Review of Quantitative Finance and Accounting, 18: 219–237, 2002
2002 Kluwer Academic Publishers. Manufactured in The Netherlands.
The Evolution of Market Efﬁciency:
103 Years Daily Data of the Dow
ANTHONY YANXIANG GU
Assistant Professor of Finance, State University of New York, Geneseo
Telephone: (716) 245-5368, Fax: (716) 245-5467
Professor of Finance, The University of Illinois at Urbana-Champaign
Abstract. Autocorrelation in daily returns of the Dow 30 Index ﬂuctuates signiﬁcantly over time and reveals a
declining trend after World War II. The relation between autocorrelation and volatility is negative and nonlinear.
The relation between autocorrelation and volume is also negative and nonlinear. Returns exhibit positive auto-
correlation during years with higher autocorrelation, and negative autocorrelation during years with lower auto-
correlation. Positive autocorrelation appears more frequently during periods of low volatility, while negative
autocorrelation appears more frequently during periods of high volatility. Current period’s autocorrelation is
related to previous period’s autocorrelation and to both the previous and the current period’s volatility and rate
of return, which implies that investors incorporate previous period’s pattern of market behavior into their trading
Key words: autocorrelation, evolution, market efﬁciency
JEL Classiﬁcation: G1
The 1940s through mid 1970s saw the formation of random walk theory and the efﬁcient
market hypothesis and their domination over ﬁnancial economics, while experiencing the
highest level of autocorrelation between daily returns of the Dow 30 Index. Over the last
15 years, many ﬁnancial economists reported evidences of inefﬁcient market but the daily
returns of the Dow have revealed the lowest level of autocorrelation.
Previous studies about the existence of autocorrelation between returns have focused on
whether the stock market is weak-form efﬁcient, none has been found to study the evolution
of the level of efﬁciency. During the last 50 years, there have been signiﬁcant advances in
information technology, investors have become more experienced, the stock market more
developed, and the U.S. economy more mature. All these changes would have some impacts
on stock market behavior.
This study looks at the changes in the level of autocorrelation in relation to the advance
in information technology over time, and analyzes the relation between return volatility,