Assessing poverty–deforestation links: Evidence from Swat, Pakistan

This article appeared in a journal published by Elsevier. The attachedcopy is furnished to the author for internal non-commercial researchand education use, including for instruction at the authors institution

and sharing with colleagues.

Other uses, including reproduction and distribution, or selling orlicensing copies, or posting to personal, institutional or third party

websites are prohibited.

In most cases authors are permitted to post their version of thearticle (e.g. in Word or Tex form) to their personal website orinstitutional repository. Authors requiring further information

regarding Elsevier’s archiving and manuscript policies areencouraged to visit:

http://www.elsevier.com/copyright

Author's personal copy

Analysis

Assessing poverty–deforestation links: Evidence from Swat, Pakistan

Shaheen Rafi Khan a, Shahrukh Rafi Khan b,⁎a Sustainable Development Policy Institute, P. O. Box 1091, Islamabad, Pakistanb Department of Economics, Mount Holyoke College, 50 College Street, South Hadley, MA 01075-1481, USA

a b s t r a c ta r t i c l e i n f o

Article history:Received 5 June 2008Received in revised form 21 January 2009Accepted 16 April 2009Available online 21 May 2009

Keywords:Ecological degradationDeforestationPovertyLocal institutionsResource management

This paper contributes to the debate on the links between poverty and forestry degradation; the view thatdue to poverty and the meeting of subsistence needs the poor use natural resources more intensively andhence cause them to degrade. Using the case of the forest rich Swat district, Pakistan, the paper addresses theissue empirically, historically, and institutionally. We do not find empirical support for the “poverty–environment nexus”, in that the poor and other income groups are equally resource dependent and alsoshow that resource degradation is not associated with poverty. Our historical and institutional analysesprovide alternative explanations for resource degradation. Selective and rotating ownership patterns, startingwith the 17th century, provided limited incentive for resource conservation. It also created tension betweende jure and de facto owners, that has persisted, and is one source of forest degradation. Ill-defined resourcerights have also exacerbated the impacts of several other factors contributing to forest degradation which iscompounded by poor management, corruption, and perverse incentives.

© 2009 Elsevier B.V. All rights reserved.

1. Introduction

Pakistan is poorly endowed with forest resources. Further, thedeforestation rate, at an annual 4.6%, is the second highest in theworld and represents a loss of ecological services in the nationalcontext.1 For forest dependent communities, it constitutes a seriouserosion of livelihoods.

Khan and Pervaiz (2001) point out that forest management inPakistan has led to the divergence of individual versus socialobjectives with regard to the use of forest resources. The mostimportant factor in this development is commercial timber extractionin which influential extractors collude with forest departmentofficials. Co-opting officials has become easier as stagnant salarieshave led to increased corruption. Forest fines and penalties have alsobecome meaningless, as they have not kept pace with rising timberand fuel wood prices.2 Attempts to right price/tax the use ofenvironmental resources is politically challenging as vested interestsresist policy reform threatening their economic profits. They also

indicate that the lack of clearly defined resource rights exacerbates theimpact of perverse incentives.

Sultan-i-Rome (2005) points to various institutional failures thathave led to forest degradation in Swat, Pakistan over the years. Theseinclude departmental malpractice, opaque resource rights, andjudicial tardiness. He describes the tension between customary andstatutory law and the progressive deterioration in conflict resolutionmechanisms over time. He concludes that extended legal, judicial andgovernance lapses have been the key factors in forest degradation.

Khan and Naqvi's (2000) qualitative analysis of the relationshipbetween poverty and resource degradation in Pakistan indicates thatthe poor are the most vulnerable to ecological degradation and yet,the absence of basic subsistence makes them predators of naturalresources thereby further exacerbating their vulnerability. They arguethat the poverty–resource degradation link reflects unavoidableresponses. This link been referred to in the broader literature as the“poverty–environment nexus.”

We add to the empirical literature by deconstructing the nexus intotwo related elements. We first explore relative resource dependenceby income group. If, as generally believed, the poor do indeed dependmore on natural resources for their livelihoods, then in principle theycould be contributing relatively more to resource degradation, even ifnot absolutely so. However this still needs to be established and that iswhat we turn to next using visual methods (satellite imagery, land usemaps, poverty and institutional maps). The use of satellite imagery atthe disaggregate level to explore the above hypotheses is also stillfairly rare in the literature.

We find no support for the hypotheses that resource degradation isinversely associated with income or that poverty and forest degrada-tion are positively associated. Yet there is considerable evidence for a

Ecological Economics 68 (2009) 2607–2618

⁎ Corresponding author. 45 Brownley Drive, Stamford, CT 06905, USA. Tel.: +1 203355 1884, +1 203 542 5292.

E-mail addresses: [email protected], [email protected] (S.R. Khan).1 Government of Pakistan/World Wildlife Fund (2000). Government of Pakistan,

Statistical Supplement (2008, p.19) shows a steady increase in forest area from 1990-91onwards to the current forest coverage of 4.2%. This increased coverage reflects theimpact of farm forestry; primary forests however are degrading rapidly as indicatedearlier.

2 Government of Pakistan, Statistical Supplement (2008, p. 62) index of wholesaleprices shows timber and firewood prices have increased respectively by 62.09 and76.28% between 2000-01 and 2006-07, the second and fourth highest among the 19non-fuel related items for which indices are reported.

0921-8009/$ – see front matter © 2009 Elsevier B.V. All rights reserved.doi:10.1016/j.ecolecon.2009.04.018

Contents lists available at ScienceDirect

Ecological Economics

j ourna l homepage: www.e lsev ie r.com/ locate /eco lecon

Author's personal copy

high level of forest degradation in the study area and so we turn to ahistorical and institutional analysis for alternative explanations. Weexplain the emergence of formal and informal systems of resourcerights, their interface, the disjuncture–or conjunction–between dejure and de facto rights, and the subsequent administrative and man-agement failures. We posit that the historical evolution of institutionsis central to understanding the current problem of rapid deforestationand that policy needs to take into account local history and insti-tutional evolution. We view this historical and institutional analysis asthe mainstay of our contribution to the literature.

Section 2 reviews the literature and conceptual issues pertainingto the poverty–environment nexus. Section 3 presents an overview ofthe case study area and its history. Section 4 discusses our researchdesign, method, data and variables and the empirical analysis is pre-sented in Section 5. We review historical and institutional explana-tions for forest degradation in Section 6.

2. The empirical debate and conceptual issues

The empirical debate on the association of poverty and theenvironment has several themes.3 The first pertains to whether ornot poverty increases environmental degradation. Second, since one ofthe mechanisms for resource degradation is the intensity of resourcedependence, one strand of this empirical literature explores the extentof resource dependence by income group. The third pertains towhether or not environmental degradation increases poverty. Thefourth views the correlations above as spurious and views povertyand resource degradation as jointly determined and caused by otherfactors. Finally, the quantitative debate above is complemented byhistorical and institutional analyses, the main focus of this paper.

Poverty, it is presumed, imposes short time horizons and riskaversion. Thus, poor people are viewed to have high pure rates of timepreference. In terms of land use, this means overgrazing pastures,shortened fallow periods, and a reluctance to invest in land improve-ments when returns occur after a long gestation period. The poor arealso viewed to be more risk averse because they possess fewer assetsand limited access to credit and insurance and so face greater con-straints to managing their risks. Thus, they would be expected to beless willing to invest in conservation.

Degradation is also viewed to result from the ecological margin-alization of the poor. They are forced into areas that are alreadyexperiencing high levels of stress and where future outcomes areuncertain. Consequently, there is a tendency to mine resources un-sustainably. Finally, there is a more questionable contention thatwould lead to similar results which is that poverty generates fatalismand an acceptance of a given situation rather than a desire and re-sultant efforts to change it.

The evidence on such links are mixed with more studies refutingthan supporting a causal link from poverty to resource degradationin general and forests in particular. Deininger and Minten (1999)develop a theoretical model that is estimated using socio-economicand physio-geographic information (including digital maps) to showlower poverty results in lower degradation in Mexico. Swinton,Escobar and Reardon (2003) draw on studies from across LatinAmerica to show that the non-poor and the poor are both at fault forresource degradation within the general context of poor incentivesfor preservation and the particular context of the low resource baseof the poor.

Other studies are more categorical in showing that resourcedegradation in general and deforestation in particular is not associated

with poverty. Based on a qualitative study of Nicaraguan hillsides,Ravnborg (2003) demonstrates, based on an investigation of fiveenvironmentally harmful natural resource management practices,that the non-poor farmers rather than the poorest are responsible.Agudelo et al. (2003), using data from a household survey in awatershed region of the Colombian Andes, show that there is nostraight forward linear association of poverty and natural resourcedegradation. The poorest have limited access to productive resourcesand it is the practices of the prosperous farmers that cause the greatestdegradation.

Jodha (1998) posits that the poor have more reason to preserveresources that their livelihood depends on and there is some evidenceto support this view. For example, Swinton et al. (2003) show thatfallowing is a practice used by the poor that reduces soil erosion andfertility loss. Similarly Scherr (2000) shows instances in which thepoor enhanced natural resource conservation.

A substantial body of the empirical literature quantifies the rela-tionship between environmental degradation and resource depen-dence by incomegroup, the assumptionbeing that greaterdependencewould lead to greater contribution to degradation. Jodha (1986) pointsto an inverse relationship between resource dependence, defined asthe percentage of total income derived from common-pool resources,and income levels. Reddy and Chakravarty (1999) confirm this findingby demonstrating a difference of almost 19% in resource dependencebetween the rich and the poor. Cavendish (2000) also supports thisfinding, but finds absolute dependence levels to be much higher thanmost other studies for both poor and rich households. In a studyexploring the benefits derived by income group from common forestsin Nepal, Adhikari (2005) finds that the richer households derivemore income from these resources than the poor both absolutely andrelatively.

Narain, Gupta, and Van de Veld (2008) document that the inverseassociation of income and dependence on common-pool resources(CPRs) are not the norm in the literature and make a case for usingpermanent rather than current household income in exploring thisassociation. They establish a U-shaped relationship between incomelevels and dependence, with such dependence declining at first andthen increasing with income. They also find that dependence is im-pacted by the quality of natural resources (that the rich draw on ahigher quality or resources), that there is a bi-modal pattern of de-pendence within groups (e.g. that the very rich do not draw on CPRs)and that overall resource dependence increases with the level ofcommon-pool biomass availability.

Studies that suggest other factors are at play include Swintonand Quiroz (2003) who use farm survey data from the PeruvianAltiplano to quantitatively show that natural resource sustainabilityvaries with management activity rather than poverty, althoughdeforestation is associated with fuel wood harvesting by the poorest.Utilizing aggregate district and provincial level data, including satel-lite imagery for poverty mapping, Dasgupta et al. (2005) concludethat demographic pressure rather than poverty are themain drivers ofdeforestation in Cambodia and Laos. Markandya (1998, 2001) aguesthat the poverty–environment association is mediated via severalfactors such as gender, policy (e.g. commodity export drives displacingtenants, capital subsidies, credit policy, social sector delivery), insti-tutions (property regimes), incentives, demographics (rural–urbanmigration, civil war and population displacement), and economicgrowth and hence is complex and the evidence in most cases issketchy at best.

These reviews suggest that the evidence is far from conclusivein establishing the poverty–degradation link and that the real issuesmay well be the nature and extent of the community's stake in thehealth and productivity of environmental resources and the techno-logical and institutional mechanisms available to safeguard theseresources. For example, Jodha (1995, 1998) contends that dilutionor disintegration of the community's stake and erosion of grassroots

3 We are grateful to one of the referees of this paper for suggesting theorganizational framework for our literature review. For more elaborate frameworksand literature reviews, refer to Duraiappah (1998) and Opschoor (2007).

2608 S.R. Khan, S.R. Khan / Ecological Economics 68 (2009) 2607–2618

Author's personal copy

level mechanisms to protect and augment it are the fundamentalreasons behind environmental resource degradation rather thanpoverty.

Essentially, Jodha claims that the resource management problemis functionally linked with cognitive space, property regimes andscarcity. Exposure to market forces and the integration into thebroader administrative and legal framework weaken traditional man-agement regimes and lead to neglect and degradation of hereditaryresource endowments. He presents alternative scenarios whereendowments are similar, but outcomes differ because economic andinstitutional parameters differ.

Schlager and Ostrom (1992) underscore the congruence of clearlydefined resource rights and sustainable resource use. Incentivesneed to be framed in a manner that promotes sustainable resourceuse; nevertheless, the risk is always present that they could becomeperverse and engender degradation. Howard et al. (1997) classifythese incentives into direct fiscal and financial incentives andindirect incentives. They suggest restructuring indirect incentivesto empower local forest dependent communities to give them aneffective stake in sustainable forest management and this is also anissue we explore.

Mogaka et al. (2001), in a study based in South and East Africa,argue that while forest resources have a high economic value for forestdependent local communities, there are significant economic costsassociated with sustainable use as well. If communities are to act asefficient actors in promoting forest conservation, economic benefitsreceivedmust be visiblymore than those obtained fromunsustainableuse. This latter risk is always present with the rapid rise in forestproduct prices.

This strand of the literature suggests that strong and effectiveinstitutions can mediate many of the tendencies that lead the poor todegrade the environment. While, there are many dimensions to aninstitutional analysis, we focus on the interface between resourcerights and perverse incentives.

3. Case study area, forest degradation and brief history

Our case study area is the region of Swat district, which is situatedin the North-West Frontier Province of Pakistan (refer to Appendix AChart I for map). As indicated in Government of Pakistan (1999), ithistorically consisted of two tracts, Swat Kohistan, which is locatedon the upper reaches of the Swat River, and the main Swat Valley.In 1917, Swat was proclaimed a state, accessed to Pakistan in 1947,and integrated administratively with the country in 1969. It is nowreferred to as Swat District.

The present district covers an area of 4000 square miles. It bordersthe Gilgit agency in the Northeast, Chitral in the North and Dir in theWest. In 1998, the district population was estimated at 1.26 millionand the annual population increase lies between 3–4%. Literacy levelsare generally low and far lower for women (25.2% male compared to3.7% female).

The Swat Valley has a Pathanmajority,most ofwhich belongs to theAkozai branch of the Yusufzai Pathans. Indigenous ethnic groupsreferred to as Kohistanismainly inhabit swat Kohistan; these comprisethe Ghauri tribe in the north and the Torwali tribe in the south. A largeGujjar community (itinerant grazers) has also taken permanent re-sidence in Swat. Agriculture and horticulture are the major sourceof income, followed by wages and salaries, and local and foreignremittances. The most important cash and fruit crops in the districtinclude wheat, maize, persimmon, tomato, onions, apple, and apricot.

The Swat Valley and Swat Kohistan are extremely rich in naturalresources. Swat once was credited for having the world's only virgindeodar (cedar) forests. Despite extensive deforestation, even todaythe district is rich in forest cover. Agricultural land, pastureland, andrangeland are found throughout the area while alpine pastures lieto the extreme north of the Swat Valley and Swat Kohistan. Table 1

above provides an overview of the change in natural resources from1969 to 2005 based on 403 responses.

As evident from the table above, forest cover has been rapidlyshrinking since 1969 when Swat state was taken over by Pakistanwhile the increase in agricultural land by 27% represents encroach-ment explained in Section 6 below.

The recent history of Swat begins with the Yusufzai Pathans whoinvaded and, subsequently, governed the region in the 16th century,defending it successfully against several waves of invaders. They main-tained a tribal set-up and by the 20th century Swat was in a state ofanarchy, with the local khans (chiefs) individually ruling the area thatfell under their sphere of influence. A dramatic turn in Swat's historycame about in September 1917, whenMiangul AbdulWadud, a YusufzaiPathan, proclaimed Swat an independent state. The British, who ruledIndia at the time, did not recognize Abdul Wadud formally, but in 1926agreed to accept him as thewali, or state ruler. Abdul Wadud abdicatedin favor of his son, Miangul Jehanzeb, in December 1949.

The walis are credited with consolidating the state. They put ef-ficient government machinery in place and undertook a large numberof development projects. The first wali, Miangul Abdul Wadud, in-stituted a land settlement system and brought the forests under statemanagement. He signed the Instrument of Accession in 1947 and thesecond wali signed the Supplementary Instrument of Accession in1954 and Swat state was finally merged into Pakistan in 1969.4 Swatresidents consider the rule of the walis as the golden period in theregion's history.5

4. Research design, method, data and variables6

The geographic focus of our study is Swat District that containstwo sub-divisions; Swat and Matta. There are 8 tehsils in the district.These tehsils are not regular revenue tehsils as in other districts in thecountry but are small administrative units that evolved when thedistrict was a state. Matta tehsil, with 13 union councils, is the mainSwat valley and Swat Kohistan, with 53 union councils, is an adjoiningvalley. Historically, these valleys have had the largest primary forestconcentrations in the country and they are the most vulnerable todegradation. We purposively selected Matta valley which has threeeco-zones (low, mid and high), defined by ecological, agriculturaland altitudinal characteristics, as indicated in the land use map(Appendix B Chart II). The numbers represent the 18 villages in thesample.

Table 1Change in quantity of natural resources (1965–2005).

Naturalresource

Increase/decrease Change (%) Increase/decrease Change (%)

Swat Valley Swat Kohistan

Conifer forests Decrease −62 Decrease −34Oak forests Decrease −75 Decrease −43Agriculture land Increase +29 Increase N/APasture land Decrease N/A Decrease N/AWater Decrease −53 N/A N/A

Source: SDPI survey.Note: the numbers in the table represent a tabulation of responses on the naturalresource questionnaire regarding degradation.

4 The first wali, Abdul Wadud, was illiterate and so the stamp from his seal (ring)represented the signature.

5 Even though our fieldwork indicated that the Walis of Swat increasingly usednatural resources, particularly forests, for political patronage, that era is nonethelessassociated with much better resource management and conservation than thesucceeding era.

6 District information is from Government of Pakistan (1999) and the rest of theinformation for this section was derived from fieldwork.

2609S.R. Khan, S.R. Khan / Ecological Economics 68 (2009) 2607–2618

Author's personal copy

We selected five villages from the low zone, four from the mid-zone, and three from the high zone. Household information wascollected from 403 households. Since our main interest was in re-source dependence and this was anticipated to be much higher in thehigh eco-zone, we selected an additional 200 households from anadditional six villages in this zone, for an extended sample of 603, forthe qualitative analysis.

For the analysis of resource dependence, information was collectedon resource income (RI) as the dependent variable and the independentvariables included income from other sources or non-resource in-come (NRI) and livestock numbers (LS), given their potential impact onresource degradation. Depending on the eco-zone, natural resources(firewood, timber, fodder and other forest products) are either pur-chased from the market or collected directly from the forests. The loweco-zone has a relatively high density of village settlements. Its locationdoes not permit easy access to natural resources, with such resourcesbeing accessed indirectly via purchases from the market. The mid-zone is defined by extensive horticultural activities and lies in closerproximity to the forests, allowing households direct access to forestresources. This zone includes a mix of coniferous and deciduous forests.The high eco-zone falls along the upper valley ranges and includeslargely coniferous forests, grazing lands and alpine pastures. The villagehouseholds are located both on the forest fringes, as well as within theforestswhere land has been cleared for residential purposes, agricultureand livestock grazing.

A household survey was pre-tested and implemented over a six-month period, from April to September 2005. Information on naturalresource management (NRM), village attributes, local history, in-stitutions and community perceptions regarding forest managementby the state was collected via a community surveys, focus groupdiscussions (FGD) and key respondent interviews. This fieldworkalso included a survey of prices, village profiles, and a natural re-source survey.7

Total household aggregate income was derived from: natural re-sources (fuel wood (wood and brush) collection, timber, fodder, non-timber forest products (NTFPs), fish, and forest royalties); agriculture(crops and orchards as an annual figure including both kharif(summer) and rabi (winter) season crops); livestock; employment;donations/transfers; forest royalties; and rental income.

5. Empirical results

5.1. The poverty–resource dependence association

As mentioned in the introduction, we quantitatively examinedpoverty–resource degradation links by deconstructing them intotwo related elements; resource dependence by income group andthe impact of resource dependence (poverty) on resource degra-dation. Resource dependence is defined as the ratio of incomefrom natural resources to income from all other sources. Naturalresource income refers to self-consumption and sale of natural re-sources (defined above) collected from government and communityforests.8

Our first objective was to assess if resource dependence is in-versely and significantly associated with other (non-resource) in-come, controlling for livestock, ownership and eco-zones. Our OLS(ordinary least squares) result (Table 2) suggests that the incomecoefficient is insignificant. Since non-resource income (NRI) mightbe endogenous, we address this potential problem using 2SLS (two-stage least squares) estimation. Since the bulk of non-resource

income is wage income, we instrument it using the human capitalmodel.9

NRI = NRI ðEd; Prof ; Age; M; LS; E2; E3Þ ð1Þwhere

Ed education measured in yearsProf profession (dummies for skilled and unskilled)Age proxy for experienceM migration statusLS livestock (numbers)E E stands for the three eco-zones (low, middle and high).

The second stage equation is the original OLS with predicted NRI:

RI = RI ðpredicted NRI; LS; E2; E3Þ: ð2Þ

Even though the fit of the first stage equation is poor, the F-statisticfrom the first stage equation demonstrates the validity of the in-strument. As shown above in Table 2, there is virtually no difference inthe OLS equation and the second stage 2SLS equation suggesting thatendogeneity is not an issue.

Eq. (2), like the OLS result, has a good fit for cross sectional analysiswith all included variables having the expected signs. The hypothesisthat resource dependence varies inversely with household income isrefuted. As expected, eco-zones have the expected signs, differences inmagnitudes, and are highly significant.10 Also as expected, livestockowned has a positive and highly significant coefficient. While wereject the hypothesis that the poor are necessarily more resourcedependent, we nonetheless explore the second link stated abovebetween poverty and forest degradation.

7 Copies of the household and natural resource instruments can be made availableon request to [email protected].

8 Communities continue to retain subsistence and royalty rights in these forestssubsequent to government take-over of the forests.

Table 2OLS and 2SLS estimation of resource dependence.

Variables/statistics Eq. (1)—2SLS(LHS/NRI)

Eq. (2) —2SLS(LHS/RI)

OLS(LHS/RI)

Constant −35.29 (−1.52) 2.52 (1.12) 2.60⁎⁎⁎(1.66)Education (years) 6.44⁎(5.09)Profession(non-agriculture=1)

25.91⁎⁎(2.18)

Age 1.26⁎(3.43)Migration(migrant from household=1)

−30.36 (−1.40)

Livestock owned (numbers) 4.24⁎⁎(1.95) 2.13⁎(5.90) 2.14⁎(6.14)Medium altitude eco-zone 16.69 (1.31) 5.77⁎(2.73) 5.79⁎(2.78)High altitude eco-zone 5.08 (0.34) 20.62⁎(8.30) 20.62⁎(8.35)Predicted NRI 0.01 (0.76)NRI 0.01 (0.93)Adjusted R2 .07 .26 0.26F-stat 5.43⁎ 36.40⁎ 37.77⁎

n 401 401 402

Notes: E1 (low eco-zone) is the excluded variable.RI = resource income.NRI = non-resource income.LHS = left hand side variable.Parentheses contain t-statistics.⁎Significant at the 1% level.⁎⁎Significant at the 5% level.⁎⁎⁎Significant at the 10% level.

9 Mean and standard deviations of all variables used in the quantitative analysis arereported in Appendix Table 1.10 For example, those residing respectively in the high and middle eco-zones deriveRs. 20,620 and Rs. 5790 more annually from resources than those residing in the loweco-zones. The average exchange rate of the Rs. for $1 was 59.86 in 2005-2006,Government of Pakistan (2008), Statistical Supplement, p. 76.

2610 S.R. Khan, S.R. Khan / Ecological Economics 68 (2009) 2607–2618

Author's personal copy

5.2. The forest degradation–poverty association11

To explore the association of poverty and resource degradation, wegenerated a relative poverty ranking of villages and overlapped itspatially with forest degradation using satellite imagery. From a listof the most frequently used global poverty indicators, we selectedthose for which data had already been collected as part of our sur-veys.12 After a number of iterations we arrived at a final set of fiveindicators including mean annual per-capita household income, em-ployment rate, mean per-capita household livestock numbers, meanper-capita rooms per household, and per-capita area of land underprivate ownership.

The individual villages were ranked for each indicator. We thenarrived at the final ranking by taking the un-weighted average of theindividual rankings with the lowest ranking representing the poorestvillage. Our main concern is the high zone where degradation is aserious problem. Here, we have a mix of both rich and poor villages,which is counter intuitive as one would, a priori, expect the moreremote areas to be poor.

There is no evidence of a poverty–environment nexus within thehigh zone. Appendix C Chart III shows the high poverty-rankedvillages (3, 4, 9, 12, 16–18) in the high zone and the red shading showsthe degraded areas in the high zone. Only two of the seven villagesoverlap with the degraded areas confirming the lack of correspon-dence between poverty and forest degradation. The absence of anexplicit relationship between poverty and resource dependence, com-bined with a lack of correspondence between zones of poverty andforest degradation, led us to explore alternative explanations for suchdegradation.

6. Historical and institutional analyses

6.1. Historical analysis13

The system of resource rights in the main Swat Valley was es-tablished after the Yusufzai Pathans invaded the valley in the 16thcentury. Customary law governing rights to natural resources wasrooted in a system introduced by a Yusufzai notable and revenueexpert, Sheikh Malli. This systemwas known as garzinda wesh; trans-lated literally, it means moveable distribution. The rationale was thatas land differed in composition, location/accessibility, fertility, andavailability of water, it was necessary to ensure equal sharing of itsbest and worst features. Land allotments were by village and re-allotments, by ballot, were carried out every 5, 7 or 10 years, as agreedmutually between the villages. The land allotments included a mix ofagricultural land, forest (zangal), pasture (warshoo), and wasteland.While the system assured socio-economic justice, it relocated entirevillages and took away the incentive to use natural resources in asustainable manner.

Within the village allotments, the entitlements to common prop-erty resources (shamilaat) that included forests, pastures, rangelands,and water were in proportion to the size of private holdings ofagricultural land. The holdings were referred to as dawtar and theowners as dawtaris. The sale and purchase of agricultural land meantthat the dawtars varied over time and, commensurately, so did thecommon property entitlements within the village.

Religious and tribal status was the basis for separate entitlementsknown as serai and the owners of such lands were called stanadars.These were community land grants and were made to holy families/persons to construct mosques as well as for self-sustenance.

A more important type of land grant, that subsequently had im-portant political ramifications in that it established a basis for here-ditary title, was to the tribal elders and this was referred to as khanserai. Originally, these grants had the functional purpose of providingthe wherewithal to entertain, a critical requisite for leadership: thecategories were daday serai (maize serving area), mela serai (guestarea), and telu serai (oil burning area in the hujra (segregated spacefor male congregation)). Serai holdings were mundai (with definedborders) and did not preclude dawtar shares.

Only the Yusufzais could be dawtaris, with ownership rights toagricultural land and hence to common-pool resources. Non-dawtariscomprised different ethnic groups, tenants, and village artisans. Theywere required to pay begar (labor tax) and qalang (grazing tax) fortenancy and grazing rights. They were allowed use rights by dawtarisin the forests but, even as owners of purchased land, had no owner-ship rights in common property. Thus the resource rights regimedifferentiated clearly between the rights of owners and users.

When Swat state came into existence in 1917, the wali (ruler)claimed ownership and custodial rights over the forests. He estab-lished a forest department to manage the forests and sanctioned a10% royalty from the commercial proceeds of timber. The wali alsoinstituted two types of timber quotas. The local quota was meant forlocal residents requiring timber for domestic use. The qaumi(people's) quota applied to residents of areas where natural resourceswere not found in abundance. It was also used to acquire timber forpublic schemes. The wesh rotations were abolished, locking in privateownership and common property entitlements spatially and in termsof shares. In the sense that dawtar shares determined the division ofroyalties, historical ownership rights persisted. As in the pre-statedays, the rights regime continued to differentiate between owners andusers. In fact, the old system continued to prevail for all naturalresources except forests. For instance, the practice of begar and qalangwas left unaltered by the wali. With regard to forests, the only changein terms of subsistence use was the qaumi quota; in other respects theold divisions and entitlements remained.

In 1926, Swat State and the colonial administration of the Govern-ment of India entered into a formal agreement to manage and con-serve forests. The agreement transferred administrative authorityof these forests to the Government of India, but the de facto controland management remained with Swat State. While the rules andregulations were drawn up by the colonial government, the walicontinued to exercise effective jurisdiction over the forests.

The independence of Pakistan in 1947, and the subsequent signingof the instrument of accession had no significant impact on naturalresource management in Swat State. Resource rights remained un-changed for both owners and users. There were no efforts from eitherthe Pakistani government (withwhom the State had notmerged at thetime) or the wali to undertake any forest or land settlements in themain Swat Valley.

A change in management practices came about after Swat Statemerged with Pakistan in 1969. The forests were declared provincialsubjects under the sole jurisdiction of provincial governments.14 In1975, the provincial government declared forests as protected, andconverted them into state property. The local and awami quotascontinued; the latter were determined solely by the forest departmentthat was susceptible to political influences. As a consequence,11 As indicated in Section 2, Dasgupta et al. (2003) and Deininger and Minten (1999)

also use spatial data. The former used aggregate district and provincial level data andthe latter indicate some problems with aggregate analysis.12 There are likely to be rich people in poor villages and vice versa but we areconcerned here with average village prosperity.13 We restrict our analysis to the main Swat-Kohistan valleys due to spaceconstraints. For more details, refer to Khan and Pervaiz (2001) and Sultan-i-Rome(2005, 2006).

14 Despite the change of management, many of the officials working in the ForestDepartment under the wali continued to hold their posts after the merger withPakistan. The implementation mechanisms were however completely overhauled.

2611S.R. Khan, S.R. Khan / Ecological Economics 68 (2009) 2607–2618

Author's personal copy

the quotas benefited a select few. Under the new system, traditionalowners were to be compensated by providing them a share in royal-ties for timber extraction. Currently, the royalty share in Matta is fixedat 60%.

The period between 1969 and 1975 was marked by confusionregarding community resource entitlements. Pre-merger, the wali'srule was authoritative, the writ of law was well established, andresource rights for communities relatively well defined. The walisclosely monitored natural resources, forests in particular, and ensuredthat no illegal felling took place. With their strict vigilance gone, therewas confusion surrounding the future rights of communities andnatural resources began to be freely exploited.

Another major change in the ownership patterns also came aboutduring the tenure of Prime Minister Zulfiqar Ali Bhutto (1972–77). Atthis time, the issue of de facto versus de jure ownership becameimportant. The period saw large-scale purchases by resident Gujjarsand land tenants of agricultural land and grazing land “bandajaat”from the landowners. The tenants and Gujjars also claimed the landthey were cultivating and grazing, encouraged by the land and ten-ancy reforms instituted by the populist Bhutto government. This alsotriggered a willingness to sell by the absentee landlords. Prior tothis development, any non-Pathan had to seek consensus of the jirga(council of village elders) or the consent of the khan (chief) to pur-chase land. The tenants had now become de jure owners and wererelieved of their obligation to pay begar and qalang to the Pathans.Land settlements, which took place in 1986, further formalized thisprocess.

6.2. Institutional analysis15

Customary and statutory law defines resource rights in the Mattatehsil. Customary law was prevalent during the pre-wali era whencommunities owned and managed the forests. Over time, the forestscame progressively under state control which managed them understatutory writ — first, under the walis and later the federal govern-ment. Thus, forest ownership, originally vested in the communities,was in time transferred to the state. While communities continue toretain rights in the forests the forest department's (FD) legal term forthem is now “concessionists” rather than rights holders.

The transition from community to state ownership and fromcustomary to statutory law has engendered conditions detrimental tothe forests and turned harmonious relations between communitiesand the state into exploitative and conflict-ridden ones. Good gov-ernance characterizes the wali era in that enforcement was both strictand fair. However, after the provincial government takeover in 1969,governance failures have become endemic and forests have degradedboth due to community and timber contractor inroads — the latteraided in no small part by forest department officials and local elders.Thus the post-merger period resulted in a rapid increase in defores-tation, stemming from institutional failure.

Community FGDs (focus group discussions) indicate that the forestdepartment is viewed as inefficient and corrupt. The discussions citedcollusion with the timber mafia and the selective application of finesand penalties for forest transgressions as manifestations of suchcorruption. The deteriorating state of resources other than the forests,

such as grazing lands, pastures and fisheries present an equally dismalpicture and this is partly due to the opaque rights that presentlygovern these resources.

Communities perceive customary law in a positive light stating apositive correlation between sustainable resource use and customarylaw. This is largely a result of the clarity of established norms, owner-ship and close monitoring of natural resources, which holds despitethe fact that customary law puts no limit on the use of resources forsubsistence.

In contrast, they have formed a jaundiced view of statutory lawand the associated forest department mandate. The reasons thatthey cite for this include excessive rent-seeking practices, bureau-cratic hurdles, and most importantly the loose writ of the forestdepartment, which results in the failure to implement rules andregulations. Discretionary powers encourage corruption and collu-sion with timber contractors.16 The forest department is under-staffed, with one forest guard being assigned, on average, to fourvillages. Delays in the release of royalties to communities have un-dermined the state's credibility. Further, the forest department istardy in educating communities about sustainable forest practicesand the potential benefits that could accrue as a result. The utility ofsuch education is self-evident when the enforcement writ is weakand the resource needs many.

The interrelationships of statutory and customary law and thecomplex and conflicting system of resource rights it has created arethe primary source of forest degradation. In this context, we identifiedfour categories of forest “owners”: the forest department (FD); com-munities with use rights; de jure owners living in close proximity totheir forest holdings; de jure owners not living in close proximity totheir forest holdings.

De jure owners living in close proximity to the land use the servicesof gujjar who normally pay qalang (grazing tax) and tenants who paybegar (free labor for using their land and forests), and theydo not claimroyalties. However, de jure owners not living in close proximity to theirland create rival ownership claims by the long-resident gujjars. In suchsituations, the latter do not pay qalang, claim royalties and, effectivelycontrol the forests thereby becomingde facto owners. These forests aredisputed and the subject of frequent litigation.

The Bhutto era (1971–1977) saw new pro-community rulingsbeing instituted. These aggravated existing tensions between com-munities and forest owners, with adverse effects on the forests. Fear-ing outright acquisition de jure owners sold their forest clearings(bandajat) to de facto owners during the Bhutto era. The nature of thesale deed was such that any area under tree cover remained theproperty of the de jure owners while uncovered area became theentitlement of de facto owners. This provided an incentive to de factoowners to fell trees, thereby laying claim to additional land eventhough it was against the spirit of the sale deed. In addition, when theprovincial government declared the forests as protected in 1975 andcompensated the de jure owners with a 60% share in royalty, the defacto owners received no such rights. On the one hand, this led tosignificant deforestation and land use change (forests were convertedto agricultural land) and on the other hand it increased tensionsbetween de jure and de facto owners.

Appendix D Chart IV characterizes the overlap between owner-ship conflict and forest degradation for four blocks and shows thatthe ownership–degradation nexus is more pronounced and overrides poverty considerations. The spatial disjuncture between dejure and de facto ownership is a key factor in degradation. The15 The institutional analysis in this sub-section is based on key respondent interviews

and focus group discussions (FGDs). The interviewees include forest departmentofficials, heads of local and foreign NGOs working in the area, and donorrepresentatives. The FGDs were with local communities across the zones. The essenceof the notes of the principal investigator, who personally led the fieldwork, iscontained in this sub-section. Other interviews and FGDs were taped and in somecases translated and transcribed. We have not used direct quotes since the text in thesecases is rough. These tapes and documents and also field reports of other field teammembers can be made available to the interested reader by [email protected].

16 One example of this is the post of “forest magistrate” who is the sole authorityresponsible for implementing any penalties against transgressors. This combined withthe rising timber prices and meaningless level of fines and penalties have acted as aperverse incentive for timber extraction.

2612 S.R. Khan, S.R. Khan / Ecological Economics 68 (2009) 2607–2618

Author's personal copy

degraded forests represent areas where de jure and de facto ownersare different because de jure owners reside at a distance from theforests. Conversely, the well-protected forests are those where dejure and de facto owners are one and the same and live close to theforests.

Shaur Block consists of compartments 1–8 and 12–22 that areprivately owned and well protected. By contrast, in compartments9–11 and 23–25, where the de jure owners reside down country andthe de facto owners (Gujjars) are proximate settlers, are in a de-graded state.

In Biha Block, compartments 1–13 are contested by the Dirresidents and are in a degraded state. The remaining compartments,11–47, are privately owned and in relatively good shape. However,the watershed dividing the two districts is subject to frequent in-cursions by the Dir residents and this part of the private forests isalso degraded.

Rohringar Block is the most degraded block in Matta tehsil.Compartments 1–5, 12–17 and 31–35 are disputed, compartments7–11 are tenant controlled, and compartments 18–30 are communallyowned. In each case, the ownership incentive for conserving theforests is weak.

Finally, in Lalku Block the Mians are the de jure and de factoowners of compartments 7–46, a well-protected forest tract. Incontrast, compartments 6–16 and 49–53 are disputed, while compart-ments 1–5 and 47 are tenant controlled. The two latter categories offorests are degraded.

Ill-defined resource rights have exacerbated the impacts ofseveral other factors contributing to forest degradation. One of themain forces resulting in the change in the demarcation of the landhas resulted from the rapid increase in population in the valleys.With demographic pressure the demand for food and cash cropsincreased and accelerated encroachments on forest land. Thesame forces also resulted in livestock intrusions. This also resultedin out-migration down country and abroad given the limited num-ber of employment opportunities in the mountain areas. The sub-sequent inflow of remittances increased the demand for modernhousing structures and fuel wood andwith that an increase in timberprices. Despite many promises, there has been no development ofalternative sources of energy such as via the tapping of hydro powerpotential.

Second, the lack of a proper demarcation of forest and non-forestland has also contributed to forest degradation. In 1986, the govern-ment of NWFP instituted a land settlement for such demarcation.However, there is a lack of coordination between revenue and forestdepartment officials. The patwaris (revenue officials) are supposedto do the annual girdawari (land use survey) but are willing to workwith influentials to change records. Thus, while the records of theforest department show forests, the revenue records show them asfarmlands. Civil courts have often settled against the forest depart-ment by reference to the de facto situation and hence have accele-rated deforestation.

Third, prior to the current conflict raging between the military andthe Taliban militants, Swat was one of the most popular cool weathertourist sites in the country.17 Over the years, this resulted in a largenumber of river front hotels built from the top to the center of themain valley and some of the side valleys. Much of the refuse andsewage went untreated into the rivers. Also due to the increasing

number of tourists over time and the relative increase in localprosperity the demand for fuel wood increased.

Fourth, one of the most destructive manifestations of mismanage-mentwere the long delays in the release of royalty shares to claimants.This induced poor communities to sell their royalty rights in advanceto rapacious timber merchants for a fraction of their real value. Withan extraction rights in hand, the merchants removed timber beyondmandated amounts in collusion with forest guards. Due to this abuseof harvesting quotas, civil society organizations working withinternational donor organizations won a rare victory in the form ofa ban on commercial logging.18 Even this backfired because it inducedhigher amounts of illicit cutting.

Fifth, a much trumpeted initiative of the military government thatassumed power in 1999 was the devolution of power to the grassrootslevel. However, there was ineffectual implementation and littlecommunity participation so that the joint forest management plansthat were to be a part of this initiative failed and local politicians in the“devolved” administrative structure became privy to the timbersmuggling.19

7. Summary and conclusion

In this paper, we explore the debates around poverty–resourcedegradation linkages; that the poor are more resource dependentand consequently contribute relatively more to resource degrada-tion. Our quantitative results show no clear association betweenincome (poverty) and resource dependence. Utilizing satelliteimagery and poverty mapping, we also demonstrate that there isno necessary overlap between poverty and forest degradation. Weturned to a historical and institutional analysis to explain forestdegradation.

Our historical analysis starting with the 16th century indicates thatselective and rotating ownership patterns provided limited incentivefor resource conservation. However, once the walis of Swat tookcontrol in the early 20th century, ownership was frozen and resourceswere protected by stringent oversight of the forest department theycreated.

When Swat was administratively merged into Pakistan in 1969,the government declared forests protected and created tensionsbetween customary and statutory law. They also did not invest indeveloping the managerial ability required to protect resources.Given the rapid rise in timber prices, the forest departmentofficials have more incentive to collude with “forest mafias” than toprotect community resources. Furthermore, the lack of coordina-tion between the forest and revenue departments and the lack ofan implemented land demarcation enables encroachment on forestland for alternative uses via corruption and accommodating courts.

The populist government that absorbed Swat exacerbated theconflict between de jure owners with property rights and tenants orde facto owners. Unless de jure owners were also de facto owners, thenature of contracts resulted in tension and forest degradation.

Swat district residents depend substantially on natural resourcesfor subsistence use and as an income source. Such dependenceunderscores the need both for defining the rights to these resourcesclearly based on an understanding of local history and institutionalevolution. It also requires instituting sound management systemsthat avoid perverse incentives. Only with such policies in place canthe current rapid rate of deforestation be avoided and sustainableresource use ensured.

17 Information gleaned from local contacts suggests an intensification of degradationpost 2005 as a result of the “talibanization” of Swat and the attack on state institutionslike public girl schools, police stations, and law courts and the subsequent militaryresponse. Under the cover of conflict, the timber mafia has trucked down vast amountsof timber with payoffs to the Taleban. The heavy air bombardment of the high forestcovered mountains where the Taleban take refuge is also taking its toll. A full accountwould have to await a survey once the conflict subsides.

18 This outcome was the result of the appointment by the military government of arespected ex head of a civil society organization as Minister of the Environment.19 For a review of devolution in various contexts including resource managementrefer to Shyamsundar (2008).

2613S.R. Khan, S.R. Khan / Ecological Economics 68 (2009) 2607–2618

Author's personal copy

Appendix A. Chart I

2614 S.R. Khan, S.R. Khan / Ecological Economics 68 (2009) 2607–2618

Author's personal copy

Appendix B. Chart II

2615S.R. Khan, S.R. Khan / Ecological Economics 68 (2009) 2607–2618

Author's personal copy

Appendix C. Chart III

2616 S.R. Khan, S.R. Khan / Ecological Economics 68 (2009) 2607–2618

Author's personal copy

Appendix D. Chart IV

2617S.R. Khan, S.R. Khan / Ecological Economics 68 (2009) 2607–2618

Author's personal copy

References

Adhikari, B., 2005. Poverty, property rights and collective action: understanding thedistributive aspects of common property resource management. Environment andDevelopment Economics 10 (1), 7–31.

Agudelo, C., Rivera, B., Tapasco, J., Estrada, R., 2003. Designing policies to reduce ruralpoverty and environmental degradation in a hillside zone of the Colombian Andes.World Development 31 (11), 1921–1931.

Cavendish, W., 2000. Empirical regularities in the poverty–environment relationship ofrural households: evidence fromZimbabwe.World Development 28 (11),1979–2003.

Dasgupta, S., Uwe, D.,Meisner, C., Wheeler, D., 2003. The Poverty/Environment Nexus inCambodia and Lao People's Democratic Republic, The World Bank DevelopmentResearchGroup Infrastructure andEnvironment, PolicyResearchWorkingPaper 2960.World Bank, Washington D. C.

Dasgupta, S., Deichmann, U., Meisner, C., Wheeler, D., 2005. Where is the poverty–environment nexus, evidence from Cambodia, Lao PDR and Vietnam. World Develop-ment 33 (4), 617–638.

Deininger, K.W., Minten, B., 1999. Poverty, policies, and deforestation: the case ofMexico. Economic Development and Cultural Change 47 (2), 313–344.

Duraiappah, A.K., 1998. Poverty and environment degradation: a review and analysis ofthe nexus. World Development 26 (12), 2169–2179.

Government of Pakistan, 1999. District Census Reports of Swat. Population CensusOrganization, Statistics Division, Government of Pakistan, Islamabad.

Government of Pakistan, 2008. Economic Survey 2007–08. Economic Advisor's Wing,Ministry of Finance, Islamabad.

Government of Pakistan /WorldWildlife Fund, 2000. Biodiversity Action Plan. Ministryof Environment, Islamabad.

Howard, C., Meijerink, G., Hoe, H., 1997. Incentives for improved forestmanagement andtree planting. Paper Submitted to the XI World Forestry Congress, Antalya, Turkey,October 13–22.

Jodha, N.S., 1986. Common property resources and rural poor in dry regions of India.Economic and Political Weekly 21, 1169–1181 July.

Jodha, N.S., 1995. Common property resources and the environmental context: role ofbiophysical versus social stresses. Economic and Political Weekly 30, 3278–3281December.

Jodha, N.S., 1998. Poverty–environmental resource degradation links: questioning the basicpremises. International Center for Integrated Mountain Development, Katmandu,Nepal.

Khan, S.R., Naqvi, 2000. The environment–poverty nexus: an institutional analysis.Sustainable Development Policy Institute: Working Paper Series, vol. 49. Sustain-able Development Policy Institute and Institute of Social Studies, Islamabad.

Khan, S.R., Pervaiz, A.N., 2001. The integration of economics into Pakistan's biodiversityaction plan — a case study. Sustainable Development Policy Institute: WorkingPaper Series, vol. 65. Sustainable Development Policy Institute and Institute ofSocial Studies, Islamabad.

Markandya, A., 1998. Poverty, income distribution and policy making. Environmentaland Resource Economics 11 (3&4), 459–472.

Markandya, A., 2001. Poverty, environment and development. Frontiers of Environ-mental Economics. InEdward Elgar, Northampton, MA.

Mogaka, H., Simons, G., Turpie, J., Emerton, L., Karanja, F., 2001. Economic Aspects ofCommunity Involvement in Sustainable Forest Management in Eastern and SouthernAfrica. IUCN (International Union for the Conservation of Nature), Nairobi.

Narain, U., Gupta, S., Van de Veld, K., 2008. Poverty and resource dependence in ruralIndia. Ecological Economics 66 (1), 161–176.

Opschoor, J.B., 2007. Environment and Poverty: Perspectives, Propositions and Policies,Institute of Social Studies: Working Paper, vol. 437. Sustainable Development PolicyInstitute and Institute of Social Studies, The Hague.

Ravnborg, H.M., 2003. Poverty and environmental degradation in theNicaraguan hillsides.World Development 31 (11), 1933–1946.

Reddy, S.R.C., Chakravarty, S.P., 1999. Forest dependence and income distribution in asubsistence economy: evidence from India. World Development 27 (7), 1141–1149.

Scherr, S., 2000. A downward spiral? Research evidence on the relationship betweenpoverty and natural resource degradation. Food Policy 25 (4), 479–498.

Schlager, E., Ostrom, E., 1992. Property rights regimes and natural resources: a conceptualanalysis. Land Economics 68 (3), 249–262.

Shyamsundar, P., 2008. Decentralization, devolution and collective action — a review ofinternational experience. In: Ghate, R., N. S., Jodha, Mukhopadhyay, P. (Eds.), Promise,Trust and Evolution:Managing the Commons of South Asia. InOxford University Press,Oxford.

Sultan-i-Rome, 2005. Forestry in the Princely State of Swat and Kalam (North WestPakistan): A Historical Perspective on Norms and Practices. In: Individual ProjectWorking Paper, vol. 6. Swat, Pakistan.

Sultan-i-Rome, 2006. Governance of forests in Swat. Troubled Times: SustainableDevelopment and Governance in the Age of Extremes. InSDPI/SAMA, Karachi.

Swinton, S.M., Quiroz, R., 2003. Is poverty to blame for soil, pasture and forest degradationin Peru's Altiplano? World Development 31 (11), 1903–1919.

Swinton, S.M., Escobar, G., Reardon, T., 2003. Poverty and environment in Latin America:concepts, evidence and policy implications. World Development 31 (11), 1865–1872.

Descriptive statistics (403 households).

Variables Mean Standard deviation

Resource income (Rs. annual) 14,281.82 21,260.99Non-resource income (Rs. annual) 71,391.49 114,981.76Livestock (numbers) 2.09 2.72Age (years) 51.14 16.13Migration (migrant from household=1) 0.07 0.26Education (years) 2.93 4.70Profession (non-agriculture=1) 0.35 0.48Low eco-zone 0.42 0.49Middle eco-zone 0.36 0.48High eco-zone 0.22 0.42

Source: SDPI survey, 2005.

Appendix Table 1

2618 S.R. Khan, S.R. Khan / Ecological Economics 68 (2009) 2607–2618