Merrill Lynch reported Thursday morning a net loss of $9.83 billion during the fourth quarter, matching the loss posted earlier in the week by Citigroup — and well below the $2.3 billion earned in the year-ago period. Losses in Merrill’s fixed income business swamped revenue gains made elsewhere, the firm said. Driving the losses for the quarter were what the Wall Street Journal reported as $16.7 billion in total write downs, although a back-of-the-napkin calculation by HW suggests that number really should be just north of $18 billion:

• $9.9 billion in to U.S. ABS CDOs;
• $3.1 billion in exposure to bond guarantors;
• $1.6 billion in subprime mortgages;
• $400 million in Alt-A mortgages;
• $500 million in mortgages outside the U.S.;
• $230 million on commercial real estate;
• $126 million on leveraged finance commitments;
• and $2.17 billion in subprime and Alt-A RMBS

(The difference between the WSJ and my own math lies in that last item — of the $2.17 billion quoted, only $869 million ended up on the balance sheet and that’s the number the WSJ used. In my mind, however, a write-down is a write-down regardless of where it’s stuffed. So I’m sticking with $18 billion.) To put the write down into industry perspective, it’s worth noting that $14.07 billion of the total write down amount is tied to U.S. residential mortgages. Ouch. Either way, the total number of write-offs for the fourth quarter ended up exceeding most analysts’ estimates (see an earlier post here), and helped drive Merrill to a full-year net loss of $7.78 billion, versus earnings of 7.5 billion in 2006. “While the firm’s earnings performance for the year is clearly unacceptable, over the last few weeks we have substantially strengthened the firm’s liquidity and balance sheet,” said John A. Thain, chairman and chief executive officer. Merrill said its net mortgage exposure fell from $49.2 billion in the third quarter to $43.6 billion in the fourth quarter, as performance in subprime and Alt-A mortgages hurt warehouse lending, whole loans, residuals and RMBS. For more information, visit http://www.ml.com.