Detecting the Olympic tourism legacy: a data-driven approach

This research applies a data-driven approach for identifying changes in time series to model the Olympic tourism legacy. Measuring potential legacy effects from mega-sporting events has been problematic in prior research. Issues revolve around how to measure tourism, how to control for pre-existing trends, and how to account for extraneous events affecting tourism unrelated to the Olympics. Most problematic for analytical models is determining when the tourism legacy begins and what is the functional form of the tourism legacy. Many of these issues interact and can confound results leading to erroneous conclusions. The seminal methodology developed by Tsay [(1988). Outliers, level shifts, and variance changes in time series. Journal of Forecasting, 7(1), 1–20] requires no prior assumption about the timing or functional forms of the outliers, therefore solving these issues and provides a framework that can be used when analysing mega-sporting event legacies. Using this methodology, the research finds limited support for a short-term Olympic tourism legacy and no support for a long-term tourism legacy.