ISSN (Print) - 0012-9976 | ISSN (Online) - 2349-8846

A+| A| A-

Estimating Loss Given Default for Indian Markets

The Basel II norms allow banks to calculate credit risk capital requirements by the standardised or internal ratings based approach. Banks using the latter need to develop methods to estimate its key components, one of which is the loss given default. This article discusses some of the issues in the estimation of the loss given default for corporate exposures in the Indian context.

COMMENTARYEconomic & Political Weekly EPW july 19, 200827Estimating Loss Given Default for Indian MarketsSoumya Kanti Ghosh, Sanjeev NewarThe Basel II norms allow banks to calculate credit risk capital requirements by the standardised or internal ratings based approach. Banks using the latter need to develop methods to estimate its key components, one of which is the loss given default. This article discusses some of the issues in the estimation of the loss given default for corporate exposures in the Indian context.The new Basel accord, expected to be operationalised by Indian banks from March 31, 2009 will allow a bank to calculate credit risk capital requirements (alternatively, the required capital adequacy ratio/Pillar 1 of the new Basel capital accord or “BaselII”) accord-ing to either of two approaches: A stand-ardised approach, which uses agency ratings (Credit Rating Information Services of India,ICRA, Credit Analysis and Research, Fifth Ratings are the key rating agencies in India) for risk-weighting assets andan internal ratings based (IRB) approach, which allows a bank to use internal estimates of components of credit risk to calculate credit risk capital. Indian banks using IRB need to develop methods to estimate key components, one of which is loss given default (LGD), the credit loss incurred net of recovery if an obligor of the bank defaults. Given that, there is a significant divergence on estimatingLGD, even in developed countries (not to men-tion developing ones), this paper discusses some of the key issues inLGD estimation in the Indian context, primarily for corporate exposures. Standardised ApproachThe crux of the Basel II accord in modelling credit risk is classifying the credit risk exposure of each lending activity in terms of appropriate risk weights. These risk weights are either predefined (as in the standardised approach) or are estimated separately for each exposure (as inIRB approach) through appropriate mathematical and statistical methods. Basel II follows a three-step paradigm on identifying credit risk capital require-ment of a bank. Banks can choose from among the following approaches to esti-mate credit risk in their portfolios: (1) Standardised approach; (2) foundation internal-rating based model; and Advanced Internal-Rating Based Model (AIRB).The standardised approach follows the same methodology as BaselI in estimating the capital requirement but makes it more risk sensitive by dividing the hitherto com-mercial borrowers (as existing in Basel I) into different categories of obligors: cor-porate, sovereign, bank, retail, real estate, and specialised lending. Each of these obligors is further subdivided into finer grades of risk classifications/five risk buckets as assessed by an external credit rating agency (ECAI). The procedure for obtaining the minimum capital require-ment for credit risk of a bank is to sum up the exposures at default (EAD) in each risk weight bucket, weighted appropriately by the relevant risk weights (Table 1) and then multiplied by the overall total capital requirement of 8 per cent. Specifically,K = EAD × RW × 0.08 …(1)where K =minimum capital requirement, EAD = exposure at default of the assets and RW = appropriate risk weights (for instance, it is set equal to 35 per cent for residential mortgages).It may be noted that under the standar-dised approach, EAD for on-balance sheet items is the nominal outstanding amount. However, in the case of off-balance sheet activities like credit derivatives and guar-antees, the EAD is adjusted through appro-priate weights to arrive at a credit equiva-lent of the contingent exposure. In the case of forwards, swaps, pur-chased options and similar derivative con-tracts, banks are not exposed to credit risk for the full face value of their contracts but only to the potential cost of replacing the cash flow (on contracts showing positive value) if the counterparty defaults. The credit equivalent amounts will depend inter alia on the (residual) maturity of the contract and on the volatility of the rates and prices underlying that type of instrument. Table 1 (p 28) provides a summary of the risk weights for different class of obligors under the standardised approach in Basel I. For the sake of convenience, we have also provided the risk weights (if any) under Basel 1 to enable a comparison.The following may be noted while allowing for risk weights under different obligors: first, regarding small and medium enterprises, in BaselII, loans extended to Soumya Kanti Ghosh ( Sanjeev Newar ( are at a leading multinational corporation present in the risk management domain. These are the personal views of the authors.
COMMENTARYjuly 19, 2008 EPW Economic & Political Weekly28small businesses and managed as retail exposures are eligible for retail treatment provided the total exposure of the bank-ing group to a small business borrower (on a consolidated basis where applicable) is less than ¤ 1 million (about Rs 6.3 crore). Second, for sovereigns, claims on multi-lateral development banks will have 0 per cent risk weight. Third, for banks, there are two options and the choice will depend on the national bank regulators. Although the standardised approach makes the banks more risk sensitive, an issue might encourage perverse behaviour in terms of risk assessment. As Table 1 shows, BaselII has recommended a lower risk weight to be applied to unrated borrowers (100 per cent) than to those rated under, say,BB- (150 per cent). This may discourage many borrowers from being rated (or obtaining a rating), as well as lower-rated borrowers from ac-cepting their ratings, if they are not rated in the investment grade. However, as a counter-measure against such possible fallout, BaselII also states that national supervisors in respective countries may consider a higher risk weight on unrated claims on corporates, if so warranted in their jurisdictions. Loss Given Cost ConceptIn the IRB approach, each bank is required to establish an internal ratings model to classify the credit risk exposure of each acti-vity (as defined in Table 1), whether it is on/off-balance sheet. The IRB approach may be the FIRB or the AIRB and is based on four key parameters used to estimate the expected loss (EL). These four key para-meters are: (i) the probability of default of a borrower over a one-year horizon (PD); (ii) the loss given default/credit loss (or 1 minus recovery) as a percentage of expo-sure at default (LGD); (iii) exposure at default (an amount, not a percentage) (EAD); and (iv) maturity (M). For a given maturity, these parameters are used to es-timate two types of EL: (1) expected loss as an amount:EL = PD*LGD*EAD; and (2) as a percentage of exposure at default: EL = PD*LGD*EAD.The principal steps to be followed by banks to estimate the regulatory capital under the FIRB approach are as follows: first, estimate the one-year probability of default (PD) for every credit transaction. The estimation of PD would be the output of an appropriately calibrated risk-rating model, which needs to be validated with historical/empirical default experience. The design and structure of models is a major exercise that calls for considerable skill and expertise in credit risk evaluation, model building, mathematical techniques and statistical testing. Basel II stipulates a lower bound on PD that is equal to 0.03 per cent, so as to create a non-zero floor on the credit risk weights (and therefore some amount of capital required to be held Table 1: Total Capital Requirements under Different Obligors in Basel II and Basel I (in %)External Credit Ratings AAA to A+ to A- BBB+ Below Unrated AA- to BB-BB-CorporatesRisk weight under Basel I 100Capital requirement under Basel I 8 8 8 8 8Risk weight under Basel II 20 50 100 150 100Capital requirement under Basel II 1.6 4 8 12 8 Sovereign AAA to A+ to A- BBB+ to BB+ to B- Below B- Unrated AA- or ECA or ECA BBB- or or ECA or ECA Rating 1 Rating 2 ECA Rating 4 Rating 7 Rating 3 to 6 Risk weight under Basel I 100Risk weight under Basel II 0 20 50 100 150 100Capital requirement under Basel II 0 1.6 4 8 12 8Banks AAA to A+ to A- BBB+ BB+ to B- Below B- Unrated AA- to BBB- Risk weight under Basel I 100 100 100 100 100 100Risk weight under Basel II – Option 1 20 50 100 100 150 100Capital requirement under Basel II – Option 1 1.6 4 8 8 12 8Risk weight under Basel II – Option 2 20 50 50 100 150 50Capital requirement under Basel II – Option 2 1.6 4 4 8 12 4Retail ¾ Individuals or Small Business, Default ¾ Revolving credit, line of credit, personal loan, lease, or small business facilityRisk weight under Basel I 100Risk weight under Basel II 75 From 150 to 50 per cent depending on level of specificprovisions againstexposureCapital requirement under Basel II 6 4 to 12Real Estate Residential CommercialRisk weight under Basel I 50 100Capital requirement under Basel I 4 8Risk weight under Basel II 35 100 (subject to certain conditions: the risk (reduced risk weight of weights may also be increased) 50 per cent possible under certain conditions)Capital requirement under Basel II 2.7 8 Specialised lending/ ResidentialProject financeObject financeCommodities financeIncome-producing real estateHigh volatility commercial real estateRisk weight under Basel I 100Risk weight under Basel II 100Capital requirement under Basel II & Basel I 8Equity CorporatesBanksSecuritiesFirmsRisk weight under Basel I 100 100 100Risk weight under Basel II 100 100 100Capital requirement under Basel II & Basel I 8 8 8Source: Basel Committee on Banking Supervision (2006).
COMMENTARYEconomic & Political Weekly EPW july 19, 200829against any individual loan, even if it happens to be of the best credit quality).Second, the EAD for on-balance sheet transactions is equal to the outstanding value of the exposure at the instant of default. The EAD is computed from the book value of an outstanding amount of a loan by applying credit mitigation factors (such as haircut – adjusted collateral, based on haircuts as defined by supervisory advice under pillarII). EAD for sovereigns is defined as “not less than sum of (a) the amount by which bank capital would reduce if exposure is written-off fully; and (b) specific provisions and partial write-offs”. For example, where a facility com-prises both a drawn amount and an undrawn amount, EAD will be calculated as (100 per cent of) the drawn amount plus 75 per cent of the undrawn balance. Thus, for a committed line of 100, with current outstandings of 60 per cent, the EAD would equal 60 + 75 per cent (100-60) or 90.Third, the regulator under the FIRB ap-proach predefines LGD. For instance, it could be set equal to 40 per cent for a loan fully secured by receivables, 45 per cent for a loan fully secured by physical, non-real-estate collateral, 50 per cent for an unsecured loan and 75 per cent for a sub-ordinated loan. The weighted average life (M) is prede-fined in the case of theFIRB approach. It is set equal to 2.5 years for loans (six months for repo transactions). Maturity is a key factor affecting the credit risk of a bond or loan. Other things being the same, the shorter the maturity of a loan, the less its underlying credit risk. The FIRB approach also assumes there exists an average default correlation (R) among individual borro-wers, which is calibrated to lie between 10 and 20 per cent, with R being a declining function of PD. Given the values for PD, EAD, LGD, R, and M, the corresponding risk weights are estimated. The internal ratings model is based on the value-at-risk (VaR) princi-ple, wherein the objective is to estimate the loss in the value of the credit portfolio over a given time period (say one year) at a given confidence level (say 99 per cent). In BaselII, capital charges (K) are designed to satisfy a portfolio-level sol-vency target (L), such that Pr [L≤K] ≥ 99.9 per cent. In other words, in 99.9 per cent of the cases the capital charges will exceed L. TheAIRB approachdiffers from theFIRB Approachin two primary ways:one, the bank uses its own LGD values (to be vali-dated by empirical evidence) in place of a regulator stipulatedLGD. This paper, as discussed earlier, simulates actualLGD values in the Indian context.Two, another adjustment to the AIRB approach is the incorporation of a maturity adjustment, which is the weighted aver-age life (ΣttPt /Σt Pt, where Pt is the mini-mum amount of principal contractually payable at time t) for all instruments. The maturity is capped at seven years to avoid overstating the impact of maturity on credit risk exposure. Key Issues in LGD EstimationLGD estimates are extremely difficult to obtain in the Indian context. There are no reliable studies in the recent past in this direction. So we did an anonymous survey among a fair number of corporates to find out the relevant de-tails (interestingly, this anonymity was neces-sary to obtain informa-tion that may not have been forthcoming in the event the identity was disclosed).This apart, the estimation of recovery information for Indian markets posed sev-eral problems. One, often the debt is re-structured and the payment schedule and rate of return are altered. Also, in several cases, debt obligations are restructured into equity. Further, often recovery hap-pens in one-to-one negotiations between debtor and debtee. Thus, information about actual recovery terms and condi-tions was very difficult to obtain. Estimating Recovery Rates in Devel-oped Countries: There are many practi-cal problems in estimating recovery rates of debt in the event of default in developed countries. Often there is no market from which to observe objective valuations and if their market prices are available, they will necessarily be within a highly illiquid market. Even if these issues are resolved there is the question of whether it is best to estimate values: (a) immediately upon announcement of default; or (b) after some reasonable period for information to become available/after a full settlement has been reached. In the Indian context, it may even take at least 10 years. For corpo-rate bonds, there are only two primary studies of recovery rates that arrive at similar estimates in theUS market. As Table 2 shows, the subordinated classes are appreciably different from one another in their recovery realisations. In contrast, the difference between secured and unsecured is not statistically significant. It is likely that there is a self-selection effect here, since there is a greater chance of recovery to be requested in the cases where an underlying firm has question-able hard assets from which to salvage value in the event of default.On an average, long-run average recov-ery rates for US corporate issues appear to be fairly stable at about 40 per cent. At the same time, recovery rates have exhibited high variability over shorter time horizons – post-default recoveries have been as high as 72 per cent in 1981 and as low as 21.7 per cent and 24.7 per cent in 1980 and 1990 respectively. Estimating Recovery Rates in Develop-ing Countries: Accurate and detailed his-torical data on post-default recoveries on loans and bonds for issuers in emerging market regions generally are sparse or non-existent. In general, however, recovery rates for defaulted emerging markets assets are much lower than those for US obligors (at around 15 per cent).In the context of emerging markets, ex-pected recovery rates depend on a number of key variables. Among those, one needs to consider: the likelihood of debt restruc-turing as opposed to outright repudiation of debt obligations; terms of restructured Table 2: Recovery Statistics by Seniority Class Carty and Lieberman (1996) Altman and Kishore (1996)Seniority Class Number Average Std Dev Number Average Std DevSenior secured 115 $ 53.8 $26.86 85 $57.89 $22.99Senior unsecured 278 $51.63 $25.45 221 $47.65 $26.71Senior subordinated 196 $38.52 $23.81 177 $34.38 $25.08Subordinated 226$32.74$20.18214$31.34$22.42Junior subordinated 9 $17.09 $10.90 - - -Par (face value) is $ 100.Source: Carty and Lieberman (1996) and Altman and Kishore (1996).

Dear Reader,

To continue reading, become a subscriber.

Explore our attractive subscription offers.

Click here

Back to Top