ABS: employment figures can’t be trusted

By Leith van Onselen

In September last year, the Australian Bureau of Statistics (ABS) released nonsensical employment numbers for August, which claimed that a record 121,000 jobs were created in a single month but contradictorily no change in the total number of hours worked. These results lead to wide spread ridicule and condemnation from analysts and commentators, prompting the ABS to undertake a detailed review and adjustments of its methodology, in particular in the way that it treats seasonality.

Today, Fairfax’s Peter Martin has reported that yesterday’s labour force figures for January, which posted a sharp seasonally adjusted increase in unemployment to 6.4%, also cannot be trusted, with new head of the ABS, David Kalisch, urging analysts to not rely on the headline figure:

David Kalisch says the media and markets should focus instead on what the bureau calls its 95 per cent confidence interval, reporting that the bureau is confident the true rate is between 6 per cent and 6.8 per cent.

It would mean reporting that the total number of Australians in jobs did something between sliding 72,200 in January and climbing 45,800. It would mean reporting that the number of Australians unemployed did something between climbing 74,900 and falling 5900. It would mean acknowledging that the bureau’s employment estimates are even less accurate than is widely believed.

…people need to recognise that it is just that – a survey, of about 26,000 households conducted once a month. “Any sense that the number is exactly whatever we report to the second decimal point is not an accurate use of those numbers,” he says…

Kalisch also notes that the ABS has been crippled from widespread budget cuts, and that the computer system used to create its labour force data is 30 years old and in desperate need of upgrading (at a cost of hundreds of millions of dollars).

As I noted last year when commenting on September’s labour force data debacle, the reduced reliability of the ABS employment statistics is hardly surprising. The organisation was forced to cut its expenditure by $50 million over three years, as part of the Federal Budget, which led to 100 staff being cut from the ABS. This followed a $20 million reduction in funding by the former Labor Government, which also saw the labor force survey sample size reduced.

So effectively, the ABS has been asked to perform its functions using: a reduced sample, less staff, and an antiquated computer system. No wonder accuracy has suffered.

A broader concern is that timely and accurate data are vital to good decision making, and the cuts to the ABS’ workforce and capability risks hampering the Government’s ability to formulate policy, the RBA’s ability to accurately read the economy and formulate monetary policy, as well as the public’s decision making and ability to evaluate policy.

In short, cutting the Bureau’s funding and capability will provide negligible cost savings to the Budget and potentially impose costs on the Australian economy over the longer-term via incorrect policy responses. Becoming less informed during what is likely to be the greatest structural adjustment affecting the Australian economy since the early-1990s recession is clearly a retrograde move.

In any even, when it comes to the ABS’ labour force data, it is best to ignore the headline seasonally adjusted figures altogether, and instead focus on the more reliable trend estimates (see below charts). These show that unemployment has been trending-up slowly and aggregate hours worked trending down, which seems appropriate given the broader state of the economy.

ScreenHunter_6055 Feb. 13 09.10
ScreenHunter_6056 Feb. 13 09.12

[email protected]


  1. It has been 12 months since I’ve been able to find the underutilisation number….even then it was well-hidden,
    but while ever it was there it made them appear honest.

    • KeenEyeKenMEMBER

      ABS 6202, Table 22
      Column DE

      Released only quarterly, the Nov-14 figure of 15% underutilisation (underemployed + unemployed) is the highest reading in 17 years – and steeply increasing.

      The labour market is cactus.

      • Thank you Ken.

        No wonder they try so hard to hide the truth….15% is a scary number, and that was before the school leavers are counted.

      • But how much of the under-utilisation rate is due to technology?

        How many (in particular youth) are out of a job because of self check outs at Supermarkets? My first job was working in a call centre when I was 18, and I highly doubt there would be that many call centre jobs even still in Melbourne!.

        Tech eliminating, and offshoring the remaining entry level jobs is the real issue here. Unfortunately people making coffees for each other looks like “a job” in the employment numbers.

      • KeenEyeKenMEMBER

        No one’s trying to hide these figures. People just dont know where to look. Most people wouldnt go past the SMH/Ninemsn website to even bother looking – hardly the fault of the ABS. Also, school leavers are already implicitly included in these figures. The Feburary release (in 4 weeks time) will see the post-xmas surge in youth unemployment, but this wont have too much of an impact on the seasonally adjusted figuers i gave mentioned earlier (which were acutally higher in Nov-14 than the 14.6% in unadjusted/original terms). The ABS is not in the business of pulling the wool over anyones eyes (never has been, never will be).
        Btw, those unadjusted/original figures are in column FG of that release.

        The figuers dont attempt to explain *why* these people are under/unemployed, just the fact that they are so. Youth unemployment is an increasingly worrying problem, and you’re probably correct in pointing towards technology as an issue – with most menial task in the firing line of automation and self-correcting algorithms. Btw, back off on the coffee makers 🙂 my barrista is an integral part of my existence (and they probably get paid more than i do. I’m putting their kids through college)

  2. mine-otour in a china shop

    Whilst I accept there is an immediate need to replace 40 year old systems, I think that statistical resources need to be allocated in a more efficient manner.

    Sure Census and social statistic developments are very important but they are coming at the expense of investment in critical macroeconomic statistics.

    We need to prioritise our data collections better – the economic cost of a policy mistake from poor data gathering will far outweigh any benefits from new social and census statistic investments.

    Across the world in Statistical offices far too much resources are being placed on 5 yearly censuses using old paper techniques. Is a full census every 5 years necessary, and can we move to a 100% electronic platform? That might make room for more investment in macro data collection.

    • Census data is used for redrawing electorate boundaries, so changing the method is a political minefield.

      • Indeed the way that politics prevents the best statistical solution to collect census data from being adopted in the US, is the topic of the linked blog post.

      • mine-otour in a china shop

        ok so just count people, age and other necessary variables in a smaller 5 year census and do the other 90% in the larger 10 year census….

      • Seems fair if the ‘every head’ count has little to no value to researchers – should be completed with the least cost possible. But I suspect that there are strong political barriers to acknowledging its flaws so directly…

    • How about we get a $100 rebate on our tax returns each year if we complete the census annually online?

      The government should have most of the data already in the modern age (between health records, medicare, tax office records).

  3. How on earth could it cost hundreds of millions to replace a computer system for an organisation of 3000 people ?

      • De-fund government, sell off assets, and when it does not function anymore….. point finger and say see I told you so…

        Skippy…. you gotta love the tombs which peddle self fulfilling prophecy.

    • May sound straight forward, but for a 30 year old computer system which is standalone, we are probably talking about ripping out and replacing antiquated network hardware, reverse engineering a database and system for which there would be limited expertise today, rebuilding a new system from scratch, etc… the number of employees is probably not relevant to the task at hand (or at least not the main contributor of costs).

      Hundreds of millions still sounds like a lot, but I don’t think you should play down the work involved without knowing exactly what it would entail.

      • Which is compounded by the necessity of running both parallel until the bugs are worked out. Shades of the Qld health problem.

        Don’t tell them about the Frankenstein systems the big banks and exchanges still operate,

      • From the article Kalisch is keenly aware of the banks’ Frankensteins: he wants to tap people with experience in those systems to assist the ABS IT makeover, probably with a view to avoiding repeating past gov’t agency snafus.

      • This is just a survey of 26,000 people.

        SurveyMonkey, email database, follow-up phone calls if necessary. Alternatively build a simple yes/no app.

        $200m is an outrage for a survey that provides 95% confidence between 6 & 6.8%.

    • It’ll cost that much if they’re buying it from an outside vendor and requires a lot of customization and modification. It’ll cost 10 times less if it’s developed in-house, however hiring more employees is simply not possible under the current political climate.

    • I agree @spleenblatt

      I’d love that contract – sight unseen I would guarantee I could do it for < $20m

      • Maybe you can. After all, one two person team is probably the most effective cost wise. But it will probably take a decade.

    • The CEO of a mid tier bank told me it was going to cost them $50m to overhaul their It system, they ended up doing it in house at a fraction of that

  4. #ProphetsOfDoom

    MARKETS up last night – Gosh, between picking their noses and scratching their arses the deadbeats are going to be unbearable today. Wait … long interest rates fell in Europe last night (they won’t know this means prices rose), so maybe they’ll get excited, for all the wrong reasons = their trademark.

    Buy those banking shares …. oh, and housing in Melb and Syd. … cheap. Somebody has to be telling them this = otherwise they’ll reach the wrong conclusions, again, the Prince of Darkness perverse demonstration so evident.

    Markets are cheap, buy … lots …. now. The World is not coming to an end just yet.

    Watch this space.

  5. The main reason that economists in the world have been incapable in their prognostications is that all the statistics that they rely on have been very wrong for a long time.
    If statisticians say there is low unemployment and next to no inflation etc, then politicians and economists knowingly make ridiculous decisions, blaming the statisticians when everything goes wrong.
    The RBA goes further…it just copies the U.S. policy (and they copy the failed Japanese policy).
    That way Glenn Stevens can just say: well, if the U.S. central bank couldn’t see it coming, how could we?.

    It is all a terrible joke and we’ll all pay heavily for our laughter…

    • “The main reason that economists in the world have been incapable in their prognostications is that all the statistics that they rely on have been very wrong for a long time.”

      @athalone. I get your sentiment. But the main reason policy makers and other have been wrong all these years is that they do not express an appropriate error rate in their forecasts/assumptions, which are much more volatile than they would have you believe.

      See Fed Budget iron ore at $120/t in perpetuity, etc.

      • “See Fed Budget iron ore at $120/t in perpetuity, etc.”—Nick

        Yes Nick, another “statistic they rely on…”

        You agree..

  6. Upgraded at a cost of hundreds of millions of dollars, now there is an opportunity to make some cash if I ever saw one.

  7. “We’re in this ridiculous death spiral, where profits are increasing as a percentage of GDP; wages are decreasing as a percentage of GDP; and we’re using the excess of the stock price to the benefit of a tiny minority of people – who then complain that there is no place to put the capital… It’s just this pernicious game of generating increasing amounts of comp for the people at the very top, while giving short shrift to the employees and the country as a whole… and stock-based compensation and stock manipulation is at the heart of it.”

    — Nick Hanauer

    Skippy…. yet some are confused… in 08 did everyone just suffer some sort of collective amnesia….

    • That’s one way to look at it, another is that the internet has flattened the world and allowed competition for low cost/low skilled work.

      Productivity gains via the internet and now robotics simply cannot be put back in the bottle.

      These jobs are not coming back no matter how much people like to blame the rich.

      I am not so sure that people feel there is nowhere to put the capital. Perhaps just not in Australia.

      • The attack on wages was purely ideological in face of the operational reality’s wrt sovereign currency – monies as far back as the 70s. This was accentuated by wonky metaphysical precepts which were utilized to rationalize increasing looting which always promised the utopian good times were always right around the corner.

        Even the Economist has noted the shareholder democracy is road kill, the equity [zip rights] never trickled down to support the investment IS-LM as was envisioned [even with forced savings via 401K and super]. Instead you got an aggregate of wealth not seen since antiquity, Romanesque as it were. Which is acerbated by willful and w/ intent broad day light accounting fraud and a plethora of other anti social activity’s.

        At the end of the day we have a employment problem accentuated by PPP over extended timelines. As you note the private sector is completely disinterested due to short timeline metrics and global capital flows seeking the path of least resistance. Yet the architects to this condition over some decades are more interested in ideological purity and poster boys for Upton quotes than addressing the problem.

        Skippy…. as in America we are suffering the machinations of the far right wing and its funding ability.

      • “Productivity gains via the internet and now robotics simply cannot be put back in the bottle.”

        Which is fine, but it will probably mean the underutilisation rate mentioned above will only grow (especially since genuine economic growth isn’t anywhere on the horizon). It’s no wonder central banks and government are so desperate to restart asset bubbles and get people borrowing money again. Unfortunately I think many Westerners are starting to wake up to the fact they can’t keep borrowing to consume.

  8. Perhaps…

    Australian Bureau of Statistics (ABS)

    should be renamed to…

    Australian Bureau of Sci-Fi (ABSF)


  9. ” A broader concern is that timely & accurate data are vital to good decision making”

    hmm, well that could explain it.

    Not much sign of this Govt. wanting anything to do with that.

    If you really want to know the number of unemployed look at the centre link numbers. They already exist on an existing computer . FFS

  10. . . . . the computer system used to create its labour force data is 30 years old and in desperate need of upgrading (at a cost of hundreds of millions of dollars).

    Surely mig could do that in a fortnight.

    With a smartphone app thrown in.

  11. “potentially impose costs on the Australian economy over the longer-term via incorrect policy responses” WTF Did they not read the small print?

    Achtung! Achtung! Streng zu Propaganda Verwendung.

  12. It is a real shame that this country can not provide statistics with some degree of accuracy. ABS stats are heavily important for those with in goverment, how can you make any good policy if your measuring stick is warped?

    It needs a overhaul, i respect ABS does many things but I feel the system is terribly inefficient in regards to job numbers etc.

  13. Who cares? Policy is being decided by ideology, not data anyway. You said it yourself, rates would be cut, CPI is irrelevant.