Mark to Model, Not Market

There is an interesting article at the Accrued Interest blog regarding mark-to-model, as opposed to mark-to-market. AI mentions that while the approach of mark-to-model and the term itself are vilified, nearly every bond portfolio is marked-to-model. Data taken from Bloomberg was used to show that there are over 150,000 different US dollar denominated corporate bonds outstanding. There are also over 1.2 million munis. TRACE data shows that less than 3% of corporates trade each day, and only about 1% of munis. Since many of the bonds do not trade frequently, there is no current market price to use for marking the bonds. Many dealers will try to find a price to keep customers happy, leaving pricing matrices with inconsistencies. While CDOs and the like are more complicated in their modeling, their use of a mark-to-model process is similar to what is often used for standard bonds.

Of course, during credit crunch times, marking is also somewhat dependent on the desperation of your neighbor. If others are getting margin calls, and need to liquidate, they may be selling at fire sale prices (or "puking up" the bonds in the delicate speak of Wall Street). Later, you come in only to find you are being marked to an artificially low price, now causing your margins to be squeezed. Maybe in this case a model, even an imperfect one, would be better than the "rational" price discovery of the markets.