Making sense of a surprising jobs spike
The labour force rollercoaster continued in August, with the largest monthly increase for employment in the history of the series. Despite an extensive investigation, the Australian Bureau of Statistics was unable to find a root cause or explanation for the remarkable surge in employment.
Last month analysts were left speechless after the unemployment rate spiked to 6.4 per cent. I flagged at the time that it may simply be a ‘rogue’ number, more noise than signal, and it might take a few months to get a good read on the labour market. Now it might take a few more months.
On a seasonally-adjusted basis, the unemployment rate fell to 6.1 per cent in August to be 0.3 percentage points higher over the year. The result beat market expectations for a more modest decline.
On the other side of the ledger, the participation rate surprised everyone by surging to 65.2 per cent in July -- its highest level since May 2012 -- to be 0.4 percentage points higher over the year. One potential reason for the recent surge in participation (as opposed to employment) is that it reflects renewed job searching efforts following the federal budget. The more likely reason, though, is simply excessive sample volatility.
The participation rate faces an uphill challenge to maintain these recent gains. Obviously the labour force data will correct at some point towards a more plausible number, and that likely means that the participation rate will be downgraded. On a longer-term basis, the participation rate is widely forecast to fall further on the back of widespread retirement among the baby boomers.
But it was the employment data that stole the show this month. Employment rose by a staggering 121,000 in August, its greatest monthly gain on record, following relatively modest growth over the past year. Remarkably, the total number of hours worked across the economy was unchanged during August.
Part-time employment rose by 106,700 in August (also a record) and it was hardly surprising that the ABS decided to investigate the outcome further. There have been some minor adjustments to its methodology in recent months, but based on its investigation the new methodology is unlikely to have caused such a surprising result.
Often these ‘rogue’ results are driven by sampling issues since the ABS relies on a monthly sample rotation that sees one-eighth of the total sample rotated out each month. When the outgoing and incoming samples have different characteristics there can often be volatile results. However, in this case the sample rotation actually lowered the estimates.
At this point, I’d normally spruik the more sensible trend estimates over the volatile seasonally-adjusted data. The trend remains your friend, but I should also offer a word of warning: the recent volatility is so great that it actually changes the interpretation of the trend data.
For example, the graph below shows the most recent data for August against what the trend would be if the data stopped in June. Trend estimates smooth through the monthly volatility but they also suffer from an ‘end-point’ problem because there is insufficient data to calculate a centred trend. Basically the ABS’ trend estimates place a disproportionate weight on the most recent (and in this case questionable) data.
The difference might not seem like a big deal but it fundamentally changes the interpretation of the estimates. Growth of almost 20,000 jobs per month (the August trend) is considered reasonably strong; by comparison, growth of 12,000 jobs per month (the June trend) points to a struggling economy.
Eventually the trend estimates will absorb the recent volatility but for the next few months the trend will be less reliable than usual.
Interpreting this month’s labour force data is obviously more difficult than usual. The noise-to-signal ratio is through the roof. How can we possibly square away the rise in employment against the estimates for hours worked? The only safe conclusion to reach is that the Australian economy did not create 121,000 extra jobs in August, but who knows what the real result was.
As for the Reserve Bank of Australia, it is firmly in wait-and-see mode. They won’t have to make any major decisions using this data. After the initial confusion and a few frustrated calls to the ABS, I expect that the RBA will simply write this month off and hope that the data returns to normal soon.