Wanted: better house-price indexes

We might want to order up a huge grain of salt to go with the latest housing stats released by the Canadian Real Estate Association (CREA). Actually, we may want to get a truck load of the stuff from a salt mine somewhere.

In a June 15 press release, CREA announced that the average price for a house sold on the Multiple Listing Service (MLS) shot up to a record high of $319,757 in May. "Over the past four months, the national MLS residential average price has recovered 16.4% from the low in January," added CREA in its press release.

This data does not gel with other data series on the economy and housing market. Last checked, Canada was still in recession: the unemployment rate in May rose to an 11-year high of 8.4%. Since the recession began 11 months ago, Canada has lost over 360,000 jobs — with more than 10% of that total coming in May. Hmmm ... how can house prices spike so much when so many people are losing their jobs or can't find work?

Meanwhile, other measures of Canadian house prices are showing declines. The Teranet-National Bank National Composite House Price Index dropped 5.8% during the 12 months to March. And, according to Statistics Canada, selling prices for new homes fell 3% over the 12 months to April (the sharpest deceleration in new home prices since the recession of the early 1990s).

The problem with using average prices in the housing market, as mentioned in my blog, is the failure to compare apples with apples and oranges with oranges. In each period, the average price is calculated from a different mix of houses. There will be differences in the proportion of houses by dwelling space, style, number of bedrooms, lot size, region, and a variety of other characteristics.

In particular, when there are a higher number of sales from regions with more expensive homes, the CREA index will signal an increase in price simply due to this shift in the mix of houses. This, indeed, is what has happened in recent months: sales in high-priced cities such as Calgary and Vancouver are rising more quickly than elsewhere and distorting the CREA index.

CREA does footnote this problem in its press releases. But why even bother calculating such a faulty indicator and reporting the results as a lead item in monthly press releases? Why not headline a less biased measure?

That would be an index like the one produced by Teranet and National Bank. Launched about six months ago, it does a much better job isolating pure price change by using the "repeat sales" approach.

Just prior to its release, TD Financial Group released a study of house prices in Canada, lamenting distortions in measurement. One of its recommendations was that "Canada needs a repeat sales home price index." They got their wish when the Teranet-National Bank index came out shortly afterward.

In the United States, indexes based on the repeat-sales formula are now mainly used. Of note is the S&P/Case-Shiller Home Price Index.

It would seem to be time for Canada to follow suit.

By the way, the S&P/Case-Shiller Home Price Index also doesn't jibe very well with CREA's house price stats. It is showing a drop of 18.7% over the 12 months to March and a 2.8% drop from February to March.

© 2009 Rogers Publishing Limited All rights reserved.