Jobs and population

Posted by

The ABS’s mea culpa regarding population benchmarks has been picked up by the Age’s Colebatch.

The only thing I have to say about this is DO NOT trust the employment numbers. Our employment survey is a household survey, which is designed to measure the unemployment rate, and it is not especially good at anything else (using the gross flows data for ratios is only ‘okay’).

This is something that regular readers of our comments section would be across. See this post and the comments for a good discussion of the labour market report’s strengths and weaknesses.

One of the things of which i am most proud of is the high quality of comments on this blog. It is clear that our readership is thoughtful and well informed.

I learn from our comments – thanks folks. Keep it up.


  1. Hey ricardo
    Like I posted before, I’d hardly call it a mea culpa. The stuff pointed out by Colebatch and Martin is misleading and over exaggerated. The employment pattern remains the same even if you readjust to erp. Peter Martin does have one point though. If the ABS wants us to focus on the e to p ratio and not the level estimates then it should change the emphasis. Having said that I know what its like trying to get things changed in the Public Service.

  2. One other thing Ricardo, why do you think that the employment survey is only useful for unemployment rate? While I agree a household survey is not as good as a payroll number for jobs figures it does have some useful features, switch between ft and pt, the associated demographics that you don’t get from payroll numbers.

    1. Two good questions:

      1/ i take the ABS’s explanation of the population stuff as vindication of the view that prior errors were dominating the jobs numbers (i believed, as did others, that the slow jobs growth reflected the ABS forcing the population to ‘grow down’ to the new forecast. They are explaining this because it is a sufficiently hot topic for there to be reasonable uncertainty about the question

      2/ the survey produces ratios that are inflated by population assumptions. The ratios are reliable, but the level estimates are only as good as the population assumptions – which are two to three quarters ahead of the demographic estimates.

      Sent from my iPad

      1. Just curious why you think the unemployment rate is the only reliable indicator and not the employment to population ratio? I know this is a view shared by the RBA just not sure why the UR takes such prominence.

  3. Also I noticed this posted on another blog

    It has comparisons of the last time the abs rebenchmarked because of population divergence. I had a quick look at the employment estimate comparisons and to me it seemed pretty minor minor shifts but patterns remain the same. Based on this I’d assume the same thing will happen this time. So its really a bit over exaggeratted…or am missing something.

    1. I think you are right to play down the revisions. They are basically meaningless, as the jobs number itself is fairly meaningless (except that it moves financial markets, and wags chins).

      I really only made a post to re-iterate the point that the household survey should be used only for the ratios. This is something we discussed here on this blog quite a few times. It is something the RBA and Treasury are across, and no serious economist would tell you different.

      The ratios are well measured – they are then inflated using population assumptions. These assumptions are frequently wrong, and are often revised. This is ‘normal’.

      Happily, the unemployment rate is the best predictor of future inflation. The e/p ratio, and other ratios (i like the hiring and firing rates from the gross flows data) have their merits – but labour supplied is what holds down wages, so that is what most folks find best when they forecast inflation. And really, that is why we care so much about the unemployment rate.

Comments are closed.