The new standard on financial instruments accounting – IFRS 9 has significantly transformed banks’existing impairment assessment to address concerns about “too little, too late” provisioning for loan losses.In our previous blog i.e. “Stage Assessment – Devil is in the detail” we discussed how lifetime PDs are used for stage assessment of instruments, apart from other nuances of assessing significant deterioration of credit quality. Entities are required to recognise an allowance for either 12-month or lifetime Expected Credit Losses (ECLs), depending on whether there has been a significant increase in credit risk since initial recognition. This significant increase in credit risk needs to be measured ideally at instrument level.
However, in some cases, especially for retail assets, this determination might be difficult to achieve and this assessment may be done at a collective level, wherein the deterioration in asset quality needs to be checked for at a portfolio level or a sub-portfolio level. It would be pertinent to note that this is relevant for Stage 2. For Stage 3 classification, since it is nearly the same as incurred loss approach, this determination would continue to be on individual assets. Consequently, for collective assessment, where it has been determined that there has been a deterioration in asset quality (on a forward looking basis), the lifetime ECLs need to be computed for that section of the portfolio. Banks may use segmentation schemes to identify slivers of portfolio that are deemed to have deteriorated, and examples of such segmentation schema include instrument type, collateral types, risk scores etc.
The measurement of lifetime expected loss can be done by either a Loss rate method or an approach that uses Probability of Default (PD) and Loss Given Default (LGD) estimates. The loss rate approach looks at historical losses suffered and uses the same to estimate forward looking losses. While the loss rate approach looks deceptively easy to implement, and at first glance it seems as if lifetime PDs and LGDs may not be needed for this approach; the truth is that while for ECL computation PDs may not be needed, to determine significant deterioration in credit quality, lifetime PDs are still very useful. The article walks through a few numerical examples of methodologies using the Loss rate approaches as well as the PD-LGD approach.
Before we talk about different modeling methodologies in loss forecasting, it is important to understand how the loss ratio is calculated. The loss rate can either be calculated at an instrument level or at a cohort level.
The table below represents historical balance, charge-offs and recovery information of a portfolio across different months. Most institutions use either the traditional historical loss rate analysis or the migration analysis for the actual loss rate calculation.
There are several modeling practices for loss forecasting across industries. Let’s discuss some of the wellknown forecasting techniques, Such as Roll rate models, Vintage loss models, Provision Matrix method, Expected loss models and Discounted Cash-flow method. Roll Rate Models, Vintage Loss Models as well as Provision Matrix method directly predict loss amount / loss ratio, whereas the Expected Loss Models predict losses by using PD as one of the components of loss prediction. The Discounted Cash-flow method predicts loss amounts as well, but it is ideal for individual assessment of losses.
Gross losses can be calculated by applying these Loss rates on the outstanding balances, which gives us a figure of $240 ($100,000 * 0.24%). To calculate Net provisions, one needs to take into account Recovery rates, Accounts that go into loss status (180+ dpd), that can be partly (or fully) recovered in the future as an outcome of internal collections or debt sales. Banks predominantly calculate and forecast recovery rates using Recovery curves, representing recoveries post charge-off (180+ in our example) across different charge-off vintages. Let us assume the recovery rate is 50%, in which case the Gross provision is $120 ($240 * 50%).
Expected Loss Models:
Significantly different from roll rate models and vintage loss curves, Expected Loss (EL) estimation is a modern modeling practice, in line with BASEL framework, developed on the basis of 3 risk parameters, namely Probability of Default (PD), Exposure At Default (EAD), and Loss Given Default (LGD) by incorporating loan-specific characteristics.
For both wholesale and retail portfolios, each of the three risk factors are modeled separately to capture the account / cohort specific behavior. Survival model can also be used for directly predicting “Time to Default” for a loan. The customers are segmented across homogeneous groups based on origination variables, such that the historical bad rates across time for each of the groups never intersect each other. For each of these groups, average PD, LGD and EAD is used to calculate Expected Loss.
Irrespective of the choice of methodologies, according to IFRS 9, the estimation of losses should incorporate not only past due information but also all relevant current as well as future credit information,
including forward-looking macroeconomic outlook. The Expected Loss methodology, which involves separate estimates of PD and LGD is perhaps the best and one of the most popular choices, the
advantages of the approach being:
That being said, it does have requirements of large amounts of data, not to mention validation and other model governance/ maintenance requirements of the upstream PD and LGD models. IFRS 9, being a
principle based guideline, does not prescribe specific methodologies for lifetime expected loss estimation, and there is no single methodology that can suit all portfolios. The choice of methodology should be an
informed one based on the availability of data as well as materiality of the portfolio, and such decisions should ideally be documented in a ‘Target Operating Model’ design document. In our next blog, we will
discuss the Target Operating model and the role of the statutory auditor in the same.
Don't miss this roundup of our newest and most distinctive insights
Subscribe to our insights to get them delivered directly to your inbox