Normative Models of Democratic Governance 2

CENTRIPETALISM AS AN EMPIRICAL CONCEPT The centripetal model is evidently closer to the centralist model than to the decentralist model, as the term itself suggests. Even so, it is distinctive. In this section we explore centripetalism as an empirical concept. Which polities are centripetal? How centripetal are they? The centripetal framework replicates certain features that are considered typical of the political landscape in Continental Europe. Thus, while contemporary United States serves as the paradigm-case for decentralism, and contemporary United Kingdom the paradigm-case for centralism, contemporary Sweden may offer the best exemplar of centripetalism. Are there also countries in the developing world that exhibit centripetal tendencies? It is not possible to explore all twenty-one dimensions listed in Table 1 on a global scale. Indeed, some of these dimensions are almost impossible to measure, even within the restrictive confines of the OECD. Even so, it may be possible to take account of those elements that are most important in establishing the governance styles and governance effectiveness of polities across the democratic world. In this section we focus on those elements of a country’s institutional profile that are measurable, exogenous (relative to other political institutions), and of greatest presumed importance to politics and policymaking. We shall refer to these factors as constitutional. They include the first four dimensions listed in Table 1: a) territorial sovereignty, b) the legislative branch, c) the executive, and d) the electoral system. Because the first two factors are closely related, both theoretically and empirically, we reduce this set to three constitutional factors. The theory of centripetalism stipulates that constitutional institutions should be unitary, parliamentary, and PR. Unitarism refers to a polity that has a central locus of sovereignty; it is both non-federal and non-bicameral — or, if there are two elective chambers, these houses are either asymmetrical or congruent (thus diluting the importance of a second chamber). Parliamentarism refers to a polity where the government is chosen by, and responsible to, the legislature; it is nonpresidential. PR refers here to an electoral system where districts are multimember and voters choose among closed party lists. In this section we show why these institutions exemplify the centripetal ideal and how they should be operationalized (according to the centripetal ideal). Thus, we address both the theoretical justification of these hypotheses as well as their empirical elaboration. Recall that the key to good government, according to the centripetal model, is to be found in institutions that successfully combine authority and inclusion within a democratic setting. Institutions must reach out to all interests, ideas, and identities (at least insofar as they are relevant to the issue at hand). And they must provide an effective mechanism for reaching agreement and implementing that agreement. This is the process of gathering-together that, we hypothesize, culminates in good government. UNITARISM Centripetalism suggests a unitary, rather than federal, form of government. Unitary rule preserves the constitutional authority of central government, while leaving open the possibility of considerable decentralization of power. Thus, when we refer to a polity as unitary we are saying that constitutional authority — sovereignty — is vested in the central (national) government, not that all decision-making occurs at the center. The crucial distinction is that power delegated from national to subnational bodies may be retrieved. In a federal system, by contrast, subnational authorities enjoy constitutional status; their power is inherent rather than delegated. (In referring to subnational authorities in this context we are referring to regional governmental bodies — states, territories, provinces, Laender — not local bodies.) The question of inclusion, the second dimension of the centripetal theory, is more complicated. Some writers argue that federal regimes are more inclusionary. However, it has also been suggested that unitary constitutions are no less inclusive than federal constitutions, and perhaps more so. Since this is a long argument, we do not dilate on the matter here. Suffice it to say that the centripetal pull toward the center of a political system is more likely to be felt in a unitary than in a federal polity. Centripetalism presupposes unitarism. We conceptualize unitarism along two dimensions: a) the degree of separation (independence) between national and territorial units, and — if any separation at all — b) the relative power of the two players (the more power the center possesses the more unitary the system). Of the many institutional factors that determine variation along these dimensions, two predominate: federalism and bicameralism. We therefore operationalize unitarism as an additive variable with two components: non-federalism and non-bicameralism, as discussed below. We understand federalism as an institutionalized division or sharing of responsibilities between a national authority and semi-autonomous regional units. Since this sharing of responsibilities takes a variety of forms and is not always formally prescribed (or is ambiguous in formal-constitutional terms), we utilize three coding categories for our nonfederalism variable: 0 = federal (elective regional legislatures plus constitutional recognition of subnational authority), 1 = semi-federal (where there are elective legislatures at the regional level but in which constitutional sovereignty is reserved to the national government), and 2 = nonfederal.48 In order to gauge the strength of unitarism we must also examine the status of territorial units within the national government. Insofar as territorial units receive special representation in 48 Principal sources employed in coding: Alvarez (1999), Derbyshire and Derbyshire (1996), Elazar (1991), Hicken and Kasuya (2001), McHenry (1997), Watts (1997), The Database of Political Institutions (Beck et al. 2000), The Political Reference Almanac (polisci.com). the national legislature — different from what would be allocated by every ballot equally — we shall consider this a violation of the principle of unitarism. In practice, wherever this special consideration obtains it is found in second (‘upper’) chambers. Thus, the second feature of unitarism concerns bicameralism, the sharing of policymaking power between two chambers at the national level. Bicameralism, like unitarism itself, must be understood across two dimensions: a) the relative power of the two bodies (symmetrical if they are roughly equal in power, asymmetrical if the lower house dominates) and b) the composition of the two bodies (congruent if the partisan distribution is roughly the same, incongruent if it is different).49 Since, like federalism, bicameralism is often a matter of degrees, we code nonbicameralism according to the predicted degree of asymmetry and incongruence: 0 = strong bicameral (upper house has some effective veto power; the two houses are incongruent), 1 = weak bicameral (upper house has some effective veto power, though not necessarily a formal veto; the two houses are congruent), 2 = unicameral (no upper house or weak upper house).50 We construct the variable unitarism by adding the scores of each country together on these two components: nonfederalism and nonbicameralism. The combination of these two dimensions is justified by the fact that they are both empirically linked (constitutional federalism is a necessary condition for strong bicameralism) and conceptually linked (the purpose of a strong second chamber is usually to protect the powers and prerogatives of subnational units).51 In a fully unitary state, territorial units (if any) have no constitutional standing, no independently elected territorial legislature, no specific policy purviews reserved to them, and minimal revenueraising authority. PARLIAMENTARISM Centripetalism implies a parliamentary, rather than presidential, system of government. Parliamentarism is a more centralized system of rule, since lawmaking power is not divided between two separately elected (and often incongruent) authorities. Parliamentary systems, we argue, are also more inclusive. The most important policymaking bodies are collegial rather than 49 Sometimes the degree of congruence is measured directly; more frequently, it is inferred by the electoral systems, electoral districts, electoral timing, and term lengths that apply to the two chambers. 50 Principal sources employed in coding: Hicken and Kasuya (2001), Patterson and Mughan (1999), Tsebelis and Money (1997), The Political Reference Almanac (polisci.com). 51 It is important to clarify that unitary government, as we use the term here, is quite different from administrative or fiscal centralization. The latter refers to a particular arrangement of powers and responsibilities between national and subnational units, involving issues such as whether revenue generation is decentralized, whether there are hard budgetary constraints on sub-national units, whether there are clear lines of authority separating national and sub-national responsibilities, whether local institutions are democratically run, and whether effective evaluative procedures are available (Bird and Vaillancourt 1998: 12-5; Burki et al. 1999; Fisman and Gatti 2000; Huther and Shah 1998; Oates 1972; Ter-Minassian 1997; Weingast 1995). Unitary governments may (and often do) institute fiscal-federal policies. individualistic; additionally, the incentive structures instituted by a parliamentary system encourage cooperation rather than defection, voice rather than exit. Thus, centripetal energy is more likely to be generated by a parliamentary constitution than by a presidential (‘separate powers’) constitution. We refer to parliamentarism as a system of government in which the executive (the prime minister and cabinet: collectively, ‘the government’) is chosen by, and responsible to, an elective body (the legislature), thus creating a single locus of sovereignty at the national level.52 Presidentialism, its contrary, is understood as a system where policymaking power is divided between two separately elected bodies, the legislature and the president. The president’s selection is usually by direct popular election, though it may be filtered through an electoral college (as in the US), and the rules pertaining to victory (i.e., by relative or absolute majority) vary from country to country. His (or her) tenure cannot be foreshortened by parliament except in cases of gross malfeasance. S/he is actively engaged in the making of public policy, and in this sense plays a political (i.e., partisan) role. In practice, between these two polar types we find many admixtures, known generically as ‘semi-presidential’ systems. Thus, we conceptualize the parliamentary/presidential distinction as a continuum with two dimensions: a) the degree of separation (independence) between president and parliament (unity = parliamentary, separation = presidential), and, if there is any separation at all, b) the relative power of the two players (the more power the president possesses the more presidential is the resulting system). We capture this complex reality with a three-part coding scheme: 0 = presidential, 1 = semi-presidential, 2 = parliamentary.53 Some notes on coding are in order. If a directly elected president exists but has no effective policymaking power, as in Iceland and Ireland, we consider the regime to be parliamentary. If a president is chosen by the legislature but enjoys a fixed term of office (cannot be removed by the legislature except in cases of gross malfeasance) and significant policymaking powers, as in Bolivia, we consider the regime to be semi-presidential. If a president is chosen by popular election, enjoys significant policymaking power, but must share power with a prime minister chosen by the legislature then we also code the system as semi-presidential. This is the most common form of semi-presidentialism, and is currently found in France, Lithuania, Poland, Russia, and the Ukraine. 52 The precise terms used to refer to these institutions vary somewhat from country to country. Sometimes, the prime minister is a ‘chancellor’ or even a ‘president.’ The important point is that s/he is chosen by, and responsible to, the legislature. 53 Principal sources employed in coding: Alvarez (1999), Delury (1999), Derbyshire and Derbyshire (1996), Diamond (1999), Golder (2003b), Hicken and Kasuya (2001), International Year Book and Statesmen’s Who’s Who (2001), Jones (1995), The Database of Political Institutions (Beck et al. 2000), The Political Reference Almanac (polisci.com), Nohlen et al. (1999, 2002). For discussion of semi-presidentialism, see Elgie (1997, PR The centripetal theory of democratic governance suggests that electoral systems, like other constitutional elements of a polity, should maximize the twin desiderata of authority and inclusion. These twin goals are best achieved when an electoral system encourages strong national parties while also maintaining low barriers to entry for new parties, strong competition among existing parties, and demographically diverse party delegations. This, in turn, mandates an electoral system that privileges inter-party choice and intra-party representation over intra-party electoral choice. Voters vote, and parties nominate. Further, the vote choice itself should be based on national, partisan principles rather than preferences for individual candidates or districtlevel concerns. (Insofar as ‘personality’ matters it should be the personality of the party leader, not the district-level candidate, that matters.) Empirically, three features of an electoral system bear critically on these issues: a) district magnitude (M), b) seat allocation rules (majoritarian or proportional), and c) candidate selection rules. The centripetal ideal-type is defined by M>1, proportional seat allocation rules, and partycontrolled candidate selection. This is the familiar list-PR electoral system— ‘PR’ for short. Let us begin with the issue of district magnitude (M). Proportional representation refers generically to an electoral system in which there is a proportional relationship between voting preferences and the allocation of seats in a legislature. To be sure, any representational system has some lower limit below which preferences cannot be translated into seats. Even if there are no statutory thresholds a de facto limit is provided by the number of seats in a legislature and by the size of the district. Thus, if a legislature has 200 members and is chosen in a single, nationwide district (M=200) then the effective threshold is roughly 1/200th of the effective electorate, even under the most proportional rules of seat allocation. We do not expect to find a perfectly linear relationship between proportionality and good governance. There is no reason to suppose, for example, that Israel and Denmark (two of the most proportional systems in the world) should experience better governance than Sweden or Spain (countries with moderately proportional electoral systems) by virtue of having more proportional seat/vote ratios. Thus, we approach that the question of proportionality through a simple, three-part coding system that recognizes important categorical differences among majoritarian systems, mixed-member majoritarian (MMM) systems (combining single-member and multi-member districts in parallel [noncompensatory] lists, as in Russia and Mexico), and proportional electoral systems (M>2 [binomial electoral systems are not considered to be proportional]). Mixed-member proportional (MMP) electoral systems, where extra seats are 1999), Nousiainen (1988, 2001), Shugart and Carey (1992), Skach (forthcoming). designed to rectify non-proportional outcomes achieved by single-member district contests (e.g., Bolivia, Germany, New Zealand, Venezuela), are classified as PR since the principle of proportionality is preserved for the system as a whole.54 Majoritarian refers here to all single-member district electoral systems, whether they employ first past the post (plurality) rules, as in the well-known Westminster system, or majority (double-ballot) rules, as in France and many former French colonies. (Block vote systems are majoritarian at the inter-party level, but not at the intra-party level, as discussed below.) Usually, a higher district magnitude offers greater choice to the elector among parties (the exception is the block vote system, discussed below). Indeed, district magnitude is the single most important factor in determining how many effective inter-party choices — choices that might culminate in electing a representative to parliament — voters will have at their disposal as they enter the ballot booth. The question of intra-party choice is somewhat more complicated. Usually, where district magnitude is equal to one, constituency members will be able to determine the nomination of candidates (perhaps with some vetting from central party headquarters), and hence exercise a degree of intra-party choice. Of course, voters in the general electorate will be prohibited from exercising any choice in this matter, under most circumstances. Where district magnitude is greater than one (M>1) it is common for candidate selection to be decided by party leaders (either at regional or national levels), with less input from the constituency parties. Thus, all things being equal, district magnitude is inversely correlated with intra-party choice. There are of course other features of an electoral system that may affect the degree to which the general electorate intervenes in intra-party choices (candidate selection). They include a) mandatory write-in ballots (e.g., the Philippines); b) multiple lists emanating from factions of the same party (e.g., Colombia, Uruguay); c) the single-transferable vote (STV); d) the alternate vote (AV); e) the single-non-transferable vote (SNTV), f) ‘open-list’ systems of PR where only preferential votes determine the order of candidates on a party’s list (as in pre-reform Italy) or where preferential voting is mandatory.55 Wherever electoral systems employ one of these preferential voting options, which we shall refer to as strong preferential voting (to distinguish them from the more usual, and less consequential, forms of preferential voting), we code such countries as ‘0’ on our three-part coding scheme. It is important to note that this re-coding affects only a handful of countries at various points in the postwar period: Australia, Brazil, Chile, Colombia, Estonia, Finland, Greece, Ireland, Italy, Japan, Jordan, Luxembourg, Malta, Nauru, Papua 54 M refers to the average (mean) district magnitude in a legislature. On MMM and MMP systems, see Shugart and Wattenberg (2001). 55 In the latter case, we assume that in order for intra-party preference voting to upstage party leaders in the selection of candidates there must be no default candidate ranking on the ballot that voters might fall back New Guinea, Peru, the Philippines, Poland, Sri Lanka, Switzerland, Taiwan, Uruguay, and Vanuatu. (It should also be noted that many of the countries listed above are, by virtue of missing data or authoritarian rule, already excluded from our statistical analyses.)56 Another important deviation from pure PR or pure majoritarianism consists of a relatively rare electoral system known as the block vote, employed in Bermuda, Djibouti, Ecuador, Jordan, Kuwait, Laos, Lebanon, the Maldives, Mali, Mauritius, Mongolia, the Philippines, Senegal, Singapore, Thailand, Tunisia, and the Virgin Islands (US) at various points in time.57 Here M>1 but seats are allocated on a winner-take-all basis, so that the party winning a plurality captures all seats in a district. Electors cast either as many votes as there are seats available (and may be allowed to cross lists) or a single vote for a party (in effect, a party-list vote).58 Note that in order to qualify as a block vote system in our typology the electoral system must be based on party lists (even if voters are allowed to exercise a preferential vote); if ballots list only individual candidates (without party affiliation) then it will be classified as a strong-preferential voting system. The block vote system is even more majoritarian than the single-member district system insofar as it adopts the winner take all electoral principle in multimember districts. However, political parties, rather than individual candidates, compete against each other. This makes it easy for party leaders to include members of key minority groups on their lists. Usually, this system is considerably more inclusive than an electoral system based on single-member districts because parties have incentives to present diverse lists, including all significant minority social groups within a constituency, and because the electoral dynamic itself is apt to be somewhat less confrontational.59 It also seems to uphold the strength and cohesion of party leadership, since candidate selection is still, for all practical purposes, a party monopoly (even where open-list voting is possible). Thus, we code block vote systems ‘1’ on our three-part scale. A final deviation from pure PR consists of polities where a majority of seats in the legislature (we refer, as always, to the lower or dominant legislature, if bicameral) are reserved for a particular social or ethnic group if such seat reservations (‘communal rolls’) involve a significant departure from the one-person/one-vote principle (i.e., if it entails significant malapportionment between the designated groups). In such cases (e.g., Fiji), PR = 0. We do not introduce other forms of malapportionment into our coding category, since these deviations generally have less severe consequences (though naturally, they do have some consequences). on as a voting cue (lists are ‘unranked’). 56 For work on intra-party preference voting and its variants (e.g., STV, SNTV) see Bowler and Grofman (2000), Grofman et al (1999), Katz (1986a), Shugart et al. (2003). 57 We exclude Fiji from this list since legislative seats are chosen from separate ethnic lists (Lal and Larmour 1997). 58 Sometimes, the single-vote system is called a ‘party block vote’ to distinguish it from the more common block vote system in which voters have as many votes as there are seats. To conclude, although the term proportional representation, strictly interpreted, refers to the proportionality in the relationship between votes cast and seats allocated, we appropriate its abbreviation (‘PR’) to refer to a distinction among electoral systems around the world that gives pride of place to list-PR, the most common form of proportional representation. Other systems are ranked lower in this coding according to their deviation from the principles articulated at the outset: high district magnitude (M), proportional seat allocation rules, and closed candidate selection rules. Thus, the coding for the PR variable is revised as follows: 0 = majoritarian or preferential-vote, 1 = mixed-member majority (MMM) or block vote, and 2 = list-PR.60 SUMMARY Thus far, we have treated the theory of centripetalism as a descriptive phenomenon. However, our theoretical interest is in the causal role of centripetal institutions on good governance. Do (democratic) countries with centripetal institutions adopt better policies and enjoy better policy outcomes than countries with centralized or decentralized institutions? In testing the centripetal hypothesis, we suppose that it takes time for institutions to exert an appreciable effect on governance outcomes. A country switching from a presidential to parliamentary system (or establishing a parliamentary system in a newly democratic or independent setting) should not expect to see immediate, dramatic changes in the quality of governance. Instead, these effects are likely to cumulate over time as the new institutional rules begin to affect actions and expectations (Gerring et al, in process). To represent this historical component empirically we create a new variable that draws upon the annual (“raw”) scores for Centripetalism described above. We assume that history matters (i.e., a country’s long-run experience with constitutional institutions is what helps drive governance outcomes), but that recent history matters more. To this end, we calculate a moving, weighted sum of the annual scores of each of our underlying explanatory factors, beginning in 1901 and ending in the observation year. We construct the weights to capture long-term historical patterns while giving greater weight to more recent years. A country’s score in 1996 would be the weighted sum of its scores from 1901 to 59 On the Mauritius experience see Mathur (1997a, 1997b). On Mali see Vengroff (1997). Countries are excluded from this classification if a majority of seats in the lower, or most powerful, house are appointed or are elected through a system of reserved seats, as in Fiji and Hong Kong. Note that where coding principles conflict, the lower coding prevails. Thus, although the Philippines employs an MMP electoral system, it also employs mandatory write-in ballots, a provision that classifies the electoral system as strong preferential vote. Hence, the Philippines is coded as 0, rather than 1, in the current period. Principal sources employed for coding: Golder (2003b), Hicken and Kasuya (2001), Massicote and Blais (1999), Reynolds and Reilly (1997), Shvetsova (1999), the EPIC Project. 60 1996. Its score in 1998 would be the weighted sum of its scores from 1901 to 1998, and so on.61 Country-years are counted in this weighted summation process (and in the empirical analysis to follow) so long as a country surpassed a minimum threshold of democracy during that year. (Centripetalism is a theory of democratic governance; it has no application within authoritarian settings.) We employ a relatively low threshold of democracy because we wish to include as many plausible cases as possible in our analysis and because the logic of centripetalism should be operative so long as there is a modicum of multi-party competition. Thus, we include a countryyear in our analysis so long as it obtains a score greater than zero — on a scale that ranges from –10 to 10 — on the Polity2 measure of democracy.62 Our intent here is not to define democracy per se, but rather to establish the empirical boundaries of the study. To the extent that our definition is too lenient, we expect that it biases the results against our hypotheses, since we anticipate that centripetal institutions will be less effective in less democratic settings. To this point, we have described the coding of three constitutional components of the centripetal theory: Unitarism, Parliamentarism, and PR. Of course, our primary theoretical interest is in their combined effect. Thus, we create a final composite score, Centripetalism, to capture the universe of democratic polities in a single measure. This is a simple sum of three components — unitarism, parliamentarism and PR — each converted to standardized units with a mean of zero and a standard deviation of one to ensure similar units and thus equal weighting when summed together to create the new measure. We employ the same historical weighting technique described above to construct our explanatory Centripetalism variable. Table 2 provides descriptive statistics and correlations for our weighted sum variables for the years used in the empirical analysis to follow. Figure 1 shows the distribution of cases for our weighted sum Centripetalism variable in 2000. A complete list of country-cases in 2000 meeting our minimal definition of democracy (N=124) and ranked by their weighted sum Centripetalism scores, along with their annual (raw) scores on all four variables, can be found in Table 3.63 We hypothesize that countries at the top of this list have better political institutions and should therefore experience better governance outcomes, ceteris paribus. The following section tests this argument empirically. 61 See Gerring and Thacker (forthcoming) for details on the construction of the weighting scheme. Marshall and Jaggers (2002). Because the Polity2 democracy score does not contain data for several countries (mostly micro-states), we impute missing values using the following alternative measures of democracy: the Freedom House Political Rights indicator (Gastil, various years), Bollen’s (1993) Liberal Democracy variable, Vanhanen’s (1990) Competition measure, and Banks’s (1994) Legislative Effectiveness I and II and Party Legitimacy variables. 63 Note that a recently democratized polity, even if fully centripetal in its current institutions, would not be high on this list because of the historically weighted and summed measures used to generate this variable. 62 Table 2: Descriptive Statistics for Cumulative, Weighted Sum Scores Moments: 1996, 1998, 2000, 2002 Unitarism (Unit) Parliamentarism (Parl) List-PR (PR) Centripetalism N Mean Std Dev 373 62.3 53.2 373 30.6 32.6 373 16.0 26.3 373 0.01 2.57 Min Max 0 200 0 100 0 100 -2.88 8.19 Correlations: 1996, 1998, 2000, 2002 Parl PR Unit Parl 0.69*** 0.44*** 0.29*** *** Sig. at .0001 level (two-tailed test) N=373 Figure 1: Centripetalism: The Distribution of Cases (2000) 40 35 25 20 15 10 5 e or M 7 6 5 4 3 2 1 0 -1 -2 0 -3 Frequency 30 Table 3: Democratic Polities of the World, 2000 Rank 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 24 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 48 50 51 51 53 54 55 56 57 58 59 60 61 62 Country Denmark Sweden Iceland Norway Netherlands Belgium Austria Israel Costa Rica New Zealand UK Finland Luxembourg Ireland Turkey Germany Japan Greece South Africa Portugal Canada Spain Mauritius Lichtenstein San Marino France Sri Lanka Italy Cyprus (G) Malta Botswana Jamaica Guatemala Trin. & Tob. W. Samoa Barbados Ecuador Nauru Thailand Australia Honduras Guyana Fiji Czech Rep. Bahamas Bolivia Malaysia Dominica Tuvalu South Korea St. Lucia St. Vin. & G. PNG Estonia Venezuela El Salvador India Slovak Rep. Suriname Uruguay Grenada Vanuatu Unit 4 4 4 3 3 1 3 4 4 4 4 4 4 4 4 0 3 4 1 4 2 2 4 4 4 3 3 2 4 4 4 3 4 3 3 3 3 4 3 0 4 4 3 3 3 1 1 4 4 4 4 4 3 4 2 4 0 4 4 4 3 3 Contemporary Parl PR 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 0 2 2 2 2 0 2 0 2 0 2 0 2 2 2 2 2 1 2 0 2 2 2 2 2 0 2 2 2 1 2 2 2 2 1 0 1 0 2 1 0 2 2 0 2 0 2 0 0 2 2 0 2 0 2 0 0 0 2 0 2 1 2 0 0 2 0 2 2 0 2 2 2 0 1 2 2 0 2 0 2 0 1 1 2 0 2 0 2 0 2 2 0 2 0 2 2 0 2 2 1 2 0 0 2 0 2 0 Cent 2.95 2.95 2.95 2.19 2.19 0.66 2.19 2.95 0.73 2.95 0.82 0.82 0.82 0.82 2.95 -0.10 1.12 0.82 0.66 2.95 -0.70 1.42 1.89 2.95 2.95 -1.05 -1.05 0.36 0.73 0.82 0.82 0.06 0.73 0.06 0.06 0.06 -2.16 0.82 1.12 -2.23 0.73 0.73 0.06 2.19 0.06 -0.45 -1.47 0.82 0.82 0.78 0.82 0.82 0.06 2.95 -0.80 0.73 -2.23 2.95 1.84 -1.40 0.06 0.06 Cum.: Cent 8.30 8.30 8.12 7.10 6.85 6.44 6.10 5.83 5.05 4.92 4.53 4.32 4.29 4.23 4.18 3.74 3.61 2.93 2.88 2.61 2.53 2.31 2.27 2.23 2.23 2.09 1.93 1.93 1.76 1.58 1.39 1.36 1.15 1.13 0.84 0.81 0.74 0.68 0.61 0.53 0.53 0.40 0.29 0.24 0.20 0.17 0.14 0.13 0.13 0.08 0.01 0.01 0.01 -0.12 -0.16 -0.16 -0.20 -0.23 -0.25 -0.32 -0.46 -0.48 Rank 63 64 65 66 67 68 69 70 71 72 73 74 74 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 123 Country Bulgaria Belize Panama Solomon Is. Slovenia St. Kitts & N. Latvia Namibia Hungary Antigua & Barb. Poland Moldova Sao Tome & P. Albania Kiribati Argentina Dom. Republic Romania Switzerland Cape Verde Nicaragua Cambodia Paraguay Lithuania Macedonia Benin Chile Colombia Andorra Bangladesh Nepal Mali Ethiopia Madagascar Marshall Islands Ukraine Mongolia Armenia Georgia Philippines Mozambique Monaco Niger Guinea-Bissau CAR Taiwan Lesotho Zambia Indonesia Russia Seychelles Malawi Palau Ghana Micronesia Iran Mexico Croatia Nigeria Senegal Brazil United States Unit 4 3 4 2 4 2 4 4 4 2 3 4 4 4 3 1 2 3 0 4 4 2 3 4 4 4 2 1 4 4 3 4 0 4 4 4 4 4 4 1 4 4 4 4 4 3 4 4 4 0 4 4 4 4 2 4 0 3 0 4 0 0 Contemporary Parl PR 2 2 2 0 0 1 2 0 2 2 2 0 2 2 1 2 2 1 2 0 1 2 1 2 1 2 2 1 1 0 0 2 0 1 1 2 1 0 1 2 0 2 2 2 0 2 1 1 2 1 0 2 0 0 0 0 2 1 2 0 2 0 1 1 2 0 0 0 2 0 1 1 1 0 1 1 0 1 0 0 0 2 2 0 0 2 0 2 1 0 1 0 2 0 0 0 1 2 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 1 0 0 0 1 0 0 0 0 Cent 2.95 0.06 -0.34 -0.70 2.95 -0.70 2.95 1.84 1.89 -0.70 1.08 1.84 1.84 1.89 -1.05 -1.56 -1.86 1.08 -3.34 1.84 0.73 1.42 -0.04 0.78 1.89 0.73 -2.93 -3.69 1.89 0.82 0.06 0.78 -2.23 -1.40 0.82 0.78 -0.29 0.78 -0.34 -3.69 0.73 0.82 0.73 0.73 -0.29 -1.05 0.82 -1.40 1.84 -2.28 -1.40 -1.40 -1.40 -1.40 -2.93 -1.40 -3.39 0.01 -4.46 -0.34 -4.46 -4.46 Cum.: Cent -0.57 -0.58 -0.60 -0.69 -0.74 -0.77 -0.83 -0.89 -0.94 -0.94 -0.98 -1.06 -1.06 -1.12 -1.12 -1.14 -1.15 -1.16 -1.18 -1.23 -1.24 -1.28 -1.28 -1.31 -1.36 -1.38 -1.42 -1.42 -1.45 -1.48 -1.55 -1.57 -1.57 -1.61 -1.61 -1.68 -1.69 -1.72 -1.75 -1.84 -1.86 -1.88 -1.91 -1.97 -2.01 -2.07 -2.10 -2.12 -2.14 -2.25 -2.27 -2.34 -2.42 -2.49 -2.50 -2.57 -2.62 -2.62 -2.63 -2.76 -2.88 -2.88 Demo: democracy score according to the Polity2 index (ranging from -10 to +10). Unit: unitarism. Parl: parliamentarism. PR: list-proportional electoral system. Cent: Centripetalism. Cum. Cent: Weighted, cumulative sum of Centripetalism. N=124. A BRIEF EMPIRICAL TEST In order to test the hypothesis that centripetal institutions improve the quality of governance in democratic polities we must identify outcome indicators that are a) measurable, b) valid crossnationally, c) a direct result of government activity (so that the outcome is interpretable as a product of political institutions), and d) clear in their implications for the quality of governance (good or bad relative to the public interest). Length precludes an extensive analysis of possible indicators (but see Gerring and Thacker forthcoming). Here, we limit ourselves to a single set of good governance indicators that have come to be viewed as relatively authoritative within the new field of governance studies. These composite indices, developed by Daniel Kaufmann and his colleagues at the World Bank, include 1) Corruption control, 2) Government effectiveness, 3) Political stability, 4) Rule of law, 5) Regulatory quality, and 6) Voice and accountability. These six indicators are now a standard component of foreign aid assessments and other policy decisions throughout the world; they also serve as common empirical measures in academic work. There is perhaps no better single source for understanding the quality of governance around the world at the present time.64 Corruption control is a composite indicator drawn from a set of international surveys that attempt to determine corruption among public officials, the effectiveness of anticorruption initiatives (Standard and Poor’s DRI); corruption among public officials (Economist Intelligence Unit); corruption in the political system as a threat to foreign investment (Political Risk Services); frequency of additional payments necessary to ‘get things done,’ corruption as an obstacle to business (World Development Report).65 Kaufmann and colleagues name this variable ‘corruption control’ — transforming a negative into a positive characteristic — so that it can be more easily compared with other measures of good governance.66 Government effectiveness measures perceptions of the quality of public service provision, including the bureaucracy, the competence of civil servants, the independence of the civil service from political pressures, and the credibility of the government’s policy commitments. Surveys attempt to determine the pro-business quality of government policy, the efficacy of government, the degree of red tape and bureaucratic efficiency (Economic Intelligence Unit); institutional rigidities that hinder bureaucratic efficiency, quality of the government’s personnel, personnel turnover lowering the quality of the bureaucracy (Standard and Poor’s DRI); the government’s ability to carry out programs, bureaucratic 64 Kaufmann et al. (1999a, 1999b, 2002). For a skeptical discussion of these indicators see Weyland (2003). Kaufmann et al. (1999b: 560). This list of questions and surveys (like those that follow) represents only the most important sources (designated ‘representative sources’) among the longer set of survey questions employed in the construction of this index. See also Kaufmann et al. (1999a). 66 The variable is referred to in Kaufmann et al.’s first work (1999a, 1999b) as ‘graft’ and subsequently (2002) 65 quality (Political Risk Services); the likelihood that when government actions violate rules a citizen can find another forum to pursue her claims, management time spent with bureaucrats, customs efficiency, general conditions of roads, efficiency of mail delivery, quality of public health care provision, efficiency in delivering services, predictability of changes in rules and laws, credibility of commitment to policies (World Development Report).67 Political stability measures perceptions of the likelihood that the government in power will be destabilized or overthrown by unconstitutional or violent means. It includes risks due to major urban riots, major insurgency or rebellion, military coup, unconstitutional government changes, political terrorism, political assassination, civil war, armed conflict, social unrest, other sorts of internal conflicts. The most important surveys used to compile this aggregate index attempt to determine the following: the risk of urban riots, insurgencies and rebellions, coups, terrorism, assassinations, and civil war (Standard and Poor’s DRI); armed conflict or war, social unrest, terrorist threat, political violence (Economist Intelligence Unit); internal conflict, ethnic tensions (Political Risk Services); terrorism, likelihood of unconstitutional government changes (World Development Report). Rule of law measures confidence in the rules of society and law-abidingness, including perceptions of the incidence of crime, effectiveness and predictability of the judiciary, and the enforceability of contracts. Surveys measure losses and costs of crime, kidnapping of foreigners, enforceability of government contracts (Standard and Poor’s DRI); corruption in banking, crime (Economist Intelligence Unit); black market, property rights (Heritage Foundation/Wall Street Journal); law and order tradition (Political Risk Services); theft and crime, confidence in authority to secure property, unpredictability of the judiciary, crime and theft as obstacles to business (World Development Report).68 Regulatory quality measures policies friendly and unfriendly to the market including price controls, government intervention in the economy, wage-price controls, tariff and non-tariff barriers to trade, restrictions on foreign capital, excessive banking regulation, and import/export regulations. Constituent surveys measure regulations that impose a burden on business, government intervention in the economy, wage-price controls, tariff and non-tariff barriers to trade, regulation of foreign capital flows, foreign and domestic banking regulation (Heritage Foundation/Wall Street Journal); import/export regulations, other regulations of business, legal restrictions on ownership of business and equity by non-residents (Standard and Poor’s DRI); regulatory obstacles to opening a new business, price controls, regulations on foreign trade, foreign currency regulations, and general uncertainty about the costs of regulations (World Development Report).69 as ‘corruption control.’ 67 Kaufmann et al. (1999b: 55). 68 Kaufmann et al. (1999b: 58). 69 Kaufmann et al. (1999b: 56). In some versions, Kaufman et al refer to this variable as ‘regulatory burden.’ Voice and accountability measures the openness of the political process and the quality of civil liberties and political rights — “the extent to which citizens of a country are able to participate in the selection of governments,” as well as the independence of the media (Kaufmann, Kraay, and ZoidoLoboton 1999b: 7). Since the theory of centripetalism is limited, by definition, to democratic frameworks — and so, of course, are our empirical tests — this indicator may be interpreted as measuring the quality of democracy within formally democratic countries. (As indicated by Table 1, there is still a significant amount of variance to be explained.) Table 4 provides descriptive statistics for each of these dependent variables. Table 4: Governance Measures: 1996, 1998, 2000, 2002 N Mean Std Dev Min Max Corruption Control 327 0.25 1.00 -1.54 2.58 Government Effectiveness 345 0.22 0.92 -1.50 2.52 Political Stability 303 0.26 0.82 -1.86 1.77 Rule of Law 349 0.30 0.74 -1.88 1.87 Regulatory Quality 337 0.27 0.94 -1.50 2.36 Voice & Accountability 372 0.55 0.71 -1.31 1.67 Limited to minimally democratic cases, where Polity2>0 Each of these six indicators draws on polls conducted by consulting groups, NGOs, and intergovernmental organizations. Some, like those conducted by the Economist Intelligence Unit, Standard and Poor’s and DRI/McGraw-Hill, Freedom House, Heritage Foundation, Political Risk Services, and the World Bank, are broad in coverage (these are referred to by the authors as ‘representative sources’). Others are limited to particular regions of the world (‘non-representative sources’). Data for each poll is drawn from respondents of two basic types: country experts and firm managers or citizens (each carrying roughly equal weight in the analysis). Expert polls represent consensus ratings agreed upon by a small number of country, sector, and regional experts convened by the reporting organization. Typically, country analysts produce an initial assessment for each country based on publicly-available information and their direct knowledge of the country. For most of the sources we report, these assessments are guided by a checklist of specific issues which analysts take into consideration when providing their initial ratings. For all sources we report, these assessments are guided by a checklist of specific issues which analysts take into consideration when providing their initial ratings [reported in Ibid.]. . . These initial ratings are then reviewed by a panel of regional and sectoral experts, who determine the final rating for each country. In addition to providing quality control on the intial assessments, the purpose of this final stage is to improve the cross-country comparability of the ratings by ensuring that countries are benchmarked appropriately.70 We adopt their recent terminology (‘regulatory quality’ (Kaufmann et al. 2002). 70 Kaufmann et al. (1999b: 4). Mass surveys, by contrast, present averages by country of the responses of a large number of respondents to a variety of questions relating to governance. Typically, survey respondents are asked to rate aspects of governance on a categorical scale. The sampling frame . . . varies. Some sources focus on the opinions of the business community, others focus on the opinions of expatriates, and some are broad-based surveys of citizens.71 Poll results (of both types) are aggregated into subject areas, according to the nature of the question. The composite index for each variable is then constructed by the authors from the individual surveys using an unobserved components model.72 Thus far, the authors have aggregated four rounds of surveys, covering 1996, 1998, 2000 and 2002. The only seriously problematic feature of the Kaufmann indicators is that they are limited to a narrow slice of time. (The polls upon which the indicators are based are simply not available for a wide enough sample of countries to extend these indicators back prior to the mid-1990s.) Furthermore, because of the standardization technique used by Kaufmann et al. and the entrance of new countries into the dataset over time, scores within countries are not directly comparable over time.73 Thus, our analysis is essentially cross-sectional rather than time-series. We employ a between-effects estimator, which averages the results from each separate cross-section. This provides the benefits of bringing as much data as possible to bear on the problem at hand without violating the highly questionable (in this instance) temporal assumptions of regression analysis. (Results do not differ appreciably from a simple pooled-OLS format.) In choosing plausible controls we have tried to be sensitive to the literature on these topics. Existing studies suggest that these governance outcomes are likely to be influenced by multiple economic, geographic, demographic, and cultural factors. While not all of these plausible influences are easily measured in a global setting we try to include all of the standard controls plus several not-sostandard controls that nonetheless may be important. Economic development is measured by GDP per capita in logarithmic form (World Development Indicators [WDI]). We include a variety of regional controls to measure whatever cultural, demographic, geographic, historical, and socioeconomic factors might be common at a continental level. These include Africa (sub-Saharan), Asia, and Latin America/Caribbean. Socialism, generally understood to have negative repercussions for the quality of governance, is measured by a dummy variable indicating that a country is or was a socialist system at some time in the twentieth century (LaPorta et al. 1999). Colonial background is measured by a dummy variable that is coded 1 if a country inherited an 71 Ibid. (4-5). Regrettably, it is not possible to obtain disaggregated survey results (question by question). Most of the data was obtained by Kaufmann and colleagues from consulting agencies, who had proprietary interests in keeping the results in aggregate form. 73 Kaufman et al. standardize scores within each observation year to facilitate crossnational comparison and interpretation. This means, however, that the same score may have different meanings in different years. 72 English legal system, and zero otherwise (LaPorta et al. 1999). Other colonial systems seem to have been uniformly “bad” for governance outcomes in the developing world, and hence are captured in this single dichotomous variable. Latitude (logged) measures the — presumably beneficial — effects stemming from being farther from the equator (Landes 1999; Sachs and Warner 1997b). Ethnic heterogeneity, generally considered to have a deleterious effect on economic and political development, is measured by a fractionalization index developed by Alesina et al. (2002). Population, which is sometimes seen as hindering development, is measured as a logged variable (WDI). Since the core/periphery location of a country may influence its economic development, as well as the diffusion of political institutions, ideas and governability, we also measure a country’s distance (in kilometers) from the nearest financial center (London, New York, or Tokyo). Mineral wealth, particular in the form of easily exploited oil and diamond reserves, is often considered to be a drag on development, perhaps even an invitation to civil war. We measure both oil and diamonds in “raw” form (i.e., as millions of barrels per day and billions of metric carats per year, respectively), as a fraction of total population (Humphreys 2004).74 Protestantism (percent of population with a Protestant religious background) is sometimes regarded as a source of good governance (Gerring and Thacker 2004). Finally, we include a control variable that measures the average value of the dependent variable across all countries, weighted by the geographic distance of each country from the country in question. The assumption is that countries lying close to one another will display similar values for extraneous reasons (culture, geography, diffusion, and so forth). By including this variable in all regressions we should be able to minimize possible spatial autocorrelation in the sample. In order to address potential problems of endogeneity, which can imply a larger causal role than warranted for the affected variables, and which can violate important assumptions of ordinary-leastsquares and related regression techniques, we employ a base-year value from 1996 for GDP per capita in all estimations. In other words, we control for the per capita income of a country at the starting point of the sample period, rather than the level of per capita income in each observation year. Arguably, corruption, rule of law, etc. may affect the economic growth and prosperity of a country, just as those factors themselves may influence governance. By using a base year measure as a control, we remove much of this potential distortion.75 Because there exists no standard benchmark model for any of these dependent variables we 74 Some indicators measure the export value of these goods as a percent of all exports or of GDP. We believe that this confuses two issues — the extent of natural resources in a country and the degree of its economic development or export orientation, which is implicit in the denominator. Since it is the first, not the second, that we wish to measure it seems preferable to employ a “raw” measure of natural resources. 75 The alternative to the use of a base year is the employment of instrumental variables in place of per capita income, but there are no viable instruments that can feasibly be used in this particular case. We anticipate that any remaining endogeneity — likely quite minimal — biases the results against our hypotheses, by “soaking up” some of the variation that might otherwise be attributed to Centripetalism. Results obtained from using contemporaneous annual scores for per capita GDP, rather than the base year, are virtually identical to those presented here. conduct two tests for each dependent variable. The first is a “full” model, including all variables discussed above. The second is a reduced-form model including only those controls that pass the threshold of statistical significance (p<0.10 in two-tailed tests),76 in the expected direction. We retain the geography-weighted control in all models, regardless of statistical significance. Arguably, there is little need to retain control variables that do not perform well, since they are of little theoretical interest for the task at hand and do not show strong results. Conversely, a fuller specification is less likely to be subject to under-specification problems and presents a more difficult test of our hypothesis. Reassuringly, results for our variable of interest, Centripetalism, are usually robust across both specifications. The results shown in Table 5 offer strong initial support for our hypothesis. The overall fit of the models is good. Each specification is significant at better than the 0.0001 level, and the R2 ranges from 0.62 to 0.85. Collectively, these models explain roughly sixty-two to eighty-five percent of the variation in the dependent variables. More specifically, Centripetalism appears to enhance the quality of governance in five out of the six dimensions tested by the Kaufmann indicators. The effect on government effectiveness is a bit weaker, passing the threshold of statistical significance only in the reduced-form model (see Table 5, column 4). But it does come quite close to statistical significance in the full specification (column 3, p<0.11). We suspect that political stability, the only outcome that shows no apparent relationship to Centripetalism, may be affected by a number of exogenous factors for which we have been unable to control (Table 5, columns 5 and 6). Centripetal political institutions, based on the evidence analyzed and presented here, appear to generate substantial improvements in the quality of democratic governance. These findings, while quite encouraging for our hypothesis, are nevertheless far from definitive. We are constrained by the nature of the dependent variable to conduct only a series of cross-sectional analyses. There are, as we have noted, many more governance indicators that might be tested. And, there are always more controls that might be employed, though it should be added that we have tested many more controls than actually appear in these tables. (Results are stable in additional specification tests.) These caveats notwithstanding, the strength, consistency and robustness of the results presented here offer encouraging evidence that centripetal institutions may generate notable improvements in the quality of governance in democratic polities around the world. 76 We retain a small number of control variables that come close to this threshold but that do not quite achieve it, if their deletion interferes with the results for other control variables. In no case did the deletion of inclusion of such a variable affect our Centripetalism variable meaningfully. Table 5: Empirical Tests Dependent variable: Corruption Control Government Effectiveness Regulatory Quality 2 5 6 7 8 9 11 12 0.071*** 0.036 0.047** -0.034 -0.017 0.059*** 0.054*** 0.054** 0.048** 0.071*** 0.062*** 0.004 0.002 0.015 0.009 0.031 0.018 0.015 0.010 0.036*** 0.026*** 0.025* 0.030** 0.391*** 0.366*** 0.390*** 0.374*** 0.339*** 0.297*** 0.417*** 0.373*** 0.273*** 0.255*** 0.255*** 0.219*** Socialist legal origin -0.178 -0.256** -0.260* -0.389*** -0.122 -0.022 -0.206* 0.038 0.048 -0.219 English legal origin 0.055 -0.103 0.117 Latitude (ln) 0.173*** Ethnic fractionalization 0.031 0.170 -0.432 Population (ln) -0.005 0.026 -0.123*** Distance from nearest financial center Oil production 0.000 -0.000 -0.000 -0.668 -0.524 -0.266 Diamond production 0.013 0.016 Africa 0.111 Asia 0.046 Latin America/Carib -0.235** -0.314*** -0.265** Protestant 0.008*** 0.007*** 0.005** Constant -2.709*** -2.481*** -2.926*** Observations Countries R-squared Prob > F 313 119 0.85 0.0000 323 123 0.85 0.0000 330 119 0.80 0.0000 (0.014) (0.038) (0.144) (0.019) (0.013) (0.033) (0.113) (0.054) (0.121) 0.151** (0.068) (0.060) (0.001) (0.295) (0.199) 0.227** (0.098) (0.036) (0.135) (0.014) (0.033) (0.112) (0.086) 0.216*** (0.075) 0.159** (0.061) (0.030) (0.010) (0.043) (0.055) (0.020) (0.008) (0.036) (0.162) (0.015) (0.044) (0.166) 0.135 0.064 0.161** 0.091* 0.096 (0.068) 0.107 -0.179 0.019 -0.046* (0.084) (0.052) (0.024) (0.074) (0.020) 0.008 0.026 -0.019 0.025 0.075 0.537** 0.259* 0.403** 0.250 0.052 0.204 0.262** 0.194* 0.086 0.099 0.053 -0.112 -0.247*** 0.148 0.345** 0.003 (0.123) (0.196) -0.360*** (0.091) (0.181) (0.107) (0.034) (0.191) (0.111) (0.086) (0.186) (0.154) (0.159) (0.130) (0.135) 0.005* 0.005** 0.006*** 0.006*** 0.000 -2.331*** -1.000 -0.928* -3.218*** -2.499*** -1.184** -0.811** -2.417*** -1.235*** 341 123 0.80 0.0000 296 107 0.64 .0000 299 108 0.63 .0000 323 120 0.85 .0000 333 124 0.85 .0000 337 121 0.67 .0000 349 125 0.65 .0000 334 120 0.64 .0000 345 124 0.62 .0000 (0.002) (0.475) (0.129) (0.622) (0.034) (0.145) (0.065) (0.024) -0.942 (0.252) 0.125* (0.000) (0.607) (0.028) (0.134) 0.069*** -0.350 (0.041) (0.039) (0.105) -0.526 (0.512) (0.014) (0.208) -0.078*** (0.000) (0.733) (0.022) 0.105 (0.207) (0.020) (0.024) 0.167 (0.102) 0.166*** (0.171) -0.109*** (0.023) -0.000 (0.146) (0.085) (0.046) (0.014) (0.019) (0.000) (0.172) (0.135) (0.026) (0.019) -0.000 (0.032) (0.160) (0.056) (0.034) (0.577) (0.030) (0.027) (0.026) (0.280) (0.000) (0.537) (0.029) (0.130) 0.179*** (0.023) (0.000) (0.442) (0.155) (0.036) (0.194) (0.021) (0.002) (0.041) (0.014) (0.098) 0.136** (0.180) (0.115) (0.015) (0.020) 0.046 (0.091) (0.063) (0.022) 10 Voice & Accountability 1 (0.020) 4 Rule of Law 0.069*** Model: Centripetalism (historical) Geography-weighted dependent variable GDP per capita, 1996 (ln) 3 Political Stability (0.003) (0.320) (0.665) (0.002) (0.479) (0.002) (0.410) (0.001) (0.293) (0.002) (0.514) (0.002) (0.390) (0.513) (0.353) Between-effects regression. Standard errors in parentheses * p<10%; p<5%; p<1% Sample Period: 1996-2002 Sample limited to countries that are minimally democratic (Polity2>0).