Content uploaded by Richard S.J. Tol
Author content
All content in this area was uploaded by Richard S.J. Tol
Content may be subject to copyright.
CLIMATE RESEARCH
Clim Res
Vol. 29: 245–254, 2005 Published October 17
1. INTRODUCTION
Climate is one of the main drivers of international
tourism, as the majority of tourists seek to relax in the
sun or the snow (Aguiló et al. 2005). We used the defi-
nition in WTO (2003), according to which a tourist
spends at least 1 but not more than 364 nights away
from home; an international tourist crosses an interna-
tional border. This study does not consider domestic
tourism, and for simplicity we refer to ‘international
tourism’ simply as ‘tourism’. The data exclude business
travel, and groups together pleasure travel with travel
for family visits, pilgrimage, education, and health.
Climate plays only a minor role in the tourism litera-
ture (e.g. Witt & Witt 1995), perhaps because its impor-
tance is so self-evident, or perhaps because climate is
believed to be constant or beyond control. Climate,
however, is changing and likely to continue to change
for decades to come. Climate change would substan-
tially affect tourism, something that is largely ignored
in the literature on climate change impacts (e.g. Smith
et al. 2001, Scott et al. 2004), even though tourism is
now the largest industry in the world and is still grow-
ing quickly. Tourism has a strong impact on the envi-
ronment (Goessling 2002), and a substantial impact on
climate, not just through the emissions of CO2(particu-
larly from air travel), but also through the direct impact
of flying (e.g. contrails).
One can approach the relationship between tourism
and climate (change) in 2 different ways, by looking:
(1) at tourists, what they (should) prefer (e.g. Scott &
McBoyle 2001, Amelung & Viner in press) or how they
behave (e.g. Maddison 2001, Richardson & Loomis
2004); or (2) at destinations and how their attractive-
ness changes with climate and management (e.g.
Abegg 1996, Craig-Smith & Ruhanen 2005, Perry
2005). Reviews of the climate and tourism literature are
given by Hamilton & Tol (2004) and Scott et al. (2005).
© Inter-Research 2005 · www.int-res.com*Corresponding author. Email: tol@dkrz.de
Effects of climate change on international tourism
Jacqueline M. Hamilton1, David J. Maddison2, Richard S. J. Tol1, 3, 4,*
1Research Unit Sustainability and Global Change, Hamburg University and Centre for Marine and Atmospheric Science,
Bundesstrasse 55, 20146 Hamburg, Germany
2Department of Economics, University College, Gower Street, London WC1E GBT, UK
3Institute for Environmental Studies, Vrije Universiteit, De Boelelaan 1081, Amsterdam, The Netherlands
4Department of Engineering and Public Policy, Carnegie Mellon University, Baker Hall 129, Pittsburgh, Pennsylvania 15213, USA
ABSTRACT: We present a simulation model of the flow of tourists between 207 countries, used to
study the impact of climate change on international tourism. The model almost perfectly reproduces
the calibration year 1995, and performs well in reproducing the observations for 1980, 1985 and 1990.
The model is used to generate scenarios of international tourist departures and arrivals for the period
2000–2075, with particular emphasis on climate change; we report variations on a single baseline
scenario (A1B). The growth rate of international tourism is projected to increase over the coming
decades, but may slow down later in the century, as demand for travel saturates. Emissions of carbon
dioxide would increase quickly as well. With climate change, preferred destinations would shift to
higher latitudes and altitudes. Tourists from temperate climates would spend more holidays in their
home countries. As such tourists currently dominate the international tourism market, climate change
would decrease worldwide tourism. Nevertheless, its effects are small compared to the baseline
projections of population and economic growth.
KEY WORDS: International tourism · Climate change impacts · CO2emissions · Scenarios
Resale or republication not permitted without written consent of the publisher
Clim Res 29: 245–254, 2005
However, looking at a particular group of tourists or a
particular destination is not enough to fully understand
tourist behaviour. Tourism, like any market, is defined
by supply and demand, by push and pull factors. Des-
tinations compete for the most lucrative tourists, and
tourists compete for the best deals. A necessary com-
ponent in the study of tourism is a comprehensive
model of tourists and destinations. For international
tourism, this implies a global model.
We built on a recent study by Hamilton et al. (2005)
where the first version of the Hamburg Tourism Model
(HTM) was presented: an econometric simulation
model of the travel patterns of tourists from 207 coun-
tries, enjoying their holidays in one of the other 206
countries. The model finds that climate change affects
international tourism, but that this effect is small com-
pared to the other changes in the industry, such as
those due to population growth and change in per
capita income. Hamilton et al. (2005) found that loca-
tions that are cool at present would attract more
tourists under global warming, and that currently
warm places would attract fewer tourists. Changes in
the strength of the pull effect of international destina-
tions, however, are tempered by changes in the
strength of the push effect of the home country.
Tourists from places that are temperate (warm) at pre-
sent would increasingly spend their holiday in their
home country, not abroad. As tourists from the temper-
ate countries of West Europe dominate the market,
total international tourism numbers would fall (relative
to a rapidly rising baseline without climate change),
and so would total distance travelled.
In this study, we extended the HTM by: (1) taking a
closer look at its ability to predict observations that
were not used in the calibration, which leads to a few
adjustments in the parameterisation; (2) using higher
income elasticities, but allowing demand to saturate;
(3) estimating emissions of CO2from international
tourism for various scenarios; (4) studying the implica-
tions of changes in international tourism for domestic
tourism.
Section 2 presents the model, its calibration and its
validation. The model and all necessary input data can
be downloaded at www.uni-hamburg.de/Wiss/FB/15/
Sustainability/HTM11.zip. Section 3 presents analyses
based on scenarios of climate, population and income.
Sensitivity analyses are presented in Section 4. Section
5 summarizes the conclusions of this study.
2. THE MODEL
We used HTM version 1.1, which models tourist
flows between 207 countries. The purpose of the HTM
is not to understand the current pattern of interna-
tional tourism—for that, we would need a more
detailed description of tourist behavior than is avail-
able to us—but rather to analyse how the current pat-
tern may change under not implausible scenarios of
future population growth, economic growth and, par-
ticularly, climate change. The inputs to the patterns
and their changes are the empirical regularities
reported in Hamilton et al. (2005). The details are
given below, along with calibration and validation pro-
cedures and results.
The basis of the model is the matrix of bilateral
tourism flows. This matrix is perturbed with scenarios
of population growth, economic growth and climate
change. The perturbations on the supply side are
perturbations on the ‘relative attractiveness’ of holiday
destinations. The perturbations on the demand side
are perturbations on the ‘number of tourists’ from ori-
gin countries. For these perturbations, we used the
same relationships as we used to construct the bilateral
tourism flow matrix.
The model was calibrated against the international
arrivals and departures data of 1995 contained in the
World Resources Databases (WRI 2000); the data were
corrected for typographic errors in the original data-
base. The WRI data are derived from World Tourism
Organisation data (e.g. WTO 2003), but presented in a
handier format. There are 3 major problems with this
dataset. (1) For some countries, the reported data are
arrivals and departures for tourism only; for other
countries, the data are arrivals and departures for all
purposes. It is impossible to correct for this, except for
data on Poland and the Czech Republic, which would
otherwise be outliers (e.g. in Poland only 12% of
departures are tourists; see Central Statistical Office
Poland [2002], available at www.stat.gov.pl/english/
serwis/polska/rocznik11/turyst.htm). (2) The data are
total arrivals and total departures; there are no data on
the origin of the arrivals or the destination of the
departures. We therefore constructed a database on
bilateral tourism flows for all pairs of countries. (3)
There are missing observations, particularly with
regard to departures.
For arrivals, we filled the missing observations with a
statistical model:
(1)
where Adenotes the number of international arrivals
at country i, Area is the country’s surface area, Tis the
annual average temperature, Coast is the length of the
coastline, and Yis the per capita income in country i.
The numbers in the main line of the equation are the
estimated parameters, the subscripted numbers are the
ln . . .
.. .
AT
iii
=+× +
−
5 97 2 05 10 0 22
097 096
7
007
Area −−×
+× +
−
−
791 10
715 10 080
221
32
303
5
0
.
..
.
.
Ti
Coast .. ln
;.
09
2
139 0 54
Y
NR
i
==
adj
246
Hamilton et al.: Climate change and tourism
standard deviation. This model was the best fit to the
observations. The data on per capita income were
taken from WRI (2000), supplemented with data from
CIA (2002); the data on area and the length of interna-
tional borders are from CIA (2002); the data on temper-
ature are from New et al. (1999). All data are available
at www.uni-hamburg.de/Wiss/FB/15/Sustainability.
After filling in missing observations, the total number
of tourists increased from 55.2 million to 56.5 million.
For departures, we filled missing observations with a
statistical model:
(2)
where Ddenotes the number of international depar-
tures from country i, Pop is population, and Border is
the number of land borders. Data on the number of
land borders were taken from CIA (2002). This model
provided the best fit, but although the fit is better than
that for arrivals, the uncertainty about the para-
meters is greater. The total number of departures was
48.2 million; therefore we scaled up all departures, so
that the total number of observed and modelled depar-
tures equalled the total number of observed and
modelled arrivals; scaling up only the interpolated
departures led to distortions, as 99 of the 207 countries
do not report departures data.
Bilateral tourism flows were derived as follows. In
keeping with the model described below, a ‘general
attractiveness index’ (GAI) for each country was con-
structed. In a first iteration, the GAI equalled the mar-
ket share of each country in world tourism. The ratio
between predicted and observed tourist arrivals was
used to adjust the GAI. The tourists of each country
were allocated to other countries according to an index
that is proportional to GAI ×distance between the 2
capital cities raised to the power –1.7 ×10–4, so that the
model reproduces the 1995 pattern of total departures
and arrivals (see Fig. 1). The model is well calibrated,
at least in arrivals, as Libya is the only country for
which model results and observations deviate graphi-
cally; the model predicts that Libya should have a mar-
ket share of 0.01 to 0.05, i.e. it should be more popular
than it is in reality.
There is only weak empirical evidence that tourists
are attracted to places with low or high population
densities. Population density is not significant in the
variants of Eq. (1) that we estimated; Maddison (2001)
reports that British tourists avoid densely populated
countries, but Lise & Tol (2002) and Hamilton (2003)
found no significant relationship for Dutch and Ger-
man tourists. Population growth is therefore assumed
to affect international tourism as a proportional in-
crease in departures. As population growth is not uni-
form over the globe and travel is partly determined by
distance, this simple assumption already creates a shift
in the pattern of international tourism (Hamilton et al.
2005).
Economic growth is assumed to affect tourism
according to Eqs. (1) & (2). That is, a country becomes
more attractive as it grows richer, with an elasticity of
0.80. A country generates more tourists as it becomes
richer, with an elasticity of 0.86. The population and
economic scenarios together produce a marked shift of
international tourism towards Asia (Hamilton et al.
2005).
The number of international trips per person is
capped at 4 per year, or 1 per season; in 1995, the ob-
served annual maximum was in Bermuda (1.57), Aus-
tria (1.55) and Switzerland (1.47); all 3 countries are
ln . . .
.. .
DT
i
i
i
Pop =− +×
−
151 018 483 10
17 05 0 17 16 82
332
422
2
009
556 10
086 023
T
Y
i
i
−×
+−
−
.
.ln .
.
.
Border
0013
2
99 0 66
.ln
;.
Area
adj
i
NR==
247
Fig. 1. Share of tourist arrivals per country observed in 1995
Clim Res 29: 245–254, 2005
small and wealthy. The choice of a maximum of 4 inter-
national holidays per year is ad hoc.
WRI (2001) presents tourism data for the period
1980–1998; 1995 is the calibration year, and the model
operates in time steps of 5 yr, so we used the model to
‘predict’ tourist numbers for the years 1980, 1985, and
1990. Since the model does not have differential equa-
tions, time can be readily inverted. There are many
missing observations in 1980, so calibrating the model
to 1980 and ‘predicting’ the following 20 yr was not an
option. Running the model back in
time requires data on population and
per capita income. Population data are
readily available from WRI (2000),
except for a few small or new coun-
tries; for these countries, we used the
population growth rates of nearby
countries and the growth rate of the
original country. Per capita income is
more problematic. WRI (2000) has
many missing observations, which we
filled with the national growth rates
from the World Bank (2003) website
(www.worldbank.org) ‘Countries at a
glance’; data for the former Yugo-
slavia were taken from the Penn
World Tables, and for North Korea
from www.inform.umd.edu/econddata/
WorkPaper/INFORUM/wp97008.pdf.
When model outcomes were com-
pared to past observations, it became
apparent that the 1995 cross-section
income elasticity of international
tourism demand (0.86) is too low. The
best fit to the observations of 1980,
1985 and 1990 is an income elasticity
of 2.57. Crouch (1995) reported the
results of a meta-analysis of tourism
demand. He found an income elastic-
ity of 1.86, with a standard deviation
of 1.78, encompassing both 0.86 and
2.57. The adjustment of the income
elasticity was the only adjustment
made to the model. Figs. 2 & 3 com-
pare modelled arrivals and departures
to observations. The model repro-
duces the arrival observations for
1995, the year of calibration, with R2=
0.9995. The model ‘predicts’ arrivals
in the other years with R2≥0.77. There
are mismatches (see below), but the
model reproduces the overall global
pattern. For arrivals, most discrepan-
cies between data and model are in
Africa and West Asia, where data
availability and quality are low. Exceptions are Finland
and Japan, which are more popular in reality than in
the model, and Australia, which is in fact less popular
than the model predicts. For departures, data and
model also often disagree for Africa. The Belgians and
the Irish travel more than the model predicts, while
people from the Nordic countries travel less.
Departures (Fig. 3) are reproduced for 1995 with R2=
0.97, and 1990 and 1985 are replicated with R2≥0.85.
For 1980, deviations between model and observations
248
10
Observed
Modelled Modelled
1980, R = 0.80
2
1985, R = 0.77
2
1990, R = 0.88
2
1995, R = 1.00
2
8
106
104
10
Observed
8
106
104
104106108104106108
1995, R = 0.97
21990, R = 0.88
2
1980, R = 0.47
2
1985, R = 0.85
2
10
Observed
8
106
104
10
Observed
8
106
104
Modelled Modelled
104106108104106108
Fig. 3. Observed versus modelled number of tourist departures in 207 countries
for 1980, 1985, 1990, and 1995
Fig. 2. Observed versus modelled number of tourist arrivals in 207 countries for
1980, 1985, 1990, and 1995
Hamilton et al.: Climate change and tourism
are larger: R2= 0.47. This is partly due to imperfections
in the model, but also because the income data for
1980 are sparse; some countries reported tourist depar-
tures, but not income. Moreover, only about half of the
modeled departures could be compared to observa-
tions. Overall, version 1.1 of the HTM has a reasonable
performance in reproducing observations over the
1980–1990 period.
3. SCENARIO ANALYSIS
Scenarios of population and economic growth are
taken from the 17-region IMAGE 2.2 implementation
(IMAGE Team 2001) of the SRES scenarios (Nakicen-
ovic & Swart 2001). We used the baseline scenario
A1B; Hamilton et al. (2005) show results for all 6 SRES
scenarios. Scenario A1B assumes moderate population
growth and rapid economic growth. More people
implies more tourists (Eq. 2). A higher per capita
income implies that people travel more (Eq. 2), and
that a country becomes relatively more attractive
(Eq. 1).
Climate change scenarios are taken from country-
specific output of the COSMIC model (Schlesinger &
Williams 1998). We used the average of the 14 general
circulation models (GCMs) in COSMIC as our middle
scenario. The effect of climate change follows from
Eqs. (1) & (2), both of which have a quadratic specifica-
tion. That is, if a cool country becomes warmer, it first
attracts more international tourists, until it gets too
warm and starts attracting fewer tourists. The turning
point lies around 14°C (annual 24 h average). Simi-
larly, if a cool country becomes warmer, fewer domes-
tic tourists spend their holidays abroad until it becomes
too warm and more tourists start travelling abroad. The
turning point lies at about 18°C (annual 24 h average).
Fig. 4 compares the growth rates of international
tourism for 3 scenarios. In the first scenario, tourism
demand does not saturate; the growth rate of numbers
of international tourists rises to more than 14% yr–1 in
2025, and then gradually falls as population and eco-
nomic growth slow; the average number of pleasure
trips in foreign countries reaches about 100 per person
yr–1 in 2075. As these results are hard to imagine, the
second scenario assumes that demand for foreign
travel does saturate, at 4 trips yr–1. The number of
international tourists rises rapidly in this scenario as
well, but not as fast as without saturation; the growth
rate is around 10% yr–1 between 2015 and 2035; these
additional tourists are primarily from Asia. After 2035,
the market saturates, and growth falls rapidly. In the
third scenario, we add climate change. As is shown in
Hamilton et al. (2005), climate change perturbs the
socio-economic scenario, but does not dominate it.
Until 2020, climate change slows the growth of inter-
national tourism, as tourists from temperate and cool
countries, particularly in Europe, stay within their own
country. After 2020, more tourists originate in hot
countries, and tourism numbers go up as they seek to
spend their holiday at cooler destinations.
Fig. 5 shows the change in departures and arrivals in
2025 for the arbitrary climate change scenario of 1°C
global warming; this scenario is towards the upper end
of the IPCC range (Houghton et al. 2001); it is used for
illustrative purposes only. As expected, climate change
would lead to a poleward shift of tourism. Countries
closer to the poles would become more attractive for
tourists. At the same time, those countries would gen-
erate less international tourists; these countries would
become more attractive to their own citizens as well as
to foreigners. Fig. 5 also shows that there will be a shift
from lowland to highland tourism; the tourism sectors
in Zambia and Zimbabwe, for instance, could benefit
greatly from climate change (if other prob-
lems are solved).
Fig. 6 plots the change in departures and
arrivals due to climate change for 2050, as a
function of the initial annual mean temper-
ature. Fig. 6 confirms the messages of
Fig. 5. In 2050, climate change would in-
crease the number of arrivals by up to 40%,
or reduce arrivals by up to 30%. Departures
would increase by up to 10% or decrease
by up to 25%. For most countries, the
changes are much smaller. The impact of
climate change should be seen in the con-
text of the baseline. Tourism numbers are
growing by several percent per year, which
is much larger than the effect of climate
change. Climate change does not reverse
the trend in tourism in any of our projec-
249
-2
0
2
4
6
8
10
12
14
16
2000 2010 2020 2030 2040 2050 2060 2070
Growth rate (% yr )
no saturation, no climate change
saturation, no climate change
saturation, climate change
–1
Fig. 4. Growth rate of international tourism according to 3 scenarios
Clim Res 29: 245–254, 2005
tions; climate change makes tourism grow more or less
faster than it would have otherwise. Some of the
largest declines (from the baseline) are seen in the
Middle East, where nascent leisure resorts may not
fare well; pilgrimage, however, is probably robust to
climate change.
The distances between the capital cities of the coun-
tries of origin and destination are used in the estima-
tion of bilateral tourism flows. The distance travelled
also determines energy use and CO2emissions. We
used an emission coefficient of 99 g CO2per passenger
km for 2000 (Goessling et al.in press); the coefficient is
assumed to fall at 1% yr–1 because of progress in
energy efficiency. Fig. 7 shows the CO2emitted by
international tourists without climate change and the
difference induced by climate change, for Scenario
A1B. Without climate change, CO2emissions increase
rapidly at first, in fact even more rapidly than the num-
ber of tourists, but stabilises later as the tourism market
saturates. At present, international air travel accounts
for some 2.4% of global CO2emissions (Metz et al.
2001); even with an optimistic rate of technical
progress of 1%, this share increases to over 30% in
2050 in Scenario A1B. If we assume that the demand
for tourism saturates at 2 (rather than 4) international
trips yr–1, then the share of emissions arising from
transport to and from tourism destinations in the total
CO2emissions is capped at 20% (the emissions scenar-
ios were built without explicit attention to international
tourism). With climate change, the upward trend is
slightly slower — again largely because the frequent
travellers from Northwest Europe stay closer to home.
The effect of climate change is in the order of 1%. This
effect is similar for the high and low saturation scenar-
ios. Towards 2075, when demand is fully saturated, the
impact of climate change beomes zero.
250
Fig. 5. Change in (a) departures and (b) arrivals as a result of a 1°C global warming in 2025
a
b
Hamilton et al.: Climate change and tourism
4. SENSITIVITY ANALYSES
The model and the results presented
above depend on a number of para-
meters, each of which is uncertain. In
Hamilton et al.(2005), we report a
sensitivity analysis on the distance
parameter, simulating a scenario in
which travel would become cheaper
over time. This greatly affects travel
patterns in Scenario A1B, but does not
much affect the impact of climate
change. Similarly, we show there that
variations in the income elasticity
have a large impact on Scenario A1B,
but much less so on the relative impact
of climate change.
Fig. 8 shows the effects of varying
climate change. In the base case, we
used the geographic pattern of tem-
perature change that is the average of
14 GCMs at 2 ×CO2(Schlesinger &
Williams 1998). As sensitivity analy-
ses, we used that average plus the
standard deviation (and minus half)
over the 14 GCMs. This roughly corre-
sponds to varying the climate sensitiv-
ity from 2.5 to 1.5°C and 4.5°C, respec-
tively. The results are as expected.
Slower climate change leads to lower
impacts of climate change, and faster
climate change to higher impacts.
Fig. 9 compares the relative impact
of climate change on arrivals and
departures for tourism demand satu-
rating at 2 and 4 trips yr–1. Although
these 2 scenarios considerably differ
in absolute numbers of international
tourists (cf. Fig. 7), the relative impact
of climate change is very similar. The
base case assumption of saturation at
251
0.50
0.40
0.30
0.20
0.10
0.00
0.10
0.20
0.30
-10 -5 0 5 10 15 20 25 30
-10 -5 0 5 10 15 20 25 30
0.15
0.10
0.05
0.00
-0.05
-0.10
-0.15
-0.20
-0.25
-0.30
Percent change in tourist arrivals
Percent change in tourist arrivals
Annual mean temperatureAnnual mean temperature
Fig. 6. Change in tourist (a) departures and (b) arrivals in 207 countries in 2050 (% of tourist numbers without climate change) as
a function of annual mean temperature
0
5000
10000
15000
20000
25000
30000
35000
1995 2000 2005 2010 2015 2020 2025 2030 2035 2040 2045 2050 2055 2060 2065 2070 2075
0
50
100
150
200
250
300
350
CO
2
(10
t)
Without climate change (left axis)
Impact of climate change (right axis)
Early saturation
Base saturation
Base saturation
Early saturation
6
CO
2
(10
t)
6
Fig. 7. Total CO2emitted by international tourism without climate change (left
axis) and the difference induced by climate change (right axis); results are for
Scenario A1B; in the base scenario, tourism demand saturates at 4 trips yr–1,
in the early saturation scenario, demand saturates at 2 trips yr–1
ab
-0.4
-0.3
-0.2
-0.1
0
0.1
0.2
0.3
0.4
-0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.
4
Base (% chan
g
e in arrivals due to climate chan
g
e)
Sensitivity (% change in arrivals due to climate change)
Low climate change
High climate change
Fig. 8. Climate induced change in tourist arrivals per country in 2050 (Scenario
A1B) for medium-level climate change versus high- and low-level climate change
Clim Res 29: 245–254, 2005
4 trips yr–1 may be ad hoc, but it does not qualitatively
influence the results.
The current version of the model is restricted to
international tourism. Domestic tourism is not included
because of the limited data availability. International
departures, however, are included, and responsive to
climate change and scenarios. We can therefore calcu-
late the number of people who would have travelled
abroad but did not. If we assume that the tourists not
travelling abroad add to domestic tourism, we get a
better idea of the real changes in tourism. Fig. 10 plots
the change in international arrivals as a function of the
initial annual mean temperature (as in Fig. 6) and adds
the number of tourists not travelling abroad. Almost
everywhere, the tourists not travelling abroad amplify
the change in tourist arrivals— substantially so in a
number of cases. Fig. 10 also shows the ratio of the
change in international tourist arrivals and the number
of tourists not travelling abroad, again as a function of
the initial temperature. In one-third of the cases, the
number of tourists not travelling abroad is greater than
the change in international tourist arrivals. In one-
tenth of the cases, the number of tourists not travelling
abroad has a different sign than the change in tourist
arrivals. This happens in the countries that have an ini-
tial temperature between 11 and 18°C, reflecting the
difference in optimal temperature for departures and
arrivals (see above).
5. DISCUSSION AND CONCLUSIONS
We have presented a simulation model of interna-
tional tourism, and developed scenarios of changes in
international arrivals and departures because of
changes in population numbers, per capita income,
and climate change. A model like this tests sensitivities
rather than making predictions. Although the model
does well in predicting out of sample, the validation
period is short compared to the ‘forecasting’ period.
252
-0.3
0.15
0.1
0.05
0
-0.05
-0.1
-0.15
-0.2
-0.25
-0.3
0.5
-0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4 0.5
0.4
0.3
0.2
0.1
0
-0.1
-0.2
-0.3
-0.25 -0.2 -0.15 -0.1 -0.05 0 0.05 0.1 0.15
High saturation (% change in departures due to climate change)
Low saturation (% change in departures due to
climate change)
High saturation (% change in arrivals due to climate change)
Low saturation (% change in arrivals due to
climate change)
Fig. 9. Relative change in the number of (a) departures and (b) arrivals in 207 countries in 2050 (Scenario A1B); base case
(demand saturates at 4 trips yr–1) compared to alternative (demand saturates at 2 trips yr–1)
-10
1
8
-10 -5 0 5 10 15 20 25 30
6
4
2
0
-2
-4
0.8
0.6
0.4
0.2
0
-0.2
-0.4
-0.6
-0.8
-1
-5 0 5 10 15 20 25 30
Percent change in tourists
Arrivals
Arrivals plus departures
Annual mean temperature
Annual mean temperature
Ratio of changes in arrivals to
changes in departures
Fig. 10. Change in tourism arrivals in 207 countries in 2050 as a function of annual mean temperature. (a) Arrivals and number of
tourists not travelling abroad in % of tourist numbers without climate change; (b) ratio of the change in arrivals to the number of
tourists not travelling abroad
b
a
b
a
Hamilton et al.: Climate change and tourism
The model shows that the past growth of interna-
tional tourism may well continue unabated in the
medium term, but will saturate in the long term. The
main driver is economic growth, and the growth of
international tourism will therefore be concentrated in
those regions with the highest economic growth. As
high-growth regions are relatively poor at present, sat-
uration will set in later there. Although intercontinen-
tal tourism will also grow, future increases in mass
tourism are likely to be concentrated close to the coun-
tries of origin, e.g. within the high-growth regions of
Asia.
Climate change would lead to a gradual shift of
tourist destinations towards the poles and up the
mountains. Climate change would also imply that the
currently dominant group of international tourists—
sun and beach lovers from Western Europe — would
travel less far, or even stay in their home country,
implying a fall of total international tourist numbers
(relative to the baseline without climate change). The
reverse is true for warmer countries; not only would
these countries attract less foreign tourists, domestic
tourists would be inclined to travel abroad for their hol-
idays. The pattern of international tourism — towards
higher latitudes and altitudes—found by Hamilton et
al. (2005) is amplified by shifts in domestic tourism:
Higher (lower) latitudes and altitudes would become
more (less) attractive to international and domestic
tourists alike.
Nonetheless, the changes induced by climate
change are generally much smaller than the changes
due to population and economic growth. Again in com-
parison with Hamilton et al.(2005), the saturation of
tourism demand introduced in this study makes the
baseline even more important relative to the impact of
climate change.
A reduction in international tourism implies a reduc-
tion in international travel and a reduction in green-
house gas emissions from international travel. This is a
negative feedback on the emission scenarios, but we
find that emissions from international tourist travel fall
by less than 1% because of climate change.
The model described in this study is, to our knowl-
edge, one of its kind. As for all early models, it has
limitations. Although the model is reasonably good at
reproducing current and past patterns of international
tourism, long-term and global studies of tourism
demand are rare — and the empirical basis of the
model is therefore weak. This is even truer for the
effects of climate change on tourist destination choice,
where the model is based on only a few studies from a
limited set of similar countries. Domestic tourism is
modelled as a residual of international tourism. The
projections neglect the fact that changes in prefer-
ences, age structure, working hours and lifestyles
would also affect tourist behaviour. Demand saturation
is included in an ad hoc manner. The spatial resolution
(national) of the model is crude, as is the temporal res-
olution (annual). A seasonal resolution would allow for
the separate analysis of sun and snow seekers, and
would allow tourists to shift their holidays not only in
space (as they do in the current model) but also in time
(from summer to spring and autumn). The model does
not extend beyond tourist numbers. Improving on all
this is deferred to future research. The results pre-
sented here demonstrate that this is a fruitful line of
research.
Acknowledgements. Comments by D. Scott of Climate Re-
search, S. Smith of the Pacific Northwest National Laboratory,
and several anonymous referees helped to improve the
manuscript. The CEC DG Research through the DINAS-Coast
project (EVK2-2000-22024), the US National Science Founda-
tion through the Center for Integrated Study of the Human
Dimensions of Global Change (SBR-9521914) and the
Michael Otto Foundation for Environmental Protection
provided financial support. All errors and opinions are ours.
LITERATURE CITED
Abegg B (1996) Klimaänderung und Tourismus— Klima-
folgenforschung am Beispiel des Wintertourismus in den
Schweizer Alpen. vdf Hochschuleverlag an der ETH,
Zurich
Aguiló E, Alegre J, Sard M (2005) The persistence of the sun
and sand tourism model. Tourism Manage 26:219–231
Amelung B, Viner D (in press) The vulnerability to climate
change of the Mediterranean as a tourist destination. In:
Amelung B, Blazejczyk K, Matzarakis A, Viner D (eds) Cli-
mate change and tourism: assessment and coping strate-
gies. Kluwer Academic Publishers, Dordrecht
Central Statistical Office Poland (2002) Tourism statistics.
Central Statistical Office of Poland, Warsaw (available at:
www.stat.gov.pl/english/serwis/polska/rocznik11/turyst.
htm)
CIA (2002) The world factbook 2002. Central Intelligence
Agency, Washington, DC (available at: www.cia.gov/cia/
publications/factbook)
Craig-Smith S, Ruhanen L (2005) Implications of climate
change on tourism in Oceania. In: Hall CM, Highham
J (eds) Tourism, recreation and climate change. Channel
View Publications, Clevedon, p 192–207
Crouch GI (1995) A meta-analysis of tourism demand. Ann
Tourism Res 22:103–118
Goessling S (2002) Global environmental consequences of
tourism. Global Environ Change 12:283–302
Goessling S, Peeters P, Ceron J, Dubois G, Patterson T,
Richardson RB (in press) The eco-efficiency of tourism.
Ecol Econ
Hamilton JM (2003) Climate and the destination choice of Ger-
man tourists. Working paper FNU-15, Research Unit Sus-
tainability and Global Change, Hamburg University and
Centre for Marine and Atmospheric Science, Hamburg
Hamilton JM, Tol RSJ (2004) The impacts of climate change
on tourism and recreation. Working paper FNU-52,
Research Unit Sustainability and Global Change, Ham-
burg University and Centre for Marine and Atmospheric
Science, Hamburg
253
Clim Res 29: 245–254, 2005
Hamilton JM, Maddison DJ, Tol RSJ (2005) Climate change
and international tourism: a simulation study. Global Env-
iron Change 15:253–266
Houghton JT, Ding Y, Griggs DJ, Noguer M, van der Linden
PJ, Xiaosu D (eds) (2001) Climate change 2001: the scien-
tific basis—contribution of Working Group I to the Third
Assessment Report of the Intergovernmental Panel on
Climate Change. Cambridge University Press, Cambridge
IMAGE Team (2001) The IMAGE 2.2 implementation of the
SRES scenarios: a comprehensive analysis of emissions,
climate change and impacts in the 21st century. RIVM
CD-ROM Publication 481508018. National Institute for
Public Health and the Environment, Bilthoven
Lise W, Tol RSJ (2002) Impact of climate on tourism demand.
Clim Change 55:429– 449
Maddison DJ (2001) In search of warmer climates? The
impact of climate change on flows of British tourists. Clim
Change 49:193–208
Metz B, Davidson O, Swart R, Pan J (eds) (2001) Climate
change 2001: mitigation—contribution of Working Group
III to the Third Assessment Report of the Intergovernmen-
tal Panel on Climate Change. Cambridge University Press,
Cambridge
Nakicenovic N, Swart RJ (2001) IPCC special report on emis-
sions scenarios. Cambridge University Press, Cambridge
New M, Hulme M, Jones PD (1999) Representing twentieth
century space–time climate variability. Part 1. Develop-
ment of a 1961–90 mean monthly terrestrial climatology.
J Clim 12:829– 856
Perry A (2005) The Mediterranean: How can the world’s most
popular and successful tourist destination adapt to a
changing climate? In: Hall CM, Highham J (eds) Tourism,
recreation and climate change. Channel View Publica-
tions, Clevedon, p 86– 96
Richardson RB, Loomis JB (2004) Adaptive recreation plan-
ning and climate change: a contingent visitation ap-
proach. Ecol Econ 50:83–99
Schlesinger ME, Williams LJ (1998) COSMIC 1.1 country-
specific model for intertemporal climate. Electric Power
Research Institute, Palo Alto, CA
Scott D, McBoyle G (2001) Using a ‘tourism climate index’ to
examine the implications of climate change for climate as
a tourism resource. In: Matzarakis A, de Freitas CR (eds)
Proc 1st Int Workshop Climate, Tourism and Recreation,
Porto Carras, Neos Marmaras, Halkidiki, Greece, 5 to 10
October 2001. International Society of Biometeorology
(available at: www.mif.uni-freiburg.de/isb/ws/report.htm)
Scott D, McBoyle G, Schwartzentruber M (2004) Climate
change and the distribution of climatic resources for
tourism in North America. Clim Res 27:105–117
Scott D, McBoyle G, Wall G (2005) The evolution of the cli-
mate change issue in the tourism sector. In: Hall CM,
Highham J (eds) Tourism, recreation and climate change.
Channel View Publications, Clevedon, p 192–207
Smith JB, Schellnhuber HJ, Mirza MMQ, Fankhauser S and 9
others (2001) Vulnerability to climate change and reasons
for concern: a synthesis. In: McCarthy JJ, Canziani OF,
Leary NA, Dokken DJ, White KS (eds) Climate change
2001: impacts, adaptation, and vulnerability. Cambridge
University Press, Cambridge, p 913–967
Witt SF, Witt CA (1995) Forecasting tourism demand: a review
of empirical research. Int J Forecast 11:447–475
World Bank (2003) Countries at a glance. World Bank Group,
Washington, DC (available at: www.worldbank.org)
WRI (2000) World resources database 2000–2001. World
Resources Institute, Washington, DC
WTO (2003) Yearbook of tourism statistics. World Tourism
Organisation, Madrid
254
Editorial responsibility: Daniel Scott,
Ontario, Canada
Submitted: February 7, 2005; Accepted: July 2, 2005
Proofs received from author(s): October 5, 2005