Consequences of Basel II for the individual SME company

1. Consequences of Basel II for the individual SME company H.A. Rijken Vrije Universiteit, Amsterdam International Conference Small business banking and financing: a global perspective University Cagliari, NYU Stern School of Business, Leeds University Business SchoolUniversity of Trieste, European Commission Cagliari, 25-26 May 2007

2. Content • Spreads from a Basel II banking perspective versus spreads in the bond market • Creditworthiness of SME vs. larger companies • Consequences for the spread of SMEs vs larger companies • Consequences for the individual SME company

3. For unexpected losses banks must hold capital: capital requirements which are calculated within the VAR concept K Prob < 1 - C 0 0 PDLGD PDLGD Assumption: no fat tails in the distribution (Mandelbrot)

4. Capital requirements are determined by the following formulas • PD is the default probability • C is the confidence level in the VAR calculations •  is the default correlation among companies. This correlation is assumed to be lower for SME companies leading to a 20% reduction in capital requirements • Parameters PD, LGD, M have to be determined by historical data

5. Relationship between PD and capital requirements differs by approach

6. Spread calculation from a banking perspective Spread = Interest income – financing costs = Expected loss + Costs related to unexpected loss + operational costs = Expected loss + costs of equity + operational costs = LGD  PD + Krequired equity rate of return + operational costs = LGD  PD + K15% + 30bp

7. Expected losses are negligible for ratings above BBB/Baa1 Expected annual loss: PD×LGD

8. Basel I bank spread = PD×LGD + KBasel I×15% + 30bp Basel I banking spread (in practice K ≈ 10–12 %) Expected Loss: PD×LGD

9. A gap exists between Basel I spread and market spread Spread in bond markets: ‘82 – ‘04 Basel I banking spread Expected Loss: PD×LGD FRICTION

10. Basel II bank spread = PD×LGD + KBasel II×15% + 30 bp Spread in bond markets ‘82 – ‘04 Basel II banking spread Basel I banking spread Expected Loss: PD×LGD

11. Average market spread 82 – ’04 fully overlaps the Basel II banking spread Possible conclusions 1. The Basle II model (KBasel II×15%) is a good proxy for the risk premium investors and bankers demand for unexpected losses. 2. The Basle II model is a good proxy how bankers price their debt. Bankers are dominant in setting the price in debt markets. 3. Basel II model parameters are chosen in such a way that a perfect match shows up between average market spreads and Basel II model spreads.

12. Perhaps the Basel II banking spread will become a standard in financial markets Basel II model is - relative easy to calculate - it is a standard set by BIS - the relative simple VAR approach fits with the investor’s intuition how to quantify (and price) credit risk Alternative explanations for the relative high spreads for A – AAA bonds - a liquidity premium of 45 bp (De Jong and Driessen, 2005) - a high asset volatility, however structural models fail to explain the gap (for discussion see Longstaff, 2005)

13. How will banks set interests rates in a (new) Basel II environment, with a special focus on SME versus larger companies? Consequences of Basel II are simulated with a “banking portfolio” consisting all firms available in the Compustat database 1. The Compustat data is used to estimate a bankruptcy prediction model → sensitivity credit scoring models to Size 2. Based on ranked credit scores equivalent S&P ratings are for all firms in the Compustat database → distribution of S&P ratings for SME vs. larger companies 3. Basel II capital requirements and Basel II banking spreads are calculated for 7 banking subportfolio’s → Basel II banking spreadsfor SME vs. larger companies taking into account the companies’ life cycle

14. To make the Compustat banking portfolio of more interest to other countries, 7 subportfolio’s by type and Size are formed Advantages of the COMPUSTAT database (compared to databases at banks) 1. complete: it contains all defaults in a specific market 2. accurate: it includes all information accurately 3. it covers a long period 1970 - 2001

15. Credit scoring models: a lot a freedom to specify these models modeltype: logit regression methodology

16. Accuracy of credit scoring models is measured by the ACR value ACR = shaded surface / 0.5 Low credit quality High credit quality

17. Three variables: profitability, solvability and Size are dominant in credit scoring models

18. Equivalent S&P ratings are defined based on ranked credit scores 16 S&P rating classes Credit score ranking Step 1: Observations with a known S&P rating are ranked by credit score. Step 2: 16 equivalent S&P ratings are defined with the same distribution as the actual S&P rating distribution Step 3: For each equivalent S&P rating class the maximum and minimum credit score is determined: [Cmin(R), Cmax(R)]. Step 4: On the basis of these intervals [Cmin, Cmax] the equivalent S&P ratings of all other observations are determined - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - Cmax NA NA Cmin

19. For stable companies LE companies are slightly more creditworthy than SME companies

20. For innovative companies differences in credit risk between LE and SME companies are larger

21. As expected loss-making companies are centered at low ratings

22. Differences between LE1 and LE2/SME become smaller 0.12% BBB+ BBB 0.41% Default rate BBB- 0.66% BB+ 1.20% 2.23% BB

23. Stable companies: Size does matter in terms of creditworthiness BBB+ BBB BBB- BB+ BB BB-

24. The lower profitability level and higher volatility level is not fully compensated by more conservative financing

25. For all credit scoring models the average equivalent S&P ratings are lower in the SME segment 9 = BBB, 8 = BBB-, 7 = BB+, 6 = BB, 5 = BB-, 4 = B+, 3 = B

26. Debt financing costs are (will be?) 120 bp higher for SME companies compared to large companies * * * * Computed on he basis of equivalent S&P rating distribution and Moody’s statistics

27. For an individual perspective the credit scoring model is relevant, not that much from a portfolio perspective standard deviations van 2 notch steps can make the difference between a B and a BB+ rating, a difference of 300bp in credit spread

28. Conclusions for the credit risk of SME company • Companies’ credit risk depends on Size, below an annual sales of 100 mln. • Accuracy of credit risk models is lower for SME companies. • Lower profitability and higher earnings volatility make SME firms more vulnerable. More conservative financing only partly compensates for this. • The creditworthiness of SME companies is more sensitive to the credit cycle. • If the Basel II model becomes the standard in credit pricing than - innovative SME companies will face higher costs of debt - loss making companies might go bankrupt more quickly • (Internal) credit rating of an SME depends strongly on the specific details of the credit rating system a bank puts in place. • management of credit risk by companies should get a higher priority

29. (Part of) Relationship banking is going to disappear in the SME segment • Internal rating systems are based on “hard” quantitative facts and become more influential in bank’s lending decisions • These systems offer more transparency within the bank and for the financial authorities (client as well ?) • Little room for the relationship manager to negotiate with the client • Relative high costs in the SME segment can be reduced • They have to be reduced to regain the Basel II investment costs