Here is a quick update on the work mohican and I did in spring 2008 around refining mohican's work tracking price changes and months of inventory. Remember that the best fit was to track half-over-half (i.e. 6 months over 6 months) price changes to a three month moving average of months of inventory (total active listings at the end of the month divided by the sales in that month).
Here is mohican's famous scatter plot for half-over-half versus 3 month moving average MOI (the red dot is June 2009's datum):
Remarkably the correlation is disturbingly accurate into the downturn. Looking forward we can see how well the model has "predicted" the next month's price movements.
Weak-minded fool that I am, I second-guessed myself and thought the benchmark would come in at $690K, though if I had stuck to the model I would have been closer. I am at a loss to explain the reason for how well this model has done in Vancouver. Other cities where I have tried to run similar analysis show nowhere near as tight a correlation between months of inventory and price changes. Since the causation underlying this model is not well understood the model is interesting but not much more -- it will track until it doesn't.
If you're flabbergasted by the market's recent strength, consider affordability has improved close to 30% in the past year (thanks for pointing this out, fish10). That's a Brobdingnagian shift, even with aggregate incomes coming under severe pressure.