Valuation Commentary - April '06

Modeling Home Prices as Dynamic Assets
by Alex Levin

The credit/default modeling efforts are receiving considerable attention these days and they will be a prime focus during the afternoon session at AD&Co.’s upcoming 2006 Annual Conference. This report summarizes some of our current views on this topic.

One factor that drives a borrower’s decision to default is the home price of his/her property. The owner with grim personal financial circumstances has an option to sell the property or to let the bank take it and foreclose. This decision will be based, primarily, on the amount of outstanding debt and the value of the house.

From the option-theoretic stand-point, the default process is a complex derivative of the home value process, among other things. Therefore, unlike many econometric models, a model for home prices needs to include random disturbances, or noise, as part of the model. Capturing volatility and correlation structures is the goal of current home price modeling efforts at AD&Co. What we are looking for is a model for randomness and a relationship, if any, between house prices and interest rates.

Factors

AD&Co. perceives that

• Individual home prices are related to a state’s (or region’s) home index plus a random idiosyncratic factor. The latter does not need to be “randomized;” its volatility lets us predict the probability of negative equity analytically.

• States’ factors have (generally, limited) correlations to the U.S. home price index (HPI), with state-level “noises.” The relationship between a state HPI and the U.S. HPI can be deemed as the relationship between a single-name stock and a market index. The Beta for each state or region and properties of the idiosyncratic factor can be backed by historical data.

>>>