approach to assess country risk, as opposed to the more traditional “top-down” meas-
ures (e.g., macroeconomic variable), was, in our opinion, an important contribution.
10.2 JAPAN. In Japan, bankruptcies are concentrated in the small- and medium-
size firms, especially those that do not enjoy the protection of an affiliated group of
companies. These groups, known as “Keiretsu,” usually involve a leading commer-
cial bank and a number of firms in diverse industries. Still, a number of larger firms
listed on the first section of the Tokyo Stock Exchange have succumbed to the nega-
tive economic reality of failure. A comparison of the business failures in Japan and
the United States may be made based on these statistics appearing in the Failure
Recordpublished by Dun & Bradstreet and Tokyo Shoko Koshinso, among others.
There have been a number of studies concentrating on failure prediction in Japan—
most were built prior to 1984. Although we will discuss just two, the reader can find
reference and discussion to at least a half dozen more in Altman (1983).
(a) Takahashi, Kurokawa, and Watase (1984). Using multiple discriminant analysis,
over 130 measures on individual firms, 36 pairs of failed and non-failed manufactur-
ing firms listed on the Tokyo Stock Exchange in the period 1962–1976 and 17 dif-
ferent model types, the authors have constructed a failure prediction model using the
following measures:
- Net worth/fixed assets
- Current liabilities/assets
- Voluntary reserves plus unappropriated surplus/total assets
- Borrowed expenses (interest)/sales
- Earned surplus
- Increase in residual value/cash sales
- Ordinary profit/total assets
- Value added (sales—variable costs)
The authors suggest that their model could be more accurate than Altman’s (1968)
because of (1) its simultaneous consideration of data from one, two, and three years
prior to failure, (2) its combination of ratios and absolute numbers from financial
statements, (3) its utilization of the cash basis of accounting from financial statements
as well as the accrual base, and (4) its adjustment of the data when the firm’s audi-
tors express an opinion as to the limitations of the reported results (window dressing
problem).
It was found that models with several years of data for each firm outperformed a
similar model with data from only one year prior to failure. Further, absolute finan-
cial statement data contributed to the improved classification accuracy and data from
financial reports prepared external to the firm on an accrual basis were more predic-
tive than those prepared from an “investment effect” or cash basis method. Adjusting
the data to account for auditor opinion limitations improved the information content
of the reported numbers and ratios. A holdout sample of four failed and 44 nonfailed
firms was tested with the selected model. The four failed firms went bankrupt in
1977, that is, the year after the last year used in the original model.
One problem with the above model might be the use of several years of data for
the same firm in order to construct a model. The authors apparently were aware of
10 • 6 BUSINESS FAILURE CLASSIFICATION MODELS