Time is Running Out for IFRS 9 – Are You Ready?

You can't outrun bad credit! – is a well-known saying among veterans of the equipment finance industry. As a result, some veterans treat rising credit risk like a coal miner reacts to gas leaks. But not all lenders are run by seasoned veterans. By taking no account of the risks in the performing portfolio, reserves proved to be quite inadequate when economic conditions caused a sharp rise in credit problems. The ALLL method, also called the INCURRED LOSS MODEL, focused on what had already gone wrong and whether enough had been set aside for that spilled milk when quite often the bigger risks were left unreserved. “Under the incurred loss rules, banks are overstating profits up front and not making prudent provisions against expected losses," states Iain Richards, Threadneedle Investments. New regulations from the International Accounting Standards Board (IASB) are meant to help lenders, and those that invest in them, to better understand the mystery of credit risk. IASB calls this new regulation International Financial Reporting Standard 9 (IFRS 9). The bad news is that provision levels are estimated to increase up to 50% under the new standard. A 50% rise in loss provisions results from the first year of implementation when reserves will have to be taken on the Expected Credit Loss (“ECL") within the entire portfolio. Below, I'll show that IFRS 9 is ultimately a better way to measure loss provisions, and how equipment finance companies can make sure that the inevitable increase in loss provisions is a realistic and supportable amount rather than a conservative and potentially inappropriate estimate.

Capital Punishment?

Let's take the following example on how IFRS 9 could raise loss provisions:

• $1 million 60 month amortizing term loan is made to a Risk Grade 4 borrower.
• Risk Grade 4 has a default range of 2.5 – 4.0%.
• The borrower is a chain of dry cleaning stores in business for 20 years.
• Public data shows a range of recovery given default of between 30% and 75%.
• There is collateral (store leases and fixtures) but no recent defaults in dry cleaners have been reported.

First, under the current rules there will be no reservation needed against this loan as long as it is current. But under IFRS 9 a provision will be required as soon as it is made. This provision will be calculated as follows:

Expected Loss = Risk Grade times Estimate of Outstanding Principal if Default Occurs times Loss Given Default.

These factors allow for interpretation and selection from a range. For example, the Risk Grade 4 indicates a default range of 2.5% to 4.0% or a range of 150 basis points. For individual lenders these ranges may be wider or narrower. The principal amount starts at $1 million but reduces as payments are made. What might be the principal due should default occur can only be estimated using historical data with like assets compiled over a number of years. Many lessors will have this data, many will not; but it is available. Recovery rates on collateral can vary by asset type, by stage of the economic cycle and by whether or not the equipment remains essential to the business. The decision to use the high end or low end of these ranges can dramatically affect the calculation of ECL.

ECL = 4.0% X $800,000 X 75% = $24,000

A conservative approach will show a $24,000 addition to loan loss provisions. But note that this has used the top end of the default risk range, an 80% assumption of remaining principal and a quite high estimate of loss given default. Furthermore, this is an amount that is straight from the bottom line. It is a reserve where none existed before. As will be seen later, not only should there be a better estimate, but there are further complications should this loan show any sign of deterioration during its life.

IFRS 9 Overview

The accounting standard has taken more than 6 years to formulate and has become more and more complex in the process. There is special treatment for financial assets such as investments, there are rules on when assets are to be excluded (“derecognized") and how probability weighting is to be used for groups of similar assets which will be of particular interest to asset-based lenders with highly homogenous portfolios.

The major element of the change is what has been referred to as the “three bucket" approach. In the first bucket go the vast majority of the loan assets for they are normal; they are expected to pay and show no signs of distress such as missed payments. These will have the calculation noted above applied using the assessed risk for the next 12 months. The second and third buckets are very different. Once a loan deteriorates (defined by the lender) then it has to move into bucket 2 where the loss provision will be based on the risk for the remaining term. This means that a term risk premium will have to be charged and this should be based on data from like assets over time. To make this calculation it will be very important to have either collected or have access to data that will enable an auditable assessment rather than a conservative estimate to be made.

Bucket 3 is for assets where there is a high probability of loss and the treatment here is very similar to that currently in place in the incurred loss model, i.e the asset has to be reduced to its estimated realizable value.

A further complexity in the IFRS 9 rules is an explicit consideration of the time value of money. Future revenue and the ECL will be discounted to reflect the present value of the cash flows using the effective interest rate of the asset at inception. As assets move through stages these calculations get performed and re-performed because the movements between stages by all assets is part of the disclosures requirements. There are also rules on asset recognition and “derecognition" that deal with assets passing through the final bucket.

Perhaps the most radical (and certainly the most controversial) change that comes with IFRS 9 is the requirement to not only estimate the future value of loans that have reached the second bucket using an economic forecast but to take the probability weighted forecast outcomes and then book the estimate in the current year. For the first time (perhaps ever) accounting requires a possible future value to be part of the historical cost results.

These rules will be scrutinized and managed by the accountants charged with meeting the regulations. Compliance should be assumed, the concern of management will be to ensure that the outcome does not provide a false picture of the real risk that exists and that the loss reserves are not materially in excess of what is realistic.

That said, every CEO will see that the underlying concept of ECL is fundamentally sound and should therefore direct attention to how to get the best management information, how to use the information for better pricing and risk management and how to avoid the imposition of excess and unjustified conservatism.

"We're Gonna Need a Bigger Boat"

In the classic movie Jaws, Police Chief Martin Brody (Roy Scheider) says, “We're gonna need a bigger boat!" Likewise, lenders are gonna need a bigger data warehouse. To achieve the best results and avoid an excess of widely variable estimates, IFRS 9 models require massive amounts of current and historical data. Most lenders will probably find that they either do not possess the amount of data needed for IFRS 9 models, or that their data collection has been inadequate to date. The risk rating process was meant for regulation and nothing else. Without a good rating scale, migration data has no value. Furthermore, many lenders have no records of prepayments, no records to demonstrate the value of covenants or other structural benefits, and no loss or default databases. Getting this data may be expensive, but the expense can bring big benefits in better decisions, better monitoring, better portfolios, and better reporting.

Protecting the Integrity of Financial Results

Let's go back to our example and outline how a loss provision estimate may be different for an equipment finance company using estimates specific to its business. Dry Cleaners are among the safest small businesses. Using the long-term default rate for Dry Cleaners (also conservative) the default risk is 285 bps. Using a deep and comprehensive default database we find that the probable exposure at default will be 50% of the principal at inception and that same default database with hundreds (perhaps thousands) of defaults of similar sized dry cleaning businesses shows that the collateral value will be closer to 60% of the principal at default. The estimate of ECL falls from $24,000 to $5,700 as follows:

ECL = (2.85%) X $0.5mm X 40% = $5,700

Imagine this calculation performed on every good asset where there is currently no reservation. The loan loss reserve could be not only vastly different, it could also be vastly wrong without the data to prove the more realistic answer. It will be critical to prepare for these outcomes for the impact on the bottom line, the senior management, the Board, and the shareholders may well be cathartic.

The problem may well be exacerbated when the other two stages (buckets) are considered. In IFRS 9, if a loan moves into stage 2 (for example by missing a payment) the increase in the required reserves can be penal. Not only will the default risk increase but also the loss reserve will have to be made for the remaining life including an adjustment for the time value of money. It would not be uncommon for the reserve to increase 200-300%. If a reasonable adjustment is to be made it will be very important to have data that can be used to measure the time (migration) risk rather than have a very high estimate imposed. Data does exist that can provide the information needed to calculate the migration risk. Seeking these out before the standard becomes the rule would be wise.

The best advice for dealing with IFRS 9 should be clear. Preparing for it means collecting or accessing the best possible data specific to your business and making sure that the auditors (and regulators) understand and agree to that data being used to provide the best estimate of ECL. The good news is that the data will be very valuable for more commercial purposes through time, but there is little doubt that the first year of this new standard will cause many surprises, with few being agreeable.

William Phelan is president of PayNet, Inc., a leading provider of credit data on small businesses in the United States. He serves on the Federal Reserve Bank of Chicago's Advisory Council on Agriculture, Small Business, and Labor, the board of directors of various industry associations, and the Research Committee for the Coalition for Responsible Business Finance, which he chairs.