On the mechanics of CMHC construction data

A quick note on the mechanics of CMHC construction data. CMHC reports three main datapoints in its Table 027-0048 Canada Mortgage and Housing Corporation, housing starts, under construction and completions in selected census metropolitan areas monthly (units): starts, completions, and under construction.

Definitions are as follows (link):

Start: For purposes of the Starts and Completions Survey, a Start is defined as the beginning of construction work on a building, usually when the concrete has been poured for the whole of the footing around the structure, or an equivalent stage where a basement will not be part of the structure.

Under Construction refers to the number of units under construction at the end of the period shown, and takes into account certain adjustments which are necessary for various reasons. For example, after a start on a dwelling has commenced construction may cease, or a structure, when completed, may contain more or fewer dwelling units than were reported at start.

A Completion is defined as the stage at which all the proposed construction work on a dwelling unit has been performed, although under some circumstances a dwelling may be counted as completed where up to 10 per cent of the proposed work remains to be done.

The graphs I show on construction activity were authored by Vancouver Housing Blogger (the blog is now almost 10 years defunct). Since there is seasonality in starts and completions, a seasonal adjustment has been performed by summing a year's worth of starts and completions, therefore the starts and completions data are filtered with a delay of 6 months. Sudden changes in starts and completions will be "averaged" so could take some time to show up. (Smarter ways of seasonally adjusting are available but whatever.)

One behaviour worthy of note is that under construction is the cumulative difference between starts and completions. If starts outpace completions, under construction rises. If completions outpace starts, under construction falls. Here is where we are today:

Starts have outpaced completions since the end of the last recession. More recently starts went plaid and under construction has gone off the charts (literally, in the case above). Since the construction process has a fair amount of fixed resources (engineering, labour, materials, inspections, etc.), this sudden surge in activity has resulted in completions lagging moreso than they have in the past (i.e. projects have been delayed), and units under construction has ballooned as starts, until just recently, have continued apace.

As the construction backlog clears, completions will increase as resources become freed up to be transferred to other projects. As completions increase, units under construction should begin to drop, but only if starts do not continue in beast mode.

Regardless, supply is coming very soon.

REBGV Sales Update Through September 2017

REBGV released their stats package through September 2017. Here are the numbers:







Sales have retrenched  and remain decidedly "average". Inventory is low. The result is a low but slowly increasing MOI (months of inventory). A long stretch of robust (or at least not pallid) new listings is still required to allow inventory to recover: new listings have returned to the point where inventory recovery is slowly occurring. These things take time, apparently.

Still no signs of a slowdown.

REBGV Sales Update Through August 2017

REBGV released their stats package through August 2017. Here are the numbers:







Sales have retrenched  and are now decidedly "average", though August was above levels seen in 2016. Inventory is low. The result is a low but increasing MOI (months of inventory). A long stretch of robust (or at least not pallid) new listings is still required to allow inventory to recover: new listings have returned to the point where inventory recovery is slowly occurring.

No signs of a slowdown.

Vancouver CMA housing construction activity --- Moar!

Moar housing is on the way:

REBGV Sales Update Through July 2017

REBGV released their stats package through July 2017. Here are the numbers:









Sales have retrenched  and are now decidedly "average", nowhere near the levels seen in 2016. Inventory is low, though new listings have recovered. The result is a low but increasing MOI (months of inventory). A long stretch of robust (or at least not pallid) new listings is required to allow  inventory to recover: new listings have returned to the point where inventory recovery is slowly occurring.

Reports from the front lines claim that certain market segments have slowed; it remains to be seen if these reports are confusing the usual summer slowdown with something more significant. Recent changes to lending and whispers of shuttered supply lines of capital from distant shores are creeping onto my twitter stream's consciousness. We shall see!

And, yes, supply is coming.

REBGV Sales Update Through April 2017

REBGV released their stats package through April 2017. Here are the numbers:

Sales have recovered and are now "average", but nowhere near the levels seen in 2016. Inventory is low and new listings have recovered in recent weeks. The result is a low MOI (months of inventory), which portends a decent spring in terms of price growth, and the MLS-HPI appears to be bearing this thesis out. A long stretch of robust new listings is required to allow low inventory to recover and it appears that new listings have returned to the point where that can start slowly occurring.

Attached housing (townhouse and apartments) continues to be stronger than single detached housing in terms of price strength as measured by the change in the MLS-HPI (see link at beginning of this post). The closing of the gap between housing types was reasonably expected in the wake of detached housing prices having had accelerated in previous years.

I expect new listings to occur at an average pace over the coming few months and inventory to grow; I was surprised by the persistence in the dearth of new listings but I also think it's a temporary phenomenon: supply is coming.

Vancouver Teranet HPI and Inventory update

Long-time readers of this blog will be familiar with the model that mohican devised that relates house price index (HPI) changes to the ratio of for-sale inventory to monthly sales (months of inventory or MOI). Below is an update to this data series with some brief analysis.

The first graph plots the time series of the 6 month change in the Teranet Vancouver HPI and REBGV MOI. The correlation is clear from direct observation.

The second graph is a scatterplot of HPI change versus months of inventory, with a time shift to optimise for highest correlation.

And finally the same scatterplot but with year-on-year HPI change with a 6 month lag.

The only likely way to get higher MOI is to have higher inventory, and the advent of higher inventory takes many quarters to materialise (inventory is currently very low). It is for this reason that I do not foresee any significant house price relief in Vancouver in the next year to 18 months.

REBGV Sales Update Through March 2017

REBGV released their stats package through March 2017. Here are the numbers:

Sales have recovered and are now "average", but nowhere near the levels seen in 2016. Inventory is low and new listings have been low, albeit new listings have recovered in recent weeks. The result is a "lowish" MOI (months of inventory), which portends a decent spring in terms of price growth (not crazy but not weak either). A long stretch of robust new listings is required to allow low inventory to recover.

Why are new listings so low? There are two reasons I think are both acting to suppress listings. First (pointed out to me by Aaron Vaillancourt) many listings manifest due to homeowners moving properties; if there are few listings then there is nothing for homeowners to buy and they will therefore not list. This is somewhat analogous to a "liquidity crisis", and the solution is to lubricate the market with more supply (which is coming later this year). 

Second with higher inventory levels the number of properties that do not clear in a given sales season are higher. Some of these properties will be listed the next season. In 2015 and 2016 the high level of sales has meant most of this inventory has cleared, pushing resultant re-listings are down.

Attached housing (townhouse and apartments) continues to be stronger than single detached housing in terms of price strength as measured by the change in the MLS-HPI (see link at beginning of this post). This closing of the gap between housing types is to be expected in the wake of detached housing prices having had accelerated in previous years.

I continue to see no evidence to suggest a significant slowdown will occur in 2017. That is not to say a significant slowdown won't occur, but I don't see any evidence supporting it.

REBGV Sales Update Through February 2017

REBGV released their stats package through February 2017. Here are the numbers:





Sales have recovered and are now "average". Inventory is low. The result is a "lowish" MOI, which portends a decent spring in terms of price growth (not crazy but not weak either). Inventory growth is very low. Whatever is holding up owners from listing is exerting upwards pressure on the market. Attached housing (townhouse and apartments) is stronger than single detached housing in terms of price strength as measured by the change in the MLS-HPI (see link at beginning of this post).

I see no evidence to suggest a significant slowdown will occur in 2017. That is not to say a significant slowdown won't occur, but I don't see any evidence supporting it.

Vancouver Teranet and Housing Inventory Correlation Update

The past few months have seen a marked cooling in price changes and sales volumes in Metro Vancouver and I just got around to updating some scatterplots that relate the change in the Teranet House Price Index to the ratio of for-sale inventory to monthly sales (months of inventory, or MOI). Here are the updated charts, showing price changes for 3 months (quarter), 6 months (half) and 12 months (year):

An interesting development late last year saw prices drop significantly when compared to the available inventory — the model predicted more price strength in the face of a dearth of available inventory than what actually occurred. Prices have moderated somewhat but are still up 16% year-on-year, an indication perhaps that last spring's buying frenzy was anomalous compared to other years. Perhaps the spike in the spring of 2016 was more of a transient event and was not truly representative of underlying demand.

On the subject of inventory, inventory is very low right now, around 8000 or so for REBGV and inventory growth is nothing short of anaemic. As the market enters the traditionally robust spring selling season the lack of inventory will tend to place upwards pressure on pricing, all else equal.  I see no significant evidence yet to suggest that prices (as measured by the Teranet HPI) are exceptionally weak compared to current MOI, in the context of the historical price-change-to-MOI correlation. That stated, with inventory this low, the market may be operating in a different state compared to most past years.

REBGV Sales Update Through January 2017

REBGV released their stats package through January 2017. Here are the numbers:

January gives us an early read on market conditions preceding the busy spring selling season. It does not appear like conditions in 2017 are similar to 2016. Sales and new listings were both slightly on the weak side of average. As a result, inventory growth has not been high, but is average. Since inventory was extremely low through all of 2016, there is a long way to climb to bring inventory back to historical levels. The lack of inventory suggests that, even with low sales volumes, prices will likely remain robust through the spring. This is based on the historical high negative correlation between months of inventory and price changes.

2017 is an unprecedented year in that both sales and inventory are low. It is unclear to me how quickly, or if, the market can return to an average level of inventory and sales this year. Inventory can only climb by new listings significantly exceeding sales (and listing expiries not being high). Unless new listings pick up or sales slow further, there is no path for inventory to return to historical levels before the end of the spring selling season. (See 2006 for a potential template.)

REBGV Sales Update Through December 2016

REBGV released their stats package through December 2016. Here are the numbers:






As 2016 draws to a close it's worthy of a look back at the year and some lessons for what happened. 2016 saw a frenzy of sales in the first half of the year (strong sales were already present in 2015 and early 2016 was arguably a continuation of a crescendo of activity). New listings were above average as well, but have dropped off in the latter part of the year. Months of inventory is negatively correlated to price changes and indeed the first part of the year experienced robust price gains with few other years as equals. 

Conditions in the second half of the year were more subdued and were trending slightly below average for sales, though nothing as severe as 2008 and 2012, years in which year-on-year price drops occurred.

At the beginning of 2016 there were signs that it was going to be a strong year for sales. The strength in fall of 2015 coupled with low inventory was enough fuel to ensure price strength through the spring. Nonetheless the strength of sales in the spring of 2016 surprised me and likely most objective observers in the RE industry. I am not fully certain the reason for the strength but I expect it was a combination of factors including: offshore capital, low mortgage rates, increasing household formation, a spike in construction activity, and inter and intra-provincial population growth. I remain unconvinced that one factor was the overwhelming determinant to 2016's sales strength.

On the policy side, 2016 has seen the federal government take more aggressive action through macroprudential guidelines via OSFI, designed to limit the amount of lending both through underwriting guidelines as well as providing a significant shift in policy with respect to unverified income loans ostensibly popular with those using foreign capital to finance house purchases. These policy shifts are likely starting to kick in now. In addition there are indications that gaps in money laundering oversight are starting to close, though I have yet to read of anything substantive on this file yet. The BC government implemented a foreign buyer tax in a rather heavy-handed fashion and the City of Vancouver has implemented an empty home tax. Both of these taxes are likely to be challenged in the courts in the coming year. 

The provincial government has announced a program where first-time homebuyers can get a loan up to just over $35,000 towards purchase of a home, though it's notable that there have been similar schemes in years past that aren't that different. (There was a recent scheme for first-time homebuyers purchasing new homes not too long ago, then there was the brief 40 year amortization and 0% down schemes from the mid-2000s, so it's not like this recent provincial scheme is anything out of line with what we've seen before.) 

There appears to be more coordination between municipal, provincial, and federal housing policy than I've seen in a while. This bodes well for ensuring some unfortunate gaps in data collection and sharing are closed.

2016 also highlighted to me how limited governments can be in terms of effecting change in housing policy. There are significant entrenched interests, overwhelmingly dominated by incumbent homeowners. Producing affordable housing policy that almost certainly means affecting the living conditions of incumbent homeowners is a tough slog. I am far from convinced that there is a significant intersection between effective affordable housing policies and popular housing policies.

What will 2017 bring? Predictions are at best statistical but here are my "most likely" predictions for 2017:

• Sales will most likely trend at the long-run average. A combination of tighter lending conditions and higher prices will likely weigh on sales this year, however there is a large 25-35 year-old demographic that is starting to form households and that should buttress numbers for the next few years.
• New listings will most likely trend higher than the 10-year average due to a slight propensity to "cash out" by those thinking of downsizing, as well as housing completions likely rising.
• Inventory will most likely climb to about 12,000 by September (we are very low now and have a long way to run).
• Prices of attached and detached should increase through the spring since there is still some scarcity. I do expect detached to increase less than attached since detached inventory is higher and prices increased more than attached the past year, so are due to retrench. Price increases will most likely be skewed to higher-quality houses that will sell in most markets.

In the long run I expect detached housing in the core to outpace denser housing forms and suburban housing in terms of land value appreciation. Pressure to increase density will continue to grow, but we're talking a many decades trend that I fully expect to continue. Does that make detached a better investment? No.

Anyways, merry new year to all!