Working Paper: NBER ID: w3911
Authors: Wenling Lin; Robert F. Engle; Takatoshi Ito
Abstract: This paper investigates empirically how returns and volatilities of stock indices are correlated between Tokyo and New York. Intradaily data are used, so that daytime and overnight returns are defined for both markets. Tokyo daytime hours overlap with New York overnight hours, while New York daytime hours overlap with Tokyo overnight hours. We find that in general Tokyo (Mew York) daytime returns are significantly correlated with New York (Tokyo) overnight returns. This suggests that information revealed during the trading hours of one market has a global impact on the returns of the other market. One exception is that after the October 1987 Crash, the Tokyo overnight returns were not significantly affected by New York daytime returns. We propose and estimate a signal extraction model with GARCH processes to determine the global factor from daytime returns. This is the problem of setting the opening price of a domestic market conditional on the foreign daytime returns. We also investigate lagged return and volatility spillovers. Except for a lagged return spillover from New York to Tokyo for the period after the Crash, there are no significant lagged spillovers in returns or in volatilities.
Keywords: stock returns; volatility; international transmission; GARCH processes; signal extraction
JEL Codes: G15
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| Tokyo daytime returns (F29) | New York overnight returns (G19) |
| New York overnight returns (G19) | Tokyo daytime returns (F29) |
| New York daytime returns (C41) | Tokyo overnight returns (G19) |
| New York returns (Y60) | Tokyo returns (F29) |
| global factor from daytime returns (C22) | overnight returns (G14) |