I realize that this title is a bit racy but I’ve become very concerned about an overcrowded trade or investment and the purpose of this article is to share my professional experiences and insights to highlight a possible future problem (probably unavoidable) imbedded in the current structure of market participants and namely that of ETF’s (Exchange Traded Funds) or ETP’s (Exchange Traded Products). I am a fan of ETF’s but I am afraid they have become too large and too influential in the underlying increasing volatility of equities they hold and the impact of the arbitrage mechanism that aligns the NAV of the ETF with the underlying security prices. I am not suggesting in this article that ETF’s will cause a market crash. What I am suggesting is that when that day comes, ETF’s will be blamed for facilitating a much larger dislocation than what would have occurred otherwise. Hear me out.
In mid-2007, while CEO of the liquid alternative business at Citi with $33 billion in assets, I wrote an article that postulated the following question, “Could a Credit Default Swap (CDS) cause a corporate bankruptcy?” A buyer of a CDS is in effect buying insurance against an issuer that is unable to make interest payments or may default into bankruptcy. Back in 2007, I was both amazed and alarmed at the growth of the CDS market. The CDS market had grown to over $45 trillion which was twice the size of the US Equity market at $22 trillion and between 8 to 10 times the size of the mortgage and US Treasury Markets at $7.1 trillion and $4.4 trillion respectively. In theory and mostly in practice, CDS’s have no linkage to the underlying issuer but rather their pricing reflects the risk preferences of market participants who are either hedging the exposure of an debt issuer owned or speculating on a issuers creditworthiness. Clearly in 2007, speculators were causing an explosion of growth in assets in the CDS market. Further, and more on this later, speculators in the CDS market are informed traders. This means that their trading and subsequent price impact in the CDS market signals either growing concern or less concern in the underlying issuer as the CDS price increases or decreases based on perceived real knowledge. It was and still is the case that just because a CDS rises in price, it does not follow that the underlying issuer’s price will decline as well.
That wasn’t the case though in 2007 and 2008. Speculators in the CDS market smelling a rout, cleverly and with malice (love that word), purchased CDS’s betting on a rising cost of insurance for the underlying issuer and then executed what is euphemistically referred to as a “Texas hedge”. The CDS buyer on significant derivative based leverage, stepped into the equity market and shorted the underlying issuer’s equity creating a self-reinforcing feedback loop that saw the entire capital structure of the underlying company collapse on the perception (in some cases real), that the issuer was in trouble. The issuer’s stock price would fall through excessive shorting. The CDS price would rise as the stock price fell and owners of the underlying bond issuer would sell as bond prices fell. This happened because the sheer size of the CDS market relative to the underlying market provided a direct incentive for CDS buyers to enter into the same sided trade in the equity markets as ‘informed traders’ causing uninformed traders (hedgers, and truly uninformed market participants) to change sides and sell alongside the CDS buyers. In a scenario that was deemed impossible, the CDS market linked to the underlying by merely acting and conveying ‘informed trading’ behavior. The lesson here is that there is information in price and in some cases can cause completely irrational behavior of market participants by signaling the perception of informed trading to uninformed traders.
While clearly not the same time period or market altogether, I see the same potential for a significant market dislocation (crash) that could be facilitated by ETF’s. Let me explain.
The global ETF (Exchange Traded Fund) market is now over $3.2 trillion by current estimates and growing. In the U.S. alone, there is an estimated $2.4 trillion. ETF’s are in most cases, passively managed or algorithmically active funds that are traded intraday. These funds (securities) own underlying equities that in most cases mimic an index. In other words, ETF’s and their trading or rebalancing behavior is uninformed. The ETF’s popularity to the investing public is obvious as it represents a very low cost, liquid alternative to owning a mutual fund or owning a basket of equities to emulate an exposure (e.g. tech) to a portfolio. ETF’s are very popular with advisors, retail investors and institutions and their design efficiency and effectiveness in delivering the desired portfolio exposure and return is proven. But here is where I grow concerned and you should be too.
In a recent academic paper written in 2015 by David, Franzonni, and Moussawi titled, “Do ETF’s Increase Volatility”, the authors empirically demonstrate some disturbing facts. First, ETF’s are less expensive to trade than the underlying equities they hold (a clear benefit to the owner). However, ETF’s are causing volatility of the underlying equities to increase for a couple of important reasons. First, there is the impact that the arbitrage mechanism has on the underlying equities. In an ETF that mimics an index or index-like portfolio, ETF price deviations from the net asset value (NAV) of the index holdings are arbitraged by traders or arbitrageurs by trading the underlying securities in the same direction as the NAV direction. What causes deviations from the NAV for ETF’s are the demand for creation of ETF’s and supply (selling pressure) of ETF’s. So this means that an ETF that owns AAPL (highest share count ownership by ETF’s) and sees its NAV trade lower to the underlying portfolio of securities will sell AAPL as arbitrageurs seek to lock in the price differential by buying the ETF and selling AAPL. As it turns out, these arbitrageurs are in most cases high frequency traders (HFT’s) who make a living working in equities dominated by ETF’s. The HFT’s, unfortunately in market participant parlance, are both informed and uninformed traders depending on any given market environment. The second reason that underlying equity volatility is increasing for ETF dominated securities is that as pointed out earlier, ETF’s have a tighter bid ask spread than the underlying stock they own in most cases. To the extent there is ETF motivated buying or selling in an underlying equity, the trading impact is greater in the underlying equity.
Now, here is where it gets interesting. As mentioned before, the U.S. ETF market is roughly $2.4 trillion in size. Of that $2.4 trillion, roughly 80% exists in the form of passively traded, index like ETF’s. The current U.S. equity market capitalization is roughly $27 trillion. So, close to 10% of the U.S. equity market capitalization is represented by ETF’s. This is clearly not as significant as the CDS market was to the underlying debt markets in 2007. But, as you’ll see, it doesn’t take a lot of informed trading to cause a market crash.
It is estimated that close to 70% (60% is estimated to be that of HFT’s alone and I’m certain there is double counting if you include ETF volume) of all volume generated on U.S. exchanges is attributable to high frequency traders and ETF’s. That is an astonishingly high figure and demonstrates the dramatic changes that have taken place to not only the percentage concentration of dominating market participants but the character of those participants as well. I am generally a proponent of HFT’s because I believe that in most orderly markets, HFT’s are liquidity providers and help facilitate other market participant trading at a relatively cheaper cost. Perhaps one way of characterizing the current market participants is by describing a typical trading day where noise traders or uninformed traders are speculating or hedging and HFT’s (both informed and uninformed)are algorithmically trading very short term anomalies (arbitrage opportunities), informed speculators are investing based on new information and ETf’s are being traded by uninformed traders because of uninformed speculation or arbitrage. What happens when this current percentage equilibrium of market participants changes dramatically as it did in 1987 and again in August 2015?
I was a young trader on the equities trading desk at Goldman Sachs during the October 1987 crash (I seem to have a knack for being around calamities). I watched my derivative colleagues putting on seemingly insane S&P 500 futures trades at incredibly steep discounts to the underlying index. In normal times when there was a significant discount of the futures to the underlying index, index arbitrageurs would step in and buy the futures and sell the underlying stocks to lock in profits from the obvious pricing discrepancy. On Black Monday, this arbitrage mechanism failed because the market participants including the index arbitrageurs perceived the sellers of the index futures (portfolio insurers and hedgers) as informed sellers and turned sellers as well. The portfolio insurers sold as the market declined and the fact that the S&P 500 futures spread was at the biggest discount it had ever seen, they just kept selling because their objective was to dynamically hedge a portfolio of existing securities as the markets fell. To the uniformed index risk arbitrageurs and other market participants, the irrational selling seemed informed and caused further market participants to sell. How much portfolio insurance trading volume was attributable to facilitating the crash? Not much as it turns out. The estimates vary but the range of portfolio insurance using S&P 500 futures was approximately 6% of all futures trading. In other words, just 6% of the trading by one type of market participant facilitated a crash. By the way, Soros did sell over 1000 S&P 500 futures at the low and Paul Tudor Jones bought them.
There have many academic papers that have been published with ground breaking market microstructure work that demonstrates pretty conclusively that crashes can be caused by very little trading by perceived (important word) informed traders.
Now, let’s examine the “Flash crash” of August 2015 which did involve ETF’s, and a more current set of market participants. While I would prefer not to go into great detail as to the reasons for this ‘crash” (I would refer readers to an excellent piece written by Blackrock titled, “US Market Equity Structure: Lessons from August 24” back in October 2015, I would suggest that the U.S. markets were limit down before the opening on the back of extreme bearish market sentiment overseas. What is fascinating about this is that due to circuit breaker and other SEC rules, it was almost impossible to get price discovery in the ETF’s and other markets for that matter. To make matters worse, the three major exchanges NYSE, NASDAQ and BATS were not coordinating price discovery in an orderly fashion. When trading opened, many ETF’s gapped down at 20% discounts based on supply and demand causing some of the more widely held underlying equity holdings to open at steep discounts to the previous close because of ETF selling. Blackrock who has significant ‘skin in the game’ with its iShares business laid out a whole host of structural change solutions that were largely adopted by the SEC. These changes include, eliminating the time periods when securities could trade without limit up and limit down (LULD) bands in place; reducing the number of trading pauses; standardizing automated re-openings following a pause in trading; and eliminating clearly erroneous execution rules when LULD bands are in effect.
However helpful these rule changes may have been in creating stability in the market pricing structure in general, it does not address how market participants interact with each other in stressful market conditions. There is no market mechanism to stop how perceived informed traders can cause other market participants to change behavior and sell alongside.
The main reason that I grow concerned about the changing market dynamics and ETF’s in particular is that ETF’s like equity prices are immediately discoverable. Ten years ago, ETF’s and HFT’s were not the dominate players and further mutual funds were not high turnover vehicles. It will not take much of a catalyst to cause a substantial market decline and I’m afraid that ETF’s will facilitate a much larger decline than would have otherwise occurred. The ETF arbitrage mechanism being what it is creates built- in inherent pressure for the underlying securities of the ETF. ETF’s represent 10% of the overall market capitalization of the U.S. market and combine that with the HFT’s, it will provide more than ample fuel for a big decline should and when that comes. ETF’s selling, followed by selling in the underlying stocks, followed by an absence of HFT participants and no liquidity or worse selling by HFT’s, gap pricing all wound up in one self-fulfilling feedback loop. That is a realistic scenario.
While I am a big fan of ETF’s, I worry when I see enormous growth in a security or theme and always think about unintended consequences of that growth. I hope I am wrong on my hypothesis, but I’ve experienced many crashes and the underlying mechanics of investor panic generally do not change.
As Michael Lewis pointed out in his most recent book, “The Undoing Project” which describes the brilliant work done by psychologists, Daniel Kahneman and Amos Tversky, behavioral finance demonstrates both damagingly and conclusively that humans make irrational economic decisions under uncertainty and risk. The asymmetry of information response and how market participants interrelate with one another is why financial panics happen more frequently than we care to admit.