An Understated Canadian Jobs Report

• Employment flatlined in February…

• ...as the unemployment rate held steady because the labour force shrank

• Wage growth accelerated

• Hours worked to weigh on February GDP, add to Q1

• Weather likely explained some of the softness in jobs and hours

• Distorted SA factors are the biggest reason to ignore this report

• The BoC should walk it off, but it doesn’t matter either way to next week’s call

• Canadian jobs, m/m 000s // UR %, February, SA:
• Actual: 1.1 / 6.6
• Scotia: 20 / 6.7
• Consensus: 20 / 6.7
• Prior: 76 / 6.6

Canada’s job market was considerably stronger than the flat headline employment change (+1.1k jobs) suggested. The lowest seasonal adjustment factor on record is subject to a biased estimation approach. Furthermore, weather played a large role in damping down jobs, hours and the labour pool which perhaps Statcan should have considered with their SA factor!

Overall, I don’t think that the Bank of Canada will be fussed by these numbers, or at least there is a very strong case for how it should not be. They are likely to cut next week to land on neutral and given uncertainty’s damaging effects. Macklem is likely to sound very guarded on the bias.

I’ll cover the numbers first, and then the reasons to ignore the report.

Details—High Breadth to the Softness

Canada only added 1,100 jobs last month. Chart 1 shows some details.

Ontario was the only province to post material job growth. All other provinces were basically flat.

Full-time jobs fell by about 20k with part-time employment up by a roughly offsetting amount.

Private sector payrolls led the way (+10k) with public sector jobs up by about 8k of which half was a gain in civil servants.

Self-employed fell by 17k which is reason to discount the soft headline jobs tally. Many self-employed jobs are valuable, but I always view it as softer quality data than payrolls.

Chart 2 shows the sector breakdown. There was very little breadth to the gains that were focused on retail/wholesale that may suffer in the next report as the expiration of the GST/HST cut may have more of an influence. Most sectors were flat. There was little breadth to the material declines that were focused on two sectors.

The unemployment rate held flat at 6.6%. That was because the labour force shrank by 17k and hence a smaller pool of available workers offset the flat headline jobs reading. The participation rate slipped by two-tenths to 65.3% for the lowest reading in four months. That is the first shrinkage in the size of the labour force of folks either employed or looking for work since September 2022, and the biggest since June 2022, all in the earlier and volatile pandemic rebound days. I'm unsure of why. Maybe tighter immigration is finally having its effects after months of expecting it to, or maybe folks got so pessimistic, snowed-in or whatever that they quit looking.

Hours worked were crushed (chart 3). They fell 1.3% m/m SA. They are tracking a gain of 1.4% q/q at a seasonally adjusted and annualized rate so far in Q1 (chart 4). So, hours worked are a negative for February GDP, but still tracking favourable for Q1 GDP given that GDP is hours times labour productivity. Canadian GDP got a productivity boost in Q4, whereas Q1 GDP is getting a boost from hours.

Wage growth surged and registered the biggest gain since October (chart 5). Wages were up by 6.1% m/m SAAR in February. That’s no5 enough to arrest the trend softening over the past four months, but it’s enough to question its durability.

WEATHER DEPRESSED JOB GROWTH

Now for the reasons to ignore the report. For one, Statcan is citing snowstorms as the reason for the drop in hours worked in February. Snow in Feb in Canada? Oh go on, shocking. But seriously, if you've been here, it has been ridiculous with much more snow than prior years. Here’s what they had to say:

"Notable winter storms buried parts of Central and Eastern Canada in snow throughout the LFS reference week of February 9 to February 15. In total, 429,000 employees lost work hours due to the weather for part of the week (not seasonally adjusted). This was more than four times higher than the average number of employees who lost work hours due to weather in February over the previous five years (96,000)"

I think weather likely held back job growth in certain sectors that can be sensitive to weather, like restaurants and bars, info/culture/recreation, construction etc.

Seasonal Adjustment Factors Are Rubbish

Chart 6 shows that the seasonal adjustment factor applied by Statcan was the lowest for any month of February on record. SA factors are calculated with a strong recency bias skewed toward the experiences of recent years. All of the lowest ones on record have been drawn from recent years and it’s unclear why that is. It could be partly due to the timing of the pandemic’s hits on the economy, the on-and-off rebounds, and the stimulus responses. It could signal something more profound given that even years before the pandemic were relatively softer on SA factors.

Chart 7 shows that the weak jobs number wasn’t really because of the seasonally unadjusted change. That was within the realm of normalcy for a month of February. Therefore, the application of a record low SA factor for the month of February was the key driver of the weak seasonally adjusted jobs print.

The added curiosity is why didn’t judgement at Statcan not override what their SA factor models are telling them especially in a month when snow and general weather were much worse than a normal month of February?

How important is the issue? Chart 8 shows that at any SA factor other than the historic low this time, Canada would have posted material job growth. 

In short, weather combined with what I think are probably unreliable seasonal adjustment factors are two very good reasons to toss this report in the round file. Added to this the fact that employment lags developments, and so there should be much more focus upon forward-looking debates like whether US trade policy will come to its senses and how that may affect trend job growth going forward.