CAN FUTURES RESEARCH CONTRIBUTE TO TOURISM POLICY.pdf

Can futures research contribute to tourism policy?. Joseph W.M. van Doorn Following a definition of the basic terms employed, the author examines the interrelationships between planning, p o l i c y m a k i n g and forecasting. The main trends in futures research are described, and some forecasting methods and techniques conventionally used in tourism studies are considered. Criteria are derived by which tourism policy- makers could measure the usefulness of forecasts presented to them. Keywords: futurestudies;tourism+ recrea- tion; foreeastingmethods This article is an edited version of a paper given to the InternationalConferenceon Winter Recreation,Ottawa, Canada, 10-15 February 1981. The author wishes to thank Dr B. Otto Schneider(AssociateProfessor in English and Linguistics,University of Barcelona, Spain)for reading,commenting and criticizing this paper. Joseph W.M. van Doorn is Associate Professorin Planningand FuturesResearch at the TwenteUniversityofTechnology,the Netherlands. He has a special interest in the relationships between futures research and policymaking, on a nationalgovernmental level, in tourism.He can be contactedat Stadhouderslaan22, 3583 JJ Utrecht,the Netherlands. 1. Good introductions to the history of futures research are: Edward Cornish, The Study of the Future, Washington, DC, (World Future Society, 1977), and Jib Fowles, Handbook of Futures Research (London, Greenwood Press, 1978). In the title of this paper an assumption is made which specialists in the field of futures research now safely take for granted, but which, nevertheless, may raise legitimate doubts in the fields of tourism and recreation. The assumption is that scientific research can be and in fact has been extended from past and present to future events. With the first societal grouping man's concern was directed towards the preservation of his possessions and the acquisition of new ones, which both implied risk-taking and hence an interest in the outcome of the actions to be undertaken. Similar to what happened in the field of medicine, where the witchdoctor had to give way to the anatomist and the physician, and where the art of healing became a science, the futures researcher has had to do away with superstition and occultism, with oracles, prophets, astrologers and fortune tellers. His tools are no longer the crystal ball, dice, entrails of animals, the stars or cards, but statistics, hard evidence and facts, and computers. The missing link between the ambiguous and paramountly applicable predictions of the oracle and our present day 'failureprone' forecasts is to be sought in the still pre-scientific approaches to insight into the future; especially into the future of whole societies, such as Thomas More's Utopia ( 1517), Bacon's New Atlantis ( 17th century), Condorcet's idea of conditional probability (1978), and Gilfillan's thesis (early 20th century) with regard to the methodology of futures research. The 1920s brought the breakthrough. In the field of methodology We have already mentioned Gilfillan. Neither should we underestimate Ossip Flechtheim's endeavours to make futurology acceptable in academic circles, as the idea that science could contribute to a better knowledge of the future was then rejected by most scholars. The vast literature in the field available today gives live evidence of a change in attitude. By the same token, we can infer from the existence and creation of scientific institutions dedicated to futures research that the legitimate claims of this new science have been recognized and met. ~ In the field of policymaking at a governmental or industrial level the futures scientist still encounters obstacles in the form of certain deep-rooted views that tend to prevail despite all theoretical advances: • The future cannot be known. Science as such has to rely on empirical facts, which are available from the past and the present, but not from an as yet non-existent future. 0261-5177/82/030149-18503.00 © 1982 Butterworth & Co (Publishers) Ltd 149 Can futures research contribute to tourism policy? • Insight into the future ispossible only by means ofnon-scientiJTc methods, such as the crystal ball, astrology, intuition, rules of thumb and hand-reading. • The future can be known i f we view it ideologically as the strict realization o f ideas conceived in the past, eg the realization of an ideal state of society in conformity with definite political ideas. Regrettably though, this view generally lacks any adequate time perspective within which the potential forecasts about this state of society have to come true (and thus could be assessed). The above notions, while they prevail, have an inhibiting effect on the futures researcher's genuine commitment. Although it is true that there are certain unpredictable events, such as natural disasters, and although tacit knowledge and political ideas have a timely bearing on futures research, eg in the Delphi method, the only way for scientists to overcome the problem of acceptance is by converting policy makers to their credo: • The future can partially be known by way o f scientific methods and techniques. Thus the assumption we implied initially is not only a challenge for the scientist, it also makes a promise to the user in the sense of partially deleting risk from any venture. So in this paper I play the role of the devil's advocate and try to make planners and users of forecasts, with regard to tourism and recreation policies, keep both their feet on the ground, by pointing out some caveats, complexities and promising possiblities in futures research. I begin by defining the basic terms such as tourism, policymaking, planning, futures research and forecasting. Then the relationships and crosslinks that exist between planning, policymaking and forecasting are explained. Further we demonstrate some trends in futures research and consider the forecasting methods and techniques conventionally used in tourism to convince policymakers. This author is not overoptimistic about the results of future-oriented research in the field of tourism, as only certain aspects seem to form the focus for scientific attention. Apart from this, short-term explorative technologies are given preference over more speculative and normative methodologies, wherever forecasting is applied, without the claims or evidence of better results. Finally, we summarize a few criteria that could be used by policymakers to measure the usefulness of the forecasts presented to them. Definitions of tourism 2. Just a few examples: C. Kaspar, in Revue de Tour&me. 1954, page 50; F. Ogilvie, The Tourist Movement (London, 1933); A. Norval, The Tourist Industry (London, Pitman, 1936); W. Hunziker and K. Krapf, Grundriss der allgemeinen Fremdenverkehrslehre (Zfirich, 1942). 150 Since the 1930s when tourism adopted the distinctive features of a mass phenomenon, a considerable bulk of the literature has been devoted to tailoring a universal definition that would cater to multidisciplinary requirements.-' Synthesizing these attempts into a typology, we can differentiate four currents: 1. 2. 3. 4. Basic definitions. Mono-disciplinary definitions. Statistical definitions. Systems analysis definitions. In the first type, the basic definitions, two elements are constantly present: a static one (the tourist stay) and a dynamic one (the journey Tourism Management September 1982 Can futures research contribute to tourism policy? to the destination area and the activities during the stay), while at the same time the non-commercial character of the overall activity will be emphasized. A fine example of this type of definition is Burkart and Medlik's: "Tourism denotes the temporary, short-term movement of people to destinations outside the places where they normally live and work, for other than business or vocational reasons, and their activities during the stay at these destinations". 3 The second type lays greater emphasis on the motives and needs underlying the temporary movement of people, bringing in concepts such as relaxation, health, recovery from previous stress, cultural and educational interests, self-realization, and so on. The order of priorities is determined by the socio-psychological, anthropological or economic bias of the source of these definitions. Cohen 4 made a major contribution to this category, not only being the first to demonstrate that the very useful term 'fuzzy set' was fully applicable to tourism, but mainly by subsequently applying the fuzzy set rules to generate a viable definition of the tourist, as we see below. Fuzzy sets are classes where there are no clear-cut distinctions between membership and non-membership. Moreover a fuzzy set can be specified by breaking down the class concept into its constituent elements or dimensions (eg the duration of the 'tour'). Deleting all non-touristic elements from the set and validating a necessary and sufficient number of constituent dimensions will enhance the generative power and hence the realism of the set. Figure 1 offers a slightly elaborated synthesis of Cohen's conceptual tree. The definition is generated by combining certain subdimensions of preference or choice that condition others to be excluded. The chain underlying Cohen's own definition is the following: 1A/2A, B/ 3AJ4B/5A,B/6B/7A; " A tourist is a voluntary, temporary traveller, travelling in the expectation of pleasure from the novelty and change experienced on a relatively long and non-recurrent round trip"? However powerful this tree diagram appears to be, we will inevitably run into difficulties if we attempt to extend it beyond its inherent limits to generate a definition of tourism on the same lines. We would commit the old fallacy of mistaking a whole for the sum and total of its elements. Tour-ism is much more complex than summing up to the n th tour-ist! We will have to tackle this problem below when trying to set up a breakdown of tourism, since Cohen's definition comes in handy in the category of the tourist as an individual. Coming back to our typology, we look at the third type. Statistical definitions are mainly used by governmental and international bodies, such as the World Tourism Organization (WTO) and the Organization for Economic Co-operation and Development (OECD). They serve highly specific purposes, eg to count the number of incoming or outgoing tourists, their spending, the length 3. A.J. Burkart and S. Medlik, and purpose of their stay, etc. Yet an imaginative futures researcher Tourism, Past, Present and Future would not be reluctant to interpret statistical definitions largely as the (London, lrleinemann, 1974), page 311. operationalization of mainly economic (in general monodisciplinary) 4. E. Cohen, "Who is a tourist? A definitions. conceptual clarification", The SocioBefore we consider the fourth type of definition, the systems logical Review, 22, 1974, pages 527analysis approach, I acknowledge that my own definition of tourism 555. 5. Ibid. is heavily indebted to this type, because it offers the widest range of Tourism Management September 1982 15 1 Can futures research contribute to tourism policy? [~imensions Subdimenstons I Voluntariness Voluntary B Figure 1. A morphological 2 Time of travel and stay- Forced by social norms" t d a y f " ' ' ~ - " - % ~ < 6 monttls " 3 Direction Round trip~,....._. One way (emigrant) 4 Distance Short ~.....I~ C Forced by coercEon Permanent (nomad) Medium Long" Repeated visit Recurrent (regular) 5 Recurrency Non-recurrent approach to the tourist role 6 General purpose Source: Adapted from Cohen, text reference 4. (_*_.) Additional subdimensions to, or deviations from Cohort's original scheme. z Specific purpose instrumental~...~ Non E(irregular)instrumental- - ~ Rest(Sec°ndhome) / / ( health, education) (business) ( pleasure) Novelty anti--Continuity and~" chonge'A"f / / stability" ~" { TOURIST t possibilities with regard to the application of sophisticated dataprocessing and assessment techniques. As we see below, close links to futures research, especially with integrative forecasting, can be established within this system. Here tourism is viewed as a series of cross-links arranged over a grid, for example: • links between tourists and the region they visit, 6 • links between the tourist and the service sector, ie the agencies responsible for transport, accommodation, recreation, catering etc, • links between tourists and the host society, • links between tourists, the service sector, the destination area and the policymaking authorities. Leiper's definition illustrates the systems analysis approach to tourism: " T h e elements of the system are tourists, generating regions, transit routes, destination regions, and a tourist industry. These five elements are arranged in spatial and functional connections". 7 Components of tourism 6. A few studies are of practical interest here: K. Przeclawski, La Rencontre des Cultures (Varsovie, 1976); L. Turner and J. Ash, The Golden Hordes: International Tourism and the Pleasure Periphery, (London, Constable, 1975); K.D. Hartmann, A uslandsreisen, dienen Urlaubsreisen die Vflkerversti~ndigung? (Starnborg, Studienkreis ftir Tourismus, 1974). 7. N. Leiper, "The framework of tourism", Annals of Tourism Research, VI (4), 1979. For the systems analysis approach, see also, c. Kaspar, "Neuere liche Erkentnisse zumwissenschaftFremdenverkerhrs-bzw. Tourismus Begriff", Revue de Tourisme, 2, 1979; and J. Jafari, Editor's page, Annals ofTourismResearch, 1977 (special). 152 V, Let us now try to break down the overall phenomenon of (mass) tourism into its major constituent parts. We ought to differentiate at least four: 1. 2. 3. 4. The tourist. The intermediate framework. The supply of tourist resources and facilities. The societal context of 1-3 (tourist, intermediate framework, tourist supply). The tourist The first category, the tourist, generates three sources of data for the forecaster. First he will collect f a c t u a l information about the tourist that is the least liable to influence from his stay, eg items such as age, political preference, profession, socio-economic status etc. The tourist has brought these from his home country and will most probably take them back unchanged. However, correlations between these factual data and recreational behaviour are worth considering and can act as 'probability-generators' for future recreational behaviour. Tourism Management September 1982 Can futures research contribute to tourism policy? Then the researcher will approach a set ofbehavioural properties which either pre-exist in the form of habits or prejudices and are subject to transformation, or else, if absent, will be generated at the tourist destination. To mention a few, we can think of attitudes towards the destination country and its inhabitants, activity patterns, likes and dislikes, etc. The third category is perhaps the most delicate to enter the forecaster's calculations. It includes certain types of adaptive behaviour in the form of decisions generated on the basis of the expected and/or new experience in the host country, eg choice of transport, accommodation, region of preference, activity schedule etc. It is up to the forecaster, who, with the planner, shares the ambitious aim to satisfy tourist needs and expectations (in the medium and the long term), to cast the delicate balance between which types of adaptive behaviour the tourist can be expected to develop, and which tourist-adaptive steps will have to be undertaken on the part of the host region. If we now, in the light of our analysis of the tourist, reconsider Cohen's definition, we find the factual, the behavioural and the adaptive aspects covered. The intermediate f r a m e w o r k The second category is vast and encompasses, apart from what is generally defined as the 'tourist industry', various semi-public and private organizations, such as consumer associations, sports federations, the boards of certain recreational facilities, camp sites, or wild life reserves. Here the researcher is interested in possible or desirable developments as a function of tourist demand, and is concerned to safeguard the flow between the tourist and the supply of tourist resources and facilities. 8 Supply o f tourist resources and facilities In this category, the supply of, the access to, and the appeal of tourist resources and facilities are the main concern of the forecaster, as well as the supply structure itself, eg the functional and spatial balance between attractions, accommodations and infra- and superstructure, He will have to find answers to questions such as: How many camp sites do we need in 1985? How many ski-lifts are necessary for a new winter resort area near Ottawa? He will also find that, in order to meet demands in this category, he will have to rely heavily on data from categories 1 and 2. Societal context o f the tourist, the intermediate f r a m e w o r k and the supply structure In this last category, an often neglected one, the forecaster's interest 8. Studieswith regard to this aspect focuses on developments in certain sectors of society that may have of tourism can be found in: C. an influence on tourism. Gearing, W. Swart and T. Var, Those developments could be derived from studies on, among Planningfor Tourism Development, Quantitative Approaches (London, others, socio-cultural changes, the energy supply-demand situation, Praeger, 1976). political affairs and technology. All those studies have to be 9. Leadingmagazines d e a l i n g undertaken to gain insight in those sectors that surround tourism. partially with the future developments of tourism and recreation, are: The results will be forecasts on, for example, tourist flow, multiJournal of Leisure Research, Leisure access hotel reservation systems, the willingness to travel and Sciences. Annals of Tourism (recreation and holiday participation), air fares (as influenced by fuel Research, Tourism Management, Journal of Travel Research, Tourist prices), changing patterns of recreational activities as influenced by Review. an increase of free spending time? Tourism Management September 1982 153 Can futures research contribute to tourism policy? Tourism as a system of four basic categories We can now develop a definition of tourism by way of the four mentioned categories: "Tourism is the composition and the result of the interactional patterns formed by three set-constituent elements: the tourists, the intermediate framework, and the supply of tourist resources and facilities, all three & w h i c h are placed in, conditioned by, and have a bearing on a definite societal context". ~0See Figure 2 for a graphic representation of this definition. The only thing we still have to do is to substitute the word tourist according to Cohen's definition. However, this will be done with one c h a n g e - - i n s t e a d of pleasure being viewed as the main motive behind the "search for novelty and change" we use the expression, "satisfying physical, cultural and recreational needs". This has been done so as to include for example health- and cultural tourism, which are not necessarily concerned with pleasure and relaxation. Both tourism and recreation could be defined by the arrangement of the same constituent element, and the same holds true for the substitution of tourist with recreationist. As a number of tourist facilities and attractions are shared equally by foreign tourists, domestic tourists and the resident population, (eg sport facilities will serve the different needs of these groups), it is thus sensible for the futures researcher to focus on the relationship between futures research and tourism in general, while treating the subject of the relationship between different types of recreation and forecasting specifically. 10. Based on a definition from the author in "Toerisme, begrip voor een begripsontwikkeling", Recreatievoorzieningen, 11 (11), Futures research, planning and policymaking: a triad 1980, and refined by B. Otto Schneider (see acknowledgment). Policies are the consequence of a decisionmaking process which has I ( I I generar,~ { Tourist } Intermeaiate framework [ T~_.c; aes,,na~,on I I ) - ) Figure 2. Tourism as a s y s t e m of four basic categories: compilation of conceptual frameworks Source: Leiper, text reference 7; Kaspar, text reference 7; Przeclawski, text reference 6; J.W.M. van D o o m , "'Toerisme en t o e k o m s t o n d e r z o e k , een heer in een te krap jasje", Recreatievoorzieningen, 9, 1 9 7 9 ; J.W.M. van D o o m , " B u r m a en het s e l e c t i e f t o e r i s m e " , Intermediair, 14 (43), O c t o b e r 1978. 154 Tourism Management September 1982 Can futures research contribute to tourism policy? the objective of modifying the present in view of the future. This is why decisionmaking bodies do need valid information about the future and the positive and negative impact of potential decisions. This information-gathering can only be done by careful research, research which will then support the policymaking processes. Let us consider an example that is common practice in many resorts - - the building of a hotel. Such accommodation already has something to do with the future: the expected flow of tourists, the necessary bed-capacities, the market share etc. Building a hotel means talking about objectives, involves data-collection and processing, and concerns several government departments and sectors at a local, regional, or national level. Policymaking should display cohesion between multi-sectoral aspects and should strive to achieve coherent and viable coordination between multi-level objectives. For example, the spatial aspect shows up with the different Zoning Acts and infrastructural needs; the economic aspect is demonstrated in employment, foreign currency and multiplier effects. The sociocultural and socioecological aspects will make their claims in balancing income in favour of a depressed region, the training of hitherto unskilled workers for hotel jobs, and the ecological imbalances created by visitors in the environment. Goals, objectives and targets are not merely the concern of the investor and/or the hotel owner/manager. The Office of Public Works, the Town Council, the police and the hotel chain -- all have their say in the whole. The objectives of each may differ widely. They may aim at the creation ofjobs, attracting congresses or a certain kind of public relations exercise for this community. And who is to coordinate all those objectives, interest groups and powers? The planner. The systematic support of policymaking by research takes place in the planning process. Planning is the basis of policymaking. Thus futures research becomes a fundamental part of the planning process. This is shown in Figure 3 in which the terms 'anticipation' and 'design' both refer to futures research, tt To conclude: in tourism, policies have to rely on a coherent set of economic, political, sociocultural and spatial objectives. These objectives have to be placed into a decision framework whose primary function is the achievement of aims with specified means in a certain period of time. Policymaking in tourism is not an exclusive task of government, but grows along the lines of cooperation with the policymaking tourist organizations (national tourist organizations, information offices, consumer associations) and the tourist industry (hotels, restaurants, tour operators, travel agencies); even pressure groups might have a say in the policymaking processes. For such a complex task, planning in tourism as a function of, and in accordance with the four categories we established in our definition, is needed. Futures research and forecasting: methods and techniques 11. J.W.M. van(Assen, Doom and Vught, Planning van F.A. van Goreum, 1978). Tourism Management September 1982 So far I have been using these two terms as if they were selfexplanatory to the non-specialist. We now see what they stand for and how they are related. Any scientific study of the future is futures research. We agreed above that for any planning, decision or policymaking process to be effective, it ought to be supported by 155 Can futures research contribute to tourism policy? L..._ AnalySIs~ ~ ~ ..... ~ 1 Pohcyimplementation Policy evo~u3t;on ~ Figure 3. Planning, policymaking and forecasting viewed as the three components of the decisionmaking process I POI'CyprepOrct,On$ DesEgn * ~ j " Action t ~Poticy decisions E,o,oo,io~ L research. Thus this research is prior to the decisions that will determine and shape the future. It processes different projections of the present into the future by means of forecasting techniques to elicit a variety of results and thus a variety of potential futures for assessment and optimization. Hence the term futures research. Forecasting techniques The supporting methodology consists of what we call forecasting techniques. Below we concentrate on the use of forecasting methods and techniques that are suitable for the field of tourism. Having differentiated four categories in this field (tourist, intermediate framework, supply and societal context), we can safely presuppose that forecasting techniques are applicable to each. Two methods of approach can be distinguished: • The database is taken from the past and the present and gives rise to 'exploratory futures research'. • The desired future itself constitutes the database, decisions being shaped by working b a c k w a r d s - 'the normative approach' to futures research. ~2 12. E. Jantsch, Technological Forecasting (Paris, OECD, 1967). 156 On the basis of this almost classical distinction we can distinguish four different forms of forecasting. They have in common that they elicit conditional probability statements, based on either of the rational models of analysis. This means that forecasts, irrespective of their form (or modality) are obtained systematically and are subject to control and testing. In exploratory forecasting the scientist is concerned with the extrapolation of trends and the search for the logical development of alternative possibilities. Apart from this form it is also scientifically sound to base one's forecasts on a blend of intuition, expertise, and generally accepted a s s u m p t i o n s - speculative forecasting. Whereas in exploratory techniques it is hardly possible to include expectations about future policy decisions, speculative techniques offer the advantage to do so by means of method-implicit procedures. Here it comes close to the ideas of normativeforecasting, although in normative forecasting the scientist starts out explicitly with the formulation of norms and values that are to be valid in the future. The procedure involves constructing a series of consistent images of the Tourism Management September 1982 Can futures research contribute to tourism policy? future and subsequently tracing the route of attainment of(access to) these images, t3 Matters become even more complex when dealing with integrative forecasting, since its procedural capacity covers all the techniques customary in the three preceding forms. The aim of these techniques is to set up consistent relational patterns among isolated forecasts to enhance the plausibility of pronouncements deriving from any other technique. Thanks to this procedurally-comprehensive approach, a functional accumulation of knowledge, time, types and sectors is achieved. In Table 1 we offer a typological synthesis view over the four forms of forecasting and the methodological tools applied in each. In Figure 4 we relate the four types of forecasting techniques with the required database and the timescale. Exploratory forecasting in tourism Time series analysis 13. J.W.M. van Doom andvan F.A. van Vught, Forecasting (Assen, Gorcum, 1978). One of the most important data for recreation as well as for tourism is to know how many people are involved. Information about flows of recreationists or tourists are the inputs into policy decisions. It is unsurprising that the majority of forecasting studies so far in tourism are devoted to demand. The local council wants to know how many recreationists will stay overnight during a particular season. This will serve as an indication about how many new budget-class hotels have to be built. The skimanufacturer is interested in the number of people per nation that will be expected to go on a skiing holiday. The national government will include into their economic budgets prognoses of the prospective receipts from tourism and recreation. So demand is exceedingly important. Demand depends on a range of factors (or variables). F'xp[oratory ! 1 Speculative [ I .... , Normative 1., 1 forecasting w i t h time- and database Tourism Management September 1982 157 Can futures research contribute to tourism policy? Table 1. Type of forecasting technique and methodological tools Explorative Type description Methodological tools Extrapolation of trends Trend setting - time series analysis Regression analysis Gravity models Historical analogy method Scenario writing Morphological analysis Search for logical alternative possibilities Speculative Probability estimates of event occurrence Implicit expectations in policy decisions Source: v a n Doom and van Yught, text see reference 13. 14. Adapted from a reportfromthe Ministry of Transport and Public Works, Department of Aviation Affairs, Schiphol/The Hague, 1980. The example given has been worked out for a real situation in the chartermarket by Holland International, Travel Group, Rijswijk, 1980. 158 Brainstorming Delbecq and Impasse Delphi Normative Explicit description of desired future states and the routes that lead to them Normative scenario writing Bayesian statistics Pattern Integrative Research into the implications of options Establishment of relational patterns among hitherto isolated forecasts Input-output models Cross-impact analysis Mapping Several explorative forecasting techniques are used to foresee the developments with regard to demand in tourism. Those techniques differ, among other things, with respect to the number of factors that are taken into account. Time series analysis merely focuses on the historical developments of one variable so as to forecast its (near) future developments; linear regression models do the same for two variables, while the multi-regression models (like the gravity model) consider demand in relation to three or more variables. Thus time series would start with just one variable: tourist arrivals, recreational receipts, or aircraft sales, etc. The forecaster can choose one of the various time series analysis techniques: B o x Jenkins, Census II, Leading Indicator etc. All these are used to break down, in one w a y or another, a time-series into seasonal, trend, cycle and random elements. N o w let us consider, by way of an example, how a Dutch forecaster, who is - - in 1980 - - interested in 1981 winter tourism development (we only take air charter development), handles the one variable time series analysis. First, he has to collect a data-series on which he can base his forecasts (see Table 2). '~ Second, as he is not sure about developments in 1980, he finds it safer to use two indicative numbers for the year 1980, an optimistic and a pessimistic one, let's say 155 000 and 175 000. N o w he uses these two figures to estimate the distribution over the winter months. He can do this with the × 11 method. Then he will attach both the new series obtained to the base one ( 1 9 6 6 - 1 9 7 9 ) in such a w a y that the effect of the new series is greater than the base series. Third, he wants to produce the 1981 forecast with upper and lower limits. As he has already two alternative 1980 figures, he will use an exponential smoothing technique for the lower limit and a linear curve-fitting technique for the upper limit, and obtain the matrix shown in Table 3. O f course our D u t c h friend could extend this exercise again to eive us monthly totals for winter 1981. But how can we assess his results? With this simple one-variable example from time-series calculaTourism Management September 1982 Can futures research contribute to tourism policy? Table 2. 1966 Data series on Dutch winter tourism development. 1970 16000 73920 1972 1974 1975 1976 1977 1978 1979 89330 107400 140000 160000 182880 182800 164000 Table 3. Possible outcomes of multiple use of time series forecasting techniques a Optimistic Pessimistic Source: Holland International, text reference 14. 1980 175 000 155 000 1981 Upper limit (linear curve fitting) 180000 169000 Lower limit (exponential smoothing) 145 000 131 000 aTheycould have used better techniques but the example is one often seen in practice. tions I wanted to emphasize that forecasts tend to become less accurate and less reliable the longer the time-period they range over. What should the tour operator, as a decisionmaker, do when he is presented with figures ranging from 131 000 to 180 000, about a year and a half ahead? Can he really base his policy on forecasts that oblige him to operate with a risk-margin of roughly 25 000 passengers? He has to make investments, buy accommodation and book plane-seats, indeed quite some time ahead! This example thus gives us some indication about the limited timescale within which time series analysis and forecasting on an exploratory basis are useful; it alludes to the problems a forecaster faces using a one-variable technique; it points towards the implications in policymaking and the 'interfering factors' that derive from lack of data, irregularities and seasonality. Although the techniques mentioned are widely used elsewhere in tourism planning, they do not seem to be of too much help to the tourism forecaster who focuses on the medium and the long term. Several authors like BarOn and Vanhove have mentioned this problem. 15 Regression models 15. BarOn, "Forecasting, theory and practice", Tenth Annual Conference Proceedings of the Travel and Tourism Research Association (1979); N. Vanhove,"Forecasting in tourism", Revue de Tourisme, 35 (3), 1980. 16. Brian H. Archer, "Forecasting demand: quantitative and intuitive techniques", International Journal of Tourism Management, 1 (I), 1980; and Vanhove, op cit, reference 15. It is easy to see, however, that the apparent inaccuracy of the forecast above is caused by the fact that here future behaviour is explained only and exclusively by way of processing the data through time. Even admitting that knowledge of past behaviour can but indicate the probabilistic structure of future behaviour, it should be clear that demand, or any other variable we are interested in, depends on a lot more factors or variables than simply time. The second type of exploratory forecasting therefore considers tWO or more variables in correlation. A well known technique here is the linear regression model, used for a two-variable relationship, eg income and holiday participation, based on the least squares method. For more than two-variable calculations, multi-variable regression models will be required. Vanhove recommends for this purpose the Artus model, while Archer presents the Askari model. ~6 As examples, or better, slight deviations from the multi-variable regression models, we could mention the so-called trip-generation models and gravity models: Tourism Management September 1982 159 Can futures research contribute to tourism policy? A multivariable demand model is specified as a functional relationship between a dependent variable and one or more explanatory variables: the object of analysis is to discover the absolute and relative degrees of influence exerted by each of the explanatory variables on the dependent variable. A Gravity model, however, is expressed in a more rigid form: the nature of the relationships, especially those concerning distance [and travel costs] is more closely specified. '7 In the literature a long list of models is presented. To mention a few with the name of the author: Armstrong, Crampon and Lesceux (gravity), Gordon (trip generation), Jud (linear regression). ~s In the Armstrong model the following variables were used to forecast tourist flows to several tourist destinations: • • • • • size of the population of the tourist generating country, income per capita (in each of the generating countries), distance from generating country to tourist destination, travel time, special relation variable, eg a common language between a generating and a destination country, • a parameter for the relative appeal of the destination countries. Using his model outcome for comparison with actual data, the results are not very encouraging. Nor do they become more promising if we allow for the extenuating circumstances that the tourist arrival statistics originated in different countries, which accounts for 17. Archer, op cit, reference 16, incongruencies in the census methodology, or the fact that the page 9. number of countries compared differ slightly with Armstrong's study 18. C. Armstrong, "International tourism, coming or going: the (see Table 4). methodological problems of forecastAlthough our criticism cannot be as severe as Vanhove's, when he ing", Futures, 4 (2), June 1972, pages speaks about trend extrapolation - - " P u r e trend extrapolations and 115-125; L. Crampon, "Gravitational model approach to travel market projections based on alternative rates of growth are lacking any analysis", Journal of Marketing, 30, background to justify future evolution ''19 - - this model at least shows April 1966; D. Lesceux, La Demande Touristique en Mediterranee (Aixthe weakness of any extrapolation method: en-Provence, 1977); G. Jud and H. • useful in the short term, Joseph, "International demand for Latin American tourism", Growth • losing power very quickly in the medium term, and Change, Jan 1974. • practically useless in the long term. 19. Vanhove,op cit, reference 15, page 5. m comparison of several explorative forecasting techniques by 20. S. Makridakis and S. Makridakis and Wheelwright may illustrate this (see Table 5). -~° Wheelwright, (eds), Forecasting (Studies in Management Sciences, Scenarios Vol 12, Amsterdam, North Holland Publishers, 1979), pages 6-9. In the forecasting techniques, arithmetic and mathematics were used. The variables were quantifiable and the results were forecasts about a certain point in time, or points in time. Nothing or little could be Table 4. A comparison of forecasts and actual data. (× 103) Used data 1967 Armstrong's Actual data forecast for 18 generating OECD countries All countries countries (1975) 1975 15 858 1 627 13 130 635 25 368 2 269 22800 971 Countries Source: Armstrong, text reference 18; and OECD, Tourism Policies in O-ECD Countries ( OECD, Paris, 1975). 160 Canada Netherlands Spain Portugal Australia Japan 160 336 409 591 13 375 2 399 28726 788 378 438 13 660 2 819 30 122 888 496 707 Tourism Management September 1982 Can futures research contribute to tourism policy? Table 5. Comparison of some exploratory forecasting techniques as to their usefulness in short-, medium- and long-term assessment Exponential smoothing X - 11 Box-Jenkins Historical analogy Regression analysis Short-term 0-3 months Medium-term Long-term 3 months - 2 years 2 years and up Fair to very good Very good to excellent Very good to excellent Poor to good Good Poor to good Very poor Very poor Very poor Poor Good to fair Poor to fair Good to very good Good to very good Poor to fair said about qualitative variables, such as the influence of the policymaking process itself, and the changes of variables are subject to time. With respect to the latter one has to keep in mind that two conditions are fundamental in time series and regression-like models: • the variables in the model used will remain unchanged in the future, • the relationships between the variables are constant. This, however, may be true for the short term, but for the medium and long term it is not. For medium- and long-term forecasts to be of any practical value to the planner, we must adjust our techniques to handle a bundle of qualitative variables denoting the expected turning points in a policy framework along a timescale as a result and extension of quantitative data processing. This is what scenarios are intended to be. In the classical sense scenarios are hypothetical sequences of events. They pretend to trace possible designs of the future and the routes that subsequently would lead towards them. In exploratory forecasting, scenario writing means moving along the scale from past-present to future, while in normative forecasting the procedure is more sophisticated: the forecaster moves from the future backwards-forwards to the desired state. To date no normative scenario exists in tourism. Principally there is no great difference between the methods of exploratory scenario writing (elsewhere called a projective scenario) and normative scenario writing (also called a prospective scenario); although in a prospective scenario description of the desired state might cause considerable difficulties. Even if problems of context can be given an acceptable solution, there will still remain certain methodological problems, eg the treatment of consistency, plausibility, and the level of aggregation as a challenge and task for futures research. It would be a fallacy to assume that the problems arising from projective scenario writing have been overcome. Due to the novelty of the technique (relatively speaking) and the difficulty in handling qualitative data with tools developed for quantitative data processing, to my knowledge only a few noteworthy studies exist. And these, at best, present a set of quantitative/qualitative trends from various fields that might constitute (or could be used as) input for scenario proper. To quote only a few examples of these inputs or 'preparatory stage-scenarios', we could mention MacGregor's, Koster's and Kahn's contributions to the conference "Tourism in the Next Tourism Management September 1982 161 Can futures research contribute to tourism policy? 21. M. Kosters, "Holland and tourism in the next decade", Tourism Planning and Development Issues, edited by Hawkins, Shafer and Rovelstad (Washington, DC, George Washington University, 1980);J.R. MacGregor, "Latin America: future scenario forecasting for the tourism industry in some of its developing nadons', Tourism Planning Development Issues, pages 429-443. 22. Kosters,op tit, reference 21, pages 51-52. 23. H. Kahn, Travel Trade News Edition, section one, X C V I (6), 1979. 24. The Big Picture, 24, ASTA travel news, Travel "79-'80, World Trends and Markets. Decade" (Washington, DC, 1979). -'t Kosters, for example, offers forecasts about some seven tourism-related fields (economy, leisure time, population, nature, space, technology and science, politics). He then tries to correlate these forecasts in a systematic framework so as to produce a weighted estimate with regard to the consequences of the correlations for tourism developments in the Netherlands. Yet he has to recognize that the direct and indirect influence of developments in different fields, with their separate impact on tourism, "in the long run remain uncertain". 22 Kahn, the father of scenario writing, has less to contribute to the future of tourism. His forecasts are either vague or so trivial that one wonders for whose benefit they have been produced.:" Just one example in this respect: Kahn foresees that in 1989 the tourism growth rate will be double the economic growth rate. This, however, is not surprising. Figures from 1969-78 on world tourism prove that this has not only been the case for quite some time already, but has even been recognized in the literature: "Apparently international tourism grows at almost twice the rate of G N P growth". -'~ Thus the borderline between simple truism and genuine forecasting is difficult to identify. Better examples that overtly point in the direction of scenarios as we defined them, are found in Baron's paper presented at the tenth annual Tourism and Travel Research Association ( T T R A ) conference in 1979. Here it must suffice to skim through just one of the examples to illustrate what kind of alternative assumptions are made and how they are arrived at (see Table 6). The assumptions made explicitly (eg the relative reduction of oil prices) and the underlying, implicit but less obvious, assumptions (eg the balance of powers in the world remaining unchanged), are always the weakest points of the exercise. The assumptions can be considered a result of a time-serialization of the database, the alternatives being triggered by means of factorization towards negative or positive developments. But there is a lack of hypothetical grading or stepwise Table 6. Scenarios based on alternative assumptions (Tourism to Thailand: scenarios 1975-1980). Field Optimistic Intermediate Pessimistic Political 1. International Improved d~tente As 1975 Increased tension, local wars Economic 2. Prices of oil and air transport Relative reductions As 1975 (relatively) Further relative increases More fare reductions, scheduled and charter As 1975 Decreasedavailability of charter and promotional fares 1 165 1 575 1 900 1 135 1 375 1 550 A i r transport 3. Fare structure Forecasts of visitors arrivals (10 3 ) 1975 1978 1980 1 185 1 895 2 550 Actual number of arrivals: Source: BarOn, text reference 15. 162 1975 1978 1980 1 180000 1 475 000 1 850000 Tourism Management September 1982 Can futures research contribute to tourism policy? progression towards one of the alternatives. Thus the scenariowriting business needs to be supported by more elaborate techniques that will enable the forecaster to improve his assumptions, to strengthen their predictive power, and to widen their scope to range over qualitative data. For this purpose we turn to the second area of forecasting speculative forecasting. Speculative forecasting This type puts a series of tools into the hands of the scientist that allows him to move towards a less quantifiable terrain and become more independent of the influence from his past-present database, thus enhancing the accuracy of his studies. We are now ready to abandon the somewhat casual assumptions made in the Thai example in Table 6 and take a look at some of these tools. In passing we mention a few names from the list of (already) conventional techniques, such as SIG (Subjective Integrated Group Processors), JAM (Judgment Aided Models) and G D S T (Group Discussion Structuring Techniques). The common objective in all these procedures is the pooling of the expertise and skills of people proficient in highly unstructured fields (eg tourism). The most famous among these techniques and perhaps the one that has been subject to the most passionate discussion and criticism is the Delphi method. The Delphi method In a Delphi study, a questionnaire dealing with a specific problem is presented to a group of experts in the field. They will answer questions, for example, about the probability and/or desirability of certain events occurring, eg the likelihood of a 100% computerized reservation system for the leading hotel chains and travel agencies in 1984. Characteristic of Delphi studies is their striving towards consensus. In several written rounds t h e - most of the time, a n o n y m o u s - experts try to convince each other, by their arguments, that certain answers are more likely than others. The result for each statement is then a statistically aggregated collective answer on which consensus was reached. A Delphi example focusing on 25. A.E. Robinson, "A return to tourism might be found in the report on the results of the international Delphi", in "A decade of achievement", Tenth Annual Conference symposium, "Tourism and the Next Decade". 25 Proceedings of the TTRA (1979). The validity of this method may be questioned and subjected to 26. Harold Saekman,Delphi serious criticism. Several papers dealing with Delphi have pointed Critique (Toronto, Lexington Books, 1975); H. Linstone and M. Turoff OUt the weaknesses and limitations. For criticism reference is (eds), The Delphi-Method: generally made to Linstone and Turoff, although to my knowledge Techniques and Applications (Reading, Sackman's Delphi Critique is even more thorough in this respect. 26 MA, Addison-Wesley, 1975). 27. Delphi-SEERTwo Delphi studies relevant to tourism were undertaken in Expertenbefragung 1978-1980 fiber Austria and Switzerland, respectively. 27But one of the first studies of F.remdenverkehrsntwicklung in this kind was carried out in Canada. Referring to the Canadian study, Osterreich (Wien, Institut t'fir Fremdenverkehrsteehnik,1980); J. it is said in the Austrian report that: "This was the first study in the Krippendorf, EineDelphi-Umfrage world that tried to estimate the future of tourism in a very f~ber die zf~kf~nftige Entwicklung des comprehensive way. It was not the aim of this study to work out Toutqsmus in der Schweiz, (Bern, 1979). measures or recommendations". :s In the Canadian study special 28. See ibid, page 51. attention was given to social trends (income structure, leisure time, Tourism Management September 1982 163 Can futures research contribute to tourism policy? the role of women, norms and values), as well as to trends in demand and environmental settings. While in the forementioned Washington exercise emphasis was laid on the societal context (our fourth category of tourism), the main core of the European studies (Austria and Switzerland) is found in the second and third categories (intermediate framework and the supply of tourist resources and facilities). In 1983 a Delphi, aimed to embrace all four categories, is due to be conducted in the Netherlands. That study will undoubtedly be indebted to the earlier studies. Integrative forecasting As was mentioned above, normative forecasting studies are almost completely absent in tourism research. This can be attributed partly to the vagueness of the concept of tourism. If one does not know exactly what tourism stands for, how can one develop a policy strategy for it? Second, in various Western European countries tourism is not attached to one single ministry but to several. In Holland at least two important ones (Economic Affairs, and Social and Cultural Affairs) and at most five claim a certain competence in questions of tourism. Last but not least the sector is denied political importance, notwithstanding its vital economic importance. Even in the U S A this is the case if we are to believe senator Daniel Inouye: Of course, neither the Administration nor its predecessors would consciously and deliberately ignore or otherwise frustrate an industry that contributes so much to our social and economic welfare. The answer has been and remains that as a Government we do not understand the industry, its magnitude and the size of its contribution to the nation's social and economic goals. 29 29. See reference 24, page 39. 30. M.O. Falani, "Air traffic forecasting: an input-output technique approach", Regional Studies, 7, 1973. 31. BarOn, op cit, reference 15; G. Taylor and M. Doctoroff, An approach to an integrated forecasting system for a national tourist office", in IUOTO, The Measurement of Tourism (British Tourist Authority, London, 1975); D. Edgell and R. Seely, "Tourism policy: a two stage model forthedevelopmentofinter- national tourism flow forecast estimates", paper presented to the Symposium, "Tourism in the next decade", Washington, DC, 1979; see also reference 2 1. 164 This attitude can be held partly responsible for the nonexistence of normative forecasting. Remembering what was said above, this form of forecasting contains integrated forecasts from various sectors, from different forecasting methods, preferably long-term studies. Yet in the course of m y paper reference has been made to the poor feasibility of longterm studies in tourism. However, there are a few good studies that comply with some criteria that define integrative forecasting. Falani's study 3° on forecasting part of the air traffic between 14 US cities through an input-output model has turned out to be so important for airport directors, airlines, pressure groups, cateringand fuel-suppliers, that it could be characterized as integrative at least in a sense. Taylor, Edgell and Baron 3~ similarly emphasize the need for integrating several techniques in one comprehensive method. The combination preferred by them and m y s e l f is the triad " T i m e SeriesDelphi-Scenario writing", in that order. Results Despite the growing file of reports on tourism forecasting, surprisingly little attention is paid to the comparison of actual data with the corresponding forecasts; and this, despite the existence of a considerable number of criteria to assess/evaluate these results. The scope of m y paper only allows for a brief mention of a few of these. Tourism Management September 1982 Can futures research contribute to tourism policy? Particularly in short-term forecasting, one should be happy when forecasts materialize. That could be proof that the technique used was valid to some extent. Usually, validation tests of post-diction and proximate variety are used. The former tries to validate historical data from one period as a function of historical data from the preceding period. The latter applies a score to the success of short-term forecasts, when materialized, in view of the long-term forecasts (see Figure 5). In this context we must emphasize the role played by criteria of plausibility (versus causation) and logical consistency (versus systems or societal consistency). But even the role, function and random behaviour of a researcher have an impact on the output. Other criteria to assess the usefulness of forecasting results could be the number of alternatives presented, the contextual stability of the forecast, the presentation itself and last, but not least, the costs. So the policymaker in tourism, whatever his place is, in government, industry or elsewhere, always has to counterweigh different criteria against his own objectives and preferences - - he will have to ask questions like: • What will the forecast cost me? • What is the term of application and what is the relative value of this term? • What data are available? • How valid is the technique proposed? • Do I need alternatives or just one answer? • Are the results plausible? • Can I use this forecast to impress my electorate or can it be used to manipulate investors, political leaders or the public? • Do I really need a forecaster? Or could I do it better myself2. Doubtless in most cases policymakers need the forecaster, but they should not blindly depend on his rulings. Decisions ought to be the result of communication, perhaps even directly embedded in a communicative or negotiation-type of planning, before both parties will benefit from one another in an optimal way. Conclusions: the bearing of forecasting on tourism In this paper we have presented a typology of tourism on the one hand, and a typology of futures research (ie forecasting) on the other. Most studies, briefly mentioned or quoted, focused on the first category of both typologies, the tourist and exploratory forecasting, or were concerned with the supply category (large companies such as in the aviation industry have their own forecasting departments). The intermediate framework proved to be less covered by research, due to its inherent dispersion and fracturing, with the exception perhaps of some big hotel chains or united travel organizations. In the last category we witnessed the first fruitful signs of combined f o r e c a s t i n g - scenario writing. Figure 5. Post-diction and proximate variety: validation tests in a time perspective Key: post-diction: a = forecast for b proximate variety: c with in accordance d Tourism Management September (d) (a) I ~ t_z 1982 (b) I ~ t., Short f e r m i~( c ) " ~"=1 ~--~'~ to ~ ' ~tr ~ . . ~ . ~ Lonq I I I I tr tz t3 t4 165 Can ftttures research contribute to tourism polic3'? Several Delphi studies which together would cover all four tourism categories, at least could have an impact on tourism and recreation policy. Yet maximum benefit could be derived from this technique when used in combination with time series and scenario writing. But, on the whole, forecasting in tourism has not yet received universal recognition as a vital aspect of planning and/or policymaking in tourism. The greater part of research in tourism refers to short-term exploratory forecasting and therefore will not substantiate the necessary support to strategic planning, while on the other hand the few 'pre-stage' scenarios presented to date move more along the lines of contingency planning than in comformity with scenarios proper. Yet in a period of continuous economic recession it is of great importance to view, through crises and disasters, a more hopeful future. However, that means a type of planning and decisionmaking that bases itself on strategic planning (or management). This planning is not widespread among the various tourism bodies and organizations. But whenever one bases oneself on norms and values set by politicians and policymakers together with their natural partners in the tourism field, integrative forecasting is needed, since this type of forecasting combines various social, political and economic trends in a normative policy framework. Recreation stands and falls with society and the forms of tourism developed in this context. It is determined by factors such as income level, leisure time, energy supply and prices, processes of individualization and social segregation, as well as by inflation, demographic structures and ecological imbalances such as p o l l u t i o n - and, not least, by the changes in all those factors or variables through time. All those aspects and factors will grow in complexity and dynamism. Besides, there is an increasing interdependence of the elements that make up the tourism and recreation system. In the decade to come tourism is bound to remain a fuzzy set. Forecasting has proved to be valid to some extent in economics, technology, demographics and a few other fields. I have tried to demonstrate in the course of the paper that it is also a handsome tool for planning and policymaking in tourism. It thus seems fair to declare forecasting to be one of the fatally neglected but predictably most vital parts of tourism and recreation policymaking in this decade. 166 Tourism Management September 1982