Canada’s housing price stats likely contributed to inflating bubble

Chart 1 Statistics Canada,Bank of Canada and Teranet-National Bank housing price indices

At the July 2014 Monetary Policy Report (MPR) press conference, Bank of Canada Governor Stephen Poloz announced the Bank would be keeping its policy rate in “neutral” for the foreseeable future. While introducing a new catchphrase – “serial disappointment” – the MPR report and the Bank governor’s comments gave short shrift to the over-heated Canadian housing market, which continues to be fuelled by historically low interest rates. Despite conceding “particularly strong” price growth over the past year and “near record-high house prices and debt levels,” the Bank insists housing is in for a “soft landing”.

While the lack of housing market information has been a popular topic of late, a closer look at the little available info on housing prices may shed some light on why the Bank has downplayed rising home prices, and why if or when the housing bust happens the Bank will say it didn’t see it coming.

The Bank’s Real Estate Market Indicators information webpage seems a reasonable place to start. It references three indicators: ‘New-housing price index’, ‘Resale-housing prices: Royal LePage’ and ‘Vacancy rate’. A link to Statistics Canada’s New Housing Price Index (NHPI) is provided. Other than brief two-sentence descriptions, no further information is provided for the Bank’s Resale Housing Price Index (RHPI) or vacancy rate. Sidebar links to ‘Graphs’ and ‘Historical Data’ provide NHPI and RHPI (but not vacancy rate) information dating back to 1992.

The StatsCan NHPI information page gets right to the index’s limitations: The NHPI is based on a survey of contractors, meaning the prices captured are contractor selling prices, not the final prices paid by new home buyers. This creates problems in adjusting new home prices for GST and other taxes, and would also underestimate new home prices during periods of general housing price inflation.

While there are more condominium apartments being built than houses (CMHC Housing Information Monthly, July 2014, Table A3-1), the NHPI only counts “single dwelling, semi-detached and row houses”. To facilitate data collection, StatsCan prefers to survey “builders who develop entire subdivisions”, further limiting the survey’s scope.

Effectively, the NHPI is a survey of tract homes in peripheral urban areas, a small share of the Canadian housing market that’s had a limited impact on surging home prices in recent years.

It’s worth noting that NHPI is the only housing price data used in the Consumer Price Index (CPI), despite the fact other housing-related CPI components (like mortgage interest) include resale homes and condos. StatsCan is well aware of the NHPI’s shortcomings; it recently proposed a major revision of the measure, and is exploring a broader Residential Property Price Index (RPPI) which would include the NHPI.

[Andrew Baldwin notes: New housing price indexes are generally less volatile than resale house prices. In a profound slump, builders will just stop building new homes if it is no longer profitable, so there is  always a floor for new housing prices that doesn’t exist in the resale market. Overbuilding in houses is also limited by the supply of residential land, but this is not such a constraint for condo developers, who can always build higher. So new housing prices are less volatile than existing housing prices, and prices of homes are less volatile than prices of condo apartments. The official CPI, even if one accepts its approach to measuring homeownership costs is reasonable, is much less volatile than it should be simply because of where its housing prices come from.]

Most Canadian housing market analysis includes at least one measure of resale homes. The Bank of Canada suggests it uses the resale index published by Royal LePage. However, the published quarterly Bank of Canada RHPI differ significantly from the quarterly Royal LePage House Price Survey stats.

While the Bank of Canada does not publish detailed methodology for its RHPI, a Bank representative noted:

“The Bank’s Royal Lepage house price series is calculated as a weighted sum of the price of detached bungalow (weight = 0.75) and the price of executive detached two-storey (weight = 0.25). The price of each type of housing is in turn a weighted sum of sub-regions, with weights set to be the sub-regional share of units sold as of a fixed date in the late 1980s. The “units” data were obtained from the Canadian Real Estate Association.”

However, the LePage survey collects “information on seven different types of housing,” only two of which are included in the Bank’s RHPI. And those two happen to be among the most expensive and least (re)sold – which makes no sense, assuming the objective is to capture the large share of the housing market not captured by NHPI.

What’s even more peculiar than the Bank’s RHPI is the fact it’s not been referenced in the closely watched Monetary Policy Report (MPR) since October 2008. And in the rare instance, it was “deflated by the total CPI” – which makes no sense, since what the RHPI includes is specifically excluded from CPI.

When asked to explain, a Bank representative quite bluntly admitted the Bank’s RHPI was inaccurate, unreliable and “almost no one, in or outside the Bank, uses this index.” Asked why this information isn’t indicated anywhere and why the RHPI the Bank now uses isn’t included instead, the representative suggested changes don’t just happen “one day” (only in government – and maybe the Bible – could 6 years = one day).

More recent Bank of Canada MPRs cite the Teranet-National Bank Composite House Price Index in place of its RHPI. The Teranet RHPI has been published since 2008 and provides 6 and 11-market composite indices dating back to March 1999. The Teranet index uses matched-pair sales based on provincial land registry data. While the index includes condominiums, it’s not clear how / whether they’re accurately captured.

Not surprisingly, the NHPI and RHPI trends are significantly different. There’s also a significant divergence between the Bank of Canada and Teranet price trends between 2003 and 2008. With respect to housing prices, Bank policy over that period would have been guided by a RHPI the Bank now regards as inaccurate and unreliable, and a NHPI that StatsCan concedes has major shortcomings.

While most analysts consider the Teranet RHPI to be fairly reliable and accurate, a reasonable argument can be made for separating out condos from other types of housing. The condo market differs from other housing markets, and has significantly driven up prices in the over-heated Toronto and Vancouver markets. A complete reworking of the NHPI is long overdue.

While it’s encouraging to see the Bank of Canada acknowledge and correct previous poor decisions, one can’t help but wonder why it designed its RHPI as it did in the first place. One also can’t help but wonder why neither the Bank’s now-discarded (yet still published and regularly updated) RHPI nor StatsCan’s current NHPI excluded condos. It’s not as if the condo craze just arrived in Canada and took everyone by surprise.

Past decisions based on faulty data by the Bank of Canada and Statistics Canada can help explain why the Canadian housing market is where it is today. [Andrew Baldwin notes: Big upward movements in actual housing prices that might have led to interest rate hikes and lower debt accumulation did not do so. Now we are left with overbuilding in many cities and big housing debts when the situation wouldn’t have been so bad with better indicators.]

Hopefully recent and proposed changes to Canada’s housing price indicators can help provide a more reliable picture of home price trends going forward. And hopefully they won’t be too late in coming.