Asymmetric Reinforcement Learning and Conditioned Responses During the 2007–2009 Global Financial Crisis: Evidence from Taiwan

Abstract

This study applies self-developed models using standardized regression analysis to investigate the learning behavior associated with investor sentiment and psychological pitfalls during the 2007–2009 Global Financial Crisis. The empirical results, which draw on reinforcement learning and classical conditioning perspectives, reveal asymmetric reinforcement learning behavior because investors report a higher tendency to avoid past unfavorable outcomes rather than repeat past favorable outcomes. The evidence also supports the conditioned response of psychological pitfalls to the Taiwanese government’s stimulus measures during 2007–2009; however, no concrete evidence supports the conditioned response of investor sentiment.