![Sample our Social Sciences journals, sign in here to start your FREE access for 14 days]({"type":"image" , "src" : "/sda/110822/TOC-Subject-Sample-2021-Social-Sciences.jpg"})
Abstract
Debt Service Capacity (DSC) is revisited within a new framework, i.e., Neural Networks (NN), an established method in the field of artificial intelligence (AI). A select group of traditional models that have been used to predict DSC is also reviewed and the logit approach is chosen for comparison with NN.
Ten measures (indices) are constructed as determinants of DSC. To keep the comparison simple, empirical applications of the two models are controlled by employing the same data sample and the same holdout and estimation groups in both models. A few drawbacks of the traditional approaches, especially the presence of multicollinearity, are pointed out. It is shown that when multicollinearity is present, the exponential of the coefficients cannot be explained as the odds ratio.
A number of tests and sensitivity analyses are performed to exhibit the attributes and the performance of the two models.