Summary: | 碩士 === 淡江大學 === 管理科學研究所 === 81 === Our government has promoted the policy of banking liberali-
zation and internationalization recently and permitted the
esta- blishment of 16 new commercial banks in 1991 and 1992. It
has made the environment of banks become more competitive. So,
it''s necessary to develope a loan credit evaluation model for
lower- ing loan risk. In reference to relevant studies, most
resear- chers have studied the samples of mixed industries and
built models with company''s original ratios due to it''s
difficult to collect credit data. However, there is a great
difference of financial characteristics among industries and
company''s origi- nal ratios are influenced by economic factor.
So this study will focus the sample on electrical and
electronics industries and build a loan credit evaluation model
with industry-relative ratios. The sample of this study
consisted of 43 companies that have made a breach of contract
and 43 matching companies that haven''t made it. The data was
collect for the years 1989 to 1992 and 23 company''s original
ratios and industry-relative ratios as vari- ables were
calculated. We use normal test at first; secondly, we analyzed
the difference of financial characteristics among com- panies;
and then whether adopted factor analysis, we used vari- able
analysis to gather determinants. At last, the logit regre-
ssion models were established with determinants and classifi-
cation correctly, type one error, type two error and weighted
efficiency were assessed. As the evidence turn out: Among the 8
loan credit evaluation models, the accuracy and weighted
efficiency of models built with industry-relative ratios and
didn''t adopt factor analysis are higher than those of models
built with company''s original ratios and adopted factor
analysis. Moreover, the accuracy of models for a single
industry are higher than those of models for mixed industries.
|