Taming Uncertainty in Financial Forecasting
Michael Cox, Ciaran Browne, Lisa Brown
Retail Credit Risk, Ulster Bank, Belfast, United Kingdom

Robust and reliable predictions of future losses are critical to the sustainability of all Financial Institutions. Uncertainty associated with transitions in economic cycles has a major bearing on the volatility of forecasts within the Financial Services sector and presents a formidable challenge to all those involved in managing credit risk.

Many of the established data-driven approaches to forecasting are reliant on the assumption that historic trends are representative of current or future events. Herein lies the problem. The macro environment combined with internal changes to policies and procedures exposes the weakness of established statistical methodologies and an alternative solution is required. Furthermore, per BIPRU 4.3.51, the UK Financial Services Authority requires Financial Institutions to find and limit errors associated with model weaknesses and to apply human judgement in order to take into account all relevant information not considered by the model.

Human judgement, in the form of expert panels, has long been deployed in the provision of guidance and formulation of conclusions on complex matters within a variety of industries. Within the Financial Services Industry, expert panels are deployed to complement algorithmic methodologies in assessing the impact of internal and external factors on future profitability.

This paper investigates established statistical approaches to financial forecasting and details their limitations in addressing the problem of uncertainty. This leads to an introduction of how expert panels can be used to control the forecasting process and ultimately lead to more robust and appropriate outcomes.

Keywords: Uncertainty; Credit risk; Expert panels; Forecasting

Biography: Ciaran Browne has accumulated 13+ years in developing statistical approaches to resolving forecasting and modelling problems within Financial Services. As Manager of Retail Credit Risk Modelling within Ulster Bank, Ciaran's main challenge is to ensure capital model predictions are adequate in an environment of increased external review and ever-changing customer behaviour.