Analysis / 03
The price of an hour of work is converging.
In 2008 an hour of labour cost thirteen times more in the most expensive EU country than in the cheapest. In 2024 it was five times. The catching-up is real and steady, and one part of the dataset shows why a number can change without the thing it measures changing at all.
The hourly labour cost is what an employer pays for one hour actually worked: wages and salaries, plus social contributions and other non-wage costs, minus subsidies. Eurostat publishes it in euros for every member state. It is the first number anyone looks at when comparing where to employ people, and it is easy to misread.
The range has narrowed every period since 2012
In 2024 the figure ran from €10.6 in Bulgaria to €54.9 in Luxembourg, around an EU average of €33.5. In 2008 the same two countries stood at €2.6 and €32.3. A logarithmic scale is the honest way to draw this: on it, equal vertical distances are equal percentage changes, and the lines of the low-cost countries visibly climb faster than those at the top.
Bulgarian and Romanian costs tripled between 2012 and 2024; Lithuanian and Latvian costs rose more than two and a half times. At the top, Denmark, Belgium and France grew by about a quarter over the same twelve years. Greece is the outlier in the other direction: an hour of work cost €16.8 in 2008 and €16.8 again in 2024.
The cheapest labour in the EU has been getting expensive fastest
Hourly labour cost in euros, log scale. Grey: the 27 member states. Black: EU-27. Red: selected country. Values exist for 2008, 2012, 2016 and each year from 2020.
Source: Eurostat, lc_lci_lev (labour cost levels, total labour cost, NACE B–S excluding O). Enterprises with ten or more employees. Nominal euros at current exchange rates. 2024–2025 provisional for Austria; 2021–2024 provisional for Finland; Belgium has no 2025 value.
The lower the starting point, the faster the growth
Plotting each country’s growth against where it started makes the pattern explicit. The fitted line slopes down, and the relationship is tight: starting level alone accounts for about three quarters of the variation in growth across the 27 countries (R² = 0.76). The spread of costs, measured as the standard deviation of their logarithms, fell from 0.74 in 2012 to 0.49 in 2024.
Two things sit behind the scatter that the scatter does not show. Costs are in euros at market exchange rates, so part of Sweden’s low figure is a weaker krona, not slower pay growth in kronor. And these are nominal values: the years after 2021 include a burst of inflation that was strongest in central and eastern Europe.
Starting level explains most of the difference in growth
Average annual growth in hourly labour cost, 2012–2024, against the 2012 level (log scale). Each dot is a member state; the line is a least-squares fit.
Source: author’s calculation from Eurostat lc_lci_lev. Fit: growth = 13.0 − 3.22 × ln(2012 cost); n = 27, r = −0.87. Unweighted: Malta counts as much as Germany. A descriptive association in nominal euros, not evidence about productivity or competitiveness.
When the law changes the number
The same dataset splits total cost into wages and everything else. In France and Sweden non-wage costs are almost a third of the total. In most countries the share has been stable for a decade. Then there are Lithuania and Romania, where it fell from above 20% to about 5%.
Employers there did not suddenly stop paying for social insurance. Romania in 2018 and Lithuania in 2019 moved most social contributions from the employer’s side of the payslip to the employee’s, and gross wages were raised to compensate. The total cost of an hour continued on its trend; only the label on part of it changed. Hungary’s smaller fall reflects successive cuts in the employer contribution rate.
It is a useful reminder for anyone comparing countries on “employer on-costs”: the split between wage and non-wage cost is a legal convention, and the total is the figure that travels.
In Lithuania and Romania the non-wage share collapsed by statute, not by economics
Non-wage costs (employers’ social contributions and other costs, net of subsidies) as a share of total hourly labour cost. Ring: 2012. Dot and figure: 2024. Red: change of more than eight points.
Source: author’s calculation from Eurostat lc_lci_lev (total labour cost and wages and salaries): share = 1 − wages ÷ total. The descriptions of the Romanian (2018) and Lithuanian (2019) reforms are background knowledge and are not part of the Eurostat dataset.
Definition. Total labour cost is compensation of employees plus taxes less subsidies, per hour actually worked, for enterprises with ten or more employees in industry, construction and services, excluding public administration. Levels for 2008, 2012, 2016 and 2020 come from the four-yearly Labour Cost Survey; later years are extrapolated by Eurostat with the Labour Cost Index and are revised when a new survey arrives.
What is not adjusted. Nothing here is corrected for prices, purchasing power, productivity or the composition of employment. A cheaper hour is not a more productive euro. Growth rates are compound annual rates between two years, twelve years apart; the dispersion measures give each country equal weight.
Convergence. A negative relationship between starting level and growth is expected when the sample is a set of economies integrating into one market, and regression to the mean exaggerates it slightly. I report the association and do not model its causes.
Data and code
- labour_cost.csv: total cost, wages, non-wage share and flags by country and year
- Raw snapshot: lc_lci_lev.csv
- build_data.py: the calculation, including the fit and the dispersion measures
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.