Financial institutions normally prefer not to have everyone think they're a shitty credit, because that can lead to doom, or MF Doom, or glitchy intimations of doom that quickly get sorted. But it can also sometimes lead to profit.
Sometimes that profit is fake, or fake-ish. When banks book a mark-to-market profit on their own credit spreads widening, that looks ... a little fake. We don't particularly object here, since it reflects a sort of economic reality, but it is probably temporary - your liabilities roll forward and eventually either you pay them off at par, in which case your DVA gains fritter down to zero through PnL, or you don't, in which case the permanency of your accounting gains are not of much interest to most people.
In any case, because it looks fake, or fake-ish, banks actually don't much abuse the privilege. Thus most of the DVA gains that banks booked last quarter were on derivatives, where US GAAP requires you to mark DVA to market, or on derivatives-looking things like structured notes where it seems more plausible than not. You don't see a lot of banks taking a lot of DVA gains on vanilla debt.
So when you have $295bn of public debt (with, I don't know, maybe a 2 year weighted average duration, whatever) and your CDS blows out by 300bps in a quarter, you don't book $18 billion of gains. You book, um, $4.5 billion. You never get to taste most of that delicious credit widening.
Now, if only there were a way for a bank to (1) get a gain on its vanilla public debt and (2) make it permanent. Like, say, this, from Bank of America's 10-Q filed today:
The uncertainty in the market evidenced by, among other things, volatility in credit spread movements, makes it economically advantageous at this time to consider retirement of issued junior subordinated debt and preferred stock. As a result of these matters, we intend to explore the issuance of common stock and senior notes in exchange for shares of preferred stock and, subject to any required amendments to the applicable governing documents, certain trust preferred capital debt securities (Trust Securities) issued by unconsolidated trust companies, in privately negotiated transactions. ... These transactions would increase Tier 1 common capital and, on an after-tax basis, reduce the combined level of interest expense and dividends paid on the combined junior subordinated debt and preferred stock. The senior notes and common stock would be recorded at fair value at issuance, which is expected to be less than the par and carrying value of the preferred stock and/or junior subordinated debt, which would result in the exchanges being accretive to earnings per common share for the period in which completed. The ultimate impact on earnings per common share is not expected to be significant for periods subsequent to the exchange and will not be known until the level of earnings per common share for the period and the exact combination of exchanged preferred stock and Trust Securities are known. We will not issue more than 400 million shares of common stock or $3 billion in new senior notes in connection with these exchanges.
It's quite a pretty trade. Note first of all that you profit by "volatility in credit spread movements" by buying back the things with the longest duration: perpetual preferred and trust preferreds, which are trading at double-digit percentage discounts to where they were issued. You replace them with a thing that in some loose theoretical way looks similar from a duration and capital structure perspective: a mix of about half common stock (400mm shares = about $3bn on a good day) and half senior notes. It's a regulatory capital improvement. And you'll be paying out less cash going forward, since the senior debt should be cheaper than the preferred and, um, about those common dividends. Sure your common shareholders will be diluted, but not so much on an EPS basis, and they'll be so thrilled with the juiced earnings this quarter that they won't be too worried about the dilution.
The best part, though, may be that BofA could improve its pricing by telling the market "we won't issue common stock just to improve our capital ratios":
Bank of America has been adamant that it wouldn't need to sell common shares for their capital positions alone, a statement it has had to defend amid investor doubts about various costs looming over the bank. But those doubts also contributed to the market reducing the prices on the bank's debt, leading to Thursday's disclosure.
And it's totally true! You didn't sell common stock to improve your capital ratios "alone." You exchanged TRUPS and preferreds for common stock, to improve your capital ratios and book an accounting gain.