Labor Displacement in Agriculture: Evidence from Oil Palm Expansion in Indonesia

Christoph Kubitza; Vijesh V. Krishna; Stephan Klasen; Thomas Kopp; Nunung Nuryartono; Matin Qaim

doi:10.3368/le.100.3.122122-0109R1

Article Figures & Data

Figures

Tables
Additional Files

Download figure
Open in new tab
Download powerpoint
Download figure
Open in new tab
Download powerpoint
Figure 1
Gendered Employment Rates at the Regency Level in Indonesia; left, Employment Rate and Sectoral Shares of Men; right, Employment Rate and Sectoral Shares of Women

View popup

Table 1

Datasets

	Year of Survey/Observation	Source
Local surveys
Farm households^a	2012; 2015; 2018	Primary data collected by authors
Spatial data
Location of palm oil mills	1922–2022	Benedict et al. (2023)
Forest cover in Indonesia	2000; 2005; 2010; 2012	Margono et al. (2014)
Large-scale oil palm plantations	2000; 2005; 2010; 2015	Austin et al. (2017)
Max. attainable yield of different crops	1961–1990 (baseline data)	Global agro-ecological zones data
National surveys
National labor force survey (SAKERNAS)	2000–2015	Badan Pusat Statistik
Tree Crops Statistics	2000–2015	Ministry of Agriculture
National village survey (PODES)	2001; 2003; 2006; 2011; 2014	Badan Pusat Statistik
Indonesian census	2000; 2010	IPUMS International database

↵a n = 683 per survey round.

View popup

Table 2

Effect of Oil Palm Cultivation on Annual Farm Household Income (2012–2015–2018)

	Village-Level Fixed Effects			Household-Level Fixed Effects
	Total Household Expenditure (log) (1)	Total Household Expenditure (log) (2)	Total Household Expenditure (log) (3)	Total Household Expenditure (log) (4)	Total Household Expenditure (log) (5)	Total Household Expenditure (log) (6)
Share of oil palm (0–1)	0.294^***	0.268^***	0.173^***	0.232^**	0.202^*	0.152
	(0.060)	(0.057)	(0.051)	(0.115)	(0.113)	(0.109)
Employed household members (= 1)		−0.605	−0.608		−0.011	−0.008
Employed household members (= 1)		(1.034)	(0.916)		(0.033)	(0.033)
Self-employed household members (= 1)		0.256^***	0.240^***		0.148^***	0.148^***
Self-employed household members (= 1)		(0.034)	(0.032)		(0.040)	(0.040)
Total farm size (ha)			0.049^***			0.038^***
			(0.005)			(0.010)
F-statistic	10.853	14.294	21.162	4.533	5.275	6.063
Observations	1,854	1,854	1,854	1,761	1,761	1,761

Note: The data source is farm-household data. Clustered standard errors at the household level are in parentheses. The dependent variable is the log of the total annual household expenditure (1,000 IDR). We control for the age and education of the household head, female-headed households, migrant households, number of women and adults, farm characteristics, distance to province capital, distance to next palm oil mill, employed household members, self-employed household members, total farm size, and year dummies. Additional covariates included in the estimation are in Appendix Table A5.

↵* p < 0.10;
↵** p < 0.05;
↵*** p < 0.01.

View popup

Table 3

Effect of Oil Palm Cultivation on Employment Status of Individuals in Farm Households (2012–2015–2018)

	Working (= 1) (Men) (1)	Working (= 1) (Women) (2)	Working On-Farm (= 1) (Men) (3)	Working On-Farm (= 1) (Women) (4)	Working Off-Farm (= 1) (Men) (5)	Working Off-Farm (= 1) (Women) (6)	Self-Employed Off-Farm (= 1) (Men) (7)	Self-Employed Off-Farm (= 1) (Women) (8)
Village-level fixed effects
Share of oil palm (0–1)	0.017	−0.159^***	0.041	−0.146^***	0.151^***	0.002	0.070^**	0.029
Share of oil palm (0–1)	(0.020)	(0.043)	(0.027)	(0.034)	(0.043)	(0.040)	(0.032)	(0.033)
F-statistic	494.760	89.602	399.988	61.836	66.373	21.417	10.062	8.658
Observations	2,759	2,610	2,759	2,610	2,759	2,610	2,759	2,610
Household-level fixed effects
Share of oil palm (0–1)	0.076^*	−0.078	0.079	−0.046	0.163	4.54e-04	0.165^**	−0.048
Share of oil palm (0–1)	(0.043)	(0.094)	(0.069)	(0.085)	(0.110)	(0.080)	(0.069)	(0.062)
F-statistic	538.625	58.762	387.982	37.355	40.157	16.408	7.176	6.762
Observations	2,693	2,548	2,693	2,548	2,693	2,548	2,693	2,548

Note: The data source is farm-household data. Clustered standard errors at the household level are in parentheses. We control for age, age-squared, student, education level, migrant households, number of women and adults in the household, farm characteristics, distance to the province capital, distance to the next palm oil mill, total farm size, and year dummies. Additional covariates included in the estimation are in Appendix Table A6 for village-level fixed effects models and Appendix Table A7 for household-level fixed effects models.

↵* p < 0.10;
↵** p < 0.05;
↵*** p < 0.01.

View popup

Table 4

Regency-Level Effects of Oil Palm Expansion on Sectoral Shares of Women (2000–2005–2010–2015)

	Share of Women Working (1)	Share of Women in Nonagricultural Sector (2)	Share of Women in Agricultural Family Labor (3)	Share of Women in Agricultural Wage Labor (4)	Share of Women in Nonagricultural Self-Employment (5)	Share of Women in Nonagricultural Wage Labor (6)
Instrumental variable
Share of smallholder oil palm area in regency (0–1)	−2.910^**	−0.162	−3.188^**	0.942	−0.639	0.438
Share of smallholder oil palm area in regency (0–1)	(1.184)	(0.630)	(1.284)	(0.626)	(0.454)	(0.286)
R-squared	0.155	0.402	0.106	0.072	0.115	0.558
Kleibergen Wald F-statistic	23.532	23.532	23.532	23.532	23.532	23.532
Observations	827	827	827	827	827	827
OLS model
Share of smallholder oil palm area in regency (0–1)	−0.731^**	0.339^***	−0.613^**	−0.011	0.186^**	0.114
Share of smallholder oil palm area in regency (0–1)	(0.348)	(0.129)	(0.301)	(0.190)	(0.093)	(0.075)
R-squared	0.235	0.409	0.216	0.116	0.185	0.564
Observations	827	827	827	827	827	827

Note: The data sources are SAKERNAS and Tree Crop Statistics. The dependent variables are shares ranging between 0 and 1. The IV and OLS estimates are reported with spatial HAC standard errors using a 100 km cutoff. The instrument is the maximum attainable oil palm yield per regency × the national oil palm expansion. We control for the mean age of working-age women, national oil palm expansion, regency fixed effects, year dummies, region trends, initial levels of population density, forest cover, hospital density, and electrification × the time trend. Initial levels are based on data from 2000.

* p < 0.10;
↵** p < 0.05;
↵*** p < 0.01.

View popup

Table 5

Regency-Level Effects of Oil Palm Expansion on Sectoral Shares of Men (2000–2005–2010–2015)

	Share of Men Working (1)	Share of Men in Nonagricultural Sector (2)	Share of Men in Agricultural Family Labor (3)	Share of Men in Agricultural Wage Labor (4)	Share of Men in Nonagricultural Self-Employment (5)	Share of Men in Nonagricultural Wage Labor (6)
Instrumental variable
Share of smallholder oil palm area in regency (0–1)	−0.372	−1.771^**	−1.053^*	2.686^***	−1.695^**	0.112
Share of smallholder oil palm area in regency (0–1)	(0.523)	(0.886)	(0.627)	(0.604)	(0.670)	(0.571)
R-squared	0.143	0.345	0.120	−0.009	−0.000	0.567
Kleibergen Wald F-statistic	23.312	23.312	23.312	23.312	23.312	23.312
Observations	827	827	827	827	827	827
OLS model
Share of smallholder oil palm area in regency (0–1)	−0.222^*	0.007	−0.401^***	0.672^***	−0.012	0.012
Share of smallholder oil palm area in regency (0–1)	(0.122)	(0.203)	(0.135)	(0.160)	(0.133)	(0.126)
R-squared	0.144	0.401	0.150	0.251	0.158	0.567
Observations	827	827	827	827	827	827

Note: Data sources are SAKERNAS and Tree Crops Statistics. Dependent variables are shares, ranging between 0 and 1. IV and OLS estimates are reported with spatial HAC standard errors using a 100 km cutoff. Instrument is the maximum attainable oil palm yield per regency times national oil palm expansion. We control for mean age of working-age men, national oil palm expansion, regency fixed effects, year dummies, region trends, and initial levels of population density, forest cover, hospital density, and electrification multiplied by time trend. Initial levels are based on data from 2000.

↵* p < 0.10;
↵** p < 0.05;
↵*** p < 0.01.

View popup

Table 6

Regency-Level Effects of Oil Palm Expansion in Land-Scare and Land-Abundant Settings (2000–2005–2010–2015)

	Share Working (1)	Share in Nonagricultural Sector (2)	Share in Agricultural Family Labor (3)	Share in Agricultural Wage Labor (4)	Share in Nonagricultural Self-Employment (5)	Share in Nonagricultural Wage Labor (6)
Men
Share of smallholder oil palm area in regency (0–1)	−0.296	0.589	−1.027^***	0.674	0.592^**	0.018
Share of smallholder oil palm area in regency (0–1)	(0.305)	(0.379)	(0.291)	(0.428)	(0.232)	(0.261)
Share of smallholder oil palm area in regency (0–1) × Share of forest cover in 2000 (0–1)	0.218	−1.713^*	1.841^***	−0.005	−1.777^***	−0.020
	(0.680)	(0.988)	(0.706)	(1.027)	(0.631)	(0.673)
R-squared	0.144	0.402	0.159	0.251	0.165	0.567
Observations	827	827	827	827	827	827
Women
Share of smallholder oil palm area in regency (0–1)	−1.777^***	0.108	−1.516^**	−0.452	0.276	−0.061
Share of smallholder oil palm area in regency (0–1)	(0.675)	(0.262)	(0.683)	(0.567)	(0.229)	(0.182)
Share of smallholder oil palm area in regency (0–1) × Share of forest cover in 2000 (0–1)	3.067^*	0.678	2.649^*	1.292	−0.264	0.512
	(1.606)	(0.599)	(1.524)	(1.307)	(0.495)	(0.437)
R-squared	0.241	0.409	0.220	0.119	0.185	0.564
Observations	827	827	827	827	827	827

Note: The data sources are SAKERNAS, Margono et al. (2014), and Tree Crops Statistics. The dependent variables are shares ranging between 0 and 1. The OLS estimates are reported with spatial HAC standard errors using a 100 km cutoff. We control for national oil palm expansion, regency fixed effects, year dummies, region trends and initial levels of population density, hospital density, and electrification × the time trend. The initial levels are based on data from 2000.