11 May 2010
One of the key issues in accounting for financial instruments is when to recognise and how to measure loan loss allowances (impairment) for receivables and loans reported at amortised cost. This affects many companies but in particular a large part of the balance sheet of most banks. It is also a key issue for investors and it is important that we understand what information investors require to fully appreciate the financial position and performance of these institutions.
At present IFRSs apply a so-called ‘incurred loss’ model whereby an allowance can be recognised, and a loan written down, only if there is evidence that a loss has been incurred and that evidence is an observable event (a ‘triggering event’) such as the default of the borrower. The problem with this approach, which has been highlighted by the financial crisis, is that no loan loss reserves can be built up in advance of a loss being incurred, even though in practice experience may show that a certain proportion of loans would be expected to default over the life of that loan portfolio. Many commentators have criticised the current approach as ‘too little, too late’. A related problem with the incurred loss model is front-loading of interest income. Prior to the triggering of a loan loss allowance interest income is recognised on the basis of the contractual interest rate being charged even though the bank may expect that for a portfolio of loans the overall return net of credit losses is likely to be somewhat lower. This results in an unrealistically high level of income being recognised during the early period of a loan.
We have proposed in a recently published draft standard that the accounting for the impairment of loans and other debt instruments recorded at amortised cost should be changed to reflect expected rather than incurred losses using an ‘expected cash flow model’. Under this approach the amortised cost and carrying amount of a loan would equal the present value of the expected cash flows from that loan (after taking account of expected credit losses) discounted at the expected yield. So, right from when a loan is first recognised credit loss expectations are accounted for. No triggering event would be required to start the recognition of a loan loss allowance and consequently a more realistic pattern of income recognition.
The mechanics and impact of the expected cash flow approach are perhaps best explained by an example. Assume that a bank grants a number of five- year fixed interest loans to its retail customers with a total principal amount of CU100 million and charges interest of 12 per cent a year. The contractual cash flows are clearly interest income of CU12 million in each of the five years plus CU100 million in year 5. This gives a contractual yield (internal rate of return of the cash flows) of 12 per cent. However, experience with similar loans and customers in the past shows that it is likely that there will be a certain level of defaults, and as a result the expected cash collection, and hence expected yield, for the bank is rather lower. Let’s assume that on average 2 per cent of loans are expected to default each year, ie 10 per cent of loans overall, and that there is no recovery on default, with the result that principal of only CU90 million is recovered in year 5. Interest income will also be reduced, although the exact amount received in each period will be determined by the timing of defaults. The expected yield for this portfolio will be about 10 per cent (it could be slightly more or less depending on the timing of defaults); a reduction in yield of about 2 per cent. One would expect that this reduction in yield is factored into decisions regarding the investment in these loans; either a decision about what contractual interest rate to charge or, if the bank is a price taker from the market, whether it wishes to participate in that market in the first place.
Application of the expected cash flow model – initial expected losses
Under our proposals this portfolio of loans would initially be recorded at CU100 million. There is no need for an immediate impairment since, although losses are expected, these are built into the interest rate being charged. (The loans are still ‘worth’ CU100 million as the loss expectations were built into the pricing.) Subsequently contractual interest is recognised as income but in each period it is reduced by a charge reflecting the reduction in yield due to expected credit losses, and as a result the net amount equals the expected yield multiplied by the carrying amount of the portfolio. The loan loss allowance is used to build up a provision against which loans are written off when eventually the bank gives up on collection.
A few investors have criticised this method of recognising initial credit losses as ‘artificial smoothing’ and have argued that reflecting losses as incurred would be better. However, we believe that our approach is consistent with the measurement of amortised cost in other areas and merely corrects an exception that has been applied through the incurred loss model. Take, for example, a stepped coupon bond issued at 100 which pays a 12 per cent coupon in the first year, reducing by 2 per cent in each subsequent year and redeemable at a 10 per cent discount. Under current amortised cost measurement, interest would be recognised on the basis of the effective yield to maturity for this security with the difference between the interest cash flow and amount recognised in profit or loss being an adjustment to the carrying amount. Effectively the expected cash flow model is exactly the same, with the exception that it uses expected cash flows net of credit losses rather than the stepped contractual cash flows. I do not believe that income is artificially smoothed in either case.
Dealing with catch-up adjustments
Clearly, estimates of future credit losses are subject to subsequent revision because actual defaults are unlikely to match expectations exactly. Under our proposals the present value of any change in expected cash flows that is subsequently identified is recognised immediately as a charge or credit in profit or loss. The adjustment results in the carrying value of a loan always equalling the present value of expected cash flows discounted at the initial expected yield. Such a catch-up adjustment would be displayed as a separate line item in profit or loss and should provide investors with valuable insight into the effect of changed credit conditions and the risk associated with a given loan portfolio. We believe that splitting credit losses into two components, in each case with full supporting disclosures, is superior to the existing incurred loss model. The two components are:
- Those initially expected credit losses that are effectively built into the interest rate charged, which are recognised as a reduction in yield.
- Those unexpected adjustments that are recognised immediately.
However, many banks and some other commentators appear to disagree with the IASB’s proposed treatment of catch-up adjustments and argue that the catch-up should be allocated (or ‘smoothed’) over the remaining life of the loan rather than recognised immediately. They suggest this would be more consistent with the treatment of the initial losses and would also reduce the volatility of gains and losses in profit or loss. However, our approach deliberately differentiates between initial expected losses and subsequent changes and applies different accounting to reflect the fact that one is effectively priced into the contractual interest rate and the other is not. Also, we believe that gains and losses should be volatile where this faithfully reflects the underlying economic position, as is the case for a change in expected credit losses, but not where volatility is false, as many argue is the case for the incurred loss model (when volatility can arise simply as a result of the realisation of initial expectations). The IASB staff have prepared a short paper and example which has more detail about catch-up adjustments and the rationale for the IASB’s approach
What are your views on the catch-up adjustment and the expected cash flow model generally?
We are particularly interested in investors’ views on the treatment of catch-up adjustments in the expected cash flow model; whether you think we have the right approach or whether you think that allocating the adjustment over the remaining life of the loan is more appropriate. We have therefore prepared a short investor survey that focuses on this question and other aspects of the model. We would really appreciate it if you could take a few minutes to fill this in. Alternatively we would welcome comments on any aspect of the proposal either by formal comment letter or by simply commenting on this article below.