The Latest S&P Case-Shiller Housing Price Indexes
Declining U.S. home prices led the way into the current worldwide economic crisis, and one sign that the crisis is abating will be when these prices begin to stabilize. More stable home prices would indicate that prices are at a point where buyers can be found and that credit is available.
The December 2008 S&P Case-Shiller Home Price Indexes (released February 24, 2009) offered no evidence that this is happening yet. Over the past year, the 20-city index fell 18.5 percent and the 10-city index fell 19.2 percent. The seasonally adjusted annualized rate of declines for December were 21.3 percent and 19.8 percent, respectively.
The only Fourth District region tracked by the Case-Shiller indexes is the Cleveland-Elyria-Mentor metropolitan statistical area (MSA), which includes at least portions of Cuyahoga, Geauga, Lake, Lorain, and Medina counties. Over the past year, the decline in Cleveland home prices, 6.1 percent, was smaller than in every other MSA included in the indexes except for two, Dallas (down 4.2 percent) and Denver (down 4.0 percent).
Cleveland’s aggregate index masks some extreme volatility in its lowest housing tier (homes valued under $116, 639 in November of 2008). In 2008 the index for this particular tier saw annualized monthly percent changes of 223.6 percent, 104.5 percent, 75 percent, -83.1 percent, and -82.9 percent. The absolute value of the index’s annualized growth rate has been less than 20 percent only twice in the past 12 months. In fact, Cleveland’s tiered-price indexes were not even included in the latest report because they were considered too statistically unreliable. A footnote stated, “After review of the data the standard errors were deemed too large and the December numbers were not believed to be reliable at this time and therefore will not be published.”
What is the source of this volatility? It is probably a consequence of the number of observations the indexes have to work with. If there are too few, averages from period to period can vary a lot. These numbers do look unusually low for Cleveland over the past year.
Case-Shiller indexes are constructed by comparing the sales prices of a single-family homes with their previous sales prices. The two prices constitute a matched “pair.” In December 2008, there were only 554 pairs in Cleveland for all three tiers. This figure is not only low compared to other cities tracked by the indexes, it is the lowest figure in the whole history of the Cleveland series, which goes back to January 1987. What is also clear is that sale-pair counts were abnormally low and somewhat damped during the past business cycle. (As in most cities, Cleveland home sales are highly seasonal, slowing during the school year and winter.) This past summer’s peak did not even rise to a more normal year’s trough, and the amplitude is sharply stunted.
The sale-pair counts for the top-10 and top-20 city indexes show a similar pattern. In addition to being seasonal, counts in both indexes show a downward trend and a decline, albeit more modest, in amplitude. But importantly from a statistical perspective, none of the other cities have as few pair counts as Cleveland. The next lowest, Charlotte, has over twice as many at 1,257. Statistically, a larger sample enables tighter bounds to be put on the estimates of the housing price index.
Why are sale-pair counts down? First, the housing market is weak, so there are not that many home transactions of any type. Then, only a subset of transactions is used to calculate the indexes. Only arm’s-length transactions, where both the buyer and seller acted in their own best economic interest, and repeat sales transactions for existing, single-family homes are selected. Filtering excludes property transfers between family members and the repossession of properties by mortgage lenders at the beginning of foreclosure proceedings. Any subsequent sales by those lenders are included, however. The data are also filtered to exclude homes that have had substantial physical changes (either major renovations or significant material damage).
Many of the Cleveland transactions are not arm’s-length. Over the past 12 months, 26.9 percent were foreclosures, according to the real-estate-information website Zillow. The figure for December is not available, but the National Association of Realtors reports that for the nation as a whole, 45 percent of December sales were foreclosure-related or otherwise distressed. The figure for Cleveland’s lowest tier is likely higher than for Cleveland overall, as it was the lowest tier that was hardest hit by delinquencies from subprime loans.
The bottom line is that match-pair house-price indexes like S&P Case-Shiller can provide valuable information about how house prices have changed over time. However, extra care must be taken in interpreting these indexes when the sales-pair counts fall to such abnormally low levels.