Introduction

Over the past decade an important new literature has emerged in comparative politics on the juxtaposition of political regimes, which Edward Gibson defines as "situations where two levels of government with jurisdiction over the same territory operate under different regimes" (2005, 103). Gibson's highly generative work on this topic has triggered a vibrant debate about how to define, and what explains, the phenomenon of subnational authoritarianism within nationally democratic regimes.¹ Even as the juxtaposition of political regimes receives the growing attention it deserves among comparativists, I argue in this article that the study of subnational politics in Latin America should also begin to examine the territorial juxtaposition of policy regimes.

If, as most scholars would agree, political and economic liberalization were the defining trends of the late 20^th century in Latin America, it is important to note that the unfolding of both trends -and not just the first- has generated a great deal of territorial dissonance between the national and subnational levels. Just as political liberalization in the form of democratization set into relief the persistence of subnational spaces of authoritarianism, so too did economic liberalization set in motion illiberal economic policy behaviors by subnational actors, as Snyder has documented in his pioneering work on re-regulation within Mexican states (2006). More recently, the national-level shift back to statism and the repudiation of neoliberal models that took place in the course of the left turn since 2000 have triggered the defense of more market-friendly projects at the municipal or departmental level, as in Argentina, Bolivia, and Ecuador. Thus, in contrast to political regime juxtaposition, where the democrats are always national and the authoritarians always subnational, policy regime juxtaposition is characterized by a more flexible dynamic in which liberals and statists can occupy either national or subnational positions.

My central claim in this article is that subnational officials in Latin America are increasingly trying to design, implement and, if necessary, defend subnational policy regimes that deviate ideologically from the content of national policy regimes. We are witnessing not just the simultaneous survival of democratic and authoritarian political regimes within the same national territory, but also the simultaneous pursuit of neoliberal and statist policy regimes by national and subnational governments within the same national territory. Put slightly differently, the ideological struggle between neoliberal and statist development models that has roiled politics in Latin America ever since independence is increasingly unfolding along territorial lines; mayors and governors, now that they are elected almost everywhere, are frequently challenging the policy preferences of elected presidents. This is important because when subnational governments are able to develop ideologically discordant policy regimes, they are in effect imposing territorial limits on the scope of the national government's policy authority. In this fashion, subnational policy regimes potentially represent a threat to one of the central state's defining attributes, namely the prerogative of imposing its preferred policies uniformly throughout the national territory (Mann 1986, Soifer and vom Hau 2008). When these subnational policy regimes manage to persist, the result is policy regime juxtaposition, a term I use to refer to the simultaneous pursuit of ideologically discordant policy regimes at different levels of government within the same country.

It is important to clarify at the outset that the emergence of this phenomenon does not necessarily require federalism, even if the kinds of powers assigned to intermediate-level governments under federalism likely augment their ability to build subnational policy regimes. Nor is it the case that only intermediate-level governments (e.g. provinces, departments, states, regions) can generate divergent policy regimes; local-level governments (e.g. cities and municipalities) can do so as well. Whenever two separately elected governments -one national and one subnational- can jointly claim governing authority over the same territory, I argue that this arrangement opens up the possibility of policy regime juxtaposition. The same is true of political regime juxtaposition; subnational authoritarianism is also a possibility in formally unitary countries like Colombia and Peru (Eaton and Prieto forthcoming, Levitsky 2014), even though the literature on this phenomenon has focused exclusively on federal countries like Argentina, Brazil, and Mexico (Behrend 2012; Benton 2012; Durazo Herrmann 2010; Gervasoni 2010, Gibson 2005 and 2010, Giraudy 2013 and 2015).² Not only is the juxtaposition of political and policy regimes not limited to federalism, but there are good reasons to suspect that the design of state institutions (e.g. federalism vs. unitarism) matters less than the underlying strength of these institutions when it comes to the kinds of policy challenges that subnational officials are able to articulate and advance. It is perhaps no coincidence that the most powerful subnational policy challenges in Latin America have emerged in countries with especially low levels of state capacity, including the three unitary countries that I draw upon in this article: Bolivia, Ecuador, and Peru.³

When we turn from the study of juxtaposed "political regimes" to the study of juxtaposed "policy regimes," we are moving from a concept that has been a central topic of study for generations of comparativists to a concept that instead has been developed largely within the American politics subfield of political science. Here I am borrowing from the debate that has emerged between two different ways of conceptualizing the (American) public policy process: "policy subsystems" and "policy regimes." For decades, much of the policy literature has emphasized the salience of "policy subsystems," alternatively labelled "policy domains" (Burstein 1991), and "subgovernments" (Berry 1989). According to Baumgartner and Jones (1991, 1045), policy subsystems result from "the ability of single-industry economic interests to insulate themselves from the influence of large-scale democratic forces through the creation of relatively independent depoliticized subgovernments." Examples of regulatory capture by single industries in the U.S. -and of the "iron triangles" that formed between businesses, bureaucrats, and legislators- can be seen in a number of policy subsystems, including for tobacco, airlines, trucking, and telecommunications (Sabatier and Jenkins-Smith 1993). On the one hand, the idea that there is no single national policy system but rather multiple and differently constituted policy subsystems appears to be compatible with my argument that subnational policy models can diverge significantly from national models within the same country. On the other hand, the development of policy subsystems as they have been documented in the U.S. case seems to require highly stable institutional environments, suggesting that this concept might not travel easily to more fluid institutional settings.

More recently, scholars in the public policy literature have criticized the focus on policy subsystems as overly narrow, and shifted to the study of policy regimes -a more capacious concept that holds greater promise for the study of policy processes in Latin America (Wilson 2000, Jochim and May 2010). According to Peter May and Ashley Jochim (2013, 428), policy regimes are "governing arrangements for addressing policy problems." Regardless of the specific substance of the policy problem in question (e.g. civil rights, environmental issues, health care), a policy regime is always composed of three main elements: "a central idea that constitutes its purpose, the institutional arrangements that structure the regime, and the interests that comprise the bases of support and opposition (May and Jochim 2013, 437)." In adapting the concept of "policy regime" from the U.S. politics literature, I use the term to refer not to any one policy issue but rather to a broader set of economic policies (e.g. taxation, regulation, privatization, and trade and financial policies) that together reflect either a state-centered or market-centered model of economic development.⁴ In this sense, I consider "policy regime" to be synonymous with "development model"; each term refers to a package of public policies and institutional practices that result from a common set of ideas and beliefs about the appropriate role for market forces and the appropriate levels and types of state intervention in economic life.⁵

This paper has three main objectives, each in the spirit of trying to broaden the study of subnational politics in Latin America so that it includes the juxtaposition of policy regimes and not just political regimes.⁶In the first section, I identify the general causes that have led to the proliferation of subnational policy challenges, some of which end up resulting in full-blown cases of policy regime juxtaposition. Why are we seeing more territorial heterogeneity within countries in terms of the pursuit of ideologically disparate economic development models? The second section assesses the importance of this trend by analyzing the chief advantages and disadvantages of policy regime juxtaposition. Moving beyond the question of why it has become more common, I ask whether juxtaposition is positive or negative -a question seldom posed in the literature on political regime juxtaposition due to a strong scholarly consensus about the superiority of democracy relative to authoritarianism. The absence of such a consensus about the superiority of neoliberalism or statism makes this a fruitful question to explore. In the third section, which is based on three countries in the central Andes (Bolivia, Ecuador, and Peru), I turn to the more difficult question of why subnational officials are able to defend ideologically deviant policy regimes in some cases, resulting in policy regime juxtaposition, but not in others. The argument that I develop -and illustrate with necessarily brief references to these three cases- straddles the state/society divide by emphasizing the importance of both institutional capacity and societal coalitions.

1. Drivers of Policy Regime Juxtaposition

Attempts by subnational officials in Latin America to mount their own policy regimes are especially striking at the start of the 21^st century because of what the 20^th century looked like in the region. Virtually all Latin American countries began that century governed by fundamentally liberal development models, followed by widespread experimentation with statism in the middle fifty years of the century, and ending with the subsequent return to market-centered approaches in the century's closing decades. Individual countries differed in the exact timing, scope, and depth of the policy reforms they adopted as part of each successive shift, but to a striking degree Latin America as a region moved in concert as national governments adopted and then shed common approaches (Corrales 2003, Sheahan 2002). One important pattern of continuity as the policy pendulum swung back and forth between liberal and statist policy regimes in the 20^th century was the unimportance of subnational actors like mayors and governors, who did not meaningfully influence the ideological debate over the market, even as this debate triggered frequent regime changes, extensive social mobilization, and in some cases high levels of political violence (Kurtz 2004; Oxhorn and Ducatenzeiler 1998). This 20^th century backdrop sets into sharp relief the protagonism that subnational elected officials are now exerting in the 21^st century, which is starting to look as a result much more like the 19^th century -a period in which subnational caudillos routinely took up arms against the center over a range of matters, including economic policy (Bethell 1985).

A number of factors, some global in their reach and some more specific to the region, help account for why we are seeing more robust and aggressive efforts by subnational officials in Latin America to fashion their own policy regimes -even if these clash with the ideological content of the national policy regime. One key driver is globalization, which has increased opportunities for some subnational regions to insert themselves into the world economy as exporters of the goods that they have comparative advantages in producing, more or less independently from their respective national governments (Piore and Sabel 1986; Scott 1998). Less fortunate subnational regions, in contrast, have been left further behind by globalization in ways that have triggered the defensive creation of distinct subnational policy regimes (Brenner 2004; Geddes 2014). What this means is that globalization is especially potent because it can trigger subnational challenges on the part of regional winners and losers (Castells and Borja 2004).

Democratization has also likely increased the scope for serious subnational policy challenges and for the possibility of policy regime juxtaposition. Democratically elected national officials may well oppose the emergence of distinct subnational policy regimes, and if so inclined they can use the full range of their powers in the attempt to attenuate policy challenges from below. Authoritarian national governments, however, are even less likely to tolerate subnational deviations, particularly in cases where authoritarian actors like the armed forces took power to fix the national economy or to pursue a particular vision of economic development (Hirschman 1979; O'Donnell 1973).⁷Even if the third wave of democratization has tended to produce thin and feeble democratic regimes more often than robust consolidated democracies (Hagopian and Mainwaring 2005), transitions to democracy in the 1980s and 1990s created permissive conditions for the generation of subnational policy challenges.

Particularly where voters can cast separate ballots in national and subnational races, and where these races are held at different times in the electoral calendar (Carey and Shugart 1992), democracy may well encourage policy regime juxtaposition. After all, voters may look to different levels of government for different policy solutions and for different levels of engagement in the economy. For instance, voters might prefer more interventionism from local governments and less interventionism on the part of the more distant national government, whose actions may be deemed to disproportionately benefit residents in other parts of the country. Or, conversely, doubting the competence of local authorities whom they can observe up close, voters may prefer greater intervention by the national government officials whom they consider to be better trained or more competent.

By definition, the ability to mount subnational policy challenges depends on some degree of decentralization, which should be considered the main driver of the tighter connection we are seeing today between ideological and territorial conflicts. The trend toward political, fiscal, and administrative decentralization that began in Latin America in the 1980s has done much to lay the foundation for policy regime juxtaposition (Eaton 2004; Falleti 2010; Goldfrank 2011). Typically taking the form of separate elections for subnational officials, political decentralization has introduced an entirely new arena for ideological contestation over the market, creating the possibility that subnational officials will win electoral mandates to pursue very different economic models relative to the national level. Fiscal decentralization, meanwhile, provides a double impetus in that it both enhances the resources that elected subnational officials can devote toward the pursuit of their preferred approaches, while also endowing them with fiscal policy tools they can use to behave in either a liberal or statist fashion (i.e. either lowering taxes to promote investment or raising taxes to finance expenditure). Administrative decentralization has also contributed to this phenomenon since it gives subnational officials real choices in terms of how much of a role they want the private sector to play in the provision of critical public services like health, education, and infrastructure.

Decentralization has undeniably set the stage for more meaningful subnational policy challenges around the world, and yet it does not necessarily imply that distinct subnational policy regimes will automatically take shape. Subnational officials can make extensive use of newly decentralized authorities without their policy choices congealing into an identifiable and coherent policy regime or challenge. The key question with respect to policy regime juxtaposition is whether subnational officials who are committed to either liberalism or statism can use the powers of their offices to act on those ideological commitments. In and of itself, decentralization does not tell us enough about the outcome of attempts by subnational officials to use their powers in this fashion. Furthermore, it is important not to overstate the significance of formal decentralization because the reality is that subnational elected officials, particularly in the context of weak central states, often enjoy significant de facto autonomy to act on their ideological preferences, even in the absence of high levels of de jure policy decentralization. A case in point is four-term Guayaquil Mayor Jaime Nebot (2000 to the present), who announced that he would embark on a process of "autonomía al andar" when confronted with national government resistance to formal decentralizing measures.

In addition to worldwide trends like globalization, democratization and decentralization, several other developments that are especially salient in Latin America have increased the likelihood of subnational policy challenges and distinct subnational policy regimes. Across the region, countries have experienced high levels of party system instability, characterized by the collapse of traditional political parties in some cases (Seawright 2012; Tanaka 2002). Party system turmoil has threatened the ability of powerful economic interests to achieve representation in the national government through the same political parties they previously relied on for this purpose. In some countries, the traditional parties that contested national elections have not disappeared altogether but rather survive at the subnational level (Cyr forthcoming; Freidenberg and Suárez-Cao 2014). Where parties that formerly represented elite interests can no longer hope to win national elections or can no longer be expected to exert great influence in national-level policymaking processes, one response might be to pay greater attention to the types of policy solutions that subnational governments can provide. Economic elites in this situation certainly have other options, including the pursuit of non-partisan mechanisms of representation at the national level (Levitsky, Loxton and Van Dyck 2016), but the greater difficulty of accessing national power via parties should throw into greater relief the potential importance of subnational policy regimes. Party system turbulence has taken different forms in Bolivia, Ecuador and Peru, but in all three cases the collapse of the national party system has increased the importance of being able to shape and defend distinct policy approaches at the subnational level.

Looking across the region brings into focus another common cause of the greater frequency of subnational policy challenges in Latin America today: ethnic mobilization. Against a backdrop of centuries of political subordination, new indigenous social movements in recent decades have dramatically expanded their capacity to influence national-level policy debates, not just by placing indigenous leaders in key bureaucratic positions but by winning presidential and legislative elections as well (Chartock 2013; Madrid 2012; Yashar 2005). While the perhaps more dramatic story here is that indigenous mobilization has led to the transformation of national policy regimes, it has also triggered the articulation of demands for alternative subnational policy regimes. This can even happen simultaneously in the same country, as in Bolivia where eastern indigenous communities have sought to enhance the policy authority of subnational units even as western indigenous groups enjoy unprecedented control over national policy levers through the dominance of the governing MAS party (Postero 2006 and 2010). More generally, where ethnic claims overlap with territorial demands for autonomy, as in the lowland areas of the Amazonian basin, the political mobilization of ethnic identities has prioritized not so much the transformation of national policy regimes as the right to pursue territorially distinct local models (Tockman and Cameron 2014).

2. Advantages and Disadvantages of Policy Regime Juxtaposition

Bracketing the question of why subnational actors are more frequently launching and defending policy regimes that clash with the national policy regime, is this a desirable trend? From a normative perspective, what are the various advantages and disadvantages of territorially juxtaposed policy regimes? An exploration of the potential costs and benefits that might be generated by policy regime juxtaposition underscores the importance of better understanding this growing phenomenon. In contrast to the literature on the juxtaposition of political regimes, which is animated by a widely-shared normative desire to promote the democratization of subnational authoritarian enclaves, the relative merit of market-oriented versus state-centered approaches has remained an open debate among scholars. As a result, greater normative discord may characterize the study of the phenomenon of policy regime juxtaposition relative to political regime juxtaposition.

There are a number of reasons to emphasize the positive effects generated by territorially juxtaposed policy regimes. First, the phenomenon reflects movement away from a one-size-fits-all dynamic in which subnational governments are not allowed to pursue development models that may elicit majority support subnationally but not nationally. In some respects, policy regime juxtaposition is simply an outcome that follows logically from one of the primary motives behind decentralization, which is that it enables policy innovation and experimentation (Cheema and Rondinelli 2007; Sabel and Zeitlin 2012). If there are strong normative reasons to support decentralization, including meaningful variation in citizen preferences across subnational units and the likelihood that local politicians have better local knowledge than do national politicians (Oates 1972; Weingast 2014), then there are likewise strong reasons to conclude that it is positive when subnational officials can craft coherent models with their policy authority.

A second possible benefit of policy regime juxtaposition is that it may enable voters at a given point in time to make comparative assessments of the two regimes in contention. When only a single, national policy regime is in place, voters can only engage in retrospective comparisons, comparing their current information about the current national model with their memories of the relative merits of the previous national model. Cross-temporal comparisons are complicated, however, not just by the quality of people's memories, but by the exogenous factors that impact the implementation of any model. How, for example, to compare the performance of statist models during the commodity boom of the 2000s in Latin America with the neoliberal model implemented during the 1980s and 1990s at a time of fiscal austerity and financial crisis? While the comparison of territorially juxtaposed policy regimes in real time suffers from other problems -namely the difficulty of comparing models across scales of government-, voters may be exposed to better and more useful information about the differences between statism and neoliberalism when they are repeatedly exposed to and reminded of these differences due to the interactions of contemporary incumbents -defending different ideological approaches- at different levels of government.

Third, policy regime juxtaposition may reduce the size and/or the frequency of the often extreme policy swings that have been so destabilizing in Latin America. Writing on the instability of economic strategies in Latin America, John Sheahan argues that "these countries change their versions of capitalism much more frequently, and more radically, than European countries do (2002, 25)." With respect to the size of past swings, policy regime juxtaposition may moderate the tendency of national governments to experiment with extreme versions of either model, whether in the form of "Gerschenkron gone awry" in the mid-20^th century (Corrales 2003) or the decision to privatize even profitable state-owned enterprises in some countries in the 1990s. Policy regime juxtaposition may have a dampening effect in either direction. With respect to the frequency of past swings, irrespective of one's ideological preferences and the value or lack thereof that one sees in the content of neoliberal or statist approaches, a case can be made that rapid and relatively short-lived changes are in themselves negative. Furthermore, not unlike democracy, which lowers the stakes of electoral contests for losers by giving them an opportunity to win in the future (Przeworski 1991), the ability to construct subnational policy regimes can lower the stakes associated with losing the national-level battle over development models. This may be especially important for economic elites, whose past support for democracy in Latin America was often highly contingent (Gibson 1996), and whose ability to protect their economic interests through subnational policy regimes might increase their tolerance for the continuation of democracy at the national level. More generally, a country's political life may become more pluralistic when subnational officials are able to defend deviant local models from the national incumbents who would prefer to impose and enforce their own models; economic pluralism may beget political pluralism.

As a final possible advantage, the pursuit of ideologically distinct models at the national and subnational levels may enable and promote convergence between neoliberal and statist approaches in the medium- to long-term. Policy regime juxtaposition can bring into focus similarities rather than differences between market-oriented and state-centered approaches, particularly at a time when environmental and indigenous groups are articulating truly revolutionary critiques of both types of approaches. In this way, policy regime juxtaposition could appeal to pragmatists who would prefer a less ideologically charged and inflexible debate over economic development, and who wish to focus the conversation instead on the identification and emulation of practical solutions and effective partnerships between the state and the private sector. Ironically, more pragmatism and less ideology might be one of the side effects of policy regime juxtaposition.

On the other hand, there are significant potential disadvantages associated with successful subnational policy challenges and the territorial juxtaposition of distinct policy regimes. First, in the context of policy regime juxtaposition, the territory circumscribed within a particular subnational unit (i.e. municipality, department, region) is in effect being governed by two different regimes, which can generate substantial contradictions. Particularly where decentralization has taken the form of shared rather than exclusive competencies over public policy fields, juxtaposition may be a recipe for chronic uncertainty. But even where there is a clearer devolution of specific policy authorities to subnational levels, the national government still retains important oversight and monitoring roles in ways that can produce confusion for the local private sector. At a minimum, policy regime juxtaposition increases transaction costs by requiring economic actors to master and stay on top of the rules and incentives emanating from two rival models. These costs could be especially onerous for foreign investors, who have less access to high-quality information about domestic political realities and work-arounds, along with fewer local relationships to help them navigate between disparate models.

A second disadvantage can be identified as the flip-side of the presumed benefits of convergence and pragmatism discussed above. Firm believers in either neoliberalism or statism would argue that the strength of the incentives created by either model is undercut or diluted by the presence of a different set of incentives at another level of government within the same country. In this view, subnational governments should be prevented from being able to pursue deviant approaches that counter the logic of national models. That subnational policy regimes might have a moderating effect should be seen as a negative rather than a positive factor, and is precisely why they should be disabled. For example, defenders of neoliberal approaches at the national level might fear the market distortions created when important subnational governments establish their own state-owned enterprises or use state-owned banks to lend to preferred borrowers. In the opposite configuration, national advocates of statism might have a harder time encouraging private sector actors to reorient their behavior toward industrialization, production for the domestic market, and/or the use of domestic (rather than imported) inputs if subnational governments are simultaneously providing vigorous support for exporting and importing activities.

Most importantly, policy regime juxtaposition in the Latin American context is likely to generate significant political conflict and consume scarce political resources that could be devoted to other pressing priorities. In the 20^th century, shifts between national development models tended to produce acute political conflict in the period leading up to the shift, and then more low-grade conflict once the new model was in place (Fitzgerald 2005). In the contemporary period, marked by greater lag effects produced by territorial resistance and the greater persistence of distinct subnational models, political conflict may be intense and nearly constant. The politics of policy regime juxtaposition are often characterized by 1) aggressive national attempts to stamp out the subnational deviation, which understandably provokes stiff local resistance, and 2) equally aggressive subnational attempts to use subnational experiences as the basis for successful efforts to overthrow the national model. In more institutionalized political settings, distinct subnational policy regimes may be less threatening to the center, and policy regime juxtaposition seems to have resulted in "cohabitation" rather than "contention" (Jones and Keating 1995). But in less institutionalized contexts, territorially juxtaposed policy regimes are likely to be the source of deep, ongoing, and even violent political conflict in ways that threaten human security, as in Bolivia and Peru and, to a lesser extent, Ecuador. In Bolivia, many lost their lives in the context of the hard-fought policy conflict between east and west that pushed the country to the brink of civil war in 2008, and many Peruvians across the country have died in local protests against the neoliberal mining policies insisted upon by the national government.

According to the argument developed so far in this paper, the juxtaposition of policy regimes is likely to become increasingly common, and the jury is very much still out in terms of whether this is good or bad for a range of important outcomes, including development, democracy, and security.

3. Policy Regime Juxtaposition as the Result of Institutional Capacity and Societal Coalitions

If we appear to be at the dawn of a new period in which subnational officials feel empowered to challenge the policy direction of the national governments in Latin America, what explains variation in the effectiveness of these challenges? Why are elected officials able to build and defend their own divergent policy regimes in some subnational jurisdictions, but not in others? If the introduction of elections for subnational officials has transformed them into more important players in their countries' ideological struggles over the market, when do they win or lose those struggles? In the attempt to uncover plausible answers to these questions, I have conducted field research over the past decade in the three Latin American countries where ideological conflict over the market has generated especially salient territorial conflicts: Bolivia, Ecuador, and Peru. In Bolivia and Ecuador, the left turn in the first decade of the 21^st century brought to power national governments intent on implementing statist policy regimes, which clashed fundamentally with the distinctly neoliberal policy regimes in place in the most powerful subnational districts in each country: Santa Cruz and Guayaquil, respectively. In Peru, in contrast, the maintenance of a profoundly neoliberal national policy regime since 1990 has triggered efforts by a variety of left-of-center regional presidents to deviate from neoliberalism and to pursue more state-centered options in the regions. In other words, where the advocates of neoliberalism and statism are located, territorially speaking, differs across these three cases; in Bolivia and Ecuador subnational neoliberals have done battle with national statists, and in Peru neoliberals at the national level have sought to stamp out subnational projects that challenge neoliberalism.

In the remainder of this article, I aim to show that two key variables are especially helpful in explaining these dynamics: the institutional capacity of the governments that subnational officials control, and the breadth and depth of the local societal coalitions they are able to build in defense of their preferred policies. By focusing on capacity and coalitions, I do not mean to suggest that they are the only important factors, simply that they likely deserve a great deal of attention in the effort to understand the contours of policy regime juxtaposition.

a. Institutional Capacity

Subnational officials are unequal in the institutional capacity of the governments they command by virtue of their direct election. When countries around the world opted recently to decentralize responsibilities to subnational governments, they adopted laws re-assigning authorities between levels of government that took place immediately upon passage of the legislation. Formal authority can be shifted from one day to the next, but the capacity necessary to make use of newly devolved authorities is another matter. According to Painter and Pierre (2005, 2), "capacity can be defined as the ability to marshal the necessary resources to make intelligent collective choices about and set strategic direction for the allocation of scarce resources to public ends." Although administrative capacity can dissipate quite rapidly when politicians decide to use the state bureaucracy for short-term political purposes (Geddes 1996), it typically takes a long period of time for this kind of capacity to accrue and accumulate. Not unlike their counterparts at the national level, for most mayors and governors hoping to deliver on their policy mandates while in office, institutional (in)capacity is mostly experienced as a given, and as a key constraint that is quite difficult to overcome within the short time horizons facing most politicians.

Until the recent adoption of decentralization, subnational governments around the world did not need to develop much capacity because centralizing traditions in most countries denied them important governing roles. The transfer of real responsibilities, however, has exposed the problem of insufficient capacity at subnational levels, as has the preference for devolution (to subnational bureaucrats) over deconcentration (which merely relocates national bureaucrats to the subnational branches of central ministries). The capacity problem, as documented by the vast literature on decentralization, takes various forms, including manifest failures in delivering more administratively challenging services, along with greater dependence on fiscal transfers due to the inability of subnational governments to extract resources locally (Bardhan 2002; Smoke, Gomez and Peterson 2007). Capacity itself, according to Wu, Ramesh and Howlett (2015, 167-168), comes in three different forms -analytical, operational, and political- all of which are necessary for subnational governments to construct and defend policy regimes. Subnational challengers must be able to devise policy actions that are technically sound (analytical capacity), align resources with policy actions to enable their implementation (operational capacity), and solicit political support for policy actions (political capacity).

I argue that prospects for the success of subnational policy challenges are enhanced if elected officials can count on relatively capable subnational governments. Subnational executives need to be able to design, finance, implement, and monitor their own policies more or less independently of the national government if they are going to be able to sustain discordant subnational policy regimes. Although institutional capacity is necessary for the construction of any coherent subnational policy regime, it stands to reason that it is especially critical for the establishment of a statist (as opposed to neoliberal) policy regime given the greater roles that subnational administrations are expected to play in this kind of model. As we know from the literature on the politics of liberalization as a national-level phenomenon, it is a mistake to overlook the critical forms of state capacity on which even market-centered economic models depend (Kahler 1990; Fukuyama 2004). This is true at the subnational level as well, where neoliberal mayors and governors are rarely indifferent to state capacity; indeed they typically argue that limiting the scope of what their governments do enables them to excel at those few responsibilities deemed critical for market-driven growth, including quality infrastructure, efficient transport systems and effective policing. And yet the kinds of professional training, technical know-how and personnel systems that are required to build a state-centered model involve a much higher level of both complexity and flexibility. Picking winners through subnational industrial policy, running subnational banks, and guaranteeing universal coverage to social services provided by subnational governments all place serious demands on the quality of subnational bureaucracies.

If the institutional capacity of the subnational state matters, where does it come from? While the comparative politics literature has focused more on the question of why state capacity varies across countries rather than across subnational units within a given country, this is beginning to change thanks to a new body of research. The answer, according to Singh (2015), lies in subnational identities or collective forms of "we-ness" that vary significantly across the national territory, and that can generate not just higher levels of public goods provisioning but deeper investments in subnational state capacity. Questioning demand-side explanations and arguing for a "decentralized path to infrastructural power," Ziblatt emphasizes the importance of the "priorities, ambitions and goals" of the professionals who work within the (subnational) state and who exhibit varying levels of professionalism (2008, 276). Others have focused on the importance of electoral competition at the subnational level. Whereas Fiszbein (1997) attributes higher levels of local capacity to more locally competitive electoral environments, Pribble (2015) adopts a more nuanced perspective, arguing that electoral competition may encourage capacity building efforts but also compromise these efforts if it generates high levels of turnover. Finally, other scholars adopt an approach closer to my own, emphasizing that subnational economic elites hold the keys for understanding how and why state capacity varies across territory. Boone (2003) emphasizes the differential bargaining power of local elites as they interacted with national state builders in Ghana, Côte d'Ivoire and Senegal, whereas Montero (2002) points to the importance of competition among regional elites, which discouraged subnational governments from creating insulated public sector agencies in select Brazilian states.

Like these authors, I emphasize the importance of the link between state capacity and regionalized economic elites. According to the argument I derive from the Bolivian and Ecuadorian cases, the historical absence of the central state in Santa Cruz and Guayaquil respectively led local business elites to invest heavily in a series of quasi-public, privately-controlled subnational institutions within the regions of their influence. Many subnational governments in Bolivia and Ecuador suffer from debilitating capacity deficits, but Guayaquil and Santa Cruz have built up relatively strong administrative structures over the course of several decades. In both cases, local economic elites directed a gradual process of capacity building over many decades that would ultimately help explain the persistence of neoliberal policy regimes in each subnational region when they came under attack from leftist presidents at the national level after 2005. As Dunlop and Radaelli (forthcoming, 2) argue, administrative capacity "is not a static set of skills, but rather it is a set of relational goods produced by interactive relations between government organizations and the wide range of governance actors beyond"- in this case, regionalized economic elites.

In Bolivia, the story begins in the 1950s with the creation of the Pro-Santa Cruz Committee (Comité Pro-Santa Cruz or CPSC). Financed and led by socially prominent families opposed to the Revolution, the CPSC identified the lack of public services in Santa Cruz as the region's chief problem and won tremendous legitimacy when it successfully led the movement to decentralize oil royalties in 1957. Even as the CPSC emerged as Santa Cruz's leading institution, the elite interests who dominated it spun off a whole panoply of other institutions that are critical in understanding the region's liberal, outward-oriented and market-centered policy regime. These include the Regional Development Corporation (CORDECRUZ), three service cooperatives (COTAS, CRE and SAGUAPAC), and a number of regionally-specific business chambers (CAINCO, CAO and FEPB-SC). Each of the pieces of what are together referred to as "la institucionalidad cruceña" (Ferreira 2010, 163) has played a distinct role in supporting the region's neoliberal policy regime: CORDECRUZ enabled economic elites to determine where and how public revenues (including oil royalties) would be invested in infrastructure and services, cooperatives demonstrated that nongovernmental social provision could result in consistently high-quality services, and business chambers offered strong and consistent support for private property and an outward-looking economic orientation (Roca 2001, 604). CORDECRUZ in particular served as an important incubator for the development of public-sector local capacity, especially since it was absorbed by the prefecture in 1994 and by the departmental government a decade later. All of this means that when conservative landowner Ruben Costas became Santa Cruz's first-ever elected governor in 2005 as part of a broader attempt to defend the department's market model from Evo Morales and the MAS, he could count on deep reserves of local administrative capacity.

A broadly similar story unfolded in Ecuador, though it began at an even earlier date in the 19^th century. As in Santa Cruz, local institutions in Guayaquil were designed by local elites to provide critical services neglected by a national government that was centered in a far-away, Andean capital (Soifer, 2016). The most emblematic local institution is the Guayaquil Charity Board (Junta de Beneficiencia de Guayaquil or JBG), founded in 1888 by leading members of the cacao oligarchy and the banks that they established, including most importantly the Commercial Agricultural Bank of Guayaquil. Financed with resources from a lottery that the national government gave it exclusive rights to manage in 1893, the JBG has offered free and subsidized social services at a wide range of high-quality private facilities. Guayaquil elites also established and controlled a number of other high performing, quasi-public institutions, including the Corps of Firefighters (Cuerpo de Bomberos), the Guayas Transit Committee (Comité de Transito de Guayas or CTG), and the Roads Committee (Comité de Vialidad). These institutions, which had a public purpose but were under the control of the local private sector, are important today not just because some of them (e.g. the JBG) continue to provide key services, but because they set the stage for the development after 1992 of privatized models of governance within city hall. Thus, when President Rafael Correa moved aggressively to eliminate Guayaquil's neoliberal policy regime after his election in 2006, Jaime Nebot as mayor (2000-2019) was able to defend it by drawing upon the city's distinctive legacy of capacity-building.

Peru represents the opposite experience, where the absence of subnational capacity has derailed efforts to build subnational policy regimes. On the one hand, insufficient capacity in Peru is to be expected given that regional governments were only created in 2002. In general, the 25 regional presidents elected in each of the four cycles conducted so far (2002, 2006, 2010, 2014) have faced deep capacity deficits, and these deficits have hindered the ability of Peru's most ambitious regional presidents to act on their non-neoliberal policy preferences through the implementation of distinct approaches subnationally. On the other hand, the Peruvian case is important because it shows not simply that capacity is hard to come by, or that it is difficult to transfer extant capacity from the national to the subnational level. The problem is more pernicious: national governments may actively try to undercut efforts to build up subnational capacity, leaving subnational governments intensely dependent on the temporary consultants who cannot effectively substitute for public sector professionals. Two important recent programs that undermined capacity building in Peru's regions are the Voluntary Contributions (Aportes Voluntarios) and Works for Taxes (Obras por Impuestos) programs developed by Presidents Alán García (2006-2011) and Ollanta Humala (2011-2016), respectively. Under the former, mining companies were asked to contribute 3.75 % of their profits to funds that they then unilaterally used to finance projects in the communities where they extract resources, thereby circumventing regional governments and ignoring their plans (Campodónico 2011). In the latter program, participating companies can finance and execute local spending projects and deduct the cost of those projects from the income taxes that they owe the national government, after which the national government reduces its revenue transfers to the subnational government by the amount equal to the project cost -all of which further inhibits the accumulation of the kind of administrative capacity that regions would need in order to build their own policy regimes.

b. Societal Coalitions

Turning from capacity to coalitions, my second central claim is that subnational chief executives simply cannot succeed on their own. If individual mayors and governors stand alone in the policy challenges they pose to the president, chances are good that they will fail. Given the tremendous advantages enjoyed by the national chief executive in terms of media attention, fiscal resources and administrative leverage, subnational chief executives must enlist the support of civil society at the subnational level. To defend discordant subnational policy regimes, mayors and governors need to construct "internal" coalitions within the districts they govern, not just by mobilizing local beneficiaries but by marginalizing detractors -including those potentially activated, funded and coordinated by ideological adversaries within the national government.

Here I draw on a long tradition of coalitional analysis within the literature on comparative political economy, according to which the content and stability of national policy regimes can be fully explained only with reference to their underlying support coalitions. This same analytical approach may hold even more promise when we shift from the national to the subnational level given the necessarily embattled status of mayors and governors who are seeking to defend discordant local models. At any level of government, the core insight holds: no economic development model is viable in the long term without a political coalition to sustain it and to defend it from opponents. Political entrepreneurs can play important proactive roles in supplying policies that they adopt before the support coalition is in place (Waterbury 1993), but to a significant extent politicians are constrained by the availability of societal partners and by the regularity with which exogenous shocks in the global economy either empower or impoverish domestic economic groups (Gourevitch 1986). According to Gourevitch's pioneering work, coalitional arrangements are more important than any other causal factor (i.e. institutional, ideational, partisan) in explaining whether European countries opted for free trade or protectionism in response to major economic crises in the 1870s, 1930s and 1970s. More recently, scholars seeking to understand the neoliberal shift that began in the 1980s and 1990s have documented the new coalitions that emerged in the developing world to sustain market-oriented economic reforms, thereby calling into question the sufficiency of technocratic insulation and bureaucratic change teams (Conaghan and Malloy 1994; Schamis 1999).

Policy regime juxtaposition depends on the construction of locally -or regionally- specific societal coalitions with organized interests that represent the socioeconomic groups who stand to gain from the policy model preferred by the subnational executive. These will vary according to specific ideological configurations: business chambers and elite-dominated civic associations are key coalition partners in the defense of subnational neoliberalism, whereas NGOs, labor unions and neighborhood associations are likely to be more important in coalitions to defend subnational statism. A particularly critical question to ask about business as a coalitional partner is the extent to which economic elites in a given country are subject or not to important regional cleavages (Gibson 1996). In my three cases, regional cleavages among economic elites have been consistently prominent in Bolivia and Ecuador, and of decreasing significance in Peru -all with observable impacts on the success or failure of subnational policy challenges (Vergara 2015). But I also argue that it is important to look at coalition-building as a discursive and not just material practice. Subnational executives have a greater shot at winning their policy struggles when they can broaden their coalitions by emphasizing the territorial as opposed to purely ideological nature of their conflict with the president. When subnational chief executives can claim that important territorial prerogatives and identities beyond ideology are under threat, and that questions of "community" are at stake, their chances of enlisting the support of a larger percentage of the local population substantially increase (Hooghe and Marks 2016). Though subnational collective identities are real and significant, my cases suggest that mayors and governors are not above manipulating these identities in the service of their economic policy battles with the national government.

In Bolivia, the political defense of Santa Cruz's neoliberal model was coordinated by CPSC President (and later Santa Cruz Governor) Ruben Costas, who tapped into deep-seated fears that the MAS would intervene in the department's service cooperatives for water, telecommunications and electricity -all of which are controlled by local elites grouped into two long-standing "lodges." But building the case for autonomy to defend Santa Cruz's neoliberal model required building a much broader coalition -one that would go substantially beyond the economic and social elites who had dominated the CPSC for nearly 50 years. In response, the CPSC systematically reached out to non-elite groups like lowland indigenous groups and workers belonging to the Departmental Labor Confederation, representatives of which were brought onto the CPSC's governing body. In a more diffuse but also demonstrable fashion, the CPSC as the ultimate arbiter of cruceñidad (i.e. what it means to be cruceño) also reached out to migrants, striking a more inclusive tone by defining cruceños as "vivientes en Santa Cruz" (e.g., those who live in Santa Cruz) rather than "nacidos en Santa Cruz" (i.e. those who were born in Santa Cruz). The success of these coalition-broadening efforts can be seen in the multitudinous town meetings (Gran Cabildos) organized to defend Santa Cruz's model in 2004, 2005, 2006 and 2008.

In Ecuador, the strength of the local coalition constructed by Guayaquil Mayor Leon Febrés-Cordero (1992-2000) and maintained by his successor Jaime Nebot since 2000 -both from the conservative Social Christian Party (PSC)- helps explain the survival of a neoliberal policy regime that President Rafael Correa has repeatedly criticized and sought to dismantle. As the former head of the powerful Chamber of Industries, Febrés-Cordero was in an ideal position as mayor to enlist the support of local business associations in his effort to remake Guayaquil in a neoliberal image, which he accomplished through privatization and the streamlining of the municipal bureaucracy. Nebot then cemented this coalition through the extensive delegation of formerly municipal responsibilities to private foundations that provided services with public monies. When the model came under attack after Correa's election in 2006, Nebot relied heavily on elite-dominated civic associations to organize and finance a major demonstration in 2008, which sought not just to critique the president's statist proposals, but also to articulate a demand for autonomy sufficient to defend the city's distinct policy regime. Nebot managed to convince working class guayaquileños and popular sector organizations that it was their city that was under attack.

Finally, limited coalitional possibilities have severely handicapped subnational executives in Peru, most of whom confront a difficult societal landscape marked by fluidity, fragmentation, and the absence of coherent civil society organizations. As numerous scholars have argued, Peru's civil society was flattened by a prolonged internal armed conflict in which the Shining Path insurgency targeted for elimination the very organizations that could have served as coalitional partners for the contemporary left. Also important are the legacies of the Fujimori regime, which closed associational space and passed market reforms that led to the relocation of industries to Lima (Gonzales de Olarte 2000; Yashar 2005). The disappearance of organized labor from numerous regional economies means that unions cannot play the coalitional roles that were critical in the emergence of "municipal socialism" in countries like Brazil (Goldfrank and Schrank 2009). Although several regional presidents have sought to foster and coordinate anti-neoliberal protests against large-scale mining, protests have been largely ephemeral and have failed to produce much of an organizational residue once protests subside. Simply put, regional presidents on the left often find that there are few established groups with whom they can partner in their attempts to challenge the neoliberal approaches insisted upon by Lima.

Conclusion

To summarize, we know more about why subnational officials are increasingly trying to build their own policy regimes in Latin America today than we do about why they succeed in these efforts, and whether it is positive or negative when they do. In most, but not all, countries in Latin America ideological conflict over the market is now characterized by much greater territorial heterogeneity, dissonance, and friction than anything witnessed in the 20^th century. Thanks to a wave of globalizing, democratizing, and above all, decentralizing changes that have empowered subnational jurisdictions, national governments can no longer so easily ignore the policy preferences of the officials who are now elected to govern these jurisdictions.

Although the territorial dimensions of ideological conflict over the market are becoming more pronounced, not everywhere will this culminate in the stable juxtaposition of policy regimes. In part this is because of the superior resources and advantages enjoyed by most national governments, which will do what they can to stamp out ideologically discordant policy regimes at the subnational level when they feel threatened by subnational challengers -as they did in Bolivia, Ecuador, and Peru. But just as important are the limitations facing subnational governments when they seek to challenge the national government. In this paper I've focused on two especially important variables -institutional capacity and societal coalitions- that help explain the outcome of attempts to build and defend deviant subnational policy regimes, which succeeded in Bolivia and Ecuador but not in Peru.

For the purposes of theory building, my country cases appear to generate contradictory findings with respect to the role of the private sector as a force that can either promote or hinder capacity building. In Bolivia and Ecuador, private sector elites took the lead in designing and running local institutions and in contributing to the accrual of local institutional capacity over time. The argument is not that private sector elites intentionally contributed to state capacity, but rather that their initial control over these originally quasi-public institutions encouraged them to invest heavily in administrative capacity that would later redound to the benefit of subnational state administrations. In Peru in contrast, the private sector in the form of transnational mining companies undermined state capacity by taking on roles in service provision that directly and indirectly displace subnational governments. Together, the three cases point to the importance of distinguishing not only between domestic and foreign owners of capital, but further disaggregating the former category to appreciate the particular preferences of private sector elites who have regional identities, locations, and commitments. Foreign and domestic capital owners alike often face weak incentives to invest in state capacity and therefore generally resist state building efforts, but regionalized economic elites may have greater cause to invest in institutional capacity, particularly if it can be limited to subnational regions in which they have great influence.

But coalitions matter just as much as capacity; their strength helps explain the persistence of neoliberal policy regimes in Guayaquil and Santa Cruz, and their absence was fatal for the construction of subnational policy regimes in Peru. Not only do my three cases speak to the importance of support coalitions, but they also show that subnational officials can deploy both material and discursive strategies in their attempts to broaden these coalitions. Materially speaking, mayors and governors in Guayaquil and Santa Cruz were able to successfully challenge the national government because of their close alliances with leading regional elites whose interests were threatened by the national turn to the left. At the same time, when subnational officials sought to broaden their coalitions beyond elites, and thereby make it costlier for the center to respond with aggression, they turned to rhetorical and discursive strategies that portrayed their conflicts with the national government as fundamentally territorial conflicts. My research in Bolivia and Ecuador suggests that these strategies are particularly appealing to subnational officials on the right of the political spectrum as they seek to defend market-oriented models, whose content typically finds greater support among elites than non-elites. If non-elite residents in subnational districts decide that they actually have more to gain from the state-centered approaches articulated by the center, then mayors and governors seeking to defend neoliberal approaches are in real trouble. In settings like these, subnational neoliberals can try to emphasize the territorial rather than ideological nature of their conflict with the center as a potentially viable way to broaden coalitions and successfully stand up to the center.