I would be interested in knowing why CanaData’s forecast of 200,000 housing starts in 2008 is so much worse than the CMHC’s forecast 214,300. The reason I ask is that CMHC has been wrong about the direction of housing starts in 2004, 2005, 2006, 2007. In each of those years the starts sets new records while the CMHC, like this year, predicted a decline.

This record of error makes it clear that the CMHC does not have any special insight into the economic forces affecting the Canadian housing market. In fact, despite the CMHC and Canadata’s forecast, Canadian housing starts rose 20.6 percent in January to a seasonally adjusted annualized rate of 222,700 units.

If you keep predicting a decline long enough, eventually you’ll be right. However, in recent years you could make a better forecast just by looking at the CMHC and predicting the opposite.

CanaData’s prediction of a 12.3% drop in housing starts is alarmingly close to the 14% drop in housing starts at the beginning of the recession in 1991. That drop began a long decline and starts did not recover to their pre-1991 level for over a decade.

The recent trend, as Mr. Carrick notes, is a leveling off. You would not expect an abrupt return to pre-2002 levels unless there was steep recession in the U.S. that had a strong effect on Canada.

Two of the other reason given for expecting a decline could just as easily be reasons to expect continued growth. If more people are staying in their home provinces because the economy there is improving, it is a reason to expect a stronger housing market in those provinces to counter the decline elsewhere. Similarly, high prices are caused by high demand, which is likely to stimulate construction.

Barring a major recession in the U.S., it seems more likely that starts in 2008 would be close the the 2007 level.