Research in finance has aggregated together cross-listed and non-cross-listed stocks and finds that, on average, value stocks outperform growth stocks, small cap stocks outperform large cap stocks, low liquidity stocks outperform large liquidity stocks and low volatility stocks outperform high volatility stocks.
But in countries like Canada where cross-listed firms are numerous, is it right to aggregate cross-listed and non-cross-listed stocks? Are these two groups of stocks homogeneous?
First, cross-listed stocks tend to be larger, more glamorous and well known than non-cross-listed stocks and are followed by more analysts. This implies that non-cross listed stocks tend to have lower multiples and attract more value investors than cross-listed ones.
Second, there is evidence that the marginal trader for Canadian cross-listed stocks is an American investor with cross-listed stocks acting a lot like U.S. ones which are driven by the different taxation and institutional framework in the U.S..
In light of this, in a recent paper using cross-listed and non-cross-listed Canadian stock market data for the period 1982-2012, I set to examine the following related questions. Do cross-listed and non-cross- firms behave similarly or there are distinct differences between them that make grouping cross-listed and non-cross-listed stocks in one basket bias results and produce unfounded generalizations and conclusions? Is it possible that the previously stated anomalies, namely the value premium, the size premium, the liquidity premium and the volatility premium, are present only in non-cross-listed stocks? Is it possible that the above mentioned anomalies recorded in the aggregate market data are a result of the fact that non-cross-listed stocks are more numerous than cross-listed stocks and thus dominate the aggregate numbers – not truly a whole market effect?
I find distinct differences between cross-listed and non-cross listed firms with size premiums, with the volatility premium differences being the most noticeable.
I also find that the driver of the size premium in aggregate data are non-cross listed firms. For example, for the full sample period, the median size premium for cross listed firms is 1.4% vs. a median size premium of 7.3% for non-cross listed firms. As far as sub-periods are concerned, the median size premium for cross-listed firms is 2% for 1982-1997 and 0.7% for 1998-2012, with the corresponding values being 5.8% and 8.7%, respectively for non-cross listed firms.
Given that non-cross-listed stocks are more numerous than cross-listed ones in the aggregate sample, this finding means that the size premium observed in aggregate data is not as generalizable as one may think.
More interesting, in light of prior evidence, is my finding with regards to the volatility or vol effect.
I find that low vol cross-listed firms have lower return than high vol firms. This contradicts findings by earlier studies that used aggregate data, whereas the vol effect for non-cross listed firms is consistent with previous evidence. For example, for the whole sample period, the median vol premium for cross-listed firms is -1.3% vs. a vol premium of 2.9% for non-cross listed firms.
As far as sub-periods, the median vol premium for cross listed firms is -3% for 1982-1997 and 0% for 1998-2012.
The corresponding values for non-cross listed firms are 4.7% and 1.6%, respectively.
In other words, the driver of the vol premium in aggregate data is the non-cross listed firms.
Given the larger number of non-cross listed firms, the non-cross listed vol effect is dominating the aggregate data giving the impression that the vol effect is pervasive. My research shows that this is not true — the disaggregation of data enables me to discern this.
In other words, my findings suggest that despite low vol investing strategies’ increased popularity in recent years, investors should be cautious.
In light of the differences I found between cross-listed and non-cross-firms, grouping cross-listed and non-cross-listed stocks into one basket biases results and produces unfounded generalizations and conclusions.