Hired, or not fired
The 2020 unemployment shock was half again as large as 2009's and healed in under a third of the time. Why — in hiring, firing, and who had a job to go back to.
In April 2020 the US unemployment rate hit 14.8% — half again as high as the 10.0% peak of the Great Recession. It was back within half a point of where it started in 20 months. 2009's took 69.
The headline poses that puzzle and cannot answer it, because a rate is a stock and says nothing about how people arrived in it. The panels below open it up: 2008–09 was a hiring collapse, slow to repair, and 2020 was a firing spike among people who were mostly coming back. Then the quieter question underneath both — a labour market that now moves less in either direction, and what that costs the people in it.
Every figure here is sourced in the fact sheet, which also carries what this data cannot settle. Data: BLS via FRED — Current Population Survey flows, JOLTS, and the Atlanta Fed Wage Growth Tracker.
Two recessions, two different events
The same two recessions from the employer's side, as monthly rates (JOLTS). These are published rates, not our arithmetic.
2008–09 was a hiring collapse. The hires rate fell from 3.7% (December 2007) to 2.8% in June 2009 — the lowest in the whole series — and did not get back to 3.7% until September 2014 — 81 months after that 3.7% starting point. Layoffs barely moved: 1.4% to a peak of 2.0% in April 2009. Almost nobody extra was being fired. Almost nobody was being hired either, for nearly seven years.
2020 was the inverse, and it is the tall spike that sets this axis. Layoffs and discharges went to 8.6% in March 2020, six times normal, and were back to 1.6% by May. Hiring spiked too — 6.1% in May 2020 — because the same people were going back.
And underneath both, the line has been getting quieter. Hires are 3.2% (July 2026), near the series low; quits are 1.9%, down from a 3.0% peak in November 2021. Fewer people are being hired, and fewer are leaving.
@andrewdun · 6d ago
On layoff, or out of work
Why the 2020 spike reversed so fast, in one comparison: how many of the unemployed were on temporary layoff — people with a job to go back to — against how many were unemployed at all.
October 2009: 1.67 million of 15.35 million — 10.9%. Nine in ten had no job waiting.
April 2020: 18.05 million of 23.08 million — 78.2%. By December 2020 it was back to 28.4%, and it sits at 11.9% today.
That is the fingerprint. A recall is not a job search, and a stock of people on layoff drains at a speed a stock of job-seekers never does. What this panel does not establish is why the recalls happened — whether closures were time-limited by design, or payroll support kept the attachment alive. The fact sheet names that as needing a source rather than an inference, and does not make one.
@andrewdun · 6d ago
April 2020, in one frame
The two monthly flows between employment and unemployment, in people. In April 2020, 17.5 million went from employed to unemployed while 1.3 million went the other way — a month with no precedent in a record that starts in 1990. By May the direction had flipped: 7.8 million found work, the highest in the series, as layoffs were recalled.
By August 2026 both are back to about 1.5 and 1.7 million a month.
Two things this panel deliberately does not do. It does not run back to 1990: these are levels, and through 2008–09 the job-finding level ROSE even as the job-finding rate collapsed, simply because the pool of unemployed people had doubled — so on a long window the chart would contradict the argument the panels above make. And the rate decomposition the fact sheet turns on (58–66% of each pre-2020 rise from job-finding, 98% of 2020's from separations) cannot be drawn here at all: those are ratios to a lagged stock, and this platform computes sums. Read them in the fact sheet.
October and November 2025 are missing from every flow series — one absent survey month costs two months of flows, because a flow joins two consecutive months.
@andrewdun · 6d ago
What the quiet costs
If a labour market with less movement in both directions were simply a calm one, staying put would be free. It is not. The Atlanta Fed tracks median wage growth separately for people who changed jobs and people who stayed. Of the 355 months since 1997 the switchers are ahead in 324, level in 12 and behind in 19 — ahead by 0.73 percentage points on average and by as much as 2.2, and never behind by more than 0.6.
As of August 2026 switchers are at 4.4% against stayers' 3.6% — a gap of 0.8 points, still there while hiring sits near a series low.
That is the cost of the quiet, and it lands on whoever is not moving. It is a gap, not a cause: people who change jobs are not a random sample of workers, so some of this premium is who switches rather than what switching pays.
@andrewdun · 6d ago
Discussion of this dashboard
Threads here are about the dashboard as a whole — its framing, what's missing, whether the constituent indicators tell a coherent story. Per-indicator discussion lives under each panel above.
Log in to start a discussion about this dashboard.
No discussions yet. Be the first.
The published US unemployment rate. Two peaks, and the taller one is the shorter.
2008–09: from 4.7% (October 2007) to 10.0% in October 2009, and not back within half a point of that starting level until July 2015 — 69 months.
2020: from 3.5% (February 2020) to 14.8% in April 2020 — and back to 3.9% by December 2021, 20 months. A shock half again as large, repaired in under a third of the time.
Nothing in this line explains that. It counts people who are unemployed; it does not say whether they arrived because nobody was hiring or because everybody was fired, and those two do not heal alike.
@andrewdun · 6d ago