A. RoL Theories

The theoretical literature does not offer a single definition of the RoL and is characterized by substantial conceptual confusion (Rodriguez, McCubbins, and Weingast Reference Rodriguez, McCubbins and Weingast2010, 1455–94). A recent review of the literature concludes that the RoL is undertheorized and that most practitioners and many scholars take the concept for granted, “equating it with the institutions and practices in those (relatively few) parts of the world where the rule of law has been largely achieved” (Hadfield and Weingast Reference Hadfield and Weingast2014, 22). Another group of scholars has criticized the idea that the RoL entails a set of institutional prescriptions. They argue that the RoL is necessarily culturally contingent, dependent on time and place, and that transplanting Western institutions in the name of RoL is bound to fail (Berkowitz, Pistor, and Richard Reference Berkowitz, Pistor and Richard2003; Brooks Reference Brooks2003; Tamanaha Reference Tamanaha2011, 209–47; Hadfield and Weingast Reference Hadfield and Weingast2014, 27). According to these scholars, a single definition of the RoL may not exist.

Another important source of contention is whether the RoL is a “thin,” formal, and procedural concept or a “thick” concept that builds in certain substantive values. The thin version of the RoL is commonly associated with Lon Fuller's classic definition, which stipulates eight procedural requirements for the RoL (Fuller Reference Fuller1969, 33–39; Rodriguez, McCubbins, and Weingast 2010, 1466–67):

1. 1. Generality, that is, conduct is stated in rules that are impartially applied;

2. 2. Publicity, that is, rules are publicly announced;

3. 3. Prospectivity, that is, rules will not be changed retroactively;

4. 4. Clarity, that is, rules are understandable for all;

5. 5. Consistency, that is, rules are not inconsistent or contradictory;

6. 6. Possibility of compliance, that is, no rules can demand conduct that is beyond the ability of those to whom they apply;

7. 7. Stability, that is, rules are stable and not subject to frequent change;

8. 8. Congruence between the rules as announced and their actual administration.

Fuller's version is thin because it makes no demand that the laws that are produced are good laws in any substantive sense. Numerous legal theorists have since built on this definition, adding further procedural and/or functional elements to his list. For example, scholars have singled out judicial independence and judicial review as further functional components of the RoL. Joseph Raz notes that “[s]ince just about any matter arising under any law can be subject to a conclusive court judgment, it is obvious that it is futile to guide one's action on the basis of the law if when the matter comes to adjudication the courts will not apply the law and will act for some other reasons” (Raz Reference Raz and Raz1979, 227). Despite these relatively minor disagreements, however, Fuller's definition remains the canonical definition of the thin or formal version of the RoL.

While the thin definition of the RoL is relatively uncontroversial, substantial disagreement exists over whether the RoL should include one or more substantive values and, if so, which ones (Rodriguez, McCubbins, and Weingast 2010, 1469). Some theorists have argued that the formal definition allows for what is sometimes called rule by law, in which the government is not genuinely constrained so long as it follows the formal requirements of the law. According to this view, true RoL requires some thicker content. Friedrich Hayek, for example, has suggested that the RoL “includes the recognition of the inalienable, individual right of man” (Hayek Reference Hayek1944, 63). As another example, Ronald Dworkin has argued that the RoL is not merely a rulebook, but that the concept is necessarily substantive and includes equality and rights (Rodriguez, McCubbins, and Weingast Reference Rodriguez, McCubbins and Weingast2010, 1470). Many other theorists have similarly added various substantive values to the definition of the RoL. Indeed, it may be exactly this indeterminacy that makes RoL an appealing concept. Rodriguez and coauthors write that the “[r]ule of law is an attractive ideal, but its attractiveness may stem mainly from its imprecision, which allows each of us to project our own sense of the ideal government onto the phrase rule of law” (Rodriguez, McCubbins, and Weingast Reference Rodriguez, McCubbins and Weingast2010, 1458; Ginsburg Reference Ginsburg2011).

Once substantive values are built into the rule of law, it necessarily becomes more contested. Let us take the example of human rights. Among the various substantive values that theorists have sought to incorporate into the RoL, respect for basic human rights is probably the most common.Footnote ³ It is perhaps also the least controversial, since the international community has articulated a set of human rights norms that are generally perceived as universal and may even have the status of jus cogens norms in international law (i.e., they are peremptory norms from which no deviation is permitted). Yet even these seemingly universal rights norms are surrounded by controversy: the universalist premise in human rights law has been criticized as an imposition of Western values on countries that purportedly prefer communal values and/or economic progress over individual liberties (Emmerson Reference Emmerson1995; Bauer and Bell Reference Bauer and Bell1999). Indeed, the somewhat tired debate over cultural relativism in international human rights law demonstrates how including substantive values in the definition of RoL opens it up to contestation.

What is more, incorporating substantive values makes the concept of the RoL harder to separate from neighboring concepts. Take again the example of human rights. A version of the RoL that includes human rights is hard to separate from the concept of human rights itself, especially since principles such as fair trial, access to justice, and a prohibition of retroactive laws are core human rights under international law. Indeed, as Joseph Raz has argued, once the notion of RoL is taken to mean the “rule of good law, then to explain its nature is to propound a complete social philosophy” with the result that “the term lacks any useful function” (Raz Reference Raz and Raz1979, 211). In other words, once substantive values are incorporated into the RoL, the concept becomes synonymous with those values and might lose analytical value on its own.

B. The Concepts Behind RoL Indicators

The same conceptual confusion that permeates the RoL literature also affects the various indicators that purport to measure the RoL (Møller and Skaaning Reference M⊘ller and Skaaning2011). Although all indicators appear to start from the core formalist ideal that contrasts the rule of law with the “rule of man” (to paraphrase John Adams's famous description of the RoL from the Massachusetts Constitution), they each add different substantive values, such as the absence of crime, human rights, substantive equality, the security of private contracts, and property rights. Indeed, not a single pair of the four indicators reviewed in this contribution defines the RoL in the same way. Of course, this diversity is unsurprising given the disagreements in the theoretical literature.

The World Bank RoL Index is probably the most well-known and most commonly used in social science research (Kaufmann, Kraay, and Mastruzzi Reference Kaufmann, Kraay and Mastruzzi2010). There is a growing consensus within the World Bank that the viability of investment and long-run economic growth depends on good governance and the quality of legal and political institutions, which caused the World Bank to start measuring RoL in the 1990s. The World Bank's RoL Index is part of a larger project, the WGI (World Bank 2014), which produces a number of indicators that capture different dimensions of good governance and are widely used by social scientists.Footnote ⁴ According to the World Bank, its RoL indicator captures “perceptions of the extent to which agents have confidence in and abide by the rules of society, and in particular the quality of contract enforcement, property rights, the police, and the courts, as well as the likelihood of crime and violence” (Kaufmann, Kraay, and Mastruzzi Reference Kaufmann, Kraay and Mastruzzi2010).Footnote ⁵ This definition demonstrates that the WGI's indicator not only includes procedural elements (e.g., whether “agents have confidence in and abide by the rules of society”) but also adds substantive concepts (e.g., security of property and private contracts). As Ginsburg previously argued, the WGI's indicator “conflates many different notions into a single concept, including both crime and contract enforcement in the same framework” (Ginsburg Reference Ginsburg2011, 271; see also Kurtz and Schrank Reference Kurtz and Schrank2007). Likewise, Thomas notes that the WGI's indicator lacks “content validity—it does not map onto a definition of the rule of law” (Thomas Reference Thomas2010, 40). Indeed, the World Bank's indicator is not only the most widely used, it is also the most heavily criticized. Appendix 1 provides more detailed information on the WGI's RoL Index.

The Heritage Foundation, a conservative US think tank, has constructed another prominent RoL index. The mission of the Heritage Foundation is to “formulate and promote conservative public policies based on the principles of free enterprise, limited government, individual freedom, traditional American values, and a strong national defense” (Heritage Foundation 2014a). The Heritage Foundation views the RoL as one of the four core economic freedoms, together with limited government, regulatory efficiency, and open markets. Compared to the WGI's indicator, the Heritage Foundation's RoL index is less focused on crime and personal security; instead, it focuses on property rights and the absence of corruption. The index assesses some formal components, such as “the independence of the judiciary, the existence of corruption within the judiciary, and the ability of individuals and businesses to enforce contracts” (Heritage Foundation 2014c). In addition, the index combines information on the security of private property rights, “the ability of individuals to accumulate private property, secured by clear laws that are fully enforced by the state,” with information on the absence of corruption. Thus, like the WGI's indicator, the Heritage Foundation's index builds substantive values into its RoL definition, albeit different ones. Appendix 1 provides more information on the Heritage Foundation's index.

Freedom House, a US-based nongovernmental organization “dedicated to the expansion of freedom around the world,” produces a third RoL measure. As part of its mission, Freedom House “speaks out against the main threats to democracy and empowers citizens to exercise their fundamental rights” and analyzes “the challenges to freedom; advocate[s] for greater political and civil liberties; and support[s] frontline activists to defend human rights and promote democratic change” (Freedom House 2014a). Unlike the World Bank and the Heritage Foundation, the Freedom House conceptualization of the RoL is heavily focused on human rights. Freedom House assigns numerical RoL scores to countries based on the following questions:

1. 1. Is there an independent judiciary?

2. 2. Does the rule of law prevail in civil and criminal matters? Are police under direct civilian control?

3. 3. Is there protection from political terror, unjustified imprisonment, exile, or torture, whether by groups that support or oppose the system? Is there freedom from war and insurgencies?

4. 4. Do laws, policies, and practices guarantee equal treatment of various segments of the population?

Like the definitions provided by the WGI and the Heritage Foundation, Freedom House's definition combines procedural elements (such as an independent judiciary and absence of abuse by state agents) with substantive elements (such as civil liberties, bodily integrity, and substantive equality). These substantive values appear to constitute at least half the information on which Freedom House bases its country scores. Unlike the WGI and the Heritage Foundation, however, the Freedom House Index does not capture the protection of private property rights or the security of private contracts. Appendix 1 provides further information on the Freedom House RoL index.

A more recent initiative in measuring the RoL comes from the WJP. The WJP is a nongovernmental organization founded with the mission “to advance the rule of law around the world,” based on the idea that “the rule of law provides the foundation for communities of opportunity and equity—communities that offer sustainable economic development, accountable government, and respect for fundamental rights” (WJP 2014, 1). Since the WJP's main goal is to promote the RoL, it is perhaps not surprising that its RoL indicator is by far the most ambitious effort to measure the rule of law globally. According to the WJP, its indicator “builds on years of development, intensive consultation, and vetting with academics, practitioners, and community leaders from over 100 countries and 17 professional disciplines” (WJP 2014, 1). The rule of law concept captured by the indicator is also the most comprehensive. The WJP (2014, 1) defines the RoL as follows:

To provide further clarity to this definition, it scores countries on nine separate factors: (1) constraints on government powers, (2) absence of corruption, (3) open government, (4) fundamental rights, (5) order and security, (6) regulatory enforcement, (7) civil justice, (8) criminal justice, and (9) informal justice (note, however, that the last component is not part of the published index).

Each of these nine factors has subdimensions, producing a total of forty-seven subdimensions (see Appendix 2 for a complete listing). As this long list demonstrates, the WJP's index combines almost all the substantive values that also feature in the other RoL indicators (WJP 2014, 8). In addition, it adds a range of features not found in any of the other indicators. The WJP notes that its conceptualization of the RoL represents “an effort to strike a balance between what scholars call a ‘thin’ or minimalist conception of the rule of law that focuses on formal, procedural rules and a ‘thick’ conception that includes substantive characteristics, such as self-government and various fundamental rights and freedoms” (WJP 2014, 5). Despite an effort to strike a balance, the WJP's RoL index arguably adopts the thickest conception of the RoL among the four RoL indicators. But it also has the distinct advantage of being modular: because there are forty-seven different dimensions, scholars can aggregate the data in different ways in accordance with their own definitions of the rule of law.Footnote ⁶

Table 1 offers a more systematic comparison of how the four RoL indicators overlap and differ. Specifically, it offers an overview of all components that are found in at least two of the four RoL indicators. Ten such components were identified. (Appendix 3 provides more detailed information on how these components are operationalized in the indicators in which they appear.) Table 1 reveals that even when we concentrate on the components for which the indicators overlap, their emphases differ substantially.

Table 1 Overview of Similarities and Differences Between Four RoL Indicators

	World Governance Indicators	Heritage Foundation	Freedom House	World Justice Project
Contract enforcement	x	x
Property rights	x	x		x
Corruption	x	x		x
Crime & violence	x		x	x
Judicial independence	x	x	x	x
Judicial efficacy	x	x	x	x
Separation of powers	x			x
Equality	x		x	x
Control over police	x		x
Fundamental rights			x	x
Other	x			x

At first glance, the WGI and WJP indicators appear to have considerable substantive overlap with the other RoL indicators. The WGI RoL indicator overlaps with at least one other indicator on all but one of the ten components (fundamental rights), and the WJP's RoL index overlaps with one or more indicators on all but two components (contract enforcement and control over the police). Importantly, however, both the WGI and WJP RoL indicators add a range of features not found in any of the other indicators. These more idiosyncratic features (listed in more detail under the category “other” in Appendix 3) include the right to petition government, the prevalence of tax evasion, administrative efficiency, the prevalence of human trafficking, the size of the parallel economy, popular observance of the law, and the degree to which official information is available on request. Table 1 also reveals that the Heritage Foundation and Freedom House indicators have a more distinctive focus. The Heritage Foundation's index overlaps with one or more indicators on only five of the ten components, and Freedom House's index overlaps with at least one indicator on six of the ten components. Notably, they are almost exact opposites of each other—Freedom House and the Heritage Foundation only overlap on judicial independence and judicial efficacy. Overall, there are substantial differences in the way the different indicators conceptualize the RoL.

C. Approaches to Measurement

In addition to defining the RoL differently, the four RoL indices use different methodologies to create their country scores. Two of the indicators—the Heritage Foundation and Freedom House—rely almost exclusively on country experts. Freedom House relies on “in-house and external analysts and expert advisers from the academic, think tank, and human rights communities” (Freedom House 2014b). According to Freedom House, “60 analysts and nearly 30 advisors” compiled its 2014 scores (Freedom House 2014b). These analysts and advisors draft country reports based on a wide range of sources, such as “news articles, academic analyses, reports from nongovernmental organizations, and individual professional contacts” (Freedom House 2014b).

The Heritage Foundation also significantly relies on expert assessments to assign country scores. Specifically, the property rights component of the RoL score is based on expert coding done by the Heritage Foundation itself. In constructing its property rights index, the Heritage Foundation relies on a standardized set of sources. For example, it based the 2014 index on (1) the Economist Intelligence Unit, Country Commerce, 2009–2012; (2) the US Department of Commerce, Country Commercial Guide, 2009–2012; (3) the US Department of State, Country Reports on Human Rights Practices, 2009–2012; and (4) various news and magazine articles (Heritage Foundation 2014c). These sources are consulted in the order they are listed, which means that the Economist Intelligence Unit's country commerce reports are the primary source. The corruption component of the Heritage Foundation's RoL index also appears to be primarily drawn from secondary expert-based sources, most notably TI's Corruption Perceptions Index.Footnote ⁷

The WGI indicator is different from the Freedom House and Heritage Foundation measures in that it does not directly rely on experts to assign RoL country scores; rather, it aggregates information from thirty-two existing variables into a single rule of law indicator. To do so, the WGI indicator uses an unobserved components model to create a measure that presumably captures the unobserved level of the rule of law that explains the joint variation in the thirty-two variables included in the model.Footnote ⁸ The variables included in the model vary considerably in their focus, but, according to the WGI's creators, all bear some connection to the RoL. They include measures of personal security and crime, the enforcement of contract and property rights, the prevalence of human trafficking, and the frequency of tax invasion, among many others (for a critique see, Arndt and Oman Reference Arndt and Oman2006, 51; Møller and Skaaning Reference M⊘ller and Skaaning2011, 49). The WGI indicator gives more weight to those variables that fit better within the underlying RoL component, thus reducing the importance of those variables less related to the RoL.

The Heritage Foundation, Freedom House, and WGI all ultimately rely on expert assessments of the RoL (the WGI indirectly relies on expert opinion by using existing variables that are based on expert coding). The WJP is more empirically ambitious than the other three initiatives. In contrast to the other indicators, the WJP supplements expert opinion with original and nationally representative household surveys for each country. Specifically, the WJP explains that its index is “constructed from over five hundred variables drawn from two novel data sources collected by the World Justice Project in each country: (1) a general population poll (GPP) conducted by leading local polling companies using a representative sample of 1000 respondents in the three largest cities; and (2) qualified respondents' questionnaires (QRQ) consisting of close-ended questions completed by in-country practitioners and academics with expertise in civil and commercial law, criminal justice, labor law and public health.” One might expect that this comprehensive approach would make the WJP's index significantly different from the other RoL indicators that rely only on expert judgment. However, it appears that the population-based questions ultimately play a small part in the overall indicator. We return to this issue in Section V.

Table 2 Pair-Wise Correlations of Four RoL Indicators

	Heritage	WGI	WJP	Freedom House	Factor Loading on Factor 1
Heritage	—	0.958	0.953	0.769	0.97
WGI	0.958	—	0.958	0.819	0.98
WJP	0.953	0.967	—	0.834	0.98
Freedom House	0.768	0.819	0.834	—	0.90

A. The RoL and Democracy

A first related but distinct concept to which we compare the RoL indicators is democracy. RoL theorists do not generally suggest that democracy is part of the RoL, and none of the four RoL indices explicitly incorporate democracy into their definitions of the RoL. Indeed, it has been suggested that the RoL is frequently in tension with democracy (Tamanaha Reference Tamanaha2008). At the same time, some of the indicators include institutional features, such as checks and balances and judicial independence, which are closely related to democracy (and sometimes even considered to be part of it). To explore the relationship between the RoL and democracy, we used the polity2 variable from the Polity IV data project (Marshall, Gurr, and Jaggers Reference Marshall, Gurr and Jaggers2014). This variable combines information on the competitiveness and openness of executive recruitment, constraints on the executive, regulation of political participation, and the competitiveness of political participation into a single numerical indicator that ranges from −10 (most autocratic) to 10 (most democratic) (Marshall, Gurr, and Jaggers Reference Marshall, Gurr and Jaggers2014, 15–17). The polity2 indicator is widely used in the quantitative political science literature.

Table 3 lists the pair-wise correlations of each of the four RoL indicators with democracy. It also lists the correlation between democracy and countries' factor scores on the first factor as obtained through our principal component analysis. The results in Table 3 reveal that the correlation between democracy and the RoL is consistently positive (ranging between 0.45 and 0.76), but not nearly high enough to conclude that the democracy and RoL indices capture the same thing. The indicator that is most strongly correlated with democracy is that of Freedom House (correlation 0.76). This is unsurprising considering that Freedom House's primary mission is to promote democracy around the world, which plausibly affects its conceptualization and assessment of the RoL.Footnote ¹⁰

Table 3 Relationship of RoL with Neighboring Concepts

	Heritage	WGI	WJP	Freedom House	Scores on Factor 1
Democracy (Polity IV, 2010 values)	0.46	0.45	0.45	0.76	0.52
Human rights (Fariss, 2010 values)	0.63	0.71	0.81	0.73	0.80
Constitutionalism (Law and Versteeg, 2009 values)	0.70	0.74	0.80	0.83	0.82
Judicial independence (CIRI, 2010 values)	0.67	0.73	0.79	0.77	0.83
GDP per capita (World Development Indicators, 2010 values)	0.77	0.81	0.81	0.61	0.80
Corruption (Transparency International, 2011 values)	0.98	0.95	0.95	0.81	0.97
Impartial government (Teorell, Dahlstrom, and Dahlberg Reference Teorell, Dahlström and Dahlberg2011)	0.89	0.86	0.90	0.80	0.90

B. The RoL and Human Rights

A second concept that is related to, but distinct from, the RoL is human rights. Although human rights are not part of Lon Fuller's procedural definition of the RoL, subsequent theorists have often suggested that respect for basic rights is a crucial component of the RoL. Indeed, both the WJP and Freedom House indicators explicitly include human rights in their definitions of the RoL. To compare the four RoL indicators with human rights, we rely on a new indicator created by Christopher Fariss, which is currently considered the gold standard in the quantitative human rights literature (Chilton and Versteeg forthcoming). Fariss's measure combines information from multiple existing human rights indicators, which are primarily based on the quantitative coding of the US State Department's and Amnesty International's annual country reports on human rights. In contrast to these other human rights indicators, Fariss's data are corrected to account for changes in the way the US State Department and Amnesty International report information on human rights abuses. His resulting indicator captures respect for basic physical integrity rights, such as the absence of torture, arbitrary arrests and detention, and extrajudicial killings (Fariss Reference Fariss2014, 297–318).Footnote ¹¹

Table 3 lists the pair-wise correlations between Fariss's human rights indicator, the RoL indicators, and countries' factor scores on the first factor. It reveals that the RoL is more closely related to human rights than democracy, but that the RoL and human rights indicators are not identical. The WJP's RoL index (0.81) is the most closely related to human rights, followed by Freedom House (0.73), the World Bank (0.71) and the Heritage Foundation (0.63). The fact that the WJP and Freedom House indices are most strongly correlated with RoL is not surprising considering that they explicitly build respect for human rights into their definitions of the RoL.

C. The RoL and Constitutionalism

A third concept related to the RoL is constitutionalism. Constitutionalism is perhaps as complicated a concept as the RoL. Many theorists have imported various substantive values into the idea of constitutionalism. In a thin version of the concept, though, constitutionalism is arguably just the degree to which a country upholds its constitutional promises, regardless of what these promises are and whether other promises should have been included (Hardin Reference Hardin, Galligan and Versteeg2013). Such a thin version of constitutionalism shares some similarities with Lon Fuller's conception of the RoL, especially his notion of congruence. One measure that arguably captures the thin version of constitutionalism is Law and Versteeg's Constitutional Performance Index. Their index captures how many out of fifteen possible rights that are promised in the constitution are actually upheld in practice (Law and Versteeg Reference Law and Versteeg2013, 863–951). It does so by contrasting the presence of fifteen de jure rights provisions with the actual realization of these same fifteen rights de facto. The result is a measure of constitutional congruence, which captures the proportion of rights promised in the constitution that is upheld in practice (Law and Versteeg Reference Law and Versteeg2013).

Table 3 presents the correlations between the Constitutional Performance Index, the four RoL indicators, and the factor scores on the first factor. It reveals that constitutionalism is fairly closely correlated with RoL (with the correlations roughly in the same range as the human rights indicator, ranging between 0.70 and 0.83), though not close enough for the two to be characterized as capturing the same thing. The closest correlation between constitutionalism and the RoL is with the Freedom House index (0.83), which is perhaps not surprising given that Freedom House's definition emphasizes judicial independence and respect for rights more than the other indicators.

D. The RoL and Judicial Independence

A fourth concept that is related to the RoL is judicial independence. As noted, judicial independence was not part of Lon Fuller's definition of the RoL, but other theorists in the procedural tradition of the RoL have subsequently added it to the list of procedural/functional conditions for the RoL. To compare the RoL indices with judicial independence, we rely on a measure by Cingranelli and Richards, which is based on the coding of US State Department's annual country reports on human rights and rates judicial independence on a three-point scale.

Table 3 presents the correlations between the judicial independence index, the four RoL indicators, and the factor scores on the first factor. The results reveal that the RoL indices are related to judicial independence, but that they do not capture the same thing. The WJP and Freedom House indicators have the highest correlation with judicial independence (0.79 and 0.77, respectively), which is unsurprising because both include judicial independence in their definitions of the RoL.

E. The RoL and GDP per Capita

We also explored the relationship between the RoL and (logged) GDP per capita. GDP per capita is not a neighboring concept to the RoL—it bears no relationship to Lon Fuller's criteria. Instead, it might be more accurate to think of GDP per capita as a possible predictor of the RoL (and the determinants of RoL are a whole separate research question, beyond the scope of this article [Gutmann and Voigt Reference Gutmann and Voigt2014]). We nonetheless include GDP per capita because Kurtz and Schrank (Reference Kurtz and Schrank2007) have shown that the WGI's governance indicators are influenced by coders' perceptions of economic growth. They claim that countries that have experienced recent economic growth are coded as possessing high-quality government institutions, a phenomenon that has been described as the halo effect (Kaufmann, Kraay, and Mastruzzi Reference Kaufmann, Kraay and Mastruzzi2010). Including GDP per capita in our analysis, therefore, has the advantage that it allows us to explore the possibility of halo effects directly.

Table 3 presents the correlations between a logged GDP per capita indicator, the four RoL indicators, and the factor scores on the first factor. The results reveal that richer countries have higher RoL scores, but that GDP and RoL are not identical. The correlation is highest for WJP and WGI (0.81). The emphasis on economic development is perhaps unsurprising for the World Bank, whose primary goal is to promote economic development, but is more surprising for the WJP, which was established for the sole purpose of measuring RoL.Footnote ¹²

F. The RoL and the Absence of Corruption

A final concept that is related to the RoL is corruption, commonly defined as the use of public power for private gain. Notably, three of our indicators explicitly add the absence of corruption to their definitions of the RoL, but even without explicitly adding corruption to the definition of RoL, the widespread prevalence of corruption in a country arguably violates Fuller's criteria of generality, publicity, consistency, and congruence. A government that is corrupt is by definition not limited by law. This is true even if the source of abuse is the private gain of rulers rather than an oppressive political program. Corruption and the RoL, therefore, are closely related concepts (Haggard, MacIntyre, and Tiede Reference Haggard, MacIntyre and Tiede2008, 210–11).

To capture corruption we rely on TI's Corruption Perceptions Index. According to TI, corruption “generally comprises illegal activities, which are deliberately hidden and only come to light through scandals, investigations or prosecutions” (Transparency International 2013b). Because corruption is hidden from the public eye, the Corruption Perceptions Index captures perceptions of corruption and does not try to measure corruption directly. To that end, it relies on thirteen secondary expert-based sources, including the Economist Intelligence Unit, Freedom House, the Political Risk Services International Country Risk Guide, the World Bank, the International Monetary Fund, and the corruption component from the WJP (Transparency International 2013a).

The TI corruption index and the four RoL indicators turn out to be nearly identical to each other. Table 3 presents the correlations between the Corruption Perceptions Index, each of the four RoL indicators, and the factor scores on the first factor. It shows that the RoL indicators by the Heritage Foundation, the World Bank, and the WJP are nearly identical to the Corruption Perceptions Index. The correlation between the Corruption Perception Index and the Heritage Foundation's RoL index is a whopping 0.98, and the correlation with the World Bank and WJP RoL indices is 0.95. The correlation with the factor scores on the first factor is 0.97. Indeed, when we add all the neighboring concepts as explanatory variables in a simple cross-sectional OLS regression model where the dependent variable is the factor scores on the first dimension, corruption is the only statistically significant predictor of the RoL at the 1 percent level. In other words, after controlling for democracy, human rights, constitutionalism, judicial independence, and GDP per capita, corruption uniquely explains the remaining variation in the RoL indices. Together, these findings suggest that whatever is discrete about the RoL indices is also captured by TI's Corruption Perception Index.

Overall, these findings suggest that the RoL indicators do capture something discrete, which means that their convergence is not driven by conceptualizations that conflate the RoL with neighboring concepts or halo effects. The notable exception for corruption, however, leaves us with another puzzle. Do the RoL indicators simply capture government corruption? Is absence of corruption the same thing as respect for the RoL?

In tackling this puzzle, it deserves mentioning that the three RoL indicators that are highly correlated with the TI index all build corruption into their definitions of the RoL. In the case of the Heritage Foundation, no less than half the indicator consists of corruption data drawn from TI, which means that the correlation is driven by identical data. The same is not true, however, of the WJP and WGI indicators.

This raises the question of whether corruption and the RoL are theoretically distinct. Importantly, conceptual connections appear to exist between the two. Although the classic discussions of the RoL focus on the risk of government arbitrariness and oppression, they do not specify the motives that might prompt such pathologies. Corruption, the misuse of public power for private gain, is surely one such motive. Thus, corruption might be an important explanation, if not the primary explanation, for the absence of the RoL. The result might be that, once operationalized in indicators, corruption and the RoL are indistinguishable from each other. Indeed, previous studies have also noted the close empirical link between RoL indicators and corruption indicators (Haggard, MacIntyre, and Tiede Reference Haggard, MacIntyre and Tiede2008, 222; Haggard and Tiede Reference Haggard and Tiede2011, 677).

Another possibility is that the indicators capture a more abstract concept that transcends the specifics of both corruption and the RoL. One such candidate is impartial government (as developed most prominently by Bo Rothstein Reference Rothstein2011, Reference Rothstein2014). According to Rothstein, both corruption and the RoL fit into a broader umbrella concept of impartiality (Rothstein Reference Rothstein2011). First, the idea of impartial government rules out all forms of corruption and several forms of particularism, including clientelism, patronage, and nepotism (Rothstein Reference Rothstein2011, 14). Second, government impartiality implies the presence of the RoL. Since laws are the most general rules of society, an impartial application of the rules requires impartial application of law, a core feature of the thin definition of the RoL (Weingast Reference Weingast1997, 245; Rothstein Reference Rothstein2011, 28). Thus, government impartiality might be a higher-order concept that connects corruption and the RoL and, accordingly, might constitute the essence of both indicators.

Teorell, Dahlstrom, and Dahlberg (Reference Teorell, Dahlström and Dahlberg2011) have attempted to measure Rothstein's notion of impartiality through a “low-cost web survey with public administration scholars” (Rothstein Reference Rothstein2011, 32). If the RoL and corruption ultimately capture impartiality, we would expect the impartiality survey questions to correlate highly with both corruption and the RoL. The last row of Table 3 presents the correlation of the various RoL measures with Rothstein's impartiality data. It shows that the correlations are indeed quite high: ranging from 0.80 for Freedom House to 0.90 for the WJP. The correlation between Rothstein's impartiality measure and TI's Corruption Perceptions Index is 0.87. These results suggest that impartiality might be the core concept captured by these indicators, especially since the impartiality data are merely a shoestring effort to capture the concept (Rothstein Reference Rothstein2014).

But we are left with another puzzle. Rothstein's concept is procedural, not substantive. That is, it captures whether government officials treat all people the same. It does not judge the substance of the treatment—whether the rules that are impartially applied are also normatively appealing. Impartiality is a thin concept, not a thick one. Thus, if the RoL indicators ultimately capture impartiality, it means that their stark differences in conceptualization (including the substantive values that they build into the RoL) fade away. Thicker conceptualizations do not matter.