Seminar 19.August 2013

Subjective Beliefs and Statistical Forecasts of Financial Risks: The Chief Risk Officer Project

Abstract

We elicit subjective belief distributions from Chief Risk Officers of major global financial risks over the next year, using incentivized and incentive-compatible scoring rules. We compare these beliefs to the forecasts from a traditional statistical model. The extent of agreement between the subjective beliefs and statistical forecasts is formally characterized, allowing risk managers to assess for themselves what confidence to place in each. Furthermore, we characterize the extent of agreement between the individuals providing subjective beliefs, to allow an evaluation of the coherence of the beliefs that experts have about these risks. Our initial results, although based on small samples in terms of numbers, point to some striking differences between the subjective beliefs of experts and statistical forecasts. The experts have a pessimistic outlook for 2013 on the risk of European and Asian equities, as well as the cost of hedging credit risk in the United States. They do not believe it is plausible to expect the prices of oil or gold to reach historic highs. In other instances, such as interest rate risk and exchange rate risk, the subjective beliefs of our CRO sample are consistent with the forecasts of statistical models, adding strength to the inferences risk managers might draw from those forecasts.