There is an I-told-you-so tizzy going on about the revelation of the BATS exchange that a system (or programming) error had resulted in the execution of about 450,000 orders at prices worse than the best bid/offer (BBO). Oh, the humanity!
A little perspective here. According to BATS, the total loss (which is a wealth transfer, mind you) totaled . . . brace yourselves now . . . $420,000. Excuse me. $420,360.
There seems to be some presumption that there was a Golden Era of trading, before the invasion of the dreaded machines, when intermediaries had hearts of gold rather than a hunger for it. When bids and offers were never violated. When there were never trading errors.
Wrong. There was no Golden Age. Nirvana is still just a band.
Think that the BBO was never violated on exchange floors? Think again. Sometimes this was inadvertent in the chaos of the pits/trading posts during active markets. Sometimes it was very advertent (is that a word?) May I remind everyone of the FBI sting on the CME and CBT, which discovered that some brokers would collude with locals to execute customer orders at off-market prices, and split the proceeds, sometimes delivered by bagmen-literally, guys passing paper bags of bills. Given the relatively crude time stamping of trading cards, it was very difficult to construct an accurate audit trail. Trades couldn’t even be sequenced with precision, and since the bid/offer were not recorded continuously or time stamped, it was impossible to see whether a trade violated the BBO. There were pit monitors who tried to keep an eye on things, but no human monitoring was capable of detecting all violations. Indeed, the sociology of the floor and the member domination of exchanges (a subject I’ll turn to shortly) meant that social pressure discouraged cracking down too aggressively, particularly on the most powerful.
Enforcing the BBO on the floor was in many ways constrained by a lack of data. If anything, current market monitors have the exact opposite problem: they are drowning in a sea of data. But as BATS shows, it is possible to go back through years of this data and pick up mistakes that cost customers about $1 per error.
Then shall we discuss out-trades? Out-trades-trading errors, where trade terms submitted by the buyer and seller didn’t match-were common on the floor. Brokers were on the hook for errors, and there were stories of brokers writing six figure checks to make a customer whole for a loss. I would not be surprised that on an inflation adjusted basis, a single broker wrote a check to a customer that exceeded $420,000 in 2013 dollars.
But this reality of the way things were doesn’t stop the hue and cry about the fallen state of today’s computerized markets. The biggest huer-and-crier (emphasis on the crying) is Jackass Joe Saluzzi of Themis Trading. Here Joe outdoes himself, by suggesting that we need to go back to the days of non-profit exchanges in order to restore public confidence in broken markets. (h/t Blivy.)
JJ apparently believes the old mutual non-profit exchanges were freaking charities, like the March of Dimes or something. Uhm, no.
Let’s look at the facts. The old mutual exchanges were cartels of intermediaries. They restricted membership in order to enhance the rents of those members. For decades, most of them ran brokerage cartels that fixed commissions. Some created monopoly privileges for some members (e.g., NYSE specialists). Others (NASDAQ comes to mind) had order handling rules that basically precluded public customers from competing with member market makers in supplying liquidity: NASDAQ was also the nexus of a flagrant conspiracy among market makers to fix spreads at supercompetitive levels.
Non-profit status had nothing whatever to do with the charitable urges of old time brokers and market makers. As I showed in research done just as the transition from non-profit to for-profit status was occurring, exchanges chose the non-profit form because the non-distribution constraint inherent in that form prevented the exchange from choosing pricing policies that transferred wealth among heterogeneous members with very specialized human capital. (An abbreviated version of the argument is here. The full version was published in J. Law & Econ. in 2000.) Electronic trading undermined the rents and the specialized assets that drove the choice of non-profit form, so the move to electronic trading in turn impelled the transformation of exchanges to for-profit entities.
In other words, non-profit form was chosen by very greedy, profit-driven individuals to protect their profits. And a good chunk of those profits resulted from the exercise of market power and the adoption of collusive arrangements by exchanges.
So spare me nostalgia. Indeed, methinks that a good deal of the nostalgia-and the related criticism of modern electronic markets-is a shriek of rage by those who profited under the old system, and are furious that someone ruined their racket.
Joe does get one thing right (cf. blind hog, acorn). He attributes the specific BATS problem, and the increased complexity of the equity markets, to RegNMS. This is correct. The information-and-linkages approach chosen by the SEC led it to adopt regulations that socialized order flow. This was done with the explicit goal of encouraging competition among trading platforms.
Another example of “be careful what you ask for: you might get it.” Pre-RegNMS, NYSE executed about 85 percent of the trades in its listed stocks, and the bulk of the remainder was executed in Third Markets which did not contribute to price discovery. NYSE was essentially the natural monopoly supplier of price discovery. Now, NYSE market share is in the low-20s, and executions are shared pretty much equally among a handful of platforms.
But this socialized order flow model requires linkages between the execution venues. It is that necessity that accounts for the complexity of the current equity markets. The interconnection imperative is directly responsible for the proliferation of order types that many find so vexing, and which indeed give advantages to specialized electronic traders. Note that the BATS error was in an order type designed to ensure compliance with RegNMS rules relating to the BBO.
You can criticize this market structure. But if you do so, you have to grasp the nettle of the fundamental trade-off. The choice is binary. You can choose to socialize order flow, or not. If you do, you get something like the current equity market structure, with intense competition among execution venues linked by a complex web of fragile connections and a proliferation of order types. If you don’t, you get something like the pre-RegNMS market structure, or futures market structure past and present. A market that tips to a single execution venue that exercises market power, either by restricting access (the old mutual model) or by charging supercompetitive prices (the new for-profit exchange model).
Those are the choices. But the debate is almost never framed that way. Instead, bat sh*t crazy people like Saluzzi (enabled by folks like the CNBC talking head in the BATS Hit video) take up all the oxygen screaming about how bad the current market structure is and retailing myths about some Golden Age that never was.
Political economy considerations almost certainly ensure that the SEC is not going to go back on RegNMS, and it is truly invested in the information-and-linkages approach it chose. So we are likely to see a continued series of controversies over problems that are inherent in the socialized order flow model. The SEC will pull its chin, and deal with each problem as it appears in an ad hoc way, just adding to the complexity as the new games are devised to exploit the new rules implemented to stop the old games that exploited the old rules. This will make for never ending media appearances for the likes of Joe Saluzzi, but don’t buy his line about the good old days. The good old days that existed when regulation effectively opted for the other binary choice (no socialization of order flow) weren’t all that great. There was a different set of problems, and a different set of people and firms profiting off the inefficiencies that inhere in the network nature of financial trading.