MARCH 22, 2011
I have to apologize for my short posts lately. I will be back to more in depth analysis shortly, but this is a very busy time for me.
I would like to very quickly discuss the cool interactive global property price feature over at the Economist. I've received a few emails from people asking why it shows such a tiny, insignificant rise in Canadian house prices compared to other nations. Here are some screen shots to show what I mean (thanks to JD for emailing them to me):
Based on this, Canada is a bastion of fiscal prudence. Well, not so fast.
The Economist simply states that their data source is 'Stats Canada'. The data line only fits with the data from the New House Price Index (NHPI) (CANSIM table 327-0005 for those who have paid access to that data).
From Stats Canada regarding the NHPI:
"In the new housing price survey, reported prices are adjusted for changes in quality of structure and land. This is important because the NHPI attempts to measure changes in price over time of identical houses in consecutive periods. Most often the estimate of the value of the quality change provided by the respondent is used to adjust the reported price to an estimate of "pure" price, unaffected by quality change."
What does this mean in plain English? It simply means that the NHPI quality adjusts its data. Let's say that next year, all new houses are built with gold shingles on the roof. Real gold. Because of this, the price of a new home sky rockets by 100K in the year. The NHPI adjusts for this improved quality to show no increase in the actual price paid. In their eyes it is now an 'apples to apples' comparison, even though it in no way reflects the fact that people just paid 100K more for their new home. However, if gold shingles became the new fad, it would require a significant expansion in debt levels to afford it, and it would likely push the median resale house price even further away from underlying measures of value like the price/rent and price/income multiples.
It also adjusts for significant differences in the average size of dwellings. We know that the average dwelling size has risen markedly in Canada. In fact it has risen from an average of 1050 to nearly double that today. House prices have risen accordingly. Yet the NHPI shows a much more muted rise in prices since it adjusts for the fact that houses are getting larger. If the average new house built next year were to double, it would not show a significant increase in the NHPI despite the fact that it would require substantially more money to purchase a new home. Bizarre!
The real-world accuracy of the NHPI
There's perhaps no better illustration of the real-world accuracy of the NHPI at calculating house price appreciation than the fact that according to the NHPI you can build a brand new home in Vancouver today cheaper than you could in 1993.....and that is in NOMINAL dollars.
Let me repeat: If you bought a new home in Vancouver in the early 90s, someone could build a comparable house today in Vancouver (including land value) for cheaper that you did in the 90s. And that's not even considering inflation.
If that strikes you as laughable (and it should), then you can see why I'm skeptical about using the NHPI as a measure of house price increase. It masks the expansion in debt, as well as the variance from measures of underlying fundamentals.
Let's not make their mistake...
As an example of how a quality-adjusted house price index can mask an unsustainable increase in house prices consider this 2004 paper by the New York Fed:
"As our analysis shows, the home price index used can have dramatic implications for one’s assessment of whether a home price bubble exists. Of the indexes available, we believe that the constant-quality new home price index is most appropriate for this assessment because it is the only one that explicitly controls for changes in quality over time. Any home price series used to assess the existence of a bubble should attempt to control for location and changes in quality."
You can probably guess their conclusion: Home prices show a significant rise based on other measures, but a much more muted rise when compared to the constant-quality NHI. We know how that ended. Let's not repeat that mistake by assuming that since people are buying better quality houses (or simply just larger houses) that they aren't also overextending themselves in the process.
Other issues with the Economist data
On one final note, the data series in the Economist picks up in 1990. In the late 80s and into 1990, the average resale home in Canada shed some 15% in a nation-wide housing recession. To pick a start data the corresponds with the end of a significant real estate correction skews data to begin with.
The bottom line is that I'm highly suspicious about the real world accuracy of the NHPI as a measure of house price appreciation. The Economist made a mistake using this data. Let's not make the mistake worse by assuming that this index in any way reflects the true state of the Canadian real estate market.