Declining jobs trend to produce some surprisingly big falls
By Garry Shilson-Josling, AAP Economist04 May 2009 5:28 PM
SYDNEY, May 4 AAP - The trend in employment will turn downward with "rogue" figures showing very steep monthly falls will grow more frequent.
At last measure in March the trend measure of the number of people with jobs estimated by the Australian Bureau of Statistics (ABS) was falling at 5,000 per month.
If that kept up for a full year it would mean an annual rate of decline of about 60,000 or 0.6 per cent.
But with gross domestic product (GDP) contracting, that downward slope will steepen.
Even if GDP flatlines over the coming year, without contracting further, it would be more than three percentage points slower than the long term average growth rate.
That in turn would imply employment growth somewhere around the same margin below its long term average of around 1.7 per cent per annum.
In other words, it points to an annual rate of decline in the number employed in the order of 1.3 per cent or 140,000, meaning a downward slope of 12,000 per month.
That's assuming the economy stops contacting, which is almost certainly too much to hope for over the short term.
The reality is likely to be considerably worse.
But even if employment only trends down at that optimistically slow rate, no-one should expect that, or something close to it, every single month.
The seasonally adjusted fall of 34,700 in March was testimony to that.
The monthly labour force survey by the ABS is subject to random variation in the sample of the population chosen to survey each month, as well as variation in the characteristics of households within that sample which respond to the survey.
As a result the ABS warns there is a confidence interval of 30,300 either side of its estimates of the movement in total employment.
Taking that into account, there was only a two in three chance that the fall in March, reported as 34,700, was actually between a fall of 4,400 and a fall of 65,000.
There was a one in six chance it was even worse and a one in six chance it was better.
On top of that sampling-related uncertainty there are other sources of volatility in the figures.
For starters, the seasonal adjustment process is imperfect, because not all seasonality is regular and stable though time.
And even if these issues could be smoothed away, it would still be a fact that economic activity does not proceed smoothly.
Sometimes the ups and downs from month to month are simply not in line with an underlying trend.
Like it or not, volatility happens.
Even with a downward trend of 12,000 a month, the occasional result showing a seasonally adjusted fall of well over 50,000 is not only possible, but highly likely.
The trick will be to resist over-reacting to these sporadic "rogue" numbers, and to be just as resistant to complacency when the figures turn out better than expected.
The ABS is due to release the April unemployment figures on Thursday.
Economist expect the unemployment rate to be about 5.9 per cent, which would be the highest since July 2003, up from 5.7 per cent in May.
For employment, forecasts are centred on a fall of 25,000, which would bring employment to a 13-month low.
But only the most naive economist would be surprised if their forecast turned out to be very wide of the mark.
