Editor's Note: Salomon Brothers has developed a new model for predicting prepayment speeds on mortgage portfolios. In this article, excerpted from a longer report, two Salomon analysts discuss housing turnover, a key element in their model.
Housing turnover may be defined as the ratio of existing single-family home sales to housing stock. This definition makes the relationship of housing turnover rates to mortgage prepayment speeds very clear.
Salomon Brothers has performed a careful analysis of housing turnover rates over the past 15 years, leading to a deeper understanding of the factors that affect home sales and to a model that may be used to project turnover rates under different interest rate scenarios.
We believe that modeling home turnover rates leads to a more dynamic and realistic depiction of the evolution of turnover and hence prepayment rates over time.
For example, if a sudden and sustained rise in mortgage rates occurs, an initial drop in home sales (and hence in prepayment speeds) will take place. But eventually, pent-up demand and a gradual adjustment to the higher mortgage rates will lead to a pickup in sales and in prepayment speeds.
The model also allows for easier sensitivity testing - testing of questions like: What happens if home inflation is significantly different from historical norms for a given level of interest rates?
Finally, users can determine for themselves if they agree with the model's projections for home turnover rates along given interest-rate paths.
Demographics and population mobility, as well as macroeconomic and social factors, all combine to influence home sales.
Affordability. This refers to the homebuyer's ability to make a monthly mortgage payment. Affordability can be approximated by the ratio of median income to the median monthly mortgage payment on a median home.
Affordability is often cited as an important predicter of home sales, and correctly so. However, the effect of affordability is subtler than it first appears. Home sales depend not only on the current affordability of housing, but also on the recent history of affordability.
For example, when mortgage rates eased after hitting all-time highs above 16% in 1981-82, the pickup in home sales was immediate. Turnover averaged 4.85% and 5.02% in 1983 and 1984, respectively, although affordability was still quite low by historical standards.
Desirability. Another socioeconomic factor that helps explain historical variations in turnover is the desirability of homeownership. We use this term to include a perception of the likely economic return from buying a home and whether it is currently prudent to do so.
In our model, the variable influencing this desirability is the prospective inflation in home prices - for which a good proxy might be a weighted average of nominal home price changes in recent years.
Consumer confidence. This is another important influence on housing activity. Though using this variable would improve the historical fit of our model, we have chosen not to use it, as we do not want to attempt to predict its future levels.
There are two intersecting populations of prospective buyers or movers: those who want to buy or trade up, and those who can afford to. The assumption is that the turnover rate is determined by the size of the intersection.
The sizes of the groups depend upon desirability and affordability, respectively. We capture these levels with an affordability factor that depends upon the median income, median home price, and mortgage rates, as we have described here, plus a desirability factor that incorporates the effects of home price inflation.
The model also accounts for pent-up demand, or the lack of demand due to past interest rates, by carrying forward an affordability "deficit" or "surplus" from previous periods.
For our projections, we make the assumption that income and home prices change at the same rate over time, so changes in affordability are just a function of changes in mortgage rates.
