When, on January 16, Citi chief executive Vikram Pandit unveiled $18 billion of write-downs at the bank’s annual results briefing, he told analysts "we want to be transparent with you on the risks we have" and promised "we will be very candid with you". But it isn’t easy making sense of the figures, understanding in straightforward terms how big a write-down the bank has taken on precisely what exposures and on the basis of what calculations from observed market prices or proprietary models.
Citi disclosed that it had direct sub-prime exposure – made up of CDOs of ABS, sub-prime loans held for securitization, financing collateralized against sub-prime and other warehoused inventory – of $54.6 billion at the end of the third quarter of 2007. It reduced this, through $18.1 billion of write-downs for the final quarter of the year and then made some small sales, to leave it with $37.3 billion of total exposure at the start of 2008.
Simple maths suggests that the bank is now carrying this toxic waste at 68% of original value. So it has hardly kitchen-sinked the problem, even though it appears to be carrying the mezzanine super-senior CDO exposure, initially valued at $8.3 billion and least likely of recovery, at $3.6 billion, implying at 43 cents on the dollar. (The bank won’t confirm this 43 cents figure.)
So wait: is that roughly one-third mark-down correct? That initial figure of $54.6 billion of net exposure comes after allowing for $10.5 billion of hedges. Question: those hedges wouldn’t be with bond insurers, now seemingly at tipping point, would they? Apparently a portion of those hedges are with monoline insurers, and Citi has taken a write-down of $935 million relating to downgrades of these hedge counterparties.
OK, let’s leave the hedges. How was the key write-down of $14.3 billion on net exposure to super-senior CDO tranches calculated? Citi says these are simply not subject to calculation from the price of observable transactions and so it uses a proprietary discounted cashflow model, incorporating several key assumptions including for the likely decline in US house prices over the next couple of years (which the bank now puts at 6.5% to 7% a year, a more pessimistic assumption than it used last November), as well as its own allowances for fraud on sub-prime loans and geographic and borrower and collateral type risk concentrations.
When this model kicks out a valuation, Citi cross-checks it against the ABX indices, even though it says there are problems in using these indices for actual valuations.
Of course, any discounted cashflow model requires another key assumption – on interest rates. Citi says it has made assumptions based on downgrades of comparable CDOs to its own – even though it might be carrying CDOS that have not yet been downgraded – and sought an analogous observable CLO rate at which to discount the cashflows.
OK. Let’s try to take another step back. Were the starting point net and gross exposure figures at the end of the third quarter the full face-value figures or were they already showing any substantial discount to par on the initial face value of these assets?
Apparently the bank had already taken marks of $1.9 billion in the third quarter but it still seems unable or unwilling to specify the exact par value of the initial exposures against which its proprietary model – using several unspecified and fluid assumptions – has now kicked out a valuation, after allowing for hedges that might, or might not, perform.
Of course, it would not be fair to single out Citi. It is a similar story elsewhere. Take UBS, which of the European banks is widely thought to have taken the most hair-shirt approach. On December 10 2007, it disclosed an additional $10 billion write-down of US-sub-prime-related securities, primarily super-senior tranches of CDOs and also RMBS. It too gave investors the strong impression that it has written down exposure so far that the problems are now taken care of. Marco Suter, chief financial officer, went so far as to hint that markets might have overshot on the downside and to speculate that "today’s write-downs might be tomorrow’s write-backs".
In fact, analysts poring over the numbers see the bank is carrying super-senior CDO positions at an average valuation of 60 cents on the dollar and RMBS, mainly AAA-rated, at 77 cents on the dollar. Further write-downs cannot be ruled out on its remaining $13 billion of super-senior CDO exposures and $16 billion of RMBS. Similar charges over a lack of transparency could just as easily be levelled at Merrill Lynch, Morgan Stanley, or in fact any bank that has announced, but not detailed, write-downs.
Throughout this crisis, banks have compounded the air of distress by a failure completely to come clean about the true size and nature of their exposures, and the exact methodology they are using to calculate the size of their write-downs.
Look at SG, which, in addition to its rogue trader issue, also revealed another bombshell of more write-downs: €1.1 billion of super-senior CDOs, €400 million on CDOs and some €550 million on monoline exposures. Its previous take on sub-prime/CDOs was a modest €1.5 billion.
Marking to market when the market is functioning normally is straightforward but now it is not. Banks have plenty of scope to mark to market, mark to model or use some sort of hybrid.
It is regrettable that the turn in the credit cycle has coincided with the introduction of fair-value accounting in the US and provided the first test of the International Financial Reporting Standards. These measures were the auditors’ response to the accounting debacle at Enron and Parmalat, and what a mess it is.
Observers’ inability to easily compare across peers could be because there is no established best practice. But what is increasingly apparent is that what we have now is little better than the financial black boxes we had before.