Economist's Commentary: June 25, 2008
High Highs and Low Lows; the Story of A Bi-Polar Index
By Ken Fears
Manager, Regional Economics
With so many housing price measures it can be difficult to estimate where the market is and where it is going. But looking back provides valuable clarity on this subject.
For several years, Robert Shiller has been forecasting a very sharp decrease in home prices, possibly even more than 30 percent. This prediction sounds terrifying. Such a substantial loss of equity could be devastating to the economy and the wealth of nearly all U.S. homeowners. It would be, if Prof. Shiller were referring to the NAR median home price.
Thankfully he is not. Prof. Shiller is referring to his own index called the Standard and Poor's/Case-Shiller index. The S&P/Case-Shiller index uses a sophisticated modeling technique developed by two academics. For this reason it has garnered much clout. However, the devil is in the details. The Case-Shiller index looks at the change in the price of the same home in just 20 markets. These markets tend to be higher priced than the national median and larger price changes are given more weight in this index. Furthermore, the Case-Shiller index looks at the average price change, rather than the median. Taken together, these factors cause the Case-Shiller index to overestimate price swings.
Historically, the Case-Shiller index underestimated the performance of the U.S. housing market relative to the NAR median during the period from 1990 through 1998 by an average of 2.2 percent per year or 18.2 percent over this 8-year period. Thereafter, the Case-Shiller index overestimated the pace of housing price growth by 3.0 percent per year up until 2006; an increase of 21.4 percent over 9 years. If this is the case, then there should be no surprise that in 2007, now that housing prices are on the decline, the S&P/Case-Shiller index is falling by substantially more than the NAR median price estimate. Indeed, it also makes clear that Robert Shiller's robust forecast for a 30 percent to 50 percent decline in home prices is specific to his index, not the market.
There are other important points to be made about the Case-Shiller index. NAR uses the median home price as a measure of the price most representative of the market. Case and Shiller opt for the average price change. The difference is telling. Take a group of numbers; 3, 5, 7, 8, and 1,000. The median of this group is 7, the middle of the set of numbers. The average is 204.6 (e.g. [3+5+7+8+1,000]/5=204.6), which is far from the central clustering of numbers. While both of these statistics provide an estimate of a representative number for the set, the average allows outliers to skew it far above or below what one might consider a representative estimate of the set of numbers. In fairness, the Case-Shiller index does attempt to remove these outliers, but any vetting process is subject to errors in methodology and execution. Another issue is that the highest priced homes tend to experience the largest price swings because their demand is driven by interest rates and more susceptible to booms and busts. Also, the Case-Shiller index puts less weight on older homes and uses the initial price of a home to weight its relevance within the index period. The combined effect of these two specifications is to give more weight to newer homes and more expensive homes; homes that tend to be more susceptible to the building cycle and interest rates. So by allowing these outliers and giving them disproportionate weight in their index, Case and Shiller have created an index that is inherently exposed to the most volatile portion of the market. More extensive critiques of the Case-Shiller index can be found elsewhere.
The Case-Shiller index, like the OFHEO housing price index and NAR median price, provides valuable insights into how the housing market is moving, but not the entire market. It focuses on the most volatile portion of the market and overestimates the price swings relative to the total market. The Case-Shiller index might be used as a proxy for consumers' expectations during the market transition as it reflects the heightened, worst-case scenario that many consumers fear during a housing turn-down. However, if one were to use this as an estimate of real activity for the nation, one would miss the mark. If banks had used the Case-Shiller index to estimate reserve requirement prior to the sub-prime crisis in 2007, banks would have been caught with even lower reserves than they were and the impact could have been catastrophic for the U.S. economy.
All three of the housing price measures mentioned in this article can be useful for measuring different portions of the market. For the best view of the national market, however, letting the data tell the story is better than the most sophisticated analytical techniques that are skewed towards only a portion of the market.
This is one in a series of commentaries by the Research staff of the National Association of REALTORS®. Read more commentaries >
Comments? Questions? E-mail NAR Research.
NAR members, learn how you can add this commentary to your Web site, blog, or newsletter. Read more >
