group 4.April 23_Travel cost_Duck hunting 2002.pdf

This article was downloaded by:[Australian National University Library]On: 4 February 2008Access Details: [subscription number 773444558]Publisher: RoutledgeInforma Ltd Registered in England and Wales Registered Number: 1072954Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

Australian GeographerPublication details, including instructions for authors and subscription information:http://www.informaworld.com/smpp/title~content=t713403176

A Travel Cost Study of Duck Hunting in the Upper SouthEast of South AustraliaStuart M. Whitten; Jeff W. Bennett

Online Publication Date: 01 July 2002To cite this Article: Whitten, Stuart M. and Bennett, Jeff W. (2002) 'A Travel CostStudy of Duck Hunting in the Upper South East of South Australia', AustralianGeographer, 33:2, 207 - 221To link to this article: DOI: 10.1080/00049180220151016URL: http://dx.doi.org/10.1080/00049180220151016

PLEASE SCROLL DOWN FOR ARTICLE

Full terms and conditions of use: http://www.informaworld.com/terms-and-conditions-of-access.pdf

This article maybe used for research, teaching and private study purposes. Any substantial or systematic reproduction,re-distribution, re-selling, loan or sub-licensing, systematic supply or distribution in any form to anyone is expresslyforbidden.

The publisher does not give any warranty express or implied or make any representation that the contents will becomplete or accurate or up to date. The accuracy of any instructions, formulae and drug doses should beindependently verified with primary sources. The publisher shall not be liable for any loss, actions, claims, proceedings,demand or costs or damages whatsoever or howsoever caused arising directly or indirectly in connection with orarising out of the use of this material.

Dow

nloa

ded

By:

[Aus

tralia

n N

atio

nal U

nive

rsity

Lib

rary

] At:

03:4

2 4

Febr

uary

200

8

Australian Geographer, Vol. 33, No. 2, pp. 207221, 2002

A Travel Cost Study of Duck Hunting in theUpper South East of South Australia

STUART M. WHITTEN & JEFF W. BENNETT, University of New South Walesand The Australian National University

ABSTRACT Wetlands in the Upper South East (USE) of South Australia yield a range ofvalues to their owners and to the wider community. One such value is that generated by duckhunting in the region. Wetland owners receive private bene ts from hunting fees, while huntersgain a non-monetary consumers surplus. These non-monetary bene ts received by duckhunters through their enjoyment of hunting in wetlands are the primary focus of this paper. Inorder to estimate these values, a travel cost survey of duck hunters participating in anorganised shoot held by Wetlands and Wildlife in February 2000 was undertaken. The extentof the private recreation bene ts so estimated is assessed in terms of their potential to stimulatethe provision of wetlands for both private and social bene ts.

KEY WORDS Travel cost method; duck hunting; wetland values.

Introduction

A large area of ecologically signi cant wetlandsup to 20 000 hais located in theregion between Naracoorte and Tintinara known as the Upper South East (USE) ofSouth Australia. The location of major wetland concentrations and the region is shownin Figure 1. The wetlands, the majority of which are located on privately owned land,provide a range of values to their owners and wider society. These include commercialbene ts from the grazing of sheep and cattle, through to the enjoyment gained bypeople who like to know that the area continues to maintain the viability of populationsof endangered species.A range of bene ts is generated from duck hunting in the wetlands of the USE.

Wetland owners who provide access to hunters for a fee enjoy a commercial bene t.Hunters enjoying their sport in the region gain enjoyment from the experience.Furthermore, it can be argued that through the maintenance of the wetlands as duckhabitat, broader wetland conservation values are provided to the wider community.However, duck hunting remains controversial because of the costs it creates formembers of the community who object on animal welfare grounds. For example, sportshooting for ducks is banned in New South Wales (NSW) in part because of concernsabout the cruelty to the ducks hunted.The results presented in this paper are aimed at providing more information about

the extent of the bene ts generated by duck hunting in the USE. This information isuseful in the formulation of government policy relating directly to duck hunting and in

ISSN 0004-9182 print/ISSN 1465-3311 online/02/020207-15 2002 Geographical Society of New South Wales Inc.DOI: 10.1080/00049180220151016

Dow

nloa

ded

By:

[Aus

tralia

n N

atio

nal U

nive

rsity

Lib

rary

] At:

03:4

2 4

Febr

uary

200

8

208 S. M. Whitten & J. W. Bennett

FIGURE 1. Location of Upper South East region and major wetlands.

the determination of plans for future wetland management in the region. It isspeci cally targeted at the estimation of values enjoyed by duck hunters.The next section of this report sets out the selection of an appropriate valuation

methodology and a brief literature review. The methodology selectedthe travel costmethodis developed in the third section of the paper, including the development andimplementation of the survey instrument used to apply the method. The resultingestimates of values resulting from duck hunting in the USE are reported in the fourthsection. Some conclusions regarding the potential usefulness of the estimated values tothe development of policies for wetland planning and management are drawn in the nal section of the paper.

Dow

nloa

ded

By:

[Aus

tralia

n N

atio

nal U

nive

rsity

Lib

rary

] At:

03:4

2 4

Febr

uary

200

8

A Travel Cost Study 209

Method selection and nature of the values to be estimated

The bene ts enjoyed by duck hunters are essentially generated outside the operation ofmarkets. They are, therefore, non-monetary bene ts. However, they are directly relatedto actions in the marketplace. For example, the decision to hunt is directly related tothe actions of hunters spending money on petrol, food, hunting fees and other items inorder to enjoy their sport.The travel cost method (TCM) is a technique that enables the valuation of

non-market-related goods such as hunting. It uses the assumption that theincurred costs of visiting a site in some way re ect the recreational value of that site(Turner et al. 1994, p. 116). By estimating this relationship, the value of thewetlands as a site for a recreational activity (e.g. hunting) can be estimated. Hence, theTCM method is suitable to the estimation of consumers surplus from hunting inthe USE.The TCM is a well-established technique that has yielded relatively consistent and

reliable results (Bennett 1995). An especially attractive aspect of the technique is theability to cost-effectively collect the data during registration for the season opening huntorganised by Wetlands and Wildlifea not-for-pro t company that owns wetlands inthe USE. A number of studies (particularly in the USA) have used the TCM togenerate estimates of hunting and shing values including: Cooper and Loomis (1993),Cooper (2000) and Offenbach and Goodwin (1994). Cooper and Loomis (1993)estimated a willingness to pay of $15.62 ($US 1993) and $26.21 ($US 1993) perwaterfowl hunting trip to National Wildlife Refuges in the San Joaquin Valley inCalifornia using an OLS- and Poisson-based model, respectively. Offenbach andGoodwin (1994) estimated a willingness to pay of $160.79 and $176.55 ($US 1994)per hunting trip for hunting trips (waterfowl and other game) in Kansas usingtwo alternative model speci cations. Cooper (2000) used non-parametric and semi-non-parametric techniques to re-analyse the Cooper and Loomis (1993) data andproduced similar estimates of consumers surplus per hunting trip to the originalestimates.The TCM involves the estimation of the relationship between the recreational

service provided (in this case duck hunting) and the purchase and use of goods andservices by duck hunters who travel to the site. The TCM assumes weak complementar-ity between the expenditure on goods and services and the recreational service (Hanley& Spash 1993). The implication of this assumption is that when consumption expendi-ture is zero, the marginal utility (and hence consumers surplus) of the good is alsozero. In other words, the consumers surplus of the furthest distant wetland visitorsapproaches zero because the costs of the visit almost equal the bene ts enjoyed fromthe trip. People living further away, with higher costs, do not visit because they wouldnot enjoy any surplus. A second assumption made under the methodology is that theutility function of duck hunters is separable (Hanley & Spash 1993). That is, thedemand for duck hunting can be estimated independently of the demand for otheractivities (both recreational and non-recreational). Finally, the TCM method usedmakes no distinction between a good hunting trip and a bad hunting trip. That is,there is no difference in the value estimated if the number of ducks shot per hunter ishigh or low. This is because the TCM methodology is based on hunters expectationsof trip quality.Speci cally, the relationship estimated is between the number of people hunting per

Dow

nloa

ded

By:

[Aus

tralia

n N

atio

nal U

nive

rsity

Lib

rary

] At:

03:4

2 4

Febr

uary

200

8


head of population and the costs of travelling to and from that site. This gives the tripgeneration function (TGF):

H 5 f (TC).

The rate of participation in duck hunting (H) is expected to fall as the costs oftravelling, potentially including the travel time (TC) increase. Hence, the number ofduck hunters per head of population is dependent on the costs of participating in thehunt. Under the assumptions of the method, the imposition, or increase, of a partici-pation fee would increase TC and reduce the rate of duck hunting. By simulating theeffects of an increase in the participation fee, TC are increased and the impacts on totalvisitor numbers estimated. That is, a demand curve (the relationship between quantityof duck hunters and price of duck hunting) is derived. The value of hunting to duckhunters (their consumers surplus) is equal to the area under the curve and above thefee charged to participate in duck hunting.

Research design

Travel costs are de ned as the costs incurred by hunters in order to participate in duckhunting in the USE. Economic theory refers to this as the opportunity costthe valueof the alternative foregone in order to participate in duck hunting in the USE. Costs canbe split between travel costs and time costs (Bateman 1995). Time costs can be furthersplit between travel time and on-site time.Bateman (1995) suggests three cost calculation options for travel costs:

(1) petrol and additional costs only (marginal costs);(2) full car costs (that is petrol, insurance, maintenance costs, etc.) and full additional

costs; and(3) perceived costs as estimated by respondents.

Use of option (2) will increase costs above (1) and hence increase the resultingconsumers surplus estimates. Bateman argues that the correct cost is that perceived byrespondents as pertaining to the visit, and this de nition was used for the study. Forexample, respondents may perceive daily insurance and maintenance costs as sunkcosts that are not traded-off against alternative uses.1

Inclusion of the cost of time spent travelling to the hunting site is debated in theTCM literature. Hunters travelling to, and participating in, duck hunting are giving upthe opportunity to participate in some other activity but they may also enjoy the trip.Bateman (1995) suggests that where time costs are thought to be important, asensitivity analysis should be conducted using values of 0.25, 0.5, 0.75 and the fullwage rate. Time spent on-site is determined exogenously. At the margin, time spenton-site is expected to generate utility equal to that from alternative activities. Hence,time spent on-site is treated as having no impact on the consumers surplus estimates.The relationship between the frequency of duck hunting and travel costs can be

analysed using either individual visitor data or data averaged across a number ofpopulation groups referred to as zones. The zonal approach de nes the visitation rateas the proportion of hunters from the population of a speci c geographic area, perpre-speci ed time period. The individual visitation rate is the number of visits perindividual per time period. A trade-off arises between the bene ts to model accuracy ofthe additional detail that is captured by the individual approach and the dif culties ofobtaining suf ciently detailed data and variability in the number of visits undertaken by

Dow

nloa

ded

By:

[Aus

tralia

n N

atio

nal U

nive

rsity

Lib

rary

] At:

03:4

2 4

Febr

uary

200

8


individuals to achieve model validity. The zonal approach was used in this applicationbecause data were collected at a single hunt and it was believed that most hunters makerelatively few hunting trips to the USE (tested via including a question about tripnumbers and destinations).A basic assumption of the TCM is that each hunter makes the trip solely to hunt

ducks in the USE. The majority of hunters are hypothesised to be travelling only toparticipate in duck hunting in the USE because duck hunting is primarily a weekendrecreational activity and because the ducks that are taken need to be frozen relativelyquickly. The presence of substitute sites can also impact on visitor demand via theirtravel cost (or price), their entry fees and their relative quality (Bateman 1995). Aquestion was included in the survey to determine whether the issue of substitute sitesis a problem.

Survey design and implementation

A TCM study of duck hunting in the USE region ideally would involve the collectionof data from all hunters over a complete hunting season. Data collection over severalseasons would be required to reduce the impact of individual hunting seasons on theoverall visitation patterns. For example, 2000 was the fth in a succession of dry years.A suitable substitute for data collection from all hunters is collection from huntersparticipating in Wetlands and Wildlife organised hunts. Wetlands and Wildlife ownsby far the largest area of wetlands suitable for duck hunting in the USE and holdsseveral large-scale organised hunts in a typical season. The reduced quality of thewetlands noted above is re ected in the number of hunters who participated in duckhunting in Wetlands and Wildlife wetlands during the year 2000 open season. Only555 hunters participated in 2000 vs an average 1156 over the past 13 years. The largestnumber of duck hunters usually participate in the season opening shoot organised byWetlands and Wildlife. A survey was therefore designed for implementation at theyear 2000 event.The intended survey methodology determines, in part, the design and structure of

the questionnaire. In order to maximise response rates and minimise respondent costit was decided that the questionnaire would be distributed to all duck hunters as partof the registration procedure at the Wetlands and Wildlife Organised Shoot. Com-pleted questionnaires would then be collected either immediately or prior to huntersleaving the event.As the survey was being distributed as part of registration procedures, minimal

introduction of the questionnaire and explanation of the purpose of the survey wasrequired in the document. Questions by hunters about the survey were answered by anassistant2 who had been briefed on the surveys purpose and structure. The question-naire consisted of the following sections:

a short preamble including who is collecting the information and the use of thatinformation;

seven questions gathering data required for a TCM analysis; a short section thanking respondents and indicating who to contact for additionalinformation; and

opportunity for additional feedback.

The basic questionnaire was based on Bennetts (1995) suggested simpli ed format.Draft versions were reviewed by Wetlands and Wildlife of cers and provided for

Dow

nloa

ded

By:

[Aus

tralia

n N

atio

nal U

nive

rsity

Lib

rary

] At:

03:4

2 4

Febr

uary

200

8


comment to local Department of Environment and Heritage of cers. Minor wordingchanges were made to the survey after comments were received.The questionnaire was designed to collect the following information:

(1) Respondents usual place of residence (Question 1).(2) Method of transport to the event (own transport, with friends or other) (Question

2).(3) The number of people with whom they had travelled (Question 3).(4) The range of potential substitute hunting areas they had accessed in the previous

12 months (Question 4).(5) Self-estimated costs of participating in the event (Question 5).(6) How long it took to travel to the event (Question 6).(7) Whether participation in duck hunting was the sole reason for the trip, and its

relative importance if it was not (Questions 7 and 8).

Response rate

A total of 294 hunters attended the year 2000 Wetlands and Wildlife OrganisedShoot. Questionnaires were distributed to hunters during registration procedures on12 February. As some groups were registered by a single group member, some huntersmay not have received the survey. One hundred and ninety-one useable responses werereceived, giving a response rate of 65 per cent across all hunters attending the shoot.

Data preparation

The zonal TCM requires estimation of the relationship between the costs of travellingto the shoot and the proportion of the population from each zone travelling to the shoot(the TGF). Prior to undertaking this analysis, the zones must be established from thedata in the questionnaire. The goal in combining respondents into zones is to establishcomposite zones containing suf cient respondents, but which are also relatively homo-geneous in terms of distance from the shoot and socio-economic composition. Ideally,each zone should also have a minimum of 30 respondents. A pragmatic response meansthat trade-offs need to be made between these goalsespecially between homogeneityand sample size. The nal zones used for the USE duck hunting TCM are shown inTable 1. In Figure 2, the travel time is plotted against the cost of attending the hunt.As shown in Table 1, the population of the zones varies signi cantly.3

The dependent variable in the TGF is the number of duck hunters per head ofpopulation from each zone. Because the Wetlands and Wildlife Organised Shoot wasattended only by males, the population for the analysis is the population of males agedover 15 years in each zone. The population for each zone was calculated using the ABScensus data from the 1996 census (ABS 1997). The visitation rate is calculated bydividing the number of visits from each zone by the number of males over 15 living ineach zone.

Travel costs

Respondents were asked to provide an estimate of their monetary costs of attending thehunt. The respondent estimate is of the marginal monetary costs of attending the hunt.The respondent estimate does not include the opportunity costs of attending the hunt.As respondents were also asked to include an estimate of travel time, a proxy for the

Dow

nloa

ded

By:

[Aus

tralia

n N

atio

nal U

nive

rsity

Lib

rary

] At:

03:4

2 4

Febr

uary

200

8


TABLE 1. USE duck hunting TCM zones

Male Mean cost ofZone population Number of respondents Mean travel(normal place of residence) over 15 yearsa hunters ($) timeb

1. Local 2845 24 152.92 1.2812. Naracoorte and districts 19 789 39 162.31 1.949

(also includes Murray Bridge)3. Adelaide 428 248 27 180.19 3.2064. Lower South East remainder 12 577 18 171.11 3.3895. Horsham and district 9893 28 189.64 3.1076. Central and Northern Victoria 125 805 18 261.67 5.0837. Melbourne and surrounds 1 297 696 14 312.86 7.1078. North Wimmera 9290 11 180.91 3.3649. Hamilton and districts 14 019 12 162.50 3.458

Notes:a Themale population is used as onlymale shooters attended the Wetlands andWildlifeOrganised Shoot.b Travel time is in hours and fractions of hours.

opportunity cost of travel time can be included. The proxy can be used to test thesensitivity of the consumers surplus estimates to alternative values of time as perBateman (1995). The time values used were 0.5 and full value of time estimated usingthe mean male weekly wage (in February 2000 from the Australian Bureau of Statistics2001) divided by the mean number of hours worked by males in 199899. The hourlywage rate calculated using this methodology was $19.09/hour. Hence, three alternativemeasures of the travel cost to participate in the hunt have been generated.The potential for multiple-purpose trips to affect cost estimates was also raised in

the section on research design. A question was included in the survey to determine theproportion of multiple-purpose trips and the relative importance of participating inthe duck hunt to the trip. Only three hunters indicated that participating in the huntwas neither the sole purpose of the trip nor very important to the trip. That is, the

FIGURE 2. Duck hunter travel time and cost.

Dow

nloa

ded

By:

[Aus

tralia

n N

atio

nal U

nive

rsity

Lib

rary

] At:

03:4

2 4

Febr

uary

200

8


trip was undertaken only because there was an opportunity to participate in the shoot.Therefore, the data were not adjusted for multiple-purpose trips, as it was consideredunlikely that the three observations would bias the results.

The travel cost relationship

The TGF is estimated by regressing the visitation rate against the mean travel cost foreach zone.4 Three alternative functional forms were investigated:

(a) Semi-log dependent: log (visit rate) 5 a 1 b travel cost.(b) Semi-log independent: visit rate 5 a 1 b log (travel cost).(c) Double log: log (visit rate) 5 a 1 b log (travel cost).

Decisions between (a) and (c) can be made using model validity, R2 and accuracy ofpredicted hunter numbers. Scaling the dependent variable as per Rao and Miller (1971,p. 107) allows likelihood ratio tests between (b) and (a) or (c). Model validity can beused to assist in making decisions between (b) and (a) or (c). Christensen and Price(1982, p. 396) note that accurate prediction of current hunter numbers is not a suitablecriterion if it leads to incorrect model speci cation and incorrect consumers surplusestimates. That is, a model that incorrectly predicts the current number of hunters butmore accurately predicts hunters responses to a change in costs may be preferred overa model that accurately predicts current hunter numbers. Hence, model validity is theprimary selection criterion and predicted current hunter numbers the secondary crite-rion.Preliminary regressions were undertaken using OLS. Predicted hunter numbers from

all models indicated that the visit rate from Adelaide was substantially different fromthe other zones and was poorly predicted (over 100 hunters vs just 27 in reality). Thiswas possibly due to differences in population preferences between capital cities andrural areas. Because of the large differences, Adelaide is also acting as an outlier onremaining predictions. A dummy variable for capital cities (Adelaide and Melbourne)was therefore included in the model to eliminate the capital city effect and improvemodel validity.The dependent variable semi-log model was not signi cantly different from the

double log model (models (a) and (c)). Model (a) was selected due to the simplicity ofestimating results and model validity over model (b). The Rao and Miller (1971) scaledlikelihood ratio test indicated no signi cant difference between models (a) and (c) butboth model (a) and model (c) were signi cantly different from (b) at the 1 per centlevel. The preliminary model estimated was:

log (visit rate) 5 a 1 b.travel cost 1 c.Cap city dummy.

With no time costs included, the OLS estimated TGF is:

log (visit rate) 5 2 2.604 2 2.10E-2.travel cost 2 2.785.Cap city dummy.(1.243) (0.007) (0.801)

Note: numbers in parentheses are standard errors.

Model R2 adjusted 5 0.847, F probability 5 0.002.

Cooper and Loomis (1993, p. 549) indicate that potential heteroskedasticity due tozones with differing population sizes is likely to be minimised by the logged dependentvariable.5 A simple test for heteroskedasticity involving the plotting of prediction erroragainst population for the logged dependent variable indicated that there was little

Dow

nloa

ded

By:

[Aus

tralia

n N

atio

nal U

nive

rsity

Lib

rary

] At:

03:4

2 4

Febr

uary

200

8


TABLE 2. OLS models for travel time cost sensitivity

Travel time cost

Coef cients Base 0.5 cost Full cost

a 2 2.604 2 2.859 2 3.065(1.243) (1.088) (0.997)

b 2 2.10E-2 2 1.69E-2 2 1.40E-2(0.007) (0.005) (0.004)

c 2 2.785 2 2.717 2 2.685(0.801) (0.770) (0.755)

R2-adjusted 0.847 0.861 0.867F (Sig.) 0.002 0.001 0.001

Note: numbers in parentheses are standard errors.

likelihood of its presence in the model. This nding was con rmed by a BreuschPagantest.In Table 2 the three models estimated in order to test the sensitivity of the results to

the inclusion of a cost for travel time are reported.The TGF can also be estimated using non-linear least squares (NLSQ).6 This

method of estimation eliminates some concerns regarding the OLS estimation of theTGF relationship.7 The results, which are reported in Table 3, are very similar to theOLS methodology used above. Consumers surplus estimates will also be calculatedfrom these estimates.

Other modelling issues

Respondents were asked to indicate where and how many times they had been huntingin ve pre-speci ed areas and to nominate any other areas in which they had been duckhunting. The proportion of hunters from each zone who had hunted in speci edregions is shown in Table 4. A majority of hunters from all zones except two (LSEremainder and Melbourne and surrounds) hunted in the USE in the last 12 months.For many hunters this would have been one or more of the 1999 Wetlands andWildlife Organised Shoots. Table 4 shows there were very few substitute sites within

TABLE 3. Non-linear least squares models for travel time costsensitivity

Travel time cost

Coef cients Base 0.5 cost Full cost

a 2 2.455 2 2.657 2 2.925(0.995) (0.858) (0.760)

b 2 2.36E-2 2 1.95E-2 2 1.61E-2(0.005) (0.004) (0.003)

c 2 2.990 2 2.876 2 2.832(0.437) (0.412) (0.404)

R2-adjusted 2 0.452 2 0.289 2 0.228

Notes: numbers in parentheses are asymptotic standard errors,R2-adjusted is not bound in [0,1].

Dow

nloa

ded

By:

[Aus

tralia

n N

atio

nal U

nive

rsity

Lib

rary

] At:

03:4

2 4

Febr

uary

200

8


TABLE 4. Proportion of hunters hunting in substitute areas by zone

Proportion of hunters hunting in each area in last 12 months

Lower SA Upper SAUSE LSE Murraya Murraya Victoria Other

Zone (%) (%) (%) (%) (%) (%) Substitutes

1 91.7 4.2 8.3 0.0 4.2 4.2 02 74.4 23.1 2.6 2.6 10.3 20.5 03 51.9 3.7 29.6 7.4 7.4 11.1 04 31.6 47.4 10.5 15.8 52.6 10.5 25 46.4 28.6 3.6 3.6 78.6 35.7 26 61.1 5.6 0.0 0.0 100.0 55.6 27 28.6 7.1 0.0 0.0 71.4 28.6 18 81.8 27.3 0.0 0.0 90.9 63.6 29 53.8 7.7 0.0 0.0 84.6 23.1 1

Notes:a Lower SA Murray is the Murray River and oodplain below Murray Bridge including LakesAlexandrina andAlbert.TheUpper SAMurray is theMurrayRiver and oodplain aboveMurrayBridge.Substitutes are de ned as more than one-third of hunters from the zone have hunted in that areain the previous 12 months.

South Australia during the last 12 months. Only South Australian hunters residentwithin the Lower South East area had a strong substitute site outside the USE (theLower South East area). The substitute variable shown in Table 4 was tried within themodel in an attempt to include the effects of substitutes. The variable was insigni cantand showed the incorrect sign. This is potentially because the duck hunting openseasons in South Australia and Victoria overlap rather than occur at the same time.The proportion of hunters who travelled with friends was also trialled as an explana-

tory variable; however, the coef cient was insigni cant. It is likely that the impacts ofthe variable were incorporated in the self-estimated travel costs as respondents wereasked to indicate the costs of their trip (rather than group costs).

Consumers surplus estimation

Estimation of consumers surplus is undertaken via a two-stage approach in most TCMapplications. The rst stage utilises the TGF to simulate demand for hunting underdifferent pricing conditions. That is, to derive a demand curve for duck hunting. Thesecond stage is to estimate the area under the demand curve at the current level ofhunting fees. For some functional forms, the consumers surplus can be estimateddirectly from the TGF (see, for example, Hanley & Spash 1993; Offenbach & Goodwin1994). In the case of the log-linear form applied in this study the average per capitaconsumers surplus is:8

CS 5 2 1/(travel cost coef cient).

Hence, for the base (OLS) relationship:

CS 5 2 1/0.210E-25 $47.73/visit.

Dow

nloa

ded

By:

[Aus

tralia

n N

atio

nal U

nive

rsity

Lib

rary

] At:

03:4

2 4

Febr

uary

200

8


TABLE 5. Sensitivity of estimates to inclusion of cost of travel time ($)

OLS NLSQ

Estimate Base 0.5 cost Full cost Base 0.5 cost Full cost

Individual CS 47.73 59.20 71.39 42.31 51.33 62.0395% upper 123.50 138.17 159.46 76.14 87.04 101.8195% lower 29.58 37.67 45.99 29.30 36.40 44.61Total CSa 14 033 17 406 20 987 12 439 15 092 18 238

Note:a Total consumers surplus for Wetlands and Wildlife opening shoot in 2000.

A 95 per cent con dence interval is calculated as follows:

2 1/(0.210E-2 2 1.96SE) , CS/visit , 2 1/(0.210E-2 1 1.96SE),

where SE 5 standard error of the travel cost coef cient.Hence the 95 per cent con dence interval is:

$29.58 , CS/visit , $123.50.

To estimate the total consumers surplus for the Wetlands and Wildlife OrganisedShoot, the per-visit consumers surplus is multiplied by the total number of partici-pants (294). Hence, the base level total consumers surplus for the 2000 shoot is$14 033. The equivalent estimate for the NLSQ method is slightly lower at $12 439 or$42.31 per visit. Similarly, the consumers surplus can be extrapolated across all duckhunting undertaken in the USE (assuming similar demand conditions and huntingquality).To estimate the sensitivity of the model to inclusion of costs for travel time, the above

calculations are repeated for the alternative models. The results of these are reported inTable 5. There are large differences between the estimates. The full cost estimate is 50per cent larger than the base model estimate. The difference between the models showsthe importance of the treatment of the costs of travel time to consumers surplusestimates. The consumers surplus estimates from the NLSQ estimation are alsoincluded in Table 5. NLSQ estimates are around 12 per cent lower than the OLSestimates and have much tighter con dence intervals.

Calculation of a net present value of duck hunting

The nal step in the analysis of consumers surplus is estimation of a net present value(NPV) of hunting bene ts that could be expected from wetlands in the USE. Tocalculate the NPV, the consumers surplus is expressed as an annuity that would bereceived for 30 years. The present value of the annuity is multiplied by the inverse ofthe selected discount rate to estimate the NPV. Estimates of the NPV are shown inTable 6.Two issues need to be clari ed prior to estimating a NPV. Firstly, the year 2000 duck

hunting seasons quality was worse than historical averages. Hence, the appropriateNPV for duck hunting on Wetlands and Wildlife properties alone will be higher thanthat estimated using year 2000 numbers. Wetlands and Wildlife provided data on thenumber of shooters attending the opening shoot and later shoots over the last 13 years.The data revealed that seasonal conditions did not affect the number of hunters at the

Dow

nloa

ded

By:

[Aus

tralia

n N

atio

nal U

nive

rsity

Lib

rary

] At:

03:4

2 4

Febr

uary

200

8


TABLE 6. Estimates of NPV for hunting ducks in the USE from the NLSQ model

First shoot W&W mean Total USE Lower USE Upper USEModel used 2000 annual total estimate sensitivity sensitivity

Base model $154 362 $606 945 $787 559 $525 039 $1 050 07950% cost travel time $187 283 $736 392 $955 527 $637 018 $1 274 035Full cost travel time $226 317 $889 874 $1 154 680 $769 787 $1 539 574Number of hunters 294 1156 1500 1000 2000

Note: NPVs calculated over 30 years using a 7 per cent discount rate.

opening shoot (294 vs an average of 286) but did impact on the overall numbers in2000 (555 hunters vs an average of 1156 across all years). Wetlands and Wildlife datawere used to estimate an average number of hunters attending, along with an upper andlower number of hunters to test the sensitivity of this assumption.Secondly, some hunters make more than one hunting trip to the USE and these trips

may be undertaken to alternative sites within the USE. For example, 69 respondents(35.8 per cent) made more than one trip to the USE. On average, these respondentsmade 3.78 hunting trips in USE wetlands in the previous 12 months and are dominatedby respondents living closer to the wetlands. Wetlands and Wildlife data verify thisassessment indicating that over three-quarters of all hunting trips to their wetlandsoccur in later hunts during the season. The long-term average of Wetlands and Wildlifehunter numbers is the base aggregate consumers surplus. In addition, 39 per cent ofUSE landholders report undertaking hunting in their wetlands (about 30 landholders)(Whitten & Bennett 1998). Hence, it is likely that well over 350 additional hunting tripsare undertaken in the USE when it is considered that friends and family often join insuch events. The third column of Table 6 is an estimate of total consumers surplus forthe Upper South East when these additional hunting trips are included. The huntersparticipating in these additional trips may enjoy a higher average consumers surplusbecause friends and neighbours often live close by (and are subject to low travel costs)and no fee is charged for participation. However, the possibility also exists that theaverage consumers surplus for these additional trips could be lower than for theWetlands and Wildlife organised hunt. For example, a small bene t from participationless even smaller transport costs leaves a small but positive consumers surplus. Hence,the consumers surplus estimate derived from the Wetlands and Wildlife organisedshoot is conservatively attributed to the estimated number of additional trips in theUSE. Sensitivity tests of the NPV of future consumers surpluses from duck hunting inthe USE are shown in the nal two columns of Table 6.The extrapolation of the survey data assumes similar travel costs and quality of the

hunting experience. This assumption is not as unlikely as it may seem, as participantsin the Wetlands and Wildlife Organised Shoot do not know which of several wetlandsthey will be allotted for hunting. Each of these wetlands may also be some distancefrom the centralised meeting point.

Estimation of producers surpluses

All hunters participating in Wetlands and Wildlife organised shoots must pay a huntingfee of $20. Most of the inputs to the hunt (such as labour, access construction andmanagement and other inputs) are donated to Wetlands and Wildlife. Hence, the

Dow

nloa

ded

By:

[Aus

tralia

n N

atio

nal U

nive

rsity

Lib

rary

] At:

03:4

2 4

Febr

uary

200

8


hunting fee is a proxy for the producers surplus, or pro t, that the hunt generates tothe wetland owner. The NPV of the future stream of producers surpluses can beestimated in a similar way to the NPV of future consumers surpluses. The NPV offuture producers surpluses, assuming duck hunting continues on Wetlands andWildlife properties at current average levels, is $286 897. As for the consumers surplusestimates, sensitivity tests can be conducted. Sensitivity tests at the 95 per centcon dence interval of the mean number of hunters participating in Wetlands andWildlife organised hunts over the last 13 years indicate an upper value of $385 921 anda lower value of $187 625 for the producers surplus.

Conclusions

The aim in this paper has been to present estimates of the values generated by duckhunting in the USE of South Australia. These values were estimated using a travel costsurvey of participants in the year 2000 Wetlands and Wildlife Organised Shoot andthe revenue generated for Wetlands and Wildlife by the shoot. The use of the TCMfacilitates estimation of the consumers surplus associated with participating in theshoot. This consumers surplus can be extrapolated (under certain assumptions) to allduck hunting in the USE.Duck hunters participating in the shoot derived an average consumers surplus of

between $42.31 and $62.03 (based on NLSQ estimates). The variation is due to arange of alternative values that could be placed on time spent travelling to the shoot.These values generate a consumers surplus of between $12 439 and $18 238 for theyear 2000 season opening event. Extrapolating these estimates across all years generatesa NPV estimate of between $606 945 and $889 874 for hunting on Wetlands andWildlife wetlands. Wetlands and Wildlife generates a further producers surplus NPVof $286 897 from fees to participate in the organised shoot. Further extrapolationacross duck hunting in all USE wetlands produces a NPV estimate of between $0.75million and over a $1 million. Hence, the total surpluses generated to wetland ownersand duck hunters from hunting in USE wetlands are likely to exceed $1 million.The estimates that are reported in this paper comprise the non-monetary values

generated to the duck hunting community and the monetary values enjoyed by wetlandowners who charge for the use of the wetlands for hunting. These values canbe incorporated with other estimates of monetary and non-monetary bene ts andcosts in a costbene t framework to assess the likely net bene ts (or costs) of undertak-ing management changes in USE wetlands. These values can also be traded off againstthe non-market costs that duck hunting imposes on some members of the widercommunity.However, a bene tcost analysis of duck hunting in the USE is far from straightfor-

ward because it involves a three-way trade-off. Firstly, wetland owners allow hunters togenerate consumers surplus by participating in hunting in wetland areas. Indeed, manywetland owners hunt ducks in their own wetland areas. The same hunting actionsgenerate a cost to the wider community via their distress over the killing of waterfowl.Secondly, wetland owners trade-off the costs of maintaining wetlands against thebene ts they are able to generate from wetlands (in part via capture of the consumersurplus generated by hunterswho may include the wetland owner). Finally, mainte-nance of wetland habitat generates other social bene ts to the community such asmaintenance of habitat and protection of endangered species that must also be traded-off against the costs of allowing hunting.9 The degree to which hunting facilitates

Dow

nloa

ded

By:

[Aus

tralia

n N

atio

nal U

nive

rsity

Lib

rary

] At:

03:4

2 4

Febr

uary

200

8


generation of bene ts from wetlands against the resulting potential biophysical costs (orbene ts) to species in wetlands is a topic for further research.

Acknowledgements

The authors would like to acknowledge and thank the assistance of Mr Paul Wain-wright and of cers of Wetlands and Wildlife. They would also like to thank Dr GeoffKerr (Lincoln University, New Zealand) for statistical comments and advice.

Correspondence: Stuart M. Whitten, CSIRO Sustainable Ecosystems, GPO Box 284,Canberra, ACT 2601, Australia. E-mail: [email protected]

NOTES

[1] Sunk costs are de ned as costs already incurred that cannot be recouped, and therefore do notimpinge on present decisions.

[2] Mr Paul Wainwright. Copies of the questionnaire are available from the authors on request.[3] Heteroskedasticity is potentially present because the unequal zonal populations generate differing

variances for each observation (see Bowes & Loomis 1980).[4] There are a number of alternative functional forms available (semi-log, double log and quadratic)

and two alternative methods of estimation (maximum likelihood and ordinary least squares).Furthermore, the non-constant variance of the dependent variable suggested that a weightedestimation procedure could be appropriate to estimate the relationship (weighted maximumlikelihood or weighted least squares).

[5] Furthermore, using weighted least squares or generalised least squares (as per Bowes & Loomis1980) would expose the methodology to the criticisms made by Christensen and Price (1982).That is, the greatest weighting is given to zones further from the hunting site by virtue of theirlarger populations. Hence, the weighting procedure suggested by Christensen and Price (1982)was not suitable for the data even if heteroskedasticity were present.

[6] NLSQ is a maximum-likelihood-based methodology.[7] The dependent variable of the TGF is both truncated and censored (Hanley & Spash 1993). It

is argued that ordinary least squares (OLS) estimates will therefore be inappropriate, but theliterature is divided on this point. For example, Smith and Desvouges (1986) and Willis andGarrod (1991) argue that maximum likelihood is more accurate, while Kling (1987, 1988) andSmith (1988) argue for OLS. See Whitten and Bennett (2001) for more information oneconometric issues.

[8] The mathematical derivation of this relationship is provided in the Appendix.[9] Other social bene ts may also be derived from any bene cial direct or multiplier impacts on local

businesses due to purchases of provisions or accommodation by duck hunters from outside theUSE region. These bene ts are not estimated by the TCM method.

[10] The proof is drawn from notes provided by Dr Geoff Kerr, Lincoln University, New Zealand.

REFERENCES

ABS (AUSTRALIAN BUREAU OF STATISTICS) (1997) Basic community pro les1996 census, AustralianBureau of Statistics (www.abs.gov.au).

ABS (AUSTRALIANBUREAU OF STATISTICS) (2001) Australia nowa statistical pro le, Australian Bureauof Statistics (www.abs.gov.au).

BATEMAN, I. (1995) Valuation of the environment, methods and techniques: revealed preferencemethods, in Turner, R.K. (ed.) Sustainable environmental economics and management: principles andpractice, London, Belhaven Press, pp. 192241.

BENNETT, J.W. (1995) The travel cost method of estimating the value of recreational use of natural areas, areport prepared for NSW National Parks and Wildlife Service, Environmental and ResourceEconomics, Gundaroo.

Dow

nloa

ded

By:

[Aus

tralia

n N

atio

nal U

nive

rsity

Lib

rary

] At:

03:4

2 4

Febr

uary

200

8


BOWES, M.D. & LOOMIS, J.B. (1980) A note on the use of travel cost models with unequal zonalpopulations, Land Economics 56, pp. 46570.

CHRISTENSEN, J.B. & PRICE, C. (1982) A note on the use of travel cost models with unequal zonalpopulations: comment, Land Economics 58(3), pp. 3959.

COOPER, J.C. (2000) Nonparametric and semi-nonparametric recreational demand analysis, AmericanJournal of Agricultural Economics 82, pp. 45162.

COOPER, J. & LOOMIS, J. (1993) Testing whether waterfowl hunting bene ts increase with greaterwater deliveries to wetlands, Environmental and Resource Economics 3, pp. 54561.

HANLEY, N. & SPASH, C.L. (1993) Costbene t analysis and the environment, Aldershot, Edward Elgar.KLING, C.L. (1987) A simulation approach to comparing multiple site recreation demand modelsusing Chesapeake Bay survey data, Marine Resource Economics 4, pp. 95109.

KLING, C.L. (1988) Comparing welfare estimates of environmental quality changes from recreationdemand models, Journal of Environmental Economics and Management 15, pp. 33141.

OFFENBACH, L.A. & GOODWIN, B.K. (1994) A travel-cost analysis of the demand for hunting trips inKansas, Review of Agricultural Economics 16, pp. 5561.

RAO, P. & MILLER, R.L. (1971) Applied econometrics, Belmont, CA, Wadsworth.SMITH, V.K. (1988) Selection and recreation demand, American Journal of Agricultural Economics 70,pp. 2936.

SMITH, V.K. & DESVOUGES, W.H. (1986) Measuring water quality bene ts, Boston, Kluwer-Nijhoff.TURNER, R.K., PEARCE, D. & BATEMAN, I. (1994) Environmental economics: an elementary introduction,Hemel Hempstead, Harvester Wheatsheaf.

WHITTEN, S.M. & BENNETT, J.W. (1998) Farmer perceptions of wetlands and wetland management in theUpper South East of South Australia, The Private and Social Values of Wetlands Research Report 2,University of New South Wales, Canberra.

WHITTEN, S.M. & BENNETT, J.W. (2001) A bio-economic analysis of potential Upper South East regionalwetland management strategies, The Private and Social Values of Wetlands Research Report 9,University of New South Wales, Canberra.

WILLIS, K.G. & GARROD, G. (1991) An individual travel-cost method of evaluating forest recreation,Journal of Agricultural Economics 42, pp. 3342.

Appendix: mathematical derivation of consumers surplus from the trip gener-ator function

The log-linear function estimate is:

Ln (VRi) 5 a 1 b.TCi 1 c.Cap city dummy

VRi 5 e(a 1 b.TCi 1 cCap city dummy),

where VRi 5 number of hunters from zone i/male population of zone i, TCi 5 travel costfor zone i, Cap city dummy 5 dummy variable for zone including a state capital.The average consumers surplus is:

CS/visit 5 2 1/b.

Proof10

CSi/person 5 E`

TCi

VRidTC 5 E`

TCi

aebTCidTC 5a

2 bebTCi

CS/visit 5CSi/personvisitsi/person

5

a

2 bebTCi

aebTCi5

1

2 b.

Hence, consumers surplus is not dependent on the zone from which the visit is made.

Documents

group 4.April 23_Travel cost_Duck hunting 2002.pdf