Fairfax’s Peter Martin today has posted an interesting article explaining some of the reasons behind yesterday’s massive surge in employment, which saw employment rising by a seasonally-adjusted 71,500 jobs over the month – the biggest monthly gain in employment since July 2000.
According to Martin, much of this gain can be explained by changes in the Australian Bureau of Statistics (ABS) survey sample as well as changes to the ABS’ estimate of the working-age population. From The Age:
The Bureau of Statistics had confirmed earlier in the day that the figures it published for February showing the unusual jump in employment of 71,500 overstated the increase, most likely by a factor of two.
To understand why, it is necessary to examine the way the figures are created.
Each month the bureau surveys about 29,000 homes. One-eighth of the group, about 3600 homes, leave the survey each month and a new 3600 are ”rotated” in.
Rarely, usually only once every one or two years, employment conditions in the new homes are quite different to those in the old. When that happens, the official employment numbers jump (or fall) even if employment itself hasn’t changed.
The houses rotated in in February were extremely different to the houses rotated out. So different that the bureau believes the rotation itself was responsible for half of the reported 71,500 surge in employment, the ”best monthly job creation result in 13 years”.
Bureau staff explained this to government officials in a briefing on Wednesday morning.
Another unusual occurrence ”amplified” the error.
To convert its survey into answers for the entire population, the bureau multiplies the result by a number based on its guess of the population. Its best guess is that Australia’s working age population rose by more than usual at the start of this year as a greater than usual number of foreign students arrived.
It reckons this further exaggerated the already-exaggerated employment growth, perhaps by another 13 per cent.
It would prefer people to look at its estimate of what it calls the trend. This shows employment climbed at a sedate pace of 11,600 in February, much less than 71,500, and just enough to keep pace with population. The unemployment rate remained steady at 5.4 per cent.
Despite the statistical noise, it looks like the labour market has bottomed based on the various marginal indicators (e.g. the ANZ jobs series). The next big test will be when the mining investment boom unwinds and mining-related jobs begin to decline, and whether the pick-up in the non-mining economy can fill the jobs void.
