House Price Index – start the count over

Real estate can be tricky to track stats for. There’s so much variation between different properties. Average sales prices get pushed back and forth based on the sales mix.

There are two sources for a more consistent tracking of house prices. One is the Teranet House Price Index, which tracks only properties that have at least 2 sales. This ‘sales pair’ approach tracks the effect of specific properties changing price each time they’re sold. There are two drawbacks to this index: It doesn’t track improvements to a property and it’s delayed by about 3 months.

The other approach is the REBGVs MLSlink HPI based on a hypothetical property that is averaged to specific traits: number of rooms, finished basement etc. The MLSlink HPI has been a consistent gauge of the local market since the mid 90s.

That’s changing now. Vancouver was the only market with a House Price Index, but the old index has been thrown aside for a new one from the CREA that is consistent between Vancouver, Fraser Valley Calgary, Toronto and Montreal.

For data-hounds, this means that you’ll not be able to directly link trends in the new index to earlier data. January 2012 becomes the zero month of this new system. Ladies and gentlenerds, start your graphs!

Sort by:   newest | oldest | most voted
Anonymous
Guest
Anonymous

http://www.thetrumpet.com/9087.7872.0.0/economy/c

Canada’s Housing Bubble Is Stretched to the Limit

Drachen
Member

If they wanted to be honest about it they'd keep reporting the old HPI for the next six months or so alongside the new. That way you know how the two methods compare.

jesse
Member

The detached HPI seemed to be tracking relatively well between the old and the new. Until 2010. Hm.

jesse
Member

"The other approach is the REBGVs MLSlink HPI based on a hypothetical property that is averaged to specific traits: number of rooms, finished basement etc."

We don't have good insight into what the benchmark measures; the methods are not published. What it seems, though, is that they use existing quantitative measures to form the index but likely do not rely on quality. That is if houses with a high level of finish sell more frequently than vanilla stock, this can push up the index higher than is warranted. The Case-Shiller methodology takes this into account in part.

In any case it's fun to watch to pass the time. IMO it's all about the sales volumes (i.e. demand); the other variables we track — inventory, benchmarks, sell/list, etc. are dependent upon this.

rp1
Guest
rp1

@jesse: Sorry, which is red and which is blue?

VMD
Member
I'm getting more and more supporters at the Chinese forums, half a year ago I was like public enemy #1 (at least amongst the realtor/house-pumpers). A few openly came out to say they had deferred home purchases last year after reading my posts. For example: "blue2013" moved to Vancouver from US in 2010 and Almost bought a house in mid 2011. However, by chance he read my posts at that time and decided to hold off home purchase. He thought he made a correct decision and thanked me for the stats/translations/reasoning. "blue2013 lived in US fro 2006-2010, he felt that the current Vancouver market felt like US in end of 2007, where home price was declining very gradually. "We all must have patience", said blue2013, "interest rate is already near rock bottom, there is not much more to keep this… Read more »
jesse
Member

@rp1: New one on top

registered
Member
registered

3 jesse Says: '….tracking relatively well between the old and the new. Until 2010."

🙂 Just like CPI. Every refinement in definition results in lower inflation. The economic outlook grows consistently rosier with increased 'accuracy'.

The same alchemy turned ketchup into a vegetable.

jumpin in
Guest
jumpin in

VMD, thanks for doing it. I wish I could to.

Must be fun though 🙂

observer
Guest
observer
@jesse We don’t have good insight into what the benchmark measures; the methods are not published. What it seems, though, is that they use existing quantitative measures to form the index but likely do not rely on quality. That is if houses with a high level of finish sell more frequently than vanilla stock, this can push up the index higher than is warranted. The Case-Shiller methodology takes this into account in part. – the methods are published and it is relatively simple hedonic index using regression analysis. Benchmarks are data driven based on what attributes are the norm – predicted values based on coefficients estimated in the model. It is not difficult, but just needs a high-powered database to sift through the data. – Hedonic is actually a better measure than Case as it estimates using the attributes. Case… Read more »
Many Franks
Guest

I never thought this day would come so soon! We've hit 200. Scott Brunton gets the honours from current media darling Toronto.

http://realestateagentsincars.tumblr.com/post/172

WS
Guest
WS

Help an outsider understand this better. What is so hard about tracking historical median sales prices (even better across quartiles) and tracking price per square feet. Yes, I know this is not ideal (can be over/understated based on mix) but it gives you a fairly decent idea of the general direction of prices and speed of incline/decline.

I have always been surprised to see average referenced in Canada before median.

N
Guest
N

In a well-buried lead, Vancouver's population growth is about 25% bellow the national average.

http://www.vancouversun.com/technology/population

jesse
Member

@observer:

"In theory, a hedonic could overcome this if it is specified correctly."

And how are you going to quantify quality in any meaningful way? It needs an entry in a database; if someone finishes a structure to a high degree the only input available is the cost, which I doubt is captured.

Case-Shiller eliminates this by looking at existing structures and should be relatively independent of sales mix. If the city generally improves quality over time than CS will miss a trend but I'm not seeing any marked change from even 20 years ago on this front.

observer
Guest
observer
@observer: And how are you going to quantify quality in any meaningful way? It needs an entry in a database; if someone finishes a structure to a high degree the only input available is the cost, which I doubt is captured. – I said in theory- in practise, one could use a dummy variable to account for it. Not perfect by any means. Case-Shiller eliminates this by looking at existing structures and should be relatively independent of sales mix. If the city generally improves quality over time than CS will miss a trend but I’m not seeing any marked change from even 20 years ago on this front. -hedonic is also independent of product mix. Proper construction of an HPI controls for product type and attributes to isolate underlying price changes through statistical methods – non-linear multiple regression methods –… Read more »
patriotz
Member

@observer:

"-assume that a person renos their property to include high quality flooring, which tends to add value to a property purchase price"

Does it? That's what everyone thought during the bubble in the US, but it turned out during the bust that renos didn't do much good for the sale price. That is, the higher prices were simply the result of the property being off the market while prices were going up.

fixie guy
Guest
fixie guy

15 observer Says: "Case-Shiller eliminates this by looking at existing structures and should be relatively independent of sales mix. If the city generally improves quality over time than CS will miss a trend …"

CS tracks the market. If the quality of an entire market elevates, that would be automatically reflected in the HPI after a short lag. Individual properties still price statistically around the HPI. Part of the problem is focuing on 'improvement', properties also wear and fall into disrepair. (Sorry, in Vancouver disrepair + granite counters = character.) Maybe it's a statistical wash.

observer
Guest
observer

@patriotz

does not change my point at all. if prices do decline after renos, then a repeat sales index would understate the price decline… essentially there is an upward bias

patriotz
Member

@observer:

I will repeat my point, which is the evidence is that renos do little to change the selling price in the first place. It's just that in rising markets people think they do.

jesse
Member

@observer: " essentially there is an upward bias"

There is an upward bias if the renovations add significant value to a property but if the renos simply return a property back to "like new" it won't bias anything; rather properties that are depreciating without upgrades will be, well, depreciating. If a neighbourhood gentrifies and a working class family dwelling built in the 1920s is renovated for a dual-income upper middle class family, well yes then we can expect more luxury. That's gentrification and a reasonable "bias" to same-sales.

BoogyBear
Guest
BoogyBear

Another problem with the CS method is time. If the re-sale of the property is more than 10 years apart, you are no longer comparing the same home. If the home today is 15 years old, what you are comparing is a 15 year old home to a 5 year old home. The error is larger in condominiums which are more sensitive to depreciation. There is no perfect method to track a general market. You can be more accurate if you track a type of home with specific physical parameters.

jesse
Member
@observer: On thing, though, that you highlight is that the Case-Shiller methodology will throw out certain property transactions that appear to be significantly deviating from the mean. This is meant to capture those that have been rebuilt or significantly overhauled to the point where they aren't really "same sales" any more. In Vancouver — as in some US markets — this may lead to understating some of the more hyperbolic sales we hear about, where the price increases so much for certain properties that it's tossed out as an inferred rebuild when it may have just been ludicrous market being ludicrous. Don't get me wrong, I like the benchmark — it's a good idea and has tracked the Teranet HPI reasonably well — but (unless I missed it) the methodology is not transparent. The Teranet website provides a very detailed… Read more »
Makaya
Member
Makaya

@Many Franks: That's awesome.

By the way, what an empty video. The guy has really nothing to say… pathetically funny!

SunBlaster
Guest
SunBlaster

New stats, are better than old stats!

In mathematics, an average, is a measure of the "middle" value of the data set. This is elementary grade mathematics, however you are not suppose to know that, otherwise you'd have no reason to pay RE agent's commissions.

Massaged data from an "official" source is present to general population to hide the reality, nothing new here. This is just hopium tactics to buy time and hope that general population doesn't catch a smell of housing dump. They know it's coming, but it can't come too fast and cause an "uncontrolled" bubble burst, rather a slow balloon deflation, because we all know balloons can be deflated.

Anonymouse
Guest
Anonymouse

@patriotz:

"I will repeat my point, which is the evidence is that renos do little to change the selling price in the first place. It’s just that in rising markets people think they do."

Which evidence?

wpDiscuz