## Sir Andrew Large: Financial instrument accounting
Speech by Sir Andrew Large, Deputy Governor of the Bank of England, to the 13th Central Banking Conference, London, 22 November 2004.
* * *
## Introduction
The recent furore over IAS 39 - the international accounting standard for financial instruments - has resulted in a situation which I think all agree is unsatisfactory. We now effectively have two versions of IAS 39, one proposed by the IASB, the other by the European Commission.
I probably do not need to emphasise that meaningful accounting standards and their effective implementation are highly relevant to any central bank's responsibilities for financial stability. They help to ensure the safe and efficient functioning of the financial system. So it is essential that accounting standards have a clear economic rationale, promote comparability and secure adequate disclosure. Only in that way can investors, creditors and others judge the positions and risks being run by financial institutions and hence exert market discipline, and perverse incentives be avoided. Together these can make an important contribution to financial stability.
I suggest later that, despite intensive work to improve accounting standards for financial instruments in recent years, fundamental issues with financial stability implications remain to be resolved. One crucial question is: 'Who and what are accounts for?' Another more specific issue is the appropriate role of fair value data. So what I want to do this morning is to discuss these issues in turn before commenting on the way forward regarding IAS 39. I should add that while IAS 39 applies to all users of financial instruments, I illustrate its implications below with particular reference to banking.
## Who and what are accounts for?
It is widely accepted that published audited accounts are prepared first and foremost for the owners of a business, that is the shareholders. However, it is often maintained that this information also meets the needs of other stakeholders, including creditors, customers, employees and so on.
And so it does up to a point. Yet there are many users who have an interest in the financial position of a firm. On the face of it, it would be rather surprising if a single presentation was ideal for all of them. Let me try to illustrate this for the financial world - where the Bank's financial stability remit runs - by looking at the respective needs of shareholders, depositors and regulators of banks.
Shareholders are particularly interested in the returns being earned by a bank (its profitability), both current and prospective. They are looking mainly at the economic value of the going concern. The focus of depositors, regulators and authorities concerned with systemic stability, on the other hand, is likely to be on whether bank liabilities will be repaid on demand or when due. Two specific examples may help to illustrate these different perspectives: the treatment of 'own-credit risk' and the valuation of sight deposits.
'Own-credit risk' arises from the 'fair value option' in IAS 39, which allows an entity to fair value not only its assets but also its liabilities. On that basis, the value of liabilities, including deposits, should fall as an entity's creditworthiness deteriorates. Here we find a difference of perspective between shareholders on the one hand and debt holders, including depositors (and by extension regulators and central banks), on the other. In an economic sense, given that the equity interest in a company cannot fall below zero, as net worth diminishes debt holders bear a growing proportion of reductions in asset values. But this approach is in tension with the understandable concern of bank depositors and regulators about the capacity of banks to repay liabilities at par when due, and who therefore want measures of net worth constructed on that basis.
A separate issue arises in arriving at a fair valuation of sight deposits . For such deposits, there is typically a difference between the contractual maturity (zero) and the behavioural maturity (the time over which sight deposits are held in practice). Since the interest paid on such deposits is almost always less than the market discount rate, once any maturity greater than zero is admitted the fair valuation of the deposit will be less than par. But once again, depositors are likely to be more interested in the financial strength of a bank assessed on the basis that deposits are ascribed their full
BIS Review 71/2004
nominal value. Moreover, any suggestion in public accounts that deposits are worth less than par could undermine the confidence on which the whole construct of banking is based.
These illustrations suggest that different users of accounts may, at least to an extent, want information produced on different bases. And the simple fact is that we will not be able to make progress if different users all assert that their way of calculating or presenting the numbers is uniquely correct. I suggest, therefore, that the users of accounts should recognise that they may in reality need to adjust them to suit their particular purposes - something ratings agencies and other analysts have done for many years. It is encouraging that the Basel Committee on Banking Supervision has adopted precisely that approach in promulgating various adjustments for regulatory purposes to data based on IAS 39.
Against that background, it would seem helpful if accounting standard setters required supplementary disclosures in published accounts which would allow different users to make such adjustments. This is important given that, unlike regulatory, fiscal and other public authorities, many private sector users are not in a position easily to obtain information beyond that put in the public domain.
Let me turn now to IAS 39 itself as it impacts on the banking industry.
## IAS 39 and the banking industry
IAS 39 has proved controversial with bankers for reasons which stem partly from specific features of banking, and partly from the implications of the wider use of fair values, particularly in accounting for hedges and in so far as they may lead to greater volatility of published profits.
## Some pros and cons of fair value accounting
The debate about the use of fair values raises a number of important issues, including the relevance of unrealised valuation changes, the appropriate assumptions on the holding period and the reliability of fair values. It also raises a question about the extent to which accounting valuations should move away from a going concern paradigm to break-up valuations, which, at least at first sight, is closer to what 'fair values' represent. But at the same time, fair values are in some ways more forward looking, since expectations about the future performance of assets and liabilities are reflected in market valuations. It will not have escaped you, however, that the move in recent years to a more securitized world, with increased capacity for rapid risk transfer, encourages the appetite for a more forward looking approach.
A fair value approach has a number of other attractions. For example, it promotes consistency of valuation of instruments in different financial sectors, something increasingly important given the fungibility between banking, securities and insurance. It will generally lead to more timely recognition of losses (for example, on bank loans), and it captures the crystallisation of market risk in non-trading positions. Moreover, fair value accounting is consistent with the increasing use of mark-to-market techniques in risk management. Such techniques in part reflect a progressive but fundamental shift from the traditional banking approach of holding assets until maturity to today's approach of managing on the basis of continually assessing the opportunity cost of maintaining the existing balance sheet. It is clearly beneficial to transparency if publicly-available accounting information as far as possible reflects the basis on which management actually run a business.
However, there are a number of complications associated with the wider use of fair values.
First, there is a question as to how one can obtain robust fair values for instruments which are not priced, even indirectly, in reasonably deep and liquid markets. Model-based valuation techniques may be used, but they may not be analogous to genuine market clearing prices. The issues here include determining the conceptual basis for valuation ('the model'), obtaining the necessary inputs, and avoiding slavish adherence to a model which may in some circumstances deliver misleading results. In other words how, and to what extent, can human judgement properly be used to modify the model?
Second, there is the question of the economic relevance of unrealised gains and losses - particularly if they are not immediately realisable. For example, while information on changes in the fair value of bank loans conveys useful economic insights, it needs to be interpreted carefully. In many cases, a gain cannot be realised 'up front' given the absence at present of developed secondary markets in bank loans - even if securitisation may be increasing in that area.
BIS Review 71/2004
## Volatility
Further concerns about wider use of fair value accounting relate to the possible implications for volatility in financial markets and in the economy more widely. In my view, there is an important distinction to be drawn between accounting rules which capture accurately the volatility inevitably present in the real world, and 'spurious' volatility introduced by the accounting rules themselves.
Few I hope would suggest that we should remove genuine economic volatility from accounting numbers. It is surely better for users to apply their own smoothing to data if they think that is justified rather than impose smoothing rules, which can both be arbitrary and may serve to obscure the underlying data, through obligatory accounting standards.
Recent experience in the US and the UK is encouraging in these respects. The widening use of fair values does not seem to have had an adverse impact on the stability of the US financial system, on the equity prices of US banks, or on banks in Europe which have adopted US or international accounting standards. And the move in the UK to a pensions standard with a strong fair value flavour (FRS 17) initially led to strong reactions, but over time I think has led to a better understanding of the actual economics of pension arrangements.
At the macro-economic level, concern has been expressed that greater use of fair values could have a procyclical impact. Banks' assets might be marked up in booms, as perceived credit risk declined, so boosting banks' capital base and capacity for further lending. However, for the concerns about intensifying economic cycles to be realised two linkages would need to operate. First, the impact on bank capital would have to increase the volume of bank lending and , second, the increase in bank lending would need to have significant implications for real activity. Neither linkage can be taken for granted, and in any case discretionary policy action, whether monetary or prudential, by the authorities may mitigate any pro-cyclical effects.
As I have indicated, we do, however, need to avoid accounting rules which introduce spurious volatility. An example, to which I will return in a moment, is the hedge accounting under IAS 39.
## The 'mixed attribute' model
The drawbacks of fair values in some contexts have led to widespread support in the banking industry for retaining the current 'mixed attribute' accounting approach. While IAS 39 requires trading instruments, derivatives and many securities to be measured at fair value, bank loans generally continue to be measured at historic cost. However, the debate about IAS 39 has highlighted some difficulties with this mixed approach as it is presently formulated.
The area of greatest difficulty has proved to be hedges which straddle the two measurement bases, where so-called hedge accounting comes into play.
One example of this is the use of interest rate swaps to manage the interest rate risk in a portfolio of fixed rate loans funded from floating rate deposits. Without modification of the mixed attribute model, such hedges could actually create, not reduce, accounting volatility, because derivatives are required to be marked to market, that is measured at 'fair value', while loans are generally booked at cost. This to me would seem perverse indeed!
Another important example arises with the use of credit derivatives as hedges. This has proved difficult more generally to fit into the IAS 39 framework.
I should add that the sheer complexity of the hedge accounting rules is also an issue, as the debate in Europe about the treatment of sight deposits and prepayments, and the problems at Fannie Mae and Freddie Mac in the US, well illustrate. We need to ensure that the rules themselves do not result in perverse behaviour as people try to find a way around them.
## The way forward
## So where do we go from here?
All of the above factors have complicated the debate on IAS 39. And in my view, the only way forward is first to take a few steps back, before work starts on re-engineering the standard. I believe that many of those interested in the debate would be supportive of this approach.
We perhaps need to look again at the process for considering, and reaching a conclusion about, the fundamental 'design parameters' some of which I have touched on today. This is no small task, but if agreement could be reached on the fundamentals, these could be stated as clear principles in a preamble to any new standard, to guide the more detailed requirements. And it might enable the widely-held vision of a less detailed and prescriptive standard to be realised, to replace the hundreds of pages of complex rules.
To address this, however, raises three questions: (i) On what is early consensus necessary? (ii) How should the transition be approached? And (iii) What does this mean in institutional terms?
## Early consensus
Early agreement seems to me particularly important in three areas.
The first is to review the conventional understanding on who the accounts are for, and clarify how the different interests of different stakeholders are to be met.
The second is to agree where the standard should be placed along the spectrum between historic cost and fair value measurement. A particularly difficult conceptual question in this area is: precisely what economic meaning is there in so-called fair valuation of instruments in which (a) there is little or no secondary market, or (b) there is a market, but a clear intent to hold an instrument to maturity?
Finally, the implications of any increase in measured volatility will need to be assessed. Here, I see a need for a process of communication and education by companies and others to shareholders, analysts and other interested parties, which should form part of the transition to any new standard.
## Transitional considerations
Once the above objectives have been agreed, it will be important to articulate where we are, and where we want to be, and then to identify the optimal migration path. Is it to be a gradual transition or a big bang? Such decisions would need to consider: the rapidly moving environment of the financial industry, and regulatory techniques; how to ensure comparability is maintained; and the costs, especially in system terms, of alternative migration paths.
## Procedures for setting international accounting standards
A final issue concerns the procedures for setting accounting standards. Concerns have been expressed by some about whether public sector interests, in particular, are adequately represented in the process. There is of course a clear public interest in meaningful reporting and adequate disclosure.
I am sure that standard setters in this, as in other areas, are alive to these issues. Equally they have and in my view quite rightly - a concern to avoid, or at least minimise, the influence of overtly political factors. In considering the future of the IASB's governance, I am sure that the Trustees will be reflecting carefully on the issue of what constitutes legitimate public policy interests and how best to respond to them.
## Concluding Remarks
To sum up, the importance of meaningful and reliable accounting information is clear. It reflects the growth of publicly-traded markets, coupled with rapid growth in innovation and complexity. The potential for opacity has therefore increased, so it is essential that accounting standards are clear and capture the economic substance. This is important for the various stakeholders, and for financial stability generally.
It is also important that the actual processes for developing accounting standards command widespread support from companies, investors, creditors and the public authorities. I have set out a few steps that might be of relevance in achieving this.
On IAS 39 specifically, it would seem to me sensible to try first to reach consensus on the fundamental issues I started with before moving to detailed drafting of a new accounting standard. This will require a continuing effort on the part of standard setters to engage all interested parties, flexibility from everyone involved and a high level of transparency within the process. A lofty goal perhaps, but important if the benefits are to be secured and we are to avoid the cacophony of today.
BIS Review 71/2004
In conclusion, although I have focused on controversies in accounting, we should recognise that we may be on the threshold of a truly significant step forward in accounting practice. From next year, companies throughout the European Union will be using the same set of high quality, international accounting standards. And there is the tantalising prospect of convergence between international and American standards. The challenge now is to realise the full potential of these changes.