try as a “developing” or a “developed” country in this survey is in the context of fail-
ure prediction and may deviate somewhat from the traditional grouping of the country.
The main characteristics of developed country models are: (1) failure prediction
studies have a long history, (2) corporate financial data are more readily available, (3)
failure is easier to identify because of the existence of bankruptcy laws and banking
infrastructures, (4) government intervention is somewhat less, but not nonexistent,
and (5) there is a more sophisticated regulation of companies to protect investors.
The developing country models are characterized by the relative absence of the above
factors. In developing countries, where free market economies have not taken hold,
a company’s failure is harder to see because of the degree of protection provided by
the government. However one may also point to similar practices in developed coun-
tries, notably the United Kingdom, Germany, Japan, to a lesser extent, and even the
United States on some rare occasions, for example, the case of Chrysler in 1980.
Exhibit 10.1 summarizes the 39 studies from 21 countries included in this survey.
We have not included summaries of nonpublished studies although we are aware of
several, for example, two from South Africa and several in languages other than Eng-
lish (e.g., Korean).
While we believe this international treatment of failure prediction models is the
most comprehensive effort to date, we recognize that some relevant works will pos-
sibly be overlooked in this survey and apologize for any omission. Note: The term
“author” or “authors” in the succeeding paragraphs pertains to the authors of the re-
spective articles, not to the authors of this review.
(b) Emerging Markets Application. One of the models presented in this chapter was
developed by Altman, Hartzell, and Peck (1995) to rate the credit quality of emerg-
ing markets corporate debt. We discuss it below in the context of Mexico—one of the
prime countries whose companies have tapped the international bond markets in re-
cent years. This application has particular relevance since the vast majority of Mex-
ican, Latin American, and emerging market countries’ corporate debt in general, is as
yet still unrated by the major rating agencies. The model is a variation on the origi-
nal Z-Score model developed by Altman (1968).
(c) Altman, Hartzell, and Peck, (1995). Most of the models presented in this chap-
ter are based on data from individual firms in a specific country and the resulting
model is unique for that country. The one exception is the model discussed in 10.1
(b) where, as noted, we used a variation on the original Z-Score model to predict dis-
tress and bond rating equivalents for emerging market corporate debt. In this case, we
advocated that a single model (Altman, Hartzell, and Peck, 1995) could be used in
any developing country and possibly for nonmanufacturing industrial firms in the
United States, as well.
In all cases, the models discussed are used to analyze individual firms. These mod-
els and the techniques used in their development (e.g., discriminant, probit, logit re-
gressions) have become extremely important and relevant as the Bank for Interna-
tional Settlements (BIS) is in the process of recommending that most banks develop
internal rate–based models (IRBs) for rating their customers’ credit risk. The so-
called Basel-2 accords are being debated as we update this article, but it is clear that
the resulting IRBs for most banks will be variations of the types of models presented
in this chapter.
A potentially important extension of these models is to use them to assess country or
10 • 4 BUSINESS FAILURE CLASSIFICATION MODELS