Failing statistics

I’ve had a series of points about New Zealand official statistics running round in my head and the list finally got long enough I thought I’d turn it into a post.

Most prominently, of course, there is the 2018 Census debacle.  Almost 16 months on there is still no data published, as the SNZ efforts to compensate for their own systematic failures by trying to fill in the gaps go on.   We still have to wait another two months before we begin to see some results at last.    Consistent with the deeply attenuated nature of public sector accountability in New Zealand, no one has resigned, no one has been sacked.  No one has even offered a genuine and heartfelt apology.  It should be simply remarkable that the Government Statistician, Liz McPherson is still in her job.  Instead, when the nature of the debacle was already apparent, McPherson was reappointed to a second term.   If she knew there were going to be problems –  eg underfunding – she had a moral obligation to have made that clear and to have considered resigning and going public if the issues weren’t dealt with.  If she didn’t know, she shouldn’t be in the job anyway.

There doesn’t seem to have been a parliamentary inquiry into what went wrong, and we still haven’t even had the report from the reviewers that McPherson herself appointed to review how her organisation has handled things (due to SNZ this month, although who knows when we  –  the public  – will see it).   One wonders if the reviewers will note SNZ’s apparent greater focus on various right-on political causes than on doing the basics well.  Probably not –  it isn’t the way to get future review-type appointments.

(Here I will largely skip over questions about whether it is really appropriate for the coercive powers of the state to compel us all to tell the state whether we are able to wash or dress ourselves.  I am seriously contemplating a rare act of civil disobedience at the next census, simply refusing to answer such grossly intrusive questions.)

But, having mentioned the SNZ priority on trendy causes, there is the Indicators Aotearoa New Zealand project I wrote about here.   Dozens and dozens of indicators about New Zealand (reminder to SNZ “New Zealand” is the name of the country), some perfectly sensible and already published, and others strange, vacuous, almost impossible to measure meaningfully (in one or two cases all three).   You might recall this extract from the table


where, for example, only Maori “spiritual health” (whatever it is) seemed to matter.  Or where if descendants of Croatian immigrants don’t speak Serbo-Croat that is somehow a problem for New Zealand, the New Zealand government, or (indeed) those individuals.  Or where our statistics thinks a ‘strong sense of belonging and connection’ to New Zealand is something they should measure.  I was born here, most of my great-grandparents were born here, and I don’t need Ms McPherson to try to tell me whether or not I’m a proper New Zealander –  even though being “a New Zealander” is not, and never will be, my primary “identity”.

Eric Crampton captures some of the lunacy of it all in a recent tweet

A couple of weeks ago a reader drew my attention to an International Monetary Fund graphic about which countries were meeting which international statistical standards (collection, publication etc).   Here is the summary chart


drawn from this page.    The two most advanced standards are SDDS (Special Data Dissemination Standard) and SDDS Plus.    There are some lower level standards (the two shades of green) and then there are the countries outside the standards altogether.  Eyeballing the map, that would be (of independent countries) Cuba, North Korea, Turkmenistan, South Sudan, Somali, and…….New Zealand.   You can read all about SDDS here –  76 countries have signed up to it since 1996.

I have some history on this issue.  When SDDS was first launched, in the internal bureaucratic discussions on such matters I never regarded New Zealand signing up as a priority, and said as much.  At the time, from memory, it was mostly for advanced countries, and we were  (a) small, (b) having no trouble attracting international buyers for New Zealand dollar securities, and (c) these were still the days in the immediate wake of our far-reaching reforms.  Why would we need to sign up to such international agency bids for relevance (might have been the gist of my sentiment). At the time, from memory, I probably still adhered to the official RB view of “who wants a monthly CPI; there are more important priorities”.  In those days, we could not subscribe to the standard because, unlike most OECD countries, we had neither a monthly CPI nor a monthly industrial production series.  We still don’t (I’m not sure if the entry rules have changed though).    Both represent fairly significant gaps, and more recently (perhaps five years ago) even the Reserve Bank came round to the view that a monthly CPI would be desirable.

At one level, our continued failure to meet the requirements for these international standards doesn’t matter very much.  No one supposes we are, say, Zambia or Tajikistan.  Our statistics are honest, even if there are significant gaps.  We still don’t have problem selling New Zealand dollar securities.    But when –  as they do –  governments and officials parrot on about “rules-based orders”, the importance of international standards, it does look at least a little embarrassing not to be part of these, not very onerous, international standards.  And economic analysts would actually use data like a CPI or an industrial production series.

And then, of course, there are other weaknesses in this area. Our quarterly national accounts numbers –  which themselves have material gaps (no quarterly income measure of GDP) –  are released more slowly than those of almost any OECD country (and, of course, are still subject to significant revisions even then).

Talking of the IMF and statistics, another reader pointed out to me recently that New Zealand seems to have been reduced to accepting technical assistance from the IMF on some aspects of our financial statistics.   Not a big issue in its own right perhaps, but I was a bit surprised nonetheless –  technical assistance (foreign aid) from the IMF has usually been something for underdeveloped and emerging countries –  especially as I knew the Bank had had a temporary secondee from the IMF a few years ago, who seemed to do a lot of work on these specific areas.  Just seems symptomatic of the not-overly-job New Zealand is doing these days around official statistics.  I guess decades of poor productivity growth really does show up in choices –  whether about cancer treatments, and things nearer to public goods such as official statistics.

My final statistics gripe for the day relates to the immigration statistics. You will recall that last year SNZ, together with Customs, MBIE, the government, and no doubt under pressure from airlines/airports etc, got rid of departure cards.  With them went one of the key short-term economic indicators (PLT migration numbers) analysts have used for decades to track short-term economic developments.  Not only is migration more important here (larger, as share of population) than in most advanced countries, but there are big cyclical fluctuations in migration (of the sort not seen in most advanced countries), largely because of the relatively free access New Zealanders have to Australia.

SNZ led the official chorus trying to tell us that the new world would be better for everyone.  It was never going to be, and I pointed this out in several posts before the final decision was made.  Sure, using passport data to work out whether or not people actually stayed (or left) long-term would produce better long-term indications of actual movements, but only with a very long lag, and in the meantime we would lose all useful short-term information on the movements of New Zealanders.  Their model estimates for the short-term were always going to have such large margins of error –  perhaps especially around turning points –  that any signal was going to be very hard to discern from the noise.  SNZ tried to tell is it wasn’t so, and when the new data starting coming out a few months ago, they continued to release prominent monthly commentaries, emphasising the signal.  And they did the same thing in each successive month even as the inevitable substantial revisions threw the numbers around.

But they seem to have finally realised that there is a problem.  On Thursday evening, I received a consultative document from SNZ inviting comment on “options for release of international migration data”.  It is a rushed affair –  they want comments within a week, much faster than normal official consultations.  I couldn’t see the document on their website but there was no indication they wanted it kept confidential either.

As I have noted to SNZ in my response, there is little sign (still) in the document that they recognise the importance of immigration changes in the short-term economic developments in New Zealand (the official who sent out the document is a demographer and the main interest seems to be in the needs of fellow demographers).

Anyway, they are now toying with dropping monthly data altogether, with releasing data only quarterly (even if there was monthly data in each release), and with dropping high frequency commentary on the net migration numbers (the latter is a move I would support –  SNZ commentary to date has fed an inappropriate reliance on highly questionable numbers).     Fortunately, they do note that they are not looking at options such as only releasing data with a six months plus lag (when the revisions have started to settle down), or release data only annually –  good of them, but extraordinary that such options even get a mention, in a country where migration data makes such a difference (including in the political debate), and where good and timely data should have a priority.

(For what it is worth, I have gone back to them urging them to keep monthly data, released monthly, with a short lag, but released straight onto the website without commentary.  The data may be poor –  and that is SNZ’s responsibility –  but there is no good reason for them to sit on data which could be made available, for analysts to make of it what they can, even recognising that the signal to noise ratio is very low.)

I could go on –  there is, after all, the breathless enthusiasm for the IDI, with little apparent thought about where such tools might lead – but won’t today.

The bottom line looks like a mix of problems.  There probably has been underspending on official statistics over the years (public goods have few vested interests to champion them), as well as some misplaced priorities (whether coming from ministers or officials). which in turn encourages a champing at the bit for apparently smarter, apparently cheaper, alternatives –  be it the Census or the migration data or whatever.  But before thinking about throwing more money at the problems, there needs to be some real accountability –  the Government Statistician in particular, but also successive Ministers of Statistics.  If we are going to do government well, two aspects of that should be serious accountability –  if you stuff up badly at the top, and especially if there is no contrition –  you should lose you job –  and doing official statistics excellently.   New Zealand is failing on both counts (and, of course, the failure on accountability runs much more broadly than SNZ),




8 thoughts on “Failing statistics

  1. If you can measure it you can manage it.
    Presumably is Serb-Croats can’t speak their grandparents’ mother tongue the government will want to do something about it.
    Otherwise, why measure it?

    Liked by 1 person

  2. Accountability has become an old fashion concept from our leaders, both private and public. Yet these same people see no hypocrisy when they demand the less fortunate in our society do what they don’t.

    At the same time it is hard to complain when we just roll over and acquiesce to such craven behaviour.


  3. What’s the point of measuring anything when the coalition has abolished targets for government performance in key areas of expenditure like health and education? If you do wish to indulge in such anal behaviour then any “statistics” should support the dogmas of the “woke” identity politicians who rule us and their prevailing groupthink.

    Liked by 1 person

  4. “I’m increasingly seeing national party politics as an irrelevance & a distraction to real politics, which is taking place in civil society, within the state apparatus and other major institutions, including business. I’m thinking this is where our world is being shaped.”


  5. The census contains too many questions for which there is no need for the extreme accuracy of surveying every person in the country as well as questions where the concept and answers are fuzzy and arbitrary.


  6. Yet if your a kiwi of Scottish or Irish descent who has a family that goes back generations, There would be no panic to make sure you knew Gaelic. At Auckland University years ago I had a lecturer who quoted polices from other countries that if you had a diaspora of more than 100 k then the Govt must fund initiatives to keep your Ancestral language alive. He said Dutch and Croatian would be taught. When I mentioned the majority celts and suggested Gaelic it went silent.


  7. For the SDDS standards NZ isn’t a signatory – but my impression was that we voluntarily disclose the same information (eg A quick google search came up with a bunch of IMF documents stating that was the case – so I’m not quite sure that map gives a fair representation of the data quality and transparency in that particular space.


    • Matt

      My reading of that RB link is that it simply says our fx reserves data complies with SDDS standards. It is several years since I was involved, but historically we did not subscribe to SDDS because we could not (SNZ didn’t produce enough data at sufficiently high frequency – from memory, one of monthly CPI or monthly IP would have been sufficient). Of course, when I say “we could not”, that in itself was a budgetary choice, about how much to spend on stats, and on which stats. Neither govt nor Tsy nor the Bank really prioritised SDDS, treating it as a “nice to have”.

      This is from the 2018 Article IV report
      “General. Data provision is adequate for surveillance. The authorities are continuing to enhance data quality and expand the range of data available, and are making progress towards subscribing to the IMF’s Special Data Dissemination Standard (SDDS).”


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s