Full Length Articles
Good mine, bad mine: Natural resource heterogeneity and Dutch disease in Indonesia

https://doi.org/10.1016/j.jinteco.2021.103457Get rights and content

Abstract

We analyze the local effect of exogenous shocks to the value of mineral deposits on a panel of manufacturing plants in Indonesia. We introduce heterogeneity in natural resource extraction methods, which helps to explain the mixed evidence found in the ‘Dutch disease’ literature. In districts where mineral extraction is relatively capital intensive, mining booms cause virtually no upward pressure on manufacturing wages, and both producers of more heavily traded and relatively less-traded manufacturing goods benefit from mining booms in terms of employment. In contrast, labor-intensive mining booms drive up local wages such that heavily traded goods producers respond by reducing employment.

Introduction

Wealth in non-renewable natural resources (such as solid minerals and oil & gas) does not always lead to sustained economic development. This observation has long inspired a debate on the existence of ‘Dutch disease’, in which natural resources crowd out the traded sector and reduce growth (The Economist, 1977; van Wijnbergen, 1984), and has led to warnings of a seemingly incurable ‘resource curse’ (Gelb, 1988; Sachs and Warner, 2001). This literature has recently moved away from cross-country studies in which endogeneity issues are harder to address and started to exploit within-country variation to minimize the influence of confounding factors.11 Using detailed county- and firm-level data for the US, several studies find contrasting positive effects of local natural resource booms, or at least no evidence for crowding out of manufacturing employment (Black et al., 2005; Michaels, 2011; Allcott and Keniston, 2018). For developing countries the evidence is more mixed and ranges from higher GDP per capita in Brazil (Cavalcanti et al., 2019) to more conflict in Colombia (Dube and Vargas, 2013), an increase in real income for households after a large open-pit gold mine in Peru increased local procurement (Aragón and Rud, 2013), localized negative effects on traded-sector employment in emerging markets (De Haas and Poelhekke, 2019), and an increase in municipal government spending in Brazil that does not translate into better public goods and services (Caselli and Michaels, 2013).

The literature has typically exploited geographic variation in natural resource wealth and time variation in world prices or giant oil discoveries, but has not distinguished explicitly between different resources or extraction techniques. We show that resource extraction techniques vary significantly in their labor intensity, and that this source of heterogeneity can reconcile positive and negative outcomes found in the literature. We analyze the local effect of a booming natural resource sector within Indonesia, which is both a major producer of a variety of natural resources that are scattered across the country and has a large and exporting manufacturing sector.

Combining detailed panel data on manufacturing plants with deposit-level data, we find that in administrative districts where mining of minerals is more capital intensive, mining booms cause an increase in plant-level manufacturing employment. Specifically, employment rises by 2.6% in a district with average mining intensity when world prices of local minerals increase by 100 log points and mining is relatively capital intensive. In contrast, mining booms in districts where mining is labor intensive reduce manufacturing employment by 1.2%. We show that this negative employment effect is driven by producers of more heavily traded manufactured goods, whereas producers of relatively less-traded manufactured goods can avoid a contraction of employment. The key mechanism, which we verify empirically, is that a booming mining sector only exerts strong upward pressure on local wages if its mining method is labor intensive. This makes local producers of heavily-traded goods less competitive during labor-intensive mining booms, while less-traded goods producers are able to pass on higher wage costs to consumers by raising prices. From the perspective of manufacturing plants, the local extraction technique can thus determine whether mining booms are good or bad.

Using novel well-level data, we control for oil & gas booms and show that these do not lead to a rise in manufacturing wages or a reduction in employment. This is consistent with oil & gas production being mostly offshore and highly capital intensive and specialized. Heterogeneity in extraction methods thus helps to explain why many studies that have focused on capital-intensive natural resource extraction such as open-pit mining or oil & gas production do not find evidence for local crowding-out effects in the manufacturing sector during natural resource booms. In terms of magnitude, the effect of mining booms in Indonesia is much larger than the effect of oil & gas booms on local manufacturing in the US (Allcott and Keniston, 2018). However, we also do not find that the reallocation between sectors and reduction in activity by more-traded goods producers leads to a decrease in total factor productivity. This speaks against foregone ‘learning by doing’ productivity spillovers as described theoretically in Van Wijnbergen (1984) and Arrow (1962) and confirmed empirically in other contexts by Ellison et al. (2010), Greenstone et al. (2010) and Kline and Moretti (2014). Therefore, our findings do not provide empirical support for Dutch disease effects in the narrow sense, in which foregone productivity gains in non-resource tradeable goods sectors slow down overall economic growth.

Our identification strategy is to correlate exogenous shocks to the value of local natural resource endowments that were discovered by the start of our sample period with local manufacturing outcomes. Informed by the literature that has questioned the exogeneity of exploration and discoveries (Bohn and Deacon, 2000; Cotet and Tsui, 2013; Arezki et al., 2019; Cust and Harding, 2020), we use exogenous world prices to infer changes in the value of natural resource deposits. We employ deposit- and well-level data to compute measures of initial endowments of minerals and oil & gas at the district level. We interact these with subsequent exogenous world price shocks and, in the case of mining, with an indicator that captures the labor intensity of local extraction methods. The mining engineering literature posits that the extraction technique is determined by the exogenous geological shape of the local deposit and not by the deposits' contained minerals or local labor market characteristics (see e.g. Hartman and Mutmansky, 2002). Using three separate databases, we empirically establish that different mining techniques (such as underground versus open-pit mining) translate into very different degrees of labor intensity. In addition, our empirical specification controls for any remaining impact of local labor market characteristics and accounts for differential trends in outcome variables across individual districts. Our annual manufacturing data contains detailed plant-specific information on all Indonesian manufacturing plants with 20 or more employees between 1990 and 2009. This enables us to control for plant fixed effects and four-digit industry-times-year fixed effects, which avoids selection effects and thereby improves identification. Furthermore, the richness of the data allows us to distinguish between manufacturers of relatively more-traded and relatively less-traded goods and thus analyze which of these suffer less or benefit more from mining booms.2

We contribute to a growing literature that analyzes the impact of natural resources in within-country settings. Data on counties and firms in the US have shown that coal, oil, and gas booms, of which the recent boom was driven by novel shale extraction techniques, have had either small or no negative effects on manufacturing. Black et al. (2005) find positive employment effects on non-tradeable sectors during the 1970s coal boom in a county-level analysis across Kentucky, Ohio, Pennsylvania, and West Virginia, but no significant effects on the manufacturing sector. A long-run study of the southern US by Michaels (2011) finds that as population increased in booming counties also local public good provision increased, with positive effects on employment in agriculture and manufacturing. Allcott and Keniston (2018) show that in a US-county with an additional oil & gas endowment of US$10 million per square mile, a natural resource boom that raises national oil & gas employment by 100 log points leads to an increase in population by 1.2%, employment by 2.8% and earnings per worker by 1.8%. The US manufacturing sector is also procyclical with oil & gas booms in resource-abundant counties, although there is some limited plant-level evidence that highly-traded goods producers contract. In terms of income per capita, however, busts can more than reverse the positive effects of booms (Jacobsen and Parker, 2016).

We add to this literature by distinguishing between different extraction methods used to take natural resources out of the ground and the relative labor intensity that this implies. By analyzing a developing country with different degrees of sectoral and regional labor mobility compared to the US, we place the results in the literature into a broader perspective. Since we find that labor mobility across districts in Indonesia is low and because limited specialization in Indonesian manufacturing likely results in more sectoral labor mobility, there may be more scope for crowding out of manufacturing. Moreover, in a developing country potentially less firms are up- and downstream to the natural resource sector than in the US, where “linkages and complementarities to the natural resource sector were vital in the broader story of American economic success” (Wright and Czelusta, 2007).

To the best of our knowledge, the effects of heterogeneity in the labor intensity of commodity extraction have only been studied in the context of conflict. Dube and Vargas (2013) find that an exogenous increase in the price of coffee (which is labor intensive in production) decreases armed conflict in Colombia because it raises the opportunity cost of fighting, while an increase in the price of capital-intensive oil production increases conflict, through increasing the gains from appropriation of oil income. These effects are consistent with a model of social conflict by Dal Bó and Dal Bó (2011) in which positive shocks to labor-intensive industries diminish conflict.

We also add to a literature that has examined a range of other outcomes during natural resource booms. These include higher local household income after a rise in local procurement by a large Peruvian gold mine (Aragón and Rud, 2013), increased wage and royalty and business income after new hydrofracturing oil & gas production (Feyrer et al., 2017), a rise in reported municipal spending after oil windfalls in Brazil that is not matched by better local living standards (Caselli and Michaels, 2013), property prices that increase due to royalty payments or decrease due to environmental risk (Muehlenbachs et al., 2015), decreased entrepreneurship in coal and heavy industry-intensive cities (Glaeser et al., 2015), increased income leading to more health care spending (Acemoglu et al., 2013), and increased crime rates (James and Smith, 2017).

Finally, by analyzing the effects of regional shocks in local districts of Indonesia, we also relate to several recent studies that have developed quantitative spatial equilibrium models that analyze the welfare consequences of regional shocks (Redding, 2016; Redding and Rossi-Hansberg, 2017; Caliendo et al., 2017; Faber and Gaubert, 2019; Fajgelbaum and Gaubert, 2020; Burstein et al., 2020). In this paper we focus on the local effects of natural resource price booms on the manufacturing sector and thus abstract from a full welfare analysis. This would ideally also take natural resource-specific factors into account. These include policy at the regional and national level such as industrial policy for the development of up- and downstream industries, and macroeconomic management of resource wealth at the national level such as fiscal rules, exchange rate management, sovereign wealth funds, and redistribution.

The remainder of our paper is structured as follows. In Section 2 we develop intuition for our hypotheses, while Section 3 provides background information for Indonesia and discusses data sources and the construction of key variables. Section 4 presents the empirical strategy, Section 5 shows the results as well as a battery of robustness checks, and Section 6 concludes.

Section snippets

Resource extraction methods and local Dutch disease

In this section we develop hypotheses that relate heterogeneity in resource extraction methods to local Dutch disease effects, for more- and for less-traded manufacturing plants.

Our point of departure is the observation that natural resource deposits differ in important ways. Oil & gas is typically pumped from wells and is capital-intensive. Mines that extract non-renewable natural resources from deposits (such as metals and solid minerals) vary in terms of which method they use for extraction.

Indonesian context

Indonesia ranks fourth in the world by population size and is both a major producer of minerals and a significant producer and exporter of manufactured goods. Therefore, the country provides an ideal testing ground. The (non-mining) manufacturing sector (ISIC Revision 3.1, divisions 15 to 37) represented 23% of GDP on average between 1993 and 2009. In 2009, Indonesia exported 14% of manufacturing output, consisting mostly of food products and beverages, wood products, rubber products, textiles,

Empirical strategy

Our main hypothesis is that the intensity of mining activity affects plant-level outcomes, and that this varies by the labor intensity of extraction techniques and the degree of tradedness of produced goods. Since we observe the location of plants at the district level we need exogenous variation in mining activity at the district level and over time. We achieve this by interacting initial mineral endowments with changes in exogenous world prices of local minerals.

We model plant-level outcomes Y

Mining heterogeneity and manufacturing outcomes

We now turn to our main results, which show that heterogeneity in the labor intensity of mining techniques can reconcile the mixed effects of mining booms on manufacturing outcomes found in the previous literature. We first distinguish the effect of labor- versus capital-intensive mining on four key manufacturing outcomes using a sample that includes both more- and less-traded goods producers (Table 1), and then study these two sub-samples separately (Table 2). For comparison, in Table 1 we

Conclusion

We estimate the impact of local mining booms on manufacturing plants in Indonesia, exploiting detailed information on natural resource deposits and introducing the different degrees of labor and capital intensity that distinct mining methods entail. We present the novel result that global resource price increases only lead to crowding out of manufacturing employment in districts where mining operations are relatively labor intensive, and only for more-traded goods producers. Producers of

Declaration of Competing Interest

None.

References (58)

  • S.T. Anderson et al.

    Hotelling under pressure

    J. Polit. Econ.

    (2018)
  • F.M. Aragón et al.

    Natural resources and local communities: evidence from a Peruvian gold mine

    Am. Econ. J. Econ. Pol.

    (2013)
  • K.J. Arrow

    The economic implications of learning by doing

    Rev. Econ. Stud.

    (1962)
  • S. Bazzi et al.

    The political boundaries of ethnic divisions

    Am. Econ. J. Appl. Econ.

    (2020)
  • M. Bertrand et al.

    How much should we trust differences-in-differences estimates?

    Q. J. Econ.

    (2004)
  • D. Black et al.

    The economic impact of the coal boom and bust

    Econ. J.

    (2005)
  • H. Bohn et al.

    Ownership risk, investment, and the use of natural resources

    Am. Econ. Rev.

    (2000)
  • A. Burstein et al.

    Tradability and the labor-market impact of immigration: theory and evidence from the United States

    Econometrica

    (2020)
  • L. Caliendo et al.

    The impact of regional and sectoral productivity changes on the U.S. economy

    Rev. Econ. Stud.

    (2017)
  • C. Cameron et al.

    A practitioner’s guide to cluster-robust inference

    J. Hum. Resour.

    (2015)
  • F. Caselli et al.

    Do oil windfalls improve living standards? Evidence from Brazil

    American Economic Journal: Applied Economics

    (2013)
  • T. Cassidy

    How forward-looking are local governments? Evidence from Indonesia

    Mimeo

    (2019)
  • T. Cavalcanti et al.

    Winning the oil lottery: the impact of natural resource extraction on growth

    J. Econ. Growth

    (2019)
  • A. Ciccone

    Economic shocks and civil conflict: a comment

    Am. Econ. J. Appl. Econ.

    (2011)
  • W.M. Corden et al.

    Booming sector and de-industrialisation in a small open economy

    Econ. J.

    (1982)
  • S. Correia

    Singletons, Cluster-Robust Standard Errors and Fixed Effects: A Bad Mix. Working Paper

    (2015)
  • A.M. Cotet et al.

    Oil and conflict: what does the cross country evidence really show?

    Am. Econ. J. Macroecon.

    (2013)
  • S. Courteney et al.

    Indonesia, Oil and Gas Atlas

    (1988–1991)
  • J. Cust et al.

    Institutions and the location of oil exploration

    J. Eur. Econ. Assoc.

    (2020)
  • Cited by (16)

    • Coal busts and urban recovery: Evidence from China

      2023, Regional Science and Urban Economics
    • Natural resources, (Mis)governance, and the lack of diversification in mining regions: A reconsideration of the Dutch disease at a local level

      2022, Extractive Industries and Society
      Citation Excerpt :

      Second, research on the Dutch disease tends to disregard the resource movement effect (Bjørnland et al., 2019); however, economic diversification depends on the availability of regional assets (Breul et al., 2021; Frangenheim et al., 2020; Hassink et al., 2019), and the concentration of scarce assets in a single industry can challenge it. Third, research examining the resource movement effect has primarily focused on the reallocation of labor (Faber and Gaubert, 2019; Pelz and Poelhekke, 2021). Consequently, other assets that are less mobile than labor for which access is geographically constrained are not considered, such as natural resources (e.g., freshwater).

    View all citing articles on Scopus

    We thank Maarten Bosker, Massimiliano Cali, Dave Donaldson, Julian Emani Namini, Aysil Emirmahmutoglu, Andreas Ferrara, Jakob de Haan, Ralph de Haas, Albert Jan Hummel, Beata Javorcik, Daniel Keniston, Peter Lanjouw, Juan Pablo Rud, seminar participants at the Tinbergen Institute Amsterdam, Vrije Universiteit Amsterdam and Oesterreichische Nationalbank as well as conference participants at the CompNet-EBRD Workshop on Localisation and Productivity in London, the CAED 2017 in Seoul, the 10th FIW conference in Vienna and the 16th EUDN PhD Workshop on Development Economics in Wageningen for helpful comments. All errors are our own. Further, we thank Beata Javorcik, Hengky Kurniawan and Menno Pradhan for providing data for this project, and Mark van der Harst for excellent research assistance.

    View full text