Survey of Employment, Payroll and Hours

This morning Statcan came out with their latest Survey of Employment, Payroll and Hours (SEPH for short). It's sort of a cousin to the Labour Force Survey (LFS), but doesn't get as much press. Among other things, the LFS has the much heralded unemployment numbers, and it also comes out mere days after the close of the month it measures, whereas the SEPH data comes out almost two months after the fact.

Unlike the LFS which is a survey of the employee, the SEPH is a survey of the employer. It includes a measure of average weekly earnings, and also of the number of people employed by corporate entities.

Particularly of late some have scoffed at the LFS unemployment figures as they believe its skewed by people who lost their jobs and are effectively unemployed, but instead list themselves as self-employed. So, lets do a comparison of the LFS and SEPH job loss numbers to get an idea of how much truth is behind those accusations.

Employment in both measures topped out last October, so this is a measure of how many jobs were lost since then (note, this makes no distinction between full-time and part-time, which in the LFS we've seen a large shift from the former to the latter over this period). As we can see, the numbers were fairly close through about February, and the SEPH losses were actually less through January... but since then we've seen a noticable decoupling, and as of September there have been roughly 40,000 more jobs lost in the SEPH then in the LFS.

We know that historically the SEPH and LFS employment figures largely track together (click for chart). This would seem to lend some credence to those questioning the unemployment figures... and just intuitively, massive recessions aren't the best time for a huge influx of people to go into business for themselves.

As an aside, judging from that big jump in LFS losses in October that could spell a pretty big jump due for the SEPH losses for that month, but we won't know that for sure until this time next month.

As mentioned earlier, the SEPH also includes a measure of average weekly earnings, so here that is. I also took the liberty of including the inflation adjusted figures for comparisons purposes, cause, well, that's just how I roll!

When adjusted for inflation, we notice that the line was fairly flat up until about 2003/2004 (other then a mysteriously sudden shift up in 1996, a change in methodology perhaps?). Since then we've seen some real growth in earnings, going from about $800, to as high as $975. That's roughly a 22% gain in a relatively short time, and particularly impressive compared to the twenty years leading up to it... conversely though, this is not as explosive of growth as some people have led us to believe.

We also notice that these two have started to track down. Unlike the number of jobs which peaked last October, earnings didn't top out until February ($965 nominal, $973 real). They've been kind of bouncing since, but generally downward since, bottoming out in August at $934 (nominal and real), but rebounding in September up to $959 (nominal and real).

Bearing in mind these numbers are subject to revision for about six months, and will change to a degree. Regardless, earnings wise we're slightly higher then a year ago, roughly 2% higher.

But that's where the problem is at looking purely at the earnings figure is that it makes no mention that while up 2%, there are over 5% fewer people earning it... and beyond that, it makes no mention that the population and labour force have continued growing over, which actually only further compounds the problem.

So, in my infinite boredom wisdom, I frankensteined my own little measure. This is the average weekly earnings multiplied by the payroll employees, then divided by the labour force size from the LFS (and multiplied by 52, to give you an annual figure, as to not be confused with the earlier weekly data).

I think this gives you a better idea of the gravity of the downturn, as it accounts for earnings, employment and the labour force... which when those items are isolated, make it hard to get an idea of the big picture. That said, this measure shouldn't be taken as gospel of anything as some included in the denominator are not included in the numerator, but assuming everything remains proportional (at least in the short term), this is good for comparison purposes.

By this measure (looking at inflation adjusted figures), it shows moderate growth in earnings through 2004, then of accelerated growth through late 2008, and then a sharp drop off... roughly 10% peak-to-trough (7% peak-to-current). We're back to about where we were in early 2007.

So, take it for what it's worth. Hopefully that gave you something to chew on for a Thursday. If you have any questions or comments, fire away.