A Theory of Model Sophistication and Operational Risk

Abstract

We study the decision making of a financial institution in the presence of a novel implementation friction that gives rise to operational risk. Operational risk naturally arises whenever the institution faces a trade-off between adopting a more sophisticated investment model and one that is less complex to implement. In practice, a more sophisticated model generates a more informative signal about an investment opportunity by relying on the latest IT infrastructure and advanced data analytics, but the use of these technologies makes it more prone to operational errors. In this case, it is no longer optimal to adopt the most sophisticated model available, and endogenous deviations from it are affected in opposite ways by the various risks the institution faces. While operational risk and market risk induce adopting a less sophisticated model, model risk induces adopting a more sophisticated one. The negative relation between operational risk and model sophistication implies that the institution’s optimal operational exposure may well become decreasing in the level of operational risk, and that its optimal market exposure becomes less volatile.

Keywords: Operational risk, model sophistication, risk exposure, risk management, market risk, model risk, Big Data, FinTech

JEL Classification: G11, G20, D90, C61