Trade‐size clustering and informed trading in global markets

• Published date: 01 October 2020
• Author: Tao Chen
• DOI: http://doi.org/10.1002/ijfe.1768
• Date: 01 October 2020

RESEARCH ARTICLE

Trade‐size clustering and informed trading in global

markets

Tao Chen

Faculty of Business Administration,

University of Macau, Macau

Correspondence

Tao Chen, Department of Finance and

Business Economics, Faculty of Business

Administration, University of Macau,

Macau.

Email: torochen@um.edu.mo

Funding information

Universidade de Macau Start‐up Research

Grant, Grant/Award Number: SRG2018‐

00115‐FBA; University of Macau, Grant/

Award Number: SRG2018‐00115‐FBA

Abstract

Based on intraday data across 41 markets, this study examines whether

informed traders exploit trade‐size clustering. Clustering trades are docu-

mented to predict price movements, to generate perpetual return impact, and

to improve informational efficiency. Collectively, these findings suggest that

the clustering strategy is leveraged by the informed to cover up their activities

in global markets. In addition, the cross‐country analysis indicates that larger

market capacity and better legal protection, as two predominant institutional

features, are associated with a lower level of informed‐trade clustering. Finally,

such negative interaction attenuates in countries with lot‐size regulations and

at bellwether stocks.

KEYWORDS

Bellwether effects,clustering, global markets, informed trading, lot sizes

JEL CLASSIFICATION

G14; G15; G41

1|INTRODUCTION

How informed traders behave in financial markets has

been extensively studied in previous literature. The

importance attached to this issue may be linked with

their actions that could exert a direct impact on the price

discovery process. This statement is valid depending on a

key postulation that sophisticated investors are inclined

to conceal trading intentions so as not to be detectable

by others, in line with the theoretical prediction by Kyle

(1985). Meanwhile, Barclay and Warner (1993) develop

the stealth trading hypothesis to explicate such behaviour

of informed traders.

Given specific trading tactics, earlier work on informed

trading reveals that these investors would deliberately

implement the stealth transaction by looking for an

appropriate time (Admati & Pfleiderer, 1988), conducting

simultaneous trades in parallel markets (Grammig,

Schiereck, & Theissen, 2001), dividing large orders into

small ones (Alexander & Peterson, 2007;), submitting

small orders over time (Keim & Madhavan, 1996), and

abusing the bias of liquidity traders (Chen, 2018). Collec-

tively, these researches appear to elucidate that both trade

sizes and timing are crucial factors when formulating the

trading strategy by sophisticated investors. To better cover

up private information, the informed would be prone to

capitalizing on multiple trades with similar sizes and

trading persistently, which ultimately leads to the occur-

rence of trade‐size clustering (TSC).

If focusing on prior studies concerning trade sizes,

informed trades are found to happen at medium sizes

because most of the cumulative price impact is attributable

to this category (Barclay & Warner, 1993). Subsequently,

Chakravarty (2001) associates this disproportionately large

cumulative price change of medium‐size transactions with

institutions. Besides, Alexander and Peterson (2007) show

that round‐size trades are closely linked with informed

investors because they have a greater price impact

Received: 18 October 2018 Revised: 16 January 2019 Accepted: 13 September 2019

DOI: 10.1002/ijfe.1768

Int J Fin Econ. 2019;1–19. © 2019 John Wiley & Sons, Ltd.wileyonlinelibrary.com/journal/ijfe 1

Int J Fin Econ. 2020;25:579–597. wileyonlinelibrary.com/journal/ijfe © 2019 John Wiley & Sons, Ltd. 579

compared with unrounded ones. However, informed short

sellers cluster less on round sizes than their counterparts

(Blau, Van Ness, & Van Ness, 2012). Recently, O'Hara,

Yao, and Ye (2014) further find that small sizes (odd lots)

are adopted by informed traders when exercising algorith-

mic and high‐frequency transactions.

Motivated by the stealth trading conjecture, this paper

first scrutinizes whether TSC is employed by informed

traders to hide their transactions in global markets. Rely-

ing on high‐frequency data in 41 markets, we seek evi-

dence to support the view that informed investors

exploit TSC. Concretely, while clustering trades push

prices toward fundamental values, temporary price

movements are dampened effectively. Additionally, the

long‐run return impact (RI) of clustering trades is quite

significant. Lastly, clustering trades are shown to drive

informational efficiency. Combined, these findings are

consistent with the idea that informed traders contribute

to TSC because their stealth actions can be camouflaged

among clustering trades.

Despite informed traders drive TSC, uninformed inves-

tors are also likely to blame for such irregular trading

behaviour. On the one hand, due to their psychological

preferences for round numbers, noise traders may gener-

ate a tendency to submit round‐size orders (Ikenberry &

Weston, 2008). On the other hand, using a rounded size

allows naïve traders to accelerate the negotiation process

and thus alleviate the reporting risk (Harris, 1991). Both

considerations would prompt uninformed traders to

engage in round‐size clustering. Given the coexistence of

informed and uninformed clustering, little attention is

paid to the differentiation with each other. To fill the

gap in the literature, this article next develops an empiri-

cal approach to extract informed‐trade clustering (ITC).

Because institutional factors of economic, cultural, and

political differences may affect the extent to which infor-

mation is disseminated, they would matter to informed

trading. Finally, we probe how these variables determine

the cross‐sectional variations of ITC. Inspired by a com-

prehensive framework proposed by Karolyi (2015), we

compile six‐dimension institutional characteristics for dif-

ferent countries. After considering firm‐specific controls

and industry dummies, market capacity and legal protec-

tion are revealed to be negatively associated with ITC. In

other words, mature market and sound legal system help

prohibit informed trades from being concealed through

clustering techniques, thus resulting in a lower level of

ITC. However, such negative influence weakens in coun-

tries with lot‐size restrictions and at bellwether stocks.

This study contributes to earlier studies in two aspects.

First, we extend the previous literature on TSC from a sin-

gle market to the global market (Alexander & Peterson,

2007; Garvey, Huang, & Wu, 2018; Garvey & Wu, 2014).

In addition, we complement these studies by developing

a new measure of TSC following the Herfindahl index.

Our method not only considers all trade sizes in the cal-

culation but also does not select arbitrage thresholds to

determine specific sizes (medium or round sizes) where

clustering trades happen. Moreover, we attempt to distin-

guish ITC from uninformed‐trade clustering, which

advances our understanding of the root cause for this

phenomenon.

Second, our paper adds to prior work on international

finance by highlighting the importance of institutional

characteristics to ITC. In related work, country‐level fea-

tures are documented to impact the firms' corporate gover-

nance (Doidge, Karolyi, & Stulz, 2007), the development of

stock markets (Djankov, La Porta, Lopez‐de Silanes, &

Shleifer, 2008), the return and liquidity commonality

(Karolyi, Lee, & Van Dijk, 2012), and the market efficiency

(Griffin, Kelly, & Nardari, 2010). By investigating ITC in

the cross‐section of stocks and markets, we offer a new

trader behaviour‐based application to this research area.

If taking into account investors' heterogeneity, the global

context enables us to generalize the findings and to over-

come the sample bias arising from one country.

The rest of the paper proceeds as follows. Section 2

elaborates the data and measurement. Section 3 presents

the empirical result on whether TSC is driven by

informed trading. Section 4 performs the cross‐sectional

analysis on ITC. The last section concludes.

2|DATA AND MEASUREMENT

2.1 |Data

The high‐frequency intraday data used in this study are

downloaded from a tick‐by‐tick database of the

Bloomberg, which is the leading information provider to

keep track of historical and real‐time transaction data in

each stock exchange around the world. If a trade occurs

in one market, the Bloomberg would register its trading

time, traded price, and trade size. In addition, the quota-

tion code is added if the trade is not a normal one. There-

fore, every record in the Bloomberg represents the

actually completed trade in the stock market. For exam-

ple, the table below illustrates how three trades for the

International Business Machines' stock in US (ticker:

International Business Machines United Nations' Equity)

are recorded in the dataset compiled by the Bloomberg.

Trading time Traded price Trade size Quotation code

Jan‐4‐2016 09:30:00 135.51 274 N.A.

Jan‐4‐2016 09:30:01 135.79 180 N.A.

Jan‐4‐2016 09:30:01 135.5 299 N.A.

2CHEN

580 CHEN