Tuesday, June 22, 2010

What Normal Looks Like

What is "normal" for Vancouver real estate? A very quick gauge is to look at a month at the peak of the annual real estate buffalo jump -- June -- and normalise it for population. This gives us sales per capita.

Note June 2010 is an estimate after 3/4 of the month's sales have been recorded. A fair estimate of sales in a "balanced" June market (where "balanced" refers to sales volume) will be about 1.5 sales per 1000 residents. Things will really get interesting if we see the 1.2 sales per kiloresident at the lower bounds of the data series, a level with historical precedent and far worse than even 2008 when prices began to fall precipitously.

8 comments:

JimTan said...

Right! I have just proposed an alternative bear theory at RE Forum. Posting just once a day and starting on June 17th, my conclusion is this,

“But, the recovery has been slower than expected and Obama has run out of time. The risk is that the right-wing win the November 2010 elections. Programs are cut instead of taxes being raised. Then, there's a possible slowdown in China. And, there's Western Europe's credit risks. A nasty coincidence!

The result is a long period of sub-normal growth. Watch that election carefully. It's a race against time. Can the States recover enough to save the Democrats?

That said, we have to be honest. The bubble bears have been wrong about Vancouver RE. The War of 2001-2008 is over. We know who won.

Now, it is time for a new generation to take ownership of the forum, and look at RE with fresh eyes.

DoDo – Your time has passed”

http://www.realestatetalks.com/viewtopic.php?f=8&t=41621

RentingSucks said...

I don't think you were right. I think bears were particularily unlucky and bulls were particularily lucky over the last few years. That's the conclusion I came too in the end. My main problem with bulls is that they are most often victims of the narrative fallacy coming up with convincing explanations of how they were right and made up reasons why the market should continue to rise. I naturally side with the bears of course because at least they've got numbers. They are at least trying to be rational. In the end there is just too much randomness for anybody to know anything for sure.

Maybe real estate prognostication based on data isn't much better than technical analysis or charting for stocks which IMHO isn't much better than a horoscope. Of course it's better than reworking history to paint yourself a genius in your own mind.

patriotz said...

JT, your theory is essentially that if something stupid has gone on long enough, it will go on indefinitely.

The world does not work like that. Not for RE or for anything else.

JimTan said...

"JT, your theory is essentially that if something stupid has gone on long enough, it will go on indefinitely."

Don't put words in my mouth.

I suggest you look in the mirror before you call other people stupid.

apakdel said...

You know, I love dots on a graph just as much as the next guy. It gives an illusion of something great, but it is ridiculously flawed, like most of my dots were in my university chem lab results.

Why? because here we are dividing something is that is exact to the last single digit at any given time (i.e. number of sales) by a number that could be off by at least 5 to 10 % (i.e. the population sample) even by most accurate measures, then claims that a 20% variation in the ratio (i.e. 1.5 to 1.3) could actually mean something (i.e. a revert back to 2008 where prices fell).

And it is rather irrationally amusing for people to cling on blips such as price fluctuation that happened in 2008, as something that could be predictor of a future event. We had no models that could predict that blip. And we had no models that could predict the run up in 2009 and now 2010. So how the hell does that piece of data have any significance?

Repeat after me: "Unless we can recreate every single variable that affect a market at any single point in time (an impossible task - single it involves psychology and perception of the masses), we cannot predict future market behavior by that past occurrence"

jesse said...

"Why? because here we are dividing something is that is exact to the last single digit at any given time (i.e. number of sales) by a number that could be off by at least 5 to 10 % (i.e. the population sample) even by most accurate measures"

If you have data to suggest the population measure is off by 5-10%, post it here. Don't just say stuff like that without backing it up. BTW a 10% error in population would be 200,000 residents. Good luck backing up your claim nonetheless.

apakdel said...

@ Jesse

If we are taking Vancouver proper population, a 5% error is 29,000 residents.

As for your doubts on at least 5-10% uncertaintity in estimating the population from the current sampling methods.... you are joking right?

Even on the consensus year, the population data is reported as "estimate". Anything between consensus years is a "projection".

No need for luck to "back-up" my claim. Since we are all referring to one source of data, here is what you wanted from from Stat Can's reports:

http://www12.statcan.ca/census-recensement/2006/ref/rp-guides/rp/sw-ep/sw-ep_p10-eng.cfm

I invite you to review the other parts that report if you really want to get an idea of how rough this population estimates are.

jesse said...

"I invite you to review the other parts that report if you really want to get an idea of how rough this population estimates are."

Perhaps a look at what I am doing with the data will help. I am correcting sales for per capita residents, nothing more. The population estimates are based on tax returns, births, deaths, and immigration, resolved with the censuses once every 5 years. The population numbers, while not completely accurate, are not significantly off in terms of estimating the trend. A look at adjustments pre and post census shows that generally the estimates are not bad.

Your cited link is for small data sets, not for large CMA pop counts where the certainty is significantly higher.

The crux of this post is to show that sales in June are typically lower than during the housing boom. A comparison between the population corrected and non-corrected sales data still shows a marked change in 2010
(that is, a 10% correction in population would change June 2010 sales per capita by 10% but this is still well below levels seen in the mid-2000s).

I think using some adjustment for total population is correct, even though it produces a relatively small effect in the end.