Thread by @JustinWolfers

Non-farm payrolls print at +2.5 million, which no-one saw coming, suggesting that most private-sector forecasters got the number wrong by 20 MILLION.

The unemployment rate is DOWN to 13.3 percent (instead of up to 20% as forecast).

There's going to be a lot to unpack here...

Real-time reax: WHAT THE WHAT???

Two obvious things to bear in mind:
1. This isn't a typical recession, and it's surprisingly difficult to even define what it means to be employed or unemployed
2. Covid has forced the BLS to adapt their methodology *far* more than they typically do, which has unknown effects.

The BLS has done an amazing job in keeping the response rate to the payroll survey so high. But the number of firms isn't fixed, and it's hard to make accurate adjustments for firms going bust.

The household survey response rate is much lower than normal and may be less reliable

Lots more technical stuff to unpack, but...

It's hard to escape the conclusion that the economy bottomed in early/mid May.

We're in a massive and deep hole, and it'll take a while to climb out, but at least the hole isn't getting any deeper.

The unemployment rate is surely understated, and even the BLS agrees that in reality it's 16% rather than 13%.

The problem is a purely technical one: If you're employed but absent from work due to covid, they're meant to count you as employed, but it's difficult to code this.

A totally serious proposition: Given the technical measurement problems that even the BLS acknowledges, journalists should refer to the unemployment rate for May as having been 16%. That's the reality of our times. (On a similar basis, unemployment was 19% last month.)

An aside (given the times we're living in): Please don't with the conspiracy theories. The BLS is being clear, upfront, honest, consistent and reliable. Measuring unemployment during a pandemic is ridiculously difficult, and seemingly minor technical issues can have big effects.

I still can't believe that @betseystevenson wrote this paragraph a whole month ago. To give context, professional forecasters predicted payrolls would fall by about 8 million this month.

(Her whole piece was excellent: washingtonpost.com/opinions/2020/…)

This is easily the most important table in the jobs report.

Good news: Roughly 3/4 of the rise in unemployment reflects temporary layoffs. There's hope many of these folks will return to work soon.

Bad news: Permanent job loss has risen by over a million since Feb.

A remarkable fact (that so far has gone unremarked): This month's rise in non-farm payrolls of +2.5 million is (easily!) the largest monthly rise ever recorded.

But it's still only one-eighth of last month's monstrous decline of -20.7 million. (Also a record.)

Over the next few months, we all need to remember that these 3 things can all be true:
1. Covid pushed the economy into a massive hole
2. It's so deep that even just a partial recovery will yield historically unprecedented growth rates
3. The economy is still in a bad bad place

Or as I would say to my freshman econ students: Don't confuse levels and changes.

The changes are going to be dramatic and positive, but that's largely because the levels are even more dramatically negative.

Lemme put today's surprising jobs numbers in perspective.

It's simple arithmetic that when millions of businesses are shuttered that employment will tank, and any reversal will lead jobs to spike.

The only real surprise was that the spike started this month, rather than next.

What do today's numbers tell us about the recovery?

Very little, I'm afraid. We knew that a partial unwinding of the shutdown would lead to a partial reversal of job losses.

The important question remains how many jobs will resume, and how quickly. That remains unknown.

Another attempt at perspective.

20 million people lost their jobs, causing unemployment to rise near Depression levels.

Today's data says 1-in-8 of those jobs returned.

But the real question is what'll happen to the rest? Even if half return we're replaying the Great Recession

Correction: private sector forecasters, on average, got today's number wrong by 10 million, not 20m. My bad. (It's still the largest forecast error ever, probably by an order of magnitude.)

A data point on how (un)informative today's payrolls number was.

Stocks rose about 2-3% on the news.

That's the sort of response you would see if payrolls were half a million or maybe a million higher than expected. In reality, today's number was 10 million higher than forecast

Trying to say this in English: If today's jobs numbers caused you to dramatically shift your assessment of the economy, you're doing it wrong.

Markets are reacting to today's jobs report as if it were pretty good news, but it's far short of great news and that seems about right.