Inventory New Construction

Over/Under Supply

For awhile now I’ve been thinking about trying to attempt to quantify the relationship between population growth and housing construction. We’ve been hearing for a year or two now about how our city is actually overbuilt to a significant degree.

This of course would come to a surprise to many as just three years ago there was an apparent serious housing shortage, and stories of tent cities propping up around the city. As it turns out that shortage was largely artificial and a result of speculative buying, and as reports such as this one from TD point out that we’ve actually been somewhat overbuilt all along.

So, today I decided to sit down and try to hammer out some kind of quantification for this… but I wouldn’t take any of this as gospel, it’s really more of a scientific wild ass guess then anything. Regardless, on with the show.

This is just basically an overview of the stats from which the subsequent findings arise. So you can examine that at your leisure.

From this data I derived some long term averages and medians to attempt to make something tangible out of that mess. Basically what I found was that for every new unit completed you needed somewhere between a 2.1-2.3 person increase in population.

Those familiar with the censuses probably know that typically Edmonton and centres like it have somewhere between 2.5-3.0 people per household (FWIW, according to the Statcan numbers I have over the last decade Edmonton has came in at 2.73).

So you may be thinking this is the big “Aha moment,” and that that alone proves we’re overbuilt… but no. That would ignore all the redeveloping/rebuilding that goes on, which I think it’s fair to assume makes up that difference. I’m not in the industry, but just as a laymen, to say 20-30% of new construction would qualify as redevelopment/rebuilding sounds reasonable.

So, now assuming that 2.1-2.3 range is appropriate lets compare how many units would be needed to absorb that population increase vs how many units were actually completed over this period.

So here we see the patterns formed by the yearly numbers (non-cumulative). No surprise, the completion numbers tend to lag the demand fluctuations… obviously it takes time to actually build the places. This also doesn’t take into account the relative inventory positions, for that we’d run the numbers cumulatively.

This, I think, is a much more telling graph as it gives a relative supply position (assuming something resembling balance in 1987 of course). It seems to give a believable pattern as one would expect some ebbs and flows as when demand is falling you’d see a large pullback in supply in subsequent years, and conversely when demand returns you’d see an overcompensation in the other direction as builders rush back in.

In any case, one would expect in balanced market conditions that the figures generally fluctuate around zero, which they do. Then we hit the boom.

What I find quite interesting is that the build up of over supply actually came about from 2003, 2004 and 2005… before the price explosion really hit. This was during a time when the market was hot and building steam but prices were still within historical means.

From 2006 until now the over supply has merely maintained… somewhere between 8,000 and 22,000 units if my factors are correct (personally I’d probably place it closer to the 2.2 curve if not a bit below, but that’s just me).

I suspect this pattern may be more due to under-reporting of migration early in the boom, thus oversupply was probably slower to build in actuality as the new construction stats are actuals whereas population/migration are estimates. Over the long term they’re adjusted to be correct, but in any given year they are prone to significant variance with what is actually experienced. But we have to go with what’s given, so I digress.

For arguments sake, lets say it’s in the 10-15,000 range. Which is a significant degree to be overbuilt, even with starts slowing it’s worth nothing that when this data cut off there were still another 11,400 units under construction.

Figuring that all in and the ratio of persons per household and that’s a couple years worth population growth even with zero subsequent starts, maybe even more as migration is slowing. It will take time to absorb all these new units, and it’s going to be a drag on the market.

Of course all that is assuming my little SWAG has any validity whatsoever… which I’m not sure I’d quite extend it, but I think it’s fair to declare it a good discussion piece at least.

Historical Prices Inventory Sales


Hope you all enjoyed your long weekend. If it was anything like mine, it was likely full of sun and way too much refined sugar.

If you’ve been reading the releases coming from the various real estate associations, you may have noticed they are now talking up the month-over-month sales and price figures more so then they have in the past.

Traditionally they usually focused on year-over-year figures… but as those are coming in as declines as of late, they are referencing those less and less. Something of a marketing move, as they’re in the business of stimulating sales, and hearing about prices and sales going down doesn’t make their message very receptive to buyers or sellers.

As a result of this, it’s also increasingly being pointed out that month-over-month gains are actually normal in the spring, as real estate for whatever reason has a lot of seasonality. So today I’m going to take a quick look at the traditional seasonal movements of prices, sales and inventory.

The 100 value is the relative value in January… and from there it behaves like percentages. So 110 = 110%… so if the price was 200,000 in January, that value would be equal to 100… so 110 would be 110% of the January price, so 220,00.

Here we see how prices typically move through the year. The blue line is the historical trend, and as you can see it usually shoots up about 7% in the first half of the year before leveling off, then drops off a couple points in the summer, and ends the year up about 5%.

We can also see that over the last year we’ve severely deviated from the norm. In 2008 it tracked normally through March, then fell off slightly through spring and then really dove throughout the second half… and 2009 thus far has been an extension of that.

Here’s a look at how prices behaved during the boom relative to historical norms. In this case normal seasonality barely registers as a speed bump, and while the boom was on through mid 2007, normal seasonality was no predictor what-so-ever. Which has also been true thus far in the bust.

Basically, seasonality is no indicator of prices during a bubble. That said, it is noticable that during the downturn, spring still does appear to be the strongest period, and falls off from there. Which seems to indicate the second half of 2009 could witness a very big decline.

Here we look at sales. Similar pattern, but in much larger proportions. Sales continually improve through May (typically more then doubling the January tallies), then going back down the rest of the year.

In 2008 we can see that it tracked a closer pattern in sales then prices. This also holds true though the boom, so while the scales may change, the general seasonal pattern remains.

It’s also notable that 2009 appears to be gaining very strongly, but this may be more of a result of an overly poor January. Again though, it is tracking quite closely to the historical seasonality.

Finally we’ll take a quick look at inventory levels. Again we’re seeing levels track fairly normal curves, 2008 went a bit higher, and 2009 is shaping up to be lower.

On the other hand, during the the boom, and the onset of the bust it looked quite different. In 2006 inventory actually tracked the opposite pattern, and declined through to the summer. This obviously caused a inbalance of supply and demand, and really fed the skyrocketing of prices.

Then in 2007 it was a very different story when inventory tracked normally, then picked up, then did some skyrocketing of its own. Which then caused a glut of inventory, and halted then reversed prices. Obviously a very volatile period there in 2006 and 2007.

So, as we can see, we should not be surprised by sales increasing in the spring, and the real estate boards out there should enjoy trumpting month-over-month gains while they can, cause come summer they’re going to have to find something else as both month-over-month and year-over-year figures will be poor.

As for some April predictions, I’ve been working on some different models but they’re hardly perfect… but I’m a glutton for punishment, so lets throw out some numbers and see how I do.

As mentioned before during a bubble prices are kind of anyones guess, but I imagine they’ll hold their prices in April. Active Inventory I’m going to say will end April around 7,900. Sales I’m going to peg at about 1,700 for April.

Though that could be a ways off, on one of the local agent blogs they are citing the EREB claiming 574 sales already this month, which by my calculations would put us on pace for almost 1,800 on the month (there were 1,830 last April)… though in the paragraph prior they talk about how much things have slowed sales wise and gave quite a low tally for the week.

So, in conclusion… who knows?! Time will tell.

Inventory Sales

Lies, Damn Lies, and Statistics

When I first started looking at the litany of stats released every month by the various outlets, it was kind of overwhelming. There are dozens of different measures, examining things from just as many angles. Even with my background in accounting and statistics, it was kind of wild.

But as I’ve worked with the numbers, and seen how they interact, I’ve found an increasingly select few stats that are actually relevant… and conversely, a growing list that are not worth the paper they’re printed on. So today I figured I’d take a quick look at the more commonly reported stats, and offer my takes on which are the wheat, and which are the chaff.

I of course reserve the right to later change my mind and completely contradict this post, when or if it becomes convenient.

Good Stats

Median Price – This I find to be the most accurate reflection of what the “average home” is worth. I don’t want to bore you with the mathematics of it, but if you follow this link you can read up on just how medians are found, if you don’t already know.

Sales – A very straight forward measure, and tells you just how many units are sold.

Active Inventory/Listings – Another straight forward measure. As its name suggests, it tells you how many properties are listed at any given time. Not always widely reported here in Edmonton, but it can always be found in the Quarterly Stats packages on the EREB site.

Between those three stats alone you can get a very good overall picture of what is going on in the real estate market.

Occasionally Interesting Stats

Average Prices – Probably the most widely reported stats, unfortunately I don’t think it’s as good a measure as median prices. I feel this way because it can be skewed a bit. For example, if in February there are 999 homes sold for $300,000 each, and one sold for $5,000,000, we end up with an average price of 304,700. So while the number might not be way off, you can see the potential for a couple high end sales to really throw off the number, especially since sales in that category can be somewhat sporadic. That’s why I prefer the median figure because I think it gives a more accurate idea of what the “average home” is worth, rather then the “average of all homes.” Though because, at least here in Edmonton, they only give average prices for condos and townhouses and not the medians, it is unfortunately the only figure available, thus it is a necessary evil to discuss them.

Days on the Market – I was tempted to put this one in the “Completely Useless” pile, but I decided not to. My problem with this stat is the way re-listings restart their clocks once they expire, which obviously will skew the data, especially in times like now when we have large inventories and a lot of expiring listings. Though, while their figures are muted, they do seem to give an idea of the general health of a market.

Completely Useless Stats

New Listings – We hear a lot about this number, and some people just love it, using it in all kinds of ratios and such… but I think it’s as useless as tits on a bull. Because of the way re-listings are included, it’s a totally bogus stat. If you want an idea of how inventory is tracking, just keep an eye on active listings and sales, that’ll give you a much better idea of what’s out there.

In conclusion, in my not so humble opinion, if you keep your eye on median price, sales and active inventory over time, you should have a pretty good idea of what is going on in the market. Most of the rest of the numbers are just static, but if you want to know how the condo and townhouse prices are doing, those average prices can be used as something of a jumping off point.

Inventory Sales

Where do they go?

Those that follow the stats released by their local real estate boards may have noticed that their inventories don’t seem to add up.

For example, at the end of December in Edmonton there were 6,316 units in inventory. Then in January there were 2,443 new listings, and 730 sales. Doing the quick math, 6,316 + 2,443 – 730 = 8,029, one would expect they’d have 8,029 units in inventory at the end of the month… but they only had 6,573.

So where did those 1,456 units go? Well, simple answer is they most likely expired, or were otherwise cancelled or withdrawn. When people sign on with a realtor to list their place, it can be for any term, many are for three or six months, but some can be for years. If that time passes without the property being sold, the listing expires and it’s up to the prospective sellers whether they want to re-list or not.

And getting back to the January totals, the truth is that in all likelihood there weren’t even 8,029 properties involved. As many would be counted twice in that figure if they expired and re-listed in the same month.

While the exact numbers aren’t included in the EREB stat packs, they are not hard to calculate since we have all the other relevant information. On another site ran by a local realtor they include such figures in their weekly updates. As you can see from charting out their figures there is always a big flood of expired listings at the end of every month.

So obviously most people sign on to list until the last day, or last business day of any given month. We can also see that come calendar year end there is an even bigger spike, probably just cause its a more significant date. While the spikes during the spring are a bit lower… probably a combination of the increased sales as well as people wanting to make sure they’re listed through the traditionally hotter sales season.

While that graph nicely shows the monthly cycle, it doesn’t really do anything to show a historical context of our current position. For that I’ll graph out the monthly totals for the last eight years.

If you’ve been following this blog, you may have noticed that the pattern followed by the Expired/Cancelled Listings is actually quite similar to that of inventory over this period. Which shouldn’t be surprising though, since obviously with increased inventory, you’re going to have more listings expiring, or corresponding decreases.

As you can see, during the boom sales rose and very few listings expired. Then as soon the market cooled, the number of expired listings quickly escalated, and much like inventory, reached record levels. Now the troughs remain at levels that the peaks didn’t even reach while the market was balanced. We now average over twice as many listings reaching their expirations as we traditionally did.

Another somewhat related stat is that of Days on Market, which is a measure of the average days a listing stays on the market before selling. I’ve charted that out here.

A correlation between D.O.M. and Expired Listing appears quite positive. Not surprising, during the boom D.O.M. were at record lows. A couple months in the first half of ’06 they reached their absolute low of 19. Now it has rebounded and is setting records at the opposite end of the spectrum, hitting 68 just last month.

Trouble is that with the number of expired listings being as high as they are, they may be further skewing the D.O.M. stat. Many units that expire get immediately re-listed, new MLS numbers and the clock starts over, so when/if they finally do sell the number is much lower then it actually should be. Thus the average D.O.M. may actually be significantly higher then they already are.

In any case, we can see from the elevated numbers of expired listings that there are still a whole lot more people looking to sell then are listed in active inventory. Whether they immediately re-list, or hold off for a few months, they’re out there and we’re a long way from supply and demand coming into balance.