Tuesday, May 11, 2010

The day the computers panicked PART 2

Should we fix it?

Up to the end of March, all the postings on this blog focused on regulatory reform; as early as January 28, one of the first postings suggested a reform that would help address this issue (see: “Efficient capital allocation doesn’t require perfect liquidity”) Nevertheless, this posting on regulatory reform is done with considerable hesitation.

When it comes to regulatory reform, nothing fits better than a quote from a Rolling Stones’ song: “You can’t always get what you want, but if you try real hard, sometimes you just might find, you get what you need.” Further, over the weekend you probably saw headline like this one from the WALL STREET JOURNAL: “Regulators Are Stumped by Drop.” “Stumped?” By what? It seems to me that regulators and politicians are going to try to obfuscate the obvious. When that happens, people get divided and emotional about anyone not going along with the party line. That’s one reason for the hesitation.

The second cause for hesitation involves disclosures. I’m not “talking my book” as the saying goes. After this posting, PART 3 will address how to make money from computer panics. It’s so easy it would be a shame to see this easy money go away. Further, mathematical models are intriguing especially when they involve financial markets. It would be equally sad to see them go away. Finally, people who develop trading systems that work can earn a rent on their effort. It’s only slightly different from people who develop any other type of systems. We let the market judge the value of other systems.

It isn’t hard to figure ways to reduce volatility. A fee per trade or circuit breakers will do the job. You might ask: why the question in the subtitle? The answer is “because any method of reducing volatility will have other consequences.”

Fees on trading have an impact that won’t be uniform across types of assets unless carefully designed. They have different implications depending on how one trades. Ultimately, they raise the cost of capital, but that could be reversed by linking them to lowering other taxes on capital.

Direct limits on vilatility will only force the volatility into other markets. That could be other markets for the same asset as occurred Thursday in selected stocks, or it could be into markets for other assets. One of the reasons for concern about limits as being discussed is a belief that many traders will shift volatility risk from equities to other asset classes where that risk can do more economic damage. Remember, equities are not the only asset class that gets volatile when computers panic, and quantitative trading cuts across asset classes. Further, limits, like fees, have to be carefully designed and cover multiple assets, not just one type of asset.

Limits also introduce another potentially dangerous risk. To illustrate how they could introduce more risk consider “trailing stop loss” orders. They are often a component of trading systems. Thursday many people learned the risk associated with those trading systems.

A “stop loss” is a limited protection. It doesn’t guarantee a price when an asset price “gaps.” That’s the same lesson learned in October 1987. Further, they don’t protect one from volatility. One can get “stopped” out of a position at a price that is lower than the price a few seconds later. In fact, the presence of a large number of “stop loss” orders increases the probability of “gapping” because once the “stop loss” thresholds have been crossed the orders all become market orders to sell.

Now, introduce limits. Limits would give people or computers a chance to go in and cancel the “stop loss” order if it had not been executed. That sounds positive. However, consider the other side of the trade. The trader who cancels the trade now has a totally un-hedged position. Granted the hedge was partial and may be failing to accomplish what the trader intended. But, as has been pointed out before in this blog and elsewhere, no hedge is ever perfect. So, by canceling the “stop loss” the trader is exposed to the risk that was hedged as well as the risk that, unbeknownst to the trader, never was hedged. That may or may not be good. But, it seems to this observer that canceling a large number of partial hedges at a time of particular volatility is a strange risk reduction strategy.

It would be naive to assume the trader whose “stop loss” has failed won’t pursue an alternative approach to limiting down side risk. When the limits kick in, would prices of protective puts shoot up to their limit? Would buying of “protective” assets spike? Would there be a flight to liquidity? Other traders would also react. Would short sellers smell blood in the water among this group of exposed longs? Would the more insightful traders find an alternative to “stop loss” orders ahead of time? If so, what would it be?

Would traders just move to a different asset class where there are no limits? All of those questions were brought up in 1987 after the October crash. Limits past after 1987 didn’t cause any disasters, but they also didn’t end computer panics. Also, historical experience seems to suggest limits didn’t reduce volatility and perhaps increased it, but, granted, other things being the same before and after 1987 would be a na├»ve assumption.

In any case, it seems very likely limits will be expanded to individual equities. That will shift volatility risk to other assets. Hopefully it won’t be Treasuries or currencies, but, quite frankly, what market needs to be more volatile?

There’s an interesting and informative sidebar issue resulting from Thursday. Let’s look at the issues surrounding the NYSE’s shift to slow trading on selected NYSE-listed stocks, Procter & Gamble in particular. It’s a perfect example of the shifting of volatility to markets that couldn’t accommodate it. However, that point is being overshadowed by differences in philosophy. As discussed below, it also looks like the markets that couldn’t accommodate the volatility want to change the issue to conceal their failure to provide an orderly market.

The issue surfaced in an exchange between Duncan Niederauer, CEO NYSE Euronext, and Bob Greifeld, CEO Nasdaq OMX Group Inc. Greifeld’s contention is that the overall volatility in the stock was increased by the Nasdaq’s inability to provide enough liquidity to accommodate an orderly handling of the volatility. He doesn’t say it that way since blaming it on the NYSE is so much more consistent with his interests. But, that’s the bottom line of his position. That seems reasonable. When all the volatility risk is shifted to one market, that market will be stressed.

Niederauer’s counter that the purpose of the NYSE is to provide an orderly market. Can’t argue with that. But, under the circumstances he seems to be overly professional in not pointing out that Nasdaq didn’t deliver an orderly market.

Here’s where the verbal exchange gets interesting, and it betrays each man’s philosophy and the market they serve. Another function of an exchange is to provide liquidity. Clearly, when the NYSE moved to slow mode it traded off providing instant liquidity for orderly market. Who is served by each? People don’t even sense instant liquidity. The slower mode and even a pause for a few minutes would hardly be noticed. By contrast, computers assume instantaneous, continuous liquidity. Put bluntly, Nasdaq would accommodate computers at the expense of people while NYSE leaned the other direction.

Next step in the analysis involves the exchange’s role as a method of price discovery. Clearly, that is a key function of an exchange. Again, Nasdaq was willing to “execute” any trade without regard to whether the price discovery was being compromised in order to accommodate continuous trading. NYSE wasn’t. If one needs proof, it will come when trades are unwound as is being discussed.

Finally, an exchange acts as a clearing house becoming the counterparty. NYSE slowed trading to ensure it could fill that counterparty role. We will see whether Nasdaq honors all trades. This last issue goes to the heart of the issue of whether an exchange walked away from the market. Slowing trading isn’t walking away from the market; it’s slowing it. By contrast executing erroneous trades and then not honoring them is walking away from at least two important responsibilities of an exchange.

The difference in philosophy revealed by the verbal exchange raises an interesting question. Do we need two different types of exchanges? One designed for computer trading with computer and one that accommodates people. People and computers could move between them, but the design of the assets would be different. It is less pie-in-the-sky than it sounds. It could be done by creating two different classes of stock. One would target those needing instant, continuous liquidity. The other would compromise instantaneous, continuous liquidity, but fulfill the ownership aspirations of some equity investors. If you think it is far fetched, remember there are ETFs that represent the same assets as traditional mutual funds although in that example the assets don’t differ much if at all. But, rather than explore that solution, let’s just hope limits aren’t viewed as a substitute for some common sense applied to the issue of fees.

No comments:

Post a Comment