John Cochrane has a new blog post summarizing recent research by Budish, Cramton and Shim on the effects of high frequency trading. The paper shows that as high frequency trading spreads, the correlation of a particular asset price across two U.S. markets (Chicago and New York) has become higher at intervals that are very short. Any price deviations across the two markets disappear in less than one second. As Cochrane puts it,
It is lovely to see the effect of "arbitrageurs" making markets "more efficient."
As an academic I enjoyed reading the post as it provides a very nice example of a clean empirical test of how high frequency trading makes the co-movements of two markets stronger. This is what we look for in academic papers, a clean test of a very simple theory that produces very credible and robust results.
But as much as I enjoyed reading the post, it also reminded me of how limited the ability of academic research is to help us understand phenomena that really matter. In this particular case, the analysis compares the price of the same security in two nearby markets (geographically but also linked by very fast communications). As communications and trade become faster and faster, price deviations between the two markets disappear in a shorter period of time. This is really nice to see but is this a big surprise? One would expect that at a minimum, very basic arbitrage opportunities do not exist in integrated financial markets. So it is reassuring that arbitrageurs help markets be more efficient but I am not sure I would go as far as saying that this is "lovely".
What would be more interesting (at least to me) is understand whether high frequency trading helps getting financial prices right. And by "right" I mean prices that are consistent with economic fundamentals, prices that do not generate volatile dynamics and bubble-type behavior. If we could prove that this is the case then I would find the result "lovely".