There's a certain intellectual beauty in the notion that big banks, or investment banks, or whatever you want to call Goldman Sachs and Morgan Stanley, should mark their whole balance sheet to market each day and report any changes in value through their income. In a business that basically consists of taking lots of shareholder and lender money and turning it into ever so slightly more money, the only thing that you should be judged on is the current market-implied value of that difference. Sometimes this can create results that are both aesthetically interesting and intuitively absurd, but the intellectual question - how much does the expected present value of your assets exceed the expected present value of your liabilities, and how did that difference change this quarter? - is one that it is noble to try to answer.
Also, if you ignore it, terrible shit happens.
So it's a little sad to learn that the only real proponents of rigorous mark-to-market among the giant publicly traded financial institutions, GS and MS, are contemplating abandoning it:
The two firms are discussing whether to reduce their use of mark-to-market accounting, in which companies immediately take profits or losses as asset values fluctuate, according to people familiar with the situation. Such swings routinely affect the bottom line at Goldman and Morgan Stanley, including in the third quarter.
Sad, but you knew this day would come. The temptation to juice earnings - or rather, to manage earnings to a smooth path of consistent quarterly growth - is just too strong for any bank to resist.
Except that, obviously, it makes no sense to use historical-cost accounting for a securities trading business - and of course GS and MS have no intention of doing so (and it wouldn't be allowed) - so their earnings will continue to be erratic. In fact, the Journal reports, GS and MS are focusing on one asset class where accounting rules let them use historical cost accounting - and where mark-to-market swings don't actually much affect the bottom line:
At least for now, any shift away from mark-to-market accounting would affect just a slice of the $1.7 trillion in combined assets at Goldman and Morgan Stanley. Mark-to-market accounting is bruising for both firms on loan commitments to large companies, requiring Goldman and Morgan Stanley to take losses even before a company borrows any money. ...
Loan commitments often are made to companies that are clients of the two securities firms to acquire or merge with other companies. The value of such loans can swing by as much as 10%.
In the third quarter, Goldman had a loss of nearly $1 billion on its fixed-income holdings, a large portion of which were loans. That pool includes investment-grade and noninvestment grade loans. The overall impact of mark-to-market accounting on the bottom line because of loan commitments wasn't material at either firm in the latest quarter, according to people familiar with the situation.
So what is actually going on?
Well, notice that "Loan commitments often are made to companies that are clients of the two securities firms to acquire or merge with other companies." True - and those bridge commitments tend to be among the biggest and the ones most subject to mark-to-market swings (since they tend to be written to companies that get downgraded when their big levered acquisition is announced, and since their risk is quite wrong-way as they only get drawn when credit conditions are so bad that the client can't take them out with public bond offerings). But, of course, loan commitments are also made for ordinary-course revolving lines of credit; it's not clear from the Journal article whether GS and MS are also contemplating reducing the use of mark-to-market for revolver commitments.
These commitments are basically always money-losing in an expected value sense. M&A related bridge loans are written with comically extensive protections for the banks to make sure that they're never drawn: the bank can normally force the acquirer to issue bonds at nosebleed rates, market them over and over again, drag the CEO on weeks of roadshows, and if necessary sell his children to raise cash before the company can draw the bridge loan. If the banks are that afraid of having to fund the loan, then that is a pretty, pretty good indication that the loan is not a good security to hold - i.e. that it's worth less than par.*
This is maybe less obvious with revolvers, but it's more true. Writing a revolver is a whole lot like writing a CDS contract (everything is a derivative!): the client pays you a fee, and in exchange, you will give it money if and when it has no money and no one else will give it any. This is a risky trade! When your client really really needs money, you don't want to give it to them! You get paid for that risk by being paid a spread. That spread is ... well, for MF Global, it was 40bps (page 67 here). Maturity mismatch blah blah blah but there was never a time you could buy CDS on MF Global for 40bps, or anywhere close to 40bps. If you think like a mark-to-marketeer - that is, if you believe that the revolver you just wrote to MF Global is a CDS contract, and you hedge it by buying CDS - you have lost a lot of money the second you signed up the revolver, even if it's never drawn.
So why on earth are Morgan Stanley and Goldman doing all of these trades that lose money, on an expected value basis, the minute they're signed up? Well, being willing to finance a merger tends to be really helpful in getting picked to advise on the merger. One-stop shopping, don't you know. Even more importantly, being in the bridge guarantees you a role in taking out the bridge, getting capital markets fees that are often more lucrative than the merger itself. And outside of M&A, big companies tend to look at the lenders in their revolver as their friends, the banks that are supporting them - and those friends are the ones who are first in line to get capital markets, derivatives, cash management, FX, and other business.
If you're Goldman Sachs or Morgan Stanley, or JPMorgan or BofA or Citi or Credit Suisse for that matter, you view these money-losing loan commitments as the crucial loss leader in getting investment banking business, more important even than paying analysts lots of money to cover nonexistent stocks. But if you're JPMorgan or BofA or Citi or Credit Suisse, you know it's a loss leader but you don't have to tell anyone - it's "held to maturity," and you mostly end up not losing any money on it.**
For GS and MS, though, you take that loss day one. And so as the M&A market ramps up, and lending becomes an important part of it again, you look less profitable than the financial supermarket banks, because you are showing the loss on your loss leader product and they aren't - or because your CFO isn't letting you take that loss and so you're missing out on banking business.
The lesson to take away from this is that GS and MS are feeling increasing pressure to win more investment banking business by lending: that is, that M&A and capital markets business is increasingly being won by banks' willingness to put their balance sheets to work, rather than by quality of advice or length of relationship or golf outings or whatever. This is sad news for those of us with an aesthetic fondness for mark to market accounting, but we'll get over it. It's worse news for the smaller independent banks - Lazard, Jefferies, whatever - who don't have the balance sheet to put to work to compete for deals that are won by lending, and who can't fix that with just an accounting change.
* These days M&A bridge fees are juicy enough that they might be positive expected value day one, actually, but that valuation is pretty volatile. In anyc case the point is that the commitment itself, day T+1 after the fees have been paid, is a bad deal for the bank.
** While the commitment has a negative expected value in a rational, mathematical world, it often works out well because of soft factors: companies actually mostly use their revolvers for working capital, not as a last-ditch just-before-default liquidity infusion, and bridges tend to get taken out before funding because CEOs aren't good at doing a clear-eyed cost-benefit analysis about selling their children. So you don't book a theoretical loss day one, and the actual loss mostly doesn't materialize. Except when it does.