Teranet and Months of Inventory Update Through June 2016

Long-time readers will remember the scatterplots this blog produced outlining the correlation between house price index changes and months of inventory. When mohican first investigated and quantified the correlation it was 2007 and even with only three years of data a decent relationship was shown. We now have about 11 years of data and the relationship has held up very well under a broad range of market conditions.

The methodology is to take two datasets, the ratio of for-sale month-end inventory and monthly sales (i.e. Months of Inventory or MOI) as published by the Real Estate Board of Greater Vancouver (REBGV), and the Teranet House Price Index, then cross-correlate them with a small amount of averaging.

I present two scatterplots here, the first being year-over-year change in prices plotted against months of inventory. Prices are lagged by six months to maximize correlation.
 
The second graph turns out to have highest correlation, relating six month price changes to months of inventory with a three month lag. It is amazing, even after the most recent strong sales activity in Vancouver that this relationship has held so well!


This correlation is what I have recently been calling the "mechanics" of the market.

The first question is surrounding causality: are MOI and price changes coincidental or does one cause another? The short answer: MOI level most likely causes price changes.

(Months of Inventory is close to a real-time dataset -- it is reported quickly a couple of days after month-end. The Teranet HPI reports based on closed sales that are delayed by some months after prices are negotiated. Not surprisingly MOI leads the delayed reporting of the HPI inputs. MOI as I report here has a 1.5 month group delay (3 month moving average), and based on conversations with several real estate agents a typical closing time is 2-3 months. Taking all this into account, MOI does indeed lead price changes, and this matches the intuition: less supply available means more competition between buyers and prices will be "bid up".)

An interesting ancillary observation here is that the dearth of inventory has been almost exclusively due to high sales volumes and not low listing levels. Homeowners have been, at least recently, slow to change behaviour in the face of significant price changes. In addition there is an inherent and unavoidable lag between construction starts and completions (about two years for multi-units, just over a year for single detached). The amount of supply has therefore been broadly fixed and has not kept pace with the sudden change in demand. The result is that MOI has fallen to unprecedented lows and prices have shot up.

Whatever the reasons for the heightened demand (theories and conspiracies abound!), high sales and low inventory has caused prices to go up. But where do we go from here?

Current MOI is around 2. Say for argument's sake we want to "soft land" the market and return price changes to around 5% per year (never mind correcting). Sales would need to stay below new listings such that MOI can climb back to an year-round average around 5. That is a significant move from the current situation and would necessarily mean significant new listing growth, falling sales, or both, just to get things to "soft land". In other words, even a soft landing will require a significant sales/listings transient to hit the market.

The danger with a bout of significant price growth is that it takes a ton of fuel just to level things out, and there is a greater risk of overshooting on the downside. Even absent an overshoot we can expect some stomach-churning along the way. Luckily we all know what overshooting looks like so it shouldn't be a surprise to anyone if that happens.