Intraday Rallies and Crashes: Spillovers of Trading Halts

• Published date: 01 October 2016
• DOI: http://doi.org/10.1002/ijfe.1556
• Author: Arie E. Gozluklu,Bei Cui
• Date: 01 October 2016

INTRADAY RALLIES AND CRASHES: SPILLOVERS OF TRADING

HALTS

BEI CUI

1

and ARIE E. GOZLUKLU

2,

*

,†

1

School of Economics and Finance, The University of Hong Kong, Hong Kong

2

Warwick Business School, Finance Group, University of Warwick, CoventryCV4 7AL, UK

ABSTRACT

This paper analyses a set of intraday rally and crash events at the firm level during the single-stock circuit breaker program and

documents the cross-sectional spillover effects of such events on non-halted stocks. We test whether such major price jumps,

and subsequent trading halts, affect related stocks through the destabilizing eme price movements that trigger the circuit brea-

kers at the firm level are accompanied by a massive surge in volume, spread and short-term volatility, which gradually revert

back to normal. Speculative strategies of arbitrageurs such as momentum and pairs trading cause cross-sectional spillovers in

volume and volatility during the trading halt. Copyright © 2016 John Wiley & Sons, Ltd.

Received 04 October 2015; Revised 12 April 2016; Accepted 12 May 2016

JEL CODE: G12; G14; G18; G28

KEY WORDS: Circuit breakers; trading halts; arbitrage; momentum; pairs trading

1. INTRODUCTION

Modern trading environment that largely benefits from the advances in technology is also more exposed to techni-

cal problems associated with automated trading. At the same time, security prices react much faster to news, thanks

to the speed of technology. Hence, it is not uncommon to see market disruptions by extreme intraday price move-

ments. On 8 July 2015, the New York Stock Exchange stopped trading owing to an internal technical issue, similar

to the Nasdaq halt in 2013.

1

Both US equity (2010) and treasury markets (2014) have experienced flash crashes,

that is, abrupt intraday price movements with a fast recovery. Such intraday events raise concerns about the correct

policy response to avoid potentially adverse effects of intraday trading turmoil on market stability.

As one of the policy measures to prevent market disruptions, market-wide circuit breakers have been in place

since the October 1987 crash (Goldstein, 2015; Goldstein and Kavajecz, 2004). Circuit breakers aim at temporarily

suspending trading under extreme market conditions to prevent excessive intraday price fluctuations. On the

unusual turbulent day, the Flash Crash, 6 May 2010, the existing market-wide circuit breaker failed to stop sharp

price downturn of individual stocks. Following the Flash Crash, the US Securities and Exchange Commission

(SEC) launched a new single-stock circuit breaker rule (SSCB, hereafter). The rule stipulates that stocks whose

price moves 10% or more in a 5-min window should be halted for a 5-min period.

2

The trading suspension can

be triggered either by a large discrete price jump or a cumulative price change over the 5-min window.

Our first contribution is a detailed analysis of such intraday events at the firm level. We look into 54 SSCB

events that involve only Standard and Poor’s 500 (S&P 500) or Russell 1000 firms and take place during the period

between June 2010 and April 2013. Unlike earlier studies that focus on the Flash Crash (e.g. Easley et al., 2011;

Kirilenko et al., 2014; Madhavan, 2012), we filter out a set of intraday rallies and crashes at the firm level that

trigger the SSCB. These events are not caused by market-wide volatility or material news, that is, news released

*Correspondence to: Arie E. Gozluklu, Warwick Business School, Finance Group, University of Warwick, Coventry CV4 7AL, UK.

†

E-mail: arie.gozluklu@wbs.ac.uk

Copyright © 2016 John Wiley & Sons, Ltd.

International Journal of Finance & Economics

Int. J. Fin. Econ. 21: 472–501 (2016)

Published online 12 August 2016 in Wiley Online Library

(wileyonlinelibrary.com). DOI: 10.1002/ijfe.1556

by the firm that affects the fundamental value, for example, stock splits or earning announcements. However, there

may be firm-specific non-material news circulating in the financial media, for example, merger/buyout talks or ru-

mours regarding a restructuring, that trigger the SSCB. We first document the trading activity around SSCB events

to understand how the trading halts affect different measures of market quality such as volume, spread and

volatility.

Our primary objective is to understand the implication of such intraday extreme price movements and subse-

quent trading halts on market stability. Our strategy is to test the cross-sectional spillover effects of trading halts

on other stocks, that is, any effect on the trading activity of related non-halted stocks.

3

In particular, we investigate

the speculative activity of arbitrageurs as one of the channels for such cross-sectional spillover effects. De Long

et al. (1990) show that in a market with feedback traders, rational speculation may be destabilizing. In light of

the De Long et al. (1990) model, the paper by Lou and Polk (2013) propose a comomentum measure, that is, an

abnormal return correlation among stocks controlling for Fama–French risk factors, which reflects the speculative

activity of arbitrageurs (Moskowitz et al., 2012). Using their measure, we first identify the set of candidate stocks

that are part of a momentum strategy. Given the lack of a fundamental anchor (Stein, 2009), such feedback trading

is more likely to have a destabilizing effect, that is, moving prices away from the fundamental value. We question

whether such arbitrage activity has destabilizing effect on market quality beyond price efficiency. We also consider

an anchored arbitrage strategy, namely, pairs trading that exploits the price discrepancy between historically similar

stocks (Do and Faff, 2010; Gatev et al., 2006) as a potential channel for spillover effects.

4

Thus, we test how the

results on the spillover effects depend on the existence of a price anchor.

Our first observation is that extreme intraday price jumps are temporary and revert back quickly in the case of

intraday crashes. Earlier studies on the Flash Crash event (e.g. Easley et al., 2011; Kirilenko et al., 2014;

Madhavan, 2012) document a similar temporary price effect. However, intraday rallies exhibit persistent price ef-

fects if there are firm-specific news on the day of the SSCB. This result is consistent with the prediction of De Long

et al. (1990) that asset prices overreact to news because of both positive feedback traders and anticipatory trades of

rational speculators.

5

We also observe that circuit breakers are triggered by abnormal trading activity, that is, ac-

tivity compared with a 30-day benchmark, in particular, high trading volume and volatility along with increased

spread just before the trading halt. If there is news associated with the firm on the SSCB day, the abnormal tradi ng

activity already starts 5 min before the trading halt. After trading resumes, the abnormal trading activity reverts

back to its normal level in about 20 min only if there is no news. The changes persist beyond half an hour in the

case of firm-specific news. The difference-in-differences (diff-in-diff) analysis shows that the abnormal trading ac-

tivity on the SSCB day is not driven by a common shock to firms with similar characteristics, such as price, market

capitalization or index membership.

Our main contribution is to identify the channel through which the abnormal trading activity is transmitted to

related non-halted stocks during the trading halt. We show that mainly the firms that are part of a risky arbitrage

activity that involves SSCB stocks are affected by the trading halts. In particular, speculative portfolio strategies

such momentum trading or pairs trading cause trading externalities during extreme market conditions. The former

strategy bets on the price trend of a group of stocks, that is, winner stocks with a better past return performance,

while the latter places speculative bets on the price convergence between stocks with similar past price paths.

The diff-in-diff analysis reveals that the SSCB trading halts cause abnormal volume and short-term volatility spill-

overs for these non-halted stocks in the arbitrage portfolio. To the best of our knowledge, none of the earlier papers

investigate the implication of such arbitrage strategies on liquidity and volatility transmission across stocks.

There is some theoretical literature that analyses the role of circuit breakers on market quality. According to the

models, circuit breakers may serve as safeguards by providing a cooling-off period that allows extra time to reas-

sess trading decisions (Ma, 1989), reduce transactional risk (Greenwald and Stein, 1991) and lower information

asymmetry in the presence of information frictions (Spiegel and Subrahmanyam, 2000). If traders incur adverse

selection costs, then introducing circuit breakers might be beneficial for market participants by reducing volatility

while increasing trading volume. However, circuit breakers might also have an adverse effect on volatility. Accord-

ing to the Subrahmanyam (1994) model, large liquidity traders might anticipate trading halts and rush to trade to

satisfy their trading needs, and hence cause the so-called magnet effect.

Earlier empirical studies show that rule-based trading halts, such as the SSCBs introduced by the SEC, affect

market activity and result in high trading volume without reducing volatility (Abad and Pascual, 2010; Ferris

INTRADAY RALLIES AND CRASHES:SPILLOVERS OF TRADING HALTS 473

Copyright © 2016 John Wiley & Sons, Ltd. Int. J. Fin. Econ. 21: 472–501 (2016)

DOI: 10.1002/ijfe