Friday, July 02, 2010

Market Microstructure and Capital Formation

In an earlier post I argued that recent changes in technology have altered the distribution of trading strategies in asset markets, with information extracting strategies becoming more prevalent at the expense of information augmenting strategies. Specifically, there has been a dramatic increase in the market share of strategies based on rapid responses to market data using algorithms and co-location facilities. One consequence is that the data itself becomes less reliable over time, resulting in greater price volatility and occasional severe disruptions. The flash crash of May 6 was a striking example. 
While my focus has been on market stability, this kind of transformation in microstructure probably has a number of other important effects. In recent testimony before the joint CFTC-SEC committee on emerging regulatory issues, David Weild has argued that one of these consequences is on the size distribution of publicly traded companies, and on capital formation more generally:
There has been a computer arms race unleashed on Wall Street by changes in regulation and technology... [This] is displacing fundamental investing with computer‐trading based strategies and has created new forms of systemic risk, a loss of investor confidence, and a disastrous decline in primary (IPO) capital formation and the number of publicly listed companies in the United States.

From 1997 to Year End 2009 there has been a 40% decline in the number of publicly listed (i.e., NYSE, AMEX and NASDAQ) companies in the United States. On a GDP weighted basis, we have seen a more than 55% decline in the number of publicly listed companies. Today’s market structure has lost the ability to support small capitalization companies and initial public offerings (IPOs) on the scale necessary to help drive the US economy. The U.S. now annually delists twice as many companies as it lists and this trend has been going on since the advent of electronic trading... the unemployment crisis in the United States has been partly caused by changes to debt and equity capital market structure and the events of May 6 may give us an opportunity to come to grips with the notion that we have entered into an era where trading interests are eclipsing fundamental investment and economic interests.

Fundamental investing, or so‐called “information increasing” activities, are being displaced by trading, or so‐called “information mining” activities. The growth in indexing and ETFs may be exacerbating this problem.

In addition, stock market structure today is geared for large‐capitalization stocks with typically symmetrical order books but disastrous for the vast majority of small‐capitalization stocks with asymmetrical order books (where there is not naturally an offsetting buy order to match against a sell order and vice versa)... The “Flash Crash” was an example of where even normally liquid securities went to a state of “asymmetry” and price discovery broke down...
[Until] all trades, quotes and other messages in all interrelated markets are tagged and traceable to the trading venue, broker and ultimate investor, and disclosed to the market, markets will not be perceived as fair... With full tagging, tracking and reporting and the application of posttrade analysis and test bed techniques such as Agent‐Based Models, regulators and market participants will... once and for all be in a position to judge the impact of other participants and to regulate and plan accordingly...
It may be time to admit that what works for large, naturally visible companies, is the antithesis of what is needed by small companies and it is these small companies that are essential to grow our markets, reduce unemployment, restore US competitiveness and drive the US economy.
I am not aware of any academic research that links market microstructure to the size distribution of publicly listed companies in the manner suggested here, and I am grateful to David for for bringing his testimony and supporting documents to my attention. The issue is clearly of considerable importance and deserving of greater scrutiny.


Update (7/2). In an email (posted with permission) David adds:
I did a presentation to the ISEEE (International Stock Exchange Executives Emeriti) at the end of April.  The audience consisted of about 25 mostly former senior stock exchange executives... I was taken aback by the reaction of people from places like the Zurich Stock Exchange, Australian, New Zealand, Bovespa and others who were of the opinion that these electronic market structures (specifically, compressed spread-trading centric electronic continuous auction markets) are hurting primary capital formation in many of their countries as well.

For me, having run strategy for investment banking, research, institutional sales and trading at a major Wall Street firm, it is pretty simple - If one can't make money supporting small cap stocks, one won't support small cap stocks...
This has had two effects:
  1. The investment banks tell issuers that they have to do a much larger ($75 million) IPO; minimum IPO sizes have increased at much faster than the rate of inflation.   
  2. Aftermarket support for IPOs has withered because issuers lose money providing it (unless the companies are much larger).
It is commonly argued that the rise of algorithmic trading has resulted in increased liquidity, although this claim is by no means universally accepted. David (if I understand him correctly) is arguing that even if liquidity has increased for some classes of securities, it has declined for others, with detrimental net effects on capital formation.


  1. This doesn't explain the rash of delistings. But I spoke with the CFO of a voluntarily delisting microcap RLEC, and his explanation was simple: compliance with Sarbanes Oxley was costing him $1m per year, or 10% of net earnings.

    Under such conditions, stable, mature, cash-positive small firms are those most able and willing to delist -- they don't need access to equity markets. When they leave, the overall quality of remaining listed micro caps falls.

    I spoke to an SEC attorney about this, and he told me that in his opinion the SEC had been captured on this issue by the major accounting firms, who see SarBox as a gravy train. This was near the end of the previous administration -- don't know if it has improved.

  2. Gemfinder, if you follow the link to Weild's statement and supporting documentation (especially page 9) you'll see that the number of IPOs raising less than $25 million fell very sharply from 1996-2000, well before Sarbanes-Oxley was enacted in 2002. I don't doubt your story but there more to this than accounting regulation.

  3. I accept that and did not intend to imply otherwise.

    My point was that the change in total listed stocks is equal to the number of new listings minus the number of delistings. Both of these are moving in the wrong direction.

    IPOs are declining, and the reasons you give are compelling.

    Meanwhile, delistings accelerated dramatically after 2002, for the regulatory reason.

    Not that regulation is unnecessary -- it's painfully obvious we have an endemic fraud problem. But adding complex reporting regimes that enrich the big accountancies (who are themselves implicated in the fraud problem) is perhaps not the highest ROI way to fix things. It also acts as a regressive tax on smaller businesses, though most fraud losses have been from large caps.

  4. Agreed - thanks for your comment.

  5. "Aftermarket support for IPOs has withered because issuers lose money providing it "

    lol - why? because spreads are too narrow? (sarcasm)...

    David's mistake is one of correlation vs causation... one of the oldest in the book. Perhaps the decline in listings has something to do with the end of the greatest leverage bubble our world has ever seen???

    see, this may be true: "Fundamental investing, or so‐called “information increasing” activities, are being displaced by trading, or so‐called “information mining” activities. The growth in indexing and ETFs may be exacerbating this problem. "

    but that conflicts with what he's saying about a lack of liquidity in smaller cap names. you can't have it both ways. HFT algos are not trading microcap stocks that trade 1200 shares a day. by DEFINITION

  6. Weild, same guy that advocates setting up an alternative market with only broker intermediated trading against market maker quotes with a dime or two spread. Phone only, no electronic execution. Basically the old school NASDAQ system. Seems like he's got an agenda.

    How about we start by killing dark pools instead and make all trades show on the consolidated tape. That's one real problem with price discovery at the moment - if the majority of trading is off market and invisible, the displayed price is much less meaningful.

  7. KD and Jim, thanks for your comments.

    KD, yes, David is arguing that decimalization and Regulation NMS had some unintended negative effects on capital formation because of the decline in spreads. Regarding the causality issue, we don't have a randomized controlled trial here to establish this, and he offers correlations along with some plausible mechanisms. Doesn't mean he's wrong, and perhaps an econometrician can look at the data and see if some causal inferences can be made using instrumental variables. I think this is worth exploring.

    Jim, you're absolutely right about David's proposals but he doesn't want to turn back the clock, he wants an alternative marketplace in which issuers could choose to list. He suspects that only small caps will choose this. Agenda or not, I think the effects he describes are worth thinking about.

  8. i feel the need to address Jim's comment on Dark Pools, because it illustrates a common misunderstanding. all trades are reported to the tape - including dark pool trades. all dark pools do is hide bids and offers, in an attempt to limit precisely the type of "information extracting" strategies Rajiv refers to in this post.

  9. KD, agreed. But dark pools may eventually give way to hidden orders on exchanges, which would have the advantage of consolidating liquidity:

    "Despite their benefits in reducing slippage, stand-alone dark pools have been controversial primarily because access to them is not universal. There is a fear that the owners of prominent dark pools will be capricious in deciding who can have access to their liquidity, and, indeed, we have seen some of this behavior in the marketplace. But this is not a long-term problem - with the recent introduction of hidden order types on the major exchanges and ECNs, stand-alone dark pools likely will shrink in volume. Most exchanges and ECNs now offer essentially the same "dark" product as dark pools, but with much greater liquidity and lower fees."

  10. yes Rajiv - I've said multiple times that I think the ideal market would be one big consolidated dark pool.

    dark pools are probably the most misunderstood and incorrectly maligned piece of the market structure.

    oh by the way, please note that "There is a fear that the owners of prominent dark pools will be capricious in deciding who can have access to their liquidity" sounds an AWFUL lot like David's alternative marketplace where issuers could choose to list...of course, in David's perfect fantasy land, they wouldn't let HFT algos trade, right? right...

  11. "I think the ideal market would be one big consolidated dark pool."

    I've been thinking along the same lines recently, especially after May 6. Algorithms can still use data on executions to make short-term price forecasts, or make experimental trades to test the depth of a market, but overall the balance should shift to longer holding periods which I believe would be a good thing.

    Could you give me a link to a post of yours on this?

  12. rajiv:

  13. Excellent post, thanks. I recommend it to any readers on this comment thread, and I'll follow up with a post of my own at some point.

  14. Hmmfph- if Sarbox encourages delisting, it also supplies a large inventory of companies to the private equity industry. So they're definitely in favor of it.

    Sarbox did nothing to remedy the Enron problem. The accounting scandal side of Enron happened because the consulting side drove auditing to be capture by the customer. There were a lot of people who switched jobs between Arthur Andersen and Enron.

    The only way to kill this problem is to destroy the structural incentives towards auditor capture. Ditto with rating agencies.