Abstract

The increase in external shocks to tourism over the past two decades has focussed attention on modelling and forecasting tourism demand cycles. This research aims to determine whether the inclusion of business cycle dynamics can improve the forecasting accuracy of the tourism demand cycle in South Africa – a developing country and tourism destination. The methodology includes a Markov switching (MS) dynamic approach, with forecasts compared to time-series approaches. The findings show that the MS and MS dynamic models are always the best or second-best forecasts over all time horizons for tourism demand from South Africa’s five main non-African markets. Keywords: tourism forecasting, business cycle, Markov switching model