A study of environmental conflict:: the economic value of Grey Seals in Southwest England

Biodiversity and Conservation 12: 2361–2392, 2003.
c 2003 Kluwer Academic Publishers. Printed in the Netherlands.



A study of environmental conflict: the economic value
of Grey Seals in southwest England
VALENTINA BOSETTI 1 and DAVID PEARCE 2, *
1

Dipartimento Informatica, Sistemistica e Comunicazione, Universita` degli Studi di Milano Bicocca,
Milan, Italy; 2 Centre for Social and Economic Research on the Global Environment, University College
London, Gower Street, London WC1 E 6 BT, UK; * Author for correspondence (e-mail:
d.pearce@ ucl.ac.uk)
Received 3 January 2002; accepted in revised form 17 December 2002

Key words: Conflict, Economic value, Externality, Fishing, Grey seals
Abstract. This paper reports an analysis of a typical case of negative bilateral externality – a situation in
which two legitimate activities, fishing and wildlife conservation, each give rise to damages to the other
party. The Cornish fishing industry believes that its annual profits are reduced by an estimated £100 000
because of the damage by seal populations to caught fish. About 80 individuals belonging to the Cornish
Grey Seal population (of about 400 specimens) are killed as a by-catch of trawling. Thus, the status quo is
clearly inefficient: seals are perceived to damage fish and fishermen definitely damage seals. The
biological dynamics of the seal population is not absolutely clear, so that a precautionary approach
requires that care should be taken to avoid the risk of damaging the population in an irreversible way.
Moreover, public opinion considers seals to be a high value ‘flagship’ species. One of the goals of any
conflict resolution should be to capture the economic value of seal conservation – i.e. to convert
conservation benefits into resource flows – and use at least part of it in order to create incentives for a
more efficient allocation of resources. The authorities should invest in seal conservation (i.e. compensating fishermen) if the benefits deriving from conservation exceed the opportunity costs of conservation.
Such a solution clearly requires that the conservation benefits be estimated. To investigate the economic
value of seal conservation a contingent valuation study is carried out. A contingent valuation study
utilises a questionnaire approach and part of the questionnaire seeks to elicit individuals’ willingness to
pay (WTP) for a change in the state of some good or asset, in this case seal conservation. Due to resource
limitations, the sample size of those interviewed in the study reported is small, so that we cannot be
extremely confident about the results. However, they are consistent with those derived from similar
studies on ‘flagship’ species. Results show a mean WTP for recreational use of seals of about £8 per
person for the option of seeing seals in a specialised sanctuary for seals recovered from accidents, and
closer to £9 for seeing seals in the wild. The annual non-use value of seals – i.e. value unassociated with
actual viewing – was found to be £526 000 in the most conservative estimation, aggregated over the Seal
Sanctuary visitors. This economic potential could be realised in several ways and used to compensate
fishermen for changing fishing techniques, targets and fishing areas. Finally, we investigate the role the
Seal Sanctuary is playing in this context and some policy suggestions are discussed.

Introduction
The present study investigates a ‘classic’ conflict scenario in wildlife conservation:
wild seals competing with local fishermen for the same resource, namely fish. Seals
are killed or injured mainly as a by-catch of trawling. Current fishing activity could
ultimately reduce seal populations below a biological threshold critical for survival.

2362
So far seals have been considered as a nuisance to fishermen, with no economic
value. However, seal watching is becoming important for the local tourism industry
and this could transform the traditional perception of these animals. Seal management has to take in consideration the trade-off between perceived costs to the fishing
industry, and benefits to the tourism industry. This paper investigates the economic
importance of seals in a particular location: southwest England.
We analyse the different economic impacts that the seal population produces. The
aim is to suggest policy mechanisms for sustainable seal management. Indeed, if the
economic value of seals in their conserved state is significant, then different
mechanisms can be used to capture this value and use it to compensate fishermen for
the catch losses, whilst conserving the seals. To estimate the conservation value of
seals, a contingent valuation survey was carried out. Contingent valuation has been
used mainly because use value is not the only possible value that tourists can attach
to the presence of seals in Cornwall. Indeed, they may also have a ‘non-use’ value
(i.e., existence and bequest values), that we have tried to capture through the
application of the contingent valuation method (hereafter, CVM). For similar
studies, see Langford et al. (1998a, 1998b).
Three parties play important and different roles in the environmental conflict
between seals and fishermen. The first is the fishing community that claims to suffer
continual losses caused by the seal population (mainly as damage to caught fish).
Fishermen threaten to react to what they see as government indifference to their
plight by solving the conflict autonomously. Cornwall and Devon are more than
average dependent on fishing activity. Moreover, while the local unemployment rate
has increased in the last decade, the new European Union limits on fishing effort are
further aggravating the economic fortunes of the industry.
The second player is the Grey Seal population that inhabits the coasts of
southwest England. It represents the southernmost reproduction site of the whole
European macro-population (together with the Britannic one). Since only recent
data on changes in population size are available, we still lack the scientific basis to
assert that the population is critically endangered. However, the limited data seem to
support the possibility that seals populations are endangered.
Thirdly, there is the tourism industry, which represents an important economic
activity in the area. Tourists have the opportunity to make non-consumptive use of
seals in their natural environment (wild seal viewing) as well as in the man-made
context of a Seal Sanctuary, in Gweek. Moreover, others who do not visit the area
derive some non-use value from the knowledge that seals are conserved. In
economic terms, both use and non-use values generate human wellbeing, and hence
their sum defines the total economic value of the wildlife resource. Therefore, all
sources of economic value have to be taken in consideration.
An estimate of the total economic value of seals (i.e. use plus non-use values)
would include the general public’s willingness to pay (WTP) for seal conservation,
not just the tourist component. However, due to the resource limitations of the study
the non-use values of the wider non-visiting population have not been estimated.
The most feasible way to capture the seals’ conservation value in terms of cash
flows would be to charge tourists going to the Seal Sanctuary or on a seal-watching

2363
trip. Therefore, the study focuses on the use and non-use values of visitors to the
Seal Sanctuary of Gweek.
There are two main categories of objectives in this study: the first being more
policy-oriented and the second more methodological. Policywise we deal with the
issue of assessing the conservation value of seals and the total costs that they impose
on the fishing industry. This enables us to highlight strategies to capture this value
and show how it might be used to compensate the fishing industry. By estimating the
total costs imposed on the fishing industry by the seal population, we also secure an
idea of the distribution of the damage among different fisheries. To this end, several
representatives of the fishing community were interviewed. Additionally, the
Cornish Fish Producers Organisation (CFPO) mailed a questionnaire to each of its
members in order to investigate the losses due to seal damage to fish. We also
evaluate the benefits linked with the seals’ presence in Cornwall. For this purpose a
contingent valuation study was carried out in order to:
• Elicit tourists’ WTP for conservation of the seal population in Cornwall. The
procedure involved interviewing a sample of tourists and generalising the results
of the study to the wider visitor population.
• Estimate the non-use value of the seals population among tourists.
These values are then used to clarify the general picture of the seals–fishermen
conflict and, hence, to propose some policy solutions.
The second set of goals involved the investigation of some economic and
econometric questions, namely:
• procedures for estimating parameters,
• influence of the level of uncertainty in respondents’ answers to different format
questions,
• issues arising from the truncation of WTP responses,
• influence exerted by information related to alternative recreational uses on the
resulting WTP,
• different forms of value and different techniques to elicit them.
The research involved in-person and telephone interviews of ‘key informants’,
in-person analysis of the different facilities for the ‘non-consumptive’ use of seals,
and an in-person contingent valuation pre-test and main questionnaire survey of
tourists.
The questionnaire was carried out in three different locations: the Seal Sanctuary
(Gweek), and the harbours of St. Ives and Dartmouth. One goal of the research was
to investigate whether users of the two different ‘goods’ (seals in the sanctuary and
seals in the wild) differed significantly in their answers. The questionnaire was
designed to incorporate three different scenarios and, therefore, to investigate
different economic aspects of the problem. The scenarios are described in detail in
the Section ‘The status quo’. The questionnaire was carried out asking a selfselected sample of people rather than a genuine random sample of the southern
England population.
We first present the background picture and describe the different ‘players’. Then

2364
Table 1. Seal population data for Cornwall.
Number of individuals per year
Pup production (Wescott 1996)
Weaned Pups (Wescott 1996)
Seals caught as a by-catch (Glen 1998)
Natural mortality (Glen 1998)

125
112
79–90
60

we describe the contingent valuation study and present the econometric analysis.
The last section is devoted to an overall discussion and conclusions.

The status quo
Even though it is a common species for the UK, the grey seal is quite rare worldwide
and the world population is estimated to be around 200 000 individuals. The grey
seal is listed in Annex 2 of the European Union Habitats Directive (English Nature
1996). This implies a responsibility towards conservation of seals. The Cornwall
site is of particular importance because, together with Brittany, it represents the
southernmost reproductive site of grey seals. Further, its location seems to be of key
significance in European migration patterns. Even though a serious census has not
been carried out yet, the Cornish grey seals population is estimated to be between
350 and 400 individuals (Wescott 1996).
Data related to population trends for the last 10 years (which would be extremely
relevant for any conservation programme) are not available, because figures started
to be collected only in 1997. The only available data are based on casual observations and, for this reason, they cannot be considered scientifically accurate. The
available data are reported in Table 1. According to these data, the Cornish
population is decreasing by 8–9% per year. At the time of writing, it is not possible
to confirm and validate such a conclusion, but according to the veterinary staff of the
Seal Sanctuary and most of the skippers working on seal-watching boats, the seal
population seems to be fairly constant.
Information on the seals–fishermen conflict is based on key interviews as well as
on the report by Glen (1998). The interaction is mainly in the form of direct
predation from the nets. Seals eat fish from the nets. They bite the tail, aiming for the
liver which is very rich in oil, required for them to store the fat that is essential to
survival in the cold waters. As a result, fish are damaged and rendered nonmarketable. Seals damage mainly monkfish and hake, which happen to be very high
value fish (monkfish is sold at £7 / kg, hake at £4.5 / kg). Fishermen using fixed nets,
mainly tangle nets, are the most affected. The seals’ targets are generally medium
sized boats (from 5 to 12 m) fishing between 6 and 20 miles off the coast. The
interactions take place mostly in the summer and are localised in ‘hot spots’, around
the Lizard Peninsula and in the area between Pendeen and St. Agnes. This
information is confirmed by the CFPO, who sent a questionnaire to each member of
the organisation in order to collect updated data on the locations, forms and nature

2365
of the seals–fishermen conflicts. At the time of this study they had received 21
replies, which represents about 10% of the CFPO Membership. Obviously, fishermen have an incentive to exaggerate fish loss and it is possible that the damages are
accordingly overstated. Nonetheless, the existence of a conflict over economic
losses is an undeniable fact.
The reported damage is unevenly spread. Seals appear seriously to affect just that
part of the Cornish fleet that is less flexible in fishing techniques and choice of
fishing area, i.e. the more vulnerable group. Moreover, the ‘‘structural changes
within the local fishing industry, towards fewer, larger boats mean that small, daily
boats are disproportionately threatened’’ (Berry 1996). This probably means that
seals, in addition to the real damage they cause, are playing the role of a scapegoat
and the ‘seal question’ is partially over-dramatised from the fishermen’s perspective.
Nonetheless, any sensible conservation policy should seriously account for the
effects of the seals on the fishing community. The lack of participation by the local
community is an enormous barrier for any conservation project to be successful.
Also, active fishing ports are a picturesque attraction of Cornwall. Therefore, the
fishing community contributes, indirectly as well as directly, to the economy of the
area. Given the localised form of the interaction between seals and fishermen, it
becomes easier to identify the fishermen mostly affected. One possible solution
could be to compensate them alone. It is quite straightforward to quantify the losses
(considering the number of fish heads in the nets). Moreover, the damage is highly
localised among the fishing areas. A second possibility might be the elimination of
isolated rogue seals. This has proved to be a successful solution in some cases and is
permitted by the Conservation of Seals Act. In fact, any order or prohibition on
killing seals does not apply ‘‘ . . . to seals killed to prevent them causing damage to a
fishing net or fishing tackle, provided the seals were in the vicinity of the gear at that
time’’ (Bonner 1989). However, as many interviews with the local fishing community and other interested parties highlighted, seals are not killed exclusively in
the proximity of fishing areas. Another proposed solution is to use contraception in
order to contain the population within optimal levels. The main drawback of such a
solution is that it would be quite expensive and would require a lot of work in the
field.
A solution that appears infeasible is a cull. Firstly, it would be strongly opposed
by public opinion. Secondly, it probably would not solve the problem because of the
other competitive species populations that might take the place of seals in the food
chain (e.g., seabirds or cetaceans). Finally, it would be just a temporary solution
because new individuals might move into the area and replace the previous
population.
The National Seal Sanctuary is located in Gweek (Cornwall) and is owned by the
same company that owns the Newquay Life Centre (Cornwall). It was instituted 40
years ago and early on mainly had just rescue and veterinary functions. Nowadays it
works also as a recreational and information centre. The Sanctuary receives about
200 000 visitors per annum and generates a minimum turnover of around £1 million.
The entry fees are £5.95 for adults and £3.75 for children. The whole staff comprises

2366
15 people. Concerning the rescue activity, the Sanctuary annually saves from 24 to
54 seal pups (half of all the pups that are born in Cornwall). Most of them come
from the Cornish coast, but a small number come from other regions such as Devon,
Dorset or the Isles of Scilly. The seals are released after about 6–10 months, even
though, in many cases, they could be released after 2–3 months. The delayed release
is due to the fact that, even though the Sanctuary is open all year around, the
business season is during the summer. Up to 6 years ago, the majority of the releases
took place in the area around the Lizard Peninsula. Now releases take place, when
possible, in the same area where the pup was found. This is because fishermen
showed a strong opposition to localised releases.
Seal watching is a significant attraction of this part of England. It can be done just
using a pair of binoculars and standing on the coastal path. The best way, however,
is to take a boat trip. Among other places, they are organised both in Dartmouth
(Devon) and St. Ives (Cornwall). Boat trips departing from the Dartmouth harbour
are very well organised. There is one big boat that can carry around 60 people and is
fully equipped in order to provide facilities for the entire family. The trip lasts
around two hours and costs £5 for adults and £4.5 for children. The boat runs from
two to seven times per week. Boat trips departing from St. Ives are less organised.
There are six boats that make up to six trips per day, each lasting around an hour and
a half. Boats are much smaller than the one in Dartmouth (from 27 to 35 feet) and do
not provide facilities in order to accommodate young children. The maximum
capacity of the entire fleet is 310 people per day. The fares are £7 for adults and £6
for children. Earnings per year are around £60 000 for the entire fleet, while the cost
per year per boat is around £4000. The industry is only marginally profitable and
poorly organised. Consumer demand has fallen in the last decades and, indeed,
around 60% of the people interviewed complained about the experience. The
skippers working for the enterprise, when interviewed, outlined three main reasons
for the decreased demand: the change in preferences of the younger generation,
economic recession, and the presence of the Seal Sanctuary. However, the results of
the contingent valuation survey suggested a very high WTP for seals in their natural
environment, suggesting that the seal-viewing industry bears some responsibility. It
should be underlined that the more frequent consumers are families with very young
children and, while they can easily take part in boat trips organised in Dartmouth,
this is not the case for the small boats departing from St. Ives.

The contingent valuation study
The survey population of the study comprises the tourists visiting the Seal Sanctuary
in Gweek, and the tourists on a seal-watching boat trip, either departing from the
harbours of St. Ives or from Dartmouth. Particular care was taken to collect a
random sample and to interview mainly householders. However, this sample is not
representative of all tourists visiting southwest England, given that it is a selfselected sample (respondents have been or are already experiencing the use of the

2367
good). Nevertheless, this allows us to gain some important insights from this
self-selected sample’s stated preferences.
There are different features of the total economic value of a natural resource. The
study tries to capture these by constructing, within the survey, different scenarios,
each aiming to collect a different component of the information needed. For a
complete version of the questionnaire, see Appendix 1. The first and second
scenarios asked about the maximum WTP to ‘use’ the resource. In particular, given
the presence of two alternative ways of ‘using’ seals, the second valuation question
aimed to estimate the WTP for the alternative that the respondent is not actually
experiencing. Thus, the maximum WTP for a seal-watching boat trip was elicited by
interviewing people at the Seal Sanctuary, and vice versa. In the last scenario, the
question asked refers to the maximum amount the respondent is willing to donate to
a conservation programme, without involving any direct or indirect use of the
resource. The conservation programme that was presented did not imply a growth in
the seal population, but simply the creation of funds to mitigate the conflict among
fishermen and seals. Seals would be able to survive and reproduce in their natural
environment rather than being killed or injured. Although this third valuation
question may capture a ‘use’ value component, if respondents belong to the sealwatching sample (deriving from the potential effect of the conservation programme
on the probability of seeing a seal during a seal-watching trip), this possibility has
been minimised through a careful formulation of the question. Two surveys were
designed for the two samples, at the Seal Sanctuary and on the seal-watching trip.
Each survey was divided into four main sections:
1. Social and environmental attitude section.
2. Descriptive section, regarding the experience (visiting the Seal Sanctuary or
going for a seal-watching boat trip) and the cost of it.
3. Evaluation section.
4. Socio-economic description section.
The attitudinal and socio-economic information on respondents was collected to
help respondents to out their responses to seal conservation in the context of other
environmental concerns and to test the significance of explanatory variables. The
descriptive section aimed to both capture important suggestions of possible management solutions and to gather information about the costs that respondents had to bare
in order to experience the service. The valuation section is divided into three
scenarios. Each of them follows the same structure and is composed of an
explanatory introduction to the scenario in order to permit a standardization of
information across respondents. The introductory part is followed by the valuation
question.
The introductory part of Scenario I is followed by two different forms of
questions to elicit the maximum WTP. The respondent is firstly reminded how
expensive the management of the Seal Sanctuary is, and similarly for the sealwatching questionnaire. The questions then emphasised how the price of the good is
not going to change, and that the aim of the survey is to collect the maximum WTP
of the respondent. The respondent is asked to cast back his / her mind to the moment

2368
he / she decided to come and assess his / her maximum WTP for the actual experience. The first valuation question is in the dichotomous choice (DC) format –
i.e. in response to a suggested ‘price’ the respondent answers yes or no. A set of
three bids was proposed on the basis of the results of a pre-test. This is followed by a
question about the level of certainty of the previous yes / no answer. The second
valuation question is an open-ended (OE) question, i.e. takes the form of a ‘what is
your maximum WTP?’, followed by a question designed to elicit the level of
certainty the respondents would associate to their preceding answer. For both
valuation questions the payment vehicle is an increase in the entrance fee.
Scenario II is designed to capture the experience of seal watching both in the
seals’ natural environment and in a man-made context. Initially, each respondent is
informed of the two alternatives and their characteristics. Two hypothetical prices
for each of the possibilities are stated. The respondent is asked which option he / she
would have chosen, given these prices. Following the first question, a second one is
asked which is identical but for the price of the alternative option. The price might
be higher or lower, according to the previous answer. This amounts to a combination
of a choice experiment approach with a double bounded dichotomous type question.
Again the payment vehicle is the entrance fee.
Scenario III aims to elicit the maximum WTP for tourists’ non-use value of seals.
The respondent is informed of the conflict situation concerning the seal population
and the fishermen, and of the resulting problem. This is followed by the description
of a hypothetical conservation plan aimed to mitigate the conflict in the most
sensible economic way. Finally, the respondent is asked whether he / she would
consider giving a donation to a conservation organisation, in addition to the entrance
fee she / he has paid. Negative answers to the referendum question are followed by a
debriefing question about different reasons for declaring a WTP equal to zero. The
role of such a debriefing exercise is to identify respondents who expressed zero
WTP as a form of protest, rather than as true WTP. As discussed in the Section ‘The
contingent valuation study’, this tool makes it possible to eliminate protest zeros
from the sample, given that they do not represent an effective value. Respondents
willing to donate are shown a range of different amounts of money, from £0p to over
£100. They are asked to:
• tick the amounts they are almost certain they would pay as a seal conservation
levy;
• put a cross by the amounts they are almost certain they would not pay as a levy;
• leave blank the amounts they are unsure whether they would pay or not.
The payment vehicle is a donation.
Finally, the socio-economic characteristics of respondents were collected to
examine the composition of the sample, as well as to analyse how the WTP varies
with these characteristics. The survey collected data on: age, gender, number of
people in the household, number of children, level of education, occupation, and
household income per year.
The fieldwork for the survey was divided into two main steps: a set of ‘key
informants’ interviews and a pre-test, and the main survey. The first step involved

2369
interviewing fishermen’s representatives – the Chief Fishery Officers of the CFPO –
and some fishermen working in Newlyn harbour. This harbour plays a fundamental
role in the whole problem, for two reasons. Firstly, in terms of value of fish landed,
Newlyn is consistently the highest in England (PESCA 1996). Moreover, fishermen
working from Newlyn are by far the most concerned about the damage due to grey
seals (Glen 1998). A set of interviews with both the veterinary and management
staff of the Seal Sanctuary in Gweek followed. Finally, useful information was
collected from several interviews with skippers working on seal-watching boat trips
in St. Ives. While this information was extremely useful in understanding the
problem and what the structure of the questionnaire should be, a preliminary pre-test
conducted on a small sample (12 individuals) was used to optimise the phrasing of
the questions and to choose a meaningful range for the values of bids in the DC
questions.
Once the survey was designed, it was carried out during 10 days of fieldwork.
Five days were spent in the Seal Sanctuary, completing 112 questionnaires; 2 days
in Dartmouth for a total of 50 questionnaires completed on board of the sealwatching trip; and, finally, 3 days in St. Ives for a total of 44 questionnaires. Each
questionnaire was face-to-face. The interviewer presented cards to help the respondent understand the questions. An introduction preceded each interview to explain
the aim of the research project as well as to capture the respondent’s full attention.
Table 2 reports the sample descriptive statistics.

Econometric analysis
The survey was designed to collect information by using different elicitation formats
and by holding a constant payment vehicle (in the first two scenarios) to test the
convergent validity of results. Internal consistency, i.e., consistency with economic
rationality, could not really be tested because of the size of the sample. To have a
consistent analysis of the data, in particular when collected via DC techniques, the
size of the sample is fundamental. Each of the two sub-samples is of about 100
individuals – too small to be totally confident in the results. Nevertheless, for the
last scenario, when regressing on the entire sample, the results are more reliable. In
almost all the cases, the explanatory variables have the expected sign, even though
the level of significance is often not reliable.
Scenario I
The WTP for this scenario was elicited by both DC and OE questions. Each
respondent was asked a DC question (the set of three bids had been calibrated
through the analysis of the pre-test results) and then to state his / her maximum WTP.
The aim is to elicit the maximum WTP, for each individual, for non-consumptive use
of seals. The reason both elicitation formats were used arises from the different
advantages and shortcomings each format presents. DC or referendum questions
present an important advantage in that they mimic everyday market transactions: the

2370
Table 2. Characteristics of the sample (total number of respondents 5 206).
Location
Seal Sanctuary
Number of respondents
Males
Age (mean)
Number of people in the household (mean)
Number of children (mean)
Education (%)
Primary
O Levels
A Levels or equivalent
First degree level
Professional qualification
Higher degree
Annual total household income after tax (mean, £)
Annual per capita income, after tax (mean, £)
Income non-response
Employment (%)
Self-employed
Employed full-time
Employed part-time
Student
Looking after the home full-time / housewife
Retired
Unable to work due to sickness or disability
Unemployed
Number of people in the party (mean)

Seal watching

112

94
51.5%

40.7
2.45
0.74

42.8
3.01
0.59
1.9
39.3
12.1
18.4
10.2
5.8

23 840
12 020

24 750
9580
5.33%
3.39
64.07
14.07
1.45
5.82
7.28
1.45
1.94

3.54

2.44

The mean income per capita and respondents’ party size are higher in the sub-sample of the Seal
Sanctuary. Similarly, the number of children is higher and this is mainly related to the difficulties of
going on a small boat with very young children.

individual faces a price and simply has to decide whether to take the good for that
price or not. For this reason it is less likely that respondents misunderstand the
question. The referendum format has been increasingly applied to take advantage of
this characteristic, and was recommended in the NOAA (National Oceanic and
Atmospheric Administration) Panel Guidelines (NOAA Panels 1992). It is as well to
note, however, that some authors have questioned the extent to which answers to
multiple bounded DC questions may be used to retrieve fully rational economic
preferences – see Scarpa and Bateman (2000) and Bateman et al. (2001).
Data collected, pertinent to each individual in the sample, were (a) a bid value
related to the entrance to the Seal Sanctuary (or seal watching), and (b) the yes / no
response. We assume that the unobserved continuous dependent variable y* is the
respondent’s true WTP for the entrance to the Seal Sanctuary (or seal watching).
Moreover, we assume that the underlying distribution of y* is a normal distribution
and that y* is conditional on a vector of explanatory variables x j (where j is the
number of explanatory variables). DC CVM data are often analysed using utilitybased models with errors distributed Extreme Value Type I which can be analysed

2371
Table 3. Single bounded model, all values certain.
Single bounded normal model

Sample mean of predicted WTP (£)

Seal Sanctuary (n 5 112)

Seal watching (n 5 94)

8.48

9.75

with logit probabilities. However, at our sample sizes these are operationally
indistinguishable from censored regression models with normally distributed errors,
so we use the latter approach, originally proposed by Cameron and James (1987).
The unobserved variable can be described, then, as y* 5 x j b1 e, while the observed
variable is an indicator that can take just the two values yes / no, which will be called
IY and IN , respectively. Therefore, the probability of observing a YES response at a
given bid amount is given by
PrsYesubidd5Prs y * $bidd5Prsx j b1´$bidd5Prs´$bid2x j bd

(1)

where bid is the bid stated in the dichotomous question. Assuming that ´| N(0, s),
we have:

S

D

S

D

S

bid2x j b
e bid2x j b
e bid2x j b
Pr ].]]] 512Pr ],]]] 512F ]]]
s
s
s
s
s

D

(2)

Analogously, we can consider the probability of saying no as
PrsNoubidd512Prs y * $bidd512Prsx j b1´$bidd512Prs´$bid2x j bd

(3)

The contribution to the log-likelihood function for each observation i, is given by
lnL5lnsIY *s12Fsbid2x j bd / sd1IN *Fssbid2x j bd / sdd

(4)

A function programme was designed to introduce this log-likelihood function. A
command programme recalls such a routine, so that STATA is able to read and
maximise the function. This procedure was repeated for each of the models.
The outputs of the maximisation algorithm are the values for the parameters that
maximise the log-likelihood function of the sample, see Appendix 2. In previous
approaches the bid of the referendum was included only as one of the regressors of
the conventional probit / logit analysis. The advantages of using this procedure, as
stated by Cameron (Cameron 1988; Cameron and James 1987), are that:
‘‘ . . . estimated coefficients (other than s) can be interpreted roughly as one would
interpret the results from an OLS regression’’ (Cameron 1988), whereas, ‘‘in the
past, empirical work using binary choice probit / logit models with referendum data
overlooked the opportunity for this simple reinterpretation of the probit / logit
parameters and therefore faced the awkward limitation of working only with
probability estimates’’.
The mean WTP was calculated for both sites using sample averages of the
covariate variables. Results are presented in Table 3.
Information about the respondents’ level of certainty was collected by introducing
a follow-up question to complete the closed-ended part of Scenario I (Ready et al.

2372
Table 4. Sample mean of predicted WTPs (£) for alternative options.

Seal Sanctuary (n 5 112)
Seal watching (n 5 94)

OE

DC, all levels
of certainty

DC, only more than 95% level
of certainty

7.93
8.89

8.48
9.75

8.13
8.84

Table 5. Model summary and coefficients for OE Model, Seal Sanctuary sample.

1998). The respondent could choose among five possibilities. The analysis was
carried out using the same maximum-likelihood function described above, excluding from the sample all individuals whose level of certainty was inferior to the 95%
level. The resulting average WTPs are listed in Table 4 together with the results
obtained from the OE data.
As can be observed in Table 4, the average WTP resulting from the analysis of a
sub-sample of the DC respondents tends to get closer to the average WTP resulting
from the OE data. It is interesting to note that the maximum WTP estimates for a
seal-watching boat trip, deriving from Scenario II (derived from the model estimated
on the Seal Sanctuary sample), are much higher. This lower value is consistent with
the general dissatisfaction about the boat trips in St. Ives.
Data from the OE question were analysed using a linear OLS regression; resulting
values of coefficients and models summary are shown in Tables 5 and 6.
In both samples the statistically significant coefficient, IBID – which represents
the bid stated in the preceding DC question – shows a positive correlation with the
maximum WTP. The anchoring to the bid level from the DC question makes the
comparison of the two elicitation methods less meaningful.
As mentioned, data were collected from a self-selected sample who had already
paid to get the good and whose base WTP was greater than zero. For this reason it
was not possible to register WTPs smaller than £4 (i.e., the minimum entry fee, for
both sites, plus zero consumer surplus). The density of a truncated variable involves
‘‘ . . . scaling the density so that it integrates to one over the range above a’’ (Greene
1997), where a is the truncation point. Analytically:

2373
Table 6. Model summary and coefficients for OE Model, seal-watching sample.

fsxd
fsxux.ad5]]]
Prsx.ad

(5)

If x is a normally distributed variable such that x| N(m, s), then
fsxd
fsxux.ad5]]]]
a2m
12F ]]
s

S

D

(6)

The graph in Figure 1 shows the density of a normal distribution, with mean and
variance determined using the previous model results, and the truncated normal
distribution.
The variable is truncated below; therefore the mean of the truncated variable will
be higher than the original mean, as expressed analytically by

Figure 1. Normal distribution and truncated distribution (a 5 4).

2374
Table 7. Results from Scenario I, seal watching.
Sample mean of predicted WTP (£); location: seal watching

Mean truncated model
Truncated Mean

OE

DC, more than 95% certainty

7.36
7.74

5.08
5.76

S D
S D

a2m
f ]]
s
E f WTPuWTP.a g 5x j b1s]]]]
a2m
12f ]]
s

(7)

In the case of the OE data, considering a truncated normal model, when the
truncation point is 4, the contribution to the log-likelihood function for each
observation i is given by:

S

D

maxWTP2x j b
1 / sf ]]]]
s
lnL5ln ]]]]]]]
42x j b
12F ]]
s

5

S

D

6

(8)

where maxWTP is the value stated by the respondent.
In the case of the DC data, the contribution of one observation to the sample
log-likelihood function is

53

lnL5ln IY

S
S

D
D

S

D S D
S D

bid2x j b
bid2x j b
42x j b
12F ]]]
F ]]] 2F ]]
s
s
s
]]]]]
1IN ]]]]]]]]
42x j b
42x j b
12F ]]
12F ]]
s
s

4 3

46
(9)

We can also consider the marginal effect of each explanatory variable (i.e., how
the value of the truncated mean varies when varying the value of the regressors). As
noted in Greene (1997), ‘‘whether the marginal effects or the coefficient b itself is of
interest depends on the intended inferences of the study. If the analysis is to be
confined to the sub-population . . . the marginal effect is of interest. However, if the
study is intended to extend to the entire population, it is the coefficients b that are
actually of interest’’. Table 7 shows the mean resulting from the truncated model
(extended to the entire population) and the truncated mean, concerning the subpopulation (parameter values are listed in Appendix 2).
Thus, the average maximum WTP to get on a seal-watching boat trip is, for the
sub-population, £7.74 according to the OE results and £5.76 according to the DC
results. The maximum WTP estimates for a seal-watching boat trip deriving from
Scenario II (when respondents in the Seal Sanctuary are asked) are much higher.
Again this probably reflects the high level of dissatisfaction with the quality of the
actual boat trip.

2375
Scenario II
Scenario II was designed to consider a different economic issue. How do respondents value the alternative non-consumptive use of seals? Does the bid function
depend on the hypothetical price of the option they are actually choosing? Additionally, the information elicited from this set of questions is used to construct an
inverse demand function for both the Seal Sanctuary and the seal-watching boat trip.
This might help managers of these economic activities to understand which portion
of the surplus they might be able to capture by raising the entry fees. Figures 2 and 3
show a visual representation of the data collected for Scenario II. Survivor functions
represent the percentage of respondents who are willing to pay at least that amount.
The elicitation format in this scenario is a hybrid form between the choice
experimental approach and closed-ended double-bounded DC method. We have
analysed the data by using a double-bounded normal model. The rationale of the
model follows from the previous single-bounded normal model. The difference is
that, now, each respondent is answering two questions and this will give more
information about the probability distribution characteristics.
In the valuation question, respondents of the Seal Sanctuary sample were asked

Figure 2. Survivor function for seal watching.

Figure 3. Survivor function for Seal Sanctuary.

2376
Table 8. Results from Scenario II.

Mean WTP (£)

Location: seal-watching WTP
relative to Seal Sanctuary

Location: Seal Sanctuary WTP
relative to seal watching

6.75

12.49

whether they would have paid BIDX to enter the Seal Sanctuary (respondents within
the seal-watching sample were asked the same question relatively to seal watching)
and BIDY to have the seal-watching trip (the question concerns the Seal Sanctuary,
for respondents in the seal-watching sample). In the follow-up question the bid for
the alternative option is increased (upBIDY, when the answer to the previous
question is YES) or decreased (downBIDY, when the answer to the previous
question is NO), see Appendix 1.
The explanatory variables in the bid function are the same as in previous cases,
except for the variable BIDX. The coefficient of the latter variable is positive and
statistically significant at the 1% level, but just in the regression related to the first
site. This is probably due to the fact that introducing a high price for one of the two
alternatives shifts the entire perspective of the respondent. The resulting marginal
effect is positive.
The contribution of one observation to the sample log-likelihood function is now:

H F S
FS
FS
FS

DG
D S
D S
DGJ

upBIDY2x j b
lnL5ln IYY 12F ]]]]
s
upBIDY2x j b
BIDY2x j b
1IYN F ]]]] 2F ]]]
s
s
BIDY2x j b
downBIDY2x j b
1INY F ]]] 2F ]]]]]
s
s
downBIDY2x j b
1INN F ]]]]]
s

DG
DG
(10)

where IYY /IYN /INY /INN are dummy variables that take value 1 if respondents answer
yes to the first and the second (higher) bid / yes to the first and no to the second bid /
no to the first and yes to the second (lower) bid / no to both bids; BIDY is the first
question bid value; upBIDY is the upper bid value; downBIDY is the lower bid
value; the unobserved variable is supposed to be normally distributed BIDY|
N(xb,s).
Resulting mean WTP values (calculated using a similar procedure as in the
previous models) are shown in Table 8; see Appendix 2 for regression results
details. Individuals interviewed in the Seal Sanctuary, when reminded of the
possibility of watching seals in their natural environment, show a higher WTP than
for the Seal Sanctuary itself.
The probability distribution of the WTP for the seal-watching boat trip, obtained
using the results from the regression, is shown in Figure 4. It is now possible to
obtain the direct demand curve. In order to design the best pricing scheme, it will be

2377

Figure 4. The demand curve for seal watching (Seal Sanctuary).

necessary to incorporate information about the cost of the activity, as well as the
carrying capacity.
Scenario III
In Scenario III the question is exactly the same for both sites, and the object
evaluated is the non-use value of seals. The form of payment is an optional, per
person, conservation fee, that would be paid on top of the entrance fee. Respondents
were shown a payment ladder and therefore the data collected are an upper and a
lower bound within which the unobservable WTP lies. The first approach to the data
is a simple visual analysis. It is useful to consider the survivor functions either for
crosses or for ticks (see Figure 5). Data from the two samples were pooled together
once the statistical significance of the variable defining the locus of the interviews
had been tested and rejected.
Table 9 reports the values for the mean, the standard deviation, the minimum and
the maximum WTP for ticks and crosses. It could be argued that using the average
tick value is a sensible conservative solution. However, a model can be designed to
estimate the average unobservable WTP and to see whether there is a relationship
with any covariate information. For example, it might be of interest to consider
whether the non-use value of seals is related to the location of the interview or not
and, therefore, whether to pool together data from the two questionnaires in a unique
sample. The problem is that there is a possibility of correlation between the income
variable and the location. As before, we consider a linear normal specification for
the model. Hence, the contribution of one observation to the log-likelihood function
is:

HS

D S

cross2x j b
tick2x j b
lnL5ln F ]]] 2F ]]]
s
s

DJ

(11)

2378

Figure 5. Survivor functions for crosses and ticks in Scenario III.
Table 9. WTP conservation fund.

Seal Sanctuary
Ticks
Crosses
Seal watching
Ticks
Crosses

N

Mean (£)

Std. dev.

Min (£)

Max (£)

112
112

2.71
5.24

5.236
7.426

0
0

33
37

94
94

2.53
6.56

4.918
12.379

0
0

33
100

Table 10. ‘Zero-responses’ reasons, card shown to respondents.
‘No Response’ reasons

Tick

1. Seal conservation is important but there are far more worthy causes to which I would prefer to
donate money
2. I cannot afford to pay anything but would do so otherwise
3. I don’t think that the problems caused by seals are severe enough to warrant paying money to
conserve them
4. I don’t believe that such a programme would be effective in helping conserve seals
5. I don’t believe that a programme of seal conservation should be funded through public
donations
6. Other (Please specify)

where F(x) is the cumulative of the normal distribution, x j are the jth explanatory
variables, and s is the standard deviation of the distribution that will be estimated by
the optimising routine together with the b parameters.
Initially, respondents are asked a referendum question. They might answer ‘no’
for different reasons, which can be usefully divided into two main subgroups. The
first is the group of effective zeros. The respondent might be unable to donate
because of his / her budget constraint or because there are more worthy causes. The
second group is defined as protest zeros: respondents are rejecting the entire
question and the underlying policy implications. In case of a zero answer the card in
Table 10 was shown to respondents.

2379
Respondents refusing to participate in the contingent market (answers 4 and 5)
were excluded from the sample. The protest zeros were 29% of the whole zeros
(31.4%) in the sample. A second issue relates to the remaining zero values which
produce a mathematical problem when working with logarithms. Several authors
have dealt with this issue by applying the Spike Model (Kristrom 1997; Reiser and
Shechter 1998a; Scarpa et al. 2001). The idea is to deal separately with the
probability of saying yes / no, on one side, and the probability of stating a value in
the positive case, on the other. The sample is divided into two sub-samples. One is
not willing to pay anything, while the second is willing to pay something and the
WTP is normally distributed.
The contribution of one observation to the log-likelihood function is

FS

D S

cross2x j b
tick2x j b
lnL5dlnp1s12ddlns12pd1s12ddln F ]]] 2F ]]]
s
s

DG
(12)

where d 5 0(1) if respondents answer yes (no). We assume that the probability of
saying no is the number of valid zeros over the total number of valid responses:

O

di
p5]]50.2880435¯28.8%
n

(13)

The value for the mean becomes:
EsWTPd5s12pdxb

(14)

The graph in Figure 6 shows the survivor function (i.e., 1 2 F(z)), estimated
using the described model, together with survivor functions for crosses and ticks. We
have tested for the statistical significance of a dummy variable defining the location
of the interview; however, as can be seen in detail in Appendix 2, the small size of
the sample makes the results from the regression scarcely reliable. Indeed, the
regression results are not statistically significant for all the coefficients except for
PARTY, which has a negative value and is significant at the 2% level of confidence.
As before, results have been used to calculate the average WTP deriving from the
estimated model, see Table 11. However, the more reliable and conservative data
that should be used to draw some conclusions is the average value of ticks.
Of interest is the value that could actually be captured in order to compensate the
fishing industry. One feasible way to obtain it is to estimate the number of paying
visitors per year, which is around 200 000. Hence, to get an extremely conservative
estimate of the actual monetary amount that might be collected we use the average
value of ticks (£2.63) and consider a number of paying visitors equal to half of the
actual annual visitors, obtaining an annual amount of £526 000.
Several results can be discussed. First, the hypothesis of weighting data coming
from referendum surveys, to take into account the level of respondents’ certainty,
has proved to be of limited value when both formats are used in the same
questionnaire, because of the anchoring effect. However, better results may be

2380

Figure 6. Different survivor functions relative to cross and tick values and to the estimated normal model.
Table 11. Possible values to be used in order to evaluate ‘non-use’ value of seals.
Mean (£)
Ticks
Crosses
Average WTP calculated on the entire sample

2.63
5.85
4.65

obtained by conducting two distinct questionnaires, each using one of the two
formats. This would require a large sample. Second, given the truncated nature of
the sample investigated, a truncated distribution model has been used to extend the
study to the entire population. The calculated average WTP is significantly lower.
However, the sample is probably too small to extend results to the entire population.
Third, results from the second scenario have shown that people in the Seal
Sanctuary are willing to pay more to see seals in their natural environment rather
than in the Sanctuary itself. Moreover, the average WTP calculated is higher than
the one obtained from the seal-watching sub-population. This could partially be a
reflection of the high level of disappointment recorded among respondents arriving
from boat trips in St. Ives. Finally, we have followed the more conservative
estimation technique and we have aggregated over the annual Seal Sanctuary
visitors the mean value of ticks to obtain a gross WTP for ‘non-use’ value of seals of
around £526 000 per year. This value can be compared, for policy purposes, to the
fishermen’s annual losses due to seals (roughly £100 000). Seals are a high value
‘flagship’ species, as the results of the study confirm. Thus, only a fraction of the
total economic value is needed to compensate fishermen for the losses they suffer.
Put another way, the benefits of a compensation scheme outweigh the costs. This
result can be contrasted with the various ‘technological’ options, including sealculling, considered so far as potential policies, and with the risk that fishermen will
‘go it alone’ by destroying seals, regardless of the law.

2381
Conclusions
The conflict between fishermen and seals is, at the present, producing an inefficient
outcome. This study shows that there is room for a more efficient reallocation of
resources. The importance of developing definitive strategies is aggravated by the
economic vulnerability of the Cornish fishing industry. Moreover, even if we are
extremely unsure of the dynamics of this seals settlement, it is clearly of key
importance in the migration patterns of the European population. Decision-makers
should consider different policies.
The first goal is to find a way to capture part of the economic value of seals and
subsequently invest in conservation. As was shown in the present study, this value is
significantly higher than the losses suffered by the fishing industry, even when
considering just the non-use portion of the total economic value. One way of
capturing this could be an optional conservation fee on top of the fees for
recreational use of seals. Fishing techniques, which are affected by and affect seals,
could be restricted in areas where seals tend to gather to mate or reproduce.
Fishermen would then be compensated for the losses. In order to implement this
strategy, the biological dynamics of the population and the seals’ territorial distribution should be studied in-depth. In specific cases, when the compensation costs
are extremely high and the damage is related to isolated rogue seals, then direct
elimination might be a more rational solution.
A second issue of importance is the role of the Seal Sanctuary. Even if effects on
the seal population are still not quite clear, it would be advisable to introduce norms
in order to regulate: the length of the captivity period for each seal pup, the
maximum number of rescued pups per year, and the location of releases.
Finally, the study has revealed a demand for seal watching that is not consistent
with the perceptions of people organising this activity. The tourists’ WTP elicited in
the present study is higher than expected. Indeed, it would be very important to
invest in improved facilities, thus capturing the proportion of consumers that have
very young children.

Acknowledgements
We are very much indebted to Brett Day and Wolf Krug of CSERGE, University
College London for advice on the original study; and to Ian Bateman of the
University of East Anglia and Ken Willis of the University of Newcastle for
comments on a penultimate draft. Special thanks go to Stephen Wescott for
invaluable help with the survey and advice on the ecology of seal populations. We
wish to thank English Nature for financing the field study costs of V.B. Finally, this
study would not have been possible without the co-operation of the Seal Sanctuary
of Gweek, the CFPO (Cornish Fish Producers Organisation) and the seal-watching
boat trip organisations in Dartmouth and St. Ives. All errors of fact and interpretation are solely our own.

2382
Appendix 1

2383

2384

2385

2386

2387

2388

2389

2390
Appendix 2

2391

2392
References
Bateman I., Langford I., Jones A. and Kerry G. 2001. Bound and path effects in double and triple
bounded dichotomous choice contingent valuation. Resource and Energy Economics 497: 1–23.
Berry C. 1996. Marine Resource Utilisation. North Sea Monitor, UK Vol. 14, pp. 6–9.
Bonner G. 1989. The Natural History of Seals. Croom Helm, London.
Cameron T.A. 1988. A new paradigm for valuing non-market goods using referendum data: maximum
likelihood estimation by censored logistic regression. Journal of Environmental Economics and
Management 15: 355–379.
Cameron T.A. and James D. 1987. Efficient estimation methods for closed ended contingent valuation
surveys. The Review of Economics and Statistics 69: 269–276.
English Nature 1996. Marine Special Areas of Conservation. English Nature, Peterborough, UK.
Glen D. 1998. Grey Seal Conservation, M.Sc. Thesis, University College London.
Greene W.H. 1997. Econometric Analysis. Prentice Hall, Upper Saddle River, New Jersey.
Kristrom B. 1997. Spike models in contingent valuation. American Journal of Agricultural Economics 3:
1013–1023.
Langford I., Kontogianni A., Skourtos M., Georgiou S. and Bateman I. 1998a. Multi-variate mixed
models for open-ended contingent valuation data: willingness to pay for conservation of Monk seals.
Environmental and Resource Economics 12: 443–456.
Langford I., Skourtos M., Kontogianni A., Day R., Georgiou S. and Bateman I. 1998b. Use and non-use
values for conserving endangered species – the case of the Mediterranean Monk seal, unpublished
paper. CSERGE, University of East Anglia, Norwich, UK.
NOAA Panels 1992. Federal Register. 58(10): pp. 4601–4614.
PESCA 1996. Delivery Plan: South West England, 1996–1999.
Ready R., Navrud S. and Dubourg R. 1998. How do respondents with uncertain willingness to pay
answer contingent valuation questions? Benefits Transfer and the Economic Valuation of Environmental Damage in the European Union. A Report to Directorate General XII of the European
Commission, Brussels, Belgium.
Reiser B. and Shechter M. 1998a. Incorporating zero values in the economic evaluation of environmental
program benefits. Environmetrics 10: 87–101.
Scarpa R. and Bateman I. 2000. Efficiency gains afforded by improved bid design versus follow-up
valuation questions in discrete choice CV studies. Land Economics 76: 299–311.
Scarpa R., Willis K.G. and Garrod 2001. Estimating WTP for speed reduction from dichotomous-choice
CV responses with follow-up: the case of rural trunk roads. Environmental and Resource Economics
20: 281–304.
Wescott S. 1996. Cornwall Biodiversity Initiative. Species Action Plan: Grey Seal. (unpublished).