Loss aversion and high stakes

Abstract

AbstractFollowing Kahneman and Tversky, studies such as those by Haigh and List, Ert and Erev and Mukherjee et al. find that loss aversion increases as stake increases. This study extends the 2011 work of Berger and Pope by analysing over 68,000 US professional basketball games played from the initial National Basketball Association (NBA) season in 1946/1947 to the 2018/2019 season, and over 69,000 National Collegiate Athletic Association (NCAA) games played from the 2007/2008 season to the 2018/2019 season. We posit that, a priori, stakes, and therefore loss aversion, will be greater for NBA teams than for NCAA teams, and higher for home teams than for away teams. Further, loss aversion is expected to be greater for favourites, that is, teams that are expected to win. We model outcomes using a digital call option. This model allows for necessary non‐linearity in the relation between halftime score and winning percentage. It also provides an analysis for which the result for home (favourite) teams is not simply the converse of that for away (underdog) teams. We find evidence of better‐than‐expected performance for NBA home teams that are behind by up to four points, and for favourites that are behind by between two and seven points. We find no evidence of this effect with respect to NBA away teams, NBA underdogs, nor for NCAA teams – whether home or away. Our results suggest that loss aversion is apparent when stakes are high.