A0583
Title: Binary choice models with multiple integrated predictors
Authors: Hsein Kew - Monash University (Australia) [presenting]
Abstract: A binary probit model with multiple integrated predictors is considered. This model is useful for predicting a binary recession using interest rates on long and short-maturity debt and private debt interest rates with different degrees of default risk. A constrained non-linear least squares estimator is considered to estimate the model's unknown parameters. Monte Carlo study shows that this estimator produces estimates with better precision for a relatively small sample size than an unconstrained, non-linear least squares estimator. This model is applied to forecast U.S. recessions, and the forecast performance is compared with binary probit models with stationary predictors in terms of in-sample and out-of-sample predictive power.
