|
Abstract:
|
Corporate bankruptcy prediction is of paramount interest in risk management. The response variable of interest, default vs. non-default is binary in nature. In this paper, we examine important predictor variables under a generalized single-index model framework. The model adopted is flexible and naturally provides interpretable index-coefficients. The univariate unknown function is estimated by polynomial splines. Least absolute shrinkage and selection operator (LASSO) penalty and SCAD penalties are used. We demonstrate through a comprehensive corporate bankruptcy database that contains both market and accounting variables, commonly used in the literature. The findings with generalized single-index approach shed new light in the bankruptcy prediction literature.
|
