sign uplog in
Want to go ad-free? Find out how, here.

David Hargreaves says the next Government needs to give a much bigger priority to ensuring we have the best statistical information possible

David Hargreaves says the next Government needs to give a much bigger priority to ensuring we have the best statistical information possible

If ever there might have been a time for Statistics New Zealand to come out and say: "Look, the figure we've got here is nonsense and we've thrown it in the bin", then surely that would have been the case on Wednesday with the unemployment figure.

The fact is that New Zealand is going into the September 19 election with an advertised unemployment rate of 4%.

This in the middle of a global pandemic that it wreaking economic havoc around the world.

I don't believe New Zealand had 4% unemployment at the end of June and I'm sure you don't either.

And yet, anybody, anywhere in the world as of today, who wants to find out what New Zealand's unemployment rate is, can look at the homepage on Stats NZ's website and be told it is 4%. Misleading? A bit.

It could be a great trivia quiz question for the future: "Which country managed to lockdown its economy for weeks on end during the Covid-19 pandemic and maintain an unemployment rate of 4%?"

In a word, ridiculous.

Let me hastily say here that I have nothing but sympathy for Stats NZ. 

How on earth do you go about measuring unemployment when the people you are contacting as your sample are all locked up, not working, and those who might have officially lost their jobs aren't in a position to actually look for new work (so technically don't count as being 'unemployed')?

Pretty damn difficult for the Stats people.

And they have explained this and have, if you will, warned people, without saying as such that they are warning people. And in the release of the unemployment information on Wednesday they went to great lengths to point out various other measures that point out the deteriorating employment situation during the quarter.

One of the most salient parts of the release is the outlining of the progression of the unemployment rate during the quarter. It started out at just 2.7% and finished at 4.9%, which surely is getting a little warmer in terms of what the actual rate might have been.

A problem in this instance is that the methodology of Stats NZ's Household Labourforce Survey is to take the average unemployment figure across the quarter.

Now, usually that's a perfectly fine idea, since it takes out strange bits of volatility that might occur during the survey period.

BUT surely in this instance when there was one huge overriding event impacting the labour market the ONLY relevant thing is to look at how your employment figures are at the END of the survey period.

The method led to madness

Stats NZ would have been perfectly justified in saying something along the lines of: "Our methodology for the HLFS produced an unemployment figure over the quarter of 4%, however, due to the difficulties of measuring unemployment during this period you are advised to disregard this figure and take note of what the unemployment was at the END of the survey period."

That would have been sensible. But that would have been political as well and I imagine probably impossible for Stats NZ to do. So very difficult for them. Meat in the sandwich.

All of the information put out by Stats NZ has the potential to be politically loaded, with none more so than the employment figures. Imagine for example the huge disappointment in the National Party parliamentary offices on Wednesday when out came the Stats release showing that unemployment had fallen! Oh, to be a fly on the wall. I wonder whether by chance a pre-written press release slamming the Government for a massive increase in unemployment had to be chucked out before this rather tepid thing was produced telling us unconvincingly to ignore the 4%?

For the Government it was perhaps an unexpected opportunity to have a bit of a boast, though personally I did not appreciate Grant Robertson's willingness to accept that 4% figure.

It all just goes to show how political these data releases can be though. And shows the very difficult position Stats NZ is in.

We need to believe

Does it matter that we have an unemployment figure we can't believe? Does it matter if we have statistics coming out we can't believe?

Well, of course it does. Because important decisions are supposed to be made around official data. How are we all supposed to make decisions if we are in effect having to look through and around official data?

The Reserve Bank for example now has to take employment into consideration in its conducting of monetary policy. The RBNZ's own forecast for unemployment in the June quarter was 7%.

Theoretically, the release of a 4% figure should have the RBNZ severely backtracking when it comes to its thinking for its forthcoming Monetary Policy Review next week (August 12).

But hey, these RBNZ people did not come down in the first shower. When they come out with their prognostications next week I'm sure they'll have in mind that the true unemployment rate in the June quarter was a whole lot closer to 7% than 4%. Their thinking will not have changed.

Our next official unemployment figure is not due to be released till November, by which point the election will be well and truly over. I'll stick my finger in the wind and say that will see an unemployment figure of 7.5% - IE not far off double the rate we apparently had in the June quarter. And there will be big headlines and the opposition political parties (whoever they may be) will go off. And the reality is there will not have been a big jump in unemployment, it will simply be that (providing we don't have another lockdown in the meantime) the Stats' method of working out the unemployment figure will function properly this time.

Give statistics higher priority

Whoever forms the Government after the September 19 election, I hope they give statistics a much bigger priority than the Government we've had for the past three years.

I had some fun when writing this, looking up on the Beehive website and some of the many and varied ministerial portfolios. I confess I had never heard of the Community and Voluntary Sector portfolio. Yes, there is one.

Do you know how many press releases this Government has put out under the title of Community and Voluntary Sector?

It has put out 16 of them in the past three years.

Do you know how many press releases it has put out under the Statistics portfolio?

It has put out 14 press releases under the Statistics portfolio. Yes, 14.

And this has been in a period in which we have seen the enveloping statistical disaster that was and is the 2018 Census - a census from which we will never have information, I believe, on which people will confidently be able to make some of the big decisions that are supposed to be made based on such information.

Build it up

Now, the Minister in charge of Statistics, one James Shaw, is perfectly, very competent, but I suspect a bit distracted, given that he's co-leader of the Greens. One of his other 'gigs' is Minister of Climate Change - and he's put out 37 press releases under that portfolio in the last 12 months alone.

I don't think anybody should be too surprised that a politician leading a Green party is prioritising, well, green, issues. 

What we do need is a Government as a whole that puts much more value on having accurate, timely, statistical data.

I would hope that the next Government makes such a thing a priority.  

As a country we shouldn't have to be making assumptions about the way things are. We should have confidence that we are measuring them correctly.

And this means building up our statistical capability so that we can all have that confidence. 

We welcome your comments below. If you are not already registered, please register to comment.

Remember we welcome robust, respectful and insightful debate. We don't welcome abusive or defamatory comments and will de-register those repeatedly making such comments. Our current comment policy is here.


Well said

The serious issue is coming to be of one of confidence in any Statistics NZ data.
The glaring question regarding the reliability of the unemployment figure along with both the Census debacle and serious questions over migration and immigration raises the issue of competence and confidence in other Stats NZ data.
One expects a margin of error in any data, but seemingly problems with methodology and reliability of data is unacceptable and troubling.

I understand that Statistics NZ (traditionally a very competent/respected Govt Dept) is well into a process of re-building its reputation, following the "resignation" of its previous CEO - who will be remembered with little affection.

Yes, reliable/timely statistics are hugely important but, please everyone, let's give SNZ a bit of time. It's coming off a very low trough right now. I understand that morale is improving across SNZ staff.


I don't think this problem of headline measures painting an inaccurate picture is just a Stats NZ problem.
Do you think France is really facing just 8.5% unemployment when it's GDP dropped by nearly 20% in the June quarter? The same country that never saw unemployment drop below 7% throughout the 2002-2007 global growth frenzy.

A good statistician understands numbers but a great one understands the limitations of those numbers.
GDP, CPI, unemployment rate, etc. should not be taken on face value but are to be read in conjunction with other metrics. E.g. falling unemployment rate in NZ for the last 5 years looked promising but when graphed alongside our falling median wage growth, it paints a worrisome picture about the skill and wage-level of recent job growth.

Unemployment does lag - it will also be late in showing the decrease in unemployment when the economy picks up again. The lag is important to remove noise. If you want a now survey then look at the bank confidence surveys - more timely but less accurate (bigger margins of error).

I would like to see data start to be published on a per capita basis instead of an unajusted nominal.

It is published just not reported as the headline. A political decision.

This is not necessarily a problem with statistics itself. It's how the data is framed and simpified to a simple metric that can be reported to the public via the media and put into Powerpoint presentations. The data sets are there (an assumption) for people to dig, but the first point is framing the data appropriately. Roy Morgan measures the unemployment rate in Australia differently to how the govt measures. The RM unemployment rate always seems at least 5% pts higher than the govt (actually in April, RM reported the unemployment rate around 24%).

I'm a bit underwhelmed by Shaw in the Statistics portfolio. Asked in the couple of weeks after the ministerial portfolios were announced whether he was disappointed to be getting statistics (and not something juicier), he said that he and the Greens wanted the stats portfolio because stats is what drives a bulk of the other government decisions, and so getting stats done in an appropriate and proper way would let them better inform their climate change aspirations as well.

But it doesn't seem like much has come from that. Perhaps being 3rd wheel in this government with Winston seemingly doing everything he can to pull the handbrake on almost everything the government tries to get done, has sucked up too much of his time so he hasn't been able to get as much done in the stats portfolio as he wanted.

what a minister would be able to achieve? They do not even understand their portfolio better than the people they suppose to lead. It is symbolic leadership at best.
And governments have a strong disdain for collecting (and releasing) information than can be used to hold them into account. Ministry of Health (under both National and Labour and as far as I know over the past 20 years), is a classic example of avoiding collecting and reporting performance measures.

This was a government, if you remember, that dropped a lot of performance measures while saying they wanted to be held accountable for results (e.g. child poverty)

It's not often that the Statistics are worse than lies and damn lies.....

sorry wrong article to comment on.

Unemployment has always been a rubbish figure. Much like using raw GDP instead of GDP per capita.

Ah yes, the CVS portfolio, run by political heavyweights such as Jo Goodhew, Alfred Ngaro and Peeni Henare.

The is a lot of value in core statistic time series. Changing methodology on an ad-hoc basis would dissolve this value. It is a lagging indicator and will show the true values over time. You can have a long-term time series or you can have short-term estimate which cannot be compared easily.

I remember a similar beat up around immigration figures. Everyone was complaining that the old figures only showed intention at arrival so they were changed to show actual long-term intention which is not finalised until 16 months later. After the switch, every complained that the short-term values were too vague. You can only have certainty in one or the other.