Wednesday, October 23, 2013

September 2013 Residential Construction Activity - Vancouver CMA

Below are CMHC starts, completions and under construction for Vancouver CMA to September 2013. Starts and completions are displayed as 12 month sums:
The first graph below is completed but unabsorbed inventory for single and semi detached properties, the second graph is showing the sum of single and semi detached under construction and completed but unabsorbed, which provides a proxy for near-term inventory of new properties of that type. (Single and semi-detached typically have construction cycle of less than a year.)
By the unabsorbed plus under construction measure, single and semi detached inventory has been at a level unseen since the early 1990s since early 2013.

Wednesday, October 16, 2013

Vancouver CMA Price to Rent Ratio

A popular ratio used to measure relative housing valuations is the "price to rent ratio" that takes a benchmark price and divides by the annual or monthly rent. This forms a type of price-earnings ratio of housing from an investment perspective. There are some significant caveats with this measure, not least the data sources used. This post will outline the various sources of data and what I believe to be the best measure of a pure price-rent ratio as well as some of the factors and limitations that are inherent in the measure.


Prices are measured in various ways, from simple average or median prices (average can be found on website, the median can be found in various places, for example the Demographia survey). Royal LePage has data going back to the 1970s to the municipality level and attempt to adjust for sales mix. Other measures include the Teranet HPI that, like the US-focused Case-Shiller house price index, uses sales pairs to attempt to adjust for quality differences in the data. A more recent measure starting to be adopted across the country is the MLS-HPI that uses a form of hedonic adjustments to perform a quality adjustment on sales. Although not using the same methodology as the Teranet model, the MLS-HPI has broadly tracked the Teranet HPI.

The MLS-HPI, Royal Lepage, and average data are broken down by housing type, broadly single detached, townhome, and condo. Teranet provides only aggregated data publicly but does track data based on housing types.


Rents are more difficult to measure, mostly because the sources of the data are not collated anywhere save through CPI measures ("rented accommodation" and "owner equivalent rent") and some other surveys that have been performed over the years. Metro Vancouver and CMHC have attempted to measure rents through broad-reaching surveys and other datasets with some success. CMHC surveys rents from a pool of purpose-built rental units and has a dataset extending back to 1992. CPI data extend back further than this.

Comparing prices and rents

Price-rent ratio, at its core, is a measure of how favourable an investment housing is at current valuations. A higher price-rent ratio generally indicates a less favourable investment and a lower price-rent ratio indicates a more favourable investment. To understand what price-rent ratio is actually indicating involves more in-depth analysis than what is possible in a blog post. Nonetheless several broad factors can and do influence price-rent ratios being higher and lower. These factors include:

  • Quality and age. A property that has a more stable income stream or requires less maintenance will generally have a higher price-rent ratio. For example a new condominium requiring little maintenance outlays and attracting high-quality dependable tenants will be a safer investment that requires less discounting. As a building ages it can become "dated" (resulting in less favourable rental terms for the landlord) or incur additional maintenance that will require cash outlays from the owner. As a unit ages its price-rent ratio will tend to fall.
  • Financing. Both short-term financing (ie 5 years or less) and longer-term financing (say a 30 year mortgage) can influence price-rent ratio. Investors will pay higher prices if the financing costs are lower (like a bond), though the caveat is that, in Canada, financing needs to be periodically renewed and that investing in property has a fixed land component that never depreciates. This means that while financing costs can decrease due to lower interest rates, if rates go higher at any point in the future (not just over the course of the loan) this benefit is lost.
  • Density. A low density property in an area undergoing rezoning or other density changes will factor in both current and future land use into its price. This means that, say, a 60 year old bungalow in a desirable higher-density area will have a high price-rent ratio. Over time, as the area continues to increase in density, the land becomes more valuable and the price-rent ratio will tend to increase over time. This means that many properties will carry a "ludicrous" price-rent ratio but that does not necessarily mean they are overvalued.
  • Rents and income. An area can be undergoing gentrification or population growth that put pressures on rents or imputed rents. This means, ultimately, that the market expects rents to increase faster than inflation and this will push up future earnings from the property. This will lead to a higher price-rent ratio.
  • Consumer surplus. If the marginal buyer is not solely focused on cash flows but on ownership this can cause prices to rise above what what an investor may be willing to pay. If owner-occupiers place an intangible net benefit on home ownership they will pay a premium relative to renting. This forces down yields for landlords, although they benefit from the capital appreciation.

There are other factors, such as raw speculation, that affect price-rent ratios. It is not immediately true that all factors that can plausibly affect price-rent ratios are unsustainable. From a net-present-value perspective there are many factors that are "value based" -- meaning future earnings from operations are reasonably expected to support a competitive inflation and risk-adjusted return -- and some that are "speculative" -- meaning future earnings are supported more by the future sales price and less by income. (My use of "value" and "speculative" should not connote good and bad, but should acknowledge how one makes one's money.)

Determining which measures to use

Determining which measures to use for the price-rent ratio involves correcting for sales mix but also correcting for other factors that can determine which factors are supported by incomes and those that are supported only by speculative activities. It can be difficult to separate certain premiums based on density increases from depreciation, finance, or rent-inflation-related factors.

Luckily, to help with this, we have a measure of quality adjustment in the form of the Teranet and MLS-HPIs that reduce the effect of depreciation and quality in our measure. In terms of density increases, this can be partly ameliorated by focusing on already-dense properties such as apartments that are unlikely to undergo significant density-increasing redevelopment in the foreseeable future. This leaves us with the residuals of rental inflation, financing, and of course the catch-all of speculation.

In terms of which rents to use, to compare a same investment over time, given we have a quality-adjusted measure of apartment prices, the best measure is to use a quality-adjusted measure of apartment rents. Here we are in luck because the CMHC rental survey measures purpose-built apartment rents over time. Since purpose-built apartment stock is generally not being replenished in any significant way (though that may change in the coming years), the rents measured on these units are de facto quality-adjusted.

So we have our chosen price and rent measures. Using the Greater Vancouver apartment MLS-HPI, dividing by the CMHC one bedroom apartment rent and normalizing (to 2005), we get the price-rent ratio that I often reference on this blog:
This measure shows the impact of rising rents on the ratio since its peak in 2008, and how elevated the ratio is compared to its value through the 1990s. This measure will be affected by the following broad factors:
  • Market expectations of future rent appreciation
  • Market expectations of future mortgage rates
  • Speculation and other non-cash-flow-supporting returns
The measure broadly corrects for:
  • Expected density increases
  • Quality and age
The price-rent ratio graph above is by no means a definitive measure of valuation. It is a measure with imprecise factors and should be used as part of a larger suite of measures to form an argument for the sustainability of current housing valuations.

As to what is a "reasonable" price-rent ratio, well, that is the question, isn't it?

Wednesday, October 09, 2013

Greater Vancouver Market Snapshot September 2013

Below are updated sales, inventory and months of inventory graphs for Greater Vancouver to September 2013. (see REBGV news releases) (click on image to enlarge)

The scatterplot of 6 month price changes and months of inventory is below. The most recent datum is the brown dot at about (MOI=6.3,price_change=3%) . The trend is roughly in line with past years.
The other graph I have added is the price-rent ratio for a Greater Vancouver apartment, using REBGV apartment MLS-HPI and CMHC-reported rents (extrapolated, as they are only reported once per year), normalized to January 2005. By this measure, price-rent is currently 15% above its value in 2005 and 27-64% above the range of values from 1992-2004. 

A caveat about this graph is that it aggregates various price tiers and municipalities into one. If higher valued apartments have appreciated more than lower valued ones this would mean price-rent on the lower end is lower than this graph would indicate (and vice versa). Put another way, price data may have been skewed by higher-tier purchases appreciating faster than the lower-tier ones. Lower tier apartments may be more likely to be rented out, in which case from an investment perspective the price-rent ratio in the graph below may be on the high side.
My estimates for August were for inventory of 16663 (actual 16155) and sales of 2353 (actual 2483) based on estimating average changes from August of years 2005-2012. Using the same technique estimates inventory and sales for October of 15605 and 2657 respectively (MOI=5.9).