ROB Insight is a premium commentary product offering rapid analysis of business and economic news, corporate strategy and policy, published throughout the business day. Visit the ROB Insight homepage for analysis available only to subscribers.
Almost as soon as Canada's near-record 95,000-job May employment surge was unveiled in Statistics Canada's labour force survey last Friday, economists began dismissing the dramatic rise as unsustainable. Others have raised an even more damning possibility – maybe it's just flat-out wrong.
One such critic is Garth Turner, long-time media personality, investment adviser, financial author, business pundit and former federal cabinet minister. In a posting on his blog this week, Mr. Turner – who is massively bearish on the Canadian housing market – questioned how the May job boom, nearly 43,000 of which came from new construction jobs, could possibly be anything but a statistical illusion. Ditto the May housing-starts numbers from Canada Mortgage and Housing Corp., which showed a surprise surge to an annualized rate of more than 200,000 units.
"Maybe it happened because the federal government gutted StatCan's budget. Maybe it was just a rogue poll … Maybe somebody lost the real one on the bus. Whatever. But the economy did not create 95,000 new jobs last month, half of them in construction," he wrote. He argued that given the year-to-date slump in housing starts and the condo market, particularly in Canada's biggest market of Toronto, the construction-led job surge is simply beyond belief.
Statistically speaking, the labour force survey could have simply been wrong. It is, essentially, a poll via telephone of 56,000 "representative households" across the country; like all polls, it has a statistical margin of error. In the case of the May numbers, Statscan indicated that the job increase of 95,000 was accurate to within plus or minus 57,400, 19 times out of 20. In other words, Statscan can say with 95-per-cent confidence that the actual job gains were somewhere between 37,600 and 152,400 last month, with 95,000 being the mid-point. But there's a 5-per-cent chance that the actual job count is outside that range.
Yet it's just as likely that the actual job gains were more than 95,000 than less in May – and being within the error range is far more likely than the slim chance that the survey was wrong altogether. And even at the negative end of the range, 37,600 jobs would still represent a solid month for job growth. (The margin of error for construction alone, incidentally, indicates a range of 8,500 to 76,900, 19 times out of 20.)
For the employment numbers, and the construction gains in particular, to be flat-out wrong, you would have to believe that the recent surge in housing-construction indicators are also statistical figments. The CMHC's May housing starts (based on an actual count of all residential construction activity in the country's major urban centres) would have to be wrong. Statscan's building permits count (derived from information provided by the country's municipalities about actual permits issued), which hit a six-month high in April and is up 21 per cent since the start of the year, would have to be wrong.
To believe in that much statistical error, from multiple sources and all at once, would require a level of doubt and suspicion akin to conspiracy theories. The job-growth numbers very well may not be sustainable, but they are even less likely to be an outright mirage.
David Parkinson is a contributor to ROB Insight, the business commentary service available to Globe Unlimited subscribers. Click here for more of his Insights, and follow him on Twitter at @parkinsonglobe .
The Globe has launched a Streetwise and ROB Insight newsletter, with content available exclusively to Globe Unlimited subscribers. Get the best of our exclusive insight and analysis delivered straight to your inbox in a daily e-mail curated by our editors. Sign up for it and other newsletters on our newsletters and alerts page .