Accountants can be creative people. Bankers can be sneaky people. Put the two together, and funny things happen.

Bloomberg's Jonathan Weil recently penned a fantastic article highlighting the flexible rules banks use to account for and value loans. Weil focuses on an accounting rule that lets banks carry loans at, frankly, whatever value management sees fit. By declaring loans as "held to maturity," banks can value loans at their amortized cost -- an assumption that can indicate the loan will be paid off in full. Only when management decides losses are probable do writedowns occur.

The alternative to this is fair value accounting, whereby loans are valued at current market prices -- the same way you and I value our stock portfolios. 

For the first time ever, banks now have to footnote loan values using both accounting measures on a quarterly basis, rather than just annually. Even if a bank uses hold-to-maturity accounting when it reports its books, it has to disclose the "fair value," or market value, of those loans as well.

Not surprisingly, as Weil shows, the disparity between the two accounting procedures can be quite large:

Bank

Stated Value of Loans

Fair Value of Loans

Difference

Tier 1 Capital

Bank of America (NYSE:BAC)

$886 billion

$822 billion

$64 billion

$191 billion

Wells Fargo (NYSE:WFC)

$799 billion

$764 billion

$35 billion

$103 billion

Regions Financial (NYSE:RF)

$91 billion

$68 billion

$23 billion

$13 billion

SunTrust Banks (NYSE:STI)

$120 billion

$106 billion

$14 billion

$19 billion

KeyCorp (NYSE:KEY)

$68 billion

$60 billion

$8 billion

$12 billion

Sources: Company filings, as of Q2 2009.

These are seriously large differences -- depending on what accounting method you use, Regions Financial is either doing great, or about to blow up.

Now, there is a decent argument for why banks should be allowed to ignore market values. If they hold the loan until maturity, all that matters is the long-term cash flows from that loan, not the current market value. And if banks were required to mark everything to market, capital requirements would be all over the map, going from overcapitalized to the brink of default, depending on the mood of the market. This is especially true during times like the post-Lehman-Brothers-failure week, when it's safe to say that nothing was accurately priced. A world where everything is always marked to market can be as dangerous as one in which nothing is.

Still, there are reasons to be suspicious of this accounting leniency:

  • Most of the time, the market value of loans should reflect a good estimation of intrinsic value. The difference between a loan's market value and the carrying value a bank designates is nothing more than accountants disagreeing with the wisdom of the market.

    That may sound like Chicagonian "no such thing as an inefficient market" extremism, but seriously: After the meltdown of the past two years, giving banks the benefit of the doubt seems insane. Banks have proven utterly incapable of valuing assets and pricing risk. Why we should suddenly assume they've mastered this field, and take them at their word, is beyond reason. Given the choice of believing the collective opinion of the market, or trusting a banker whose bonus is contingent on reporting desired results, the choice should be obvious.
  • Banks can, and do, log gains when the market value of default risk rises on their debt. The assumption is that when a bank's debt falls in value, the bank could theoretically repurchase it at a discount, calling the difference between par and the repurchase price "income." Even if the bank doesn't repurchase the debt, it can still report the difference as a sort of magical gain. Such accounting shenanigans have become big hits with banks like Citigroup (NYSE:C) and B of A in recent quarters.

The absurdity here is impressive: Banks can simultaneously say "We choose to ignore market values on our assets, because we don't feel the market has a clue what they're really worth," and "We choose to take the market at its word and mark our liabilities at whatever it says they're worth." When you're trying to meet your quarterly numbers, this setup can be quite convenient.        

If anything, these accounting discrepancies should drive home an important point: Banks' books are enormously subjective. Their value is largely based on the opinions of people whose livelihoods rely on these values. If you believe in the power of incentives, that's a scary thing.

It's easy to look at banks today and say "Such-and-such is cheap because it's trading below book value!" This may be true, and it may not be. It's true insofar as you take stated book value at face value. Depending on what accounting measure you choose to use, that's no sure thing.

For related Foolishness: