A deadline almost nobody met
Directive (EU) 2023/970 obliged all 27 member states to write pay transparency into national law by 7 June 2026. Among its provisions: an employer must tell an applicant the pay level or range before the first interview, normally in the vacancy itself, and may not ask what the candidate earned before.
Only four states transposed in full by the deadline — Slovakia, Italy, Lithuania and Malta. Three more were partial. Ten had a draft; ten had nothing. Germany, Spain and Sweden were in the last group. The Netherlands, Sweden, Czechia and Denmark have since confirmed dates in January 2027. The Commission declined to extend the deadline and has signalled infringement proceedings.
What the job ads actually say
That is the law on paper. Here is the market. Of the 15 EU cities in our aggregate today, eight display no advertised median salary at all: Dublin, Berlin, Munich, Lisbon, Milan, Stockholm, Copenhagen and Brussels. Seven display one.
The interesting part is that the two lists do not line up. Milan shows no advertised pay, and Italy is one of the four states that transposed on time. Madrid and Barcelona do show pay, and Spain had not even published a draft. Whatever is driving disclosure in these postings right now, it is not the legal status of the member state. Law changes the obligation; it has not yet changed the copy.
| City | Advertised median | National status |
|---|---|---|
| Milan | none shown | transposed on time |
| Dublin | none shown | partial |
| Brussels | none shown | partial |
| Berlin | none shown | delayed, no draft |
| Munich | none shown | delayed, no draft |
| Stockholm | none shown | delayed to Jan 2027 |
| Copenhagen | none shown | delayed to Jan 2027 |
| Lisbon | none shown | — |
| Madrid | EUR 28,500 | delayed, no draft |
| Barcelona | EUR 37,500 | delayed, no draft |
| Lyon | EUR 32,000 | — |
| Amsterdam | EUR 40,500 | delayed to Jan 2027 |
| Paris | EUR 50,000 | — |
| Vienna | EUR 65,000 | — |
The same silence, in remote work
Pay is not the only thing the ads leave out. Across the same 3,927 openings, 15% are flagged remote-eligible, and that average conceals two markets: 17 cities sit at exactly zero, 14 at half or more.
An independent reading points the same way. US telework has held near 24% for two years while job postings continue to tilt on-site — employers advertising for the office while their own staff work otherwise. Two different instruments, one finding: the posting is a weaker description of the job than we treat it as, on pay and on place alike.
Meanwhile, demand rotated hard toward engineering
The macro backdrop is cautious. US employers added 44,000 jobs in July, against about 75,000 expected and 98,000 in June. Postings are down year on year across major markets — roughly 23% in the US and 25% in France. Global unemployment is forecast flat at 4.9%, with the ILO putting the wider jobs gap at 408 million.
Our own pool moved the other way, and its composition moved faster than its size. Openings rose to 3,927, up 1.6% in a day. But Python went from 26 to 40 mentions and TypeScript from 10 to 17, while Amazon Web Services and JavaScript entered the top eight and Photoshop and Illustrator dropped out of it entirely. Anthropic and Rippling entered the top five hirers; Databricks and Block left. This is the two-speed market in one snapshot: broad caution, narrow and intense demand.
- Skills named in listings, 6 to 7 August 2026
- Leadership — 69 → 52down
- Python — 26 → 40up
- Communication — 40 → 31down
- TypeScript — 10 → 17up
- Project Management — 14 → 16up
- Amazon Web Services — 15new to the top eight
- Accounting — 10 → 15up
- JavaScript — 13new to the top eight
A correction, and what it is worth
Yesterday this desk reported that Britain's advertised medians were inverted — London lowest of the three UK cities at GBP 40,000, Birmingham highest at GBP 46,500. Today the same pages read London GBP 54,500 and Birmingham GBP 30,000. The finding did not weaken; it reversed, in twenty-four hours.
The explanation is the subject of this piece. A median computed over the few listings that disclose pay moves violently when that handful changes. So we are retiring day-over-day salary comparisons: this desk will report advertised pay as a level with its disclosure rate attached, not as a trend. When eight of fifteen European cities publish nothing, the honest headline is about the silence, not about the number.
Every market figure here is a page you can open. Start with the market by city and read your own.
The markets in this piece, live
Earlier editions
- · 12 min readNew Zealand has more under-25s but fewer in workFrom July to September 2026, 353,000 of the 690,200 people aged 15 to 24 in New Zealand were employed, against 393,000 of 639,700 in the same quarter of 2022. That is 40,000 fewer employed in a population 50,500 larger, and an employment rate of 51.1% against 61.4%. Over the same four years the count of unemployed people of that age is 34,600 higher and the count outside the labour force is 55,800 higher. Among people aged 25 to 54 the employment rate went from 85.9% to 84.0%. All of it comes from one file, the OECD's infra-annual labour statistics for New Zealand, and none of it says what any young person outside the labour force is doing, or why.Read the analysis
- · 15 min readCosta Rica's unemployment fell 44% while its employment rate fellBetween July to September 2019 and July to September 2026 the number of unemployed people in Costa Rica fell from 278,108 to 154,736, which is 123,372 fewer, and the unemployment rate fell from 11.4% to 6.5%. Over the same seven years the share of people aged 15 and over who were employed fell as well, from 54.7% to 51.3%. The count of the employed is 47,716 higher, the population aged 15 and over is 359,112 larger, and the count of people outside the labour force is 434,768 higher. All of it comes from one household survey, the Encuesta Continua de Empleo of the Instituto Nacional de Estadística y Censos, and none of it says where any unemployed person went.Read the analysis
- · 15 min readHong Kong has 137,300 more women in work than menFrom April to June 2026 Hong Kong counted 1,891,700 women and 1,754,400 men as employed, a lead of 137,300 for women. The same survey publishes a second count that excludes foreign domestic helpers, and in it the order reverses: 1,541,100 women and 1,749,300 men, a lead of 208,200 for men. The two counts differ by 350,600 women and 5,100 men, which is this edition's subtraction and not a published number of helpers. Women's labour force participation rate was 52.2% in the first count and 47.3% in the second, and men's was 61.1% in both. Both counts come from the Census and Statistics Department, foreign domestic helpers are employed under its definition, and neither count is a correction of the other.Read the analysis



