The Patronage Patchwork Village Brokerage Networks and the Power of the State in an Indonesian Election

This article analyses the roles played by patronage and brokerage in an Indonesian district election by focusing on variation in village-level results. Through interviews with village elites, we found that highly uneven village voting patterns were strongly influenced by varying patterns of patronage distribution. The winning candidate (the son of the incumbent) used a state-centred form of patronage involving pork barrel and club goods that provedmore effective than gifts distributed through social, specifically religious, networks. Also critical were effective and trustworthy community-level brokers to deliver patronage, with retail vote-buying especially vulnerable to broker predation. Among the four categories of brokers we identified, state brokers—notably village heads—were especially effective at marshalling votes. Our findings underline the importance of patronage in Indonesian elections and the centrality of brokers in clientelistic systemsgenerally. They alsohelp explain thedominationof formerbureaucrats in electoral contests in regional Indonesia.

aspinall and as'ad Bijdragen tot de Taal-, Land-en Volkenkunde 171 (2015)  It has become a widely accepted observation, almost a truism, in the study of contemporary Indonesia that patronage-cash, goods, jobs, contracts, or other material benefits distributed in exchange for political support-is the glue that holds the political system together.Patronage, it is also widely agreed, is particularly important in local elections.But how do we move beyond such general observations to more precise judgements?For example, are some forms of patronage more effective than others, and under what circumstances?And how do we account for the significant variations in results that candidates using patronage strategies can obtain across geographic locations?This article proposes one novel approach to begin answering such questions by presenting the results of a meso-level study of village election results and patronage networks in South Kalimantan Province.Rather than focusing on the macro level of candidates and their strategies, or the micro level of individual voters, our focus is instead on the connective tissue that lies between the two.That connective tissue is to be found in the communities where voters live.
Accordingly, we sent a team of researchers to three dozen villages in the district of Tanah Laut (Map 1), not long after it had carried out a bupati (district head) election in 2013.Our team members interviewed village elites about what factors had influenced the election results in their communities.Although our questions were open-ended and we were interested in any strategies the candidates might have used, we particularly focused on their patronage and networking efforts.We tried, in other words, an approach that is simple yet surprisingly rarely used in research on patronage politics in Indonesia and more generally: we endeavoured to determine to what extent variation in election results was a consequence of varying patterns in patronage delivery.We focused on the intermediate level of the villages, in part precisely because it is relatively straightforward to identify variation in election results at this level.Indeed, the electoral map of Tanah Laut District resembled a patchwork (see maps 2 and 3) in which even neighbouring villages recorded very different results despite having similar socio-economic and cultural profiles.We were also interested in this meso level for a theoretical reason: our understanding of patronage delivery suggests that it is mid-level brokers who play a critical role in ensuring effective delivery of patronage to voters around elections, and in ensuring that recipients reciprocate with their votes.Put simply: no brokers, no patronage.
Although, as with any method, this one had its limitations, we arrived at findings that contribute significantly to existing literature on patronage politics in Indonesia.Four findings stand out.The first two concern patronage, which we define, following Shefter (1994:283 n. 3; see also Hutchcroft 2013) as 'a divisible benefit that politicians distribute to individual voters, campaign workers, Bijdragen tot de Taal-, Land-en Volkenkunde 171 (2015)  or contributors in exchange for political support' .First, our study confirms that patronage can indeed count a great deal in Indonesian elections, and not just in a general sense but also in producing varied voting results.We found evidence of recent patronage delivery by candidates in all but one of the villages that comprised our case studies, and by multiple candidates in 30 of the 36 villages.In the only village where there was no evidence of recent patronage delivery, the candidate who won there promised a future pork-barrel project should he be elected.Perhaps most dramatically of all, we found evidence of patronage delivery by the candidate who achieved the highest vote in the village concerned in 34 of the 36 villages, showing a high correlation between patronage delivery and electoral victory.These findings provide a degree of empirical evidence of patronage that is largely missing from the more anecdotal accounts that predominate in existing literature on Indonesian local politics.
Second, we found that not all forms of patronage counted in the same way or to the same degree.In particular, pork barrel projects and club goods delivered through the lowest level of the government apparatus had greater impact than the other methods we identified as being common-individual and club goods provided through social networks (notably religious organizations) and retail vote-buying (though for reasons we explain below we are less confident of our findings in regard to vote-buying).
Our next findings concern brokerage.The third finding is that while patronage mattered, so too did the brokers by whom it was delivered.The patchwork of election results in the district not only mapped closely onto candidates' patronage delivery at the village level, but also onto their ability to recruit effective and reliable brokers there.Vote-buying efforts were especially prone to failure if poor-quality brokers were recruited.Fourth, it was not only the reliability of individual voters that mattered, but their type, too.We identified four categories of brokers as having played a role in this election (party, state, community, and market), but found that it was the state brokers whose access to patronage was connected to their occupancy of low-level positions in the state structure-notably village heads-who were most prominent in determining the election's outcomes.Taken together with our finding about the efficacy of state-centred patronage, our findings point toward the techniques of power maintenance that bureaucratic actors use to entrench themselves in local politics in Indonesia.
We present these findings and arguments in several sections.First, we locate our study in the context of existing literature on local electoral politics in Indonesia and on brokerage as it pertains to the study of patronage.Second, we provide a background to the election in Tanah Laut.In a third section, we begin to describe our empirical findings by outlining the state-centred club Bijdragen tot de Taal-, Land-en Volkenkunde 171 (2015)  goods and pork barrel projects that were decisive in this election.Fourth, we investigate a category of patronage that has been much remarked upon in the Indonesia literature and that was common in Tanah Laut: collective and individual gifts distributed through social networks, specifically grass-roots religious organizations.Fifth, we evaluate the role of retail vote-buying in this election, noting how patchy this method was and how vulnerable it was to broker betrayal.Sixth, we investigate the role of brokers in the election, emphasizing their centrality in determining smooth patronage delivery and thus influencing electoral outcomes.In a final section, we return to our major findings and highlight implications of the study for future research.

Locating the Research Empirically and Theoretically
In the wake of the post-Suharto twin transitions of democratization and decentralization, over the last decade and a half literature on local politics in Indonesia has greatly expanded (for book-length overviews, see Aspinall and Fealy 2003;Choi 2011;Erb and Sulistiyanto 2009;Hadiz 2010;Schulte Nordholt and Van Klinken 2007).Early on, a major concern was identifying the sociopolitical backgrounds of the politicians who were winning competitions for local office, especially as governors of provinces, bupati (heads of districts), and walikota (mayors of cities), at first via indirect elections in local parliaments and, from 2005, through direct elections of regional government heads, or pilkada.Scholars wanted to ascertain whether democratization was reconstituting local elites and allowing access to new entrants, especially those representing formerly marginalized groups.From early on, studies established that this was not generally the case, but often with rather broad brush strokes.Typical was Hadiz's (2004:711) conclusion that 'the decentralization process has been effectively captured by predatory interests' which had been 'nurtured under the Soeharto regime's formerly vast, centralized system of patronage' (see also Choi 2011:14-5).More recently, such findings have been amended by scholars pointing to growing pluralism in local elites (Choi 2014).Others have teased apart the catch-all category of 'elites' to identify which subgroups are particularly dominant in local politics.A critical finding here, relevant for our analysis of Tanah Laut, is that a high proportion of aspirants for political office in the regions have backgrounds in the state bureaucracy (Buehler 2010(Buehler :275, 2013;;Mietzner 2010:178).
Having established the elite backgrounds of local powerholders and aspirants, scholars have also turned their attention to the power resources that candidates mobilize, and their connections to circuits of financial and bureau-Bijdragen tot de Taal-, Land-en Volkenkunde 171 (2015) 165-195 cratic power.Although much of this literature uses the indigenous Indonesian term 'money politics' , scholars often mention patronage distribution strategies such as vote-buying and distribution of club goods, though usually in passing or anecdotally in local case studies focusing on other issues (for instance, Hidayat 2009:129-32;Hadiz 2010:120-33).Even so, some scholars have drawn on such studies to argue that patronage is a defining feature of local politics (for instance, Van Klinken 2009;Aspinall 2013).Another strong finding is that political parties often play a relatively marginal role in organizing local electoral contests (for instance, Buehler 2009;Mietzner 2010:177-8;Tomsa 2009) and that personal networks are instead often central.In one widely cited study from South Sulawesi, for example, Buehler and Tan (2007:58) pointed to 'efforts of power brokers-such as religious figures, large landholders, and bureaucrats-who could lobby in the subdistricts and who often distributed goods to the people' .In short, scholars have frequently identified patronage and brokerage as being critical to Indonesian local politics, even if they do not always explicitly connect their analyses to comparative literature on such topics.
Much of this literature has been produced by scholars favouring case study research, where the unit of analysis is a particular electoral contest and where the researcher uses elite interviews, media analysis, ethnographic methods, and similar techniques to map the power resources and strategies used by candidates (see for instance various chapters in Erb and Sulistiyanto 2009).Often scholars conduct more than one such study to identify common patterns (for instance, Choi 2011), or to identify different modes of campaigning (one excellent example is Tans 2012).While this research has produced rich findings, and identified important shared features of local politics (for instance, 'patronage is important'), it has been less effective at testing more specific propositions (for instance, 'under what conditions is patronage important?').This article presents one attempt-though we stress many other methods are possible-to move this research agenda forward.
In our research in Tanah Laut, we were fundamentally interested in examining the extent to which patronage swayed voter choices in the villages we studied.Additionally, we were motivated by two main lines of inquiry.First, are some forms of patronage more common-and more effective-than others, and under what conditions?Patronage, defined above as divisible benefits distributed in exchange for political support, comes in different forms.Let us mention three varieties that are common in electoral mobilization in Indonesia (Aspinall 2014a) and which, as we shall see, featured in Tanah Laut.Pork barrel projects (Stokes et al. 2013:12) are geographically targeted programmes designed to persuade the residents of a particular locale to support a party Bijdragen tot de Taal-, Land-en Volkenkunde 171 (2015)  or candidate, or to reward them for having done so (in Indonesia they often take the form of road, bridge, or similar construction projects).Club goods (Kitschelt and Wilkinson 2007:11) are also for collective benefit, but recipients are not defined geographically but by membership of a formal or informal organization (a prayer group, a sports club, a youth organization, et cetera).Vote-buying (Schaffer 2007) entails the gifting of goods or cash to individuals, in exchange for their votes.Each method has costs and benefits for politicians, most of which relate to the critical questions of reciprocity and contingency, which lie at the heart of clientelistic exchange (Hicken 2011:291-2; Kitschelt and Wilkinson 2007:10-2): politicians hope or expect that voters will repay them for the benefits they hand out, but it can be very difficult to enforce this exchange.For example, club goods can raise problems of free riding-a beneficiary group may receive a collective gift from a politician, but some or all of its members may not feel obliged to repay the politician with their votes if, for example, they receive individual cash payments from a rival.On the other hand, individual vote-buying is notoriously difficult to enforce if the ballot is truly secret (Schaffer and Schedler 2007).Different sorts of patronage will thus typically require different sorts of brokers to facilitate and monitor these exchanges.
This brings us to our third question: Are some brokers more important, and more effective, than others?Because patronage involves exchange, patronage politicians require intermediaries between them and voters in order to determine voters' preferences, deliver the benefits, and to enforce or encourage compliance with 'the deal' (Chattharakul 2010;Krishna 2007;Stokes et al. 2013;Wang and Kurzman 2007).Although the role of intermediaries has long been stressed in literature on clientelism, earlier works tended to stress the hierarchical nature of clientelistic relationships and, especially, the dependence of clients on patrons in the context of relatively stable agrarian societies (Scott 1972).Much of the new literature emphasizes the independent agency of brokers and their capacity for preferences that diverge from their patrons, for example when misdirecting or expropriating part of the cash or goods they are supposed to distribute to voters (Stokes et al. 2013:20;Kitschelt and Wilkinson 2007:8).Brokers may be critical to patronage politics, but they can also sabotage it.Aspinall (2014b) discusses this proposition with regard to Indonesia, highlighting the role of 'opportunist brokers' whose primary goal is to benefit materially from a political campaign.
In much of the comparative literature on clientelism, especially that drawing on Latin American cases, brokers are assumed to be party workers (see, for example, Stokes et al. 2013:19).However, as Berenschot (2015) has recently argued, drawing on a comparison of India and Indonesia, there is nothing Bijdragen tot de Taal-, Land-en Volkenkunde 171 (2015)  intrinsic to patronage distribution that requires parties to play the central role.Berenschot (2015) sketches out a typology of brokers, which we build upon and adapt here.Although this list is not exhaustive, based on previous Indonesian research we can speculate that at least four categories of brokers might be important.First, as in much of the world, are party brokers-though, as already noted, in Indonesia personal networks and informal 'success teams' are often more important.Second, success team members are often tokoh masyarakat (community leaders or notables)-such as religious or adat (customary) leaders-who exercise social influence by way of informal social ties and functions.We might think of such individuals as constituting community brokers.Much of the literature on Indonesian electoral politics emphasizes the role of such actors: Aspinall (2014b:551) refers to Indonesian success teams as 'social network machines'; Allen (2012:122) writes of 'the relative power of prominent local notables' in Indonesian politics.A third category are state brokers, such as bureaucrats or low-level elected officials; for reasons we shall explore below, political candidates throughout Indonesia are keen to draw such actors into their success teams (Berenschot 2015;Tans 2012).Finally, though this category does not exhaust the list, we might also identify market brokersindividuals such as landlords, plantation owners, construction contractors, distributors, or other entrepreneurs whose influence is derived from the economic dependence that tenants, employees, sub-contractors, and others may have on them.While candidates' wealth and other endowments will greatly affect their abilities to recruit brokers of any sort, the mix of brokers that occurs in any particular election campaign is likely to strongly reflect the patronage strategies adopted by candidates.
In sum, though clearly patronage counts in Indonesian electoral politics, the literature so far has largely been reliant on case study methods that are less well suited to testing under what circumstances patronage is most effective, what forms of patronage predominate, and how it might be affected by variants of brokerage.In the research summarized in this article, we experimented with a different approach, by lowering our focus of analysis from the macro level of candidates and their campaigns to the meso or village level.We studied just one election: the 2013 bupati race in Tanah Laut District (see Map 1).In that regard our approach accords with the case study analysis that has hitherto predominated.However, candidates and core campaign organizers were not our central focus, though we did interview them where possible.Instead, our unit of analysis was the villages, precisely because candidates recorded highly varied results at this level (maps 2 and 3).We wanted to explore the reasons for this patchwork of results and, specifically, to examine whether variation in patronage distribution and brokerage contributed to it.To our knowledge, no study has taken a single election and systematically analysed the role of patronage delivery in producing such varied outcomes.1So how did our research proceed?As already mentioned, we sent a team of researchers to interview community leaders in a sample of 36 of the 135 villages in the district.Our researchers interviewed a total of 342 informants.We chose a mixture of villages.Twenty were villages where three of the four bupati candidates had won with their largest percentages (ten villages where the winner achieved the highest percentage of the vote, eight won by the second-placed candidate, and two villages won by the third; the fourth candidate did not win the highest vote in any village and we disregarded him).In addition, 16 villages were selected at random from across the nine subdistricts in Tanah Laut.
In our interviews, we specifically targeted village elites, such as village heads and other village officials, party activists, religious leaders, midwives and other health workers, teachers, and other informal community leaders (tokoh masyarakat), precisely because our understanding is that such actors often act as vote brokers in the success teams that mobilize voters and deliver cash or goods to them at election times.Our researchers used semi-structured interviews to identify the key electoral actors in these villages, the strategies they used, and the forms of patronage delivered, as well as these village elites' perceptions of the reasons for the election outcome in their village.To ensure balanced information we made a strong effort to interview individuals who supported more than a single candidate (though this proved difficult in some villages where support was strongly concentrated on just one candidate).Overall, this research strategy proved effective and we were able to turn up a wealth of details about patronage, clientelistic exchange, and network-building during the election.
candidates.In a closely fought contest, the son, Bambang Alamsyah, won with a narrow margin of 60,573 votes, or 40.7% of the total, beating his nearest rival, Atmari, with 57,348 votes or 38.5%.As with many such elections of local government heads, however, we noted significant variation in community-level voting patterns.
Tanah Laut is in many ways a rather typical corner of an undistinguished Indonesian province.As with South Kalimantan Province as a whole, the population of the district (308,000 in 2012) is largely rural, with only 22% living in the one sizeable town, the district capital Pelaihari.Agricultural production is dominated by rice and corn, with a number of secondary crops.There is a sizeable plantation sector, with large, commercial oil palm estates, but also some smallholders growing oil palm and rubber.More unusually, there is also significant mining-mostly coal but also some iron ore and other mineralsconcentrated in the eastern subdistrict of Kintap.Large, private companies dominate, but there is significant small-scale artisanal (and largely illegal) mining.
aspinall and as'ad Bijdragen tot de Taal-, Land-en Volkenkunde 171 ( 2015)  Local politics in South Kalimantan is also typical of most Indonesian provinces: there is the usual fragmented political party map, and most elected local government heads are former bureaucrats (in early 2015, nine of the 13 bupati and mayors in the provinces were former civil servants).The province is distinguished, however, by its wealth of resources, being the third-largest area of coal production in the country.As a result, much political life centres on resource rent-seeking, with bureaucrats, politicians, security officials, and others seeking access to mining permits and the opportunities for official revenues and illicit payments these provide.Several local mining magnates exercise major influence on local politics in the province, both directly by leading several of the major political parties or heading district governments, or indirectly by sponsoring political candidates (Susanto 2007;Morishita 2011).With Tanah Laut being the third-largest coal-producing district in the province, with a little over 19 million tons of production in 2011 (bps 2012:231), and having significant unexploited reserves, the political economy of mining became an important part of the backdrop of the election discussed in this article.
The post of bupati was held between 2003 and 2013 by Adriansyah, a bureaucrat who first became prominent through a position in Bartamara, a local government company with mining interests based in the neighbouring district of Banjar.In this regard also Tanah Laut was not unusual, reflecting the continuing dominance by former bureaucrats in local elected office throughout Indonesia, though in Adriansyah's case his access to mining funds must have made him particularly well resourced.It is widely understood that through his position Adriansyah built up a war chest which he then used to buy support from the local council for the election in 2003.For much of his decade in office, Adriansyah operated as a typical patronage politician, with control of palm oil and mining licenses in his district being a major source of funds.Late in his term he was investigated, and subsequently charged, for a bribery case involving the issuing of a mining license to a local businessman who subsequently became mayor of Banjarmasin, the provincial capital.In April 2015, after the research upon which this article was based, he was arrested by the Komisi Pemberantasan Korupsi (Corruption Eradication Comission) in a hotel in Bali while allegedly in the midst of transacting a us$ 50,000 bribe concerning a mining operation in Tanah Laut.Yet Adriansyah was also locally regarded as an effective government leader.In particular, he supported a programme of agricultural development and assistance to farmers' groups, and prioritized rural infrastructure.For example, the length of asphalted district roads increased from 361km in 2003 to 641km over his two terms, while the proportion of district roads classed as being in a 'good' condition increased from 46 to 70% (bps 2004:211-2, 2013:251-2).He also developed a political Bijdragen tot de Taal-, Land-en Volkenkunde 171 (2015)  base in the Partai Demokrasi Indonesia-Perjuangan (pdi-p, Indonesian Democracy Party-Struggle), becoming head of the South Kalimantan board of the party.
When his second term (the maximum allowable) was drawing to a close, Adriansyah chose an option that is common in Indonesian local politics: dynastic succession (Buehler 2013).Adriansyah's son, Bambang Alamsyah, was young and lacking in career achievements before being elected to the district parliament (Dewan Perwakilan Rakyat Daerah, or dprd) in 2009.However, as the branch head of pdi-p, the party which achieved the highest vote in the district, he automatically became the speaker of the dprd.From this position he was able to build links with key elites in the district, especially at the village level, by officiating at the launch of local development projects (though his advantageous position also made him vulnerable to accusations of dynastybuilding, which informants agreed alienated some voters in the election, especially in the district capital).Bambang's running mate was Sukamta, a bureaucrat and faithful servant of Bambang's father as head of various local government agencies, most recently the Badan Kepegawaian Daerah (bkd, Regional Public Service Agency) and the Badan Pemberdayaan Masyarakat dan Pemerintah Desa (bpmd, Agency for the Empowerment of Village Communities and Governments).Both positions were strategic: through the former, Sukamta controlled allocation of bureaucratic positions; through the latter, he allocated rural-development assistance and regularly interacted with village heads and other rural elites.
Bambang's main rival in the election was his father's deputy, Atmari.A personal note was thus added to the rivalry in this election, though one which is very common in Indonesian elections, where local government heads often face off against their deputies in second-term elections.Atmari and Adriansyah had long worked closely together in the bureaucracy, and (according to informants close to the two men) during his first term, Adriansyah had promised Atmari that he would eventually support him as his successor.Atmari thus viewed Adriansyah's decision to back his son as a betrayal.As relations worsened between them, according to many sources, Atmari was excluded from government decisions and access to programmes.This may partly explain why he chose as his running mate a wealthy but uncharismatic figure, Muhammad Nur, the son of Haji Abidin, a mining magnate and head of the South Kalimantan branch of the Gerindra party.Other candidates who threw their hats in the ring were never considered serious contenders: Abdul Wahid (who attained 15 % of the vote) was another former bureaucrat and head of the district branch of pan (Partai Amanat Nasional), while Amperansyah (6 %) was an independent candidate.
The candidates were nominated and campaigns progressed, according to a pattern that will be familiar to students of local elections in Indonesia (for more detailed accounts of the way party coalitions are assembled, see Mietzner 2009Mietzner , 2010)).Bambang was nominated by pdi-p and the Golkar party, while Atmari was backed by a coalition of smaller parties.As is typical, the candidates had to pay for these nominations (according to an informant close to the Atmari camp, Bambang paid 17 billion rupiah (about us$ 1.7 million) to purchase the support of Golkar, while the parties who nominated Atmari were paid a total of around 40 billion, or us$4 million) (we were not able to verify these amounts with the candidates concerned).However, each candidate could not rely on these parties to campaign wholeheartedly for them, and instead constructed ad hoc 'success teams' (Aspinall 2014b).These teams in turn recruited large numbers of coordinators, intermediaries, and brokers in every subdistrict and, where possible, every village.As noted above, however, the results achieved by these teams varied greatly.Even neighbouring villages sometimes voted very differently.Moreover, these variations did not correlate with differences in ethnic composition, income levels, or livelihoods.Although the largest ethnic group in Tanah Laut is the local Banjar, there are also significant populations of Javanese and Madurese transmigrants, as well as Bugis along the coasts.We found that these ethnic differences had no appreciable impact on voting patterns.How, then, do we explain the variation?Below, we lay out the main findings of our village-level investigations.

Projects and Pork
While the winner was assisted by his father's record of promoting agricultural development, this history of achievement was strongly backed by-and entwined with-patronage.In 90% of the villages won by Bambang that we surveyed, he and his father had targeted delivery of development assistance in the weeks and months leading up to the election, using both formal government programmes and private donations.They distributed benefits, most commonly hand tractors, seedlings, and fertilizer to farmers' cooperatives (in the terms introduced above, club goods) and road repairs and other smallscale infrastructure to beneficiary villages (pork barrel projects).Bambang was also greatly advantaged by his father's position as incumbent and his ability to call upon village officials (notably village heads) to mobilize voters, a point we return to below.In sum, this was a state-centred campaign of a sort that has been described by Tans (2012:10) as characteristic of local 'mafias' which 'control local state institutions' .
In one sense, it is not surprising that what was in effect the campaign of an incumbent would stress local development achievements, given that the development of their region is, in effect, the major programmatic goal of any local leader in Indonesia.However, a large body of literature (especially from Latin America) points to how government 'programmes' are often subverted to meet clientelist purposes and handed out in part with an expectation of reciprocity: schemes are devised to ensure that supporters of the government leader or ruling party benefit disproportionately and/or with an implicit (or sometimes explicit) expectation that programme beneficiaries will reward the governing party or coalition with their votes (see, for example, Diaz-Cayeros, Estévez, and Magaloni 2012; Manor 2013).A similar blending of programmatic politics with clientelist delivery is common in Indonesia, and was pronounced in the Tanah Laut election.
The memory of Adriansyah's term as bupati, especially his agricultural achievements, strongly contributed to Bambang's victory.In a large number of the villages that our team visited informants attributed the overall outcome to his father's development achievements and to hopes that Bambang would continue those efforts.This was especially the case in the subdistricts of Kurau and Bumi Makmur, which are rice-growing areas located about 40 km from the district capital.As a result of concerted development programmes over the last decade, featuring improved immigration, technical assistance, and aid to farmers' groups (including provision of improved varieties, hand tractors, and subsidized fertilizer), in much of this rice belt farmers had increased their harvests from once to twice a year, greatly improving their income.
But Bambang (and his father) did not passively rely on this record to sell itself.During his campaign, Bambang claimed that he personally visited 96 out of the district's 135 villages (he claimed he lost in only one of the villages he visited, though our team found others).According to his own account, he would tailor what he said on these visits to prior surveys of village needs conducted by team members.Instead of making general promises, he would say 'very specific things in response to their needs […] here is the programme, here is the money' .2Our informants also told us that Bambang's father, while still bupati, often accompanied Bambang on these visits or invited his son to functions where agricultural assistance was being distributed or development projects being launched.These events were memorable for many local residents.In particular, Bambang had a reputation for getting his hands dirty by working in the fields alongside farmers on such occasions.Although these were presumably stage-managed events, they had a strong impact, contributing to Bambang's man-of-the-people image, despite his elite background.
Especially important were the targeted assistance packages that were disbursed on these visits.These came in two forms.The first was clubs goods in the form of gifting of agricultural equipment (typically, hand tractors) and seeds, Bijdragen tot de Taal-, Land-en Volkenkunde 171 (2015)  seedlings, and fertilizer to farmers' groups.These assistance programmes were especially important in Kurau and Bumi Makmur subdistricts, where they reinforced Adriansyah's previous record of agricultural development and where Bambang won with 43 and 46% of the vote respectively.The second form of largesse was pork barrel projects in the form of small-scale village infrastructure projects, typically the repair, asphalting, or widening of village roads, but also the building of facilities such as village halls, gates, fences, sports fields, and the like.
In nine out of the ten villages where Bambang received his highest vote (in each of these villages he received over 60% of the vote) we found evidence of strong mobilization of this sort in the months leading up to the poll, with Bambang and his team visiting these villages and distributing development assistance and projects, and local elites ascribing his success to these efforts (in the one that was an exception, there were other forms of patronage distribution-notably vote-buying and an influential local businessman strongly backing Bambang's campaign).In eight of the nine villages, too, we found evidence he was backed by the village head.Moreover, in nine out of the ten randomly selected villages we visited where Bambang won with a plurality of votes we also found evidence of such pork barrel and club goods mobilization by Bambang, though often with lower intensity.In contrast, we found evidence of this strategy in only five of the 16 villages we surveyed where Bambang did not win.In other words, we found no evidence of such mobilization in 11 of the 16 villages where Bambang lost, whereas it was absent in only two of the 20 villages where he won.Overall, this was our strongest finding: village development projects and targeted agricultural assistance were critical to achieving political success in this election.
By definition, this strategy is one that is open only to political candidates who have emerged through the local bureaucracy, and especially to the incumbents who control it, and its effectiveness is therefore likely to be one reason for the continuing hold that former state officials exercise on local politics, as observed by Buehler and others (it should be noted, too, that legislators are increasingly trying to be able to use this approach, via discretionary constituency development funds known as dana aspirasi or 'aspiration funds'; see Mahsun forthcoming).
According to informants, the distribution of these projects followed a typical pattern: Bambang would either personally visit a village or send a representative, hold a discussion with village leaders, ascertain their needs, and then invite them to send project proposals to the relevant district government agency.He would promise to help them in ensuring a swift and positive response (of course it was presumably his father, still serving as bupati, who Bijdragen tot de Taal-, Land-en Volkenkunde 171 (2015)  was able to ensure positive outcomes).To cite just one example, in the village of Raden, in Kurau, one informant explained that Bambang is very extraordinary, his leadership is very good, because in this village he has built a village hall, an official residence for the midwife, and has also put in street lighting.But he did this after we made proposals that were submitted to the Community Welfare Bureau.Pak Bambang helped us to get them through the process.
Here, the initial meeting took place well in advance of the election, several months before Bambang's candidacy was formally announced.In about half of the cases, the projects were accelerated and fully implemented prior to voting day; in about an equal number of cases, the project was either initiated with a promise it would be completed if Bambang won, or its delivery was made contingent upon his victory.
Critically, many village elites and voters appreciated this delivery of concrete improvements in their villages and contrasted it unfavourably with Atmari's efforts.One informant, in a village which Bambang won with a large majority, explained: 'It was Pak Bambang who promised he would build a road for the farmers so they could get to their dryland and rice paddies.All we heard from Pak Atmari's success team was his vision and mission [the standard Indonesian phrase for an election platform] about what his work will be like in the future and how he wanted to improve the welfare of the poor.'At least in the view of village elites, voters wanted tangible benefits.
One important factor that emerges from the above description is the negotiated nature of these interactions, pointing to the important role of agency on the part of village-level vote brokers.Indeed, in some places, elites were openly calculating in their approach.In the village of Kait-Kait Baru, village officials themselves took the initiative to send project proposals to all the candidates, but they found Bambang's response to be the most satisfactory: We sent in a proposal to Bapak Bambang [for assistance in providing materials with which] the community here could build a fence around the cemetery and for cement to repair our mushollah.Everything we sent to him in the form of proposals was fulfilled by Bapak Bambang.That's why he won in our village.We are materialistic.Whoever gives the most, that's who we vote for.This tactic mostly involved politicization of the district's development procedures, with projects funded out of the government budget.On a few occasions, Bijdragen tot de Taal-, Land-en Volkenkunde 171 (2015) 165-195 however, our village informants told us the projects were personal gifts from Bambang and his father, paid out of private funds.While this is not impossible (Bambang made many private gifts of other sorts, as we see below), it seems equally likely that voters were misapprehending the source of the benefits bestowed upon them (in other words, local government projects were sold to them as gifts from Bambang and his father).This was itself an important finding: many of our village elite informants simply did not distinguish between private and public goods for their villages.

Religious Gifts
As already noted, many analyses of Indonesian politics stress the social-network aspect of local-level campaigning, including the prominent role played by local religious authorities in connecting candidates to voters.This element was also all but ubiquitous in Tanah Laut, though it was trumped by the statefocused mobilization discussed in the preceding section.
A strategy that was used by all candidates, but most assiduously by the second-placed Atmari, was distribution of club goods to religious institutions and groups.These gifts-which we found evidence of in 26 out of our 36 sample villages-came in three forms.The first was donations of sums of money (the largest we encountered was 25 million rupiah, or about us$ 2,500) to mosques, prayer halls (langgar), or religious boarding schools (pesantren) in the villages, in order to renovate, expand, or improve these facilities.Sometimes cash gifts were replaced or supplemented by in-kind support, such as sacks of cement.The second form was the distribution of small gifts, typically headscarves, sarung, prayer mats, and the like (sometimes combined with small cash gifts) to congregations at religious meetings attended by candidates.Typically, these gifts would be distributed after the candidate attended a prayer meeting, often taking the form of a sholat hajat (a prayer requesting the fulfilment of a wish or goal, this time electoral victory).A third and final form of religious gift was the distribution of sacrificial cattle (sapi qurban) to villages at the time of the annual Idul Adha (Feast of the Sacrifice) festival, an occasion when wealthy members of the community donate goats, sheep, or cattle to be slaughtered, cooked, and shared out among the needy and other community members.The festival had occurred in late October 2012, about six months prior to the election, exactly the time when candidates were starting to promote themselves, prompting an outburst of charitable giving.
The campaign by Atmari was particularly marked by such religious gifting.In 23 of the 36 villages in our sample we found evidence of Atmari providing aspinall and as'ad Bijdragen tot de Taal-, Land-en Volkenkunde 171 ( 2015)  donations to mosques, prayer halls, and religious schools, and/or distributing items to members of devotional groups.At the same time, in the vast majority of villages he visited, he attended a prayer meeting run by a village-level religious group (for more on these groups, see below).According to one of his supporters he also gave sapi qurban to 'hundreds' of prayer halls throughout the district.3However, this campaign was not particularly effective: Atmari won the vote in only ten of the 23 villages where we identified such gift-giving, and lost in the other 13.Those ten villages were certainly a large proportion of the 14 villages in our sample that were won by Atmari; but in all but a couple of these villages our informants ascribed his victory to other factors.His religious gifting in these places seemed to reinforce other bonds that he had established, and which were in themselves the primary explanation for his victory.
Why did Atmari choose this approach, and why was it ineffective?It was not a case of a distinction between a religious candidate and a more secular one.According to one of his associates, Atmari did not have a reputation as being particularly devout prior to the election, instead he 'only started to host majelis taklim [religious study groups] at his house as he was thinking of nominating himself' (interview, 27 January 2014).Rather, the key difference was in political resources.Atmari's rival, Bambang, had access to the most powerful networks in the district.Not only did Bambang's father, Adriansyah, control the bureaucracy and was able to mobilize village officials to support his son (see below), the two strongest political parties in the district, pdi-p and Golkar, also supported him.In these circumstances, Atmari had few alternative networks available to him.The political parties which supported his candidacy did so because they were paid, as noted above, and lacked strong village branches.Only a few village chiefs supported him.
In these circumstances, Atmari turned to a type of village-level, informal religious institution that proliferates in South Kalimantan, as in many parts of Indonesia.Usually called kelompok pengajian or majelis taklim, these are groups consisting of the more devout people in the village, who meet regularly-typically on Friday evenings-to recite Quranic verses, listen to religious sermons, or otherwise engage in devotional activity and learning (for one recent account of majelis taklim, see Winn 2012).Most such gatherings bring together a few dozen people, typically drawn from older residents.Sometimes a group will invite a religious expert from a nearby village, or further afield, to lead the prayers and give a sermon.The groups are often run by an individual from the village who develops a district-wide or, rarely, province-wide reputa-Bijdragen tot de Taal-, Land-en Volkenkunde 171 ( 2015)  tion for religious knowledge, preaching, and sometimes other mystical abilities, such as healing, and who receives invitations to preach elsewhere.Such groups have multiplied throughout Indonesia over the last decade or so, and are, as one of our informants put it, often treated as an 'entry door' for political candidates seeking access to village networks.This seems to have been Atmari's approach, in the absence of viable alternative strategies.One of the members of his campaign team explained it this way: It was a political strategy.What he did was come to the majelis taklim in the village and carry out sosialisasi [literally: socialization, that is, promote his candidacy].That's more effective than just giving assistance.The idea is that from that pengajian they would spread his ideas around among the other villagers and pass on his appeal.That's the idea: you visit the pengajian group and from there it spreads.4 The problem for Atmari, however, was twofold.In the first place, the informal religious networks were just as open to other candidates as they were to him.Accordingly, we found evidence of Bambang making donations to religious institutions and networks in 13 of the 36 villages, including many that were also supported in this way by Atmari.We interviewed several influential local kyai (religious scholars) and other religious leaders, who were approached by several candidates for their blessings, and in several villages local pengajian groups, mosques, or langgar received donations from two or more candidates.Several villages likewise received donations of more than one sapi qurban.In many places, this competitive religious gifting simply cancelled itself out.No doubt, gifting helped provide candidates with reputations as pious benefactors of the Islamic community, but it rarely provided a winning edge.
A second problem was that although Atmari often found an appreciative audience-and votes-among these prayer groups, their members were a minority in each village.Even if their close relatives voted for Atmari, this was often insufficient to neutralize other benefits, such as village projects, provided by Bambang and his father.Atmari, it appears, chose this strategy because he lacked more attractive alternatives; not surprisingly, it failed to deliver strongly for him.
This finding, though based on a single study, potentially has broader implications for the study of Indonesian politics, where religious networks play a critical role in electoral politics but have an influence that habitually falls short aspinall and as'ad Bijdragen tot de Taal-, Land-en Volkenkunde 171 (2015)  of expectations (evident, for example, in the continuing inability of Islamic parties to gain a large share of the national vote).The fact that state-centred patronage so easily trumped distribution through religious networks in this election suggests that the latter strategy might often be a 'weapon of the weak' , a patronage option pursued primarily by politicians who lack access to more effective, state-based networks.We lack space to explore this proposition fully, but this might even be one possible explanation for the continuing difficulties faced by Islamic parties in Indonesian elections.

Vote-Buying
Our researchers also found evidence of vote-buying-the provision of cash or other gifts (in addition to gifts provided in a religious context, discussed already) to individual voters-in 22 of the 36 villages.Although these findings exceeded our expectations, we still need to treat them cautiously given the notorious difficulties of finding accurate data on this topic.National survey data show relatively high social acceptability of vote-buying, but the practice is illegal, and accusations of vote-buying by Bambang's camp had been a focus of a Mahkamah Konstitusi (Constitutional Court) challenge by Atmari a couple of months before our field research.Some of the village elites we interviewed were likely themselves involved in vote-buying efforts and thus had an incentive to downplay its significance.It is thus possible that some of the 14 villages where we found no evidence of vote-buying were false negatives, and that our informants also understated the extent and impact of the practice when we did find it.Indeed, in the few cases where our initial research failed to turn up a strong explanation for an outcome, we revisited the village in question, interviewed different informants, and found evidence of vote-buying.
With this caveat in mind, our findings are nevertheless revealing in three critical respects.The first concerns the patchiness of the vote-buying effort.Although, as explained above, both leading candidates were wealthy, there was no clear regional pattern.In exactly half (11) of the villages where we found vote-buying, it was conducted by just one candidate; elsewhere, two or more participated.We found evidence of vote-buying by candidates in villages where that candidate was already strong (for instance, because of family ties), in villages where no candidate had a strong claim, and in villages that were strongholds of rivals.Moreover, in most villages where vote-buying occurred only part of the population received payments.Amounts distributed to villagers also varied (informants spoke of sums of between 25,000, 50,000 and 100,000 rupiah, roughly us$2.50 to us$10).
Bijdragen tot de Taal-, Land-en Volkenkunde 171 (2015)  Second, this patchiness was apparently related to difficulties candidates faced in choosing trustworthy and reliable brokers.As already alluded to, this problem is recognized in the comparative literature: in order to effectively buy the vote, candidates need to choose intermediaries with local networks and influence who can back the delivery of cash with intimate knowledge of the recipients; such people, however, often have interests that diverge from the candidate, including in keeping a proportion of the funds for themselves.In Tanah Laut, the candidates called on various persons to fulfil these brokerage roles during vote-buying: village officials, religious leaders, health and education workers, village youths, party activists, and so on.But, again, Bambang had a structural advantage.His father's incumbency meant his vote-buying effort could be built around the category we have called 'state brokers' , specifically village heads, who are typically the most influential figures in the villages.The Constitutional Court case prepared by Atmari's team provides insights into how Bambang's campaign was organized in this regard (see below).Atmari's efforts were more ad hoc, and in at least several villages he lacked access to influential local figures.Thus, some village informants reported instances of outsiders turning up and handing out cash and gifts along with requests to support Atmari; all agreed that such efforts were ineffective because recipients did not feel at all obliged to carry their promises through at the ballot box.
In all but two of the 11 villages we identified where only one candidate engaged in vote-buying, it was that candidate who won.In a handful of villages, informants readily told us that voters were cash-sensitive and voted for whoever paid them.Usually, however, cash gifts were most effective when backed by other factors that gave the candidate an edge in that village, such as support of a local notable or a major project or donation.
A third finding, perhaps most unexpected of all, was that vote-buying itself was also-in about half a dozen villages-counterproductive.This was not, however, because voters were hostile to the practice.Rather, they reacted against it when they believed it was being carried out unfairly.In several villages, brokers evidently did not distribute all or part of the money they were expected to pass on to the voters.We have direct evidence of this in only one village, where a village head cheerfully admitted that he spent the 15 million rupiah given to him by Bambang's campaign for voters on other purposes: 10 million to repair a village road, and 5 million for a drinking party for village youths.In other places, informants told us that they had heard that money had been 'passed down' to a village success team member but was then not distributed onward, or only distributed to people who were related to, or lived close to, the person concerned.In some places, team members were Bijdragen tot de Taal-, Land-en Volkenkunde 171 (2015)  seen as profiting handsomely from such embezzlement-in one village they bought new motorcycles.
Intriguingly, in most of these cases informants told us that such manipulation made the village swing against the candidate concerned, as a form of protest at the greed of his brokers.This occurred in the village where Bambang recorded his highest vote; according to one informant, Bambang's team distributed the relatively small sum of 25,000 rupiah to most voters there: 'Although it was small, what's important is that everybody got the same.This was better than it being big, but all we heard was the distant thunder.'This informant was referring to the fact that a village elder was believed to have received a large sum from Atmari, but did not distribute it.In such places, informants blamed candidates for choosing the 'wrong people' for their teams or choosing people who were not 'efficient' , 'effective' , or 'trustworthy' .Such reactions point to a wider moral economy surrounding vote-buying, in which distributing cash or other gifts is not only seen as a positive attribute of a generous leader, but where fairness in delivery is also regarded as a mark of that leader's integrity and competence.It is not only that candidates need brokers to make patronage strategies work, but also that poor-quality brokers could make those strategies backfire badly.

Vote Brokers
The candidates' ability to connect to local networks and influential brokers was, accordingly, an important factor in explaining many village results.In about a quarter of our villages, informants pointed to either a family tie with the candidate or a single prominent local figure as being critical in swaying the vote.More broadly, the brokerage pattern matched closely onto the patterns of patronage distribution we have already identified.
State brokers were especially associated-not surprisingly-with the statecentred strategy employed by Bambang.He was much advantaged in his networking by his ability to mobilize civil servants and, especially, village heads.It is not unusual in local-government head elections in Indonesia for incumbents to turn the bureaucracy into an electoral machine.As Tans (2012:24) notes, in such cases '[t]he primary tool for manipulating the bureaucracy [is] the executive's right to reassign civil servants' .Throughout Indonesia, it is common for incumbent district heads to transfer civil servants to postings in remote locations, or in less well-funded institutions, if they do not support the incumbent's re-election effort.We heard many anecdotes of this technique being used in Tanah Laut, and Bambang's ease in delivering projects to villages in Bijdragen tot de Taal-, Land-en Volkenkunde 171 (2015)  the months leading up to the election demonstrates that he was able to put his father's continuing hold over the bureaucracy to good electoral use.Especially significant in voter mobilization was Bambang's-or more to the point, his father's-ability to mobilize village heads.Village heads in Indonesia are elected, and are therefore less vulnerable to being dismissed or reassigned by a district head than civil servants.However, they are often connected by informal patron-client ties to the incumbent, and their ability to access village development projects and assistance programmes depends on their ability to foster good relations with him or her.5 Their ability to represent the village and speak on its behalf in negotiations with candidates over projects that will benefit the community as a whole, as the discussion above indicates, also makes them especially influential as vote brokers.Part of their effectiveness, meanwhile, comes from the discretionary power they have when it comes to determining which village residents get access to government programmes: for example, they issue letters that determine whether poorer residents are eligible to various cash transfer and subsidy schemes (Berenschot 2015;Mulyadi 2013).As a result, villagers are especially liable to listen to their directives.
Accordingly, village heads were critical to Bambang's success in some villages.As already noted, they backed Bambang's candidacy in most of the villages where he ran a successful state-based patronage campaign emphasizing pork barrel projects and agricultural development.But village heads were not only involved in brokering deals about club goods and pork barrel projects, they also played a critical role in vote-buying.The documentation of Atmari's unsuccessful Constitutional Court challenge contains much relevant detail, including testimony from one village head that he took part in a meeting with ten others in the official residence of Adriansyah, who gave each of them 12.5 million rupiah (around us$1200) and told them to distribute it to residents and direct them to vote for his son (Mahkamah Konstitusi 2013:26).
Village heads were not the only network resource candidates could draw upon.Variations in village-election results in Indonesia often result from candidates' local ties (Clark and Palmer 2008:37); accordingly, in a number of villages (six out of our sample of 36) a candidate did well because of family connections.The third-placed candidate, Abdul Wahid, won only in two villages: Kuringkit, near where he grew up and where his brother had been village head, and Muara Asam-Asam, where many of his wife's family lived.
aspinall and as'ad Bijdragen tot de Taal-, Land-en Volkenkunde 171 (2015)  He came a close second in Tabanio village, where his wife was from (in each place he reinforced these ties with regular visits and donations).Bambang also had local relatives, several holding village government positions, and he won in all such villages we encountered.Atmari, born in East Java, lacked local family ties, but in some places those of his running mate compensated for this: he received one of his highest results, just over 60 %, in Sungai Pinang village, the birthplace of Haji Abidin, Muhammad Nur's father, and a place where Haji Abidin owned a large rubber plantation which employed many residents.
Party brokers were named as playing a significant role in village-level campaigns in about half (17) of our villages, which figure was higher than we had expected.But their role was generally secondary to village heads and other village elites, and they tended to be foot soldiers rather than key power brokers.One exception was in Takisung Subdistrict, one of only two won by Atmari, where Amtari was much assisted by a member of the district parliament, Abdi Rahman, a resident of the village of Ranggang (one of our randomly selected villages) who had a reputation for helping residents in need.Switching from a minor party to Muhammad Nur's Gerindra in the 2014 legislative election, he threw his influence behind the Atmari-Nur campaign.Although informants in two villages said that his endorsement was itself important, it was exercised alongside vote-buying: a series of sholat hajat were run in 12 villages in the subdistrict, to pray for Atmari's victory, and residents who attended each received 50,000 rupiah.
This example points to the importance of supra-village power brokers: individuals whose influence extended over more than one village, and sometimes over an entire subdistrict.Many such individuals were market brokers: local entrepreneurs whose political influence was directly connected to their economic power.For example, in the village of Sungai Bakar, Bajuin Subdistrict, where Atmari won over 60% of the vote, a wealthy local businesswoman was in charge of a scheme where locals who collected 100 photocopies of id cards of voters who pledged to vote for Atmari were paid 2,000,000 rupiah (around us$200).Bambang, too, could call on local connections.For example, in the coal-mining subdistrict of Kintap (the only subdistrict where he won an absolute majority of the vote, albeit narrowly with 51 %) he was greatly assisted by Haji Ibus, a wealthy local businessman, pdi-p politician, and long-time associate of Adriansyah.Described by informants in his home village of Kintap Kecil (one of our selection of ten villages where Bambang recorded his highest vote) as being in equal measure respected (in part for a history of generous donations to religious institutions and community activities) and feared (in part for running a network of toughs, whom he used to assert control in the rough-and-Bijdragen tot de Taal-, Land-en Volkenkunde 171 (2015)  tumble of the local mining industry), Haji Ibus was critical to the success of Bambang's campaign in the subdistrict.Although such figures turned up only a few times in our study, when they did, they generally turned out to be especially influential.
Community brokers, by contrast, appeared in virtually every village success team, but often failed to carry decisive weight on their own.In every village we studied, informants spoke of religious leaders, youth leaders, and the ubiquitous tokoh masyarakat-meaning here, roughly, village elders-as being the key actors in success teams.As already discussed, coordinators of majelis taklim and other religious leaders were targeted by all candidates, but especially by Atmari, as brokers for club goods directed at religious congregations, but they rarely succeeded in swinging the votes of an entire village in the desired direction.Even so, there were a few places where a religious leader commanded great respect: the village of Mekar Sari in Kintap Subdistrict went to Atmari because it was the home of an influential religious leader and paranormal, Haji Sutomo, who also headed the local Gerindra party chapter and who facilitated a major donation from the Atmari camp for the village mosque.In Alur village, Haji Mukri, a famous religious leader in charge of a pesantren, swung this village to a 67% vote for Atmari.
If local-leadership connections counted at the subdistrict and village level, they could also be influential at the sub-village level.Although not our focus, our field researchers often picked up stories indicating that voting patterns varied significantly within a single village, with very different outcomes in different polling stations (typically there were between two and four of these in a village).Sometimes the pattern was that residents who lived close to the village head voted for Bambang, but those who lived near a local religious leader leaned toward Atmari.But this was not a consistent pattern, with everything depending on the personal connections linking village notables and candidates.
An intriguing example was the village of Handil Babirik in Bumi Makmur, which was won narrowly by Atmari, unlike most villages in this subdistrict.Here, the village head was strongly in the Atmari camp: he received two vehicles from the Atmari campaign for use by the village government in the several months leading up to the election (they were returned after Atmari lost; the understanding was that he would have kept them had Atmari won), and he testified for Atmari at the Constitutional Court.Accordingly, Atmari carried two polling stations in the part of the village where the village head and most of his relatives lived; the other two polling stations went to Bambang, with residents being grateful for hand tractors and other gifts distributed in the run up to the election.Time and again, we found our informants mapping results at the sub-village level in this way: explaining varying polling-station results by pointing to an influential individual who lived in a particular neighbourhood, and/or by pointing to neighbourhood variations in experiences of vote-buying and giftgiving.Our focus on village rather than sub-village variation does not allow us to explore such neighbourhood clientelism in detail; it does, however, suggest it would be a potentially rewarding topic for future study.
Finally it should be noted, in keeping with our earlier observations, that local community leaders and brokers often had considerable autonomy in determining which candidates they wanted to support.In some cases, local leaders had long-standing clientelistic relations with one of the candidates.In others, our informants recalled how village heads and other community leaders had carefully weighed up which candidate offered the greatest advantages for the village and directed voters accordingly, a phenomenon that has been observed elsewhere in Indonesia (Clark and Palmer 2008:37-9) as well as in other parts of the developing world (for instance, Krishna 2007:148).We have already mentioned one village where officials sent project proposals to all candidates, and directed residents to vote for Bambang when he responded most generously.This was an extreme case, but in many of the places where Bambang and his father delivered or promised village improvements prior to the election, it is evident that village leaders consulted with residents, determined that they would benefit most if Bambang was elected, and made their preference known.Thus, in the village of Tirta Jaya one village official explained that '[i]t was the village apparatus which itself took the initiative to choose a pair [Bambang and his running mate Sukamta] and [then we] directed and coordinated the residents [so they would vote accordingly]' .They did, so afterwards Sukamta visited the village and acceded to the elders' request to provide heavy machinery to enable them to enlarge the village cemetery.We might think of such instances as representing a form of consensual voting, mediated by state brokers.Not surprisingly, this pattern will tend to confer a strong incumbency and bureaucracy advantage in local electoral races.

Conclusion: A Patchwork Politics
The study presented here is far from being the final word on patronage politics in Indonesia.It has clear limits, most obviously those that derive from it being based on a single district.Nevertheless, we hope we have demonstrated that meso-level studies of village-level results and networks are one way of furthering the study of patronage politics, not least because they focus at the interme-Bijdragen tot de Taal-, Land-en Volkenkunde 171 (2015)  diate level in patronage chains where the success or failure of patronage politics is typically determined.They also allow us to get beyond broad generalizations about the power of patronage to more precise evaluations about what varieties of patronage are most effective and under what conditions.
Our approach confirmed that the 2013 district head election in Tanah Laut, like so many in Indonesia, was a poll in which patronage was king.As we have seen, the patchwork of varied village-level results obtained by the leading candidates can largely be explained by variations in the patronage benefitswhether collective or individual-they provided in those locations.More specifically, our study higlights the critical role of brokerage.In rural Indonesia, patchy electoral outcomes of the sort we have analysed in this article are commonplace.This is because electoral competitions in such circumstances are not merely straightforward contests of financial power in which the wealthiest candidates win, but contests in which candidates require effective networks of brokers to deliver these benefits and ensure that the beneficiaries reciprocate with their votes.Gifts provided by candidates to villagers might not be viewed as binding unless they are backed up by the words of a locally authoritative community leader.As Clark and Palmer (2008:39) observe, 'the voiced opinions of the high status members of society' can carry influence in Indonesian elections, and many voters are 'strongly guided by the opinions of neighbors and especially the guidance of village elders' .
But if brokers are critical to electoral success, this does not mean that all brokers are equally effective.One of our main findings concerns how difficult it can be for candidates to find reliable intermediaries.In many villages, candidates lacked agents they could trust to deliver cash or gifts.Voter perceptions that brokers were hoarding such material rather than distributing it backfired on some candidates, accounting for poor support in some villages.We are left with an inescapable image of how unevenly electoral influence can be distributed, and how imperfectly a campaign strategy can be executed, even when the candidates are wealthy and powerful on the local scene.
However, we are far from suggesting that power resource views of Indonesian local politics are redundant.As it turned out, overshadowing all the subtleties of this election campaign were the brute facts of incumbency and bureaucratic mobilization.What gave the victor his winning edge were statecentred forms of patronage and assiduous networking with a group we have described (following Berenschot 2015) as state brokers, especially village heads.By utilizing the resources of the local government to provide targeted development projects and other benefits, and by relying on village heads to mediate and provide their blessings for such deals, Bambang and his father seized victory, in the face of accusations they were establishing a local dynasty.Other aspinall and as'ad Bijdragen tot de Taal-, Land-en Volkenkunde 171 (2015)