Roy Morgan: ABS job figures “defy belief”

From Roy Morgan this afternoon and presented without comment:

Yesterday’s ABS unemployment estimates (showing January unemployment dropping to 5.1%) defy belief — and common sense. The constant stream of companies announcing retrenchments in the early stages of 2012 are a clear sign that the ABS figures are diverging from reality.

Roy Morgan’s unemployment estimate (10.3%, up 1.7% in January — released on February 3, 2012) provides a more accurate picture of what is happening on the ground in the Australian economy — and more importantly — what is happening NOW!

The key differences are outlined in this note (reproduced below*) we include in each of our monthly Roy Morgan unemployment releases. Most specifically, the third and fourth paragraphs in, namely: The ABS only classifies an unemployed person as part of the labour force if, when surveyed, they have been actively looking for work in the four weeks up to the end of the reference week and if they were available for work in the reference week.

This creates a natural lag on the ABS estimates whereas the Roy Morgan estimates are taken at that point in time. If the respondent is out of work and looking for work — we classify them as unemployed.

The ABS also uses a seasonal adjustment process to determine their final figure. In January this would have a larger than usual downward effect. Roy Morgan estimates show unemployment in January rising 9/11 years going back to the turn of the century. So ‘on the ground’ more people are looking for work after New Year than before — hardly surprising. The ABS seasonal adjustment process no doubt takes this “static” (school leavers, recent uni grads etc.) out of the picture — and there is no doubt it brings the ABS number down each January.

As mentioned before though — one of the key differences between the ABS estimates and the Roy Morgan estimates is the immediacy of the figures. Roy Morgan estimates are immediate, whereas to show up as unemployed in the ABS figures you may have lost your job in November or December last year. The ABS unemployment estimates constitute a ‘lagging’ indicator.

The Roy Morgan Unemployment estimate is obtained by surveying an Australia-wide cross section with face-to-face interviews. An unemployed person is classified as part of the labour force if they are looking for work, no matter when. The results are not seasonally adjusted and provide an accurate measure of monthly unemployment estimates in Australia.

The Australian Bureau of Statistics Unemployment estimates are obtained by mostly telephone interviews. Households selected for the ABS Survey are interviewed each month for eight months, with one-eighth of the sample being replaced each month. The first interview is conducted face-to-face. Subsequent interviews are then conducted by telephone.

The ABS classifies an unemployed person as part of the labour force only if, when surveyed, they have been actively looking for work in the four weeks up to the end of the reference week and if they were available for work in the reference week.

The Australian Bureau of Statistics Unemployment estimates are also seasonally adjusted.

For these reasons the Australian Bureau of Statistics Unemployment estimates are different from the Roy Morgan Unemployment estimate.

Comments

  1. Well I know that an ‘independent’ survey with no Government association is the one I would (and do) trust.

    RM vs ABS is like the trust issue difference I would have between The Dalai Lama and Gillard 🙂

    • RM are alos motivated to clear their names here. I don’t think they implying the ABS is bent, just lagging. Which seems fair enough. To be honest I don’t think they did a great job of defending themselves. The issue is not seasonal adjustment, it’s the degree of divergence in the raw figures, as UE showed this morning.

      • HnH whilst I agree completely that the ABS data lags, I think that data interpretation and statistical analysis can show what you want it to show. As an example, remove a few outliers that shift the outcome in the preferred(?) direction.

        Having done quite a lot of statistical analysis as part of a PhD, I know I can conclusively prove to you, graphically, that “increasing global temperatures has led to the decline of Caribbean Pirates” 🙂

        It’s how you use the data

      • When I worked at the applied maths department of one of our tier 1 unis I had people from other departments coming to us with requests to find statistical methods that could prove their hypotheses.

      • And failing to follow protocol. All hypothesis and methodology, including which statistical tests that will be used, have to be declared before the experiment …

      • JPK, that they were from medical departments is very interesting and something I suspect Rumples would find intriguing.

      • Zentao I reckon many of reports we read now provide the preconceived view and frame the hypothesis and statistical structure around the desired outcome.

      • That makes perfect sense to me Sebastianbear – I know if I was a Caribbean pirate and things got too hot for my liking I’d pack it in and retire under the shade of a palm tree 🙂

    • I don’t really get this line of thinking. Given everyone’s complaining about “independent” firms like RP Data and APM’s property data, I assume that you see ABS house prices as the more objective?

      Why is RM more independent than these private firms and the ABS? Surely they make money through people/firms buying their survey results and services. One could make the case that they’re incentivised to produce headline-worthy results simply to sell their services. It’s just the same argument the RP Data are incentivised to spin or even falsify data to help sell it to RE agents.

      I actually think all those theories are conspiracy rubbish, but don’t see why RM is “independent”, even compared to the ABS, while all others are suspect. Maybe because their data is bearish?

  2. Wow, talk about a stir. I know people have questioned the ABS stats before, but their latest release has really gotten everyone scratching their heads with skepticism. Like RM have said (and I in a previous post), this could be chalked up to the ABS being a month or so behind.

    I’ll give it to next month and if there’s further divergence, I’m going to start assuming that someone is getting something wrong somewhere (not saying who just yet).

  3. I agree with your comment H&H. All my dealings with the ABS have been good and the people are helpful and do call me back so while I think there are holes in some of their methods I think they do the best they can.

    You only need to look at retail in any Westfield site in Melbourne to see closures and way less people about. If we all move to WA/QLD this will be sorted.

  4. I’ve been unemployed for 4 years and am sure they are yet to count me. I don’t get “Newstart” so I suppose that makes a difference. You only count if you are getting that benefit.

    • dumb_non_economist

      RodZone,
      That’s not correct. I was advised early this morning that my services are now surplus to requirements. I called up the ABS after reading the first blog on the ABS U/E figures to find out who is counted as unemployed. I was advised it is purely by survey, the fact I will not be “registered” with C/Link doesn’t matter.

      To Everyone Else,
      I doubt many here really believe that the ABS distort the figures. I feel the requirements to be measured as UE prevent a real indicative result of UE levels. Somewhat what I feel about the BS CPI figures. They have been massaged over the years to meaningless!

      • Thanks and commiserations. I now spend my time travelling and investing here and there. Pretty happy.

      • DNE, sorry to hear that. Good luck, I do hope the stats that show WA has the lowest unemployment rate are accurate. BTW BHP WAIO need around 600 at this very moment – don’t know your field.

  5. Of course the data is wrong.

    All the high profile stories of retrenchments are multiplied times over in the rest of the economy.

    Free trade alone has cost us countless real jobs.

    We might have more lawyers and economists but we sure have a lot less manufacturing and farming jobs.

  6. So we have two data sets showing conflicting results. Hardly a new concept.

    The key point for me is that I understand the ABS data to be consistent with the headline figures calculated elsewhere in the world. Therefore, it’s the ABS figures that we should compare to other economies on an ‘apples-apples’ basis.

    Having said that I accept the structural differences in the two methodologies, and I can make up my own mind as to the likely direction of travel from here, given all the hard and anecdotal evidence available.

    For me, this is just another case of certain constituents focusing on the present (current data prints) and not on what in aggregate the data implies about the future outlook.

    • Good points Ben, but with regards to your first point, while different stats showing conflicting results is common enough, there has been pretty good correlation between the RM and ABS stats over the past (accounting for the higher volatility in the RM stats). But in the last couple or so releases, we have seen a clear divergence emerge, much larger than what we have seen in the past which is naturally raising alot of questions. Again, it may be a bit of an abnormal blip and next month could see the gap close, but when compared to past correlations, this latest gap does stick out like a sore thumb.

      • Yeah that’s a good point. I don’t think it’s worth expending any time arguing which is ‘right’ or ‘wrong’, but you are right there is an unusual divergence at the moment, which presumably will be largely resolved in the coming months!

  7. Wasted OpportunitiesMEMBER

    I have an unrelated mystery for everyone…I had a slightly tin-foil reaction and I’m interested if anyone can shed light on it.

    A strange thing happened to our company today – someone set up a fake job ad on SEEK in our name. We got an email for a new SEEK account, saying welcome etc, with all the correct details for our company (name, phone number, address, email etc.), except it was addressed to a “Rachael” and we don’t have a Rachael.

    The ad was for an electrical engineering position in our environmental engineering company. Within 30 minutes we had several genuine applicants for this fictitious job opening.

    My feeling is that the only person to gain from this exercise would be someone interested in artificially boosting the SEEK new job ads index. We contacted them and the position was removed immediately.

    Is it possible someone is getting paid to create false positions? It seems crazy, but I can’t see the point otherwise.

    • Indeed it is possible someone is getting paid to create false positions for SEEK. They are probably living in the Philipenes and can be hired with the utmost of ease for as little as $3 an hour on this mind blowing site …

      http://www.odesk.com

      I discovered this today via another MB post, and out of curiosity posted a project job for statistical analysis paying $10 an hour for up to 5 hours work. I have had almost 100 responses since this morning.

      Hard to see how this trend is not going to cut a vast swathe through a lot of white collar professions in this country.

      • Some time back I posted a link to work by Dalia Marin (Prof Economics, U Munich, International Trade) wherein she forecast growing outsourcing/offshoring of white collar jobs primarily due to cost to emerging economies giving examples of routine legal work, insurance, much back-office, engineering/technological even in the future standard bureaucratic functions of government. Quite a read.

      • Given the ease with which I just hired a British educated, Kenyan residing actuary for five hours work at 1/10 the price of a Sydney based actuary, I don’t doubt many professions are facing down some serious competition for their overpaid salaries. (My own included).

      • If we outsource everything, what jobs will be left behind here?
        .
        IMHO, outsourcing is not a sustainable business model. Sooner or later, either the cost arbitrage will disappear or your local customers will have no job to pay for whatever product/service you have to offer.
        .
        US outsourced manufacturing to China. Now they have to borrow money from the Chinese in order to continue buying said Chinese products 🙂 .. No wonder, former champions of globalisation now squeal about “decoupling” and need for “increased Chinese consumption”.

        PS: I say all this, despite being a product of outsourcing myself 🙂

      • Mav.

        Pt 1.

        I don’t know. Survival of the fittest, someone always prepared to produce cheaper, and so on and on.

        Pt 2. It may in fact in sustainable indefinitely, but expensive for initial setup – see above.

        Pt 3. We have all loved and enjoyed the benefits of globalisation…until some sectors have been bitten… Generally, I would say an outright benefit. How to manage the negative, intelligently, the trouble. We all enjoyed it until…mmmmm let me think. The end of credit. Is globalisation founded on faulty ground?

        Pt 4. Mav, almost every argument here, pared back, is in a similar position. You recognise it. Kudos.

  8. Both data series are good. There is no ‘manipulation’ going on. There are no ‘conspiracy theories’ in play. They just happen to measure different things and, although they diverged sharply in January, both series tend to move together.

    Let’s wait for a few more months of data before jumping to any conclusions. It is likely that the series will line up again in that time – be it RM’s pulling back, ABS’ jumping up, or the most likely outcome in my opinion, something in between.

    • Well said Leith.

      I’d add that both surveys have a huge margin of error and bounce around a lot around the trend. Large revisions are common also, and when last months data is revised down (as it was for December) it can amplify the headline number.

      I don’t believe there’s any manipulation going on from either party, but the ABS is unlikely to issue a press release defending its data!

    • 2 Different methodologies, both with limited sample sizes – fluctuations occur, comparing them both has limited value. Best off comparing each one with itself over time and leaving it there.

      Throw in Confidence levels ( rule of thumb +- 5% ) and really there is nothing special about this statistically at all.

  9. Roy Morgan says:

    The constant stream of companies announcing retrenchments in the early stages of 2012 are a clear sign that the ABS figures are diverging from reality.

    According to Gittins! that makes Morgan a weak-minded fool:

    You’d need the steel-trap mind of an economist to say to yourself: ”These stories I’m hearing about layoffs here and there are sad news for the individuals involved, but they don’t really prove anything. To make a balanced assessment of what’s happening in the labour market I need aggregate statistics, not anecdotes – and the last stats I saw said that, overall, employment is growing sufficiently to hold the unemployment rate steady at 5.2 per cent.”

    Gittins: Now with a steel-trap mind of an economist!

    Perhaps we can call him “steel-trap” from now on?

    • Steel-trap mind.. with predictive powers of a goose and long-term memory capacity of a gold fish (in order to forget past statements/predictions)

    • Roy Morgan has made the classic mistake of saying “Carefully collected statistics do not match up with what newspapers publish to sell papers, therefore the carefully collected statistics are wrong.” It’s the equivalent of saying “Well I keep reading about crimes in the Daily Telegraph, so any statistics showing crime is going down are wrong.”

      These announced retrenchments are a drop in the ocean compared to the thousands upon thousands of people who quit jobs and start new jobs every week. Retrenchments only get reported in the paper if they’re done by major corporations in their hundreds or more. Meanwhile, firing by small and medium businesses never get reported. For that matter, neither do hirings.

  10. So how much does the decline in the participation rate explain the divergence?

    Roy Morgan would be sort of vindicated if there was in fact a trend of baby boomers retiring earlier than they had hoped to, and anecdotally I’ve heard of retrenched gen-x and gen-y going back to full-time study to re-skill for something that’s still in demand.

    Either/both of these factors would increase government welfare burden and lower productive output, even if we are officially in the low 5% of unemployment.

  11. If the ABS requires you to be unemployed for four weeks then the job losses from Jan-Feb won’t be reported in full until the ABS’ release in mid May.

    You’re gonna have to wait!

  12. I wonder if Roy Morgan have ever assessed the affect that modern communications preferences might be having on the reliability of their services. When they conduct telephone surveys, I assume they only contact landline phones. What happens to households which have opted for mobile only communications?

      • It is a good question but it sounds like it implies that ownership of a landline phone increases the chances of unemployment versus those who own a mobile but no landline.

      • Not intended that way, Lef-tee. More of a general question about all their surveys. My guess, if I was forced to make one, is that it would be likely to bias their unemployment numbers slightly the other way, if anything.

  13. There might be copyright issues with the display picture! 🙂
    If I am not mistaken, its a screen shot of a famous 70’s/80’s bollywood actor Shashi Kapoor. Anyways, it certainly doesn’t affect the unemployment data! haha

  14. Good point Alex.
    My sister uses Skype and her iPhone to talk to people. She doesnt own a landline.
    The wife and i have a landline but almost never use it.

  15. Both the ABS and RM found a spike in unemployment in January. However, the ABS recognises that unemployment levels can fluctuate due to seasonal factors, and so adjusts its data accordingly. Thus its seasonally adjusted series attempts to measure the actual underlying trend.

    RM chooses not to take seasonal variation into account. Why not?