Estimated Cumulative Default Rates by Category of Credit Ratings- Results of JCR Cumulative Default Rate Estimation Model for Individual Companies-
JCR News
Estimated Cumulative Default Rates by Category of Credit Ratings- Results of JCR Cumulative Default Rate Estimation Model for Individual Companies-
Jan. 19, 1999
JCR announced today its estimated cumulative default rates (CDRs) by category of credit rating, calculated by the JCR Cumulative Default Rate Estimation Model for Individual Companies.
Each of the estimated CDRs below is an average of calculated default rate of individual companies classified by category of credit ratings assigned by JCR for the period of fiscal 1991(ending March 31,1992) to fiscal 1997, which are calculated by the Model. JCR intends to improve the objectivity of its credit ratings through continuous development or revision of this kind of model, while it engages in in-depth researches and analyses on industries as well as individual companies by analysts.
Estimated CDRs based on JCR Model for Individual Companies (%)
Rating Year 1 Year 2 Year 3 Year 4 Year 5
AAA 0 0 0 0.03 0.04
AA 0.13 0.27 0.41 0.49 0.62
A 0.21 0.43 0.67 0.85 1.1
BBB 0.26 0.55 0.86 1.15 1.48
BB 0.31 0.66 1.06 1.44 1.89
up to B 1.26 4.18 6.62 10.03 13.27
Notes:
1) An estimated CDR is an average for each category. For credit rating categories having few samples, the rate was statistically estimated.
2) The CDRs by category of credit ratings are not actual default rates in the past for companies rated by JCR. They do not always correspond to future default rates, either.
JCR Cumulative Default Rate Estimation Model
The Model is a general-purpose model formulated based on default samples covering wide-ranging listed and over-the-counter companies, and making use of many financial data. The Model has the following characteristics:
1) First, the Model is formulated to reflect actual default rates peculiar to Japanese companies by selecting samples in the way that is presumed to lead to meaningful estimates of them. Second, the diverse information obtained from relevant financial statements is incorporated in the Model to enhance the reliability of its estimation.
2) Negative net worth companies are added to the default sample as quasi-default. This is because JCR considers that an estimated default rate based on the Model which treats negative net worth as default would be more appropriate for the purpose, in view of an increasingly severe financial environment surrounding Japanese companies, under which it will become harder for negative net worth companies to survive. In this sense, it may well be said that a rate produced by the Model is considered to be an estimated probability of an occurrence of default of a company.
3) The input data for the Model are financial data on a company under review. These are selected out of more than fifty financial indicators using multivariate analysis. They consist of various elements which cover a wide-range of financial indicators considered to be necessary to extract the financial characteristics of defaulted companies: financial ratios, figures in financial statements, and trends of them.
JCR Cumulative Default Rate Estimation Model for Individual Companies
On January 19, 1999, JCR announced its estimated CDRs, which were produced by the JCR Model, classified by category of its credit ratings covering listed and over-the-counter companies other than banks, nonbank financial companies and insurers rated by JCR. The development process and features of the Model are described briefly below.
1. Development Process of the Model
This model is formulated so that it enables to estimate a cumulative default rate for the duration of one to five years of a listed or over-the-counter company belonging to the manufacturing or non-manufacturing sector of industry (excluding trading, railway, airline and utilities companies). A discriminant which differentiates defaulted companies from sound ones was sought out by tracing back to one through five years prior to a default using discriminant analysis. Making use of the logistic regression analysis, JCR estimates a default rate by the conversion of a calculated figure into the estimated default rate.
The sample of defaulted companies consists of those listed and over-the-counter companies which were actually defaulted and were in the state of negative net worth (quasi-defaulted ones). JCR infers the parameters for the conversion based on past data.
2. Features of the Model
(1) Actuality-based Estimate of Default Rate
In formulating the Model, JCR includes the state of negative net worth as quasi-default so that its estimated CDRs can better reflect expected realities of Japanese companies falling into default in the future. This is because JCR sees that the probability of companies with negative net worth to continue their business activities will become very low in view of a severer environment in prospect for corporate finance. JCR considers that the Model into which this factor is incorporated would better reflect realities of default by Japanese companies.
JCR's default sample, including quasi-default, is selected from companies listed or registered as over-the-counter companies in 1980 and thereafter. The calculated yearly default rates are examined for the formulation of the Model (see Table 1).
(2) Input Data Selected from Wide-ranging Financial Indicators
The input data for the Model are financial data of individual companies. JCR selects 53 financial indicators from those which are generally used by JCR, and then adopts 4 to 9 variables (financial indicators) through discriminant analysis that is a method of multivariate analysis. The financial indicators selected for the Model are many in number and consist of those which include ratios, absolute numbers, and trends from financial statements, relevant to the extraction of the financial characteristics of companies in default.
Table 2. Financial Indicators
Size 7
Stability 15
Growth 3
Profitability 11
Efficiency 8
Debt Service 9
Total 53
(3) Estimated Cumulative Default Rate Consolidating Versatile Information
Estimation formulas for the Model are developed for manufacturing and non-manufacturing industries (excluding general trading, railway, airline and utility companies) respectively for each year of the duration. The input data for the Model, therefore, are selected independently from those financial indicators which reflect the financial conditions of defaulted companies and sound ones for each year of the duration according to a discriminant analysis. As a result, a CDR inferred by the Model becomes an estimate that reflects more versatile financial information with the advance of years.
The financial indicators adopted for the formulas of the Model for each of the 5-year duration are as follows.
Table 3 indicates that financial indicators distinguishing defaulted companies from sound ones are those indicating soundness, profitability and debt-service capability. It is verified that there is no strong correlation (multicollinearity) between the indicators adopted for the Model. Accordingly, the Model is considered to be statistically stable.
3. The General-purpose Model
The results of application of the Model by category of credit ratings to the companies rated by JCR show significant distinguishability according to the rating category. It should be noted that this Model is so formulated as to be able to apply to listed and over-the-counter companies in general. JCR considers that the above-mentioned distinguishability shows that the Model is general-purpose and consistent with JCR's credit ratings. Naturally, the Model is developed so as to estimate default rates of individual companies generally, not for the purpose of explaining or inferring JCR's category of credit ratings as is understandable from the development process mentioned above.
Figure 1 below shows the frequency (percentage) of the estimated CDRs calculated by the Model covering over 17,600 listed and over-the-counter companies, which are applied to the Model, for the period of fiscal 1991 to 1997. It can be recognized that averages of estimated default rates of individual companies increase with the elapse of years, and the variance increases the wider as the prediction becomes the longer in duration with each graph shaping unimodal.
Figure 1. Distribution of the estimated CDRs
( for the period of fiscal 1991 to 1997 )
