Thursday, December 25, 2008

Accrued Interest Takes on M2M

 Mark-to-Market: Discussion minus the Zealotry is an attempt to get at a reasonable approach to M2M. AI brings up a number of points. Obviously, a lot of examples are didactic and don't reflect anything that is being done now.

Example 2:
Let's say that we have two firms, both have made loans to XYZ Retailer. But one is a bank which has made a traditional loan, and the other is a brokerage which holds a private placement bond. The broker almost certainly has to mark that loan to market, but the bank may not.
And in both cases, the rapid changing liquidity premium in the market place alters the "mark" for this asset. By this I mean, say the retailer is performing reasonably well, and thus the risk of non-payment remains remote. Given the weak economy, its obvious that the risk has increased by some degree, but given the extremely weak liquidity across fixed income products, the larger portion of the assets price decline would reflect liquidity. If the firms don't intend to trade the loan, is the changing liquidity premium relevant?
This is a problem where similar assets are booked at different prices. Here I would note that the bank would typically book the loan as a loan (intended to be held to maturity) rather then "loans held for sale." However, the brokerage's core business is trading and not originating and holding loans or similar assets to maturity. The bank is required to hold an "allowance for loan losses" on its balance sheet, which is an estimate. The broker's private placement bond doesn't trade, so credit derivatives would likely be an input into valuation. Note that there really isn't a direct market price for this asset. You have dueling models, one based on a bank's historical losses, judgment, and bank specific accounting guidance and the broker using a pricing model with the "observable" input being a credit derivative. As AI has noted, the bond price may include a liquidity premium which is clearly inappropriate for an institution that doesn't need or want liquidity -- the bank is simply going to hold and amortize the loan to maturity. If the broker really intends to sell the bond, then it should book to an estimate of today's market price. If the broker intends to keep the bond until maturity, then it has an issue, but not with accounting. From my perspective, you can have two different prices for a similar asset in this situation without a major problem. There might be "financial reporting" arbitrage opportunities. However, no one ever claimed that accounting should adopt modern portfolio theory.

Example 3:
There are other problems. Say you are a bank that has a private loan to a company with traded CDS contracts. Your best mark-to-market estimate would be to price the loan based on the cost of hedging out the credit risk. But in many cases, the CDS and cash bond markets have decoupled. Many bonds are trading a drastically wider levels than the CDS market, owing in part to easier funding of CDS. Take Amgen, where cash bonds are trading at a LIBOR spread of nearly 300bps, but the CDS are around 90bps. On a 10-year loan, that implies a valuation differential of about 15 points!

So here again, we have a situation where two firms can use "market" prices to price non-marketable assets, and come up with wildly different valuations. We hear mark-to-market and assume that the "market" is some kind of observable thing. But that is just not the case.
In this case, if the bank holds it as a loan, m2m doesn't come into play, so it is a moot point.   It looks like the CDS market is "broken" in that it doesn't model reality very well.  A synthetic bond would yield less then a natural bond.  That would imply an arbitrage situation where a firm would sell the synthetic bond or CDS contract and buy the underlying.  That is, if markets worked.  Which raises the point of why one would consider a "broken" market -- one that violates the arbitrage free assumption -- as a better representation of reality then other estimates. 
The idea of taking a real, natural loan and goofing around with a pricing model using credit derivatives seems silly.  The firm that owns bonds has to deal with the fluctuation in market price as part of the luxury of owning an asset with a real market price.  The bond is a true level 1 asset and has always and will always be booked at the market price.  This isn't m2m -- it's just accounting.  No one has ever asserted that securities that are traded on an exchange or in a relatively liquid market *not* be marked to market to my knowledge.  Here we would have the loan booked at $X (say par), a theoretical pricing model using CDS's that may value the loan at $X + something (but is never used), a real bond selling for a discount in a bona fide market, selling for $Y (say a lot less then par), and a synthetic bond (cash + CDS) that could be purchased for $Z (more then par).  In this case it is simple -- ignore credit derivatives.  I would like anyone to find examples where a company complained about booking real bonds at market prices.  This isn't M2M, it is vanilla GAAP that has been around forever.  

AI Concludes:
But what's the alternative? Those that are calling for an end to mark-to-market are out of their mind. First of all, there is no clear alternative. Second, we have enough trouble trusting firms' balance sheets as it is. Imagine if mark-to-market were suddenly suspended!
It's very important to note that booking listed stocks and bonds is vanilla GAAP and has nothing to do with m2m, FAS 133, and FAS 157.  No one wants to change this aspect of accounting.  Until we can find this straw man, people need to cool their jets.

However, most critics of m2m are people that, say, hold commercial real estate CDO's and don't want to book them at a huge haircut because of a credit derivative index.  They tend to have a legitimate beef, since there is a basis differences between their security and the index.  There are, once again, dueling models and arguing that credit derivative indices are a market is a stretch.  It is one input into a level 2 or level 3 pricing model.  To my knowledge, they don't want to book these at par.  They have their own models and they don't want to be forced to use credit index based models which may be worse then alternative models.  There is a problem in that these critics always want to book higher asset values, and in a declining market, are usually wrong.  At this point in the cycle.  


No comments: