I’m not particularly fond of seasonally adjusted data. So when CREA released September’s figures showing that the national resale housing market actually rebounded slightly from the previous month, ( the first monthly increase since the spring) I was a little skeptical. Especially in light of non-seasonally adjusted year-over-year sales that were showing a -15.1% decline Canada-wide.
Scotiabank (Global Economic Research, Closing Points, October 15, 2012) has more insight on the matter:
Were Canadian existing home sales data better than expected today (+2.4% m/m), bad (-15.1% y/y), neither? The answer lies in the gloomy recesses of the seasonal adjustment process. The data on new home sales released today saw the largest ever upwards adjustment to the raw ‘unadjusted’ quantity of existing home sales – by far.
The amount of existing home sales that we saw in the seasonally adjusted month-on-month number that flashed across Bloomberg screens and appeared in the Canadian media was adjusted higher by an aggregate 13.7% (NSA: 32,192; SA: 36,601).
In contrast, in September 2011 the raw number was adjusted up by 1.2% (NSA: 37,903, SA: 38.383) and in 2010 the raw September number was adjusted up by 1.4% (NSA: 34,003; SA: 34,493).
The highest ever previous adjustment was +10%. Why the very high adjustment? CREA economists attributed it to a lower amount of trade days than normal this past September (i.e., too many weekends).
When coupled with other factors, too few weekdays resulted in this month’s mixed signals after the data had been seasonally adjusted. How do we interpret the data? While there’s strong logic to the methods employed by CREA – including the adjustment of housing turnover data to account for trade days – the result is that this month’s month-on-month number needs to be taken with a grain of salt and viewed in the larger context of the year-on-year number, which was very low, and the year-to-date number, which has slowed to crawl at 1% in September compared to last year.
BMO takes a different position by suggesting that we not put as much emphasis on the year-over-year data. They followed TREB’s lead by placing most of the blame of the softening sales figure on the extra weekend:
Well, this is a curious one. Canadian existing home sales were down 15.1% from year-ago levels in September, as expected, but that translated into a 2.5% increase from the prior month in seasonally
adjusted terms. The issue here is that there were just 20 weekdays this September, compared with 22 a year ago, and that 9% drop alone accounts for a big chunk of the year-on-year decline. Still, while the underlying trend in sales is likely not as dire as the headline yearly drop would suggest, there is little doubt that sales have taken a step back in recent months (Source)
RBC is completely out to lunch with their assessment that “the monthly increase may indicate that the rule changes’ cooling effect may be running their course faster than we anticipated.”
However they still expect “that intense negative pressure will persist in the near term before easing by year end. The Canadian market is in transition to a more moderate and sustainable pace following a decade of exceptional growth. We expect both resale activity and home prices in Canada to be flat or to slightly decline in 2013. Local markets where affordability is poor, such as Vancouver, will be under the most stress and face greater declines.” (Source)