NZ GDP at record highs?

It has been a long time since I’ve closely followed New Zealand macroeconomic statistics – but I was a bit surprised seeing the headline “GDP jump of 14 per cent completes NZ’s ‘V’-shaped recovery“.

Cliffs notes to this post – it was a strong result, but September likely wasn’t the strongest quarter on record, and real economic activity is still probably down a little bit. You gotta be careful with seasonal adjustment during “shock and turning points” – and Stats NZ did actually even tell us this in their release!

Update: As Economissive notes I can’t read and this was up on a year ago – good shout this was a dumb thing to miss. Dang that is a strong GDP result!

Reading the article makes it seem as if the New Zealand economy made up its COVID loses and was back to where it was prior to the pandemic. The only warnings in the text was that this may be due to “pent up demand” which will cool somewhat.

And this was a strong result – contrary to the article the increase in activity (including revisions to prior quarters) put activity above RBNZ or Treasury expectations.

But it is also not the whole truth.

Lets look at the GDP news release from Stats NZ. You’ll notice that, relative to a year ago, September quarter economic activity was down 2.2%. How could this be the strongest quarter on record if it is 2.2% lower than it was a year earlier? [Note the article says that activity was up on a year earlier, which must have come from comparing the seasonally adjusted figures – not the actual numbers).

It is true that the seasonally adjusted series is at an all time high. But (largely, igorning trading day effects) September quarters are the same in seasonal terms. So what has happened?

To figure this out we need to ask how we seasonally adjust. Seasonal adjustment involves taking the actual data and estimating what a “trend”, “seasonal”, and “irregular” component of the data is. Seasonally adjusted data then removes the seasonal component. However, to estimate a trend we need an idea of where data is going in the future – as a result if there is a turning point in the data, or a massive shock, the “end points” of the data become a bit biased.

Now Stats NZ is awesome and will have done a lot of work to try to clean up these estimates and deal with end point bias. But at some level a degree of bias is unavoidable – and as new data is released the seasonally adjust figures will be revised. In fact it is very likely that this specific figure will be revised down a bit through time.

In this way, the RBNZs forecast that economic activity would come in towards its late 2018/early 2019 level is probably about right. The Treasury forecast that it would be around its 2017 levels was probably a bit weak – but at the same time they are expecting growth in December, when the lack of tourist spending and slow down in durable spending growth from the September binge will likely swamp the benefit of Auckland not being in lockdown.

But ultimately who knows – it is danged hard to forecast a wild shock like this, and people have done a pretty good job over the second half of this year. My old workplace Infometrics has been doing a pretty awesome job all crisis long from my reading.

All I’m trying to say here is the following. Dude, always cross-check against the annual changes when making claims from seasonally adjusted data – if you get a result that says “strongest quarter ever” and it is down on a year earlier then there should be flashing red lights going off. And when Stats NZ put this at the very top of their news release:

This indicates that the strong growth in the September 2020 quarter was not enough to fully make up for the economic impact of COVID-19 and the measures taken to contain it.

We should probably include that in our news stories … 😉