Fv and procyclicality draft jim dillon version 1.0
1. DRAFT
Preliminary Thoughts on
Fair Value Accounting & Dampening Procyclicality
August 21, 2010
While the key medium term drivers of the financial crisis were, in my opinion, excessive leverage, the rating
agency process, housing policy, funding maturity mismatches and misaligned incentives, what almost pushed the
market over the edge was the procyclicality that turned a very bad situation into an extreme liquidity crisis.
In Dodd-Frank there are no specific mechanisms or policies to thwart procyclicality. (Section 616 of the bill
requires that federal banking regulatory agencies “…shall seek to make such capital requirements countercyclical,
so that the amount of capital required to be maintained by a [holding company or depository institution] increases
in times of economic expansion and decreases in times of economic contraction, consistent with the safety and
soundness”, however, it’s not clear what form or substance those requirements will entail.)
On July 16 the BIS released for comment “Countercyclical capital buffers: exploring options” which proposes that
capital levels be increased when there is “excess aggregate credit growth” (based on, for example, the trend of
the credit-to-GDP ratio) which, in the judgment of regulators, is contributing to potential systemic risk.
In effect this is a build up of “reserves” in pre-bubble times (note: it’s not just in ordinary “good times”, it’s when
credit growth is deemed to be “excess”) which would be released at the onset of a financial crisis.
The problems I see with this BIS approach include:
The decision by regulators to increase capital requirements may be subjective (“Overall, our empirical
analysis indicates that a fully rule-based mechanism may not be practicable at this stage. Some degree of
judgment, both for the build-up as well as the release phase, seems inevitable.”) and may be strongly
resisted (politically and otherwise) by banks who are looking to grow further during an expansion –
whether credit growth is “excessive” or not.
It’s also at the individual discretion of regulators within each national jurisdiction.
Therefore the US could adopt it while the UK doesn’t despite similar growth in both
economies, putting the US at a competitive disadvantage – at least over the short term.
As proposed, it’s also a system-wide constraint within each country. Therefore a bank
which has been prudent in its underwriting and risk-management would have the same
requirement to increase capital as its riskier peers.
In my opinion it would be more effective if the mechanism to dampen procyclicality was non-judgmental – based on
the current level of our political discourse, things may have to get pretty bad before the proverbial punch bowl is
take away…. and by then it may be too late.
FAIR VALUE ACCOUNTING
BIS WORKING PAPERS No 317 Countercyclical capital buffers: exploring options July, 2010
”Accounting conventions can have a first-order effect on the dynamics of financial strains and on the interaction
between losses, asset quality and liquidity constraints.”
Last week I attended a presentation by the New York State Society of Certified Public Accountants on
“Understanding the FASB's Financial Instruments Proposal” which prompted an idea for a possibly more efficient
approach for dampening procyclicality.
However, I should make two disclaimers at this point:
2. 1. I’m not an accountant; and
2. while I’m very concerned about Fair Value accounting increasing the risk of a downward procyclical spiral,
there are a few of elements in the FASB proposal that may dampen the procyclical elements of marking
assets to market including:
the proposed Fair Value adjustment of Liabilities and
the probable migration to Level 3 valuation for many assets.
As a number of market observers have commented, Fair Value accounting is inherently procyclical. In particular,
Fair Valuing of assets under certain extreme circumstances can lead to a contraction of capital for some market
players, forcing additional asset sales which depress asset values across the broad market, producing a negative
feedback loop and ultimately a liquidity crisis.
Fair Value accounting, as currently defined, can also increase the likelihood and severity of a future asset bubble
by immediately taking Fair Value gains into capital, leading to excessive credit creation at a time when risk
aversion is typically very low.
Like many other regulations and methodologies (ie: VaR), Fair Value works well 90% of the time. But, as we move
into the tails of the distribution, there is an increasing likelihood that adverse unintended consequences can occur.
Therefore one question might be: can Fair Value be modified where it still works well in the 90% range but is more
effective/less damaging in the tails?
A related observation might be that Fair Value and accounting in general are microprudential tools while the major
risks we face are systemic. I would suggest that when there is a conflict between micro and macroprudential
objectives, macroprudential considerations should prevail – but every effort must be taken to ensure that the
microprudential perspective is also robust and it adequately protects investors, consumers and other stakeholders
of individual financial institutions.
MY CONCEPTUAL PROPOSAL
First, I’m not suggesting we return to/stay with amortized cost accounting (Note: the most contentious element of
FASB's Financial Instruments Proposal is the treatment of loans; I’m not distinguishing between loans and
securities in this paper). Unquestionably there is a need for financial statements to reflect changing economic
realities. However, it’s not clear that a quarter-end snapshot valuation is the optimal approach.
Fair Value accounting inherently presumes the market is efficient. While this may be true over the longer term, as
we have recently learned, market efficiency – as difficult as that is to define – can severely break down in the short
term, especially during a liquidity crisis.
Therefore, a more representative asset valuation approach would be to use an historical moving average (subject
to certain constraints). This would preserve the benefits of transparency since the current asset values that feed
into the moving average could either be explicitly listed in the footnotes or could certainly be estimated based on
the change in the moving average.
Additional analysis would need to be done to backtest this approach. In some respects this is a classic cost vs.
benefit analysis. The benefit is assumed dampening of the procyclical impact arising from overshooting markets
while the cost is a possible delayed recognition of a major structural change in asset values which is not
temporary.
The problem with the current approach is the potentially exaggerated impact on capital when we are in the tails of
the distribution. This is due in large part to the fact that GAAP policies and procedures are imbedded in the
regulatory process (in a similar way that ratings from the Nationally Recognized Statistical Rating Organizations
are also imbedded in banking regulations).
3. Like so much of finance – and life – behavioral reactions matter a lot. Even if the entire market knows what the
underlying inputs are (ie: “current” asset values), how the inputs are used under FASB regs to ultimately arrive at
a measurement on capital can be significant.
Reportedly FASB is also very much aware of the behavioral aspect of accounting. According to one industry
source, the driver behind having proposed Fair Value changes flow through the income statement and not just
OCI is to give greater visibility and emphasis in the minds of investors to the Fair Value adjustment. (It would
seem that FASB, in addition to keeping score, wants to influence how the score is used and interpreted.)
It’s also interesting to note that under GAAP, the Fair Value of an asset is defined as “…the amount at which an
asset could be bought or sold in a current transaction between willing parties, other than in liquidation”. Therefore
the valuation doesn’t apply when the institution is insolvent – just when you really need it to be accurate.
In addition, using a weighted average also may reduce the benefits to certain players of manipulating the market
for their own gain.
I have not attempted to quantify the optimal period over which asset values should be averaged. As a
conservative starting point, I would suggest that the default term should be four quarters. An exception would be
firms with very short term liabilities; in the extreme, if a firm were funded 100% with overnight CP, asset values
should not be averaged.
Another computational issue would be how to reflect the values of a dynamic portfolio where a firm may only have
held some assets for a few weeks or months.
A variation on my historical four-quarter averaging would be a concept called Mark-to-Funding; this would be
forward looking where there are two alternative prices for an asset: today’s market price and the discounted
present value of the future earnings stream. In normal times these two prices are nearly the same. But in a
liquidity crisis the market price falls substantially below the present value. Under a mark-to funding approach, a
weighted average of the market price and present-value price would be used where the weights would depend on
the weighted average maturity of the institution’s funding.
I find this approach appealing as well. However one reason to prefer the historical moving average computed
over a defined time period such as four quarters is that it’s more transparent and less computationally complex
than Mark-to-Funding.
SUMMARY
Similar to the BIS Working Paper, this is not a finalized proposal; its intent is to start a conversation.
I’m suggesting that we should keep capital requirements rules-based (ie: not judgmental) and slightly adjust the
accounting methodology for computing capital, against which the capital regulations are applied.
Of course, the opposite approach would be to maintain Fair Value as-is (even when we’re operating in the tails)
and have the regulators lower (or otherwise modify) capital requirements depending on the circumstances. While
quantitatively these approaches might be structured to achieve similar numerical results re: minimum capital, I
fear there would be a greater chance of a negative behavioral reaction to the more volatile Fair Value approach
It’s instructive to consider a quote from William Isaac, former Chairman of the FDIC in Congressional testimony
before the House Committee on Financial Services on March 12, 2009 “If we had followed today’s approach
during the 1980s, we would have nationalized nearly all of the largest banks in the country and thousands of
additional banks and thrifts would have failed. I have little doubt that the country would have gone from a serious
recession into a depression.” Unfortunately, it often seems that academic theory inappropriately dominates
pragmatic considerations
4. Finally, I will acknowledge that modifying accounting concepts and practices will be difficult to get enacted. But, if
such a modification reduces systemic risk and is balanced and rational it should merit serious consideration. As
also noted in the July, 2010 BIS Working Papers No 317, “….reducing the sensitivity of the minimum capital
requirement is an important element of a credible countercyclical buffer scheme.”
PLEASE SEND COMMENTS ON THIS CONCEPT DRAFT TO:
Jim Dillon at jvd422@gmail.com