Merrill's $3.4 billion balance sheet bomb

In three short weeks, the Wall Street giant's losses grew from $4.5 billion to nearly $8 billion. Fortune's Peter Eavis shows what went wrong.

Peter Eavis, Fortune senior writer

NEW YORK (Fortune) -- What is the balance sheet of Merrill Lynch really worth? Depends to a large extent on who's in charge of valuing the company's large holdings of risky securities.

Wednesday, Merrill (Charts, Fortune 500) reported third quarter earnings that contained $7.9 billion of losses on collateralized debt obligations (CDOs), which are complex debt securities, and junk mortgages. What shocked the market was that only three weeks ago Merrill estimated losses of $4.5 billion on these sorts of assets. What on earth happened that caused the brokerage to suddenly find another $3.4 billion of losses? One cause was that Merrill gave the job of valuing these securities to a group of people who turned out to have a much more conservative view on these assets' true worth.

The revelation of extra losses clobbered Merrill's stock, which fell $3.76, or 5.6%, to $63.36 Wednesday.

The human element in balance sheet valuation emerged during Merrill's communications with the public about its ugly third quarter. On a conference call Wednesday, Merrill CEO Stanley O'Neal said that over the past few weeks, the brokerage's new head of fixed income, David Sobotka, had been part of "a collective review" of the troubled securities, which resulted in "more conservative valuation assumptions" and the larger loss number.

A person familiar with how the losses were booked says that the valuation committee that included Sobotka took a more conservative stance than the people that previously had responsibility for valuing CDOs and junk mortgages. The fixed income business was previously headed by Osman Semerci, who has left Merrill.

Attempts to contact Semerci were unsuccessful at publication time.The market may never hear Semerci's view of what happened and how assets were valued under his charge, and it is important to remember that it is in a brokerage's interest to blame problems on a former executive, rather than those still running the show, like O'Neal himself.

When asked whether Merrill's valuation methods gave substantial leeway for executives to reach markedly different conclusions, company spokeswoman Selena Morris pointed out that O'Neal had said that the loss increase was part of a collective review, indicating that it wasn't driven by one person or a small number of people.

While blame may never be properly apportioned at Merrill, one thing is clear: the brokerage's problems have reignited the debate over whether Wall Street's balance sheets can be trusted.

Compared with 10 years ago, Wall Street firms hold far more securities and financial instruments that don't trade regularly, which means they are valued according to in-house estimates and not so much by market prices. The extra $3.4 billion of losses at Merrill will only deepen fears that brokerages and banks have been overvaluing their assets to avoid taking the correct amount of losses at the appropriate time.

To be sure, Merrill was more exposed than its peers like Goldman Sachs (Charts, Fortune 500) and Bear Stearns (Charts, Fortune 500) to CDOs, which have been hit particularly hard. And many banks and brokerages that have reported third quarter earnings have said they feel their valuations are correct.

But the extra $3.4 billion of losses at Merrill shows how easy it is for valuations to change. The Merrill loss adjustment thus flies in the face of the view, presented by most banks, that their valuations rely less on subjective human contributions and far more on the results of rigorous computer models.

Indeed, on the conference call, the Merrill CFO, Jeff Edwards, gave a helpful insight into the valuation process, and it shows how large the human factor can be.

He said that Merrill didn't change the methodology it used to value the securities that took the big losses. Edwards noted that Merrill's methodology produced a range of valuations for the assets in question. So, what actually changed after Oct. 5, when the bank estimated that it would report the lower $4.5 billion of losses on CDOs and junk mortgages? After Oct. 5, Merrill chose to opt for valuations that were at the "significantly more conservative end of the range," according to Edwards.

This is big. It shows that executives had enough leeway under Merrill's approach to choose a very different end result. Different to the tune of $3.4 billion.

Sobotka, the new head of fixed income, has good reason to go for a more conservative set of valuations. It makes it easier for his business to show improvement in the coming quarters, though Merrill execs didn't sound that confident on the Wednesday call that there wouldn't be more losses on CDOs and junk mortgages.

Yes, Merrill has stumbled more than others -- it wasn't prepared for the credit crunch, its internal loss estimates were too low, and to cap it off its credit rating was cut Wednesday by Moody's and Standard and Poor's.

But it did the market a great service: It showed just how dependent Wall Street balance sheets can be on arbitrary human judgments. Top of page