Wednesday, July 23, 2008

OFHEO HPI and long term trends.

The new OFHEO house price index came out yesterday. At the moment, I like OFHEO better than Case-Shiller (I worry that Case-Shiller is now giving too much weight to REO properties in its index). In any event, it is fun to play with some long term trends.

According to yesterday's OFHEO press release, since 1991, house prices have risen about 4.5 percent per year; since 2000, they have risen by 5.5 percent per year, even taking into account the recent decline. This means that between 1991-2000, prices rose about 3.6 percent per year (take (1.045^17/1.055^8)^(1/9)-1).

Suppose that 3.6 percent is the long-term nominal house price growth trend. By how much are house prices overvalued? The answer is (1.045^17)/(1.036^17)-1= .158. So house prices would have to fall by about another 13.6 percent immediately to stay in line with the long term nominal trend, after which they should rise by 3.6 percent per year.

Alternatively, if house prices just stayed flat for four more years, they would return to their long-term trajectory--assuming that the trajectory before the year 2000 was the long-term trajectory.

3 comments:

TStockmann said...

The rate of nominal increase seems to me completely useless as predictive trend line. I can't even think of an economic housing model that would justify it.

Richard Green said...

OK--CPI growth between 1991 and 2000 was a little under 2.6 percent per year; since 2000 it has been about 2.9 percent per year. This implies that house prices need to fall less and/or take less time to return to normal trend, assuming constant real interest rates, etc.

I don't take trend analysis that seriously--I am just trying to have a little fun. But others seem to base all their views on housing around trend analysis. As I have said before, I think that house prices have fundamentally returned to something pretty close to normal, but psychology and seizures in the lending market may cause overshooting on the downside.

gaius marius said...

I think that house prices have fundamentally returned to something pretty close to normal, but psychology and seizures in the lending market may cause overshooting on the downside.

i agree on the probability of an overshoot, dr green -- but what analysis leads you to believe that housing prices have normalized?

fwiw -- and i realize this is but anecdotal, but it is illustrative of the general condition -- i live in suburban chicago, renting a house. i pay $2000/mo, property taxes are $550/mo. my rental payment then would support (ignoring upkeep/insurance/etc) a $1450/mo payment.

at today's 30-year fixed rate (~6.5%), that would support a $230,000 loan. with 20% down, call the purchase price $290,000 -- and generously, as we are excluding all expenses but taxes.

this house sold in 2005 for $460,000. houses in the neighborhood still list for $410,000.

i agree very much with paul mcculley when he says that bottom will be struck when positive carry becomes compelling -- when capital is again rewarded for being tied up in the asset by being able to turn a decent income through renting. in chicago, we are a VERY long way still from that point -- indeed some 30% by my anecdote. that also happens to be approxiamtely as far as price-to-income ratios in chicagoland remain from historical norms.

i'm not saying i have to be correct, of course -- but i am interested in your or any methodology that sees this price level as representative of the eventual equilibrium.