Forecasting Tourism Demand: Methods and Strategies
Sample Chapter, continued
Indeed, the discovery by German physicist Wemer Heisenberg in the 1920s renders this approach futile. His 'uncertainty principle' states that it is impossible to determine both momentum and position of a subatomic particle at the same time, only the probabilities of each (Paulos, 1991: 119-20). One consequence is that there is no deterministic theory that we can rely on to precisely forecast tourism futures of interest to us.
At the other end of the spectrum, as indicated in Figure l. l, are those future events that are essentially random, that is, each possible occurrence of an event has the same probability. Flipping a coin or choosing a card from a well shuffled deck are examples. The numbers selected in a lottery are designed to be random as well. In tourism, whether the next person to enter a restaurant is a male or female is a random event under most circumstances.
1. The future is totally predictable (i.e., unalterable) implying sound forecasts are useless.
2. The future is totally unpredictable (i.e., random) implying sound forecasts are impossible.
3. The future is somewhat predictable and somewhat alterable implying sound forecasts are useful and feasible.
These quite uncertain events are, by definition, impossible to forecast with any acceptable degree of accuracy. Consequently, forecasting them is not a worthwhile endeavor.
Fortunately, there is an alterative to these pessimistic views of the future that gives hope to forecasters. That is, future events important to tourism operations are somewhat predictable and somewhat changeable. We can predict events with probabilities significantly greater than zero and markedly less than 100 per cent. And these events can be affected by other events, including our own actions. This is the view adopted in this book.
This is also the hope of marketers and managers: that we can infer enough about the future to choose certain actions to shape it toward our preferences. Some call this 'inventing the future'. We obtain these inferences from reviewing the past. A forecasting method is simply a systematic way of organizing information from the past to infer the occurrence of an event in the future.
The two extreme views suggest a warning. we make a mistake if we invest heavily in trying to achieve a near perfect forecasting method. If we can achieve such, the results will be useless. Rather we must expect that our forecasts are not going to hit the mark each time, and this is good news for those trying to invent the future. The other caveat is that we can find a forecasting method that will tell us something useful about future tourism in most cases. That is, it will increase our probability of making an accurate forecast.
Bernstein (1996:7) points out that these two extremes are independent of Whether we try to quantify past patterns or use subjective means of indicating possible futures. Even if we could build a mathematical model that accurately reflects past behavior we would have no reason to believe that a true knowledge of the future is in our hands. Bernstein (ibid.) quotes Nobel laureate Kenneth Arrow (1992):'our knowledge of the way things work, in society or in nature, comes trailing clouds of vagueness' Vast ills have followed a belief in certainty'. Tourism forecasters beware!
Forecasting definations
Forecasting is fundamentally the process of organizing information about a phenomenon's past in order to predict a future.
A phenomenon is simply, A fact or event that appears or is perceived by one of the senses or by the mind' as The New Shorter Oxford English Dictionary (Brown, 1993), defines the word. We can organize information about a phenomenon's past in many ways' One way is to manipulate objective, quantitative data by mathematical rules' Another is to analyze the opinions of experts about the phenomenon, past and future. Both of these classes of forecasting methods are treated in this book.
Many who study the future believe as Herman Kahn did' that there are .alternative futures, that the future is not a single inevitable state, but change can evolve in strikingly different ways' (Coates and Jarratt, 1989: xi). In effect we can .invent' a future by making changes in the present. Forecasting, then, allows us to predict a single future or a set of futures, each associated with a different set of postulated changes.
In summary at its most basic, forecasting. Takes historical fact and scientific knowledge . . . to create images of what may happen in the future' (Comish,|911:51).This book describes much of the realm of scientific Knowledge that has been applied to tourism demand in the last quarter of a century.
This book focuses on ways of forecasting the behavior of tourism markets, that is, demand for some tourism product' This product may be a hotel room, a restaurant meal, a visit to a destination, a day at Walt Disney World, or even a trip away from home. In most cases, we want to know how many consumers there will be, how many units will be sold, how much will be spent on the purchases, or any combination of these.
This book will use the term, tourism demand forecasting, to indicate a process designed to reduce the risks of tourism marketing and other management decisions through the use of forecasting. This is assumed to be synonymous with tourism market forecasting.
Other Definitions
There are other terms that are not specifically related to tourism but are basic to understanding and discussing forecasting techniques:
Data point: an individual value in a time series.
Data series: same as a time series.
Forecast time series: a time series of future values produced by some method.
Historical time series: refers to the time series of past values.
Observation: same as a data point reminding us that each data point must be observed and measured, introducing the possibility of error.
Time series: an ordered sequence of values of a variable observed at equally spaced time intervals.
Variable: any phenomena that can be measured; usually refers to all of the data points associated with it.
It is important to understand the special use of the word 'past' in the above definitions. To a forecaster, the past includes all periods for which reasonably final values are available. Future time, on the other hand, includes time that has passed for which we do not yet have reasonably complete and accurate data, as well as time not yet encountered. Some time series of interest to tourism forecasters may run three to six months behind actual time, so that we may not know what happened to tourism demand in 1999 until 2001. This produces the odd (to non-forecasters) habit of forecasting periods that have passed us by but for which relevant measures are not available. This is the use of 'future values' that is implicitly adopted in this book: if the measure has not been developed for it, then the time period lies in the future for forecasting purposes.
Additional terms will be defined as they arise in this book. The glossary at the end of the book is designed to provide a complete listing of important terms used herein and their definitions.