Canadian Interest Rates Were Raised On Junk Employment Data

Canadian Interest Rates Were Raised On Junk Employment Data

Canadian real estate prices are very closely related to employment numbers. They don’t just impact potential buying pools, they also strongly influence interest rates. However, what happens if we we’re bullish on job numbers, and we’re reading them wrong? Most wealth managers and fund researchers don’t rely on the employment numbers most of you recieve. The numbers the media parades are less useful in finance, especially when determining a short term outlook. Here’s how to read job numbers, and get a better picture of how the economy is evolving.

Things You Need To Understand

There’s a couple of things you’ll need to know today, seasonal adjustment and how job numbers are born. First, seasonal adjustments are a statistical method for removing a seasonal bias from a time based series of data. It’s a method that was developed 70 years ago, to compensate for the change in weather related employment, like construction. In 2018, there isn’t all that many jobs that cease to exists for the whole winter.

Now don’t get me wrong, seasonally adjusted data makes charts easier to read. However, it’s not an accurate report when reporting a single month’s data point. The Wall Street Journal’s Daniel Quinn Mills pointed out the issue with this a couple of years ago. The issue he presents, is that policy decisions are now made much more rapidly than they were more than half a century ago. A couple of seasonally adjusted job reports are all people need to make policy changes. This can have huge consequences when using seasonally adjusted data.

The second thing you need to know is how job numbers are created at Statistics Canada (a.k.a. StatsCan). They use a survey of 55,000 people, and project that for the population. It’s a fairly large sample, and they’re pros at getting distribution right. However, using 55,000 people to get information about a population of over 36 million people, does have its problem. This will also be clear when we compare them to ADP’s numbers.

You Heard That Jobs Increased 79,000

Let’s go through the jobs reported in December from StatsCan, which is the bullish report most of you heard. StatsCan reported a seasonally adjusted increase of 79,000 jobs in December. This brings their estimate of the total number of employed individuals to 18,648,000. Unemployment also declined 0.2 points, down to 5.7%. It’s a really good report, and it spread confidence through the economy. It also influenced the Bank of Canada’s decision to raise interest rates. But here’s the thing…

Unadjusted, Canada Lost 6,700 Jobs In December

Using the unadjusted data, gives you a quicker read on the economy. Unadjusted, Canada lost 6,700 jobs in December. This means there were actually 6,700 less people working that month.  That brings the unadjusted estimate of total employment to 18,560,400. That’s a total of 87,600 less people working, than in the “adjusted” data. These losses might eventually show in seasonally adjusted data, but it would be softened, and delayed. So don’t be surprised when we get a soft job report in March.

They both add up to a positive employment change, but the seasonally adjusted data is 87,900 jobs higher. Think of it this way, would over shooting your income by 18% be okay? Probably not. Source: Statistics Canada.

ADP Job Numbers, Vs StatsCan

Back to that survey, vs a larger sample set of data. ADP, the payroll behemoth, operates payrolls for over 2.2 million Canadians. This makes them a prime candidate for giving better job numbers. Which is great, because they do publish their own job reports – non-finance folks just don’t use them. In addition to having a larger data set, they also have hard data – not a survey based approach like StatsCan. These reports have a larger sample set, and less of a chance of human error. There’s a reason these numbers are more popular with people that understand the economy.

Source: Statistics Canada, ADP.

Canada Lost Over 7000 Jobs According To ADP

ADP’s job numbers are even worse for December. Canada lost 7,088 jobs, slightly more than StatsCan’s unadjusted numbers. This brings the total ADP estimate to 16,094,853 employed Canadians, about 13% lower than the StatCan number. We have a pretty good reason to believe that the numbers we base our policies on, are overly optimistic.

There’s nothing wrong with using seasonally adjusted numbers, they’re easier to use when reading long-term trends. When looking for month by month updates however, it doesn’t tell you what you’re looking for. If we lost 6,700 jobs in December, does it make us feel better if the seasonally adjusted number is 79,000 higher? Probably not. The fact of the matter is, less people were working in December. But don’t let that ruin a good ole fashioned, interest rate hike. It just means the economy is more vulnerable to the impact of that rise in interest rates.

Like this post? Like us on Facebook for the next one in your feed.



We encourage you to have a civil discussion. Note that reads "civil," which means don't act like jerks to each other. Still unclear? No name-calling, racism, or hate speech. Seriously, you're adults – act like it.

Any comments that violates these simple rules, will be removed promptly – along with your full comment history. Oh yeah, you'll also lose further commenting privileges. So if your comments disappear, it's not because the illuminati is screening you because they hate the truth, it's because you violated our simple rules.

  • Tristan 6 years ago

    Every Canadian should probably keep this point in mind: It’s not a bank economist’s job to give you an accurate read on the economy. Chief economists at banks are there to ensure the environment in which credit is being issued, is healthy.

    Healthy to them, means you’re willing to take out a mortgage. These mortgages are insured by tax payers, so they have the ability to be optimistic on these numbers. If they’re right, they make more money. If they’re wrong, they make more money. The only potential loser in this scenario is taxpayers.

    Harper’s government hid the bailout the banks received after the great recession. This made Canadians more confident in job numbers. However, we won’t be so lucky with a hidden bailout this time. Every single country/bank/fund is watching, to make sure that doesn’t happen.

    • Alistair McLaughlin 6 years ago

      The so-called “bail-out” was announced in Parliament, and approved by the all-party Finance Committee. It even had a fancy name – the Insured Mortgage Repurhcase Program (IMPP). The Department of Finance included regular updates on its website every time the CMHC made more money available to repurchase mortgages from the banks. Far from hidden, it was shouted from the roof-tops.

      • Alistair McLaughlin 6 years ago

        Insured Mortgage Purchase Program, not ‘Repurchase’ as I originally stated.

      • Marcus R. 6 years ago

        They weren’t exactly honest about it though. The “liquidity support” that three of the Big Five received was greater in value than that of the whole bank’s equity.

        I think a big problem is, 10 years ago, we only had a few news outlets, and we depended on them to understand how these things worked. Today we can look back, and see that liquidity support sounded like they would pick up any lending slack they needed, but in actuality it was an investment.

        Banks need to be responsible for their own risks. They engage in reckless lending behavior, because tax payers will pick up the bill for mistakes.

        • Alistair McLaughlin 6 years ago

          Maybe if you include central bank lending it was bigger. But we should not confuse central bank lending with government purchase of mortgages. $108 billion of mortgages were purchased. I’d be surprised if the market cap of the big 5 was less than that in 2009, but maybe I’m wrong. Their stocks did get pummelled.

      • Justin Thyme 6 years ago

        TD Canada Trust has more branches in America than it does in Canada. Mre of its business is international than domestic.

  • Michael 6 years ago

    Can I seasonally adjust my mortgage payments when I can’t make them? I’ll give the bank 18% less, and they can just round up.

    We give our government way too much credit, and think they’re much more reliable than they are. When China’s seasonal adjustments round these things out, it’s heck from Canadians about reliable governments. When our government adjusts this data to suit their narrative, we give ourselves a pat on the back for living in the best country in the world.

  • Trader Jim 6 years ago

    We had a giggle about the adjusted data as well. Most people don’t realize that research firms send out the unadjusted data for us. History won’t remember the loss of jobs in December. However, a trade desk will certainly the impact in the first quarter.

    I 100% agree that seasonally adjusted employment numbers are the wrong numbers the media focuses on. Seasonal adjustments just help Canadians not feel like the job market is as volatile as it is. Fact of the matter is, a lot of those jobs that are created in the summer, disappear in the fall. Adjusting that data across the year, doesn’t make a whole lot sense.

  • C 6 years ago

    Another Better Dwelling conspiracy. I think the government knows what they’re doing. If job numbers were off, they would have no reason to mis-report them. Sometimes a cigar is just a cigar.

  • EC 6 years ago

    My personal favourite is the inclusion of self-employment numbers in job numbers. Rising self-employment is typically not a hallmark of a healthy economy and they typically come with marginal incomes. A good chunk of jobs added to the economy in StatsCan reports over the last few years, including the last report, had been been in the self-employment category.

    • Ahmed 6 years ago

      If my waiter is a singer, dancer, and actor, does that count as four Canadian jobs? 🤔🤣

    • Jungle 6 years ago

      Good point. I agree self employment is not always regarded as strong, but I don’t know if it’s all bad either.

      I know an old friend who started self-employed building fences and decks. Surprisingly, he did very well and grew it into small business. Even adjusting for seasonal work, he was making above average income.

      Same with cleaning lady who starts self-employed, throws up a website, gets more jobs, etc.

      So not all bad and realize not everyone can work for fortune 500 corp.

      Anyway, does anyone know if the ADP numbers show self employment?

      • bluetheimpala 6 years ago

        ADP uses their payroll data as their sample set. As self employed individuals would not use a payroll service I think it is fair to assume ADP data does NOT include any self employed individuals.

        • Zhara 6 years ago

          Had to look it up to confirm. Looks like ADP uses the Stats Can Labour Force Survey, in addition to their own numbers to get a better view. So technically they do include self-employed individuals. Their data just pours some water on Stats Can’s numbers, which gives me even more confidence in their set vs Stats Can. (also, is it Stat Can or Stats Can?


          • Justin Thyme 6 years ago

            Two things from your link.


            After aggregating matched pairs by month, the data is then seasonally adjusted using the well known X-12 ARIMA method. This method is a filtering process that cleans the series of nonlinearities such as trading day effects, as well as detects and corrects for extreme observations. Statistics Canada’s methodology for payroll estimation also incorporates the use of the X-12
            method. Deseasonalized trends for the industrial cells are recalculated with each new month of data. To prepare the data for estimation, growth rates for employment are computed by calculating the matched employment growth in every industry.

            So ADP seasonally adjusts their data, then deseasonalizes it. Huh?

            Second, ADP extrapolates their data to the general population based on StatsCan ratios. However, they can only extrapolate from the sectors they gather data on, Since they do not sample self-employed, they cannot extrapolate this to the general population.

            That is, their extrapolation to the general population based on StatsCan segment ratios includes ONLY those segments ADP samples.

      • Dana 6 years ago

        Less credit available less offer for building decks, cleaning services, dinning out…. 😉 Just saying…

  • Jungle 6 years ago

    Excellent post, but remember there’s more context to the latest rate hike.

    The BOC is trying to get out of the hole and normalize rates, before household debt and housing explodes.

    We’re still in somewhat heavy stimulus mode, when it appears economy is doing fine.

    • Ben 6 years ago

      True. But this adds another layer to the terrible decision to increase rates. Macquarie, a big investment bank, pointed out that 30% of GDP growth was from auto and housing this week. 40% of nominal growth in GDP was based on rising home prices being leveraged. This is a disaster setup, and makes me think someone that knows what they’re talking about needs to take the wheel. Unfortunately, Canadians think government qualifications are a genetic thing.

      • MH 6 years ago

        So that’s what you see in these numbers? We need more of the same?

        To me it looks a lot like a ponzi scheme of economy and those do not end well no matter what you do. BOC may deserve their share of criticism but it has no good moves in the current situation, which is much bigger than your preferences for extremely low interest rates, and it is worth remembering that supporting the bubbles is not in their mandate.

        • David C. 6 years ago

          Wouldn’t this have been better addressed if the government built in mandatory changes to the GDS/TDS levels instead? The stress test across the board was a great idea, but it didn’t necessarily require a change to interest rates to accomplish this. No?

          • MH 6 years ago

            People tend to see the events through the lens of their personal needs. I can’t speak for BOC but I am rather sure that this hike has nothing to do with the housing bubble. I think that they would have gladly kept the rates steady but they simply had no choice thanks to the set of factors related to federal financing needs. Nothing personal.

        • bluetheimpala 6 years ago

          Here here!

      • EC 6 years ago

        BOC’s decision to increase rates was not terrible – it was the right decision to make. Assets are still inflating and they will continue to do so until interest rates start going up. Yes, there will be casualties, there always are in such situations, but we’re flirting with an even bigger disaster if BOC doesn’t start making use of the reset button. It’s been long overdue and that’s part of the problem.

      • Alistair McLaughlin 6 years ago

        You’re quite right about it being a disaster set-up. But that is an argument in favour of the increase, not against it.

        40% of GDP being supported by borrowing against rising home prices sounds like a fire that needs to be doused. With water, not with gasoline.

  • bluetheimpala 6 years ago

    Wouldn’t call the data junk; forecasts from China can be complete junk, we’re just selective in what we communicate to the public. And they’ve been smoothing these number for decades, ipso facto if we always look at adjusted numbers when designing policy, and not flip flop to build a case, is this really a problem? Many businesses use adjusted data.

  • Alistair McLaughlin 6 years ago

    I’m old enough to remember rates going up even in the face of rising unemployment and recessionary conditions. Sometimes rates go up because they need to go up, and sometimes those reasons exist even when jobs and the economy are suffering.

    In the early 80s it was because of inflation. In the late 80s and into 1990, it was again inflation. In 1993 to 1995, it was the fact that foreign bond buyers would not touch Canadian bonds except at a huge discount (or put another way, a huge yield premium) compared to US bonds. That was due to our excessive and persistent deficits and also political instability (Quebec referendum campaign).

    Seriously, if we think the BoC shouldn’t raise rates now, then they should never raise them again, because I’ve seen rates go up a lot more than this against the background of far weaker jobs and economic indicators. It was painful but necessary medicine, and we survived. The reward was price stability once growth returned; even housing was relatively stable for a time. That changed after 2003 or so once central banks lost their way and forgot the lessons of the past. But because the inflation went into the housing market, it didn’t show up in the CPI. So they just kept rates low, and we now sit atop a massive housing bubble and onerous household debt levels.

    A recent Better Dwelling column made the point that regular recessions are not only a good thing, but necessary. Up until the late 1990s, central banks believed the same, and behaved accordingly. (Nearly every recession endured since the 60s was intentionally engineered by central banks.) Sometimes recessions are preferable to overheating and the excesses that can build up. Maybe we need one now to force debt levels lower. But it won’t be a mild recession by any measure. That bus left long ago. The deleveraging will be brutal and grinding.

    • bluetheimpala 6 years ago

      Great comment. Nice to have some history and real-life account…if we forget our past we forgo out future.

  • CJ Ray 6 years ago

    Rates should not have been reduced that last 1/2% to begin with. I was never sure what they were thinking about by doing that. I guess that’s why I’m not head of BOC. It made some shyster richer and some regular folk overly exposed.

    • Jungle 6 years ago

      Looking back, do you think the boc regret doing those two cuts in 2015? The boc just shifted risk into something else.

  • Jungle 6 years ago

    If they started raising rates when the fed did we might have not seen such a blow off top for housing and debt. Seems they kept low for too long.

  • Justin Thyme 6 years ago

    First, ADP is much more biased than is StatsCan. It’s sample population is not adjusted in any way to map the demographics of Canada as a whole. The fact that there is such a huge discrepancy between a non-random ADP sample and the random StatsCan sample is certainly evidence of this. StatsCan will pick up scenarios that ADP does not cover, and these could easily be a big sub-population.

    ADP captures ALL of a two million population, but samples NONE of the other 16 million population.

    Secondly, I seriously doubt that Poloz made his decision based on this data. He might use it for the general public consumption but his decision is based more on the US market.

    The recent disasters have destroyed more than US$300 BILLION of structure. This has to be replaced, and quickly. The American manufacturing plant is no where close to being able to supply it. So NAFTA or not, high tariffs (which will e paid for by the American consumer) or not, there is tremendous export potential for Canada. And the disasters are still coming in. The California mudslides, for example. One study I saw clearly demonstrated that these disasters damaged or destroyed more housing that has to be replaced than the entire Canadian new housing market.

    These continuing US climate-change disasters are going to be a tremendous stimulus for the Canadian economy. NAFTA is just a side show, that will determine how high inflation runs in the US.

    • Zachary 6 years ago

      The first part is wrong. ADP uses their data to validate the Stats Can data. That is, they us the Labour Force Survey information, then check it against their system.

      The result is a much more complex validation of the random samples. It’s a far superior method, which is why it’s the primary method of employment data in the US. Canadians seem to be overly confident in information that comes from the government, and it’s very strange.

      We have little reason to believe they produce better information. You can see StatCan’s political bias, when they lead with adjusted numbers, which are next to useless. You can also see it with the foreign buyer data, which they misinterpreted in their report.

Comments are closed.