JUMPS, NEWS, AND SUBSEQUENT RETURN DYNAMICS: AN INTRADAY STUDY

• Author: Yuewen Xiao,Xiangkang Yin,Jing Zhao
• Date: 01 August 2020
• Published date: 01 August 2020
• DOI: http://doi.org/10.1111/jfir.12223

The Journal of Financial Research Vol. XLIII, No. 3 Pages 705–731 Fall 2020

DOI: 10.1111/jfir.12223

JUMPS, NEWS, AND SUBSEQUENT RETURN DYNAMICS: AN INTRADAY

STUDY

Yuewen Xiao

University of Shanghai for Science and Technology

Xiangkang Yin

Deakin University

Jing Zhao

La Trobe University

Abstract

We detect jumps in a high‐frequency price series of exchange‐traded funds (ETFs)

that track the broad indexes of U.S. equity markets. Although many jumps (43%) are

related to macroeconomic news, more jumps (57%) are not. No‐news jumps are

followed by significant return reversals for at least 60 minutes. The return dynamics

after news‐related jumps vary with the news characteristics. Scheduled‐news jumps

are followed by reversals, whereas unscheduled‐news jumps are followed by

momentum. Whether related to news or not, negative jumps are followed by stronger

return reversals than are positive jumps.

JEL Classification: G12, G14

I. Introduction

Extreme price movements (i.e., jumps) are prominent phenomena in financial markets.

Understanding jump behavior is crucial to addressing fundamental finance issues, such

as asset pricing and the return–risk relation. This is because jump risk and diffusive risk

are compensated differently (Liu and Pan 2003), and jump risk premiums represent a

major part, nearly 100% at times, of the total premium (Santa‐Clara and Yan 2010).

Extreme price movements are also linked to market health (Jawadi, Louhichi, and

Cheffou 2015), and research on price jumps facilitates the detection and diagnosis of

potential market instability. We focus on exchange‐traded funds (ETFs) that track

broad market indexes and investigate the relation between ETF price jumps and the

release of macroeconomic news, as well as the return dynamics of these ETFs after

jumps.

This research is supported by the Shanghai Planning Office of Philosophy and Social Science (Grant No.

2019BJB023). We thank the anonymous referee and seminar participants of Deakin University and Monash

University for valuable comments, the discussant and participants of the 32nd Australasian Finance and Banking

Conference, and Guoliang Ma for helpful suggestions on promoting the coding efficiency. Yuewen Xiao also

thanks Deakin University for providing a supportive and stimulating working environment during her visit.

705

© 2020 The Southern Finance Association and the Southwestern Finance Association

News is considered a dominant driver of price movements because it revises

investors’information set, contributes to their expectation formation, and influences

market sentiment. In a recent study, Jiang and Zhu (2017) use jumps in stock prices as a

proxy for large information shocks. However, assuming a one‐to‐one correspondence

between news events and jumps is questionable because not all information shocks

trigger extreme price movements and vice versa. For example, Evans (2011) shows that

approximately one‐third of intraday jumps in the S&P 500 E‐Mini, 10‐year U.S. Treasury

bond, and euro–dollar foreign exchange futures can be linked to macroeconomic news

announcements. Piccotti (2018) finds that only a small portion of jumps in spot foreign

exchange markets are motivated by prescheduled macroeconomic news releases. We

study this issue in the context of U.S. equity markets, which are represented by a sample

of ETFs. With a comprehensive news database provided by RavenPack News Analytics

(RPNA),

1

we find 38,137 novel announcements of macroeconomic news over our sample

period from the beginning of 2002 tothe end of 2016. Based on intraday price data from

Thomson Reuters Tick History (TRTH), we find 212,439 jumps across 205 sample ETFs

over the sample period. Among them, 91,192 (43%) jumps are related to news

announcements in our database and 121,247 (57%) are not. This confirms that news can

be a significant driver of price jumps in the securities markets but that not all jumps are

attributable to news. Furthermore, news events are much more frequent than are jumps,

indicating most news releases do not trigger jumps. For instance, the interval between

14:00 and 14:15 is the typical time when the Federal Open Market Committee (FOMC)

makes its announcements. Only 3,648 jumps, fewer than 18 jumps per ETF, occurring

during this interval are news related, although there are 1,141 macroeconomic news

releases during the same interval.

Price jumps that are not related to news about economic fundamentals

2

may

reflect investor overreaction and market friction. Our empirical evidence shows that

for at least 60 minutes, subsequent returns after no‐news jumps reverse strongly.

Moreover, this return reversal is much stronger than the reversal following a normal

15‐minute interval that does not contain jumps. Thus, our findings can be interpreted

by short‐run market irrationality, which leads to substantial deviation from the

fundamentals. Nevertheless, over a longer period, markets are efficient in that

mispricing is gradually corrected and prices are pulled back to the fundamentals.

Illiquidity market friction effects or the leverage effect can also cause return reversals

following no‐news jumps; this explanation is partially supported by our illiquidity

analysis in Section V. Furthermore, if we look into jumps more closely by separating

positive jumps from negative jumps, the return reversals following negative no‐news

jumps are the strongest. More broadly, whether news related or not, return reversals

after negative jumps are stronger than after positive jumps.

3

1

News data from RPNA are used in numerous studies, including Kolasinski, Reed, and Ringgenberg

(2013), Drake, Guest, and Twedt (2014), Lin, Massa, and Zhang (2014), Dang, Moshirian, and Zhang (2015),

Massa et al. (2015), Bushman, Williams, and Wittenberg‐Moerman (2017), and Kelley and Tetlock (2017).

2

In this article, we call these no‐news jumps.

3

For positive news jumps, subsequent returns exhibit continuation.

706 The Journal of Financial Research