An investigation of the resource curse in Indonesia

Abstract

We investigate the effect of resource dependence on district level income in a rare within-country study for Indonesia, one of the largest resource producing countries in Asia. We follow 390 districts between 2006 and 2015, consider four alternative measures of resouce dependence, and instrument for the potential endogeneity of each using historical measures of oil, gas and coal reserve locations, and changes in the physical production of each resource. Using annual fixed effects and first differenced regressions with and without various instruments, we find no evidence of a "resource curse". Instead, we find robust evidence across all models that dependence as measured by mining's share of output is positively associated with district real per capita income. We find a similar positive relationship between dependence as measured by the share of district government revenues from oil and gas or mining overall, and income in our most credible specifications with instruments. For example, a standard deviation increase in change in district government dependence on oil/gas revenues increases real per capita income by 16 percent over a nine year period.

Resource dependence and the causes of local economic growth: An empirical investigation*

“…For our first step, we estimate the overall (reduced form) effect of resource dependence on real GRDP per capita using a first difference estimator. Analogous to fixed effects for two periods, first differencing should control for stable but unobserved district characteristics that may also be affecting income, and replicates the method of Hilmawan and Clark (2018), albeit for a slightly shorter difference in time (2007 to 2015) to accommodate subsequent analysis.…”

“…Yet since that known abundance was determined by central government or international corporate exploration efforts, it should affect subsequent district level growth mainly through extraction. 11 Our second type of instrument is the difference in physical oil, natural gas, and coal mining production from 2007-2015, again as constructed by Hilmawan and Clark (2018) following Caselli and Michaels (2013) for Brazil, and Cust and Rusli (2016) for Indonesia. The specific instrument we use of each type is customised to each resource dependence measure, and will be tested for relevance and overidentification using tests developed for Stata by Schaffer, Baum, and Stillman (2003).…”

“…1 The sheer size and variation in resource dependence across Indonesia, coupled with the improved availability of its district level data since its move to decentralisation between 1999 and 2004, make the country an excellent test case for the effects of resource dependence. Both Cust and Rusli (2016) and Hilmawan and Clark (2018), using panel regression and instruments, have found that district government revenues from oil and gas are positively associated with real per capita income. Hilmawan and Clark (2018) also find positive effects on growth of mining's share in district gross regional domestic product (GRDP).…”

“…Both Cust and Rusli (2016) and Hilmawan and Clark (2018), using panel regression and instruments, have found that district government revenues from oil and gas are positively associated with real per capita income. Hilmawan and Clark (2018) also find positive effects on growth of mining's share in district gross regional domestic product (GRDP).…”

“…In this paper, we seek to test the extent to which the four causal mechanisms above can explain the positive overall effect of resource dependence on per capita income found for Indonesia by Cust and Rusli (2016) and Hilmawan and Clark (2018). We continue this literature's recent practice of using panel methods, multiple resource dependence measures, and instruments for all resource dependence measures.…”