|
|
|
| The Learning Mechanism of Chinese Institutional Investors: Rational Bayes or Simple Reinforcement Learning？ |
| SHAO Xinjian, WANG Huiqiang, WANG Xingchun, QIN Jiaqi
|
School of International Trade and Economics, University of International Business and Economics;
School of Finance, Nankai University |
|
|
|
Abstract The theoretical literature on financial economics usually assumes that rational investors follow the Bayesian learning method—that is, if an investor's personal experience cannot contain new information, it cannot affect the investor's belief-updating process. Professional institutional investors have inherent advantages over individual investors in terms of information acquisition and processing, investment decision-making mechanisms, and internal governance methods, and are therefore considered rational investors. However, a key question remains: do professional institutions follow rational Bayesian learning or reinforcement learning in their decision-making on investments？
Given the random allocation of investors' profit and loss generated by the lottery system used to issue initial public offering (IPO) investments in China, this paper uses IPO lottery allocation data provided by the Shenzhen Stock Exchange to empirically test the effect of experience on the decision-making behavior of institutional investors. First, it tests whether institutional investors' previous earnings experience affects their tendency to participate in subsequent IPOs. Then, it tests whether the previous loss experience of institutional investors affects their quotations for subsequent IPOs. Finally, it examines whether the personal characteristics of fund managers can influence the effect of previous loss experience on institutional behavior.
The main conclusions of this research are as follows. (1) Institutional investors do not completely follow a rational Bayesian learning method but do follow a naïve reinforcement-learning method. Specifically, institutions that are allocated the winning bid at an early stage are significantly more likely to participate in subsequent IPO subscriptions than institutions that are not. In addition, the higher the rate of return of new shares issued in the previous period, the stronger the positive relationship between the allocation experience and subsequent participation in IPO subscriptions. (2) The income experience of institutional investors promotes the renewal of their beliefs and produces more optimistic valuation beliefs than pessimistic valuation beliefs. Specifically, the higher the return on new stocks in the previous period, the higher are the price increases given by the institutions that receive stock allocations through lotteries when they participate in the valuation of subsequent new shares, and these price increases are significantly higher than those given by the unsuccessful control group. (3) Experience may influence institutional behavior via the intensive learning process of institutional investment managers. Regarding the personal characteristics of fund managers, research on the mechanisms of adjustment of institutional behavior shows that rich long-term experience, high education level, and competition between multiple fund managers moderate the effect of profit and loss experience on institutional behavior.
The main contributions of this research are as follows. (1) This research focuses on the experience of professional institutional investors, who are typically regarded as rational investors, and finds that the behavior of institutions is significantly affected by a naïve reinforcement-learning process. Thus, a good result reinforces the behavior that an organization takes to achieve the result and may cause its valuation beliefs to become relatively more optimistic. (2) Empirical research on experience and behavior usually faces the endogenous problem of investor experience (Choi et al., 2009; Chiang et al., 2011). This article instead draws on the random test opportunities provided by the lottery system in China's IPO distribution (as these better alleviate the endogenous problems of experience) and obtains more credible inferences about causality than other studies. (3) Mainstream studies in the field of IPOs usually focus on the issue and pricing efficiency of the inquiry system and auction system, and theoretical and empirical studies usually ignore the effect of institutional investors' learning mechanisms on pricing efficiency. In contrast, based on research on the background of China's IPO auction system, this paper finds that an institution's experience has a significant effect on its future IPO purchasing tendency and price level. This provides a new perspective for re-examining over-quoting and the “uncertainty of investor participation in the IPO auction system,” which is a concern noted in the theoretical literature (Sherman, 2005; Jagannathan et al., 2015).
|
|
Received: 17 December 2018
Published: 09 May 2022
|
|
|
|
|
|
|
|
