I am re-posting this blog in response to the redfin post: http://blog.redfin.com/blog/2008/10/that_was_a_doozy.html
The Case-Shiller method for home determining housing prices is another glorious product of the 1980s; while it may lack the flair of Madonna and the early years of MTV (back when it played music), its consistent benchmarking and dependence on the “repeat sales method” can only be dwarfed by the Rubik’s cube.
Now that the quota for 80s references has been hit, let’s get into the “nitty-gritty” of the Case-Shiller index…
The Case-Shiller index has been seen as a trusted tool for housing prices in the United States since the 1980’s when it was introduced by Case, Shiller, and Weiss’s research principals. The technique is built around the “repeat sales method” – which uses data on properties that have sold at least twice, in order to capture the true appreciated value of each specific sales unit. Sales pairs are designed to yield the price change for the same house, while holding the quality and size of each house constant. This methodology is considered by many as the most reliable means to measure housing price movements and is used by other home price index publishers, including the Office of Federal Housing Enterprise Oversight (OFHEO).
I cannot argue that the “repeat sales method” works well in well established markets with normal turnover. In places like Greenwich, CT, where the only thing that changes owner to owner in a home is the region that their oriental rug comes from, this methodology is perfect. Nonetheless the methodology fails to properly identify the emerging markets and places that are becoming hot spots to buy. It only takes into consideration places that have sold on the market, not homes that are just listed, and places with relatively no turnover get overlooked. This is an obvious observation when you look at all that the different home sales/home types that the methodology excludes.
The methodology will always exclude from its calculations all new construction, as well as properties with substantial physical changes. Furthermore transactions (or similar transactions), where the property type designation is “changed” (e.g., properties originally recorded as single-family home are subsequently recorded as condominiums), suspected data errors where the order of magnitude in values appears unrealistic, homes that sell more than once within six months, and condominiums and co-ops are specifically excluded!
Data related to homes that sell more than once within six months are excluded from the calculation of any indices. It is assumed that the transactions 1) are not arms-length, 2) precede or follow the redevelopment of a property, or 3) are fraudulent transaction. This means all those flipped homes with their trendy interior designs and staged IKEA furniture are completely excluded from the calculations. This makes it almost impossible for areas that are going through gentrification to be properly priced using the Case-Shiller model.
Take for example, New York City. Thus far in 2008 there have been 12,579 co-op sales all of which play no role in the evaluation of the Metropolitan Statistical Area (MSA), to which New York City belongs. That means, in 2008 if co-op sales remain constant, approximately 17,150 co-op sales will not be factored into the data used for the Case-Shiller methodology. That is a lot of excluded records! New York City is clearly a place where condo and co-op sales are much more prevalent then single family homes, but as long as NYC has more single family homes selling for a higher price than the last time that home was sold, the index will go up. Even if the new condos or older co-ops are selling at a much lower price then the single family homes. This clearly illustrates that the index is upwardly biased, and ignores the fact that first time home buyers who usually buy condos or co-ops, provide the bump for condo and co-op owners to purchase a single family home. Thus the sales prices of condos and co-ops are a good indicator of what the spending power will be for buyers that are leaving the city and looking to buy in the surrounding suburbs (i.e., where all the single family homes are). To me, ignoring a key economic indicator like this is no different then throwing out the baby with the bath water…
I understand that no model is perfect, but is the Case-Shiller method fatally flawed? No, because if you live in an area where there are only single family homes, and where sellers don’t rush the sale by going under market value, and there happens to be no major changes to any property sold, and the only thing firefighters in this place have to do is rescue cats in trees, the model is perfect….