"continued from the previous page"

The quality of the ratings by rating agencies

It is relevant to investigate whether the agencies' assessments of the credit risk associated with the bond issuers are in accordance with the actual conditions.

The rating of a bond issue must take into account both the probability of default and the size of the loss in the event of default. The principle of the higher the rating, the lower the credit risk, must apply. The use of ratings is thus based on the fact that they provide a consistent ranking of the credit risk on liabilities. In this connection the level of the credit risk at the individual grades of the rating scale is less important.

For several years Moody's and Standard & Poor's have published historical data to illustrate the probability of default by rated issuers within various timeframes. Table 3⁶⁾ presents data for private business enterprises and central governments which have issued bonds or raised loans rated by Moody's. The Table clearly shows that the lower the rating of the enterprise, the greater the risk of default.

However, the average values in Table 3 conceal that the risk of default varies considerably in both the short and medium term.⁷⁾ This variation can be attributed to such factors as the effect of the business cycle on the issuer's ability to repay the loan. In principle, ratings do take into account the expected impact of present and future cyclical fluctuations on the ability to repay, but often any problems will only become significant in a recession.

It is important that the rating agencies pursue a reasonably consistent rating policy which is not subject to frequent changes. For investors it is important that the meaning of a rating is as stable as possible over time.

Investors and lenders who use ratings must likewise be aware that high ratings are changed less frequently than low ratings. If debt of high quality is subject to frequent changes, over a relatively short timespan this debt could shift from a very low risk of default to a position with a quite considerable risk. The value of a rating as a debt-ranking tool would thus be eroded. The stability requirement for high ratings in fact appears to be fulfilled, cf. Table 4, which shows the percentage annual change in Standard & Poor's rating of private business enterprises. Annual movements of more than one grade are rare.

While it is easy to relate to the rating agencies' historical data on the risk of default it is considerably more difficult to evaluate whether a rating adequately reflects the expected loss in the event of default. Moody's has sought to solve this problem by setting the loss ratio as the difference between par and the price at which a defaulted liability was traded in the market one month after the default took place. However, this method is extremely sensitive to the development in interest rates and in the economy. Moody's indications of average loss ratios are also given with very broad confidence ranges.

On evaluating Tables 3 and 4 it is noteworthy that the data basis is dominated by US companies, so that the data cannot immediately be taken to express the rating agencies' ability to rate central governments or business enterprises established in countries whose institutional and legal systems deviate significantly from the conditions in the USA. Experience from the past year's financial and economic crisis in a number of East-Asian countries indicates that no universal interpretation can be applied to ratings by rating agencies.

A clear positive correlation can be observed in the US capital market between an issuer's rating and the interest rate required of the issuer in conjunction with bond issues, measured by the interest-rate differential to US government bonds, cf. Chart 1 and Box 3. The ranking by credit risk undertaken by the rating agencies can thus also be observed in the market. It is important to understand that even if the rating scale comprises more than

"to be continued on next page"

Fodnoter