1. International migration statistics have traditionally focused on immigration, but accurate data on emigration is equally important. Emigration influences population size and structure, labour markets, and long-term demographic change. Nations need to understand the scale and characteristics of people leaving, both to inform policy and to support international cooperation when populations and labour markets are trans-national.

2. Recognizing this, as well as the challenging nature of measuring emigration, the Conference of European Statisticians (CES) established the Task Force on Measuring Emigration in 2023. The Task Force sought to review current practices, highlight innovative methods, and provide recommendations for improving the quality, timeliness and comparability of emigration statistics.

3. To understand the emigration measurement landscape, the Task Force invited countries to participate in a survey gathering information across several themes. Across 11 responding countries, the Task Force found wide variation in how emigration is measured.

(a) Data sources. Countries combine multiple sources such as censuses, household surveys, border-crossing records, population registers, and tax or social security data. Each has strengths and limitations;

(b) New and alternative sources. Several countries are experimenting with tax data, linked administrative datasets and machine learning models to produce faster or more detailed estimates;

(c) Breakdowns produced. Most countries can report emigrants by age, sex, and citizenship. Far fewer can report by country of destination, occupation, or reason for emigration ─ leaving important policy gaps;

(d) Timeliness. Producing official statistics can take from 1 month (Latvia) to 3 years (Canada). Some countries use modelling to release provisional figures while waiting for final data;

(e) Methods used in estimation emigration. Australia, New Zealand, and the UK use statistical models and machine learning to classify migrants provisionally. Canada combines several administrative datasets with adjustment factors to improve coverage and uses modelling across their outputs.

4. Overall, the integration of multiple data sources may improve timeliness without compromising quality.

5. One approach is the use of mirror statistics ─ using immigration records from destination countries to estimate emigration from the origin country. Immigration data are often more reliable because migrants are usually well captured by the arrival jurisdiction. Additionally, where registers are involved, there is a high chance that immigrants register in their new country but they are less likely to deregister at home.

6. Findings include:

(a) Usage: 9 of 11 countries already use mirror statistics in some form; 10 want to expand their use;

(b) Barriers: Inconsistent definitions of migration, lack of timely data, lack of international frameworks for sharing individual-level data and limited data-sharing agreements hinder wider adoption. Limitations in the published disaggregation can also be a barrier;

(c) Case Studies:

(i) Italy provides detailed immigration statistics to help other countries (e.g. Romania, Albania, Poland) improve their emigration estimates;

(ii) New Zealand collaborates closely with Australia, sharing aggregate data established from linked border-crossings data to track flows;

(iii) Poland uses mirror statistics to estimate the number of Polish citizens abroad, revealing underestimation in national data.

7. International cooperation and harmonisation of definitions would enable expanding the value of mirror statistics.

8. Measuring emigration is inherently uncertain. To build trust, countries increasingly focus on evaluation and transparency:

(a) Many compare national data with external sources (e.g. Canada with U.S. records, Albania with Italian statistics, Mexico with the American Community Survey);

(b) Some, like Mexico, New Zealand, and the UK publish uncertainty ranges or confidence intervals for their estimates, particularly for provisional estimates;

(c) Others, like Australia and Italy, uses regular population censuses or integrated demographic registers to validate estimates;

(d) Clear communication of limitations, revisions, and methodological choices helps users interpret migration data responsibly.

9. The task force has made the following key recommendations:

(a) Use multiple sources: No single dataset captures emigration well; combining registers, surveys, tax data, and border data gives a fuller picture;

(b) Explore new data: Administrative and “signal” data (tax, health, education) can improve timeliness; future work may include mobile phone, social network or financial data;

(c) Leverage mirror statistics: Share immigration data across borders, starting with key partner countries, to strengthen emigration estimates;

(d) Link datasets: Integrating sources can improve accuracy and enable more detailed breakdowns;

(e) Report on uncertainty: Publishing confidence intervals, revisions, or quality notes helps build trust with users;

(f) Promote international collaboration: Harmonised definitions and data-sharing agreements are essential for consistent global migration statistics.