The Washington Post's WonkBook column reports that the market's vertiginous 1,000-point cliff-jump was caused by the failure of traders to take steps to stabilize the market despite obligations to do so:
Securities and Exchange Commission Chairman Mary L. Schapiro said several widely discussed causes of the violent swings in the market were unlikely to have played an important role. These include a "fat-finger" erroneous trade that started it all; hacking or terrorist activity; unusual trading in shares of Proctor & Gamble, a component of the Dow Jones industrial average; and futures trades linked to the Standard & Poor's 500-stock index.
In prepared testimony for a congressional panel probing the market turmoil, Schapiro said that firms known as "liquidity providers" stopped buying many of the stocks that were suffering the largest declines last Thursday.
Some of these firms -- which are part of major banks -- have an obligation under securities rules to act only in ways that stabilize the market. But other firms, which focus on make lightning-fast trades based on computer algorithms, do not have such obligations.
Some firms that had previously been active participants in the markets withdrew their liquidity after prices declined rapidly. Schapiro said these firms may have followed the rules in doing so.
But she added that regulators must reexamine whether liquidity providers should able to retreat from the markets at times of severe volatility.
There is more detail on this in Securities and Exchange Commission (SEC) Chairman Mary Schapiro's May 11 Testimony before a House Committee. She noted the rise of "high-frequency trading" (HFT) firms:
The exchanges and other trading venues have adopted highly automated trading systems that can offer extremely high-speed, or “low-latency,” order responses and executions. The average response times at some exchanges, for example, have been reduced to less than 1 millisecond. Many exchanges also offer individual data feeds that deliver information concerning their orders and trades directly to customers. To further increase speed in transmitting market data and order messages, many exchanges also offer co-location services that enable exchange customers to place their servers in close proximity to the exchange’s matching engine.
Highly automated trading systems have helped enable a business model for a new type of professional liquidity provider that is distinct from the more traditional exchange specialist and over-the-counter (“OTC”) market maker. In particular, proprietary traders now use high speed systems by submitting large numbers of orders that can result in more than 1 million trades per day by a single firm. These proprietary traders often are labeled as engaging in high-frequency trading (“HFT”), though the term does not have a settled definition and may encompass a variety of strategies in addition to passive market making.
HFT traders can be organized in a variety of ways, including as a proprietary trading firm (which may or may not be a registered broker-dealer and member of FINRA), as the proprietary trading desk of a multi-service broker-dealer, or as a hedge fund (all of which are referred to hereinafter collectively as a “proprietary firm”). Other characteristics often attributed to proprietary firms engaged in HFT are: (1) the use of extraordinarily high-speed and sophisticated computer programs for generating, routing, and executing orders; (2) use of co-location services and individual data feeds offered by exchanges and others to minimize network and other types of latencies; (3) very short time-frames for establishing and liquidating positions; (4) the submission of numerous orders that are cancelled shortly after submission; and (5) ending the trading day in as close to a flat position as possible (that is, not carrying significant, unhedged positions over-night). Given the competitive pressures to maximize their speed of trading, HFT firms typically will attempt to streamline the code for their trading algorithms. However, every check and filter in that code reduces its speed, creating a tension.
HFT is one of the most significant market structure developments in recent years. Estimates of HFT volume in the equity markets vary widely, though they often are 50 percent of total volume or higher. By any measure, HFT is a dominant component of the current market structure and is likely to affect nearly all aspects of its performance.
Schapiro noted that New York Stock Exchange (NYSE) "circuit-breakers" that kick in at 10, 20 and 30 percent Dow Jones Industrial average (DJIA) market declines were not activated; the NYSE accounts less than half US trading volume. She added:
It is important to recognize that severe market disruptions in the form of precipitous price declines are not exclusively associated with automated trading. Disruptions are caused by a glut of sellers willing to trade at any price, combined with the near or total absence of buyers at a particular instant in time (who may themselves be influenced by the wave of sell orders crashing on the market). In these circumstances, prices can decline precipitously, as they did in many stocks on May 6.
Severe market disruptions have occurred throughout financial market history in a wide variety of market structures. For example, the U.S. equities markets declined by 22.6 percent on October 19, 1987 in a market structure that was dominated by a single manual trading venue. More recently, of course, and particularly since the implementation of Regulation NMS, the U.S.-listed equity markets have become much more automated and much faster. Nevertheless, they generally were able to continue operating smoothly even through the global financial crisis that reached a peak during the autumn of 2008. Accordingly, the inability of the equity markets to maintain fair and orderly trading in many stocks on May 6 is profoundly disappointing and troubling.
Deeper Historical Factors Underlying the Market's Plunge. Hardly mentioned in the general circulation public press are historical factors, rooted in economic incentives and cultural change, that contributed to the mini-crash. Starting in the 1960s institutional trading volume began to vastly increase.
Historical Factors: Economic Incentives. At the time, markets were maintained primarily by specialist holding franchises to trade specified stocks; floor traders added liquidity, without assigned stock portfolios. As institutional volume mounted, they and the block trading firms (like Goldman Sachs & Bear Stearns) complained that specialists lacked sufficient capital to maintain an orderly market.
Schapiro's testimony noted this change:
The events of May 6 implicate a number of issues raised in the Market Structure Concept Release. For example, it asked whether the current market structure appropriately minimizes the short-term volatility that can be so harmful to long-term investors. It asked whether the relatively good performance of the market structure in 2008 indicated that systemic risk was appropriately minimized in the current market structure and, if not, what further steps the Commission should take to address systemic risk. Finally, it noted the dominant role of HFT firms in today’s market structure and observed that they had largely replaced the role of specialists and market makers with affirmative and negative obligations for market quality. More specifically, the Market Structure Concept Release asked whether there is any evidence that proprietary firms increase or reduce the amount of liquidity provided to the market during times of stress. It also discussed various types of short-term trading strategies, including “directional” strategies, such as “momentum ignition,” that could present serious problems in today’s market structure by exacerbating short-term volatility.
Thus, the market was becoming more a trader's than an investor's market. Traders seek to profit by high volume, low-cost trading ISO small profits; investors seek to profit over the longer term by investing in securities with low turnover. The more trader volume increased, relative to investor volume, the harder it was for specialists & floor traders to stabilize the market. Block trading firms had more capital, and provided some measure of stability, but with No legal requirement to do so. Even on the investment side, share prices were skewed by institutional investment preference. Thus the "Favorite Fifty" stocks made stars of high-risk technology firms like Control Data, Burroughs, and Sperry-Rand. Lots of solid, less glamorous (and less risky) companies saw their shares neglected.
In 1975 the longstanding system of fixed commissions was abolished, and negotiated commissions became the order of the day. Large firms who traded in huge volumes got steep discounts fairly quickly; the retail investors lost favor at larger firms. Unless you had a few million to invest, you got passed over. Put simply, big firms rarely served smaller investors. Commission income was the primary incentive to do so. Absent good commission income, firms have little incentive to serve the needs of retail investors.
Historical Factors: Cultural Change. When I worked in Wall Street (1969 - 1973 at Goldman Sachs & 1973-1974 at Drexel Burnham Lambert) it was a very different place than it is today. The GS I worked at would not have tolerated the behavior of GS today; the Drexel I worked at was destroyed by buccaneers who ran the firm into the ground after its founder retired, getting into serious legal troubles. Cultural change killed Drexel & damaged Goldman. That change was one from a service culture to a predator culture.
Drexel was destroyed by misuse of junk bonds. These were a financing vehicle long extant to enable less favored companies to raise capital, albeit at higher interest rates than more creditworthy firms, instead of being shut out of capital markets entirely. Michael Milken and the late Frederick Joseph used them not simply to give new companies access to market capital. They also used them to bust up established firms and sell the pieces for trading profit; some managements were stodgy and deserved to be broken up, but others did not. To those who attended Drexel's annual "Predators' Ball" these differences did not matter.
Goldman took longer to fully embrace a predatory culture. When it did, the basis of customer relations fundamentally changed, much for the worse. In a service culture, customers are assets, and merit treatment that encourages them to keep their business with the firm. In a predator culture, customers are targets, milked for whatever cash they can be separated from, and if they leave the firm seeks the next mark. Abuses in a service culture, such as "churning" accounts did happen, but the firms who did this knew it was wrong, and rules existed that could punish egregious violators. In a service culture, firms entered into what lawyers call a quasi-fiduciary ("as if faithful") relationship: customers were entitled to fair treatment--at minimum, protection from predatory conduct by the firm.
In a predator culture, a firm can knowingly market defective securities to clients, under pure caveat emptor, asserting that sophisticated investors are supposed to know how to evaluate instruments. (Goldman is alleged to have done this; the firm denies it.) This is akin to the infamous "bear raids" of the 1930s (the way, on top of bootlegging, Joseph Kennedy made his fortune): Speculators sold securities short--selling stock you do not own, and buying it back later--while spreading rumors of bad news about a company, then bought the stock back lower to cover the short, upon which the bad news was revealed as phony. (A short seller borrows shares to deliver to the buyer, then after buying the shares back, delivers them to the lender of the shares originally sold.) Such predatory behavior gave rise to the securities laws & regulations under FDR. The "uptick rule" for short sales was designed to prevent this; in many markets today there is no uptick rule, and thus predatory selling prevails.
Even traders, a generation ago, obeyed strict codes of behavior--an unwritten code of honor that went beyond the rules written by the authorities. Thus, at Goldman's trading desk, the other firms were put into three categories: (1) firms who did not trade; (2) firms who traded and played by the rules--unwritten as well as written; (3) firms that traded and ignored unwritten rules.
A firm in the first category was not a competitor, and thus all its business was customer business. Even in a market sell-off, trading firms of the better sort would buy more than the minimum amount required (100 shares in those days) even knowing they would lose money, in anticipation of getting more customer business upon which they could make more money later.
Firms in the second category were competitors, and were entitled nonetheless to fair treatment. In a sell-off, the good-firm trader would buy the minimum 100 shares, sometimes even 200 of, say, a 1,000-share order from a competitor who was selling. If the competitor said that he was executing a customer order rather than trading for his own account, the good-firm trader would buy additional shares, perhaps 500. And it was never acceptable at good firms to run around behind another firm's sales and try to unload your own trading position. (Running around a customer firm's order was a hanging offense; you would never see that firm's business again.) You expected, of course, the same courtesy in return.
Then there was the third
type of firm, which ignored unwritten rules (and was more likely to
break written ones as well). When it sold to the better firms, those
firms
bought 100 shares only no matter what kind of order it was. You only
showed such sellers your customer
business as a last resort, after trying the better class of firms,
because those sellers would run around behind your order and cost your
customer money by forcing the market down as they sold. Goldman, then,
was one of the better firms. It obeyed the unwritten rules.
Today, increasingly, firms like Goldman that once were leaders seem to behave as if there are no rules whatsoever. The result is a predatory, insider's trading market. It's not that rules were never broken in the old days, but then punishment was a real risk, and rule-breakers knew they were breaking the rules.
Arbitrage: Market Stabilizer Becomes Market Menace. Arbitrage is the buying and selling of securities or their equivalents. Arbitrage traders--arbitrageurs--seek to make profits by serving as market intermediaries. When, a generation ago, the world was still separated geographically, with telephone & Telex links only, trades rarely happened in real-time; rather they were executed hours later, often overnight if between foreign markets. Arbitrage served the function of creating reasonably uniform prices, so that the customer in New York, London, Tokyo, or even smaller markets like Philadelphia, was reasonably assured of getting a fair market price. By aiding market liquidity and making prices fairer, arbitrage served a socially valuable function.With the advent of automated trading markets became even more efficient, as price disparities between markets further narrowed. But in the 1980s so-called "program trading" at times triggered steep price changes, as automated programs inundated markets with more volume than those charged with stabilizing them could rapidly absorb. That was an early augury of far worse to come.
Automated electronic trading between computers--independent of contemporaneous human intervention--now is estimated to account for 61 percent of US market trading volume, up from only 30 percent just five years ago. According to an article in the January 2010 Issue of Technology Review, "Trading Shares in Milliseconds" (purchase required via TR home page), some trading algorithms (formulas adapted for specific tasks) execute 1,000 trades per second. At that speed, 60,000 trades can be executed in one minute; if each trade is at a share price of $20, that would equal $1.2 billion dollars of volume in a single minute--with no contemporaneous human intervention. Defenders say that ten years ago a retail investor paid $150 to execute a 500-share trade, with a ten-cent spread between bid & ask quotes, whereas today the retail investor pays a penny with micro=spreads.
Of this problem Schapiro said:
For example, we must ask whether the general, market-wide circuit breaker provisions that currently are on the books (none of which were triggered on May 6) need to be revised. I note that a vitally important element of the market-wide circuit breakers is that they apply across all stock and options trading venues and all venues for trading equity security-related futures, because markets for all equity security-related products are closely linked.
I believe that we also must consider the various types of time out mechanisms that can be applied to individual stocks. Although the prices of many stocks on May 6 declined in proportion with the broader market decline that occurred in securities and futures index products, the prices of many other individual stocks declined much, much more (before snapping back largely to the prices at which they were trading prior to the precipitous decline). At this point, the root cause of the sudden disappearance of liquidity in many stocks is unclear. One potential explanation is the inherent nature of algorithmic liquidity providers. In today’s highly automated markets, proprietary trading firms provide much liquidity electronically through algorithms that are programmed to respond to events without human intervention. Such algorithms typically are developed by studying historical trading conditions and identifying patterns for profitable trading. Algorithms may be programmed to shut down trading when events no longer line up with the patterns that they are designed to exploit. Stated another way, algorithms may be very effective in adding liquidity in normal trading conditions, but may be inherently ineffective in adding liquidity when dealing with highly unusual events such as occurred on May 6.
Unlike pre-coded algorithms, people have the capacity, flexibility, and creativity to assess and respond to highly unusual events. Consequently, we must consider whether today’s highly automated markets need additional time out mechanisms to deal with unusual events that may lead to a sudden loss of algorithmic liquidity sufficient to satisfy the demand for liquidity. For example, we are considering whether all trading venues in the national market system should be subject to a requirement to stop trading for a brief period of time on a stock-by-stock basis when prices move beyond normal trading patterns. The time period should be sufficiently long for traders to assess trading conditions (or to assess the operation of algorithms).
She noted again the decline in obligations of traders:
For example, in the past, professional liquidity providers with the best and fastest access to the markets were charged with affirmative and negative obligations to promote market quality. One of the most significant negative obligations was a restriction on “reaching across the market” to take out quotations and thereby drive prices up or down. Many of the most active and sophisticated traders in today’s market structure are not subject to any obligations with respect to the nature of their trading. If active trading firms exploited their superior trading resources and significantly contributed to the severe price swings on May 6, we must consider whether regulatory action is needed to address the problem.
And there is more to worry about.
One key factor is that the financial services industry and the government did not convene any committee to look strategically at network failure risks after the 1998 collapse of Long-Term Capital Management, a hedge fund with less than 200 employees that held world finance hostage. Increasingly, markets are operating inside the human decision-making cycle. One trader argues that market orders--to buy or sell at whatever price prevails when submitted--may lose favor. One finance maven argues for banning market sell orders entirely.
These "hyper" factors explain the phenomenon: (1) as noted above, the hyper-speed of networked computers allows microsecond trading; (2) hyper-leveraged positions triggering cascade selling on minuscule declines in asset values; (3) hyper-complexity of financial instruments whose value is hyper-opaque to human (investor/trader) understanding, thus lengthening the time needed to make rational decisions on valuations & investment & market risk. Add to that hyper-inflated compensation schemes decoupled from fortunes of underlying firms, thus removing incentives to manage risk prudently.
Modern networks and finance are interlinked in potentially lethal ways because all modern networks are Faustian bargains: In return for unmatched economic efficiency in interconnecting large numbers of end points via a limited number of nodes, they offer heightened risk of catastrophic failure. They are: (a) accessible to countless users; (b) global in reach--connecting points across the globe; (c) programmable by software algorithms like those noted above; and (4) fragile--at times difficult to diagnose for errors and then to repair.
It seems that at certain critical crisis points, financial networks are at serious risk of operating like the proverbial Doomsday machines of nuclear deterrence; in response to certain enemy acts, the Doomsday device retaliates by destroying everything, even at risk of doing so by mistaken response to accidental launch. Fortunately, the nuclear strategists rejected such devices. The financial folks should rein in market devices with potential Doomsday automated potential.
Bottom Line. There are disturbing signs that increased reliance on automated computer trading, coupled with fewer traders bearing responsibility for market stability, are creating a growing risk of an avoidable market crash. This problem needs urgent attention. And the deeper problems of re-aligning economic incentives with productive economic behavior, and of restoring a service culture in lieu of predation, will take wise legislation and regulation, commodities seemingly in short supply these days.
Letter from the Capitol, LFTC, Economy, Conservative Politics
Comments