Is Case-Shiller the Best Measure of U.S. Housing Prices?

As I read another press release on the Case-Shiller index ("the Index") for the month of August, I wonder whether this index is the best measure of assessing national home price trends. I ask this question after looking at the 20 cities in the index. It appears to me, after looking at the map on the Standard & Poor’s website for the Index, that the cities in the Index are located in areas particularly hard-hit by the housing crisis.   Seven of the cities are located on the West Coast, four on the East Coast, two in Florida and four in the Midwest.   Only Dallas is represented in the Southwest and Denver is the only centrally located city. I am sure it is coincidence, but this housing/financial crisis is most affecting the blue-collar Midwest (Detroit and Cleveland, due to the auto and manufacturing industry), and the residential overbuilding and reckless financing that occurred primarily on the West Coast, particularly Phoenix, Las Vegas and southern Florida. These areas include at least six cities in the Index, or 30% of the Index.   These areas also saw the highest home-price gains from 2000 through 2006.

I also looked at the individual cities in the Index. The 20-city Index tracks sales in 20 metropolitan areas. The 20 areas are: Seattle, Portland, San Francisco, Los Angeles, San Diego, Las Vegas, Phoenix, Denver, Dallas, Minneapolis, Chicago, Detroit, Cleveland, Boston, New York, Washington D.C., Charlotte, Atlanta, Tampa and Miami.

I wondered if these metropolitan areas represented the largest populated metropolitan areas in the country? Are these the most populated cities in the country?  I went to the US Census Bureau’s website and researched the most populated metropolitan areas and cities. The following table is the results:

Las Vegas is number 28, Denver is number 26, Seattle is number 24 and Boston is number 23.

Next, I looked at the 20-most populated metropolitan areas. Again, according to the US Census Bureau, the following table represents the 20 most populated metropolitan areas:

 

Five of the largest metropolitan areas of the United States are excluded from the Case-Shiller index. The Las Vegas metropolitan area has a population of 1,836,333, which is approximately number 28, and Cleveland-Elyria-Mentor is approximately 25^th with a population of 2,096,471. 

I don’t know how much different the Index would be if Houston was included and Las Vegas was excluded, for example, or if any other substitutions were made. In my opinion, the monthly 10 and 20 city composite index, for whatever reason, is weighted toward the areas that are, and have been particularly affected by this housing crisis. This is another example that individuals should carefully read and understand the economic statistics that are reported by the media every day.

The S&P - Case-Shiller U.S. National Home Price Index is a much more accurate index of housing prices. Below is an analysis of this Index since Q1 2000:

According to this Index, the 1^st half of 2008 saw an 8.9% decline in national home prices. YoY, from Q2 2007 to Q2 2008, the decline was 15.4%. The 20-city composite Index reported a 16.6% drop.

Comments

[Eric,Virginia:]

Regarding which MSAs are included and which are excluded, being a former Economist with the U.S. Bureau of Labor Statistics, we chose our Metro areas based on 4 tiers: Tier 1 were the largest MSAs, and hence, "certainty" picks (100% of the MSAs in tier 1 were selected). Tier two were the second largest group of MSAs, so, 80% of tier 2 MSAs were selected, tier 3 MSAs, 50%, and a much smaller # of tier 4s. Not knowing exactly Shiller's tier structure, I'm guessing (from the MSA pop. stats) that the five largest MSAs (NY-Philly) are the tier 1s, #6-15 MSAs are tier 2 (Houston-Seattle), tier 3 MSAs starts with #16 (Minneapolis), and so forth. Something like that.

The point is that if you pick according to your tier structure, and you select the non-certainty tiers randomly (i.e., Houston has an equal chance of being excluded as, say, Atlanta), then you have a statistically valid sample.

Of course, with only a 20 MSA sample size, you do have a larger sampling error, and hence, a greater likelihood of the data being skewed, as you have pointed out. And, you rightly point out that the National survey is the most accurate because, of course, the sampling error is smaller. But, for whatever reason, the 10 and 20 city indexes do seem to get all the press.

2008 Oct 29 09:25 AM

[John Corey:]

The index specifically excludes certain properties or types of property. Anything that has not been sold twice is out (i.e., all new build) plus condos and other attached property. The focus is on single family residential based on what the methodology page said the last I read it. This is neither good or bad in some ways. It just might not be what people assume. The index explains all the details on the website.

Now, if you overlay the fact that condos are out with the fact that some city centers have a heavy concentration (space is limited so housing is mostly stacked) that can explain why NYC with be out. It also tends to distort what people will see in Miami vs. what the index claims is going on in Miami.

2008 Oct 29 10:43 AM

[User 288155:]

"Five of the largest metropolitan areas of the United States are excluded from the Case-Shiller index."

If you scroll down to page 9 on the link below, you'll see exactly which metropolitan areas are included.

www2.standardandpoors....

2008 Oct 29 01:01 PM

[netreality:]

And also the fact that properties that sold at least twice over the given time period is also a filtered pool. Many many houses turn over only once a decade or less and would be excluded. These are usually the ones in the more desireable neighborhoods, better school districts. Parents don't want to move and disrupt children's schools. Properties in good school districts hold their values much better than the opposite. Case in point: Palo Alto, CA. Very low inventory, prices still rising today! However East Palo Alto, gang ridden, crappy schools, tons of turnover, foreclosures. 1 mile and a world apart, but represented MUCH higher in volume in the Case-Shiller San Francisco Bay Area index.

2008 Oct 29 01:45 PM

[John Preston:]

First flaw is not the cities or MSA's...its the nature of the data....CS tracks JUMBO properties....those with prices above what Fannie and Freddie finance...

There has been no adequate mortgage financing for these properties ALL YEAR long....that distorts the data...along with the other valid concerns discussed above...

In normal markets, in my area of AZ, we see perhaps 1/3 cash buyers and 2/3 of the buyers financing these jumbo property purchases. In the past 5 months..we have seen far fewer transactions...down about 60% yoy, with cash buyers accounting for 95% of the transactions...we are missing those jumbo buyers who chose to finance, or must finance....

The CS index is at best a trend indicator....

2008 Oct 30 09:43 AM

[sumosama:]

great info, all! Thank you!

2008 Oct 30 12:15 PM

[GUMP:]

C/S also does not track condo units, etc., only single family homes. I believe a much more accurate view of the housing market is produced by a company called Radar Logic. Their index is used by a number of investment banks to track, and trade, housing markets across the country. It consists of 25 MSA's and includes condos. Full information can be found here:
www.radarlogic.com/

2008 Oct 30 12:27 PM

[kurt walter:]

Can you buy a residential property, put 20% down and get a positive cash flow in rental income after all expenses? If you can't, the property is overpriced. If yes, you might have a good deal.

2008 Oct 30 12:31 PM