Information share and its predictability in the Indian stock market

• Date: 01 October 2019
• Author: Madhusudan Karmakar,Sarveshwar Inani
• DOI: http://doi.org/10.1002/fut.22041
• Published date: 01 October 2019

J Futures Markets. 2019;39:1322–1343.wileyonlinelibrary.com/journal/fut1322

|

© 2019 Wiley Periodicals, Inc.

Received: 18 October 2018

|

Revised: 11 June 2019

|

Accepted: 11 June 2019

DOI: 10.1002/fut.22041

RESEARCH ARTICLE

Information share and its predictability in the Indian stock

market

Madhusudan Karmakar

1

|

Sarveshwar Inani

2

1

Finance & Accounting Area, Indian

Institute of Management, Lucknow, Uttar

Pradesh, 226013, India

2

Finance & Accounting Area, Jindal

Global Business School, OP Jindal Global

University, Sonipat, Haryana, India

Correspondence

Madhusudan Karmakar, Indian Institute

of Management, Prabandh Nagar, IIM

Road, Lucknow, Uttar Pradesh, 226013,

India.

Email: madhu@iiml.ac.in

Abstract

The study investigates price discovery in the Indian stock market and finds that

spot market plays a dominating role in price discovery when it is estimated for

the entire period as a whole. However, periodic measures of price discovery

suggest that it does not remain the same throughout the period, but varies with

time. Panel data analysis also indicates that spot market is more efficient in

price discovery for majority of size and sector panels. Finally, while market

state‐related variables are found to impact information shares in a majority of

the cases, macroeconomic announcements rarely predict the price discovery.

KEYWORDS

component share, information share, macroeconomic news, market state variables, modified

information share, price discovery, sectoral effect, size effect

JEL CLASSIFICATION

G12; G13; G14

1

|

INTRODUCTION

Which market impounds new information faster into prices and contributes more to price discovery—the futures or the

spot market? Assuming the fact that the futures market provides greater liquidity, lower transaction costs, and fewer

restrictions, a trader would like to trade profitably on a given piece of information in the futures market. Futures

market, accordingly, is more likely to incorporate information efficiently than the spot market, and, hence, contribute

more to the price discovery process. In contrast, if there are some restrictions to entry into the futures market, traders

possessing the information would prefer to trade stocks rather than futures. Consequently, information should first be

reflected in the spot market, implying that the spot market would play a more dominant role in the price discovery

process.

The issues of the relative contributions of the spot and futures markets to the price discovery process are of great

importance and have consequently received considerable attention in literature. Although voluminous studies have

investigated the price discovery process in developed countries, the subject remains relatively unexplored in the context

of India (to be surveyed briefly in the following paragraphs). Moreover, a comprehensive approach to investigate price

discovery is seen to be missing in the Indian studies. This study aims to fill this gap by comprehensively investigating

price discovery dynamics in the Indian stock market. More specifically, the study examines whether it is the futures or

the spot market in India that contributes more to the price discovery process. We explore the relative price discovery at

both the market and the individual stock level for the overall period as a whole. We also look into whether the relative

contribution of the spot and futures markets to the price discovery process remains the same or varies over time. As

observed in other markets, price discovery is likely to be sector‐specific and may also be size‐dependent. Recognizing

the above possibility, we investigate whether sector and size of firms affect price discovery in the Indian stock market.

We also examine the possible determinants of information shares to better understand where and when price discovery

takes place.

As mentioned above, several researchers have investigated the price discovery process using data mainly from

advanced countries. Researchers initially used the vector autoregressive (VAR) model or vector error correction model

(VECM) to examine the price discovery process; however, their attention was mainly restricted to studying the lead–lag

relationship between the spot and futures markets (Chan, 1992; Garbade & Silber, 1983; Herbst, McCormack, & West,

1987; Kawaller, Koch, & Koch, 1987; Schroeder & Goodwin, 1991; Stoll & Whaley, 1990). Much later, some researchers

used the popular models, notably the Gonzalo and Granger (1995) and Hasbrouck (1995) models to investigate the

relative contribution of price discovery to closely related markets (Booth, So, & Tse, 1999; Chu, Hsieh, & Tse, 1999;

Hasbrouck, 2003; Ivanov, Jones, & Zaima, 2013; So & Tse, 2004; Tse, 1999; Tse, Bandyopadhyay, & Shen, 2006; Yang,

Yang, & Zhou, 2012).

Another strand in literature examines the impact of size and sector on price discovery. However, we find that only a

few attempts have been made in this regard. Two important studies are Narayan and Sharma (2011) and Narayan,

Sharma, and Thuraisamy (2014), which propose a panel data model of price discovery, using panel cointegration and

the panel VECM (PVECM) method, to investigate how size and industry impact the price discovery process.

Some researchers have studied price discovery and examined the determinants of information shares. Mizrach and

Neely (2008) are the first to demonstrate that market state‐related variables, such as spread, traded contracts, and

volatility can explain price discovery shifts between the US treasury spot and the futures market. Following Mizrach

and Neely (2008), other researchers also conduct similar studies for different financial markets (Chen & Gau, 2010;

Chen, Chung, & Lien, 2016; Fricke & Menkhoff, 2011; Frijns, Gilbert, & Tourani‐Rad, 2015; Frijns, Indriawan, &

Tourani‐Rad, 2015). Besides examining the impact of the state‐related variables, some of these studies also investigate

the effect of macroeconomic news on price discovery.

As previously mentioned, most studies have been conducted in the context of developed countries, particularly, the

United States. However, very few studies have investigated the price discovery process using data from the Indian stock

market. Some studies (Karmakar, 2009; Rajput, Kakkar, Batra, & Gupta, 2012) have examined the relationship between

the spot and futures markets using daily S&P CNX Nifty spot and futures prices. However, they have simply used the

VAR model and VECM, and their approach focuses merely on the lead–lag relationship between the spot and futures

markets. Using models developed by Gonzalo and Granger (1995) and Hasbrouck (1995), some researchers have

investigated price discovery taking high‐frequency data from the Indian stock market. Kumar and Tse (2009)

investigate the average monthly price discovery over all stocks based on both the models from January 2004 to

December 2004, and observe that the relative contribution of spot and futures markets to the price discovery does not

remain the same throughout the year but varies with time. However, they do not investigate price discovery at the

individual stock level. On the contrary, Kumar and Chaturvedula (2013) examine the relative price discovery of spot

and futures markets at the individual stock level for the period between January 2004 and March 2007. However, while

they examine price discovery for the entire period, they do not investigate whether it is time‐varying at the individual

stock level. Aggarwal and Thomas (2011) examine price discovery from March 2009 to August 2009; however, their

emphasis is slightly different as they test the role of liquidity and price movement in the price discovery process

between the spot and futures markets.

From the above discussion, it becomes clear that none of the above studies investigate whether price discovery is

time‐varying at the individual stock level. Also, they use data from the past decade, and two of them (Aggarwal &

Thomas, 2011; Kumar & Tse, 2009) use data for a very short period. The Indian economy, in the current decade, has

achieved a remarkable GDP growth rate to an extent that it is now projected as the world's fastest growing major

economy. In keeping pace with the rapid economic growth, there has been remarkable progress in single stock futures

(SSFs) trading in the Indian market, making the National Stock Exchange (NSE) in India the most vibrant SSF market

in the world. Therefore, it would be interesting to more comprehensively examine price discovery using recent data

from the vibrant stock market of the world’s fastest growing major economy. Hence, we investigate price discovery in

the Indian stock market using high‐frequency data for a period of 5 years, from January 2012 to December 2016.

The present study differs from the previous studies that use data from the Indian stock market in several aspects.

First, we use all the three standard econometric approaches, that is, the component share (CS) method of Gonzalo and

Granger (1995), information shares (IS) of Hasbrouck (1995), and MIS of Lien and Shrestha (2009), whereas the other

studies (Aggarwal & Thomas, 2011; Kumar & Chaturvedula, 2013; Kumar & Tse, 2009) only use one or, at the most two

standard approaches, or simply use the VAR model and VECM (Karmakar & Inani, 2019; Rajput et al., 2012). We also

use all the three popular techniques on high‐frequency data of a relatively longer and more recent period than the

KARMAKAR AND INANI

|

1323