How Derivatives and Risk Models Really Work: Sociological Pricing and the Role of Co-Ordination

Abstract

In this paper I discuss financial models created to price complex derivatives. I argue that their development can only be understood with reference to the purposes for which they have been created, and to the institutional environment in which they have evolved. I show via stylized computer simulations that the timing of losses from using the 'wrong' model can be very different, depending on whether the trader adopts the market consensus model (even if wrong), or not. This and other institutional factors encourage model uniformity, and penalize the trader or quant who tried to develop a 'better' model at odds with the market. I argue that this gives rise to a game of 'sociological pricing co-ordination', a game that can be sustained despite the apparently model-independent reference point given by the terminal payoff. Given the timing of the losses, these models can be successful for the quants and traders who develop them, but provide few 'cognitive spin-offs', and they can be seriously misunderstood by outsiders such as regulators.

Keywords: derivatives models, interest models, structured products

JEL Classification: A14, G00, G21, G24, O31