Economists have taken aim at the reliability of Australia’s jobs data

Photo: Robert Cianflone/Getty Images.

For all the time and money spent on producing it, nobody, and I mean nobody, trusts the reliability of the seasonally adjusted data in Australia’s monthly jobs report, released by the ABS.

That fact has been reinforced by the commentary from Australia’s economic community relating to the ABS’ latest offering, with the figures presented for September literally all over the place.

Before we hear about the distrust in the ABS’ figures from some of the most respected, and smart, financial minds in the country, here’s a recap of what the ABS dished up for September:

Full-time employment plunged by the most in over five years while part-time employment soared. Unemployment fell to a three-year low, thanks primarily to male labour market participation collapsing to the lowest level on record. Oh, and Tasmanian unemployment plunged 0.7 percentage points to 6.5%.

The list of anomalies could go on, and on, and on.

After trying to digest the data and make something of it, even Australia’s economic community — usually guarded about criticising one of their own — appears to have had enough, presenting a less-than-glowing assessment of the figures, some more politely that others.

Here’s a selection of the commentary that we’ve read. It really does offer a damning assessment of what is the single-most important data release in Australia each and every month. Our emphasis in bold.

John Peters, Commonwealth Bank

The latest data also revealed a curious result in relation to the participation rate. The participation rate (% of population employed or unemployed) has actually fallen by 0.4% over the past 2 months – one of the biggest declines on record. It sits at odds with the leading indicators of participation. Big moves are typically a sign of survey problems. The participation rate will fall if you have underestimated the level of employment. So the bottom line is to take the latest headline jobs prints with a grain of salt.

The mixed September labour force report with the huge size of some of the reported moves raises a degree of scepticism about the results. The ABS goes some way in addressing this scepticism by noting in its latest report that the new group added into the survey had a lower employment intensity than the group that rotated out. This would have biased down jobs growth. There is a problem in QLD in particular.

Stephen Walters, Australian Institute of Company Directors

Today’s employment report for September was a dog’s dinner. Total employment fell again (the second straight monthly decline), dragged down by another plunge in full-time job places. But, the jobless rate fell (after revisions) to a three-year low, although this was only because more people left the labour force. Hours worked across the entire workforce rose, but the underemployment rate fell. All up, with the main indicators all over the place in what we now know is a very volatile data set, it’s extremely difficult to draw firm conclusions from today’s release.

Paul Dales, Capital Economics

There are two reasons, though, why we can’t rely on these data. First, job growth is probably stronger than it looks as September’s data will have been dragged down by the end of the temporary contracts of those people employed by the Census in August. That may have reduced it by around 10,000.

Second, there is reason to believe that job growth is worse than September’s figure suggests. The ABS stated that it has altered the headline figure because the incoming rotation sample for Queensland was “considerably different to the rest of the Queensland sample”. Part of the survey sample changes every month, but it is unusual for the ABS to actively reduce the influence of the incoming sample. Somewhat unhelpfully, when we phoned the ABS it wouldn’t say which direction it tweaked the data! But since the state breakdown shows that employment in Queensland fell by 4,100, we’re guessing that fall was smaller than would otherwise have been the case.

Michael Turner, Royal Bank of Canada

For September’s report, the ABS warns that a new part of the survey to rotate in from Queensland was “noticeably different in its labour force characteristics to the group that it replaced”. The ABS has therefore “temporarily reduced the influence of this rotation group for September estimates”. In short, it appears the ABS is struggling to take survey responses at face value, and is almost ignoring some of them. This is obviously less than ideal, and underscores widely held concerns over the veracity of the data on a month-to-month basis.

Ivan Colhoun, National Australia Bank

Overall a weaker-than-expected report, but once again with a heightened degree of uncertainty about the exact signals of the various series given a number of very unusual – and large – monthly moves.

In spite of the veracity questions, the trends are not likely to allay the RBA’s fears about the momentum of the labour market.

Paul Bloxham, HSBC

Other labour market indicators have not shown this kind of deterioration; business and consumer confidence are holding up at solid levels and job advertisements have continued to rise. The official labour market survey has also faced difficulties with changing seasonal patterns during the past few years, which mean that the other labour market measures may be giving a better read of actual conditions.

Not exactly glowing, are they?

At it’s October monetary policy meeting, the RBA board noted that “there remained a degree of uncertainty about the momentum in the labour market”.

Uncertainty. Something that was delivered in spades yet again today.

Don’t expect this debate to go away anytime soon. Something needs to change in order to prevent the possibility of a costly, and damaging, policy misstep.

Jacqui Jones, Program Manager at the Labour and Income Branch of the Australian Bureau of Statistics, provided the following comment in response to a request from Business Insider for information on the decision to reduce the influence of the incoming Queensland rotation group:

Following recommendations from the Labour Force reviews in late 2014, the ABS has increased the range of detailed analysis of underlying survey dynamics. The September 2016 incoming rotation group in Queensland was the most significantly different rotation group since that time, and required the ABS to use a treatment within the established methods.

The ABS has published the best estimate for September 2016 and will not publish an alternative.

The ABS also notes that trend estimates provide the best measure of the underlying behaviour of the labour market, and these are not subject to the inherent sampling variability of seasonally adjusted data.

The push towards trend estimates is a relatively new development from the ABS, with the group previously reporting on the seasonally adjusted data in its monthly press releases.