Tracking home prices in San Diego

Submitted by Econbrowser

Earlier this week, I explained why real estate prices, rather than interest rates or credit workouts, are the critical determinant of how bad the foreclosure problem is going to become. Today I discuss some of the alternative measures of real estate prices that we might look at, illustrated using the latest numbers for my own community here in San Diego.

The most commonly reported statistic is the median price of homes sold during a given period. This has the advantage that we can obtain detailed and up-to-date numbers from real estate networks. The main drawback is that the numbers can be seriously contaminated by a composition effect– you don’t know whether individual homes have fallen in price or if people are simply buying more homes in areas that have always been cheaper. For example, between 2005:Q3 and 2006:Q3, the median sales price was reported to have gone up in every single region in the U.S., and yet the national median sales price went down.

An alternative to using the median sales price is provided by the S&P/Case-Shiller Home Price Index. This attempts to control for composition by using only homes for which one knows the price at which the same house sold on two different dates. A crude version of the idea behind the index would just take the total dollar amount required to purchase a group of homes that sold in 2007 divided by the total dollar amount that the identical group of homes sold for in 2005 as an indicator of how much the price of a “typical” home went up between 2005 and 2007. The actual methodology is more sophisticated than this, refining the above with information about the separate price changes between 2005-2006 and 2006-2007, and also uses the assumption that the longer the time period between recorded sales, the higher the measurement error in the data is likely to be. The data on individual sales prices used to construct the S&P index are taken from local official recording offices in assorted U.S. communities.

The Case-Shiller methodology also forms the basis for the OFHEO House Price Index, whose home price data come from the conforming mortgages handled by Fannie Mae and Freddie Mac. The data base is more broad in some respects (e.g., OFHEO includes the whole country rather than just the particular communities covered by the S&P index), and more narrow in others (e.g., OFHEO excludes noncomforming loans such as jumbo and subprime).

The graph above compares these three measures for San Diego over the last several years. The yellow line plots the median sales price; these data are available weekly from Housing Tracker. The fuchsia line is the monthly S&P/Case-Shiller index, while the dark blue line plots the quarterly index from OFHEO. Note the latter two are index numbers, and have been normalized so that in August 2005 all three measures report the price at $545,000, which was the median sales price reported by Housing Tracker for that month. This normalization is used in order to facilitate visual comparison of the three measures on a single graph and to provide a base of reference for interpreting the numbers.These three measures seem to give very different answers as to how far prices have fallen in San Diego. According to OFHEO, prices are down only 2.6% from August 2005. The S&P/Case-Shiller index puts the decline at 17.0% (logarithmically), while the median sales price implies a 29.7% logarithmic decline. (One reason I prefer to use logarithmic changes here rather than the more standard percent change is that the sum of the logarithmic change over two periods is exactly equal to the single logarithmic change over the entire interval, an additivity property that I’ll be making use of below but which would not quite hold for percent changes.) A similar qualitative disparity is observed between the national OFHEO, S&P, and median price numbers (though quantitatively much less dramatic than these numbers for San Diego), as discussed in an article in today’s Wall Street Journal.

Part of the discrepancy in the case of San Diego appears to be due to how up-to-date the estimates are. The latest OFHEO numbers describe 2007:Q3, while S&P/Case-Shiller goes through November 2007 and the median data are available through February 7. S&P/Case-Shiller implies that an additional 6.1% drop occurred over the two more recent months that the latest S&P index includes but OFHEO does not. The median measure reports that an additional 8.3% drop has occurred in the 12 weeks of data more recent than anything included in the S&P. Thus if OFHEO and S&P have continued to decline at the same rate as observed in the most recently available data, the three indexes would record that San Diego real estate prices have so far fallen by 17.0%, 25.3%, or 29.7% since August 2005, respectively.

The different source data for the S&P and OFHEO indexes may also account for some of the difference. For example, if you look at these two indexes over a much longer period, one sees a tendency for S&P to record a stronger dip in the 1994 downturn than OFHEO did, which might accurately reflect the fact that nonconforming mortgages were hit harder by the 1994 credit tightening than were conforming mortgages. But the same disparity should prove to be much more dramatic in the current episode.

All of which leads me to conclude that the relatively benign numbers reported by OFHEO through 2007:Q3 significantly underestimate the size of the problem we’re already in, both locally and nationally.