Unity in diversity? The creation of new local governments in a decentralising Indonesia

Abstract

This paper evaluates the proliferation of local governments concurrent with Indo-nesia's 2001 decentralisation. Largely static for decades under the New Order, the number of municipalities (kota) and districts (kabupaten) has increased by half, from 292 before decentralisation to 434 in 2003. Most of the increase is off-Java. This represents a fundamental change in Indonesia's subnational administrative, political and fiscal landscape. We present a new dataset that elucidates the characteristics of these new kota and kabupaten, and review political, fiscal and economic incentives for creating new jurisdictions. We find that geographic dispersion, political and ethnic diversity, natural resource wealth and scope for bureaucratic rent seeking all influence the likelihood of regional splits. Since jurisdictional changes will affect local governance and service delivery, we stress the importance of providing an effective institutional framework for the creation – or future amalgamation – of local governments as part of a strategic approach to consolidating Indonesia's decentralisation.

Notes

World Bank Office, Jakarta ([email protected], [email protected], [email protected]). The views in this paper do not necessarily reflect those of the World Bank. Special thanks go to Sylvia Irawan and the Ministry of Home Affairs for clarifications regarding the creation of new regions. Kadjatmiko, Melina Nathan and Paul Smoke provided helpful comments. We would especially like to thank our two anonymous referees. The authors wish to acknowledge the World Bank Dutch Trust Fund for Strengthening Indonesia's Decentralisation for supporting the work of Fitrani. This paper is part of the Indonesian Decentralization Empirical Analysis (IDEA) project of the World Bank Office, Jakarta.

Significant instances of jurisdictional fragmentation include Kenya, Nigeria (whose districts almost doubled in number from 301 to 589 by the early 1990s), Uganda and South Africa (which has in turn seen some consolidation). As Crook (2001) points out, the effectiveness of decentralisation programs – for example, in poverty reduction – depends on the politics of central–local relations and the general regime context. The motivations for creation of new jurisdictions can be quite telling in this regard.

The term ‘Big Bang’ refers to Indonesia's implementation of far-reaching fiscal, political and administrative decentralisation measures almost overnight at the start of 2001 (Hofman and Kaiser 2002).

• *The second figure for 2004 (434) is for districts approved for 2004, the first (410) for those confirmed by law for 2004.

• Source: Authors' 2004 district database; local governments shown exclude those within DKI Jakarta.

‘Parliament Approves 25 New Districts and Cities: Growing Concern about Usefulness, Implications’, GTZ Decentralisation News (40), 7/2/2003: 1.

The territorial division of Indonesia dates back to the Dutch colonial era. Historically, the dynamics of the relationship between the central government on Java, with its more established administrative structure, and the much larger land areas of the Outer Islands, have shaped the various types of government organisation (MacAndrews 1986). Before decentralisation, provinces were known as level one (Dati I) and local governments as level two (Dati II). The Dutch introduced the system of provinces on Java in the 1920s, with each composed of regencies (residentien/regentschappen and districten) (Gonggryp 1992/1934). However, the administrative subdivision of the Outer Islands lagged behind that of Java. The imposition of Javanese administrative patterns, often conflicting with traditional forms of local community organisation, has been a source of long-standing discontent outside of Java.

Interestingly, some early attempts to revise Law 22 appear to have been characterised less by re-centralisation than by an attempt to strengthen the executive relative to the legislature (Ellis and Du Sautoy 2001). These were retracted in the face of opposition from regional representatives and parliaments.

Provinces were largely bypassed in terms of power and resources in the decentralisation. Under Law 22/1999 (article 4), there is also no longer a hierarchical relationship between provinces and local governments. East Timor, Indonesia's 27^thprovince in the pre-decentralisation era, seceded in 1999. Since then the number of provinces has increased from 26 to 30. A controversial decision in 1999 to split Papua/Irian Jaya into three provinces (McGibbon 2003) has been the subject of a court challenge by Papuan provincial government leaders; controversy continues over the Constitutional Court's decision on 11 November 2004 confirming the 2003 establishment of West Irian Jaya province, but ruling against the proposed creation of Central Irian Jaya province.

The implementation of institutional arrangements for the lowest tier of government, the village (desa) and neighbourhood (kelurahan), have now been delegated to the districts. However, especially in the case of villages, national laws prescribe accountability arrangements that include an elected council (BPD), which in turn can hold an elected village head accountable.

Before Law 22/1999 was enacted, the creation of new regions was regulated under Law 5/1974 (article 72). Government Regulation 129/2000 on the Establishment of New Regions (Persyaratan Pembentukan dan Kriteria Pemekaran, Penghapusan, dan Penggabungan Daerah) sets out the criteria for both the creation of new regions and the amalgamation of regions. Proposals for new regions are supposed to be scored on a number of economic, socio-cultural and socio-political criteria, as specified by Regulation 129/2000, although in practice this does not appear to have been a guiding or binding constraint. The criteria are scored on the basis of 19 indicators and 43 sub-indicators. Each indicator is scored on a scale of 1–6, with the respective weighting of the indicators generating a potential score of 100. A regional split appears to be vetoed (ditolak) if more than 10 indicators receive a score of 1, but granted (lulus) if the score is above a minimum, although this does not appear to be specified.

After the second (2000) amendment to the constitution the president had no effective veto on laws approved by the DPR.

Our district dataset – available on request – tracks the legacy of splits from the original 292 local governments in 1998 to the 434 ratified by end of 2003. In the initial DAU allocations for 2004, 410 regions received DAU, although this was subsequently pro-rated to cover all districts. We identified all new local governments, their originating local governments, and the respective legal instruments. The full dataset covering all 434 local governments contains one identifier for each of the five splitting waves. A ‘1’ denotes an original (induk) region, whereas ‘2’ and above indicates a derivative ‘child’ (anak) region. From this sequence of digits we can read whether a region was ever subject to a split, as well as its respective parent region. A sequence of five zeros indicates that the local government boundaries remain unchanged. Positive digits denote the region type (induk or anak), the existence of a split and the timing of the split or splits a region experienced.

For the April 2004 elections, the General Election Commission (KPU) requested funds for election committees for 438 local governments (‘Komisi Pemilu Anggaran Rp 2,9 Trillion untuk 2004’, Tempo, 22/5/2003: 5).

Banten split from West Java, Gorontalo from North Sulawesi, Bangka Belintung from South Sumatra, and North Maluku from Maluku. Papua was split into three provinces, but West Papua is the only new province to have become operational.

An exception is the city of Bandung, which expanded its boundaries into the district of Bandung in the 1990s. Indonesia's more than 4,000 subdistricts have also experienced some boundary changes.

Our focus here is on the motivations for creating new local governments, as opposed to provinces. Not only have there been far fewer provincial splits, but the dynamics seem to differ across these cases.

Strictly speaking the field appears to be offshore, although a limited amount of revenues have been going to the province through special arrangements. Expected contracts in the West Natuna gas field are projected to generate billions of dollars in revenue for the Indonesian government (Laksamana.net, The Politics and Economics Portal, ‘Oil & Gas: Get Human, ExxonMobil Told', 27/5/2002).

Huzrin was convicted of stealing an alleged half-million dollars, and was relieved of his post (‘Supreme Court Rejects Huzrin's Appeal’, JP, 15/4/2004). The national legislature had formed a special committee to draft a bill on the formation of Riau Islands province and passed this as Law 25/2002, although it had not yet been approved by the Riau provincial legislature. Members of the committee argued that approval from the provincial head or regional legislature was not actually needed (‘Thousands Demonstrate for Riau Island Province’, JP, 1/5/2002).

Evidence from local political jurisdictions in the United States suggests that communities are willing to sacrifice economies of scale to avoid racial heterogeneity in their jurisdiction; similar effects for income heterogeneity were less pronounced (Alesina et al. 2000).

These revenue categories and their amounts and allocation rules are specified under Law 25/1999. The provincial revenue sharing relationship for natural resource transfers dictates that one-third of oil and gas natural resource revenue goes to the producing local governments, one-fifth to the province, and one-third equally to all other local governments (Hofman et al. 2004). Natural resource sharing in the ‘special autonomy’ provinces of Aceh and Papua typically involves additional allocation criteria.

For example, Collier and Hoeffler (2002) suggest that the root cause of international secession movements may be not ethnic identity per se, but rather that a sense of identity may be stoked and exploited by politicians to gain economic advantage.

Our data are drawn from a range of sources, including the 2000 population census, the National Socio-Economic Household Survey (Susenas) and the 2000 Village Potential Statistics (Podes). Ideally, further work will attempt to map these characteristics to both beginning and end reference periods (with 292 and 434 localities, respectively), although this has proved very challenging.

^aRegions B and C are examples of pre-decentralisation splitters, having split from region A between 1998 and decentralisation in 2001. Region D is an example of a postdecentralisation splitter, giving rise to regions E and F between 2001 and 2004. Broken arrows denote the possibility that a district splits into more than two districts, and that districts resulting from a split may split again.

It is worth bearing in mind that behind these figures there is a great deal of diversity. Ten local governments had populations of over 2 million, led by kabupaten Bandung in West Java with 4.1 million inhabitants. Twenty-one local governments had populations of less than 100,000, with only 24,000 inhabitants in the Acehnese city of Sabang.

Group identities, and their manifestations in social and economic processes, are complex and difficult to measure. Since these measures are self-reported, however, they may be indicative of locally relevant identities.

For example, the proposed splitting of kabupaten Mener from kabupaten Central Aceh seems to have been triggered by the fact that the Gayo ethnic group is concentrated in the north-eastern subdistricts.

Note that the diversity of the district (E _i ) will always be larger than that of the subdistrict (E _i ) using the Theil measure, so that the difference between the two (E _i – E _j ) will be positive.

Empirical results using the ELF and the E measure were consistent.

Ideally we would have liked to construct a measure to determine whether districts became more homogeneous owing to a split. Data limitations prevented this, however.

The poverty share measures the fraction of the district population that falls below the poverty line, with per capita consumption (from the Susenas) as the welfare measure. The poverty gap (or poverty depth) index of poverty is based on the average amount by which the poor's consumption falls below the poverty line.

Local governments have the option to choose ‘lean’ or ‘bloated’ organisational structures. One reason cited by some informants for localities choosing larger bureaucracies is that they yield more management jobs.

Our data on the political composition of local governments is based on the share of votes received by the 11 top parties nationally in the 1999 elections. These account for over 80% of votes on average across local governments, but range from 92% on Java to only 66% in Maluku/Papua. We are seeking to update these data to include the shares for all parties and/or to obtain the hardcopy records of party affiliation for actual seats in local DPRDs (KPU 2002). Of the 48 parties that contested the 1999 elections, 46 are actually represented among the more than 10,000 local legislative members, some with only one representative. Evans (2003) provides a useful graphical overview of the local-level results of the 1999 elections, but does not elucidate how the jurisdictional definitions of 1999 were mapped to the 359 local jurisdictions presented.

The national electoral system provided for an over-representation of off-Java districts in the 1999 elections. Golkar, the party in power before decentralisation, was overrepresented outside Java, potentially giving it an incentive to create more electoral districts off-Java.

Somewhat worryingly, elsewhere we found some evidence (from a special survey of 150 localities) that citizens perceive KKN in major local political processes, such as the election of regional heads and acceptance of their annual accountability speeches, to be higher in regions that split after decentralisation (Hofman and Kaiser 2004).