Saturday, 21 July 2018

Statistical trickery

Arguing the toss

Another substantial surge in jobs was reported for June, with total employment up by a seasonally adjusted 50,900 in the month.

This led to the usual quibbles about whether the figures are 'right' or not. 

This seldom happens when the number are below expectations.

But when there's a beat...every time!

There are even occasional calls to ditch the seasonally adjusted figures entirely in favour of the smoother trend data, but I'm not sure I buy that.

What are the trend figures, after all, but a representation of the same numbers from the survey? 

The further you disaggregate the figures the more prone they are to throwing out unusual results. 

For example, Macquarie highlighted an oddity in that the bulk of jobs growth over the past four months was accounted for by youth employment, especially from the female cohort.

The detailed labour force figures at the regional level are similarly prone to volatility, which is why the ABS reports annual average figures in addition to the original survey results. 

Labour market improving

For all of the grouching, the labour force has continued to tighten gradually over the past few years. 

For the monthly doubters, the seasonally adjusted unemployment rate has now fallen by more than one percent from 6.38 per cent in October 2014 to 5.37 per cent in June 2018.

If you prefer to use the trend rate, then that's plotted below for you too. 

The unemployment rate is now at the lowest level since 2012. 


Another piece of pessimistic statistical trickery is to highlight the underutilisation rate while neglecting to report the more telling volume measures of underutilisation.

In other words, yes lots of people want more hours of work, but the numbers of extra hours wanted has been declining over the past four calendar years too. 


This is good news, and it should be embraced as such!

Welcome to the recovery.