Cyclical shocks and spatial association of Indonesia's district‐level per capita income

• Published date: 01 September 2022
• Author: Mitsuhiko Kataoka
• Date: 01 September 2022
• DOI: http://doi.org/10.1111/asej.12277

ORIGINAL ARTICLE

Cyclical shocks and spatial association of Indonesia’s

district-level per capita income

Mitsuhiko Kataoka

Graduate School of Business, Rikkyo University, Tokyo, Japan

Correspondence

Mitsuhiko Kataoka, Graduate School of Business, Rikkyo University, 3-34-1 Nishi-Ikebukuro, Toshima-ku, 171-

8501-0, Tokyo, Japan.

Email: mkataoka@rikkyo.ac.jp

Funding information

Japan Society for the Promotion of Science, Grant/Award Number: Type c: 17K03723

Abstract

Our exploratory spatial data analysis covered Indonesia’s district-level per capita

incomes for 2004–2018 and found statistical evidence of a weak but monotonically

increasing positive spatial association. The spatial income clusters/outliers were

scattered nationwide and expanded geographically. Applying the filtering method, we

found that regional cyclical shocks significantly influenced spatial association, largely

in resource-rich districts, and identified the locations of persistent spatial association

that were immune to shocks. We also specified new development targets, showing the

adjacent coexistence of low-income agrarian clusters with high-income mining clusters

in the undeveloped Papuan provinces. The overall national average is moving toward

the narrowing of inter-district income disparities. However, nearly half of all districts

have widened the income disparities with their neighbors; shocks increased income

gaps rather than narrowing them.

KEYWORDS

cyclical shocks, exploratory spatial data analysis, Geary’sc, Indonesia, Moran’sI, quantile

regression, spatial autocorrelation

JEL CLASSIFICATION

R11, R12, R58

DOI: 10.1111/asej.12277

© 2022 East Asian Economic Association and John Wiley & Sons Australia, Ltd.

Received: 10 August 2020; Accepted: 20 April 2022

Asian Economic Journal. 2022;36:261–287. wileyonlinelibrary.com/journal/asej 261

1|INTRODUCTION

Spatial association (autocorrelation) is the degree of similarity between one

region and its nearby regions.

1

Exploratory spatial data analysis (ESDA) is a

collection of techniques for the conventional statistical analysis of geographic

information; it describes the patterns of spatial association, identifies their

location and magnitude, and visualizes spatial distributions (Anselin, 1995;

Fischer & Wang, 2011; Kondo, 2015; Messner et al., 1999; Tamesue

et al., 2013). Global indicators of spatial association (GISA) analyze spatial

association as a whole, whereas local indicators of spatial association (LISA)

specify the location and magnitude of which regions with similar values are clus-

tered, dispersed or randomly distributed (Anselin, 1995).

As the largest archipelagic nation, Indonesia has the world’s fourth-largest

population, scattered natural resources and extraordinary industrial diversity.

2

It has been beset by serious regional income disparities since its independence

(Akita & Lukman, 1995; Kataoka, 2018; Vidyattama, 2013,2014). To address

this issue, the government has implemented various policies, including several

5-year national development plans and integrated economic development zones.

Furthermore, since 1999, Indonesia has implemented major changes in its inter-

governmental budget allocation system, adopting a more decentralized fiscal

regime (Kataoka, 2018). However, outcomes are still far below the target levels.

The sub-national structure consists of four tiers: provinces (Indonesian: pro-

vinsi), regencies (kabupaten)/cities (kota), sub-districts (kecamatan) and urban

villages (kelurahan)/rural villages (desa).

3

Regional data are published at the

province and regency/city levels. Hereinafter, this study defines the second sub-

national tier as a district.

Figure 1illustrates the spatial distribution of the per capita gross regional

domestic product (GRDP) of 514 districts in 2018 by quartile.

4

The figure shows

that high-income districts are distributed unevenly. The majority of the highest-

quartile income districts are located in resource-rich provinces such as Riau;

Riau Islands; East, South and North Kalimantan; North Sumatra; Papua; the

leading business districts on Java Island; and the major commercial centers off

1

The terms ‘spatial association’and ‘spatial autocorrelation’are used interchangeably.

2

Indonesia’s major natural resource products vary by province: oil and natural gas in Aceh, Riau, Riau Islands

and East Kalimantan; coal in North Sumatra, East Kalimantan and South Kalimantan; and gold and copper in

Papua and West Papua.

3

The nation has experienced a territorial split at the provincial and district levels, associated with the

decentralization process, for the past two decades. According to Indonesia’s Central Bureau of Statistics (Badan

Pusat Statistik, BPS, 2018), the nation has 34 provinces, 416 regencies, 98 cities, 6543 sub-districts, 919 urban

villages and 74 517 rural villages.

4

Most interregional inequality studies focus on GRDP as a proxy for per capita regional income and as an

indicator of development (Vidyattama, 2013). This study uses the terms ‘per capita GRDP’and ‘per capita

(regional) income’interchangeably.

262 KATAOKA