Tourism Management 33 (2012) 971e977
Contents lists available at SciVerse ScienceDirect
Tourism Management journal homepage: www.elsevier.com/locate/tourman
Recreational benefits from a marine protected area: A travel cost analysis of Lundy Dong-Ryul Chae a, Premachandra Wattage b, *, Sean Pascoe c a
Gyeongnam Development Institute, 248 Youngji-ro, Changwon-si, Gyeonsangnam-do 641-728, South Korea Centre for the Economics and Management of Aquatic Resources (CEMARE), University of Portsmouth, St. George’s Building, 141 High Street, Portsmouth PO1 2HY, UK c CSIRO Marine and Atmospheric Research, Australia b
a r t i c l e i n f o
a b s t r a c t
Article history: Received 28 June 2010 Accepted 10 October 2011
Marine Protected Areas (MPAs) have been proposed in many countries as a means of conserving parts of the marine environment. In some cases, MPAs may also confer recreational benefits. In this paper, a travel cost model is used to estimate the non-market recreational benefits arising from the Lundy Island Marine Nature Reserve (MNR). The estimated mean consumer surplus for visiting Lundy was found to range from £359 to £574 per trip. The designation of No Take Zone (NTZ) has also contributed to higher consumer surplus values. This result provides a strong economic justification for the designation of MPAs for recreational as well as conservation purposes. Crown Copyright Ó 2011 Published by Elsevier Ltd. All rights reserved.
Keywords: Recreational benefits Marine protected areas Travel cost method Lundy Island
1. Introduction Marine Protected Areas (MPAs) are specially designated zones of the sea that are designed to restore marine ecosystem to the original state by excluding all detrimental human activities. While often imposed for purely conservation purposes, MPAs can also result in economic benefits to users of the marine environment. Economic benefits of MPAs can be largely divided into two parts e fisheries and non-fisheries benefits. In the case of fisheries, many studies have demonstrated both theoretical and empirical benefits to fisheries from MPAs (Ami, Cartigny, & Rapaport, 2005; Dugan & Davis, 1993; Lewison, Crowder, Read, & Freeman, 2004; Sumaila, 1998; Zeller & Russ, 1998). Studies concerning the non-fisheries benefits of MPAs have mainly dealt with nature based recreation and tourism within large-sized tropical coral reef areas such as the Great Barrier Reef Marine Park in Australia (Driml, 1997; Hill, Rosier, & Dyer, 1995; Rouphael & Inglis, 1997; Valentine, Birtles, Curnock, Arnold, & Dunstan, 2004) and the Florida Keys National Marine Sanctuary in the USA (Bhat, 2003; Leeworthy, Wiley, English, & Kriesel, 2001; Park, Bowker, & Leeworthy, 2002). This study aims to examine the potential benefits of marine nature based tourism in the UK arising from the designation of a MPA. This information will be useful for policy makers when planning the designation of future MPAs in line with international
* Corresponding author. Tel.: þ44 23 9284 8508; fax: þ44 23 9284 8502. E-mail address:
[email protected] (P. Wattage).
commitments to protect the marine environment (e.g. under the UNEP Convention on Biological Diversity of 1993 and the World Summit for Sustainable Development of 2002). The study focuses on recreational demand in the Lundy Island marine nature reserve (MNR). This contains the first no take zone in the UK water. A travel cost model was estimated using count data regression techniques in order to measure the non-market recreational benefits of the Lundy MNR. 2. Lundy Island MNR 2.1. Background Lundy is an island three miles long by half a mile wide, standing four hundred feet out of the sea, situated in the Bristol Channel (UK e Lat: 51 100 N; Long: 4 400 W) (See Fig. 1). Currently, the Island is owned by the National Trust and is financed, administered and maintained by The Landmark Trust. There are about 20 permanent residents working on the island, and 23 buildings that can be rented by holiday visitors. Lundy has been inhabited since prehistoric times. Nomadic hunters and fishermen were the earliest people to live on Lundy. Archaeologists presume Lundy was a summer fishing base during the later Mesolithic Period (The National Trust, 2002). Since medieval times, it has been managed by private ownership and the tenants made their living by agricultural, stock farming, fisheries activities amidst other forms of livelihood (Langham, 1994). The last joint owners sold Lundy to the National Trust on 1 October 1969 for £150,000 and the National Trust leased Lundy to the Landmark
0261-5177/$ e see front matter Crown Copyright Ó 2011 Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.tourman.2011.10.008
972
D.-R. Chae et al. / Tourism Management 33 (2012) 971e977
Fig. 1. Lundy Island location.
Trust for a period of sixty years. Thereafter, the Landmark Trust has expanded and improved facilities around the island including electricity and water supply, visitor accommodation and restoration of old buildings. The main transportation to Lundy is a shuttle ferry e the MS Oldenburg e with a capacity of 267 passengers and 20 tons of freight. Annual passenger numbers on the MS Oldenburg were approximately 17,000 in 2005. Other visitors to the island can use charter boats based at the local harbour of Clovelly and Appledore. However, with the operation of full winter helicopter service since 2003, visitor numbers have increased recently (Friends of Lundy, 2004). By the end of the 20th century, the island was designated into several types of special or protected areas, reflecting different environmental and other assets of this area. It was originally established as a Voluntary Marine Nature Reserve (VMNR) in 1973 by the local naturalist society (Jones, 1999). In 1986, Lundy Island and the sea area around Lundy were designated as the UK’s first statutory Marine Nature Reserve (MNR). Recently, it was designated as a Marine special area of conservation (MSAC) under the European Union’s Habitats Directive legislation as part of the Natura 2000 Network. On January 2003, the east side of Lundy MNR was designated as the first no take zone (NTZ) in the UK jointly proposed by the Devon Sea Fisheries Committee and English Nature to alleviate pressure on fish and shellfish stocks and to restore wildlife.
2.2. Recreational activities on Lundy Island Visitors are attracted by the outstanding natural conditions of Lundy, well protected ecological environment, and historical remains scattered on the island. Lundy provides opportunities for various recreational activities from simple relaxation to expert leisure activities that demand special equipment and skills. The underwater natural environment around Lundy affords optimum conditions for both recreational and research purpose divers. First of all, the seawater around Lundy Island is pristine, because of the distance from sources of pollution on the mainland. Secondly, Lundy is well-known for its plentiful and varied marine flora and fauna due to the geological and topographical features. The ocean area around Lundy is a transitional zone where the Gulf Stream mixes with cooler north Atlantic water, and so is a suitable habitat for both cold water and warm water organisms. The seabed around Lundy contains a multiplicity of habitats including rocky
reef, kelp forest, sandy and muddy bed. In addition, the shipwreck sites around Lundy are also an important attraction for divers. Wildlife watching is also a popular tourist activity. Visitors can see jellyfish, grey seal and turtles with the naked eye from several places and in the summer months, dolphins, porpoises and basking sharks as well. The island is also famous for a variety of its birdlife, especially the Puffin; the name Lund-ey is Norse for Puffin Island (Langham, 1994). The results of research by the Lundy Field Society demonstrate that the natural condition of Lundy is good for birds; 1) the cliffs around Lundy provide summer breeding sites for several seabirds; guillemots, storm petrels, puffins and Manx shearwaters and 2) the plateau provides breeding sites for curlew, lapwing, meadow pipit, skylark and wheatear while oystercatchers breed along the coastal fringes. Lundy is well-known as a special place for cliff climbers and walkers. According to the Lundy information map, there are 29 climbing routes varying from gentle slabs to steep cracks and the west coast cliffs named Devil’s Slope are particularly good for climbing. There are two suggested walking routes on Lundy, although many tourists prefer to wander around the island. The island also contains many historical sites of interest to tourists. The archaeological remains distributed around the island provide another large component of the recreational activities on Lundy. Various prehistoric remains including hut circles, walls and associated field boundaries are scattered around the island. A 13th century castle, known as Marisco Castle, standing at the southern end of the island is a representative ruin of medieval times. A total of three lighthouses constructed on Lundy, the Old Light and two new lights (North and South Light) are significant within the context of ocean history. The Old Light, the highest lighthouse in Britain, was built in 1820, but replaced in 1897 by two new lights close to sea level at the north and south ends. The two lights were converted to solar power in 1991 and are still in operation. 3. Study design This study employs the travel cost method to model the recreational demand of Lundy Island and to calculate the non-market benefits gained by Lundy visitors. The travel cost method is a revealed preference approach that is based on the premise that visitors must have been willing to pay at least what they did pay in order to visit the island, so the benefits of the visit must have at least exceeded this amount. The demand for the recreation can be estimated by comparing the number of visits by individuals from different locations with varying travel costs. 3.1. Data collection The data used in the study were obtained from an on-site survey of recreational visitors on Lundy Island. Face-to-face interviews were conducted during the period of JulyeAugust 2005, mostly onboard MS Oldenburg and sometimes on the island itself. A total of 161 survey questionnaires were collected, but not all of these were usable for the travel cost analysis due to several reasons. The most common reason for the non-usability of responses was nonresponse for some key questions involving household’s income level. As participation in the survey was voluntary, interviewers could not ensure that all questions would be answered household income levels. From the 161 completed surveys, 86 responses were included in the final econometric analysis. Information collected from the remainder of the interviews was still usable for other parts of the study (not presented in this paper). The questionnaire consisted of four parts, with a total of 44 questions. These were a mixture of yes/no, multiple choice, and open-ended questions. The first part of the questionnaire was
D.-R. Chae et al. / Tourism Management 33 (2012) 971e977
designed to enquire about the tendency for visiting coastal areas e.g. where they go, what they do, how often and how long they stay for and their considerations regarding deciding where to visit. The second part asked about the respondents’ knowledge and opinions about the Lundy protected area. The next part included questions relating to the number of trips to Lundy taken by the respondents, the number in their group, the type of transportation used, whether the trip was multi-purpose or not, etc. The last part included demographic questions such as the departure point of this trip, household income, working hours in a week, year of birth, education level and so on. 3.2. Regression models for count data The travel demand model was estimated using count data regression techniques. In contrast to ordinary least-squares (OLS) regression, count data models emphasise the non-negative, integer nature of data on the number of trips taken and are most useful when the counts are small (Englin, Holmes, & Sills, 2003). The Poisson regression model, first used by Shaw (1988), has been used in a number of recreational demand analyses. For example, Creel and Loomis (1990) compared the estimates of different count data models for deer hunting data in California; Groger and Carson (1991) employed standard and truncated Poisson and negative binomial models to estimate fishing demand in Alaska; Englin and Shonkwiler (1995) developed a long-run recreational demand model for hiking trails in the Pacific North West and Bhat (2003) used a combined model of travel cost and contingent behaviour to estimate the recreational value of reef quality improvements in the Florida Keys marine reserve. A Poisson distributed random variable N with parameter l is defined for all non-negative integer numbers 0, 1, 2, ., n such that:
PrðN ¼ nÞ ¼
ln n!
el
(1)
The parameter l > 0 is both the mean and variance of the random variable N, such that
EðNÞ ¼ VarðNÞ ¼ l
(2)
The log-likelihood function for the standard Poisson regression model is given by
lnL ¼ s0 l þ Y 0 N b s0 ln½Yi
(3)
s0
where is N 1 sum vector and the logarithmic and factorial functions are element-by-element (Creel & Loomis, 1990). A Poisson regression model treats visits as a discrete random variable and assumes the probability of observing that a specific number of visits can be described with a Poisson distribution (Smith, 1988). The demand for Lundy recreation can be formulated as:
Yi ¼ f ðPi ; Xi ; bÞ þ mi
(4)
where Yi is the trip demanded by the ith individual to visit the Lundy Island. Pi is the travel cost associated with visiting the ith recreation site, Xi is the vector of explanatory variables, b is a vector of unknown parameter and mi is the error term. The model represents an individual frequency of travel, rather than a zonal frequency travel cost. 3.3. Model variables 3.3.1. Key variables The dependent variable of the model is the number of visits to Lundy Island in the last three years along with seven independent
973
variables, which are Travel Cost, Income, Friend of Lundy, Biodiversity, Age, Educational level, and NTZ. These variables were initially selected from the survey questionnaire for the estimation of the recreation demand model. As far as the influencing factors are concerned, recent econometric studies of tourism demand have shown that the tourist’s income, tourism price (travel cost) and exchange rates are the most important determinants of tourism demand (Song & Li, 2008). The key variable is the travel cost, which is the estimate of the household’s return trip cost incurred in visiting the Lundy Island. This can be divided into two parts e the cost from the respondent’s home to the port of departure to the island (i.e. Bideford or Ilfracombe), and the cost from the mainland (i.e. the port) to Lundy. For the purposes of the study, three separate travel costs were calculated (Table 1). Travel costs were not collected from the survey as it is unlikely that respondents could accurately estimate the cost of getting to Lundy. Travel costs were, instead, derived from other information collected in the survey. The point of origin was used to estimate the distance travelled, and this was in turn used to estimate the basic motoring costs. Travel distance and time were calculated by the electric map service, RAC Route Planner (www.rac.co.uk/web/ routeplanner/). The motoring costs per mile (pence per mile) were derived from data provided by the AA Motoring Trust in 2005. Two different rates were used, reflecting different assumptions about the motoring costs. The first rate included only the estimated basic variable costs (fuel and tyre wear), while the second rate included service costs, replacement parts and tolls (see footnote to Table 1). All visitors were assumed to use a medium class petrol car with a purchase price when new of between £13,000 and £20,000. A third measure of travel cost included a measure of the cost of travel time. The opportunity cost of travel time, which represents the value of the time taken travelling, is a sensitive issue in recreational demand studies. While most travel cost studies have considered the opportunity cost of travel, these have been applied in different ways (e.g. Cesario, 1976; Hagerty & Moeltner, 2005; Larson, 1993; McConnell, 1992; McKean, Johnson, & Walsh, 1995; etc.). There is no consistently agreed rule as to how these costs should be applied. In this study, TC1 and TC2 exclude the opportunity cost of travel time. It is assumed for TC3 that this cost is equivalent to 30% of the visitor’s wage per working hour. The value of time spent travelling can be problematic; as such there is no strong consensus on the appropriate measure. Cesario and Knetsch (1976) suggested that the opportunity cost of travel time would be some proportion of the wage rate and used 60% of the wage rate in a recreation analysis. Analysing TC3 in this study, we felt that 60% is somewhat higher and arbitrarily used 30% of wage rate. The respondent’s level of annual household income before tax was also included in the model. Actual income was not asked in order to increase the response rate. Instead, respondent were asked to choose one of 12 income levels ranging from ‘£0e£5000’ to ‘more
Table 1 Composition of the three alternative travel cost estimates. Travel cost
Components included in travel cost
TC1 TC2 TC3
Fare of ferry or helicopter þ basic motoring costa Fare of ferry or helicopter þ total motoring costb TC2 þ opportunity cost of travel time
a
Petrol þ tyres ¼ 13.34 pence per mile. Petrol þ tyres þ service labour costs þ replacement parts þ parking and tolls ¼ 20.54 pence per mile. b
974
D.-R. Chae et al. / Tourism Management 33 (2012) 971e977
than £100,000’. Hence, this is an ordinal categorical variable rather than a continuous variable. The concerns and preferences for biodiversity on Lundy were asked from all respondents and incorporated into the model. Three questions relating to biodiversity were asked in the survey (i.e. Biological diversity of marine ecosystem, Biological abundance of marine ecosystem and Biodiversity/abundance of birds, mammals and insects), with the respondent giving a value of between 5 (very strongly) and 1 (never) levels for each question. The biodiversity variable, therefore, ranged between 15 and 3. An implicit assumption with this variable is that each component of the biodiversity measure is equivalent, hence can be added to produce a single variable for the data analysis. Other socio-economic variables related to the education level, age and attitudes of the visitor. The Education level was a categorical variable with five levels relating to the respondent’s final education. The Age of the visitor was calculated as 2005 (the year of the survey) minus the year of birth. Friend of Lundy was a binary variable with a value of 1 if the respondent was a member of the ‘Friend of Lundy’ which is a society to fund the protection and conservation of Lundy, with a value of 0 (zero) otherwise. NTZ was the other binary variable with a value of 1 if the respondent agreed with the designation of the no take zone and a value of 0 (zero) otherwise. 3.3.2. Variables used in the model The selection of suitable variables for the model was made using standard procedures of regression and correlation analysis. In the case of the Lundy recreation model, it was necessary to apply nonparametric estimation techniques to detect correlations among variables because the model contained several categorical variables. Spearman’s rank correlation analysis and chi-squared test were used to estimate correlation coefficients. According to the correlation analysis the Number of Visits (Y) with Travel Cost, Income, Biodiversity and Education level have negative signs while Friend of Lundy, Age, and NTZ have positive signs. The negative relationship with the dependent variable (Y) and respondent’s income yields a strange result in travel cost analysis. The presence of multicollinearity has been checked from the results of correlation analysis among independent variables. From the results of the correlation coefficient it was confirmed that there are some considerable perfect linear relationships between some independent variables. Firstly, respondent’s Education level seems to have a perfect linear relationship with all three levels of Travel Cost. Secondly, Friend of Lundy variable also seems to have a perfect linear relationship with Education level and NTZ. For this reason, two independent variables, Education level and Friend of Lundy were excluded in the final analysis. Thus, only five-independent variables are included in the model and the functional form expressed as:
Y ¼ expðb0 þ b1 X1 þ b2 X2 þ b3 X3 þ b4 X4 þ b5 X5 Þ þ mi
CIA World Fact book, 2007 est.
Variables
Mean
Std. dev
Maximum
Minimum
Y X1
1.430 130 157 225 5.76 11.07 42.48 0.17
1.476 53 63 96 2.66 2.73 11.50 0.38
10 269 331 508 12 15 65 1
1 42 42 50 1 3 19 0
X2 X3 X4 X5
Number of visits TC1 TC2 TC3 Income Biodiversity Age NTZ
4. Results 4.1. Summary of survey results The gender-ratio of respondents in the sample was roughly equal. About 58% of the respondents were in their forties and fifties, with 31% under forty and 12% over sixty. The respondents’ household income level ranged relatively evenly from grade 1 (less than £5000) to grade 12 (more than £100,000), with an average value of 5.3. This means the average household income level of respondents was in the range £30,000 to £39,999. Around 70% were educated to higher than degree level. Groups represented by the respondent, on average, consisted of 2.1 adults and 0.9 children (i.e. an average of 3 in each group). Lundy visitors, as a group, specially preferred the marine environment for nature based recreation and had a high interest in nature conservation. Around 30% of respondents visited coastal areas for recreation between 11 and 20 days a year and 21% visited more than 30 days. The respondents strongly considered water quality and climate/weather when deciding where to go for marine recreation. Around two thirds of the respondents (66%) knew that Lundy Island has been specially managed as a Marine Nature Reserve by the UK government, although relatively few (around 4%) were a member of the ‘Friend of Lundy’ group. Most (91%) respondents agreed that the coastal area around Lundy was especially abundant, clean and protected compared to other areas, and most (97%) agreed that it should continue to be protected. For many of the respondents (69%), the trip was their first visit to Lundy. The main reasons for their trip was relaxation (61%) and to observed wildlife (23%). Most (80%) were day visitors. Those who stayed on the island (20%) tended to stay for more than one night. For roughly two thirds of the respondents (64%), Lundy was the only destination of this trip. All but two respondents were UK residents (the other two coming from Germany and Australia).
4.2. Poisson regression results
(5)
The variable values of interest to the econometric recreational demand analysis are presented in Table 2. The estimated average travel cost varied from £130/trip to £225/trip depending on what factors were included in the cost estimates. In Table 2, the mean value of the income variable is 5.76 which informs the respondents’ average household income ranges between £30,000e£39,999. According to the National Statistics Online, this ranks four-fifth level of income per household 2005/06 in the UK. That means the average income of the Lundy visitors are above the national average. In addition, the mean value of age in the sample is 42.48 which also slightly higher than 39.6 of median age1 in the UK.
1
Table 2 Summary of the variables used in the regression model.
The regression model for the Poisson distributed data was estimated using Maximum Likelihood Estimation techniques. The computer software Limdep (version 8.0) was used to estimate the models. Three Poisson regression models were estimated e one each for each of the travel cost estimates. The estimation results of the recreational demand models are given in Table 3. All three models performed relatively well based on their c2 statistic. The signs on the coefficients were largely invariant to the choice of travel cost used and in most cases the estimated coefficients are statistically significant at the 95 percent confidence interval except Age variable. Each estimated coefficient represents the relationship between the variable and number of visits to Lundy in a given period. In all cases including the income variable, the sign of the estimated coefficient was consistent with the results of correlation analysis, or a-priori expectations.
D.-R. Chae et al. / Tourism Management 33 (2012) 971e977 Table 3 Poisson regression results with different travel cost levels applied. Model 1
Model 2
Constant 1.72027*** 1.79475*** Travel cost 1 0.00436** Travel cost 2 0.00416*** Travel cost 3 Income 0.14119*** 0.13355*** Biodiversity 0.06648** 0.06676** Age 0.012179 0.011447 NTZ 0.584515** 0.567652** Criteria for assessing goodness of fit 2 c 33.605*** 35.4207*** Log-likelihood 110.5162 109.6084 function Overdispersion test (regression based test) g ¼ mi 3.669 4.317 0.919 1.275 g ¼ m2i
Model 3 1.37921***
975
considered constant, and therefore are effectively captured in the constant term b0. If we define X1 as the current travel cost, consumer surplus for access is the area under the demand curve for all prices greater than X1 This can be given by
ZN 0.00286** 0.10035** 0.05944* 0.01477* 0.507861**
X1 ¼ X1
"
¼
33.4862*** 110.5756
***, **, and * indicates the coefficient is statistically significant at the 1%, 5% and 10% level respectively. N ¼ 86.
For example, the negative coefficient of the travel cost variable indicates that people with lower travel costs visit more frequently, consistent with economic theory. The NTZ dummy variable came out with a positive coefficient as expected. In other words, respondents who agree with the designation of the no take zone visit Lundy more frequently than the others. Similarly, the positive relationship with age suggests that older people are more likely to visit the site than younger people, although this result was also not statistically significant, indicating that no strong relationship exists between age and the number of visits. The negative sign of the coefficient on the income variable was not consistent with various past travel cost studies. This suggests that high income groups are less likely to visit the Lundy MNR. Similarly, the negative coefficient on the Biodiversity variable suggests that there is a negative relationship between the number of visits and the degree to which respondents’ value biodiversity on Lundy. This is an interesting result, as it suggests that the main attraction of Lundy is not related to its biodiversity. While some people visit Lundy for its biodiversity, many more visit Lundy more often for other reasons. To detect under or overdispersion in the Poisson regression model, the overdispersion test was conducted as proposed by Cameron and Trivedi (1990). Overdispersion occurs when the conditional variance of the dependent variable is bigger than its conditional mean, and underdispersion occurs when the conditional variance of the dependent variable is smaller than its conditional mean. From Table 3, the estimated values for m (mu) in all three models is much lower than 0, indicating a large degree of underdispersion in the Poisson regression analysis.
4.3. Estimating WTP and consumer surplus Travel costs indicate what a visitor has actually paid to visit a site, not the total that they would be willing to pay. Willingness to pay is a better indicator of the total value that individuals place on the environmental goods. The difference between willingness to pay and the actual cost is known as consumer surplus. The consumer surplus of household i for access to the Lundy MNR can be estimated as the area under the utility constant demand curve for the site. The demand curve for travel to the site can be expressed as Y ¼ eb0 þb1 X1 , where Y is the number of trips and X1 is the cost of travelling to the site each trip. For the purposes of estimating consumer surplus to a particular site, all other factors can be
eb0 þb1 X1
#N
b1
eb0 þb1 X1
N
¼
b1
X1
eb0 þb1 X1
b1
b0 þb1 X N 1
¼ 3.841 1.058
eb0 þb1 X1 þb2 X2 þ/þb5 X5 þ 3
Consumer Surplus ¼
e b0 þb1 X1
b1
¼
Y
(6)
b1
where b1 < 0 is the estimated regression coefficient relating to the travel cost and Y is the average number of trips observed. Given this, the consumer surplus during the given period is estimated to range from £328e£500 and consumer surplus per trip range from £229e£350 (Table 4). The consumer surplus represents the non-market benefits accrued over and above those indicated by the travel costs. As a result, the total benefit of the trip can be estimated as the sum of the two measures. However, as the trip did incur costs, the net economic benefit is the consumer surplus only. 4.4. Underdispersion problem and the Gamma model The result of the standard Poisson analysis shows there is an underdispersion problem in the data. As a result, the standard errors may be biased. The negative binomial analysis, which is used most frequently as an alternative to standard Poisson, can manage the overdispersion problem but cannot be applied in the case of underdispersion. According to Winkelmann (1995), the Gamma model can be used for both over and underdispersion. The distribution of this model is written as:
f ðyÞ ¼ Gðay; bÞ Gðaðy þ 1Þ; bÞ and
1 Gðay; bÞ ¼ GðyaÞ
Zb
mya1 em dm ; n ¼ 1; 2; .
0
where the integral is the incomplete gamma function. The expected value of this distribution is given by
EðyÞ ¼
N X
Gðai; bÞ
i¼1
Assuming that
b ¼ exp x0i g a
Table 4 Estimated consumer surplus and total benefits. Cost assumptions
TC1 TC2 TC3
Mean TC
CS per trip
WTP per trip
CS for three years
WTP for three years
(A)
(B)
(A) þ (B)
(C)
(A) 1.43 þ (C)
£130 £157 £225
£229.4 £240.4 £349.7
£359.4 £397.4 £574.4
£328.0 £343.8 £500.0
£513.9 £568.2 £821.8
976
D.-R. Chae et al. / Tourism Management 33 (2012) 971e977
This parameterisation yields the regression Eðsi jxi Þ ¼ expðx0i gÞ where s is the waiting time between events. To express the estimated coefficients in terms of their effects on the event counts y, and make them comparable to those obtained from the negative binomial or Poisson regression models, a simple strategy is to hold all explanatory variables constant at their means, and to compute ^ Dx, where x is the remaining explanatory variable, and the DY= change is defined by a unit increase at the mean value (in the case of continuous variables), or by a change from 0 to 1 in the case of dummy variables (Winkelmann, 2003). In the analysis of Lundy data, the Gamma model in Limdep version 8.0 is utilised to manage the underdispersion problem and the results of this model are given in Table 5.
Table 6 Impact of the NTZ designation on number of visits and total consumer surplus. Model 2 (TC2)
NTZ ¼ 0 NTZ ¼ 1 Difference
Model 3 (TC3)
No trips
CS
CS/trip
No trips
CS (£)
CS/trip
1.32 2.18 0.86
£221.1 £365.5 £144.4
£168.02 £168.02 e
1.33 2.20 0.87
£324.3 £536.8 £212.4
£244.31 £244.31 e
group, and all members did not incur the costs themselves (e.g. children did not incur any actual cost as it was paid by their parents). 5. Discussion
Table 5 Regression estimators by Limdep Gamma model with different travel cost levels applied. Parameter
Model 1
Constant 1.76437*** Travel cost 1 0.00389*** Travel cost 2 Travel cost 3 Income 0.1242*** Biodiversity 0.05868*** AGE 0.010661 NTZ 0.509371*** Criteria for assessing goodness of fit Chi-square 16.86467*** Log-likelihood function 102.0839 Consumer surplus estimation CS per trip £257.1 Alpha 2.49653***
Model 2
Model 3
1.8346***
1.45655***
0.0037*** 0.11718*** 0.0588*** 0.009993 0.492108***
0.00254*** 0.0899*** 0.05234** 0.012973* 0.440799***
19.05679*** 100.0800
16.5786*** 102.2863
£270.3 2.65624***
£393.7 2.48313***
***, **, and * indicates the coefficient is statistically significant at the 1%, 5% and 10% level respectively.
The results of the Gamma model are very similar to those of simple Poisson in which all signs of coefficients are the same, but estimated CS is slightly higher than that of a Poisson regression. The estimated value of alpha, dispersion parameter, in each model indicates that the underdispersion was handled successfully.
4.5. Impact of the no take zone (NTZ) The impact of the NTZ on the value of the trip to the island can be determined from the regression model by estimating the number of visits given the average travel cost for the two groups e those agree with the designation of NTZ (i.e. NTZ ¼ 1) and those who disagree (i.e. NTZ ¼ 0). From the regression model, those who agreed with the designation of the NTZ would undertake nearly 65% more trips over a three year period than those who did not, all other things being equal (Table 6). The additional consumer surplus generated over the period examined, as a result of the increased number of visits was estimated as £144/visitor group based on TC2, and £212/visitor group based on TC3. The average consumer surplus per trip, however, was unaffected by the designation of the NTZ. Extrapolating this to a total value is less straightforward, as the travel costs used in the analysis related to groups of various sizes, the assumption being that, for a family, the decision to visit the Island would be based on the total cost of the family trip. An individual’s consumer surplus is not just the total divided by the average number in the group, as the marginal travel cost of the additional group member is only the cost of the ferry trip and the travel time forgone (i.e. not the transport cost, which was already incurred by the first passenger). Further, the decision to travel to the Island was not necessarily made by all members of the family
This study is particularly significant because it was the first approach to carry out an economic analysis of tourism benefits from MPAs in the UK, where tourism demand for unspoiled natural area is currently rising. The estimated consumer surplus of recreation in the MNR was very high compared with other recreational travel cost studies (e.g. see Rosenberger & Loomis, 2001; Willis, 1991). This may be due to several factors. Petrol costs in the UK are higher than many other countries and the ferry trip to Lundy is also relatively expensive, making the total travel costs higher than that experienced in most other studies. Further, the estimated demand curve was found to be highly inelastic (i.e. jb1j1), and this induces high consumer surplus for the site. In most other studies, alternative similar recreational activities were more available, resulting in more elastic demand and lower consumer surplus. As Lundy is a unique recreational experience in the UK, a small number of visitors are willing to pay a relatively high cost to visit the island. The existence of multi-purpose trips can also result in the recreational benefits being overestimated. An implicit assumption in the analysis is that the costs were incurred exclusively to visit Lundy. The 39% of multi-purpose trip observed in this study suggests that there can be some degree of overestimation, but as most visitors’ primary objective was to visit Lundy, this is not likely to cause a large distortion. Similarly, multi-destination trips were not a major concern as visitors’ objective was only to visit Lundy. A further difficulty in deriving estimates of consumer surplus is the problem of “group” rather than individual travel. While group size averaged two adults and one child, there was considerable variation around these means. Group costs were therefore a function of group size, adult/child mix and distance travelled. Further, while the ferry costs were incurred individually, other travel costs (e.g. fuel) are incurred only once regardless of number of travellers. Deriving a meaningful individual cost was not feasible, as it was not considered appropriate to divide the travel costs by the number of travellers. An implicit assumption of the analysis is that the decision-making unit is the group, and hence the consumer surplus relates to the group rather than individuals. Group size was not incorporated as a variable into the econometric model. Larger groups would have higher travel costs and would, therefore, be assumed to make fewer visits. Despite the high value generated by visiting the island, the majority of the visitors did not revisit Lundy for several years. This factor contributed to the underdispersion of the dependent variable. This phenomenon might be explained by poor accessibility of the island and lack of incentives to induce repeated visits to Lundy given the high cost of access. As a consequence, the distribution of visitors’ residences tends to be evenly spread out across the UK rather than diminishing the further away from the site (i.e. as would be expected assuming a gravity model).
D.-R. Chae et al. / Tourism Management 33 (2012) 971e977
The negative coefficient on the income variable is a further surprising outcome, which suggests relatively lower income respondents had a tendency to visit Lundy MNR more frequently during the last three years. This suggests that UK based ecotourism is essentially an inferior good. Given its relative remoteness to most parts of the UK, high income earners would have a substantial opportunity cost of travel time. Many overseas (European) locations could be reached in the same or less time, particularly from the major centres. In contrast, lower income groups face greater financial restriction but relatively lower opportunity cost of travel time. In this regard, domestic MPA tourism can be a substitute good for international tourism for lower income earners. The study was unable to capture all users of the island, with the scuba divers being the main group excluded in the analysis. A similar survey of scuba divers was also undertaken, with 23 responses. However, these were not adequate to be added into the regression model. The study was also unable to capture the views of all visitors visited the island over a period of full year. In addition, the sample use in this study was quite small and the data were collected from only the sample of visitors which gives rise to the issue of “endogenous stratification”. Future studies on travel costs should avoid these sample related problems at the planning stage of the survey. The results of the study suggest that there can be considerable recreational benefits generated by the creation of marine parks. However, considerable additional research is necessary before these values are used to justify additional marine parks in the UK. The high value associated with Lundy may reflect its uniqueness as a marine park in the UK. Additional parks may divert Lundy visitors, resulting in a shift in the demand curve for Lundy tourism and a decrease in consumer surplus. As a result, a new park would be expected to have an average lower consumer surplus than those estimated in this study and also reduce the benefits generated by Lundy. Whether the total combined benefits would increase or decrease requires further consideration.
References Ami, D., Cartigny, P., & Rapaport, A. (2005). Can marine protected areas enhance both economic and biological situations? Comptes Rendus Biologies, 328(4), 357e366. Baht, M. G. (2003). Application of non-market valuation to the Florida Keys marine reserve management. Journal of Environmental Management, 67, 315e325. Cameron, C., & Trivedi, P. K. (1990). Regression based tests for overdispersion in the Poisson model. Journal of Econometrics, 46, 347e364. Cesario, F. J. (1976). Value of time in recreation benefit studies. Land Economics, 52(1), 32e41. Cesario, F. J., & Knetsch, J. L. (1976). A recreation site demand and benefit estimation model, regional studies. Taylor and Francis Journals, 10(1), 97e104. Creel, M., & Loomis, J. (1990). Theoretical and empirical advantages of truncated count data estimators for analysis of deer hunting in California. American Journal of Agricultural Economics, 72, 434e441. Driml, S. M. (1997). Bringing ecological economics out of the wilderness. Ecological Economics, 23(2), 145e153.
977
Dugan, J. E., & Davis, G. E. (1993). Application of marine refugia to coastal fisheries management. Canadian Journal of Fisheries and Aquatic Science, 50(10), 2029e2042. Englin, J. E., Holmes, T. P., & Sills, E. O. (2003). Estimating forest recreation demand using count data models. In E. O. Sills, & K. L. Abt (Eds.), Forests in a market economy (pp. 341e359). London: Kluwer Academic Publisher. Englin, J., & Shonkwiler, J. S. (1995). Estimating social welfare using count data models: an application to long-run recreation demand under conditions of endogenous stratification and truncation. The Review of Economics and Statistics, 77, 104e112. Friends of Lundy. (2004). Winter helicopter service takes off, Friends of Lundy news letter. Newsletter 2004, 1. Groger, J. T., & Carson, R. T. (1991). Models for truncated counts. Journal of Applied Econometrics, 6, 225e238. Hagerty, D., & Moeltner, K. (2005). Specification of driving costs in models of recreation demand. Land Economics, 81(1), 127e143. Hill, G., Rosier, J., & Dyer, P. (1995). Tourism development and environmental limitations at Heron island, Great Barrier reef: a response. Journal of Environmental Management, 45(1), 91e99. Jones, P. J. S. (1999). Marine nature reserves in Britain: past lessons, current status, and future issues. Marin Policy, 23, 375e396. Langham, A. F. (1994). The island of Lundy. Phoenix Mill: Sutton Publishing Ltd. Larson, D. M. (1993). Separability and the shadow value of leisure time. American Journal of Agricultural Economics, 75(3), 572e577. Leeworthy, V. R., Wiley, P. C., English, D. B. K., & Kriesel, W. (2001). Correcting response bias in tourist spending surveys. Annals of Tourism Research, 28(1), 83e97. Lewison, R. L., Crowder, L. G., Read, A. J., & Freeman, S. A. (2004). Understanding impacts of fisheries by catch on marine megafauna. TRENDS in Ecology and Evolution, 19(11), 598e604. McConnell, K. E. (1992). On-site time in the demand for recreation. American Journal of Agricultural Economics, 74(4), 918e925. McKean, J. R., Johnson, D. M., & Walsh, R. G. (1995). Valuing time in travel cost demand analysis: an empirical investigation. Land Economics, 71(1), 96e105. Park, T., Bowker, J. M., & Leeworthy, V. R. (2002). Valuing snorkeling visits to the Florida Keys with stated and revealed preference models. Journal of Environmental Management, 65(3), 301e312. Rosenberger, R. S., & Loomis, J. B. (2001). Benefit transfer of outdoor recreation use values: A technical document supporting the forest service strategic plan 2000 revision, (Gen. Tech. Report RMRS-GTR-72). Fort Collins, CO: USDA Forest Service, Rocky Mountain Research Station. Rouphael, A. B., & Inglis, G. J. (1997). Impacts of recreational scuba diving at sites with different reef topographies. Biological Conservation, 82, 329e336. Shaw, D. (1988). On-site samples’ regression: problems of non-negative integers, truncation and endogenous stratification. Journal of Econometrics, 37, 211e223. Smith, V. K. (1988). Selection and recreation demand. American Journal of Agricultural Economics, 70, 29e36. Song, H., & Li, G. (2008). Tourism demand modelling and forecasting e a review of recent research. Tourism Management, 29, 203e220. Sumaila, U. R. (1998). Protected marine reserves as Fisheries management tools: a bioeconomic analysis. Fisheries Research, 37, 287e296. The National Trust. (2002). The archaeology and landscape Lundy, (a field guide). The National Trust. Valentine, P. S., Birtles, A., Curnock, M., Arnold, P., & Dunstan, A. (2004). Getting closer to Wales e passenger expectations and experiences, and the management of swim with dwarf minke whale interactions in the Great Barrier Reef. Tourism Management, 25, 647e655. Willis, K. G. (1991). The recreational value of the forestry commission estate in Great Britain: a ClawsoneKnetsch travel cost approach. Scottish Journal of Political Economy, 38(1), 58e75. Winkelmann, R. (1995). Duration dependence and dispersion in count data models. Journal of Business & Economic Statistics, 13(4), 467e474. Winkelmann, R. (2003). Econometric analysis of count data (4th ed.). Berlin: Springer. Zeller, D. C., & Russ, G. R. (1998). Marine reserves: patterns of adult movement of the coral trout (Plectropomus leopardus (Serranidae)). Canadian Journal of Fisheries and Aquatic Science, 55, 917e924.