Linear-Disriminant Model

[vibhakary]

Hi David,

One of the drawbacks mentioned for Linear-Discriminant Model is 'Centralized Business Loans Data are not available'. Why do we need Centralized Business Loans Data for Linear-Discriminant Model?

Thanks,
Vib

 

[David Harper CFA FRM]

Hi Vib,

In brief, to reduce sample bias. These multivariate models like Altman's Z produce coefficients based on analyzing an actual historical sample of defaulted (versus non-defaulted) firms. The linear coefficients are only as reliable as the sample. It's not just sample size, it's the type of firms/loans in the sample (large or small firms? senior or junior debt? industry?). So, the idea is that with a centralized DB, a superior sample would lead to a more relevant set of coefficients...David