Conducting Event Studies With Asia-Pacific Security Market Data

Abstract

We investigate the effectiveness of standard event study procedures with Asia-Pacific security returns data. Monte Carlo simulation experiments reveal that standard event study test statistics often exhibit modest misspecification under a null hypothesis of no abnormal performance. This holds for both parametric and nonparametric tests. Test misspecification is severe with clustered event dates or a doubled event date returns variance.JEL Classification: C12, C14, C15, G14, G15 Conducting Event Studies With Asia-Pacific Stock Market Data AbstractWe investigate the effectiveness of standard event study procedures with Asia-Pacific security returns data. Monte Carlo simulation experiments reveal that standard event study test statistics often exhibit modest misspecification under a null hypothesis of no abnormal performance. This holds for both parametric and nonparametric tests. Test misspecification is severe with clustered event dates or a doubled event date returns variance.

Event studies: A methodology review

“…In simulation experiments based on New York stock exchange data, Boehmer et al (1991) and Corrado and Truong (2008) report that this test statistic is well specified with good power properties.…”

“…The most successful among these tests were the nonparametric sign and rank tests advanced in McConnell and Muscarella (1985), Corrado (1989), Lummer and McConnell (1989), Zivney and Thompson (1989), and Corrado and Zivney (1992). Well-known studies of this type are Chandra et al (1992Chandra et al ( , 1995, Cowan (1992Cowan ( , 1993, Wasley (1993, 1996), Cowan and Sergeant (1996), Giaccotto and Sfiridis (1996), Affleck-Graves et al (2000), Dombrow et al (2000), Seiler (2000), Bartholdy et al (2007), and Corrado and Truong (2008). These studies generally concluded that sign and rank tests are well specified and provide an improvement in test power compared to standard parametric tests.…”

“…More recently, Bartholdy et al (2007) found that rank and sign tests outperform parametric tests with security returns data from the Copenhagen stock exchange. In a comprehensive study of event study test statistic performance with data from eleven Asia-Pacific stock exchanges, Corrado and Truong (2008) found that standard parametric event study tests can be poorly specified with these returns data, but nonparametric rank and sign tests are both well specified.…”

“…However, like the Patell T-test, the Boehmer-Musumeci-Poulsen T-test may not perform well when the distribution of security returns data departs markedly from a normal distribution. Corrado and Truong (2008) found that the BMP T-test tends to reject a true null too often when used with data from the American stock exchange, Nasdaq, and stock exchanges in the Asia-Pacific region. Bartholdy et al (2007) report a similar finding with security returns data from the Copenhagen stock exchange.…”

Market Reactions to Increased Reliability of Sustainability Information

“…While the metric proposed by Brown and Warner (1985) is a parametric test statistic, the metric developed by Corrado (1989) is a rank-based nonparametric test statistic. For a discussion of the statistical advantages and attributes of these specific test statistics, see Warner (1980, 1985) and Corrado and Zivney (1992), Corrado and Truong (2008).…”