Garry Shilson-Josling, AAP Economist
(Australian Associated Press)
Most economists are expecting a fall in employment in December, but that depends on an interpretation of recent volatility in the figures that may turn out to be wrong.
Forecasts by economists surveyed by AAP range from a fall of 25,000 to a rise of 15,000, after seasonal adjustment, with a median of a drop of 5,000.
But the fog of uncertainty shrouding the estimates due from the Australian Bureau of Statistics on Thursday is even denser than usual.
There are plenty of ways for “rogue” numbers to crop up, including survey sample variations, irregular or changing seasonal patterns and changes to the way the bureau collects the responses.
It’s not known for sure which of these factors, or something else, has affected the numbers over 2014 and 2015, but it is clear that the regular seasonal pattern has gone haywire.
So the bureau may be adjusting for a seasonal pattern that’s no longer there.
This year, the fall between July and August in the unadjusted estimate of employment was about what would have been expected.
But the subsequent rise in September, typically one of the strongest monthly gains in the year, was unaccountably weak.
Then, instead of the mild falls in October and November that would usually follow the big September bounce, the unadjusted estimate of employment headed strongly higher.
And the bureau, allowing for a seasonal pattern that had gone into hiding, reported big seasonally adjusted rises in employment in those two months.
It was, as a result, the biggest two-month seasonally adjusted rise in employment since 1988.
It didn’t really happen, of course.
The labour market was strong in 2015, but not that strong.
But that doesn’t help economists much.
The key question is whether, after all the perplexing monthly moves, the estimate of employment in November was “right”.
Most economists think not – which is why most expect a fall some describe as a correction after rises in employment in the October and November surveys were accounted for by only a small section of the survey sample.
But the argument that employment was definitely overestimated in November is not as much of a sure thing as it seems.
Skating around the dodgy-looking estimates for September and October, there was a 2.0 per cent rise (before seasonal adjustment) from August to November.
Rises of about that magnitude happen every four years or so, on average, so it’s hardly out of the ordinary.
In fact it looks fairly standard for a labour market enjoying reasonably strong jobs growth up to that point.
And if there’s nothing much wrong with the November number, then the strong jobs market means there should be a continuation of solid employment growth, something near or even some way beyond the upper end of the forecast range, rather than the correction many economists anticipate.
Either way, it will be an interesting set of numbers.