
Steven DeSanctis, Head of U.S. small cap strategy at Merrill Lynch, has tried to answer all these questions in a report published today. The report tries to analyze the impact the ETFs, specifically the $16 billion iShares Russell 2000 Index Fund (IWM), have on small cap stocks.
According to the study, ETFs have had a substantially effect on trading volume. They impacted both the relative and absolute performance, with managers struggling to beat the benchmarks when ETF flows were strong, and outperformed when the flows were weak.
In 2011, the study estimates that the IWM, on an absolute basis each day, saw $384.2 million dollars of flows, which included both inflows and outflows. The report then proceeds to calculate the theoretical dollar amount traded each day by each name by taking the current IWM holdings and their respective weights in the ETF and multiplying each name by the $384.2 million. The average number of shares that could be hypothetically traded by the ETF on a daily basis is then deduced by dividing this theoretical dollar amount by the stock’s current price. Finally, this average number of shares is divided by each stock’s 52-week average daily volume to get the percent of trading volume potentially traded by the IWM. Based on market cap bucket, the universe is quintiled, and breaking it up by size bucket, the study shows that ETF could be 7.7% of the volume for names under $250 million compared to 3.7% for companies with market caps above $1 billion.
DeSanctis then goes on to analyze correlations, and according to him, in the fourth quarter of 2011, the pair-wise correlation in the Russell 2000 was 54.5%. The report reveals that since the IWM was introduced in 2000, pair-wise correlations in small caps have been on the rise. Post the introduction of IWM in 2000, pair-wise correlations in small caps have risen substantially, averaging 25.0%, which is much higher than the average of 5.3% recorded prior to the introduction of the IWM in the second quarter of 2000. The report concludes that ETFs have contributed significantly to the increase in pair-wise correlations.

