Biological Conservation 168 (2013) 31–39

Contents lists available at ScienceDirect

Biological Conservation

journal homepage: www.elsevier .com/ locate /biocon

Developing a spatially-explicit, sustainable and risk-based insurancescheme to mitigate human–wildlife conflict

0006-3207/$ - see front matter � 2013 Elsevier Ltd. All rights reserved.http://dx.doi.org/10.1016/j.biocon.2013.09.017

⇑ Corresponding authors. Address: Department of Biological Sciences, NationalUniversity of Singapore, 14 Science Drive 4, Singapore 117543, Singapore.

E-mail addresses: beckyshuchen@gmail.com (S. Chen), ted.webb@nus.edu.sg(E.L. Webb).

Shu Chen a,b,⇑, Zhuang-Fang Yi c,d, Ahimsa Campos-Arceiz e, Ming-Yong Chen f, Edward L. Webb a,⇑a Department of Biological Sciences, National University of Singapore, 14 Science Drive 4, Singapore 117543, Singaporeb Institute of Zoology, Zoological Society of London, Regent’s Park, London, England NW14RY, United Kingdomc Key Lab of Tropical Ecology, Xishuangbanna Tropical Botanical Garden, Chinese Academy of Sciences, Mengla 666303, Chinad China and East Asia Node, World Agroforestry Center (ICRAF), Kunming 650223, Chinae School of Geography, University of Nottingham Malaysia Campus, Jalan Broga, 43500 Semenyih, Selangor, Malaysiaf School of Life Science, Yunnan University, Kunming 650091, China

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

Article history:Received 9 March 2013Received in revised form 16 September 2013Accepted 20 September 2013

Keywords:Asian elephantChinaCompensationHevea brasiliensisRiskRubberXishuangbanna

Insurance may encourage coexistence between farmers and wildlife by reimbursing farmers’ losses.China introduced an insurance scheme to mitigate human–elephant conflict in Xishuangbanna Daiautonomous prefecture in Yunnan Province, where elephants cause damage to rubber plantations. How-ever, recent experience has suggested that the present insurance system exhibits poor performancerelated to funding shortfalls, undervaluing of plantations and insufficient payouts, and by limiting com-munity involvement. To address these shortcomings we conducted attitude surveys with farmers, anddeveloped an actuarial (risk-based) insurance model for rubber loss that incorporated spatially-explicitrisk of depredation and net present value of rubber at damage, in order to calculate fair payouts at villageand town levels for the year 2011. Farmers were largely dissatisfied with the current insurance system,and their level of satisfaction was associated with the compensation ratio (percentage of lost rubberreimbursed by insurance). The illustrative results based on 2011 rubber loss data revealed high variabilityin risk and therefore payouts (and further, premiums) and that fair insurance payouts would be approx-imately five times the current levels. To improve compensation and support long-term program sustain-ability, we considered an insurance cost-sharing mechanism that incorporated shared payments fromgovernment, rubber farmers, and Chinese tourists. We found that multiple stakeholders were willingto pay for elephant conservation, which could make significant contributions to insurance premiums overthe long term. Importantly, this proposed insurance model could be broadly applicable to livestock andlong-lived cash crop compensation systems.

� 2013 Elsevier Ltd. All rights reserved.

1. Introduction

Human–wildlife conflict (HWC) is an important driver of popu-lation declines of many threatened species (Banerjee et al., 2013;Woodroffe et al., 2005). In order to conserve species that generateconflict with people, there is a need to secure the rights and liveli-hoods of rural residents who face the conflicts, bear the cost im-posed by wild animals, and may want the ‘‘pests’’ to beeradicated (Dickman, 2010). To reduce hostility of farmers towardswildlife, and the resulting retaliatory or pre-emptive killing, com-pensation schemes may be implemented, often in combinationwith other mitigation such as deterrents and barriers to prevent

damage, and awareness campaigns to increase people’s toleranceto wildlife (Nelson et al., 2003; Woodroffe et al., 2005).

Compensation—the reimbursement to individuals or their fam-ilies who have experienced wildlife damage to crops, livestock, orproperty, or who have been injured, killed, or physically threatenedby wildlife (Nyhus et al., 2005)—can incentivize coexistence,especially when HWC is chronic and the economic impact ofwildlife-attributed loss is substantial. While several successfulcompensation schemes have been reported across a range ofhuman–wildlife settings (e.g. Maclennan et al., 2009; Nyhus et al.,2005), some compensation programs have failed to encouragecoexistence and sometimes even worsened HWC (e.g. Gussetet al., 2009; Mishra, 1997; Naughton-Treves et al., 2003; Ograand Badola, 2008). Reasons for failure are various and includeinadequate compensation, lack of sustainable funding and thecreation of incentives detrimental for conservation—so-called‘moral hazards’ that may include over-reporting of losses (Bulteand Rondeau, 2007, 2005; Dickman et al., 2011; Nyhus et al., 2005).

32 S. Chen et al. / Biological Conservation 168 (2013) 31–39

Insurance is a form of compensation that requires participantsto pay a premium for their involvement (Dickman et al., 2011); thismechanism is gaining recognition as a potential improvement totraditional compensation, for several reasons. First, if based onactuarial (risk-based) analyses, insurance can promote fair pay-ments by incorporating in the premium the spatially-explicit riskof HWC with the fair market value of the insured goods. For exam-ple, HWC incidents could potentially be linked to a predictable setof habitat features such as wild prey/food availability and live-stock/crop accessibility (Hill, 1997; Naughton-Treves, 1998; Red-path and Thirgood, 1999; Sitati et al., 2005, 2003; Stahl et al.,2002). Moreover, net present value (NPV) could be incorporatedto estimate an actual/potential cost of losing an animal or long-lived cash crop (e.g. rubber, oil palm, coconut), including theexpected future profitability (Yi et al., 2013), Considering risk het-erogeneity and fair market value could lead to spatially variable,effective and equitable premium estimates and payouts. Third, aninsurance-based compensation system could be sustainable if itwere supported by multiple stakeholders, such as communityfunds and locally-generated wildlife revenues through, for exam-ple, ecotourism. Local generation of funds reduces dependenceon external sources, allocates responsibilities for conflicts amongstakeholders, and spreads risk among households when wildlifedamage to individual farmers is highly stochastic and unpredict-able (Dickman et al., 2011; Madhusudan, 2003). For example, acost-sharing mechanism has been used to mitigate human-snowleopard conflict in Pakistan and India (Hussain, 2000; Mishraet al., 2003). Finally, insurance may also prevent moral hazardsby generating incentives for farmers to protect their assets and re-duce false claims as premiums increase with risk, and becausefarmer’s eligibility for insurance can be made conditional on theiradoption of best practices for damage prevention (e.g. Woodroffeet al., 2005).

1.1. Human–elephant conflict in southern China

Driven by the transition to a market-driven economy since the1950s, lowland forest conversion into rubber (Hevea brasiliensis)plantations has become the hallmark of tropical southern China,especially in Xishuangbanna Dai autonomous prefecture. Xishu-angbanna is home to China’s remaining 190–236 wild Asian ele-phants Elephas maximus (Chen et al., 2012). The expansion ofrubber has occurred at the expense of wildlife habitat and the sim-plification of local livelihoods (Fu et al., 2010; Li et al., 2009; Ziegleret al., 2009), and has led to an increase in human–elephant conflict(HEC) owing to frequent damage to rubber plantations. Elephantdamage to rubber is characterized by the killing of seedlings andyoung saplings as elephants pass through the plantations to raidagricultural crops (rice, maize, banana). Feeding on rubber seed-lings and saplings may also happen, as Malaysia reported rubberbeing raided by wild elephants to take in minerals deficient in theirdiets (Blair et al., 1979). Because of the high economic value of rub-ber, depredation by elephants inspires extreme local hostility(Chen, 2012). Farmers are least tolerant of damage to high-valuecash crops (Messmer, 2000), so the increase in the market valueof agricultural land further worsens the HEC found in China. From1918 to 2005, 67 elephant deaths and 7 injuries were recorded asthe result of retaliatory killing and poaching, and from 1991 to2004, 21 human deaths and 91 injuries were attributed to ele-phants (Chen et al., 2012).

A host of technical measures to mitigate HEC have been imple-mented in China since the1990s, including electric fences, trenchesand walls (Chen et al., 2012). However, those attempts provedinfeasible to use at a large scale in Xishuangbanna for various rea-sons. Building and maintaining trenches and walls incur high costs,may have short lifespans in the mountainous landscape, or may

not allow for expansion of private agricultural land. Electric fences,while effective to mitigate HEC in many situations, have often beenpoorly designed and require maintenance, making them ineffectiveand/or costly. Moreover, fences erected around protected areascould disrupt elephant movements through critical conservationcorridors in Xishuangbanna, which link Mengla and Shangyongprotected areas; and Shangyong and Nam Ha (Lao PDR) nature re-serve (Xi, 2009).

In 2009 Chinese authorities in Xishuangbanna began incorpo-rating insurance as a HEC mitigation tool, by replacing the tradi-tional technical tactics with compensation to farmers foragricultural losses caused by elephants. Wildlife mitigation fund-ing from the national government to Xishuangbanna prefectureis budgeted towards payments to a private insurance company.Once crop damage occurs and is reported, insurance agents areresponsible for loss verification and compensation.

Although the existing insurance scheme has brought positivechanges of local attitudes towards wild elephants (Chen, 2012;Chen et al., 2012), recent experience has suggested that centralizedinsurance exhibits poor performance by limiting communityinvolvement, by not utilizing other conservation funds to supportthe payments, and by not considering the spatial heterogeneitiesof conflict intensity and market value of crops (e.g., compensationfor each lost rubber tree is US$ 2/tree regardless of age or produc-tivity; Chen, 2012). Furthermore, funding shortfalls for the HECinsurance scheme threatens its long-term sustainability. For exam-ple, in 2010 there was a budget of US$ 450,000 for the HEC insur-ance scheme; and in 2011 there was a budget of US$ 1 million forall HWC insurance (although the amount allocated to HEC was notspecified in the governmental budget, HEC compensation ac-counted for 87% of the actual cost of the HWC insurance in thisyear; Chen et al., 2012). The allocated budget, in any case, wasinsufficient to cover HEC-compensation, which was as much asUS$ 693,000 in 2010 and US$ 1,371,000 in 2011 (Chen et al.,2012). The insurance company, thus, lost money in both yearsand therefore an improved system to fund this insurance schemeis needed to help mitigate HEC in China.

The costs of rural residents living alongside threatening wildlifemay be shared by the wider beneficiaries, such as tourists (Dick-man et al., 2011). Tourism, as one of the ‘‘non-consumptive butsustainable’’ uses of wildlife, can help to achieve conservationgoals by offsetting local costs of coexistence and enhancing sup-portive conservation attitudes. As a charismatic flagship species,elephants attract tourists and may support a successful and sus-tainable ecotourism project (Kruger, 2005). In Xishuangbanna,tourist visits to Elephant Valley, a famous elephant-orientedattraction, nearly doubled from 2007 to 2012 (0.68 million in2007, 0.83 million in 2008, 1.04 million in 2009 and 2010 and1.21 million in 2011; Xishuangbanna Nature Reserve Bureau,unpublished data). At present, tourism revenues are not allocatedto HEC, nor has there been assessment of tourists’ willingness topay for wild elephant conservation. Tourism revenues may repre-sent a significant opportunity to relieve the financial deficiencyof the current insurance scheme in Xishuangbanna.

In this paper, we propose an actuarial insurance system to helpmitigate the conflict between farmers and wild Asian elephants inXishuangbanna. Our objective in this study was to develop an im-proved insurance-based compensation system by taking into ac-count the spatial heterogeneity of conflict intensity and marketvalue of rubber trees, and by distributing the costs of HEC mitiga-tion across multiple stakeholders who have a vested interest in theprogram’s success. We calculated the full-indemnity insurancepayouts at both the village and town level based on spatially-expli-cit HEC risk and rubber NPV at damage, estimated insurance pay-outs for HEC in the area, and used the results along withwillingness-to-pay data to recommend a sustainable cost-sharing

S. Chen et al. / Biological Conservation 168 (2013) 31–39 33

insurance scheme jointly paid by the government, rubber farmers,and tourists.

2. Methods

2.1. Study area

This study was conducted in Xishuangbanna (19,700 km2,21�090-22�330N, 99�580-101�500E) on the southernmost margin ofYunnan province, bordering Laos and Myanmar (Fig. 1). Xishu-angbanna has very high biodiversity—it is included as part of theIndo-Burma biodiversity hotspot (Myers et al., 2000) and is definedas one of the key Biodiversity Conservation Corridor Initiative focalareas in Greater Mekong Subregion (Xi, 2009). Due to agricultural(mainly rubber) expansion, most of the remaining forest is locatedwithin the boundaries of the Xishuangbanna Biosphere Reserve,which consists of five discontinuous nature reserves in a matrixof different land-use practices (Li et al., 2007) (Fig. A.1). Our studywas carried out in Shangyong Nature Reserve (NR), which bordersLao PDR, harbors 60–80 wild Asian elephants (up to 30% of allremaining elephants in China, Chen et al., 2012), yet which facesintense HEC (Chen, 2012) and involves complex trans-border coor-dination. This study focused on 26 villages of three townshipsaround Shangyong NR that reported recent cases of HEC (Fig. 1).

2.2. Development of a spatially-explicit HEC risk map

We interviewed key informants from Xishuangbanna WildlifeDepartment to obtain information on villages next to ShangyongNR that had experienced HEC between 1991 and 2012. From those

Fig. 1. Map of study area. Top: the geographical location of Shangyong Nature Reserve inthat reported cases of human–elephant conflict between 1991 and 2011.

interviews, 26 villages were identified, and subsequent interviewswere conducted with village heads and patrolmen regardingwhether the village experienced HEC in the previous two years.Among the 26 villages, 12 reported elephant crop raiding occurringbetween January 2010 and January 2012.

In the 12 study villages, we randomly selected a minimum of20% of households in each village (n = 208). From January to March2012, semi-structured interviews were conducted in the local lan-guage, with male or female household heads who worked on thefarms on a daily basis. Surveys captured information on householdsocio-demographic characteristics; rubber mortality by elephantsduring the year 2011; perceptions towards the current insurancescheme; and willingness to pay (WTP) for rubber insurance pre-mium (in CNY/mu, later converted into US$/ha) and to share thecosts with government if full indemnity were guaranteed. ForWTP, each respondent was asked to vote ‘‘yes’’ or ‘‘no’’ for anascending sequence of bids in the form of a double-boundeddichotomous choice question. If the respondent voted ‘‘yes’’ for abid, the next bid was raised, and this process continued until therespondent voted ‘‘no’’ or the highest bid value in the bid list(CNY 60/mu or about US$ 134/ha) was reached. When the respon-dent voted ‘‘no’’ the willingness of the respondent to pay the pre-vious bid was re-confirmed. In addition to WTP, farmers providedinformation on the age, total number of rubber trees, and the num-ber of trees killed by elephants in each plantation for the year prior(2011). Mortality reporting was limited to only the year prior to re-duce recall bias. Subsequently, with the help of experienced farm-ers and forest patrolmen in each village, GPS points were collectedat the locations of the most rubber mortality reported in 2011(n = 7–51 per village).

Xishuangbanna, Yunnan Province, China. Bottom: distribution of the study villages

34 S. Chen et al. / Biological Conservation 168 (2013) 31–39

For each study village, the percent rubber loss of all interviewedhouseholds for 2011 was used to represent that village’s risk to ele-phant depredation. Depredation risk (i.e., conflict risk) was thenmapped and simulated with spatial kriging interpolation in ArcGIS10.1 (ESRI, Redlands, CA, USA).

2.3. Modeling annual payouts for elephant attacks

In a previous study, Guo et al. (2006) used NPV to evaluate themarket value of rubber in Hainan Province, China, applying the fol-lowing equation:

NPVrubber ¼XT

6

RðlatexÞt � ð1þ rÞT�tþRðtimberÞT �XT

0

Ct � ð1þ rÞT�t

" #=½ð1þ rÞT �1�

ð1Þ

where R(latex)t is the revenue from latex tapping in the year t whichis determined by latex production that is age-correlated and influ-enced by management regime and environmental factors; T is therotation age of rubber; R(timber)T is the final harvest income fromtimber; Ct is the annual cost; and r is the social discount rate, whichis a measurement used to guide the resource and public investmentallocations, reflecting relative valuation between today’s well-beingand the one in the future (Zhuang et al., 2007). We selected a socialdiscount rate of 8%, as suggested by the Asian Development Bank(Zhuang et al., 2007).

For this study, the cost of rubber damage (Cdamage) pertained to(1) the cost of replanting rubber seedlings killed by elephants, and(2) the loss of NPV over final years due to delay of latex harvest be-cause all trees—including those replanted after loss—are harvestedat the same time (Fig. A.2). We assumed no loss of timber revenueat harvest. The equation took the form:

Cdamage ¼Xt0

0

Ct0 � ð1þ rÞT�t0=½ð1þ rÞT � 1� þ NPVT�t0 ð2Þ

where t0 is the age of rubber attacked and NPVT�t0 is the loss of NPVwhen the plantation is harvested prior to the end of the tree’s life-span. However NPV is also affected by elevation (Yi et al., 2013), sowe produced a linear regression that included elevation, NPV andage at damage. To do this, we calculated a hypothetical Cdamage foreach plantation in the study by Yi et al. (2013), according to ageat damage (1–4 years) and the previously calculated NPV(Table A.1). A linear regression was then produced, with plantationelevation and age at damage as predictors, and Cdamage as the depen-dent variable. The final regression took the form:

CdamageðUS$=haÞ ¼ 2607:351þ 1810:583 age

� 3:638 elevation ð3Þ

Data from the field surveys were entered into this regression toestimate the village-specific Cdamage, where age was the weightedaverage age at damage across all respondent households in the vil-lage, and elevation was the average elevation of all GPS points ta-ken at damage sites in the village. Cdamage (US$/ha) was alsoconverted to a Cdamage value per tree, based on a typical rubberplanting density of 495 trees/ha (Yi et al., 2013).

The annual (year 2011) payout was calculated as Cdamage multi-plied by the percent rubber loss of 2011. This was converted to avillage-level payout by multiplying the per-ha annual payout bythe projected total area of rubber plantations in the village (themean area for sampled households multiplied by the total numberof households in the village). Town-level payout was also calcu-lated by summing up the gross payouts of villages from the sametown.

2.4. Attitudinal analysis of farmer’s perception towards currentinsurance scheme

Responses from the questions on attitudes towards the com-pensation scheme (see Section 2.2) were ordinal, from the mostnegative (1: Current insurance is worse than the previous HEC mit-igation measures) to the most positive (4: I am fully satisfied withthe current insurance scheme and it should be applied widely). Weapplied classification inference trees (CIT) to the data to identifyvariables shaping farmer perception towards the current insurancescheme. CIT is a nonparametric technique that employs a statisticalstopping criterion to split the data set into binary groups repeat-edly based on the association between covariates and response,until the recursion needs to stop and generate an optimal decisiontree (Giam et al., 2011; Hothorn et al., 2006, 2011). The CIT wasconstructed in R using the package ‘‘party’’ (Hothorn et al., 2011)and we set Monte Carlo P < 0.10.

2.5. Tourist surveys

In March 2012, a tourist survey was conducted at the ElephantValley, a popular tourist attraction in Xishuangbanna, to examinetourists’ willingness-to-pay for HEC mitigation. A total of 210 inde-pendently sampled tourists (123 male, 87 female) were surveyedby semi-structured questionnaires to capture information on theirsocio-demographic characteristics, level of knowledge about ofHEC, and their WTP for HEC mitigation to conserve wild elephantsin China. Most respondents were Chinese tourists from provincesoutside Yunnan (n = 160, 76.2%), and a small proportion of touristswere from overseas (n = 5, 2.4%). The remainder of tourists wasfrom Yunnan province.

3. Results

3.1. Farmer perception towards the current insurance scheme

Overall, only 3.8% (n = 8) of the 208 interviewed farmers werefully satisfied with the current insurance scheme. Most complaintsabout shortcomings were due to ‘‘unreasonable’’ compensationsuch as reimbursement of a fraction value of the market price, inef-ficiency of damage verification, non-timely payment, and low com-pensation ratio (percentage of lost rubber claimed by respondentsthat was reimbursed by the insurance company), followed by rudeservice and perceived unfairness.

In general, the village of residence and compensation ratiodetermined farmer satisfaction towards the current insurancescheme (Fig. 2). About 85% of the farmers in Villages 1 (Shangzhon-gliang), 2 (Manfen_Shang), 7 (Xiazhaongliang), 11(Dashujiao) and12 (Zhongnanxi) perceived the current insurance as worse thanthe previous HEC mitigation measures, or that insurance did notbring any improvement. Villages 7, 11 and 12 received no compen-sation in 2011. In the other study villages, 80% of rubber holderswho had received more than 22% compensation (i.e. a compensa-tion ratio >22%) perceived insurance more positively, e.g. thatinsurance helped to reduce loss effectively, albeit with some short-comings. In contrast, attitudes among those who received 622%compensation were equally positive and negative.

3.2. Rubber loss dynamics and expected payouts

In 2011, elephants reportedly killed a total of 13,071 rubbertrees across the 208 respondent households. Approximately 92%were less than 5 years old and <1% were older than 8 years. Theoverall weighted average age at damage was 2.9 years (Table 1).

Fig. 2. Conditional inference tree explaining farmer perceptions towards the present insurance system to compensate elephant conflict in Xishuangbanna, China. The p-valuerepresents the significance level of the split, and values along node axis are the ‘cut-off’ points for splits. Histograms in terminal nodes are farmers’ ordinal responses definedas follows: (1) current insurance is worse than the previous HEC mitigation measures; (2) insurance does not bring any improvement; (3) insurance helps to reduce losseffectively albeit with some shortcomings and (4) I am fully satisfied with the current insurance scheme and it should be applied widely. The y-axis represents the probabilityof each response.

Table 1Percent rubber loss, Cdamage and expected annual per-ha payout of villages.

Village Percent loss (Risk) (%) Mean elevation of damage (m) Weighted average age t0 Cdamage (US$/ha) Cdamage (US$/tree) Payout (US$/ha)

Shangzhongliang 6.0 858 3.4 5624.1 11.4 339.7Manfen_Shang 5.5 774 2.2 3773.8 7.6 207.6Manfen_Xia 3.6 774 2.2 3773.8 7.6 136.2Hetu 2.1 832 2.7 4449.6 9.0 92.4Nanping 7.0 776 2.6 4388.2 8.9 307.5Hongxingdui 0.7 805 2.8 4801.6 9.7 34.6Xiazhongliang 0.4 711 3.2 5886.7 11.9 26.2Dongyang 0.7 702 4.0 7224.7 14.6 48.8Xiaolongha 1.0 825 2.2 3499.9 7.1 35.9Zhongnanxi 0.0 896 2.0 2968.3 6.0 0.3Dashujiao 0.5 694 4.3 7831.4 15.8 39.8NanxiXincun 1.6 896 4.2 6881.7 13.9 107.1Mean 2.6 795 2.9 4892.5 9.9 126.2

S. Chen et al. / Biological Conservation 168 (2013) 31–39 35

Conflicts were unevenly distributed among study villages, andthe risk of loss ranged from 0.0% to 7.0% of the village’s rubber trees(Table 1, Fig. 3). Villages in the west suffered considerably higherlosses than in the east; risk was also associated with proximityto the reserve border. The greatest rubber mortality from wild ele-phants occurred at 600–950 m elevation.

Based on the elevation, NPV and age-corrected linear regres-sions, the mean cost of elephant-related rubber damage was US$9.9/tree (US$ 126.2/ha, Table 1), i.e. about five times higher thanthe current insurance payment of US$ 2/tree provided to farmers(Chen, 2012).

Commensurate with risk, higher expected rubber payouts werefound at villages to the west of Shangyong NR than to the east (Ta-ble 2, Fig. 3). For the year 2011, the annual expected payout of vil-lages ranged from US$ 0.3/ha to US$ 340/ha, corresponding to totalvillage payouts of US$ 65 to US$ 245,000 (Table 2). At the town le-vel, the estimated gross payouts were US$ 9628 (Mengpeng), US$17,004 (Shangyong) and US$ 471,612 (Mengman) (Table 2, Fig. 3).

3.3. Willingness to pay for insurance

A large proportion of rubber farmers (n = 126, 61%) believed areasonable market-based compensation could mitigate their antip-athy towards elephants and increase their tolerance of coexistence.

Under the assumption of future full loss indemnity if insurancepremiums were shared between the government and community,61% of respondents were willing to pay for insurance, with a meanWTP of US$ 36.2 ± 42.2/ha. There was high variation in willingness:out of the 208 respondents, 39% expressed an unwillingness to payfor insurance; 10% were willing to pay 50–100% of the expectedrubber payout; and 20% were willing to pay more than the full ex-pected rubber payout (Fig. 4). A households’ willingness to contrib-ute any amount to insurance was positively associated with theamount of rubber lost to elephants in 2011 (Table A.2).

The mean village WTP ranged from US$ 2–60/ha, contributingfrom 1.9% to more than 100% of gross village payouts, assumingall households would pay the mean WTP value for their village.The mean town-level WTPs were US$ 27.3–29.8/ha, contributing14.4–100% of the expected gross rubber payout at the town level(Table 2).

In general, Chinese tourists were familiar with, and willing tocontribute to, elephant conservation through financial support ofinsurance. A total of 68% of the 210 interviewed tourists knewabout HEC, while 32% expressed unfamiliarity with the issue. Tour-ists generally supported wild elephant conservation and 90.5%were willing to help the mitigation of HEC in Xishuangbanna bycontributing a donation of income to elephant funds (Fig. 5). Themean WTP per respondent was US$ 12.5 ± 13.2, which represents

Fig. 3. Spatial risk of elephant depredation around Shangyong Nature Reserve in 2011and the anticipated annual insurance payouts. Top: risk gradients of rubber loss toelephants and the per-ha insurance payouts of villages (US$/ha). Bottom: the anticipated goss payouts of towns (US$).

Table 2Gross rubber payouts of villages and towns.

Villages Households Land area(ha)

Unit payout(US$/ha)

Gross payout(US$)

Mean WTP of farmers(US$/ha)

Gross WTP of villages(US$)

Contribution of WTP to grosspayout (%)

Mengman TownShangzhongliang 136 721 339.71 244,931 14 10,094 4.1Manfen_Shang 127 430 207.56 89,251 29 12,470 14.0Manfen_Xia 95 352 136.23 47,953 48 16,896 35.2Hetu 131 476 92.38 43,973 14 6664 15.2Nanping 23 130 307.45 39,969 60 7800 19.5Hongxingdui 74 160 34.6 5536 14 2240 40.5Total 586 2269 – 471,612 – 67,691 14.4

Mengpeng TownXiazhongliang 76 297 26.19 7778 48 14,256 >100Dashujiao 17 35 39.78 1392 10 350 25.1Total 93 332 – 9171 – 9628 100

Shangyong TownDongyang 35 82 48.82 4003 60 4920 >100Xiaolongha 44 155 35.85 5557 33 5115 92.0Zhongnanxi 30 218 0.30 65 14 3052 >100Nanxi Xincun 25 169 107.11 18,102 2 338 1.9Total 134 624 – 27,727 – 17.004 61.3

36 S. Chen et al. / Biological Conservation 168 (2013) 31–39

0.23% of the 2011 per-capita GDP for China of US$ 5447 (The WorldBank n.d.).

4. Discussion

Fair compensation is important to make HWC insuranceschemes successful (Nyhus et al., 2003), yet our results clearly

show that in Xishuangbanna the present payout system of US$ 2per depredated rubber tree is severely inadequate. According toour model, the average payout should be about five times higherand vary according to age at death and site-specific NPV. In a sur-vey of Xishuangbanna farmers, 65% of them prioritized compensa-tion as a future HEC mitigation measure (Chen, 2012), and 61% ofrubber famers in this study stated that they would have more

Fig. 4. Contributions (%) of rubber farmers’ willingness to pay (WTP) to theexpected insurance payout (n = 208). The percetages in the x-axis represent theratio of farmers’ WTP to the calculated insurance payouts they are expected to pay.

Fig. 5. Histogram of Chinese tourists’ willingness to pay for wild elephantconservation. Survey taken in Elephant Valley (n = 210).

S. Chen et al. / Biological Conservation 168 (2013) 31–39 37

positive attitudes towards wild elephants if fair-market compensa-tion were provided, thus supporting economic compensation as apotential HEC mitigation strategy. Moreover, we found substantiallocal and regional willingness to pay for elephant conservation viainsurance, provided that full compensation would be assured. Wetherefore conclude that the current compensation scheme utilizingflat premiums and payouts could be transformed into a more equi-table and economically sustainable actuarial insurance scheme by(1) incorporating spatial heterogeneity of HEC risk and rubber cropNPV in the calculation of both premiums and payouts, (2) creatinga system to share the cost of premiums among different stakehold-ers (government, farmers, and tourists), and (3) achieving long-term buy-in of all stakeholders.

We found that rubber depredation by elephants around Shang-yong NR was an idiosyncratic event, as it is elsewhere (Campos-Arceiz et al., 2009; Hoare, 1999; Sitati et al., 2003), and hence eco-nomic losses varied greatly across space. Consequently, the needfor spatially variable payouts (and likewise premiums) is evident.However, the spatial dynamics of elephant conflict likely vary tem-porally as well (Campos-Arceiz et al., 2009; Chiyo et al., 2005), andso must be incorporated into a multi-year spatiotemporal riskmodel. The case study utilized here was generated on one year ofrubber loss data based on recall of farmers, and is therefore onlyillustrative and not predictive. Nevertheless it demonstrates the

inadequacy of a flat premium and payout system, and lays a solidfoundation for implementation and improvement going forward,as more depredation events are recorded in spatiotemporal dat-abases. It will be crucial in the case of Shangyong NR, as withany system utilizing a spatiotemporal risk-based insurancescheme, to determine whether rubber losses are spatially predict-able across years so that premiums could fairly reflect HEC risk.Fluctuation of rubber price could also affect the estimates of NPVand insurance payoffs. A fair system will require significant invest-ment by the government to introduce a continuous, systematizedHEC monitoring, reporting, and economic analysis system, and byinsurance providers to incorporate those data into dynamic andtransparent calculations of risk-based premiums.

Long-term economic sustainability is a major concern whenestablishing a HWC mitigation insurance scheme (Dickman et al.,2011; Nyhus et al., 2005). In this study we calculated the gross pay-out, which when adjusted for risk should correlate with premiumspaid by villages. However, premiums will also include insuranceadministration costs, profit margins and underwriting charges, aswell as the dynamic effects of timing of cash flows, risk of inflation,and uncertainty of basic economic state variables (Kraus and Ross,1982). Despite our inability to calculate what premiums might befor the villages, the premium associated with an annual gross pay-out of approximately US$ 500,000 for these three townships (Ta-ble 2) would exceed the entire 2010 insurance budget for HEC inXishuangbanna prefecture of US$ 450,000 (Xishuangbanna NatureReserve Bureau, unpublished data). The 2011 governmental budgetfor all wildlife damage was US$ 1 million, highlighting serious con-cern for the economic sustainability and social equitability of theprogram under current conditions.

Thus, our model proposing a partnership between government,local inhabitants and tourists in the formulation of a premiumcost-sharing insurance mechanism could make a significant contri-bution towards fair-market payouts in Xishuangbanna. The dataindicate a large and heretofore untapped source of funding fromboth internal and external stakeholders, which could offer a signif-icant level of support to the program. From our illustrative casestudy, WTP interviews indicated that approximately 18.5% of theentire 2012 insurance premium of Shangyong NR could be contrib-uted by farmer groups, and in low-risk areas of Mengpeng andShangyong, rubber farmers could sustain their own insurancescheme (assuming continuance of risk levels).

In addition, the number of visitors to Elephant Valley has nearlydoubled from 2007 to 2012 (0.68 million in 2007, 0.83 million in2008, 1.04 million in 2009 and 2010 and 1.21 million in 2011;Xishuangbanna Nature Reserve Bureau, unpublished data), indicat-ing substantial potential for tourist-based funding for the mecha-nism. Based on an average of 1 M visitors, the WTP of US$ 12.5/person could raise nearly US$ 12.5 million per year through theextraction of a percentage of entrance ticket fees; although imple-mentation of this strategy must make more conservative projec-tions of expected revenue as the WTP is unlikely to correspondto actual payments. This locally-generated tourist fund could ad-dress the current financial deficiency and support the overall insur-ance sustainability, and this form of conservation payment(Dickman et al., 2011) would successfully spread the costs of livingalongside wild elephants to the wider beneficiaries of conserva-tion. One obvious caveat that could limit the direct implementa-tion of this model elsewhere, is that a significant source ofrevenue through tourism in Xishuangbanna lies within the sameregion as the HEC. Other HWC areas with low potential for ecotour-ism may not have this potential source for insurance payments,and thus a larger portion of total premium payments would fallon local people and the local government, or require novel mecha-nisms to attract funding. One such novel mechanism could beengaging large plantation companies (e.g. oil palm companies in

38 S. Chen et al. / Biological Conservation 168 (2013) 31–39

Malaysia and Indonesia, pineapple companies in Thailand, tea andcoconut companies in India and Sri Lanka) to create funds to sub-sidize HWC premiums for smallholders as part of the companies’corporate social responsibility (CSR) programs.

Distributing insurance payments across numerous stakeholderscould—if managed properly—create incentives to promote long-term sustainability of the scheme. For example, farmers have amajor stake in securing their crops and thus demonstrated awillingness to undertake some responsibilities to protect theirown assets and contribute to insurance premiums. Moreover,because premiums could be tied to risk, farmers might not havean incentive to over-report damage as (a) such behavior could leadto higher premiums in future years, and (b) premiums would bevillage-based and therefore over-reporting could be discouragedthrough village enforcement and social norms.

Importantly, sustainable development of any shared-cost insur-ance system requires buy-in from all participants (Treves et al.,2009). In Xishuangbanna, 39% of farmers indicated an unwilling-ness to pay into the system, largely because of mistrust towardsthe insurance firm and low probability of their rubber plantationsbeing damaged by elephants; and 39% of farmers indicated thattheir antipathy towards elephants was unlikely to improve evenwith fair compensation, mainly due to fear, trauma and other im-posed intangible costs. This echoes experiences in several Africancountries, where compensation programmes for elephant damagesuffered from several deficiencies (African Elephant SpecialistGroup, 2002). An important first step to developing long-termbuy-in by local participants is assurance of equitable and timelypayouts. Our results showed that the major dissatisfaction of farm-ers to the present insurance scheme was the underperformance ofthe insurance firm, characterized by a low compensation ratio(<22%) associated with payment insufficiency and inefficiency. Im-proved transparency on the part of the insurance companies wouldbe required in order to increase confidence of local farmers. Inaddition, improved accountability overall, through the develop-ment of transparent and responsive oversight mechanisms, gov-ernment agencies could enforce regulations applied to insurancecompanies, to conduct on-the-ground monitoring so as to avoidmoral hazards associated with faulty claims, and to offer transpar-ency in allocating tourism-based revenues for elephant conserva-tion (Nyhus et al., 2005).

In order to increase local buy-in, decreasing damage is a prere-quisite for its success in the long run (Swenson and Andrén, 2005),and economic compensation alone is not necessarily enough tochange people’s perceptions and tolerance towards wildlife(Naughton-Treves et al., 2003). This would require ancillary invest-ment from local villages and the government, to reduce direct hu-man–elephant encounters, such as the use and propermaintenance of small-scale barriers around crops, construction ofbiological corridors to link isolated reserves, or reforesting low-profitability rubber plantations in high conflict areas to create nat-ural forest buffer zones, such as the marginally profitable rubberfarms in Nanping Village that are exposed to the highest risk andrequired >US$ 300/ha annual payout. The eligibility for insuranceshould depend on farmers’ commitment to adopt a set of ‘bestpractices’ for damage prevention. For example, the human-wolfconflict mitigation programme is much more effective in Swedenwhere farmers are required to comply with good herding practicesin order to be eligible for compensation, than in Norway wherethere is no such requirement (Swenson and Andrén, 2005). Socio-economic analyses can be included to help understand the reasonsfor low tolerance to wildlife among a part of the population and todesign actions to increase it when appropriate (Banerjee et al.,2013). Buy-in on the part of insurance companies would includeclear commitment to ensure fair, rapid and transparent claims re-sponse and payouts.

Finally, to operate this cost-sharing insurance scheme effec-tively, greater clarity is needed, and agreement reached, on the roleeach stakeholder will play. For example, the local government maytake on the responsibility of raising funds from the national gov-ernment and tourists, and to ensure that the funding is appropri-ately disbursed to not only insurance subsidies, but also tosupport education, outreach, and other mitigation activities (e.g.elephant trenches, walls or electric fences where appropriate andeffective). Insurance firms may be obliged to provide services suchas evaluating annual risk and cost, calculating actuarial premiums,providing compensation standards and allocating central funds tovillages, all with improved accountability through third-partyauditing. Finally, responsibility for raising community funds,administration of central funds and timely damage investigationfor reporting to claims investigators could be decentralized andmanaged by communities through ‘village insurance committees’,which would be trained by, and work in collaboration with, theinsurance firm. These mechanisms could enhance the efficiencyof the insurance programme through the low-cost adjudication, re-duce the administration cost of the insurer and incentivize villag-ers. Similar community-based insurance mechanisms havecontributed to successful human-snow leopard conflict mitigationin Baltistan (Pakistan) (Hussain, 2000) and India (Mishra et al.,2003).

While insurance is not the panacea to resolve HWC (Bulte andRondeau, 2005; Nyhus et al., 2005; Wagner et al., 1997), develop-ing a multi-stakeholder insurance compensation system based onquantitative analysis of both risk and economic valuation of prop-erty loss, would make a significant contribution to enhancing coex-istence of wildlife with rural communities. Such a model could beexported to other systems where elephants or other large herbi-vores produce damage on long-lived, high-value cash crops suchas rubber and oil palm plantations throughout Southeast Asiaand coconut plantations in India and Sri Lanka.

Acknowledgments

We sincerely thank Kay Arnold and Ian Mellshop who providedfunds for this study; and the Xishuangbanna Wildlife Departmentfor coordination and participation in this fieldwork. We thank theElephant Conservation Group (ECG) and Nasharuddin bin Othmanfor their initial help to shape the study. We also appreciate adviceand feedback on numerous aspects of analysis and the manuscriptfrom R. Carrasco, M. Rao, R. Corlett, G. Blackham, D. Friess, and S.Howard. Y.F. Xie, Z.J. Li, Q.K. Dao and X.J. Wang offered great helpin both study design and field work.

Appendix A. Supplementary material

Supplementary data associated with this article can be found,in the online version, at http://dx.doi.org/10.1016/j.biocon.2013.09.017.

References

African Elephant Specialist Group, 2002. Review of compensation schemes foragricultural and other damage caused by elephants. Technical Brief. Human–Elephant Conflict Working Group, African Elephant Specialist Group, IUCN,Gland, Switzerland.

Banerjee, K., Jhala, Y.V., Chauhan, K.S., Dave, C.V., 2013. Living with lions: theeconomics of coexistence in the Gir Forests, India. Plos One 8, e49457.

Blair, J.A.S., Boon, G.G., Noor, N.M., 1979. Conservation or cultivation: theconfrontation between the Asian elephant and land development inPeninsular Malaysia. Land Development Digest 2, 25–58.

Bulte, E.H., Rondeau, D., 2005. Research and management viewpoint: whycompensating wildlife damage may be bad for conservation. Journal ofWildlife Management 69, 14–19.

S. Chen et al. / Biological Conservation 168 (2013) 31–39 39

Bulte, E., Rondeau, D., 2007. Compensation for wildlife damages: habitat conversion,species preservation and local welfare. Journal of Environmental Economics andManagement 54, 311–322.

Campos-Arceiz, A., Takatsuki, S., Ekanayaka, S.K.K., Hasegawa, T., 2009. The human–elephant conflict in sountheastern Sri Lanka: type of damage, seasonal patterns,and sexual differences in the raiding behavior of elephants. Gajah 31, 5–14.

Chen, S., 2012. Mitigating Human–Elephant Conflicts in Xishuangbanna, China.Department of Biological Science, National University of Singapore, Singapore.

Chen, M.Y., Hong, G.Z., Li, Z.L., Dong, Y.H., Yang, Y.Z., 2012. Studies of Conflictbetween Elephant and Human being in China (in Chinese). Yunnan Science andTechnology Press, Kunming.

Chiyo, P.I., Cochrane, E.P., Naughton, L., Basuta, G.I., 2005. Temporal patterns of cropraiding by elephants: a response to changes in forage quality or cropavailability? African Journal of Ecology 43, 48–55.

Dickman, A.J., 2010. Complexities of conflict: the importance of considering socialfactors for effectively resolving human–wildlife conflict. Animal Conservation13, 458–466.

Dickman, A.J., Macdonald, E.A., Macdonald, D.W., 2011. A review of financialinstruments to pay for predator conservation and encourage human–carnivorecoexistence. Proceedings of the National Academy of Sciences of the UnitedStates of America 108, 13937–13944.

Fu, Y., Chen, J., Guo, H., Hu, H., Chen, A., Cui, J., 2010. Agrobiodiversity loss andlivelihood vulnerability as a consequence of converting from subsistencefarming systems to commercial plantation-dominated systems inXishuangbanna, Yunnan, China: A household level analysis. Land Degradation& Development 21, 274–284.

Giam, X., Ng, T.H., Lok, A.F.S.L., Ng, H.H., 2011. Local geographic range predictsfreshwater fish extinctions in Singapore. Journal of Applied Ecology 48, 356–363.

Guo, Z.M., Zhang, Y.Q., Deegen, P., Uibrig, H., 2006. Economic analyses of rubber andtea plantations and rubber–tea intercropping in Hainan, China. AgroforestrySystems 66, 117–127.

Gusset, M., Swarner, M.J., Mponwane, L., Keletile, K., McNutt, J.W., 2009. Human–wildlife conflict in northern Botswana: livestock predation by EndangeredAfrican wild dog Lycaon pictus and other carnivores. Oryx 43, 67–72.

Hill, C.M., 1997. Crop-raiding by wild vertebrates: the farmer’s perspective in anagricultural community in western Uganda. International Journal of PestManagement 43, 77–84.

Hoare, R.E., 1999. Determinants of human–elephant conflict in a land-use mosaic.Journal of Applied Ecology 36, 689–700.

Hothorn, T., Hornik, K., Zeileis, A., 2006. Unbiased recursive partitioning: aconditional inference framework. Journal of Computational and GraphicalStatistics 15, 651–674.

Hothorn, T., Hornik, K., Strobl, C. and Zeileis, A., 2011. party: A Laboratory forRecursive Partytioning. R package version 0.9. Retrieved from <http://www.cran.r-project.org/package=party>.

Hussain, S., 2000. Protecting the snow leopard and enhancing farmers’ livelihoods: apilot insurance scheme in Baltistan. Mountain Research and Development 20,226–231.

Kraus, A., Ross, S.A., 1982. The determination of fair profits for the property–liabilityinsurance firm. The Journal of Finance 37, 1015–1028.

Kruger, O., 2005. The role of ecotourism in conservation: panacea or Pandora’s box?Biodiversity and Conservation 14, 579–600.

Li, H.M., Aide, T.M., Ma, Y.X., Liu, W.J., Cao, M., 2007. Demand for rubber is causingthe loss of high diversity rain forest in SW China. Biodiversity and Conservation16, 1731–1745.

Li, H.M., Ma, Y.X., Liu, W.J., 2009. Clearance and fragmentation of tropical rain forestin Xishuangbanna, SW, China. Biodiversity and Conservation 18, 3421–3440.

Maclennan, S.D., Groom, R.J., Macdonald, D.W., Frank, L.G., 2009. Evaluation of acompensation scheme to bring about pastoralist tolerance of lions. BiologicalConservation 142, 2419–2427.

Madhusudan, M.D., 2003. Living amidst large wildlife: Livestock and cropdepredation by large mammals in the interior villages of Bhadra TigerReserve, south India. Environmental Management 31, 466–475.

Messmer, T.A., 2000. The emergence of human–wildlife conflict management:turning challenges into opportunities. International Biodeterioration &Biodegradation 45, 97–102.

Mishra, C., 1997. Livestock depredation by large carnivores in the Indian trans-Himalaya: conflict perceptions and conservation prospects. EnvironmentalConservation 24, 338–343.

Mishra, C., Allen, P., McCarthy, T., Madhusudan, M.D., Bayarjargal, A., Prins, H.H.T.,2003. The role of incentive programs in conserving the snow leopard.Conservation Biology 17, 1512–1520.

Myers, N., Mittermeier, R.A., Mittermeier, C.G., da Fonseca, G.A.B., Kent, J., 2000.Biodiversity hotspots for conservation priorities. Nature 403, 853–858.

Naughton-Treves, L., 1998. Predicting patterns of crop damage by wildlife aroundKibale National Park, Uganda. Conservation Biology 12, 156–168.

Naughton-Treves, L., Grossberg, R., Treves, A., 2003. Paying for tolerance: Ruralcitizens’ attitudes toward wolf depredation and compensation. ConservationBiology 17, 1500–1511.

Nelson, A. Bidwell, P. and Sillero-Zubiri, C., 2003. A review of humane elephantconflict management strategies. People and Wildlife Initiative. WildlifeConservation Research Unit, Oxford University, Oxford. Retrieved from<http://www.peopleandwildlife.org.uk/crmanuals/HumanElephantConflictP&WManual>.

Nyhus, P., Fischer, H., Madden, F., Osofsky, S., 2003. Taking the bite out of wildlifedamage: the challenges of wildlife compensation schemes. Conservation inPractice 4, 37–43.

Nyhus, P.J., Osofsky, S.A., Ferraro, P., Fischer, H., Madden, F., 2005. Bearing the costsof human-wildlife conflict: the challenges of compensation schemes. In:Woodroffe, R., Thirgood, S., Rabinowitz, A. (Eds.), People and Wildlife: Conflictor Coexistence? Cambridge University Press, Cambridge, pp. 107–121.

Ogra, M., Badola, R., 2008. Compensating human–wildlife conflict in protected areacommunities: ground-level perspectives from Uttarakhand, India. HumanEcology 36, 717–729.

Redpath, S.M., Thirgood, S.J., 1999. Numerical and functional responses in generalistpredators: hen harriers and peregrines on Scottish grouse moors. Journal ofAnimal Ecology 68, 879–892.

Sitati, N.W., Walpole, M.J., Smith, R.J., Leader-Williams, N., 2003. Predicting spatialaspects of human–elephant conflict. Journal of Applied Ecology 40, 667–677.

Sitati, N.W., Walpole, M.J., Leader-Williams, N., 2005. Factors affecting susceptibilityof farms to crop raiding by African elephants: using a predictive model tomitigate conflict. Journal of Applied Ecology 42, 1175–1182.

Stahl, P., Vandel, J.M., Ruette, S., Coat, L., Coat, Y., Balestra, L., 2002. Factors affectinglynx predation on sheep in the French Jura. Journal of Applied Ecology 39, 204–216.

Swenson, J.E., Andrén, H., 2005. A tale of two countries: large carnivore depredationand compensation schemes in Sweden and Norway. In: Woodroffe, R., Thirgood,S., Rabinowitz, A. (Eds.), People and Wildlife: Conflict or Coexistence?Cambridge University Press, Cambridge, pp. 323–339.

The World Bank, n.d. Web. 02 Aug. 2013. GDP per Capita (current US$). Retrievedfrom <http://www.data.worldbank.org/indicator/NY.GDP.PCAP.CD/countries>.

Treves, A., Wallace, R.B., White, S., 2009. Participatory planning of interventions tomitigate human–wildlife conflicts. Conservation Biology 23, 1577–1587.

Wagner, K.K., Schmidt, R.H., Conover, M.R., 1997. Compensation programs forwildlife damage in North America. Wildlife Society Bulletin 25, 312–319.

Woodroffe, R., Thirgood, S., Rabinowitz, A., 2005. People and Wildlife, Conflict or Co-existence? Cambridge University Press, Cambridge.

Xi, J., 2009. Valuation of ecosystem services in Xishuangbanna biodiversityconservation corridors initiative pilot site, China. In: Greater MekongSubregion Core Environment Program. Asian Development Bank, MetroManila, Philippines.

Yi, Z.F., Cannon, C.H., Chen, J., Ye, C.X., Swetnam, R.D., 2013. Developing indicators ofeconomic value and biodiversity loss for rubber plantations in Xishuangbanna,southwest China: a case study from Menglun township. Ecological Indicators:http://dx.doi.org/10.1016/j.ecolind.2013.03.016.

Zhuang, J., Liang, Z., Lin, T., Guzman, F.D., 2007. Theory and practice in the choice ofsocial discount rate for cost-benefit analysis: a survey. Asian DevelopmentBank, pp. 1–50.

Ziegler, A.D., Fox, J.M., Xu, J.C., 2009. The rubber juggernaut. Science 324, 1024–1025.