194. The accuracy and reliability of long-term emigration estimates are critical for understanding population dynamics. In addition to ensuring the production of good quality emigration statistics, evaluation also identified the limitations of statistics to help prioritize research and development and to help users interpret the statistics.

195. This section explores the practices of the 11 NSOs in assessing the quality of their long-term emigration estimates. NSOs employ various methods to evaluate the quality of their estimates, with approaches differing significantly across countries. While some NSOs actively compare their estimates to external data sources for validation, others rely solely on internal methodologies.

196. It also highlights the notable and various efforts to quantify uncertainty in their estimates using external data sources, such as diaspora statistics, mirror data, and household surveys, to identify discrepancies and enhance accuracy.

197. Many NSOs compare their estimates of long-term emigration to other data sources to evaluate the quality of their estimates. These include Albania, Australia, Canada, Hungary, Latvia, Mexico, New Zealand, Poland and the United States. In contrast, other NSOs, such as Italy and the United Kingdom do not use external data sources for comparison.

198. Albania compares its long-term emigration estimates annually with mirror statistics from Italy and Eurostat, using indicators such as citizenship, age, and sex. These comparisons, along with household surveys, help identify discrepancies and improve the accuracy of emigration estimates. The data sources are used individually or combined to adjust the estimates using an indirect method, which is also applied to adjust the back series for the intercensal period. Additionally, population projections and mirror statistics are used to model migration by age, ensuring the reliability of the estimates. The comparisons are conducted every year to evaluate and enhance the quality and accuracy of emigration estimates by identifying discrepancies and inconsistencies in the data.

199. Australia assesses its emigration estimates using census counts and the population rebasing process. Every five years, Australia conducts a census to rebase its population estimates, which helps correct any errors that may accumulate during the intercensal period. Since international migration, including emigration, is a key driver of population growth, any inaccuracies in migration data could lead to larger intercensal differences. However, historical data shows that absolute intercensal differences have averaged only around 0.25% of the population across the last seven censuses, highlighting the high reliability of migration estimates. Additionally, the primary source of migration data comes from border crossing information, which is rigorously quality-assured before being used in the estimates. Furthermore, preliminary migration estimates are continuously compared with final, directly calculated data as it becomes available, ensuring that any discrepancies are quickly identified and addressed.

200. Canada compares its estimates with multiple international and Canadian data sources, including U.S. visa data, the American Community Survey (ACS), Canada Post's movers file, the Registration of Canadians Abroad (ROCA), tax data, the residual method, and the Census Undercoverage Study. Trends, levels, and potential biases are regularly analysed using Power BI dashboards, while considering the concepts, the strengths and the limitations of each source. Canada also provides a regularly updated methodological guide. Specific studies are also conducted to evaluate different elements of emigration statistics, such as the accuracy of the adjustments made to consider the impacts of the COVID-19 pandemic.

201. To assess the accuracy of its emigration estimates, Hungary compares its macro-level data with mirror statistics from countries of destination. By evaluating the quality of its data against these external sources, along with analysing migration trends, Hungary ensures that discrepancies or inconsistencies are identified.

202. Latvia compares its long-term emigration estimates annually with mirror statistics from Denmark, Finland, Sweden, Norway, Spain, the Netherlands, Austria, Iceland, Germany and Ireland. These mirror statistics, along with population estimates, are used to calculate long-term emigration estimates. Initially, emigration estimates are broken down by country of residence. Information for each emigrant, such as the country of residence, is derived from the Register of Natural Persons of the Office of Citizenship and Migration Affairs (OCMA). In the case where an individual has no record of residence in another country (meaning their record indicates Latvia as their country of residence), they are assigned the same country as similar emigrants based on demographic characteristics. When analysing long-term international emigration and immigration with the OCMA, the data at individual level is compared with the corresponding population data. In cases where data discrepancies are observed, adjustments are made based on case studies.

203. Mexico assesses its emigration estimates through a statistical quality assessment process, which involves verifying data consistency and identifying potential systematic errors. This process includes cross-referencing internal sources by evaluating their consistency and historical trends, as well as comparing them with external sources, such as censuses and surveys from other countries, particularly the American Community Survey (ACS). Key indicators used for these comparisons include the emigration level, distribution by sociodemographic characteristics, destination country, duration of stay abroad, and the main reasons for emigration. These comparisons are conducted regularly, between each census or survey period, to ensure the data is updated and accurate. To further assess the quality of emigration estimates, —particularly in specialized surveys such as the National Survey of Demographic Dynamics (ENADID, by its Spanish acronym)—precision statistics are applied to evaluate the reliability of the published figures. These include the standard error, coefficient of variation, and 90% confidence intervals.

204. New Zealand uses diaspora data from other countries to assess coherence, with high-level comparisons. For instance, the Australian Bureau of Statistics (ABS) reports on New Zealand-born people living in Australia, reflecting migration trends, and with an age-sex breakdown which is valuable for understanding the diaspora's composition. Overall, these comparisons suggest no need for adjustments or methodological reviews. Additionally, New Zealand publishes uncertainty estimates and highlights revisions between provisional and final emigration (flow) estimates. This includes standard errors from model uncertainty in its provisional estimates, using the XGBoost model with a rolling training window for sampling. Provisional estimates are accompanied by 95 per cent confidence intervals, reflecting the level of uncertainty in the data, and historical revisions to these estimates are tracked over time. As more border-crossing data becomes available, provisional estimates are updated monthly, enhancing the accuracy of migration figures. Final estimates, which are derived from linking border arrival and departure data, do not include confidence intervals, as they are considered to have negligible uncertainty. However, small errors may still occur due to data linkage issues.

205. Poland applies mirror statistics to identify underreporting in emigration figures. The analyses are based on data regarding residents with Polish citizenship in selected European countries, which are the main destinations for Polish emigrants (such as the United Kingdom, Germany, the Netherlands, Norway, Ireland, Belgium, the Czech Republic, Spain, Italy, Iceland, Greece, France, Denmark, Switzerland, Sweden, and Austria). It is important to note that for most of this data, no distinction is made between the type of stay (long-term or short-term).

206. The United States has been conducting research using mirror statistics with Mexico and Canada, and with other countries for US-born emigrants.

207. Countries that do not use external data sources for comparison in assessing the quality of their emigration estimates use other ways to evaluate the quality of their estimates. For instance, Italy evaluates the quality of its emigration estimates using the cancellation data from the central population register (ANPR), which tracks demographic events continuously. They rely on an integrated framework for micro-demographic accounting and statistical population registers, to produce more accurate migration statistics. This system aligns with EU regulations and improves the consistency of migration data. Specifically, the system ensures the consistency of individual migration trajectories, treating untraceable cases as emigrations when appropriate, thus enhancing the reliability of outflow estimates.

208. The United Kingdom provides a composite measure of uncertainty is provided for net migration. To measure uncertainty, they quantify uncertainty associated with the individual steps in this statistical system, for example, adjustments, modelling, and survey-based components. This uses a simulation methodology that is outlined in this "Measuring uncertainty in international migration estimates" working paper. The final stage combines simulated samples to create an uncertainty interval for immigration, emigration, and net migration. Simulated samples are produced for three main national groupings: EU, non-EU, and British nationals. However, these samples currently capture only some of the uncertainty sources. The simulated estimates for immigration and emigration are summed in the order returned by the simulation, introducing additional uncertainty to the composite measure. They acknowledge that the current uncertainty intervals do not account for all sources of uncertainty, and ongoing research intends to include more sources into the measures for administrative-based migration estimates.

209. Mexico, New Zealand and the United Kingdom are the only consulted NSOs that produce estimates of uncertainty or confidence intervals for long-term emigration.

210. Effective communication about the quality of emigration statistics is essential for ensuring transparency, helping users interpret the data appropriately and fostering trust in official statistics.

211. The consulted NSOs adopt various strategies to convey the methodologies, quality, limitations, and reliability of their emigration statistics, adapting their approaches to meet the needs of diverse stakeholders and national contexts.

212. This section examines how NSOs communicate the quality of their emigration estimates in a transparent fashion. It highlights various practices such as publishing uncertainty measures, methodological documentation, and quality assessments.

213. International migration is the most challenging component to measure in Albania, as a population register has not yet been established. To address this, the country relies on the best available proxy data, primarily gathered from the migration module of the labour force survey, combined with population projection rates. A technical guide is also publicly available.

214. Australia communicates concepts, data sources and estimation methods through the Overseas Migration methodology. Additionally, the methodology used for final rebased population estimates is available , providing a framework to evaluate emigration estimates during the intercensal period. To further assist users in assessing the accuracy of emigration estimates, preliminary modelled estimates are compared with the final, directly calculated estimates once they are available. Each revision is then published, enabling users to directly compare the updated estimates with previous versions.

215. Canada releases comprehensive methodological documents to communicate the methods used and the quality of its estimates at the provincial/territorial level and for sub-provincial areas. Following each census, a detailed methodological document is also, released. In addition, ad hoc studies, such as the Technical Supplement: Production of Demographic Estimates for the Second Quarter of 2020 in the Context of COVID-19, are published to inform users on specific topics related to emigration statistics. Methodological documents and presentations are also produced for provincial and territorial focal points, though these are for internal use only.

216. Hungary communicates the quality and estimation methods through the Meta data information. Additionally, Hungary provides detailed explanations in various studies, such as The migratory impact of COVID‐19: The role of time and distances in the migration decisions of Hungarians during the COVID‐19 pandemic. These resources offer valuable insights into the methodologies and data quality used in Hungary's estimation processes.

217. Italy ensures the quality of its emigration estimates through certified statistical processes that align with European Reference Standards. These processes provide detailed data quality cards for all statistical processes, which include quantitative indicators such as non-response rates, timeliness, and the length of the time series. These cards allow users to assess the quality of the statistical process and the data produced. Additionally, Italy’s integrated MIDEA-ANVIS system fits into this quality path by guaranteeing high-level demographic statistics. This approach ensures that emigration estimates are based on robust statistical methods, promoting transparency and trust in the data produced.

218. Latvia publishes all information related to data quality on the Official statistics portal. Documentation outlining the methodology on long-term migration is available, along with, a detailed document that describes the methods used to produce population statistics and long-term emigration estimates.

219. Mexico communicates the quality and reliability of its long-term emigration estimates to the public and data users through technical notes, tables of results with their respective precision statistics, methodological documents, metadata based on microdata.

220. To help users in assessing the quality of these estimates, New Zealand provides comprehensive documentation, including Migration Data Transformation, which outlines the methodology for shifting from intentions-based statistics to more accurate outcomes-based estimates. Additionally, each monthly release (International migration: monthly report) highlights revisions and explains how data accuracy improves as migration patterns become more certain. By consistently offering uncertainty measures and detailed methodological explanations, New Zealand ensures that users can interpret emigration data with a clear understanding of its limitations and reliability.

221. Poland offers quality assessments through publications on emigration data and census results. A dedicated document has been published that discusses the methodology and provides an evaluation of the quality of the estimates. The document is not fully translated into English, a summary in English is included.

222. The United Kingdom provides user guides and updates on methods and assumptions. Extensive documentation on the quality of data sources, including the quality assurance of administrative data, is made available to users. A technical user guide is published, explaining the methods, assumptions, and adjustments used. Additionally, a research progress update is published alongside each long-term international migration report to inform users about the latest research and any changes to the methods.

223. The United States publish emigrants estimates as part of the Net International Migration estimates and are not available separately to data users.

224. Evaluation and communication are fundamental to producing high quality long-term emigration estimates, given the reliance on indirect measurement methods and the increasing complexity of international migration patterns. Transparent communication and rigorous quality assessment are essential to help ensure data reliability, support accurate interpretation, and foster trust among users and stakeholders.

225. While practices differ across countries, the collected experiences of NSOs reveal a range of approaches. The analysis carried out by the task force serve to highlight important lessons and identify existing limitations, with the goal of improving both the evaluation and communication of emigration statistics. From that perspective, the following points need to be considered:

226. The integration of multiple data sources for validation. By combining administrative records, survey data, and other indirect indicators, NSOs can validate their statistics to produce more reliable estimates. This validation includes the strengths and limitations of each data source and, as a result, enhances the credibility of emigration estimates and allows for ongoing quality monitoring. It is recommended to pursue multiple data source validation strategies and consider adopting interactive tools, such as dashboards, to enhance monitoring and improve the validation process;

227. The use of external validation sources such as mirror statistics and diaspora data from key destination countries to cross-check and improve their emigration estimates. These comparative methods help identify underreporting or data inconsistencies that can have a bigger impact on emigration data from the source country. Countries should be aware of the risks when using mirror statistics to evaluate their emigration estimates as circular validation may occur. However, it is recommended to promote the use of such international data sources to validate and refine emigration statistics;

228. Clear and accessible communication of data quality is another essential component. Most NSOs offer some form of methodological documentation, although the level of detail and accessibility varies. Publishing clear technical guides, metadata, and visualization tools, allows users to better assess the quality and limitations of the statistics. This also includes regularly discussing with users to ensure the relevance of these products;

229. The importance of publishing quantitative uncertainty measures such as confidence intervals, precocity errors or explanatory notes. These indicators can provide users critical insight into the reliability of emigration estimates and improve transparency. It is recommended to expand and systematize the publication of uncertainty measures even if they are preliminary as these indicators improve transparency and foster better informed decision-making in a timely manner;

230. Since emigration involves at least two countries, international collaboration also plays an increasingly important role in strengthening the evaluation and communication of emigration statistics. Case studies demonstrate that regular exchange of experiences, data sharing and alignment of methodologies and best practices among NSOs can support continuous improvement and innovation. It is recommended to engage in peer collaboration, both regionally and internationally. This fosters knowledge exchange and supports the identification of gaps and opportunities for improvement in a context where international migration is increasingly complex.

231. Innovation and flexibility in the evaluation and communication of emigration statistics should be encouraged. Increasing the use of existing practices such as dashboards, increased social media presence, and clearer, more concise messaging for the public, can enhance accessibility and engagement.

232. With the emergence and accelerating uptake of Generative AI the consumption of official statistics, including migration statistics, will increasingly be via AI layers. These may be through AI incorporated into search engines, or the public and customers of the statistics directly using Large Language Model (LLM) driven AI tools for sourcing their data and information. Ways to make emigration outputs “AI ready” should be explored, with particular attention given to the fact that LLMs typically consume textual information (including the likes of written outputs) more readily compared to standard data outputs.

233. In conclusion, a continued emphasis on external validation, uncertainty quantification, methodological transparency, international collaboration and innovation will be essential to improving the quality, reliability and usability of emigration statistics.