Displacement-induced inequalities in Koto Panjang resettlement villages of Indonesia

ABSTRACT

We evaluated the impact of the displacement project on land and income inequalities in Koto Panjang resettlement villages of Riau Province of Indonesia. The displaced households received equally compensated land before the project’s operation. We took that displacement as a treatment and conducted a with-and-without project comparison to the changes in the household economy, putting the non-displaced households as counterfactual. Using the 2003 and 2013 Indonesian Agricultural Household Income Survey, we found that the displaced areas had equally distributed land but unequally distributed income. In addition, both land and income levels were higher in the displaced than in the non-displaced.

KEYWORDS:

• Koto Panjang dam project
• inequality
• displacement
• land compensation
• impact evaluation

Introduction

Developments with displacements for dam constructions in Indonesia have happened since the 1980s. The dams aimed at meeting the increasing demand for electricity and irrigation in Indonesia. More than 70,000 people have been displaced by the construction of dams in Kedung Ombo, Saguling and Cirata (Nakayama, 1998; Nakayama et al., 1999; Picciotto et al., 2000; Sunardi et al., 2019). While dam constructions may be inevitable, the displacements threatened people’s livelihoods and many cases showed only a small fraction of resettlers recovered their livelihood (Andrianus, 2017; Scudder, 1997). Hence, as Biswas and Tortajada (2001) pointed out, despite the undoubtedly important role of dams, among the crucial issues facing the development of large dams is how best to minimize their adverse impacts and promote equity issues.

Cernea (1997) identifies eight impoverishment risks when people are forced to be displaced, including joblessness and landlessness. Two studies on impoverishment risks related to the Koto Panjang project showed conflicting results. JBIC’s evaluation of people’s livelihood conditions in 2004 found almost 70% of the displaced were worse off after resettlement (JBIC, 2004). This finding supported studies of Imhof (2003) and Walhi (2009), which claim the development of dams worldwide destroys traditional lifestyles and cultures. In contrast, Karimi et al. (2005) conducted a similar evaluation in Koto Panjang resettlement areas and found, in contrast, that the displaced households experienced an increase in income and produced more commodities after displacement. A similar better-off situation after resettlement also appeared in the New Tehri Town of India after the construction of the Tehri Hydro Power Project (Reddy, 2018).

Those studies produced conflicting results, possibly due to a lack of valid counterfactuals. Comparing people’s perceptions to indicate the change in living conditions before and after displacement as a measure of impact is subjective and dependent on emotional feeling about the project. Specifically, their method fails to account for welfare differences between those with and without the project. In fact, proponents and opponents of the project tend to use anecdotal information to support the dogmatic statements they wish to make (Biswas & Tortajada, 2001).

Our study avoided subjective judgement in conducting an impact evaluation of the displacement programme. We use the Indonesian Agricultural Household Income Survey, which contained household’s land ownership and annual income in 2003 and 2013, to compare inequalities between the displaced villages and their closest non-displaced counterparts as counterfactual. Hence, the novelty of our study lies in its ability to conduct a social impact evaluation of the Koto Panjang project with and without, as well as before and after the dam construction.

Location overview

The development of the Koto Panjang Dam was to meet the increase in the demand for electricity in Indonesia. Located in the middle of Sumatra, this project was expected to supply 114 MW of hydroelectric power, which was about 20% of electricity power in the area (JICA, 1984). Since the beginning of construction, the dam was expected to have a substantial social and environmental impact. Hence, JBIC extended Japan’s ODA loan under Indonesia’s commitment to pay sufficient attention to such impact (JBIC, 2004). The dam project, which commenced construction in 1992 and operation in 1998, displaced at least 4886 families from eight villages in Riau Province and two villages in West Sumatra Province, representing between 17,000 and 23,000 people (Imhof, 2003; JBIC, 2004). The resettlement process was commenced in 1993 and completed in 2000 mainly in the XIII Koto Kampar subdistrict of Kampar Regency (Fujikura & Nakayama, 2013; Karimi et al., 2009).

Table 1 shows the eight villages of Riau Province and the two villages of West Sumatra Province that were relocated into sixteen new resettlement villages in the neighbourhood of Koto Panjang. The relocation took place in 1993–1995, and based on the 2000 Indonesian Population Census, there were 5149 total households in the resettlement areas (JBIC, 2004). According to Karimi et al. (2009), each family was to be compensated with the following in their new villages: (a) 2 ha of rubber plantation, (b) 0.4 ha of land to grow food, (c) 0.1 ha of land for a 36 m² house with a courtyard, (d) living allowance for two years and (e) monetary compensation for all properties given up as a result of dam construction. Further, every village was to be provided with (a) a drinking water supply, (b) a school, (c) a health care centre, (d) a mosque and (e) a marketplace. There was a significant change in the primary source of income after resettlement, where the contribution of rubber plantations dropped from 60% before the resettlement to 20% after resettlement, which might affect the economic conditions at the new villages.

Table 1. Resettlement villages: relocation time and source of income.

Old village	New village	Year of relocation	Household (2000)	Primary source of income
				Before (%)	After (%)
A. Riau Province
Pulau Gadang	Pulau Gadang	1993	333	Rubber (72)	Rubber (70)
	Koto Masjid	1993	259	Rubber (77)	Rubber (53)
Batu Bersurat	Ranah Sungkai	1995	337	Rubber (84)	Rubber (22)
	Lubuk Agung	1995	200	Rice (42)	Rubber (90)
	Batu Basurat	1995	522	Rubber (46)	Others (n.a)
	Binamang	1995	178	Rubber (47)	Others (n.a)
Pongkai	Pongkai Baru	1994	200	Rubber (87)	Others (n.a)
	Mayang Pongkai	1994	259	Rubber (51)	Palm (94)
	Pongkai Istiqomah	1994	187	Rubber (90)	Others (n.a)
Tanjung Alai	Tanjung Alai	1994	313	Rubber (53)	Rubber (30)
Muara Takus	Muara Takus	1994	244	Rubber (87)	Rubber (18)
Koto Tuo	Koto Tuo	1994	599	Rubber (84)	Others (n.a)
Muara Mahat	Muara Mahat Baru	1994	477	Rubber (32)	Palm (73)
Gunung Bungsu	Gunung Bungsu	1995	241	Rice (83)	Rubber (54)
B. West Sumatra Province
Tanjung Pauh	Tanjung Pauh	1993	450	Rubber (55)	Rubber (16)
Tanjung Balik	Tanjung Balik	1993	350	Rubber (34)	Rubber (6)
Total			5149	Rubber (60)	Rubber (20)

Notes: Three major sources of income in the area are rubber plantations, rice production and palm plantations. ‘Others’ denotes a variety of sources of income, each less than 2%. According to Karimi et al. (2009), each family was to be compensated with the following in the new villages: (a) 2 ha of rubber plantation, (b) 0.4 ha of land to grow food, (c) 0.1 ha of land for a 36 m² house with a courtyard, (d) living allowance for two years and (e) monetary compensation for all properties given up as a result of dam construction. Further, every village was to be provided with (a) a drinking water supply, (b) a school, (c) a health care centre, (d) a mosque and (e) a marketplace.

Source: JBIC (2004) and Walhi (2009).

In 2001 the World Bank introduced the Operational Policy or Bank Procedure OP/BP 4.12 for Involuntary Resettlement, in which its objective was to improve the displaced livelihoods and standards of living or at least to restore them to pre-displacement levels or levels prevailing before the beginning of the project’s implementation, whichever is higher (World Bank, 2001). The implementation of the procedure, especially the compensation procedure, has been challenged by the displaced people both domestically and internationally (Cernea, 2003). Before the procedure took place, the first lawsuit in a local court in Indonesia was in June 1998 regarding project compensation, where ten households of Tanjung Balik village of West Sumatra appealed to the Tanjung-Pati District Court against PLN (the state electricity company). Another lawsuit was made by 67 families of Tanjung Pauh village of West Sumatra against the Home Secretary, Ministry of Agriculture and PLN. In 2000 and 2001, respectively, the judgements for these cases were made. Although the payment for parts of unpaid compensation was ordered, many of the claims were rejected (Sumi, 2004, as cited in Karimi et al., 2005).

After the procedure was signed, representatives of 3861 displaced people filed a lawsuit in the Tokyo District Court in September 2002 demanding that the Japanese government, JBIC, JICA and Tokyo Power Service Co. (TEPSCO) restore the affected rivers and compensate the displaced by as much as 5 million yen (about $42,000) per person (Imhof, 2003; Karimi et al., 2005). The lawsuit was filed due to inadequate prior consultation with the affected people about the impact of project implementation, a decrease in standard of living after relocation, the crisis of water supply at relocation sites, destruction of society and culture and a lousy impact on biodiversity.

Methods and data

We used the Theil, Atkinson and Gini indices to measure inequalities following Jenkins (1999) and Ridwan et al. (2018). For a population of households $\mathit{i}=1,2,3,\dots ,\mathit{n}$ each with land or income $y_i$ let $\mathit{m}$ be the arithmetic mean of income or land. The Theil index is given by $\mathit{T}=\frac{1}{n}\sum _{\mathit{i}=1}^n\left(\frac{y_i}{\mathit{m}}\right)\mathit{log}\left(\frac{y_i}{\mathit{m}}\right)$, the Atkinson index by $\mathit{A}=1-\frac{1}{\mathit{mn}}\sum _{\mathit{i}=1}^n\mathit{log}\left(y_i\right)$, and the Gini index by $\mathit{G}=1+\frac{1}{n}-\left(\frac{2}{m{n}^2}\right)\sum _{\mathit{i}=1}^n\left(\mathit{n}-\mathit{i}+1\right)y_i$. The Theil and Atkinson are decomposable into, within and between group inequalities but the Gini is not.

To conduct a ceteris paribus comparison, we treated all households in the XIII Koto Kampar subdistrict of Kampar Regency as the displaced households (the treatment group) and those in its closest neighbouring subdistrict of Kampar Kiri as their non-displaced counterparts (the control group). Figure 1 shows the location of the two neighbouring subdistricts. According to the 2000 Population Census, the XIII Koto Kampar and the Kampar Kiri were the first and the second largest subdistricts in the Kampar Regency with an area of 2319 km² and 2194 km², and a population density of 14 and 22, respectively. The census also showed that the average household size in the two subdistricts was both four persons per household (Central Bureau of Statistics, 2002). Therefore, we believe that these two subdistricts can be treated as an apple-to-apple comparison.

Figure 1. Location map.

Three maps with the following as a subset of the previous. The map of Indonesia showing Riau Province in Sumatra Island is followed by the map of Kampar Regency and the map of the subdistricts surrounding the Koto Panjang Dam Project.

To compare land and income inequalities we used the 2003 and 2013 Indonesian Agricultural Household Income Survey data sets, which were part of the National Agricultural Census (Central Bureau of Statistics, 2003, 2013a). The Government of Indonesia has carried out a national agricultural census since 1963 and the microdata access can be requested at https://silastik.bps.go.id/v3/index.php/site/login/. The 2013 Census was the sixth census themed around providing agricultural data for a better future for farmers. Each survey was cross-sectional, so we have two consecutive cross-sectional surveys to see the changes in occupied land and income in the non-displaced and the displaced areas. As a part of the agricultural census, samples were collected using stratified random sampling at each regency (Central Bureau of Statistics, 2013d). The data set contains information on occupied land and the income of farming households in all regions of Indonesia. The Central Bureau of Statistics (2013c) recorded that in 2013 Riau Province had 568,000 agricultural families, of which 76,000 were in Kampar Regency. Each agricultural household in Riau Province occupied 2.64 ha of land on average, and that in Kampar Regency was 2.40 ha.

Table 2 presents the statistics of relevant variables comparing the displaced (the treatment group) and the non-displaced (the control group). In the 2003 Indonesian Agricultural Household Income Survey data set, there were 193 household respondents, consisting of 84 households in the displaced areas and 109 households in the non-displaced areas. Whereas, in the 2013 Indonesian Agricultural Household Income Survey data set, there were 110 household respondents, consisting of 44 households in the displaced areas and 66 households in the non-displaced areas.

Table 2. Summary of statistics.

	2003			2013
	Non-displaced	Displaced	Non-displaced vs. displaced	Non-displaced	Displaced	Non-displaced vs. displaced
A. Baseline demographic
Female = 1	0.08	0.10	−0.13	0.11	0.09	0.02
			(0.04)			(0.06)
Years of schooling (year)	10.50	12.92	−2.92***	11.70	10.89	0.84
	[6.04]	[3.87]	(0.76)	[4.92]	[5.29]	(1.00)
Working experience (year)	30.35	23.37	6.98***	29.78	29.22	0.56
	[15.23]	[11.28]	(1.98)	[1.64]	[2.27]	(2.80)
Agribusiness as primary income = 1	0.90	0.81	0.09*	0.93	1.00	−0.08**
			(0.05)			(0.03)
Having non-agribusiness income = 1	0.09	0.40	−0.31***	0.17	0.07	0.10
			(0.06)			(0.06)
B. Observed outcome
Land owned (ha)	1.78	2.21	−0.44***	2.32	2.64	−3.24
	[0.73]	[1.17]	(0.14)	[2.25]	[2.28]	(4.41)
Annual income (Rp million – at 2003 price)	20.33	19.22	1.11	25.68	52.16	−26.48*
	[6.18]	[9.21]	(1.11)	[23.43]	[10.67]	(16.33)
Log (land)	9.67	9.89	−0.22***	9.69	9.77	−0.07
	[0.56]	[0.49]	(0.07)	[0.87]	[1.24]	(0.21)
Log (income)	9.88	9.79	0.09*	9.85	10.35	−0.50***
	[0.29]	[0.40]	(0.05)	[0.80]	[0.83]	[0.16]
Observations	109	84	193	66	44	110

Notes: Standard deviations (sd) are in brackets and standard errors (se) are in parenthesis. ***, **, and * denote significance at the 1%, 5%, and 10%, respectively.

Source: Authors’ calculations.

On a baseline demographic, we compared the households based on gender of the head, their years of schooling, their working experience, their primary source of income, and their non-agribusiness income. In 2003, four of the five baseline demographics were significantly different: the displaced areas had more schooling, fewer years of working experience, less dependence on agriculture and more non-agribusiness income than their non-displaced counterparts. In 2013, however, none of the five baseline demographics had a significant difference between the displaced and the non-displaced, except that the displaced had a slightly higher number of households stating agriculture as their primary source of income. This phenomenon is interesting since the displaced areas became more dependent on agriculture after resettlement. There must have existed occupational changes towards agribusiness after resettlement took place in the displaced areas.

Result and discussion

We focus our analysis on the land occupation and income of the households in the treatment and control areas. The statistical summary in Table 2 shows the land occupation and income of the displaced and non-displaced. The average land occupied by households was higher in the displaced areas compared to the non-displaced areas both in 2003 and 2013. The average land occupation in 2003 was about 2.21 ha in the displaced areas but only 1.78 ha in the non-displaced. In 2013, the land occupation increased to 2.64 ha in the displaced areas and 2.32 ha in the non-displaced. A possible source of increased land owned by the displaced was due to land compensation provided by the project to each displaced household. Moreover, as Karimi et al. (2005) cited, those households who had owned land and houses before the resettlement received additional compensation. Nevertheless, although land occupation was always higher in the displaced areas, the differences between the displaced and non-displaced were no longer statistically significant in 2013.

The household income in the displaced areas was initially lower than in the non-displaced, but after resettlement, the income in the displaced was higher. In 2003, the average income in the displaced areas was about Rp 19 million a year, and in the non-displaced was about Rp 20 million measured at the current price. Interestingly, in 2013 the average income of the displaced was more than double of that of the non-displaced. The average income in the displaced areas jumped to about Rp 52 million a year, while in the non-displaced areas was only about Rp 26 million a year measured at the constant price of 2003. Income differences at the level were insignificant in 2003 but statistically significant in 2013. In addition, income differences at the percentage level were significant both in 2003 and 2013. In 2003, the annual income was 0.09% lower in the displaced areas, but in 2013 it was 0.5% higher than in the non-displaced counterparts.

A possible source of rising income in the displaced areas was agricultural value added. The number of households that processed their agricultural products was considerably higher in the displaced than in the non-displaced: 2.7% in the displaced compared to only 1.1% in the non-displaced (Central Bureau of Statistics, 2013b). This figure was consistent with the occupational shift towards agricultural businesses for the displaced after resettlement. Should this trend persist, we would predict that while land occupation remained stable, income would be substantially higher in the displaced areas.

The finding of higher income in the displaced areas was supported by people’s perceptions about their current economic conditions contained in the data sets. Table 3 shows respondents’ perceptions about their current economic circumstances in the two observed years. In 2003, only 25% of respondents in the displaced areas revealed that their current economic condition was at least better than the previous year, compared to about 50% of respondents in the non-displaced. Conversely, in 2013, more than 18% of respondents in displaced areas were at least better than the previous year, compared to only less than 14% of respondents in non-displaced areas. In other words, while in 2003 fewer respondents in the displaced felt better off, in 2013 more respondents in the displaced revealed their economic conditions were better off compared to their non-displaced counterparts. Furthermore, in 2013, only 25% of respondents in the displaced, compared to more than 36% in the non-displaced, revealed that their current economic condition was worse off than the previous year. Thus, in general, the displaced households were better off in terms of economic conditions than the non-displaced. Our results confirmed the results from Karimi et al. (2005) who also found an increasing income after displacement for the Koto Panjang project. Their study revealed that in three of the four resettlement villages they surveyed, the majority of people had more income than before relocation. Only in one village did more people experience a decrease in income after resettlement.

Table 3. The distribution of households and current economic condition.

Condition	2003			2013
	Non-displaced	Displaced	Total	Non-displaced	Displaced	Total
Much better	2.75	1.19	2.07	1.52	0.00	0.91
Better	48.62	23.81	37.82	12.12	18.18	14.55
Same	28.44	75.00	48.70	50.00	56.82	52.73
Worse	17.43	0.00	9.84	30.30	20.45	26.36
Much worse	2.75	0.00	1.55	6.06	4.55	5.45
Totals	100.00	100.00	100.00	100.00	100.00	100.00
Observations	109	84	193	66	44	110

Note: The question asked was the perception of household about their current economic condition relative to that in the previous year.

Source: Authors’ calculations.

In terms of inequality, we measured the distribution of occupied land and income in the displaced and non-displaced areas for the two time periods, as shown in Figure 2. In 2003, both the distribution of land and the distribution of income were better in the non-displaced areas than in displaced areas, but in 2013 the distribution shifted. With this new equilibrium, the distribution of land was more equal, but that of income was more unequal in the displaced areas than in the non-displaced. The more equal land distribution in the displaced is understandable since the project compensated each household with an equal size of productive land before displacement took place. In fact, land equality should have been perfect in the displaced areas. Thus, since land equality in the displaced areas was imperfect, a transfer in land ownership among households within villages must have occurred.

Figure 2. Land and income inequality between displaced and non-displaced.

Two groups of figures, (a) contains Land Lorenz Curves and (b) contains Income Lorenz Curve. A1 and A2 are the Land Lorenz Curve of the displaced and the non-displaced in 2003 and 2013, respectively. B1 and B2 are the Income Lorenz Curve of the displaced and the non-displaced in 2003 and 2013, respectively. The displaced uses a broken dash line and the non-displaced uses a solid line. Source: Authors’ calculation.

In contrast to the distribution of land, the distribution of income was more unequal in the displaced than in the non-displaced areas in 2013, adding to the fact that the level of income was higher in the displaced. This condition implies that the dam project succeeded in improving the level of income but failed to promote income equality. Although each displaced household was compensated with an equal amount of productive land, the agricultural business opportunities might vary across locations. As indicated by Karimi and Taifur (2013), the alternative sources of income were more diverse in the displaced areas, especially in the successful resettlement villages.

Table 4 provides a more detailed picture of the inequalities. The three measurements confirmed that at the initial equilibrium in 2003 households’ land and income were more equally distributed in the non-displaced than in the displaced areas. The Gini index for land was 0.22 in the non-displaced, compared to 0.26 in the displaced, and the Gini index for income was 0.15 in the non-displaced but 0.21 in the displaced.

Table 4. Inequality measures.

Measurement	2003			2013
	Non-displaced	Displaced	Total	Non-displaced	Displaced	Total
Gini
Land	0.22	0.26	0.24	0.45	0.39	0.43
Income	0.15	0.21	0.18	0.42	0.54	0.50
Theil
Land	0.09	0.12	0.11	0.36	0.42	0.39
Within			0.10			0.39
Between			0.06			0.00
Income	0.04	0.08	0.06	0.30	0.51	0.45
Within			0.06			0.39
Between			0.00			0.06
Atkinson
Land	0.11	0.11	0.11	0.30	0.34	0.32
Within			0.11			0.32
Between			0.00			0.00
Income	0.04	0.08	0.06	0.26	0.40	0.36
Within			0.06			0.34
Between			0.00			0.03
Observations	109	84	193	66	44	110

Source: Authors’ calculations.

The distribution of land and income slightly changed at the new equilibrium in 2013. The land was more equally distributed in the displaced than in the non-displaced areas, opposite to that of its initial condition. In contrast, the income became more unequally distributed in the displaced, reinforcing that of its initial condition. The Gini index for land was only 0.39 in the displaced but 0.45 in the non-displaced. The Gini index for income was 0.54 in the displaced but only 0.42 in the non-displaced. Both Theil and Atkinson decompositions show that within-village inequality was the main source of land and income inequality. This decomposition result indicated that the cause of inequality was not in the physical disparities between the displaced and non-displaced but the heterogeneity among households within the villages, consistent with Ridwan et al. (2018).

Overall, the displacement project successfully increased households’ occupied land and the annual income of the displaced compared to the non-displaced. However, the project adversely affected the distribution of income. In other words, the Koto Panjang dam development boosted economic growth with the cost of raising inequalities in income. Unequal income distribution in the new villages was possibly rooted in a mismatch between the quality of compensated land and the skills of household members. Since, at the beginning of the resettlement period, the land was the only productive asset and was difficult to be converted into other forms of asset, some displaced households must have adjusted their source of income towards agricultural sectors, which finally helped them survive and prosper.

Reddy, Sarkar et al. (2022) provided an interesting observation on the existing land and income inequality in other developing countries. The economy of rural India was characterized by a high land inequality but a low-income inequality. This similar situation is related to our prior displacement period, where land inequality was higher in the displaced areas. Their study claimed that the low-income inequality in a high land inequality area was due to wide-open job opportunities in the non-land-based income. In our study, a possible cause of an increase in income and its inequality in the displaced areas was the open opportunity for an occupational shift towards agricultural sectors.

Robustness check

We conducted the variance ratio test to see whether the differences in land and income inequality were significant between the two groups in two periods, as presented in Table A1 in the supplemental data online. First, the variance differences in land ownership were significant in 2003 but insignificant in 2013. While in the beginning the land Gini index of the displaced areas was robustly higher than that of the non-displaced, the land inequality was proven to be convergence at a higher level magnitude. The interesting fact about land distribution is not that the non-displaced and the displaced had converged in land distribution but rather that the displaced who began with equal land distribution due to compensation, after a period of time, went to an unequal situation.

Second, the variance differences in income were significant in both 2003 and 2013. These results indicate that the higher income inequality in the displaced areas compared to the non-displaced was robust. The income inequality between the areas tended to diverge at a higher level magnitude. The starting point for income distribution in the displaced areas was that each household relied on a temporary instalment provided. The project then assumed the resettlers utilized the compensated land for their long-term source of living. Most likely, the economic challenges faced by some resettlers at their new location made them find a new and better source of living.

We check whether displacement causes an increase in economic outcomes. We regressed income and land on displacement status with and without controlling for demographic characteristics. We used the difference-in-difference (DD) method to check the impact of displacement on land owned and income earned (see Reddy et al. (2022) for a detailed explanation of the DD method).

$y_i={\beta }_0+{\beta }_1\mathit{Yea}{r}_i+{\beta }_2\mathit{Displace}{d}_i+{\beta }_3\mathit{Displaced}\times \mathit{Yea}{r}_i+\sum _{\mathit{j}=4}^7{\beta }_j{X}_{\mathit{ji}}+{\upsilon }_i$ where, for household $\mathit{i}=1,2,\dots \mathit{N},$ the dummy Year variable equals 1 for the year of 2013 and 0 otherwise; the dummy variable $\mathit{Displaced}$ equals 1 for the displaced household and 0 otherwise; and $X^{\prime }\mathit{s}$ are the baseline demographic controls as in Table 2. The variable $\mathit{y}$ is the outcome of interest: land owned and annual income. The coefficient of interest, ${\beta }_3,$ measures the impact of displacement on the outcome of interest.

The regression results in Table 5 show the impact of displacement on two variables: land owned and annual income. Regressions in columns (1) and (2) are those without and with demographic controls, respectively. The regression result in A shows that the displacement had no significant impact on land owned. Meanwhile, the result in B shows a significant impact of displacement on annual income, both without and with demographic controls. Displacement increased the average annual income by about Rp 27.8 million, confirming the findings from Karimi et al. (2005).

Table 5. DD regression results of displacement on land ownership and income.

	(1)	(2)
A. Dependent variable: land owned
Year	.5461*	.5368 **
	(.2425)	(.2441)
Displaced	.4408	.4742*
	(.2257)	(.2425)
Displaced × year	−.1166	−.1239
	(.3775)	(.3949)
Demographic control		X
R-square	0.04	0.06
B. Dependent variable: Annual Income (Rp million, 2003 = 100)
Year	.5350	.5258
	(.6601)	(.6704)
Displaced	−.1105	−.1218
	(.6144)	(.6658)
Displaced × year	27.588***	27.773 **
	(10.276)	(10.844)
Demographic control		X
R-square	0.07	0.07
Observations	303	303

Notes: Model (1) and (2) are without and with demographic controls, respectively. Annual income was measured in million Rupiahs at constant 2003 price. Demographic controls are female, years of schooling, years of working experience, agribusiness as the main source of income, and non-agribusiness income. Robust standard errors are reported in parentheses. ***, **, and * denote significance at the 1%, 5%, and 10% level, respectively.

Source: Authors’ calculations.

Table 6 summarizes our research findings, comparing land and income of the displaced and non-displaced households in 2003 and 2013. First, land owned remained higher in the displaced areas than in the non-displaced throughout the period; and the DD regression results confirmed no significant change in land owned due to displacement. Nevertheless, the displacement reduced land inequality. Second, annual income was initially lower in the displaced areas compared to their non-displaced counterparts but it was finally higher in the displaced. The DD regression results confirmed that the displacement increased annual income. However, the displacement also increased income inequality.

Table 6. Summary of findings.

Criteria	Displaced vs Non-displaced		Change^a
	2003	2013
Land owned	>	>	–
Land inequality	>	<	↓
Income earned	<	>	↑
Income inequality	>	>	↑

Notes: ^a Change reflects the impact of resettlement on the displaced, after taking into account the counterfactual of the non-displaced and the results from Table A1 in the supplemental data online and Table 5.

Conclusion

This study reveals two significant findings. First, after displacement, the displaced households, on average, owned more land than the non-displaced, and the land Gini indices were similar in both areas. A possible source of increased land owned by the displaced was due to land compensation provided by the project to each displaced household. In addition, the households who had owned land and houses before the resettlement received additional compensation. Nevertheless, given that the displaced households were equipped with a nearly perfect land distribution at the time of displacement, a transfer of land ownership must have occurred between villagers. Otherwise, the land inequality should have been much lower in the displaced areas.

Second, the displaced households earned better income than the non-displaced; and the income Gini indices were higher in their areas. Open opportunities to shift the occupational choice towards more profitable agricultural sectors after displacement might have contributed to the rise in income in the displaced areas. Given that all displaced households received, in principle, an equal amount of compensated land, this occupational shift most likely triggered not only higher income but also higher inequality. Therefore, policymakers need to be more aware of the income inequality results and put in place safety nets to ensure that the disadvantaged do not descend into poverty.

This study is limited to cross-sectional data in 2003 and 2013. Consequently, we have been constrained from analysing unobserved heterogeneity at the household level, which possibly contributed to the livelihood success. Future research should focus on longitudinal data to better pinpoint the impoverishment risks involuntary resettlers face.

Acknowledgments

An earlier version of this paper was presented at the First Economic, Law, Education, and Humanities International Conference (ELEHIC), held in Padang, Indonesia, on 14–15 August 2018, titled ‘Inequality and Economic Structure of the Displaced: A Household Study in Indonesian Koto Panjang Electric Dam Area’. We thank the Institute of Research and Community Service (LPPM) Universitas Andalas for the funding under Kontrak Penelitian Skim Klaster Guru Besar Number 10/UN.16.17/PP.RGB/LPPM/2018.

Supplementary material

Supplemental data for this article can be accessed at https://doi.org/10.1080/07900627.2024.2341268

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This work was supported by the Universitas Andalas [10/UN.16.17/PP.RGB/LPPM/2018].