When statisticians fall out – employment data becomes a battleground

Former Chief Statistician Bill McLennan, has again weighed into the debate about Australia’s revised employment statistics and their accuracy. McLennan has continued his criticism of the Australian Bureau of Statistics (ABS), the organisation he headed from 1995 to 2000 (from 1992 to 1995 he was also the head of the United Kingdom’s Central Statistical Office) suggesting the revised seasonal employment data is inaccurate and misleading.

Because of a major revision to the methodology for calculating the unemployment rate, the ABS’ data has become unreliable and may not return to normality for perhaps two years. This is critically the case for the seasonal analysis of the data, which realistically can only be determined after there is sufficient data to confirm what is seasonal and what is a one-off variation.

McLennan is difficult to challenge on these points and he is joined by many other economists in his criticism of the seasonal employment data.

For its part, the ABS maintains a fairly dignified stance, seeking to explain its approach is to improve the series and re-establish its integrity. It concedes some of these points, but also says that the nation needs the data, at the highest level of integrity it can maintain.

After all the argument is done, Callam Pickering (Business Spectator) has the answer anyway.  He wrote:

"Readers would be wise to ignore those [seasonal] estimates and place greater emphasis on the less volatile trend figures. The trend may be based on some dodgy data, but by smoothing the figures we can hopefully identify some meaningful trends.” AFR, 15th October, 2015

Pickering makes a simple and valid point. If the seasonal measurements of employment are not working, refer to the trend measures to at least work out the direction employment is headed. For an update – using employment trend data – see the next item in Statistics Count.

Perhaps the real issue here is not the situation involving the employment data, but rather the decision making in the ABS. If a data series is not working adequately, or has been changed and does not have a sufficient time series on which to be confident, perhaps it should not be placed on the shelf for two or more years?

Employment data is important and sensitive. Policy makers use it to set and adjust policy. Businesses use it as a forward predictor of future demand growth. Markets use it, among other tools, to settle equities prices. In the light of that sensitivity, perhaps the ABS would be better placed conceding the absence of confidence and relying on the data with greatest reliability.

Wheeling out the former heads of organisations or departments to confirm how terrible they are now is a time-honoured tradition that appears to have no end of willing participants. It seems the challenges of the modern day statistician are not enough to keep the predecessor quietly being thankful its not them in the chair.

The modern ABS is a struggling organization which costs money to operate every year, despite its fee for service and semi-commercial focus. Some view this as inappropriate, but where would we be if the nation had no general access to information to help it make decisions?

Every business leader knows, you cannot manage what you cannot measure. The aphorism applies to countries, as much as it does companies.

It is to be hoped that the ABS gets its seasonal employment data series back on track. We need the data, not the argument.