Date posted: 19/08/2020 5 min read

Why we should push for more near-real-time numbers

COVID-19 kick-started new ways to quickly measure the economic pulse of our nations. And that’s a good thing.

In Brief

  • Big economic numbers such as GDP or balance of payments are complex concepts that take time to calculate to get an accurate result.
  • During COVID, both the ABS and Stats NZ started releasing preliminary data to provide snapshots of economic activity.
  • As yet, there is no commitment these innovations will continue beyond COVID.

It’s the end of a June financial year and the senior leadership team under its new CEO is meeting in early July to go over the year’s results.

“Here’s the thing,” says the CFO. “We won’t really know how we went until late September. Maybe mid September if everything goes OK. Of course, there’ll be revisions to the numbers in December, but our investors are used to that.”

There is a long pause.

“What have we got in the meantime?” queries the CEO.

“Glad you asked. We should have sales next month. And headcount.”

“And the results from our new overseas consultancy service?”


“A balance sheet?”


The CEO leans back in her chair.

“Guys,” she says, “I need to talk to the CFO. Alone... ”

If you think that’s fanciful, then you weren’t up with the pre-COVID state of play of official statistics. Our treasurers – Josh Frydenberg in Australia and Grant Robertson in New Zealand – would have been desperate to know, as soon as possible, what had happened to March quarter GDP during the height of the first COVID outbreaks, but didn’t get to know until 3 June (Frydenberg) or 18 June (Robertson).

They won’t have the June quarter outcomes until 2 September (Oz) and 17 September (NZ).

Other data also came with lags, particularly in New Zealand where retail sales and unemployment – two of the bigger moving parts in the COVID cycle – are available only quarterly.

To be fair, counter-COVID policy measures were taken in both countries without knowing the full scale of the macroeconomic damage. “It’s really bad, whatever it is, so let’s push back hard” was as effective an approach as anything more informed, even if there had to be the occasional policy readjustment along the way.

Why getting industrial-strength numbers take time

Even if you had a highly pressing need to get them faster than usual, it isn’t the easiest thing in the world to speed up the production of GDP or the balance of payments.

The big economic numbers are complex concepts – I looked up the International Monetary Fund’s manual on compiling GDP to the latest specifications, and it runs to 722 pages – and because it requires a wide range of data inputs, the convoy tends to sail at the speed of the slowest ship.

Although every nerve in your body may be screaming to make quick-and-dirty approximations to get a number out there faster, the risk is that you introduce inconsistencies between the old industrial-strength numbers and the new cobbled-together ones.

Comparability – over time within one country, and across countries – is still important. It does you no good to think you have weathered the COVID storm better than somewhere else if all that’s happened is you haven’t measured the cost as carefully as they have.

“It does you no good to think you have weathered the COVID storm better than somewhere else if all that’s happened is you haven’t measured the cost as carefully as they have.”
Donal Curtin

“Unreliable early estimates should never be published,” the UN’s manual on quarterly GDP sternly says, “as providing inaccurate signals about the current state of the economy can create more problems than benefits for the users and undermine the credibility of compilers.”

All that said, the bad news is that pre COVID, neither Australia nor New Zealand was in the best shape when it came to highly timely cyclical data. Indeed, if you were in a grouchy mood, you might well ask why the experience of the global financial crisis hadn’t already prompted a stronger focus on near real-time monitoring of the business cycle.

“Pre COVID, neither Australia nor New Zealand was in the best shape when it came to highly timely cyclical data.”
Donal Curtin

The good news is that – despite the logistical challenges of organising workflow through COVID – both countries have done a lot to fill in the gaps.

Near real-time data now available

The Australian Bureau of Statistics (ABS) and Stats NZ have upped their game no end. The ABS, for example, has run new surveys of businesses and households to see how COVID-19 has affected them, and accelerated the release of its existing suite of statistics. (It now releases preliminary estimates of monthly retail sales and monthly merchandise trade, for example, rather than waiting for 100% coverage before publishing.)

Stats NZ has a nice data portal that combines its own provisional and preliminary data series and a variety of other near-real-time data from other agencies. Daily electricity demand, for example, gives a very strong feel for the hit to economic activity. You can clearly see that the trough of New Zealand’s COVID impact was mid-April.

Both countries now have a weekly measure of employment – a huge improvement on our previous understanding. In Australia’s case, it is available with only a fortnight’s delay, and it shows that Australia’s COVID trough was also mid-April.

Another innovation has been the New Zealand Activity Index (NZAC), put together for New Zealand by Treasury, the Reserve Bank and Stats NZ, which while “not an interim estimate of GDP” (as Treasury notes), nonetheless “summarises eight monthly indicators of economic activity covering consumer spending, unemployment, job vacancies, traffic volumes, electricity generation, economic outlook and manufacturing expectations. It will be updated around 14 days after the end of each month.”

It’s as good a near-real-time indicator of overall economic activity as we’re ever likely to get.

These are great innovations, and deserve even more credit for being rolled out in difficult times. It’s a pity, though, that it took a crisis for them to see the light of day.

Those terrific weekly job estimates, for example, are based on the data that employers were providing in any event to the tax authorities (in Australia, for example, from the Single Touch Payroll system).

Promptly recycling this ‘administrative’ data to give a more timely read on the labour market should always have been higher on the agenda. Even now, it’s not a given that they will be maintained beyond COVID. Stats NZ says it has no commitment beyond October 2020.

That would be a shame. There’s a role for readers here: both the ABS and Stats NZ are keen to know what users think of this new information. Log on to their sites and tell them. Let’s not go back to square one.