Economic troubles and lender profit motives have shrunken variable discounts beyond expectations. Banks are now commonly quoting prime rate, for example, with little discounting.
Once the last few holdout lenders with P-.50% disappear, discounted variables could move towards P-.25%…or worse. Some lenders even suggest that prime or prime plus could be the new normal.
Meanwhile, aggressive brokers are selling five-year fixed rates at 3.25% or less. That’s an unusually low 50 basis point premium to a variable. A spread that tight doesn’t come around often, and it makes you rethink all of the research suggesting variables are the way to go.
Popular research indicates that people have saved money on variable-rate mortgages:
- 88.6% of the time:
- Source: Floating Your Way to Prosperity, 2001, Moshe Milevsky, PhD; Study period: 1950-2000
- 90.1% of the time:
- Source: Moving Mortgages, 2008, Moshe Milevsky, PhD; Study period: 1950-2008
- 83% of the time:
- Source: Variable Rate Roller Coaster Still Hard to Beat, 2011, Douglas Porter and Benjamin Reitzes, BMO; Study period: 1975-2011
Odds like that make some people question the sanity of going fixed.
But there’s a little more to the story.
More importantly, assumptions are key when it comes to rate studies. Two important factors have impacted the research quoted above:
- A multi-decade bias towards falling rates
- Use of posted rates (instead of discount rates)
“Interest rates have been trending downward for two decades,” BMO Capital Markets Senior Economist Benjamin Reitzes told us in a recent interview. By default, he says, that’s tilted the table more in favour of variables than it otherwise would be.
Looking ahead, rates are no longer able to drop over one percent. The most we can realistically hope for is an extended period of horizontal rate movement. (The BoC can still cut rates slightly, but the European and American crises and sub-2% core inflation won’t delay hikes forever.)
As a result, Reitzes says, “Going forward, borrowers won’t see the same advantage to variable rates as they have in the past 25 years”
The second factor that’s largely ignored when citing rate research is the actual mortgage rates used for backtesting. Each of the three studies above uses posted rates in their historical analysis.
Reitzes states that this practice distorts the results somewhat. “Discounts off posted rates were not as prevalent historically.” Nowadays, however, “Most people get a (rate) discount if they are credit-worthy borrowers.”
That matters, because the rate discount you get obviously impacts the likelihood of your mortgage outperforming other options.
Here’s an example.
- If you look at data from 1970 to 1995, the average difference (spread) between 5-year fixed and variable rates was 126 basis points.*
- The average difference today is roughly 50 basis points.
That’s a remarkable 76 basis points lower than historical rate spreads. That makes a huge difference in research conclusions.
If you theoretically backtested with the same spreads as today (i.e., 25 bps off prime for variables and 204 bps off posted for 5-year fixeds), you’d find that fixed rates outperform considerably more often.
According to Milevsky, “…The historical probability of doing better with the floating rate mortgage…hovered around 70% to 80%” when the borrower used deep discount rates (based on a 1965-2000 study period).
Using today’s discounts, that 70-80% drops to just 53%, based on our findings from 1970 to 2006. (Obviously today’s spreads would not have applied historically but, as Milevsky maintained in his research above, that is beside the point.)
In other words, the fixed/variable decision would have been a coinflip, based on today’s spreads.
(Click to enlarge)
This isn’t meant to imply that fixed rates now have an insurmountable edge. If the Bank of Canada drops rates unexpectedly, a variable could easily beat all other terms over the next five years.
A variable may also prevail for other reasons. See:
That said, if the BoC’s next rate move is up (which is the highest probability outcome, say economists), the boring old 5-year fixed could certainly outperform. That’s true even when compared to a variable with payments set at the 5-year fixed rate. (We’ll post a scenario like this soon.)
The nice part is this: If you go fixed and variables end up winning, you’ll likely be out far less money than in most prior years.
* Data source: Bank of Canada. (We chose 1970-1995 because 1970 is as far back as we have clean 5-year fixed rate data, and 1995 was before rate discounting started taking off. Yes, people actually used to pay posted rates.)
Note: If you’re already in a discounted variable, the conclusions drawn here may not apply to you. For guidance on locking in, always consult a mortgage professional.
Rob McLister, CMT