Officially unemployment dropped in the June quarter - from 4.2 at the end of March to 4 per cent at the end of June.
That is extremely weird and, intuitively, just plain wrong.
We knew the labour data was going to be weird but this morning's release was an OMG moment for those following it closely.
ANZ senior economist Liz Kendall puts it bluntly: "Today's data massively understates the weakness that was prevailing in the labour market in Q2, due to a measurement issue".
It sure does.
Thousands of people have lost their jobs. Latest numbers show more than 212,000 people are now on a Jobseeker benefit - up more than 100,000 on the same time last year.
One big, simple reason for the low number is that the Household Labour Force Survey provides an average unemployment number from across the quarter.
As ANZ's Kendall notes, unemployment fell as low as 1.5 per cent during the initial weeks of the quarter, when we were all in lockdown.
That's because those out of work during that period didn't meet the requirement to be officially unemployed.
To be categorised as unemployed by StatsNZ, a person must:
• not have a job,
• be available to start work, and
• have been actively seeking work in the past four weeks or be due to start a new job in the next four weeks.
If you were locked down, you just weren't looking for work.
The figure did rise through the month and was around 6.2 per cent by the final week of the Household Labour Force survey.
"This steady rise partly reflects measurement issues resolving, and job losses genuinely rising as the crisis unfolded," Kendall says.
That number feels intuitively close to where we are at - although the sample size for just one week of the survey means that figure should also be taken with a grain of salt.
To smooth things out StatsNZ has released an extended unemployment figure which includes those who did not meet the classification for official unemployment reasons related to Covid-19.
That gives us an average of 4.6 per cent for the quarter - up just slightly on the 4.2 per cent March figure.
What, if anything does that tell us?
As Ben Udy at Capital Economics says, it "shows that the Government's generous wage subsidy protected many jobs".
The wage subsidy policy worked, probably better than anyone expected.
The multibillion-dollar costs of that policy are a debate for another day.
But of course when it comes to an end on September 1, most economists expect we'll see another bump up in unemployment after that.
Unemployment is still forecast to peak somewhere between 8 and 10 per cent by most economists.
It's unlikely they'll be revising forecasts too far based on today's data.
They might however be having a look at their timing for the peak.
Treasury's Budget forecast was for the unemployment rate to hit 8.3 per cent in June.
Subsidies have cushioned the blow from Covid and have probably delayed the low point in the economic cycle.
But with borders closed for the foreseeable future, there's few who'd argue that we're through the worst yet.
"Currently around 430,000 workers are on the extended subsidy," Westpac senior economist Michael Gordon says.
"We don't know what portion of those jobs are at risk once the wage subsidy ends, but they could far outweigh the 11,000 jobs lost to date."
He, like most economists, sees the underutilisation rate as a more robust economic indicator this quarter.
The employment measure showed 11,000 fewer employed people than last quarter, down to 2,665,000.
"The broader underutilisation rate gives a better guide to what life was like under lockdown," he says.
"This measure rose from 10.4 per cent to 12 per cent, the biggest increase since the series began in 2004. This measure captures those who would like a job but were not actively looking."
The unemployment figure from the HLFS is still a good one.
It dates back to 1986 and allows robust comparison with previous recessions - like 1991 when unemployment hit a record 11 per cent.
Ultimately we just have to accept that the last quarter was one of the weirdest in New Zealand's history from an economic point of view.
The distorting effect of lockdown is also going to do weird and not-so-wonderful things to the next round of GDP data.
But relatively speaking we remain in better shape than our international peers - who also face some measurement issues and rises in underutilisation rate.
New Zealand's 4 per cent June Quarter rate compares to 7.4 per cent in Australia, 11.1 per cent in the US and 12.3 per cent in Canada.
For all the statistical anomalies and serious worries ahead, "today's labour market data was certainly welcome to the extent they indicate a stronger-than-expected starting point," says ASB senior economist Mike Jones.
"We expect the unemployment rate to rise from here but concede that the peak may be lower than first thought."