The popularity of index funds (ETFs) has growth exponentially since year 2000. The low cost and transparency appeal of ETFs have attracted an ever growing number of ETF providers, products as well as investors:
Graph provided by iShares/BlackRock
Retail investors have an easy gateway to gain exposure to every asset class starting from U.S. government Treasuries (TLT), though broad range of equity ETFs down to more obscure asset classes like REITS (VNQ) or commodities (DJP).
With this rise in popularity more assets are diverted from conventional stock picking mutual funds in order to make room for ETFs. As a result, the mutual funds are losing shares while interest in ETFs is growing proportionally:
The mutual funds space, our prehistoric dinosaur, is dying a slow death to make room for more indexing products but there is one question no one seems to be asking: What are the consequences of this global shift in appetite?
Distortions in Commodities Markets
Recently I came across an investment manager's investor letter pointing out the distortions in futures markets caused by the increase in use of index funds in the commodities space. Intrigued, I decided to confirm his comments with my own research. According to Marit.com the assets under management (AUM) in the commodities ETF space grew 22% in 2012 to a record number $201 billion:
The theory is that the increase in number of futures ETFs, which buy the near term contracts (1 and 2 months ahead), causes a permanent and increased steepness of the futures term structure (contango). It happens due to the constant need of the index fund to roll the near expiring contracts into the longer dated contracts in order to prevent taking physical delivery of the contracts. As a consequence, a commodity with high physical demand and low relative supply can now trade in a contango rather than a natural state of backwardation. As a result, the cost of rolling a contract is much higher and the ability of using the futures term structure to read the demand/supply imbalance has been compromised.
This theory can be confirmed by a positive correlation between the changes in steepness of the futures term structure and the changes in a dealer's net long open interest (data provided by CFTC) which represents that increase in index fund holdings. According to Lake Hill Capital Management the correlation (R squared) between the average steepness of term structure and average dealer's net long open interest between June 2006 and March 2013 was 51.6%, which confirms the above theory.
Distortions in Equity Markets
The equity markets do not have a term structure so what kind of distortion could possibly the rise in index funds have on the stocks, one might ask? The answer is simple: higher correlations, lack of diversification and valuation ambivalence. When a unit of an ETF is created, the underlying index stocks are being bought on a ratio basis independent of their value or merit causing an increase in correlation between the "good" and "bad" stocks within the index. The more money that is in an ETF, the more profound the impact it has on the underlying stocks' correlations to each other. To test our hypothesis, we looked at two peak-to-trough-to-peak periods (2000-2007 and 2008-2013) in order to get similar market cycle time frames with comparative volatility and drawdown levels:
Additionally, we know that the assets in ETFs almost doubled from $800 billion at the end of 2007 to close to $1,600 billion in 2012. We selected a random sample of 20 stocks from the S&P 500 Index across all sectors and calculated an average cross correlation between the stocks for each time interval:
It clearly showed that during 2008-2013, when the stock market went through a similar peak-to-trough-to-peak period as 2000-2007, the cross correlation of S&P 500 stocks was much higher. What is even more profound is the fact that between 2010 and 2013 - a period after the spike in volatility which usually coincides with spike in correlations - the cross correlation was 0.41. In a similar period 2004-2007 the cross correlation was only 0.13, further confirming our hypothesis.
If the initial assumption of increased assets in ETFs causing higher correlations between stocks is correct, then the next distortion has to be the market's valuation ambivalence. If correlations between stocks go up, then it should be more difficult for active managers to outperform their benchmarks and for equity market neutral managers to produce positive returns. Indeed, the HFRX Equity Market Neutral Index had an average annual return of 3.24% over the period 2000-2007, compared to an average annual return of -1.85% for the period 2008-2013.
The growth of assets in index funds has had a significant impact on the markets and the way they operate. The distortions created by the use of ETFs in the form of exaggerated steepness of the futures term structure, higher roll costs, increased correlations and valuation ambivalence are here to stay, so one need to adapt to the new normal. The paradigm is changing and it might be that stock picking is an art of the past. It might be that alternative hedge fund strategies trying to produce alpha in a market neutral fashion are slowly getting suffocated by the increased correlations. It might be time to focus on dynamic asset allocation rather than single security selection.