1 Expected credit loss assessment by banks This article aims to: Present the key components of a probability of default-based approach for computation of ECL on term loans. With the implementation of Indian Accounting Standard (Ind AS) 109, Financial Instruments, financial institutions will move from a standardised and regulatory approach to an Expected Credit Loss (ECL) model for recognising an impairment allowance on their financial assets. This is expected to have a significant financial impact and would also entail extensive changes to systems and processes used for credit monitoring and analysis. Implementation of the ECL approach may require the application of sophisticated measurement techniques and estimates based on the nature of an institution s financial asset portfolio. In our previous article, we highlighted the concepts of significant increase in credit risk and default. Ind AS 109 requires the financial institutions to define these in the context of their loan portfolios. Other factors to be considered while computing ECL include the expected life of the exposure, stage allocation, forward looking information and discounting at an appropriate rate. In our previous ECL article, we also considered how banks may use an internal credit risk rating model to identify a significant increase in credit risk and perform a stagetransfer assessment for their loan assets. Ind AS 109 does not mandate any specific model to be adopted for ECL computation, however, financial institutions are required to adopt methodologies which are commensurate with the size, complexity, structure, economic significance and risk profile of their exposures. This will determine the level of sophistication the financial institutions need to incorporate in their ECL methodology. A model based on estimating the Probability of Default (PD) associated with loan exposures is a methodological framework for estimating ECL. This article aims to present the key components of a PD-based approach for computation of ECL on term loans.
Accounting and Auditing Update - Issue no. 15/2017 2 Key characteristics of loan Bank M has extended a term loan on 1 April 2017 to company P, a manufacturer of mobile phones in India, for importing machinery. The details of the loan are below: Details Amount of loan Rate of interest Period Repayment terms Effective Interest Rate (EIR) Collateral Marketability of collateral Other contractual terms Classification and measurement of loan Particulars INR100,000,000 10.5 per cent per annum 5 years The loan is repayable in five equal annual instalments on 31 March each year, commencing from 31 March 2018. 10.92 per cent Bank M holds a charge over the new machines imported by company P. These are valued at INR100,000,000 on 1 April 2017. On the basis of market data available on 31 March 2018, the bank would be able to recover approximately 80 per cent of the book value of the asset (which represents their economic value). Company P depreciates the machine on a straight line basis over a period of five years. There are a number of companies manufacturing phones using similar technology, hence Bank M considers that it would be able to sell the machine within a short period of time, if necessary, in order to recover the collateral value. The company does not have an option to extend the period of the loan, however, it may prepay the loan after repaying the second instalment on the loan. The bank does not have an option to demand payment of the loan before its due date. The bank expects to hold the loan until maturity. Further the contractual cash flows arising from the loan are solely in the nature of principal and interest on principal outstanding (SPPI). Hence, the bank has classified and subsequently measures the loan at amortised cost. An analysis of companies in the manufacturing sector with similar risk characteristics as company P does not indicate a pattern of prepayment for their mediumterm borrowings. Generally, these companies are expected to repay their borrowings over the contractual term of the loans. As per its internal credit risk rating system, which considers various factors including the credit risk of the borrower and other macroeconomic indicators, on initial recognition, bank M has computed the forward lifetime PD of the loan as 1.5 per cent. Bank M further expects the loans to have a low cure rate and that the amount of default, if any, would be recovered through the sale of the underlying collateral. On 31 March 2018, company P has made timely payment of the interest and principal instalment due on the loan. The forward lifetime PD of the loan is computed as 1.8 per cent (considering current economic conditions in the industry, and their expected future impact on the company s performance, bank M has determined a marginal increase in the forward lifetime PD of the loan) and bank M determines that there has been no significant increase in credit risk since initial recognition. Bank M has assessed that a PDbased approach to measurement of ECL would meet the requirements of Ind AS 109.
3 Accounting issue Ind AS 109 requires ECL to be computed as a probability weighted estimate of credit losses over the expected life of a financial instrument. It, however, does not prescribe a single approach or method to measure ECL. Therefore, the bank s management should carefully analyse the risks and characteristics of its financial assets and adopt suitable methods/ models for ECL computation based on the availability of reasonable and supportable information without undue cost and effort. This would require the exercise of significant judgement and adds new responsibility, considering the high estimation uncertainty surrounding the computation of impairment allowances. In this article, we consider how bank M defines the parameters for computing ECL using a PD-based model. We also highlight how the bank should incorporate the impact forward looking information on impairment allowance. Accounting guidance Figure 1: ECL computed using a PD-based model Default Broader definition as compared to regulatory literature Definition should be consistent with that used for internal credit risk management Need to be assessed based on quantitaitve, qualitative and backstop indicators Aligned with definition of credit-impaired PD EAD LGD Discount rate Expected Credit Loss (ECL) Estimate of the likelihood of default over a given time horizon. Estimate of exposure at a future default date, taking into account expected changes in exposure after reporting date. Difference between contractual and expected cash flows, including from collateral, expressed as a percentage of EAD Generally the Effective Interest Rate (EIR) applicable to the financial asset Forward looking information Macroeconomic assumptions and forecasts Emerging issues and uncertain future events Should neither be optimistic or pessimistic (Source: KPMG in India s analysis, 2017) Analysis The PD-based model estimates ECL based on three principal parameters, the PD, Exposure at Default (EAD) and Loss Given Default (LGD). Other factors to be considered when assessing and using these components to calculate ECL are: The term structure and other contractual features of the asset The stage allocation, i.e. whether there is a significant increase in credit risk Forward looking information (including macro-economic factors); and Discounting based on the EIR. These parameters are analysed below. Default In order to quantify parameters such as PD or LGD, a bank is required to define or establish a policy to identify default. This is also relevant in staging the financial asset, i.e. to determine whether there has been a significant increase in credit risk associated with that asset. Ind AS 109 specifies certain backstop indicators stating that a default is considered to have occurred when there has been a delay of over 90 days in repayment. In this case study, bank M also considers other qualitative factors to define what would be considered as a default on the loan to company P.
Accounting and Auditing Update - Issue no. 15/2017 4 Period of exposure and probability of default A. Period of exposure Prior to determining the PD, it is important to assess the period over which it will be determined. Generally, the period of exposure is based on the contractual terms of the financial asset, which include the borrower/lender s ability to extend/prepay the loan. In this case, company P does not have the option to extend the loan, and bank M cannot demand repayment of the loan prior to the due date. However, company P may prepay the loan anytime after payment of the second instalment, on prior intimation to the bank. In this case, the period of exposure may be calculated on the basis of historical behavioural information of company P, adjusted for forward looking information. Where historical behavioural information of the specific borrower is not available, the bank may consider behavioural information relating to entities of similar size and characteristics, operating in Figure 2: Lifetime PD term structure 1 the same sector. The bank has analysed companies operating in the manufacturing sector, with similar risks and characteristics as company P. In addition, the bank considers other macroeconomic, forwardlooking information, for example, the probability of a reduction in future interest rates, which may cause company P to prepay the loan. Based on its analysis, the bank assesses that the company is unlikely to prepay its term loan. Accordingly, it determines the period of exposure to be five years from the date of inception of the loan (i.e. until 31 March 2022). B. Probability of default PD is an unbiased estimate made by a lender on the likelihood of the loan not being repaid by the borrower within a particular time period. It is computed on the basis of the bank s assessment of the credit history of the borrower and the nature of the loan, adjusted for forward looking information. PD is used in both, calculating ECLs and assessing whether there has been a significant increase in credit risk (for a staging assessment). For the purpose of computing ECLs, a PD term structure is required to be developed. A PD term structure shows the estimated PD of a borrower at each point in time for the lifetime of the exposure. When estimating ECL, banks would be required to consider a 12-month PD term structure for loan assets in stage 1, and a lifetime PD term structure for assets in stage 2 or 3. In the current case, the lifetime PD of the loan is estimated by the bank as 1.5 per cent on initial recognition and 1.8 per cent on the reporting date (31 March 2018). This represents the estimated probability of a default occurring over the remaining life of the instrument computed at each point in time- i.e. on initial recognition and on the reporting date. A PD term structure developed on initial recognition should be adjusted for current conditions and future macroeconomic factors at each reporting date. It is represented by the chart below: 2.5 PD 2 1.8 1.9 Probability of default 1.5 1.5 1 1.4 1.2 PD Linear (PD) 0.5 0 0 April-17 March-18 March-19 March-20 March-21 March-22 Time horizon (Source: KPMG in India s analysis, 2017) 1. The above chart represents a lifetime PD term structure for the loan, however, since the loan is in stage 1, a 12-month PD term structure will be required to be developed. A 12-month PD is a part of the lifetime PD, hence, bank M should extract a 12-month PD term structure from the above chart.
5 In the current case, on the reporting date, there is no significant increase in the credit risk, and no adjustments are required to the PD term structure. The loan asset is assessed as stage 1 and the bank would be required to determine the risk of default occurring in the next 12 months (at each reporting date) based on the 12-month PD structure. Exposure at Default (EAD) EAD is a point-in-time measure of the loan exposure after considering contractual and/or behavioural cash flows in the given time horizon. The time horizon would depend upon the stage of the loan. In the given case, since the loan asset is in stage 1, 12-month ECL is required to be computed. For this, the EAD as on 31 March 2019 (i.e. 12-months from the reporting date) is computed on considering the principal and interest repayments during the period. The bank estimates that company P would make timely repayments (on the basis of its previous track record), of principal and interest thereon as on 31 March 2019. Hence, the EAD as on that date is estimated as INR60,000,000. Loss Given Default LGD is an entity s estimate of the present value of the loss/cash shortfalls expected as a result of a default, as on the date of default. While calculating LGD, an entity needs to identify all components of collections/recovery of cash flows and project their values and the timing of their receipt. The LGD is usually expressed as a percentage of the EAD. In case of secured loans, the amount of cash flows that are expected from foreclosure are cash flows that the entity actually expects to receive in the future. Timing of these cash flows is critical to the computation of LGD. Entities may need to refer to a historical trend and adjust this for current and future economic conditions to make an unbiased computation of LGD. In the current case study, the cure rate of the loan is estimated as nil by the bank. However, in case of default, the bank would be able to dispose off the collateral and recover approximately 80 per cent of its book value immediately. On 31 March 2019, the EAD of the loan and book value of the machine are both estimated at INR60,000,000. Hence, the LGD as on that date would be 20 per cent of the EAD. Discount rate The measurement of ECL should also reflect the time value of money. Hence, cash shortfalls associated with default should be discounted to the reporting date. An entity should maintain consistency between the rate used to recognise interest revenue and project future cash flows and the rate used to discount those cash flows. Generally, this is the EIR of the financial instrument or an approximation thereof. In the current case study, the EIR of the loan is 10.92 per cent, which is also considered as the discount rate to estimate ECL at the reporting date. Forward looking information The measurement of ECL reflects an unbiased and probability-weighted amount based on a range of possible outcomes reflecting an entity s own expectations and incorporates forecasts of future economic conditions, including observable market information. In arriving at the estimate of ECL, an entity is required to use all reasonable and supportable information available at the reporting date, without undue cost or effortabout past events, current conditions and forecasts of future economic conditions. Entities may incorporate forward looking information in computing their impairment allowance by establishing a relationship between the macroeconomic forecasts and ECL (for example, an increase in Gross Domestic Product (GDP) of the country may result in a reduction in ECL). In doing so, an entity could develop a base scenario, i.e. a most likely path for a set of macroeconomic variables that together form a forward-looking view of the economy - for example, a base case estimate of GDP at 6 per cent 2 would result in no change to the estimated PD at initial recognition of the loan. Multiple scenarios could then be projected based on positive and negative deviations from the base scenario, and their impact on ECL or a component of ECL estimated. For example, a positive scenario may be that a 0.5 per cent increase in GDP would reduce the PD by 0.25 per cent. Conversely, a negative scenario may be that a 0.5 per cent reduction in GDP would increase the PD by 0.75 per cent. ECL may then be estimated on the basis of a weighted average of these scenarios. An entity should exercise judgement to determine what constitutes reasonable and supportable forwardlooking information that is relevant to the financial instruments to which the impairment requirements are applied. In the current case, bank M needs to incorporate forward looking information from a variety of sources into various parameters of the model in order to reasonably estimate its expected loss on this loan. Once the bank has estimated the various parameters described above, it should estimate the ECL on the basis of the formula: ECL = PD*EAD*LGD 2. for illustrative purpose only
Accounting and Auditing Update - Issue no. 15/2017 6 Consider this While there are various methods that can be used for computing ECL, banks should adopt the method specific to their requirement based on availability of data, sophistication of credit risk management systems and characteristics of the loans in their portfolio. Banks may consider aligning the definition of default with the regulatory definition, so that there is no conflict between the two. They should also ensure that there is consistency of information used for Ind AS 109 purposes and that used for organisational budgeting, risk management and regulatory purposes. Generally, for regulatory purposes the PD is computed through-the-cycle, i.e. in cycle neutral economic conditions. However, the PD required for Ind AS 109 purposes is at a point-in-time (i.e. PD in current economic conditions). Hence, when using PDs calculated for regulatory purposes as a starting point, banks should make appropriate adjustments in order to be in compliance with Ind AS 109.