Tuesday, November 04, 2008

What Do House Price Indices Currently Mean?

At the Berkeley-UCLA conference last week, audience members queried Bob Shiller and Chip Case about how they adjusted their price index for foreclosure sales. Chip's answer was that they had a regression that contained a foreclosure dummy, but that only paying customers S&P-Case-Shiller customers got to see it.

This is a problem. CS is a repeast sales index, with the idea being that by looking at houses that sell twice, and seeing how their value changes across time, one has a constant quality house price change. The problem with including foreclosures is that the constant quality feature is almost certainly eliminated--foreclosed houses are almost certainly undermaintained (there are newspaper reports about this--I would be curious to know if there is a rigourous study), meaning that they are no longer constant quality houses. This feature will bias estimate downward.

My colleague Chris Redfearn notes another problem with the current index. For the LA area, the index is being disproportionately influenced by Riverside and San Bernardino counties, where foreclosure sales have produced big upticks in sales volumes. In stable neighborhoods with financially stable households, people are simply not putting their house on the market, so the relative stability is not reflected in the index. If one looks at any real estate web site, one will find that the number of listings in Santa Monica, for instance, is quite small.

All of this suggests that the CS Index is currently biased downward. While there can be no doubt that house prices here in Southern California have fallen a lot, they have almost certainly not fallen as much as the CS index suggests.


Anonymous said...

Well the people that create and market the index certainly know about this defect. Why do you suppose they don't put a big flashing disclaimer on it? S&P produces this index. Is it any surprise there are questionable aspects of both the intent and formulation of the index?

As for its relation to housing values. I think the bias is moot since prices will continue to plummet. The index has served its credible and highly documented purpose.

Anonymous said...

The S&P/CS Los Angeles HPI includes no Riverside/SanB transaction data - it's 100% LA and Orange counties.

All indexes have flaws, guys, and housing is challenging to measure - re: impact of foreclosure sales on the index, I'd say you're off the mark again (here again, the freely-availble S&P/CSI methodolgy doc is instructive). Of course, when lenders' REO sales comprise a significant portion of the sample, prices fall (and adversely impact values of nearby homes that were not just sold by a bank/lender). How meaningful would a housing index be - especially in a market such as this one - if home values pertaining to post-foreclosure bank sales were NOT included??? I would suggest that in such a construct, the index would be artifically high, biased upwards and thus not properly measuring this admittedly ugly market.

Anonymous said...

" I would suggest that in such a construct, the index would be artifically high, biased upwards and thus not properly measuring this admittedly ugly market."

And making used house salespersons very happy to point to it but very unhappy when trying to price a home accurately while ignoring foreclosed sales in a market riddled with foreclosures rapidly driving prices ever downward. Until the amount of foreclosed homes come more in line with low historic numbers, denigrating C/S charts is about which gallon of water actually caused the sinking of the ship and how full the ship was at the time. Pointless now that it has sunk. It's all about the number of foreclosures. Analyze that to get a handle on where this is headed.

Anonymous said...

That is an interesting response. I can see where you are coming from, good input.


Anonymous said...

Someone really decided to put on their thinking cap, great going! It’s fantastic to see people really writing about the important things.