Village economic opportunity, forest dependence, and rural livelihoods in East Kalimantan, Indonesia

World Development Vol. 33, No. 9, pp. 1419–1434, 2005� 2005 Elsevier Ltd. All rights reserved

Printed in Great Britain
0305-750X/$ - see front matter
doi:10.1016/j.worlddev.2004.10.006www.elsevier.com/locate/worlddev

Village Economic Opportunity, Forest Dependence,

and Rural Livelihoods in East Kalimantan,

Indonesia

SONYA DEWI, BRIAN BELCHER and ATIE PUNTODEWO *

Center for International Forestry Research, Indonesia

Summary.— The changing role of forests in people’s livelihoods in frontier areas is important fromthe perspective of poverty alleviation and forest conservation. This study explores the link betweenexpanding economic opportunities, forest dependence, and welfare in 73 villages. Village economicoptions, forest cover, and land suitability for agriculture and forestry are determining factors ofpeople’s well-being. Increased accessibility to markets and deforestation are strongly associatedwith economic diversity at the village level. Increased economic diversity, larger areas of forests,more intensive land use, higher endowments of agricultural land and forest, and higher village pop-ulation are related to increased well-being.
� 2005 Elsevier Ltd. All rights reserved.
Key words — Asia, Indonesia, spatial analysis, geographical targeting, income portfolio, ruralnonfarm employment

* This paper was developed from a poster presented at

the International Conference on Rural Livelihoods,

Forest, and Biodiversity, held in Bonn, Germany, in

2003. We thank the editorial committee for this invita-

tion. The paper has benefited from insightful and cons-

tructive comments and suggestions from Arild Angelsen,

William Sunderlin, Lini Wollenberg, Benoit Mertens,

and two anonymous reviewers. The research was sup-

ported by grants from the Canadian International De-

velopment Agency (CIDA) and CIFOR’s core donors.Final revision accepted: October 25, 2004.

1. INTRODUCTION

Increasing interest in rural poverty allevia-tion has resulted in a new focus on the ‘‘for-est-dependent poor’’ and on the actual andpotential contributions of forests to livelihoods(Angelsen & Wunder, 2003; Arnold, 2001;Dove, 1993; World Bank, 2001; Wunder,2001). Forest-rich areas are frequently associ-ated with rural poverty. Remote locations,rugged terrain, low population densities, limitedcommunications and transportation infrastruc-ture, infertile soils, and difficult climates act asconstraints that limit forest harvesting and theconversion of forests to agricultural land. Thesame factors limit the economic opportunitiesof people living in the area (Ashley & Maxell,2002; Wunder, 2001). Development in frontierareas in the form of resource extraction pro-jects, infrastructure development, or industrialdevelopment, for example, rapidly changesboth the physical and the economic landscapesand the corresponding opportunities for localpeople (Brookfield, Potter, & Byron, 1995;Padoch & Peluso, 1996). Analysis of the rela-tionship between such developments and rural

141

livelihoods is needed to better understand theirimpact and to target interventions aimed atimproving livelihoods in forest areas.Some argue that forest resources can be put

to work to help improve the livelihoods of thepoor (Scherr, White, & Kaimowitz, 2002),while others believe that forests have a limitedpotential to contribute to poverty reduction(Wunder, 2001). Part of the discrepancy be-tween these views originates in differentassumptions about the scope for creating newopportunities for rural people to take advan-tage of forest resources. Typically, the poor

9

1420 WORLD DEVELOPMENT

residents of forest-rich areas have not had avoice in deciding how to use forests and havelacked opportunities to transform forest re-sources into wealth for themselves. They tendto lack ownership and control over the re-sources, and do not have access to the marketsor the skills and contacts needed to take advan-tage of opportunities. And, while there is a glo-bal trend toward devolving natural resourcemanagement rights and responsibilities to localcommunities, this is usually in degraded forestareas and in many cases has resulted in in-creased state control (Edmunds & Wollenberg,2003). Governments are reluctant to give upcontrol over valuable revenue sources. Thereis a tendency to assign rights to well-connectedinvestors as in Sarawak (Cooke, 2002), Kali-mantan (Dove, 1993, 1996; Dove & Kammen,2001), and Thailand (Vandergeest & Peluso,1995). Decisions about forest use and invest-ments in forest areas are often determined byoutsiders interested in short-term extraction offorest resources, with little concern about thelong-term economic impact on local people.Large-scale resource extraction, forest conver-sion, and other development projects in forestareas tend to provide little direct benefit to for-est-dependent communities, and more com-monly have a negative impact as localpeoples’ access to resources is curtailed andthe physical resource base is degraded (Brook-field et al., 1995; Potter & Lee, 1998).On the other hand, development in forested

areas might be expected to create new opportu-nities for generating employment, income, andwealth. The World Bank (2002) identifies lim-ited land and market opportunities as a majorconstraint to poverty reduction. Road building,for example, increases people’s access to facili-ties and resources and also reduces transportcosts. Several studies in rural Latin Americafound that factors outside the agriculturalsector raised demand for nonfarm goods andservices and resulted in increased local incomesand the accumulation of capital for investment.Investment further increases rural nonfarmwages and self-employment through produc-tion and expenditure linkages (‘‘economictransformation’’) to reduce the incidence ofpoverty (Reardon, Berdegue, & Escobar,2001). Benefits arise in the form of improvedtransportation facilities, better market access,new markets for agricultural products, oppor-tunities for petty trading, and some employ-ment, at least in the short term (Brookfieldet al., 1995).

Efforts to encourage equitable and durableimprovements in human well-being need totake account of the direct and indirect liveli-hood impacts of resource-extraction projects,and the impacts of broader development ef-forts. In this paper, we examine these issuesas they have been played out in East Kaliman-tan, a forest-rich and fast-changing part of Bor-neo. Agriculture is the mainstay in the area,based primarily on a shifting cultivation systemin which regenerating forest is used to replenishsoils and control weeds, and the secondaryforest is managed for a variety of economicproducts (Lahjie, 1996; Sardjono & Samsoedin,2001). Forests are very important in local live-lihoods. Many valuable products are harvestedfrom the forest and rivers for direct consump-tion and for sale. We use the term ‘‘agro-forestry’’ throughout the paper to define thisintegrated system, in which households relyon agriculture and forestry both simultaneouslyand sequentially for their livelihoods.The main drivers of land use change in the

area are forest concessions, timber and estatecrop plantations, mining, road building by pro-jects and public investment, and migration(both government sponsored ‘‘transmigration’’and independent migration) (Brookfield et al.,1995; Padoch & Peluso, 1996). We investigatethe impact of forest-based and infrastructuredevelopments on economic opportunities andwelfare in the study area. We focus on the rela-tionships between village-level income diversity(as an indicator of economic opportunity) andvillage welfare. We look at the relationship ofthose characteristics to the location of indus-trial resource extraction projects, roads, andother economic features like forest cover,land-use suitability, and political and economiccenters. This allows an empirical test of the roleof these various factors in determining welfare.We use the village as the unit of analysis fortwo reasons: First, communities in the areatend to be small (mean = 150 households, butmany have fewer than 100) and relativelyhomogeneous, and the main drivers of changeoperate at a scale larger than a single commu-nity. The main differences in income opportuni-ties, welfare, and other factors of interest arefound across communities rather than withinthem. Second, for an analysis of spatial pat-terns, we need to cover an area large enoughto capture a range of stimuli. This approachis consistent with the geographic targeting ofsmall administrative regions advocated by Big-man and Fofack (2000). We supplement the

VILLAGE ECONOMIC OPPORTUNITY, FOREST DEPENDENCE, AND RURAL DEPENDENCE 1421

village-scale analysis with household level dataand case studies to explore how different de-grees of forest dependence at the household le-vel are reflected at the community level.

2. STUDY AREA

The study area includes the contiguous Kabu-patens (districts) of Paser and Kutai Barat inthe Indonesian province of East Kalimantan,covering 1,111,107 ha (Figure 1). The area isundergoing a rapid change. Previously heavilyforested, it has been deforested by large-scaleresource extraction, small-scale agriculturalexpansion, and two episodes of large-scale for-est fire, one in 1982 and the other after thestudy period, in 1997–98. Marked improvementin transportation and communication infra-structure in recent years, in-migration fromother areas of Indonesia, and new markets foragricultural and forest products and for laborhave led to important socioeconomic changes.There are clear gradients from villages thathave been strongly affected by development tovillages that remain remote and less affectedby modernization.The most important activities driving change

in the study area are logging (forest conces-sions), 1 timber plantations, 2 coal mining 3,oil-palm plantations, 4 smallholder rubberplantations 5, and transmigration projects, 6 aswell as road building 7 (Figure 1). All exceptsmallholder rubber planting and roads competedirectly with local people for resources by limit-ing access and/or by depleting or degrading theresource base. Some employment is generatedby these activities, but there has been a tendencyto hire from outside the local area. Local peopledo not tend to have the skills and experience or,according to some employers, the work ethic, tocompete for anything but low-paid jobs. Thereare projects to also develop roads, improve mar-ket access, and create new markets for locallyproduced supplies, and some services. Somecompanies are required by law to support com-munity development in their surrounding areas,although there are few demonstrable impacts.Table 1 summarizes information about eachproject in the study area.The boundary of the study area was defined

based on data availability. Cloud cover in satel-lite images is a key limiting factor in this high-rainfall area; no ‘‘cloud-free’’ images wereavailable. Projects located outside the bound-aries of the study area were included as influ-

ences in the analysis. These include a goldmine, a coal mine, an oil-palm plantation, arubber plantation, some transmigration sitesin West Kutai, and oil-palm plantations in thePaser district.

3. DATA

(a) Spatial data

Geo-referenced data were collected from avariety of sources, including (i) topographicand infrastructure maps produced by the Na-tional Coordinating Agency for Surveys andMapping in 1991; (ii) Land Systems and Suit-ability maps (Regional Physical Planning Pro-gram for Transmigration—RePPProt, 1982);and (iii) maps of company locations producedby the Ministries of Forestry and Transmigra-tion in 1997. Land-cover maps were producedusing unsupervised classification of LandsatTM images for 1992 and 1996, with three clas-ses: (i) mature forest; (ii) young secondary for-est; (iii) nonforest land. This classificationscheme allows an accurate and clear classifica-tion because each of the three classes has a dis-tinctive signature, and it fits well with ourdefinition of deforestation. We calculated thedeforestation rate as the area of mature forestand young secondary forest in period 1 thatchanged to nonforest in period 2, divided bytotal forest in period 1 per village area. Small-holder deforestation patches were distinguishedfrom others by using spatial characteristicsincluding size and location. In shifting cultiva-tion systems, the length of the fallow period iscommonly used as an indicator of land useintensity (Inoue, 2000). In this study, a higherproportion of new agricultural plots createdfrom young secondary forest (i.e., land thathad been cultivated too recently to permitmature forest to regenerate) was taken as anindicator of higher land use intensity. An indexof land use intensity was calculated as theproportion of agricultural plots cleared fromyoung secondary forest relative to total small-holder deforestation patches.A road map was developed by combining the

official national maps, Landsat TM 1992, andmaps produced by companies. Access to mar-kets and other features were calculated in termsof travel time. Accessibility was weighted by thequality of roads. Four classes of roads wereidentified: (i) the trans-Kalimantan highway,which is a surfaced road built and maintained

Figure 1. Map of study area showing villages, transportation network, and large-scale projects.


by the province to connect district capital citiesin Kalimantan; (ii) district roads, which are sur-faced roads, some of which were built in con-junction with transmigration projects; (iii)mining roads, built by mining companies,

which are unsurfaced roads but usually of agood quality to accommodate heavy equip-ments; and (iv) logging and plantation roads,which are low-quality roads built by compa-nies, and often not usable during the rainy sea-

Table 1. Summary of projects within the extent of the study area

Projects Area (ha) District Established References

Logging concessions 326,416 Paser and West Kutai Varies Integrated Forest FireManagement (IFFM) map

Oil-palm estates 13,141 Paser Mid-1980s IFFM mapTimber plantation 29,414 Paser and West Kutai 1992 IFFM mapCoal mine 15,998 Paser 1986 Kompas, August 28, 2001


son. Class (iv) roads have a shorter life, quicklybecoming unusable without regular mainte-nance.Officially, districts are subdivided into sub-

districts based on watershed boundaries. Vil-lage boundaries are based on local custom,are not recognized by the state, and are notmarked on any available map. We estimatedthese boundaries by drawing polygons throughmidpoints between neighboring villages usingthe Thiessen polygon technique. This roughlyapproximates the customary method of defin-ing boundaries, 8 and, judging from field obser-vations and interviews, it gave a reasonableestimate. Still, we needed to be cautious ininterpreting the results as incorrect boundariesinfluence several variables.

(b) Village-level socioeconomic dataand household survey information

Secondary, village-level socioeconomic datawere obtained from the Agricultural Censusof the Central Board of Statistics: Village Po-tential, 1993 9, the latest 10-yearly census dataavailable. These data were used to developindices of village-level economic diversity andwell-being and are described below.Primary data were collected using structured

interviews of household heads in randomly se-lected households covering between 15% and50% of households in each of 10 villages in1997–98 (Figure 1). The villages were selectedto represent a range of points along a gradientof distance to projects and other developments,stratified by district. Questionnaires were de-signed to capture information on demograph-ics, assets, income by source, and land uses.These data give us detailed village case studiesto compare with the main analysis based onthe Agricultural Census data. The comparabil-ity may be compromised by the five-year lag be-tween the Agricultural Census (1993) and thehousehold survey. However, we found a goodcorrespondence between similar data in thetwo data sets.

4. METHODS

(a) Indices of economic diversity andwell-being at the village level

We developed two indices of village eco-nomic diversity and peoples’ well-being at thevillage level.

(i) Village economic diversity index (EDI)Economic diversity at the village level is de-

fined in terms of heterogeneity of incomesources among households in a village. Inremote rural locations such as our case studyarea, where agriculture and forestry are the pre-dominant economic activities and where thereare few alternative opportunities, economicdiversity is low. It is hypothesized that develop-ment will lead to new employment opportuni-ties and therefore increased economic diversity.We developed a village EDI using the

Shannon–Weaver diversity index, which is ameasure used commonly in ecology andinformation theory. 10 The index is formulatedas

EDI ¼ �X

pi

lnðpiÞ;

where pi is the proportion of households in avillage that rely primarily on each main incomesource and i = 1, . . . ,n, where n is the numberof classes of main income sources. A householdis classified into one category based on its mainincome source and pi’s add up to one. The in-dex ranges from 0 to ln(n). In a village whereall households have the same main incomesource, EDI = 0. Where there is an even distri-bution of all possible main income sourcesamong households in a village, EDI = ln(n).In this case, with four possible main incomesources, perfect heterogeneity would give anEDI of ln(4). 11

The Agricultural Census of the IndonesianCentral Board of Statistics 1993 aggregatesmain income sources into four classes: agricul-ture, mining, wage, and services. This usefully


groups sources of income as farm, mining, andnonfarm wages. 12 It should be noted that theindex is sensitive to the number and definitionof classes used. In our study area, peopletypically integrate agriculture and forestryactivities in an indigenous agroforestry system,and we follow this in classifying these activitiesas one.Because agroforestry is the most common

main source of income, the EDI of a village willbe increased when some households derive theirmain income from sources other than agrofor-estry. The index is used as a measure of increas-ing economic opportunities in an area wheresuch opportunities are typically limited. A dif-ferent index would be needed in urban and otherareas with higher levels of diversity at time zero.

(ii) Well-being at the village level (VDI)The second index is a measure of people’s

well-being at the village level, based on theirhealth, education, and wealth relative to othervillages in the study area. It loosely followsthe idea of the UNDP Human DevelopmentIndex (HDI), which uses several socioeconomicindicators relating to income per capita, health,and education to compare average nationallevels of welfare. 13

The Village Development Index (VDI) is cal-culated as follows, based on the AgriculturalCensus Data:

Table 2. List

Variables

Economic diversity index Measure ofof househol

Village development index Measure ofin terms of

Forest cleared over total Proportionthe total cle

Population Village popuTime to district Distance (inTime to subdistrict Distance (inTime to logging Distance (in

to the closesTime to transmigration Distance (in

to the closesProvincial road, district road,mine road, logging/plantation road

Density of pand logging

Forest 92 ProportionDeforestation 92–96 Deforestatio

regrowth toAgrosuitability ProportionLand-use intensity Proportion

nonforest la

VDI ¼ ððh� hminÞ=ðhmax � hminÞþ ðe� eminÞ=ðemax � eminÞþ ða� aminÞ=ðamax � aminÞÞ=3;

where h is the infant survival rate; e is theschool enrollment rate for children of age 7–11 years; a is the assets (motor bikes andmotor boats) per household, 14 and subscriptsmax and min refer to the maximum and mini-mum values in the area. The minimum andmaximum possible values of VDI are 0 and 1,respectively. The index gives an equal weightto health, education, and wealth. For our data-set, Cronbach’s alpha reliability estimate ofVDI was 0.65. This is close to the commoncut-off point of 0.7 for demonstrating internalconsistency in the index data (Nunnaly, 1978).

(iii) Relationships between EDI and VDIVillage economic diversity, measured by

EDI, is expected to increase as new opportuni-ties are created by improved transportationinfrastructure, product markets, labor markets,and land and resource availability. (The last isaffected by deforestation.) The VDI is hypothe-sized to be determined by economic opportuni-ties, measured by the EDI, as well as by theavailability of agricultural and forest resources,and agricultural practices. Population density isalso expected to contribute to the VDI in this

of variables

Descriptions

heterogeneity of main income sourcesds in a villagedevelopment or well-being at the village levelhealth, education, and wealthof area cleared from forest overared area by smallholderslationhours) from the settlement to the district capitalhours) from the settlement to the subdistrict capitalhours) from the settlementt active logging companyhours) from the settlementt transmigration siterovincial road, district road, mine road,/plantation road (km/ha)of village area covered by forest in 1992n rate from 1992 to 1996 (changes from forest andnonforest land divided by forest cover area in 1992)of suitable area for agroforestry over total areaof area of young secondary forest in 1992 changed tond in 1996 over total smallholder deforestation

Table 3. Spatial error model of village EDIa

Variable Coefficient z-Value

Provincial road 206.31 5.17**

District road 54.13 3.41**

Mine road �8.65 �0.41Logging/plantation road 23.16 2.11*

Time to logging 4.10E�03 0.43Time to transmigration 0.02 3.35**

Deforestation 1992/1996 2.27 3.55**

Lambda �0.38 �2.79**

a With maximum likelihood estimation, number ofobservation = 73, R2 = 0.82, LIK = 5.62, AIC = 2.76.*Significance level of 0.05.**Significance level of 0.01.


remote area, through increasing economicinteractions. We also included proximity tothe district and subdistrict capital as this vari-able measures access to the local center of tradeand services, and also to the government ser-vices such as health care, education, and infra-structure development.

(b) Spatial statistical analyses of EDI and VDI

We explore the empirical relationships: (i) be-tween village economic diversity and distancesfrom projects and developments and (ii) amongwell-being at the village level, village economicdiversity, and other factors using regressionanalysis of spatial data and village-level socio-economic data (Table 2). We use Space Stat,a software designed to build spatial economet-ric models, in conjunction with ArcView, aGIS software. Space Stat provides a set of diag-nostic tools to determine an appropriate modelto handle a particular form of spatial depen-dence when it exists (Anselin, 1999). In orderto incorporate spatial dependence in the ana-lysis, a neighborhood matrix is generated byconsidering villages to be connected when theyshare boundaries and a transportation link(either road or river).We further explored the links between EDI/

VDI and the role of forests in livelihoods usingthe household survey data from 10 villages.We calculated the proportion of cash incomeearned by each household from agroforestryproducts (including rattan, honey, fruit, game,fish, other forest products, and rice) for eachsample village. Households that earn more thanhalf of their cash income from agroforestryactivities are classified as agroforestry-depen-dent households. Village-level agroforestrydependence is then measured as the proportionof village households that are agroforestrydependent. The income portfolio for three se-lected villages is described to illustrate thebreadth of income-earning activities.

5. RESULTS AND DISCUSSION

(a) Factors associated with village EDI

Spatial statistical analysis was run on thesubset of variables that gives the best-fit spatialerror model (Table 3), with EDI as the depen-dent variable. Figure 2 shows the map ofEDI. The distances to plantations (timber, rub-ber, and oil palm), to mines and to transmi-

gration sites are strongly and positively corre-lated with one another. Therefore, we includedonly one variable—distance to transmigrationsite—in the analysis to avoid a collinearityproblem. This collinearity reflects the deliberatedevelopment of particular economic activitieswithin transmigration projects. In the studyarea, the main transmigration area was devel-oped in conjunction with an oil-palm planta-tion and processing factory and, to a lesserextent, with timber plantation and rubber pro-jects. Transmigrants were provided with an oil-palm plantation to manage, and the oil-palmcompany has actively encouraged neighboringvillages to cultivate oil palm in the so-called‘‘plasma’’ area (the area around the ‘‘nucleusestate’’).The best-fit spatial error model suggests that

unexplained factors between neighboring vil-lages are strongly correlated. Table 3 showsthat better road access increases EDI. The pro-vincial road gives increased opportunities forincome and employment, presumably by link-ing villages to the broad network of marketsfor different products and labor, improvingaccess to information and facilitating in- andout-migration. Villages connected by provincialroads show high values of economic diversity.District roads have a similar but less pro-nounced effect. This was expected, as accessibil-ity, in terms of the quality of the road and theplaces it connects, is less than that providedby the provincial road. The finding that roadsare associated with a higher village economicdiversity is consistent with findings from LatinAmerica in which access to road infrastructureand closeness to towns are shown to be robustdeterminants of rural nonfarm employmentand incomes (Reardon et al., 2001).

Figure 2. Village economic diversity index map.


Logging/plantation roads also lead to a higherEDI, but the effect is smaller than that of otherroad classes. This class of road typically pro-vides access to smaller markets, both for prod-ucts and labor, and mainly to markets directlyrelated to the activities of the resource extrac-tion company and its employees. Mining roadsdo not show statistically significant relation-ships with EDI, which can be explained bythe fact that local people were not permittedto use them during the study period.While logging/plantation roads are associ-

ated with a higher village economic diversity,

proximity to logging companies does not affectEDI. This implies that these projects do notprovide sufficient economic opportunities forlocal people to adopt them as a main sourceof income. This is consistent with reports thatin Borneo most labor for logging companieswas hired from outside the area (Brookfieldet al., 1995). Logging roads are not perfectlycorrelated with logging companies because theroads may remain after the company (and asso-ciated economic activities) has left the area.The EDI correlates positively with deforesta-

tion (including conversion to agriculture and to

Figure 3. Village development index map.


other activities) during the 1992–96 period.This may indicate that some activities that re-sulted in deforestation also provided nonfarmincome opportunities for local people.

(b) Well-being at the village level (VDI)

The map of VDI is presented in Figure 3,and the best-fit spatial error model of VDI is

Table 4. Spatial error model of VDIa

Variable Coefficient z-Value

Economic diversity index 0.19 3.70**

Agrosuitability 0.27 7.86**

Land-use intensity 0.15 2.39*

Forest 92 0.53 6.79**

Population 8.87E�05 5.39**

Time to district capital �0.03 �1.13Time to subdistrict capital 0.01 0.38Lambda �0.44 �3.30**

a With maximum likelihood estimation, number ofobservation = 73, R2 = 0.98, LIK = 63.65, AIC =�113.31.*Significance level of 0.05.**Significance level of 0.01.


presented in Table 4. A higher village economicdiversity leads to a higher VDI. In the casestudy area, increases in EDI generally meanthat a higher proportion of households in a vil-lage rely on the nonagricultural sector for theirmain source of income. The proportion of landcleared by smallholders from land that was ayoung secondary forest (relative to the amountcleared from mature forests) is positively corre-lated with VDI. This is mostly related to theavailabilities of fertilizer and herbicide and landpressure, which are both driven by market andeconomic opportunities.A larger extent of land suitable for agrofor-

estry practices leads to a higher VDI. That is,the traditional system provides a relativelygood living if there is an adequate resourcebase. A large proportion of village land withforest cover, measured at the beginning of thestudy period, has an even larger positive coeffi-cient, showing that the forest has contributed tohigher welfare.While, on the one hand, a higher initial

forest cover and land quality for agroforestryare associated with higher welfare, on theother hand a higher EDI (shift toward lessagroforestry-dependent communities) andhigher land use intensities are also associatedwith a higher welfare. The role of forest inthe EDI–VDI context is intriguing. Whiledeforestation correlates positively with EDIand with VDI, a larger proportion of villagearea initially covered by forests also increasesVDI. This shows two different possible devel-opment paths. The EDI may increase as theforest cover declines, especially where suitableland for agroforestry is limited. Or agrofor-estry can serve as the main source of liveli-hood while maintaining a high forest cover

when the village has sufficient suitable land.As discussed earlier, opportunities for locallivelihoods to improve in the ‘‘poverty elimi-nation’’ sense (see Sunderlin et al., this vol-ume) based on agriculture and forestry havebeen limited, but appropriate interventions atthe project and policy level might help (e.g.,Scherr et al., 2002).Given the fact that the area has a low popu-

lation density overall (mean = 13.95 persons/sq. km), it is not surprising to find that the pop-ulation has a highly significant positive coeffi-cient. Small isolated villages are unable togenerate economic dynamism.

(c) Agroforest dependencies/role of forestsin village samples

For the 10 village samples (Figure 1), theEDI (from the census data) tends to increaseand then decrease as agroforest dependence(from aggregated household survey data)decreases (R2 = 0.84). On the other hand, theproportion of village area initially covered byforests seems to be the major factor determin-ing the proportion of households that areprimarily dependent on forest products andagroforestry with a very strong, quadraticrelationship (R2 = 0.93) (Figure 4).Agroforest dependence shows a weak qua-

dratic relationship with VDI. High agroforestdependence is associated with a relatively highVDI. The VDI declines as agroforest depen-dence declines, but then they increase together,with the least agroforest-dependent villageshaving the highest VDI. The two villages withhigh agroforest dependence and a high well-being index have large areas of good qualityforests, but there are obviously other variablesinvolved in determining people’s well-being atthe village level (Figure 5).

(d) Agroforest dependence case studies

Modang, Muser, and Legai, the three villagesin the lower end of the agroforest dependenceaxis have a low forest cover, but a high EDIrank. At the other extreme, Besiq and RantauLayung villages have a high forest cover anda high dependence on agroforestry and a lowEDI. Sample villages with an intermediate levelof agroforest dependence and forest cover di-verge in their VDIs such as Kendisiq, MuaraSiram, and Muara Lambakan have a lowVDI while Kasungai and Muara Nilik have ahigher VDI.

Proportion of forest cover in 19921.0.9.8.7.6.5.4

Agro

-fore

st d

epen

denc

ies

1.2

1.0

.8

.6

.4

.2

R2 = 0.93

R Layung

Muser

Modang

M Siram

M Nilik

M Lambakan

Legai

Kendisiq

Kasungai

Besiq

Figure 4. Scatterplot of proportion of forest cover in 1992 and agroforest dependencies.

Agro-forest dependencies1.21.0.8.6.4.2

Villa

ge D

evel

opm

ent I

ndex

.70

.68

.66

.64

.62

.60

.58

.56

.54

.52

R Layung

MuserModang

M Siram

M Nilik

M Lambakan

Legai

Kendisiq

Kasungai Besiq

R2=0.29

Figure 5. Scatterplot of agroforest dependencies and village development index.


We can illustrate this with the examples ofthe villages of Legai, Muara Lambakan, andBesiq (Figure 6). Legai represents a village witha high EDI, low forest cover, and a high VDI.It is located in the Paser district, very close tothe trans-Kalimantan highway and to the sub-

district capital. We observed some intensifiedagricultural practices here with the use of herbi-cide, pesticide, and short fallow periods. Arti-sanal gold mining is the primary source ofincome, followed by agroforestry (rattan, cof-fee, rubber, and other crops), nonfarm income

Figure 6. Mean household income portfolio in three villages.


(trading, wage labor, services), and forest prod-ucts (wild rattan, logs).Muara Lambakan, also in the Paser district,

is a remote village with no industries nearby,

but with a history of several episodes of loggingby different logging companies. Most of the vil-lage area is unsuitable for intensive farming dueto thin topsoil and a rough terrain. This village


represents the group with a moderate agrofor-est dependence, moderate forest cover, lowEDI, and a low VDI. The largest proportionof income is from forest-based activities andthe largest part of that comes from game andfish. Income from farming is slightly lower thanthat from forest. Off-farm income as rattanworkers in a few large gardens in the same vil-lage comprises a substantial proportion ofincome. Nonfarm income contributes almost20% of the total cash income.Besiq village, in West Kutai, has a large area

of forest, high agroforest dependence, low EDI,and a relatively high VDI. The area was re-mote, with no road access 15 and river accesslimited during the dry season. Forest products,including rattan, rubber, game, fish, and honey,provide the main source of cash income. Rattanagroforest and domestic animals contribute asmall proportion of total cash income, followedby nonfarm income, which adds very little tothe total cash income.

6. CONCLUSIONS AND IMPLICATIONS

The indices developed and the analysesconducted have proven useful to explain rela-tionships between various driving forces andvillage welfare and economic diversity. The ap-proach could also be used as a tool for sce-nario-based land use planning at the districtlevel. Poor data availability was the main con-straint we faced in using this approach. Withactual village boundaries rather than approxi-mations, the analysis could have been moreaccurate. Cloud-free satellite images would givea more comprehensive data set to work with.Also, more frequently collected secondary datafrom government statistics would allow the useof time-series analysis in conjunction with thespatial analysis to cover both temporal andspatial dynamics.Under the conditions prevailing during the

study period, the local people were highly dis-advantaged. In general, health service and edu-cation quality were generally low, and the localpeople were not involved in major resourceextraction decisions. They had little opportu-nity to benefit from the rich natural resourcesin the areas where they lived. This may changewith more decentralized governance (since2001), but it is too early to assess the impactof that. As forest resources have been depleted,people have turned to alternative activities. Avariety of externally controlled projects (mainly

resource exploitation projects, but also transmi-gration) created different opportunities andconstraints for local people.The analysis shows that large-scale commer-

cial logging did not benefit the local people.Mines, oil-palm plantations, and transmigra-tion projects created some new markets for lo-cal products that replace traditional activities,but these projects do not show trickle-down ef-fects outside the agricultural sector exceptthrough improvement in the transportationnetwork. These projects often focus on onemain economic activity such that in the areasaround palm-oil estates and transmigrationareas, there is a low economic diversity.Roads are very important to village eco-

nomic diversity, with a higher quality, better-connected roads being associated with a higherdiversity. Mining roads are the exception.They do not result in local economic diversitybecause they are not open for public access inthe study area. Forest conversion to otheruses is generally associated with increased vil-lage economic diversity, again with bothpositive (pull) and negative (push) factorsinvolved.Generally speaking, well-being, as measured

by education, health, and assets, increases withmore village economic diversity. However, vil-lage economic diversity is not the single mostimportant determinant of well-being. Higherlevels of forest resources and suitable land foragroforestry are also associated with higherwelfare. Relatively remote, well endowed forestvillages with limited economic alternativesshow a high well-being relative to other villagesin the area being studied. A good forest endow-ment allows people to live well at or near thesubsistence level. But opportunities for forest-based poverty elimination, in the sense of liftingpeople safely out of poverty, have been limited.The worst-off villages are those with poor re-source endowments and limited alternative in-come-earning opportunities.But the future need not mirror the past. Rich

resource endowments could be used in otherways, with more benefits accruing to localpeople. Forest and agricultural livelihoods canpotentially support relatively high levels ofwelfare. Improving local people’s access toresources in their vicinity and their capacityto transform them is critical for enabling themto attain better health, education, and otherwell-being improvements. More investment ininfrastructure, markets, and other factors thatsupport village economic diversity should be


encouraged. In areas where land suitable foragriculture and forestry are less available,

increasing economic opportunities is even moreimportant.

NOTES

1. Logging roads make areas of forest/land more easilyaccessible and reduce transportation and marketingcosts for local products (Dove, 1996). Logging campscreate temporary markets for products and labor,though it is a common practice in the study area tohire workers from outside areas. Logging activities alsocompete for forest resources with local people; typically,local people have been prohibited from entering conces-sion areas.

2. State policy to assign ‘‘degraded lands’’ for large-scale timber plantations often targets (deliberately ornot) managed forest gardens and secondary forests thatare part of local peoples’ agroforestry systems, resultingin a direct competition for land and, often, conflict.These plantations were usually operated in conjunctionwith logging concessions, and the plantation concessionswere used as ‘‘licenses’’ to clear-cut in the guise ofclearing for planting (see, e.g., Manurung, 2001). Ben-efits for local people are similar to those available fromlogging concessions. Roads are built, which leads to theemployment of some villagers mainly as nonskilledworkers, and temporary markets are created for localproduce (Brookfield et al., 1995). In our study area,timber plantations were established in conjunction withthe transmigration program.

3. The study area has rich coal deposits. Mining inIndonesia has been strictly controlled by the centralgovernment, with a top-down approach that has beencriticized for its negative impacts on the local people.Conflicts between mining companies and local peopleare common (Down to Earth, 2001; World Bank,2001). Mining competes for access to forest lands, andprohibitions against trespassing are strictly enforced.While mining concessions may be smaller than forestconcessions, the forest cannot be expected to recoverafter the end of mining activities. Roads built bymining companies are better made and bettermaintained than logging roads, and markets forlabor, supplies, and services are larger and morelong-lasting than those created by logging and timberplantations.

4. As with timber plantations, development of oil-palmestates implies a direct competition for land with small-scale farmers (Manurung, 2000). Positive contributionsto the area include road construction, development ofprocessing facilities, and creations of markets.

5. The main rubber producers in Indonesia are small-holders. There have been attempts to support the sectorby providing planting materials, fertilizer, pesticides,equipment, and land titles, as well as cash payments tocover subsistence expenses during the period before thetrees mature. Two important projects in the vicinity ofthe study area were the Tree Crop Smallholder Devel-opment Project funded by the World Bank (IBRD),started in 1974, and by the Asian Development Bank,started in 1992 (Budiman, 1999).

6. Transmigrants were typically settled in purpose-builtcommunities, often on lands claimed by local people.This resulted in competition and some conflict for forestresources and land between transmigrants and the localpeople. As part of these projects, new roads, as well ashousing, schools, and health facilities were built. Sometransmigration projects in the study area were developedin conjunction with timber plantations, oil-palm plan-tations, and processing factories. The projects creatednew markets for various products and for labor. Also,there is usually some technology transfer as in-migrantsbring different ideas and approaches that may beadopted by local people.

7. The Trans-Kalimantan Highway, the most signifi-cant road development in the area, was built in the mid-1980s by the provincial government to connect majorcities in Kalimantan. Many of the transmigration roadswere built around the same time. Project-related roadshave been built (and abandoned) as and when needed forcompanies’ operational purposes such that from thevillagers’ point of view, they only provide a temporaryaccess to resources and sometimes to the market. Thehistory of road building in the area is not welldocumented.

8. Local village heads reported that boundaries weredefined by agreement, using natural landmarks close to amidpoint that fairly divided resources between twoneighboring villages.

9. Village Potential data include human capital(demography, workforce), natural captial (forest, mine,wetlands, etc.), social captial (institutions, networks),and produced capital (health facilities, school buildings,infrastructure, mills, factories, etc.). These data werecollected by village administrative officials and thencompiled by the Central Board of Statistics (BPS)


(Suhariyanto, K. pers. com. 2004). The data were cross-checked by our enumerators with several key informantsand were found to be consistent.

10. The Shannon–Weaver index measures diversity in away very similar to that of the Simpson index(Baumgartner, 2002). Using the Shannon–Weaver index,our data are normally distributed, while computationusing the Simpson index shows a tendency to departfrom the normal distribution, that is, with insensitivity inthe lower diversity.

11. This index is very different from the diversity indexproposed by Chang (1997), which measures diversity ofhousehold income portfolios and its contribution to thetotal, and from diversification indices presented inWagner and Deller (1998), which measure economicdiversity and industrial linkages at the state level in theUnited States. This index does not capture diversifica-tion at the household level, but rather at the aggregatelevel, in this case, the village. Therefore, our EDI doesnot link explicitly with the household income portfolioor household economic diversification that is oftendiscussed in the livelihood literature (e.g., Ellis, 2000;Reardon et al., 2001). Using the index as a measure ofeconomic diversity one assumes that improved oppor-tunities generated by external developments and thecapabilities of people to take advantage of thoseopportunities reduce the dominance of agriculture inpeople’s livelihoods at the village level.

12. The BPS’s 10 income categories are agriculture(fishery, crop, cattle ranching, fish farming, forestry),mining (including sand and rock), wage earning, craftsand home industries, electricity and water, construction,trading, transport, finance, services, and others. Exceptfor agriculture, mining and wage earning, the last eightcategories were aggregated into one class called ‘‘ser-vice’’ for the purpose of our analysis. This was donebecause the data of our study area showed low

frequencies in most of those eight categories. For ourstudy area, ‘‘wage labor’’ includes labor in construction,rattan harvesting, oil-palm harvesting, rubber tapping,sawmilling, logging, etc.

13. The HDI has been criticized on several counts. Inparticular, Hicks (1997) has pointed out the problem ofdistributional inequality within countries, which isoverlooked in the HDI. We share Hicks’s concern aboutdistribution and the need for disaggregating the popu-lation. However, as we are using much smaller units ofanalysis (small villages as opposed to national popula-tions), inequalities are much less pronounced. Our VDIand the HDI are not comparable as they are measured atdifferent scales with different data for different purposes.

14. We use the number of motorcycles and motor boatsper capita as a proxy for wealth. Barham, Takasaki, andCoomes (1999) argue that the wealth of households intropical rainforest areas is best measured by assets thatare related to livelihood strategies. In their study, keyassets are land, productive capital, and nonproductiveassets that together provide a basis for producingsubsistence and cash income, a buffer against bad times,and a better life style. In our study area, boats andmotorcycles perform all three functions. People rely onprivate transport to travel to forests and farms for theirproductive activities, to transport their products to themarket and at times to generate income throughproviding transport services. People will buy an extramotorcycle or a boat in good times and sell during badtimes, using these goods, in effect, as a ‘‘savingsaccount.’’ Motorcycles and boats also have a role in‘‘conspicuous consumption,’’ as a common way todemonstrate wealth. There are other assets that wethink should be important in measuring wealth (e.g.,chainsaws), but these are not included in the datacollected by BPS.

15. A road was opened after the study period.

REFERENCES

Angelsen, A., & Wunder, S. (2003). Exploring thepoverty-forest link: Key concepts, issues and researchimplications. Occasional Paper No. 40. Center forInternational Forestry Research, Bogor.

Anselin, L. (1999). SpaceStat Version 1.80 User Guide,Urbana Champaign.

Arnold, M. (2001). Forestry, poverty and aid. OccasionalPaper No. 33(E). Center for International ForestryResearch, Bogor.

Ashley, C., & Maxell, S. (2002). Rethinking ruraldevelopment. Development Policy Review, 19(4).

Barham, B. L., Takasaki, Y., & Coomes, O. T. (1999).Rain forest livelihoods: income generation, house-hold wealth and forest use. Unasylva, 198.

Baumgartner, S. (2002). Measuring the diversity of what?And for what purpose? A conceptual comparisonof ecological and economic measures of biodiversity.

Bigman, D., & Fofack, H. (2000). Introduction andoverview. In D. Bigman & H. Fofack (Eds.),Geographical targeting for poverty alleviation: Meth-odology and applications. Washington, DC: TheWorld Bank.

Brookfield, H., Potter, L., & Byron, Y. (1995). In placeof the forest: Environmental and socioeconomic trans-formation in Borneo and the Eastern Malay Peninsula.Tokyo: The United Nations University.

Budiman, A. F. S. (1999). Partnership of rubberproducer association with smallholders. Paper


presented at the Agricultural Producer Organizations:Their Contribution to Rural Capacity Building andPoverty Reduction, Washington, DC.

Cooke, M. F. (2002). Vulnerability, control and oil palmin Sarawak: Globalization and a new era? Develop-ment and Change, 33(2), 189–211.

Dove, M. R. (1993). A revisionist view of tropicaldeforestation and development. Environment Conser-vation, 20.

Dove, M. R. (1996). So far from power, so near to theforest: A structural analysis of gain and blame intropical forest development. In C. Padoch & N. L.Peluso (Eds.), Borneo in transition: People, forest,conservation, and development (pp. 41–58). KualaLumpur: Oxford University Press.

Dove, M. R., & Kammen, D. M. (2001). Vernacularmodels of development: An analysis of Indonesiaunder the New Order. World Development, 29(4),619–639.

Down to Earth (2001). Communities and companies.Down to Earth Newsletter, 50(August).

Edmunds, D., & Wollenberg, E. (Eds.). (2003). Localforest management. London: Earthscan Publications.

Ellis, F. (2000). Rural livelihoods and diversity in devel-oping countries. New York: Oxford University Press.

Inoue, M. (2000). Mechanism of changes in the Ken-yah’s swidden system: Explanation in terms ofagricultural intensification theory. In E. Guhardja,M. Fatawi, M. Sutisna, T. Mori, & S. Ohta (Eds.),Rainforest ecosystems of East Kalimantan: El Nino,drought, fire and human impacts. Tokyo: Springer-Verlag.

Lahjie, A. M. (1996). Traditional land use and KenyahDayak farming systems in East Kalimantan. In C.Padoch & N. L. Peluso (Eds.), Borneo in transition:People, forests, conservation, and development(pp. 150–161). New York: Oxford University Press.

Manurung, T. (2000). Pembangunan perkebunan kelapasawit di Indonesia: Ancaman terhadap hutan alam.Warta Fahutan Online, 1–6.

Manurung, T. (2001). Potret pembangunan hutan tana-man industri di Indonesia. Paper presented at the

Seminar Nasional: Hutan dan Lingkungan, Tema:‘‘Era Menanam: Membangun Kelestarian Hutan danLingkungan’’. Auditorium Rektorat IPB, KampusDarmaga, Bogor.

Nunnaly, J. C. (1978). Psychometric theory. New York:McGraw-Hill.

Padoch, C., & Peluso, N. L. (Eds.). (1996). Borneo intransition: People, forests, conservation, and develop-ment. New York: Oxford University Press.

Potter, L., & Lee, J. (1998). Tree planting in Indonesia:Trends, impacts and directions. Occasional Paper,18, 1–76.

Reardon, T., Berdegue, J. A., & Escobar, G. (2001).Rural nonfarm employment and incomes in LatinAmerica: Overview and policy implications. WorldDevelopment, 29(3), 395–409.

Sardjono, M. A., & Samsoedin, I. (2001). Traditionalknowledge and practice of biodiversity conservation:The Benuaq Dayak community of East KalimantanIndonesia. In C. J. P. Colfer & Y. Byron (Eds.),People managing forests: The links between humanwell-being and sustainability. Resources for theFuture and Center for International Forestry Re-search (CIFOR).

Scherr, S. J., White, A., & Kaimowitz, D. (2002).Making markets work for forest communities. Wash-ington, DC: Forest Trends and CIFOR.

Sunderlin, W. D., Angelsen, A., Belcher, B., Burgers, P.,Nasi, R., Santoso, L., et al. (this volume). Liveli-hoods, forests and conservation in developing coun-tries: An overview. World Development.

Vandergeest, P., & Peluso, N. (1995). Territorializa-tion and state power in Thailand. Theory andSociety, 24.

World Bank (2001). Indonesia: Environment and naturalresource management in a time of transition. Wash-ington, DC: The World Bank.

World Bank (2002). A sourcebook for poverty reductionstrategies: vol. 1. Washington DC: World Bank.

Wunder, S. (2001). Poverty alleviation and tropicalforests: What scope for synergies. World Develop-ment, 29(11), 1817–1833.

Documents

Village economic opportunity, forest dependence, and rural livelihoods in East Kalimantan, Indonesia