Benefit rate diverges from Unemployment rate

Posted by

My seasonally adjusted estimates suggest that the DEEWR benefit rate has been trending down, while the unemployment rate has been trending sideways.

We know from a speech by Dep Gov Battellino that the RBA was looking at this measure as a check on the rise in the unemployment rate in Q3.  I suspect that this will make the Bank more comfortable that the labour market is balanced.


    1. Check the three series on a chart – the RM measure is very volatile. It is the only data point we have for Jan just now, so it is new information, but i think it is more likely that the ABS rate remains flat in jan.

      Sent from my iPad

  1. Are you sure this divergence can’t be explained by other effects? There has been a fairly large decrease in the participation rate recently. Could that explain the difference between the unemployment rate and the benefit rate? Given that those who leave the workforce won’t be claiming benefits?

    The divergence is not all that large compared to some that have occurred previously.

    1. I should have trimmed the sample to focus on the more recent data – but yes there have been larger divergences, typically when the macroeconomy is swinging more sharply.

      The DEEWR data is interesting as it is a population – not a sample. However, benefit recipients may not report that they were looking for work in the sample period, and not everyone who is looking for work is eligible for benefits, so obviously the two can diverge for micro reasons.

      I suspect the participation / labour market data are distorted by changing holiday shopping patterns – so would guess that the part rate will correct.

      The macro signal in the benefit and unemployment rates should be similar – It will sure be interesting to see how this pans out.

      FWIW labour market data is looking better than the ABS data.

  2. @TheLorax, If you want ppl to take your comments seriously I wouldnt quote Roy Morgan. His numbers are a joke. The sample he uses is tiny (less than 1/10th of the ABS sample) and the methods he uses are questionable. The only reason he gets any attention is because journo’s get lazy from time to time and its easy to drum up a ‘story’ about how bad it really is and the official numbers don’t tell the real story. Much like the cost of living pressures ‘story’.

    @drpage, the drop in the part rate was a blip confined to a very spefic demographic (young women) that will most likely correct itself in time. I think its a bit of a wait an see moment. Various indicators have been suggesting softness for a while now but so far not too much has come through to the unemployment rate.

    1. The male participation rate has fallen by 1.1% over the past year (from 72.8% to 71.9%). Quite high by historical standards and on a par with drops during recessions in early and early 90s.

      Not surprising given that there has been no net job creation over the past year.

      1. I’d agree there has been a downward trend in part rate but hardly dramatic (i thought you were refering to the lastest print). Those are Seas. Adj. numbers your quoting, the trend figures say the same thing but its less dramatic with all the noise removed. Plus its not surprising the part rate has been trending down since the peak of late 2010 given the cooling of stimulus spending.

        1. Me too. I guess it is an industry thing – so focused on month to month innovations (which is dumb as they are more noise than signal). I expect some of it is demographic – i will have a look and get back.

          Sent from my iPad

  3. Just be careful on the participation rate. The ABS actually measures employment growth and the unemployment rate. The participation is merely the difference between the two. It is not measured.
    analysts look to see both employment and unemployment are sending the same signal.

    The ABS labour force survey is confidential so a person could say something in that and report something different in the other. Aslo beware the definition of unemployment! It is international however.

    Agree on Roy Morgan on ‘problems ‘ they have with their numbers.

    The main problem is the mining boom is yet to materialise as strong as many thought before such as RBA and Treasury. Fiscal policy , some might argue , is too tight!!

    1. Have to disagree on your statement on participation rate. It is measured. (Emp + unemp)% civ pop is part rate. Plus the warnings on unemployment is unnecessary as the definition has stayed essentially the same since the survey started ( barring a small chnage to the qualification period of willing and able to work)

      Also the mining boom is largely inconsequential to the Labour Market (directly anyway). The mining Industry employs relatively insignificant amounts of people directly, although an argument could be mounted on indirect employment but I doubt it would be significant, happy to be corrected though.

      1. My understanding is that the ABS forecast the population (the demographics survey lags the labour market report) and that the labour market report generates a bunch of ratios which are inflated by these population assumptions.

        The participation rate can get moved by survey rotation (sample moves to an old surburb, or one with lots of teenage kids), or over the medium term changes in the stratification of the population – as in last night’s US lmkt report, where participation fell as the census turned up more out of the labour force 55+ (they had assumed that they had died). So employment went up, unemployment fell, and the part rate also fell.

        The blunt fact is that none of these series (jobs, part rate) add information to an inflation forecast. They are relevant indicators of how good or bad your micro settings are, but they are not (generally) very interesting for monetary policy.

        For the purpose of monetary policy, it is the unemployment rate you want.

        Sent from my iPad

  4. I agree in terms of monetary policy, the Labour Market data is just one of many indicators. But I have to disagree with you about the UR being the indicator of choice…well I dont think it should be whether or not it is I don’t know. The reason being is that historically its fairly volatile…not surprising really given its only about 2% of the sample. Its really uncharateristic for it to be so stable. I’m guessing they take a broader view of the data and hoping that they understand the perils of putting too much weight on month to month changes.

    In terms of the population projections used in the figures yeah I agree there are problems with this but I’m guessing the noise it adds is no more than the varibilty around the monthly estimates. For example the standard errors around the monthly movements are if I remember around 25k for emp. So to be 95% certain the move is significant the move would have to be greater than 50 k. Half a percent either way on population growth is tiny compared to this.

    Also the chances that sample rotation would have any impact on the estimates is tiny. For example if a whole lot of teenagers are selected the weight assigned to them is reduced because there are more of them, thats how pop benchmarks work. Second the ABS now uses composite estimation which is supposed to lessen the impact of sudden sample changes due to sample rotation.

    1. My experience is that many indicators get revised to the unemployment rate – it turns out that it is pretty well measured. But you have to keep in mind that it is a survey that is designed to measure the national unemployment rate, and that the more you break it apart (by gender, age, region etc) or look at the subcomponents (jobs, part rate) the less accurate it is.

      As for monetary policy, i think that the RBA spends a lot of time trying to figure out what activity and capacity actually are – and for sure they look at a very broad range of data.

      Personally, i like factor models for this reason.

      Sent from my iPad

  5. The participation is a residual.They do not actually measure it.
    It is employment and unemployment that is measured. The RBA looks at both to see they are consistent.

    Agree on the rest

    1. Sorry, maybe Im arguing semantics here Or I dont really understand your argument but it aint a residual (ie left over), its the sum of two groups that are measured. How is that not a measure? If you were arguing that the NILF component was not measured I would agree as it is just the ppl that dont fit employment or unemp classifications. Yes maybe they dont directly measure part rate but they dont directly measure unemp rate either. All that is truely a direct measure are the direct counts i.e. simple counts of emp unemp and the leftover nilf, everything else is derived, i.e. pop benchmarks feed into estimates and they are estimates in and of themselves

  6. They do not ask people if they are still in the workforce. They ask them if they are working or looking for work. Hence we get figures for employment and unemployment.

    Once you get the first two numbers then you get the participation rate, hence it is a residual

  7. Sorry still missing your point. By your reasoning all the figures are residuals, i.e. you need population estimates and composite estimation regression values to calculate the weight and in turn the aggregated level estimates. Therefore by your reasoning all estimates are residuals.

    And as for your original point that the ABS directly measures the unemployment rate, well by your own definition the unemp rate is a residual as well. The unemp rate is the proportion of ppl that are classified as unemployed as a % of the labour force i.e unemployed % unemployed + employed. In your own words “Once you get the first two numbers you get the…” unemployment rate. So I really dont understand your point.

  8. sorry bob.

    We have a survey. the ABS measures those in employment and those that are not.

    It is from that the participation rate is derived.
    no the U rate is simply the unemployed ( which we already know) as a % of the population 9 which we already know). It isn’t a residual.

    The participation rate is not measured.

    1. I’m surprised at your misplaced confidence. You are wrong. Go check the definition from the 6103.0 concepts sources and methods…I’ll save you some time

      Calculation: Unemployment Rate = Unemployed persons/Labour Force

      Labour = employed + unemployed…

      Residual argument = bubkas

    2. Okay, i think we are at cross purposes. Let’s get some maths to clear things up.

      I understand, and think bob is saying, that the survey collects number of employed (N), unemployed but looking (U or unemployed), and those unemployed and not looking (NILF).

      From these data the ABS derives a bunch of ratios:

      Urate = U / (N+U)

      Prate = (N+U) / (N + U + NILF)

      Nrate = N / (N+ U + NILF)

      These ratios are then inflated by population estimates to get the so called ‘jobs numbers’.

      The survey unit is the physical establishment (house, hostel, hospital, etc), so the abs does not know the proportion of the population they are sampling, and it varies from month to month.

      In these terms, OYM, what is it you are claiming?

      Sent from my iPad

  9. Also while your checking on the definition, have read at how the estimates are actually calculated and you will realise that your assumptions on how thing are measured are wrong.

    I find Its always handy to know what your talking about…so do a bit of reading, 6102.0, 6103.0 are a good start and if there to heavy theres a Dummies guide on the 6202.0 website that puts it in plain English.

  10. Thank you for every other great post. Where else could anyone get that kind of information in such an ideal manner of writing? I have a presentation subsequent week, and I am on the search for such information.

Comments are closed.