Information Percolation in Dark Markets

Using ideas from stochastic process theory (percolation), Darrel Duffie, Gustavo Manso and collaborators have recently proposed that under certain conditions dark markets fully aggregate information as do centralized exchanges (for the most recent work, click here ). Dark markets are settings where order submission and trading is bilateral and private, so that other participants would not know whether negotiation took place, if it ended in a trade, and at what price. We are running laboratory tests of this theory. The experiments provide qualified evidence in support of information percolation. Prices aggregate all available information but not in the strict sense of the theory – that eventually everyone trades at prices equal to expected payoff conditional on the aggregate information. Prices instead fluctuate within a narrow no-arbitrage band centered around the aggregate (average) private signal. Notably, participants were not aware that they all traded close to the fully revealing price.

Our results should inform the debate around recent legislation (Dodd-Frank Act in the U.S. and MiFID II Reform in Europe) that aims at providing transparency (and centralized clearing) to dark markets, or even disallowing them altogether.