The truth about unemployment in Australia

A spurious analysis of labour force data may lead many analysts to conclude that the Australian economy is improving, but a closer look will reveal quite a different conclusion.

Usually, a fall in the unemployment rate indicates that the economy is on the mend. However, a quick look at the details leads to a different interpretation -- including the realisation that the Australian unemployment rate didn’t fall at all.

On a seasonally-adjusted basis, the unemployment rate declined to 5.8 per cent in March, beating market expectations, which shot the Australian dollar to its highest level since November. But had the market focused on the correct data, they would have reached a very different conclusion.

Month after month, the market incorrectly assesses the labour market because they focus on the wrong data.

The Australian Bureau of Statistics makes it very clear that the trend data is more reliable than the seasonally-adjusted estimates. It recommends that analysts focus on the trend, but every month it is largely ignored.

Using the seasonally-adjusted data, the ABS is 95 per cent confident that employment growth was somewhere between a 39,000-person fall and a 76,000-person rise during March.

They are also 95 per cent confident that the unemployment rate was somewhere between 5.2 and 6 per cent. Wide confidence intervals are a common feature of seasonally-adjusted data, which is precisely why the ABS recommends a greater focus on trend estimates.

Last month the ABS was at pains to remind people that the strong growth in February was due to a change in sample.

With the exception of myself and a few others, its warning was largely ignored (Why the jobs figures don’t add up, March 13). It was universally reported that employment rose by around 47,000 in February and almost universally ignored that 37 per cent of that growth was completely imaginary and a result of a change in sampling methodology by the ABS.

Fast forward a month and analysts are misinterpreting today’s data as well.

The unemployment rate did fall by 0.2 percentage points in March but so did the participation rate. Most of the improvement to the labour market was driven by people either giving up on finding a job or retiring.

Graph for The truth about unemployment in Australia

By comparison, the trend data is providing a very clear signal that the labour market has yet to pick up. On a trend basis, the unemployment rate was largely unchanged at 6 per cent in February and the participation rate remained at 64.7 per cent.

Employment was up by 14,200, which is evidently enough to meet population growth, but not a sign of strength. Full-time employment rose by 10,900 and part-time employment rose by 3,300.

Graph for The truth about unemployment in Australia

There is one caveat I will make about the trend estimates for employment: the sample change in February was so large that it will throw some trend estimates out for a few months.

I’d wager that full-time employment growth is a little weaker than the graph above shows, while part-time employment is a little stronger. That would be consistent with the trend prior to the sampling change.

Many will declare that labour market growth in February and March is a clear indication that the economy is back on track, and that the RBA should consider raising rates.

But the moment you consider the underlying trend it becomes obvious that this would be premature. I know only too well the emphasis that the RBA places on the labour market trend and it won’t be so easily fooled.

The labour force survey is, in my opinion, the most important Australian data release. At the very least it is the release that is most important to the Australian people. Analysts have an obligation to assess the data correctly and too often they fail to do so. I don’t know whether it is simply laziness or ignorance or some unfortunate combination of the two, but it is mindboggling how often economic data is misconstrued in Australia.