Sitting on the banks of the Rio Grande is Truth or Consequences, a New Mexico town formerly known as Hot Springs but renamed after a television game show to increase tourism in the city. When analyzing housing prices, many analysts proceed through truth or consequences moments due to the large variety of home price data in the market. The most well known price index is the Standard & Poor’s Case-Shiller HPI. However given that the two monthly Standard & Poor’s Case-Shiller composite indices cover only 10 or 20 cities, some have remarked that they are not representative of the nation as a whole. How does Case-Shiller compare against a similar national price index?
Among the existing price indices, the CoreLogic HPI is the most similar index to S&P Case-Shiller. The two indices are constructed using a value weighted repeat sales method and both include distressed sales. The major difference is that the CoreLogic HPI covers over 580 metros, compared to the S&P Case-Shiller composite HPI which, depending on the index covers 10 or 20 cities for the monthly index .
Despite the differences in geography, the year-over-year changes in both indices are remarkably similar (Figure 1). For the time periods they cover, the 10 and 20 City indices averaged 3.9% and 3.2%, respectively and both were within one-tenth of one percent of the CoreLogic HPI for the same evaluation period . While on net the average price growth was similar over the cumulative time horizon, on an absolute basis the 10-City Index differed by an average of 2.0 percentage points and the 20-City Index differed by 1.3 percentage points. This means that the Case-Shiller indices exhibited larger swings in prices by those percentages for any particular month, but that they offset on a cumulative temporal basis. While there are moderate differences in absolute changes for any one month, the cumulative differences over time are quite small. The key takeaway is that despite only covering 10 or 20 large metropolitan areas, the Case-Shiller is representative of the national market.
One unique distinguishing feature of the CoreLogic HPI is that it contains tiers that include or exclude distressed sales and the two indices can be used to gauge the impact of distressed sales on price changes. There is a long line of literature on the impact distressed sales on the level of home prices, such as nearby prices or the differences in aggregate geographies. However, there is much less data or research on the impact of distressed sales on home price changes. In other words, it is well known that distressed sales can significantly negatively affect the level of overall prices, but it is less well known how large the impact is on overall price changes over time.
Three things stand out from examining the rate of appreciation of all transactions versus transactions that exclude distressed sales. First, prior to 2004 the two series were nearly identical despite going through two economic recessions in 1991 and 2001 and one housing recession in 1990 (Figure 2). Second, the peak impact of foreclosures on the two different price series occurred between 2005 and 2010. Beginning in 2004, overall prices began to accelerate at a faster rate than prices that excluded future distressed transactions by two percentage points at the peak in December 2005. How can it be that when future distressed sales are excluded, prices increases are higher? It is because homes whose prices were accelerating the most in 2004 and 2005 were the ones most likely to go into foreclosure in subsequent years . In other words, they were overvalued the most and hence had the highest risk of future default. As the market collapsed, foreclosures became a major drag on home prices by a negative six percentage points at the peak month of distressed sales in January 2009. Third, since 2009 as the market began to slowly heal and normalize, the impact of including distressed sales had a positive impact on prices by nearly 3 percentage points in mid-2010 and mid-2012. Since then the impact has dissipated and as of May 2013, the gap has shrunk to one percentage point and is expected to slowly dissolve . One hypothesis for the high level of appreciation is that it has partly been driven by the subsequent sales after a distressed sale. If that hypothesis was true, we would observe that the CoreLogic index that includes distressed sales would be appreciating at a higher rate than when we exclude distressed, but those differences are very small, averaging only one percent in mid-2013.
In conclusion, the differences between the Case-Shiller and Corelogic HPIs year-over-year changes are moderate for any particular month and very small over longer-time horizons. Analyzing the impact of distressed sales on prices reveals what we intuitively knew; distressed sales had a substantial impact on prices during the bust. However, it revealed two interesting issues; the overvaluation impact of distressed sales in the mid 2000s, and that the impact of distressed sales on home price increases in 2013 is minor. Given the currently small influence of distressed sales on changes in prices, it is clear that prices are currently rising due to fundamental factors, such as the chronic lack of inventory, rising demand and high affordability.
 There are other differences such as different definitions of metro geography, different treatment on interval and geographical weights for the repeat pairs, and different inclusion of property types. There are also more minor differences such as the filtering and exclusions criteria for outliers. There is also the CoreLogic Case-Shiller quarterly index which has much broader coverage than the S&P Case-Shiller monthly indices, but in this post, we only evaluate the 10 and 20 city indices vis-à-vis the CoreLogic index.
 The 10 City Index comparison was from January 1988 to May 2013 and for the 20 City Index it was from January 2001 to May 2013.
 In the repeat pairing process, both transactions are excluded, so in 2005 many of the transactions that were occurring in the overall price index in later years became distressed, but at the time this was not known. In hindsight, it’s clear that those properties were overvalued.
 Analyzing several large hard hit markets reveals that the national trends generally remain true at the CBSA level.