Data Disaggregation in Demand Forecasting

Abstract

It is assumed in tourism demand forecasting that the disaggregation of data is useful in terms of country of origin and also in terms of purpose of travel (Smith and Toms, 1967; Blackwell, 1970; Martin and Witt, 1989a). The primary disaggregation by country is useful for determining regional forecast flows and the disaggregation by purpose of visit has been considered potentially useful for increasing forecasting accuracy given the flows have different characteristics (Turner, Kulendran and Pergat, 1995; Morley and Sutikno, 1991). It is also possible to disaggregate on the basis of age and gender. It has been assumed (because no research has disaggregated by age and gender) in previous research that the gender and age composition of flows is a reflection of the total population and therefore exhibits the same time-series characteristics. This may not be the case, however.

This study uses data for tourist arrivals into Korea to test the assumptions that further disaggregation of data on the basis of gender and age is not needed, and to further examine whether disaggregation by purpose of visit is worthwhile, when the purpose is to forecast total country arrivals. Quarterly data from 1994 to 2003 are used with the estimation period 1994–2001 and the post-estimation period 2002–2003. The conclusion from the study is that total arrivals forecasting is not more accurate when the data used is the sum of forecast disaggregated series, as opposed to direct forecasts of total arrivals.