Abstract
This study constructs composite leading indicators to predict turns in the growth cycle of hotel occupancy for hotels in Hong Kong with different tariffs, as well as for all hotels in Hong Kong. First, a composite leading indicator for each of the top five source markets - mainland China, Taiwan, Japan, the USA and South Korea - was constructed using selected economic variables. Second, a composite leading indicator for all hotel categories in Hong Kong was constructed by combining the leading indicators constructed for each source market, using tourism market share and the cross-correlation coefficient as weightings. Third, the combined composite indicator was compared with the published OECD composite leading indicator and the OECD business survey index, which were constructed using the same weighting methods. In order to identify the best weighting method and to select the best composite leading indicator for different tariff categories of hotels, this study assessed the probability forecasts from the logistic regression leading indicator models using the quadratic probability score (QPS). The result shows that the composite leading indicator combined with tourism market share provides more accurate forecasts than the composite leading indicator combined with the coefficient of correlation.