The problem with forecasting house prices

The value of a house should, in equilibrium, be its capitalized rents. Its capitalization rate is (roughly) the after-tax required rate of return, plus depreciation and expenses less expected appreciation. We may write this out as:

r + m - pi.

The r and the m are relatively easy to measure. Households in the conventional conforming market can borrow at an after-tax mortgage rate of about 5 percent right now, and with very little equity, they can borrow at around 6 percent (the cost of interest plus mortgage insurance). Let's add a risk premium and put the total around 7 percent (this is probably a bit high). Depreciation and expenses will run around 2 percent of house value. So the value of a house is the value of its rent divided by 9 percent less expected appreciation.

Now let's see what happens when we let expectations about future prices rangs from an increase of 5 percent in a year to a decrease of 5 percent in a year. If expected prices increase five percent, values are 25 times rent (rent/.04); if the decrease 5 percent in a year, values are about 7 times rent. So a ten percentage point deline in expectations could cause house prices to fall by two-thirds.

I think this is part of the current problem. On the one hand, after roughly 2002, prices in many markets shot up well past the 25 to 1 point (on a quality adjusted basis), in part because of very low interest rates, and in part because of unrealistic expectations. Now that prices are falling (as inevitably they would), expectations have reversed, although it is not yet clear by how much.

Some months ago, when others were writing that the bottom was coming, I wrote that I didn't know when the bottom of the housing market was coming, and neither did anyone else. I wish I had been wrong.