Annals of Tourism Research, Vol. 35, No. 4, pp. 1053–1073, 2008 0160-7383/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved. Printed in Great Britain
www.elsevier.com/locate/atoures
doi:10.1016/j.annals.2008.09.001
DELINEATING TOURISM Defining the Usual Environment Robert Govers Etienne Van Hecke Peter Cabus University of Leuven, Belgium Abstract: The ‘usual environment’ is a central concept in the UNWTO’s definition of tourism. Many countries apply rule-of-thumb distance measures to delineate its boundaries, which is rather problematic, because theoretically, the ‘usual environment’ consists of a selection of places, rather than one space with come concentric boundary. The concept becomes even more arduous in highly urbanized countries, as threshold distance values are hard to determine. Therefore, this paper will report on a exploratory analysis using various data sources. It leads to a more informed choice of spatial demarcation, although it is still inadequate in the light of theoretical considerations. Nevertheless, the approach should be of interest to those working in tourism and tourism statistics in particular. Keywords: usual environment, spaces, places, satellite accounting. Ó 2008 Elsevier Ltd. All rights reserved.
INTRODUCTION National statistical bureaus use different definitions of what constitutes a traveler’s ‘usual environment’ in terms of the statistical boundaries that are applied in tourism economic impact assessments (Instituto de Estudios Turı´sticos and The Canadian Tourism Commission 2003; Smith 1999). The choice of how to define the ‘usual environment’ has particularly become an issue since the earlier developments towards an international methodological framework for tourism satellite accounting commenced in the early 1990’s (UNWTO 2002) under which tourism is defined as: The activities of persons traveling to and staying in places outside their usual environment for not more than one consecutive year for leisure, business and other purposes not related to the exercise of an activity remunerated from within the place visited (UNWTO 2001:13).
Robert Govers is Assistant Professor in the Master in Tourism and Project Manager in the Flemish Center for Tourism Policy Studies, University of Leuven (Celestijnenlaan 200E, 3001 Heverlee, Belgium, E-mail ). Etienne van Hecke is Professor in Social and Economic Geography. Peter Cabus is director of the Flemish Center for Tourism Policy Studies; Associate Professor in the Institute for Social and Economic Geography; and Senior Researcher in the Flanders Social and Economic Council. 1053
1054
R. Govers et al. / Annals of Tourism Research 35 (2008) 1053–1073
Hence, essential in this definition is the ‘usual environment’, which, as a concept, is required in order to distinguish tourists from residents (or others such as for instance commuters, students or local leisure seekers) within a location. At the same time, this element is difficult to define in precise terms. ‘‘Generally speaking, it corresponds to the geographical boundaries within which an individual displaces himself/herself within his/her regular routine of life’’ (UNWTO 2001:14). Many statistical bureaus apply a distance threshold to delineate the ‘usual environment’, but of course the question then becomes ‘‘what is an appropriate threshold to use for defining tourism?’’(Smith 1999:137). This is a key issue, because it can have a dramatic effect on tourism statistics and the measurement of the impact that the phenomenon has on a region or nation, in economic as well as other terms. As Smith states (1999:137): ‘‘Despite the importance of ‘‘usual environment’’ in its definition of tourism, the [UN]WTO does not specify how the concept should be operationalised. Instead, statistical agencies are free to decide how this concept should be defined within their own social, political and economic context’’. That importance of this operationalization is clearly illustrated in tourism satellite accounting. Within the system of national accounts, tourism satellite accounting aims to isolate the economic impact of tourism by confronting tourism expenditure on the demand side with the production in tourism-related industries on the supply side. In this way, the economic contribution of tourism-related activities to the economy in general and relevant industries in particular, can be assessed. The starting point in tourism satellite accounting is to estimate total tourism expenditure of visitors. Considering the above definition of tourism, visitors include day-trippers, for whom in particular the concept of ‘usual environment’ is essential, in order to distinguish between routine and non-routine expenditure (assuming that, in general, people that spend the night away from home, do not do this within their ‘usual environment’, hence, this becoming a less relevant issue when looking at overnight visitors). In the context of the Canadian definition of tourism, Smith (1999) estimated the effect of halving the ‘usual environment’ threshold to 40 km or doubling it to 160 km. ‘‘The results indicate that cutting the threshold would introduce a large number of low-value, routine trips that do not appear to conform to the spirit of the WTO definition’’ (Smith 1999:137). Of course, the reverse could be argued, that when the threshold value is increased, a large number of close-by but non-routine trips that do conform to the spirit of the WTO definition, are not taken into account when estimating the impact of tourism. In other words, this is a paper about who to include and who to exclude from the collection of tourism statistics and hence analysis of tourism trends. Following Smith (1999), a more intense international discussion on the definition of ‘usual environment’ was provoked by several papers that appeared more recently, such as by Rogers (2002b; 2002a). They illustrate the complexity of the problem. There are basically three dimensions to the concept of ‘usual environment’: distance (or
R. Govers et al. / Annals of Tourism Research 35 (2008) 1053–1073
1055
the crossing of administrative boundaries); frequency of visit, and; definitions of places where people perform routine activities (home, work, study, shop, etc.). Also, there are two basic ways of determining what constitutes someone’s ‘usual environment’ in survey research; endogenous versus exogenous. In an exogenous assessment of the ‘usual environment’ the researcher must collect data regarding visitors’ location of residence, place of work or study and other areas frequently visited. The researcher will have to tell respondents what is considered to be ‘frequent’. Secondly, the researcher will have to set some distance threshold to determine if a visitor to a certain place can be defined as a tourist or a local visitor as illustrated above, depending on where the person lives, works or studies or which other places are visited frequently. Another way to deal with the issue of distinguishing between tourism and recreation is to leave it up to the visitor him or herself to asses if the location visited is within his or her ‘usual environment’ or not. This is referred to as the endogenous approach and seems to be the preference of the WTTC, which, under guidance of Oxford Economic Forecasting has as its ‘‘philosophy to let those who travel determine with their data when and where to draw the line between ‘usual environment’ and Travel & Tourism’’ (WTTC 2002:3). Official institutions in countries like Belgium or the Netherlands are reluctant to apply exogenous approaches applied elsewhere, because, unlike many early adopters of tourism satellite accounting, they are very small countries with relatively large highly urbanized and densely populated areas. Distance measures of fifty to eighty kilometers, such as applied by countries like Sweden (OECD 2000:257), Australia (Australian Bureau of Statistics 2002), the US (Okudo and Planting 1998:11) or Canada (Meis 1999:7), therefore seem inappropriate. In fact, for many small countries, such distance measures would imply that domestic tourism does not exist, as residents will cross international borders when they travel over 80 kilometers from home (in order not to complicate things even further, the issue of ‘borders’ is left out of the discussion, assuming that people that cross international boundaries are by definition traveling outside of their ‘usual environment’, although, of course, this is highly contestable). In order to assess what would be a good approach to delineate the boundaries of tourism in highly urbanized countries or regions, this paper will report on a exploratory analysis, which will be performed on survey data that collect both endogenous as well as exogenous measurements of the ‘usual environment’ of visitors to certain tourist attractions in Belgium. To put this touristic/recreational behavior into perspective, results will be contrasted with the findings of a household survey which measures mobility among consumers in relation to routine, day to day, weekly or monthly activities such as shopping, sports and other leisure activities. The authors will provide insight into the way in which endogenous assessments of the ‘usual environment’ correlate to exogenous approaches, and what would be the best approach and definition of the ‘usual environment’ in small densely populated countries. The methods and results should be of interest to any scholar
1056
R. Govers et al. / Annals of Tourism Research 35 (2008) 1053–1073
or practitioner involved in tourism, as it helps us to get a better understanding of the very foundation of the socio-geographic phenomenon of tourism, beyond the case-study presented here. Also, it provides a framework of analysis that could be applied elsewhere in order for other countries or regions to define their boundaries of tourism. However, first, the authors would like to briefly position the search for the ‘usual environment’ definition within current geographical literature on space, places and flows. It will lead to further conclusions about the need for additional research in order to embed this tourist concept in the new theoretical geographical approaches. USUAL ENVIRONMENT AND GEOGRAPHY The term ‘usual environment’ is not common in the geographical literature. Space and place together define the nature of geography (Tuan 1974). Place is a unique contextual (Aase 1994; Harvey 1973) entity and has a history and meaning: ‘‘Places incarnate the experiences and aspirations of people’’ (Tuan 1974). Space is a more abstract concept than place. Spaces have areas and volumes, places have space between them (Cresswell 2004). Yi-Fu Tuan (1977) has likened space to movement and places to pauses-stops along the way. When humans invest meaning in a portion of space and then become attached to it in some way, it becomes a place. This is the most straightforward and common definition of place—a meaningful location (Cresswell 2004). In order to become a place, repetition is the essence. In carrying out our daily routines we repeatedly go from one point to another, following established paths, so that in time a web of nodes and their links are imprinted in our perceptual systems affecting our expectations. A ‘habit field’ is thus established, where place is a ‘‘construct of experience’’ (Tuan 1974). Following this logic the part in physical space that can be termed ‘usual’ is determined by people’s individual experience, sense and construction of space. Through his research into people’s perception of space, by analyzing mental maps amongst other things, Lynch (1960, 2000) concluded that only small areas are well perceived, but that structures of cities or larger geographical areas are discerned by situating landmarks and pathways within space. Besides markers that stand out because of the individual’s personal experience or because they are well-known common place images (large monuments, parks, or other prominent structures) Lynch (1960, 2000) also identified limited areas (around places of residence, shopping or work) that people are very familiar with. In addition, lines that form linkages between major markers or that correspond to well traveled routes between place of residence and other frequently visited localities, create structure in space. Possibly there are ‘known’ areas that correspond to places of residence of family members or friends and there might be one or more places outside residential areas that are part of people’s space awareness, particularly as these places are linked to recreational activities. Also here, linkages are identified where distance is not always accurately perceived.
R. Govers et al. / Annals of Tourism Research 35 (2008) 1053–1073
1057
The literature discussed above already provides an initial indication of what the relationship is between ‘usual environment’, space and place. Being the result of an identification process it seems that the interpretation of the term ‘usual environment’ is better explained in the context of ‘places’ as opposed to ‘spaces’. Indeed the ‘usual environment’ is not a spatial continuum, but rather a collection of places with varying surface areas, that might not border onto each other and depending on peoples’ action spaces (Saey 1994), they are situated at different levels of scales. However, the unique contribution of this paper is not that it wants to force the reader to decide between space and place; but rather, the notion that the usual environment as a concept really provides an interface between the notions of space and place as they relate to tourism. In trying to explain the spatial organisation of society there are several (older) theories. It would lead us too far astray to provide great detail on all of these, but relevant references would include Christaller’s (1933) central place theory, where the market reach of urban functions is the main organizing principle. The oldest model (of 1826) is Von Thu¨nen’s core-periphery/concentric rings model, based on marginal productivity. Although these and other theories can still help us to understand spatial patterns of relationships, more recent views, linked to the process of globalization, add new elements. Strangely enough the abovementioned traditional dichotomy between space and place is reaffirmed in an era of globalization as a result from what can be named the global-local paradox (Conti 1993). Despite a progressive globalisation, there is at the same time a resurrection of the local level (many authors such as Porter (1990; 1998), Storper (1995; 1999), Swyngedouw (1989) and Uitermark (2002)). Identification processes and the existence and creation of cultural marks or identifying features are crucial for community coherence and become, at the same time, important elements with which the territory is ‘selling itself to itself ’ and to the outside world (Castells 2002; Ray 2001). Localisation based on identification is the counterpart of globalisation based on (economic) functionality (Castells 2002). What is different, however, is the increasing importance of the space of flows in understanding the space of places. Spaces of flows are increasingly determined by fast connections (ICT, Internet, High Speed Train, airports) ‘‘The higher the value of people and places, the more they are connected in interactive networks; the lower their value, the lower their connectivity. In extreme cases, some places are by-passed by the new geography of networks’’ (Castells 2002:551, see also; Taylor 2004; 2005). As this situation is obvious on macro (global) and meso (regional) level, this also plays a role at the micro-level (e.g. within one city where some places are connected and others not). Within the spaces of flows, people can jump places and scales (Uitermark 2002) and as it is the case for other agents (e.g. firms), people use places in space in a networked way (Cabus 2001; Castells 1996), facilitated by fast (virtual) connectivity. What makes tourism and the tourist unique in the networked global system (Sheller and Urry 2004), is that tourism and mobilities are not
1058
R. Govers et al. / Annals of Tourism Research 35 (2008) 1053–1073
two separate entities. Rather they are part and parcel of the same set of complex and interconnected systems (Hall 2005; Hall and Mu¨ller 2004). These systems involve networks of ‘host-guest-time-space-cultures’ that stabilise certain places as ‘places to play’. Places are not fixed but are themselves ‘in play’ in relation to multiple mobilities. Furthermore, places are being made and remade in order to draw in and to capture people on the move. The global mobilities presuppose the growth of ‘tourism reflexivity’ that ensures that increasing numbers of places around the world monitor, evaluate, and develop their ‘tourism potential’. This reflexivity involves identifying a place’s location within the contours of geography, history and culture that swirl around the globe, and locating their actual and potential material and semiotic resources. The existence and increasing importance of the spaces of flows, where places get their meaning from their connectivity to other places, complicates the assessment of ‘usual environments’. From a traditional geographical point of view ‘usual environment’ can be defined as a combination of mostly contiguous physical daily action spaces (work, shopping, recreational school, etc,). In a network society defining the ‘usual environment’ in that way becomes problematic because neither the geographical discontinued spaces of flows, nor the possible virtual dimension of the daily action space, where place images are increasingly created online and through other media (Govers 2005), are taken into account. From a more sociological point of view, (Usher 2002:45) points to the fact that there is no linear relationship between familiarity (usual) or strangeness (non-usual) and geographical distance. He suggests the familiar and unfamiliar are reconfigured and re-ordered, and that increased (en)counters with strangeness—direct or indirect—can result in enhanced understanding and sociality as much as increased alienation and/or hostility. Further research is needed to deepen the relationship between the geographical concepts of space, place and flows and ‘usual environment’. It is clear that defining ‘usual environment’ through average distances traveled, as is suggested by many national tourism authorities or statistical agencies, lacks theoretical embedding in current views on geography. Such distances suggest a space that could be interpreted as ‘usual environment’, while reality is different, as people are familiar only with parts (places) of that space, while at the same time there might be places outside of it that could be considered as ‘usual environment’ too, if the average distance traveled was ignored. In the latter case, it concerns more distant places that are nevertheless frequently visited and have become familiarized and hence are perceived as ‘usual environment’, as could be indicated by travelers when asked through endogenous questioning. To elaborate on these issues and to come up with a more informed decision on how to delineate the ‘usual environment’ in densely populated countries such as Belgium, this paper will therefore explore various empirical routes. As this exercise aims to assess statistical boundaries for the application of tourism satellite accounting, and the network behavior of tourists still has to be assessed properly, we
R. Govers et al. / Annals of Tourism Research 35 (2008) 1053–1073
1059
also apply—as is the case in other countries—contiguous approaches and not a networked approach. The Usual Environment in Belgium Hubert and Toint (2002) provide a general framework explaining the routine traveling behavior of Belgians, split between the non-active (elderly, students, unemployed, non-working) and employed population. Among the working population, the primary journey, the one between the places of residence and work, is also the longest (17.2 km). In proximity of the place of residence, there is a small trip in the morning, escorting children to school and/or doing minor grocery shopping. In the evenings shorter trajectories are traveled for reasons of leisure activities. For the non-employed population, which represents two thirds of total mobility, the average displacement is the result of less univocal traveling behavior, because the people involved belong to various types of groupings, representing diverging behaviors. The main trip among the non-employed population involves shopping, visiting friends or relatives, or recreation (with an average trip length of 19.8 km). This trip is slightly longer than the average commuter journey. But also among this non-employed population, morning trips are observed involving escorting or shopping and evening patterns include shopping, recreation, and visiting friends or relatives. Both among the employed and the non-active / non-employed population, primary routine journeys might be ambiguous involving multiple motives, such as escorting the children on the way to work, for instance. This results in an increased average trip length. To expand on this common knowledge of mobility in Belgium, in order to get some insight into how to define the ‘usual environment’ in regions such as these, this paper will now contrast two research studies, looking at mobility from two opposing perspectives, i.e., routine and non-routine traveling behaviors. Visitor Attraction Survey A central research objective within the Flemish Centre for Tourism Policy Research is to provide a complete picture of the significance of visitor attractions within Flanders (the Northern Dutch-speaking part of Belgium). This assignment breaks down into two parts: to identify visitor attractions that cause mobility of consumers moving outside of their ‘usual environment’ and that involves investments in time (distance traveled/time away from home) and resources (expenditure, investments and subsidies); and, both on the demand and supply side, measure variables that are able to signal trends in terms of economic effects, socio-cultural changes and special impacts. Within this context, the empirical data relevant for this paper was collected. Since 2003, the Flemish Centre for Tourism Policy Research has listed all visitor attractions in Flanders that fulfill the following
1060
R. Govers et al. / Annals of Tourism Research 35 (2008) 1053–1073
requirements: An identifiable physical location; Open to the public for at least two months in a year; A separately recognizable organizational structure; The core product is wholly and partly touristic/recreational in nature; A minimum of five thousand visitors a year. In 2004 the above inventory contained 285 attractions, subdivided into: 160 museums; 29 water and attraction parks; 21 historic residencies and monuments; 14 gardens, parks and nature reserves; 12 sightseeing attractions; 11 zoos; 7 castles and fortresses; 4 theme parks; and 29 miscellaneous. In order to initially assess the economic, spatial and socio-cultural impact of these attractions, a large scale visitor survey was conducted in 2004. This also included endogenous and exogenous assessments of visitors’ perceptions and scaling of their ‘usual environments’. Study Method. Between May 20 to October 4, 2004, tour parties and families were handed coupons at visitor attraction entrances. The coupons allowed visitors to partake in a prize draw contest either via telephone (a computer-response 0900-number), internet or by returning the coupon in the mail. Contest participants were subsequently asked to also participate in the research study. Online respondents were channeled through to an online survey and therefore able to immediately provide their feedback. Telephone or mail respondents received a return call within 12 days, asking them to complete a computer assisted (web survey linked) telephone interview. Contest enrolment was independent of the survey, but the prices offered (free flights and weekend trips) were meant to encourage visitors to register and subsequently take part in the survey. In total, 162 visitor attractions out of 285 handed out coupons. This 57% participation rate was considered highly satisfactory, particularly as the spread across categories of attractions and Flemish provinces was decidedly representative (see Flemish Centre for Tourism Policy Research (Cabus, Govers, Lievois and Keulen 2005)). Survey questions relevant to this paper, besides personal details, included the respondents’ postal code (to accurately measure distance traveled against the known postal code of the visitor attraction), frequency of visits to the particular attraction, and ‘usual environment’ perception. The latter question was formulated as ‘‘the visit to attraction XYZ was outside my normal environment; that means: outside the geographical boundaries within which I travel as part of my normal routine; i.e., not in the immediate vicinity of my home, work or place of study?’’ Response categories included: completely true; true; neither true nor untrue; not true; completely untrue; no idea. Sample. Close to eighty percent of the visitor attractions listed by the Flemish Centre for Tourism Policy Research submit their monthly visitor numbers to the institute. In the period between October 2003 and September 2004 this amounted in a total number of 15.2 million visitors (or rather visits; taking into account the possibility of multiple visits by the same person). This is an underestimation of the actual visitor numbers for all attractions, as 62 of the 285 attractions do not submit their data, but as most of these 62 attractions are relatively small it is assumed that this involves only a minor underestimation.
R. Govers et al. / Annals of Tourism Research 35 (2008) 1053–1073
1061
In total, 7,210 visitors participated in the prize draw contest. 12% of the contest participants was excluded from the survey as the twelve days post attraction visit maximum survey period was exceeded. This twelveday deadline was set because of recall considerations. Among the remaining group of contest participants a high level of cooperation was achieved with 87.5% of participants joining the survey. 300 respondents completed less than half the survey and 140 respondents completed the online survey after more than four weeks. As these groups were excluded, the resulting sample contained 5,110 respondents. After data quality checks, 4,600 questionnaires were retained for data analysis. This sample provides insight into the profile of 23,000 attraction visitors, as the survey pertained to tour parties (families) as a whole. The average tour party included five persons, which is close to the average group size of overnight leisure tourists in Flanders (4.6) according to De Bruyn, Nijs, and Lammens (2005). The highest absolute number of respondents is generated by museums (close to 48%, but note that also more than one out of two visitor attractions is a museum). Museums and zoos are generally overrepresented, while theme parks and sightseeing attractions are underrepresented. Nevertheless, a good sample across most types of attractions was achieved. Ultimately, an overall response rate of 1.5 in one thousand visits was reached. This number climbs to three out of one thousand visits when only the summer months, the period in which the survey was conducted, are considered (more than half of all visitor attraction trips occur in the months of June through August). Of course, non-response is hard to measure as it is not known how many coupons were handed out and noticed by recipients. Nevertheless, a satisfactory sampling across age groups, social classes and types of tour parties was achieved. 47.6% of respondents was female. Results. The average distance that attraction visitors travel is 47 kilometers, the median being 34 kilometers. There is a concentration of visits to attractions at relatively limited distance; up to 20–25 kilometers with a maximum frequency at 0-5 kilometers (11%). At 5–10 kilometers this percentage is already reduced to 8% and continues to decline as distance increases. There are very few visitors that travel more than 150 kilometers on the day to visit an attraction, which is normal for day excursions. However, the distribution is not the same across different types of attractions. To theme parks, castles and fortresses, visitors travel the longest distance, while this is the shortest to ‘water and recreation parks’. This is affected by the nature of the attraction and the availability of supply across Flanders. As far as the perceived location of the attraction relative to peoples’ ‘usual environment’ is concerned, a strong link with distance is observed. The proportion of visitors that perceives the attraction to be located within the ‘usual environment’ declines with distance traveled. Or, in other words, the endogenous measurement scale that assesses the perceived location of the place visited, relative to peoples’ ‘usual environment’, correlates with distance traveled.
1062
R. Govers et al. / Annals of Tourism Research 35 (2008) 1053–1073
Table 1. Visitor Attraction Location and Usual Environment Perception Distance (km)
Visit frequency (per year)
Total
0–3
4–11
>11
75% 62% 47% 34% 23% 14% 15% 6%
82% 77% 57% 31% – – – –
89% 71% 55% 35% – – – –
Total
23%
48%
59%
26%
Proportion of total visitors
88%
6%
6%
100%
0 to <5 5 to <10 10 to <15 15 to <20 20 to <25 25 to <30 30 to <50 50 to 100
78% 65% 49% 34% 22% 15% 16% 7%
Note: The percentages indicate the proportions of respondents, within the relevant distance \ frequency category, that perceive the attraction to be within their ‘usual environment’. In order to ensure sufficient observations per distance category, larger classes are used from 30 km onwards.
Table 1 shows that, as an attraction is visited more frequently, it is more likely to be perceived to be located within respondents’ ‘usual environment’. However, this is misleading, as visit frequency is dependent on distance and attraction type, two variables that themselves are related (as argued above). Per distance category, the change in proportion of respondents that perceive the attraction to be within their ‘usual environment’ according to visit frequency, clearly shows less variation or even irregular patterns (see distance categories 5 to <10, 10 to <15, and 15 to <20 in Table 1). This illustrates that distance, more than visit frequency, affects peoples’ perception of their ‘usual environment’, when it comes to visiting tourist attractions. All respondents that visit the same attraction more than three times a year travel less than 20 kilometers to get there (see empty cells in Table 1). This seems to indicate, that some natural average threshold distance value exists at somewhere around 20 kilometers. Consumer Consumption Survey In order to broaden the perspective on the ‘usual environment’ considerations in the above visitor attraction survey and to more accurately approach the issue of distance, it is useful to compare the above results with another major consumer behavior survey that measures consumption related mobility. A survey among 20,000 Flemish households attempted to map out all locations visited for regular or irregular consumption purchases and use of services, such as for instance restaurants, cinema’s or hair dressers. Locations were differentiated based on visit frequency and distance traveled. These variables allow for a comparison with the above visitor attraction survey.
R. Govers et al. / Annals of Tourism Research 35 (2008) 1053–1073
1063
Study Method. Within the framework of a large research study concerning the hierarchy of urban centers and their spheres of influence, conducted by the Institute for Social and Economic Geography of the University of Leuven and the Se´minaire de Ge´ographie of the Universite´ de Lie`ge, a written survey inquired about the places where consumers purchase various goods (convenience, shopping of specialty goods) or consume services (such as food and beverage, entertainment, cultural, health or educational services). For all types of goods and services, respondents were able to indicate three purchase locations in decreasing order of visit frequency, described as visited ‘most of the time’, ‘often’ and ‘sometimes’. In addition, the survey also gauged place orientation in a ‘synthetic’ manner, without distinguishing the types of goods or services, but including an indication of decreasing visit frequency. The question asked: ‘‘In general, which place(s) do you visit in order to purchase goods, consume services or indulge in leisure activities, and how often?’’ Respondents were able to list five places, with, for each, four visit frequency categories: ‘roughly weekly or more’, ‘roughly monthly (6 to 15 X/year)’, ‘several times a year (2 to 6 X/year)’, or ‘once a year or less’. In contrast with the other questions this involved concrete frequencies. The survey was conducted between 1995 and 1996; i.e., not very recent, but comparison of the results with 1975 data, showed stability of the findings as far as urban and regional spheres of influence of cities and towns is concerned. Hence, the data were considered to be a valuable source of comparison in relation to the visitor attraction survey and useful for the purpose of this paper. Nevertheless, the Belgian Directorate General for Statistics and Economic Data (2008) reported that both the total mobility in Belgium as well as the mobility in private vehicles in Flanders in terms of total vehicle mileage has increased with roughly 14% between 1995 and 2005. Even though this is a significant increase, the authors feel that it is not the kind of change that would significantly impact the type of use that we make of the 1996 data nor the conclusion that we draw from them. Sample. The survey was completed by roughly 20,000 households in Flanders and administered through all local schools, correcting for sampling of elderly, which indicates a sample of about one percent of all households, spread across all municipalities in Flanders, guaranteeing a geographically highly representative sample. Results. Table 2 lists the distances traveled (in kilometers) to locations visited for purchase of goods or services according to decreasing level of importance and visit frequency. The negative correlation between distance and visit frequency is apparent. Because of the way in which the survey was administered and presented here in Table 2, it follows that the table displays a diagonal structure, i.e., distance increases as location importance and visit frequency decrease. However, these are averages; exceptions are conceivable. For instance, when the person is working there, a large city resident might visit the city centre weekly and only call on the local suburban shopping area on a monthly basis, as facilities and supply there are limited and more specific. The general observation though is that consumers tend to visit the closest
1064
R. Govers et al. / Annals of Tourism Research 35 (2008) 1053–1073
Table 2. Average Distance (km) of Locations Visited for Consumer Purchases Roughly weekly or more
Roughly monthly (6 to 15 X/year)
Several times a year (2 to 6 X/year)
Once a year or less
3.6
10.1
–
–
6.0 7.4 9.3 9.9
11.7 14.0 16.7 19.0
19.6 22.5 26.2 29.3
– 33.4 35.7 39.0
4.7
13.6
24.8
37.0
Location 1 (visited most often) Location 2 Location 3 Location 4 Location 5 (visited least often) Total
pole most frequently only then to visit further poles in decreasing frequency. Table 3 lists distances traveled according to trip purpose and frequency. The frequency categories in the above Table 2 were determined quantitatively, while in Table 3 they are based on survey response categories as interpreted by respondents. Again it is apparent that visit frequency declines as distances increase. For a-periodic purchases distance traveled is obviously larger than for periodic purchases. As far as services are concerned, it is observed that sports are practiced in a tight local area, while visits to theatre, opera or concerts are less localized. Place in the Municipal Hierarchy. The above results indicate that it seems to be insufficient to operationalise the ‘usual environment’ purely on distance criteria. Besides the clear link with the type of purchase of goods or services, it is certainly necessary to embed the concept within the notions of structural space, for instance the settlement system such as displayed in Figure 1 (Van der Haegen, Pattyn and Cardyn 1982; Van Hecke 1998).
Table 3. Average Distance Traveled (km) According to Trip Purpose and Frequency Trip purpose
Periodic (semi-customary) purchases Less frequent (a-periodic) purchases Sports Cinema Theatre, opera, concerts Restaurant
Location visited most of the time
Location visited often
Location visited sometimes
9.2
16.1
21.9
12.0
19.8
25.1
4.2 11.3 15.7 9.1
8.1 20.0 27.9 15.8
12.1 26.6 36.2 22.4
R. Govers et al. / Annals of Tourism Research 35 (2008) 1053–1073
1065
Figure 1. The Belgian Settlement System
The statistical distribution of distances traveled for the purchase of consumer goods and services is clearly skewed. Among the total population in Flanders, many citizens live in larger cities and their agglomerations (34%) or their suburbs (14%) and in small towns (21%). These urbanites acquire their consumer goods and services at close proximity to their home (as opposed to the 31% of Flemish that live in rural areas). The average distance traveled for the purchase of goods and services therefore depends on the position of place within the settlement system (see top Table 4). The higher in the municipal hierarchy, the smaller the distance residents travel for their most frequent purchases. For less frequented locations this hierarchy becomes less relevant. In these cases, inhabitants of conurbation suburbs might travel shorter distances (for instance to visit the city centre), while in reverse, city or town centre residents are more likely to travel to another more distant town or city. For all hierarchical levels of place of residence distances increase as frequency decreases, but distances are generally longer for cities and towns than for their agglomerations, suburbs and other municipalities. When comparing distances in the various tables, one could deduct that the average distance traveled to familiar locations, where on a regular basis, roughly monthly, goods or services are purchased, is about 14 kilometers. However, all these tables indicate distances traveled between places, not zones or areas (spaces). To make the conversion possible from ‘places’ to ‘spaces’ consumer flows within urban spheres of influence were examined. From this, one can deduct the proportion of residents that travel for purchase of goods and services
1066
R. Govers et al. / Annals of Tourism Research 35 (2008) 1053–1073
Table 4. Average Distance Traveled (km) for Consumer Purchases and Spheres of Influence Hierarchical lever
Major cities Agglomeration major cities Regional cities Agglomeration regional towns Suburbs Commuter catchment areas Small towns Other municipalities Total
Roughly weekly or more
Roughly monthly (6 to 15 X/year)
Several times a year (2 to 6 X/year)
4.0 4.1 3.4 3.7 5.0 5.3 4.4 5.1 4.8
19.8 9.6 19.2 10.7 12.0 13.3 18.2 14.4 13.6
28.8 17.2 37.4 28.0 21.1 22.2 29.8 25.6 24.8
% residents that travel Of a major city Of a regional town Of a smaller town Total
within urban sphere of influence (level of small town supply) 90% 68% 497 Km2 89% 62% 378 Km2 81% 31% 176 Km2 86% 47% 230 Km2
% residents that travel Of a major city Of a regional town Total
within regional sphere of influence (level of regional city supply) 95% 81% 65% 1,384 Km2 93% 69% 42% 855 Km2 94% 74% 50% 968 Km2
48% 29% 12% 22%
within urban (level of small town supply) or regional (level of regional town supply) spheres of influence according to frequency and type of sphere (regional and large towns also have a sphere of influence at the level of small town supply because they also provide lower order goods and services) (see Table 4). Weekly flows for the purchase of goods and services remain within the urban sphere of influence in 86% of the cases. Because smaller towns provide less facilities and services than larger cities they have a smaller sphere of influence. Therefore the proportion of residents of smaller towns that stay within their urban sphere of influence for the purchase of goods and services is lower than in the case of larger cities. Monthly trips cross the borders of urban spheres of influence for over half the population on average, but more so for residents of smaller towns. Even half the population travels beyond the regional spheres of influence two to six times a year, a quarter does this on a monthly basis. Lastly, the second column of Table 4 allows for spheres of influence to be considered in terms of distances. Converting surface areas into distances produces an average radius of 8.5 kilometers for urban spheres of influence (7.5 for smaller towns) and 17.5 kilometers for regional spheres of influence. Attraction Visits in the Urban Hierarchy Table 5 allows attraction visitors to be placed within the settlement system. As one considers higher level urban spheres of influence, the
R. Govers et al. / Annals of Tourism Research 35 (2008) 1053–1073
1067
Table 5. Distances Traveled and Usual Environment Perceptions of Attraction Visitors % attraction Average Average distance distance visitors traveled traveled from outside within own sphere of influence sphere of sphere of influence influence
Within urban sphere Small towns Regional cities Major cities Total
of influence 12% 18% 33% 18%
(level ‘small 4.6 km 5.0 km 5.1 km 4.8 km
town’) 55 km 57 km 59 km 58 km
Within regional sphere of influence (level ‘regional town’) Regional cities 29% 10.4 km 63 km Major cities 37% 6.7 km 62 km Total 31% 10.3 km 62 km
Perceived to be within usual environment Distance % for Distance % for traveled visitors traveled visitors by these outside by these within visitors own visitors own sphere of sphere influence of influence
81% 73% 68% 73%
4.2 km 4.6 km 4.8 km 4.5 km
17% 16% 11% 16%
27 km 33 km 41 km 31 km
57% 63% 58%
6.9 km 5.3 km 6.9 km
14% 9% 13%
38 km 48 km 38 km
proportion of attraction visitors from the own sphere of influence increases. This is obvious as the number of attractions within such spheres is likely to be higher, but also particularly because the surface area of such higher level spheres of influence is larger (see Table 4). The same can be observed for regional spheres of influence. Within regional spheres of influence the average distance traveled to visitor attractions is also larger at 10.3 kilometers, roughly double the average distance traveled within urban spheres of influence at ‘small town level’. Because the urban spheres of influence of major cities are larger than for smaller towns, it is not surprising that the proportion of attraction visitors that perceive themselves to be within their ‘usual environment’, is smaller within the sphere of influence of a larger town, than it is for a small town. Distances traveled by attraction visitors from various types of spheres hardly vary, but this is not contradictory. Because the sphere of influence surface areas for major cities are larger, attractions are also located in the suburbs that are less frequently visited. The proportion of attraction visitors that travel to an attraction within the same regional sphere of influence and that perceive themselves to be within their ‘usual environment’ is again smaller and this for the same reasons. Regional spheres of influence are on average larger than urban spheres of influence, so the chance that residents visit an attraction within the own sphere of influence, but outside the personal ‘usual environment’ in enhanced. Some trips to attractions outside the own sphere of influence can still be perceived to be within peoples ‘usual environment’, but the percentages are small, i.e., 16% for urban spheres of influence and 13% for regional spheres of influence. The average distance traveled in these cases is significantly larger, being the result of a wide distribution of values corresponding to diverging perceptions of individuals.
1068
R. Govers et al. / Annals of Tourism Research 35 (2008) 1053–1073
Nevertheless, distances traveled in these cases are smaller than the average distance traveled to attractions outside the sphere of influence and outside the ‘usual environment’. One explanation for this is that attractions can also be perceived to be located within ‘usual environments’ that correspond to places of work, study, holiday resorts, near second homes or other places frequently visited.
CONCLUSION This paper has argued that the very foundation of tourism and how it is defined by means of the ‘usual environment’ concept is rather unstable and hence untenable. Because, any attempt to establish supposedly objective boundaries is, presently, wide-open to interpretation in the context of tourism statistics and satellite accounting. This is a particularly thorny issue, as the delineation of tourism can easily be shifted by readjusting local or national subjective definitions of the ‘usual environment’ (Smith 1999). This paper has therefore attempted to tackle this issue from both a theoretical socio-geographical perspective of places and spaces, as well as an empirical perspective, by contrasting touristic (visitor attraction) and regular consumption behavior, utilizing existing data sources. The latter has been exemplified through the case of Flanders, Belgium, but the methods used as well as the theoretical discussion could be equally applicable to any other entity, region or nation. It should therefore be of interest to any scholar or practitioner involved in tourism. The analysis of empirical data has clearly shown that trips to visitor attractions take place at varying distances from the place of residence. It seems rather obvious, considering the nature of recreation and tourism as the latter may play in remote destinations, in contrast to the former, possibly, in close proximity of one usual habitat (e.g., a midget golf course). The perception concerning places visited relative to peoples’ perceived ‘usual environment’ is distance related, as expected. The fact that attractions at larger distances are still perceived by some to be located within their ‘usual environment’, suggests individual differences in perception or personal circumstances (such as former places of residence, places of work, study, or second homes). The reverse, where attractions close to the place of residence are regularly perceived to be located outside of the ‘usual environment’, at first glance, seems odd. However, this phenomenon can be explained. Within urban or regional spheres of influence, people often travel centripetal, to the centre of town for the purchase of goods and services (which is often the case in a European context, but might be different in other parts of the world). Attractions are often not located downtown (with the exception of many museums and monuments), but on other locations within the sphere of influence; places that might not be visited frequently for work, school, shopping or otherwise. Hence, such places might easily be perceived to be located outside of the ‘usual environment’ when they are only visited for occasional recreational activities.
R. Govers et al. / Annals of Tourism Research 35 (2008) 1053–1073
1069
The Flemish consumption survey shows that for monthly purchases of goods and services, consumers visit places that are located between 10 to 20 kilometers away from the place of residence (on average 13.6 kilometers away). One might assume that consumers perceive places that they visit on a monthly basis as ‘familiar’. But a ‘familiar place’ is not necessarily part of a person’s ‘usual environment’. Again, here there are interpretation issues. Places within larger spaces might be perceived as being part of the ‘usual environment’ or not and it is logical that even at distances between 10 to 20 kilometers parts of the population perceive attractions to be outside of their ‘usual environment’ while others think otherwise. It is possible that an attraction located close to a shopping mall is perceived to be part of the ‘usual environment’, while this is not the case for a recreation park on the edge of an agglomeration. It is linked to the complications and impediments which researchers encounter in the theoretical debate about, amongst others, the exact definition, and by extension the role and meaning of ‘places’, in contrast to ‘spaces’, and vice-versa, for example, in the applied networked way people represent and subsequently consume places in space, as presented in the introduction of this paper. Here we also point to the different conceptual frameworks, with the central place theory on the one hand and the network approach on the other. While the first starts from an average attitude of consumers, resulting, in a European context (with in most cases a medieval city network still dominating commercial and service offerings), in a hierarchical centripetal model of central places (cities) with their spheres of influence, the second refers to the way individuals use places in space in the context of time-space compression (Harvey 1989) advantages that networks such as the TGV and budget airlines afford consumers and tourists. From a conceptual perspective, this paper seems to contribute to the idea that the concept of usual environment provides an interface between the concepts of space and place as they relate to tourism. It also shows that, following Smith (1999), this concept requires much more attention in the formulation of a system of tourism satellite accounts as its vagueness can have an immense impact on the results of tourism impact measurements. Why invest in impact assessments the way we have, if we can manipulate the results any way we want by just changing our usual environment definition? In order to start resolving this issue, this paper has made contributions both conceptually as well as pragmatically. In the latter sense the paper has found a way to determine the ‘usual environment’ specification in terms of surface areas. This was facilitated by the transition from distance measures to spheres of influence. Regional spheres of influence seem most appropriate. 74% of Flemish consumers purchase goods and services on a monthly basis within the regional sphere of influence. The average distance traveled is 13.6 kilometers, which could be reconciled with the average radius of regional spheres of influence, which is 17.5 kilometers (considering the concentration of retail shopping and services in city centers, it is logical that the average distance traveled is smaller than the average radius). These distances are also in agreement with distances traveled for
1070
R. Govers et al. / Annals of Tourism Research 35 (2008) 1053–1073
the purchase of goods and services in places that are visited ‘most of the time’ or ‘often’. It indicates a ‘usual environment’ threshold value between 15 and 20 kilometers. While the ‘un-usual’ environment threshold value in the Flemish visitor attraction survey is found to be positioned in the range between 20 and 25 kilometers, 20 kilometers seems to be a logical common threshold value to base the Belgian tourism satellite accounting approach on, combining this with endogenous measurement approaches. Although we must emphasize that, in the light of above theoretical discussions and variations in findings, such a standard distance threshold is still far from ideal, it seems to be the only practical solution for delineating the ‘usual environment’ from a statistical data collection perspective. When this is combined with endogenous approaches, it allows for corrections based on individual variations in perceptions of what is ‘usual’ and what not. Also, 20 kilometers seems to be a small value compared to internationally applied standards, but this study has provided the evidence to support the argument that this is an appropriate criterion in a densely populated and highly urbanized region such a Flanders. Future research might suggest other methodologies in order to provide evidence in support or opposition of these arguments or otherwise apply this research in other countries and regions in order to facilitate comparative analysis. The paper provides a methodological framework meant to define the usual environment in a manner that regions and nation states would deem appropriate for delineating tourism-related activities. The authors hope that their paper shall raise the debate for the urgent need to generate an evidence based ‘usual environment’ definition, i.e., a standard that goes beyond the current opacity, without enforcing common statistical thresholds that would ignore regional or national circumstances that have been shown, through the discussion in this paper, to be highly significant both on a local and global scale. The paper hopefully stimulates a rethink of the way we conceptualize usual environment in the context of places and places and the manner in which tourism impacts upon them. Perhaps, it generates a spark to ignite collaboration that would lead, one day, to the development of a global standard. Acknowledgments—The authors wish to thank their good colleague from the Rotterdam School of Management, Frank M. Go, for his help in editing the paper in its final stage.
REFERENCES Aase, T. 1994 Symbolic Space: Representations of Space in Geography and Anthropology. Geografiska Annaler: Series B, Human Geography 76(1):51–58. Australian Bureau of Statistics 2002 Australian Tourism Satellite Account. (Appeared in: Australian Economic Indicators, September 2000). Retrieved on November 5, 2003 from http://www.abs.gov.au/Ausstats/.
R. Govers et al. / Annals of Tourism Research 35 (2008) 1053–1073
1071
Belgian Directorate General for Statistics and Economic Data 2008 Mobiliteit (Mobility). Belgian Federal Government Service for Economics, SMEs, Retail Trade and Energy—Directorate General for Statistics and Economic Data. Retrieved on June 16, 2008 from http://statbel.fgov.be/ figures/d37_nl.asp#2. Cabus, P. 2001 The Meaning of Local in a Global Economy. European Planning Studies 9(8):1011–1038. Cabus, P., R. Govers, E. Lievois, and A. Keulen, 2005 Betekenis van attracties in Vlaanderen: Een gevarieerde sociale activiteit met een belangrijke economische impact (The Significance of Visitor Attractions in Flanders: A varied social activity with a relevant economic impact) (Tourism Research Papers No. 9). Leuven, Belgium: Flemish Centre for Tourism Policy Research. Castells, M. 1996 The Rise of the Network Society. The Information Age: Economy, Society and Culture, Number 1. Cambridge, Massachusetts: Blackwell Publishers. 2002 Local and Global: Cities in the Network Society. Tijdschrift voor Economische en Sociale Geografie 93(5):548–558. Christaller, W 1933 Die zentrale Orte in Su¨ddeutsland. Jena. Conti, S. 1993 The Network Perspective in Industrial Geography: Towards a Model. Geografiska Annaler. Series B, Human Geography 75(3):115–130. Cresswell, T. 2004 Place, A Short Introduction. Malden, USA: Blackwell. De Bruyn, R., V. Nijs, and M. Lammens 2005 Toerisme in Cijfers 2004 (Tourism in Numbers). Brussels: Flemish Tourist Board. Govers, R. 2005 Virtual Tourism Destination Image: Glocal identities constructed, perceived and experienced. Erasmus Research Institute of Management, Ph.D. Research in Management Series, Number 69. Rotterdam: Rotterdam School of Management Erasmus University. Hall, C., and D. Mu¨ller 2004 Tourism, Mobility, and Second Homes: Between Elite Landscape and Common Ground. Channel View Publications. Hall, C. 2005 Tourism: Rethinking the Social Science of Mobility. Pearson Education. Harvey, D. 1973 Social Justice and the City. London: Edward Arnold. 1989 The Condition of Postmodernity: An Enquiry into the Origins of Cultural Change. Oxford: Basil Blackwell. Hubert, J.-P., and P. Toint 2002 La Mobilite´ Quotidienne des Belges. Namur, Belgium: Presses Universitaires de Namur. Instituto de Estudios Turı´sticos, and The Canadian Tourism Commission. 2003 Research on National Practices Defining the Usual Environment: Basic Findings (Document presented at the fourth meeting of the Committee on Statistics and Macroeconomic Analysis in Tourism). Madrid, Spain: World Tourism Organisation. February 3–4. Lynch, K. 1960 2000 The Image of the City. (27th ed.). Cambridge, Massachussetts: The MIT Press. Meis, S. 1999 The Canadian Experience in Developing and Using the Tourism Satellite Accounts. Canadian Tourism Commission. (World Conference on the Measurement of the Economic Impact of Tourism). June 15–18. Nice, France. Retrieved on November 5, 2003 from http://www.wto.org/english/tratop_e/ serv_e/nicepaper_e.pdf.
1072
R. Govers et al. / Annals of Tourism Research 35 (2008) 1053–1073
OECD 2000 Measuring the Role of Tourism in OECD Economies: The OECD Manual on Tourism Satellite Accounts and Employment. Retrieved on November 5, 2003 from http://www.canadatourisme.com/tsa/english/assets/Measuring_ Tourism_part3.pdf. Okudo, S., and M. Planting 1998 US Travel and Tourism Satellite Accounts for 1992. Survey of Current Business 78(July):8–22. Porter, M. 1990 The Competitive Advantage of Nations. London Basingstoke: The Macmillan Press Ltd. 1998 On competition. Boston: Harvard Business School Press. Ray, C. 2001 Culture Economies: A Perspective on Local Rural Development in Europe. Newcastle upon Tyne: Centre for Rural Economy. Rogers, J. 2002a Have You Crossed the Line? A Discussion of Measurement Challenges in Leaving the Usual Domestic Environment. In Proceedings of the 33rd Annual TTRA Conference. Capitalizing on Travel Research for Marketing Success, pp. CD-ROM. Travel and Tourism Research Association. 2002b Crossing an Administrative Boundary: A New Approach to Leaving the Usual Domestic Environment. In Enzo Paci Papers on Measuring the Economic Significance of Tourism, Vol. 2, pp. 7-20. World Tourism Organisation. Saey, P. 1994 Van Ratzel tot Wallerstein: Zienswijzen op de Sociaal-Geografische Structuur van de Wereld. Tijdschrift van de Belgische Vereniging voor Aardrijkskundige Studies 63(1):47–86. Sheller, J. and M. Urry, eds. 2004 Tourism Mobilities: Places to Play, Places in Play. London: Routledge. Smith, S. 1999 How Far is Far Enough? Operationalizing the Concept of ‘‘Usual Environment’’ in Tourism Definitions. Tourism Analysis 4:137–143. Storper, M. 1995 The Resurgence of Regional Economies, Ten Years Later: The Region as a Nexus of Untraded Interdependencies. European Urban and Regional Studies 2(3):191–221. 1999 The Resurgence of Regional Economies, Ten Years Later: The Region as a Nexus of Untraded Interdependencies. In The New Industrial Geography: Regions, Regulations and Institutions, T. Barnes and M. Gertler, eds., pp. 23–53. London and New York: Routledge Studies in Modern World Economy. Swyngedouw, E. 1989 The Heart of the Place: The Resurrection of Locality in an Age of Hyperspace. Geografiska Annaler. Series B, Human Geography 71(1):31–42. Taylor, P. 2004 World City Network. A Global Urban Analysis. London and New York: Routledge. 2005 Inter-City Relations within Contemporary Globalization, Lecture on the Colloquium Marking the Completion of Professor Peter J. Taylor’s Tenure of the International Francqui Chair Gent: 13–14 January. Tuan, Y.-F. 1974 Space and Place: Humanistic Perspective. Progress in Geography 6:211–225. 1977 Space and Place: The Perspective of Experience. Minneapolis: University of Minnesota Press. Uitermark, J. 2002 Re-Scaling, ‘Scale Fragmentation’ and the Regulation of Antagonistic Relationships. Progress in Human geography 26:743–765. UNWTO 2001 Tourism Satellite Account: Recommended Methodologoical Framework.
R. Govers et al. / Annals of Tourism Research 35 (2008) 1053–1073
1073
Luxembourg, Madrid, New York, Paris: Eurostat, OECD, United Nations World Tourism Organisation and United Nations Statistics Division. UNWTO 2002 Basic Concepts of the Tourism Satellite Account (TSA). United Nations World Tourism Organisation. Retrieved on November 17, 2003 from http:// www.unwto.org/statistics/basic_references/index-en.htm. Usher 2002 Putting Space Back on the Map: Globalisation, Place and Identity. Educational Philosophy and Theory 34(1):41–55. Van der Haegen, H., M. Pattyn, and C. Cardyn 1982 The Belgian Settlement System. West European Settlement Systems (Vol. 22) pp. 251–368. Heverlee: Acta Geographica Lovaniensia. Van Hecke, E. 1998 Actualisering van de Stedelijke Hie¨rarchie in Belgie¨ (Updating the Urban Hierarchy in Belgium). Tijdschrift van het Gemeentekrediet 205(3):45–76. Von Thu¨nen, J. 1826 Der isolierte Staat. Rostock. WTTC 2002 Methodology for Producing the 2002 WTTC/OEF Travel and Tourism Simulated Satellite Accounts (Methodology and Documentation): World Travel and Tourism Council, Oxford Economic Forecasting.
Submitted 16 April 2007. Resubmitted 6 July 2007. Resubmitted 10 August 2007. Final Version 20 August 2008. Accepted 3 September 2008. Refereed anonymously. Coordinating Editor: Neil Leiper
Available online at www.sciencedirect.com