April 15, 2012
Three years ago, when virtually nobody had yet heard of High Frequency Trading, Zero Hedge wrote “The Incredibly Shrinking Market Liquidity, Or The Upcoming Black Swan Of Black Swans” in which we asked “what happens in a world where the very core of the capital markets system is gradually deleveraging to a point where maintaining a liquid and orderly market becomes impossible: large swings on low volume, massive bid-offer spreads, huge trading costs, inability to clear and numerous failed trades?” Subsequent to this, our observation was proved right on both an acute (the May 6, 2010 Flash Crash), and chronic (the nearly 50% collapse in average daily volumes since the 2008 top) secular basis. And while we are not happy to have been proven correct in this particular forecast, as it ultimately means the days of equity capital markets in their current configuration are numbered, we now note that none other than Morgan Stanley’s Quantitative and Derivative Strategies released a note which, with a three year delay, effectively predicts the end of capital markets in a world where every declining retail participation (another topic we have been hammering for the past 3 years as it is only the most natural response to a world in which not only equities are openly manipulated by central banks, but in which perpetrators for massive market disturabances are neither identified nor prosecuted) is replaced by artificial high frequency trading churn, which never was and never will be a true liquidity provider on a long-term basis.
To wit from Morgan Stanley: “In our mind, many of the approaches to algorithmic execution were developed in an environment that is substantially, structurally different from today’s environment. In particular, the early part of the last decade saw households as significant natural liquidity providers as they sold their single stock positions over time to exchange them for institutionally managed products… While the time horizon over which liquidity is provided can range from microseconds to months, it is particularly shorter-term liquidity provisioning that has become more common.” Translation: as retail investors retrench more and more, which they will due to previously discussed secular themes as well as demographics, and HFT becomes and ever more dominant force, which it has no choice but to, liquidity and investment horizons will get ever shorter and shorter and shorter, until eventually by simple limit expansion, they hit zero, or some investing singularity, for those who are thought experiment inclined. That is when the currently unsustainable course of market de-evolution will, to use a symbolic 100 year anniversary allegory, finally hit the iceberg head one one final time.
How does Morgan Stanley frame their analysis? First, MS notes the ever increasing ownership of the stock market by big institutions, as retail investors took a back seat to investment allocation decisions, a secular theme until 2008, which however has subsequently plateaued:
Asset management has become increasingly institutionalized over the years. Individuals have outsourced their wealth management to institutions, whether pension funds, insurance companies or investment advisors. These, in turn, invest mostly in institutionally managed products such as mutual funds, ETFs, or long/short vehicles. The net result of this is that the vast majority of investable assets are held through institutionally managed vehicles. Exhibit 1 shows the evolution of ownership vehicles of corporate equity in the US. 37% of the USD 22tn of corporate equity is held by ‘Households and Nonprofits’ now, down from 50% at the turn of the century. This segment includes endowments and foundations, as well as on-shore hedge funds. Arguably, these should be counted as institutional investors as well. This means that direct household ownership of corporate equity is substantially below this figure.
For a universe of large-cap stocks4, Exhibit 4 shows the evolution of the percentage of ownership attributable to 13F filers and mutual funds since Dec 2001. This data corroborates Exhibit 1 on the increase in institutional ownership – on average, institutional ownership increased from 54% in March 2000 to 81% at the end of 2011.
Following the rapid growth of institutional asset management, however, the pace of increase in institutional ownership has slowed since 2008. We see this as one of the key drivers of the change in market structure and liquidity sourcing opportunities in recent years.
As more and more “equity capital” was concentrated into the hands of fewer and fewer people, the only logical outcome took place:
Trading decisions have become more concentrated as asset management has institutionalized. There are fewer decision makers (fund managers) now than in a world where management is dispersed across households. The size of their parent orders, on the other hand, has grown. The basic set up of the market – in terms of a continuous auction limit order book supplemented with ‘upstairs’ solutions – has not changed. The details of the implementation have adjusted, of course – such as competition in execution venues, new order types, and greater use of technology.
One of the most significant results of the tension between fewer market participants and larger parent order sizes is that the share of ‘real’ trading volume has declined by around 40% in the last five years. In Exhibit 5, we show the average proportion of quarterly trading volume that is attributable to changes in the 13f filings of each institution. We use this as our definition of ‘real’ trading volume. We also calculate the trading volume from our separate dataset of mutual fund holdings. We aggregate ‘buys’ (positive position changes) and ‘sells’ separately for each institution and mutual fund, on a per-stock basis, and calculate the average percentage of volume across stocks in our universe.
As a result, reports of the market’s evaporating volume are not greatly exaggerated.
In our mind, these numbers constitute a lower bound on the amount of ‘real’ trading volume in the market (defined as the trading volume from market participants that hold assets for longer periods).
The share of real trading volume shows three distinct phases. From 2001 to 2006, institutional buys accounted for 27% of total trading volume on average, while sells accounted for 20%7. The asymmetry between buys and sells reflects the growth in the institutional share of ownership over the period (see Exhibit 4). From 2008 to 2011, we see a significantly lower share of trading volume – buys represent just 16%, and sells 13%, a drop in market share of almost 40%.
The 2007/2008 period represents a transition period, with a rapidly declining share of ‘real’ trading volume. This period coincided with rapidly increasing overall trading volume (Exhibit 6) – in 2006, the ADV was 4.8bn shares, while in 2008 it was 8.8bn shares. Volume has since declined again – YTD ADV is 6.9bn shares.
While it will not come as news to any of our regular readers, the disappearance of retail investors has meant the incursion of electronic trading in the form of relentless rise in HFT dominance.
Throughout the last decade, the share of institutional trading volume by each institution type has been remarkably constant (Exhibit 7). This means that the reduction in institutional share of overall trading volume between 2007 and 2008 was not due to a reduction in trading activity by any one institution type, but rather due to the introduction of a new type of trading volume.
A potential reason [for the drop in our measure of the share of ‘real’ institutional trading volume] is that institutional execution strategies have made liquidity more challenging to find. Concentration in assets under management has led to larger order sizes. One of the responses to this has been automation in execution strategies. The algorithms used tend to split parent orders into smaller child orders. As a result, we find that block trades, which made up around 30% of trading volume in 2003, accounted for just over 5% of trading volume in 2011 – see Exhibit 8. At the same time, the average trade size has fallen to around 250, from more than 1,000 back in 2003. Both data series are based on NYSE listed stocks.
Next Morgan Stanley explains precisely why the current market is no longer fit to deal with the existing roster of players, fit for a previous iteration of capital market topology such as that which prevailed when Reg-NMS was conceived, but certainly not the current one, especially if retail continues to withdraw from trading equities and invest its cash forcibly into that other terminally epic bubble – bonds.
In our mind, many of the approaches to algorithmic execution were developed in an environment that is substantially, structurally different from today’s environment. In particular, the early part of the last decade saw households as significant natural liquidity providers as they sold their single stock positions over time to exchange them for institutionally managed products. Adjusting the institutional execution strategy to capture this liquidity was a rational thing to do.
This institutionalization of asset management is mostly done by now, as we showed in Exhibit 4. As a result, execution strategies that were calibrated on the earlier market environment may no longer be optimal. The rise in trading volume since 2007 (when the growth in institutional ownership leveled off) reflects the growing challenges of sourcing liquidity. The way this has been resolved is through the introduction of more ‘market making’ activity in the form of liquidity provider trading.
Let’s repeat that for the cheap seats: “As a result, execution strategies that were calibrated on the earlier market environment may no longer be optimal“ and we could in theory just end it here.
We all know that the bulk of HFTs close each day flat to avoid overnight holding risk, which they do by increasing churn amongst each other to unprecedented levels, in the process generating massive momentum swings as every player piggybacks on either side of the move. End result:even Moran Stanley admits that churn is not liquidity, and that the inability of HFT to carry inventory and have a longer-term bias is the fatal flaw in the current market topology, precisely as we warned back in April 2009!
The risk-carrying capacity of these providers is limited. If natural liquidity does not materialize, they may trade with another intraday liquidity provider to manage their inventory. This is particularly true if the institutional parent orders are larger and hence typically longer lasting.
From here, everything else follows:
Typical market-making liquidity provisioning strategies can be modeled as mean reversion strategies. If liquidity demand is persistently one-sided (such as in the case of large parent orders), it is rational to flatten the market maker position faster if the risk-carrying capacity is limited. In the absence of natural liquidity on the other side, this will often be through a trade with another intraday liquidity provider. The net result is that the the amount of trading volume that is attributable to this segment of the market will increase.
Thus: lim investing time horizons approaches 0 as HFT -> infinity
While the time horizon over which liquidity is provided can range from microseconds to months, it is particularly shorter-term liquidity provisioning that has become more common. This is partially a reflection of the changing nature of the default liquidity provider – ‘High-Frequency Trader’ is a commonly applied term.
Unfortunately, the “High Frequency Trader” is NOT, as explained, a liquidity provider in the conventional sense: it is an ultra-short time horizon churn facilitator and liquidity extractor (when the meager rebate for providing liquidity does not offset the capital holdings risk) and nothing else. Which is why just like the Fed has become the artificial lender of last resort in a regime that is unsustainable and where central banks are forced to grow their assets exponentially (as shown on Zero Hedge) just to preserve the flow so very needed to keep equities from collapsing, so HFT has become the artificial provider of fake liquidity.
The problem is that just like the half lives of central bank intervention, so the incremental benefits of ever greater HFT penetration are becoming less and less.
What happens next? Here Morgan Stanley, while trying to be diplomatically correct, comes to precisely our conclusion – trade while you can.
In our view, many of the changes in the market environment – such as the decline in trading volume – are secular. The trade from household direct share ownership to institutionally managed ownership has happened, removing one of the natural sources of liquidity. At the same time, the micro-efficiency of the market in identifying and exploiting liquidity demand, exemplified by the growth in intraday liquidity provisioning strategies, is here to stay.
What are the implications for institutional execution strategies? The first implication is a re-evaluation of parent order sizing. Liquidity for institutional trades is now ultimately sourced from other institutions for the most part, rather than from households. The share of trading volume from these institutions has been falling by almost 40% over the last five years. This means that the amount of liquidity we can reasonably expect to source in the market should also fall by a similar amount. For example, we find that the upper limit of the percentage of ADV that can be traded in a VWAP-type strategy without undue price impact is typically around 4-5% now, versus 10-15% in the period before 2007.
The second implication is that execution strategies have to focus on maximizing the likelihood of being a liquidity provider. This has always been the objective of portfolio managers (‘Buy Low, Sell High’). Within the institutional asset management process, that has not always been as central to the execution strategy. The assumption has been that liquidity will be available in the markets, and that the cost of demanding that liquidity (the market impact cost) was small relative to the alpha potential over time. As the composition of trading volume changes, this assumption has become more problematic. Having urgency constraints (e.g. having to finish a trade at a particular time) becomes increasingly costly relative to the alpha potential.
Where Morgan Stanley stops short is the logical next question: what will detour this transition to a market driven by quantized incremental binary decision-making, aka RISK ON, RISK OFF, where with every passing day, we get greater and greater volatility shifts? The answer: nothing, unless of course, for some reason retail investors do come back, however with Lehman, the Flash Crash, MF Global, central planning, forced media propaganda telling everyone “it is a once in a lifetime opportunity to buy”, even as markets in real terms are still down nearly 40% from 2000, retail has had enough of the rigged stock market casino.
Simply said: they are done.
Hence HFT will have no choice but to become a greater and greater role in equity trading, pardon, churning. Until one day, by logical extrapolation, only HFT is the marginal setter of prices, with no regard for value, logic or analysis, and a price-determining function set purely by historical precedent yet a precedent which will be no longer applicable in the least as the paradigm shift to a conceptually different “market” will have then happened. Or said otherwise: “large swings on low volume, massive bid-offer spreads, huge trading costs, inability to clear and numerous failed trades“… just as we predicted back in April 2009.
Just as simply said: with its advent, HFT sowed the seeds of its own self-cannibalization.
Which also goes back to another key concept, and arguably the biggest flaw in all of modern economics: it is never about the stock. It is all, and always has been, about the flow. Last week Goldman tacitly acknowledged it for the first time. Expect more and more economic hacks to follow suit.
The irony that ties it all together, is that if indeed for some reason retail investors do come back, and do pile their over $8.1 trillion in fungible money currently stored under the electronic cushion as we described in This Is Where “The Money” Really Is – Be Careful What You Wish For, which in turn would also unleash the titanic wall of money hidden behind the Shadow Banking dam wall (at last count about $35 trillion contained among the custodial holders of all securities why are quietly swept into the shadow banking system’s re-re-rehypothecated pseudo asset pyramid and regulated by exactly nobody), which no conventional economic theories account for, yet which as Ben Bernanke this week, and Zero Hedge for the past 2 years, has been warning is the real catalyst of the (hyper) inflationary spark, then the Fed will be powerless to stop the biggest avalanche of empty artificially created fiat currency ones and zeros to ever hit the monetary system in the history of the world since Weimar. Only this time it will have the added benefit of HFT to accentuate every move imaginable as cash transitions from an inert form to an active, asset managed one.
But this is far beyond what one learns in Econ 101, which is why we will have to wait at least another 3 years before the Morgan Stanleys and all other bandwagon chasers of the world close the loop on what we are (and have been for a while) warning right now.
In the meantime, we are confident readers will enjoy the supreme irony: in their attempt to perpetuate the insolvent status quo farce, the central planners are now forced to choose between the terminal Scylla and Charbydis: a pyrrhic Schrödinger [alive|dead] market, or an even more pyrrhic Schrödinger [alive|dead] monetary regime.
We hope they choose wisely.