CREDIT RISK MODEL for Banking Counterparties. Fernando Zimet, CFA Banco Central del Uruguay March 2013

Transcription

1 CREDIT RISK MODEL for Banking Counterparties Fernando Zimet, CFA Banco Central del Uruguay March 2013

2 Agenda 1. Introduction. 2. Financial Indicators. Credit Ratings. Credit Default Swaps (CDS). 3. Credit Risk Model. Overview. Inputs. Construction. Backtesting. Further Research. 4. Example.

3 Agenda 1. Introduction. 2. Financial Indicators. Credit Ratings. Credit Default Swaps (CDS). 3. Credit Risk Model. Overview. Inputs. Construction. Backtesting. Further Research. 4. Example.

4 Introduction Considering credit ratings as the only source of information to select deposit counterparties might be insufficient: In a world of financial constraints and rapidly changing environments, ratings might be out of date. Credit ratings highly questioned because of severe misjudgements of financial instruments by the major credit ratings agencies (Standard & Poor s, Fitch Ratings and Moody s). Main guideline: include updated market information in the credit assessments.

5 Introduction Requirement: A model combining public accounting data with quickly observable and up to date unbiased information. Solution: Construct a model which combines credit ratings with market information about credit risk assessments (CDS). Desirable characteristics: Quickly observable. Easy to update in a context of limited human resources. Complete summary of all the available information.

6 Agenda 1. Introduction. 2. Financial Indicators. Credit Ratings. Credit Default Swaps (CDS). 3. Credit Risk Model. Overview. Inputs. Construction. Backtesting. Further Research. 4. Example.

7 Financial Indicators: Credit Ratings Despite all the cons, credit ratings should not be disregarded from the analysis. Why? Specific credit analysis of each institution made by recognized professionals. The market uses them, so the credit quality converges. Based on the same type of accounting calculations as banking supervision institutions do. Easy to obtain and process.

8 Financial Indicators: CDS Credit Default Swap (CDS): Credit derivative contract between two counterparties. Characteristics: Buyer makes periodic payments to seller. Payment from seller is triggered by a credit event defined in the contract (i.e.: default). Both may own the underlying or not. Arbitrage: CDS premium/price should converge to the credit spread of a zero coupon corporate bond with the same maturity: Corporate Bond + CDS = T-bond Corporate Bond T-bond = - CDS CDS converges with the perceived risk of an instrument or institution. Liquidity concerns are not relevant because they are a minor part of the pricing of instruments.

9 Agenda 1. Introduction. 2. Financial Indicators. Credit Ratings. Credit Default Swaps (CDS). 3. Credit Risk Model. Overview. Inputs. Construction. Backtesting. Further Research. 4. Example.

10 Overview The main idea of the model is to combine the information included in credit ratings with market information contained in CDS (50% weight for each group of information). If any counterparty doesn t meet a minimum threshold in any of both measures, the early warning model will classify them as high risk entities. A credit index is constructed with these two inputs. If any counterparty doesn t meet a minimum threshold for the aggregate index, it will be classified as a high risk entity. With the counterparties that were not eliminated previously, a ranking is elaborated.

11 Inputs Credit Rating indicator construction: 1) Express Fitch, Moody's and S&P ratings as numbers according to the table. 2) Calculate an unweighted average for both, short term and long term indicators. 3) Construct a synthetic indicator by weighting 70% for short term ratings and 30% for long term ratings. Fitch S&P Moody s Valor F1+ A-1+ P-1+ 6 F1 A-1 P-1 4 F2 A-2 P-2 2 AAA AAA Aaa 6 AA+ AA+ Aa1 5 AA AA Aa2 4 AA- AA- Aa3 3 A+ A+ A1 2 A A A2 1 A- A- A3 0

12 Inputs Credit Default Swaps (CDS) indicator. Basic indicator: Senior 5 year CDS in home currency expressed in bps. Adequate liquidity. Includes short and long term implied default probabilities. Because of arbitrage arguments, CDS in different currencies are comparable. For symmetry, the indicator used is the inverse of the CDS (1/CDS).

13 Steps: Model construction 1) Adjust a normal distribution to both, CDS and credit ratings indicators. The institutions located under the 20% more fragile tranche are assigned to the red alert group. High Risk Group Credit Ratings, CDS If tested empirically, a normal distribution is rarely rejected for both indicators separately. The universe used is the one composed by the major financial institutions with a credit rating of at least A+ (minimum credit rating tolerated).

14 Model construction Steps: 2) Both CDS and credit ratings indicators are standardized using the following formula: Z = ( X µ ) σ where Z stands for the standardized parameter, X for the original data, m for the mean and s for the standard deviation of the original data.

15 Steps: Model construction 3) A summary index is constructed by weighting 50% each indicator for each institution. 4) A normal distribution is adjusted to the summary index, which is divided in 20% probability tranches. The five tranches have the following allocation. 5) Institutions eliminated in step 1 is placed in the high credit risk group. Group I II III IV V Credit Risk Low Medium low Medium Medium high High V IV III II I

16 Model construction Exceptions: Canadian banks CDS markets are not developed. Instead, we use as a proxy of credit risk the spread between corporate and government bonds of the same duration.

17 Model construction Historical allocations by the model are tested for banks that notoriously had financial problems: Three major banks liquidated during the 2008 financial crisis They would have been classified as group V, during all of the period starting Jan The biggest private bank bailed out during the financial crisis Would have been classified as groups III or IV until September,18th 2008, moving to group V on that date.

18 Model construction Further developments: Improve the CDS information platform: from Bloomberg to Markitt. Check market news for unexplained movements in CDS premia. Restrict the maturity of time deposits with financial strength and not only the amount of them.

19 Agenda 1. Introduction. 2. Financial Indicators. Credit Ratings. Credit Default Swaps (CDS). 3. Credit Risk Model. Overview. Inputs. Construction. Backtesting. Further Research. 4. Example.

20 Example Supose a bank with a senior 5 year CDS of 110 bps with the following credit ratings: Short Term Long Term Fitch F1+ AA- S&P A-1+ AA- Moody s P-1 Aa1 Using the table shown before, the credit ratings can be translated into numbers and averaged. Short Term Long Term Fitch 6 3 S&P 6 3 Moody s 4 5 Average Rtg 5,33 3,67 The rating score is 30% long term rating + 70% short term rating. In this case: 0,3*3,7+0,7*5,3=4,8

21 Example Summarizing institutional and market data: CDS 1/CDS Credit Rating Institution 110 0,91 4,8 Market data Mean 176 0,76 4,4 Std. Dev ,39 0,8 STEP 1 Check if the institution is not in the 20% most fragile of each category. Accumulation of 20% probability according to the normal distribution: Mean 0,84 Std. Dev. 1/CDS: 0,76 0,84*0,39 = 0,43 Credit Rating (CR): 4,40 0,84*0,8 = 3,72 In both cases, the institution has better values than the thresholds we can continue into step 2. Otherwise, we would clasify the institution as group V immediately.

22 Example Summarizing institutional and market data: CDS 1/CDS Credit Rating Institution 110 0,91 4,8 Market data Mean 176 0,76 4,4 Std. Dev ,39 0,8 STEP 2 We standardise the institution's values: CDS: Z CDS =(X-µ)/σ = (0,91-0,76)/0,39 = 0,38 CR: ZCR=(X-µ)/σ = (4,8-4,4)/0,8 = 0,55 STEP 3 The summary index is constructed: Index = 0,5*ZCDS+0,5*ZCR = 0,5*0,38+0,5*0,55=0,47

23 Example STEP 4 The index's mean is 0 by construction, and it's standard deviation won't be too far away from 1 (in this case we will suppose it is 1). According to this figures, the tranche selection is the following. Tranche Min Max Allocation I 0,84 + inf 15% II 0,25 0,84 10% III -0,25 0,25 5% IV -0,84-0,25 3% V - inf -0,84 0% This institution, with a synthetic index of 0,47, would be ranked as part of group II.

24 THANK YOU!