The Political Economy of Public Reform Adoption: Patterns in Twenty Indonesian Districts

ABSTRACT

New Public Management (NPM) reforms are often perceived as technical, with little attention to political and institutional factors. Comparing choices in 20 Indonesian districts, we explore which of four political economic factors influence uptake of NPM-based service delivery reforms. We find that democratic political competition laid the groundwork for political alliances, patterns of patronage, and party provision of benefits that condition reform choices. State-led policy entrepreneurship was evident from education agency technocrats. Public sector modernization may have increased orientation towards performance, with education reforms adopted by districts already achieving relatively good sectoral results. Health reforms were more common in districts providing greater opportunities for citizen participation. The complex interactions among the factors argue for working within these realities, rather than seeing them as impediments to be avoided in a drive for reforms. Education technocrats’ dominance, bolstered by central policy priorities, argues for more nuanced mechanisms for meeting national goals to avoid crowding out responsiveness to local citizens.

INTRODUCTION

Prescriptions and related reform programs to address performance weaknesses have long had their foundations in the principles and practices of New Public Management (NPM). NPM has several variants; we focus on the strand that combines elements of a governance ideology that values limited government and a managerial reform agenda with associated tools that apply market and private-sector principles to governments’ fulfillment of their responsibilities, in particular their public service delivery tasks (Hood 1991; Ferlie et al. 1996). The core perspective of NPM on performance problems is that they derive from ineffective principal-agent relationships, which create weak incentives for performance and limited government responsiveness to citizens. NPM's primary remedies for public sector performance failings and flawed principal-agent relationships include sharpened accountability, contracting in and contracting out, clearly specified objectives and measures, performance-based budgeting, and close monitoring and evaluation.¹

In developing countries, NPM reforms have been central to the public sector modernization and “good governance” programs of international donors since the 1980s (Batley and Larbi 2004; Manning 2001). Thirty years of experience with reform efforts across a wide range of countries have produced a vast literature documenting progress and problems, exploring causal explanations, evaluating successes, and offering remedies to address failures, constraints, and unanticipated consequences (Brinkerhoff 2008). Among the common findings across a wide variety of both reforms and countries is that, while reforms tend to be conceived of as largely technical in content, successful implementation requires attention to political and institutional forces. These factors have come to be grouped under the broad rubric of political economy.

Indonesia is among the countries that have pursued performance-enhancing public sector reforms that embody NPM principles. In a previous study, we analyzed the range of performance-based reforms to improve service delivery that Indonesia has implemented (Brinkerhoff and Wetterberg 2013). We found that, while many of the reforms are in place, they have not been implemented as designed, often due to complex political relations between central and district actors. In this article, we build on that work to focus on reform choices among decentralized local municipalities. We seek to answer the question: Which political economic factors influence the uptake of NPM-based reforms to improve service delivery? Our aim is to offer insight into which of the political economy explanations posited in the development literature are relevant to reform choice in a recently decentralized, post-authoritarian context, such as Indonesia. Our focus is thus not on the implementation or effects of reforms, but rather on the antecedent process of reform choice. The analysis contributes to the debate over the causal pathways through which decentralization was intended to work and whether we see these operating as expected. It also aims to enhance understanding of how political economy factors can be taken into account so that implementation of reforms can avoid derailment and achieve intended results.

Our investigation is exploratory, using limited data drawn from reform choices in the early stages of the Kinerja program (funded by the US Agency for International Development), which works in 20 districts in four provinces: Aceh, East Java, South Sulawesi, and West Kalimantan. The analysis is complemented by other studies that investigate the political and institutional dynamics of governance and reforms in Indonesia.

Of the four factors considered—political competition, state entrepreneurship, public sector modernization, and civil society activism (Grindle 2007a)—we find that government-led policy entrepreneurship is the most frequent driver behind reform choice.² Especially in sectors where they are bolstered by national policies, technocrats often dominate reforms, reinforcing prior high performance and investment in their own sector. Where there have been past modernization reforms, however, and citizen participation is relatively stronger, reforms more often focus on sectors where there has been past underinvestment.

The article proceeds as follows. After introducing four key elements of political economy analysis, we briefly and selectively overview the literature on explanations for performance-enhancing behaviors, with a focus on decentralized, local-level public officials. We then turn to the Indonesian experience and examine how the political economy factors identified in the literature apply to the Kinerja project's target municipalities’ adoption of public service improvement. We next discuss key findings, and offer some concluding thoughts on interventions to enhance public service delivery performance.

Political Economy of Service Delivery Reform

Political economy combines politics with economics to examine the underlying dynamics of the relationships among states, markets, and society, and to explain outcomes. Politics concerns the contestations, bargaining, and power distributions among these three that determine “who gets what” in order to achieve some set of goals. Economics focuses on exchange relations in terms of resources, transaction costs, information, efficiencies, and incentives. Political economy analysis, then, focuses on actors and institutions operating within a particular structural context (Fritz et al. 2009). Actors pursue their interests, operating within the rules and roles that institutions (formal and informal) establish, subject to a range of structural variables (e.g., level of development, natural resource endowments, urban–rural/center–periphery relations, historical legacies, and so on).

Such analysis can lead to the identification of drivers of change, as well as constraints and obstacles (Bunse and Fritz 2012). In a review of the political economy of service delivery, Wild et al. (2012), for example, examine how “political market imperfections”—various types of dysfunctional relationships between citizens and politicians—can have negative impacts on access, quality, and equitable availability of public services. Mcloughlin and Batley (2012) explore the connection between political-economic factors and service delivery, identifying, for example, favorable institutional conditions (e.g., political settlements and a social contract that create positive incentives) and political circumstances (e.g., the dynamics of national and local competition, and electoral cycles) that can lead to improved service delivery. Batley et al. (2012) highlight emerging research on causal chains in the opposite direction, focusing on the feedback effects of service delivery on politics, which suggests the possibility of the emergence of virtuous cycles.

With regard to decentralized local governments’ decisions to improve provision of public services, four main explanations have been advanced: democratic political competition, state-led policy entrepreneurship, public sector modernization, and citizen participation. In her study of municipal government performance in Mexico, Grindle (2007a) finds entrepreneurship by elected officials to be the most important explanatory factor. She also notes, however, the contribution of the other possible explanations, which enabled officials’ reform initiatives by “the expansion of opportunities for competitive elections in a more democratic context and … a considerable amount of capacity building and citizen demand making” (22). Each of these explanations is associated with an active stream of investigation and debate, which we touch on only briefly here. For each of the main hypotheses, however, we introduce variants that could potentially account for reform choices (Table 1). Variants derive from our literature review and capture a selection from the full range of possible sub-hypotheses related to our data.

TABLE 1 Proposed Explanations for Reform Choice, with Variants

Main Factor	Variants	Support in Kinerja Districts
A. Democratic political competition
A1.	Candidates compete for individual votes by offering needed public goods, and providing them after election.	No data. Other research has shown no evidence of direct elections affecting responsiveness to health and education priorities in Indonesia.
A2.	The introduction of political competition provide opportunities for building constituencies and new political alliances, which may promote public goods provision, but can also increase elite capture.	Supported (weakly). Anecdotal data of elected officials prioritizing/blocking reforms based on priorities of constituent groups. Officials without constituencies or political alliances may be stymied in reform efforts.
B. State-led policy entrepreneurship
B1.	Entrepreneurship is driven by officials’ ideas, leadership skills, and strategic choices.	Not supported. Level of interest in reforms is not primary driver of reform choice.
B2.	Entrepreneurship is conditioned on actors’ positions in broader structures, but reforms reflect their own, local priorities.	Supported. Generally more powerful education technical agency appears to have “right of first refusal” for reform choice. If reforms do not correspond to interests, choice passes to other sectors.
B3.	Entrepreneurship by local actors is driven by and a reflection of national priorities.	Mixed support. National policy priorities strengthen position of local actors but do not determine reform choices, which are driven by local circumstances.
C. Public sector modernization
C1.	Reforms are motivated by need to remedy local shortcomings in past performance and investment.	Mixed support. Education reform choice reinforces relatively high past performance and investment. Health reform choice tends to compensate for relatively low past investment.
C2.	Prior policy changes to specifically improve responsiveness are critical to reform choices.	Mixed support. Health reform choice more likely in districts with prior modernizing reforms. No pattern observed with education reform choice.
D. Citizen participation
D1.	Creating opportunities for interactions between officials and citizens promotes reforms.	Mixed support. Health reform choice more likely in districts with more opportunities for citizen participation. No pattern observed with education reform choice.
D2.	More general strengthening of civil society (through CSOs) leads to influence over reform choices.	Mixed support. Health reform choice somewhat more likely in districts with stronger health CSOs. No pattern observed with education reform choice.

Democratic Political Competition

Many scholars identify political competition at multiple levels as the key contributor to responsive government, arguing that, in truly democratic circumstances, officials have compelling incentives to provide the policies and services that citizens want or face electoral ouster (Hecock 2006; Ponce-Rodríguez et al. 2011). Competition among political parties, and across local jurisdictions, is posited to create pressures on elected officials to perform effectively. This is the long route to accountability (World Bank 2004); candidates compete for votes by offering needed public services and must provide them if they hope to be re-elected (Variant A1, Table 1). Major analytic threads (both theoretical and empirical) focus on how democratic decentralization improves allocative efficiency through matching services with citizen preferences, increases service production efficiency and cost recovery, and affects intergovernmental fiscal relations (e.g., Lockwood 2009; Shah and Thompson 2004; Faguet 2004; Tiebout 1956).

However, various analysts note the gap between theory and practice. For example, in a comparative study of Uganda and the Philippines, Azfar et al. (2001) find little evidence of local election voting being driven by service-delivery concerns. Boulding and Brown (2012), in a study of Brazilian municipalities, note the intervening role of financial resources on political competition and social spending. They find that municipalities with few resources have low voter turnout and incumbents who consistently fail to win reelection, whereas, in well-resourced municipalities, incumbents are able to mobilize substantial numbers of voters and tend to win by wide margins. Cleary's (2007) study of Mexican municipalities also finds a weak to nonexistent link between elections and government performance, and identifies institutional barriers to performance, such as term limits and the dynamics of national party politics, which create perverse incentives against elected officials focusing on citizens’ demands and needs. Thus, as an explanation for responsive service delivery, political competition is mediated by a number of structural and institutional variables that affect public officials’ behaviors (e.g., financial resources and national party politics).

In a post-authoritarian context, the introduction of democracy also provides new bases for political legitimacy that can influence the prospects for service delivery reforms (Variant A2, Table 1). Cleary (2007) posits that it is not the degree of electoral competitiveness that is associated with responsiveness, but rather that elections encourage officials to listen to citizens’ viewpoints. Instead of a direct relation between competitiveness and responsiveness, Cleary argues that elections prompt officials to build constituencies and encourage citizen participation in decision-making, which can improve government performance. Of course, influential constituencies may dominate both electoral processes and, as a result of their support for winning candidates, mechanisms intended to facilitate citizen participation, resulting in elite capture, rather than broad-based responsiveness (Ackerman 2004; Alatas et al. 2013; Casey et al. 2012; Dasgupta and Beard 2007; Fritzen 2007; Lund and Saito-Jensen 2013).

State-Led Policy Entrepreneurship

Reformist technocrats pursuing change agendas can stimulate adoption of performance-enhancing policies and programs. A growing literature on leadership and individual change agents frames the elements of policy entrepreneurship. Andrews et al. (2010), for example, characterize leadership for development as creating “change space” by fostering acceptance of new policies and practices, granting authority for change, and building capacity. Leonard (2010) demonstrates that, within almost any institutional context, committed leaders can create and/or expand space for change. Among the well-known examples of performance-enhancing innovation that derived from reform-oriented actors is the participatory budgeting process that emerged from the Brazilian city of Porto Alegre and has been widely applied in other countries (Baiocchi 2003). This explanation focuses on the interests of public-sector actors in positions of power and authority, their perspectives on and commitment to reform, the incentives they face, and their ability to build coalitions to move their agendas from plans to action.

The balance varies, however, between local officials’ agency in driving reforms and the influence of structures that empower them. At one end of the spectrum, explanations attribute changes primarily to officials’ skills in strategically marshalling resources and networks to promote their ideas (Eyal 2000; Wallis 1999; Variant B1 in Table 1). A second variant (B2) acknowledges that local officials’ agency is critical to reforms, which then reflect their knowledge and priorities. Importantly, however, opportunities for entrepreneurship are determined by reformers’ embeddedness in broader social structures and political networks (Evans 1995).

Central–local relations play an important role in influencing whether centrally driven policy reforms are implemented as intended, particularly the configuration of power relationships between central and regional/local elites (Manor 1999; Crook 2003). As Das Gupta et al. (2004) note, central governments can exercise their power over sub-national levels to support the achievement of national development objectives, such as poverty alleviation or reduction in maternal deaths. Booth (2011a), drawing on findings from a research program on African politics and power, notes that local service delivery success is linked to top-down discipline and motivation. In Indonesia, where general revenue allocations (Dana Alokasi Umum or DAU)³ on average cover over a third of provincial budgets and two-thirds of district budgets (Lewis and Smoke 2011), Olken (2005) found that increasing the likelihood of audits by a central government agency reduced corruption in local governments. A third variant (B3) of the state-led policy entrepreneurship explanation thus favors the broader political context over local officials’ agency, arguing that the former drives the latter's opportunities for reform, as well as reform substance (see Hypothesis 5 in Leonard 2010).

Public Sector Modernization

Decisions to restructure public organizations, adopt new management tools and accountability systems, privatize selected services, and so on, can create the conditions and incentives that lead to enhanced performance. This is the premise driving politicians and other public officials to seek to modernize their public sectors. In many developing countries, these actors look to international donors for help and advice. Donor-supported public sector reform programs are largely based on NPM principles and toolkits (Manning 2001). These programs provide technical inputs to build capacity, along with associated training and resource transfers. In addition, access to capacity, training opportunities, and resources can serve as incentives to local officials to pursue modernization reforms. These incentives are likely to be particularly strong in countries with higher levels of aid dependence.

A large literature has debated the pitfalls of the “one best way/best practices” conceptualizations of policies, organizational structures, and processes that have been associated with public sector modernization, particularly as operationalized by international donor-funded programs (Evans 2004; Pritchett and Woolcock 2004; Brinkerhoff 2008). One aspect of the debate has to do with whether the reform agenda associated with public sector modernization exceeds the capacities of countries undertaking reforms to carry through with them, which has fueled the “good enough governance” discourse (Grindle 2007b). The political economy perspective delves beyond the narrow technical capacity explanation. The extent to which modernization leads to enhanced public sector performance and better service delivery relies fundamentally on whether reforms align or are at odds with the objectives, interests, and incentives of political actors in the country.

In Indonesia, the most far-reaching modernization reforms have been the series of decentralization laws passed since 1999 (see next section). Part of the motivation behind decentralization has been to improve responsiveness to local needs (Faguet 2004). Reform choices could thus be driven by a general prioritization of sectors plagued by past poor performance or underinvestment (Variant C1, Table 1). Alternatively, reform choice could reflect further efforts by local governments to improve responsiveness through specific reforms to enhance transparency, accountability, and citizen participation (Leonard 2010; Variant C2 in Table 1).

Citizen Participation

Another analytic stream explores how the participation of citizens and their cooperation, or confrontation, with government leaders are primary drivers of improved performance.⁴ NPM market-driven reform metaphors recast citizens from passive recipients of public services doled out by government to individual consumers/customers—“users and choosers” (Barzelay 2001; Cornwall and Gaventa 2001). Good governance and public sector modernization reforms have expanded the customer idiom to incorporate citizens as rights-bearers, which reframes accountability (Joshi and Houtzager 2012). Such changes have created space for citizens, individually and/or collectively, to engage with local authorities as empowered “makers and shapers” of services-related decision making (Cornwall and Gaventa 2001). Which citizens have access to, and influence on, decisions also matters. The political landscape shapes patterns of participation and, importantly, influences the extent to which citizen–politician/service provider relations are clientelist or patronage-based (Kitschelt and Wilkinson 2007). For example, as Bardhan and Mookherjee (2006) argue, the effectiveness of decentralizing service delivery depends critically on the degree to which local elites capture benefits. Debates around the effectiveness of participation in encouraging responsive and accountable service delivery signal the importance of mechanisms that enable citizens to have access to the decentralized decision-making space already populated by politicians, technocrats, and providers. As Ribot (2004) points out, an important question is whether or not decentralized structures and processes put in place accountability mechanisms and incentives for local governments to enfranchise communities. Another question is whether citizens and communities have the capacity and the motivation to exercise an accountability role, given the information asymmetries and power differentials between them and service delivery professionals. In a range of countries, citizens have been engaged in monitoring service delivery performance in one way or another, using tools such as citizen report cards, absenteeism tracking, community “contracts,” and membership on boards and committees (e.g., hospitals and schools). Some studies have shown that community monitoring can have a measurable impact on services (e.g., Björkman and Svensson 2009). One variant (D1 in Table 1) of the citizen participation hypothesis suggests that it is such “invited spaces” (Cornwall and Coelho 2007) that consciously create opportunities for officials and community-members to interact, which motivates reforms (Ackerman 2004; Goldfrank 2002; Yackee 2015).

An alternative explanation (D2 in Table 1) is that reforms ensue from a more general strengthening of civil society, representatives of which can then “claim space” through more confrontational tactics to push for reforms (Cornwall and Coelho 2007; see also Fox 1996; Gaventa and Barrett 2012; Woolcock 1998). Cleary (2007) identifies several drivers by which increased citizen participation can stimulate more local government responsiveness. He cites the influence of participation via shared social networks among government officials and their constituents, expanded information availability concerning citizens’ needs and demands, and incentives for responsiveness to avoid the stress of interaction with angry citizens.

NPM Reform Adoption and Implementation in Indonesia

With the fall of the Soeharto regime, in the wake of the 1997–98 Asian financial crisis and widespread public dissatisfaction, Indonesia began a transition period of ambitious and far-reaching democratic governance reforms. Among them was a rapid decentralization program, beginning in 1999, which progressively transferred the bulk of authority for fiscal and legislative decisions, as well as service delivery, to the district level. Public sector reform programs, funded by international donors, supported the country's ambitious decentralization, and sought to increase administrative efficiency and effectiveness and to reorient the public administration toward citizen responsiveness. These reforms were grounded in NPM principles of shortening routes of accountability, increasing efficiency of public service delivery, and improving effectiveness through performance standard-setting and measurement (Brinkerhoff and Wetterberg 2013).

In parallel to decentralization, Indonesia undertook electoral reforms. Freedom House has classified the country as an “electoral democracy” since 1999, but electoral scope and practice has changed substantially since then (Mietzner and Aspinall 2010). By 2009, voters could directly elect their president, governor, and district head, as well as vote for legislative representatives at each level.⁵

Progress with decentralization has been substantial, but not without problems, as would be expected with change of such a magnitude (Rapp et al. 2006). Among the issues identified have been both gaps and overlaps in the legal framework, unclear amounts and timing of intergovernmental transfers, muddied service expectations and responsibilities, and weak information and feedback flows, all of which affect the principal-agent⁶ relationships among various levels of government, service providers, and citizens (Lewis and Smoke 2011).

While local officials have gained experience with decision making since the late 1990s, their orientation towards performance varies. Declines in some public services, such as health, are evident, while there has been performance improvement in others, such as education (Kristiansen and Santoso 2006; Suryadarma 2012). Some observers attribute declines in service availability and quality to problems of elites capturing service benefits, as well as corruption, and tend to be pessimistic about prospects for the success of NPM-based public sector reforms that Indonesia has undertaken to date (Blunt et al. 2012; Rosser et al. 2011). Others express guarded optimism, indicating the growth of grassroots challenges to entrenched interests and more willingness to experiment with governance innovations (Antlöv and Wetterberg 2013). Eckardt (2008, 6) notes the role of local decision makers in influencing paths to improvement or to remaining in the thrall of ingrained clientelism and elite domination:

Local government performance increasingly varies depending on the extent to which local government took advantage of the opportunities offered by decentralization. A significant number of local governments have forged ahead with reforms and became the locus for innovative forms of governance and better public services. Elsewhere, however, local political contexts became charged with malfeasance, fraud, lack of accountability and failure to respond to the needs of local residents.

The categories of political economy factors reviewed earlier offer four possible explanations for the variation that Eckardt and others have noted in local government adoption of reforms and pursuit of improved performance. To explore the contours of these four explanations, in the next several sections we examine a subset of district governments participating in the US Agency for International Development's (USAID's) Kinerja program.

The Kinerja Program

The Kinerja program works in 20 districts in four provinces (Aceh, East Java, South Sulawesi, and West Kalimantan⁷) to “solidify the links between stimulation of demand for good services through active civil society engagement and improved local government response” (RTI International 2010, 2). The executive in each district leads a consultative process to select a set of services to target for improvement in the first year of reforms.⁸ Another sector is chosen for the second year of Kinerja participation.

Reform sectors include health, education, and local economic development. Specific reforms addressed are exclusive and immediate breastfeeding and safe delivery; school-based management, teacher distribution, and school unit cost analysis; and one-stop shops (OSS) for business licensing. While health and education were possible reform choices in all districts, OSS selection was limited to those that had not previously received similar support.

Within each sector, the project works with local governments, civil society, and service delivery units (schools, clinics, and business licensing offices) on specific interventions, chosen for their alignment with national policy priorities and/or demonstrated effectiveness. Kinerja also includes a series of cross-sectoral interventions designed to create incentives for improved local service delivery performance by giving citizens a more effective voice in public service delivery, supporting performance management systems in local governments, and increasing competition through benchmarking, competitive awards, and public information (RTI International 2011).

DATA AND METHODS

We take as our starting point the reform package chosen for year 1 in each district. Grouping districts by program choice, we then assess the extent to which the four explanations of reform adoption align with patterns of Kinerja reforms, drawing on data collected under the project's purview and on Indonesian government statistics. We benefit from Kinerja's program design, which specified random selection of 20 districts in four provinces. The randomized selection, in some cases, “resulted in selected treatment districts with low political will, low capacity, major pending corruption cases, or resulted in the selection of [newly established] regions … that may hamper achievement of service delivery outcomes” (RTI International 2011, 4–5). While site selection presented challenges for program implementation, it bolsters our analysis through variation on the explanatory variables of interest, as well as on other characteristics.

We rely on a comparative case analysis of 20 randomly selected districts to identify which of the hypothesized variables relate to the choice of sectoral reforms. In Gerring's (2004, 343) typology of case studies, our analysis can be characterized as Type II, as it breaks the primary unit (Indonesia) into sub-units (districts) that are subjected to synchronic covariational analysis. Limiting the comparative analysis to a single country holds constant national policies, including decentralization, but the cross-district analysis allows for local variation in the factors of interest. Given that the data come from only four provinces, however, the findings are not representative of Indonesia as a whole.

As mentioned earlier, districts participating in Kinerja are given a choice between three reform packages. The dependent variable is each district's selection of priority sectoral reforms, whether education, health, or business licensing.⁹ For data on the explanatory factors, we rely on three studies¹⁰ commissioned by Kinerja and carried out during February–June 2011 (see the following), the same period during which districts were selecting reform packages (RTI International 2011, 6). Observations reflect either the same time period as the program choice (Stakeholder Analysis) or the time leading up to it (Local Budget Analysis and Index.) The observed values on the explanatory factors are therefore not a subsequent result of the districts’ program choice, reducing concerns about reverse causality. There are, of course, innumerable other factors that could affect program choice independently or in combination with those emphasized here. Unable to control for and/or gather information on all alternative explanations, we focus on those identified in the literature summarized earlier. In addition to political and institutional data associated with each variant, however, we also present underlying economic and social information (discussed in the beginning of the next section).

Data on the explanatory factors were drawn from the three studies listed below. No consistent measures of political competition were available in these data sets; we therefore rely on secondary information complemented by examples recorded in the Kinerja Stakeholder Analysis to consider the effects of democratic political competition. Throughout the article, we also draw on the Stakeholder Analysis for illustrative examples and interview respondents’ and focus group participants’ explanations of local power relations and reform choices. These examples are intended to complement the data presented in the tables and figure.

1. Kinerja Local Budget Analysis (LBA) (SekNas Fitra and The Asia Foundation 2012a)

   The LBA collects a series of district budget and planning documents to identify trends in local governments’ policy and planning choices in the health, education, and public works sectors. The methodology, which also assesses the efficiency and effectiveness of budget choices, was used in 42 districts in 2009 and 2010 before its 2011 application in the 20 Kinerja districts. We draw on secondary data collected for the LBA to present contextual data on the Kinerja districts and relative levels of sectoral expenditures.

2. Kinerja Local Budget Index (LBI) (SekNas Fitra and The Asia Foundation 2012b)

   The LBI (companion research to the LBA) gauges the extent to which principles of transparency, participation, accountability, and gender equality have been integrated into four stages of local budget processes (planning, discussion, implementation, and public accountability) in each of the 20 Kinerja districts. The methodology relies on scoring 101 items (44 for transparency, 15 for participation, 20 for accountability, 22 for gender equality) based on the availability, adequacy, and quality of key budgeting documents and activities in the health, education, and public works sectors. Each item is given equal weight in calculation of an overall index, as well as a separate index for each of the four domains of interest. We use the LBI indexes created for transparency, participation, and accountability to gauge the level of public sector modernization in each district (Figure 1), as well as the indicator for the number of budget planning meetings held. For Figure 1, note that we categorized degree of adoption relative to the distribution of index scores across the Kinerja districts, whereas the original LBI analysis assessed progress relative to the maximum possible score on each index.

   Figure 1 Degree of adoption of transparency, participation, and accountability reforms, by district and program choice 2011. Source: Adapted from Kinerja Local Budget Index (SekNas Fitra and The Asia Foundation, 2012b).

   

3. Kinerja Stakeholder Analysis (Faculty of Social and Political Sciences 2011; Kemitraan 2011)

   The Kinerja Stakeholder assessments were carried out in 2011 using a qualitative research methodology combining focus groups and interviews with key informants in 10 districts in Aceh and East Java. These data are used in the analysis below to compare different types of state and non-state actors' interest in and power to enact reforms in the relevant districts. Researchers from the political science department at Universitas Gadjah Mada developed the methodology. It gauges stakeholder power (defined as capacity to influence success/failure of programs through relevant policy authority, funding, and networks) and interest (based on stated policy priorities and past activities) in the education, health, and business licensing sectors. While the methodology rated power and interest for each of Kinerja's distinct health and education reforms (i.e., unit cost analysis, teacher redistribution, and school-based management in education, and safe delivery and exclusive breastfeeding in health) in some districts, combined ratings for each sector are presented here.

Our approach and broader relevance of the analysis are constrained by the available data. With 20 districts and four sets of possible explanations, we report on patterns observed in case comparisons of single variables, rather than attempting to assess possible effects involving multiple variables. The data related to state-led policy entrepreneurship and citizen participation are especially limited, as the Kinerja Stakeholder Analysis was only carried out in 10 districts. The analysis thus relies on descriptive comparisons to explore the relevance of the proposed explanations, as the small sample size precludes quantitative bi- or multivariate analysis.

PATTERNS OBSERVED IN KINERJA DISTRICTS

Before considering the range of explanations for reforms proposed in the literature, we present economic and demographic characteristics that could underlie program choices (Table 2). Grouped by reform choice, few consistent patterns are apparent in these data. Districts in Aceh are overrepresented among districts choosing healthcare reforms, while no districts in South Sulawesi chose the health package. Several of the districts choosing health reforms also had higher population growth rates than those choosing education; it is possible that Kinerja's maternal and child health interventions are particularly salient in such districts (but note that two high-population growth districts chose OSS.) On average, poorer districts (as measured by gross domestic product per capita and the national poverty index) chose education reforms, while the richest selected the business licensing package. However, the districts opting for healthcare reforms had the lowest average Human Development Index (HDI) score (67.94), but it was not dramatically different from the average for districts choosing education reforms (69.36). Both averages were also below the 2008 national average of 71.17 (Biro Pusat Statistik 2014). Further investigation is needed to understand whether these economic and human development variables simultaneously relate to program choice and our explanatory variables, but we are unable to undertake such analysis with the available data and limited number of districts. We therefore cannot rule out omitted variable bias. Acknowledging these limitations, we focus on identifying patterns in the variables related to posited political economy explanations, while allowing for the possibility of other viable causal explanation.

TABLE 2 Characteristics of Kinerja Treatment Districts, by Program Choice 2011

District	Province	Program Choice	Population 2011	Pop. Growth 2009–10 (%)	HDI 2008	GDP/Cap. 2008 (Rp)	Poverty Index
Kota Probolinggo	E. Java	education	216,967	(5.86)	73.29	16,624,884	0.47
Luwu	S. Sulawesi	education	332,863	1.43	72.96	8,316,081	0.94
Bulukumba	S. Sulawesi	education	394,757	0.00	69.87	6,941,865	0.92
Sekadau	W. Kalimantan	education	181,377	0.40	66.13	5,626,878	1.75
Bengkayang	W. Kalimantan	education	214,785	2.31	66.81	9,396,527	1.73
Jember	E. Java	education	2,346,498	0.80	63.71	8,277,226	1.04
Barru	S. Sulawesi	education	165,900	1.79	69.54	7,578,582	0.67
Simeulue	Aceh	education	80,279	(2.51)	68.6	5,079,591	1.61
Luwu Utara	S. Sulawesi	education	287,606	(10.68)	73.15	7,423,320	0.91
Aceh Tenggara	Aceh	education	178,852	1.03	70.99	6,711,370	1.21
Melawi	W. Kalimantan	education	179,586	4.80	67.91	4,714,035	2.01
AVERAGE FOR DISTRICTS CHOOSING EDUCATION			416,315	(0.59)	69.36	7,880,942	1.21
Sambas	W. Kalimantan	health	496,116	(0.07)	63.73	9,554,534	1.85
Bondowoso	E. Java	health	736,530	3.90	61.26	5,863,871	1.28
Bener Meriah	Aceh	health	121,870	6.47	69.77	10,939,707	1.06
Aceh Singkil	Aceh	health	102,213	(0.28)	68.12	5,785,169	1.12
Kota Singkawang	W. Kalimantan	health	186,306	4.84	68.02	11,568,122	0.97
Kota Banda Aceh	Aceh	health	224,209	5.64	76.74	16,956,137	0.34
AVERAGE FOR DISTRICTS CHOOSING HEALTH			311,207	3.42	67.94	10,111,257	1.10
Kota Makassar	S. Sulawesi	one-stop shop	1,339,374	5.31	77.92	20,793,760	0.32
Probolinggo	E. Java	one-stop shop	1,095,370	4.90	61.44	11,450,917	1.25
Tulungagung	E. Java	one-stop shop	989,821	(0.22)	72.45	13,124,066	0.40
AVERAGE FOR DISTRICTS CHOOSING LICENSING			1,141,522	3.33	70.60	15,122,914	0.66

Source: Adapted from background data for Kinerja Local Budget Analysis (SekNas Fitra and The Asia Foundation, 2012a).

Democratic Political Competition

Finding reliable information on electoral competition is challenging in many countries (Boulding and Brown 2012); consistent and current data for the Kinerja districts have also proved elusive. For efforts at measurement, see, for example, Tadjoeddin (2012) and Toha (2009). Neither of these studies provides data with appropriate time frames or sufficient variation for use in analysis of Kinerja districts. Case studies from across Indonesia offer conflicting accounts of the competitiveness of local elections (Mietzner and Aspinall 2010). While there is a fertile landscape of political parties—especially compared to the effectively single-party system of the New Order—and electoral reforms have created opportunities for independent candidates, educational requirements and substantial campaign costs favor entrenched elites (Buehler 2010).

Quantitative analyses by other researchers do not point to a prominent role for political competition as an explanation for Indonesian reforms. While the direct election of district officials has been shown to increase district expenditures overall, there is no discernible effect on responsiveness to objective local needs in the health and education sectors (Skoufias et al. 2011; see also Sjahrir et al. 2014).¹¹ These authors suggest that increases in expenditures are driven by pre-existing conditions, reinforcing inequalities in wealth and service provisions across districts. Also, higher expenditures are particularly likely before the next election and after the election of non-incumbents.

The basis of political competition may be changing in some regions. Palmer (2010) argues that much Indonesian voting behavior is motivated by short-term benefits provided by candidates. He distinguishes benefits motivating 2009 voters in Aceh from other provinces, however, in that they rewarded parties (Partai Aceh and Partai Demokrat) for providing collective benefits (autonomy and peace), rather than individual candidates supplying private benefits (patronage.) Whether this tendency carries over to subsequent elections is an open question, but it may motivate officials to shift toward providing public goods as a means of gaining reelection.

At least at this early stage of Indonesian democracy, the degree of electoral competition may matter less for reforms than the new opportunities for political alliances, bases of legitimacy, and constituent lobbying created by elections (Rosser et al. 2011). Anecdotal data from the Kinerja districts suggest how these political openings may affect reforms. In one poor, devout district in East Java, for example, kyai (religious leaders) are essential to mobilizing voters affiliated with mosques and religious schools. The current bupati (district head) has cultivated these ties, establishing a quarterly discussion forum for 300 kyai to provide input to district technical agencies. Due to the local religious community's greater interest in Kinerja's health package (compared to education or licensing), maternal and child health reforms were chosen in this district, in spite of persistent problems with uneven teacher distribution and the bupati's demonstrated interest in education reforms.

State actors’ bases of legitimacy have also changed with democratic elections. Alliances among elected executives, legislators, and heads of agencies from the same party are noted in several districts as underlying reforms. In such contexts, officials without an electoral base are at a substantial disadvantage, as noted in Jember district, where the bupati and his deputy were forced from office to face corruption charges shortly after their reelection. The head of the provincial audit agency was named as a replacement, but has been unable to enact desired reforms due to lack of political alliances with legislators (Dewan Perwakilan Rakyat Daerah or DPRD, People's Regional Representative Assembly) and heads of the planning and technical agencies (Faculty of Social and Political Sciences 2011). Prior to 2005, all bupatis based their authority on gubernatorial appointments but the advent of local democracy has undermined this source of political legitimacy.

In summary, democratic elections have affected reform choices in the Kinerja districts. Rather than the straightforward relationship between competitiveness and responsiveness posited in some parts of the literature (Variant A1, Table 1), however, elections have likely had less direct effects. New constituencies, alliances, and means of patronage resulting from electoral reforms produced varied, and possibly contradictory, effects across districts (Variant A2).

State-Led Policy Entrepreneurship

To proxy for policy entrepreneurship, Tables 3A–C show state actors’ interest and power to affect reforms in the education, health, and business licensing sectors for the 10 districts with available data.¹² In the following, we consider, in turn, patterns in levels of interest, power, and these variables’ relationships to reform choice.

In terms of these actors’ interest in the reforms introduced by Kinerja, executives show consistently high interest in education (Table 3A) and in establishing one-stop shops (Table 3C), with more varied attention paid to health (Table 3B). Technical agencies are generally quite interested in relevant sectoral reforms. In contrast, district legislators (members of the Dewan Perwakilan Rakyat Daerah or DPRD, People's Regional Representative Assembly) have widely varying levels of interest in education and OSS reforms, but almost all appear to be indifferent to healthcare reforms. Particularly in the education sector (Table 3A), legislators have different levels of interest in the reform options offered by Kinerja; for example, in Bener Meriah, the relevant DPRD commissioners show high interest in establishing standard costs, but low interest in school-based management and balancing teacher distributions.

TABLE 3A State Actors’ Interest in and Power to Enact Education Reforms, by District and Program Choice 2011

District	Program Choice	EXECUTIVE		TECH AGENCY		LEGISLATIVE
		Interest	Power	Interest	Power	Interest	Power
Kota Probolinggo	education	high	high	high	high	medium	high
Jember	education	high	medium	med/high	high	med/low	med/low
Simeulue	education	–	–	high	high	med/low	med/low
Aceh Tenggara	education	high	med/high	high	high	low	low
Bondowoso	health	high	high	medium	high	high	high
Bener Meriah	health	–	–	med/high	med/high	low/high	low/high
Aceh Singkil	health	–	–	high	high	med/high	med/high
Kota Banda Aceh	health	–	–	high	high	med/low	med/low
Probolinggo	one-stop shop	high	med/high	high	medium	low/high	med/high
Tulungagung	one-stop shop	–	–	–	–	–	–

TABLE 3B State Actors’ Interest in and Power to Enact Health Reforms, by District and Program Choice 2011

District	Program Choice	EXECUTIVE		TECH AGENCY		LEGISLATIVE
		Interest	Power	Interest	Power	Pnterest	Power
Kota Probolinggo	education	high	high	high	high	low	high
Jember	education	medium	medium	high	medium	medium	high
Simeulue	education	–	–	low	high	low	med
Aceh Tenggara	education	high	high	high	med	low	high
Bondowoso	health	medium	high	high	high	low	high
Bener Meriah	health	high	high	high	med	low	high
Aceh Singkil	health	low	high	med/high	med	low	high
Kota Banda Aceh	health	–	–	med/low	high	med	high
Probolinggo	one-stop shop	low	low	high	low	low	low
Tulungagung	one-stop shop	high	high	high	high	medium	medium

TABLE 3C State Actors’ Interest in and Power to Enact Licensing Reforms, by District and program Choice 2011

District	Program Choice	EXECUTIVE		TECH AGENCY		LEGISLATIVE
		Interest	Power	Interest	Power	Interest	Power
Kota Probolinggo	education	–	–	–	–	–	–
Jember	education	–	–	–	–	–	–
Simeulue	education	–	–	–	–	–	–
Aceh Tenggara	education	–	–	–	–	–	–
Bondowoso	health	–	–	–	–	–	–
Bener Meriah	health	–	–	–	–	–	–
Aceh Singkil	health	–	–	–	–	–	–
Kota Banda Aceh	health	–	–	–	–	–	–
Probolinggo	one-stop shop	high	medium	high	medium	medium	high
Tulungagung	one-stop shop	high	high	high	high	high	high

Note: – indicates no data available.

Source: Adapted from Kinerja Stakeholder Analysis (Faculty of Social and Political Sciences, 2011; Kemitraan, 2011).

For the executives in each district, power often, but not always, coincides with their level of interest in the sector. Among others, the Kota Probolinggo executive has high interest in health and has also demonstrated high influence over policy and programs in this sector, while executives in Jember and Probolinggo have medium and low interest-power combinations, respectively (Table 3B). In only four cases (Jember education (high interest-medium power, Table 3A), Bondowoso health (medium interest-high power, Table 3B), Aceh Singkil health (low interest-high power, Table 3B), and Probolinggo licensing (high interest-medium power, Table 3C) is there a complete disparity between executives’ levels of interest and power in a sector.

For technical agencies, however, there are sharp distinctions between sectors. The education technical agency (Dinas Pendidikan) stands out as powerful, even in districts where non-education reforms are chosen (Table 3A), and as having more power than the corresponding health agency (Dinas Kesehatan) in many districts (comparing technical agencies’ power across Tables 3A–B).

In terms of differences in reform choice, districts choosing education reforms appear to have both high interest and power from the education technical agency (Table 3A). While interest and power are generally high for this type of actor, three out of four districts with lower interest and/or power chose reforms in other sectors (Bondowoso, Bener Meriah, and Probolinggo). For example, in spite of high commitment to education from the executive and legislative branches, Bondowoso district chose health reforms. This outcome may be explained by the relatively low interest shown by Dinas Pendidikan in Kinerja reforms. A similar pattern is observed in Probolinggo, where the education agency is relatively weak, and in Bener Meriah, where it was rated medium/high on both power and interest.

Kabupaten Jember would appear to counter this pattern, as education reforms were chosen in spite of only medium/high interest from Dinas Pendidikan. The technical agency in this district is distinct from its counterparts, however, for being the most powerful actor in the sector due to the lower legitimacy of the appointed executive and political ties to the DPRD (as noted earlier). Also important to the program choice are recent cuts in district funding for education. The agency's prominent position, combined with a need to maintain programs in the face of shrinking budgets, may explain both the ability and motivation of Dinas Pendidikan to sway program choice in its favor (and away from a weaker health agency), despite relatively low interest in the specific reforms offered by Kinerja (Faculty of Social and Political Sciences 2011).

These findings argue for an important role for both the power and interest of technocrats in reform choice (Variant B1, Table 1). In contrast to Mexico's relatively weak institutional environment, where entrepreneurial elected mayors drove reforms and technocrats changed with elections (Grindle 2007a), Indonesia has a strong civil service tradition that has proven stubbornly resistant to change (Antlöv et al. 2010; Priyono et al. 2007).¹³ While the stability of the Indonesian civil service may prove advantageous for the sustainability of reforms once they are made, it has also slowed and complicated efforts to increase responsiveness to local needs. The World Bank has argued that Indonesians are “increasingly experiencing the ways in which existing civil service policies, rules and practices impede institutional reforms aimed at improving performance within all kinds of public functions from service delivery to policy making” (World Bank 2009a, 6). For example, in spite of 2003 bureaucratic reforms to shrink the civil service and more efficiently allocate salary budgets, unchanged hiring practices led to the number of civil servants increasing by almost 20% between 2006 and 2008 alone (World Bank 2009b; see also McBeth 2014).

As noted earlier, legislators show variable interest in Kinerja-specific reforms; their power also varies widely. The DPRD may be largely irrelevant to program choice. Districts with high DPRD interest in education tended to choose non-education packages (Table 3A; the four districts with high interest in education chose health or OSS), while those with lower interest from legislators selected education reforms (all districts with education reforms had low or medium DPRD interest in education). Compared to education, DPRD power is generally higher in the health sector (Table 3B; seven districts have high DPRD power in health), but interest is low or medium across all districts.

Finally, for licensing reforms, there is little information available for comparing the role of policy entrepreneurship in different districts (Table 3C). For the two districts with information, however, there is fairly consistent interest and power to enact licensing reform from all state actors.

In summary, the technical agency appears to be the most prominent policy entrepreneur pushing for reforms in the education sector. High interest from powerful bupatis also matters, but does not vary greatly (across districts for which information is available). Therefore, the entrepreneurship shown by the education agency seems to explain the choice of education reforms, given high executive interest in Kinerja's reforms.

In contrast, no such pattern emerges in the health sector (Table 3B). Here, there is more varied interest from the executive, a less powerful technical agency than in education, and generally low interest from the DPRD. In the Kinerja districts, some examples can be found of innovative Dinas Kesehatan, such as paying for malnutrition programs with corporate social responsibility funds from local firms (Kota Probolinggo) and close collaboration with a reproductive health NGO (Kota Banda Aceh) (Faculty of Social and Political Sciences 2011; Kemitraan 2011). In spite of innovative activities, however, state-led entrepreneurship does not provide a consistent explanation for districts choosing health reforms.

These patterns call attention to the role of national policy in facilitating local state entrepreneurship (Variant B3, Table 1). In general, education is a more popular program choice than health or licensing reforms in Kinerja districts, and Dinas Pendidikan is consistently more powerful than its health counterpart. Part of the explanation behind these outcomes may be the national government's attention to the sector. Since 2003, national legislation has required 20% of both national and district budgets to be allocated to education (Suryadarma 2012). While a substantial portion of these funds goes to wages for teachers and agency staff, large allocations to the education sector have likely strengthened Dinas Pendidikan's relative position and given programs in the sector additional legitimacy.¹⁴

While an interested education agency supported by national policy appears to push program choice towards reforms in this sector, the Kinerja program choices are not uncontroversial (particularly teacher redistribution programs) and may not be enacted as intended. In several districts in East Java, local elites have successfully lobbied elected officials (both DPRD and bupatis) to block efforts of education agencies to move well-connected teachers to less desirable rural posts (Faculty of Social and Political Sciences 2011). In these cases, the entrepreneurship of empowered technocrats is thwarted by the electoral leverage that organized elites hold over elected officials (Variant B2, Table 1).

Public Sector Modernization

Program choice does not appear to be motivated by poor performance (Variant C1, Table 1). In fact, districts tend to choose reforms that reinforce relatively good past performance, at least in education (Table 4). Of the 11 districts that selected education reforms, seven were performing relatively better in this sector than in the health sector. The same pattern is not seen in the health sector. Performance by districts selecting these reform packages was mixed. These patterns likely reflect districts’ performance management systems’ tendency to use input indicators for decision making (Mimba et al. 2013). Performance data collected on outcomes and outputs are passed on to higher levels of government, but are not used in routine reporting at the local level.

TABLE 4 Performance on Education and Health Indicators Relative to District Average, by District and Program Choice 2011

District Name	Program Choice	EDUCATION		HEALTH
		Net Enrollment Rate	Dropout Rate	% Excl. Breast-Feeding	% Assisted Delivery
Kota Probolinggo	education	+	+	–	+
Luwu	education	+	+	–	–
Bulukumba	education	+	–	–	–
Sekadau	education	–	–	+	–
Bengkayang	education	+	+	–	+
Jember	education	+	+	+	–
Barru	education	–	–	–	–
Simeulue	education	–	+	–	–
Luwu Utara	education	+	+	–	–
Aceh Tenggara	education	+	–	+	–
Melawi	education	+	–	+	–
Sambas	health	+	+	+	–
Bondowoso	health	–	–	–	–
Bener Meriah	health	+	–	–	+
Aceh Singkil	health	+	–	+	–
Kota Singkawang	health	–	–	–	+
Kota Banda Aceh	health	–	+	–	+
Kota Makassar	one-stop shop	–	–	+	+
Probolinggo	one-stop shop	–	–	–	–
Tulungagung	one-stop shop	+	+	+	+

Note: +indicates performance better than province average for Kinerja treatment districts.

− indicates performance worse than province average for Kinerja treatment districts.

Source: SUSENAS (Biro Pusat Statistik, 2011).

Past funding allocations seem to influence choices in a similar way as past performance (Table 5).¹⁵ While relative expenditures on education and health tend to co-vary (i.e., districts spending above/below the provincial average in one sector tend to do the same in the other sector), districts choosing education reforms appear to be reinforcing past investment patterns. Of 11 districts selecting education packages, seven are already spending more than the provincial per capita average on education. Three of them are also spending below the provincial per capita average on health. In only one district (Melawi) are expenditures on health higher than the provincial average, while lower on education. Instead of choosing reform packages to compensate for relatively low past investments, education reforms appear to go to areas that are already favoring this sector.

TABLE 5 Per Capita Expenditure by Education and Health Technical Agencies Relative to Province Average, by District and Program Choice 2011

District	Program Choice	Education	Health
Kota Probolinggo	education	+	+
Luwu	education	+	–
Bulukumba	education	–	–
Sekadau	education	+	–
Bengkayang	education	+	–
Jember	education	–	–
Barru	education	+	+
Simeulue	education	+	+
Luwu Utara	education	+	+
Aceh Tenggara	education	–	–
Melawi	education	–	+
Sambas	health	+	–
Bondowoso	health	–	–
Bener Meriah	health	+	+
Aceh Singkil	health	+	+
Kota Singkawang	health	+	+
Kota Banda Aceh	health	+	–
Kota Makassar	one-stop shop	–	–
Probolinggo	one-stop shop	–	–
Tulungagung	one-stop shop	+	–

Note: +indicates expenditure higher than province average for Kinerja treatment districts.

− indicates expenditure lower than province average for Kinerja treatment districts.

Source: Adapted from background data for Kinerja Local Budget Analysis (SekNas Fitra and The Asia Foundation, 2012a).

As for past performance, the pattern is less clear for health. Although fewer districts overall choose these reforms, there are some signs that those opting for safe delivery and exclusive breastfeeding programs have lower past investments in the health sector. None of the districts selecting health reforms are spending more than average on health, while spending less on education (all are either below average on health and above on education, or above/below in both sectors.)

For the three districts choosing licensing reforms, two had previously established OSS (Probolinggo and Tulungagung, both in 2008). In both districts, the OSS is noted as “not working optimally,” due to continuing problems of overlapping regulations and jurisdictions across agencies, resulting in substantial wait times and continued back-and-forth between agencies to obtain permits (Faculty of Social and Political Sciences 2011, 11 [Tulungagung]). State actors may thus be pushing for continued improvement in their already established OSS, seeking to build on prior policy choices rather than undertaking new reforms. Notably, continued investment in this sector may be prioritized over any concerns about poor performance in health and education. Two of the three OSS districts are performing below the provincial average in health and/or education (Table 4), and all three are spending below the provincial average in one or both sectors (Table 5.) Interviews confirm this interpretation. For example, the Probolinggo bupati explained that health and education have received a lot of attention in the past; he asserted that current priorities thus need to reflect a focus on the previously neglected economic sector (Faculty of Social and Political Sciences 2011) (in spite of consistently below-average performance and investment on all health and education indicators in Tables 4 and 5).

General governance reforms to encourage transparency, responsiveness, and accountability of government officials are also expected to improve service performance by empowering citizens with information on state decisions and performance, opportunities to make their needs heard, and means of correcting shortcomings (Variant C2, Table 1). Figure 1 reports on indices combining each district's past adoption of modernizing reforms (such as pro-reform regulations, meeting procedures, and information-sharing mechanisms) to enhance these general governance principles. For the Kinerja districts adopting the education program, there are no clear relationships across the three categories of modernization reforms, as these districts include both high performers on all three measures (such as Kota Probolinggo and Bulukumba), as well as others that consistently rank at the bottom (for example, Luwu and Jember). In contrast, all districts choosing health reforms have made high levels of progress in at least one of the governance categories.

Citizen Participation

On average, districts choosing health packages were holding more meetings to gather input to government planning before Kinerja reforms were chosen (Variant D1, Table 1); three of these four districts have held the highest reported number of planning meetings (Table 6). Across the Kinerja districts, there are also reports of executives establishing regular meetings for input and to encourage dialogue between state actors and citizens in the 10 districts for which data are available. In addition to the Bondowoso bupati's quarterly kyai meeting, both Kabupaten Probolinggo and Kota Probolinggo hold weekly “coffee mornings” at the district executive's office. The latter district also has formed a Healthy City Forum to improve waste and sanitation (Faculty of Social and Political Sciences 2011).

While creating space for interactions between officials and civil society, the existence of these meetings does not necessarily result in citizen influence over decisions. Greater openness at early stages of the planning process rarely extends to participation in decisions regarding budgets and legislation (Antlöv and Wetterberg 2013). Among the Kinerja districts, Jember has a School Committee Communication Forum that has been invited to preliminary meetings in the education sector planning process, but its members are not privy to substantive decisions. Education-sector civil society organizations (CSOs) in Kota Probolinggo voice similar complaints, as do medical professionals’ organizations in Bondowoso (Faculty of Social and Political Sciences 2011). In these cases, decisions remain in government hands and participation fails to empower citizen groups.

Some evidence suggests, however, that non-state actors in districts choosing health packages have both relatively higher interest in the sector and more power in driving reforms than their counterparts in other sectors (Variant D2, Table 1). Whereas CSOs generally hold little sway (see power columns across Tables 6A–C), health organizations in several of the districts selecting healthcare packages stand out as having consistently medium or high levels of authority in the sector prior to the choice of Kinerja reforms (Table 6). This research cannot assess whether stronger non-state actors have pushed for more participatory planning meetings, or whether the existence of the meetings strengthened CSOs. Whichever the case, stronger health advocates and more planning meetings are associated with choosing Kinerja's health reform packages.

TABLE 6A Number of Planning Meetings Held and Non-State Actors’ Interest in and Power to Enact Education Reforms, by District and Program Choice 2011

District	Program Choice	# Planning	EDUCATION CSOs		MEDIA		RELIGIOUS LEADERS
		Meetings	Interest	Power	Interest	Power	Interest	Power
Kota Probolinggo	education	1	high/low	low	–	–	–	–
Jember	education	0	high/low	medium/low	low	high/low	–	–
Simeulue	education	1	–	–	–	–	–	–
Aceh Tenggara	education	5	med/low	low	low	high	–	–
Bondowoso	health	8	medium	low	low/med	high/med	medium	high
Bener Meriah	health	8	low	low	–	–	–	–
Aceh Singkil	health	1	high	medium	–	–	–	–
Kota Banda Aceh	health	8	medium	low/high	med/high	med/low	–	–
Probolinggo	one-stop shop	1	low	low	–	–	high/low	high/low
Tulungagung	one-stop shop	1	–	–	–	–	–	–

TABLE 6B Number of Planning Meetings Held and Non-State Actors’ Interest in and Power to Enact Health Reforms, by District and Program Choice 2011

District	Program Choice	# Planning	HEALTH CSOs		MEDIA		RELIGIOUS LEADERS
		Meetings	Interest	Power	Interest	Power	Interest	Power
Kota Probolinggo	education	1	low	low	low	medium	–	–
Jember	education	0	low	low	medium	medium	–	–
Simeulue	education	1	–	–	low	low	–	–
Aceh Tenggara	education	5	high	med/low	high	med	low	high
Bondowoso	health	8	med/high	low/high	medium	medium	high	high
Bener Meriah	health	8	–	–	–	–	low	low
Aceh Singkil	health	1	high	med/low	high	high	–	–
Kota Banda Aceh	health	8	low/high	med/low	low	low	–	–
Probolinggo	one-stop shop	1	high	medium	low	medium	–	–
Tulungagung	one-stop shop	1	med/high	med/high	medium	medium	–	–

TABLE 6C Number of Planning Meetings Held and Non-State Actors’ Interest in and Power to Enact Licensing Reforms, by District and Program Choice 2011

District	Program Choice	# Planning	BUSINESS ORGS		MEDIA
		Meetings	Interest	Power	Interest	Power
Kota Probolinggo	education	1	–	–	–	–
Jember	education	0	–	–	–	–
Simeulue	education	1	–	–	–	–
Aceh Tenggara	education	5	–	–	–	–
Bondowoso	health	8	–	–	–	–
Bener Meriah	health	8	–	–	–	–
Aceh Singkil	health	1	–	–	–	–
Kota Banda Aceh	health	8	–	–	–	–
Probolinggo	one-stop shop	1	high	low	–	–
Tulungagung	one-stop shop	1	high	low	medium	low

Note: – indicates no data available

Sources: Kinerja Local Budget Index (SekNas Fitra and The Asia Foundation, 2012b) and Stakeholder Analysis 2011 (Faculty of Social and Political Sciences, 2011; Kemitraan, 2011).

In districts where the education agency shows less interest in reforms, or has less power to enforce them, there may be opportunities for non-state actors to push for other reforms. For example, in Bondowoso, CSOs, the media, and religious leaders all show interest and have some say in the health sector. The high number of planning meetings held in the district may complement the relatively strong position of non-state actors in pushing for health reforms (Faculty of Social and Political Sciences 2011).

As noted earlier, health program choice aligned with past underinvestment, in contrast to education and OSS reforms, which reinforced past performance and expenditures. Through increased interactions with officials during planning meetings and in other fora, interested and empowered citizens may be calling the state's attention to health services’ shortcomings and holding them accountable for improvements.

KEY FINDINGS

Acknowledging the limitations of our data, we seek, in this section, to offer an overview of the relevance of the four possible political economic explanations for reform adoption identified in the literature. Patterns across Kinerja districts suggest that factors associated with all four political economy explanations contribute to understanding which reforms are adopted (Table 1). Although we present only anecdotal data to support it, other studies have demonstrated empirically that the introduction of democratic political competition has laid the groundwork for new political alliances, patterns of patronage, and party provision of benefits that condition reform choices (Buehler 2010; Mietzner and Aspinall 2010; Palmer 2010; Skoufias et al. 2011). State-led policy entrepreneurship is evident from technocrats in the education agency. Public sector modernization may have increased orientation towards performance, as education reforms have been adopted by districts that were already achieving relatively good results in the sector. More openness towards citizen participation coincides with health reforms; districts that provide more opportunities for civil society to attend planning meetings tend to choose such reforms.

Interactions among several of the political economy elements suggest that it matters not just whether districts took advantage of opportunities offered by decentralization (Eckardt 2008), but also which aspects of decentralization dominate. Where local technocrats’ entrepreneurship has been strengthened by supportive policies and alliances with elected officials (e.g., education and OSS), reform choices and performance orientation differ from districts where modernization reforms have increased opportunities for citizen participation (e.g., health; see Tables 3456, Figure 1).

Patterns in the Kinerja districts also suggest the continued, but perhaps unintended, effects of national policy choices on decentralized reforms. While mandated budget allocations, such as those for education, are an important tool for communicating central priorities to the local level, national actors need to integrate financing, policies, and enforcement to create additional incentives to address local needs. In a context such as Indonesia, with a history of top-down policies and programs implemented by largely unaccountable and non-responsive technocrats, such countervailing policies are especially important. Decentralization reforms were, in part, designed to increase responsiveness, but, as the Kinerja analysis implies, national policy priorities may fortify technical actors who tend to use their strengthened position to reinforce past performance and expenditure rather than assessing local circumstances and changing course to address areas of past underinvestment and poor performance.

In sectors such as education, where technical agencies’ power to steer reform choice is buttressed by national budget choices, other studies have suggested that the strong leadership provided by the local agency may be beneficial once reforms are being implemented (Di Mascio and Natalini 2013, 163). However, the agency's dominance also runs the risk of marginalizing citizens’ influence over reform choice to sectors or issues where state actors show low interest. Rather than increased state responsiveness and collaboration with citizens that were envisioned with decentralization, such outcomes reflect continued state dominance of local priorities. In Jember, for example, one education CSO representative explained that his organization had shifted focus away from advocacy and towards developing its own model of service delivery. In his view, there was little to be gained from continuing to criticize government policy and budgeting in this district, where the education technical agency dominates local decision making (Faculty of Social and Political Sciences 2011). In the Kinerja districts, there are some indications that technocratic reforms can be stymied by elected leaders intervening on behalf of organized constituencies, but such (patronage-based) actions are far from the ideal of efficient state-society interactions envisioned in public sector modernization efforts.

We offer the following summary observations on the reform patterns in the districts studied. Pursuit of reforms is enhanced when (1) local officials are disposed to support reforms; and (2) when decentralized sectoral technical agencies have the power and motivation to adopt reforms. Bupatis and the leaders of sectoral technical agencies have significant leeway to decide which specific reforms to pursue in the absence of strong institutional constraints imposed by central level authorities. Grindle (2007a), citing Hirschman (1981), makes the distinction between policy problems and their solution that are “chosen” by officials and their agents versus those that are “pressing,” where decision makers perceive that they must take them on as a high priority. This distinction is helpful in interpreting the patterns across the Kinerja districts, where there did not seem to be a strong relationship between poor prior performance—which would suggest a pressing problem—and selection of a sectoral reform. The menu of reforms offered through Kinerja may not have extended to districts’ highest priorities. For example, in the project's mid-term evaluation, some officials pointed out that teacher distribution is important, but the most concerning issue is the availability of qualified teachers (Social Impact 2013). Most reforms, then, were chosen based on political economy factors other than an urgent need to address performance issues.

It could be expected that decentralization would provide citizens with a role in determining which problems are pressing, and motivating government to respond. However, in the Kinerja districts, there is little consistent evidence that citizen participation has motivated local officials or agencies to pursue reforms. While CSOs appear to be able to inject their views on needs and preferences into service delivery planning, their views are rarely reflected in budget or policy choices, particularly in the face of a dominant technical agency. Where there is more balance among agencies, there may be more space for both citizen participation and progress on governance reforms. CSO advocacy influenced the choice of Kinerja health reforms, such as in Bondowoso, where there were more planning meetings, the health agency had close ties to a local CSO, and the bupati was connected to NU, which favored MCH interventions over education reforms (Faculty of Social and Political Sciences 2011). Creating a culture of democratic accountability remains in its early stages in Indonesia, where patronage relationships between government and citizens continue to prevail, and ones based on citizens’ rights to demand good performance are emergent at best. The election of more district officials who owe their victories to a broad base of rank-and-file citizens rather than local elites (Rosser et al. 2011), combined with local performance management systems that collect and use outcome data (Mimba et al. 2013), could bolster civil society's power to push for needed services.

CONCLUSION

Like Grindle, we find the presence and actions of entrepreneurs located in district-level public bureaucracies to be the most important factor in explaining patterns of reforms. In contrast to entrepreneurial elected mayors in Mexico, however, reform choices in the Kinerja districts are most often driven by unelected technocrats empowered by national policy priorities, in spite of democratic decentralization intended to increase responsiveness to local needs. In districts where the generally dominant education agency cedes the opportunity for reforms, the effects of public sector modernization and citizen participation emerge.

This study thus sheds light on the political economy of decentralization. The objectives of decentralization in Indonesia, following the classic rationale, are twofold: to increase administrative efficiency and effectiveness, and to enhance accountability and responsiveness to citizens (Faguet 2004; Lockwood 2009). Our findings point to incompatibilities and tensions between these two objectives. Under conditions of partial decentralization—that is, where local governments do not have full autonomy regarding service delivery and revenue raising and spending—entrepreneurial local officials have incentives to exploit the available policy and operational space to pursue their interests under the aegis of centrally mandated efficiency and effectiveness measures. Such incentives, however, can accord a “back seat” to citizen accountability and responsiveness. Thus, achieving these twin decentralization objectives calls for multiple and varied levers for change to bring about improved performance.¹⁶

At a general level, the conception of the political economy of service delivery solely in terms of constraints and impediments to reform that need to be eliminated or skirted needs to shift. The focus needs to incorporate the realities of political economy in ways that seek to align reforms “with the grain,” as Booth (2011a) terms it, to identify local sources of commitment to, and incentives for, change and mobilize them to move forward. More specifically, in post-authoritarian, recently decentralized countries, national government must reconsider blanket national mandates. While tempting politically and administratively, rigid central prescriptions distort sub-national power structures and crowd out local priorities. Instead, national goals can be achieved through more nuanced mechanisms that focus on regions with more severe shortcomings, thereby aligning central and sub-national needs. For example, the mandatory budget allocation for education in Indonesia could be relaxed where education goals have been met and/or where health indicators require urgent attention. Such actions could free up both budgets and political space for other sectors, and for citizens to express their needs, edging closer to fulfilling the dual goals of decentralization.

ACKNOWLEDGEMENTS

The authors gratefully acknowledge access to data and documentation for the districts studied provided by the US Agency for International Development's Kinerja Program (Cooperative Agreement No. AID-497-A-10-00003). We also appreciate comments on earlier drafts from Elke Rapp, Jana Hertz, Hans Antlöv, Jennifer Brinkerhoff, Steven Kelman, seminar participants in Prague, Durham, and Jakarta, and two anonymous reviewers. The views and interpretations in this article are solely those of the authors and should not be attributed to USAID or any of the Kinerja partners.

Additional information

Notes on contributors

Anna Wetterberg

Anna Wetterberg ([email protected]) is a social science research analyst at RTI International (Research Triangle Institute). Dr. Wetterberg has more than 15 years’ experience working on research and operations in international development, particularly in Indonesia. Her current research interests include state-society relations, local governance, and frontline service delivery. Dr. Wetterberg earned her PhD in sociology from the University of California, Berkeley, and her MA in international development policy from Stanford University.

Derick W. Brinkerhoff

Derick W. Brinkerhoff ([email protected]) is a Distinguished Fellow in International Public Management with RTI International (Research Triangle Institute) and associate faculty member in the Trachtenberg School of Public Policy and Public Administration at George Washington University. His recent research focuses on citizen engagement, public–private partnerships, decentralization, and governance. He is coeditor of Public Administration and Development and a fellow of the National Academy of Public Administration.

Notes

Sahlin-Andersson (2001) discusses the emergence and spread of NPM as a reform template with common elements, citing the role of the Organization for Economic Co-operation and Development's Public Management Committee, known as PUMA, as a disseminator. The World Bank played an analogous role in developing countries (Brinkerhoff 2008). Manning (2001) offers a detailed treatment of the limitations of the NPM template in the developing world, noting, among other issues, weak institutional preconditions and resistance from political leaders.

There is an extensive literature on policy entrepreneurs. We define government-led policy entrepreneurs as bureaucrats or technocrats who seek to solve policy, organizational, and/or service delivery problems by introducing new ideas and practices (Roberts and King 1991; Mintrom 1997; Mintrom and Norman 2009).

In addition to the DAU, districts receive varying combinations of sectoral and special-purpose transfers from the central government.

Participation, as it is used here, goes beyond citizens’ roles as voters to include holding bureaucrats and elected leaders accountable for service quality between elections, as well as involvement in identifying and solving problems.

The national-level legislative assembly is the Dewan Perwakilan Rakyat (DPR), while the corresponding bodies at provincial and district levels are Dewan Perwakilan Rakyat Daerah (DPRD) I and II, respectively. In the rest of the article, discussion of the DPRD refers to district legislators.

We are not using “principal-agent relationships” as a reference to agency theory and its emphasis on contracts (Fama and Jensen 1983), but to point to the information asymmetries and accountability issues inherent in decentralized service provision. For an elaboration, see Chapter 3 of the 2004 World Development Report (World Bank 2004).

In 2012, the program's scope was expanded to include West Papua, but districts from that province are not considered here.

Indonesian districts include kabupaten (rural regencies) and kotamadya (urban municipalities). Kabupaten are led by a bupati (regent), while kotamadya executives are called walikota (mayors).

While the health and education packages were available to all districts, the choice of business licensing reforms was limited to districts that had not previously received similar technical assistance.

Neither of the authors was involved in these studies.

At the national level, Toha (2009) finds that electoral competitiveness of districts does not account for increases in central transfers for public goods.

Power is defined as capacity to influence success/failure of programs through relevant policy authority, funding, and networks; interest is gauged through stated policy priorities and past activities (Faculty of Social and Political Sciences 2011).

Although note the recent passage of Law 5/2014 on Civil Service Reform. See McBeth (2014) for a description of the legislative effort.

Sectoral transfers are also available from the Ministry of Health to fund operational costs for preventive care and health promotion, but are generally smaller than grants for education. For example, in 2011, health transfers were, on average, one-third the size of those for education in the Kinerja districts (SekNas Fitra and The Asia Foundation 2012, 9). There are also no nationally mandated minimum allocations for district health budgets.

Note that Table 5 shows overall health and education expenditures, rather than specific allocations relevant to Kinerja reforms (for example, antenatal care). If such expenditures were available, they might show different patterns than those observed in the general data.

For an analysis of Kinerja's efforts to improve accountability and responsiveness to citizens after selected reforms were underway, see Wetterberg, Hertz, and Brinkerhoff (2015).