Warning: The data on which much of this column is based comes from a survey with a very wide margin of error.
Wait. Don’t go. I recognize that this is the age of Big Data and therefore no one need waste her or his time on shoddy information. But I’m talking about the latest estimate of hiring and firing by Statistics Canada. Judging by the initial reaction to those numbers Friday morning, I’m guessing some of you were unaware that the figures are little more than rough approximations of what’s happening in the labour market. They are helpful, but no one should be using them to form conclusions.
And yet so many of us do.
StatCan’s newest survey of the labour market definitely looks bad on the surface. Some 88,000 people lost their jobs in January, the report said. The decline marked the end of 17 consecutive months of hiring; worse, it was the biggest drop since January 2009, a time when many feared a repeat of the Great Depression. A bad omen, surely.
Capital Economics, a research shop in Toronto, called the release a “game changer.” Bank of Montreal said the terrible news means we can “dismiss” any notion of the Bank of Canada raising interest rates in March, and that an April increase now is a “long shot.”
Too many tweeters to count declared that the data proved that Ontario’s minimum-wage increase was the job killer they said it would be. StatCan said part-time positions plunged by 147,000, the most ever, and a lot of them were in Canada’s biggest province. Those were precisely the kinds of jobs the critics of Premier Kathleen Wynne’s policy said would be the first to go. Another bad omen, surely.
The assertions could turn out to be true, but let’s be clear about something: The asserters are guessing. The data simply aren’t reliable enough to do anything else. Those of us who get paid to circulate facts and arguments about the economy do a poor job of explaining the quality of the data with which we work. That’s especially true of StatCan’s monthly survey of the labour market, which is based on a poll of approximately 56,000 households in the middle of every month. There’s only so much a poll of that size can tell you definitively about a working population of about 20 million people.
That’s why StatCan nudges readers of the Labour Force Survey to consider the data over a period of time, not in isolation. Generally speaking, the agency is only confident that it has the direction of the change in employment correct when there has been a shift of at least 46,000 positions in one direction or the other. Anything less is statistically insignificant.
So what can be said about the state of the economy based on the latest jobs estimates?
Rather than a precipitous drop in economic activity, the figures suggest the jobs data at the end of 2017 exaggerated somewhat the economy’s strength. Many economists assumed that was the case. Ahead of Friday’s release, the three-month average monthly increase was about 60,000, too much for an economy that had slowed to an annual rate of growth of 1.7 per cent in the third quarter. Now, the three-month average is a gain of about 19,000, a more realistic number.
The unemployment rate stayed below six per cent for a third consecutive month in January, matching a streak that has happened only once before in records that go back to 1976. That’s a sign of strength, as a jobless rate that low suggests the economy is operating near “full employment,” or the level at which economists estimate that additional hiring would begin to put upward pressure on inflation.
Another sign of strength: The average hourly wage jumped 3.3 per cent in January from a year earlier, the biggest increase since 2015 and the fifth consecutive month that wages gained in excess of two per cent on an annual basis.
The Bank of Canada might indeed wait until the summer to raise interest rates, but it’s unclear that weakness in the labour market will be the reason. Policymakers have emphasized that they will be waiting for evidence that wages are accelerating; StatCan’s data offer them some. What central bankers are lacking is clarity about the future of the North American Free Trade Agreement. That uncertainty is the most important factor in determining the path of interest rates.
As for the effects of Ontario’s new minimum wage, it simply is too soon to say whether the decline in January was the result of a revolt by employers.
To some extent, the drop in part-time employment reflects a migration to more secure employment. The number of Ontario residents between the prime working ages of 25 and 54 who said they had full time jobs increased by 29,200 in January, and was about 46,000 higher than a year earlier. And overall hours worked were little changed from December, suggesting output was unaffected. So maybe the higher minimum wage destroyed some part-time work, but there’s just as much evidence that it had no impact on the broader economy. In fact, Ontario looks better than it has in years.
But you can be sure that the small-business lobby, ideologically inclined economists and Wynne’s partisan opponents will circulate the unflattering headline numbers from the latest jobs report. That makes advertising the weaknesses of those figures all the more important; not because they might play a role in Wynne’s political future, but because voters could end up making decisions based on false assumptions.
StatCan could do a better job of explaining the limits of its pursuit of high-frequency data, perhaps by including a disclaimer at the top of its reports instead of at the bottom. Meanwhile, those of us who know better should resist sexing up the results for the sake of more dramatic stories and tweets. We don’t discuss polls without including the margin of error. Why should writing about economic data be any different?