Chapter 16: When Policy Conflicts with the Data Science: Predicting the Cost Overrun of the 2032 Brisbane Olympic Games

No Olympics in recent history have come in on or under budget, with past Olympics having overrun their budgets by an average of 172% (Flyvbjerg et al., 2021). Growing concerns among taxpayers about the expense have led many cities to drop out of the bidding process for the Olympics: for the 2024 Olympic Games, bids were withdrawn by Boston, Budapest, Hamburg, and Rome (Okada & Greyser, 2018). Sports infrastructure, including the main stadium, is a major cost category for the Olympic Games. Building or renovating venues to Olympic specifications can leave cities with heavy costs of long-term maintenance as well as the cost of initial construction (Baade & Matheson, 2016). In practice, cities tend to overestimate the size and suitability of existing venues, leading to underestimated costs and budget overruns (Preuß et al., 2019). The Brisbane 2032 Olympic bid says that Brisbane’s Olympics are confident of breaking even, and avoiding building stadiums that end up as white elephants. But does the choice of Olympic stadium make a difference to the final cost overrun? This study finds that cost overruns for past Olympics have been predicted using a few characteristics of the main stadium: its age, its distance from the city centre, and its seating capacity. This has important policy implications for selecting stadiums to be used at the Olympic Games and particularly for the Brisbane 2032 Olympic Games where a stadium has been earmarked for selection without data-driven evidence to support the decision.