In most countries only a small percentage of firms has been rated to date. The lack of rated firms is mainly due to two factors. Firstly, an external rating is an extremely costly procedure. Secondly, until the recent past most banks decided on their loans to small and medium sized firms (SME) without asking for the client's rating figure or applying an own rating procedure to estimate the client's default risk. At best, banks based their decision on rough scoring models. At worst, the credit decision was completely left to the loan officer.
Since learning to know its own risk is costly and, until recently, the lending procedure of banks failed to set the right incentives, small and medium sized firms shied away from rating. However, the regulations are about to change the environment for borrowing and lending decisions. With the implementation of the New Basel Capital Accord (Basel II) scheduled for the end of 2006 not only firms that issue debt securities on the market are in need of rating but also any ordinary firm that applies for a bank loan. If no external rating is available, banks have to employ an internal rating system and deduce each client's specific risk class. Moreover, Basel II puts pressure on firms and banks from two sides.
First, banks have to demand risk premia in accordance to the specific borrower's default probability. Table 10.1 presents an example of how individual risk classes map into risk premiums (Damodaran; 2002) and (Füser; 2002). For small US-firms a one-year default probability of 0.11% results in a spread of 2%. Of course, the mapping used by lenders will be different if the firm type or the country in which the bank is located changes. However, in any case future loan pricing has to follow the basic rule. The higher the firm's default risk is the more risk premium the bank has to charge.
Rating Class (S&P) | One year PD (%) | Risk Premia (%) |
AAA | 0.01 | 0.75 |
AA | 0.02 - 0.04 | 1.00 |
A+ | 0.05 | 1.50 |
A | 0.08 | 1.80 |
A- | 0.11 | 2.00 |
BBB | 0.15 - 0.40 | 2.25 |
BB | 0.65 - 1.95 | 3.50 |
B+ | 3.20 | 4.75 |
B | 7.00 | 6.50 |
B- | 13.00 | 8.00 |
CCC | ![]() |
10.00 |
CC | 11.50 | |
C | 12.70 | |
D | 14.00 |
Second, Basel II requires banks to hold client-specific equity buffers. The magnitudes of these buffers are determined by a risk weight function defined by the Basel Committee and a solvability coefficient (8%). The function maps default probabilities into risk weights. Table 10.2 illustrates the change in the capital requirements per unit of a loan induced by switching from Basel I to Basel II. Apart from basic risk determinants such as default probability (PD), maturity and loss given default (LGD) the risk weights depend also on the type of the loan (retail loan, loan to an SME, mortgages, etc.) and the annual turnover. Table 10.2 refers to an SME loan and assumes that the borrower's annual turnover is 5 million EUR (BCBS; 2003). Since the lock-in of the bank's equity affects the provision costs of the loan, it is likely that these costs will be handed over directly to an individual borrower.
Rating Class | One-year | Capital | Capital |
(S&P) | PD (%) | Requirements | Requirements |
(%) (Basel I) | (%) (Basel II) | ||
AAA | 0.01 | 8.00 | 0.63 |
AA | 0.02 - 0.04 | 8.00 | 0.93 - 1.40 |
A+ | 0.05 | 8.00 | 1.60 |
A | 0.08 | 8.00 | 2.12 |
A- | 0.11 | 8.00 | 2.55 |
BBB | 0.15 - 0.40 | 8.00 | 3.05 - 5.17 |
BB | 0.65 - 1.95 | 8.00 | 6.50 - 9.97 |
B+ | 3.20 | 8.00 | 11.90 |
B | 7.00 | 8.00 | 16.70 |
B- | 13.00 | 8.00 | 22.89 |
CCC | ![]() |
8.00 | ![]() |
CC | 8.00 | ||
C | 8.00 | ||
D | 8.00 |
Basel II will affect any firm that is in need for external finance. As both the risk premium and the credit costs are determined by the default risk, the firms' rating will have a deeper economic impact on banks as well as on firms themselves than ever before. Thus in the wake of Basel II the choice of the right rating method is of crucial importance. To avoid friction of a large magnitude the employed method must meet certain conditions. On the one hand, the rating procedure must keep the amount of misclassifications as low as possible. On the other, it must be as simple as possible and, if employed by the borrower, also provide some guidance to him on how to improve his own rating.
SVMs have the potential to satisfy both demands. First, the procedure is easy to implement so that any firm could generate its own rating information. Second, the method is suitable for estimating a unique default probability for each firm. Third, the rating estimation done by an SVM is transparent and does not depend on heuristics or expert judgements. This property implies objectivity and a high degree of robustness against user changes. Moreover, an appropriately trained SVM enables the firm to detect the specific impact of all rating determinants on the overall classification. This property would enable the firm to find out prior to negotiations what drawbacks it has and how to overcome its problems. Overall, SVMs employed in the internal rating systems of banks will improve the transparency and accuracy of the system. Both improvements may help firms and banks to adapt to the Basel II framework more easily.