Press Room

If nothing else, banks' third-quarter results demonstrate the complexities of the seemingly simple dictum that all assets (and dare we say, liabilities) should be marked to market. Comparing bank results is complicated by lingering doubts as to whether the marks were determined in the same fashion--whether they truly represent current market conditions, or cling to some concept of fundamental value in more normal conditions for the portions of the balance sheet in not-so-liquid markets.

Accounting guidelines (such as FASB Statements No. 115 and 133 under U.S. GAAP and IAS 39 under IFRS) have been very clear that most assets on banks' balance sheets must be marked to market (with the exception, for example, of originated or purchased loans held for investment and classified as held-to-maturity) to reflect their value under current market conditions rather than their intrinsic value under more stable market conditions. There is considerable discretion, however, in how firms can interpret those guidelines.

The recently announced or preannounced third-quarter results for the brokers and many large global banks offer very little clarity on the subject. Concerns about the write-offs on funded and unfunded leveraged loan commitments provide a good example of the complications. The information in table 1 shows the charges taken on a gross basis, before fees and hedges, and the current exposures of a sample of banks for which information is publicly available. Simply calculating a ratio of the write-off to the current exposure is misleading. This is because there was a big difference between the starting portfolio at the end of the second quarter and at the end of the third quarter.

During the quarter, many loans were either sold off, failed to materialize, or were funded by other banks (some banks record the commitment as if the entire amount of the loan was to be theirs to fund when in reality they may ultimately share the loan with other participating banks). Other commitments were originated on current market terms that would produce no meaningful discounts. The portfolio at the beginning of the quarter would be a better starting point, but not everyone reported it. For those for which Standard & Poor's Ratings Services has the information, we estimate that the average write-off taken on the initial portfolios was about 2.8% (ranging between 1.8% and 4.7%) before fees and hedges. Fees tended to offset about 40% of the gross losses. The apparent differential between the discounts taken by different banks can be accounted for as much by differences in the makeup of the commitment portfolios (how many loans actually materialized or were sold at par, and the quality of the particular loans) and differences in reporting periods (those who marked loans in September benefited from a slight recovery in the markets) as by differences in conservatism of the marks. Anecdotally, discounts taken on loans that were originated before June and required a write-down were about 5%.

In evaluating the discounts, we take comfort from the loan market not being completely illiquid. Some loans are being sold, providing real benchmarks for pricing. Some banks have reportedly sold loans above the price at which they were marked. We also gain comfort from the fact that market price indications for liquid loans published by S&Ps LCD were in the area of 5% in September, or comparable to the anecdotal evidence on the marks, and that subsequently some loans have even traded above par. Given the liquidity that seems to be coming back, at least to the U.S. market, we believe these issues are behind the banks and that the logjam will be cleared in the next few months.

More difficult to evaluate are losses related to the warehouses of mortgage-related loans and other securities that were awaiting securitization or sale. The investment and universal banks have had very different types of assets and asset classes, ranging from whole loans to super-senior tranches of securitizations. Disclosures on outstandings were far more spotty than they were for the leveraged loans, and the asset category was less consistently defined, so that calculating a percentage write-off is not possible. Similarly, the losses on mortgage-related securities portfolios included everything from highly rated tranches to mezzanine tranches of CDOs that were illiquid even in the best of times. The losses on the two types of exposures were larger than those taken on leveraged loans. The largest was Merrill Lynch's $4.5 billion, largely on super-senior tranches of CDOs. The denominators of any discount ratio often were not disclosed, but would not be meaningful due to the differences in asset mix. We believe that greater transparency on the nature of the securities, the valuation assumptions used, and some indication of the range of valuations by type of security would be very helpful in allaying market concerns, and should be included in the 10-Qs.

The extent of the difficult-to-value securities being held is indicated by the disclosures on the Level III assets in the brokers' recently filed 10-Qs. These increased substantially from the prior quarter as the firms had to transfer assets out of the Level II category into Level III. The banks report that they tried to use current prices at which the assets could be sold when quotes were available. Where models were used, they were based on assumptions that produced a match for the observable prices for similar securities. In some cases, independent pricing services were used, but these could be based on an analysis of the fundaments for underlying assets and may not represent the price at which the assets could be liquidated. In our opinion, it is still an open question as to whether these securities have been similarly if not realistically marked by all players to reflect current market conditions. Consequently, we expect further mark-downs in future quarters.

We have not seen a shift of assets to accrual books that do not require a mark from marked-to-market books, as some suggest has been happening. The commercial banks that do not ordinarily mark unfunded commercial loan commitments to market because the loans would ordinarily sell at par, marked them because the losses were "probable and estimable." The commercial loans above the desired hold level have remained in the marked-to-market category. They will need to be sold as market conditions permit because they represent single-loan exposures that are too high, which is a concentration risk. Some of the loans represent multibillion-dollar exposures for each of the banks involved. These are too large for even the largest balance sheets to risk. If real credit events were to occur, compounding the issues that are now only a matter of loan pricing, then the 5% discounts could turn into 20% or even more, which would be unacceptable risk. For the securities portfolios, we have not yet seen any migration from trading books to held to maturity. A shift of loans held for sale to held for investment sometimes occurs, but with the requisite mark to market at the time of the transfer.

Given the extremely difficult conditions under which the valuations are being made, we believe there has been a reasonable effort to take marks to market on a variety of exposures. Hopefully the lion's share of the discounts have been taken. No CEO would want to have to report another quarter of such losses, so there is every incentive to try to get as much as possible into third-quarter results. However, we do not think every bank used the same methodologies and assumptions on the securities portfolios or warehouses, so some adjustments will likely have to be made.

We have not changed ratings because of any concerns about securities valuations. The downgrade of UBS and a change in outlook for Bear Stearns and Merrill Lynch reflected concerns about risk management and risk appetite.