International Review of Economics and Finance 8 1999 375–397
The intraday relation between return volatility, transactions, and volume
✩
Xiaoqing Eleanor Xu
a
, Chunchi Wu
b,
a
School of Business and Administration, St. Louis University, 3674 Lindell Blvd., St. Louis, MO 63108, USA
b
School of Management, Syracuse University, Syracuse, NY 13244-2130, USA Received 5 February 1998; accepted 16 July 1998
Abstract
In this article, we examine the relation between return volatility, average trade size, and the frequency of transactions using transaction data. Consistent with Jones, Kaul, and Lipson
1994. Review of Financial Studies, 7, 631–651, our results show that the frequency of trades has a high explanatory power for return volatility. However, contrary to their finding, we find
that average trade size contains nontrivial information for return volatility. The positive relation between return volatility and average trade size is more significant for actively traded stocks.
Furthermore, return volatility exhibits significant intraday variations. It is found that the effect of trade frequency on return volatility is much stronger in the opening trading period.
1999
Elsevier Science Inc. All rights reserved.
JEL classification: G10; D82
Keywords: Information asymmetry; Trade size; Frequency
1. Introduction
Financial researchers have paid a great deal of attention to the relation between trading volume and stock return volatility. Empirical studies have generally found a
positive relation between return volatility and transaction volume.
1
Karpoff 1987 conducted a comprehensive survey on the relation between price changes and trading
volume. Schwert 1989, Gallant et al. 1992, and Foster and Viswanathan 1993 found a positive correlation between stock return volatility and trading volume.
The positive correlation between volatility and volume supports the Mixture of
✩
An earlier version of this paper was presented at the 1997 Eastern Finance Association Meetings. Corresponding author. Tel.: 315-443-3549; fax: 315-443-5389.
E-mail address : cwusyr.edu C. Wu
1059-056099 – see front matter
1999 Elsevier Science Inc. All rights reserved. PII: S1059-05609900029-5
376 X.E. Xu, C. Wu International Review of Economics and Finance 8 1999 375–397
Distribution Hypothesis MDH. Originally developed by Clark 1973 and subse- quently extended by Tauchen and Pitts 1983, Lamoureux and Lastrapes 1990, and
Anderson 1996, the MDH postulates that return volatility and volume are positively correlated because both are related to the underlying information flow. The strong
contemporaneous correlation between trading volume and return volatility docu- mented by previous studies may well reflect their joint dependence on the underlying
latent events.
Recently, Jones, Kaul, and Lipson 1994; hereafter JKL documented striking evi- dence for the role of the frequency of trades in determining the volatility of returns.
They found that the positive relation between volatility and volume actually reflects the positive relation between volatility and the frequency of transactions.
2
On the other hand, the size of trades has virtually no information beyond that contained in
the frequency of transactions. Based on this finding, they concluded that the number of transactions contains all the pertinent information to the pricing of securities.
The finding that the size of trades contains no information is puzzling. Financial theories, represented by both the competitive and strategic models, have predicted a
positive correlation between the size of trades and volatility. In competitive models Pfleiderer, 1984; Grundy McNichols, 1989; Holthausen Verrecchia, 1990; Kim
Verrecchia, 1991, the trade size is shown to be positively related to the quality of information, and insiders prefer to trade a large size at any given price. In strategic
models with a monopoly informed trader Kyle, 1985; Admati Pfleiderer, 1988; Foster Viswanathan, 1990, a camouflage of one large trade into several small-
sized trades tends to obscure the relation between volatility and trading size. However, in a more realistic setting with multiple informed traders, Holden and Subrahmanyam
1992 showed a similar positive relation between trading size and the quality of information possessed by informed traders. Thus, both competitive and strategic mod-
els predict a positive relation between the size of trades of informed traders and the quality of their information and hence a positive relation between average trade size
and price changes. The predictions of these theoretical models are apparently at odds with JKL’s finding.
Furthermore, it has often been argued that it takes volume to move prices. Technical securities analysts use volume data extensively to predict future price movements.
Grundy and McNichols 1989 demonstrated the informational role of volume and its applicability for technical analysis. Blume et al. 1994 showed how trading volume,
information precision, and price fluctuations are related. They found that traders who use the information contained in volume obtain higher-quality private signals than
traders who do not. The finding that average trade size contains no information would seem to be inconsistent with the volume-based technical trading activities observed
in security markets.
In this article, we attempt to provide further evidence on the role of the size of trades in return volatility at the intraday interval. Like JKL’s 1994 study, we analyze
the relation between transactions, average trade size, and stock return volatility. However, our study differs from theirs in several key aspects. First, while they used
daily data, we use intraday data. The use of intraday data allows us to better capture the relation between volume and stock price dynamics as predicted by microstructure
X.E. Xu, C. Wu International Review of Economics and Finance 8 1999 375–397 377
theory. A main objective of our study is to see if JKL’s results hold at the intraday interval. Also, it will be interesting to see whether the volatility–transaction relation
varies over intraday periods. Second, we use the generalized method of moments GMM in empirical estimation.
The GMM model invokes much weaker distributional assumptions on stock returns. Within the GMM framework, the disturbance terms of the volatility regressions can