Programs trades and trade regulation: An evidence of the Korean securities market

• Author: Woo‐Baik Lee,Steven J. Jordan,Cheoljun Eom,Jong Won Park
• Published date: 01 January 2020
• DOI: http://doi.org/10.1002/fut.22056
• Date: 01 January 2020

J Futures Markets. 2020;40:44–66.wileyonlinelibrary.com/journal/fut44

|

© 2019 Wiley Periodicals, Inc.

Received: 20 September 2018

|

Accepted: 11 August 2019

DOI: 10.1002/fut.22056

RESEARCH ARTICLE

Programs trades and trade regulation: An evidence of the

Korean securities market

Cheoljun Eom

1

|

Steven J. Jordan

2

|

Woo‐Baik Lee

3

|

Jong Won Park

4

1

School of Business, Pusan National

University, Busan, Republic of Korea

2

Econometric Solutions, Tampa, Florida

3

Department of Business Administration,

Korea National Open University, Seoul,

Republic of Korea

4

College of Business Administration,

University of Seoul, Seoul, Republic of

Korea

Correspondence

Jong Won Park, College of Business

Administration, University of Seoul,

Seoulsiripdae‐ro 163, Dondaemun‐gu,

Seoul 02504, Republic of Korea.

Email: parkjw@uos.ac.kr

Funding information

2017 Research Fund of the University of

Seoul, Grant/Award Number:

201704251037

Abstract

This study addresses the effects of program trade regulation during large market

moves. To address this issue, we analyze the effect of sidecars (halts that only

affect program trades) on trade imbalance using Korean intraday data. We find

that sidecars, as currently designed to halt all program trades, are not effective

at controlling trade imbalance around volatile markets. Resolution of trade

imbalance is more effective when program trade is unrestricted. Program trade,

at least a subset, provides liquidity when it is at a premium. We conclude that

current sidecars should be more carefully crafted as some program trades are

market stabilizing.

KEYWORDS

KOSPI 200, program trade halts, program trades, sidecar, trade imbalance

JEL CLASSIFICATION

C14, G12, G13, G22

1

|

INTRODUCTION

Sidecar is one of the trade halts to control for temporary large price moves potentially caused by program trading.

Given the prominent role program trading

1

has been assigned as a potential cause of market instability, it is a bit

surprising that there exist only a few studies that specifically address program trade halt rules and their

effectiveness at controlling large market moves.

2

Natural questions of interest are: Why is it important to study

program trading halts separately? What can we learn from examining these halts that we cannot learn from the

studies that document the effects of trading halts in general? Given that trade halts occur when markets

are experiencing large price moves, and this is specifically true in our program trade halt data by construction,

liquidity will be at a premium. When the price of liquidity is high, one scenario is that sophisticated traders are

one likely group to step in and try to capture the temporary increase in profits from providing liquidity. Another

scenario, as in the recent flash crashes, program tradescreatestheliquiditysupplybytradingheavilyin

one direction in an attempt on the response of the news. An example of this is that the recent “tweet crash”on

1

It should be noted that the terminology “program trading”has different definitions in different contexts. To enact regulation, program trading has to be defined according to observable criteria.

Definitions usually consist of the number of different assets traded or whether various markets are traded simultaneously. In this paper, we use the KOSPI market data of Korea Exchange (KRX). KRX

defines program trading as following types of investment strategies; (a) index‐arbitrage trading is defined as a set of trading activity which includes buying (or selling) basket of constituent stocks in

KOSPI 200 index and selling (or buying) KOSPI 200 index futures or options for the purpose of gaining profits by taking advantage of instant price discrepancy between KOSPI 200 stocks and KOSPI

200 futures or options and (b) non‐index‐arbitrage trading is defined as either buying or selling basket of 15 or more constituent stocks in KOSPI 200 index at the same time by the same investor.

2

The few papers we found that addressed program trading rules for large market move all examined the NYSE Rule 80A. However, Rule 80A is a trade inhibitor (we classify trade regulating

mechanisms into halts and inhibitors; halts completely stop targeted trading, while inhibitors allow trade under different rules), thus program trading still exists under most implementations, only the

rules of trade change.

April 23, 2013 where twitter was hacked and news of a bomb in the White House caused heavy program trade

selling.

3

Thus, one would want to study whether program trades are demanding or providing liquidity before

eliminating them from such markets. In addition, regulators have specifically targeted program trade rules as a way

to control excess market moves. Given the recent spate of flash crashes, program trade halt rules are a potential

tool to regulate short‐term program trading.

To address this issue, we analyze the effect of program trade halts on trade imbalance using Korean intraday data.

The Korean market provides an advantageous setting to study whether program trades regulation has achieved its

objective to relieve price pressure from program trades compared to the US market and clean tests can be designed to

analyze the effectiveness of program trading regulation during large market moves (see Section 3.1). The sidecar of KRX

only stops program trades, allowing all other types of trades to be executed. Thus, studies using the KRX data can

address the effect of program trades (index‐arbitrage or non‐index‐arbitrage trade) on market quality. We can explore

the importance of the various program trade types on market quality. This trade‐type dimension has not been explored

in the prior empirical literature. Our measure of trade initiator imbalance (|TIM|) is the absolute value of the trade

imbalance in Chordia, Roll, and Subrahmanyam (2002, 2008).

4

Several studies show that TIM affects market quality

proxies such as liquidity, price change, and price convergence making it an important market characteristic to

understand in markets with large price moves (see Section 4.1).

Our research contributes insight into three important questions. First, what are the characteristics of TIM for

different trade types (i.e., program, non‐program, index‐arbitrage, and non‐index‐arbitrage trades) around program

trade halts? To answer this, we analyze and compare imbalances surrounding sidecar events. We measure the TIM of

all KOSPI 200 stocks by trade type. We analyze all sidecar events, the up‐market‐sidecar subsample, and the down‐

market‐sidecar subsample. We control for market dynamics and firm characteristics. Our initial result is that the

resolution of TIM is greater for program trades than for non‐program trades. We also document that non‐index‐

arbitrage trades experience higher TIM resolution than index‐arbitrage trades. In addition, we find that there is an

increase in trade volume after a sidecar, indicating pent‐up trade demand.

Second, a fundamental question in finance is how information is embedded into price. We explore two possible

hypotheses. According to no arbitrage, information can be transmitted via the action of arbitrageurs engaging in index‐

arbitrage trades. In this scenario, mispricing is corrected by simultaneously taking a long position in the undervalued

market, while shorting the overvalued market. An alternative hypothesis is that there is a smart‐money effect. In this

scenario, some traders have better private information and target individually mispriced assets to take advantage of this

information advantage. These “smart trades”push the mispriced asset to equilibrium. To discern between these two

competing theories, we take all program trades and classify them as either index‐arbitrage or non‐index‐arbitrage

trades. The information transfer mechanism that is more important should have a larger TIM resolution across the

sidecar event. We find that the smart‐money effect is stronger than the index‐arbitrage effect.

Third, we explore how markets correct themselves in the absence of a sidecar implementation, that is are program

trade halts necessary? We study this question for all program trades and for index and non‐index arbitrage separately.

To answer this question, we use our pseudo‐sidecar sample to test if extreme market moves are associated with similar

imbalance reduction patterns in the actual‐sidecar sample. If so, then the sidecar is of questionable utility as markets

are adjusting on themselves. In the pseudo‐sidecar control, we use the same trade types during a different time period

with similar market dynamics. A priori given that mean reversion of price exists in markets, we would expect larger

market moves and larger TIM to have on average larger corrections. Thus, we expect to find a larger correction in TIM

for the actual‐sidecar events. However, we find the opposite. The pseudo‐sidecar events have a larger correction

compared to when program trading is halted. We conclude that actual sidecars are, on average, inhibiting the market’s

self‐regulating mechanisms. That is, the sidecar is not necessary to observe the reduction in TIM associated with a large

market move. The market self‐adjusts via its own internal mechanisms. Therefore, program trade halts reduce the

market’s capacity to adjust for large trade imbalance during large market moves.

3

The Dow Jones industrial average fell more than 150 points after the fake Twitter posting, then quickly recovered and the S&P 500’s value lost over $136 billion in less than 3 min plunge (see https://

business.financialpost.com/investing/dow‐jones‐plummets‐then‐recovers‐after‐fake‐ap‐tweet‐of‐explosions‐at‐the‐white‐house). Another event occurred on April 17, 2013 when the German DAX, the

French CAC, and the UK FTSE all experienced a flash crash in less than 2 min. The DAX lost 2.6% over this very short time period (see http://thekeystonespeculator.blogspot.com/2013/04/flash‐crash‐

dax‐germany‐cacfrance‐and.html).

4

Although Chordia et al. (2002) labeled the difference between buyer‐initiated trades and seller‐initiated trades order imbalance, the market microstructure literature recognizes that these measures

are trade imbalance, not order imbalance, measures. For example, Boehmer and Wu (2008) state: “Strictly speaking these imbalance measures reflect net trading imbalances. To be consistent with

prior literature we use the term order imbalance in the text.”

EOM ET AL.

|

45