I continue to work with some CREA data to find historic relationships between house prices and other factors. I posted earlier on the historic correlation between house prices and per capita GDP, which shows significant overvaluation on a nation wide basis. Today we’ll look quickly at inflation and then at some supply and demand factors.
As a reminder, this is dealing with aggregate Canadian data. This is not to suggest that house prices everywhere are significantly overvalued. Certainly there are a handful of particularly bubbly locales, but it is increasingly evident that even the more ‘reasonably’ priced areas are at the top of their long-term historic norms when measured against underlying fundamentals.
House prices and inflation
It was bubble guru and Yale economist Robert Shiller who famously calculated that house prices in the US have done little more than pace inflation over the past century.
In fact, the most extensive historic housing data set in the entire world verifies this finding. Price data for the Herengracht Canal in Amsterdam extends all the way back to the early 1600s. Remarkably, real (inflation-adjusted) house prices have moved around a long-term trend line over the past 350 years. That is to say that houses essentially pace inflation over time. Source: Piet Eichholtz as cited by Steve Keen
Note that periods of under performance relative to inflation were eventually corrected with rapid moves to the upside, while periods of over performance relative to inflation resulted in an eventual correction to the down side. Note however that this can happen via a reduction in house prices or a stagnation in house prices while inflation catches up (the ‘soft landing’ thesis, if you will).
Here in Canada, inflation is often measured using the Consumer Price Index which tracks changes in a basket of consumer goods. If house prices typically track broad measures of inflation over a long enough time horizon, how have we fared here in Canada?
Cumulative house price gains have far exceeded the cumulative growth in the CPI over the past 30 years. Now it should be acknowledged that the starting point makes a great deal of difference in this case as it could be argued that we may have started from a period of significant house price undervaluation which would imply that the growth in housing relative to inflation is merely “catching up”. This is a valid point, though the massive divergence between the two trends certainly argues against the notion that house prices are simply catching up.
Of note, many have criticized the CPI for understating inflation. As John noted when we discussed the New House Price Index (NHPI), it is this index that is used as the input into the CPI measure, arguably understating true inflation. I went ahead and plotted the same data series but this time I assumed that CPI had understated inflation by 50%. There is no rationale for choosing 50% as an adjustment other than to simply illustrate how significantly government stats would have to be adjusted over each of the past 30 years in order to compensate for the massive divergence. Here is the result:
It would essentially mean that house prices would have lagged inflation over much of the previous 30 years before rapidly rocketing past the inflation trend line in the past 5 years.
House prices and population growth
I want to now take a moment and discuss two often-cited arguments to support house prices at current levels:
1) Immigration and population growth will continue to drive prices.
2) Canada has not experienced the level of overbuilding experienced by the US.
Let’s start with the first statement. While immigration and population growth can lead to higher property prices, they do so only under the relatively rare instances that the housing industry either cannot build houses fast enough to meet the demand or that current land restrictions prohibit building in highly desirable locations. This second point is the primary focus of Demographia, the organization that releases the annual affordability rankings of international housing markets.
Aside from those two isolated instances, population growth in and of itself should have very little impact on aggregate house prices. Nevertheless, let’s see data:
Here we see cumulative house price increase since 1980 plotted against increases in total population and growth in immigration.
If we broke down the annual change in house prices and plotted them against the annual change in total population and immigration as a percent of total population in each year, we see the following:
Note that in each case the correlational coefficient is negative, strangely enough, though both are insignificant. In essence, these findings seem to mirror the findings of Steve Keen who found that immigration and population change had little impact on house prices in and of themselves.
One of the most commonly used arguments in support of the stability of our housing market relative to our southern neighbours is the supposed lack of overbuilding in Canada. It is alleged that rampant speculation and overbuilding led to the supply glut that we currently see south of the border.
To validate this statement, I plotted the annual rate of housing starts per 10,000 people in both Canada and the US all the way back to 1980. I found population data from Stats Canada and the World Bank, and housing starts data from CMHC and the National Association of Home Builders (NAHB). Here are the results:
As you can see from the data, the US hasn’t built more homes than Canada on a per capita basis since 2001. Over the entire 30 year period, housing starts per 10,000 averaged 61 in Canada and only 54 in the US.
It appears from this data that the “massive overbuilding” argument is highly suspect.
More to follow…