Open this photo in gallery:

Excluding health care and public administration, Canada’s job market has been shrinking since last August.Tijana Martin/The Globe and Mail

Canada’s job market is flatlining as the trade war takes its toll, with payroll employment growing by only 0.2 per cent annually in June, according to Statistics Canada.

In fact, payroll employment peaked in January on a seasonally adjusted basis and has yet to reclaim that high. And if just two leading sectors, health care and public administration, are excluded, Canada’s job market has been outright shrinking since last August.

Canadian economy sheds almost 41,000 jobs in July, unemployment rate remains steady

The number of factory jobs fell 8,400 on a month-over-month basis, as manufacturers scaled back operations in the face of U.S. President Donald Trump’s tariffs on imports from Canada. Based on the payroll data, manufacturing employment has fallen every month since December, wiping out 26,600 jobs since then.

Meanwhile, employment at retailers suffered the next largest decline, dropping by 8,100 positions between May and June.

On an annual basis, health care and public administration have driven the vast majority of Canada’s meagre job gains.

However, even on that front, Thursday’s report offered a note of caution. The number of job vacancies in health care fell 22,000 from the year before, the biggest decline of any sector.

Statistics Canada’s Survey of Employment, Payrolls and Hours (SEPH) paints a far grimmer picture of the job market than the more closely watched Labour Force Survey (LFS), which is more current. (LFS job numbers for August will be released next Friday.) In last month’s LFS report, employment notched down on a monthly basis, but was still 1.4 per cent higher than the same month a year earlier.

However, the SEPH, as a measure of actual employer payrolls, is regarded as a more accurate gauge of labour market health than the LFS, which surveys individuals about their employment status.

A big reason for the discrepancy between the two measures is the population estimate underlying the LFS “has been slow to react to the actual slowdown in Canadian population growth,” said Brendon Bernard, senior economist at Indeed Canada, giving the LFS job count an undue lift.

What’s more, because the SEPH counts the number of jobs rather than the number of people who say they’re working, even if Canadians are picking up second jobs to make ends meet, that’s still not resulting in overall job growth.

Decoder is a weekly feature that unpacks an important economic chart.