04cbe2206199d2381c37c6846018c0b6.ppt
- Количество слайдов: 22
Presentation to National Housing Bank Risk Scoring and Risk Based Pricing of Home Loans ICRA Management Consulting Services Limited April 20, 2012 New Delhi © IMa. CS 2012 Printed 15 -Mar-18 Page 1
Risk based pricing enables better risk management Risk Identification 1. A rating model or scorecard will discriminate good and bad borrowers 2. Identify risks in the property (collateral) Risk Measurement 1. Estimate credit losses through models 2. Compute credit risk premium from risk grading Risk Mitigation 1. Manage anticipated credit losses by provisioning and risk based pricing 2. Maintain capital to absorb adverse losses © IMa. CS 2012 Printed 15 -Mar-18 Page 2
Losses Credit losses can be divided into expected loss and unexpected loss Unexpected Loss Expected loss Year Expected Loss Average loss in the course of business Managed by pricing and provisions Unexpected Loss Frequency Peak losses in excess of expected loss Managed with capital cushion Credit Loss © IMa. CS 2012 Printed 15 -Mar-18 Page 3
Expected loss is the average loss anticipated in the course of business 1. Forecast of average level of credit losses a firm reasonably expects to experience in a year 2. One of the cost components of doing business 3. Managed by pricing and provisioning 4. For e. g. a. On an average, out of 100 AA borrowers, two of them default at the end of a normal business year b. On an average 10% is the loss in the realisation of the asset 5. Rating model or scorecard will help estimate expected loss scientifically Expected Loss = Probability of Default * Loss Given Default © IMa. CS 2012 Printed 15 -Mar-18 Page 4
Unexpected loss is a peak loss that exceeds expected loss 1. Peak losses do not occur every year but can potentially be very large 2. Capital acts a cushion to absorb unexpected losses 3. Losses can exceed expected losses due to reasons like a. Economic slowdown, higher interest rates leading to more defaults b. Correction in property prices leading to negative equity Capital required = Exposure * Risk Weight Risk weight is based on loan amount and LTV © IMa. CS 2012 Printed 15 -Mar-18 Page 5
Risk based pricing Riskbased pricing The cross-subsidy 1 Good credits overpriced Interest Rate Subsidy 2 Typical Pricing 2 Bad risks under-priced 1 Low High Risk q Good accounts subsidize poor credit risk accounts q Risk based pricing can mitigate problem of adverse selection © IMa. CS 2012 Printed 15 -Mar-18 Page 6
Risk Adjusted Return on Capital Employed Risk Adjusted Return Income Interest Income Fee and other non interest income Expenses Interest Expense Origination and servicing costs Risk Adjustment Provisions / Expected Loss Capital Required (regulatory) / Employed (economic)
Cost heads considered in pricing… as % of exposure Cost of Funds Cost of borrowing Direct and Indirect Costs q Loan origination and servicing cost + q Other overheads Provisions Maximum of q Existing provisions apportioned q Expected loss computed as PD * LGD * Exposure Opportunity Cost of Regulatory Capital q Hurdle rate based on Ro. E * q Regulatory capital © IMa. CS 2012 Printed 15 -Mar-18 Page 8
Risk based pricing – an example 12. 00% 10. 00% 0. 50% 0. 90% 0. 10% 10. 90% Cost of capital Processing fee Lending rate 8. 50% 8. 00% 6. 00% 4. 00% 2. 00% 0. 00% Cost of funds Overhead cost Credit risk premium © IMa. CS 2012 Printed 15 -Mar-18 Page 9
Credit risk premium depends on the rating of the borrower Assumptions a. Cost of funds 8. 50% b. Overhead cost 1. 00% c. Processing Fee 1. 00% d. Regulatory capital 12% e. Return on Capital 10% f. Risk weight for home loans g. Cost of capital = d * e * f Quality of borrower Credit risk Excellent Negligible 0. 30% Good Moderate Poor PD 50%, 75% 0. 60%, 0. 90% LGD Expected loss = PD* LGD Risk based pricing 10% 0. 03% 10. 43% Low 1. 00% 10% 0. 10% 10. 50% Medium 2. 00% 10% 0. 20% 10. 60% High 5. 00% 10% 0. 50% 10. 90% *LGD is assumed as 10% as per Basel guidelines Processing fee is amortised over 10 years © IMa. CS 2012 Printed 15 -Mar-18 Page 10
Benefits of a rating model 1. Decision to lend – reduce adverse selection problem 2. In case of lending for a poor credit worthy borrower, what additional collateral to be sought 3. Measure risk and price loans in a scientific manner 4. Achieve consistency across the organisation 5. Perform analysis of portfolio using risk scores, drivers of risk in the rating model © IMa. CS 2012 Printed 15 -Mar-18 Page 11
Explanatory Variables in the Home Loan Model Qualitative Cost of Living Quantitative Family structure Joint/Nuclear Skill level Years of Banking Years of Experience Quality of Borrower Age Others Marital Status No. of Residence dependents type Loan to Value Fixed Obligation / Income EMI/NW Loan Amount Income © IMa. CS 2012 Printed 15 -Mar-18 Page 12
Quantitative Indicators Fixed Obligation to Income Ratio (FOIR) 35. 0% 30. 0% 25. 0% 20. 0% 15. 0% 10. 0% 5. 0% 0. 0% FOIR 29. 5% 8. 5%0. 5% Less than 25% 9. 0% 0. 5% 25%-40% 28. 6% 0. 7% 0. 5% 40%-60% Relative Frequency Higher the FOIR, lower is the capacity of the applicant to absorb the negative shock in net income. Hence, higher the FOIR, lower is the ability of the applicant to meet unforeseen expenses. 1. 6% 24. 5% 60%-80% Greater than 80% 1. 8% 1. 6% 1. 4% 1. 2% 1. 0% 0. 8% 0. 6% 0. 4% 0. 2% 0. 0% Default Rate 1. From the data it is observed that if FOIR exceeds 60%-80%, default increases 2. The optimum range for lending in terms of most favorable default experience is the 40%-60% © IMa. CS 2012 Printed 15 -Mar-18 Page 13
Quantitative Indicators Loan to Value Ratio (LTV) 50. 0% 45. 0% 40. 0% 35. 0% 30. 0% 25. 0% 20. 0% 15. 0% 10. 0% 5. 0% 0. 0% LTV 47. 5% 1. 6% 27. 5% 17. 5% 2. 0% 0. 0% Less than 25% 0. 4% 5. 5% 0. 2% 25%-40% 40%-60% Relative Frequency Lower the LTV, greater is the applicants contribution towards the asset i. e. loss in event of default increases for the applicant. 1. 0. 6% 60%-80% Greater than 80% 1. 8% 1. 6% 1. 4% 1. 2% 1. 0% 0. 8% 0. 6% 0. 4% 0. 2% 0. 0% Default Rate The optimum range in terms of most favorable default experience is the 60 -70% 2. Default rates increase sharply when LTV is greater than 80% © IMa. CS 2012 Printed 15 -Mar-18 Page 14
Quantitative Indicators Age of the Borrower 40. 0% 35. 0% 30. 0% 25. 0% 20. 0% 15. 0% 10. 0% 5. 0% 0. 0% 1. 8% 1. 6% 1. 5% 1. 4% 1. 2% 19. 0% 1. 0% 0. 9% 0. 8% 0. 7% 0. 6% 6. 4% 0. 4% 3. 6% 0. 2% 0. 8% 0. 4% 0. 0% Less than 25 -30 30 -40 40 -50 50 -60 60 -70 Greater 25 than 70 34. 1% 35. 7% Relative Frequency The lower age bracket and the higher age brackets appear more 1. Default Rate The 40 -50 years age bracket seems to be the safest prone to default © IMa. CS 2012 Printed 15 -Mar-18 Page 15
Other important factors which should be considered for appraisal Credit Track Record Past credit record depicts the attitude of the person in honouring his credit obligation. “Wilful default” are one of the causes for a number of defaults. Nature of Asset In Housing Segment, assets gradually appreciate with time unlike many other assets [Cars, white goods, etc. ]. The chance of negative equity will be lesser and Loan to Value ratio will improve over the period of time. Collateral Security Additional collateral security lowers the net exposure of the bank. It increases the applicants contribution in the asset thus effectively reducing loan to value ratio If the Collateral Security is high, in case of default by the applicant, the Loss Given Default will be lower © IMa. CS 2012 Printed 15 -Mar-18 Page 16
Rating models – Does it really work ? q 6967 out of 9092 customers correctly classified -77% accuracy (73% accuracy within first 3 grades – refer blue color last column)
Cumulative NPAs Balance business flexibility with asset quality improvement Risk Grade 10 9 8 7 6 5 4 Cumulative Lending 3 2 1 The objective is to strike a balance between business objectives (so that not too many cases are rejected) and potential NPA reduction. © IMa. CS 2012 Printed 15 -Mar-18 Page 18
Formation of pools – cost effective way of managing risks Pool 1 Pool 2 Pool 3 Pool 4 Source of Income Salaried Self Employed LTV 50% - 75% >75% FOIR < 40% - 60% > 60% Expected Loss 0. 1% 0. 25% 0. 3% Interest Rate 11% 11. 10% 11. 15% 11. 45% © IMa. CS 2012 Printed 15 -Mar-18 Page 19
Risk based pricing of mortgage loans – USA 1. Interest rates are determined based on a number of factors like a. Loan Type, Loan Amount b. Property Type, Property Use, Property Location c. Credit Score and History – one of the most important factors d. Debt to Income Ratio e. Appraised Value/Purchase Price f. Loan to Value/Purchase Price g. Documentation Type Example: FICO score >=760 score can fetch 0. 375% rebate FICO score 680 -719 will have no fee/rebate FICO score 660 -679 will incur 0. 25% cost Illustration source: http: //www. thetruthaboutmortgage. com/mortgage-dictionary/risk-based-pricing-loan/ © IMa. CS 2012 Printed 15 -Mar-18 Page 20
Risk based pricing system in various countries USA/ Canada UK Australia Interest Rates are linked to credit scores and internal rating models Interest rates are linked to internal rating models FICO score is widely used in US Internal rating models and scorecards are used widely than external credit scores to calculate credit risk and interest rates Equifax's Score. Power and Trans. Union's credit score are popular in Canada Experian and Delphi scores Risk based pricing notice to are also referred to be given to consumers mandated by regulation External credit scores used to decide whether loan to be approved or not and set limits © IMa. CS 2012 Printed 15 -Mar-18 Page 21
Discussions
04cbe2206199d2381c37c6846018c0b6.ppt