Methodology: how we use official data

Sources

We use these sources. Each page names the dataset and the year behind every figure.

• International Labour Organization (ILOSTAT): median monthly earnings of employees by education level and by occupation group, and unemployment rates by education level. Licence CC BY 4.0.
• OECD, Education at a Glance: earnings of adults by education level relative to upper secondary education, and earnings and employment rates by field of study. Licence CC BY 4.0.
• World Bank, World Development Indicators: population, GDP per person at purchasing power parity and tertiary enrolment. Licence CC BY 4.0.
• US Bureau of Labor Statistics, Occupational Employment and Wage Statistics (OEWS): national employment and annual wage percentiles by occupation. Public domain.
• UK Department for Education, Graduate labour market outcomes (LEO): earnings of UK graduates by subject one, three, five and ten years after graduating. Open Government Licence.
• European Commission, ESCO: official occupation names, descriptions and essential skills in 28 languages.

Rules we apply

For each country and measure we use the latest year that has the values a table needs, from 2015 onwards. We use figures for both sexes combined. We exclude values that the source flags as unreliable and we show the source's own notes, such as breaks in a series. We never fill gaps with estimates: a missing value is shown as a dash.

We publish a country page only when at least two official data blocks are available for it, or one block for a popular study destination. Countries without enough official data are not shown.

What the main measures mean

Tertiary or advanced education means short-cycle tertiary, bachelor's, master's and doctoral levels (ISCED 2011 levels 5 to 8). The degree premium is the percentage difference between the median monthly earnings of employees with advanced education and the median for all employees in the same country and year.

PPP$ (purchasing power parity dollars) adjust for differences in prices between countries, so they are the fairest way to compare pay across countries. Local-currency figures are shown as the source publishes them.

Unemployment rates are for adults aged 25 and over unless a table says young people aged 15 to 24.

Careers and occupation groups

Each career is matched to an ISCO-08 occupation code through ESCO, and to a US SOC 2018 code where a clean match exists. Most countries do not publish pay for single job titles, so the ILO earnings tables on career pages are for the whole ISCO-08 major group the career belongs to, for example all professionals. We label these tables clearly. US pay on career pages is for the matching SOC occupation only.

Degree subjects and fields of study

Each subject is classified using UNESCO's ISCED-F 2013 fields of education. OECD outcome data is published for broad or narrow fields rather than individual subjects, so a subject page shows the field it belongs to and says so. UK figures come from the matching UK subject code in the LEO data.

Limits you should know about

Official statistics are the best evidence available, but they have limits.

• Averages hide wide variation between individuals, universities, regions and industries.
• Earnings data covers employees; the self-employed are usually excluded.
• Years differ between countries, and some surveys change their methods over time.
• A high degree premium can reflect a shortage of graduates rather than the value of a particular degree.
• Data describes the past. Labour markets change, especially in fast-moving fields.

Updates

We refresh the data when the sources publish new releases and show the update date on every page. Our data files, the scripts that build them and the full data are available from the Open data page.