Last year I did a number of blog posts criticizing the consensus view that Japan had fallen into a recession. Many people assumed that two consecutive declines in real GDP meant that Japan was in recession. I argued that the two consecutive quarter criterion was not reliable, for instance RGDP did not fall for two consecutive quarters in the 2001 recession in the US. The NBER is considered the official arbiter of business cycle dating in the US, and they look at a wide range of indicators.
In the US, unemployment always rises sharply during recessions, and I believe that's the stylized fact that people should focus on. Obviously people are free to define recessions as they wish, but our models of recession are typically aimed at explaining big increases in involuntary unemployment.
At the time, some commenters suggested that unemployment was a lagging indicator, and that my claim of no Japanese recession was premature. Here's the Japanese unemployment rate over the past 10 years.
As you can see, unemployment does increase in an actual recession like 2008-09. However in the three subsequent pseudo-recessions the rate of unemployment did not increase. In the most recent case, Japanese consumers bought lots of goods right before the April 1, 2014 sales tax increase, and GDP fell immediately afterwards. Firms didn't lay off workers, knowing that it was simply a timing issue, not an actual recession. Massive natural disasters like the 2011 tsunami also don't impact unemployment in a big, diversified economy.
Today one can find many news reports that Canada has fallen into recession. My prediction is that a year from today we'll know that this was also a pseudo-recession. Unlike in 2008-09, Canada will not see a significant rise in its unemployment rate (which has been stuck at 6.8% for the past 6 months.)
Canada's RGDP fell due to a weak commodity sector. However commodities are land and capital intensive, not labor intensive. So relatively few jobs are lost when commodities decline. Even Texas is still gaining jobs. The fall in Canada's real GDP is an optimal response to the negative commodity shock, and should not be addressed with monetary policy. Indeed in the past I've suggested that major commodity exporters should target something like total labor compensation, or average wage rate, rather than NGDP.
For big diversified economies like the US and China, it probably doesn't matter. In those two cases a significant fall in RGDP would probably represent a recession (indeed even a small increase in RGDP in China might be viewed as a recession.)
To summarize, don't obsess about words such as "recession". Words can get in the way of meaning. The concept we are interested in is a general fall in economic activity across a wide range of industries, leading to a big rise in unemployment. Between January 2006 and April 2008 we saw US housing construction fall in half. But that was not an economy-wide phenomenon; it simply represented the after-effects of misallocation of resources into housing (or at least misallocation is the standard view, Kevin Erdmann would argue otherwise.) But regardless of what caused the housing slump, unemployment merely rose from 4.7% to 5.0% over 27 months. That's not significant, as recessions in the US are caused by monetary shocks that reduce NGDP, not reallocation of resources. When actual recessions occur, unemployment rises by at least 200 basis points
PS. If you don't accept my definition of recession, then call a generalized fall in business activity a "banana". Then my claim is that tight money causes bananas. (Does anyone recall where I got the term 'banana'?)