Bankruptcy Modelling

Over a long time, bankruptcy prediction has been the focus of considerable academic effort. For

example; a number of papers by Altman using expost univariate analysis where we could observe,

for example, that companies which had greater capacity to absorb interest payments were less likely

to fail. The difficulty though is in applying these results to exante prediction which becomes very

prone to type I/II errors. Altman went on to develop a multivariate approach to bankruptcy

prediction (Zscore model). Similar effort by Beaver (1968), Zmijewski (1983), Ohlson (1980) and

many others. The relatively high misclassification rates were still often found to be present. This may

have inspired the development of socalled marketbased models (which drew their inspiration from

the BlackScholes option pricing model) and which have been operationalised in real commercial

models which have been (and are) used by banks. Are these marketbased models any better though?

What is the evidence?

Suggested Reading would include, but not be limited to:

Penman Ch. 20

Ohlson (1980)

Taffler (1984)

Agarwal & Taffler (2007, 2008)