Implementation of the expected credit loss model for receivables
Implementation of the expected credit loss model
Case study for IFRS 9
Implementation of the expected loss model according to IFRS 9 is a challenge for many companies. In an example of application for the simplified approach to trade receivables, we show how implementation might look in practice and which strategies are advisable for automation.
The new impairment model under IFRS 9 provides for allowances for expected credit losses, marking a shift away from the previous approach based on incurred losses. Financial reporting thus moves closer to forward-looking credit risk management and means that a model is required to measure credit risks for all financial assets not measured at fair value.
In today's article, we focus on the implementation of the simplified approach, which is used for items such as trade receivables and contract assets according to IFRS 15. We have already described the basic general conditions in an earlier newsletter.
IFRS 9 does not provide any specifications on the design of the model. In practice, there are two main approaches to determine ECLs (expected credit losses):
- Allowance matrix based on an entity's internal, historical credit loss data and past due receivables
- Valuation model that uses probabilities of default
Common to both approaches is that they are probability-weighted and must be adjusted for forward-looking, macroeconomic factors ('forward-looking information'). In many cases, the entity's historical data is not representative, because due to economic development or the business model, the data shows only small credit losses. However, IFRS 9 does not allow pure extrapolation from past results; this means that even customers with a good credit rating have a certain percentage-amount probability of default. In other cases, the accounting data available in the financial accounting system (ERP) does not allow granular evaluation of historical default data. For these reasons, it is advisable to use a valuation model that determines the ECLs based on probabilities of default that are applied to the receivables portfolio.
We will now use an example of application to look at how this model can be implemented in three steps.
Example application impairment model
1. Design of model requirements
First, the required input parameters must be defined and the corresponding availability of the data examined. Other information sources apart from the ERP are risk management systems, e.g. in receivables management, where databases typically exist. The key information for the model is:
- Carrying amounts of trade receivables and contract assets;
- Contractual terms;
- Customer names and addresses;
- Ratings or scorings;
- Probability of default (PD).
Rating/scoring information and probability of default for the customer base is often not available, especially at international, heterogeneous entities without ongoing risk management. Scoring providers, rating agencies and credit insurers can help out here, as they are already prepared to deliver this data custom-fit for credit risk purposes.
It's often worth casting a glance to the treasury's existing market data information system, which for many listed companies offers ratings and probabilities of default at industry level, which can at least be used for an initial quantification or for an individual analysis of major customers.
A structured approach is also advisable because the entity must use comments in the notes to explain the input data, assumptions and methods individually used to determine the loss allowances.
2. Specification and data collection
The second step focuses on how data is actually collected how it is integrated into IT systems. The basis for measurement is initially the risk exposure, in this case the carrying amount of the receivables (exposure at default, EAD). Risk-minimising security measures such as credit insurance or Hermes guarantees can either be used to directly reduce the exposure or integrated via a weighting factor at the end (see below). It is important to bear in mind any deductibles and other conditions that result in residual risks remaining at the entity.
The contractual term is a key factor in the measurement of the risk exposure. The empirically provable principle applies: the longer the term, the higher the risk of default. Inversely, the risk is reduced accordingly for corporates that typically have short payment terms. The payment terms for each customer or remaining terms of receivables recognised and the payment plans for contract assets are often not available quickly when needed and, since they are an integral part of the model, should be collected at an early stage. This can be done on a standardised basis through the ERP system or via a request through one of the local entity's reporting packages.
For entities with a heterogeneous customer basis or small customers, IFRS 9 allows individual risk portfolios to be created, called clusters. Based on homogeneous risk characteristics such as region, industry, size or historical payment behaviour, homogeneous risk clusters can be formed which can then be analysed as whole for measurement. This grouping also allows the data volume to be reduced. For reasons of materiality, a partly generalised analysis may be possible.
The data stated thus far has been exclusively of an internal nature; now, the important final step in our example uses data external to the company: customer ratings/scorings, and (in particular) the probability of default as a percentage. This PD must be allocated to the respective customer or risk cluster in a customised way, and calibrated to the corresponding term. Furthermore, a confirmation should be obtained from the data provider stating which factors (especially forward looking information) have been taken into account to determine the figures, as this must be disclosed in the notes. The classification of risks is also important for disclosures in the notes according to IFRS 7.
3. Implementation and recognition
If the internal company data on customers, receivables details and collateral has been collected, grouped into suitable risk clusters (if applicable) and supplemented with external data on ratings and probabilities of default, all information is in place to measure the expected losses.
To do this, the formula typically used in practice is:
ECL = EAD * PD * LGD
[Expected credit losses = exposure at default * probability of default * loss given default]
LGD (loss given default) denotes the share of losses, i.e. the actual receivables loss in the event of customer default, or what is expected to be irrecoverable from among the assets in insolvency proceedings.
An entity has an unsecured receivable of EUR 100 million owed by a customer with a remaining term of one year, a one-year probability of default of 1% and a loss given default of 50%. This results in expected credit losses of EUR 0.5 million (ECL = 100 * 1% * 0.5). For reasons of materiality, discounting is disregarded in this example.
The expected credit losses are recorded in profit or loss on initial recognition in an allowance account for the respective item in the statement of financial position and updated at every reporting date. On top of the ECLs, specific allowances will continue to be recognised if certain 'loss events' have occurred, as was the case under IAS 39.
Integration into processes and systems
In order that the determination of impairment according to IFRS 9 doesn't remain a mere theoretical tract, the optimal way of integrating the model into the accounting-relevant processes and IT systems should be examined.
Depending on the ERP landscape and the group structure, possible options for the entry process are top adjustments in the consolidated statement of financial position and a 'push-down' into the local ERP systems. Process efficiency and risk of error must be gauged when these options are considered. For example, in practice, initial data collection and modelling occurs in group accounting, which uses standardised reports to centrally import the data for the loss allowances into the ERP or the reporting system. Local entities can access the specified data through these systems and, using a consistent logic, apply them to the local balances and make corresponding entries.
Challenges for automation arise in practice especially for data that has not previously been utilised for financial statement purposes, for example customer details or contractual terms. To avoid additional manual queries and resource-consuming data mapping, the goal should be to uniformly integrate in the ERP or in the reporting packages all data that must be regularly collected for financial statements and notes. Implementation should involve early coordination with the auditor, not only regarding the methodology, but also as regards integrating the data into the accounting-relevant processes, IT systems and the internal control system.
Depending on data availability and interfaces to external providers, fully automated impairment solutions can now already be implemented, ensuring high data quality and efficient processes.
Source: KPMG Corporate Treasury News, Edition 81, June 2018
Author: Christian Pfeiffer, Manager, Finance Advisory, email@example.com
© 2021 KPMG AG Wirtschaftsprüfungsgesellschaft, eine Aktiengesellschaft nach deutschem Recht und ein Mitglied der globalen KPMG-Organisation unabhängiger Mitgliedsfirmen, die KPMG International Limited, einer Private English Company Limited by Guarantee, angeschlossen sind. Alle Rechte vorbehalten. Für weitere Einzelheiten über die Struktur der globalen Organisation von KPMG besuchen Sie bitte https://home.kpmg/governance.