The blame game

The subprime crisis would make a fantastic pulp murder mystery: bodies of funds, banks and senior executives are piling up. Whodunnit? Perhaps it was the foolish borrowers, the duplicitous lenders, or the greedy securitisers. But it could have been the gullible investors, the naive rating agencies, or the complacent regulators and central banks. Amid all the finger-pointing, name-calling and intrigue, one group - the accountants - looked set to slip out of the back door unnoticed. That has changed in recent months, with mammoth full-year and first-quarter losses focusing attention on what exactly those numbers mean, how they are calculated, and whether the accounting standards played a part in exaggerating the scale and accelerating the speed of the crisis.

Wayne Upton, director of research at the International Accounting Standards Board (IASB), is not too upset: "Harry Truman used to have an expression - if you want a friend, buy a dog. Whenever we have a crisis, everyone looks around for someone to blame, and that's the way a standard-setter's life always is."

That makes it sound like any gripes about accounting are just a case of scapegoating - but try telling that to the head of credit portfolio management at one large US bank, who says his firm is sitting on credit default swaps that are referenced to illiquid leveraged loans. "I'm required to fair value them and I have no way of doing it. There is no market for these things - none. So how do you do it? It's just a guess," he says.

Or tell it to AIG's former chief executive, Martin Sullivan, who - prior to his resignation in mid-June - had seen his company report an $11.1 billion writedown in March, and railed against accounting standards, contending that the company's own estimates reckoned the worst-case loss on those assets was more like $900 million.

These cases illustrate two distinct criticisms. Fair-value accounting works by reporting changes in the market value of financial instruments as though they were profits or losses. The first complaint is that when markets are illiquid and instruments are not trading, it can be difficult to work out what their fair value is. As a result, reported values are at best a hazy estimate and, at worst, are cooked up or manipulated by reporting institutions to suit themselves. The second criticism is more serious, and claims market prices become detached from fundamental values during periods of illiquidity. In other words, fair-value accounting forces institutions to report exaggerated losses, which fuel market panic, creating a destructive downward spiral of vanishing liquidity, falling prices and ever-greater writedowns (Risk November 2007, pages 23-25).

In its most recent half-yearly Financial Stability Report, published at the start of May, the Bank of England lent some credence to this second criticism. By analysing the probability of default in subprime mortgage loans, the central bank calculates that securities containing these assets can be expected to eventually lose $170 billion over the next five years. In contrast, the report points out the family of indexes that banks have used to derive market values during the crisis - the ABX - implies losses of around $380 billion, or more than twice the figure produced by credit analysis. The report concludes that "using a mark-to-market approach to value illiquid securities could significantly exaggerate the scale of losses that financial institutions might ultimately incur. It will exaggerate to an even greater extent the potential damage to the real economy these losses might inflict."

Standard-setters do not buy this argument. Their first retort is to ask why - if the risk of losses on mortgage-backed securities is so exaggerated - banks are not hoovering up the assets with a view to making some hefty relative-value gains. Their other response is more principled: fair value provides an objective standard for companies to follow. Ron Lott, research director at the US standard-setter, the Financial Accounting Standards Board (FASB), says: "Fair value is what it is. It's a market price, or an estimate of a market price, based on what people in the market are willing to pay on that day. It isn't necessarily some real, intrinsic, inherent value."

But should this real, longer-term value be the goal of the accounting standards? The IASB's Upton thinks not: "This argument has been out there a long time and it's usually taken up by a bank that has had a big writedown and wants an excuse. Essentially, they say 'I'm not going to sell into this market. I don't want to sell into this market. If I can just hold on, it will be OK'. But without fair-value information, the people to whom the bank is making that claim - the analysts and investors - have no idea whether it really has that kind of staying power in a market that is falling and falling."

If banks genuinely believe market prices are producing wildly overstated writedowns, they should simply say so in their financial statements. Upton argues there is nothing preventing institutions from adding a commentary to their numbers, explaining why their assets are worth more than the market currently thinks and inviting analysts and investors to use their own judgement about whether the company can hold on to those assets until prices rebound.

Allies

The standard-setters may have been on the defensive in recent months, but they have allies: "Almost every financial analyst anybody has bothered to ask," says Upton.

Pauline Wallace, a partner at PricewaterhouseCoopers (PwC) in London, is also in favour of the current rules: "Fair-value accounting standards mean you are pinning bank earnings back to a market. Just because we've had some big shifts in numbers this year, it doesn't mean the process is broken and, frankly, nobody has come up with a better way of presenting the underlying cashflows and the risks in assets. I'm a believer."

So are some bankers, who see a flip-side to the argument that fair value exaggerates losses and has accelerated the spread of the crisis. During a webcast organised by the FASB in early June, Matt Schroeder, global head of accounting policy at Goldman Sachs in New York, argued that by carrying asset price deterioration into profit and loss, fair-value reporting has actually enhanced financial system stability: "There are real economic losses here, and to say they're not happening is not the case. One of the things fair-value accounting has forced banks to do is take those losses and raise capital more expeditiously than the industry used to."

Tushar Morzaria, chief financial officer for JP Morgan's investment banking business in the Europe, Middle East and Africa region, makes the same claim and uses the spectre of Japan's 'lost decade' to illustrate the alternative: "In that situation, banks were sitting on stacks of non-performing loans that were reported at their historical value, and it just dragged on and on until finally being flushed out of the system. During the past 12 months, you've seen banks taking much quicker and more appropriate action to shore up their capital position, and I think it's a healthy thing for investors and banks to face the truth in these circumstances."

But truth can be elusive - and that is where the first criticism comes into play. With the market for many structured credit instruments locked down or trading only occasionally - and then at huge discounts - banks have been forced to estimate fair values, and doubts have grown about how accurate these estimates are. For their part, banks complain they are being compelled to wildly underestimate the value of their assets. Conversely, some cynics in the analyst community charge that banks are using the opacity of the estimation process to their own ends, by overstating their asset values or, in some cases, exaggerating writedowns to get them out of the way quickly.

"I mark my credit to market, but it's not an accurate basis for accounting," says the US bank's head of credit portfolio management. "It's not a reflection of earnings - it just leads to absurdities and tremendous volatility that is not reflective of the true risk I've taken."

So, how does the process work? US accounting standards separate fair values into three types: level 1 assets, which are those where reliable, quoted prices are available from an active market; level 2 assets, which are not traded actively so some degree of modelling is required - but all the significant inputs to those models are taken from external, observable sources such as proxies, comparable assets and services that aggregate price data from individual banks; and level 3 assets, which are those for which one or more significant inputs are unobservable - in other words, they rely heavily on the bank's own estimates and assumptions. The IASB's accounting standards don't have the same, clear categorisation, but they operate on the same principles.

Currently, the problem assets at the centre of the crisis - such as collateralised debt obligations (CDOs), residential mortgage-backed securities, leveraged loans and credit default swaps referenced to any of these instruments - fall primarily into level 2. JP Morgan, for example, had just over $1 trillion in level 2 assets at the end of 2007, and only $71 billion in level 3. The huge imbalance is because the accounting rules insist on the use of observable inputs over unobservable assumptions wherever possible. So, if a bank has to value a structure containing subprime exposure, it is expected to use the ABX index or some other external value as an input, rather than rely on its own internally generated numbers. Unfortunately for banks, the sheer number of people shorting the index means the ABX has been massively exaggerating the losses expected on those structures for much of the crisis, as the Bank of England noted in May.

"The problem for banks is that although these things are still trading, they're trading at a very bizarre price," says PwC's Wallace. "It's not quite so bad now, but at one point, the ABX was suggesting anyone holding the underlying instruments was going to be paid just the next few interest payments and lose everything else. It got to a stage where it didn't reflect the underlying cashflows."

That is not all, according to the US bank's head of credit portfolio management. He charges that consensus pricing services are not capturing the actual prices seen in the market. "Everyone gets these pricing feeds. It's an accepted pricing service, so I look at what those mark at. But then I go and look at our loans and we start marking where we can actually sell them and the two values bear no relation to each other at all," he complains. "There is no real market. People have curves they submit to these services, but no-one really sees a market for these names. It's all a bit of a fiction."

If proxies are both exaggerating credit risk and failing to capture genuine market prices, can banks just ignore them and go to level 3? In theory, yes, says Wallace: "If you can demonstrate the observable input is not the right basis for a valuation, then you've got to find something that is. If you were looking at CDOs, for example, you'd have to demonstrate the underlying assets in the deal you were trying to value were very different from the assets that underlie the instruments that make up the ABX."

Raising eyebrows

In some cases, that is exactly what banks are doing - but it has raised eyebrows among regulators. Since October last year, a team of regulators chaired by Patrick Amis, head of accounting affairs at France's Commission Bancaire, has been working under the auspices of the Committee of European Banking Supervisors (CEBS) on a report about valuation challenges in illiquid markets. The paper was published by CEBS on June 18. One of the issues it raises is a concern about the use of consensus pricing services and the need for consistency in the way banks use indexes such as the ABX. Prior to the crisis, Amis says, some banks saw consensus pricing services as observable inputs, to be used in conjunction with level 2 instruments. Today, in some cases, the same inputs are being reclassified as unobservable, allowing banks to move to a level 3 treatment. "There is a strong need to clarify how fair values should be calculated in illiquid markets," says Amis, who notes that this work is currently being tackled by the IASB.

If banks have some leeway over the inputs they use, the weights they give those inputs and when they can apply their own models, could they be manipulating their reported figures? PwC's Wallace, who advises auditors that worked on bank financials during the crisis, does not think so: "I've seen no evidence of that happening. People have crawled all over this stuff more than they have ever crawled over it before, so I actually think we have better information to make those judgements than we've ever had before."

JP Morgan's Morzaria says the bank has "a very rigorous and sophisticated process to make sure we do get to our best estimate of fair value, so it's not simply a matter of a single person on the dealing floor making a judgement. It's reviewed across different departments and from different perspectives to ensure we get to the right answer." He also says he would be surprised if other institutions were trying to play with the accounting rules.

The debate about the appropriateness of fair value in illiquid markets is bound to rumble on, but Morzaria says JP Morgan is in favour of maintaining the status quo. That is a view shared by the Commission Bancaire's Amis, who says he thinks the current accounting model - with its mixture of fair-value and other approaches - works reasonably well. He adds most regulators would not be in favour of the long-mooted expansion of fair value to encompass a greater array of assets, provided the existing system prevents potential abuses. But any suspicion banks are trying to game the system could prompt standard-setters to push for a full fair-value model, Amis warns.