Commission bancaire - Recommendations on Accounting for Financial Instruments and Similar Items 

OVERVIEW

1. The following is the comprehensive response of the Commission Bancaire (the CB) and the Banque de France (the BdF) to the request for comments on the Draft Standard "Recommendations on Accounting for Financial Instruments and Similar Items" developed by the Joint Working Group of Standard Setters (JWG).

2. These comments are intended to provide the IAS Board with the analysis and recommendations of the CB and the BdF regarding the application of fair value accounting (FVA) to financial instruments, and in particular regarding the specific implications of the Draft Standard for the banking industry.

3. The CB and the BdF acknowledge that the Draft Standard is a very important document which contributes significantly to the international debate on the development of improved accounting standards.

CONCLUSIONS OF THE CB AND THE BDF

4. The request for comments from the JWG distinguished between comments on conceptual issues related to fair value accounting (FVA) on the one hand, and comments on specific technical details on the other hand, the latter being solicited by means of an extensive questionnaire.

5. The CB and the BdF's response covers both areas. The conceptual issues are discussed in the following paragraphs, and detailed responses to the questionnaire are provided in the Annex. Nevertheless, the CB and the BdF attach the greatest importance to the conceptual issues, since they pose the most substantial problems relating to the application of FVA to banks.

6. The CB and the BdF's comments on the conceptual issues focus on (i) the overall relevance of FVA to financial instruments held in banks' banking book, (ii) the use of fair value-based information within banks for risk management purposes and for reporting on financial performance, (iii) the implications of FVA for financial stability and transparency, and (iv) the impact of FVA on capital adequacy regulation. The additional issue of the accuracy and reliability of FVA for financial instruments in the banking book, while highly significant, is not dealt with in as much length in this response, both because the CB and the BdF consider the issue of relevance to be of greater fundamental importance, and because the issue of accuracy has already been thoroughly addressed by several working groups' with which the CB and the BdF are associated.

Working group of the Banking Advisory Committee of the European Commission. Working group of the Basel Committee on Banking Supervision. Working group of the Conseil National de la Comptabilite.

I. RELEVANCE OF FVA FOR FINANCIAL INSTRUMENTS INCLUDED IN BANKS'

BANKING BOOKS

7. One of the most important issues raised by the Draft Standard is the relevance of figures derived from full FVA measures of valuation to the proper use and understanding of banks' financial statements. In this context, the IASB Framework pointed out that "...information must be relevant to the decisionmaking needs of users" (see Framework, para. 26). There are substantial doubts as to whether the full fair value approach proposed by the JWG satisfies that criterion, when applied to the balance sheet and income statement of a commercial bank.

8. The relevance of FVA for financial instruments held in connection with banks' trading activities is not questioned.

9. In contrast, the relevance of FVA for the financial instruments held in the banking book (i.e. held-to-maturity investments, loans originated by banks, receivables and core deposits) is highly questionable. Significant questions can be raised involving (i) the balance sheet, (ii) the income statement and (iii) the goingconcern principle.

10. The JWG proposal is based on the assumption, as a matter of principle, that a market price, determined by an efficient market, is available for every financial instrument. This assumption is expressed in paragraph 71 of the Draft Standard, which defines the fair value of a Financial Instrument as an estimate of its market exit price. This exit price is assumed to be the basis for evaluating all financial instruments.

11. Furthermore in paragraph 1.8(a) of the Basis for Conclusions concerning relevance, the JWG states that one of the most important conceptual advantages of fair value measurement is that it reflects "...the market's assessment of the effects on financial instruments of current economic conditions... when they take place. This follows from the expectation that the fair value of a financial instrument determined in competitive, open market economies embodies all available information up to the measurement date. "This confirms the JWG's assumption that the FVA approach is driven by a trading perspective of accounting, and also underlines the JWG's view that the valuation of financial instruments should reflect changes in current economic conditions in real time.

12. These assumptions are appropriate for the trading book activities of banks, which involve actively buying and selling publicly traded securities and related instruments with the goal of profiting from short-term variations in market prices. "Fair value" would appear to be an appropriate basis for valuing and accounting for such trades.

13. But it is questionable whether these assumptions are equally appropriate for banking book assets, which are not intended to be sold or settled in the short term, and for which market prices are not readily available. The goal of the lending business is not to profit from short-term changes in economic conditions, but rather to earn recurrent income from the margin between loan revenue and deposit expense. Charges on loans are calibrated to cover costs over the lifetime of the lending relationship. These costs consist of interest expense (including interest on core deposits, which are typically provided at a low and stable interest rate, since the objective of depositors is to have a cash reserve at their disposal rather than to obtain a market yield), administrative expenses (which are recognised on an accrual or cash basis), credit risk (which financial markets generally cannot estimate accurately), and the cost of own funds (which depends on the long-term risk-appetite of the bank's shareholders). Thus the costs incurred in the banking book have little or no correlation with short-term fluctuations in financial market conditions.

14. Moreover, banking book customers (especially retail borrowers, but also corporate borrowers) generally do not behave in the same fashion as professional traders. Because traders are interested in short-term profit, they decide whether to maintain or liquidate their positions from moment to moment as a function of current market conditions. Banking book customers typically have longer term financial objectives (e.g., financing a residence, funding a long-term business plan) which are less sensitive to variations in market conditions.

15. For loans and other items held to maturity, the amount due at maturity is more relevant to both the bank and the borrower-than transient swings in the liquidation value of the position. This is particularly true when the banking business is long term and cash flows are matched, as is the case for much of the banking book. When an instrument is held to maturity, its value is defined by the conditions when it was originated, and not by the conditions prevailing when it is revalued.

16. The CB and the BdF conclude that FVA, as defined by the JWG, does not reflect the true economic value of the banking book. The historical cost method currently used to evaluate the banking book-while it could be improved is more appropriate.

17. In proposing FVA for all financial instruments, the JWG in effect treats different business lines as equivalent; whereas in fact the fundamental differences between the trading and banking books call for different accounting treatments.

18. FVA is particularly problematic when applied to core deposits and the own credit risk. The JWG's proposed accounting treatment fails the test of relevance because it does not include the value of funds that core depositors can be expected to deposit in their accounts in the future. This omission cannot be satisfactorily remedied, since the optional character of future core deposits is extremely difficult to model with precision. This point is explained in more detail in the following paragraphs.

19. The Draft Standard does not treat financial assets and liabilities consistently. Paragraph 92 of the Draft Standard states that "values that are included in a market exit price that are not directly attributable to the rights and obligations that constitute a financial instrument should not enter into the determination of the fair value of that financial instrument." Subsequent paragraphs (93, 94, and 336 to 339) specify that, for other transactions, expected future deposited funds should not be taken into account to determinate the fair value of core deposits.

20. This is inconsistent with the proposed treatment of the asset side of the balance sheet, which requires all expected future cash flows or changes in cash flows due to embedded options (such as early repayment options) to be included in the fair valuation of assets.

21. Depositors have the option of depositing additional funds in their accounts. This option, which is a feature of the initial contract, should be taken into account in the fair valuation of core deposits. In practice, the vast majority of core depositors exercise their option to deposit additional funds. Consequently, FVA valuations that are based exclusively on the runoff of the initially deposited amount will differ very substantially from the actual economic value of the account.

22. The CB and the BdF conclude that a FVA approach to core deposits would, in principle, have to take into account the amount and timing of future deposits, which is not allowed in the Draft Standard. This conceptual drawback of the JWG proposals would have a very significant impact on the balance sheet of credit institutions.

23. However, the behaviour of core depositors is, by definition, relatively independent of short-term variations in market interest rates. It is therefore difficult to see how an accounting system based on fair value could accurately measure the value of core deposits.

24. The CB and the BdF also criticise the effect of fair value measurement of financial liabilities, as proposed in the Draft Standard, when the own credit standing of an enterprise (including banks) deteriorates. The fair value method leads to revaluation of the debt issued by the company, leading to the fact that the more the rating of the company deteriorates, the more the value of its debt decreases and the more its equities increases as the difference of revaluation is accounted for in the profit and loss account. A bank would then see its equities and regulatory capital reinforced by the deterioration of its financial situation. But in real terms a decreasing fair value of debts will not, in this case, represent an improvement in substance as the enterprise is still under the legal obligation to reimburse the nominal value of its debt. Even if its debt is sold to another creditor at a discount, the new creditor keep the right to request the repayment of the nominal value. Furthermore, this situation is completely contradictory with jurisdictional regulations, which usually make shareholders' rights be subordinated to the complete settlement of all creditors' claims. Thus, the fair valuation of own credit risk is not only counter-intuitive, but also generates clearly untrue and misleading information that could lead creditors and shareholders to erroneous decisions. This would not be in accordance with objectives laid down in the IASB Framework (par. 26) about relevance.

25. The Draft Standard would require every change in the fair value of a financial instrument to be taken immediately to the income statement. This approach has been widely criticized as failing to distinguish between realized and unrealized gains, and thereby allowing the distribution of unrealised gains. The CB and the BdF share this criticism: it is highly imprudent to take to the income statement what may be transient paper changes (especially increases) in asset values, as if they were realised.

26. This problem is compounded by the unreliability of "mark-to-model" estimates of the fair value of unmarketable banking book assets. Thus the issue of reliability, which is most usually brought up in the context of balance sheet valuation, is also highly relevant to the income statement.

27. There is a strong possibility that fair value would be interpreted as the liquidation value of a financial instrument. For trading book assets, it is reasonable to assume that liquidation values and the value of continuing to hold the assets are closely aligned, but this is not generally true of banking book assets, which are highly illiquid and therefore difficult to liquidate quickly for more than a fraction of their value if held to maturity.

28. Similar problems arise in applying FVA to acquisitions: in particular, the value of goodwill is not taken into account in the financial statement under full fair value accounting. This result is inconsistent with the principle that fair value should reflect the price between unrelated enterprises in an arm's-length exchange motivated by normal business considerations.

29. A related problem has already been discussed above: core deposits. FVA would value core deposits at their liquidation value, as the JWG only considers their contractual term (i.e. payable on demand), ignoring future transactions that would renew the value of the deposits. This accounting treatment does not reflect the true economic value of core deposits. More generally, the core business of the banking book consists of managing the mismatch between contractually short term liabilities and longer term assets. This mismatch is in fact largely apparent, since the banking relationship with core depositors is in fact long term. By ignoring such long-term relationships, FVA fundamentally underestimates the value of the banking-book.

II. USE OF FAIR VALUE-BASED INFORMATION WITHIN BANKS FOR RISK MANAGEMENT

PURPOSES AND REPORTING ON FINANCIAL PERFORMANCE

30. Most banks do not currently use fair value-based information as the basis of their risk management processes. In fact, banking books are generally managed to optimise the long-term margin income generated by stable assets and liabilities. As discussed above, the revenues and costs associated with financial instruments in the banking book have little or no correlation with short-term changes in market conditions. Serious questions must be asked about the wisdom of adopting accounting methods that do not correspond to the methods that banks actually use to monitor and assess their risks.

31. In paragraph 1.8 (d) of the Basis for Conclusions, the JWG responded to this objection by stating that "...fair value reflects the effects of management's decisions to continue to hold the assets and owe liabilities, as well as decisions to acquire or sell assets and to incur or settle liabilities;" Thus, the JWG believes that the value of financial instruments is determined by economic conditions and remains the same whether management decides to liquidate the instruments or hold them to maturity. This leads to the conclusion that a single valuation method is justified for all financial assets, and that unrealised gains should be recognised in the income statement.

32. But is the value of a financial instrument really independent of the bank's strategy? It is intuitively obvious that different holding periods imply exposure to different kinds of risks. A financial instrument held for a short period of time is exposed primarily to market risk, the expected future cash flows depending mainly on financial market conditions that affect the instrument's market price. The same financial instrument held for a long period of time will be more sensitive to credit risk, as the probability of default generally increases over time, whatever the current rating of the loan. At the same time, market risk becomes less relevant as the holding period increases, since short-term fluctuations in market prices are poor predictors of the eventual state of the market at a distant maturity date. This uncertainty diminishes as the repayment date approaches, but in that case the market value converges to the historical cost.

33. Furthermore, FVA focuses on the effect of interest rates on banking book values at a particular point in time, but does not convey the sensitivity of the banking book to interest rate movements. In practice, information on sensitivity is more important. Management assesses the bank's sensitivity to interest rate risk on a global basis using tools such as gap analysis. This provides management with a more useful measure of the interest-rate risk profile of the bank than would be provided by a snapshot of interest rate exposure at a single point at time. There is no reason to believe that the interests of analysts and investors would differ on this point.

34. By the same token, FVA is of limited relevance in the reporting of banking book activities in the bank's financial statements. Readers of the financial statements have a stronger interest in recurrent margins in the banking book than in short term fluctuations in value. Because lending is a long-term business, short-term fluctuations in market spreads should not be taken as an indication of deterioration in credit quality, either for a single loan or for a entire loan portfolio. The source for such a variation could be broad macroeconomic factors which are not directly linked to a single borrower or a group of borrowers. Internal systems for managing risk in the banking book therefore focus primarily on detecting possible deterioration in credit quality, rather than on following market rates. In this context, the CB and the BdF note that credit risk measurement and management may evolve over time (at least at large and sophisticated banks) as a consequence of the proposed recognition for capital adequacy purposes of internal ratings-based methods.

III. FINANCIAL STABILITY AND TRANSPARENCY OF FVA

35. The traditional activity of banks, which constitutes the banking book, consists in recycling short-term savings at variable rates into medium- and long-term loans at fixed rates. Large commercial banks have for a long time developed expertise and skills in managing this transformation. They play an important role as intermediaries between other economic participants, who are less able to manage these mismatches efficiently. This intermediation is a key element in the efficient functioning of the overall financial system;

36. Under fair-value accounting, the valuation of banking-book assets and liabilities would fluctuate in response to short-term fluctuations in market interest rates. This would result in an artificial increase in the volatility of the earnings and capital generated by banking book activities. This increase in volatility is artificial as the volatility that is present in markets for traded instruments does not exist in markets for loans and deposits. And this volatility is all the more artificial since provisions of the Draft Standard related to the valuation of core deposits are inappropriate, leading to an asymmetrical accounting treatment between assets and liabilities.

37. Banks can be expected to respond to this increase in volatility by reducing their apparent, but misrepresented risk exposure in this area, transferring liquidity and interest rate risks (through short-term loans at variable rates) to other economic participants who do not have the skill to manage these risks efficiently. Banks could also charge their banking book customers an extra premium. The result would be an overall decline in intermediation, with serious implications for the real economy.

38. The tendency towards disintermediation would have particular negative effects during periods of economic stress, precisely when the need for liquidity from the bank system is greatest, as the reluctance of banks to recognise a misrepresented risk exposure could lead to a credit crunch. Furthermore, fair value accounting would lead to a uniformity of behaviour (herding behaviour) of both financial market participants and the banking community. This is a key issue, as the equilibrium price of assets is always the result of the interaction of different parties having opposite needs and objectives. The market needs `contrarians' namely participants that are willing and able to act as counterparts and to provide liquidity when the market needs it, and more specifically when the usual liquidity providers are not able to fulfil this role. Otherwise, a uniform behaviour will result in an even more increasing volatility.

IV. IMPACT OF FVA ON CAPITAL REGULATIONS

39. The conceptual framework of FVA would necessitate sweeping changes in accounting, and therefore in the calculation of capital requirements, which are based on data derived from accounting. FVA would recognise more latent gains, regardless of the liquidity of related financial instruments and the volatility of the gains over the holding period.

40. FVA would also substantially affect the supervisory review process, which seeks to determine whether the capital available at a given bank is adequate to cover the risks it incurs, taking into account its current exposures, its prospective income, and the quality of its risk management and internal controls. The CB and the BdF considers it essential that banking supervisors should be closely involved in developing new accounting standards that could be used to value financial instruments in order to assure the consistency between accounting methods and the objectives of the prudential regulation, promoting sound management practices and conservative valuation.

RECOMMENDATIONS AND ALTERNATIVE PROPOSALS

41. Fair value accounting has serious disadvantages when applied to bank loans and other instruments in the banking book:

There is, generally speaking, no secondary market for bank loans, and their fair value cannot be determined with any precision by "mark-to-model' systems. Thus, errors in measurement are likely to outweigh any theoretical improvement in valuation.

Furthermore, the use to internal valuation models will result in non comparable figures from one bank to another.

FVA does not take into consideration the principle of prudence, since it treats latent profits and latent losses similarly. A prudential attitude does not recognize latent profits except on marketable instruments.

It does not reflect the economic reality of the way that banking-book assets and liabilities are managed in the long term. Management is driven principally by long-term decisions about credit quality and concentration and revolves around the fostering of customer relationships over the life of the contract, and not by reference to short-term changes.

42. The CB and the BdF conclude that, fair value accounting is appropriate only for the trading activities of banks, consistent with the so-called "mixed model" resulting from the implementation of IAS 39.

43. The CB and the BdF recognise that this mixed model needs improvement. We agree with the necessity to better incorporate unmaterialized credit losses and interest-rate risk into asset valuations, but we consider that this can be done in ways that are more reliable and more consistent with banking-book practice: i.e., through dynamic provisioning for unmaterialized losses, and assessment of interest rate risk in the banking book as a whole using stress-testing.

Dynamic provisioning for unmaterialized credit losses

44. As mentioned above, there is no secondary market for bank loans in European countries, and thus no way of observing the "market price" of a loan. Fair-value accounting of bank loans would therefore have to rely on models that estimate the present value of the cash flows generated by each loan. In practice, this presents difficult technical problems:

While it is relatively straightforward to model the principal and interest payments due on a loan, it is much more difficult to model non-interest income such as late payment fees.

Cash flows are highly sensitive to the exercise of embedded options (such as prepayment options on loans), which are notoriously difficult to model.

For instruments denominated in foreign currencies, fair-valuation requires estimates of future exchange rates, which introduces an additional source of error.

In theory, the appropriate discount rate varies from one instrument to the next, depending on the instrument's effective maturity (which is complicated by embedded options), and on the risk premium that corresponds to the instrument's riskiness.

45. A more practical approach to dealing with unmaterialized losses is to adjust the value of the banking book using dynamic provisioning.

Under current accounting principles, provisions are made for losses as they materialize, as evidenced by asset impairment, asset depreciation or liabilities appreciation, or a strong presumption of degradation in the value of a commitment.

Dynamic provisioning recognises (1) that some fraction of a currently unimpaired portfolio can be expected to deteriorate in the future, and (2) that the magnitude of these "expected but unmaterialized" losses over the lifetime of the portfolio can be predicted based on statistical analysis of similar portfolios.

46. Current accounting principles are often criticized as permitting latent credit losses to remain hidden, since recognition of the losses is delayed until signs of deterioration are evident. Dynamic provisioning eliminates the basis for this criticism.

47. Furthermore, dynamic provision is based on historical experience rather than on arbitrary assumptions regarding prepayment rates, yield curves, discount rates, etc. Consequently, there is less scope for bias and manipulation.

48. Dynamic provisioning is consistent with the principle of prudence, in that it adjusts for latent losses but not for latent gains.

49. Finally, dynamic provision would be far simpler to implement than fair-value accounting. The IRB architecture under the pillar 1 of the New Capital Accord already requires banks to estimate the probability of default and the loss given default. These are precisely the parameters that are needed to estimate expected loss.

Interest rate risk

50. Under fair-value accounting, the valuation of banking-book assets and liabilities would fluctuate in response to short-term fluctuations in market interest rates. This would result in an artificial increase in the volatility of banks' earnings and capital, which would tend to make them more reluctant to play their traditional role of intermediation when credit conditions deteriorate.

51. Furthermore, fair-value accounting focuses on the value of the banking book at a particular point in time, but does not convey the sensitivity of the banking book to interest rate movements. The latter is arguably a better indication of the interestrate risk profile of a financial institution.

52. Our proposal would be to require banks to publish a stress-test analysis of the interest-rate sensitivity of the banking book. The analysis could be based on nonaccounting data presented in the notes to the accounts, such as sensitivity and duration. This proposal is fully in line with pillar 2 of the New Capital Accord (which calls for banks to have internal systems that measure the effect on economic capital of a standardized interest-rate shock) and pillar 3 (which requires public disclosures to facilitate investors' and market participants' assessment of the banks' interest rate risk (IRR) profile in the banking book).

53. The current draft of the New Capital Accord requires disclosures regarding the bank's risk management structure for overseeing IRR, the nature of IRR, key assumption in its measurement, the use of hedging programs, and internal measurement systems. The New Accord also recommends disclosures on sensitivity analysis employed with regard to key assumptions, their effect on results, and the use of stress test scenarios. Through these disclosures, the readers of the bank's financial statement can appreciate the interest rate sensitivity, which is not possible with fair-value accounting.