A plain-English guide to the Beveridge curve.

Every JOLTS release lands in two parts: the headline numbers (openings, hires, quits, layoffs) and the curve that economists actually look at. The curve is the Beveridge curve, named for the British economist William Beveridge, and it plots the unemployment rate on the horizontal axis against the job-vacancy rate on the vertical. The curve has been the cleanest single picture of US labour-market tightness since 2020. This is the explainer on what it is, what its shape tells you, and where today's JOLTS print places the US economy on it.

The shape, and what it means

Plot every monthly observation of US unemployment (u) against the monthly job-openings rate (v). The points trace out a downward- sloping curve. Higher unemployment goes with lower vacancies; lower unemployment goes with higher vacancies. That is what you would expect from a frictional labour market: when the economy slows, firms post fewer openings and more workers are out of work. When the economy booms, openings rise and unemployment falls.

The downward slope is the easy part. The interesting question is where on the curve the economy sits, and whether the curve itself has shifted. Shifts in the curve are diagnostic. An outward shift, more vacancies for the same level of unemployment, says matching efficiency has deteriorated. An inward shift says matching has improved.

The picture

The 2022 outward shift, and why it mattered

Through 2020 the US Beveridge curve traced out a fairly stable hyperbola: the product u·v sat around 15, meaning a 5 percent unemployment rate paired with a 3 percent vacancy rate and a 3 percent unemployment rate paired with a 5 percent vacancy rate. The relationship was tight enough that economists used it as a coincident indicator of labour-market slack.

In 2021 the curve shifted outward by historical standards. Vacancies climbed to almost 7 percent at unemployment near 3.5 percent; the historical curve said vacancies of about 4 percent were normal at that unemployment rate. That outward shift was the empirical fact behind the 2021-2022 "labour shortage" narrative. Matching efficiency had deteriorated: firms wanted workers, workers wanted jobs, but the rate at which the two were finding each other was lower than the prior fifty years of data implied.

Three explanations got the most academic traction. First, sector composition: vacancies were concentrated in healthcare and leisure-hospitality while unemployment was spread across sectors. Second, geographic mismatch: COVID-era remote work and moving disrupted the location side. Third, the surge in "poaching vacancies", openings posted to hire workers already employed elsewhere, which the St. Louis Fed has documented rose to as much as 80 percent of all openings during the period.

What u* = √(uv) means

The Blanchard-Diamond formulation of the curve gives a clean shorthand for the "natural" rate of unemployment, written u*, which the labour market would settle at in a steady state. The formula is:

u* ≈ √(u · v)

where u and v are measured as fractions of the labour force. If u = 4.3 percent and v = 4.3 percent, then u* ≈ √(0.043 · 0.043) ≈ 4.3 percent. When u and v are roughly equal, you are sitting on the natural rate. When v > u (vacancies higher than unemployment), the economy is overheating; when u > v, it is slack.

The vacancy-to-unemployment ratio v/u is the desk's shorthand for the same idea. The Fed cares about v/u because it tracks wage pressure with a six-to-twelve-month lead. Above 1.5, you are in clearly tight territory; below 1.0, in clearly slack territory. The ratio peaked at roughly 2.0 in 2022, the highest in the JOLTS history, and has fallen to roughly 1.0 by early 2026. That is the data-not-dots thesis in a single number.

Where today's print sits

The April JOLTS release pushed openings to 7.6 million, against an unemployment rate that held at 4.3 percent. That places today's point at roughly u = 4.3 percent, v = 4.3 percent on the curve, right at the natural-rate intersection. The point sits close enough to the pre-pandemic fitted curve that the "outward shift" can be considered largely reversed.

The interesting wrinkle, the one the headline misses, is that hires fell to 5.1 million while openings rose. That is a matching story: firms are advertising more, but the realised flow into work is slowing. In Beveridge-curve terms, that does not move the curve much (matching frictions are captured by the curve's position relative to the trend), but it does say that the labour market is "frozen" rather than tight or slack. Frozen-market regimes are notable because they can suddenly snap to either side: a few firms cancelling openings turns a tight market into a slack one in two or three months.

What the curve does not tell you

1. Causation. The Beveridge curve is a reduced-form summary of matching, not a structural model. It tells you where the economy is, not why.
2. Composition. A single (u, v) point hides the sectoral and geographic distribution. The St. Louis Fed's decomposition by vacancy type is essential when the headline number is ambiguous.
3. Lead-lag. The vacancy series tends to lead employment by one to two quarters in turning points. Reading a single month's point as if the labour market is in equilibrium misses the dynamics.
4. Wage pass-through. v/u tracks wage growth with a long lead, but the relationship is not mechanical. Wage formation depends on the bargaining structure, union density, and minimum-wage regimes; the same v/u in two countries produces different wage prints.

Related reading

• Today's analysis applies this directly: JOLTS sent a mixed signal. The dollar held its ISM gains.
• The framework reading the labour data into a one-number summary: A plain-English guide to economic surprise indices
• The original thesis the curve supports: The dollar is weaker on the data, not the dots
• How wage pressure connects to inflation gauges: A plain-English guide to PCE vs CPI