Econ 101 for Wall St: The Economics of HFT
There's been a ton of discussion about high-frequency trading and market microstructure lately. Here's a great summary in the NYT.
What's missing from the conversation is the basic economics at hand. Here's a summary.
The problem is two kinds of efficiency
- and the second outweighs the gains from the first.
If algos can discover higher values, that's fine. It unlocks allocative efficiency
: goods are allocated to their highest valued, or most productive, uses.
The problem is that the market for
algos is not efficient. It is a big, fat market failure happening in real time. Flash orders displayed to some but not others are an information asymmetry that subverts subvert competitive efficiency
. The rebates (read: bribes) that exchanges grant for "providing liquidity" are just a side payment that does exactly the same thing: they create a new source of scale economies for the deepest-pocketed traders, that are essentially switching costs to not defect to rival platforms.
What's the point of subverting competitive efficiency? To enable not price discovery, but price discrimination. Authentic price discovery is prevented by side payments, that ensure that prices float well above value. Flash orders ensure it.
What is the biggest problem with this arrangement? Prices will never
be forced down to marginal cost. In fact, the dynamics are these: algos compete to discover higher values, unlocking allocative efficiency. Yet, without competitive efficiency, the algos capture a larger and larger share of that allocative efficiency.
In the end, this market microstructure eats itself: the returns to people on either side of the trade vanish.
Sound familiar? It should
. Liquidity that needs rebates isn't real liquidity. Just like SUVs subsidized by cheap gas aren't real transportation. Just like pharma companies that rely on subverting markets don't provide real healthcare.
It's just business as usual in our awesomely lame Ponzi zombieconomy.
Want fries with that unemployment check?