Dutch Disease? Firm-level evidence from Indonesia Jim Cust NRGI and OxCarre Tornn Harding NHH Norwegian School of Economics Pierre-Louis Vézina King's College London March 28, 2016 Cust et al. Dutch Disease March 28, 2016 1 / 29
What's Dutch Disease and why would it matter? Does oil and gas extraction make manufacturing less competitive by increasing wages? This is important as manufacturing may be special in terms of productivity spillovers (Matsuyama, 1992) Cust et al. Dutch Disease March 28, 2016 2 / 29
How does one catch the Dutch Disease? Corden and Neary (1982): Extra wealth appreciation of real exchange rate contraction of traded sector Allcott and Keniston (2014): Whether manufacturing contracts during a resource boom depends on whether: local manufacturing wages rise prices are exogenous, i.e. don't rise Cust et al. Dutch Disease March 28, 2016 3 / 29
Is Dutch Disease real? Smith (2014): Wages and employment increase in oil-dependent countries during the oil price boom in the 1970s Little on rm-level Dutch Disease eects Allcott and Keniston (2014) is one recent exception Cust et al. Dutch Disease March 28, 2016 4 / 29
This paper What is the eect of oil and gas extraction on wages across Indonesian rms? Our contribution: Firm-level evidence from an emerging economy An improved identication strategy Cust et al. Dutch Disease March 28, 2016 5 / 29
Recent and relevant papers to keep in mind Cavalcanti et al. (2015): Compare Brazil municipalities where oil was discovered to municipalities with drilling but no discovery Oil discoveries signicantly increase GDP per capita Positive spillovers to services and manufacturing GDP Cust et al. Dutch Disease March 28, 2016 6 / 29
Recent and relevant papers to keep in mind De Haas and Poelhekke (201X): Look at rms in emerging economies Firms report more constraints to doing business if they are located within 20km of active mines These constraints include access to educated workers Cust et al. Dutch Disease March 28, 2016 7 / 29
What about papers on Indonesia? Cust and Rusli (2014) look at scal transfers related to oil production in Indonesia They boost local GDP Cust (2014) examines the labor market eects of mines in Indonesia Shift of employment from the traded manufacturing sector to the non-traded services sector Cust et al. Dutch Disease March 28, 2016 8 / 29
Indonesia is a resource-rich country Natural resource rents (% of GDP) 0 10 20 30 40 1970 1980 1990 2000 2010 Indonesia World Source: World Development Indicators, World Bank. Cust et al. Dutch Disease March 28, 2016 9 / 29
Indonesia is a resource-rich country Source: The Observatory of Economic Complexity Cust et al. Dutch Disease March 28, 2016 10 / 29
Source: Wood Mackenzie Cust et al. Dutch Disease March 28, 2016 11 / 29 Indonesia is a resource-rich country Barrels per day (thousands) 0 100 200 300 1990 1995 2000 2005 2010 OIL GAS
Introduction Literature Data Empirical design Results References Indonesia is a resource-rich country Oil and gas total production 1990 2008 (2512.51,2493062] (21,2512.51] (0,21] [0,0] No data Note: Around 85% of districts have no oil or gas production in any period. 35% of production elds are o shore but can be linked to a closest district. Source: Wood Mackenzie. Cust et al. Dutch Disease March 28, 2016 12 / 29
Ups and downs in oil and gas production among non-dry districts Kernel density 0 2 4 6 8 1.5 0.5 1 Total production growth 1995 2000 2005 Cust et al. Dutch Disease March 28, 2016 13 / 29
Data Indonesian Manufacturing Census 1990-2008 All plants with more than 20 employees (>30,000 plants) We look at wages, exit rates, and prices using core products' unit values As in Amiti and Davis (2012), wages are the log of the average rm-level wage, dened as the total wage bill divided by the number of workers Cust et al. Dutch Disease March 28, 2016 14 / 29
Table: "Two-digit industries, ISIC rev. 3" isic_2d isic_2d_name N 15 Manu. of food products and beverages 7617 16 Manu. of tobacco products 1544 17 Manu. of textiles 3646 18 Manu. of wearing apparel; dressing and dyeing of fur 3112 19 Tanning and dressing of leather; Manu. of luggage, handbags, saddlery, harness and footwear 922 20 Manu. of wood and of products of wood and cork, except furn.; articles of straw and plaiting mat. 2186 21 Manu. of paper and paper products 709 22 Publishing, printing and reproduction of recorded media 990 23 Manu. of coke, rened petroleum products and nuclear fuel 232 24 Manu. of chemicals and chemical products 1384 25 Manu. of rubber and plastics products 1869 26 Manu. of other non-metallic mineral products 2579 27 Manu. of basic metals 406 28 Manu. of fabricated metal products, except machinery and equipment 1218 29 Manu. of machinery and equipment n.e.c. 650 30 Manu. of oce, accounting and computing machinery 266 31 Manu. of electrical machinery and apparatus n.e.c. 427 32 Manu. of radio, television and communication equipment and apparatus 380 33 Manu. of medical, precision and optical instruments, watches and clocks 240 34 Manu. of motor vehicles, trailers and semi-trailers 425 35 Manu. of other transport equipment 453 36 Manu. of furniture; manufacturing n.e.c. 3122 37 Recycling 86 Total 34463 Cust et al. Dutch Disease March 28, 2016 15 / 29
Identication Oil and gas extraction may take place where institutions and infrastructure are best, and this also favors manufacturing We need exogenous variation in the location and timing of oil and gas windfalls This comes from luck in oil and gas discovery's interaction with the real oil price Cust et al. Dutch Disease March 28, 2016 16 / 29
Identication This means we focus only on districts where exploration took place Cavalcanti et al. (2015) suggest districts with exploration but no discovery are a valid counterfactual We exploit both randomness in success of oil exploration and yearly variation in oil prices in our identication Cotet and Tsui (2013) also exploit randomness in success of oil exploration to look at eect of oil wealth on conict Cust et al. Dutch Disease March 28, 2016 17 / 29
Production is inuenced by luck but luck does not increase with exploration drilling Oil and gas production 0 5 10 15 Success rate 0.2.4.6.8 1 0.2.4.6.8 1 Success rate Regression coeff: 3.24. Robust s.e.:1.20 0 5 10 Exploration wells (log) Regression coeff:.011. Std. error clustered by year and Kab:.017 Cust et al. Dutch Disease March 28, 2016 18 / 29
Wages Average wage 7 7.5 8 8.5 9 9.5 Average wage 7 8 9 10 1990 1995 2000 2005 1990 1995 2000 2005 Luck in exploration No luck Oil and Gas No Oil and Gas Cust et al. Dutch Disease March 28, 2016 19 / 29
Exit rates Exit rate 0.05.1.15 Exit rate 0.05.1.15 1990 1995 2000 2005 1990 1995 2000 2005 Luck in exploration No luck Oil and Gas No Oil and Gas Cust et al. Dutch Disease March 28, 2016 20 / 29
Number of rms Nb of firms 2000 2500 3000 3500 4000 1990 1995 2000 2005 Nb of firms 5000 10000 15000 20000 1990 1995 2000 2005 Luck in exploration No luck Oil and Gas No Oil and Gas Cust et al. Dutch Disease March 28, 2016 21 / 29
Manufacturing output Manufacturing output (1990=100) 0 500 1000 1500 2000 Manufacturing output (1990=100) 0 500 1000 1500 2000 1990 1995 2000 2005 1990 1995 2000 2005 Luck in exploration No luck Oil and Gas No Oil and Gas Cust et al. Dutch Disease March 28, 2016 22 / 29
Cross section of rms (1) (2) (3) (4) (5) (6) Wage growth Wage growth Wage growth 2000s Wage growth 2000s Exit Exit Discovery rate 0.033 0.027 0.037 0.033-0.030-0.018 (0.017) (0.014) (0.019) (0.019) (0.036) (0.036) Constant 0.111 0.120 0.259 (0.013) (0.014) (0.026) N 14763 14763 11731 11731 16025 16023 R2 0.001 0.008 0.001 0.008 0.001 0.012 Industry FE N Y N Y N Y Kabubpaten-clustered s.e. in parenthesis. * 0.10 ** 0.05 *** 0.01. Cust et al. Dutch Disease March 28, 2016 23 / 29
Cross section of rms Electrical Machinery And Apparatus N.E.C. Machinery And Equipment N.E.C. Tobacco Products Other Transport Equipment Coke, Refined Petroleum Products And Nuclear Fuel Paper And Paper Products Rubber And Plastics Products Textiles Food Products And Beverages Publishing, Printing And Reproduction Of Recorded Media Chemicals And Chemical Products Products of Wood, Cork, and Straw Leather Luggage, Handbags, Saddlery, Harness And Footwear Furniture; Manufacturing N.E.C. Basic Metals Fabricated Metal Products, Except Machinery And Equipment.4.2 0.2.4.6 Effect of exploration luck on wage growth Cust et al. Dutch Disease March 28, 2016 24 / 29
Panel identication Y it = α i + δ jt + βl i P t + u it (1) L i is luck, i.e. the number of non-dry wells above total wells dug in the kabupaten of rm i P t is the oil price in year t in 2010 USD α i and δ jt are rm and industry-year xed eects Standard errors are clustered two ways: By rms to take into account potential serial correlation By kabupaten-year to adjust for any Moulton bias, as the treatment is more aggregated than the outcome. Cust et al. Dutch Disease March 28, 2016 25 / 29
Panel (Reduced-form) - Only explored Kabupatens (1) (2) (3) (4) Wage Wage production Wage non-production Unit value Avg success rate x Oil price 0.190 0.255 0.116 0.067 (0.043) (0.050) (0.231) (0.093) N 48194 48194 48194 46981 R-sq 0.80 0.91 0.80 0.85 Notes: Panel regressions with rm and industry-year xed eects. Standard errors clustered by rm and kabupaten-year in parentheses. ***p<0.01, **p<0.05, *p<0.10. Cust et al. Dutch Disease March 28, 2016 26 / 29
Panel (Reduced-form) - Only explored Kabupatens (1) (2) (3) (4) (5) (6) Exit Entry Product drop Product intro Output Empl Avg success rate x Oil price -0.018-0.006 0.018-0.001 0.156 0.073 (0.032) (0.004) (0.028) (0.029) (0.057) (0.025) N 48194 48194 48194 48194 48193 48194 R-sq 0.33 0.28 0.54 0.55 0.94 0.96 Notes: Panel regressions with rm and industry-year xed eects. Standard errors clustered by rm and kabupaten-year in parentheses. ***p<0.01, **p<0.05, *p<0.10. Cust et al. Dutch Disease March 28, 2016 27 / 29
Panel (Reduced-form) - Only explored Kabupatens (1) (2) (3) (4) Labor Prod. % exported Share imported inputs Intermediate inputs Avg success rate x Oil price 0.186-3.447 0.012 0.123 (0.063) (1.820) (0.008) (0.062) N 48193 48194 47876 48192 R-sq 0.81 0.69 0.87 0.92 Notes: Panel regressions with rm and industry-year xed eects. Standard errors clustered by rm and kabupaten-year in parentheses. ***p<0.01, **p<0.05, *p<0.10. Cust et al. Dutch Disease March 28, 2016 28 / 29
Conclusions We nd some evidence of Dutch Disease symptoms in Indonesia Firms face higher wages in oil and gas districts But rms are not more likely to exit Firms adapt by raising productivity Cust et al. Dutch Disease March 28, 2016 29 / 29
Allcott, H. and D. Keniston (2014). Dutch disease or agglomeration? the local economic eects of natural resource booms in modern america. American Economic Review. Amiti, M. and D. R. Davis (2012). Trade, rms, and wages: Theory and evidence. The Review of Economic Studies 79(1), 136. Cavalcanti, T., D. D. Mata, and F. Toscani (2015). Winning the oil lottery: The impact of natural resource extraction on growth. Technical report. Corden, W. M. and J. P. Neary (1982). Booming sector and de-industrialisation in a small open economy. The economic journal, 825848. Cotet, A. M. and K. K. Tsui (2013). Oil and conict: what does the cross country evidence really show? American Economic Journal: Macroeconomics 5(1), 4980. Matsuyama, K. (1992, December). Agricultural productivity, comparative advantage, and economic growth. Journal of Economic Theory 58(2), 317334. Smith, B. (2014). Dutch Disease and the Oil and Boom and Bust. Technical report. Cust et al. Dutch Disease March 28, 2016 29 / 29