The employment decline exists – let’s just be careful interpreting it

Over at Whale Oil I spotted this piece by the NBR which discusses the claims of “40,000 job losses in manufacturing” – a claim I think I’ve heard from the Greens in the past, I’ll include a link if someone gives me one 🙂 .  [Update Link here, along with a pointer that it is from QES for 2008 to now.  Adjusting for seasonal variation this specific claim does look a little exaggerated in terms of filled jobs (but I still don’t think it is necessarily a misleading claim in this context) – I’m also not sure why we’d ignore the 2005-08 period given this is when the real exchange rate issues really kicked into gear.]

The figures quoted by Whale Oil suggest that the drop was only half that over the past 17 quarters.  However, the people who have been claiming that employment in the manufacturing sector is down by around 40k over time are not wrong.

For that, let’s jump on the Statistics New Zealand site and look at the HLFS (Household labour force survey).  Below I have a graph of annual average employment from 2004 onwards:

Between it’s peak in March 2005 and now, average annual employment is indeed down by 44,125 people.

Now this measure isn’t perfect – but it is better than the QES.  The HLFS measures “employment”, the QES measured the number of jobs (people can have multiple jobs) and it excludes self-employment and some areas such as agriculture.  Self-employment is pretty important in a industry filled with small firms like NZ manufacturing, so we have to keep that in mind! [Note: For some context, the level of annual QES jobs are currently 77% of the level of annual HLFS employment in manufacturing – even though there are multiple jobs for some employees.  Normally this ratio is closer to 80%, but it does indicate that some true underlying jobs/employment may be missed by the QES 🙂 ]

Of course, we need to think about “why” employment is shifting out of manufacturing and what that means before we can say anything about it – there is no fixed “lump” of jobs, and the structure of the economy does change.  Eric Crampton notes that in this tweet, pointing to this post on the manufacturing industry.

But in of itself, the number is not bogus – it is straight from the HLFS! it could have come straight from the HLFS 😉

Update:  In terms of the NBR pieces conclusion – it is true there has been a secular decline over a long period of time.  Understanding why in this case is important – and exciting.  Global manufacturing employment is falling, could we be about to experience the “manufacturing revolution” version of the “agricultural revolution”?  I hope so, and I hope that if this is the case then unlike the agricultural revolution society/government helps people with the transition 🙂

Update again:  I’ve noticed people from all the sides blaming government X or Y – a structural trend in the industry, a downturn in global manufacturing, and seizing up of credit conditions to the point where we haven’t been building, are not factors I would blame on any politicians.  And I’m notoriously harsh on politicians.

  • Rob Hosking

    The point of my story – one of them – wasn’t that the employment decline doesn’t exist: far from it.

    The starting point was I keep hearing the line that the current govt has lost 40,000 jobs in manufacturing and this just doesn’t seem to be in the data.

    They’d stopped using the 40,000 figure after the last time I queried it, but then it came out again yesterday.

    So I looked at the 17 quarters from the first full quarter the current lot have been in power, and by way of comparison looked at the previous 17 quarters.

    I should add the last time I asked the Green office where they were getting their numbers from they said it was the QES.

    So I looked at the QES filled jobs and FTEs, for the two lots of 17 quarters either side of 31 Dec 2008.

    There wasn’t a radical difference either way, as the story showed.

    The one thing I’m puzzled about is your HLFS data and mine – yours is annual, mine is quarterly, but even so they don’t seem to match at all.

    Haven’t had time to look at this yet but will do so.

    The wider point I’d like to make is of course that jobs aren’t the sole, or even the main, measure of success in an industry, and it worries me we seem to get caught into talking as though they are – at least in manufacturing, although funnily enough not in other industries.

    cheers

    Rob H

    • Cheers, as I noted at the end in terms of your actual piece I appreciated the longer term structural focus you had going – it is an important point that is forgotten!

      In Infoshare the series I was using was:

      Persons
      Employed by Sex by Industry, ANZSIC06 (Qrtly-Mar/Jun/Sep/Dec)

      Total Both Sexes

      Manufacturing

      And given I was working off the Whale Oil post, rather than a comparison to governments, I just went from the “peak” in Mar-05. It makes sense, as it was at this point a lot of factors combined which would have been expected to “boost the real exchange rate” … which is the path we are looking at in terms of causation!

      Of course, what would be best is if we could just look at the LEED data – that kicks the HLFS employment and QES jobs data out of the water in terms of accuracy. Just not timely.

    • pete

      So, in your NBR story, when you say “Dr Norman is just making numbers up”, you _knew_ that they actually came from the QES? That really is astoundingly dishonest.

      • I don’t think that is really fair on Rob – I don’t think he is being dishonest at all. If the claim from the Greens is truly based on the last 17 quarters, and if they are only using the QES as they are worried about data issues with the HLFS (following the earthquake say), then Rob has found that the figure is not 40k. In that case, saying it is 40k when it isn’t is just “making it up”.

        Journalists, and economists, tend to get annoyed by people with vested interests misusing data – and so I fully appreciate his sentiment. I posted here, not in response to his article itself, but just when I saw others stating that a 40k figure was just a lie in general. Hence why I popped up a graph that gave it some context, in terms of a measure that I view as a bit more appropriate (although I am then talking about employment numbers rather than “jobs” in that context) 🙂

        • pete

          The difference between 2008Q2 and 2012Q3 in the QES is 40,000. If Rob wants to argue that that’s cherry picking and that the last 17 quarters are a more appropriate measure then that’s fine. But instead he’s accused Dr Norman of making up numbers.

          • Fair enough.

            I’ll be honest that I’m not really about the politics of it all. I just want to know why the change happened and what it means.

            Would be neat if, like, all the “finance ministers” had to write a (say) 4,000 word essay (with graphs) to discuss the economy over the last decade. With something like that, it would be easier to see what the different political parties mean when they discuss things!

            • pete

              That would be fantastic.

            • Rob Hosking

              This is a superb idea.

          • Rob Hosking

            “Cherry picking” though is a weasel word in this context, because, firstly, Norman’s statement was headed up “manufacturing languishes under four years of National’ and then says “There’s no signs of clawing back any of the 40,000 jobs lost in the manufacturing sector since 2008.”

            Question: would anyone else, reading this, not presume the 40,000 jobs was since the change of government?

            I genuinely want an answer to that question, by the way.

            Anyway, I started trying to find the 40,000. And I couldn’t find them.

            If Norman’s statement had said “since June 2008 – oh and, until Q3 last year even though there’s been two sets of updated data since then” I would have – but then I would also pointed out Norman was deliberately using outdated data and ignoring the more recent data.

            And I would have used the term “misleading” to describe it, because it would have been misleading.

            The bigger issue here is this: something Matt touched on – politicians misusing data.

            In this case, its been used to talk up a “crisis” that isn’t.

            NZ has a horrible tendency to conduct economic arguments in apocalyptic doom and gloom tones and it is extraordinarily damaging.

            I lived through a lot of it in the 1970s-90s and its is corrosive and destructive – and, when politicians do it, shallow and irresponsible.

            It contributed, I believe, to a sense of despair about this country’s prospects.

            It really fires me up when I see politicians (and other commentators) doing it.

            By all means, discuss how you would change the policy settings, regulate or tax differently, and why – all that.

            But don’t manufacture crises which aren’t there: don’t magnify our difficulties.

  • Pingback: The employment decline exists – let’s just be careful interpreting it | Fifth Estate()