How Canada’s Unemployment Adjustments Artificially Lowered The Rate

Do you recall other peak economic cycles? Think back to 2007, 2000, 1989… or further back if you remember them. Those were nothing compared to the current economic prosperity. Aside from last year, no other July has ever reported a smaller share of unemployed Canadians. It’s an employee’s economy, where they call the shots, and employers listen. No one needs to even really look for a job.

Circle back to your image of 2007, 2000, or 1989. Does the current environment feel the same? Most people we’ve encountered, with the exception of politicians, tend to disagree. Cities are filled with lines, but they aren’t for trendy restaurants or shopping. They’re for food banks and to apply for entry-level, low wage jobs. The best paid young adults can’t afford a home in their cities. Even with a sharp correction, it’s unlikely most ever will. The picture of reality doesn’t quite resemble the data, does it?

What if the data was wrong, and the quick-to-react policymakers of  the 2020s acted on a skewed read? Notable experts warned this happens after recessions due to seasonal adjustments. Post-recession economic activity fails to follow typical seasons, resulting in overstated growth. That’s most likely the case with Canada’s unemployment rate, which has been driving policy to address issues that aren’t present. 

“When the anecdotes and the data disagree, the anecdotes are usually right. There’s something wrong with the way you are measuring it”

— Jeff Bezos, astronaut

What The Heck Are Seasonal Adjustments? 

Seasonal adjustments are a statistics filter used to eliminate predictable seasonal shifts. They first popped up around 70 years ago, when snow meant you physically couldn’t get to work. Today it’s used for everything from employment to home sales. Rather than looking at retail sales crash from December to January, you now have a smooth line. Easy peasy to read, without much noise.

Smooth charts are nice, but what’s predictable? Sure, as stated above, retail sales fall from December to January, because my pitch for a two-month Christmas got a cold reception. But, and I’m just spitballing here, but what if a pandemic delays a busy season for an industry? The demand is shifted into a non-seasonal window, and is interpreted as a massive surge. A deceptively large increase fuels other distortion patterns. Not just behavioral (“sales are so high, we need to buy now”), but also comparison patterns. There’s a reason there’s no such thing as seasonally adjusted oil prices. 

Seasonally Adjusting Unseasonable Economic Periods Doesn’t Work

Notable experts have warned that both seasonal and unadjusted data need consideration. The EU central bank notably warned there’s little research on the impact of seasonal adjustments and the business cycle. They found a recession’s impact on seasonal adjustments is delayed, leading to distortions.

Wacky Euro bankers. What do they know? Well, US Federal Reserve researchers warned that unseasonal factors would distort post-pandemic readings. They cite the Great Recession as an example, where employment was overstated. Consequently, it led to perplexed policymakers when strength faded a year later. 

The Fed researchers cautioned analysts this will occur post-pandemic. The readings will then flip the other way, as the distortions create a reaction. This will potentially cause an overreaction and misunderstanding of the economy. They explain that manual data-intervention can help, but it’s impossible to figure out the seasonal shift until well after the fact. 

In an economy where policymakers react almost instantly, that creates a problem. Prominent Harvard Business School (HBS) Professor Daniel Quinn Mills suggests trashing seasonal adjustments. His argument is that policymakers make decisions much faster today than 70 years ago. Consequently, those decisions are made on monthly data that may not reflect reality. Elected officials might as well drive around, and guess.

We know a high school econ teacher probably told you to only use seasonally adjusted data. Now it’s hard to trust experts from the EU central bank, the US Federal Reserve, and Harvard. What do those folks know, right? Anyway, humor us and let’s compare the unadjusted employment data. 

Canadian Unemployment Is On The Rise, But How Much of It Is Seasonal?

Canada’s seasonally adjusted unemployment rate climbed for a third consecutive month. It rose 0.1 points to 5.5% in July, and is 0.6 points higher than a year ago, when it was at the record low. 

In contrast, the unadjusted rate climbed 0.7 points to 5.9% in July, equal to the annual increase. July’s unadjusted increase was 40% larger than the one seen last year, and more than double the size a year prior. Possibly a little too much seasoning on the adjusted rate, so let’s take a closer look.

Let’s hop into our time machine and go back to when we were naive fools—2019. Unadjusted monthly growth for July was the same size as this year. It’s also just slightly higher (+0.1 points) than the median for July over the 5 year period ending in 2019. Ditto with the 10 year median. This year’s move might seem like an unseasonable surge, but it’s totally normal.

The issue is the overstated strength in the years following the pandemic. It’s literally what Fed researchers warned would happen.

Just to get a baseline, let’s look at July over time. The below chart shows seasonally adjusted and unadjusted employment rates since the 80s. Prior to the mid-90s, seasonal adjustments printed a higher unemployment rate. Afterwards, it switches to understating the unemployment rate.

Canadian Unemployment Rate

The seasonally adjusted and unadjusted rate of unemployment for July.

Source: Statistics Canada; Better Dwelling.

A few issues worth a dive, including the influence of overstating prosperity for Gen X and beyond. However, today’s takeaway is that seasonal adjustments are generally consistent over time. That’s a good thing, since predictable patterns require consistency.

Canada’s Record Low Unemployment Rate? Likely A Mistake

First, let’s discuss pandemic operating distortion. A shutdown at the end of March meant a period that is normally busy would suddenly have no activity. This pushes economic activity into a period where the economy is expected to slow. Some focus was placed on ensuring unemployment wasn’t overstated at the time. How that impacts the seasonal pattern in the future is a little harder to figure out. In other words, the end of the year looked strong in a restricted  economy.  

Let’s focus on the distortions that would occur in 2021, as per the Fed warning. The below chart shows the variation of seasonal adjustments from unadjusted data. It’s clear there’s a bit of an overcorrection, trimming more than usual from January, March, and August. It boosts unemployment slightly more than usual in June, September, and October. However, the bias is clearly towards overstating a lower unemployment rate. 

Impact of Seasonal Adjustments On The Unadjusted Employment Rate

The change to the unadjusted employment rate when seasonally adjusted, in percentage points.

Source: Statistics Canada; Better Dwelling.

Canada’s miraculously low record unemployment rate is a point worth highlighting. In June and July 2022, it fell to 4.9%, a rate the country has never before seen. This fueled discussions about labor shortages, in turn boosting wages and interest rates. June typically sees a much larger increase, and July a much smaller decline.

Below is an isolated look at the impact of seasonal adjustments on July. To say the seasonal influence has a much bigger than usual impact would be an understatement. 

Seasonal Adjustments Are Having An Unusually Large Influence On The Unemployment Rate

The percentage change to the unadjusted employment rate when seasonally adjusted for July.

Source: Statistics Canada; Better Dwelling.

The good news is the rise in Canada’s unemployment rate is likely smaller than thought. The bad news is the unemployment rate was likely understated, due to how hard it is to adjust data. That means the record will likely be revised in coming years. Canada’s labor market is still very strong, just not as strong as many were led to believe. 

To clarify, this isn’t saying the data is politically manipulated or Statistics Canada messed up. Stat Can has the impossible task of figuring out how seasonal patterns change. They don’t get to say, “we’ll report it in a few years when the trend is a little more clear.” 

However, Prof Mills’ point still stands. Policymakers make decisions much faster than they did 70 years ago. Stat agencies have to make seasonal adjustments, even in periods where the seasonal influence has shifted. Perhaps, permanently shifted. The public still faces the consequences of decisions made by policymakers that didn’t understand the data. Data that may or may not represent an accurate picture of reality during a critical period for an economy.