What is the Case Schiller Index in Real Estate?

The Case-Schiller Index is a widely quoted estimate of residential real estate trends in over 20 metro areas. It was developed by Karl Case and Robert Shiller, who were frustrated with the accuracy of the "median prices" for housing that were commonly used previously to measure the price changes in various markets. It has been adopted from it's original format to encompass a much larger geography over the years, and has been in widespread use since the mid nineties. It is used by mortgage lenders, rating issuers, and other industries, to attempt to assess the risk in any measured housing market.

How does Case Schiller Index Measure Home Values

The Case Schiller Index measures home values in a unique conceptual way. It looks at prices of the same houses over time when they are re-sold. So if a home sold in 1995 for $200,000, and was resold in 2005, for $250,000, then it increased in value by 50,000, or by 2.226% each year for 10 years. From that point, and looking at many resales, they can make an attempt to understand what is happening in the markets that they track.

How Accurate is the Case Schiller Index in Measuring Home Values

Well, that's a good question! The only way to really know is to have a better estimate for such a thing, but since the Case-Schiller was primarily measured against "median home prices" in terms of accuracy, it easily won out. Why? Because "Median Home Prices" is a very poor way to measure how expensive real estate is.

What are the potential problems with the Case Schiller Index and Median Home Prices

There are problems with how Case Schiller

measures housing markets, but they

aren't talked about often.

The problem with Median Home prices is that it measures both how expensive housing is as well as which houses are selling. In residential real estate, not all houses sell at the same rate. Sometimes the "bottom" of the market is seeing high transaction volume, and sometimes the "top" of the market is seeing higher transaction volume than is typical. We actually just went through this period quite recently in 2010, and the reason was a large separation between the rates on Jumbo mortgages and normal residential loans. When the banks went into the credit crunch at the end of 2009, they decided they needed to charge higher rates for so-called Jumbo mortgages. These are mortgages that are so large, the government won't insure them against default. Normally, Jumbo mortgages are 1 or 2 percentage points higher than typical government loans, but when things got crazy it was closer to 5 or 6 points for a jumbo. This huge "spread" in interest rates, combined with the extra government incentives for first-time buyers (who were given money to buy inexpensive homes), meant that the number of "low" priced homes sold went way up - at the same time the number of high priced homes went way down. This had the affect of drastically moving the "median" home price in most markets, even though prices didn't budge very much in this area. Although this is a drastic example, it's quite common to see the top and bottom of residential markets move in different directions, and that's why Median Home Prices is an especially poor way to judge home prices.

 

Now the Case-Schiller is better. But it's not perfect either. First of all, it relies on a relatively small amount of data - instead of looking at all the home sales in an area, it only looks at ones that have "re-sold". And it basically (as far as I can tell) assumes nothing about the home has changed in a material way. Unfortunately, the problem there is homes are always changing, even when they aren't changing! A roof gets older every year, and the closer it gets to the end of it's life, the more of an impact it has on the home itself. At a minium, there's cosmetic dating, infrastructure aging and improvements, expansions, lot reductions (people taking a big lot and selling it off will impact the property). Basically, lots and lots of changes that are happening to many of these properties. Maybe Case Schiller works better because these things "even out" over time, but until there's a better metric to compare it to, we're stuck with this one.

 

Is there a better way to measure how expensive real estate is?

I would argue there is. Appraisers use it all the time, and since their job is evaluating properties value, it may make sense to look at it. It's the metric called dollars per square foot, or $/sq. ft. It has it's problems too - especially over small data sets - but the larger the data set is, the more accurate it tends to get. Prices per square foot jump around a lot by house style and age, but the distribution and sales patterns of those homes tends to be stable (most towns aren't adding new houses very quickly, so the "mix" of homes in the town is stable), so it does usually work itself out. First off, it moves around much less than "average median prices", because the differential between what buyers pay on a square foot basis for small properties and large properties isn't very big, numerically speaking, so it tends to be much more stable. And unlike Case-Schiller, it takes into account property improvements (or lack there of) in a real time basis. For example, if I had a 1000 sq foot home that I could get $200,000 for, and then I added a big family room of 300 sq feet, my 1300 sq foot home is now probably worth closer to $240,000. My $/sq foot actually goes down in this example, but not by much, and probably correlates well with the actual market change. In my book, how much you have to pay a square foot for a home determines how expensive a town is, so it's what I use most often.

Why aren't their any indexes that use $/Sq foot?

Well, there aren't any popular ones. But the real estate websites like Trulia and Zillow are starting to collect that data. I just looked at Ashland, for example and saw that they have the data, but it's not smoothed out very well. (It means they aren't looking at enough data points for each point).  It's worth saying that both pricing models require that certain variables "average out" and so neither is perfect.  I do like $/sq foot because you can do it over smaller data sets and smaller geographic areas quite accurately, which is what I need to do!

 

So in Summary, there's no perfect metric out there, and there are a lot of biases built into any numerical summary.  Be careful what you infer from the headlines!

 

Do Good Things Today!

Matt Heisler

*All information is posted in good faith and is assumed to be reliable, but may rely on third party information sources.