A new ordinal mixed-data sampling model with an application to corporate credit rating levels

In this paper we propose a new ordinal logistic regression model (OLMIDAS) that allows the inclusion of independent variables at higher frequencies than that of the dependent variable. A simulation study shows that our proposed model can find the true patterns in the data. In an empirical study we apply OLMIDAS to the prediction of corporate credit rating levels and compare its performance to classical logistic regression models with an annual aggregation of the higher-frequency variable, such as ordinal logistic regression and multinomial logistic regression. We find that OLMIDAS outperforms the classical logistic regression model while providing additional knowledge of the structure of the higher-frequency explanatory variable.