Market‐wide overconfidence and stock returns

• Published date: 01 January 2024
• Author: Qiang Chen,Yu Han,Ying Huang
• Date: 01 January 2024
• DOI: http://doi.org/10.1002/fut.22462

Received: 16 April 2022

|

Accepted: 9 September 2023

DOI: 10.1002/fut.22462

RESEARCH ARTICLE

Market‐wide overconfidence and stock returns

Qiang Chen

1

|Yu Han

2

|Ying Huang

2

1

School of Economics and Department of Digital Economy (Digital Economy Research Institute), Shanghai University of Finance and Economics,

Shanghai, China

2

School of Economics, Shanghai University of Finance and Economics, Shanghai, China

Correspondence

Yu Han, School of Economics, Shanghai

University of Finance and Economics,

Shanghai, China.

Email: hanyu@163.sufe.edu.cn

Funding information

Fundamental Research Funds for the

Central Universities,

Grant/Award Number: 2023110139

Abstract

In this paper, a novel measurement of overconfidence over the market is

developed based on the size of ambiguity (the confidence of investors in

information). The proposed measure of market‐wide overconfidence is

consistent with the predictions motivated by prior literature. It has a

significant negative association with the next‐month market excess return.

Associations between the overconfidence measure and riskier portfolio returns

behave stronger and last longer, implying a risk‐taking proclivity of

overconfident investors.

KEYWORDS

ambiguity, market excess return predictability, overconfidence measurement

1|INTRODUCTION

The extant literature provides extensive evidence for the existence of overconfidence in the financial markets (e.g., Chou &

Wang, 2011;Chuang&Lee,2006;Statmanetal.,2006)andtheeffectsitbrings(e.g.,Danieletal.,2001;Odean,1998;

Scheinkman & Xiong, 2003). However, prior empirical studies mostly focus on the implications contained in the

overconfidence hypothesis. To the authors' knowledge, there have been no indicators reflecting the aggregate level of

overconfidence over the market. In this paper, a novel measure is proposed to capture the size of market‐wide overconfidence.

Overconfidence is viewed as investors overestimating the precision of their information (Odean, 1999; Scheinkman

& Xiong, 2003). They believe that their information is more accurate than it is. Therefore, an intuitive measurement of

overconfidence degree is to measure the difference between the accuracy of investors' information they believe (or the

confidence they have) and the actual accuracy of the information (or the confidence they “should”have). In the strand

of literature related to decision‐making, ambiguity emerges in situations wherein the probabilities of each relevant

event that may occur are a prior unknown, implying that decision‐makers are uncertain about the data‐generating

process itself.

1

Accordingly, the degree of ambiguity can reflect the accuracy of the information and is usually

associated with decision‐makers' confidence in their information.

2

Ellsberg (1961) informally described ambiguity as

the nature of one's information concerning the relative likelihood of events …a quality depending on the

amount, type, reliability and ‘unanimity’of information, and giving rise to one's degree of ‘confidence’in

an estimation of relative likelihoods. (p. 657)

J Futures Markets. 2024;44:3–26. wileyonlinelibrary.com/journal/fut © 2023 Wiley Periodicals LLC.

|

3

1

Ambiguity is different from risk. Risk refers to a situation with a single certain probability distribution and uncertain outcomes.

2

If decision‐makers are absolutely confident about their information, they will be certain about a single data‐generating process, and the degree of

ambiguity is zero.

Numerous studies link ambiguity to confidence. Ilut and Schneider (2014) model shocks to agents' confidence about

the future as changes in ambiguity. Zhao (2017) interprets the size of ambiguity as investors' confidence. Brenner et al.

(2015) provide a model and experimental evidence to show that ambiguity affects the decision‐maker's confidence and

suggest that ambiguity may play an important role in explaining overconfidence behavior. Consequently, expected

ambiguity at the beginning of 1 month can gauge how accurate investors believe their information about the upcoming

month is or how confident they are, and actual ambiguity at the end of this month can gauge the actual accuracy of the

information or the confidence investors “should”have. Then, it is natural to measure the overconfidence (i.e., the

overestimation of the information precision) as the difference between the end‐of‐month actual ambiguity and the

beginning‐of‐month expected ambiguity.

3

This measure thus reveals how investors are confident about their

information on this month (specifically, this month's return distribution model in our study).

In this paper, the methodology proposed by Brenner and Izhakian (2018) is applied to calculate the ambiguity in the

stock market and options market about the return distributions. According to Izhakian (2020), this measure is

independent of risk and attitudes and is empirically applicable. It has been used in several empirical research

associated with stock market ambiguity (e.g., Augustin & Izhakian, 2020; Brenner & Izhakian, 2018; Izhakian &

Yermack, 2017; Izhakian et al., 2022). We extend this method to the options market, considering the forward‐looking

feature of the options market, to help extract the expected ambiguity. The ambiguity in the options market thus

represents the expected ambiguity in the market, while the ambiguity in the stock market represents the actual

ambiguity. Therefore, the market‐wide overconfidence measure can be obtained. Existing studies also develop other

approaches to measure the size of ambiguity (e.g., Driouchi et al., 2020; Viale et al., 2014), but we argue that the

method applied in Brenner and Izhakian (2018) is much more convenient to be implemented in the stock market and

options market simultaneously.

The values of the proposed overconfidence measure time‐series in this paper are mostly positive, indicating that

investors in the market are normally overconfident. This is consistent with the argument of existing literature like

Daniel et al. (1998) and Hirshleifer and Luo (2001) that overconfident investors can survive and dominate the markets

eventually. We first investigate the relationship between the proposed market‐wide overconfidence measure and the

market trading activity. Gallant et al. (1992) and Hiemstra and Jones (1994) use aggregate share volume to measure

trading activity. However, with the development of the financial market, the number of outstanding shares has risen

noticeably, making raw share volume a poor measure of trading activity. Lo and Wang (2000) and Statman et al. (2006)

argue that turnover is an appropriate indicator for trading activity. Therefore, we use the turnover of an exchange‐

traded fund designed to track the Standard & Poor 500 (S&P 500) stock market index, the iShares Core S&P 500 ETF

(IVV) (2000–2021), to represent the market trading activity. In line with previous studies like Gervais and Odean (2001)

and Chuang and Lee (2006), we document a positive association between trading activity and the proposed measure. If

the market‐wide overconfidence measure increases, the market will trade more aggressively over the current month.

This conclusion provides support that this measure can reveal the size of market‐wide overconfidence to some extent.

We next examine the influence of overconfidence on stock market excess returns. Previous studies like Barber and

Odean (2000), Barber and Odean (2001), and Hsu and Shiu (2010) use data sets from brokerage firms to conclude that

overconfident individuals tend to perform poorly in investment. In contrast to previous research, we concentrate on

aggregate market returns. The empirical evidence in our study suggests that market‐wide overconfidence is also

hazardous to future market excess returns. An increase in overconfidence leads to a decrease in the next‐month market

excess returns. We argue that this pattern is a mispricing correction. A high level of market‐wide overconfidence

implies that investors overestimate themselves and tend to make aggressive decisions, causing prices to rise

temporarily. When the overconfidence fades, however, prices revert to the fundamentals, implying a negative

relationship between the size of overconfidence and next‐month market returns. We also find that the negative

relationship between our proposed overconfidence measure and the next‐month market returns is more robust in

economic expansion periods. Besides, an out‐of‐sample test is conducted to show that the proposed measure indeed

carries incremental information about the future market excess return.

In addition, prior literature suggests that overconfident investors have a proclivity toward taking more risks and

trading relatively risky assets (e.g., Chuang & Lee, 2006; Liu et al., 2010). With the proposed overconfidence measure,

we directly examine the relationship between our measure and different portfolio returns to provide new evidence for

3

If the expected ambiguity is much less than the actual ambiguity, that means decision‐makers overestimate the precision of their information more

and tend to be more overconfident.

4

|

CHEN ET AL.