Overnight returns of industry exchange‐traded funds, investor sentiment, and futures market returns

• Published date: 01 June 2022
• Author: Yun‐Huan Lee,Tzu‐Hsiang Liao,Hsiu‐Chuan Lee
• Date: 01 June 2022
• DOI: http://doi.org/10.1002/fut.22321

Received: 6 July 2021

|

Accepted: 3 February 2022

DOI: 10.1002/fut.22321

RESEARCH ARTICLE

Overnight returns of industry exchange‐traded funds,

investor sentiment, and futures market returns

Yun‐Huan Lee|Tzu‐Hsiang Liao|Hsiu‐Chuan Lee

Department of Finance, Ming Chuan

University, 250 Zhong Shan N. Rd., Sec. 5,

Taipei, 111, Taiwan

Correspondence

Hsiu‐Chuan Lee, Department of Finance,

Ming Chuan University, 250 Zhong‐Shan

N. Rd., Sec. 5, Taipei 111, Taiwan.

Email: hclee@mail.mcu.edu.tw

Funding information

Ministry of Science and Technology of

Taiwan, Grant/Award Number: MOST

109‐2410‐H‐130‐009

Abstract

This study investigates whether investor sentiment estimated by overnight

returns of industry exchange‐traded funds (ETFs) affects Volatility In-

dex (VIX) futures and stock index futures returns. Our empirical results in-

dicate that high overnight returns of industry ETFs are associated with

sentiment‐based trading. The results also show that investor sentiment, as

measured by the relative comovements of overnight returns of industry ETFs,

Granger‐causes VIX futures and stock index futures returns, but not vice versa.

Finally, investor sentiment, as measured by the relative comovements, dis-

plays statistically and economically significant out‐of‐sample predictive power

for VIX futures and stock index futures returns.

KEYWORDS

industry ETFs, investor sentiment, overnight returns, stock index futures, VIX futures

1|INTRODUCTION

The role of investor sentiment in financial markets is a topic of substantial research interest. Prior studies have

proposed various top‐down and bottom‐up approaches to measure investor sentiment (e.g., Aboody et al., 2018; Baker

& Wurgler, 2006; Berkman et al., 2012; Da et al., 2015; Guo et al., 2019; Huang et al., 2015).

1

The sentiment indices

constructed from market‐level variables proposed by Baker and Wurgler (2006), Da et al. (2015), and Huang et al.

(2015) are categorized as a top‐down approach, while those defined by Berkman et al. (2012), Aboody et al. (2018), and

Guo et al. (2019) are based on individual stock overnight returns and are thus a type of bottom‐up approach. On the

basis of the conceptual idea of the bottom‐up approach, this study employs overnight returns of industry exchange‐

traded funds (ETFs) to measure investor sentiment. It first investigates whether overnight returns of industry ETFs are

associated with the share demand of optimistic overnight individual/retail investors and then examines the impacts of

investor sentiment, as measured by overnight returns of industry ETFs, on subsequent Chicago Board Options

Exchange's Volatility Index (VIX) futures and stock index futures returns. In contrast to previous studies that employ a

bottom‐up approach, this study focuses on the industry level rather than the firm level, and hence provides a less time‐

consuming approach to construct measures of investor sentiment.

This study argues that overnight returns of industry ETFs might be correlated with the industry demand of

optimistic overnight individual/retail investors. Prior studies have found that individual stock overnight returns can

serve as a measure of firm‐level sentiment (see Aboody et al., 2018; Berkman et al., 2012). As indicated by Lou et al.

(2019) and Akbas et al. (2021), the presence of overnight and intraday clienteles can explain high overnight returns

J Futures Markets. 2022;42:1114–1134.wileyonlinelibrary.com/journal/fut1114

|

© 2022 Wiley Periodicals LLC

1

The studies cited are representative and do not constitute an exhaustive list.

relative to the rest of the trading day. Specifically, the overcorrection hypothesis proposed by Akbas et al. (2021) states

that persistent daily reversals with high opening prices are associated with optimistic overnight retail traders and

daytime arbitrageurs to overcorrect a prolonged sequence of positive overnight returns. In a similar vein, overnight

returns of industry ETFs might be correlated with the industry demand of optimistic overnight individual/retail

investors for two main reasons. First, retail investors pay more attention to industry information than firm‐specific

information because information collection and analysis across industries is less time‐consuming than examining

thousands of stocks (see Jame & Tong, 2014). Second, Broman (2016) suggests that industry ETFs allow individual

investors to easily buy and sell popular investment styles at a cost lower than their basket trading of underlying

securities. Motivated by this stream of the literature, if industry‐specific attention‐triggering events lead to higher

buying price pressure by individual/retail investors around the market opening, the opening prices for industry ETFs

will be overpriced, and thus high overnight returns of industry ETFs are associated with the industry demand of

optimistic overnight individual/retail investors.

Using nine SPDR industry ETFs, our empirical results for daily portfolios of industry ETFs show a strong tendency

toward positive (negative) overnight returns and negative (positive) trading day reversals. Overnight and trading day

portfolio returns have the opposite signs, indicative of an intraday reversal in daily portfolio returns. Additionally, the

evidence for daily portfolios shows that the absolute magnitude of overnight returns and daily reversal is larger for

positive overnight returns than negative overnight returns (see Akbas et al., 2021). For monthly portfolios, the evidence

indicates that industry ETFs with relatively high overnight returns over the prior month have, on average, relatively

high overnight returns as well as relatively low intraday returns in the subsequent month. These findings suggest that

high overnight returns of industry ETFs are associated with overnight buying pressure by optimistic retail investors.

Then, this study employs three approaches to construct investor sentiment measures using overnight returns of

industry ETFs and explores whether these measures of investor sentiment are correlated with VIX futures (V

F

), E‐Mini

S&P 500 index futures (

S

P

), and E‐Mini Dow Jones (

DJ

) index futures returns. Since the work by Whaley (2000), the

VIX has been viewed as the “investor fear gauge”and thus is associated with investor sentiment; however, the

VIX cannot easily be traded. On March 26, 2004, the CBOE introduced VIX futures to address demand for hedging

purposes. Frijns et al. (2016) indicate that higher investor fear (or pessimistic investor sentiment) results in an increase

in hedging demand using VIX futures and that VIX futures returns are negatively associated with stock index futures

returns. According to the contagion effect, the recurring overnight price pressure induced by optimistic overnight retail

traders for industry ETFs is likely to spill over to market‐wide futures markets, thus causing persistence in futures

returns.

On the basis of a vector autoregression (VAR) model, we find that investor sentiment estimated by the relative

comovements of overnight returns of industry ETFs Granger‐causes VIX futures returns and stock index futures

returns, but not vice versa. The VAR results also indicate that investor sentiment, as estimated by the relative

comovements, has a negative effect on subsequent VIX futures returns and a positive effect on subsequent stock index

futures returns. Finally, investor sentiment, as measured by the relative comovements, improves the out‐of‐sample

predictive ability for VIX futures returns and stock index futures returns.

This study contributes to the literature in several ways. First, it provides a possible explanation for high overnight

returns of industry ETFs and thus allows the development of a less time‐consuming approach to construct measures of

investor sentiment by employing overnight returns of industry ETFs. Lachance (2021) shows that unusually high

overnight returns of ETFs are distorted by market microstructure effects. However, as suggested by Lou et al. (2019)

and Akbas et al. (2021), high overnight returns can be explained by the fact that heterogeneous investor clienteles tend

to initiate trades at different points in a day. Following Lou et al. (2019) and Akbas et al. (2021), our evidence shows

that high overnight returns of industry ETFs are associated with optimistic overnight retail traders. As such, this study

provides another approach to construct measures of investor sentiment using overnight returns of industry ETFs.

While Guo et al. (2019) construct investor sentiment by employing the overnight returns of all stocks listed on the

NYSE, Amex, and Nasdaq, it is impossible for individual/retail investors, and even institutional investors, to calculate

investor sentiment using their bottom‐up approach, as it examines thousands of individual stocks. Instead, the ap-

proach suggested in this study uses only nine industry ETFs to estimate investor sentiment, making it more feasible for

constructing an index of investor sentiment for individual/retail and institutional investors. This study thus extends the

literature by proposing a simple way in which to construct investor sentiment based on overnight returns of in-

dustry ETFs.

Second, this study contributes to the existing literature by examining the out‐of‐sample forecasting performance of

VIX futures returns based on investor sentiment, as estimated by overnight returns of industry ETFs. Ballestra et al.

LEE ET AL.

|

1115