Paper by Daniel Beunza and David Stark
Abstract:
Financial models pose a cognitive paradox. As a powerful form of codified knowledge, models allow their users to interpret complex information in an uncertain world. But models can also blindside their users by locking them in the cognitive schema encoded in the models. Our ethnographic study of an equity derivatives trading room reveals that arbitrageurs overcome this paradox by introducing dissonance in their daily calculations. Arbitrageurs compare the outputs of their models with the estimates made by their rivals, themselves obtained by using models in reverse. This form of reflexive modeling distributes calculation across rival arbitrage funds, enhances search and increases returns, but it can also lead to financial disasters. Reflexive modeling differs from Granovetter’s embedded action in that it entails a calculative activity centered on formulae and numbers. It differs as well from Callon’s disentanglement in that it emphasizes how social relations make calculation possible.
Not to be missed, table 1: Arbitrage disasters, 1990-2003
This has been long discussed in the economics literature as “herd behavior” and as “informational cascades”.
Economist aren’t troubled by the fact that it entails a calculative activity ;)