Centrelink payments trend upwards

Cross posted with permission from Mark the Graph:

DEEWR publishes monthly data on its payments. The April data was released Wednesday.

Based on this advice from Centrelink, it looks like two of these payment data series – Youth Allowance Other and New Start Total Recipients – can be added together and used as a rough proxy for the number of people who are unemployed (noting that the DEEWR payments data suffers from subtle and not so subtle series breaks as the eligibility requirements for payments change over time).


Collectively these series added together look like the following.

These unemployment payment figures compare with the original ABS count of the number of unemployed people as follows.

Applying a seasonal decomposition to the DEEWR data, we get the following trend and seasonally adjusted series.

Looking at the seasonally adjusted series since September 2011, the number of unemployed persons on DEEWR payments has been rising.

Interestingly, this is quite a different story from the last few months of the unemployed persons data from the ABS.

I am more inclined to favour the payments data over the labour force survey data as the best indicator on what is happening in the labour market.

The trend monthly and annual growth of people on unemployment payments is as follows.

Just for the fun of it, let’s look at the most recent bit of that data in detail … unemployment benefits payments are trending upwards.

Thank-you: to the DEEWR Income Support Information Section for providing me with the payment statistics back to late 2001.



  1. russellsmith55

    Looks like there is a clear cyclical pattern of people taking up benefits from November/December until March, then steady decline until the next November/December period. I’m sure there’s a well known reason for this? I know retailers tend to hire up on casuals a few months before Christmas then let them go after the rush has passed.

    Do the official ABS figures reflect this?

    • Norammly as people leave school in Nov/Dec they get payments then get jobs in the new year heading to March.

    • The ABS Original unemployed persons series (see chart #4) exhibits similar but more pronounced pattern of 12-month seasonal variability as the DEEWR payments data.

    • Most of the teachers I knew back in the day would collect benefits through their off-period.

      Made me laugh. They were on 50k (10yrs ago) and getting an extra couple of k of centrelink.

      There’s also the standard end of school/uni cycle for students. Layoffs tend to happen at this time. Most government departments love to lay casuals off on the 20thish of December.

      • What is an off-period for teachers? Are you suggesting they quit their jobs and instead of getting paid annual leave and accruing long service leave they opt to get welfare? That’s silly. Also, when you say ‘lay casuals off’ over the holiday period, do you mean to say there is less casual work when government departments and just about every other business largely close down and run on skeleton staff during the Christmas holiday period?

  2. Empire Investing

    I am certainly no chartist, but the trend since 2010 looks downwards to me, not upwards. The recent spike also seems to be part of a repeated, periodic variation. While I believe employment weakness exists, the data appears to be at odds with the post title.

    • I’ll have a hand at answering this.

      The medium-term trend is certainly down, but the point of the article was the last few months, which the trend is only just starting to reflect. FWIW, the trend is a long-term weighted average. I assume 13 months, taking the centre point, and 6 either side. The latest point doesn’t have 6 points to the right of it (as those values are in the future), so the latest are taking the current and the last 6 months, weighted towards the current value. Because of this, the trend line will be revised – not just next months, but the most recent 6 months, and they will keep getting revised until there are 6 more recent values. I would definitely expect it to point up soon.

      Second question is regarding the periodic variation. This is another way of saying “seasonality”, and has been removed from the 5th chart onwards. So the bottom charts, which show a clear uptick, are seasonally adjusted. In other words, no this is not due to the repeated periodic variation.

      • Empire Investing

        Thanks Nathan, although I am Dave BD regarding the attempt to forecast a longer term trend over such a short time frame (6 months). I guess the other question is what impact population growth would have on the numbers, or has this already been accounted for?

  3. reusachtigeMEMBER

    Now, if only the Disability Pension was included – that’s where the real unemployment is hiding!

    • Jumping jack flash


      I would love to see a breakdown, and trends of DSP recipients since 1996 by disability category.

      • reusachtigeMEMBER

        All long-term welfies strive for a disability classification and it’s not too hard to get one – just act mental and have a limp – double whammy! It’s a win/win, ie, the welfie gets a steady income, albeit tiny but larger than newstart, and they don’t have to do anything or apply for anything or clean parks and the government wins because it gets them off the unemployment stats.

        Too easy!!

    • I would love to look at the number of recipients of the Disability Support Pension month-by-month. If anyone knows of a public data source, please let me know.

    • I’ve met people with no disability happy to be on DSP. Reckon I would be too if I could be bothered.

  4. So the total unemployed is trending DOWN from 2008… you can’t make a trend from 6 months data when there is that much noise (‘seasonal variation’) in the data.

    It would be more useful to show this as a percentage (to include population changes). As population has increased since 2008, the downwards trend of % unemployment should be even more obvious.

    Time to stop letting anecdotal evidence and the wealth effect from tempering your judgement. The data is right there for you!


    • “you can’t make a trend from 6 months data when there is that much noise (‘seasonal variation’) in the data.”

      Correct, which is why he has done it on the seasonally adjusted series.

  5. Excellent post, and I think this looks like it sits roughly in the middle of the Roy Morgan and ABS unemployment stats, but with a hint that things are starting to get bad. I would expect, based on this, that either the participation rate will continue to fall and UE rate will stay low, or UE rate will start to rise.

    • “…I think this looks like it sits roughly in the middle of the Roy Morgan and ABS unemployment stats…”


      Have another look at the fourth and eighth charts. They show a close correspondence between the ABS and DEEWR data. Roy Morgan’s data, otoh, is lost in space somewhere.

      • yes, I’m looking at the 8th chart. The timescale is pretty long, so just look at the last little bit. The last 6 months on that chart shows a clear divergence with ABS heading down, and DEEWR levelling off, with the suggestion in this post that it is going to head up soon. Charts 6 and 7 drill into that – they look like mirrors!

        • Nathan,

          My main point was that these two independently derived measures correspond fairly closely, but are at complete odds with the Roy Morgan survey. For March, the ABS estimate is 630k unemployed and the DEEWR data indicate 665k unemployed. Not the same, but close.

          OTOH, Roy Morgan reckons its 1.12m, almost double!!

          Furthermore, in the year to March the ABS data show that unemployment has increased by 6%, the DEEWR data show a small decrease of 0.6%. Roy Morgan reckons unemployment has jumped up 30%!

          Regarding the recent divergence between the ABS and DEEWR data. Unless I am mistaken, there exists a qualifying period before you can receive Newstart/Youth Allowance benefits. Thus the DEEWR data will always lag the ABS data.

          This is particularly noticeable if you look at what happened in 2009 on chart 8.

    • The DEEWR data has three categories of Youth Allowance: (1) Full-time students; (2) Apprenticeships and (3) Other. I have only used the other category. But I accept even this is a bit speculative, but probably better than not including it.

      • My understanding is that (1) is the old Austudy, though I could be wrong. Nonetheless, if you include ‘(3) Other’ and exclude the other two that should be fine.

  6. Just speculating, but in the last few months I have heard several anecdotal reports of rents in Sydney being hiked by 20% to 40%. Social security payments are partly based on how much rent you are paying, and so if this trend is widespread then that might be a big part of the reason why the unemployment numbers and unemployment payments are moving in different directions.

    As to why rents have gone up so much, I suspect quite a few property investors have realized that they might not get the capital gains that they were hoping for, and are trying to get their rental properties to pay for themselves.

    Not quite sure what the consequences of a widespread major increase in rents would be, but it is an interesting question to think about…

    • Also wouldnt cost of living increases also be included, doesnt the dole figures get increased each year by the rate of inflation?

    • The thing about rents is, landlords can put their prices up as much as they like, if other people in the area keep theirs the same then nobody will take up the lease.

      Not sure what it’s like in Sydney but I can do a search on any real estate website (domain.com realestate.com etc) and get over 100 rental properties that fit my search criteria for location, size and maximum rent per week. Even if half of those doubled their rent it wouldn’t affect me.