Friday, March 9, 2012

Why Rent-to-Resale Calculations are Bunk

We read a great blog post recently by John Pasalis of Realosophy: click here
John touched on 'rent multiples' and that several reports on the Canadian housing market have used this measure to determine that Canada and several areas within it are "overvalued". The hypothesis is that the gap between the average rent and the average resale price has been expanding and that the average rent is a baseline for affordability, and that an area is valued ‘correctly’ or the area is balanced if the average rent and average price move in parallel. John puts it in slightly different terms:

The theory is that house prices should be appreciating at a similar rate to rents.  If house prices start to appreciate at a much faster rate than rents it suggests that house prices are overvalued... ...Eventually, more people start renting until rents rise and/or house prices fall to a level where there is no material financial benefit to renting over buying.

Many of the charts for major Canadian centres shows that the gap between the average resale price and average rent has widened and the conclusion naturally is that housing in overvalued. Not sure why this doesn't suggest that rents are undervalued, and a major upward correction in rents is imminent?

Regardless, this methodology is fatally flawed for several reasons: it does not factor in the age, size and exact location of the units in question. Secondly, the source of the rental information has been called into question, especially with the major expansion of the private rental market, the inclusion/exclusion of heat, hydro, water, cable etc. A third major factor is the existence of rent controls in certain areas. A fourth factor is the existence of a low interest rate environment that allows the average home owner to purchase a larger, more expensive home (or simply get in the market) with a lower monthly payment.

The key point to take out of the above is the first one. We are a firm that looks at averages frequently, but they can often mask or misrepresent data due to outliers and other factors. And consequentially, WHEN DRAWING CONCLUSIONS BASED ON THE RELATIONSHIP BETWEEN TWO VARIABLES, ONE MUST CONSIDER A THIRD (or fourth, fifth, sixth, etc) VARIABLE THAT COULD BE INFLUENCING THE RELATIONSHIP.

We'll provide a simple example using sports. Let's say I took three NBA teams (Spurs, Lakers, Magic) and looked at their average rebounds per game in a year and their winning percentage in those years. Wow, the relationship shows that a major improvement in rebounds per game in 97-98 led to much higher wins for the Spurs that year. We also saw rebounds and wins jump for the Lakers and 96-97, but both dropped in 04-05. There was a major improvement in 04-05 in rebounds and wins for the Magic.

Therefore we draw the conclusion that the more rebounds you get, the more wins your team will get. The Raptors see this data and go out and sign 5 guys that are 7'3" tall. The next year the Raps dominate the league in rebounding but win only 5 games, what! They should have seen an increase in wins right? Wrong, the five players can't cover anyone on the outside, can't play defence, but because they are tall they get lots of rebounds. They can't shoot, so they keep grabbing their own offensive rebounds and missing again!
So what went wrong in our analysis above? There was a third factor that contributed both to the increase and decrease in rebounds and wins in the above example, the Spurs added Tim Duncan in 04-05, Shaq came to the Lakers from 1996 to 2004 and the Orlando Magic added Dwight Howard in 2004 and all added new dynamics to their teams including shot blocking, scoring, passing, and confidence, in addition to their rebounding skills.

This is a typical example given in any economist's first year Econometrics class. The economists making these 'overvalued' statements should pull that textbook back off the shelf and take a look at it again. In economics you must 'control' outside variables, keep them as constant as possible so you can really determine the relationship between the two variables you are looking at.

In terms of rent multiples, you could potentially have a beach city building tons of 2,000 sf rentals overlooking the ocean and small 1,000 sf bungalows on outskirts of town and you compare that to a town building a bunch of 3,000 sf McMansions on estate lots and then small rentals on the outskirts near an industrial park, the rental multiples will be very different for those to areas, which one is really overvalued?

Although our data (from our UrbanRental report) is limited, having only begun collecting data for our reports over the past two years, we looked at the relationship between index rents and index prices in the Toronto CMA condominium apartment market. To avoid having the data skewed by location or age of product, we looked at ‘matching pairs’ of data only, dividing the average price psf by the average rent psf at a specific building. We looked at the weighted average (by total resale and rental transactions) of our rental multiple for some of the top municipalities and the Toronto CMA overall by looking at data points for the same buildings in Q4-2010 and Q4-2011. Using the index price and index rent, further controls for the size of the suites. In this example we were able to control for the location, age and size of the units. However, there are a few measures we couldn't control for: the floor that the unit is on (condo rentals tend to be on lower floors), parking (more units are rented without parking than resale units), renovations (resale units tend to be renovated more often than rental suites), terraces (units with large terraces or balconies tend to be owner occupied), etc, etc. See the table below.

If the hypothesis of a ‘balanced’ market held true, the rent multiples would remain the same year-over-year, however all the major municipalities increased with the exception of Richmond Hill. Therefore all markets are, in effect, less affordable than last year if one believes that rental rates are the baseline for affordability. The Toronto CMA is 3.1% less affordable (or overvalued) in comparison to last year.

Scarborough’s rent multiple increased by 10% in Q4-2011 over Q4-2010, the largest increase among the municipalities above, followed by Vaughan at 7.4% and Mississauga at 6.6%. The former City of Toronto increased by just 1.4%.

Urbanation would find it hard to believe that Scarborough condominium apartment units became much more overvalued than the former City of Toronto in 2011 and that Richmond Hill units became more affordable. According to John Pasalis, even Windsor is currently overvalued based on the metrics discussed earlier.

At Urbanation we do not believe that pricing will go up forever, or that Toronto's condominium market is not showing potential signs of future weakness, however is not because of an "out of whack" rent multiple. Look to Urbanation for reasoned, unbiased data on Toronto's condominium market from the firm that has tracked it longer than any other firm.