Analysis / 01
Europe’s hiring squeeze did not end. It moved south.
In 2025 the EU had roughly the same balance of vacancies to unemployment as in 2019. That flat average conceals thirteen national labour markets that became tighter and thirteen that eased, and the two groups sit in different parts of the map.
A Beveridge curve plots two rates against each other: the share of the labour force that is unemployed and the share of posts that are vacant. When an economy recovers, the point travels up and to the left: fewer people looking, more jobs open. Where a country sits on that plane says more about how hard it is to hire there than either number does alone.
I followed that point for every EU country with usable data over sixteen years. The question was simple: after the post-pandemic hiring rush, did European labour markets return to where they were, and did they all do it in the same way?
The EU cooled through vacancies, not through unemployment
From 2013 to 2019 the EU-27 moved along a clean curve: unemployment fell from 11.6% to 6.8% while the vacancy rate rose from 1.2% to 2.3%. The pandemic interrupted that for a single year. By 2022 the vacancy rate had reached 3.0%, more than the 2.7% that a curve fitted to the previous decade would give at that level of unemployment.
What happened next is the interesting part. Between 2022 and 2025 the vacancy rate fell by almost a third, to 2.1%, and unemployment did not rise. It edged down to 6.0%. The 2025 point lies below the pre-2020 curve, which would put vacancies near 2.8%. Employers withdrew openings; they did not, in aggregate, shed people.
The EU-27 travelled out along its curve, overshot in 2022 and came back underneath it
Job vacancy rate against unemployment rate, one point per year. Black: 2010–2019. Red: 2019–2025. Choose a country to see its own path.
Source: Eurostat, une_rt_a (unemployment rate, ages 15–74) and jvs_a_rate_r2 (job vacancy rate, NACE B–S, all enterprise sizes, annual average). Grey curve, EU-27 only: least-squares fit of log vacancy rate on log unemployment rate, 2010–2019 (n = 10, R² = 0.87); a description of that decade, not a structural estimate.
The same sixteen years, country by country
The panels below use logarithmic scales on both axes, so equal distances are equal proportional changes and countries with very different levels can share one frame. The two faint diagonals join points with the same ratio of vacancy rate to unemployment rate, 0.1 and 1. A path that climbs towards the top-left corner is a labour market getting tighter.
Northern and central markets turned back after 2022; several southern ones kept climbing
Unemployment rate (horizontal) and vacancy rate (vertical), both on log scales, 2010–2025. Grey: to 2019. Red: 2019–2025, ending at the red dot. Every panel spans unemployment of 1.7–32% and vacancy rates of 0.3–8%. Ordered from tightest to slackest in 2025.
Source: Eurostat, une_rt_a and jvs_a_rate_r2. Denmark is not published for this breakdown. Series start later for France and Croatia (2012), Italy (2016) and Malta (2017); France has no 2020 value and Greece no 2013 value.
Who tightened and who eased
Dividing one rate by the other gives a single index of tightness. For the EU-27 it was 0.34 in 2019 and 0.35 in 2025. Almost nothing moved. Underneath, the index rose in thirteen countries and fell in thirteen.
The risers are mostly in the south: Greece, Cyprus, Italy, Spain, Malta, Portugal, Croatia and Slovenia, joined by Bulgaria, Lithuania, Ireland, France and the Netherlands. Greece is the largest change. Its unemployment rate halved, from 17.9% to 8.9%, while its vacancy rate almost tripled, from 0.6% to 1.7%. It is still a slack market by northern standards, but it is no longer in a different category.
The fallers are central and northern: Czechia, Finland, Romania, Estonia, Germany, Sweden, Hungary, Latvia, Luxembourg, Poland, Austria, and marginally Belgium and Slovakia. Germany’s index fell from 1.14 to 0.68. Czechia, which in 2019 had three vacant posts for every point of unemployment, shows the steepest fall; part of its 2025 drop coincides with a break in the vacancy series, and the 2024 value (1.27) is the safer comparison.
In 2025 only the Netherlands and Malta still had a vacancy rate above their unemployment rate.
A flat EU average, made of opposite national movements
Ratio of job vacancy rate to unemployment rate, 2019 (ring) to 2025 (arrowhead), log scale. Red arrows: tighter. Blue arrows: looser.
Source: author’s calculation from Eurostat une_rt_a and jvs_a_rate_r2. Italy and Malta are included from their first available year; Denmark is absent. 2025 vacancy rates are provisional for Bulgaria, Spain, Hungary, Ireland, Italy, Malta and the Netherlands, and Czechia has a series break in 2025.
What this does and does not show
It shows that the distance between Europe’s tight and slack labour markets has narrowed since 2019, and that the EU-wide cooling after 2022 took the form of fewer openings instead of more unemployment. For anyone planning where to recruit, the old shorthand of a tight north and a slack south is a poorer guide than it was six years ago.
It does not show why. Annual national rates cannot separate weaker demand from better matching, changes in the mix of sectors, migration, or changes in how firms advertise posts. The pattern is an observation, and I have not tested any explanation for it here.
The two rates have different denominators: unemployment is a share of the labour force, the vacancy rate is vacancies divided by occupied posts plus vacancies. Their ratio is therefore an index, not a count of vacancies per unemployed person. It is comparable over time within a country and, more cautiously, between countries, because vacancy surveys differ in coverage. Eurostat flags the French vacancy series and the recent Spanish and French unemployment series as differing in definition.
Vacancy rates cover industry, construction and services (NACE B–S), enterprises of all sizes. Nothing is seasonally adjusted because the data are annual averages. Missing values are left missing; no series was interpolated or extended.
Data and code
- beveridge.csv: the table behind all three figures, with Eurostat flags
- Raw snapshots as retrieved: une_rt_a.csv, jvs_a_rate_r2.csv
- build_data.py: every transformation, in about forty lines for this piece
Eurostat data are reused under the European Commission’s reuse policy (CC BY 4.0), with the source acknowledged. Calculations and any errors are mine.