Friday, August 20, 2010

Stock exchanges: stop the robots

Regulators must stop the robots from ruining our stock exchanges, according to democrat Senator Ted Kaufman.

In a 5 August letter to SEC chairman Mary Shapiro, Kaufman warns that some algorithmic trading harms long term investors and dominates the equity markets in a way that brings no social utility. He describes some murky algo practices such as trading on advance information about other people’s orders, collecting unearned liquidity rebates and unnecessary intermediation, to support his case.

The Senator has picked up a useful ally in Gillian Tett, who used her FT column today to back his call for a review of high frequency trading. Tett explains his argument that the benefits of the narrow spreads which high frequency trading supposedly brings may be outweighed by hidden costs, leaving only a “thin crust” of liquidity that evaporates in bad times, as with the flash crash of 6 May.

In an separate but related story, the paper reports that two day-traders in Norway were indicted on charges of market manipulation after they stumbled across – and exploited - a trading pattern that tricked programme trading computers into raising their prices. Despite the breach of market rules, plenty of people cheered the duo for beating the machines: “Robots are designed to push the market but when someone pushes the robots it is suddenly a criminal offence,” was one comment on a Norwegian chat forum, according to the report.

Behind both these stories lies an unease about the growing dominance of robot trading. While not quite an android invasion, the increasing use of computers to arbitrage short-term price anomalies is a sinister development whose effects are not well understood. But it accounts for nearly 70% of US equity trading, making it an important source of revenue for stock exchanges.

As with all arbitrage, programe trading is useful up to a point. Price discovery, an appearance of liquidity and narrow spreads are seen as good for markets. But they shouldn't be confused with the equity market’s primary function of allocating capital to the most productive use. Common sense says that if the market were made up entirely of robots chasing price anomalies, it would not do its job very well.

Similarly, the experiences of the 6 May flash crash and before that the sub-prime sell-off, LTCM and Black Monday tell us that an appearance of liquidity from computerised trading is not the same as a truly deep market of buyers and sellers with different views.

The danger is that if robot trading ends up undermining investors’ confidence in the market, it will frustrate the market's core functions of providing a mechanism for collective investment in large undertakings (capital formation), and attracting long-term investors for that purpose. There is already evidence that equity investors are preferring bonds and commodities, with US equity mutual funds reporting net redemptions in July, for example.

Kaufman proposes a number of technical fixes, such as charging high frequency traders per message rather than per trade (some apparently send 100s or 1000s of buy and sell messages for every order filled, sometimes to "spoof" other algorithms), and a number of regulatory fixes including an emphasis on deeper, less fragmented markets. He doesn't mention the obvious solution of a transaction tax, but perhaps he's just being politic, since Geithner unfortunately rejected FTTs in the US last year.

Algo trading raises a similar concern as passive and active fund management, where some fear that the growth of low-cost index funds will smother the market's ability to pick winners and allocate capital effectively. Computer trading and passive funds direct huge quantities of dumb money without contributing much to asset allocation. But while the fund market can correct itself (active funds simply lower their prices and raise their game until they are competitive again), once robots have smothered the market like pondweed there's not a lot other investors can do to fight back.

We shouldn't forget that stock exchange executives need to make a profit too and keep their companies strong in a consolidating sector, hence their recent investments in HFT and dark pools. But to survive in the long-term, many years after Xavier Rolet steps down as CEO of the London Stock Exchange or Reto Francioni as head of Deutsche Bourse, exchanges need to deliver first and foremost for the equity market.

A quick look at Deutsche Boerse's website illustrates the problem. It lists its core values as integrity, transparency, precision, reliability, innovation and a focus on the needs of customers, investors and staff. These are fine values but it's strange that efficient capital formation is not among them.

It looks like we'll have to rely on regulators like Shapiro to save exchanges by making them perform their core economic function first, and make an honest profit - with or without robots - second. Like utilities, they are part of the infrastructure of finance. We readily accept that private water and energy companies serve us better when they compete within profit and capex limits set by public regulators. Why should stock exchanges be any different?

Update 23 August: This FT article reports that stocks are not responding as they ought to earnings news as macro-investing and programme trading have increased. Stocks in the S&P 500 are 60-70% correlated to each other today, up from 30% in the 1930s, making the ancient art of stock picking relatively pointless, at least until certainty returns to the market.

1 comment:

  1. Gillian Tett's article in FT's "Insight" today tells us that:

    "(Senator) Kaufman argues that it is not good enough for the SEC just to tackle the symptoms of the flash crash (say, by imposing circuit-breakers); he wants to address the root causes too, by making markets more “truly” liquid. And he offers a pretty sensible set of proposals for doing this: regulators should create incentives to move more trading on to regulated exchanges; they should monitor strategies of high-frequency traders; they should force HFT entities to pay a fee per message, not per trade; and HFT entities should also hold more capital."

    I welcome the new emphasis in the Senate on ROOT causes rather than on superficial symptoms capturable in a trivializing phrase. But I vehemently oppose creating incentives to move trading to regulated exchanges. There is a necessity for central registration of derivative contracts, of course. And we all acknowledge that the trading community en masse has been guilty of too much gambling and not enough investment in the economy as a whole. So introducing an incentive to trade centrally is only going to put more money into a community that has yet to rein in those of its members who indulge in rampantly dysfunctional gambling. Unfortunity, this community doesn't in general rein in its more conscienceless bounders; rather, it seems to adulate them.

    What, as members of a dynamic economy we all need -- in my strong opinion as an experienced executive coach with clients including a manager of trading for a Canadian bank -- is that ALL trades be registered and in this age of computers, there's no excuse not to have a central registry. Additionally what we need is that ALL derivative trades be filtered automatically in this central registry, so that the capital, energy, and potential for rationality on a human scale of derivative trading be harnessed for the good of the economy as a whole.

    Filtered and harnessed? How might those very desirable tasks be accomplished in practice?

    Well, for over a year now there has been a strong move in Europe and one now gathering momentum on the US West Coast to institute a tax on financial transactions (an FTT). Unfortunately that movement appears to be being driven by haste, impatience, or raw vengeful anger rather than by intelligent, fearless, and effective determination to harness speculative investment skill sets and resources for the good of the economy as a whole.

    The FTT idea would be a good one if it were to discriminate rationally between functional and dysfunctional trading in derivatives and synthetic derivatives. How that can be done is the subject of a paper entitled "Why a 'Calibrateable dsFTT' makes sense: Beyond the Toronto G20 and the Dodd-Frank Bill". "dsFTT" stands for differentiated speculative Financial Transaction Tax, and the reasons why a dsFTT makes sense are explained at the link, where practicable suggestions for how a dsFTT may be calibrated in a rational way (again from the point of view of the economy as a whole) are also made.

    Senator Kaufman was clearly recognizing that high-frequency computer-generated trading, which is, I believe, virtually always in short-term contracts, does require, for the good of all of us and especially of the capital community, to be regulated boldly and effectively. There's nothing bolder than a central registry and an FTT tax, and nothing more likely to assure a rational harnessing of the skills of speculative investment to help plough the fields of the whole economy, rather than only the self-erotic portion of it that speculation so often has been, than a calibrateable dsFTT. IMO, of course.

    Angus Cunningham