Tourism demand modelling and forecasting: How should demand be measured?

Haiyan Song, Gang Li, Stephen F. Witt, Baogang Fei

Research output: Journal article publicationJournal articleAcademic researchpeer-review

202 Citations (Scopus)


Tourist arrivals and tourist expenditure, in both aggregate and per capita forms, are commonly used measures of tourism demand in empirical research. This study compares these two measures in the context of econometric modelling and the forecasting of tourism demand. The empirical study focuses on demand for Hong Kong tourism by residents of Australia, the UK and the USA. Using the general-to-specific modelling approach, key determinants of tourism demand are identified based on different demand measures. In addition, the forecasting accuracy of these demand measures is examined. It is found that tourist arrivals in Hong Kong are influenced mainly by tourists' income and 'word-of-mouth'/ habit persistence effects, while the tourism price in Hong Kong relative to that of the tourist origin country is the most important determinant of tourist expenditure in Hong Kong. Moreover, the aggregate tourism demand models outperform the per capita models, with aggregate expenditure models being the most accurate. The implications of these findings for tourism decision making are that the choice of demand measure for forecasting models should depend on whether the objective of the decision maker is to maximize tourist arrivals or expenditure (receipts), and also that the models should be specified in aggregate form.
Original languageEnglish
Pages (from-to)63-81
Number of pages19
JournalTourism Economics
Issue number1
Publication statusPublished - 1 Jan 2010


  • Forecasting accuracy
  • Hong Kong
  • Tourist arrivals
  • Tourist expenditure

ASJC Scopus subject areas

  • Geography, Planning and Development
  • Tourism, Leisure and Hospitality Management


Dive into the research topics of 'Tourism demand modelling and forecasting: How should demand be measured?'. Together they form a unique fingerprint.

Cite this