by Calculated Risk on 1/17/2008 12:06:00 PM

Last July, OFHEO economist Andrew Leventis wrote: A Note on the Differences between the OFHEO and S&P/Case-Shiller House Price Indexes. The OFHEO note suggested that the primary reason for the difference between the national Case-Shiller and OFHEO price indices is geographical coverage (not the loan limitations for OFHEO).

Now Leventis has completed a more thorough analysis removing the geographical coverage by focusing on 10 MSAs: Revisiting the Differences between the OFHEO and S&P/Case-Shiller House Price Indexes: New Explanations

The results are surprising, and the implications significant. Leventis found:

The empirical estimates suggest that, while the causes of divergence may have differed in previous periods, most of the current gap is generally attributable to three factors: OFHEO’s use of home price appraisals, differences in how much weight is given to homes that have lengthy intervals between valuations, and variations in price patterns for inexpensive homes with alternative financing.
emphasis added

Leventis analyzed the inclusion of non-agency financed medium and high prices homes, but this didn't have a large impact. This is contrary to the common view that the difference between OFHEO and Case-Shiller is because of the conforming loan limit.

OFHEO provides a Purchase Only index that eliminates the first factor (the use of appraisals). The second factor - the differences in weighting certain homes - is somewhat technical. But the third factor is clearly important:

The depressing effect of the inclusion of low-priced houses without Enterprise-related financing raises many questions. Some of these houses were undoubtedly financed with subprime mortgages and thus one might wonder whether some of the effect somehow relates to turmoil in that market. For example, subprime homes may be clustered in neighborhoods with relatively intense recent foreclosure activity. While this analysis attempted to rule out such “neighborhood effects” at the zip code level, zip codes are large areas and analysis of smaller geographic regions (e.g., census tracts) might reveal more localized differences. Another plausible explanation is that borrowers with subprime loans may not have spent as much on home improvements, maintenance or repair. If these types of expenditures were lower for subprime borrowers, then deprecation rates may have been greater for the homes with subprime financing.

A review of the impact of adding the low-end, non-Enterprise properties to OFHEO’s dataset suggests that, during the latter part of the housing boom, these properties may have appreciated significantly more than Enterprise-financed properties. Accordingly, it seems these properties are different from Enterprise properties in ways that are correlated with price trends.

This suggests that one of main differences between OFHEO and Case-Shiller was that Case-Shiller included many non-agency homes financed with subprime loans. These homes saw more appreciation during the boom, and are now seeing larger price declines.

Whatever the reasons, the Case-Shiller index seems to more accurately reflect the current price declines in the housing market, as opposed to the OFHEO index. And this has significant implications for the economy.

The Fed uses the OFHEO index to calculate the changes in household real estate assets. If the OFHEO index understated appreciation during the boom that means households have MORE real estate assets, and more equity, than the current Flow of Funds report suggests.

That sounds like good news, but ... that also means that during the housing boom, the wealth effect was larger, and the impact on GDP greater, than current estimates. This also means - if OFHEO understated appreciation - that the negative wealth effect, and the drag on GDP, will be probably be greater than expected during the housing bust.