ABS abandons seasonal adjustment for latest jobs data

By business reporter Michael Janda


The Australian Bureau of Statistics is taking the rare step of abandoning seasonal adjustment for its latest employment data.

The ABS uses seasonal adjustment, based on historical experience, to account for the normal variation between hiring and firing patterns between different months.

However, after a winter where the seasonally adjusted unemployment rate swung wildly from 6.1 to 6.4 and back to 6.1 per cent, the bureau released a statement saying it will not adjust the original figure for September for seasonal factors.

It will also reset the seasonal adjustment for July and August to one, meaning that these months will also reflect the original figures.

Excluding any other revisions, that change will take July's previously reported unemployment rate of 6.4 per cent back to 6 per cent, and August's 6.1 per cent jobless rate also down to 6.

Participation - the proportion of those aged over 15 in work or looking for it - will also have remained steady over winter at 64.7 per cent, rather than jumping between 64.8 and 65.2 per cent.

The biggest change will be to August's outsized jobs growth of 121,000, which will be revised back down to a more typical, but still strong, 32,100.

Market economists, and even the Reserve Bank, have been expressing doubts about the reliability of figures that showed huge job losses and gains over the past couple of months.

July's figures showed a jump in unemployment to a 12-year high of 6.4 per cent, despite an estimate of just 300 jobs being shed, while August reversed that rise with a record 121,000 jobs estimated to have been added.

The ABS says the seasonal pattern usually evident in July, August and September was not apparent this year.


Unemployment data explained

ABC business reporter Michael Janda explains how the jobs data are calculated and what to look for in the figures.


It is commissioning a review, with external input, to developing an appropriate method for seasonal adjustment in October's data and beyond.

"It is critical that the ABS produces the best set of estimates that it can so that discussion is on what the estimates mean, and not the estimates themselves," said acting Australian Statistician Jonathan Palmer in a statement.

The ABS has effectively repeated the old economist adage that "the trend is your friend", saying it will continue producing its trend estimates and encourages users to focus on this more stable data.

That is certainly what many market analysts will be doing, with RBC Capital Markets economist Michael Turner saying the unadjusted figures are also unreliable.

"In terms of gauging the health of the labour market, using unadjusted data is not ideal given there is still likely some seasonality present, just that the ABS is not sure how much," he wrote in a note on the announcement.

Well whoever it is, they seem to have attributed the same kind of accuracy to seasonal fluctuations in employment as the supposed climate gurus have to history.
Are you sure they haven't snuck a few of the redundant ones in there? Maybe Tim Flannery?

Now there's a bloke we could rely on for our economic forecasts.

Too much cutting of funds and staff numbers means the ABS surveys are not large enough so fluctuate and have a higher margin of error.

We need to fund the CSIRO, BuMet, ABS and the ABC properly. This govt is definitely anti–science so I expect defunding will continue.

The pseudo part of current climate scientists is being put in to practise

"

Redmond Neck wrote on Oct 9^th, 2014 at 10:06am:


Yes - he blamed Labor for not upgrading the ABS computer systems.... you mean they were broken?  You mean the old systems didn't work?  Does this mean the figures have always been wrong?

The mind boggles.... I used to do phone surveys for the ABS and I compiled my data by hand....... it is simply not that hard.

I used to do an overtime survey.... and an industrial disputes one.... multi-skilled....

That's what seasonal adjustment is for, to smooth out the figures and make the trend clearer.
You've got to pursue one line of statistics or the other, ie raw data or seasonally adjusted. A mish-mash of both is misleading.

The climatescientists are doing a fine job.

http://tiny.cc/dfmhnx