Today’s ABS regional unemployment data contain bad news for the sluggish Far North economy. Loose Change observes (Ugly unemployment):

The unemployment rate in the Far North for February is back above 10% with the latest ABS regional numbers released today. The disappointing number and recent trend is likely to focus attention on the regional economy ahead of the upcoming elections.

I agree with this observation, but I think it’s worth waiting a few more months before concluding there is an adverse trend in the Far North labour market. First, in February 2011 the Far North unemployment rate was also in double digits at 13%, so I wonder if there are seasonal factors (e.g. school leavers) at play. Second, there is a large degree of sampling error at the regional level for the ABS labour force data. Over the last twelve months, the Far North unemployment rate has ranged from 6.4% to 10.8%, which suggests to me the data are very noisy. Consider the size of the labour force survey. The ABS notes:

The Labour Force Survey is based on a multi-stage area sample of private dwellings (currently approximately 29,000 houses, flats, etc.) and a list sample of non-private dwellings (hotels, motels, etc.), and covers approximately 0.33% of the civilian population of Australia aged 15 years and over.

That’s 29,000 households across Australia. If the Far North is home to around 1.25% of the Australian population, we’d expect the ABS surveys around 360 households in the Far North. This sample size is insufficient to generate robust sample statistics.

My estimate of the sampling error (standard error), using a simple formula (see margin of error Wikipedia entry), for the Far North unemployment rate (assuming it’s currently around 8%) is 1.4 percentage points. But, adopting the 95% confidence interval that statisticians generally require, this means the margin for error around the Far North unemployment rate is around +/- 3 percentage points. I’m now wondering if the ABS should bother reporting the regional labour force data.