Title: Optimizing credit gaps for predicting financial crises: Modelling choices and tradeoffs
Authors: Mohammad Jahan-Parvar - Federal Reserve Baord of Governors (United States) [presenting]
Daniel Beltran - Federal Reserve Board (United States)
Fiona Paine - MIT Sloan (United States)
Abstract: The purpose is to bridge the academic and policy debates on the role of credit gaps for predicting financial crises, by integrating the modelling choices associated with trend-cycle decomposition methods into the design of crises early warning indicators (EWIs). We evaluate how the performance of EWIs is influenced by the choice of trend-cycle decomposition methods for constructing credit gaps (including the smoothness of the underlying trend), and by the policymaker's preference over false positives and false negatives. For the most common trend-cycle decomposition methods used to recover credit gaps, we find that optimally smoothing the trend improves the tradeoff between false positives and false negatives of the resulting EWIs, and thus their out-of-sample performance. The out-of-sample performance improves further once we consider a preference for robustness of the credit gap estimates to the arrival of new information.