estimating the benefits of beach recreation: an

Pertanika J. Soc. Sci. & Hum. 3(2): 155-162 (1995) ISSN: 0128-7702© Universiti Pertanian Malaysia Press

Estimating the Benefits of Beach Recreation: An Applicationof the Contingent Valuation Method

NIK MUSTAPHA R. ABDULLAHDepartment of Natural Resource Economics

Faculty of Economics and ManagementUniversiti Pertanian Malaysia

43400 UPM Serdang, Selangor Darul Ehsan, Malaysia

Keywords: benefit, beach recreation, contingent valuation method, dichotomous choice, Iogit,probit, willingness-to-pay

ABSTRAK

Kajian ini melaporkan penggunaan kaedah penilaian kontingen pilihan dua aliran untuk menganggar faedahdaripada rekreasi pantai di Port Dickson. Modellogit dan probit digunakan untuk menganalisis data dan nilaikesanggupan membayar telah dikira daripada anggaran kebarangkalian maksimum. Nilai median kesanggupanmembayar untuk rekreasi pantai di Port Dickson berkisar di antara &\155.01 dan RM534.80 masing-masinguntuk purata pendapatan di antara RM404.56 hingga RM3933.50. Nilai ini boleh digunakan oleh penggubaldasar untuk mengira jumlah faedah tahunan yang diperolehi oleh sumber pelbagaiguna ini.

ABSTRACT

This study presents an application of the dichotomous choice contingent valuation method to estimate thebenefits of beach recreation in Port Dickson. The logit and probit models were used to analyse the data andthe willingness-to-pay values were computed from the maximum likelihood estimates. The median willingnessto-pay value for beach recreation in Port Dickson ranged from RM55.01 to RM534.80 for mean income ofRM404.56 to RM3933.30, respectively. These values can be used by policy-makers to compute the annual totalbenefits obtainable from this multiple-use resource.

INTRODUCTION

Research in outdoor recreation has generated alot of interest to economists as it provides toolsfor the economic evaluation of natural resourcesin recreational use. These tools were developedin response to problems which arise in themanagement of natural resources. Despiteconsiderable work on the economics of outdoorrecreation to measure benefits of national parks(Clawson 1959), sport fisheries (Brown et al.1965), lakes (Burt and Brewer 1971), endangeredspecies (Bowker and Stoll 1988) and other typesof natural resources, there is a paucity ofliterature on measuring benefits of recreationalbeach resources. Beach renourishment, publicacquisition of beachfront and beach carryingcapacity have become important policy issues in

practically all coastal states. Critical to thesepolicy issues is the evaluation of these beachrecreational resources.

The most popular method of measuring thebenefits of outdoor recreational resources is thetravel cost method (TCM). It occupies a majorplace in the applied research programmes ofresource and environmental economists (Smith1993). Some of the works utilizing thistechnique on Malaysian data are by Wan Sabri(1987) and Ahmad Shuib (1993). However, noresearch carried out in this region has everapplied the alternative technique of non-marketvaluation - the contingent valuation method(CVM) to value outdoor recreationalresources. Research to measure the benefitsof beach recreation using either techniqueis almost nonexistence in this country.

Nik Mustapha R. Abdullah

Although some authors have utilized TCMto evaluate beach recreational demand in FIOIida,the results obtained mixed responses fromeconomic studies (Bell and Leeworthy 1990;Shaw 1991; Hofand King 1992). The purpose ofthis study is to make the first attempt of theestimation of benefits from beach recreation inPort Dickson using the contingent valuationmethod. A dichotomous choice form ofcontingent valuation is applied to quantifyindividual economic surplus associated with thepreservation of beach recreation in TelokKemang, Port Dickson.

The Contingent Valuation MethodThe contingent valuation method has been inuse for over 30 years and is now the mostfrequently used approach to estimate nonmarketeconomic values. It is a highly developed surveyapproach to nonmarket valuation. The surveytries to elicit how people would respond tocertain hypothetical changes in environmentalresources. It is a powerful tool for valuingnonmarket goods, so much so that a panelestablished by the ational OceanicAtmospheric Administration ( OAA) in theUnited States co-chaired by Professors KennethArrow and Robert Solow issued a report whichconcluded that "CV studies can produceestimates reliable enough to be the startingpoint for a judicial or administrativedetermination of natural resource damages including loss of passive-use value" (Carson etat. 1993a).

Considerable research has been carried outusing CVM for estimating values for publicdecisions. Some of the more popular studies areby Randall et at. (1974); Bishop and Heberlein(1979); Hanemann (1984); Seller et al. (1986);Bowker and Stoll (1988); and McConnell (1990).The different kinds of work done using CVMcan be found in Mitchell and Carson (1989). Arecent bibliography (Carson et al. 1993b)lists over 1400 contingent valuation studiesand papers from over 40 countries coveringa highly varied empirical applications ofcontingent valuation.

There are several forms of the contingentvaluation method. In this study the dichotomouschoice technique for CVM is used. Therespondents were asked to pay a certain level ofmembership fee in order to use the recreational

resOurces. They were then asked to accept orreject the offer to pay for the use of thesefacilities. The dichotomous choice CVM hasseveral advantages over other forms. It is simpleto administer and respondents are not facedwith intricate bidding schemes and do not haveto contemplate exact values for the resource.The respondents have to respond "yes" or "no"to the dollar offer. In addition, this approachprovides analysis which is consistent with utilitytheory (Hanemann 1984).

However, the dichotomous choice modelprovides researchers with limited informationfrom the respondents and therefore requiresmore sophisticated statistical and estimationtechniques to analyse the qualitative responses.In addition, this technique requires theappropriate range of offer levels of the goodsbeing valued and this has to be determined inadvance in order to capture it in the data foranalysis. In most cases these offers arearbi trarily set and usually done in an ad hocmanner. Although CVM is argued to sufferfrom theoretical biases such as strategic andselection bias (Cummings et at. 1986), butinterviewer bias and starting point biascommon to iterative bidding are minimized inthis dichotomous choice approach (Bowkerand Stoll 1988).

METHODOLOGY

Hanemann (1984) has proposed the theoreticalmodel from which Hicksian compensating andequivalent surplus measures are obtained fromdichotomous choice, discrete response data. Inthis paper Hanemann's approach is followed ina willingness-to-pay framework to obtain ameasure of individual's equivalent surplus. It isassumed that the individual derives utility fromparticipation in beach recreation and moneyincome. Variable r represents recreation, wherer=l if the individual is able to recreate and r=Oif he is not. Income is represented by Y andother attributes of the individual which mayaffect his preferences are denoted by vector d. Ifhe can use recreational facilities the individual'sutility is U

I= U(1,Y,d); ifhe cannot, his utility is

Uo

= U(O,Y,d). Since there are unobservablerandom components to an individual's utilityfunction, utility is treated as a random variablewith a parametric probability distribution withmeans v (O,Y,d) and v (1,Y,d), and the

]56 PertanikaJ. Soc. Sci. & Hum. Vol. 3 No.2 1995

Estimating the Benefits of Beach Recreation

where the vector d has been suppressed, then

v U, Y; d) = exj

+ (3Y , ex > 0; j = 0, 1 (3)

The willingness-to-pay probability may bewritten as:

stochastic components e,_ , which are independentand identically distributed (i.i.d) randomvariables with zero means.

When confronted with a price, $A, toenter and use the recreation area, anindividual will pay the amount only if

Estimation Procedure

Central to the whole exercise is the estimationof parameters which define the willingness-topay probability function (FTJ). Several quantitativeresponse models have been used by researchersand this includes the linear probability model(LPM), the logit model (LM) and the probitmodel (PM) (Judge et al. 1980; Amemiya 1981;Maddala 1983). In this paper, a transformationapproach using an index variable Zi (3,representing the utility difference [equations(4) and (6)] is employed. The bigger Zi areassociated with higher probabilities that the event(Y) will occur, i.e. Y=I. Therefore, there exists am~notonic relationship between the probabilityof the event occurring and the index variable.Under such circumstances, the true probabilityfunction would resemble a distribution function.The two most commonly used distributionfunctions are the normal and the logistic,resulting in the probit and logit models.

In many economic applications, however,the standard logistic distribution function,

Estimating the parametric probabilityfunction FTJ (~v (A)) allow us to obtain estimatesof the desired welfare measure.

(1)

(2)PI = FTJ (~v)

where ~v = v (1, V-A; d) - v (0, Y; d) andFTJ is the probability function for the error.If the argument ~v is a utility difference,then the binary response model can beinterpreted as the outcome of a utilitymaximizing choice (Hanemann 1984).

Hanemann (1984) also suggested explicitspecification of the non-random component ofthe indirect utility function to be:

v (1, V-A; d) + E1 > (0, Y; d) + Eo

which gives

v U, Y; d) = exj

+ (3 In Y, (3 > 0; j = 0,1 (5)

and the statistical discrete choice model becomesP = FTJ (ex + (3A) where ex = (exo - ex l )·I .The other suggested functional form IS

(9)( I)1+---e(-Xi~)

Study Area

This study was carried out in Port Dickson,Negeri Sembilan. The whole district of PortDickson has an area of 572.63 sq. km and apopulation of 101707 people. This is the onlydistrict in Negeri Sembilan that faces the Straits

is assumed for FTJ. The logistic distribution closelyapproximates the normal and is relatively easierto estimate (Judge 1980). Both the logit andprobit models are used in this study. Theparameters for the binary response models caneither be estimated using generalized leastsquares (Pindyck and Rubinfeld 1976) ormaximum likelihood method (Capps and Kramer1985; Hanemann 1984). In this study the latterestimation technique is employed. Twodifferent specifications for the index variableZi = Xi(3 for equations (4) and (6) are usedhere. Each specification is estimated usingboth logit and probit models.

(6)

(4)

(8)FTJ (E\IIJ) = 0.5

From these specifications and assuming thatthe equivalent surplus is random with aprobability distribution (GE (A)), estimate ofequivalent surplus is obtained by using theexpected value of E, E~IN where

EMN = f Fry (L1v(A)) dA (7)o

Alternatively, an approximation of equivalentsurplus may be obtained by using the medianvalue, E~lD' of the distribution GE (A), where

PertanikaJ. Soc. Sci. & Hum. Vol. 3 No.2 1995 157


of Malacca and has about 54 km of beachstretching from Sungai Sepang in Selangor toSungai Linggi in Melaka. Telok Kemang beach,12 km from Port Dickson town, is one of themore popular beach recreation sites in WestMalaysia. Its sandy beach has attracted andcontinues to attract millions of holiday-makersto this area, especially on weekends and publicholidays, much more than any other beachresorts elsewhere. AI; many as 30,000 picnickersflock to this area on a given weekend. It isstrategically located between Kuala Lumpur inthe north and Singapore in the south, easilyaccessible from the north-south highway and iswell-served with good roads and public transport.

Source of DataThe data were collected through on-site surveysin November 1991 using a structuredquestionnaire. A total of 188 domesticrecreationists, all ofwhom were day-trippers, wereinterviewed at the beach recreational site and 169questionnaires were used for further analysis. Thestatistics of the respondents are shown in Table 1.

About 52.7% of the respondents weregovernment servants. The average monthlyincome of the respondents was about MR940.The majority of the respondents were 21-39years, age category and travelled about 75 kmto the beach. Each interviewee was informedof the purpose of the study with respect to theclub membership, the facilities available inthe study area and the format used in thecontingent valuation technique. Following thebriefing, the respondents (the head of the

group or family) were asked to give a "Yes" or"No" response to the question. Example:

"If the annual membership fee costsRMx in 1991, would you have joinedthe club so that you could havecontinued to use the beach recreationalarea throughout the year"?

where x ranged from RMI to RM500. Thesevalues were arbitrarily selected as they representa "reasonable" amount of entrance fee to manyprivately managed recreational areas in thiscountry. Furthermore, there was no comparativevalue to refer to as this was the first attempt atusing CVM in valuing benefits of beachrecreation in this region.

RESULTS

Equation (4), represented by Model 1 andequation (6) in Model 2, were estimated usingboth the logit and probit models. Thelikelihood estimates of these two specificationsare shown in Table 2.

The estimates of both logit and probitmodels for the two specifications (Models 1and 2) did not differ very much in terms of thesummary statistics and parameters significance.All variables had the expected results and weresignificant at 1% level. This finding is consistentwith earlier work in the binary dependentvariable case (Bowker and Stoll 1988). Model 2,however, had a slight edge over Model 1 interms of McFadden-R2 and it was argued to bemore preferable than the latter model as"income effect" was present, as expected in the

158

TABLE 1Some descriptive statistics of the respondents

Variable N Mean Std. Deviation Max Min

Income (RM) 176 939.7 620.42 4800.0 200.0

Group/Family

size 188 5.4 3.14 32 2

Distance

travelled (km) 171 75 49.7 350 1.0

Expenditure (RM) 183 57.8 108.9 999.0 1.0

PertanikaJ. Soc. Sci. & Hum. Vol. 3 No.2 1995


utility model. Based on these criteria, model 2was chosen for the subsequent measure ofwelfare from the consumers' willingness-to-payfor beach recreation in Port Dickson.

Welfare Measurement

From the estimates of Model 2 given in Table 2,the equivalent surplus welfare measures can becomputed. The mean values were calculated bynumerically integrating the area under eachestimated willingness-to-pay function over therange of offer amount. Income was set as itssample mean. The computed mean and medianvalues for income groups are presented in Table 3.The mean for willingness-ta-pay ranged fromRM63.83 to RM620.58 for a mean income ofRM404.56 to RM3933.30 and RM71. 74 toRM597.48 for the same mean income in logitand probit models, respeGtively. The medianwillingness-to-pay, meanwhile, varied fromRM55.01 to RM534.80 for the logit specificationand from RM60.54 to RM588.36 for the probitmodel. For all income groups, the mean

willingness-to-pay was found to be about 15%higher than the median equivalent surplusmeasures and this is consistent with the findingof Hanemann (1984).

However, the logit specification in Model 2performed slightly better than the probit modelin terms of McFadden-R2 (Table 2). Based purelyon this criterion, the mean and the medianwillingness-to-pay derived from the logit modelcan be considered as a more reliable measure ofwelfare. Nevertheless, one has to make a choicebetween which measure of VVTP to use. Froman economic point of view, the selection ispurely a value judgement. Hanemann andBowker and Stoll preferred the median WTPas a welfare measure as it is statistically morerobust than the mean WTP measure. For thisreason the median VVTP values will be used inthis study to represent the equivalent surplusmeasure for beach recreation in Port Dickson.

The median WTP computed from thelogit model varied from RM55.01 for a meanincome of RM404.56 to as high as RM534.80

TABLE 2Parameter Estimates for Dichotomous Choice CVM Models for Port Dickson Beach

Modell Model 2

Variable Logit Probit Logit Probit

Constant 1.1818 0.7016 1.1879 0.6391

(4.372)* (4.363)* (4.388)* (4.130*)

Price -0.0120 -0.0067

(-6.244)* (-6.9864)*

pricelog (1- -.-- ) 0.3181 4.5834

ll1come(5.708)* (6.535)*

McFadden R2 0.336 0.331 0.366 0.344

Log-likeli hood

ratio 76.18 75.09 83.07 78.08

% of correct

prediction 0.83 0.78 0.81 0.81

N 169 169 169 169

Note: t - statistics are given in the parenthesis

* - significant at 1% level

PertanikaJ. Soc. Sci. & Hum. Vol. 3 No.2 1995 159

for a mean income of RM3933.30. Thisestimate is, however, for one individual andexpressed in an annual basis. Taking a veryconservative estimate of RM55 which by nomeans reflects the average value of WTP, asan individual annual benefit of using beachrecreational resources in Port Dickson, thetotal annual benefit from beach recreationcan be very substantial if the total annualvisits to this area is taken into account.

Suppose there were 300,000 visitors to thePort Dickson beach in 1993, and using MR55 asan example of a willingness -to-pay measure forthis recreational resource, then the annual totaleconomic benefits that can be obtained fromthis resource comes to about RM1.65 million.This value represents both consumers'consumptive and non-consumptive use of theresource. If for some reason, the beachrecreation area is turned into some commercialproject or put to other uses, then the annualopportunity cost of this new development willbe the total economic benefits that thisintangible resource is capable of generatingfor that particular year. These economicbenefits carefully discounted must then becompared with benefits arising from newprojects to justify the rational use of the sameresource.

CONCLUSION

Attempts at valuing outdoor recreationalresources in Malaysia have been sketchy and onan ad hoc basis. Previous works used the indirectvaluation technique - the travel cost method tovalue the economic benefits of the theseresources. This study presents an alternativevaluation technique, the contingent valuationmethod (CVM) and is the first attempt at CYMto value beach recreational resources in Malaysia.The dichotomous choice form used in CVM ispurported to reduce some inherent biases thatoccurred in the CYM techniques and it isconsistent with consumers' utility maximizingbehaviour.

The binary response models estimated bylogit and probit specifications and the maximumlikelihood estimates were satisfactory andconsistent with expectations. The medianwillingness-to-pay (WTP) measure was chosento represent the reliable equivalent surplusvalues for beach recreation in Port Dickson.The median WTP computed from the logitmodel ranged from RM55 for a mean incomeof RM404.56 to RM535 for a mean income ofRM3933.30. These values represent reasonableestimates of economic benefit per individualper year obtainable from the resource.However, since this is the first attempt to use

160 PertanikaJ. Soc. Sci. & Hum. Vol. 3 No.2 1995


CYM in valuing beach recreation in Malaysia,the results and conclusions must be interpretedwith caution until further such studies arecarried out. Taking a very conservative estimateof median WTP value at RM55, the annualtotal economic benefits derivable from thisresource can be very substantial if one takesinto account the total number of visitors toPort Dickson beach for a particular year. Thisvalue should reflect the "true" value of theresource as it includes both consumptive andnon-consumptive aspects of this multi-purposeresource. From a policy-making point of view,the total economic benefits could be used as abench mark for alternative development ofthe area to make it economically attractiveand justifiable.

However the shortcomings in using CYMare biases with regards to the responses bythose interviewed. Although a lot of effort hasbeen undertaken in past years to minimizethese biases, much more refinement is neededto improve this method. More work is neededespecially in this part of the world with differentcultural settings, before using CYM to get"reasonable" and reliable estimates of value ofnon market goods for policy uses in this country.

ACKNOWLEDGEMENTSThis research was partially funded by UniversitiPertanian Malaysia through its short-term researchgrant. The author would like to thank Mrs.Normarziana bt. Hasan for the analysis of data.

REFERENCESAHMAD SHUIB.1993. Faktor pesaing di dalam

penilaian permintaan rekreasi luar. Pertanika].Soc. Sc. Hum. 1(2): 199-204.

A'vIEMlYA, T. 1981. Qualitative response models: a survey.journal of Economic Literature 19: 1483-1536.

BELL, F.W. and V.R. LEEWORTHY. 1990. Recreationaldemand by tourists for saltwater beach days.journal of Environmental Economics andManagement 18: 189-205.

BISHOP, R.e. and T.A. HEBERLEIN. 1979. Measuringvalues of extramarket goods. American journalof Agricultuml Economics 61: 926-930.

BOWKER, J.M. and J.R. STOLL. 1988. Use ofdichotomous choice nonmarket methods to

value the whooping crane resource. Americanjournal of Aglicultuml Economics 70: 372-381.

BROWN, W.G., SINGH and E.N. CASTLE. 1965. Neteconomic value of the Oregon salmon-stell headsport fishery. journal of Wildlife Management 29:266-279.

BURT, O. and D. BREWER. 1971. Estimation of netsocial benefits from outdoor recreation.Econometrica 39: 813-827.

CAPPS, 0., JR. and R.A. KRAMER. 1985. Analysis offood stamp participation using qualitativeresponse models. AmericanjoU"lnctl ofA!§icultumlEconomics 67: 49-59.

CARSON, R.T., .F. MEADE and V.K. SMITH. 1993a.Introducing the issues. Choices. Second Quarter:5-8.

CARSON, R.T., N. CARSON, A. ALBERJNl, N.E. FLORESandj. WRJGHT. 1993b. A Biblio!§llPhy ofContingentValuation Studies and Papers. La Jolla, CA:Natural Resource Damage Assessment, Inc.

CLAWSON, M. 1959. Methods of Measuring the Demandforand Values ofOutdoor Recreation. Washington, D.e.:Resources for the Future, Inc.

CUMMINGS, R.e., D.S. BROOKSHIRE and W.D. SCHULZE(eds). 1986. Valuing Public Goods: An Assessmentof the Contingent Valuation Method. Totowa NJ:Rowan & Allanheld.

HANEMANN, W.M. 1984. Welfare evaluation incontingent evaluation experiments withdiscrete responses. American journal ofAf5Iicultuml Economics 66: 332-341.

HOF,j.G. and D.A. KING. 1992. Recreational demandby tourists for saltwater beach days: Comment.journal of Environmental Economics andManagement 22: 281-291.

JUDGE, G.G., W.E. GRIFFTTHS, R.e. Hill. and T.e. LEE.1980. The Theory and Practice of Econometlics.New York: Wiley.

MADDAL~, C.S. 1983. Limited Dependent and QualitativeVariables in Econometlics. Cambridge: CambridgeUniversity Press.

MCCONNELL, K.E. 1990. Models for referendum data:The structure of discrete choice models forcontingent valuation. journal of EnvironmentalEconomics and Management 18: 19-34.

MITCHELL. RC. and R.T. CARSON. 1989. UsingSwveysthe Value Public Goods: 77le Contingent Valuation

Pertanika J. Soc. Sci. & Hum. Vol. 3 TO. 2 1995 161


Method. Washington, D.C.: Resources for theFuture, Inc.

PINDYCK, R.S. and D.L. RUBINFELD. 1976. EconometricModels and Econometric Forecasts. ew York:McGraw-Hill.

RANDALL, Au.N, B.C. rVES and C. EASTMAN. 1974.Bidding games for valuation of aestheticEnvironmental improvement. Journal ojEnvimnmental Economics and Management 1:132-149.

SELLER, C., J.P. CHAVAS and J.R. STOl.L. 1986.Specification of the logit model. The case ofvaluation of non market goods. Journal ojEnvironmental Economics and· Management 13:382-390.

SHAW, D.W. 1991. Recreational demand by touristsfor saltwater beach day: Comment. Journal ojEnvillmmental Economics and Management. 20:284-289.

SMITH, V.K. 1993. Nonmarket valuation ofenvironmental resources: An interpl-etiveappraisal. Land Economics 69: 1-26.

"VAN SABRI, W.M. 1987. Forest recreation usepatterns, user behaviour and recreational valuein Malaysia. Ph.D. Thesis, University of Wales,Bangor.

(Received: 14 May 1994)

162 Pertanika.J. Soc. Sci. & Hum. Vol. 3 No.2 1995

estimating the benefits of beach recreation: an

Documents