Monday, January 08, 2007

EEM and FXI: The Split Personality of Emerging Market ETFs

In an earlier post, I observed that the emerging markets ETF (EEM) traded very similarly to the S&P 500 ETF (SPY). Both, for example, displayed an upward trend during the period from the market close to the next market open, but actually lost ground during the U.S. day trading session, from open to close. Let's take a closer look at this pattern and bring in a third market, the iShares FTSE/Xinhua China 25 Index Fund (FXI).

Going back to 2005, we see that SPY has gained 25.31 points due to U.S. overnight change (close to open), but has lost 5.64 points during its day sessions (open to close). The pattern is even more pronounced in FXI: from close to open, the ETF gained 70.13 points, but during the U.S. day session (open to close) it lost 19.65 points. And with EEM, the contrast is even more stark: from close to open, it gained 73.31 points, but from open to close, it has lost 30.36 points.

All of this makes me wonder how traders are pricing emerging markets during U.S. trading hours when the bourses for those emerging nations are, for the most part, closed. Going back to 2005 (N = 506 trading days), the correlation between day session change in SPY (open to close) and day session change in FXI has been .69. The correlation between SPY and EEM during the day session has been .75. This suggests that, during the U.S. day session, traders are pricing the ETFs for emerging markets by the only benchmark they have available: U.S. stocks. That raises the possibility that, in a relative information vacuum--when the exchanges in those emerging countries are closed--ETFs could be mispriced due to U.S. daytrading. Such mispricing could lead to exaggerated selling during the day session, which is corrected in subsequent U.S. overnight trade.

Could a trader exploit such an inefficiency? I went back to the start of my data in FXI (October, 2004) and note that, when the day session in FXI has been up (N = 268), the next day in FXI (close to close) has been up by an average of only .04% (138 up, 130 down). When the day session in FXI has been down (N = 295), the next day in FXI (close to close) has been up by an average of .21% (169 up, 126 down). This suggests that the emerging markets may be reversing selling that occurs during the U.S. day session.

Clearly this is but an opening step in ongoing research into these patterns. What seems clear is that the emerging markets ETFs are performing very differently during the U.S. day session (open to close) than during the U.S. overnight period (close to open). It also appears that the price changes in the emerging markets ETFs during U.S. trading hours are highly correlated with the price changes in U.S. shares.

In short, perhaps we're trading those derivatives as if they're American stocks, even though their underlying assets are not American. It is not clear to me why events that impact U.S. stocks, from trendline breaks to news reports to company earnings to economic releases, should exert a similar--and even exaggerated--reaction from ETFs that hold emerging country and Chinese assets. The result has been a split personality in emerging market ETFs: a bullish performance from close to open and a bearish one from open to close.