Hayekian Rational Turbulence: 15-Oct-2014 US Treasury versus 24-Aug-2015 US Stocks
On October 15, 2014, after an early morning release of weak US retail sales data, the benchmark 10-year US Treasury yield experienced a 16-basis-point drop and then rebounded to return to its previous level between 9:33 and 9:45 a.m. ET. The major US regulators were sufficiently disturbed by this event to prepare a Joint Staff Report about this episode. I blogged about this report last month arguing that there was nothing irrational about what happened in that market on that day.
Now compare that with what happened to the S&P 500 stock market index on August 24 and 25, 2015 in response to bad news from China. On the 24th, the market experienced the following before ending the day down about 4%:
A fall of about 5% within minutes of the open
A rise of about 3.5% over the next hour or so
A fall of about 1.5% over the next hour or so
A rise of about 2.5% over the next hour or so
A fall of about 3.75% over the next three hours or so
The market was a little less erratic the next day, rising 2.5% before falling 4% and ending about 1.4% down.
I see similar phenomena at work in both episodes (15-Oct-2014 US Treasury and 24-Aug-2015 US Stocks): the market was trying to aggregate information from diverse participants in response to fundamental news which was hard to evaluate completely. In Hayek’s memorable phrase, prices arise from the “the interactions of people each of whom possesses only partial knowledge” (F. A. Hayek, “The Use of Knowledge in Society”, The American Economic Review, 35(4), 1945, p 530).
Sometimes, the news that comes to the market is such that it requires the “interactions of people” whose beliefs or knowledge are somewhat removed from the average, and these interactions can be achieved only when prices move at least temporarily to levels which induce them to enter the market. The presence of a large value buyer is revealed only when the price moves to that latent buyer’s reservation price. A temporary undershooting of prices which reveals the knowledge possessed by that buyer is thus an essential part of the process of price discovery in the market when fundamental uncertainty is quite high. To quote Hayek again, “the ‘data’ from which the economic calculus starts are never for the whole society ‘given’ to a single mind which could work out the implications, and can never be so given.” (p 519).
Hayek’s insights are timeless in some sense, but today seventy years later, I venture to think that if he were still alive, he would replace “people” by “people and their algorithms”. Algorithms can learn faster than people, and so sometimes when the algorithms are in charge, the overshooting of prices needs to last only a few minutes to serve their price discovery function. That is conceivably what happened in US Treasuries on October 15, 2014. Sometimes, when the evaluation and judgement required is beyond the capability of the algorithms, human learning takes over and overshooting often lasts for hours and days to allow aggregation of knowledge from people whose latency is relatively long.
Posted at 1:29 pm IST on Wed, 26 Aug 2015 permanent link
Comments