Investor sentiment and the risk–return relation: A two‐in‐one approach

• Published date: 01 January 2024
• Author: Darren Duxbury,Wenzhao Wang
• Date: 01 January 2024
• DOI: http://doi.org/10.1111/eufm.12427

DOI: 10.1111/eufm.12427

ORIGINAL ARTICLE

Investor sentiment and the risk–return

relation: A two‐in‐one approach

Darren Duxbury

1

|Wenzhao Wang

2,3

1

Newcastle University Business School,

Newcastle University, Frederick

Douglass Centre, Newcastle Helix,

Newcastle upon Tyne, UK

2

The Business School, Edinburgh Napier

University Craiglockhart Campus,

Edinburgh, UK

3

School of Finance, Shandong University

of Finance and Economics, Jinan,

Shandong, China

Correspondence

Wenzhao Wang, The Business School,

Edinburgh Napier University

Craiglockhart Campus, 219 Colinton Rd,

Edinburgh, EH14 1DJ, UK.

Email: w.wang@napier.ac.uk

Abstract

Traditional finance theory posits a positive risk–return

relation, but empirical evidence is inconclusive. Retail

investor sentiment has long been viewed as a distorting

factor, while more recently institutional investor

sentiment is thought to play a role. We examine the

separate and joint impacts of retail and institutional

investor sentiments on the risk‐return relation. We

find, at both market and firm levels, the risk‐return

relation is more likely to be distorted by the two

investor‐type sentiments jointly, rather than separately.

We further find a cross‐sectional pattern, with the

risk‐return relation being more sensitive to investor

sentiment for stocks with specific characteristics.

KEYWORDS

beta‐return relation, institutional investor sentiment,

mean–variance relation, retail investor sentiment, risk‐return

relation

JEL CLASSIFICATION

G12, G14, G41

Eur Financ Manag. 2024;30:496–543.496

|

wileyonlinelibrary.com/journal/eufm

EUROPEAN

FINANCIAL MANAGEMENT

This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and

reproduction in any medium, provided the original work is properly cited.

© 2023 The Authors. European Financial Management published by John Wiley & Sons Ltd.

We thank John A. Doukas (the editor) and two anonymous reviewers for extremely insightful and helpful comments

during the review process. We are grateful to Gbenga Adamolekun, Min Du, Qiaohui Liang, Antonios Siganos, Lan

Xiang, participants at Brown Bag Research Seminar (2022) and Taishan Finance Forum (2022) for helpful comments.

Responsibility for errors remain with the authors.

1|INTRODUCTION

Finance theory posits a positive risk–return relation, but the empirical evidence is not always

supportive. Departures from this theorized relation suggest asset mispricing and are indicative

of market instability. Investor sentiment has been shown to distort the risk–return relation

(Antoniou et al., 2016; Yu & Yuan, 2011) and hence has a potentially destabilizing impact on

financial markets, with retail investor sentiment conventionally viewed as the irrational

culprit,

1

while institutional investors have been largely viewed as rational, informed traders

capable of bringing the markets back into line. Findings in Cohen et al. (2002), for example,

support the stabilizing impact of institutional investors driving stock prices toward

fundamental values. In contrast, other studies offer evidence of the destabilizing impact of

institutional investors due to herding (Cai et al., 2019) and/or positive feedback trading

(Nofsinger & Sias, 1999). Bohl et al. (2009) argue that this can only be construed as indirect

evidence and so as such does not necessarily imply that institutional investors destabilize stock

prices.

2

On the contrary, they suggest that institutional investors may stabilize stock prices,

especially if they counter irrational retail investor sentiment. We directly test this proposition

by examining not only the separate impacts of institutional and retail investors' sentiments on

the risk–return relation, but also their joint impact. In doing so we respond to recent

developments in theoretical models (Sheng et al., 2022) and call for greater attention to be given

to the interplay between investor types (Spyrou, 2013).

The standard financial framework theorizes a positive risk–return relation, showing that

bearing high (low) risk should be compensated by high (low) returns (Merton, 1973,1980).

Empirical findings diverge, however, with evidence of positive, negative, and mixed relations.

3

Behavioural studies highlight the importance of investor sentiment in the determination of

asset prices. De Long et al. (1990a), for example, propose a model in which informed investors

trade together with noise investors and report that the latter's participation triggers systematic

risk, imposing limits on arbitrage and leading to persistent mispricing (see, also, Palomino,

1996; Shefrin & Statman, 1994). The theory is supported by empirical evidence from the US

market as well as global markets, revealing a persistent negative impact of investor sentiment

on stock market returns (Baker & Wurgler, 2006,2007; Brown & Cliff, 2004,2005; Da et al.,

2011,2015; Lemmon & Portniaguina, 2006; Schmeling, 2009; Wang et al., 2021,2022). The

influence of investor sentiment has been linked to the likelihood of stock market crises (Chen

et al., 2020; Zouaoui et al., 2011).

4

Yu and Yuan (2011) combine the mean–variance relation

and investor sentiment, arguing that first, retail investors are unsophisticated and, hence, likely

to misestimate the variance of returns, distorting the positive mean–variance relation as

theorized, and second, retail investors are more willing to trade when feeling optimistic than

pessimistic due to limits on short selling (Barber & Odean, 2008). Therefore, one can observe

1

DeVault et al. (2019, p. 985) note that this traditional view has been explicitly repeated in the nearly 70 years following the pioneering

work of Drew et al. (1950), citing Shleifer and Summers (1990), Lee et al. (1991), Neal and Wheatley (1998), Nagel (2005), Barberis and

Xiong (2012), and Da et al. (2015), among others, by way of example.

2

Using the Polish stock market as the sample, Goodfellow et al. (2009) report that institutional investors do not exhibit herding.

3

See, Campbell (1987), Baillie and DeGennaro (1990), Baker et al. (2011), Geoffrey Booth et al. (2016), Brandt and Kang (2004), Brandt

and Wang (2010), Campbell and Hentschel (1992), Fiore and Saha (2015), French et al. (1987), Glosten et al. (1993), Guo and Whitelaw

(2006), Harvey (2001), Ludvigson and Ng (2007), Lundblad (2007), Pástor et al. (2008), Rossi and Timmermann (2015), Scruggs (1998),

Turner et al. (1989), Wang et al. (2017), and Whitelaw (1994).

4

Other studies demonstrate a more specific impact of event‐driven sentiment on stock markets, including sunshine (Hirshleifer &

Shumway, 2003), aviation disasters (Kaplanski & Levy, 2010), sporting events (Edmans et al., 2007), war (Hudson & Urquhart, 2015),

religious events (Gavriilidis et al., 2016), air pollution (Lepori, 2016), and music (Edmans et al., 2021).

DUXBURY and WANG EUROPEAN

FINANCIAL MANAGEMENT

|

497

that during high‐sentiment periods, the positive mean–variance relation would be distorted by

the elevated presence of noise traders, while amid low‐sentiment periods when noise traders

stay in the sidelines, a positive mean–variance relation is present,

5

which is confirmed by their

empirical results. Similarly, at the firm level, Antoniou et al. (2016) argue that retail investors

trade more over high‐than low‐sentiment periods. However, their elevated trading

concentrates on high‐beta stocks (Barber & Odean, 2000,2001; van Binsbergen et al., 2022),

leading high‐beta stocks to be overpriced, which is followed by low returns and hence, the

collapse of the capital asset pricing model (CAPM) and distortion of the positive beta–return

relation. Conversely, the CAPM would be more likely to hold during periods when retail

investors feel pessimistic and so trade less, which is supported by their empirical results.

Emerging evidence, however, appears to challenge the conventional wisdom that retail

investors are solely to blame for irrational markets while institutional investors are sophisticated

and less susceptible to noise and biases in trading. A number of theoretical underpinnings

emerge in the literature. First, institutional sentiment trading is partially driven by reputational

effects. Institutional investors, like investment managers, tend to follow other managers even

when they have opposing fundamental information, such that their performance is close to the

average and so theydo not stand out from the crowd (Chelley‐Steeley et al., 2019). As pointedout

in Scharfstein and Stein (1990), if fund managers do not adopt sentiment trading and if their

performance is worse than that of other fund managers, they may be classified as

underperforming managers even when their performance is at times better, suggesting that

institutional investors have a strong motivation to employ sentiment trading and to follow others

as they aim to obtain average profits rather than stand out and risk their reputation (Dasgupta

et al., 2011; Prendergast & Stole,1996; Rajan, 2006; Trueman, 1994). Graham (1999), for example,

reports that when analysts have high reputation or low ability, or if their private information

contradicts strong public information, they tend to herd lest they stand out and perform worse.

Another manifestation of the reputational effects, as posited in Chelley‐Steeley et al. (2019),

is that institutional investors hope to retain their clients and deter them from moving to rival

institutions. If their clients exhibit sentiment while institutional investors do not follow or trade

on sentiment, it is conceivable that such clients will close their positions with the managers,

who consequently face declining revenues. In an attempt to maintain their clients, therefore,

institutional investors may choose to trade on sentiment, albeit originating initially from their

clients rather than themselves. More recently, examining three components of institutional

5

Following the two‐agent model in De Long et al. (1990a), Yu and Yuan (2011) provide a theoretical model proposing that the

mean–variance relation in stock markets follows:

⋅

⋅⋅









































{}

() ()

E

RσRS

=

+

−1

+,

ex ex

S

NNK

α

N

NNK

NK

NNKNKiN

NNK

εiα

NKiN

NK ηi

εi

N

NK NKiN

NK

εi

121

0

1+2−

11

1+2−

2−

1+2−

1

2−=1+1

1+2−1

(1+ )

0

1

2−=1+1

−

1+

1

2−

1

2−=1+1

−1

1+

where N=N

1

(the number of informed traders) + N

2

(the number of sentiment traders); S

0

is the stock price at time 0;

Kis the number of sentiment traders whose short‐sale constraints are binding; ε

i

is the noise in investor i's variance

estimation; η

i

is the noise in investor i's expectation estimation; and αis the CARA risk‐aversion coefficient. The

mean–variance relation is largely determined by two factors: (i) the average inverse risk‐aversion attitude of stock

market participants (i.e., [·]), and (ii) the average of stock market participants’stock holding (i.e., {·}). Yu and Yuan

(2011) posit that high investor sentiment undermines the positive mean–variance relation by reducing the average

stock holdings of market participants (i.e., {·}) and by increasing the average of the inverse risk‐aversion attitudes of

stock market participants (i.e., [·]). See, the appendix of Yu and Yuan (2011), for detailed derivation and proof.

498

|

EUROPEAN

FINANCIAL MANAGEMENT

DUXBURY and WANG