Wednesday, 26 March 2014

Thou shalt consult - Day 1

I spent today at Statistics New Zealand's User Forum in Wellington. I didn't know, until Government Statistician Liz MacPherson mentioned it in her opening speech, that the forum is actually required in legislation: I looked it up, and sure enough s10(1) of the Statistics Act 1975 says that "The Statistician shall, at least every 5 years, convene a meeting of users of official statistics to review user needs for such statistics".

At first I thought this was rather odd, and so did some of the people I talked to, in the sense that you'd think most agencies would have the wit to consult with important stakeholders from time to time without needing to be formally instructed, and I put the statutory requirement down to the micromanagement style of 1970s legislation. But then as I thought about it, I began to wonder if it wouldn't be a good idea if all of the public sector had the same formal requirement to consult. Many do it off their own bat, of course - Stats itself has a good record of consulting users and suppliers - and public sector stakeholder surveys of one type or another tend to feed into the strategic planning process and the writing of Statements of Intent and the like, but it's probably not such a bad idea to require agencies to go back to the basics and get their customers' views on what they most want delivered.

In any event it was a fascinating day (there's another day of it tomorrow), and I did my little bit to help it along by chairing one of the concurrent sessions. Here are a few of the things that got my attention.

First, the enthusiasm of Maurice Williamson, who's the Minister of Statistics. I've heard him before on the value of statistics - he gave an excellent speech in Wellington, for example, at the launch of Stats' digitised set of Yearbooks - and it's good to have a Minister who's connected the dots between maintaining a good statistical service and making solid, evidence based, policy decisions.

Liz MacPherson's speech mentioned our high rating on the Global Open Data Barometer. I'd never heard of the thing before, but then one of the benefits of forums like this is that they point you to new sources you weren't aware of. The Barometer "explores countries' readiness to secure benefits from open data, the publication of key datasets, and evidence of emerging impacts from OGD [Open Government Data]", says its website, and we score a highly respectable 4th out of 77. There's a whole range of official data made freely available on - news to me, I have to confess, and why it's all on a non-Stats website I don't know, but all that said it's good that we're making far more available, for free, than many other countries.

Graeme Wheeler, the Reserve Bank governor, gave us his "if only" list  of the macro stat gaps he'd most like to see filled - sectoral balance sheets ("who owes what to whom" as he described them), an income version of quarterly GDP, more frequent rebasing of GDP so as to get earlier warning of structural change happening in the economy, and a monthly CPI. Amen to all of that, I say, especially the income measure of GDP. If you're relatively new to this issue, the situation is that we have got two of the three ways of measuring GDP (output, and expenditure) but not the third (income), and the third way is the one that generates some important information that you don't get the other two ways, notably (my particular interest) corporate profits.

Grant Barnes, who heads up strategy and research for Auckland Council, talked mostly about the links between Auckland's strategic plan and the data needed to feed it and monitor it, which was all fine. Along the way he put up a remarkable graph showing the projected population growth in Auckland (heading for 2 million people) and the projected population growth in other parts of New Zealand (relatively insignificant). I'm not an "everything into Auckland" booster, but if that's where we're headed, then we'd better grease the wheels so that it happens as smoothly as possible.

Alan Johnson, principal researcher for the Salvation Army, started off with a proposition which I initially thought was old-style class warriorish - the data that gets collected is by and large the data of most value to the rich and powerful, was the gist of it - but as he went along he made a good fist of arguing that social statistics in general, and distributional data in particular, don't get a fair look in. Pretty much everyone I talked to was impressed. The Sallies have produced their own State of the Nation report to show the data they think are important.

Keith Ng, who engagingly described himself in the forum bumph as a "data nerd" and is probably best known as the guy who discovered that WINZ kiosks had accidentally exposed access to confidential WINZ files, was excellent on how people visualise data and how to present data in ways that best press people's buttons. I think that I know a thing or two about getting economic data across to non-specialists, but Keith told me stuff I didn't know. I'd like to point you to some more of Keith's resources, but he doesn't seem to be blogging much anymore at Public Address and his data visualisation website ChewyData is still an early work in progress. But he still struck me as someone who had intelligent things to say about data presentation.

Another thing I hadn't heard of before was Wiki New Zealand, which Lillian Grace, its founder, talked about. As it says on its website, "Wiki New Zealand brings data together in one place and in accessible formats. Topics are presented from multiple angles, wider contexts and over time. Presenting this data in similar forms invites users to compare, contrast and interpret it easily and without bias". Have a look: I've only had a quick play with it, and my initial reaction is that its upside, and its downside, is that you might find anything there. If you're a fox (you know a lot of small things), you'll love it. If you're a hedgehog (you know one big thing), you might want something more structured.

Big overarching themes? A few. Data isn't information, or intelligence (but you knew that), and turning it into information or intelligence is a skill in itself. In the session I chaired, for example, Klaas Stijnen, managing director of Montoux, talked about a service called Zuno which, in essence, turns the data from the Annual Enterprise Survey into benchmarking information that enables a company to figure out how it is travelling against other firms in its sector.Timeliness is becoming a really significant issue - always was, but more so now. A lot - a lot - of people rely on various bits of the Census data, more than I would have thought. And there's a lot of clever work going on which meshes different Stats data sets to create something bigger than the sum of its parts, for example by combining demographic or economic data with geolocation data, as Tim Ryan from Critchlow demonstrated.


  1. Wandering Albert Ross27 March 2014 at 11:35

    You might want to have a look at NZPC's report on regulatory institutions and practices, chapter 8, where they conclude that whether regulators should have a formal requirement to consult depends on the specific circumstances. Actually that is pretty much what they conclude about every aspect of regulatory design

  2. Great comment, thanks. I have to fess up and say that thus far I've been dipping in and out of the NZPC regulatory stuff, eg, and haven't read the whole thing yet. I'll get onto it