Blog - Investing Notes

August 13, 2007 - Exploiting Global Plunge Contagion?

Reader David Zaitzeff poses the following question:

"In the last several weeks, there have been several times when the Dow Jones Industrial Average (DJIA) was down a lot, and Asian stock markets followed it down the next day. How reliably do Asian stock markets follow sharp drops in the U.S. stock market?"

To answer this question, we first examine the overall relationship between the U.S. stock market (DJIA) and Asian stock markets (Hang Seng and Nikkei 225). Then, we focus on what happens in Asian stock markets the day after sharp drops in the U.S. market. Using daily closing levels of the DJIA, the Hang Seng index and the Nikkei 225 index for 12/31/86-8/9/07 (roughly 5300 trading days), we find that:

Since the three stock markets under consideration have different operating schedules, we adopt the following rules for aligning index level data:

1. If the U.S. market is closed, it cannot affect Asian markets, so we exclude such dates from the analysis.
2. If an Asian market is closed for multiple days during which the U.S. market trades, we assume that the influence of the U.S. market on the next trading day for the Asian market is cumulative.

The following scatter plot depicts the relationship between the daily change in the Hang Seng index and the prior day change in the DJIA over the entire sample period. The Pearson correlation between these two series is 0.37, indicating that the Hang Seng index follows the DJIA to some degree. The R-squared statistic is 0.14, suggesting that DJIA behavior explains 14% of next-day Hang Seng index behavior. Excluding the outlier in the lower left corner of the plot (10/23/87-10/24/87) reduces the Pearson correlation to 0.35 and the R-squared statistic to 0.12.

The relationship between same-calendar-day changes in the DJIA and the Hang Seng index is much weaker, with Pearson correlation 0.14 and R-squared statistic just 0.01, supporting a view that the U.S. market leads rather than lags the Chinese market.

Is the the Hang Seng index lag effect more reliable for some ranges of DJIA changes (for example, plunges) than others?

The next chart modifies the preceding one by ordering the changes in the DJIA from most negative to most positive, starting with the 10.4% drop on 10/23/87 and ending with the 6.3% gain on 7/24/02. (The Chinese market was not open the day after the DJIA plunged 22.6% on 10/19/87.) The horizontal axis is not time-sequential. The chart shows that Hang Seng index daily behavior is fairly noisy (unreliable) with respect to all ranges of prior-day DJIA behavior. There is a hint that the relationship may be most reliable at the extremes.

Does a close-up on the "plunge" end indicate reliability?

The next chart is a close-up of the left end of the prior one, including only those data pairs for which the DJIA falls by at least -2.5%. The average next-day Hang Seng index return for this subsample is -2.7% with standard deviation 4.9%. Excluding the extreme 10/23/87-10/24/87 data point, the average next-day Hang Seng index return is -2.2% with standard deviation 3.1%. These statistics suggest that persistently shorting the Hang Seng index on the day after DJIA plunges would be profitable over the long term. However:

1. The Hang Seng index tends to gap down at the open after DJIA plunges, limiting the degree to which a trader can exploit its "follow-the-leader" statistics. Using open-to-close rather than close-to-close changes in the Hang Seng index results in an average next-day Hang Seng index return of -0.7% after the specified DJIA plunges. In other words, a large component of the plunge contagion is an untradable overnight or weekend adjustment.
2. There are only 65 relevant plunges in the DJIA over more than 20 years, and these plunges are very sporadic (13 in 1987, none during 1992-1995, 11 in 2002 and none during 2004-2006). Annualized results of a plunge contagion strategy, even if one could capture overnight gaps, would therefore be modest.

The relationship between daily changes in the Nikkei 225 index and prior day changes in the DJIA is somewhat weaker than that found above for the Hang Seng index, as follows:

• The correlation between daily changes in the Nikkei 225 index and prior-day changes in the DJIA is 0.34 (compared to 0.37 above). The R-squared statistic is 0.12 (compared to 0.14 above), indicating the prior-day DJIA changes explain 12% of daily changes in the Nikkei 225 index.
• For the 71 observations when DJIA daily returns are -2.5% or less, the average next-day return for the Nikkei 225 index is -1.5% with standard deviation 2.4%. Excluding the most extreme outlier (10/19/87-10/20/87) reduces the average Nikkei 225 index loss to -1.3%. Using open-to-close rather than close-to-close changes in the Nikkei 225 index further reduces the average Nikkei 225 index loss to -1.1%.

In summary, Asian stock markets do to some degree follow the U.S. market, including its sharp drops. However, U.S. market plunges are rare and clustered, the Asian market responses to these plunges are noisy, and untradable overnight/weekend gaps down on Asian markets substantially limit exploitation of any "plunge contagion."

For similar research, see Blog Synthesis: Some Trading Indicators. Consider also browsing formal research on financial markets contagion by searching on key words such as "contagion" and "stock" in the Social Sciences Research Network eLibrary Database.