Pinners Hall 105-108 Old Broad Street London EC2N 1EX tel: + 44 (0)20 7216 8947 fax: + 44 (2)20 7216 8928 web: www.ibfed.org 9th September 2009 Comment letters International Accounting Standards Board 1st Floor 30 Cannon Street London EC4M 6XH United Kingdom Dear Sir Re: Request for information - Impairment of financial assets: Expected cash flow approach The International Banking Federation (IBFed) welcomes the opportunity to provide you with comments from the global banking community regarding the feasibility of an expected cash flow approach for measuring impairment of financial assets. The IBFed is the representative body for a group of key national banking associations in Europe, the United States, Australia, Canada and Japan. The countries represented by the Federation collectively represent more than 18,000 banks with 275,000 branches, including around 700 of the world’s top 1000 banks which alone manage worldwide assets of over $31 trillion. This worldwide reach enables the Federation to function as the key international forum for addressing legislative, regulatory and other issues of interest to the global banking industry. Based on our members’ experience of applying the current incurred loss guidance contained within IAS 39, we welcome proposals which allow us to apply a greater level of professional skill, experience and management judgement to an inherently subjective area. A move towards a robust, expected loss approach which is pragmatic and operational in the context of a modern bank and which allows for the recognition of losses earlier in the credit cycle than is currently the case is welcomed by our constituents. Further, we welcome proposals which simplify impairment guidance such that there is a single model of impairment that is equally applicable to all types of financial assets. We do, however, have two significant concerns: 1 We do not consider the expected cash flow approach to be operational and our member banks are unlikely to be able to implement it at a reasonable cost. We do not consider the time frame being proposed for this project to be reasonable. The proposed approach represents a very significant departure from current practice and we consider the Board should instead be proposing an interim measure addressing the incurred loss issue thereby allowing sufficient time for the Board and its stakeholders to mutually develop a more considered and satisfactory model over an appropriate timescale. The practicality of the proposal Whilst appearing conceptually simple when applied to a single fixed rate product, the proposed model becomes far too complex when applied to portfolios of products. Banks generally maintain their records based on outstanding balances and evaluate loans for impairment based on credit losses. Thus, systems for estimating cash flows for all loans, whether on an individual loan basis or on a portfolio of products basis, would need to be developed. While the current collective basis of impairment is based on portfolios of loans with similar risk profiles, we foresee enormous practical difficulty in applying the expected cash flow approach to any portfolio where the population of the portfolio constantly shifts due to originations, redraw facilities and early redemptions. Experience in the United States to estimate expected cash flows to comply with AICPA Statement of Position 03-3 (a process limited to purchased impaired loans and securities) has shown this process to be extremely challenging. For example, prepayment assumptions must be broken out at a minimum by product, geographic region and underwriting stratum and require detailed estimates of yield curves that can be highly speculative. Expanding this process to entire portfolios is a daunting challenge that appears to provide little incremental benefit. The complexity increases when variable rate products are considered. The very fact that the IASB was only able to provide guidance on how the model might be applied to a variable rate product after the initial request for information (RFI) had been released is evidence enough of the conceptual difficulties involved – without even getting to the stage of considering how the model might be applied in practice to a loan book of thousands or tens of thousands of variable rate loans. The model is also premised on the collection and availability of expected cash flow data that forecast the amount of, and the period in which, credit losses arise. The majority of banks do not currently base their impairment models on such data – particularly with respect to portfolios of lower balance loans. 2 Time frame and scale of proposed change The system and process changes required to implement the expected cash flow approach is better characterised as “completely new systems” rather than “adjustments to current systems”. Banks’ systems and process for measuring impairment are generally some of the largest and most complex processes the bank maintains as part of its financial reporting process, and any wholesale change is likely to carry a very significant expense and take considerable time to design, test and implement. What issue is being addressed? We note the various criticisms of the current incurred loss approach detailed in the RFI. While these criticisms are open to debate, we believe the key issue causing most concern internationally is that the current guidance has prevented banks from providing against losses early enough in the credit cycle. We believe that addressing this issue in a pragmatic way should be the objective of this review of impairment by the IASB. The proposed model is too complex to be operationally feasible and would be far too expensive for the international banking community to implement, particularly when compared to the perceived benefits to shareholders and other users. We recommend that the board investigate alternative approaches to addressing the perceived failings of the current incurred loss model. These approaches may involve modifying the proposed expected cash flow model to make it more operationally feasible, modifying the current incurred loss approach such that it becomes an expected loss approach, or designing a model based on an expected loss approach without resorting to expected cash flows. For example, we believe that a model which projects losses that are “foreseeable over a horizon that is reasonably estimable” would be more operational, yet provide investors with a practical review of current results. Modifying the proposed expected cash flow model The implementation and application difficulties posed by this model appear not to be a complexity of the model per se, but rather a feature of the expected cash flow approach in general. Banks do not collect the required data and do not currently have the processes and systems to collect and manage it. Further, banks will also be unable to predict the timing of the missing cash flows, a requirement which is inherent in the calculation of the effective interest. As such we do not believe there are significant simplifications that can be made. On a conceptual basis, the expected cash flow model appears simple when applied to a single fixed rate product; however, when applied in volume, it may not be able to be simplified and still function as a measurement of impairment. The model appears not to be scalable in terms of application to portfolios or to instruments of increasing complexity. 3 Modifying the incurred loss approach The concept of what is “incurred” places significant artificial restrictions on the factors and data that would otherwise feed into our model and inhibits the use of our professional skill and judgement concerning the adequacy of the provisioning. In a nutshell, this means that at times our experience and application of judgement indicates to us that losses have occurred, are currently occurring or will shortly occur in a portfolio, but we are prevented from providing for them because of the accounting. Our strongly held view is that simply addressing the “incurred” concept and thereby removing these artificial restrictions will pragmatically address the issue at hand and will have the following advantages: Banks may continue to generate impairment provisions using the systems, data and processes they currently have in place. Any changes, if there are any, would be smaller than those involved in moving to an expected cash flow approach, with smaller associated cost. Such smaller changes could be implemented in a timelier manner than would be the wholesale changes required by the IASB proposal. The text in paragraph 59 of IAS 39, which states that “losses expected as a result of future events, no matter how likely, are not recognised”, should be removed. This text should be replaced by guidance that all available information needs to be factored in to determine the required provisioning, allowing management to use its experienced judgement to provide for losses that: may be recognised in the absence of the “trigger event” concept, which arises due to the incurred loss approach, are inherent in a portfolio but for which objective evidence of a loss event having occurred cannot be obtained; and/or are expected to be incurred in the near future based on analysis of market, economic, industry and other relevant data. An alternative expected loss approach An approach to measurement of impairment that allows for provision against expected losses, but which is not based upon and does not require the collection of expected cash flow data, appears to be a feasible alternative. Such an approach may use the drivers of probability of default, loss given default and exposure to determine an expected loss over the remaining life of the portfolio. The probability of default 4 for this calculation may be an economic cycle average. This approach, as opposed to the expected cash flow approach, allows portfolios of loans with similar risk profiles to be constructed and for these to form the basis of the collective impairment exercise. Consistency with regulatory measures of impairment Our members believe that the basis upon which expected loss impairment provisions are calculated for regulatory and for financial reporting purposes should be substantively the same. The ability to leverage the same data sources, systems and processes for both regulatory and financial reporting would reduce banks’ compliance costs and increase financial / regulatory comparability. Transition Although we support removing the restrictions relating to provisioning, our members have also expressed concern that a transition period is needed for converting to a new model. Banking institutions have been prohibited for a number of years from expanding their provisions to include expected losses, and consideration should be given to providing a reasonable transition period for implementing the new model. In summary, we believe the expected cash flow approach overcomplicates and hugely expands the cost to preparers of addressing this “too little too late” issue. Additionally, given the system and process changes that would be required it greatly inflates the time taken to implement the change and bring it to market to benefit users. We believe there are pragmatic alternative expected loss approaches available which the Board should consider prior to developing the forthcoming Exposure Draft on impairment. We would be pleased to discuss our concerns with you at any time. Yours sincerely, Sally J Scutt Michael Domann Managing Director Chair, IBFed Accounting Working Group 5