Democracy, political partisanship, and state capacity in Latin America

Abstract

Social scientists have often neglected, or not sufficiently explored, the role of political factors in shaping state capacity. When they did, they mostly focused on key institutional features of political regimes, especially democracy. In this paper, we broaden this approach: besides the institutional traits of democracy, we analyze how governments and their ideologies influence state capacity. In particular, we assess the impact of democracy and executives’ partisanship on a composite index of state capacity, based on political order, administrative ability, and extractive capacity. To this end, we apply a pooled cross-sectional time-series model to 18 Latin American countries between 1995 and 2009. Our findings suggest that, in recent years, state capacity in the region was significantly affected by both democratic features and the ideological bearing of elected governments.

Introduction

This paper argues that both democracy and political partisanship have a bearing on state capacity. We tested this relationship for Latin America, between 1995 and 2009, and found that democracy weakens the negative influence of authoritarianism on stateness, but it is insufficient to promote state strengthening. In addition, we explored the potential causal link between changes in political partisanship and stateness. Although several qualitative studies have underpinned the mechanisms that relate partisanship to a stronger state capacity (Slater, 2008; Besley and Persson, 2009; Kirby, 2009), we are among the first to test this hypothesis quantitatively, finding generally supportive evidence: our investigation opens new avenues of research into the topic and originally contributes to the literature on state capacity.

A number of studies have analyzed the relationship between democracy and state capability both in wider geographical contexts and in particular regional settings. Adserà et al. (2003) found that higher political accountability improves state capacity by lowering corruption and the risk of expropriation of property and by strengthening bureaucratic quality and the rule of law. Besley and Persson (2009) noticed that along with the incidence of war and other more specific variables, the historical weight of parliamentary democracy is a remarkably stable predictor of both the legal and the fiscal capacity of the state. Other scholars, finally, claim that regimes with regular elections and other political freedoms will eventually produce stronger (or, more accurately, better funded) states (Schmitter et al., 2005: 11). The role of democracy, however, is still controversial as, historically, developmental states in Asia existed under authoritarian regimes (for instance in Taiwan and South Korea). ¹ Bratton (2008) found that in Africa democracy appears to be empirically connected with such components of state capacity as rule of law, control of corruption, effectiveness, legitimacy, and accountability. Slater (2008), finally, looked at three Asian countries (Malaysia, Indonesia, and the Philippines): he concluded that competitive elections have facilitated state building in the region by stimulating the formation of stronger political parties; a more vigorous state commitment to citizen registration; and the imposition of centralized authority over societal strongmen.

In Latin America, democracy has been analyzed mostly as an outcome of state weakness. This political regime does not appear in major accounts of determinants of state capacity: Centeno (2009), for instance, suggests that the main determinants of state capacity have been specific historical legacies; (non-state) institutional quality, as the degree of confidence in public and private institutions; economic and social inequality; and globalization. Cardenas (2010), however, identified political inequality as one of four factors that potentially explain the exceptionally low state capacity in the region, along with economic inequality, interstate conflict, and civil war. He adds, however, that the positive effects of democracy are undermined in the presence of high economic inequality. Likewise, in their analysis of the determinants of welfare states in a series of world regions that include Latin America, Haggard and Kaufmann (2008) argued that, ceteris paribus, democracy promotes more progressive social welfare states: democratic rule provides incentives for politicians to reach the poor and those more exposed to risk, and opportunities for the latter to organize, thus strengthening the nature and scope of a crucial state capacity – providing citizens with fundamental social services. Whether or how democracy affects state capacity in Latin America, however, remains an open question. Acemoglu and Robinson (2008) noticed that, in spite of an unambiguous trend toward greater democracy, state capacity has remained particularly low in the area, for the persistence of de facto powers. Even if de jure institutions (political rights, elections, and checks on the executive) are established, the political balance may not change. A similar conclusion was reached by Cheibub (1998), for whom, over the period from 1970 to 1990, the government’s extractive capacity was not much stronger in democracies than in authoritarian regimes. There are, in sum, several counter-arguments to the notion that democracy always benefits stateness. Democratization and polarized competition can destabilize a state, and democratic petitions may overload and weaken political institutions (Hagopian and Mainwaring, 2005). Rather than contributing to legitimizing state institutions and improving their effectiveness the adoption of formally democratic institutions may actually produce a deterioration of state authority and public services.

An often overlooked premise to these reflections is that most states that eventually proved successful in achieving a robust stateness did not have the necessary capacities or the right bureaucracies to do so at first: they built them (UNRISD, 2010: 259). A political leadership committed to fast growth and equality must usually reach a political settlement with domestic actors to define the direction of public policy and then create a developmental and welfare-enhancing bureaucracy to support it (Haggard and Kaufmann, 2008). In a similar way, state capacity may be purposely undermined: the weak state capacity of Latin America (to levy taxes, in particular) is also due to powerful social groups that in some countries successfully opposed increases in the tax burden (Huber and Stephens, 2012: 41).

Redistributing social and political power in favor of subaltern groups is important in order to build bureaucratic capacities that support both growth and redistribution: the active inclusion of such groups in the political process enables states to frustrate pressures from dominant elites, which oppose redistribution and tax revenue increases. Mobilized subaltern groups press for an expansion of public provisions and the institutional capacity necessary to deliver administratively demanding universal public services such as public housing and high-quality education and healthcare. If these arguments are correct, left governments seem well suited to strengthen state capacity. Left governments are committed to lifting people out of poverty, reducing inequality, and providing quality public services, particularly education and health. In order to be able to do this, they have to increase tax collection, which is a basic indicator of state capacity. They also want to get away from clientelism, because their competitors on the right are usually quite good at getting voters by these means and they prefer an effective, non-corrupt, bureaucracy that treats people impartially, as citizens with rights, not as supplicants who are asking for a favor, and can be asked for something (like a vote) in return. Moreover, they want to avoid capture of the state by elites and private interests, because this will derail their program of providing quality services for all citizens, from law enforcement to child care.

Governments of other ideological bents, however, may also be interested in strengthening state capacity: order and legality, for instance, are a central concern of right of center parties, as crime and violence are inimical to business investments and growth and are most often targeted against wealthy individuals. Thus, in countries such as Colombia, executives in the right (and center) have strengthened political order and the overall capacity of the state by reducing political violence: they successfully fought rebel organizations, paramilitary groups, rural and urban vigilante bands, and private security services, and they were able to contain more or less organized crime and lower the number of homicides and kidnappings (Feldman, 2012: 743). More generally, right-leaning governments may uphold state capacity as, in weaker states, enforceable property rights, duly supported by smoothly functioning legal institutions, are inadequate and widespread corruption is prominent. These limitations, in turn, generate insecurity and unpredictability, which are inimical to new investments, the creation of jobs, and the development of business (Fukuyama, 2007; De Soto, 1989).

In addition, once democracy is established, ideological orientation crucially interacts with electoral competition. When election results are disputed, irrespective of ideological orientation, parties may be driven to undertake policies that they might otherwise disregard, for the exclusive purpose of winning elections (Haggard and Kaufman, 2008: 360). Policies may be undertaken to capture the independent voter at the center and will be more conservative, or progressive, than a purely ideological stance implies. Accordingly, left-wing and right-wing governments become more similar: the former cannot pursue their preferred policy of welfare state expansion because of budget constraints, and the latter cannot pursue their preferred policy of cutbacks because their constituency has grown attached to the welfare state’s programs (Schumacher and Vis, 2009). In Uruguay, for instance, a broad social alliance backed by the left Frente coalition was able to exercise a wide-ranging veto power against neoliberal pension reforms. Politicians from traditional parties, among which the conservative Blanco party, fearing the loss of the pensioners, either backed the reform or did not take a position. The existing welfare system enjoyed wide political support, not only among organized stake-holders but also within the broader electorate: politicians bent on reforming the welfare system had to reckon with both (Luna, 2006). In short, whether in Latin America governments on the left have contributed to state capacity more than their counterparts on the right is a question better answered empirically.

This paper is organized as follows: in the next few pages, we will define our main dependent, independent, and control variables. In the following paragraph, we will submit our major hypotheses and outline the mechanisms by which our variables are related. Next, we will estimate the effect over time of democracy, executives’ partisanship, and other sociodemographic factors on different measures of stateness in Latin America. Finally, we will illustrate and discuss our findings, with references to the political experience of the region. Conclusions, as usual, will wrap up the analysis.

State capacity

The concept of state capacity, and its operationalization, is problematic and heavily contested in the literature (Studies in Comparative International Development 2008, special issue; Revista de Ciencia Política 2012, special issue). ² In fact, there is greater scholarly agreement on key dimensions of the state than on how to operationalize such dimensions or the notion of the state itself (Carbone, 2013: 6). We believe that a definition of state capacity should be centered on a basic, narrow scope of state functions to avoid the problem of conflating causal and constitutive relationships, thus using causes or outcomes as indicators of the concept (Hanson and Sigman, 2013). We define state capacity, broadly, as the ability of state institutions to effectively implement official goals (Sikkink, 1991). This approach eludes normative conceptions about what the state ought to do or how it ought to do it – for instance, in the aftermath of the neoliberal turn in Latin America and the cutbacks of state intervention and bureaucracy, this definition should avoid confusing minimal but capable states, as Chile, with those that are fundamentally weak (Soifer, 2012: 86–87). More specifically, our definition of state capacity relies on Mann’s ‘infrastructural power’ concept – that is, ‘the institutional capacity of a central state, despotic or not, to penetrate its territories and logistically implement decisions’ (2008). Central to his notion, then, is ‘the question of the state’s authority over territory’, as well as ‘whether governments can implement policies, including the provision of public goods’ (Fortin, 2010: 656).

Empirical studies of the state typically highlight at least one of three constituent elements (Soifer, 2012; Hanson and Sigman, 2013): the enforcement of a degree of internal political order through a monopolistic control over the means of coercion; the presence of a basic administration [a ‘usable bureaucracy’ in the words of Linz and Stepan (1996): 11]; and the ability of the state to extract revenue from their citizens (North, 1981). ³ The coercive dimension is crucial in Weber’s (1965) definition of state, as an organization that possesses a monopoly on the legitimate use of physical force within its territory. The enforcement of internal political order by the state includes the capacity to defend its frontiers, uphold domestic order, and enforce law and policies by curbing violence. Administrative capacity refers to the ability to plan and develop policies and to produce and deliver public goods and services. This, in turn, requires technical competence, professional civil servants, a curbing of public corruption, and an effective reach across state territory. In particular, a professional bureaucracy tends to be somewhat autonomous from political pressure and displays an established mechanism for recruitment and training. In relation to this dimension, we believe it is important to avoid conflating a government’s policy choices with the ability of states to implement these policies – that is, to distinguish between the ability to administer from the services themselves (Soifer, 2012: 591). For this reason, in measuring this dimension, we avoid looking at the outcomes of public goods and services. Extractive capacity, finally, refers to the ability of the state to raise taxes and extract revenue from its citizens, and implies a series of competences and skills that are crucial for a definition of overall state capability: means to access population; instruments to gather and organize complex information; the possibility to count on honest civil servants; and ways of ensuring popular compliance with tax policies (Hanson and Sigman, 2013: 4). For extractive capacity, as well, particular attention should be paid to distinguish the policy choice to tax from the ability of the state collection apparatus to collect the assessed taxes.

Measuring state capacity is equally challenging. We agree with Hendrix (2010: 283) that any single variable is unlikely to adequately capture the multidimensional concept of state capacity. However, many composite indexes of state capacity also include potential causes (as the lack of democracy) and predicted consequences (such as humanitarian disasters) into their definitions (Mata and Ziaja, 2009; Gutiérrez, 2011; Soifer, 2012). The ideal measure of state capacity, in short, should consist of an index with multiple number of indicators, which carefully avoids conflating key dimensions of the dependent variables with its potential causes and effects. On the basis of these considerations, we resolved to use the State Capacity Dataset (1960–2009) elaborated by Hanson and Sigman (2013). This data set measures state capacity based on the three main dimensions discussed above: extractive, administrative, and coercive capacities. These in turn are measured by 24 main indicators. In detail, administrative capacity is assessed through four sets of data: the International Country Risk Guide bureaucratic quality index; additional indexes related to administrative capacity and civil service quality; ⁴ a measure of census frequency, which signifies both the ability to collect data and effective territorial reach; and an indicator of contract intensive money, which represents the state’s capacity to regulate economic exchange. Overall, extractive capacity is first measured by tax revenues as a proportion of GDP. As argued above, however, this information by itself is unable to distinguish between policy choices and extractive capacity. Thus, additional measures were added, as the proportion of tax revenue coming from income, property, and domestic consumption taxes relative to revenue originating from international trade, as with customs duties. The greater the former, which are administratively more complex and require a more structured bureaucratic apparatus, the higher the expected level of extractive (and administrative) capacity. ⁵ Other indicators, finally, determine the degree to which revenue collection is connected with context-bound expectations. For instance, the Arbetman-Rabinowitz et al. data set (2012) gauges the ratio of actual tax revenue relative to an expected tax yield given a country GDP per capita, mineral production, exports, and other factors. Coercive capacity, to conclude, is also measured by four groups of data. First, there are data on military expenditures and personnel. The relationship of these figures with coercive capacity, however, is problematic as a large military force may indicate war or insecurity, which significantly weaken state capacity. Thus, additional data have been added, showing whether and how much the state itself is involved in perpetrating violence, has a monopoly on the use of force, and is present in the territory, by looking at the artificiality of state borders or the extent to which its land is mountainous, and therefore more difficult to reach. The authors of the data set, to conclude, identify, by way of latent variable analysis, a series of underlying factors that signify both overall state capacity and more specific aspects of this concept (Hanson and Sigman, 2013). ⁶

Determinants of state capacity: regime and partisanship

With democracy, we refer to a political regime characterized by universal suffrage, the protection of key civil and political rights (including the presence of alternative sources of information), and free, transparent, and competitive elections. In addition, formal democratic institutions must be sovereign (Dahl, 1989). Thus, an analysis of democracies should exclude hybrid or ‘electoral authoritarian’ regimes, as they do not hold free and fair elections (Levitsky and Way, 2002). Democracies, however, may also be incomplete and partial: ‘defective’ democracies offer only limited guarantees for political rights (Merkel, 2001); ‘illiberal’ democracies provide inadequate protection for civil rights and the rule of law (Zakaria, 1997); in ‘delegative’ democracies, finally, elected officials are scarcely responsive to citizen preferences, inadequately constrained by other agencies of government, and insufficiently respectful of the rule of law (O’Donnell, 1993). We resolved to use minimalist notions and measures of democracy, rather than more substantive ones, to avoid mixing up attributes of political authority and state qualities (as its capacity) we wanted to disentangle. Consequently, we eliminated from our procedural measure of democracy, based on the Polity IV data, the component of the scale that refers to conflict (lack of political order). ⁷ The resulting scale runs between −2 (full autocracy) and +7 (full democracy). ⁸

Partisanship refers to ideal worldviews and the instruments held necessary to accomplish them: in Latin America, as in more advanced industrial countries, parties may be classified based on a left/center-left vs. right/center-right dichotomy. ⁹ We follow the classification suggested by Coppedge (1997) and updated by Huber et al. (2012). Coppedge’s experts classified parties, along this dimension, into six categories: left, center-left, center, center-right, right, and personalist. Thus, parties of the right (Brazilian ARENA) presented themselves as, or appealed to, heirs of traditional elites, fascists, neo-fascists, or the military with a conservative message (Coppedge, 1997: 8). Coppedge classified as center-right (Argentine UCD) parties that ‘targeted middle- or lower-class voters in addition to elite voters, by stressing cooperation with the private sector, public order, clean government, morality, or the priority of growth over distribution’. In turn, he classified as centrist (Argentine UCR) those parties that ‘stressed classic political liberalism, the rule of law, human rights, or democracy, without a salient social or economic agenda’. Also included in this category are ‘governing parties whose policies are so divided between positions both to the left and to the right of center that no orientation that is mostly consistent between elections is discernible’. Going to the other side of the partisan spectrum, center-left parties (Venezuelan Acción Democrática) ‘stress justice, equality, social mobility, or the complementarity of distribution and accumulation in a way intended not to alienate middle or upper-class voters’. Left parties (Partido Socialista de Chile; any Communist party) ‘employ Marxist ideology or rhetoric and stress the priority of distribution over accumulation and/or the exploitation of the working class by capitalists and imperialists and advocate a strong role for the state to correct social and economic injustices’. ¹⁰ Finally, parties are classified as personalist if they ‘base their primary appeal on the charisma, authority, or efficacy of their leader rather than on any principles or platforms, which are too vague or inconsistent to permit a plausible classification of the party in any other way’ (Peruvian Cambio 90 or Bolivian Movimiento al Socialismo).

We use the instruments outlined above to measure the partisan orientation of governments in Latin America during periods of democratic rule [Executive Partisan Balance (EPB)]. ¹¹ When EPB increases, the ideological orientation of the executive moves toward the left (Huber et al., 2012). In order to account for the necessary time needed to change state capacity through the formulation and passing of the relevant legislation, the training or re-training of civil service officials, and the like, the variable is cumulated for the 15 years preceding the year of observation. ¹² Finally, a set of standard control variables, both economic and social, has been added to our model. Stateness difference controls for the weight of past state capacity and has been measured by calculating the difference in state capacity levels over the previous 14 years. Economic development, indicated by GDP per capita at current USD in 1994, is taken from the World Bank, as are country size and oil rents; ethnic fractionalization is measured by using the data presented in the article Fractionalization by Alesina et al. (2003).

Hypotheses

Democracy is thought to reduce violent conflict by providing institutionalized channels of communication with political opponents, by offering to incorporate them into the debate, and by conceding to some of their political and social demands. ¹³ The rationale for this strategy, over and above the intrinsic desirability of democracy, is that by making the government more accountable, citizens have less cause for violent opposition. Accordingly, people will be more likely to identify with democracy and the state. Opening up a closed media will also allow greater public scrutiny of poorly performing areas of state function. Creating space for independent civil society permits advocacy groups to monitor and critique state performance and work together with the state to offer new policy ideas. Finally, by making politicians, and at least indirectly also administrators, accountable, democratic procedures and sanctions help control arbitrary power and the diffusion of corruption (Carbone, 2013). However, we do not rule out that the relationship between democracy and stateness may assume a non-linear form: different levels of democracy may have a diverse impact on stateness as suggested by part of the literature (Bäck and Hadenius, 2008). Thus, our first hypothesis is that more complete and mature democratic governments will improve state capacity in either a linear or a quadratic form (Hypothesis 1).

The evidence and scholars’ opinions on the role of partisanship in democratic governments are mixed. Comparative historical evidence indicates that incumbency of left parties is key for welfare state development (Huber and Stephens, 2012). More specific studies have also established a strong relationship linking robust mass mobilization and leftist-party rule to the development of an infrastructurally capable welfare state. ¹⁴ In Latin America, finally, recent left governments, consistent with their major political objectives, have made efforts toward deepening democracy and citizens’ rights, rebuilding state capacity, freeing while regulating markets, and creating an adequate technical and political environment for competent policymaking (Bresser-Pereira, 2001). Other studies, however, point at the strengthening of state capacity under right-leaning governments (Rangel, 2005; Fukuyama and Colby, 2011). Finally, scholars have underlined a process of ideological convergence by major political parties and executives that blurs partisan and policy differences. As a result, we leave an initial answer to this issue to our empirical analysis (Hypothesis 2).

To conclude, we add a number of standard control variables. First, we posit that current levels of state capacity are influenced by past levels. It is reasonable to assume (and extensively recognized) that state capacity changes slowly, and that past levels of state strength strongly affect current ones; thus, we added a stateness difference variable to our model. Second, we introduce economic development to test a possible spurious relationship between our main variables: both democratic governments and state capacity might be the product of economic development over time. For Schmitter et al. ‘the richer and more developed the country, the more it will spend on public goods provided by its state’ (2005: 7). Income increases may foster state capacity because richer countries can be expected to afford better institutions and administrative structures and many variables correlated with income, such as schooling levels or urbanization, may decrease the social tolerance of corruption (Pellegrini and Gerlagh, 2008: 255). Saylor (2012) also argues that the pursuits of private profit can be a powerful state building motive: when hunting profit during booms, export-oriented actors regularly seek new state-supplied public goods, whose provision promotes the expansion of state capacity. All these mechanisms imply an independent effect of the level of economic development on state capacity. Thus, we insert a specific hypothesis (Hypothesis 3).

All else being equal, excessively large territories may prove hard to establish institutionalized rule on, given that vertical integration and horizontal coordination becomes more difficult (Herbst and Mills, 2006). Thus, we expect larger countries to be associated with lower state capacity (Hypothesis 4). Ethnic fractionalization is associated with a challenging environment for institutions in balance between inclusion of minority groups and preservation of governing effectiveness and efficiency (Ben-Meir, 2006). In countries where fractionalization is high, citizens have low confidence in political institutions (Sojo, 2011) and the democratic process is less likely to take hold. Accordingly, it is reasonable to hypothesize that where fractionalization is high it will be negatively related to stateness (Hypothesis 5). ¹⁵ Finally, excessive dependence on oil exports weakens state capacity by loosening the tie between tax burden and the right to political representation (Hypothesis 6). The mechanisms connecting oil rents and state capacity are at least two: first, large oil rents (when both the state and democracy are not yet mature) weaken state capacity because they exempt the state from building an efficient tax extraction system. Oil-producing countries tend to have weaker state apparatuses than their income levels would predict, because rulers have less need to invest in the bureaucratic capacity to collect taxes. Thus, they are unusually detached from and unaccountable to the general population, and their populations, in turn, are less likely to demand political responsibility and representation in government. Oil wealth also produces greater spending on patronage, which, in turn, weakens existing pressures for representation and accountability (Karl, 2004). ¹⁶

Data, methodology, and findings

Our paper focalizes in a diachronic way on stateness in Latin America. Although the choice of the subcontinent ensures comparability within a most similar research design, it also represents an ideal testing ground for our main hypothesis – the impact of political factors on state strength. In Latin America, democracies are abundant and differ in quality (as opposed, for instance, to the Middle East), whereas political parties follow ideologies that have been classified along a right-to-left continuum perceived to be meaningful for the structuring of politics (as opposed, for instance, to the worldviews of most African and Asian parties). The presence of political beliefs that generate consistent and coherent policy choices, even if less easily identifiable than in the developed West, is vital for an investigation of the potential links between political parties acting in a democracy, partisanship, and state capacity, and validates the election of the subcontinent to assess our research questions. Our analysis comprises all Latin American countries with population above 2,000,000, observed over the 1995–2009 period: we initially considered 22 countries in total, 11 belonging to Central America and 11 to South America. Missing data problems forced us to exclude four of them from our final report. ¹⁷

As we rely on a short series of observations obtained from many countries over time, we adopted a pooled cross-sectional time series. ¹⁸ The advantage of longitudinal panel information, compared with cross-sectional information, consists in its potential for an analysis of social, political, and economic dynamics at different levels. To test whether a fixed effects or random effects (RE) model is more appropriate for our data set, we applied the Hausman test. ¹⁹ On the basis of test results, we used different RE longitudinal regression models on a long unbalanced panel data set, using the software STATA. ²⁰ In principle, state capacity could cause both more democracy and the rise of left. It could cause higher-quality democracy by prompting political actors to choose to play within a set of institutions effectively enforced rather than opting for non-institutional means of political competition. In addition, the poor will press the state and the political arena for redistribution when they are confident that existing state institutions are sufficiently capable to implement their preferred policies. Thus, support for parties promising increased service provision (i.e. the left) varies with state capacity for any level of inequality. In conclusion, as the relationship between democracy and stateness raises reciprocity questions, we tested for the endogeneity of democracy. Our results confirm that our model is free from such problem. ²¹

We then proceed to a visual analysis of the values of our dependent variable (Figure 1). During the period of observation, our measure of state capacity has increased especially in countries where the quality of democratic governments has been intermediate to strong, and the left has exercised power for significant periods of time, as in Chile, the Dominican Republic, Uruguay, Peru, and Costa Rica. However, less-prominent increases in state strength also occurred where governments were controlled mostly by parties of the right, as in Guatemala and Colombia. ²² In contrast, decreases in state capacity took place in five countries, some of which have been governed by personalist executives, such as Argentina and Bolivia, whereas in others a left executive was in power only shortly, such as in Ecuador (since 2007) and Paraguay (since 2008). Only Brazil represents a clear case of a left-leaning country characterized by a declining state strength. ²³

Figure 1 Differences in stateness (1995–2009).Note: State capacity data set (see Hanson and Sigman, 2013).

To explore the impact of our independent variables, we develop three models (Table 1). In the first two, we focus separately on levels of democracy and the impact of executives’ partisanship, along with control variables. Subsequently, we estimate the aggregate bearing of these variables on the dependent variable. More precisely, in the first model we regressed the indicator of state capacity against our level of democracy index (in linear and quadratic forms, see Table 1) controlling for context variables: a lagged (14 years) stateness variable; level of economic development in 1994; the size of the country; oil rents; and ethnic fractionalization. Contrary to expectations, results on democracy suggest a non-linear but negative impact during the period of observation (R ²=42%). In short, the negative bearing of less-democratic governments fades progressively away when political performance improves and democracy matures, but a negative effect still persists. A positive effect is detected, on the other hand, when we regress executive partisan balance (EPB) against state capacity levels (see Table 1): increasing the presence of the left in government also increases state capacity levels (R ²: 30.5%).

Table 1 The effect of democracy and partisanship on stateness in Latin American

Note: We have applied a random effects generalized least squares regression.

*P<0.10; **P<0.05; ***P<0.01; ****P<0.001.

In model 3, finally, we incorporate levels of democracy and EPB, along with our control variables: democracy and EPB are still significant and unfold the above-mentioned effects, whereas state capacity levels are also increased by the difference of stateness variable. Finally, levels of economic development, land size, the proportion of oil in the economy and ethnic diversity have the expected influence, but fail to reach statistical significance. Figure 2 describes in detail the interaction affecting our democracy variable: levels of democracy have a significant and negative impact on the probability of increasing the levels of stateness. This effect, as noticed above, decreases as democracy takes hold, but democracy, by itself, has been insufficient to advance state capacity in Latin America between 1995 and 2009. When countries become fully democratic (score 5 on the chart), ²⁴ finally, the variable loses significance.

Figure 2 Marginal effect of democracy.Note: The figure displays the predictive margins with 95% confidence interval of democracy (Polity IV×Polity IV) reported in Table 1 model 3.

In sum, between 1995 and 2009, the effect of our political regime variables on state capacity has been significant and negative, but increasing levels of democracy have weakened its intensity. We find that, at the end of the last decade, higher competitiveness and openness of executive recruitment, as well as more substantial constraints on the chief executive, have been unable by themselves to promote state capacity. Our study empirically confirms a series of investigations that underline this critical relationship. On the one hand, Latin American democracies have been generally unable to solve the problem of socio-economic inequality and, on occasions, have even failed to remove glaring political inequalities, both of which represent severe obstacles to more capable states (Huber et al., 1997; Cardenas, 2010). Moreover, de jure democratic institutions, such as political rights and elections, may be established, but the political balance, strongly influenced by de facto powers, may not change: when the former only reflect the latter, democracy is associated to a decay in state authority (Cheibub, 1998; Acemoglu and Robinson, 2008). In Latin America, the recurring opposition of the most powerful segments of society to social and fiscal reforms aimed at expanding public services and at distributing their costs more equally has considerably constrained the tasks and responsibilities of local states, preventing them from developing more completely both their structures and their functions (Huber and Stephens, 2012). Democratization and polarized antagonism, finally, may undermine state institutions, and democratic requests may occasionally overburden and destabilize political and state undertakings (Hagopian and Mainwaring, 2005).

We also find that political partisanship affects stateness. Our conclusions are partly in line with, and complete, those by Cardenas (2010), Haggard and Kaufmann (2008), Acemoglu and Robinson (2008), and Hagopian and Mainwaring (2005). Cardenas claims that the weakness of the state in Latin America depends on pervasive economic inequality: economic inequality, however, is also a reflection of left parties’ weakness, and we illustrated the mechanisms relating these parties to state capacity. For Haggard and Kaufmann, democracy may strengthen the state by promoting more progressive social welfare systems. In Latin America, in turn, more progressive social welfare systems have been usually associated with the presence of a stronger left: left governments have more incisively promoted redistributive welfare and prevented the purposeful boycotting of crucial state powers such as tax extraction, which encroached upon the privileges of local elites. ²⁵ Thus, left parties have fostered both a deepening of democracy and a containment of social and economic privileges, which have contributed to a strengthening of state capacity: it is through the crucial intermediation of these parties that the gap between political and social powers has been reduced, and state strength has somehow increased in the region. ²⁶ In particular, a ‘new left’ emerged in the subcontinent at the end of the 1990s, which accepts the essential principles of market economics, while endorsing reforms such as the enactment of welfare plans for the least advantaged, a revived concern for public safety, a more active role for the state as supervisor and arbitrator between capital and labor, the growth and upgrading of public services, and the inauguration of a more progressive tax system (Barrett et al., 2008: 22). ²⁷

In Chile, for instance, the Lagos (2000–06) and Bachelet (2006–10) administrations sponsored important reforms in the area of public health, education, and social security, which resulted in an unprecedented growth of social benefits aimed at its poorest citizens; passed a series of laws on probity in the Public Administration (2003) and transparency (2009); reduced the number of civil servants directly designated by the executive; and introduced a Senior Management Service System (Alta Dirección Pública), whose positions were filled through public competition, thus making civil service careers more professional and meritocratic. The armed forces’ special prerogatives over elected officials, in addition, were drastically reduced (Bertelsmann, 2010). ²⁸ In the welfare sector, in particular, President Lagos promoted a Regime of Explicit Health Guarantees (Plan de Acceso Universal con Garantías Explícitas, also known as Plano AUGE), whose objective was to expand the quality and availability of public sector health services, especially for lower-income groups. In its original form, the plan directed to deliver medical care to people meeting certain age requirements and suffering from one of a set of specified diseases. The AUGE plan set new standards for both the quality and the quantity of the services offered, establishing maximum waiting times for the provision of medical treatment and ensuring that the necessary activities, procedures, and technologies would be provided only by registered and accredited healthcare professionals (Missoni and Solimano, 2010). This reform, finally, was to be accompanied by an information system and by studies to assess compliance monitoring and impact assessment. In short, a series of institutions and practices were deployed and strengthened, which have been usually associated with capable states.

What characterizes these policies is the political will to mold the state as a mechanism to address the huge social deficits caused by neoliberal policies and, in some cases, to re-establish an institutional infrastructure to seek to ensure a more responsive, democratic, and well-functioning public administration. ²⁹ Thus, although the ‘new left’ governments ‘have opted for the strategy of “bending and moulding” the existing political institutions and economic model rather than doing away with it … it may be that when the margins are bent and molded enough, they will change their relation to the core to the extent that the core itself will become something different and better’ (Panizza, 2005: 730). To sum up, the political agenda of the left in the region has generally tended to develop, both historically and more recently, some of the abilities and resources usually associated with state capacity. In particular, ‘new left’ governments have strengthened the state’s role in social provision, implementing programs that are more extensive in their reach and that provide better and more complete benefits (Kirby, 2009). Specifically, these governments have supplied a democratic and inclusive social contract, although within the confines of market-oriented, export-led growth (Grugel and Riggirozzi, 2012). ³⁰

These conclusions do not rule out that right governments might strengthen state capacity. In Colombia, for instance, state capacity increased under a right executive, whereas other factors that according to the literature contribute to this result, such as democratic quality, were absent or weakened. Although the actual extent of their success remains unclear (Feldmann, 2012), right-wing governments contributed to restore political order, in relation to both political and common violence. Since Uribe’s peace talks with the AUC (Autodefensas Unidas de Colombia) started in 2003, hundreds of paramilitary leaders have been jailed and a dozen key chiefs were handed over to the United States. In short, a demobilization process was initiated that translated into the neutralization of a formidable military force. Homicide rates have also plunged from ~28,000 in 2002 to 15,459 in 2010. Tax extraction capacity has likewise improved, increasing from 1990 (10.9% of GDP) to 2009 (17.8% of GDP) after reaching a peak of 18.2% between 2005 and 2007 (Gómez et al., 2012: 13). ³¹ Civil service performance, finally, benefited from a 2004 law that sought to put an end to 5 years of legal ambiguity in which provisional appointments reached 38% of the career personnel, and new merit-based criteria were introduced for entering the civil service (Grindle, 2010: 22). In short, right governments improved especially political order and were able to expand overall state capacity.

Conclusion

Our study adds to a topic of research – namely, the impact of democracy on state capacity – which has hardly been addressed thus far. Although the bearing of democracy on state capacity has been discussed and empirically analyzed by (a limited number of) previous studies, we offer a novel investigation on the role of partisanship in the process, outline the mechanisms by which democracy and EPB affect state capacity, and measure their effect on state strength. In Latin America, left governments have facilitated the active political inclusion of less-favored social groups, by empowering states to resist pressures from dominant classes opposed to income redistribution or more intense tax extraction, thus strengthening administrative capacity and overall stateness. At least during the period covered by this study, these goals were not accomplished by all democratic governments in Latin America and democracy, by itself, has been unable to unchain this process. The recent appearance of ‘new left’ governments seems to have fueled a novel phase of state strengthening, which counters the downsizing of state capacity that took place in previous decades. The extent and historical relevance of this shift needs to be fully clarified: we have documented, however, its empirical thrust and underlined its theoretical relevance. Although political parties of different convictions have also been able to strengthen state capacity, by implementing different policies in relevant areas of stateness, we maintain that, the latest expansion of left governments has resulted in stronger states, with higher administrative capacity and higher levels of political order.

Our results open intriguing avenues of research on the consequences of democracy and complement similar studies on the consequences of left parties’ power in the area (Huber et al., 2006). In addition, our approach tackles a major criticism levied against democratization studies: too often political scientists have focused on the formal requisites of democracy and overlooked the constellation of de facto social and economic powers that decisively shape the political trajectories and fortunes of people in the area. Our analysis takes both aspects into account: the strategy of left parties conjugates broader political participation with more extended social and political programs for the least advantaged and with an improvement of domestic political order. State capacity is a crucial instrument by which fuller political participation may translate into valuable social gains for the weakest in society. By granting the political incorporation of formerly excluded social groups and strengthening the capacity of the state to intervene in society to reduce social and economic inequalities, left parties make democracy more mature and, at the same time, inhibit the factual powers that effectively restrain progressive social change, thus reducing the gap between ‘formal’ and ‘substantial’ democracy, which allowed to promote stateness. Accordingly, our analysis carries major practical implications: if state capacity may be directed and stimulated politically, developing countries and governments may hope to strengthen their capacity to tackle the problems they have traditionally faced, from poverty to illiteracy and from political violence to public corruption, thus improving their chances to advance their social and economic conditions, while deepening the participatory and representative nature of their respective political regimes.