The impact of climate change on tourism in Germany, the UK and Ireland: a simulation study
- 7.3k Downloads
We downscale the results of a global tourism simulation model at a national resolution to a regional resolution. We use this to investigate the impact of climate change on the regions of Germany, Ireland and the UK. Because of climate change, tourists from all three countries would spend more holidays in the home country. In all three countries, climate change would first reduce the number of international arrivals—as Western European international tourist demand falls—but later increase numbers—as tourism demand from increasingly rich tropical countries grows. In Ireland and the UK, the regional pattern of demand shifts is similar to the international one: tourism shifts north. In Germany, the opposite pattern is observed as the continental interior warms faster than the coast: tourism shifts south.
KeywordsInternational tourism Domestic tourism Climate change Regional impacts
Climate is a crucial resource for tourism. Climate change would have a profound impact on tourism. This was largely ignored in earlier impact studies (Smith et al. 2001)—but a series of papers has emerged recently (for reviews see Scott et al. 2005; Hamilton 2006; Hamilton and Tol 2004). These studies of the impacts of climate change on tourism either investigate the fate of a single region—ignoring the wider context of national and international competition for tourists—or study tourism at the national level—ignoring that tourists tend to be highly spatially concentrated. This study is a first step towards reconciling the local and global aspects of the impacts of climate change on tourism.1
We use an econometric simulation model of domestic tourism in countries and of international tourist flows between 207 countries (Bigano et al. 2005). We combine that model with the subnational data-sets of domestic and international tourism of Bigano et al. (2004),2 and we develop a downscaling method that is consistent with the assumptions in the country model. We apply this to Germany, the UK and Ireland; these are countries for which we have good data and with which we are familiar. There are notable differences between the three countries: Germany is large, and its climate is part maritime and part continental; Great Britain is large, and its climate is maritime but with a pronounced north–south gradient; Ireland is small with a fairly homogenous maritime climate; the island of Ireland is shared between the Republic of Ireland and the UK. The differences between these three countries allow for speculative extrapolation to the regional impacts of climate change on tourism in other countries. Germany, Ireland and the UK account for 24.5% of all international tourist departures. However, the innovation of this paper lies in the regional downscaling within the countries. Together, the three countries receive 7.5% of all international arrivals, and they generate 6.3% of all domestic tourists (Bigano et al. 2004).
The paper is structured as follows. The second section presents the model used. The third section shows selected results at the national and regional level. The fourth section concludes.
The model consists of one component predicting tourism flows between countries and domestic tourism in countries. This is described in the second part of this section. A second component which scales national tourist numbers down to the regions of Germany, Ireland and the UK is described in the third part of this section. But first we will discuss the advantages and limitations of this approach in the following section.
Advantages and limitations of simulating tourism flows
The Hamburg Tourism Model (HTM) was developed based on the identification of several gaps in the existing literature on the impact of climate change on tourism. Existing studies have looked only at particular countries, particular environments or particular activities. Often these studies examined either inbound tourism, outbound tourism or domestic tourism. This means that substitution between destinations and between domestic and outbound tourism had been completely overlooked. Substitution, however, is a vital issue, as climate change will occur globally and not just in particular countries or regions. In addition, there is the issue of climate as a push factor, which is related to substitution. Not only can the attractiveness of a country’s climate be a factor in the decision to take a holiday there or in the decision when to have a holiday there, the climate, or rather the relative attractiveness of the climate, of the origin country is also a factor in influencing whether to have a domestic holiday or to travel abroad. The HTM tries to fill these gaps by estimating the demand to travel from all countries of the world, estimating the shares of domestic travel and international travel and estimating the demand for each of the countries in the world. In comparison with traditional demand estimation, which typically uses only economic indicators, the HTM uses environmental indicators such as area, length of coastline and temperature to predict demand.
The limitations of the HTM have been discussed in other articles using the simulation model (Hamilton et al. 2005a, b; Bigano et al. 2005) as well as in review articles (Bigano et al. 2006; Goessling and Hall 2006a, b). These can be summarised as follows. The tourism data used is problematic. Ideally, we should use monthly data broken down by trip purpose. Unfortunately, such data is only available for a few countries and even then it is incomplete. The World Tourism Organisation, which compiles the national tourism data used to produce the core equations of the HTM, also uses the same data to make forecasts until 2020 (WTO 2001). For some of the countries in the HTM data for arrivals or departures had to be estimated. Another limitation is that climate is only proxied by temperature in the equations that run the HTM. Climate is of course more complex than just temperature and tourists are not just interested in temperature. The complexity of climate and the interdependency of the various variables used to measure climate make it difficult to include more than one variable in a statistical analysis. Temperature, however, is the one variable that is statistically significant in all econometric studies of climate and tourism demand, and also the one variable that is always available in climate data sets—whether historical data or projections. A case study of German leisure tourists found that only 12% of tourists did not find temperature important (Hamilton and Lau 2005). Moreover, climate assessments using the Tourism Climate Index such as that by Amelung et al. (2007) also rely heavily on temperature data. The temperature data used in the HTM studies is the average for the whole country, which is an improvement on other studies where data for capital cities has been used. Another problem with HTM, indeed with any simulation model of social phenomenon, is that the future drivers of demand are uncertain. The scenarios used to drive the HTM include changes in population size and in GDP. While these factors are known to influence demand, other factors such as disposable household income and available leisure time may also affect demand. Unfortunately, data on such factors is only available for some countries and there are no long-term scenarios; therefore, these aspects cannot be modelled at present. The impact of random events, such as acts of terrorism or natural disasters, can, of course, not be modelled by the HTM.
There are several other shortcomings of the HTM. Its resolution is crude. It does not distinguish seasons, nor classifies tourists by age, income or trip purpose. Spatially, the model is restricted to countries. This paper only improves on the last shortcoming by downscaling to regions.
Amelung et al. (2007) is the only paper that is similar to ours, in the sense that their analysis is global with a reasonable amount of regional detail. Using a tourism climate index and two scenarios of climate change, they find that climatically ideal conditions for tourism are likely to shift towards the poles. Furthermore, they find that some destinations will see their peak season move away from the summer to the shoulder seasons and that destinations at higher latitudes could experience a longer summer season. Their analysis is, however, restricted to the supply side of tourism and the reaction of demand to the predicted changes in the supply of the tourism product (i.e., climatic attractiveness) is only speculated upon by Amelung et al. (2007). The tourism climate index used is mostly based on biophysical principles of human comfort. Other aspects of climate that have been decided to be important for tourists without any empirical evidence are also included. While the index may capture climate in a more complex way than the HTM does, the index ratings of the suitability for tourism are set arbitrarily and are again not based on empirical studies.
Tourism flows at the national level
We use version 1.2 of the Hamburg Tourism Model (HTM), an econometric simulation model of tourism flows in and between 207 countries. Version 1.0 is described by Hamilton et al. (2005a), version 1.1 by Hamilton et al. (2005b) and 1.2 by Bigano et al. (2005). The econometrics are inspired by Maddison (2001), Lise and Tol (2002) and Hamilton (2003), while the data are as in Bigano et al. (2004).
HTM works as follows. The number of tourists that a country generates depends on the size of the population and of average income. The share of domestic in total tourism depends on the climate in the home country and on per capita income. Climate is proxied by the annual mean temperature. A number of other variables, such as country size, were included in the estimation, but these factors are held constant in the simulation. International tourists are allocated to all other countries on the basis of a general attractiveness index, climate, per capita income in the destination countries, and the distance between origin and destination. Again, other explanatory variables were included in the regression for reasons of estimation efficiency, but these are held constant in the simulation. The number of international tourists to a country is the sum of international tourists from the other 206 countries. See Bigano et al. (2005) for further details.
The core equations are estimated using 1995 data, and the model is further calibrated, so that the model almost perfectly reproduces the observations on the number of domestic tourists, international arrivals, and international departures. More convincingly, the model also reproduces international arrivals and departures for the years 1980, 1985 and 1990. Data from these years were not used to calibrate the model. The correlation between observed and modelled international arrivals in 1995 is almost perfect. For the other years, the correspondence between observations and modelled values is never below 92%. For 1985, 1990 and 1995 the correspondence between observations and model results for international departures is between 91 and 94%. Only for 1980 does this drop to 79%. See Bigano et al. (2005) for more information on validation.
The Hamburg Tourism Model and its results depend on a number of parameters, each of which is uncertain. Hamilton et al. (2005a) present sensitivity analyses of the effect of changing the distance parameter (travel costs) and the income elasticity. They find that changing these parameters has a major effect on the baseline results but only a minor effect on the impact of climate change on international tourism. Similar sensitivity analyses are carried out for domestic tourism (Bigano et al. 2005). Changing the parameters for temperature and the scenarios of climate change has a greater impact on the results (Bigano et al. 2005; Hamilton et al. 2005a, b). Another issue is the saturation point of demand. Currently demand saturates at 12 trips per person per year. Hamilton et al. (2005b) find that changing the saturation point for international trips affects the absolute numbers of international tourists but the relative impact of climate change is similar.
The model shows that with climate change countries at higher latitudes and altitudes will become more attractive to tourists, to both domestic tourists and those from abroad. Tourists from the north west of Europe currently dominate international tourism,—the Germans, the Irish and the British together account for 25% of the international tourist market—which implies that the world total of international tourist numbers initially falls because of climate change. The model also shows that the effect of climate change is much smaller than the combined effects of population and economic growth, at least for most countries.
The Hamburg Tourism Model operates at a national scale, tracing domestic tourism in 207 countries and international tourism flows between those countries. In addition to analyses at the national level, the data presented in Bigano et al. (2004) allows us to look at a finer spatial resolution. For the majority of the regions, the resolution of the data is at the NUTS2 level. For Germany, this coincides with the administrative regions (Regierungsbezirke) within the larger states. The smaller states, however, are represented at the NUTS1 level, which is equivalent to the federal state (Bundesland). In total there are 40 regions. For the UK, we have data for 34 NUTS2 regions and London and Northern Ireland (both NUTS1). For Ireland, we have data for 8 NUTS2 regions, which are equivalent to the regional authorities.
We cannot increase the resolution of the model. Although we have data on where tourists go at a regional resolution, we do not have data on where they are from. This prevents us from re-estimating the empirical relationships that underlie HTM.
Instead, we downscale the national results of HTM to the 84 regions. The downscaling method distributes the national numbers of domestic and international tourists over the constituent regions. For the year 1995, this allocation is identical to the observed distribution.3 We split the allocation into a climate component, C, and an “all other factors” component, O. For instance, 3.5% of all international tourists in Germany visit Stuttgart (and its surroundings). The climate component equals C = 0.22T − 0.00791T 2. The other component is set such that CO = 3.5%, that is, O = 0.035/(0.22T 1995 − 0.00791T 1995 2 ). The values of C and O are calculated for every region. O is held constant over the simulation period. As the values of C change due to climate change, the market share will alter.
As the climate varies from north to south and from east to west, even a uniform warming of the British Isles and Germany would lead to a regionally differentiated pattern of climate change impacts on tourism. However, a uniform warming is not expected. We use the regional climate change scenarios from ATEAM for Germany and from UKCIP for the British Isles.
For Ireland regional data was only available for international arrivals. The main market is the South West, featuring the Blarney Stone and the Ring of Kerry, with just over a quarter of all tourists visiting there. The Georgian capital Dublin, with its pubs, museums and shops, is the second market, with a share of 17%. Apart from the relatively small markets of the Midlands (2.7%) and Mid East (4.9%), the other regions have markets shares of between 10% and 15% (see Fig. 14). As the distribution of domestic tourists is assumed to be the same as that of international tourists, the results discussed below apply for both.4 Nevertheless, it must be borne in mind that the distribution of international and domestic tourists is not likely to be homogenous.
In Ireland, for the low climate change scenario, for both domestic and international tourists, the border counties and the eastern seaboard (but not Dublin) become slightly more attractive, and the rest of country slightly less so. The change is less than a tenth of a percent. Under the high scenario tourists will increasingly visit the border counties, and the rest of the country will see a slight reduction in market share. Again the changes in market share are not higher than a tenth of a percent, except for the Borders which gain by almost a fifth of a percent. As climate change is largely homogenous over the relatively small island of Ireland, the spatial reallocation of tourists is small.
Discussion and conclusion
Previous studies of the impact of climate change show that there would be a shift of tourism towards the poles—if data and model are resolved at the country level. At a finer resolution, this is not necessarily the case as shown here for Germany. On the other hand, the global trend holds for the UK and Ireland. Not only does the regional topography of countries have to be taken into consideration, but also the regional climate change scenarios, which may not necessarily predict homogenous changes over the country. Particularly, the continental interior is likely to warm faster than the ocean board—which explains the difference between Germany on the one hand and Ireland and the UK on the other.
Here we have presented the regional effect of climate change as changes in market share. On the whole these changes are relatively small. This is particularly the case for Ireland, as both climate and climate change are fairly homogenous there. Nonetheless, changes of, for example, half a percent or quarter of a percent can still be of considerable significance in absolute numbers. This is particularly so for domestic tourism. For example, in Germany in the base year of 1995 there were 80 million domestic tourism trips. A change of half a percent leads to an increase or a reduction of 400,000 tourists per year. Although the regional changes caused by climate change may be significant in absolute terms, the increase in tourism caused by population and economic growth will be more important.
All models have weaknesses. The model simulates the development of total tourism demand by country, changes in destination choice, and hence tourism demand in countries and regions. What is not considered in the model is the capacity of countries and regions to meet demand. We assume that Say’s Law holds, that is, supply will meet demand. Apart from capacity restrictions, some countries may not be willing to meet the demand. Further limitations of the model are the focus on annual flows as opposed to seasonal flows. With the availability of better tourism data in the near future the model could be extended to simulate monthly or quarterly flows of tourists. We further assume that tastes and technologies are constant. These limitations are shared between the regional and national versions of the model. At the regional resolution, the major shortcoming is that destination is regional but origin is national. This reflects the lack of data rather than an inherent limitation of the model. The implication is that our measure of distance, an important variable in destination choice, is distorted, particularly for domestic tourism. All this is deferred to future research, as is the replication of the analysis presented here for all other countries.
In previous studies, we showed that the impact of climate change on national tourism is substantial. Here, we show that regional patterns do not follow trivially from national patterns—particularly in large countries. As tourism is so important economically, this justifies further research to alleviate the caveats listed above.
Note that weather and climate are different. Climate is what you expect, and weather is what you get. Climate, that is the expected weather, is important for planned holidays. Weather is important for behaviour during holidays, and for spontaneous (short) trips, although it may also distort climate perceptions.
Both domestic and international tourism data include holidays, business travel, visits to friends and relatives, and pilgrimage. Tourists stay at least 24 h at their destination, and at most 365 days. Note that labour migrants, particularly seasonal ones, often masquerade as tourists.
Note that we have regional data for domestic and international tourists for the UK, but international tourists only for Ireland; in the latter case, we assume that domestic tourists have the same regional preferences as international tourists.
Note that the absolute size of the domestic market is seven times that of the international market.
Discussions with Andrea Bigano, David Maddison and Katrin Rehdanz and comments by two anonymous referees helped to improve the paper. Jenny Behm, André Krebber and Nele Leinert helped with the data. Financial support by the EC-DG Research (DINAS-Coast) and the Hamburg University Innovation Fund is gratefully acknowledged. All errors and opinions are ours.
- Bigano A, Hamilton JM, Lau M, Tol RSJ, Zhou Y (2004) A global database of domestic and international tourist numbers at national and subnational level. Fondazione Eni E. Mattei Working Papers 3.05, Milano, ItalyGoogle Scholar
- Bigano A, Hamilton JM, Tol RSJ (2005) The impact of climate change on domestic and international tourism: a simulation study. Research Unit Sustainability and Global Change Working Paper FNU-58, Hamburg University and Centre for Marine and Atmospheric Science, HamburgGoogle Scholar
- Hamilton JM (2003) Climate and the destination choice of German tourists. Research Unit Sustainability and Global Change Working Paper FNU-15 (revised). Hamburg University and Centre for Marine and Climate Research, HamburgGoogle Scholar
- Hamilton JM (2006) Tourism, climate change and the coastal zone. University of Hamburg Faculty of Economic and Social Sciences Doctoral Thesis, Hamburg. Retrieved from http://www.sub.uni-hamburg.de/opus/volltexte/2006/3062/
- Hamilton JM, Lau MA (2005) The role of climate information in tourist destination choice decision-making. In: Gössling S, Hall CM (eds) Tourism and global environmental change. Routledge, LondonGoogle Scholar
- Hamilton JM, Tol RSJ (2004) The impacts of climate change on tourism and recreation. Research Unit Sustainability and Global Change Working Paper FNU-52. Hamburg University and Centre for Marine and Climate Research, HamburgGoogle Scholar
- Hamilton JM, Maddison DJ, Tol RSJ (2005b) The effects of climate change on international tourism. Clim Res 29:255–268Google Scholar
- Scott D, Wall G, McBoyle G (2005) The evolution of the climate change issue in the tourism sector. In: Hall CM, Higham J (eds) Tourism, recreation and climate change. Channel View Publications, Clevedon, pp 44–60Google Scholar
- Smith JB, Schellnhuber HJ, Mirza MMQ, Fankhauser S, Leemans R, Lin E, Ogallo L, Pittock B, Richels RG, Rosenzweig C, Tol RSJ, Weyant JP, Yohe GW (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, pp 913–967Google Scholar
- WTO (2001) Tourism 2020 vision—Europe. World Tourism Organisation, MadridGoogle Scholar