The Effect of State–Local Complementarity and Local Governance on Development: A Comparative Analysis from Post-War Guatemala

World Development Vol. xx, pp. xxx–xxx, 2016 0305-750X/Ó 2016 Elsevier Ltd. All rights reserved. www.elsevier.com/locate/worlddev

http://dx.doi.org/10.1016/j.worlddev.2016.01.005

The Effect of State–Local Complementarity and Local Governance on Development: A Comparative Analysis from Post-War Guatemala MATTHEW T. KLICK* University of Denver, USA Summary. — What accounts for variation in human development levels across similar communities? Why, for example, have some equally poor, indigenous, highland communities in Guatemala made advances in health and education while others have stagnated or regressed? This paper argues that, contrary to an emerging recognition of the role of ‘‘ordinary citizens” in peacebuilding or crime reduction, human development requires the complementarity of non-state leaders and government resources. Results are demonstrated through a quantitative analysis of all of Guatemala’s 334 municipalities, combined with a qualitative analysis, including over 250 key informant interviews and focus group participants from four paired communities throughout the Western Highlands, which effectively traces the implementation of a widely acclaimed government anti-hunger program. This paper underscores the variability in local governance present across short distances in Guatemala, with dramatic influence on local development outcomes. In most communities, the role of traditional leaders and other non-state actors is increasingly crowded out by political conflicts exacerbated by growing, and shifting political party and religious affiliations. In these instances, development governance has broken down and development resources are distributed along locally derived lines of patronage versus need. Elsewhere, however, where state actors have identified and incorporated the legitimacy, informal authority, and networks of traditional leaders at the village level, the implementation of the anti-hunger program is visibly more effective. The role of local, informal leaders is pivotal in legitimizing the program, mobilizing citizen participation, and overseeing more equitable distribution of key resources. The paper concludes with an exploration of what explains variation in the quality of local governance across space and its implications for development practice. Ó 2016 Elsevier Ltd. All rights reserved. Key words — human development, local governance, complementarity, state–society relations, Latin America, Guatemala

1. INTRODUCTION: SPATIAL VARIATION OF HUMAN DEVELOPMENT IN THE GUATEMALAN HIGHLANDS

it. The irrelevance of state presence at the local level for development, for example, is striking, but identifying causal relationships hinges instead on qualitative work in Guatemala. Findings are based on fieldwork which spanned over six months and include the results of key informant interviews and focus groups held in hard-to-access hamlets throughout the targeted municipalities (N = 254; 251 individuals, 146 focus group participants, 108 discrete interviews). A Qualitative Comparative Analysis (QCA) of four Highlands communities—chosen specifically in order to control for factors like climate, access to roads, or regional economic hubs—helps isolate significant explanatory variables, while contextualizing others, like the role of social discord in each community for instance. A process-tracing stage drills more deeply still, and examines how the implementation of an internationally acclaimed government anti-hunger program (El Pacto Hambre Cero or the ‘‘Zero Hunger Pact) differs across three communities. This component, building on the others, elucidates which local factors distinguish those more ‘‘successful” communities from those coping with especially poor development indicators and experiences. A plausible explanation for discrepancies across similar communities might be that differing levels of state presence yield different degrees of service provision and thus distinct outcomes. In contrast, between government corruption and anemic capacity everywhere, differences in local development

Guatemala’s 36 year long civil war, unfulfilled land reforms, unimplemented peace accords, and now drastically high rates of criminal violence (coupled with continued high rates of impunity) each contribute and intertwine to hinder more equitable development gains nationally. These factors—along with what some call a dysfunctional state apparatus and a structural racism that severely disadvantages the large indigenous population—remain critical to explaining Guatemala’s stalled prosperity. This paper, however, probes a different and mostly overlooked phenomenon altogether—why communities of otherwise similar historical, geographical, and demographic backgrounds within Guatemala have diverging experiences with, and outcomes from, 20 years of post-war development. Some communities, for example, are doing better in reducing infant mortality (Figure 1), hunger or illiteracy rates, while otherwise very similar communities continue to struggle along the most basic indicators of wellbeing. What accounts for these peculiar differences across communities of similar geography, history, and ethno-linguistic background? What, in other words, explains variation in human development outcomes within rural western Guatemala? In order to answer this puzzle the paper utilizes three overlapping methods, both quantitative and qualitative. The project begins with a regression analysis of all 334 Guatemalan municipalities, exploring the relationship between levels of ‘‘state density,” local spending, crime, and other local factors compared with myriad development metrics, including their changes over time. The quantitative component, however, does more to enrich and underscore the puzzle than answer

* This research was supported in part by the Josef Korbel School’s Doctoral program in International Studies and a travel grant from the Foundation for Global Scholars. I thank the three anonymous reviewers for their commentary and suggestions. Final revision accepted: January 6, 2016. 1

Please cite this article in press as: Klick, M. T. The Effect of State–Local Complementarity and Local Governance on Development: A Comparative Analysis from Post-War Guatemala, World Development (2016), http://dx.doi.org/10.1016/j.worlddev.2016.01.005

2

WORLD DEVELOPMENT

Figure 1. Variation in infant mortality rates (2011) across select Guatemalan highland communities. The communities span a belt across southern Quiche´ Department and into Totonicapa´n and are used to illustrate the peculiar discrepancies in development metrics across similarly poor, rural communities.

profiles might instead be the result of either different levels of social cohesion at the local level, or the influence of traditional leadership and intact traditional norms that mobilize citizens or provide additional accountability. Neither, however, explains observed outcomes by themselves. This paper argues that spatial variation in human development and hunger in the Western highlands of Guatemala is explained most by the degree to which formal delivery and government implementation utilizes, and indeed complements, local forms of legitimate, and influential local leadership, or what is heretofore called state–local complementarity. This is conceived, below, as differences in local de facto governance. (a) Significance of spatial variation in local development outcomes The phenomenon of spatial variation of human development is significant for several reasons. First, Guatemala has received notable external assistance since the 1996 Peace Accords were struck—through the United Nations Development Program (UNDP), World Bank organizations, foreign NGOs and through bi-lateral aid. 1 Despite outside assistance and capacity enhancement programs, however, and despite an alphabet soup of ministries and a revolving door of antipoverty and development initiatives, variation in outcomes at the community level suggests that poorly-understood local conditions thwart well-intentioned and even comprehensive development programs—thus squandering resources and limiting both scope and impact. This paper explicitly analyzes the implementation of development programing over similar communities, underscoring its complexity, and striving to tease out which local factors explain development disparities. Second, spatial variation in development outcomes confounds basic assumptions of much development programing—that rural areas, or all indigenous groups, for example, have identical needs and/or identical obstacles to development. Literature increasingly underscores how locally sited

factors, whether the role of social cohesion at the village level (Kaplan, 2013) or deeply-entrenched resistance to outside influence (Pugh, 2005), significantly influence post-war outcomes across space. Acknowledging the significance of local factors is especially relevant in a country like Guatemala, where the breadth and scope of state presence is notoriously limited, and where more spatially variegated political processes have been well documented. These include the relevance of indigenous forms governance, legal-pluralism, and competing claims to sovereignty—over life, land, or local authority (Godoy, 2006; Larson, 2008; Sieder, 2011; Stepputat, 2015; Yashar, 1998). But whereas local communities might have leeway in affecting episodes of violence or on crime, in particular, this paper also explores the limitations that local actors face when the dependent variable is something as complex as human development. None of the communities under investigation in this paper experience excellent outcomes in development, nevertheless the variation across similarly poor communities is stark (Table 1). As noted above, this paper argues that a critical factor explaining variation is the degree to which state actors, with important material resources, work with, and even defer to, both official and informal authorities at the community level who mobilize citizens and distribute resources equitably and more transparently. Crucially, this paper finds that development gains are best when informal authorities are expressly incorporated into the implementation of a key development program. This is a rare occurrence, however, and the toxic combination of state mismanagement, deep distrust of state authorities as well as local political divisions and rivalries routinely thwart what might have been state–local complementarity. Nevertheless, lessons from Guatemala empirically demonstrate how subtle, local political dynamics and shifting state–society relations have a direct bearing on livelihoods in developing countries, with implications for evolving development practice.

Please cite this article in press as: Klick, M. T. The Effect of State–Local Complementarity and Local Governance on Development: A Comparative Analysis from Post-War Guatemala, World Development (2016), http://dx.doi.org/10.1016/j.worlddev.2016.01.005

THE EFFECT OF STATE–LOCAL COMPLEMENTARITY AND LOCAL GOVERNANCE ON DEVELOPMENT

3

Table 1. Summary of human development indicator (HDI) scores across Guatemalan municipalities Variable

Obs

Mean

Std. Dev.

Min

Max

HDI_2005 HDI_2005, where elevation is > 6,500 feet above sea level

331 85

.591 .571

.089 .103

.306 .306

.828 .784

2. CONCEPTUALIZING SPATIAL VARIATION IN DEVELOPMENT: STATES, SOCIETIES, AND GOVERNANCE (a) Spatial variation In 1889, Charles Booth (1902) created a map depicting the relative prosperity of Victorian London by city block, color coding each block on a scale ranging from ‘‘vicious poverty” to ‘‘wealthy,” or the ‘‘servant keeping class.” 2 But contemporary work on sub-national variation in development and its drivers is generally limited. Gennaioli, La Porta, Lopez-deSilanes, and Shleifer (2011) analyze a database of 1,569 subnational regions spanning 110 countries. In a cross-regional analysis of geographic, institutional, cultural, and human capital determinants on regional development, the authors find that human capital, in particular education, is mostly associated with ‘‘regional development”—which is, quite minimally, reported as GDP per capita, and thus not particularly insightful from a human development perspective. Within equally poor sub-regions of Guatemala, for example, certain communities have made improvements in health and education that would not be predicted by income (Klick, 2013). Given the relative paucity of the literature dedicated to the questions of spatial variation in development at the village level, however, this paper situates itself more within debates of state and society. (b) Fragile states and development This debate centers on whether the state, as conventionally conceived, remains an essential instrument through which basic resources can be equitably delivered, or a hindrance. The idea of the largely autonomous, Weberian state continues to dominate post-war policymaking, leading to the now ubiquitous ‘‘fragile” and ‘‘failed state” nomenclature and statebuilding doctrine. Post-war states, it is argued, have either been dismantled or otherwise ravaged by civil conflict to a degree that their internal capacity to conduct essential operations like collecting taxes, or delivering even basic services is undermined (Rotberg, 2003). Diminished capacity erodes the legitimacy of state actors, instigating a negative cycle that, unless reversed, reinforces fragility. Thus either high rates of criminal violence, as we see in Guatemala, or non-state actor use of violence, is inherently a reflection of diminished state capacity, which is best reversed by externally-led, or assisted, statebuilding (Lake, 2010). Though conceptions of statebuilding are increasingly nuanced, it still assumes that legitimacy, peace, and development flows automatically from state capacity, which is established rather technically. Statebuilding also privileges a normative understanding of the unitary, rational state (coupled with an assumption that it will dominate all other political orders), while reinforcing a dichotomous understanding of the ‘‘true” state, versus the ‘‘non-state” (Albrecht & Kyed, 2011, p. 14).

(c) Post-weberian and society-centric revisions Scholars are revising the conception of the modern state as a ‘‘democratic and capitalist state governed by the rule of law” on both normative and analytical grounds (Migdal & Schlichte, 2005; Risse, 2011). Thomas Risse argues that a more relevant research question for development scholars working on fragile states would be: ‘‘who governs for whom, and how are governance services provided under conditions of weak statehood? (Risse, 2011, p. 4). Reconceptualizing the problematique as one of governance under conditions of limited statehood, versus state fragility or state failure, per se, is more analytically fruitful at the local level, in particular, where ‘‘various combinations of state and nonstate actors ‘govern’” (Risse, 2011, p. 11). Governance, or the ‘‘interactive processes of multistakeholders (including government) in order to resolve common problems” (Saito, 2008, p. 6), acknowledges, but otherwise de-emphasizes the exclusive role of the state in coordinating or producing binding rules or collective goods (Risse, 2011, p. 9). Recent scholarship, for example, demonstrates that social services are routinely and robustly delivered by nonstate actors (Cammett & MacLean, 2014). A ‘‘governance approach” thus discards the dichotomous depiction of strictly state and society, or public and private, incorporating into the analysis, instead, the influence of ‘‘different constituent members including public, private, and civil organizations in order to resolve common political, economic and social issues” (Risse, 2011; Saito, 2008, p. 6). Field observations in areas of weak state presence demonstrate how state authority is routinely interpreted, mediated, or molded by local actors. In some cases, this may yield improvements in peace and security, even while further undermining the authority of the state in remote regions (Chopra, 2009; Menkhaus, 2008). In east Africa, variations in ‘‘statemaking” by central governments are seen in part as a function of rural interests and political struggles within the countryside, rather than simply a function of state capacity (Boone, 2003). Another strand of literature, formed mostly as a critique of the ‘‘liberal peace” and modern peacebuilding and postconflict statebuilding operations generally, explores ‘‘hybrid” governance. In the best cases, hybridity is a natural order of governance that blends locally resonant forms of traditional leadership (the role of chiefs for example), with formal-legal systems of representation and legislation that provides sustainability and durability (Boege, Brown, Clements, & Nolan, 2008; Brown, 2007). In more fraught cases, hybridity is the de facto result of local communities actively resisting liberalism and externally-imposed authority (Pugh, 2005; Richmond, 2006). The notion that communities ardently resist state expansion is reinforced by the historical work of Scott (1998, 2009). The desires of states to control and order everything from natural resources to recalcitrant populations, Scott argues, diametrically clashes with local values in different contexts, undermining state influence at the local level. Moreover, local populations have demonstrated far greater agency than

Please cite this article in press as: Klick, M. T. The Effect of State–Local Complementarity and Local Governance on Development: A Comparative Analysis from Post-War Guatemala, World Development (2016), http://dx.doi.org/10.1016/j.worlddev.2016.01.005

4

WORLD DEVELOPMENT

commonly assumed—in resisting state encroachment and in forging autonomy. Development literature is also exploring how state and society interact in order to determine development outcomes that vary within countries. Much of this work is focused on social capital as a dependent variable, and whether or not it can be engineered from outside to improve local governance (Adveenko & Gilligan, 2015; Altschuler & Coralles, 2013; Khan, Rifaquat, & Kazmi, 2007). Crucially, additional literature has isolated yet more nuanced criteria around which such programs, or service delivery more generally, succeeds or fails to improve livelihoods at the local level. In China, the presence or absence of ‘‘solidary groups” that reward officials with ‘‘moral standing” for their effective public service explains discrepancies in infrastructure development, despite the absence of more formal democratic accountability mechanisms (Tsai, 2007). In South India (Karnataka province), informal customary village councils (CVCs) are commonly augmenting the role of more formal village Gram Panchayats, in roles such as dispute resolution and even social welfare provision (Ananthpur & Moore, 2010). In other regions (Rajastan and Madhya Pradesh), social capital helps explain differences in economic development, community peace, and democratic participation, but positive results hinge most on the presence of a ‘‘mediating agency,” or whether a capable and effective interlocutor exists from village ranks to coordinate between state and village actors (Krishna, 2002). In each of the above cases, local conditions are highly relevant, but are otherwise limited in their influence over local development outcomes without the presence of an ‘‘associational environment,” or where ‘‘top down efforts are coordinated with bottom-up mobilization” (Ananthpur, Malik, & Rao, 2014, p. 3; Fox, 1992). This line of research provides a plausible hypothesis for what explains variation in development in Guatemala, but critical differences in context and background distinguish this case study. The communities under study were never subjected to projects or experiments in building social capital, and reflect instead endogenous variations in local governance and development outcomes that vary dramatically over short distances. Moreover, rather than examining outcomes through the lens of social capital, social spending, infrastructure or even income, this paper seeks to understand why similar communities have yielded such dramatic differences in observable development metrics themselves.

explore the relationship between a diverse set of independent variables and dependent variables that represent a range of human development indicators, including human development index (HDI) scores as well as hunger, infant mortality, and illiteracy rates as robustness checks. The explanatory variables compare in essence rival hypotheses to the idea of state–local complementarity. They also reflect state-centric, state–society, and society-centric theories of development, and thus a combination of theoretical propositions discussed above, as well as factors specific to the Guatemala context—like the lingering influence of the armed conflict 3 as well as an effort to capture how variation in local organization across space relates to development. In sum, the model under investigation is reflected by the following equation: Y i ðhuman developmentÞ ¼ ai þ b1 x1 ðstate presenceÞ þ b2 x2 ðlocal spendingÞ þ b3 x3 ðconflict intensityÞ þ b4 x4 ðlocal social organizationÞ þ ei where ai = the intercept and ei = the residual error. The results reveal mostly a thorough disconnect between macro drivers of development—including state presence, local social spending, and political parties—and well-being outcomes. At the same time, there is some indication that local factors, in contrast, have some bearing on development outcomes, though the direction of causality remains unclear and findings ultimately revolve mostly around subsequent ethnographic work in the field. (i) Variation in human development index scores A logical starting point for a quantitative exploration of Guatemalan communities is to regress independent variables on 2010 HDI levels (HDI_extrap), which is a figure extrapolated from HDI scores from two previous years (2000 and 2005), but which have not been reported at the municipal level since. 4 Widespread variation in HDI scores among communities with limited state presence is apparent in Figure 2. A preliminary global model explores the relationship between human development (HDI_extrap) and state density (SDI_2009), 5 local government spending (the level of local spending on social services per inhabitant, Gasto_Mun), conflict intensity during the armed conflict (Conf_Intens), social organization (as represented by official denunciations of public

3. CAUSAL PATHWAYS TO DEVELOPMENT VARIABILITY IN HIGHLANDS GUATEMALA In order to investigate whether there is a causal relationship between the nature of village-level governance arrangements and human development outcomes, the project unfolds over three overlapping, complementary stages—a quantitative overview gives way to a qualitative comparative analysis, then process-tracing component. (a) Quantitative analysis The quantitative component serves as a ‘‘first glimpse” into the relationships between possible drivers of human development across Guatemala, and variations thereof. Utilizing a novel database that combines data from the municipal level across all 332 communities—which originates from highly dispersed and sometimes obscure sources—this component utilizes ordinary least square (OLS) regressions in order to

Figure 2. 2010 HDI plotted against the log-transformed state density variable.

Please cite this article in press as: Klick, M. T. The Effect of State–Local Complementarity and Local Governance on Development: A Comparative Analysis from Post-War Guatemala, World Development (2016), http://dx.doi.org/10.1016/j.worlddev.2016.01.005

THE EFFECT OF STATE–LOCAL COMPLEMENTARITY AND LOCAL GOVERNANCE ON DEVELOPMENT

officials and police by citizens in each community reported by the Guatemalan Human Rights Ombudsman, relative to population size, labeled Denperpop), and finally electoral alignment (Elect_Align, or whether a municipal mayor is from the same party as the President as of the 2011 elections, where 1 = alignment and 0 = no alignment). 6 This preliminary model is, as a whole, statistically significant (p = .000) but substantively insignificant, or at least limited in substantively explaining variation in HDI across communities (Table 2). Though several variables in this model (log_SDI, Conf_Intens and Denperpop) are statistically significant, coefficients in general are very small (Model 1). Subsequent models are each significant and progressively more robust, but only tinker with the strength of each explanatory indicator and coefficients remain generally quite small. For the moment, state density is positively correlated with HDI, and significant, though the effect is minimal. Local social spending is never a significant variable. Conflict intensity during the civil war (Conf_Intens) is negatively correlated with HDI (and statistically significant with 99% confidence in each of the models). Finally, per capita rates of officially recorded denunciations of public officials and police officers (Denperpop) has a positive effect on HDI which is both statistically significant (p < .05 in Models 1, 2, and 3, p < .01 in models 4 and 5) and by far the strongest of any variable. Interpreting this poses some problems. On the surface, high levels of denunciations would indicate a community racked with corruption, abuse of power, and a frustrated citizenry. This paper assumes that denunciations indicate on some level an element of locally-based social organization, however. In a culture of fear and distrust, as is widely reported to permeate Guatemala (Azpuru, 2011), denunciations might instead represent the determination of local actors to organize and protest, or a local level of impunity for reporting, rather than actual levels of official abuse (which is arguably high everywhere). Additionally, a simple bivariate regression reveals a negative correlation between denunciations and declining homicide rates (n = 331, coefficient = .034, p = .8253). 7

5

Though not statistically significant, denunciations nevertheless do not reflect levels of violence, and are thus a response to something other than crime. Moreover, the effect of denunciations and conflict intensity suggest the influence of locally-relevant factors on development outcomes. This holds true across a range of robustness checks discussed below, while the influence of state presence ultimately disappears altogether. (ii) Accounting for population size Given that larger communities are frequently regional economic hubs, and places where citizens can access certain services more easily than in rural locales, the relative size of a community may disguise effects that influence development outcomes. The same models as above are therefore run again, but this time only after excluding larger municipalities (or those municipalities with a population greater than 25,000 inhabitants) (Table 3). Both conflict intensity and denunciations remain statistically significant, with conflict intensity particularly robust (p < .01 in each of the models) if small, while the effect of denunciations remains the most dramatic—with coefficients raging from 1.77 to 2.59 (Table 7). Peculiarly, the state density sign changes from positive to negative in this test while becoming dramatically stronger as well, though it no longer remains statistically significant. In the very least, the density of state offices and resources has no bearing on the human development outcomes of smaller municipalities in Guatemala, while some element of local conditions, including social organization and protest as suggested by the amount of official denunciations, does affect results. (iii) Robustness checks: illiteracy and hunger HDI scores are an inherently crude metric of development and, in the case of rural Guatemalan communities, subject to measurement and reporting errors. If individual indicators of education and health reveal similar patterns, however, a more robust image of the relationship between state density, conflict, and social organization can be formed. Indeed,

Table 2. Ordinary least square regression: variation in HDI_extrap Predictor variable

Log_SDI Gasto_Mun Conf_Intens Denperpop Elect_Align

Model 1 (N = 278) Coefficient (standard error)

Model 2 (N = 277) Coefficient (standard error)

.0497943* (.0229995) .0000268 (.0000237) .0000918** (.000217) 1.867374* (.9201119) .0054443 (.0130069)

.0562005* (.0233884) .0000287 (.0000212) .0000953** (.0000225) 1.805071* (.9214902) .0060596 (.0130156) 3.48  106 (2.35  106)

Elev

Model 3 (N = 277) Coefficient (standard error)

.0000314 (.0000239) .0000919** (.0000226) 1.956041* (.9274082) .0095903 (.0130458) 2.44  106 (2.34  106)

Hom_Diff Constant F(df) R2 Adjusted R2

.5011801 7.28 (5) .1180 .1018

.4670603 6.44 (6) .1252 .1507

.6401602 6.46 (5) .1065 .1570

Model 4 (N = 325) Coefficient (standard error)

Model 5 (N = 325) Coefficient (standard error)

.046398* (.0218491)

.0587264** (.0212306)

.0000884** (.0000224) 2.644132** (.7782323)

.0000903** (.0000192) 2.475925** (.0086536)

.0009367* (.0004267) .510152 13.10 (4) .1407 .1300

.4770062 15.67 (3) .1278 .1196

Note: **p < .01; *p < .05; Log_SDI represents the log-transformed state density index score, Gasto_Mun is local social spending, Conf_Intens is civil-war related victim count per community, Denperpop equals the number of official denunciations per capita, Elect-Align is a binomial variable indicating whether the Mayor and President are of the same party, Elev equals elevation and Hom_Diff is the difference in homicide rates at the municipal level between 2000 and 2011.

Please cite this article in press as: Klick, M. T. The Effect of State–Local Complementarity and Local Governance on Development: A Comparative Analysis from Post-War Guatemala, World Development (2016), http://dx.doi.org/10.1016/j.worlddev.2016.01.005

6

WORLD DEVELOPMENT Table 3. HDI_extrap when population <25,000: ordinary least square regression Predictor variable

Log_SDI Gasto_Mun Conf_Intens Denperpop Elect_Align

Model 1 (N = 138) Coefficient (standard error)

Model 2 (N = 137) Coefficient (standard error)

.0829566 (.051696) .0000533* (.0000279) .000218*** (.0000685) 1.894503* (.000142) .0090263 (.0188681)

.0855112 (.0535125) .0000525* (.0000282) .0002169*** (.0000693) 1.908834* (1.012271) .009434 (.0190546) 8.21  107 (3.63  106)

Elev

Model 3 (N = 137) Coefficient (standard error)

.0000401 (.0000272) .0002148*** (.0000697) 1.767816* (1.014379) .0067685 (.0190937) 5.09  107 (3.55  106)

Hom_Diff Constant F(df) R2 Adjusted R2

.88775961 4.65 (5) .1497 .1175

.8987034 3.83 (6) .1503 .1111

.6421548 4.04 (5) .1336 .1005

Model 4 (N = 138) Coefficient (standard error)

Model 5 (N = 163) Coefficient (standard error)

.0195352 (.0425373)

.0094573 (.042135)

.0002142*** (.0000664) 2.57166*** (.900458)

.0002174*** (.0000666) 2.594271*** (.9036133)

.0029894 (.0020327) .7081681 5.99 (4) .1316 .1096

.680653 7.21 (3) .1197 .1031

Note: ***p < .01; **p < .05; *p < .10; Log_SDI represents the log-transformed state density index score, Gasto_Mun is local social spending, Conf_Intens is civil-war related victim count per community, Denperpop equals the number of official denunciations per capita, Elect-Align is a binomial variable indicating whether the Mayor and President are of the same party, Elev equals elevation and Hom_Diff is the difference in homicide rates at the municipal level between 2000 and 2011.

regressions of illiteracy and changes in hunger in communities smaller than 25,000 inhabitants confirm initial findings. Specifically, only conflict-affectedness and denunciations have significant relationships with illiteracy at the municipal level (p < .01) (Table 4). Testing the same predictor variables against changes over time in chronic hunger (Table 5) further reinforces these findings. Chronic hunger rates are based on government statistics that are the result of two different reports—the 2002 and 2009 National Height Census of Primary School Students (Censo Nacional de Talla en Escolares de Primer Grado) (Ministry of Education, 2001, 2009).

These regressions cannot capture the effects of the existing Hambre Cero anti-hunger program (discussed in more detail below), but myriad other government interventions were introduced in the period between the two hunger censuses—including Mi Famila Progressa (My Family Progresses, now Mi Bono Seguro), which was designed with the express purpose of addressing childhood hunger, maternal health, and school attendance—and the Bolsas Solidarias program (now Mi Bolsa Segura) which was a direct transfer of cash and basic foodstuffs directed at the poorest families. In sum, a quantitative overview suggests that it is community-level factors that better explain discrepancies in

Table 4. 2011 illiteracy rates where population <25,000: ordinary least squares regression Predictor variable

Log_SDI Gasto_Mun Conf_Intens Denperpop Elect_Align

Model 1 (N = 128) Coefficient (standard error)

Model 2 (N = 128) Coefficient (standard error)

3.433755 (5.612491) .0010112 (.0029575) .0310563*** (.0071164) 354.5368*** (.000142) .9415009 (2.007523)

6.370267 (.5.717079) .0004693 (.0029302) .0297665*** (.00705) 356.4175*** (109.9833) .761388 (7.982922)

Elev

.0007952 (.0003841)

Model 3 (N = 128) Coefficient (standard error)

.0002829 (.0028542) .0295369*** (.007054) 354.3001* (110.0757) .8456734 (1.983439) .0006891 (.0003724)

Hom_Diff Constant F(df) R2 Adjusted R2

20.65968 6.96 (5) .2220 .1901

8.326414 6.67 (6) .2486 .2114

27.62416 7.74 (5) .2409 .2098

Model 4 (N = 151) Coefficient (standard error)

Model 5 (N = 151) Coefficient (standard error)

2.762708 (4.771261)

.4567894 (.4.792278)

.028844*** (.0070779) 337.0204*** (104.0205)

.0293737*** (.00072238) 347.3542*** (106.1332)

.5853974*** (.2175466) 23.71136 10.00 (4) .2150 .1935

30.13987 10.47 (3) .1761 .1593

Note: ***p < .01; **p < .05; *p < .10; Log_SDI represents the log-transformed state density index score, Gasto_Mun is local social spending, Conf_Intens is civil-war related victim count per community, Denperpop equals the number of official denunciations per capita, Elect-Align is a binomial variable indicating whether the Mayor and President are of the same party, Elev equals elevation and Hom_Diff is the difference in homicide rates at the municipal level between 2000 and 2011.

Please cite this article in press as: Klick, M. T. The Effect of State–Local Complementarity and Local Governance on Development: A Comparative Analysis from Post-War Guatemala, World Development (2016), http://dx.doi.org/10.1016/j.worlddev.2016.01.005

THE EFFECT OF STATE–LOCAL COMPLEMENTARITY AND LOCAL GOVERNANCE ON DEVELOPMENT

7

Table 5. Difference in chronic hunger (2002–09) Predictor variable

Log_SDI Gasto_Mun Conf_Intens Denperpop Elect_Align

Model 1 (N = 279) Coefficient (standard error)

Model 2 (N = 278) Coefficient (standard error)

2.362724* (1.286329) .0019976 (.0013256) .0024055* (.0012494) 169.8314*** (.51.45253) .4716094 (.7269776)

1.956537 (1.305827) .0021079 (.0013267) .0026269** (.0012544) 166.0358*** (51.46353) .4320683 (.7263941) .0002198* (.0001311)

Elev

Model 3 (N = 278) Coefficient (standard error)

.0020133 (.0013282) .0027448** (.0012548) 160.728*** (.51.45882) .5565181 (.723793) .0002554 (.0001292)

Hom_Diff Constant F(df) R2 Adjusted R2

.13.53678 5.86 (5) .0969 .0804

11.37744 5.36 (6) .1061 .0863

5.355074 5.96 (5) .0987 .0822

Model 4 (N = 326) Coefficient (standard error)

Model 5 (N = 326) Coefficient (standard error)

1.974551 (1.234109)

2.057491* (1.190307)

.0027783** (.0010797) 180.5699*** (43.94241)

.0027659** (.0010771) 181.6888*** (43.66712)

.0062565 (.0241011) 12.54552 7.14 (4) .0817 .0702

12.76723 9.52 (3) .0815 .0729

Note: ***p < .01; **p < .05, *p < .10; Log_SDI represents the log-transformed state density index score, Gasto_Mun is local social spending, Conf_Intens is civil-war related victim count per community, Denperpop equals the number of official denunciations per capita, Elect-Align is a binomial variable indicating whether the Mayor and President are of the same party, Elev equals elevation and Hom_Diff is the difference in homicide rates at the municipal level between 2000 and 2011.

development, rather than formal, state influence. Difficulties in interpreting causality and accounting for omitted variables, however, force a more ethnographic approach to understanding variations in development outcomes across similar communities. This was approached through a combination of a qualitative comparative analysis across four rural communities and process-tracing of the implementation of a lauded government run anti-hunger program. (b) A qualitative comparative analysis (QCA) of four highland communities QCA methodology permits systematic comparison across an intermediate amount of cases in order to test the necessity and sufficiency of carefully coded causal conditions. Its main purpose is to allow for such comparisons, while simultaneously acknowledging within-case complexity (Rihoux & Ragin, 2009; Speer, 2011). Specifically, QCA methods account for multi-causality and equifinality, or the possibility of unobservable interactions between explanatory variables, and limitations in a researcher’s ability to adequately make perfectly controlled comparisons (George & Bennett, 2005). Field site communities for this study were initially selected from afar based on dramatic differences between them in their changes in HDI over time (2000–05), even after first controlling for elevation, presence of the state, and geography (Table 6). Elevation can drastically affect climactic and thus growing conditions over a short distance in Guatemala, which could be reflected in health or living condition indicators. Communities were selected that varied in their respective heights above sea level within one standard deviation of the country average. 8 Moreover, and without diminishing the daily struggles of lowland Guatemalan communities, highland communities struggle most with poverty and chronic hunger in particular. The departments where the four communities are located— Quiche´ and Totonicapa´n—are the worst performing in the country, with rates of chronic hunger that affect in excess of

60% of schoolchildren. 9 One community under focus, Santa Marı´a Chiquimula in Totonicapa´n, reports a staggering rate of 75.5% (Ministry of Education, 2009). Based on over 6 months of field work in Guatemala, and over 100 semi-structured interviews and a dozen focus groups (another 148 participants), the presence of potentially rival hypotheses—from conflict-affectedness, local spending, and social cohesion (or rather social discord, and separately the effectiveness of civil society)—were coded as either ‘‘high” or ‘‘low” in each of the four communities (Appendix B). Moreover, the QCA includes a ‘‘complementarity” variable that was otherwise unavailable quantitatively, coded based on the observation of state and local coordination over developmentrelated decision-making and distribution of services. These variables were again compared with multiple indicators of human development, including rates of chronic hunger and illiteracy rates in particular. The results of QCA are tabulated in the form of a ‘‘truth table” (Table 7) in order to more systematically compare across cases, to decipher patterns and relationships between clusters of variables, and to observe the relationships between new variables, like complementarity, and development outcomes. (i) QCA results: community variables take shape The most obvious pattern to emerge is a correspondence between high levels of internal social discord I, or where internal political or confessional divisions at the community level, and weak development outcomes, chronic hunger in particular (top left, Figure 3). Communities having made medium progress on hunger (no communities made high progress) experience low levels of social discord uniformly. Social spending (A) is less obviously important, though the only community with high levels of social spending did indeed make greater headway in combating hunger. Similar results stem from an examination of civil society, in that the only instance of more highly mobilized civil society (B) corresponds to better results. The role of conflict intensity (D) is surprisingly irrelevant among these four communities, as it is distributed evenly across better and poorly performing

Please cite this article in press as: Klick, M. T. The Effect of State–Local Complementarity and Local Governance on Development: A Comparative Analysis from Post-War Guatemala, World Development (2016), http://dx.doi.org/10.1016/j.worlddev.2016.01.005

8

WORLD DEVELOPMENT Table 6. Initial selection criteria, summary Location (elev in ft)

HDI (2005)

Santa Marı´a Chiquimula (6,975) Santa Lucı´a la Reforma (6,013)

Totonicapa´n Department .380 .427

Patzite´ (7,545) Zacualpa (4,875)

.489 .494

SDI (2009)

D Health HDI (2000–05)

.16 .16

.024 +.118

.16 .21

.453 +.083

Quiche´ Department

Table 7. QCA table Presence of:

Social spending? (A = high, a = low) ‘‘Strong” local civil society? (B = high, b = low) Social discord? (C = high, c = low) Conflict affected? (D = high, d = low) Complementarity (E = high, e = low) D Chronic hunger (2002–09) D Illiteracy (2009–11)

Community Sta Marı´a Chiquimula (6975’)

Sta Lucia la Reforma (6013’)

Patzite´ (7545’)

Zacualpa (4875’)

a b C d e 1.7% (Low) 9.5% (Med)

a B c D E 3% (Med) 20% (High)

A b c d e 6.8% (Med) 7% (Med)

a b C D e 1.4% (Low) .2% (Low)

communities. Finally, the observance of state–local complementarity (E) and its effect on outcomes is not dramatic, but like civil society and social spending, it too appears only among those communities performing better. Analyzing the results with respect to changes in illiteracy (top right, Figure 3) yields similar results. Since more communities did better in this respect over time, however, results shift slightly. The effects of social spending (A), civil society (B), social discord (C), and complementarity (E) all diminish slightly compared with their apparent effects on hunger, while the influence of conflict intensity gains in strength. After combining hunger and illiteracy (bottom of Figure 3), the most consistent results are apparent among the worst performing communities—where limited social spending (a), an anemic civil society (b) and high social discord (C) are each 100% consistent. Moreover, the absence of complementarity is consistent across all poorly performing communities. Among the better performing communities, results are less consistent, but nevertheless suggestive. Higher social spending (A) and a ‘‘strong” civil society (B) reappear. Only one community, of those doing better across both hunger and illiteracy, exhibits observable social discord (C). Conflict (D) is again distributed evenly across both poor and better

performing communities, making it a surprisingly non-relevant variable in determining current development outcomes in these select communities. Finally, like social spending and social discord, complementarity (E) reappears and only among those communities doing well. Also, the only community with a ‘‘High” performance in any of the development metrics (remote Santa Lucı´a la Reforma reports a 20% reduction in illiteracy) exhibits clear state–local complementarity, despite higher rates of poverty and extreme poverty than any of the other communities. Crucially, the QCA suggests that social discord at the community level inhibits development. The degree of division at the village level, however—including emerging political rivalries within small, rural, and ethnically homogenous communities—has been largely overlooked by development literature and has limited bearing on contemporary development policy. With respect to complementarity, the QCA does not reveal its clear deterministic function in improving development outcomes, but it has teased out its influence—along with local social spending and the strength of locally-based civil society—as a potentially critical facilitating condition that results in development-oriented governance and improved outcomes at the community level.

Chronic Hunger ab C de Low ab C De aB c DE Med/High Ab c de

Δ Illiteracy abCDe Low abCde Med/High Abcde aBcDE

Chronic Hunger and Illiteracy Combined abCde Low abCDe abCDe aBcDE Abcde Med/High abCde Abcde aBcDE Figure 3. QCA summary.

Please cite this article in press as: Klick, M. T. The Effect of State–Local Complementarity and Local Governance on Development: A Comparative Analysis from Post-War Guatemala, World Development (2016), http://dx.doi.org/10.1016/j.worlddev.2016.01.005

THE EFFECT OF STATE–LOCAL COMPLEMENTARITY AND LOCAL GOVERNANCE ON DEVELOPMENT

The case of Santa Lucı´a la Reforma is especially compelling. It is remote, very poor and was more impacted by the armed conflict than the other communities, with the exception of Zacualpa, yet it has outperformed its immediate neighbor, Santa Marı´a Chiquimula, across an entire range of development indicators including chronic hunger. It is also the community with the greatest reductions in illiteracy among the four communities of the interior highlands—another unlikely outcome. Yet how this happens is still unclear. (c) Process-tracing from governance to development In order to more precisely account for these differences, while also accounting for endogeneity and omitted variable bias, this paper ultimately hinges on an final methodological component: a process-tracing stage with the express purpose of observing the sequences, processes, and agent-to-agent linkages that constitute local development governance. Specifically, I examine the implementation at the village level of a key, award-winning government anti-hunger program, El Pacto Hambre Cero. 10 For this component, focus groups were conducted in remote aldeas (or hamlets within the municipality surrounding the municipality’s population center) in addition to conducting key informant interviews with local religious leaders, politicians, government agency representatives, indigenous leaders, and civil society. Additionally, I participated in meetings and events—formal and informal—that were designed to affect the scope and nature of Hambe Cero’s implementation. This increased the number of observations considerably, providing rich material for description as well as an opportunity to test the observable implications of state–local complementarity, and the competing implications of social discord, harmony, legacy of conflict, and civil society. In this section, I especially dwell on the contrasting experiences of Santa Marı´a Chiquimula, and its immediate neighbor, Santa Lucı´a la Reforma, though draw from fieldwork in other communities as well to illuminate points where helpful. The wildly different political environments across the two communities underscore the variability in governance possible over short distances in rural Guatemala, as well as the possibility of complementarity to improve service delivery and, more importantly, development outcomes themselves. (i) Indigenous, rural and poor, but not alike Santa Lucı´a la Reforma (la Reforma here forth) is located at 6,013 feet above sea level (1,833 m) in Totonicapa´n department—a remote and infrequently visited region with a population 98% indigenous (K’iche´ Mayan). It was in fact once part of Santa Marı´a Chiquimula municipality (Chiquimula here forth), and thus shares much in terms of historical, cultural, and even familial ties with its now neighbor. Despite higher rates of extreme poverty (55.4%, versus 35%), and equal rates of state density (a very low .16), la Reforma has outstripped its neighbor in every development metric besides income—whether illiteracy, infant mortality, HDI scores, chronic hunger, and even homicides. This has occurred, moreover, despite recording four times as many victims of conflict and violence during the civil war (a statistically significant drag on development as demonstrated earlier), and more recently, spending half as much on social services per inhabitant than its neighbor. Several factors explain these diverging outcomes. First, La Reforma, has visibly less social discord influencing its local decision-making. In Chiquimula, widening divisions between Catholics and Evangelicals (and increasingly between

9

Evangelical sects themselves) have now overlapped with political party affiliation. The result is a toxic political milieu and the collapse of Hambre Cero oversight and effective implementation. Second, the once influential role of traditional leaders in Chiquimula, organized around customs of voluntary service (voluntario instituciona´l), has quickly eroded. It has been usurped by party politics, growing disinterest among youth, and even the indirect influence of regional indigenous organizations like the powerful 48 Cantons of Totonicapa´n, which takes an ardent anti-state approach to local leadership and politics. Third, the persistent relevance of traditional leaders in La Reforma, combined with committed state actors and the absence of divisive party politics allows for the complementarity between state-directed resources and locally-led mobilization. State actors worked with municipal leadership to minimize the spillover of confessional divisions. State and municipal leadership coordinated with traditional leaders to enroll citizens in health and education programs, and to guarantee more effective distribution of resources throughout the municipality’s hinterlands. These factors are explored in more detail below. (ii) New social discord in a homogenous community: the death of governance Chiquimula is located at 6,975 feet (2100 m) above sea level in the interior of Totonicapa´n. Chiquimula is both one of the poorest communities in Guatemala and one of the ‘‘hungriest,” making it in essence one of the communities most affected by chronic hunger in the entire Western hemisphere. Despite this a shared need, social discord has wracked the community, and made effective response impossible. As noted above, Chiquimula is divided both along political party lines as well as new religious divisions. These have hardened as the two have since overlapped. The locally dominant National Unity of Hope party (Unidad Nacional de Esperanza, or UNE) and its leaders are generally Evangelical Protestant, while the affiliates of the President’s Patriot (Patriota) party remain loyal to the more traditional Catholic Church. Evangelical Christianity has been steadily gaining in popularity in Guatemala for decades, according to some the result of a systematic recruitment campaign that capitalizes on mass communication and entertainment, as well as overlaps with traditional beliefs that make it appealing to indigenous followers (Althof, 2014; Schultze, 1992; Turek, 2015). The growing relevance of party affiliation at the local level, meanwhile, might seem in conflict with Guatemala’s infamously weak party system, where parties form and collapse spontaneously around personas rather than coherent platforms (Jones, 2011; Schneider, 2012), but it is not. Party politics at the community level represent more the unstable quid pro quo of simple vote buying by regional party organizers in exchange for support of local ambitions, than it does any sort of grassroots organizational prowess. In Zacualpa, the highly contentious three-term mayor has represented three different parties. In another community, the exmayor monopolizes development resources despite his electoral loss, supplied by regional elites tied to his party. The newly elected officials meanwhile try in vain to implement competing programs and patch-gaping budget holes. Following Patriota’s electoral loss in Chiquimula in 2011, but simultaneously bolstered by the party’s presidential victory, the Ministry of Social and Economic Development (MIDES) and agricultural development ministry (MAGA) moved their respective headquarters from the municipal building

Please cite this article in press as: Klick, M. T. The Effect of State–Local Complementarity and Local Governance on Development: A Comparative Analysis from Post-War Guatemala, World Development (2016), http://dx.doi.org/10.1016/j.worlddev.2016.01.005

10

WORLD DEVELOPMENT

to different corners of town. Based on candid focus group feedback, basic resources are systematically distributed with the express purpose of vote-buying and an anti-mayor smear campaign, rather than need. In more remote corners of the municipality, where hunger rates are among the highest anywhere, not a single focus group participant had heard of Hambre Cero. Elected officials, in turn, increasingly monopolize the political process of decision-making. The most prominent local NGO, which had been running a successful maternal health education program with help from the municipality, is increasingly passed over and its funding cut off. The mayor created a competing maternal health program and appointed a family member as its lead. Potential jobs in municipal offices are used increasingly for patronage and to empower family members and political supporters at the expense of merit-based practices. One municipality worker confided that he was asked to either leave his post, or pay what amounted to several months of salary. Historically influential actors have also been steadily squeezed out of the decision-making process by elected officials and party competition. Village mayors (alcaldes comunitarios), the respective village development councils (consejos comunitarios de desarrollo or COCODES), long-standing local NGOs with a development focus, and even the council of elders (T’zolojche’) had comprised a more communal, if imperfect, governance network. Village elders had only one year earlier openly chastised state officials for their mishandling of a direct transfer program that made national news, 11 but are now absent from development decision-making in any capacity. Ironically, the indirect influence of a more formal, wellfinanced, and well-organized regional indigenous organization—the 48 Cantons of Totonicapa´n—further complicates Chiquimula’s political environment. The organization has continually tried to build a following within Chiquimula (it is already prominent in the department capital, the city and municipality of Totonicapa´n). Its ardent resistance to state encroachment creates yet a further barrier to cooperation with state actors in the field—an observation lamented to me by Hambre Cero’s regional coordinator. Though patronage in local government is unsurprising, the increasingly delimited battle lines drawn in Chiquimula have disintegrated traditional patterns of governance by disempowering formerly influential figures, including traditional authorities. (iii) Partnerships between state and society: complementarity and development governance at the local level Things are very different in la Reforma, 30 km away. First, different government agencies organize themselves around Hambre Cero, as intended. Interview participants from government agencies routinely describe working in conjunction with other offices in order to accomplish specific targets. In a crucial monthly meeting, the Hambre Cero coordinator (a state employee), to which mostly everyone deferred, criticized a colleague from the government’s literacy program (CONALFA) for not securing better participation with local partners in two of the ten outlying villages. This accountability is in stark contrast to anything witnessed in Chiquimula. Second, state and local (formal) leadership work together to bridge community divisions, if simply for the purposes of enrolling citizens for health and education programs. Both the mayor and Hambre Cero coordinator separately recounted efforts to actively bridge the gap between traditional church followers and the emerging Pentecostal movement in la Reforma, for example. Cuesta, each told me separately, cuesta mucho. . . ‘‘It’s been very challenging”. . . ‘‘But,” the mayor added, ‘‘there really is no other option.” Separately, the mayor

organizes a monthly Hambre Cero meeting, noted above, which brings together all participating actors. This process, which is also part of the program’s design, has become so politicized elsewhere that it is rarely, or never, held. Third, and essential, cooperation between state and local actors is visibly apparent. The semi-formal village leaders from throughout la Reforma (alcaldes comunitarios) are active and invested in development matters. In one meeting between only the village leaders themselves, organized around market day in the municipal center, they discussed exactly how best to work with the local Hambre Cero coordinator in order to identify crisis situations, and families most in need. In another, the local director of the government entity in charge of registering citizens and issuing new IDs (RENAP), visited the alcaldes and implored them for their help in mobilizing citizens, alleviating people’s fears, and even helping arrange transportation to the office where they were digitizing registrations and issuing new identification cards. The leaders solemnly agreed. In subsequent interviews, it became clear that this is an established role of the alcaldes, who had already been bringing people to the local health clinic for vaccinations and ante-natal checkups, and working with state officials in order to build trust. Moreover, the RENAP liaison and others confided that, without the assistance of the alcaldes, they could never reach their target population in the most rural locales, and that forcing it otherwise would be in vain. At the municipal level, the mayor and alcaldes each describe cordial relations, but also relative independence from one another, in contrast to Chiquimula where the Mayor had cowed village leaders in his effort to assert his authority vis-a`-vis party enemies. This level of deference by formal leadership to informal, more customary actors was pivotal in at least implementing Hambre Cero over its ‘‘last mile.” 4. DISCUSSION: EXPLORING ROOT CAUSES OF VARIATION IN GOVERNANCE Governance writ large in Santa Lucı´a is inclusive of nonstate actors, while complementarity between government resources and locally-based traditional leadership explains its superior development outcomes compared with similar communities. Indeed, complementarity between state resources and local actors explains the gulf between Chiquimula and la Reforma, but why it exists in one community and not another is less clear. Tsai (2007) observes that the presence of ‘‘solidary groups” in rural China that both encompass the public, and embed local officials, explains increased spending on infrastructure developments at the village level. Unlike the Chinese context, however, social boundaries and political boundaries clash rather than overlap in rural Guatemala, as divisions between Catholics and Protestants (and between Evangelical sects further) divide communities, and in places, politicians as well. These divisions are actively overcome in la Reforma by a combination of an activist mayor and a committed state coordinator, working in conjunction to limit the impacts of a clerical division on Hambre Cero implementation. This points to two factors that are in la Reforma’s favor. First, there is no political rivalry between the ‘‘ownership” of Hambre Cero and the mayor. The President’s political party (Patriota) and the mayor’s are the same. Though this need not determine a negative outcome—indeed in several communities examined it is not relevant—it no doubt facilitates cooperation between state and formal local leadership under conditions of weak accountability. In other communities, including Chiquimula, party identification underlies the meltdown of

Please cite this article in press as: Klick, M. T. The Effect of State–Local Complementarity and Local Governance on Development: A Comparative Analysis from Post-War Guatemala, World Development (2016), http://dx.doi.org/10.1016/j.worlddev.2016.01.005

THE EFFECT OF STATE–LOCAL COMPLEMENTARITY AND LOCAL GOVERNANCE ON DEVELOPMENT

cooperation between state actors, with critical resources, and the municipality, with disastrous consequences for the distribution of essential goods. Second, the influence of effective and committed ‘‘streetlevel bureaucrats” is apparent. The principal Hambre Cero coordinator in la Reforma spurs the participation of younger workers from other government offices, works in conjunction with the mayor to bridge potentially calamitous confessional divisions, and helps organize the monthly meetings of area stakeholders that is required under Hambre Cero (but which is rarely takes place in the other communities). Critically, these meetings also incorporate the participation of semi-formal indigenous leaders from outlying aldeas, with whom the Hambre Cero coordinator has a positive reputation and effective rapport. This last phenomenon is a distinguishing feature of la Reforma and its development governance. As already described, effective implementation in la Reforma hinges considerably on the mostly unsung influence of informal, traditional village leadership. Village leaders in essence approved Hambre Cero, and subsequently mobilized citizens to enroll in government services, and assisted in the coordination of its implementation in remote areas. Ananthpur and Moore (2010) witnessed a similar phenomenon in south India, where customary village councils (CVCs) worked synergistically with legal-formal elected village councils. Their observations that CVCs with no permanent source of income, and no control of valuable resources, and limited opportunity to abuse power were less threatening to formal leadership and facilitated cooperation rings true in Guatemala as well. Though it explains less why la Reforma’s village leadership remained more relevant than in Chiquimula. Violence during the civil war does not explain this variation, as la Reforma experienced considerably more than its neighbor, and was the site of a military garrison. Instead, I argue, based on qualitative observations, that the degree of political party rivalry, where most intense, has effectively crowded out the role of traditional elders and village leaders. Where political rivalry has a longer history, village leaders have been coopted or turned over so as to serve the sitting mayor. In Zacualpa, for example, where political rivalry has at least a 12-year history, the traditional indigenous mayor (alcadı´a indı´gena) was dismissed out of hand by interviewees as nothing more than a symbolic artifact. Parallel COCODES that supported the mayor were reportedly established as well. In Chiquimula, the intensity of party rivalry has usurped the role of traditional leaders that only a year earlier had effectively protested the state’s development programing. Anecdotally, workers familiar with both communities concluded that the

11

difference between la Reforma and Chiquimula was the persistence of the well-grounded Maya tradition of ‘‘service” in la Reforma. Finally, and more contentious, the influence of outside indigenous groups in Chiquimula might also worsen the situation. The 48 Cantons of Totonicapa´n is a powerful local organization with tremendous influence in the department’s capital and immediate surroundings. There, the organization has effectively barred the implementation of formal village councils (COCODES) that were formed as part of a 2002 decentralization bill, and an effort to respond to the demands of the 1996 Peace Accords. The organization argues that COCODES are nothing but a tool of the state to usurp indigenous influence over local matters. Though the COCODES system is corrupted throughout much of the highlands, local leaders, including indigenous leaders in Chiquimula, are leery of this approach and its implications. And despite the Canton’s power, chronic hunger in the aldeas remains brutally high under its watch. The organization has consistently tried to influence Chiquimula’s politics, and encourage an alliance across municipalities. In the Chiquimula municipal offices, the influence of the 48 Cantons is ardently resisted. Meanwhile the presence of the 48 Cantons was undetected in la Reforma where, despite its proximity, it is much harder to access from the department capital. 5. CONCLUSIONS Under conditions of minimal state presence, local governance is a dynamic mix of formal and informal authority that can create a dead space in which top-down development programs, no matter how sophisticated, are twisted, corrupted, or stopped dead. But this mix can provide opportunities as well, to marry the strengths of certain actors (with resources, especially) with the legitimacy and local relevance of others. This variation in governance across short distances complicates how development implementation de facto takes place, and moreover how development practice might be perceived under conditions of weak states here forward. This paper suggests, however, that there is ample opportunity for a ‘‘development as diplomacy” approach to implementation under these shifting contexts, and to create nodes of dialog between factions, including the state, or with local non-state interlocutors. This is a hopeful scenario, admittedly, given the rarity of organically generated state–local complementarity, but it embraces the reality of local conditions that development practice must contend with.

NOTES 1. Including $1.758 billion in IRBD loans as of 5/31/2012, and an average of approximately $95 million in recent years from the US government, consisting of $28 million dedicated to ‘‘health” in 2012, $9 million in education and social services and another $19 million for economic development for a country of almost 15 million people, see: http://foreignassistance.gov/OU.aspx?OUID=205&FY=2012&AgencyID=0&budTab=tab_Bud_Planned&tabID=tab_sct_Peace_Planned (accessed 7/5/2013). OECD DAC official development assistance (ODA) totaled an additional $289 million in 2011—second only to Haiti in the region (http://stats.oecd.org/Index.aspx?datasetcode=TABLE2A). 2. See http://booth.lse.ac.uk/static/a/4.html.

3. I coded conflict intensity based on the number of victims tied to each community as recorded by the 1999 Memoria del Silencio report, published by the Historical Clarification Commission which was established by the 1996 Peace Accords, and which is still the authoritative source on civil-war related violent events. I manually counted all victims, regardless of perpetrator (guerilla or army) or crime (whether arbitrary execution, torture, or sexual violence) that took place or originated in each community. By this count, conflictrelated violent incidents of QCA communities were: Santa Marı´a Chiquimula (36), Santa Lucı´a la Reforma (146), Zacualpa (661) and Patzite´ (44). I calculated an ‘‘average” score of 96 victims/community for comparison.

Please cite this article in press as: Klick, M. T. The Effect of State–Local Complementarity and Local Governance on Development: A Comparative Analysis from Post-War Guatemala, World Development (2016), http://dx.doi.org/10.1016/j.worlddev.2016.01.005

12

WORLD DEVELOPMENT

4. The use of extrapolated HDI scores is an attempt to overcome the endogeneity that would stem from regressing 2005 numbers on 2009 state density scores, as well as to maximize the availability of two individual years by creating a trend, rather than relying on a single point in time. Naturally, this score by itself is insufficient to represent the state of development in a community and is used only as a starting point, before comparing with results from other development indicators. The similarity of results across regressions does suggest a level of validity in using this method, however.

5. State Density Index scores were calculated by UNDP Guatemala (2011) (see Appendix A).

6. Local social spending per inhabitant (Gasto_Muni) is reported by the United States Agency for International Development (USAID) and Instituto Centroamericano de Estudios Fiscales (ICEFI) (2009) ‘‘Atlas del Gasto Municipal” (Municipal Social Spending Atlas). The Human Rights Ombudsman was created by the 1985 constitution as an arbiter and of human rights abuses, though its director is approved by Congress, raising

questions of its independence and credibility (see: http://www.pdh.org.gt/). SDI_2009 is log-transformed to create log_SDI, in order to more normally distribute the otherwise clustered nature of state density figures. 7. Hom_Diff reports the change in homicide rates during 2003–12 at the municipal level. 8. I initially selected six communities, including two in the Department of Solola´. My fieldwork there indirectly informs findings presented here, but are otherwise excluded because of their proximity to more developed tourist centers and cash income that skew results. 9. Quiche´ (63.9%) and Totonicapa´n (69.4%) are listed as suffering from ‘‘very high” rates of chronic hunger officially. 10. See: http://www.hancindex.org/. 11. Gobiernos indı´genas contra alcaldes electos en Totonicapa´n,” El Periodico, http://www.elperiodico.com.gt/es/20120725/pais/215555/ (accessed 7/2013).

REFERENCES Adveenko, A., & Gilligan, M. J. (2015). International interventions to build social capital: Evidence from field experiments from Sudan. American Political Science Review, 109(3), 427–449. Albrecht, P., & Kyed, H. M. (2011). Introduction: Non-state and customary actors in development programs. In P. Albrecht, H. M. Kyed, D. Isser, & E. Harpe (Eds.), Perspectives on involving non-state and customary actors in justice and security reform (pp. 3–22). Rome: International Development Law Organization. Althof, A. (2014). Religion and society: Divided by faith and ethnicity: Religious pluralism and the problem of race in Guatemala. Berlin: Walter de Gruyter. Altschuler, D., & Coralles, J. (2013). The promise of participation: Experiments in participatory governance in Honduras and Guatemala. London: Palgrave Macmillan. Ananthpur, K., Malik, K., & Rao, V. (2014). The anatomy of failure: An ethnography of a randomized trial to deepen democracy in rural India Policy research working paper 6958. The World Bank Development Research Group Poverty and Inequality Team. Ananthpur, K., & Moore, M. (2010). Ambiguous institutions: Traditional governance and local democracy in rural south India. Journal of Development Studies, 46(4), 603–623. Azpuru, D. (2011). Political culture of democracy in Guatemala in 2010: Democratic consolidation in the Americas during hard times, IX study of democratic culture of Guatemalans. Nashville, TN: Vanderbilt University (Retrieved from: http://www.vanderbilt.edu/lapop/guatemala/ 2010-political-culture-English.pdf). Boege, V., Brown, A., Clements, K., & Nolan, A. (2008). On hybrid political orders and emerging states: State formation in the context of ‘‘fragility”. Berghof Research Center for Constructive Conflict Management, Handbook. (Retrieved from: http://www.berghof-handbook. net/documents/publications/boege_etal_handbook.pdf). Boone, C. (2003). Political topographies of the Africa state: Territorial authority and institutional choice. Cambridge: Cambridge University Press. Booth, C. (1902). Life and labour of the people in London (Vol. 1). London: Macmillan. Brown, M. A. (Ed.) (2007). Security and development in the Pacific Islands. Social resilience in emerging states. Boulder, CO & London: Lynne Rienner. Cammett, M., & MacLean, L. M. (2014). Introduction. In M. Cammet, & L. M. MacLean (Eds.), The politics of non state social welfare (pp. 1–16). Ithaca, NY: Cornell University Press. Chopra, T. (2009). When peacebuilding conflicts with statebuilding: Notes from the Arid Lands of Kenya. International Peacekeeping, 16(4), 531–545. Fox, J. (1992). The politics of food in Mexico: State power and social mobilization. Ithaca, New York: Cornell University Press.

Gennaioli, N., La Porta, R., Lopez-de-Silanes, F., & Shleifer, A. (2011). Human capital and regional development. National Bureau of Economic Research, Working paper 17158. George, A. L., & Bennett, A. (2005). Case studies and theory development in the social sciences. Cambridge: MIT. Godoy, A. S. (2006). Popular injustice: Violence, community and law in Latin America. Stanford, California: Stanford University Press. Guatemalan Commission for Historical Clarification. (1999). Guatemla, memory of silence. Tz’inil Na’tab’al. Report of the Commission for Historical Clarification. Guatemala. Jones, M. P. (2011). Weakly institutionalized party systems and presidential democracy: Evidence from Guatemala. International Area Studies Review, 14(4), 3–30. Kaplan, O. (2013). Protecting civilians in civil war: The institution of the ATCC. Colombia Journal of Peace Research, 50(3), 351–367. Khan, S. R., Rifaquat, Z., & Kazmi, S. (2007). Harnessing and guiding social capital for rural development. New York: Palgrave MacMillan. Klick, M. (2013). More than legitimacy: Spatial variation in human development in post-war Guatemala. Journal of Peacebuilding and Development, 8(3), 7–22. Krishna, A. (2002). Active social capital. New York: Columbia University Press. Lake, D. A. (2010). Building legitimate states after civil wars. In M. Hoddie, & C. Hartzell (Eds.), Strengthening peace in post-civil war states (pp. 29–52). Chicago: University of Chicago Press. Larson, A. M. (2008). Indigenous peoples, representation and citizenship in Guatemalan forestry. Conservation and Society, 6(1), 35–48. Menkhaus, K. (2008). The rise of a mediated state in Northern Kenya: The Wajir story and its implications for state-building. Afrika Focus, 21(2), 23–38. Migdal, J. S., & Schlichte, K. (2005). Rethinking the State. In K. Schlichte (Ed.), The dynamics of states: The formation and crises of state domination (pp. 1–40). Burlington, VT: Ashgate Publishing. Ministry of Education. (2001). Segundo Censo Nacional de Talla en Escolares del Primer Grado de Educacio´n Primaria de Guatemala. Guatemala C.A. Ministry of Education, Secretaria de Seguridad Alimentaria y Nutricional. (2009) Tercer Censo Nacional de Talla en Escolares del Primer Grado de Educacio´n Primaria de Guatemala. Guatemala C.A. Pugh, M. (2005). The political economy of peacebuilding: A critical theory perspective. International Journal of Peace Studies, 10(2), 23–42. Richmond, O. (2006). The problem of peace: Understanding the ‘‘liberal peace”. Conflict, Security and Development, 6(3), 291–314. Rihoux, B., & Ragin, C. C. (2009). Configurational comparative methods: Qualitative comparative analysis and related techniques. London: Sage. Risse, T. (2011). Governance in areas of limited statehood: Introduction and overview. In T. Risse (Ed.), Governance without a state? Policies

Please cite this article in press as: Klick, M. T. The Effect of State–Local Complementarity and Local Governance on Development: A Comparative Analysis from Post-War Guatemala, World Development (2016), http://dx.doi.org/10.1016/j.worlddev.2016.01.005

THE EFFECT OF STATE–LOCAL COMPLEMENTARITY AND LOCAL GOVERNANCE ON DEVELOPMENT and politics in areas of limited statehood (pp. 1–35). New York: Columbia University Press. Rotberg, R. (2003). State failure and state weakness in a time of terror. Washington, DC: Brookings Institute Press. Saito, F. (2008). Decentralization and local governance: Introduction and overview. In F. Saito (Ed.), Foundations for local governance (pp. 1–24). Leipzig, Germany: Physica-Verlag. Schneider, A. (2012). State-building and tax regimes in Central America. New York: Cambridge University Press. Schultze, Q. J. (1992). Catholic v. protestant: Mass mediated legitimation of popular evangelicalism in Guatemala. Public Relations Review, 18(3) , 257–263. Scott, J. C. (1998). Seeing like a state: How certain schemes to improve the human condition have failed. New Haven: Yale University Press. Scott, J. C. (2009). The art of not being governed: An anarchist history of upland southeast Asia. New Haven: Yale University Press. Sieder, R. (2011). Contested sovereignties: Indigenous law, violence and state effects in postwar Guatemala. Critique of Anthropology, 31(3), 161–184.

13

Speer, J. (2011). Participatory governance, accountability, and responsiveness: A comparative study of local public service provision in rural Guatemala. Unpublished dissertation. Stepputat, F. (2015). Formations of sovereignty at the frontier of the modern state. Conflict and Society: Advances in Research, 1, 129–143. Tsai, L. T. (2007). Solidary groups, informal accountability and local public goods provision in rural China. American Political Science Review, 101(2), 355–372. Turek, L. F. (2015). To support ‘‘a brother in Christ”: Evangelical groups and U.S.–Guatemala relations during the Rı´os Montt Regime. Diplomatic History, 39(4), 689–719. UNDP Guatemala. (2011). ‘‘Guatemala: hacia un estado para el desarollo humano”. Informe nacional de desarollo humano 2009/2010. United States Agency for International Development (USAID) and Instituto Centroamericano de Estudios Fiscales (ICEFI). (2009). Atlas del Gasto Municipal, Guatemala: USAID/ICEFI. (Retrieved from: http://issuu.com/maijala/docs/01_atlas_portadacompleto). Yashar, D. (1998). Contesting citizenship: Indigenous movements and democracy in Latin America. Comparative Politics, 31(1), 23–42.

APPENDIX A. STATE DENSITY INDEX CALCULATION Components

Education

a. Presence of state dependents b. Bureaucracy (per population)

Coverage Teachers

c. Budget (per capita) Sub-indices

Q/person iedu = l(xi1)

Health

Other

Sub-Inidices

Level of service Personnel, Minister of Public Health and Social Services (MSPAS) Q/person isal = l(xi2)

Numer of ‘‘dependants” Employees

ia = l(x1j) ib = l(x2j)

Q/person iotr = l(xi3)

ic = l(x3j) IDE = l(xij)

APPENDIX B. QCA CODING CRITERIA PER VARIABLE Presence of:

Coding criteria:

Social spending (A = high, a = low) ‘‘Strong” local civil society (B = high, b = low)

Social discord (C = high, c = low) Conflict affected? (D = high, d = low) Complementarity (E = high, e = low) Improvements in Hunger (DV) D Illiteracy (DV)

ICEFI/USAID ‘‘Atlas del Gasto Social Municipal”: % of total budget per inhabitant spent on ‘‘social functions” = above (high) or below (low) mean (234.54 Quetzales/person) Qualitative observations and interviews: What is the presence and density of locally-resonant forms of civil society or traditional authority? More crucially, can these organizations affect change, mobilize citizens, and spur deference by other actors? Qualitative observations and interviews: What is the extent and nature of social divisions within a community, if any? How do they manifest? Historical experiences with the civil war: Above (high) or below (low) the mean conflict intensity ‘‘score” (96 victims/community). This is augmented by qualitative data gained through interviews Process tracing: Observation of lead civil society actors working, or not, with government officials to implement anti-hunger program elements Beginning with the difference between Guatemala’s 2nd and 3rd ‘‘hunger census” (which measures stunting) scores, differences as a percentage and compare across communities are calculated. Changes in illiteracy rates during 2009–11 based on official statistics and compared with mean (1.6%) and SD (5%). Greater than 2 SDs = ‘‘High”

Available online at www.sciencedirect.com

ScienceDirect Please cite this article in press as: Klick, M. T. The Effect of State–Local Complementarity and Local Governance on Development: A Comparative Analysis from Post-War Guatemala, World Development (2016), http://dx.doi.org/10.1016/j.worlddev.2016.01.005