The Home Price Index - what it is and why TREB is pushing it more
Monday Sep 25th, 2017
The MLS® Home Price Index ("HPI") has been emphasized a bit more by the Toronto Real Estate Board ("TREB") recently, and I've been seeing it mentioned in the media a bit more of late. This index is supposed to work like the Consumer Price Index ("CPI") that most are familiar with. The CPI tracks changes in price of a "basket of goods and services," while the HPI tracks a "basket of home attributes." In both cases what's in the basket remains the same, and the index changes strictly due to changes in price.
What's the purpose of the HPI?
In the GTA real estate market in the last several months, average prices have changed a ton. Yes, prices have declined from springtime highs, but a lot of that movement in the average price can be attributed to the change in mix away from expensive detached homes and towards more inexpensive condo apartments. To show how much of a factor the mix change can be, I posted a chart in my July TREB stats blog that showed 416 average prices for July were down 19% from April. The casual reader of a real estate article might think "OMG, my house value has dropped almost 20%!". But as you see below, nobody's home dropped that much, as the actual range was -7.3% to -17.0%. The extra percentage points were due to mix.
The allure of HPI is it takes mix changes out of the equation, contributing to a more "apples-to-apples" number.
How does HPI work?
I mentioned it's like the CPI. So the HPI does have a base period where the indices are equal to 100: January 2005. If you're really interested in this, there are a lot of FAQs in TREB's initial news release of the HPI back in February of 2012. Note the value for the composite index was already 143.6 at that time (i.e. 43.6% higher than the January 2005 value of 100). The August 2017 composite index for the City of Toronto was 243.7. These index numbers all tie back to prices. If the HPI goes up 10% it means that the prices have gone up 10%.
I had a rough understanding of how the HPI worked, but even after reading the TREB news release about the introduction I still had several questions. I made a few calls to TREB and CREA to get a better understanding. This post is intended to save you the effort of making those phone calls :)
So, for starters, the HPI sets a "benchmark home" for each housing type. Their are four benchmark home types TREB reports each month on the second last page of its publicly-available statistical package: Market Watch.
- Single-Family Detached (freehold detached houses)
- Singe-Family Attached (freehold semi-detached houses)
- Townhouse (freehold attached/row/townhouses and condo townhouses)
- Apartment (condo apartments)
The latter two would include other ownership (e.g. co-operative) where you pay monthly maintenance fees.
Straight from TREB's news release, "A “Benchmark home” is one whose attributes are typical of homes traded in the area where it is located, with one benchmark being generated for each supported sub-area and home type. Benchmark property descriptions are based on median values for quantitative property attributes (e.g. above ground living area in square feet), and the most commonly occurring value (i.e. modal value) for qualitative attributes (e.g. basement is not finished). The attributes of Benchmark homes remain constant over time, allowing for an apples-to-apples comparison of price over time."
Here is some example attributes:
- Number of rooms above the basement level
- Number of bathrooms & half-bathrooms
- Square footage for main living & basement areas
- Whether it has a fireplace and/or finished basement
- Lot size
- The age of the property
- How the home is heated
- Foundation, flooring, siding & roofing types
- Whether the property has waterfront or panoramic view
- Whether the property has been sold previously (newly constructed and previously unsold, or repeat sale)
- Proximity to shopping, schools, hospitals, police stations, churches, sports centres, golf courses, parks, and transportation (including the train station, railways, and airports)
How is the benchmark calculated?
Where they mentioned "median values for quantitative" attributes, look at the first few bullets. If there were 5 houses and the number of bedrooms were 3, 3, 3, 4, and 5, the median value (same number of values above or below) would be 3 (though average is 3.6). If it was square footage of 1500, 1800, 2000, 2500, and 3000, then the median value is the 2000 sqft house (though the average is 2160).
Where they say the "modal value for qualitative attributes" an example would be how the home is heated. If the five houses had 2 with forced air natural gas furnace, 1 with electric baseboards, 1 oil furnace, and 1 with hydronic radiators, the benchmark house would use natural gas since it had the most responses with 2 (even though it wasn't in the majority).
So the benchmark home from my examples would have 3 bedrooms, 2000 sqft, and forced air natural gas heating. They would do this benchmarking exercise for each of the 4 house types mentioned, and it goes down to its community level. A community to TREB is what you and I might call a neighbourhood. There are 144 of those in Toronto, for example, Little Portugal, Annex, South Riverdale, and Junction Area.
So instead of looking at prices where the average square footage of homes sold could change month to month from 2160 up to 2500 then down to 2250, the HPI just concentrates on movement in prices of the 2000 sqft benchmark home it figured is most typical of the area.
These benchmark attributes use a 5 year sample size and are revised annually. The weighting of how much each attribute contributes to price (e.g. say square footage above grade contributes 21%) works on a rolling 3 year basis.
What is the HPI Composite?
Remember those 4 benchmark home types above? Well, the composite index takes a weighting of those 4 different benchmark homes. Once it's fixed, It uses that same weighting every month. So again, the only thing that varies are the prices - unlike TREB's average prices which have weightings that vary month to month depending on actual sales. In April across the GTA, detached homes were 49.1% of sales. In August that was down to 40.6%. That's going to drag down average prices. If the composite number is 44% detached homes, it stays at 44% every month.
The prices used are supplied by each local board using actual MLS® transactional data.
How do the benchmark prices compare to average prices?
the first chart below is average monthly sold prices for 416 detached homes dating to 2012 and the HPI is underneath. You'll notice that the average prices are way more volatile in two regards:
- the amplitude of the lines
- the seasonality
From December 2016, the average price of a detached home jumped up by $295K (22.3%) to the peak, and has declined $391K (24.7%) to August.
From December 2016, the benchmark price of a detached home increased by $190K (18.3%) to the peak, and has declined $129K (10.5%) to August.
You can see the average prices rise in spring, drop in summer, and rise again in fall in the averages. I wrote a blog post here showing why prices decline in the summer months. It turns out that bigger and/or better homes are generally made available for the spring and fall markets. The HPI benchmark isn't subject to those swings so in most years you see the price steadily climb.
Since 2012 when the MLS® Home Price Index was introduced for Toronto, the 416 average price for a detached house has exceeded the benchmark price for single family detached houses in each and every month. The range has been 8.3% to 42.4% more, with the average and median both rounding to 22% more.
Why would TREB want to use HPI more now?
Here is the year-over-year ("YoY") change in HPI benchmark for single-family detached homes in the City of Toronto:
You see that for August it's showing as 16.7% above last August's prices. In the August TREB Market Watch report, 416 detached houses showed as being down 1.2% YoY. Not as confidence-inspiring. On the flipside, March on this chart is up 24.1%, while per the March TREB Market Watch it was up 32.8%. So in general, the HPI isn't going to have as crazy swings, and it's a more apples-to-apples view. That's why TREB is a fan. Cynically speaking, it does seem funny I feel like I'm noticing more emphasis now when it's convenient to their narrative of "don't worry, things actually aren't that bad." But maybe I'm just looking out for HPI mentions more now?
For further reading and viewing of HPI statistics
This was meant to be a basic introduction. If you didn't know what HPI was, now you do, and if you only knew a bit about it, hopefully you know a little more. If you'd like to dig into further, here you go:
You can look up many of these stats for the 13 CREA board that run an HPI (and represent 52% of national re-sale transactions) at www.homepriceindex.ca. There, you may want to play with the HPI Tool where you can customize your own charts looking at different markets and indeces.
TREB's HPI page can be found here. It has all of the monthly 3 page releases going back to the introduction in February 2012.
If you're super-keen you can check out this 24 page PDF on HPI Methodology.
With this introduction at hand, I hope to introduce some HPI pages into my monthly 416 market statistics look-in on my blog. And I plan on inserting some new HPI charts in the full package of monthly charts I release on my SlideShare.