For the past couple of months I have been cautioning our readers against making any big conclusions about Toronto’s real estate market based solely on changes in average prices. Toronto has seen a big decline in the number of sales of high end homes this year which has been exaggerating the decline in prices. For more on this read my previous posts Toronto Land Transfer Tax Exaggerates Housing Price Decline and Making Sense of Toronto’s Real Estate Decline in October.
At a national level, economists from TD Economics noticed that national average prices were being skewed down because of steep declines in sales in British Columbia, where average home prices are the highest in Canada. Last month TD Economics published a report titled A Different Look at Canadian Home Prices where they introduced the TD Home Price Index as a more accurate way to measure changes in national prices.
I invited one of the authors, Economist Grant Bishop to answer a few questions about their report.
John Pasalis: Hi Grant, thank you for taking the time to answer a few questions about your report. Can you start off by explaining why Canada needs a Home Price Index? What’s wrong with using average prices to track national house prices?
Grant Bishop, TD Economics: National-level average prices are distorted by swings in sales volumes across different markets. By applying consistent weights, an index strips away these market-specific swings and is a better indicator of the value of housing nationwide.
Presently, house prices at the national level are reported as an average across all properties sold in the month, quarter or year. The problem is that the average house price is reported as representative of the average value of the housing stock. It is only a measure of those houses sold; not those that never went on market. Certain high-priced or low-priced markets may experience relatively high or low sales volumes in any given period. The number of sales is highly variable between cities and over time. Sales volumes can thereby generate unrepresentative swings in observed house prices.
An aggregate measure at the national level should provide a consistent picture of how home values are changing over time. The index should be representative of the stock of housing rather than just those houses that are sold. An index that removes the effect of momentarily hot markets is necessary to remove the distortion from these temporary high sales volumes.
As a stylized example, consider two cities each of 100 homes. In city A, all houses are worth $200K. In city B, all houses are worth $100K. The average value of housing is then $150. However, consider a period where 30 houses are sold in city A at $200, and 10 houses are sold in city B at $100K. The observed average price is then (30 x 200K + 10 x $100K)/(30 + 10) = $7000K/40 = $175K. Because more houses were sold in the higher value city A, the average price was distorted upwards.
As a more concrete example, from the 1981 census, Vancouver constituted 9.3% of all owner households in our 24 city sample. From the 2006, census Vancouver constitutes 10.5%. Despite interprovincial migration, a given city’s proportion of homeowners moves relatively slowly over time. Compare this with sales volumes: In July 2007, Vancouver represented 8% of all national sales. By July 2008, Vancouver’s sales were down to 5% of the national average. Vancouver is a high-priced city and its lower sales volumes diminish its weight in the national-level average, distorting this measure downwards. In contrast, an index ensures constant weighting so that the national-level measure better represents the price of the overall stock of housing.
As a macroeconomic indicator, home values are very important. Canadians hold around 35% to 40% of their wealth in their principal residence. Home values then have a substantial impact on households’ decisions to spend, work, and invest. It’s important that we’re consistently gauging this indicator. Moreover, even though people buy houses locally, not nationally, the impact of such reported swings can create misperception about the value of one’s own home, inducing more pessimism or optimism than is warranted.
CREA’s [Canadian Real Estate Association] average of house prices is certainly not wrong. It measures what it measures. However, as a measure of the value of housing, we contend that an index with constant weights is a more consistent gauge. An average of sales is prone to swings in different markets and thereby misstates the actual average value of the stock of housing.
John Pasalis: How does the change in average prices reported by CREA compare to the change in TD’s Home Price Index?
Grant Bishop, TD Economics: CREA’s house price in October was down 10.9% Y/Y while the TD HPI showed a less steep decline of 4.6%. From 2002 until early 2008, the year-over-year percent changes within the TD HPI were generally consistent with CREA’s average. Over that period, the level of CREA’s average generally exceeded that of the TD HPI. For instance, when house prices peaked in May 2008, the TD HPI records a $340,046 price while CREA reports $345,362. However, the greatest – and most important - difference has been during the post-2007 period of deceleration and decline. But turning points are when accuracy counts most. With house prices now in a year-over-year decline, it’s critical to gauge the magnitude of the fall consistently.
John Pasalis: The TD Home Price Index does not control for unit-type shifts (e.g. changes in sales volumes of high end homes vs. starter condos). If you could take unit-type shifts into account, what impact do you suspect that would have on the Index.
Grant Bishop, TD Economics: Using repeat sales to control for quality across time is an important feature of the S&P/Case-Shiller index that is used in the United States. The TD HPI strips away the distortion caused by sales volumes in different markets. However, we agree that controlling for unit types and quality is key. Given the downturn in the market, it is likely that we would see a lower decline if controlling for unit types. Specifically, higher quality homes will likely be held off market to a greater degree and new homeowners will likely look down market. This would mean that the average price from observed sales would be biased downwards.
We do project a movement towards higher density and lower cost units. In a large centre like Toronto, this has both cyclic and structural elements: Over the downturn, income growth will stagnate and younger cohorts especially will look towards cheaper options. Retiring boomers are prone to downsizing. Although the latter may buoy the luxury condo market, their new condo will likely be cheaper than the single-detached that they’re cashing in. Of new housing, we project a more rapid fall in singles construction than in multiples. Overall, the near-term trend should be for “hamburger rather than steak”.
In the data, we see Toronto overall sales declining year-over-year (-22% in August and a further -7% in September), but these were led by declines in sales of single-detached units (-24% in August and a further -9% in September). Townhouses and condos sales have also declined year-over-year but not quite to the degree. As well, although there’s a decline in price story as well, the proportion of Toronto homes selling over $500K have declined from 18% of sales to around 14%. Indications are that those who can wait to sell are doing so.
John Pasalis: What do you feel is the key take away message from your report?
Grant Bishop, TD Economics: Firstly, Canada needs better housing data, and consistent measures of prices. Homes are very important to families, and housing plays a large role in the economy.
Secondly, especially in a period of high volatility, it’s important to measure indicators consistently. Any statistic can be biased by its measurement and is key to consider what is really being measured.
John Pasalis: What’s your outlook for Canada’s real estate market over the next year?
Grant Bishop, TD Economics: Over the coming year, we forecast a price decline of -6.1% nationwide under our base-case forecast and a -10.7% under our pessimistic scenario. The greatest declines will be seen in the overbuilt markets of Western Canada where the shock to commodity prices will markedly depress income growth.
Between our base-case and pessimistic forecasts, the devil’s really in how credit turmoil is resolved and plummeting consumption rebounds stateside. Canada is being buffeted by the three C’s (credit, commodities, and cross-border trade) and this is going to cause stagnation in income growth for Canadians over the coming year. The high erosion of affordability shows that prices had disconnected from incomes and this is now being reined in rapidly. In the long-run, house prices can’t exceed what people earn, have saved, or banks are willing to lend on the basis of their future earnings. We expect prices to return to a more affordable level that is more in line with income growth and interest rates.
John Pasalis: What’s your outlook for Toronto’s real estate market over the next year?
Grant Bishop, TD Economics:Our forecasts for Ontario are for a decline of -4.5% under a base-case and -9% under a pessimistic scenario. By sales, Toronto comprises 40% to 50% of the province’s real estate market. The province and its major city are going to feel severe strains from the downturn in exports and a sagging manufacturing sector. However, the GTA should nonetheless fare better than the outer Golden Horseshoe, where there are signs of oversupply. Affordability in Toronto is within reasonable limits but prices will nonetheless feel some significant downwards pressure.
December 2, 2008Market |