What's behind those jobs numbers?

Yesterday's labour market figures, and particularly the employment outcome and the participation rate, came as an unwelcome surprise to everyone. The consensus expectation amongst economic forecasters had been that employment would rise by 0.4%, whereas it actually fell by 0.4%, and the participation rate (the proportion of the population in the labour force), which had been expected to hold up at its historically high 69.3%, dropped back to 68.6%. Given that a falling participation rate is usually taken as a sign of a weaker labour market, since people tend to leave the labour force when they become less confident that jobs are available,  the employment and participation numbers taken together showed an unexpectedly soft jobs market in the September quarter.

On the other hand, though, there were some oddities in the numbers. Employment certainly went down, but the number of filled jobs, and the total number of hours worked, both rose during the quarter, and those increases would tend to suggest that the labour market was a bit better than previously.

The different outcomes got me wondering about how these three measures - employment, filled jobs, hours worked - have behaved over time. Here's the answer, since the start of 2000.

Generally they move together, as you'd expect. There is the odd occasion, as in this latest September quarter, when employment falls but the number of jobs and the hours worked rise (March 2000, September 2006, September 2012). There is even the odd occasion where the opposite happens - employment rises but jobs and hours fall (June 2005, March 2008, June 2009, June 2015). But as a rule they tell the same story, and quarters like this June (employment up, the others down) and this September (employment down, the others up) are the exceptions.

What's happened, I reckon, is that there was clearly some general slowing of the economy earlier this year, with an impact on the labour market, but  we're also seeing the impact of quite a lot of noise in the data. Over the long run (back to early 1989, when the jobs and hours series start), all three measures have almost exactly the same average growth rate: employment, jobs and hours have each grown, on average over the long haul, by 0.4% a quarter. But there's a lot of volatility in the three numbers: if you look at the standard deviation of each one, for employment it is 0.6%, for jobs 0.8%, and for hours worked it's 1.0%. As a rough rule of thumb, even if the underlying 0.4% hasn't changed at all, two thirds of the time you're going to see employment numbers between -0.2% and +1.0%, jobs numbers between -0.4% and +1.2%, and hours numbers between -0.6% and +1.4%.

The overall lesson is that you're probably best advised not to get fixated on any one of the three measures: employment is somewhat less volatile than the others, and to that extent it's more of a reliable pointer than the other two, but they're best considered (a) in the round and (b) on timeframes longer than a single, possibly unrepresentative, quarter.

If, for example, you look at 2014 as a whole, the average quarterly increase in employment was +0.89%, the average increase in filled jobs was +0.61% and the average increase in hours was +0.74%. In the first three quarters of this year, the same averages were +0.12% (employment), +0.56% (jobs) and +0.77% (hours). So you'd conclude, overall, that there has been some modest slowdown: employment on its own would point towards a reasonable slowdown, but the other two point to little or none. That's not at all surprising, given the effect that falling dairy prices were having at the time. A modest slowdown, but still ongoing growth in employment, is also exactly what you see in the employment component of the ANZ's business survey, so it all fits together nicely.