Canadian Jobs — A Messy Walk in the Park

• Canada added a small amount of jobs due to temporary election hiring…

• ...but jobs were lost ex-election effects

• An historically low seasonal adjustment factor biased the outcome

• Historically low SA factors will soon turn toward historically high SA factors …

• ...and it’s unclear why earlier distortions are lingering, beyond merely mechanistic calculations of SA factors

• Markets paid little attention to the figures

• Scant implications for the Bank of Canada

• Canadian jobs, m/m 000s / UR %, April, SA:
• Actual: 7.4 / 6.9
• Scotia: 25 / 6.7
• Consensus: 5.0 / 6.8
• Prior: -32.6 / 6.7

Going through this report is like watching where you step in a poorly maintained dog park. I don’t trust the headline or the details and think it changes nothing for the Bank of Canada.

On the surface, it looked like 7.4k jobs were created last month which was in line with the consensus guess. Some details are shown in chart 1 and I’ll elaborate upon others.

ELECTION HIRING PROPPED UP THE HEADLINE AS EXPECTED

The first issue is that the election effect was about what I had expected. Public administration jobs were up by 37.1k which reflects an election hiring effect given that two of the advance polling days appeared in the reference week. As Statcan noted, “The increase was mostly in temporary work and coincided with activities associated with the federal election.” That made the effect similar to the prior three elections in that sense (chart 2).

EX-ELECTION HIRING WAS WEAKER THAN EXPECTED

Then everything else fell apart. The ex-election component was weaker than I had guessed, but with a strong caveat. Jobs excluding public administration fell by about 30k. I’ll come back to other details after explaining why I don’t trust the report.

SEASONAL ADJUSTMENTS ARE COOKED

The caveat is that Statcan applied the weakest seasonal adjustment factor on record compared to all other like months of April in history (chart 3). This reflects a recency bias drawn from pandemic and post-pandemic distortions. 

Why does that matter? Because it’s the entire reason for the weakness ex-election effects. Seasonally unadjusted hiring in April was in line with historical norms for like months of April (chart 4). And if we remove public admin jobs where the election hiring occurred this month and do so for every like month of April in history, then the seasonally unadjusted job gain was still perfectly in line with historical normal for like months of April (chart 5).

At any other SA factor for like months of April in history, there would have been notably higher job creation. Chart 6 shows the range of m/m changes in jobs at different SA factors that have been used for past months of April. And so, we’re being asked by the data agency to take on faith that they know what they’re doing with such out-of-sample SA factors. 

Given the reliance on distorted SA factors, Statcan put some effort into explaining why they went with the lowest one on record, right? Wrong. No mention. Nothing, nadda, zilch, trust us, it’s what the model spits out and they don’t think about whether it’s reasonable or not. Absent an effort at some explanation this is what we’re left to conclude.

The appendix to this note presents the historical SA factors for all other like months in history. Statcan has been using historically low SA factors for like months of January through May, and then they shift to historically higher SA factors for like months of June through August before being tempered in September through to November and then start to weigh on hiring again in December.

I’m not quibbling with SA factors per se and fully understand SA 101 after three decades in the biz. I do, however, question whether we should be as reliant upon historic outliers for two-thirds of the months in a year over recent years compared to history for those same months. For instance, is the pandemic really carrying such lasting influences on SA factor distortions in 2025 as it did in its depths and in the initial stages of recovery? Or are we just skewing the SA factors to averages that are still too heavily weighted toward what may have driven unusual patterns of activity in prior years?

DETAILS

Because I don’t trust the overall report, take the following details with a mountain of salt.

Full-time jobs were up 31,500 and part-time jobs fell 24k. Most of the full-time was in temporary election jobs.

Chart 7 shows the sector breakdown of the change in jobs. Goods sector hiring was down 33k and led by a 30.6k drop in manufacturing jobs. 

Services hiring was up 40.3k but only because of election hiring.

Within services, wholesale/retail was down 26.8k, business/building/other support services fell 14.5k, health care and social assistance was down 7.8k jobs, and accommodation and food services down 8.5k. There were some gainers, but not enough to offset the losers, and so it's not just the tradeables sector that was weak

Wages were up by 3.1% m/m SAAR which is a bit of a pick-up from the prior month's 0.0% but still softish (chart 8).

Hours worked were up 4.5% m/m SAAR which is a plus for April GDP, but hours are tracking a rise of only 0.7% q/q SAAR in Q2 which is not great for Q2 GDP so far (chart 9).

The unemployment rate moved up two-tenths to 6.9%. That was driven by a 46.7k rise in the size of the labour force after two weak months. That is probably a temporary effect. The main driver of the higher UR in Canada is likely to reverse the next month. It was driven by a temporary addition to the labour force due to election hiring, against a trend of tighter immigration policy and weaker population growth. Elections always temporarily pull in marginally attached workers, like retirees, who are keen on getting a few days of paid work at polling stations etc. So, don't read anything into the rise in the UR imo.

BANK OF CANADA IMPLICATIONS

I don’t think anything has changed for the BoC and only partly given the low trust factor for the report.

The BoC’s mandate is inflation. 2% over the medium-term. The tension remains between growth/jobs versus 2% over the medium-term while still at neutral. That tension is driven by four things that will take time to inform.

• One is what happens to inflation through retaliatory tariffs and whether they are a one-off price level effect, or a catalyst to renewed inflation (ie: sustained price changes). That depends upon inflation expectations, pass through effects, how behaviour may change etc.
• Second is uncertainty toward how supply side shocks hit Canada. Will Canada import negative supply chain hits with higher prices due to the connection to US supply chains given how product shortages loom stateside and may ripple through N.A. supply chains? And/or will this be offset by, say, China dumping into Canada which my hunch leans toward viewing as a minor effect given little Canada-China trade?
• Third is whether the demand side gets hit harder than the supply side or vice versa, and timing the relative effects.
• Fourth is how other considerations evolve. For one, we expect notable fiscal stimulus in a June budget that will be passed before summer recess. The Carney administration only needs token help from the opposition parties in other to pass a major stimulus package. The June 4th decision follows the appointment of cabinet, the speech from the throne, further establishment of policy priorities, and traction toward a budget. Let’s please not repeat the massive combined monetary and fiscal policy overshoot that was delivered the last time around!

Governor Macklem has been very clear about the difficulty between balancing downside risks to growth including jobs versus achieving their inflation mandate amid some retaliatory tariffs, imported supply chain pressures, and a desire to avoid misreading the net demand and supply pressures that they totally misread the last time.

And so that's likely why your curve and CAD reactions are miniscule. So far. I think markets are reading the initial reaction correctly although it's also possible that's because they're not looking beneath the distorted headline (by election hiring), yet if they did, they'd look at distorted underlying changes thanks to SA factors.

It’s not that it’s entirely unreasonable for markets to be pricing another 50bps of cuts through year-end. It’s just that I wouldn’t push it further at this point, and think that the uncertain effects on inflation plus the likely substitution of fiscal easing for monetary easing make the outlook more complex and uncertain than existing market pricing.

APPENDIX—EMPLOYMENT SEASONAL ADJUSTMENT FACTORS FOR ALL MONTHS