I'm personally confused that we could even ask this question.

How could you not have multiple exchanges? They don't sprout out of magic beans; they're markets. Inherently, they're groups of people deciding to do business with each other. Almost the only thing that makes them substantively different from a farmer's market is standardized contracts.

In the stock market of the 1950s, if you were a naive corporate buyer or seller, something like 10 different kinds of market participants could (and would) fleece you on every transaction. You probably couldn't even get a quote without getting hit by some kind of scam.

How much better would the markets be today if we had taken the 1950s system and granted it a monopoly?

Actually, it IS just one big distributed exchange since Reg NMS requires each ATS to route an order to a rival with a better price.

This is good, because new ATS's dont need to worry about network effects and can compete on features.

On the other hand, this required functionality could limit the set of possible new features. For example, it may be difficult to implement some way to discretize or slow down markets to remove any unnecessary advantages conferred on HFT traders, because of this mechanism.

NASDAQ got caught in a price-fixing scandal in the mid 1990's. Part of the fall-out from that was Reg ATS, an SEC regulation that permitted "alternative trading systems" who eventually became ARCA (bought by NYSE Euronext), INET (bought by NASDAQ OMX), BATS, and Direct Edge. These four venues have spawned an enormous amount of innovation in the market space that NYSE and NASDAQ simply wouldn't have done without competition. Electronic order matching, market-maker rebate schedules, and equal opportunity for displaying liquidity all came from the ATS's.

Without the market fragmentation, we'd still have "specialists" controlling stock prices physically from the floor of the stock exchange.

Thanks - alternative trading systems can surely only come into existence if they offer a clear advantage, and of course this drives the evolution of the markets. That's great.

My question: how much do network effects matter. Internet guys always assume network effects are a primary force in marketplaces. I'm an Internet guy, so I can't help but have this bias.

It's fascinating to me that there are so many ATSs. What's the endpoint? Will we eventually see hundreds of ATS's, each with some benefit valued by some arcane subset of traders? Or can an individual ATSs incorporate the range of improvements, and through network effects crowd out the others?

The four ATS's I mentioned all became full exchanges. ARCA and INET got bought by existing exchange operators, while BATS and Direct Edge were cleared by the SEC to be exchange operators. These four all exist after a massive wave of consolidation among the ATS's from the past decade.

There are also dark pools, whose innovation is that they don't display quotes. Every large brokerage has a dark pool. I wonder if there will be consolidation there; it's anyone's guess at this point.

As for network effects, Reg NMS requires an exchange to route an order to a rival if the rival has published a better price. So network effects are less important than you might think outside the dark pools. The features that really attract market makers are things like fees and likelihood of getting a fill (ie, priority rules).

Thanks! I didn't realize that exchanges need to route orders to any other exchange with a better published price. That's a great rule. I assume that most ATS's are located within a short distance of each other.

Really helpful post. Thanks for the information.

> "I assume that most ATS's are located within a short distance of each other."

... if only. Reg NMS is a massive headache. This sort of price convergence would be better left to the arbitrageurs. Let the market deal with the light speed issues instead of trying to regulate physics.

Fascinating. Even more interesting. Thanks for the information.

> What's the benefit of multiple exchanges?

Free open markets, anyone? Multiple exchanges ensure that the markets are not controlled by any one exchange, which can use its monopolistic advantage to increase costs by levying high fees etc.

Furthermore, if one exchange does something shady like front-running, or preferential information sharing with its "partners", then people can just move to another exchange. Basically, multiple exchanges make the markets more honest.

One benefit of multiple exchanges is that company owners can choose to offer trading of shares of the company they own on whichever exchange they think would do their company best.

Some people in this thread have expressed that they think a discrete-time market (where trades are executed at certain time ticks, instead of continually) would be beneficial. If these people were to IPO their company, they could choose exchanges that use this approach.