Could a Resource Export Boom Reduce Workers’ Earnings? The Labour-Market Channel in Indonesia

Abstract

For a decade from 2000, Indonesia underwent a natural-resource export boom. Aggregate income rose, but real labour earnings stagnated. Employment rose, too, but mainly in low-skill sectors with predominantly informal employment arrangements. In this article, we reveal causal connections from the aggregate phenomenon of Dutch disease to these labour-market outcomes. We first explain broad sectoral trends, and then, integrating data from several national surveys, investigate sources of variation in boom-era labour earnings. We use instrumental variables to address issues of endogeneity and selection in earnings equations. After controlling for individual and district features, we find that the intensity of palm oil production—palm oil having been a key booming resource export—robustly predicts diminished formal employment, and that lower formality, in turn, robustly predicts lower earnings. Our findings establish causal linkages absent from prior studies, and so provide a structural dimension to ongoing debates over persistent poverty, rising inequality, and the lack of educational progress in Indonesia.

Selama satu dekade sejak 2000, Indonesia mengalami ledakan komoditas berbasis sumber daya alam. Penghasilan secara keseluruhan menunjukkan kenaikan namun penghasilan riil pekerja tetap stagnan. Lapangan pekerjaan juga bertambah, namun kebanyakan terjadi di sektor-sektor rendah keterampilan yang cenderung informal. Dalam tulisan ini, kami menemukan hubungan kausalitas dari fenomena agregat Dutch Disease terhadap apa yang terjadi di pasar tenaga kerja. Para penulis menjelaskan tren sektoral secara luas. Selanjutnya, dengan mengintegrasikan data dari beberapa survei nasional, kami menelaah berbagai sumber perbedaan pada penghasilan di era ledakan komoditas. Setelah mengisolasi pengaruh fitur-fitur individu dan daerah kabupaten/ kota, penulis menemukan intensitas produksi kelapa sawit sebagai kunci ekspor ledakan komoditas secara tepat memprediksikan penurunan lapangan pekerjaan formal dan bahwa selanjutnya, tingkat pekerjaan formal yang menurun tersebut secara tepat memprediksikan penghasilan yang lebih rendah. Penemuan penulis memberikan informasi hubungan kausalitas yang sebelumnya absen pada studi-studi yang lalu. Dengan demikian, tulisan ini memberikan dimensi struktural terhadap perdebatan berkelanjutan mengenai kemiskinan, meningkatnya ketimpangan, dan kurangnya kemajuan pendidikan di Indonesia.

Keywords:

• Dutch disease
• palm oil
• formal employment
• education
• poverty

JEL classification:

• O15
• J21
• F16
• I32

Notes

1. See http://www.indexmundi.com/agriculture/?country=id&commodity=palm-oil.

2. Trade data taken from Statistics Indonesia (BPS), the central statistics agency: http://bps.go.id.

3. In Indonesia, as elsewhere in the developing world, purchases of services (including housing, health, transport, and personal services) make up a larger share of expenditures for urban households than for rural ones, and for wealthier households relative to poorer ones. Since skilled workers are more highly concentrated in cities and in the upper deciles of the income distribution, it follows that higher prices of services will have a greater effect on their cost of living. Other things being equal, this will further reduce the real skill premium.

4. These probabilities are based on shares calculated using data from Sakernas (National Labour Force Survey). If we use data from the Indonesia Family Life Survey (IFLS), the probabilities are 8%, 16%, 39%, and 64%, respectively.

5. . This simple model also ignores variation in individual ability or motivation that might result in unequal rates of selection into schooling. Failure to account for selection effects imparts an upward bias to estimates of returns to schooling. Because variation in individual ability or motivation is largely unobservable, a wide range of instruments have been proposed and tested. We return to this topic in the discussion of our econometric model.

6. Basic data taken from the Bank for International Settlements (http://bis.org).

7. Of course, one important contrast between a resource export boom and an investment collapse is that the former raises total income whereas the latter reduces it. It is therefore clear that of these influences on manufacturing growth, the resource export boom was dominant, having sustained a real GDP growth rate of about 5% per year in the post-crisis era.

8. More information about Sakernas is available at http://microdata.bps.go.id/mikrodata/index.php/catalog/sakernas.

9. See http://databank.worldbank.org/data/reports.aspx?source=1266.

10. See http://www.rand.org/labor/FLS/IFLS.html. The description that follows paraphrases text in Strauss et al.’s (2009) overview.

11. Labour-market reports break sectors into agriculture, forestry, hunting, and fisheries; mining and quarrying; manufacturing; electricity, gas, and water; construction; wholesale and retail trade, restaurants, and hotels; transport, warehousing, and communication; financing, insurance, real estate, and business services; and community, social, and personal services.

12. The IFLS records both parents’ educational attainment. These are highly correlated, however, so we use only maternal educational attainment.

13. Of course, there are significant markers of educational attainment (such as completion of a schooling stage), so using discrete educational levels rather than years allows for non-linear returns. We also estimated these, but the results add little to the empirical story so we omit them.

14. This is because the probability of formality in this sample is 25.4%. So the expected formality premium is 73.9%, or 291 × 0.254.

15. The rising trend in inequality pre-dates by several years the adoption of some important policy reforms thought by some to have increased inequality—such as the relaxation of fuel subsidies.