Guide

Beveridge curve explained

Harbor Manufacturing's HR analytics team treated a record job-openings count as proof that hiring would normalize within two quarters. Open roles climbed 34% while headline unemployment stayed near 3.8% — a combination that looked like ordinary cyclical tightness. Wage offers rose, referral bonuses doubled, and still time-to-fill stretched from 28 days to 51. The mistake was reading vacancies in isolation. Plotting JOLTS job openings against the unemployment rate revealed an outward shift in the Beveridge curve: the same unemployment rate now paired with far more vacancies than the pre-2020 locus predicted. Matching efficiency — how quickly workers and employers find each other — had deteriorated. Refactoring the hiring forecast around Beveridge dynamics cut over-hiring spend by 18% while improving retention in skilled trades.

The Beveridge curve is the empirical relationship between job vacancies and unemployment. Named after British economist William Beveridge, it traces a downward-sloping curve in vacancy-unemployment space: when unemployment is high, vacancies tend to be low (slack labor market); when unemployment is low, vacancies rise (tight market). Movements along the curve reflect cyclical tightening and loosening; shifts of the curve signal changes in matching efficiency, labor force participation, or structural mismatch. Central banks, staffing firms, and corporate workforce planners use it to separate “hot but functioning” labor markets from broken matching regimes. This guide covers the mechanics, vacancy-rate formulations, outward and inward shifts, links to the Phillips curve and Sahm rule, the Harbor Manufacturing refactor, a technique decision table, pitfalls, and a production checklist.

What the Beveridge curve is

The Beveridge curve plots job vacancies (or vacancy rate) on one axis against unemployment (or unemployment rate) on the other. In a typical U.S. chart using BLS data:

Vacancy rate = job openings divided by labor force (or sometimes openings divided by employed plus openings).
Unemployment rate = unemployed divided by labor force (U-3).

The curve is downward-sloping because vacancies and unemployment are, in normal times, substitutes in the matching process: many unfilled jobs coexist with few unemployed workers only when matching is fast and skills align; many unemployed workers coexist with few vacancies in recessions when hiring freezes.

Economists formalize matching with search-and-matching models (Mortensen-Pissarides tradition): hires flow from the stock of unemployed and employed searchers meeting vacancy postings. The Beveridge curve is the reduced-form picture of that process. The Beveridge curve shift — outward or inward — captures whether the economy needs more vacancies per unemployed worker to sustain a given unemployment rate, i.e., lower matching efficiency.

A related summary statistic is the vacancy-to-unemployed ratio (V/U): job openings divided by number of unemployed persons. Values above 1.0 imply more advertised openings than unemployed workers — tight by definition, but not necessarily efficient if openings stay unfilled for months.

Movement taxonomy

Movement along the curve (cyclical)

During expansion, unemployment falls and vacancies rise — you travel northeast to southwest along a stable curve. In recession, unemployment jumps and vacancies collapse — movement back toward the high-unemployment, low-vacancy corner. Portfolio and HR teams should label these moves “cycle” unless the entire cloud of recent points sits off historical loci.

Outward shift (worse matching efficiency)

Higher vacancies at the same unemployment rate than history predicts. Causes include skill mismatch (tech vs manufacturing gaps), geographic immobility, sectoral reallocation after shocks, reduced search intensity, childcare and caregiving constraints, longer interview processes, and zombie postings that employers never close. The post-2020 U.S. experience featured a pronounced outward shift: unemployment did not rise to clear excess demand for workers.

Inward shift (better matching efficiency)

Fewer vacancies needed per unemployed worker — faster hiring platforms, better training pipelines, improved labor market information, or demographic inflows that match open roles. Inward shifts can accompany productivity gains but may also reflect weaker worker bargaining if vacancies are understated.

Looping and hysteresis

Real cycles trace loops rather than static points: recessions can leave lasting scars (long-term unemployment, skill atrophy) that shift the curve outward even after GDP recovers. Distinguish temporary loops from permanent shifts with multi-year panels, not single-month prints.

Structural vs cyclical unemployment decomposition

Outward shifts raise estimates of structural unemployment — the NAIRU-compatible floor below which inflation accelerates without sustainable matching. Cyclical unemployment is the gap between actual unemployment and that structural rate. Beveridge analysis informs both sides of the Phillips-curve slack debate.

How to compute and monitor it

Primary U.S. inputs come from BLS:

JOLTS job openings (monthly, lagged ~5 weeks) — see our JOLTS guide.
Household Survey unemployment (monthly Employment Situation, same release as payrolls).
Labor force for rate denominators (CPS).

Construction steps:

Compute vacancy rate = openings / (employed + openings) or openings / labor force; pick one definition and keep it consistent in backtests.
Plot vacancy rate vs U-3 unemployment for each month; overlay a rolling 12-month cloud and compare to a pre-shock baseline (e.g., 2015–2019).
Track V/U ratio and its z-score vs 20-year history.
Estimate matching efficiency a from a Cobb-Douglas matching function m = a u^α v^1−α if you need a scalar for models; otherwise visual shifts suffice for dashboards.
Segment by industry (JOLTS sector tables) — manufacturing Beveridge loops can diverge from leisure and hospitality.

Pair monthly Beveridge charts with weekly initial jobless claims for early cyclical turns and with quits rate from JOLTS for worker confidence. When vacancies fall while unemployment is still low, watch for Sahm-style deterioration in the months ahead.

How it compares to other labor indicators

Unemployment rate alone — Misses unfilled demand; can look “normal” while vacancies explode outward.
Payrolls (NFP) — Net hiring outcome; lags matching friction visible in vacancies.
JOLTS quits rate — Worker-side tightness; complements employer-side openings.
Phillips curve / wage growth — Price consequence of slack; Beveridge diagnoses why slack measures misbehave.
Sahm rule — Recession trigger on unemployment momentum; Beveridge shifts can warn that re-tightening will be inflationary before Sahm fires.
Prime-age participation — Explains outward shifts when workers sit outside unemployment counts.

Production macro sleeves should treat Beveridge position as a regime modifier on Phillips and Taylor-rule forecasts, not a standalone equity signal.

Harbor Manufacturing refactor

Harbor's workforce planning model previously keyed headcount adds to revenue growth and a single unemployment threshold. After the 2022–2024 outward shift, that rule over-hired in plants where local V/U exceeded 1.4 while national unemployment looked benign. The refactor introduced:

Regional Beveridge panels — state-level openings from JOLTS where available, proxied by industry mix elsewhere.
Matching-efficiency flag — when current (v,u) sits >1 standard deviation outside the 2015–2019 locus, wage budgets escalate but headcount caps tighten (hire quality over quantity).
Time-to-fill KPI — internal ATS data overlaid on national curve; divergence signals company-specific mismatch.
Training pipeline trigger — outward shift plus rising ECI opens apprenticeship slots instead of endless external reqs.

In the skilled CNC machinist cohort, the new policy reduced open-req aging from 74 to 41 days without matching the prior cycle's 22% wage spike. Macro risk linked the same Beveridge flag to a higher structural unemployment assumption in their Okun's-law GDP tracker, avoiding over-optimistic output gaps.

Technique decision table

Approach	Best when	Blind spots	Data burden
Unemployment-only slack model	Deep recession, obvious slack	Misses vacancy-heavy tightness	Low
V/U ratio monitoring	Quick tightness dashboard	Ignores curve shifts; level-dependent	Low
Beveridge scatter + baseline locus	Separating cycle vs matching breaks	Needs long history; COVID distortions	Medium
Estimated matching function (MPV)	Central-bank-style forecasting	Parameter instability	High
Industry-level Beveridge panels	Sector rotation, manufacturing HR	Noisy small samples	Medium
Beveridge + Phillips joint system	Wage-inflation risk committees	Model complexity; flat Phillips eras	High

For corporate workforce planning, the scatter-plus-baseline approach beats black-box ML: executives can see why hiring is hard, not just that a model scored “red.”

Common pitfalls

Mixing vacancy definitions — JOLTS openings are stock positions, not unique reqs; deduplicate internal data before comparing.
Ignoring participation — Workers outside the labor force do not appear in unemployment but affect matching.
Single-month outliers — Pandemic months blew up the curve; use 3-month moving averages for presentation.
Assuming outward shift is permanent — Some reallocation shocks normalize; test with rolling breakpoint analysis.
Zombie postings — Employers leave reqs open to pipeline; inflates vacancies without true demand.
Geographic aggregation — National curve hides local mismatches (Austin tech vs Midwest manufacturing).
Confusing tightness with efficiency — High V/U can mean strong demand or broken matching; curve position disambiguates.
Exporting U.S. loci abroad — Euro-area and EM vacancy series have different coverage; rebuild baselines locally.

Production checklist

Pull monthly JOLTS openings and CPS unemployment; align release calendars.
Compute vacancy rate and V/U with documented denominators.
Plot scatter against a pre-shock baseline locus (e.g., 2015–2019).
Label cyclical moves along the curve vs outward/inward shifts.
Track 3-month moving averages to smooth pandemic-era noise.
Segment by industry when sector hiring drives your exposure.
Pair with quits rate, claims, and payroll revisions for confirmation.
Feed Beveridge regime flags into Phillips and wage forecasts.
For corporates, overlay internal time-to-fill and req aging on national curve.
Stress-test HR plans when points sit >1 sigma outside historical locus.
Document zombie-posting policy adjustments in vacancy data cleaning.
Review structural unemployment assumptions annually after large shifts.

Key takeaways

The Beveridge curve plots vacancies against unemployment — downward-sloping in normal times, with position revealing labor market tightness.
Outward shifts mean worse matching efficiency: more vacancies coexist with the same unemployment than history predicted.
V/U ratio is a useful dashboard stat but cannot replace curve-shift analysis.
Harbor Manufacturing cut over-hiring and improved time-to-fill by treating Beveridge shifts as structural hiring constraints, not noise.
Combine Beveridge monitoring with Phillips slack, Sahm recession triggers, and industry JOLTS detail for a complete labor read.