September 14, 2009

Regulatory Perspectives on the Changing Accounting Landscape

Governor Elizabeth A. Duke

At the AICPA National Conference on Banks and Savings Institutions, Washington, D.C.

Good morning. I want to thank the AICPA for inviting me to speak at this year's Conference on Banks and Savings Institutions. Over the years, this conference has served as an important forum for the exchange of views on regulatory matters, emerging accounting issues, new accounting standards, and new auditing considerations.

This year's conference comes at a critical time: Regulators and policymakers around the world are now evaluating changes to practices and structures to address weaknesses revealed by the recent financial crisis. At the same time, accounting standard setters are proposing changes that will, in turn, affect regulatory standards. These changes, along with those made by regulators and policymakers, will help determine the speed and the durability of the global financial system's revitalization.  Further, the accounting and regulatory changes made now will help shape future business models for financial institutions and thus influence credit availability. It is important to ensure that these changes facilitate, not hinder, the decision-making processes that support financial intermediation and economic activity.

I would like to spend my time with you today talking about current and proposed accounting standards that will, in my opinion, have the greatest impact on the operation and supervision of the U.S. banking system. Before I begin, I should define for you my perspective on these matters. Given my background as a community banker, I feel it is crucial that an accounting regime directly link reported financial condition and performance with the business model and economic purpose of the firm. It is difficult for me to comprehend the value of an accounting regime that doesn't make that link.

As a regulator, I focus on the viability of individual financial institutions and the financial system as a whole. To be frank, it has been frustrating to try to assess that viability when the value of an asset is based on the nature of its acquisition rather than the way in which it is managed or the way in which its economic value is likely to be realized.

And finally, as an economic policymaker, I fully understand the integral role that financial institutions play in the overall performance of our economy. Equally important are the roles played by those that trade and those that lend and by the securitization markets. And I believe a legitimate case can be made for differences in accounting treatment between them to facilitate financial intermediation and economic activity. You might have guessed by now that I would like to talk primarily about fair value and loan reserve accounting.

I should also remind you that the views I express are my own and do not necessarily reflect the thinking of my colleagues on the Board of Governors or Board staff.

Relevance and Reliability
I think it might be useful to discuss current and proposed accounting standards by first considering the concepts of relevance and reliability.

In terms of relevance, the measurement principle should reflect the manner in which entities actually use financial instruments. In this regard, the business model and risk-management approach taken by the reporting entity--as well as the way in which the value of the instrument itself is likely to be realized--should be factored into the measurement determination.

If the business model is predicated on the trading of financial instruments for the realization of value, or other strategies that essentially focus on short-term price movements, then fair value has relevance. In the trading business model, reporting fair value focuses risk management on short-term price movements and in most cases incentivizes management to define the organization's risk appetite and to mitigate risk through hedging or other means. Fair value also incentivizes the entity to raise and maintain capital at a level sufficient to cover the price volatility of its assets. For example, if the business model is an originate-to-distribute model, then fair value has relevance.

In contrast, if the business model is predicated on the realization of value through the return of principal and yield over the life of the financial instrument, then fair value is less relevant. Consider, for example, a bank that finances the operations of a commercial enterprise. The realization of value will come from the repayment of cash flows. Risk management is based on an assessment of the borrower's creditworthiness and the entity's ability to fund the loan to maturity. In this case, the accounting should incentivize the entity to maintain sufficient funding to hold the instrument to maturity and to hold a sufficient amount of capital to cover potential credit losses through the credit cycle, preferably in a designated reserve. Indeed, the use of fair value could create disincentives for lending to smaller businesses whose credit characteristics are not easily evaluated by the marketplace.

Admittedly, some have used the business model argument to manipulate accounting results. But the actions of those entities do not diminish the relevance of the business model to the measurement principle. Indeed, over time if the valuation model is not relevant to the business model, the business model itself is likely to change. Rather, the lesson to be learned from such manipulation is that we--preparers, users and auditors of financial statements--need to be vigilant in evaluating actual business practice, and restrict the use of particular measurement principles to the relevant business models.

To this end, safeguards should be implemented to eliminate a firm's ability to overstate gains or understate losses by switching back and forth between business models or by reclassifying assets from one business segment to another. For example, from a regulatory perspective, assets in a financial institution's liquidity reserve, by their nature, imply utility through sale and, therefore, should be valued at market price.

In terms of reliability, the measurement principle should reflect the ability of all types of entities to calculate a value within a reasonable range of confidence throughout the economic cycle and the life of the financial instruments. There is a good deal of reliability when the fair value of a financial instrument is observable in an active market.You accountants refer to these observable inputs to fair value as level 1. As you leave the active markets and get into the so-called level 2 and level 3 inputs to fair value measurements, an entity's ability to reach a consistent fair value or an estimate of fair value within a reasonable range of values for a particular financial instrument significantly diminishes. As the recent financial crisis has shown us, a financial instrument's fair value can vary widely among entities in similar markets. And the existence of wide variability in valuation models makes comparisons between entities difficult if not suspect.

The reliability of amortized cost is not as questionable. Amortized cost simply is the amount paid to acquire a financial asset, adjusted for any unaccreted discount or unamortized premium. All entities calculate amortized cost using the same formula. Of course, amortized cost is not a panacea. Entities purchase assets at different times and the timing of expected cash flow changes can result in different measurements for the same financial instruments. If the receipt of future payments is in doubt, impairment must be estimated.

The use of a reserve for credit losses helps distinguish between contractual amounts due and payment uncertainty created by economic or borrower-specific conditions. Current accounting standards permit credit reserves only for losses likely to be realized in the short term. Lenders regularly adjust credit standards to achieve life of loan profitability given through-the-cycle estimates of credit loss. They should similarly be able to estimate through-the-cycle reserves as reliably as short term likely losses.

The Stress Test
Accounting treatment issues are critically important in the regulatory evaluation of financial institutions' safety and soundness. The Supervisory Capital Assessment Program (popularly known as the stress test) provides a window into the likely future of bank supervision as well as the accounting issues encountered by regulators in evaluating capital adequacy. The stress test was a simultaneous, horizontal review of the 19 largest financial institutions in the United States. The review was led by the Federal Reserve, but conducted jointly with other federal banking regulators. In essence, we focused on three key pieces of information--pre-provision net revenue, potential losses, and final equity capital.

Pre-provision net revenue and potential losses were estimated under two different economic scenarios, a baseline scenario and one that was more stressful. Importantly, losses in the trading book were estimated using indicators of financial stress and market volatility while losses in the loan book were estimated using economic indicators to assess probability of default and projections of asset prices to estimate loss severities. In this way, the stress test held to the principle of relevance, while statistical history provided us with some measure of reliability. The stress test was a forward-looking exercise and losses were estimated over a two-year horizon: 2009 and 2010.  Given that the two-year horizon was likely to be one of the most stressful time periods in our history, the credit loss estimates would come closer to approximating through-the-cycle losses than the reserve amounts calculated under current accounting standards. 

Three of the banks that participated in the stress tests had acquired significant loan portfolios through business combinations.  Given the elimination of the pooling-of-interest method, the advent of AICPA Statement of Position 03-3 (SOP 03-3) accounting,  and the prohibition on presenting an allowance for any acquired pools of loans, it was very challenging to determine the amount of credit risk that was already captured in the carrying amount of these acquired loans.  Essentially, we had to determine the level of credit risk that would have been present under the pooling-of-interest method and adjust those amounts under the economic scenarios.  Adjusting our measurement of these portfolios for pre-provision net revenue, loan loss, and capital proved quite challenging.  I would expect it to be similarly challenging for analysts and investors to make similar adjustments in order to set benchmarks and compare performance, of an individual bank over time or between two or more banks.  This is a case where I believe we have experienced a reduction in transparency and have lost valuable credit information on acquired loans.  

I recognize that the Loan Loss Disclosures project at the Financial Accounting Standards Board (FASB) is an attempt to provide some of this information by (1) requiring more disclosure of credit risks generally and (2) requiring the disclosure of the total carrying amount and the total unpaid principal balance of impaired FAS 114 loans for both loans with and without a related allowance for credit loss.  However, reserve coverage and loss ratios are calculated using amounts shown in the financial statements.  Allowing different treatment of portfolios acquired through business combinations and originated portfolios will make comparisons difficult and will make norms and averages less meaningful over time.

In calculating capital requirements, we treated off-balance-sheet entities as if they were carried on the balance sheet.  We consulted with the FASB for the most current thinking about what would be consolidated under the accounting rules they were finalizing.  

To recap, in the stress test we analyzed 19 financial intermediaries engaged in a mix of trading and lending businesses.  We conducted this analysis on behalf of the U.S. taxpayer who had become a primary investor in these institutions.  And we analyzed them in a way that, in my opinion, best characterizes the risk and performance of the entities:

  • We evaluated trading assets on a fair value basis;
  • We evaluated loan assets on the basis of expected credit loss through an adverse cycle;
  • We evaluated assets based on the way they were managed rather than the way they were acquired; and
  • We included assets held both on- and off-balance sheet.

And finally, we used this analysis to estimate the capital buffer needed to protect the entity in an adverse scenario. We published both our methodology and our findings. While I am normally firmly opposed to making public confidential supervisory information, in this case, we were able to respond to high levels of uncertainty and speculation by publishing our findings. Marketplace confidence seemed to rise upon the publication of results and the subsequent successful capital raises by the firms.

Now that some measure of confidence has been restored and financial strains are receding, it is time to turn our attention to the lessons learned in the crisis. And to ask: How can we prevent future crises? Accounting standard setters, regulators, and policymakers around the world are discussing and proposing preventative measures. Now the challenge lies in integrating those changes smoothly and seamlessly.

Accounting Issues Identified During the Crisis
A number of groups have analyzed the role that accounting played during the crisis and have made recommendations to strengthen accounting standards and the standard-setting process. Although accounting was not the cause of the financial crisis, certain accounting measures, such as the use of fair value accounting for illiquid financial instruments and the impairment model for loans and debt securities, have drawn considerable attention. Throughout the crisis, there were considerably fewer actual market transactions available for use as reference prices for fair values. At the height of the crisis, there was such little market activity that serious consideration was given to abandoning the use of fair value for a period of time.  Even now, the debate continues about whether fair value is the appropriate measurement attribute for debt securities and other financial instruments, particularly in less active markets.

Similarly, there were serious concerns about the approaches used to determine the impairment of loans and certain debt securities during the crisis. Some argued that the approaches available inhibited firms from recognizing credit losses on loans sooner and artificially required the recognition of losses on debt securities. Standard setters responded by providing guidance on the determination of fair values in the stressed market environment and the determination of financial instrument impairment. This was a very challenging period for financial statement preparers, users, standard setters, and regulators.

FASB and IASB Approaches Under Consideration
Standard setters are now actively engaged in the discussion of the appropriate accounting principle for measuring financial instruments. Currently FASB and the International Accounting Standards Board (IASB) are pursuing measurement approaches that diverge in important ways.

FASB's approach would measure all financial instruments (assets and liabilities) at fair value through the income statement or other comprehensive income. This would mean that an entity's business strategy for investing in securities and originating loans would not be taken into account. FASB is willing to disclose the amortized cost of these financial instruments on the balance sheet along with the fair value, but there is a catch. The FASB approach would modify the income statement to include all changes in fair value whether or not they are included in other comprehensive income. Reflecting market value fluctuations of all assets through the income statement would significantly increase the volatility of reported bank earnings, likely leading to changes in risk-management practices. At the extreme, this approach could incent all financial intermediaries to adopt a trading or investment banking business model.

On the other hand, IASB's approach would measure financial instruments at amortized cost if they have characteristics of a basic loan and are managed on a yield basis. Under the IASB approach, characteristics of a basic loan are fairly narrowly defined. For example, only the most senior tranche of an asset securitization might qualify for amortized cost. Similarly, any loan with any unusual provision might not qualify for amortized cost.  This approach would measure all other financial instruments at fair value predominantly through the income statement. There is a narrow exception for fair valuing certain strategic equity investments through other comprehensive income.  Consistent with the FASB approach, the IASB approach would modify the income statement to include all changes in fair value whether or not they are included in other comprehensive income.

By now, you have gotten the picture that both of the approaches under consideration would constitute a significant departure from current practice. Both of these proposals raise a number of concerns for me:

  • From the standpoint of relevance and reliability, the FASB and IASB approaches would not accurately reflect the traditional commercial banking model. Indeed, the imbedded incentives would actually favor "originate to distribute" rather than "originate and hold" lending models.
  • Second, we have very little actual experience in fair valuing liabilities. Using fair value for liabilities introduces a new set of incentives and risk exposures for management.
  • Thirdly, no market currently exists for non-government-guaranteed, small business loans. The lack of fair value information for these types of loans could actually discourage small business lending.
  • Fourth: Smaller banking companies likely will incur substantial costs and experience great difficulty in applying the new standards. But will financial statement users see any real benefit?
  • And finally, the two Boards are planning to exchange views and work products, but are not duty-bound to achieving a single converged standard. If the approaches are implemented along different timelines in the United States and abroad, they could bring the two sets of standards further apart and possibly incent some governmental bodies to mandate an approach in order to level the playing field.

Retaining Both Lending and Trading Models
My preference would be for standards that recognize both lending and trading business models for financial intermediation even when they exist within the same firm. My wish list would include the following:

  • Trading assets shown at fair value with market value gains and losses recognized through the income statement;
  • Assets held for secondary liquidity shown at fair value with market value gains and losses recognized in the capital account through other comprehensive income;
  • Assets held to maturity and managed for yield and return of principal over time shown at amortized cost with a reserve reflecting life-of-loan or through-the-cycle potential credit losses; and
  • For business combinations, identical accounting treatment for acquired assets and similarly managed assets on the acquirer's balance sheet.

Regulatory Changes
In terms of regulatory changes, our current regulatory capital framework needs to be revised to ensure that banking organizations have a level of capital sufficient to facilitate lending, while also ensuring safe and sound operation throughout the economic cycle. Work is underway to develop an approach that would allow banks to retain more capital in good economic times and to allow this excess or buffer to be reduced as the economic cycle worsens. The goal is to have a level of capital that is sufficient to support lending, while maintaining safety and soundness. The challenge is to develop an appropriate target for this excess amount and to identify the right economic trigger for determining when this excess should be reduced. This is a delicate balance.

In addition, the elements that we consider to be tier 1 capital in our current framework need to be revised. Since our framework starts with components of equity capital as measured under generally accepted accounting principles (GAAP), we are carefully evaluating every element of regulatory capital that is treated differently in regulatory capital than in GAAP. For example, GAAP equity includes accumulated other comprehensive income (AOCI) and the current regulatory framework neutralizes the impact of certain items in AOCI such as unrealized holding gains and losses on available-for-sale debt securities.

And finally we are coordinating capital standard setting with our counterparts in other countries. To the extent that GAAP and accounting standards in those countries are different, our capital definitions may also differ.

Impact On Securitization
Finally, I'd like to offer a quick word of caution on accounting for off-balance-sheet items and the future of securitization markets.

Our financial system has become dependent upon securitization as an important intermediation tool. During the crisis, securitization markets ground to a halt. The Federal Reserve's Term Asset-Backed Securities Loan Facility (TALF) has helped restart activity in some markets, such as securities backed by auto loans or credit card receivables. But the CMBS market is still very weak and the market for newly issued, private-label RMBS remains closed. And although the TALF has been successful, it is a short-term facility that was only intended to give the markets and policymakers time to restructure the securitization model to make the securitization markets more viable going forward.

The recent G20 agreement calls for a retention of risk, or "skin-in-the-game" approach for asset securitizations.  It also calls for higher capital standards and a leverage ratio for all banks. If the risk retention requirements, combined with accounting standards governing the treatment of off-balance-sheet entities, make it impossible for firms to reduce the balance sheet through securitization and if, at the same time, leverage ratios limit balance sheet growth, we could be faced with substantially less credit availability.  I'm not arguing with the accounting standards or the regulatory direction. I am just saying they must be coordinated to avoid potentially limiting the free flow of credit.

We will learn more about the impact of the new accounting rules on securitization activities as banking organizations implement the new standards. We will also learn more about the impact of our regulatory capital regime on securitization activities as we evaluate the responses to our proposed changes to the regulatory capital guidelines. In the past, accounting rules and regulatory capital guidelines have been drivers for how the securitization model has been structured. As policymakers and others work to create a new framework for securitization, we need to be mindful of falling into the trap of letting either the accounting or regulatory capital drive us to the wrong model. This may mean we have to revisit the accounting or regulatory capital in order to achieve our objectives for a viable securitization market. A healthy economy needs an array of tools for financial intermediation and we need to be careful not to be overly punitive to this particular tool. We just need to focus on providing the appropriate incentives, oversight, and accountability.

The financial crisis has certainly highlighted the need for a safe and stable financial system. To promote confidence and attract capital to the system, we need financial statements that provide maximum insight into the financial condition and risk positions of financial intermediaries. We need supervisory oversight and regulatory constraints, such as regulatory capital, that provide safeguards and incentives that support our objectives of prudent provision of credit and sustainable economic activity. And the accounting and supervisory frameworks need to recognize and support all viable forms of financial intermediation regardless of whether it occurs in the traditional lending model, the trading model, the securitization model, or some other business model.  In this regard, I believe it is important to show in the statements themselves the numbers needed to construct ratio analysis between firms or of individual firms across time.

Finally, I believe that accounting standard setters, regulatory bodies, and lawmakers have a vested interest in working together to ensure the oversight mechanisms, reporting frameworks, and other elements of the revitalized financial system operate in a manner that is both stable and efficient.

Last Update: September 14, 2009