Abstract
The purpose of this paper is to test for proportionality and linearity of the bond ratings data using adjacent categories logit model. In the analysis of bond rating, the ordered logit models and/or ordered probit models are being widely used for credit analysis. Although these models require the strong assumption of the proportionality and the linearity, these assumptions have not been verified in the previous research of this field. In this paper, we tested these assumptions using Japanese bond ratings data by the non-proportional and additive adjacent categories logit model. Consequently, we found that these assumptions were both not satisfied.
