Forecasting Tourist Arrivals
This is the first in a series of articles written to show how Economics can be useful in business. The focus here is on using a multiple regression analysis to develop a predictive model. In this case it was for a tourist arrivals. The approach lends itself to many other applications.
Tourism is big business. It is the lifeblood – directly or indirectly for many firms. The importance goes beyond business – tourist destinations thrive or whither on arrivals. The Greek isles, West Indies, Seychelles, Mauritius and Cape Town are obvious examples.
Tourism can be a difficult business. A recession in a source country can decimate arrivals. Growing competition from other destinations, albeit slower, is also a challenge. Even conditions at the destination – tsunamis or droughts – affect perceptions and arrivals.
So, if you are in that business, or a government that depends on tourists, you need to anticipate possible future changes. This allows you position yourself to limit short term damage or benefit from booms and address the challenges in the longer term.
Economics can help.
This short note outlines an approach used by StratEcon to bring science to forecasting tourism arrivals by developing an econometric based forecasting model. This is, conceptually, straightforward and there are three steps.
First, is to identify important origin countries – the countries that tourists come from. This kind of information is readily available. This country selection can also be used strategically so that countries which are currently targeted as new markets or are part of the opening of new air routes, for example, can also be included. The purpose is to give insight into how internal and external factors affect these targeted tourist arrivals.
Second, is identify and measure tourist pull/push factors.
Pull factors are those conditions that are unique in the destination (where your business is) that attract international tourists. For the most part, these are given and change slowly. The factors that are important are those that can change rapidly and affect tourist attitudes and therefore arrivals. These may include, for example, a drought or adverse political changes. They would also include changing attractions of competing destinations.
Push factors are those in the origin country that give impetus to out-bound tourism. These are likely to include change in income, relative exchange rates, extreme weather events (very cold Northern winters and the need for sun shine), etc.
In economic jargon these push and pull factors are the so-called ‘independent’ variables. They are used in a step-wise regression analysis to determine the impact of each independent variable on the dependent variable. The dependent variable is the tourism arrivals from an origin country. The equation would be expressed as:
Tourist Arrivals (Germany)
=f(β1_GDP(GM),β2 €/(Local Currency),β3_ECB Interest Rates,β4 …..,u)
The step-wise regression would be run for all selected source countries (with Germany as an example). The β1 to βn (betas) are the correlation coefficients of each independent variable to the dependent variable. In other words, β1 explains the degree to which changes in German GDP affects German arrivals in the tourist destination – where your business is located. Other independent variables are likely to include exchange rates, interest rates, and the attractiveness of competing destinations.
There is a health warning – the analysis is conceptually straightforward. The detailed analysis does need some specialist knowledge to include, for example, all standard econometric tests to identify multi-collinearity and autocorrelation.
Third, is to develop an econometric forecast model using the beta estimates of all independent variables for all origin countries. So, for example, the regression betas may be estimated on data between 2000 and 2018. The forecast part of the model would then use real-time information – GDP in 2019 – to forecast tourist arrivals by the end of 2019.
Outputs (this is the important bit). There are, at least, two important outputs.
Forecast tourist arrivals and expenditure. As the information on the independent variables becomes known – German 2018 Q1 and Q2 GDP, exchange rate, etc – they are used with the relative betas to forecast likely tourist arrivals from Germany over the next 12 months. These, in conjunction with known spending patterns, and possible changes, would give an expenditure estimate as well. The forecast would be taken further by using forecasts of the independent variables for Germany. There are many institutions, like the IMF and OECD, that make forecasts of country GDP, interest rates, exchange rates, etc. These would be used to forecast tourist arrivals over the next 12 to 36 months. The same process would be followed for all selected origin countries. The results would give a forecast of expected, total, tourist arrivals and expenditure in your region over the next three years.
Scenario or ‘What If’ Analysis. A completed forecasting model of this nature is a perfect tool for a scenario or ‘what – if’ analysis. There are many examples of such ‘what – if’ questions. For example, what is the likely impact on tourist arrivals of a Italian sovereign debt default or a German recession? Another example would be an understanding of the consequence of changes at the destination – political shocks or extreme weather conditions.
I hope this short guide helps you grow your tourist reliant business, resort, island or country.