The age-performance relationship for a cognitive-intensive task: Empirical evidence from chess grandmasters

Abstract

To gain insight into the age-performance relationship (APR) for cognitively demanding tasks, we analyse the APR for the task of playing chess from a theoretical and empirical perspective. We set up a game-theoretic model showing that for risk-neutral players who are homogeneous in their linear payoffs, the APR can be estimated with either players' game outcomes or their Elo ratings. This result is empirically substantiated with data on outcomes of games played at an annual international chess tournament (1970–2021), and with players' Elo ratings. Further, the findings support the estimation of the APR with Elo ratings using a model that controls for player fixed effects and period random effects, which is also in accordance with the theoretical model. Next, we show that on average chess performance peaks in grandmasters’ early-thirties and declines thereafter. Implications of the findings for a labour market characterised by an increase in cognitively demanding jobs and individuals working longer are discussed.