Some 33% of US stock trading activity takes place outside of exchanges. Four years ago, only 20% of trading did. What does that mean? Now, only two-thirds of trading actually sets the price of a stock that zips along the ticker, down from 80% in 2007. Where has all the trading gone?

Call it high frequency trading or automated market making or any number of other monikers, but trading isn’t happening on the exchanges. Instead, they might be traded within a dark pool, through a crossing network, or absorbed within the platform of a broker-dealer. The prices in these venues, however, don’t make it to the tape, and we are left with a question: is that a good thing?

The trend owes much the infiltration of technology in developed markets. Technology made markets electronic, reduced latency, and enabled algorithmic approaches to trading. Decimalization emerged from an SEC rule-change in 2001. Trading went from a business of sixteenths to a business of pennies or less. It narrowed spreads and made profits less comfortable for traders of the human variety. Service providers naturally responded with high-speed market data, and the markets got faster, with bigger and bigger volumes and smaller and smaller orders.

Inexorably, trading shifted from the center to the edge.

Driving much of the change – the “make or take” model. Island ECN introduced the idea of make-or-take pricing in 1998. Rather than charge both sides of the market a trading commission, Island chose to reward those who offer liquidity and charge those who take it. Those who offer liquidity do so through standing limit orders are paid a rebate when these are filled. Those who take liquidity pay an access fee to submit market orders. It fomented an groundswell of high frequency trading operations, driven increasing price competition, and left the exchanges with US equities trading businesses that now represent 10-15% of their overall revenues.

Now a question – is it a good thing?

Zerohedge put together a thoughtful piece today on the impact of maker-taker automated market making on price discovery, price improvement, and the hidden cost of failed execution.

Prior to the Flash Crash, a group of academics from CMU, Georgetown and USC published a paper in February of 2010 on Equity Trading in the 21st Century. Among the contributors was Lawrence Harris, former Chief Economist at the SEC. Together they argued that though the maker-taker model may appear to provide price improvement, it’s only an illusion. It looks like a good price because best execution standards have not kept up and do not incorporate the full complement of costs, such as the rebates paid to liquidity providers.

According to Justin Schack, vice president of market structure analysis at agency brokerage Rosenblatt Securities (Reuters, 25 Jan 2011), “There definitely has been an increase in the level of debate over internalization lately, and the exchanges seem to be lobbying more actively for limits that would bring internalized flow back on-exchange.”

Says Robert Colby, counsel at Davis Polk & Wardwell and former deputy director, Trading & Markets, SEC (1993 to 2009), “The markets could get to a point where there’s so much internalization of orders that the price formation mechanism is not accurate anymore.” (Reuters, 17 December 2010).

Is it a good thing?