Learning about unprecedented events: Agent-based modelling and the stock market impact of COVID-19

Davide Bazzana; Michele Colturato; Roberto Savona

doi:10.1016/j.frl.2023.104085

Learning about unprecedented events: Agent-based modelling and the stock market impact of COVID-19

Financ Res Lett. 2023 Sep:56:104085. doi: 10.1016/j.frl.2023.104085. Epub 2023 Jun 8.

Authors

Davide Bazzana^{1

2}, Michele Colturato³, Roberto Savona⁴

Affiliations

¹ Department of Economics and Management, University of Brescia, via San Faustino 74/b, 25122 Brescia, Italy.
² Fondazione Eni Enrico Mattei, Corso Magenta, 63, 20123 Milan, Italy.
³ Department of Mathematics, University of Pavia, Via Ferrata, 5, 27100 Pavia, Italy.
⁴ Department of Economics and Management, University of Brescia, C/da S. Chiara 50, 25122 Brescia, Italy.

Abstract

We model the learning process of market traders during the unprecedented COVID-19 event. We introduce a behavioural heterogeneous agents' model with bounded rationality by including a correction mechanism through representativeness (Gennaioli et al., 2015). To inspect the market crash induced by the pandemic, we calibrate the STOXX Europe 600 Index, when stock markets suffered from the greatest single-day percentage drop ever. Once the extreme event materializes, agents tend to be more sensitive to all positive and negative news, subsequently moving on to close-to-rational. We find that the deflation mechanism of less representative news seems to disappear after the extreme event.

Keywords: Agent-based model; Asset pricing model; Heterogeneous expectations; Representativeness; Unprecedented events.