
017)
(0.0
017)
(0.0
022)
(0.1
458)
[102
]
−0.1
774
−0.0
008−0
.005
5a−0
.000
50.
1143
0.27
47a−0
.421
20.
7603
(0.1
255)
(0.0
018)
(0.0
017)
(0.0
020)
(0.2
684)
(0.0
391)
(0.3
175)
[102
]
Illite
racy
rate
27.7
599a
0.09
080.
3875
a0.
1598
——
3.65
100.
3860
(5.3
080)
(0.0
988)
(0.0
960)
(0.1
265)
(9.1
492)
[116
]
15.0
234a
0.03
130.
2490
a0.
0381
8.32
58−1
2.06
60a
96.5
462a
0.66
90(5
.131
6)(0
.067
4)(0
.071
8)(0
.090
6)(1
3.16
5)(1
.222
5)(1
0.48
8)[1
16]
Infra
stru
ctur
equ
ality
inde
x−1
.662
5b−0
.036
4a−0
.046
9a−0
.029
7b—
—9.
1259
a0.
3774
(0.7
231)
(0.0
081)
(0.0
085)
(0.0
113)
(0.6
915)
[54]
1.28
09b
−0.0
112c−0
.013
1−0
.003
00.
4773
1.17
32a−3
.384
9b0.
8059
(0.6
039)
(0.0
057)
(0.0
092)
(0.0
057)
(1.0
769)
(0.1
847)
(1.5
471)
[54]
Size
ofpu
blic
sect
or
Tran
sfer
san
dsu
bsid
ies
as%−1
2.66
3a−0
.086
0b−0
.109
0a−0
.072
9b—
—20
.169
4a0.
3414
ofG
DP
(1.9
090)
(0.0
331)
(0.0
281)
(0.0
347)
(2.7
997)
[88]
Con
tinue
d
254 The Journal of Law, Economics, & Organization, V15 N1
Tabl
e5.
Con
tinue
d
Dep
ende
ntV
aria
bles
Inde
pend
entV
aria
bles
Eth
nolin
guis
ticC
atho
licM
uslim
Oth
erLa
titud
eLo
gG
NP
Inte
rcep
tA
dj.
R2
fract
iona
lizat
ion
deno
min
atio
nspe
rcap
ita[N
]
−0.5
782
0.03
95−0
.000
30.
0333
28.2
789a
0.86
17c−8
.248
40.
6950
(2.0
731)
(0.0
292)
(0.0
298)
(0.0
319)
(3.7
928)
(0.4
963)
(4.9
815)
[88]
Gov
ernm
entc
onsu
mpt
ion
as−2
.242
3−0
.096
2a−0
.063
9a−0
.085
1a—
—22
.859
4a0.
1932
%of
GD
P(1
.377
9)(0
.018
1)(0
.019
3)(0
.026
4)(1
.555
6)[1
00]
0.10
52−0
.075
0a−0
.032
3−0
.068
3b−0
.096
81.
1052
b12
.162
0a0.
2564
(1.5
478)
(0.0
192)
(0.0
227)
(0.0
279)
(3.1
882)
(0.5
137)
(4.4
892)
[100
]
SO
Es
inth
eec
onom
yin
dex−0
.908
30.
0036
−0.0
061−0
.006
1—
—4.
7480
a0.
0396
(0.7
474)
(0.0
089)
(0.0
085)
(0.0
122)
(0.7
444)
[100
]
−1.2
360
−0.0
026−0
.002
3−0
.006
7−4
.947
4b0.
6518
b1.
7897
0.12
77(0
.807
4)(0
.010
9)(0
.010
1)(0
.012
9)(2
.008
9)(0
.250
4)(1
.920
1)[1
00]
Pub
licse
ctor
empl
oym
ent/
−4.0
507a
−0.0
764a−0
.074
0a−0
.076
1a—
—11
.622
9a0.
4694
tota
lpop
ulat
ion
(0.6
329)
(0.0
158)
(0.0
135)
(0.0
146)
(1.4
097)
[100
]
−1.3
935
−0.0
498a−0
.042
2a−0
.045
4a2.
3145
0.94
98a
0.80
200.
6332
(0.6
972)
(0.0
128)
(0.0
121)
(0.0
140)
(1.7
412)
(0.2
490)
(2.2
609)
[100
]
Polit
ical
freed
om
Dem
ocra
cyin
dex
−3.0
046a
−0.0
550a−0
.086
5a−0
.069
2a—
—10
.859
3a0.
3026
(0.9
543)
(0.0
112)
(0.0
111)
(0.0
129)
(0.8
900)
[125
]
Con
tinue
d
The Quality of Government 255
Tabl
e5.
Con
tinue
d
Dep
ende
ntV
aria
bles
Inde
pend
entV
aria
bles
Eth
nolin
guis
ticC
atho
licM
uslim
Oth
erLa
titud
eLo
gG
NP
Inte
rcep
tA
dj.
R2
fract
iona
lizat
ion
deno
min
atio
nspe
rcap
ita[N
]
0.55
12−0
.020
4c−0
.035
7a−0
.025
9c0.
8540
1.73
78a−6
.501
6a0.
5741
(0.9
931)
(0.0
120)
(0.0
133)
(0.0
134)
(2.5
435)
(0.3
968)
(2.2
928)
[125
]
Pub
licrig
hts
inde
x−1
.532
3a−0
.012
7b−0
.043
2a−0
.025
4a—
—7.
2391
a0.
3510
(0.5
202)
(0.0
050)
(0.0
054)
(0.0
069)
(0.4
178)
[152
]
−0.0
755
−0.0
006−0
.029
1a−0
.013
2b2.
6524
a0.
4192
a2.
0483
0.47
55(0
.551
6)(0
.005
2)(0
.006
6)(0
.006
7)(0
.941
2)(0
.154
8)(1
.298
0)[1
52]
Ord
inar
yle
ast
squa
res
regr
essi
ons
ofa
cros
s-se
ctio
nof
coun
tries
arou
ndth
ew
orld
.Th
ede
pend
ent
varia
bles
are
clas
sifie
din
tofiv
edi
ffere
ntgr
oups
:(i)
inte
rfere
nce
with
the
priv
ate
sect
or;(
ii)ef
ficie
ncy;
(iii)
outp
utof
publ
icgo
ods;
(iv)s
ize
ofth
epu
blic
sect
or;a
nd(v
)pol
itica
lfre
edom
.R
obus
tsta
ndar
der
rors
are
show
nin
pare
nthe
ses.
The
inde
pend
entv
aria
bles
are:
(1)
anav
erag
eof
five
indi
cato
rsof
ethn
o-lin
guis
ticfra
ctio
naliz
atio
n;(2
)R
oman
Cat
holic
popu
latio
n(a
sa
%of
tota
l);(3
)M
uslim
popu
latio
n(a
sa
%of
tota
l);(4
)oth
erde
nom
inat
ions
whi
chis
the
resi
dual
popu
latio
nw
hich
does
notr
epor
tto
prof
ess
Pro
test
ant,
Cat
holic
,orM
uslim
relig
ions
(as
a%
ofto
talp
opul
atio
n)(th
eom
itted
popu
latio
nis
the
Pro
test
antp
opul
atio
n);(
5)la
titud
e;(6
)log
GN
Ppe
rcap
ita.
a Sig
nific
anta
t1%
leve
l.b
Sig
nific
anta
t5%
leve
l.c
Sig
nific
anta
t10%
leve
l.
256The
JournalofLaw,E
conomics,&
Organization,V15
N1
Table 6. Government Performance, Legal Origin, and Religion
Dependent Variables Independent Variables
Ethnolinguistic Socialist French German Scandinavian Catholic Muslim Other Latitude Log GNP Intercept Adj. R2
fractionalization legal origin legal origin legal origin legal origin religions per capita [N]
Interference with the private sector
Property rights index −0.7013b −1.3894a −0.7900a 0.9918a 0.9416c 0.0041 −0.0037 −0.0003 — — 3.9053a 0.3557
(0.3041) (0.3932) (0.2730) (0.2052) (0.5101) (0.0066) (0.0068) (0.0060) (0.5176) [124]
0.3843 −1.2522a −0.6531a 0.0519 −0.1921 0.0062 0.0024 0.0049 1.7985a 0.3892a −0.0431 0.6408
(0.2661) (0.3048) (0.1870) (0.2383) (0.4238) (0.0044) (0.0050) (0.0044) (0.5801) (0.0929) (0.7176) [124]
Business regulation index −1.0988a −1.2348a −0.3404 −0.1500 −0.0095 −0.0004 −0.0034 0.0040 — — 3.2539a 0.2299
(0.2759) (0.3420) (0.2277) (0.2375) (0.4170) (0.0053) (0.0055) (0.0053) (0.4318) [124]
−0.2661 -0.8376a −0.2491 −0.9027a −0.6382 0.0030 0.0039 0.0084c 0.1859 0.4570a −0.7207 0.5479
(0.2479) (0.2812) (0.1633) (0.1810) (0.4003) (0.0043) (0.0046) (0.0043) (0.5908) (0.0914) (0.6776) [124]
Top marginal tax rate in 1.4378 14.9193a 6.5827 10.5544c 11.1248 −0.1657 −0.0586 −0.0791 — — 45.8541a 0.2039
1994 (5.7018) (4.6331) (4.4391) (5.8712) (9.6287) (0.1233) (0.1103) (0.1152) (10.121) [79]
13.9829b 13.9606b 7.6698c 6.4841 7.1961 −0.0962 0.0159 −0.0215 16.8895 2.9196 8.9102 0.3229
(6.5626) (6.0011) (4.1466) (5.9912) (7.6966) (0.0908) (0.0918) (0.1023) (10.964) (1.8765) (15.468) [79]
Efficiency
Corruption −2.2008a −0.1559 −0.5023 1.4193 2.6055 −0.0115 −0.0204 −0.0087 — — 7.6444a 0.3237
(0.8273) (0.6641) (0.5516) (0.8955) (1.6767) (0.0203) (0.0189) (0.0180) (1.8791) [114]
Continued
TheQ
ualityofG
overnment
257
Table 6. Continued
Dependent Variables Independent Variables
Ethnolinguistic Socialist French German Scandinavian Catholic Muslim Other Latitude Log GNP Intercept Adj. R2
fractionalization legal origin legal origin legal origin legal origin religions per capita [N]
0.5909 −0.0732 −0.2364 −0.0929 0.9132 0.0001 −0.0014 0.0067 4.3158a 0.7725a −1.3701 0.5818
(0.8821) (0.5916) (0.4404) (0.5779) (1.1987) (0.0149) (0.0143) (0.0134) (1.4094) (0.1984) (2.1381) [114]
Bureaucratic delays −2.0774a −2.0465a −1.0801a 0.1384 −2.3447a −0.0350a −0.0442a −0.0322a — — 9.1862a 0.6430
(0.6394) (0.5910) (0.4012) (0.4674) (0.8156) (0.0107) (0.0105) (0.0106) (0.8196) [54]
−0.2225 −0.7470 −0.5488c −0.0030 −1.6180b −0.0224b −0.0264b −0.0165c 0.1007 0.6362a 2.0056 0.7885
(0.6107) (0.6091) (0.3206) (0.3308) (0.6397) (0.0090) (0.0107) (0.0089) (1.1806) (0.2220) (2.0288) [54]
Tax compliance −1.4229b −1.9495a −1.4780a −0.5448c −1.4599b −0.0090 −0.0038 0.0071 — — 5.0747a 0.5201
(0.5966) (0.3106) (0.3036) (0.3203) (0.5636) (0.0079) (0.0086) (0.0075) (0.6132) [48]
−0.6969 −0.9764b −1.0886a −0.5867b −1.0331c −0.0056 0.0054 −0.0013 −1.0134 0.4711a 0.7286 0.6283
(0.6030) (0.4741) (0.2587) (0.2714) (0.5601) (0.0070) (0.0087) (0.0074) (0.7531) (0.1430) (1.5921) [48]
Avg. government wages / 0.9124 −0.7845a 1.2156a −0.0643 −0.3760 −0.0037 0.0045 0.0058 — — 1.3803c 0.5010
GNP per capita (0.8598) (0.2731) (0.3086) (0.2873) (0.6509) (0.0074) (0.0077) (0.0089) (0.7166) [47]
0.5150 −1.8937a 1.1686a 0.4324 −1.3567c −0.0200b −0.0204c −0.0091 2.4471b −0.8031a 8.5803a 0.6672
(0.7203) (0.5136) (0.3380) (0.3049) (0.7655) (0.0087) (0.0101) (0.0103) (0.9250) (0.1872) (1.8527) [47]
Output of public goods
Log of infant mortality 1.2610a 0.5676b 0.4244a −0.7441a −1.4558a −0.0056 0.0045 −0.0036 — — 3.4664a 0.5255
(0.1805) (0.2751) (0.1571) (0.2692) (0.2510) (0.0038) (0.0033) (0.0038) (0.2827) [151]
Continued
258The
JournalofLaw,E
conomics,&
Organization,V15
N1
Table 6. Continued
Dependent Variables Independent Variables
Ethnolinguistic Socialist French German Scandinavian Catholic Muslim Other Latitude Log GNP Intercept Adj. R2
fractionalization legal origin legal origin legal origin legal origin religions per capita [N]
0.4020a 0.1408 0.2609a 0.0983 −0.4970b −0.0053b 0.0000 −0.0053b −0.4252 −0.4754a 7.4267a 0.8637
(0.1300) (0.1727) (0.0794) (0.1268) (0.2120) (0.0025) (0.0024) (0.0025) (0.2836) (0.0415) (0.3683) [151]
Log of school attainment −0.7840a 0.3276a −0.2593b 0.1521 0.3324 0.0004 −0.0069b −0.0008 — — 1.8889a 0.5322
(0.1499) (0.1058) (0.1065) (0.1416) (0.2476) (0.0032) (0.0031) (0.0034) (0.2801) [102]
−0.1821 0.4050a −0.1914a −0.2251b −0.1966 −0.0000 −0.0045 0.0011 0.0249 0.2942a −0.4628 0.7974
(0.1199) (0.0964) (0.0643) (0.0916) (0.1932) (0.0025) (0.0023) (0.0028) (0.2611) (0.0389) (0.3354) [102]
Illiteracy rate 27.0753a 7.7474 8.9148c −12.881b — −0.0490 0.2737a 0.0884 — — 8.5013 0.4226
(5.4717) (10.479) (4.9304) (6.1284) (0.1245) (0.1229) (0.1544) (9.7099) [116]
12.5505b −10.404 6.8149c −5.0399 — −0.0497 0.1873b 0.0537 5.7156 −12.2460a 99.6097a 0.6878
(5.0585) (9.8843) (3.5124) (4.6052) (0.0874) (0.0867) (0.1077) (12.903) (1.3433) (10.965) [116]
Infrastructure quality index −2.6254a −3.3972a −1.1956a 0.4224 −3.1489a −0.0482a −0.0587a −0.0458a — — 11.5389a 0.6711
(0.6199) (0.7994) (0.4079) (0.6621) (1.1454) (0.0131) (0.0126) (0.0127) (1.1681) [54]
0.2361 −1.8375b −0.3693 0.2153 −2.2470a −0.0284a −0.0328a −0.0204b 1.6779 0.7915a 1.4527 0.8647
(0.5079) (0.7730) (0.2823) (0.4692) (0.7525) (0.0091) (0.0109) (0.0087) (1.0857) (0.2180) (1.9191) [54]
Size of public sector
Transfers and subsidies as % −10.429a 10.0018a −1.6490 0.7443 5.8750 −0.0259 −0.0514 −0.0422 — — 15.2889b 0.4266
of GDP (2.2704) (2.7618) (2.1624) (3.1120) (6.5351) (0.0782) (0.0697) (0.0682) (6.4337) [88]
Continued
TheQ
ualityofG
overnment
259
Table 6. Continued
Dependent Variables Independent Variables
Ethnolinguistic Socialist French German Scandinavian Catholic Muslim Other Latitude Log GNP Intercept Adj. R2
fractionalization legal origin legal origin legal origin legal origin religions per capita [N]
−0.3402 6.1352b 0.3590 −3.6369 −0.5825 0.0218 −0.0087 0.0157 24.1427a 1.5940a −11.3810b 0.7297
(2.0479) (3.0242) (1.2501) (2.3440) (4.3890) (0.0415) (0.0393) (0.0394) (3.9456) (0.5324) (5.3205) [88]
Government consumption as −3.0761c −5.0546a −0.7166 −1.7008 0.2411 −0.0843b −0.0558c −0.0714c — — 22.7980a 0.2316
% of GDP (1.5586) (1.5113) (1.1828) (1.9574) (3.2478) (0.0332) (0.0303) (0.0383) (2.6229) [100]
−0.4569 −5.4747a −0.4151 −3.4240c −1.4461 −0.0706b −0.0339 −0.0576 3.1099 0.9854c 12.5943b 0.3074
(1.5940) (1.4245) (1.0868) (2.0800) (3.0986) (0.0312) (0.0315) (0.0370) (3.4085) (0.5706) (5.3128) [100]
SOEs in the economy index −1.7574b −5.0152a 0.0354 0.3548 −2.0452 −0.0098 −0.0189 −0.0091 — — 6.2642a 0.2450
(0.7749) (0.5294) (0.5594) (1.0483) (1.4665) (0.0180) (0.0161) (0.0177) (1.5382) [100]
−1.5762c −4.5385a −0.0379 0.0848 −1.9666 −0.0094 −0.0137 −0.0089 −1.8264 0.3692 3.8933 0.2648
(0.8258) (0.6178) (0.5697) (1.0463) (1.3848) (0.0184) (0.0165) (0.0181) (1.9240) (0.2416) (2.2718) [100]
Public sector employment / −3.3400a −0.3848 −1.1109b −0.6880 8.0477a −0.0075 −0.0149 −0.0147 — — 6.2029a 0.5978
total population (0.6646) (0.9318) (0.4833) (0.9384) (1.8996) (0.0167) (0.0143) (0.0159) (1.4515) [100]
−0.9215 0.6805 −0.7417c −2.5817b 6.9815a 0.0042 0.0068 0.0038 −0.1045 1.2912a −5.3766b 0.7657
(0.5762) (0.8417) (0.3940) (1.6003) (1.6004) (0.0124) (0.0117) (0.0141) (2.0524) (0.2835) (2.3324) [100]
Political freedom
Democracy index −3.5236a −4.9840a −2.5819a 0.4810 1.5197 −0.0081 −0.0464 −0.0302 — — 9.0226a 0.4262
(1.1404) (0.8777) (0.8671) (1.5157) (2.0742) (0.0265) (0.0247) (0.0250) (2.3289) [125]
Continued
260The
JournalofLaw,E
conomics,&
Organization,V15
N1
Table 6. Continued
Dependent Variables Independent Variables
Ethnolinguistic Socialist French German Scandinavian Catholic Muslim Other Latitude Log GNP Intercept Adj. R2
fractionalization legal origin legal origin legal origin legal origin religions per capita [N]
−0.2370 −4.2093a −2.1704a −2.4051b −1.2381 0.0031 −0.0203 −0.0085 3.2446 1.4940a −5.1883c 0.6511
(1.0095) (0.9607) (0.6412) (1.1647) (1.6795) (0.0217) (0.0212) (0.0216) (2.6246) (0.3303) (2.9477) [125]
Political rights index −1.5088a −1.0721 −0.7744c 0.9945a 0.6585 −0.0006 −0.0327a −0.0154 — — 6.5948a 0.3792
(0.5672) (0.7895) (0.4081) (0.3667) (0.6596) (0.0092) (0.0090) (0.0096) (0.7466) [152]
−0.2595 −1.3386c −0.7521c −0.4058 −1.5218b −0.0003 −0.0305a −0.0137 3.9678a 0.3138c 2.9411b 0.5005
(0.5705) (0.6793) (0.3711) (0.4524) (0.6528) (0.0084) (0.0086) (0.0087) (1.0809) (0.1631) (1.4209) [152]
Ordinary least squares regressions of a cross-section of countries around the world. The dependent variables are classified into five different groups: (i) interference with the private sector;(ii) public sector efficiency; (iii) output of public goods; (iv) size of public sector; and (v) political freedom. Robust standard errors are shown in parentheses. The independent variables are:(1) an average of 5 indicators of ethno-linguistic fractionalization; (2) a set of “legal origin” dummies (French, Socialist, German, Scandinavian and the omitted dummy being English); (3) RomanCatholic population (as a % of total); (4) Muslim population (as a % of total); (5) residual population (as a % of total); (5) other denominations which is the residual population which does notreport to profess Protestant, Catholic, or Muslim religions (as a % of total population) (the omitted population is the Protestant population); (6) latitude; (7) log GNP per capita.aSignificant at 1% level.bSignificant at 5% level.cSignificant at 10% level.
The Quality of Government 261
theory would suggest. Alternatively these results may mean that people in highEF countries are less productive, perhaps because of private rent seeking andwarfare, and that low income in turn adversely influences government. Becausethe first interpretation seems to us to be more straightforward, we interpret theEF evidence as supporting the political theories of institutions.13
4.2 Legal OriginThe regressions in Table 4 also assess the influence of legal origin on governmentperformance. To begin, the results of the effects of the socialist legal originare exactly as predicted by the political theory. Compared to those in commonlaw countries, and even controlling for per capita income, latitude, and EF,government in countries of socialist legal origin are more interventionist acrossthe board (have worse protection of property rights, more intrusive regulation,and higher tax rates) as well as being less efficient (lower score on bureaucraticdelays and tax compliance, though not on corruption). Of interest, they havelower relative wages of public sector employees. Socialist origin countrieshave higher infant mortality, though not controlling for how poor they are, andpoorer infrastructure quality, even controlling for per capita income. However,they do very well on school attainment, especially controlling for how poor theyare—some evidence that their egalitarianism or militarism benefits education.Turning to the size of government, socialist origin countries have sharply highertransfers and subsidies and more state enterprises, but also lower governmentconsumption, than common law countries ceteris paribus. In general, then, wesee that socialist law is associated with more interventionism, less efficiency,bigger government transfers, and less democracy—consistent with the obviouspolitical story that socialist policies serve to enhance the power of the State.
The results on the French legal origin are equally striking. Compared tocommon law countries, French origin countries are sharply more interventionist(have higher top rates, less secure property rights, and worse regulation). Theyalso have less efficient governments, as measured by bureaucratic delays andtax compliance, though not the corruption score. French origin countries payrelatively higher wages to bureaucrats than common law countries do, thoughthis does not buy them greater government efficiency. French origin countriesfall behind common law countries in public good provision: they have higherinfant mortality, lower school attainment, higher illiteracy rates, and lowerinfrastructure quality. There is not much evidence of a difference in the patternof government spending, except that French origin countries have lower publicemployment. Finally, French origin countries score worse on our democracymeasures than the common law countries. Most of these results hold bothwith and without per capita income and latitude controls. As predicted bythe political theory then, the state-building intent incorporated into the designof the French legal system translates, many decades later, into significantly
13. EF may reduce productivity because it undermines human capital accumulation. To theextent that EF does that through undermining public education, however, the direct effect is on thegovernment institution of public education rather than on productivity.
262 The Journal of Law, Economics, & Organization, V15 N1
more interventionist and less efficient government, less political freedom, andevidently less provision of basic public goods.
The results on the German legal origin suggest that these countries are rathersimilar to those of common law origin, controlling for income and latitude.These controls are particularly important because German origin countries aregenerally located in central Europe and East Asia and tend to be relatively rich.Finally, compared to common law countries, Scandinavian origin countries aresharply more interventionist (in fact, the pattern of coefficients here is similarto that for socialist countries), though—with a notable exception of lower taxcompliance—they do not have less efficient governments. Scandinavian origincountries also tend to have better public goods and higher public spending thancommon law countries. Controlling for per capita income and latitude reducesthe difference, although in the area of public employment, the Scandinavianorigin countries are out on a limb, even controlling for income. These resultspoint to important differences between civil law countries of different origins.In particular, the German and Scandinavian evidence—while consistent withthe political theory of institutions—is not nearly as striking as that for countriesusing French law.
The contrast between the socialist, common law, and French origins, how-ever, is very significant and persistent across the variety of measures. Gov-ernments in the socialist law countries are the most antimarket, and those inthe common law countries the most promarket, with governments in Frenchorigin countries in the middle. The fact that this political heritage matters somuch for government performance supports the political theory of institutions,especially when combined with the evidence on the importance of EF.
4.3 Religious AffiliationTable 5 reports the effects of religious affiliations on government performance.Recall that we exclude from the regression the percentage of each country’spopulation reporting to be Protestants. Note first that the results on EF inTable 5 are similar to those in Table 4: EF has a sharp negative influence ongovernment performance, but this influence is captured by the lower per capitaincome (and closeness to the equator) of high EF countries.
Table 5 reveals that both Catholic and Muslim affiliations of the populationare associated with worse government performance, though, as in the case withEF, both of these influences generally become insignificant once per capitaincome and latitude are controlled for.
Begin with the Catholic affiliation. With the notable exception of the marginaltax rate, Catholic countries are generally more interventionist. Their govern-ments are also less efficient, including more corrupt, but better paid. Catholiccountries do worse on public good provision than Protestant countries, they havesmaller transfers, government consumption, and public sector employment, andare significantly less democratic. Virtually all of these effects, however, becomeinsignificant once we control for per capita income and latitude. As with EF,there are two stories. The one we find less plausible suggests that Catholics areless productive, and this translates into inferior government. The more plausible
The Quality of Government 263
argument is that the worse-functioning governments of the Catholic countriesreduce income, with the result that the adverse effect of the religious affiliationon the quality of government is in part captured by per capita income.14
The effect of the Muslim affiliation is generally in the same direction asthat of the Catholic affiliation, only stronger. Muslim countries tend to havemore interventionist and less efficient, but better paid, governments, althoughthese effects are generally captured by their lower per capita income. Thereis a remarkably large negative effect of the Muslim affiliation on public goodprovision, which appears to be strong enough to survive the income and latitudecontrols.15 Muslim countries also tend to have smaller governments (thoughless so with controls), and be sharply less democratic, even controlling for theirpoverty. Muslim affiliation thus has a pronounced adverse effect on governmentperformance. If we take the view that the amelioration of this effect whenper capita income is included is a consequence of the adverse effects of badgovernment on income, the problem to be explained only gets bigger.
Finally, we note two side issues that come up in these regressions. First,the results on relative government wages show that both Muslim and Catholiccountries pay higher relative government wages than Protestant countries do,but their bureaucracies are still less efficient. The story with government wagesand corruption is thus more complicated than the World Bank, inspired by theEast Asian experience, would argue (World Bank 1995). In many Catholic andMuslim countries, officials are paid relatively well, but still misuse their power.There are more basic factors determining government efficiency than just therelative pay of the bureaucrats.16
Second, compared to Protestant countries, both Catholic and Muslim coun-tries exhibit both poor government performance and low tax rates combinedwith small government transfers and consumption. This evidence, as well asall the other evidence we have presented, seems squarely inconsistent with theidea that a good government is a small government. Rather, poorly functioninggovernments tend to be relatively small, and collect fewer taxes, whereas well-functioning governments tend to be much larger, presumably at least in partwith the consent of the governed. This view is consistent with the finding thatthe one activity that poorly functioning governments do more of is operate statefirms—a redistributive rather than wealth-creating function. This view is alsoconsistent with historical research (Brewer, 1988) and with the recent evidencefrom the postcommunist transition (Johnson, Kaufman, and Shleifer, 1997).
14. As with ethnic fractionalization, particular religious affiliations may adversely affect humancapital accumulation, which in turn reduces income. As with EF, one way in which this happensis that religion undermines public education, so the causation is through government.
15. Since we are looking at the outcome of public good provision, as opposed to just governmentexpenditure on it, our evidence is inconsistent with the theory that Catholic or Muslim religionssubstitute for the lack of public provision of education through their own provision.
16. An alternative interpretation of these findings is that in Catholic and Muslim countries generaleducational attainment is low, and hence the bureaucrats are relatively overpaid because they arebetter educated relative to the rest of the population than are the Protestant bureaucrats.
264 The Journal of Law, Economics, & Organization, V15 N1
4.4 Law or Religion?The evidence in Table 5 does not tell us whether the negative influences ofCatholic and Muslim affiliations on government performance are a consequenceof troubled political history or of culture. We have made a number of a prioriarguments suggesting that cultural theories can work through politics, and that apolitical interpretation of the religious variables may be more appropriate. Thatis, the use of religion for political purposes in Muslim and Catholic countries,and the destructive competition between Church and State in Catholic countriesin particular, may have shaped policies in ways that ended up being quite hostileto market development. It is possible that such politics worked by adverselyaffecting tolerance, as in Landes (1998), or trust, as in Putnam (1993) andLa Porta et al. (1997b), but the heritage that adversely affects governmentperformance seems to be largely political.
However, as we noted earlier, legal origins and religious affiliations are cor-related with each other. To understand the facts better, as well as to makefurther progress in interpreting the evidence, we would like to know which setof variables, if any, wins out in a horse race. As a final step therefore, we runregressions that include EF, legal origins, and religious affiliations in the samespecification. Table 6 presents the results.
We can summarize the findings without discussing each regression individ-ually. EF survives as an important determinant of government performance.Moreover, as a general rule, legal origins—particularly socialist and French—continue to exert significant adverse influence on government performance,even controlling for religion. Religious variables, however, generally becomeinsignificant. Catholic affiliation loses its statistical significance almost always(exceptions are bureaucratic delays, infrastructure quality, and government con-sumption), as does the Muslim affiliation (exceptions are bureaucratic delays,school attainment, illiteracy, and infrastructure quality). Statist laws are thusa more robust predictor of poor government performance than interventionistreligions.
If we take religion as a proxy for cultural influences on government, andEF and law as proxies for political influences, this evidence provides furthersupport for the political theories. In our data, these political variables providethe most pervasive, and relatively clearly interpretable, influence on governmentperformance.
4.5 Other FactorsThere are two variables—colonial status and continent—that we have not in-cluded in our statistical analysis which arguably qualify as potentially exoge-nous determinants of government performance, and which are often used in thecross-sectional studies of economic growth. For completeness, we make somecomments about these variables.
Several writers measure whether a country is a former colony, and, if so,what country it was a colony of (Barro, 1996b; Treisman, 1997; Hall andJones, 1999). The usual justification for using this variable is that the colonizerhas transplanted into the colony some of its key institutions, such as religion
The Quality of Government 265
and law, which in turn influenced subsequent development. We already includereligion and law as independent variables, and hence measure these possiblecolonial influences directly rather than indirectly. As important, we have foundit difficult to identify the colonial status of particular countries. For example,Barro (1996b) does not classify Hungary, Czechoslovakia, or Italy as formercolonies, even though each was at times (at least partially) controlled by the Hab-sburg empire, and the former two were arguably colonized by the Soviet Unionas well. Because of such difficulties, we stick to our more direct measures.
From the theoretical viewpoint, the justification for using continents as deter-minants of government performance is also problematic. Various writers findslow growth in Latin America and Africa (Barro, 1991; Easterly and Levine,1997), but the significance of the continent effects is attributed to the omittedinstitutional factors, such as religion or, as Easterly and Levine (1997) arguedirectly for Africa, EF. Since we measure the political and cultural influenceson performance directly, the case for including the continent effects is weak.We also include latitude in the regressions, which is the one theoretically jus-tified measure of geography. Still, we have rerun our regressions controllingfor continents. These controls have predictable effects. The Africa dummyin particular weakens but does not eliminate the effect of EF, and the LatinAmerica dummy weakens but does not eliminate the effects of French laws orCatholicism. Again, we take this evidence to mean not that the continents havean independent influence on government performance, but rather that they serveas proxies for the more fundamental determinants of the quality of governmentdiscussed in this article.
5. ConclusionWe have set out to examine whether, from the perspective of promoting devel-opment, the quality of government across countries varies in systematic ways.We assess the quality of government using proxies for interventionism, publicsector efficiency, quality of public good provision, government size, and politi-cal freedom. The data show clearly that, using these measures of performance,the quality of governments varies systematically across countries. Rich nationshave better governments than poor ones. Ethnolinguistically homogeneouscountries have better governments than the heterogeneous ones. Common lawcountries have better governments than French civil law or socialist law coun-tries. Predominantly Protestant countries have better governments than eitherpredominantly Catholic or predominantly Muslim countries. These results tendto be consistent across the many measures of government performance we used.
These results present clear evidence of systematic influence of historicalcircumstances, as captured by ethnolinguistic heterogeneity, legal origins, andreligion, on government performance. Government performance is surely inpart determined by economic development, but it is also shaped by systematicvariation in the histories of individual countries. This conclusion, we believe,is strongly supported by the data.
The results are more difficult to interpret as supporting a particular version ofpolitical or cultural theories of institutions. Taken as a whole, however, the re-
266 The Journal of Law, Economics, & Organization, V15 N1
sults support the political theories, since ethnolinguistic heterogeneity and legalorigin remain extremely important factors shaping government performance,even controlling for religion. In addition to showing that history matters, webelieve that we have provided some evidence that it is political history thatmatters most clearly. At the same time we would not interpret our evidence asoutright rejecting the cultural theories of institutions.
Finally, we have consistently found that the better performing governmentsare larger and collect higher taxes. Poorly performing governments, in contrast,are smaller and collect fewer taxes. This result does not of course imply that it isoften, or ever, socially desirable to expand a government of a given quality, butit tells us that identifying big government with bad government can be highlymisleading. The question of how the better governments get to be bigger, orvice versa, remains open for future work.
Appendix A
Summary statistics
Variable Number Mean Standard Minimum Maximumof obs. Deviation
Interference with private sector:
Property rights index 149 3.2752 1.1500 1.0 5.0Regulation index 149 2.7047 0.9480 1.0 5.0Top marginal tax rate 82 41.390 12.2812 0.0 66.0
Efficiency:
Corruption index 126 5.6532 2.2945 0.1786 10.0Bureaucratic delays 60 4.6872 1.4454 2.0292 7.7760Tax compliance 49 3.1571 1.0019 1.43 5.05Avg. government wages / 63 1.9111 1.2816 0.1 7.1GNP per capita
Output of public goods:
Log infant mortality 196 3.7207 0.9020 1.9792 5.2444Log school achievement 106 1.3934 0.5900 0.1484 2.4356Illiteracy rate 128 31.0057 22.5418 1.8 81.3Infrastructure quality 60 5.5284 1.8732 1.5 9.1521
Size of public sector:
Transfers and subsidies as 90 9.1837 8.3419 0.1 37.2% of GDPGovernment consumption as 104 15.028 4.9606 6.68 33.8% of GDPSOEs in the economy 104 4.3654 2.2212 0.0 10.0Public sector employment as 124 4.3773 3.2256 0.4 17.3973% of population
Continued
The Quality of Government 267
Appendix A Continued
Variable Number Mean Standard Minimum Maximumof obs. Deviation
Political freedom:
Democracy index 161 3.6739 3.8922 0.0 10.0Political rights index 209 4.6029 2.2361 1.0 7.0
Determinants:
Ethno-linguistic 161 0.3264 0.3006 0.0 1.0fractionalizationEnglish legal origin 212 0.3443 0.4763 0.0 1.0Socialist legal origin 212 0.1651 0.3721 0.0 1.0French legal origin 212 0.4339 0.4967 0.0 1.0German legal origin 212 0.0330 0.1791 0.0 1.0Scandinavian legal origin 212 0.0236 0.1521 0.0 1.0Protestant population as 206 14.6932 23.3400 0.0 99.8% of totalRoman Catholic population as 209 32.4077 36.1883 0.0 99.1% of totalMuslim population as % of total 209 21.7399 35.2773 0.0 99.9Other populations as % of total 206 30.9192 31.8005 0.0 100.0
Economic Development:
Latitude 209 0.2810 0.1885 0.0 0.8Log GNP per capita 186 7.2858 1.3572 4.6471 10.1517
268 The Journal of Law, Economics, & Organization, V15 N1
App
endi
xB
Inde
pend
entv
aria
bles
byco
untry
Cou
ntry
Cod
eE
thno
lig.
Rel
igio
nLa
titud
eLo
gG
NP
Fra
ctio
n.Le
galo
rigin
(per
cent
ofpo
pula
tion)
per
capi
ta
Eng
lish
Soc
ialis
tFr
ench
Ger
man
Sca
ndin
.P
rote
stan
tC
atho
licM
uslim
Oth
erde
n.
Afg
hani
stan
afg
0.44
840
01
00
0.00
0.00
99.3
00.
700.
3667
5.26
44A
lban
iaal
b0.
0017
01
00
00.
000.
0020
.50
79.5
00.
4556
6.06
38A
lger
iadz
a0.
2937
00
10
00.
000.
5099
.10
0.40
0.31
117.
4568
Am
eric
anS
amoa
asm
10
00
076
.00
20.6
00.
003.
400.
1578
8.18
64A
ndor
raad
o0
01
00
0.50
99.1
00.
000.
400.
4700
Ang
ola
ago
0.77
280
01
00
19.8
068
.70
0.00
11.5
00.
1367
6.40
80A
ntig
uaan
dB
arbu
daat
g0.
0000
10
00
042
.20
10.2
00.
4047
.20
0.18
928.
4185
Arg
entin
aar
g0.
1769
00
10
02.
7091
.60
0.20
5.50
0.37
788.
1177
Arm
enia
arm
01
00
00.
000.
000.
0010
0.00
0.44
447.
2282
Aru
baab
w0
01
00
8.00
86.0
00.
006.
000.
1367
Aus
tralia
aus
0.11
281
00
00
23.5
029
.60
0.20
46.7
00.
3000
9.33
27A
ustri
aau
t0.
0332
00
01
06.
5088
.80
0.60
4.10
0.52
449.
3520
Aze
rbai
jan
aze
0.00
000
10
00
0.00
0.00
93.4
06.
600.
4478
6.87
73B
aham
as,T
hebh
s0.
0000
10
00
047
.20
25.5
00.
0027
.30
0.26
838.
8955
Bah
rain
bhr
10
00
00.
900.
8095
.00
3.30
0.28
898.
9605
Ban
glad
esh
bgd
0.00
001
00
00
0.20
0.20
85.9
013
.70
0.26
675.
0582
Bar
bado
sbr
b0.
0733
10
00
033
.20
5.90
0.20
60.7
00.
1456
8.27
13B
elar
usbl
r0
10
00
0.00
14.0
00.
0086
.00
0.55
897.
9818
Bel
gium
bel
0.36
380
01
00
0.40
90.0
01.
108.
500.
5611
9.35
66B
eliz
ebl
z0.
4091
10
00
013
.20
66.8
00.
0020
.00
0.19
067.
0908
Ben
inbe
n0.
6831
00
10
02.
8018
.50
15.2
063
.50
0.10
335.
7114
Ber
mud
abm
u0.
0000
10
00
027
.00
18.6
00.
0054
.40
0.35
789.
6338
Bhu
tan
btn
0.43
751
00
00
0.00
0.00
5.00
95.0
00.
3033
5.98
07B
oliv
iabo
l0.
5994
00
10
02.
3092
.50
0.00
5.20
0.18
896.
2779
Bos
nia
and
Her
zego
vina
bih
01
00
04.
0015
.00
40.0
041
.00
0.48
89
Con
tinue
d
The Quality of Government 269
App
endi
xB
Con
tinue
d
Cou
ntry
Cod
eE
thno
lig.
Rel
igio
nLa
titud
eLo
gG
NP
Fra
ctio
n.Le
galo
rigin
(per
cent
ofpo
pula
tion)
per
capi
ta
Eng
lish
Soc
ialis
tFr
ench
Ger
man
Sca
ndin
.P
rote
stan
tC
atho
licM
uslim
Oth
erde
n.
Bot
swan
abw
a0.
3775
10
00
026
.80
9.40
0.00
63.8
00.
2444
7.13
06B
razi
lbr
a0.
0558
00
10
04.
0087
.80
0.10
8.10
0.11
117.
5254
Bru
nei
brn
0.50
001
00
00
1.10
3.20
64.2
031
.50
0.04
7810
.045
5B
ulga
riabg
r0.
1157
01
00
00.
400.
5010
.60
88.5
00.
4778
7.62
32B
urki
naFa
sobf
a0.
5467
00
10
01.
609.
0043
.00
46.4
00.
1444
5.33
79B
urun
dibd
i0.
0133
00
10
04.
9078
.30
0.90
15.9
00.
0367
5.17
79C
ambo
dia
khm
0.13
350
10
00
0.10
0.10
2.40
97.4
00.
1444
5.35
66C
amer
oon
cmr
0.85
200
01
00
18.1
035
.00
22.0
024
.90
0.66
676.
4922
Can
ada
can
0.37
621
00
00
29.6
046
.60
0.60
23.2
00.
6667
9.43
92C
ape
Verd
ecp
v0.
3750
00
10
03.
0095
.90
0.00
1.10
0.17
786.
1755
Cay
man
Isla
nds
cym
10
00
092
.40
2.60
0.00
5.00
0.21
44C
entra
lAfri
can
Rep
ublic
caf
0.78
560
01
00
50.0
033
.10
3.20
13.7
00.
0778
5.65
65C
had
tcd
0.66
620
01
00
11.6
021
.00
44.0
023
.40
0.16
675.
0508
Chi
lech
i0.
0506
00
10
01.
9082
.10
0.00
16.0
00.
3333
7.52
52C
hina
chn
0.23
330
10
00
0.00
0.00
2.40
97.6
00.
3889
5.99
46C
olom
bia
col
0.05
580
01
00
0.90
96.6
00.
202.
300.
0444
6.93
96C
omor
roIs
land
com
1.00
000
01
00
0.10
0.10
99.7
00.
100.
1344
6.09
81C
ongo
cog
0.66
930
01
00
24.9
053
.90
0.40
20.8
00.
0111
6.66
13C
osta
Ric
acr
i0.
0532
00
10
05.
8090
.50
0.00
3.70
0.11
117.
2865
Cot
ed’
Ivoi
reci
v0.
8565
00
10
04.
7018
.50
24.0
052
.80
0.08
896.
5819
Cro
atia
hrv
01
00
00.
4076
.50
1.20
21.9
00.
5011
7.91
81C
uba
cub
01
00
00.
8032
.00
0.00
67.2
00.
2367
Cyp
rus
cyp
0.30
001
00
00
0.60
1.30
18.5
079
.60
0.38
898.
6090
Cze
chR
epub
liccz
e0
10
00
4.60
39.2
00.
0056
.20
0.54
948.
0889
Den
mar
kdn
k0.
0275
00
00
195
.20
0.60
0.20
4.00
0.62
229.
5613
Con
tinue
d
270 The Journal of Law, Economics, & Organization, V15 N1
App
endi
xB
Con
tinue
d
Cou
ntry
Cod
eE
thno
lig.
Rel
igio
nLa
titud
eLo
gG
NP
Fra
ctio
n.Le
galo
rigin
(per
cent
ofpo
pula
tion)
per
capi
ta
Eng
lish
Soc
ialis
tFr
ench
Ger
man
Sca
ndin
.P
rote
stan
tC
atho
licM
uslim
Oth
erde
n.
Djib
outi
dji
0.71
430
01
00
0.20
6.70
90.6
02.
500.
1256
Dom
inic
adm
a0.
5000
10
00
08.
3089
.80
0.00
1.90
0.14
787.
1474
Dom
inic
anR
epub
licdo
m0.
0108
00
10
01.
4096
.60
0.00
2.00
0.21
116.
8041
Ecu
ador
ecu
0.32
540
01
00
1.90
96.4
00.
001.
700.
0222
6.89
46E
gypt
,Ara
bR
epub
liceg
y0.
0231
00
10
00.
200.
2081
.80
17.8
00.
3000
6.28
51E
quat
oria
lGui
nea
gnq
0.62
500
01
00
4.90
71.3
00.
5023
.30
0.02
225.
9369
Erit
rea
eri
00
10
00.
0033
.00
0.16
67E
ston
iaes
t0
10
00
66.0
02.
000.
0131
.99
0.65
568.
1819
Eth
iopi
aet
h0.
6771
00
10
03.
800.
7031
.40
64.1
00.
0889
4.99
23Fa
roe
Isla
nds
fro0
01
00
99.8
00.
100.
000.
100.
6889
Fiji
fji0.
8000
10
00
039
.10
9.00
7.80
44.1
00.
2000
7.34
58Fi
nlan
dfin
0.10
500
00
01
93.1
00.
100.
006.
800.
7111
9.40
51Fr
ance
fra0.
1455
00
10
02.
4076
.40
3.00
18.2
00.
5111
9.42
29Fr
ench
Poly
nesi
apy
f0
01
00
54.0
039
.40
0.00
6.60
0.16
678.
4025
Fren
chG
uian
agu
f0
01
00
3.90
87.0
01.
008.
100.
0444
7.67
04G
abon
gab
0.79
670
01
00
18.8
065
.20
0.80
15.2
00.
0111
8.20
50G
ambi
a,Th
egm
b0.
7804
10
00
00.
401.
9084
.80
12.9
00.
1476
5.60
13G
eorg
iage
o0
10
00
0.00
1.00
11.0
088
.00
0.46
677.
1468
Ger
man
yade
u0.
0438
00
01
046
.40
35.0
00.
0218
.58
0.56
6710
.151
7G
hana
gha
0.70
611
00
00
25.8
018
.70
15.7
039
.80
0.08
895.
8978
Gre
ece
grc
0.07
780
01
00
0.10
0.40
1.50
98.0
00.
4333
8.34
88G
reen
land
grl
00
10
098
.60
0.10
0.00
1.30
0.80
00G
rena
dagr
d0.
0000
10
00
013
.20
64.4
00.
2022
.20
0.13
417.
7341
Gua
delo
upe
glp
0.00
000
01
00
1.90
95.1
01.
101.
900.
1794
7.84
29G
uam
gum
10
00
016
.40
79.5
00.
004.
100.
1476
Con
tinue
d
The Quality of Government 271
App
endi
xB
Con
tinue
d
Cou
ntry
Cod
eE
thno
lig.
Rel
igio
nLa
titud
eLo
gG
NP
Fra
ctio
n.Le
galo
rigin
(per
cent
ofpo
pula
tion)
per
capi
ta
Eng
lish
Soc
ialis
tFr
ench
Ger
man
Sca
ndin
.P
rote
stan
tC
atho
licM
uslim
Oth
erde
n.
Gua
tem
ala
gtm
0.47
670
01
00
4.90
94.0
00.
001.
100.
1700
6.81
43G
uine
agi
n0.
7598
00
10
00.
101.
1069
.00
29.8
00.
1222
6.17
12G
uine
a-bi
ssau
gnb
0.85
000
01
00
0.60
10.2
038
.30
50.9
00.
1333
5.26
66G
uyan
agu
y0.
2378
10
00
018
.00
18.0
09.
0055
.00
0.05
566.
2404
Hai
tiht
i0.
0644
00
10
012
.80
82.6
00.
004.
600.
2111
5.54
58H
ondu
ras
hnd
0.09
740
01
00
2.60
95.8
00.
101.
500.
1667
6.37
62H
ong
Kon
ghk
g0.
2368
10
00
07.
507.
900.
5084
.10
0.24
618.
9795
Hun
gary
hun
0.06
510
10
00
21.6
053
.90
0.00
24.5
00.
5222
7.82
21Ic
elan
dis
l0.
1000
00
00
196
.60
0.70
0.00
2.70
0.72
229.
5796
Indi
ain
d0.
7422
10
00
01.
101.
3011
.60
86.0
00.
2222
5.52
15In
done
sia
idn
0.69
060
01
00
4.80
2.70
43.4
049
.10
0.05
566.
1649
Iran
irn0
01
00
0.00
0.10
97.9
02.
000.
3556
7.89
99Ira
qirq
00
10
00.
001.
8095
.80
2.40
0.36
67Ire
land
irl0.
0904
10
00
01.
1095
.30
0.00
3.60
0.58
898.
7672
Isle
ofM
anim
y1
00
00
20.9
010
.00
0.00
69.1
00.
6017
Isra
elis
r0.
3271
10
00
00.
201.
008.
0090
.80
0.34
788.
8921
Italy
ita0.
0389
00
10
00.
4083
.20
0.10
16.3
00.
4722
9.17
11Ja
mai
caja
m0.
0125
10
00
055
.50
9.60
0.10
34.8
00.
2017
7.14
16Ja
pan
jpn
0.00
990
00
10
0.90
0.60
0.00
98.5
00.
4000
9.58
34Jo
rdan
jor
0.02
970
01
00
0.30
1.70
93.0
05.
000.
3444
7.39
47K
azak
stan
kaz
01
00
02.
003.
0047
.00
48.0
00.
5333
7.55
71K
enya
ken
0.82
701
00
00
19.3
026
.40
6.00
48.3
00.
0111
5.73
04K
iriba
tiki
r0.
5000
10
00
045
.30
49.0
00.
005.
700.
0139
6.69
48K
orea
,Sou
thko
r0.
0000
00
01
012
.20
3.90
0.00
83.9
00.
4111
8.03
38K
orea
,Dem
ocra
ticR
ep.
prk
01
00
00.
000.
000.
0010
0.00
0.44
44
Con
tinue
d
272 The Journal of Law, Economics, & Organization, V15 N1
App
endi
xB
Con
tinue
d
Cou
ntry
Cod
eE
thno
lig.
Rel
igio
nLa
titud
eLo
gG
NP
Fra
ctio
n.Le
galo
rigin
(per
cent
ofpo
pula
tion)
per
capi
ta
Eng
lish
Soc
ialis
tFr
ench
Ger
man
Sca
ndin
.P
rote
stan
tC
atho
licM
uslim
Oth
erde
n.
Kuw
ait
kwt
00
10
00.
102.
1095
.10
2.70
0.32
569.
5581
Kyr
gyz
Rep
ublic
kgz
01
00
00.
000.
0070
.00
30.0
00.
4556
6.94
97La
o,Pe
ople
’sD
em.
R.
lao
0.25
000
10
00
0.20
0.80
1.00
98.0
00.
2000
5.57
03La
tvia
lva
01
00
014
.10
18.2
00.
0167
.69
0.63
338.
0794
Leba
non
lbn
0.11
400
01
00
1.00
36.2
037
.40
25.4
00.
3722
7.49
89Le
soth
ols
o0.
2098
10
00
029
.80
43.5
00.
0026
.70
0.32
566.
0115
Libe
rialb
r0.
8031
10
00
018
.60
1.90
21.2
058
.30
0.07
006.
1408
Liby
alb
y0.
1214
00
10
00.
100.
2098
.10
1.60
0.27
788.
7087
Liec
hten
stei
nlie
00
01
010
.70
87.9
00.
001.
400.
5233
Lith
uani
altu
01
00
080
.00
0.00
0.62
227.
8258
Luxe
mbo
urg
lux
0.21
670
01
00
1.20
93.0
00.
005.
800.
5494
9.83
36M
acao
mac
00
10
00.
508.
800.
0090
.70
0.24
56M
aced
onia
mkd
01
00
01.
000.
5030
.10
68.4
00.
4611
6.75
11M
adag
asca
rm
dg0.
0627
00
10
022
.00
26.0
01.
7050
.30
0.22
225.
6334
Mal
awi
mw
i0.
6224
10
00
031
.50
27.6
016
.20
24.7
00.
1478
5.06
79M
alay
sia
mys
0.61
041
00
00
1.40
2.80
49.4
046
.40
0.02
567.
4685
Mal
dive
sm
dv0.
0333
10
00
00.
000.
1099
.90
0.00
0.03
506.
5399
Mal
im
li0.
8086
00
10
00.
200.
7080
.00
19.1
00.
1889
5.25
71M
alta
mlt
0.10
330
01
00
0.50
97.3
00.
002.
200.
3944
8.18
21M
arsh
allI
slan
dsm
hl1
00
00
0.00
0.00
0.10
00M
artin
ique
mtq
0.00
000
01
00
2.20
95.5
00.
102.
200.
1600
7.98
17M
aurit
ania
mrt
0.27
000
01
00
0.00
0.30
99.4
00.
300.
2222
5.98
18M
aurit
ius
mus
0.70
850
01
00
0.90
31.2
016
.40
51.5
00.
2241
7.54
14M
ayot
tem
yt0
01
00
0.30
0.90
98.8
00.
000.
1389
Mex
ico
mex
0.17
410
01
00
1.20
94.7
00.
004.
100.
2556
7.70
09
Con
tinue
d
The Quality of Government 273
App
endi
xB
Con
tinue
d
Cou
ntry
Cod
eE
thno
lig.
Rel
igio
nLa
titud
eLo
gG
NP
Fra
ctio
n.Le
galo
rigin
(per
cent
ofpo
pula
tion)
per
capi
ta
Eng
lish
Soc
ialis
tFr
ench
Ger
man
Sca
ndin
.P
rote
stan
tC
atho
licM
uslim
Oth
erde
n.
Mic
rone
sia,
Fed.
Sta
tes
fsm
10
00
049
.20
45.6
00.
005.
200.
0728
7.57
56M
oldo
vam
da0
10
00
0.00
0.00
0.00
100.
000.
5222
7.20
97M
onac
om
co0
01
00
4.60
90.7
00.
004.
700.
4827
Mon
golia
mng
0.07
370
10
00
0.00
0.00
1.40
98.6
00.
5111
6.94
17M
oroc
com
ar0.
3480
00
10
00.
000.
2099
.40
0.40
0.35
566.
6175
Moz
ambi
que
moz
0.78
630
01
00
6.80
31.4
013
.00
48.8
00.
2017
4.71
98M
yanm
arm
mr
0.38
400
10
00
3.20
0.90
3.60
92.3
00.
2444
Nam
ibia
nam
0.72
831
00
00
64.2
019
.10
0.00
16.7
00.
2444
7.45
53N
epal
npl
0.45
001
00
00
0.00
0.00
3.00
97.0
00.
3111
4.99
77N
ethe
rland
snl
d0.
0634
00
10
042
.40
42.6
01.
0014
.00
0.58
119.
3772
Net
herla
nds
Ant
illes
ant
0.13
890
01
00
9.20
87.2
00.
103.
500.
1350
8.70
41N
ewZe
alan
dnz
l0.
1476
10
00
037
.90
18.7
00.
0043
.40
0.45
568.
9693
New
Cal
edon
ianc
l0
01
00
18.1
072
.50
4.00
5.40
0.23
678.
6326
Nic
arag
uani
c0.
0992
00
10
04.
4094
.70
0.00
0.90
0.14
446.
3723
Nig
erne
r0.
7329
00
10
00.
000.
2087
.90
11.9
00.
1778
5.61
40N
iger
iang
a0.
8567
10
00
015
.80
12.1
045
.00
27.1
00.
1111
6.36
70N
orw
ayno
r0.
0699
00
00
197
.80
0.30
0.10
1.80
0.68
899.
6846
Om
anom
n0
01
00
0.10
0.10
98.9
00.
900.
2333
8.25
75Pa
kist
anpa
k0.
6216
10
00
00.
800.
5096
.80
1.90
0.33
335.
7316
Pana
ma
pan
0.19
080
01
00
5.20
85.0
04.
505.
300.
1000
7.50
04Pa
pua
New
Gui
nea
png
0.80
271
00
00
58.4
032
.80
0.00
8.80
0.06
676.
6082
Para
guay
pry
0.41
110
01
00
1.90
96.0
00.
002.
100.
2556
6.96
60Pe
rupe
r0.
4316
00
10
02.
7095
.10
0.00
2.20
0.11
117.
0173
Phi
lippi
nes
phl
0.72
380
01
00
3.80
84.1
04.
307.
800.
1444
6.36
57Po
land
pol
0.03
900
10
00
0.10
81.0
00.
0018
.90
0.57
787.
5977
Con
tinue
d
274 The Journal of Law, Economics, & Organization, V15 N1
App
endi
xB
Con
tinue
d
Cou
ntry
Cod
eE
thno
lig.
Rel
igio
nLa
titud
eLo
gG
NP
Fra
ctio
n.Le
galo
rigin
(per
cent
ofpo
pula
tion)
per
capi
ta
Eng
lish
Soc
ialis
tFr
ench
Ger
man
Sca
ndin
.P
rote
stan
tC
atho
licM
uslim
Oth
erde
n.
Portu
gal
prt
0.00
250
01
00
1.10
94.1
00.
004.
800.
4367
8.23
80P
uerto
Ric
opr
i0.
0267
00
10
05.
1091
.50
0.00
3.40
0.20
178.
2965
Qat
arqa
t0
01
00
0.90
1.20
92.4
05.
500.
2811
9.74
41R
euni
onre
u0
01
00
0.50
96.3
02.
201.
000.
2340
Rom
ania
rom
0.12
200
10
00
5.80
4.90
1.20
88.1
00.
5111
7.27
93R
ussi
anFe
dera
tion
rus
01
00
00.
001.
4011
.30
87.3
00.
6667
8.07
05R
wan
darw
a0.
0609
00
10
011
.60
55.6
08.
6024
.20
0.02
225.
3901
Sal
vado
r,E
lsl
v0.
0514
00
10
02.
4096
.20
0.00
1.40
0.15
006.
6524
Sao
Tom
ean
dP
rinci
pest
p0.
0000
00
10
02.
2092
.40
0.00
5.40
0.01
115.
9841
Sau
diA
rabi
asa
u1
00
00
0.10
0.10
98.8
01.
000.
2778
9.10
06S
eneg
alse
n0.
7789
00
10
00.
105.
6091
.00
3.30
0.15
566.
1754
Sey
chel
les
syc
0.00
000
01
00
1.10
89.3
00.
309.
300.
0483
7.94
20S
ierr
aLe
one
sle
0.81
301
00
00
4.80
2.20
39.4
053
.60
0.09
225.
5518
Sin
gapo
resg
p0.
3215
10
00
02.
604.
7017
.40
75.3
00.
0136
9.02
33S
lova
kR
epub
licsv
k0
10
00
8.40
74.0
00.
0017
.60
0.53
787.
9918
Slo
veni
asv
n0
10
00
0.00
71.4
01.
5027
.10
0.51
118.
8777
Sol
omon
Isla
ndsl
b0.
5714
10
00
039
.80
19.1
00.
0041
.10
0.08
896.
3531
Som
alia
som
0.07
911
00
00
0.00
0.00
99.8
00.
200.
1111
4.64
71S
outh
Afri
caza
f0.
8310
10
00
039
.00
10.4
01.
3049
.30
0.32
227.
6137
Spa
ines
p0.
2745
00
10
00.
1096
.90
0.00
3.00
0.44
448.
7537
Sri
Lank
alk
a0.
3257
10
00
00.
406.
807.
2085
.60
0.07
785.
9197
St.
Kitt
san
dN
evis
kna
0.00
001
00
00
52.0
08.
200.
0039
.80
0.19
118.
0191
St.
Luci
alc
a0.
5833
10
00
06.
3088
.30
0.00
5.40
0.15
037.
9153
St.
Vinc
entA
ntill
esvc
t0.
0000
10
00
040
.70
19.4
00.
0039
.90
0.14
616.
9234
Sud
ansd
n0.
5122
10
00
00.
104.
4073
.00
22.5
00.
1667
5.89
57
Con
tinue
d
The Quality of Government 275
App
endi
xB
Con
tinue
d
Cou
ntry
Cod
eE
thno
lig.
Rel
igio
nLa
titud
eLo
gG
NP
Fra
ctio
n.Le
galo
rigin
(per
cent
ofpo
pula
tion)
per
capi
ta
Eng
lish
Soc
ialis
tFr
ench
Ger
man
Sca
ndin
.P
rote
stan
tC
atho
licM
uslim
Oth
erde
n.
Sur
inam
esu
r0.
7500
00
10
036
.60
36.0
013
.00
14.4
00.
0444
7.50
38S
waz
iland
swz
0.00
001
00
00
33.9
010
.80
0.10
55.2
00.
2922
6.68
70S
wed
ensw
e0.
0650
00
00
168
.40
1.40
0.10
30.1
00.
6889
9.60
34S
witz
erla
ndch
e0.
3076
00
01
043
.20
52.8
00.
303.
700.
5222
9.86
67S
yria
nA
rab
Rep
ublic
syr
0.09
480
01
00
0.20
1.30
89.6
08.
900.
3889
7.01
88Ta
iwan
tai
0.25
510
00
10
3.00
1.50
0.50
95.0
00.
2589
8.60
52Ta
jikis
tan
tjk0
10
00
0.00
0.00
85.0
015
.00
0.43
336.
4547
Tanz
ania
tza
0.89
021
00
00
11.2
028
.20
32.5
028
.10
0.06
675.
1980
Thai
land
tha
0.35
691
00
00
0.20
0.40
3.90
95.5
00.
1667
6.86
97To
gotg
o0.
7285
00
10
06.
1029
.30
17.0
047
.60
0.08
895.
7328
Tong
ato
n0.
0000
10
00
061
.90
17.9
00.
0020
.20
0.22
227.
1199
Trin
idad
and
Toba
gotto
0.23
131
00
00
13.2
035
.80
6.50
44.5
00.
1222
8.21
23Tu
nisi
atu
n0.
0703
00
10
00.
000.
1099
.40
0.50
0.37
787.
0194
Turk
eytu
r0.
1636
00
10
00.
000.
1099
.20
0.70
0.43
337.
3620
Turk
men
ista
ntk
m0
10
00
0.00
0.00
87.0
013
.00
0.44
447.
0838
Uga
nda
uga
0.83
581
00
00
1.90
49.6
06.
6041
.90
0.01
115.
6250
Ukr
aine
ukr
01
00
00.
000.
000.
0010
0.00
0.54
447.
7826
Uni
ted
Sta
tes
ofA
mer
ica
usa
0.20
901
00
00
43.6
030
.00
0.80
25.6
00.
4222
9.61
80U
nite
dA
rab
Em
irate
sar
e0.
0000
10
00
00.
300.
4094
.90
4.40
0.26
679.
9490
Uni
ted
Kin
gdom
gbr
0.10
631
00
00
16.1
013
.10
1.40
69.4
00.
6000
9.16
54U
rugu
ayur
y0.
0667
00
10
01.
9059
.50
0.00
38.6
00.
3667
7.73
73U
zbek
ista
nuz
b0
10
00
0.00
0.00
88.0
012
.00
0.45
566.
9233
Vanu
atu
vut
0.54
411
00
00
54.6
016
.90
0.00
28.5
00.
1778
6.97
03Ve
nezu
ela
ven
0.05
250
01
00
1.00
94.8
00.
004.
200.
0889
8.03
99Vi
etna
mvn
m0.
1176
01
00
00.
203.
901.
0094
.90
0.17
785.
3148
Con
tinue
d
276 The Journal of Law, Economics, & Organization, V15 N1
App
endi
xB
Con
tinue
d
Cou
ntry
Cod
eE
thno
lig.
Rel
igio
nLa
titud
eLo
gG
NP
Fra
ctio
n.Le
galo
rigin
(per
cent
ofpo
pula
tion)
per
capi
ta
Eng
lish
Soc
ialis
tFr
ench
Ger
man
Sca
ndin
.P
rote
stan
tC
atho
licM
uslim
Oth
erde
n.
Virg
inIs
land
svi
r1
00
00
43.0
033
.60
0.00
23.4
00.
2022
Wes
tern
Sam
oaw
sm0.
0000
10
00
076
.30
21.3
00.
002.
400.
1483
6.76
01Ye
men
Rep
ublic
bye
m0.
0122
00
10
00.
100.
0099
.50
0.40
0.16
675.
6168
Zaire
zar
0.87
230
01
00
29.0
048
.40
1.40
21.2
00.
0000
5.79
21Za
mbi
azm
b0.
8294
10
00
031
.90
26.2
00.
3041
.60
0.16
676.
1321
Zim
babw
ezw
e0.
5986
10
00
021
.40
14.4
00.
9063
.30
0.22
226.
4040
a Eth
nolin
guis
ticfra
ctio
naliz
atio
nfo
rGer
man
yis
the
inde
xca
lcul
ated
fort
hefo
rmer
Wes
tGer
man
y.b
Eth
nolin
guis
ticfra
ctio
naliz
atio
nfo
rthe
Rep
ublic
ofYe
men
isth
eav
erag
eof
the
form
erN
orth
and
Sou
thYe
men
.
The Quality of Government 277
ReferencesAlesina, Alberto, Reza Baqir, and William Easterly. 1999. “Public Goods and Ethnic Divisions,”
114Quarterly Journal of Economics.Banfield, Edward. 1958.The Moral Basis of a Backward Society. New York: Free Press.Barraclough, Geoffrey. 1972.The Medieval Papacy. New York: Harcourt, Brace and World.Barrett, David B., ed. 1982.World Christian Encyclopedia: A Comparative Study of Churches and
Religions in the Modern World, AD 1900–2000. New York: Oxford University Press.Barro, Robert. 1991. “Economic Growth in a Cross-Section of Countries,” 106Quarterly Journal
of Economics407–443.. 1996a. “Democracy and Growth,” 1Journal of Economic Growth1–27.. 1996b. “Determinants of Economic Growth: A Cross-Country Empirical Study,” Working
Paper Series 5698, National Bureau of Economic Research., and Jong-Wha Lee. 1994. Unpublished dataset for a panel of 138 countries. Harvard
University.Bates, Robert. 1981.Markets and States in Tropical Africa. Berkeley: University of California
Press.Beasley, William. 1972.The Meiji Restoration. Palo Alto, CA: Stanford University Press.Becker, Gary. 1983. “A Theory of Competition Among Pressure Groups for Political Influence,”
98Quarterly Journal of Economics371–400.Boycko, Maxim, Andrei Shleifer, and Robert W. Vishny. 1995.Privatizing Russia. Cambridge:
MIT Press.. 1996. “A Theory of Privatization,” 106Economic Journal309–319.
Brewer, John. 1988.The Sinews of Power. New York: Knopf.Bruk, S. I., and V. S. Apenchenko, eds. in chief. 1964.Atlas Narodov Mira. Moscow: Miklukho-
Maklai Ethnological Institute at the Department of Geodesy and Cartography of the StateGeological Committee of the Soviet Union, USSR Academy of Sciences.
BERI (Business Environmental Risk Intelligence), Center for Institutional Reform and the InformalSector (IRIS). 1995.BERI’s Operations Risk Index. College Park: IRIS, University of Maryland.
Central Intelligence Agency. 1996.CIA World Factbook(published online).Coleman, James. 1990.Foundations of Social Theory. Cambridge: Harvard University Press.David, Rene, and John Brierley. 1978.Major Legal Systems in the World Today. New York: Free
Press.DeLong, Bradford, and Andrei Shleifer. 1993. “Princes and Merchants: Government and City
Growth Before the Industrial Revolution,” 36Journal of Law and Economics671–702.Demsetz, Harold. 1967. “Toward a Theory of Property Rights,” 57American Economic Review
Papers and Proceedings347–359.De Toqueville, Alexis. [1858] 1955.The Old Regime and the French Revolution. New York:
Doubleday.Easterly, William, and Ross Levine. 1997. “Africa’s Growth Tragedy: Policies and Ethnic
Divisions,” 112Quarterly Journal of Economics1203–1250.Elster, Jon, Claus Offe, and Ulrich Press. 1998.Institutional Design in Post-Communist Societies.
Cambridge: Cambridge University Press.Ertman, Thomas. 1997.Birth of the Leviathan. Cambridge: Cambridge University Press.Finer, Samuel. 1997.The History of Government, Vol. I–III. Cambridge: Cambridge University
Press.Freedom House. 1996.World Survey of Economic Freedom 1995–1996: A Freedom House Study.
New Brunswick, NJ: Transaction Publishers.. 1996.Freedom in the World: The Annual Survey of Political Rights and Civil Liberties
1995–1996. New Brunswick, NJ: Freedom House.Fukuyama, Francis. 1995.Trust. New York: Free Press.Grossman, Sanford, and Oliver Hart. 1986. “The Costs and Benefits of Ownership: a Theory of
Vertical and Lateral Integration,” 94Journal of Political Economy691–719.Gunnemark, Erik V. 1991.Countries, Peoples, and Their Languages: The Linguistic Handbook.
Gothenburg, Sweden: Lanstryckeriet.Gwartney, James, Robert Lawson, and Walter Block. 1996.Economic Freedom of the World,
278 The Journal of Law, Economics, & Organization, V15 N1
1975–1995. Vancouver: The Fraser Institute.Hall, Robert, and Charles Jones. 1999. “Fundamental Determinants of Output per Worker Across
Countries,” forthcoming inQuarterly Journal of Economics.Hayek, Friedrich. 1944.The Road to Serfdom. London: G. Routledge and Sons.Holmes, Kim R., Bryan T. Johnson, and Melanie Kirkpatrick. 1997.1997 Index of Economic
Freedom. Washington, D.C.: The Heritage Foundation.Huntington, Samuel. 1968.Political Order in Changing Societies. New Haven: Yale University
Press.. 1991.The Third Wave. Norman: University of Oklahoma Press.
Jaggers, Keith, and Ted Robert Gurr. 1996.Polity III: Regime Change and Political Authority,1800–1994[computer file]. Ann Arbor: Inter-university Consortium for Political and SocialResearch [distributor].
Johnson, Simon, Daniel Kaufmann, and Andrei Shleifer. 1997. “The Unofficial Economy inTransition,” 2Brookings Papers on Economic Activity159–240.
Johnson, Stuart. 1975.Frederick the Great and His Officials. New Haven: Yale University Press.Jones, Eric. 1981.The European Miracle. Cambridge: Cambridge University Press.Kamen, Henry. 1997.The Spanish Inquisition. New Haven: Yale University Press.Knack, Stephen, and Philip Keefer. 1995. “Institutions and Economic Performance: Cross-country
Tests Using Alternative Institutional Measures,” 7Economics and Politics207–227.. 1997. “Does Social Capital Have an Economic Payoff? A Cross-country Investigation,”
112Quarterly Journal of Economics1251–1288.Landes, David. 1998.The Wealth and Poverty of Nations. New York: W. W. Norton.La Porta, Rafael, Florencio Lopez-de-Silanes, Andrei Shleifer, and Robert W. Vishny. (1997a).
“Legal Determinants of External Finance,” 52Journal of Finance1131–1150.. (1997b). “Trust in Large Organizations,” 87American Economic Review Papers and
Proceedings333–338.. 1998. “Law and Finance,” 106Journal of Political Economy1113–1155.
Lindbeck, Assar. 1997. “The Swedish Experiment,” 35Journal of Economic Literature1273–1319.Mack Smith, Denis. 1997.Modern Italy: a Political History. Ann Arbor: University of Michigan
Press.Marx, Karl. 1974 [1872].Das Kapital. London: Lawrence and Wishart.Mauro, Paolo. 1995. “Corruption and Growth,” 110Quarterly Journal of Economics681–712.Montesquieu, Charles de Secondat. 1748.The Spirit of the Laws. Paris.Muller, Siegfried H. 1964.The World’s Living Languages: Basic Facts of Their Structure, Kinship,
Location, and Number of Speakers. New York: Ungar.North, Douglass. 1981.Growth and Structural Change. New York: W. W. Norton.
. 1990. Institutions, Institutional Change, and Economic Performance. Cambridge:Cambridge University Press.
Olson, Mancur. 1965.The Logic of Collective Action. Cambridge: Harvard University Press.. 1982.The Rise and Decline of Nations. New Haven: Yale University Press.. 1993. “Dictatorship, Democracy, and Development,” 87American Political Science Review
567–576.. 1996. “Big Bills Left on the Sidewalk: Why Some Nations are Rich, and Others Poor,”
10Journal of Economic Perspectives3–24.Pipes, Richard. 1974.Russia under the Old Regime. New York: Charles Scribner’s Sons.Political Risk Services.International Country Risk Guide. East Syracuse, NY: Political Risk
Services, Institutional Reform and Informational Sector, various years.Putnam, Robert. 1993.Making Democracy Work: Civic Traditions in Modern Italy. Princeton:
Princeton University Press.Rauch, James. 1995. “Bureaucracy, Infrastructure, and Economic Growth: Evidence from U.S.
Cities During the Progressive Era,” 85American Economic Review968–979., and Peter Evans. 1997. “Bureaucratic Structure and Bureaucratic Performance in Less
Developed Countries,” mimeo, University of California, San Diego.Roberts, Janet. 1962. “Sociocultural Change and Communication Problems,” in Frank A. Rice, ed.,
Study of the Role of Second Languages in Asia, Africa, and Latin America. Washington, D.C.:
The Quality of Government 279
Center for Applied Linguistics of the Modern Language Association of America, 105–123.Schiavo-Campo, Salvatore, Giulio de Tommaso, and Amitabha Mukherjee. 1997. “An International
Statistical Survey of Government Employment and Wages,” inGovernment Employment andPay: A Global and Regional Perspective Series, Public Sector Management and InformationTechnology Team, Technical Department for Europe, Central Asia, Middle East and NorthAfrica, World Bank.
Shleifer, Andrei, and Robert W. Vishny. 1993. “Corruption,” 108Quarterly Journal of Economics599–617.
. 1994. “Politicians and Firms,” 109Quarterly Journal of Economics995–1025.
. 1998.The Grabbing Hand: Government Pathologies and their Cures. Cambridge: HarvardUniversity Press.
Smith, Adam. 1976, [1776].An Inquiry into the Nature and Causes of the Wealth of Nations.Chicago: University of Chicago Press.
Statistical Abstract of the World. 1995. New York: Gale Research, Inc.Tilly, Charles. 1990.Coercion, Capital, and European States. Berkeley: University of California
Press.Treisman, Daniel. 1997. “The Causes of Corruption: a Cross-National Study,” mimeo, UCLA.United Nations. 1995.Demographic Yearbook. New York: Department of Economic and Social
Affairs, Statistical Office, United Nations.Weber, Max. 1958.The Protestant Ethic and the Spirit of Capitalism. New York: Charles Scribner’s
Sons.Weingast, Barry. 1995. “The Economic Role of Political Institutions: Market-Preserving
Federalism and Economic Development,” 11Journal of Law, Economics & Organization1–31.
World Bank. 1995.Bureaucrats in Business. Oxford: Oxford University Press.. 1997.World Development Indicators 1997[CD-ROM]. Washington, D.C.: World Bank.
World Economic Forum. 1996.The Global Competitiveness Report 1996. Geneva: World EconomicForum.
Worldmark Encyclopedia of the Nations, 8th ed. 1995. Detroit: Gale Research.