EUROPEAN COMMISSION

Brussels, 20.1.2023

SWD(2023) 11 final

COMMISSION STAFF WORKING DOCUMENT

IMPACT ASSESSMENT REPORT

Accompanying the document

Proposal for a

REGULATION OF THE EUROPEAN PARLIAMENT AND OF THE COUNCIL on European statistics on population and housing, amending Regulation (EC) No 862/2007 and repealing Regulations (EC) No 763/2008 and (EU) No 1260/2013

{COM(2023) 31 final} - {SEC(2023) 38 final} - {SWD(2023) 12 final} - {SWD(2023) 13 final} - {SWD(2023) 14 final} - {SWD(2023) 15 final}

Table of contents

1.Introduction: Political and legal context

2.Problem definition

3.Why should the EU act?

4.Objectives: What is to be achieved?

5.What are the available policy options?

6.What are the impacts of the policy options?

7.How do the options compare?

8.Preferred option

9.How will actual impacts be monitored and evaluated?

Annex 1: Procedural information

Annex 2: Who is affected and how?

Annex 3: Analytical methods

Glossary

1.1.    Introduction: Political and legal context

Eurostat and the European Statistical System

Eurostat, the statistical office of the European Union (EU), ensures the production of high-quality, comparable European statistics 1  under Regulation (EC) No 223/2009 2 (the ‘Statistical law’) and the statistical principles, notably those laid down in the European statistics Code of Practice 3 . These rules and principles aim to ensure, among other things, the independence, impartiality, objectivity and reliability of European statistics, and through those objectives public trust in the statistics. The main uses of European statistics are to serve EU policy design and implementation and monitoring, and their main users are EU institutions.

The European Statistical System (ESS) is the partnership between Eurostat and the national statistical institutes (NSIs), as well as other national authorities responsible for developing, producing and publishing European statistics in each Member State. This partnership also includes the European Economic Area (EEA) and European Free Trade Association (EFTA) countries, i.e. Iceland, Liechtenstein, Norway and Switzerland. The ESS functions as a network in which Eurostat’s role is to lead the way in the harmonisation of statistics in close cooperation with national statistical authorities, which collect data and compile statistics for national and EU purposes. To fulfil this role, Eurostat issues statistical regulations and methodological guides, organises expert groups, and assesses the quality of statistics and Member States’ legal compliance. In line with the EU’s principles of subsidiarity and proportionality, each ESS member develops a statistical system suitable to their individual institutional context, while still following the common rules. The ESS also coordinates its work with candidate countries, other Commission departments at EU level, and international organisations, such as the United Nations Economic Commission for Europe (UNECE) and other United Nations (UN) bodies.

Eurostat’s activities are further influenced by overarching policies such as the EU’s Better Regulation agenda 4 , which promotes open and transparent EU decision-making and evidence-based decisions, and European Commission President Ursula von der Leyen’s six political priorities, among them ‘An economy that works for people’, ‘Promoting our European way of life’ and ‘A new push for European democracy’ 5 . Implementing, monitoring and assessing these priorities requires impartial and objective data – that is, official statistics.

Policies for the EU’s population and citizens

Under Article 9 of the Treaty on European Union (TEU), every national of a Member State, in addition to the national citizenship, is also a citizen of the EU. To define and implement policies and activities benefiting the EU population and citizens in the areas of the EU’s powers, as laid down in Articles 2 and 3 of the Treaty on the Functioning of the European Union (TFEU), EU institutions need timely, reliable, detailed, harmonised and comparable European statistics. To observe the principle of non-discrimination in all its activities and the individual citizens’ rights as enshrined in Articles 10 and 19 of the TFEU and the Charter of Fundamental Rights of the European Union (CFR), EU institutions need a reliable and comparable count of the whole population of the EU. Moreover, the Commission is required to monitor and report on the EU demographic situation in line with Article 159 of the TFEU. EU institutions must also have accurate and comparable population figures at their disposal for administrative and procedural purposes, e.g. qualified majority voting in the Council.

The Treaties oblige the European Parliament and the Council to adopt measures for producing official statistics where necessary for EU policies (Article 338 of the TFEU). Over the past three decades, many EU policy areas have seen strongly increasing and evolving needs for complete, coherent, comparable, reliable and regular European statistics on population, demography and international migration 6 to support evidence-based policymaking. For instance, long-standing policy needs that already led to the current legislation concerned economic, social and territorial cohesion, migration (including labour market integration and wider developments) or challenges related to ageing. Policy needs that have emerged more recently concern more detailed geographic patterns, including on migration, EU and regional mobility, urban/rural integration at local levels, monitoring the European Green Deal, and equality and non-discrimination.

European statistics on population

Population and demographic statistics are among the most popular data Eurostat produces, and they are important for almost every area of policy. For instance, EU economic, social and territorial cohesion policies, labour market and integration policies, equality policies and the EU’s long-term economic and budgetary projections have been relying on these statistics over the past two decades. However, these statistics also provide essential input for public research and informing the society at large. Finally, population statistics are very relevant to the general public because they describe facts and events that concern each individual.

As part of this initiative and as illustrated in Figure 1 , European statistics on population (ESOP) cover official European statistics on population, demographic events and migration, statistics from the population and housing censuses, and the various indicators based on these statistics. Eurostat has published statistics in these areas since 1960, when the first survey on the size and structure of the active population in Member States at that time was introduced. Since then, population statistics have been produced mainly by taking results from a direct enumeration of the population during population censuses and interpolating intermediate periods with information on population changes from civil registration administrative systems (on births, deaths and migration). This means the burden of production has always been mostly on statistical authorities, except for the censuses that interviewed nearly the entire resident population, typically at decennial intervals. The ongoing move from traditional field censuses to combined or even fully register-based modes minimises the burden of production for the general public and moves it onto the statistical authorities.

Until 2007, Member States transmitted all related data on a voluntary basis, which resulted in inconsistencies and lack of completeness or timeliness (see evaluation of the current legislation) 7 . Therefore, after a previous intervention these statistics are now based on legislation adopted between 2007 and 2013. First, Regulation (EC) No 862/2007 8  (Migration Regulation) set out requirements on migration statistics in line with the action plan for the collection and analysis of Community statistics in the field of migration 9 . It includes statistics on: (i) immigration to and emigration from Member States, including from the territory of one Member State to that of another Member State and between a Member State and the territory of a non-EU country; (ii) the citizenship and country of birth of persons usually resident in the territory of the Member States; and (iii) statistics on citizenship acquisitions. Regulation (EC) No 763/2008 10  (Census Regulation) then set out common rules for providing comprehensive census data on population and housing in the EU every 10 years. This ensured compilation of detailed data on specific demographic, social and economic characteristics of persons, families and households, as well as on housing at a national, regional and local level. Finally, Regulation (EU) No 1260/2013 11  (Demography Regulation) laid down common rules for European demographic data, including data requirements on population stocks and vital events such as births and deaths. This Regulation also obliged Member States to provide the Commission (Eurostat) with harmonised data on the total national population to be used as weights for the qualified majority voting in the Council.

All three regulations specified definitions, data variables and periodicity of statistics. Data breakdowns (e.g. demographic or geographical) and detailed tabulations were defined mostly in implementing acts. Eurostat currently collects data from Member States according to data requirements specified in these statistical regulations. Where the legislation does not cover the EU’s needs for statistics, Eurostat seeks the agreement of Member States to initiate voluntary data collections. For instance, statistics on marriages, divorces, legally induced abortions, infant mortality and loss of citizenship are currently entirely voluntary. Various breakdowns of live births and deaths (e.g. marital and activity status, educational attainment) and of international migration are voluntary as well. To produce regional population projections that are important for the regional and cohesion policies and the European Semester, Eurostat collects voluntary data on regional international migration and interregional migration in the EU.

In 2020, to respond to data needs in the context of the COVID-19 pandemic, Eurostat and Member States set up the voluntary data collection on weekly deaths that also enables the publication of statistics on excess mortality. The evaluation has shown that regular use of voluntary data collections to address important user needs is not sustainable because timeliness and completeness of statistics at EU level cannot be guaranteed as needed. Many Member States compile and publish more extensive national population statistics than currently collected at EU level. However, there are differences in national practices and data availability.

European statistics on population – the backbone of social statistics

Population statistics are the backbone of all social statistics, as they provide the most accurate and up-to-date reference information on the entire population and its basic demographic characteristics. Such a population framework with very good coverage and location information is indispensable for more detailed annual population estimates, data collections based on samples, and regional analysis. Population estimates are also needed to obtain per capita indicators in statistics. Population statistics provide the input for preparing population projections for EU long-term economic and budgetary projections specifically and EU economic, social and cohesion policies more broadly.

In 2014, to comply with emerging statistical needs, the Commission (Eurostat) began the process of modernising social statistics. This led to adopting a common legal framework for European statistics on persons and households, based on data at individual level collected from samples of persons and households 12 in 2019. This framework is fundamental for laying a coherent foundation at European level for data collections on samples. This initiative on European statistics on population is the second core element of this modernisation process. Early high-level support for the initiative in the ESS was expressed in the 2017 Budapest Memorandum 13 , which endorses action on a flexible response to changing needs, further harmonisation of concepts and definitions, and expanding annual data collection, including on migration and geographic detail.

European statistics on population and the global data and governance environment

At a global level, there is action to coordinate and harmonise population and housing censuses through the World Population and Housing Census Programme, which is developed and updated on a 10-yearly basis under the coordination of the United Nations Statistics Division (UNSD). The current 2020 decennial programme encourages all countries to hold at least one census between 2015 and 2024 and take into account international and regional recommendations on data quality, methods, concepts and definitions. The programme is supported by a set of principles and recommendations for population and housing censuses 14 that provides extensive guidance to countries on the design and implementation of the census. Regional and topical guidelines on the census and different methodological and implementation issues are produced, such as the Conference of European Statisticians’ (CES) recommendations for censuses of population and housing 15 . These CES recommendations are compatible with the world census programme but are focused on the specific circumstances and needs of countries in Europe. Looking ahead to the next 2030 worldwide census round, the foundations of which are currently being prepared, international recommendations are shifting further towards more efficient and versatile production systems, for instance with the 2018 UNECE guidelines on the use of registers and administrative data for population and housing censuses 16 and with the UN Handbook on Registers-Based Population and Housing Censuses 17  (draft in progress). These developments also address lessons learned from the COVID-19 pandemic, which showed that modern official statistics systems are needed that minimise dependencies on responses from individual contacts while becoming more efficient to react to ad hoc public needs.

Demography and migration statistics are other areas for which there is fruitful international statistical cooperation. The UNSD issues guidelines on demography, vital events and migration statistics 18 , and collects data leading to the publication of the annual Demographic Yearbook 19 . Eurostat has integrated the UNSD data request in its data collection, thus helping 43 European countries to avoid transmitting the same data twice to international organisations. In the area of migration statistics, Eurostat also has a close partnership with the International Organisation for Migration, the United Nations High Commissioner for Refugees and UNICEF on methodological, promotional and technical assistance matters 20 .

Furthermore, official statistics are being developed and produced in the context of a public data environment that is currently evolving rapidly, given the ongoing digitalisation of all sectors of public life. More and more often over the past two decades, statistical office complement traditional methodological approaches with opportunities offered by Big Data to increase topical relevance, timeliness of data and efficiency. This initially started with the increasing use of administrative register information that is becoming richer and more accurate – a transformation that is still ongoing and not equally achieved in all Member States. In addition, more and more new sources are becoming available (e.g. satellite, social network and mobile operator data), including sources held in the private sector. A modern governance framework including official statistics should embrace all these developments. For instance, the recently adopted Commission proposal for a Data Act 21 provides enablers for business-to-government (B2G) data sharing under certain conditions, including for official statistics. This new opportunity would increase the potential benefits from a modern data environment and incentivise further modernisation of population statistics 22 .

Current issues with European statistics on population

As recognised by the evaluation, statistical data on the European population – including demographic and migration events, information on families, and households and housing arrangements – are vital for evidence-based policymaking. In the European context, high-quality statistics across Member States are essential to support many policy areas and initiatives across the EU. Apart from the long-standing use cases mentioned above, four of the six Commission priorities for 2019-2024 (footnote 5) have expressed clear needs for specific relevant European population statistics as data evidence for policy, namely ‘A European Green Deal’ (renovated, energy-efficient buildings), ‘Promoting our European way of life’ (fundamental rights), ‘A new push for European democracy’ (long-term vision for rural areas, rights of persons with disabilities) and ‘An economy that works for people’ (‘Union of equality’). Moreover, the final proposals of the Conference on the Future of Europe 23 have noted a need for further EU-level efforts to collect such data.

Appropriate data evidence for EU policies should be complete, coherent and comparable at EU level as well as timely and frequent enough to support policymaking effectively. A long-standing practical example is cohesion funding, which is allocated on the basis of regional population size and demographic patterns. If the underlying statistics are not comparable across all EU regions, this can distort the allocation of funding. Therefore, it is important that the data collected across Member States are coordinated and consistent to have relevant and comparable data to support policy initiatives and comply with the EU’s aims and objectives. This includes ensuring consistent statistical definitions and data collection methods, including the periodicity, timeliness, completeness and required detail of statistical publications at EU level.

In terms of periodicity and timeliness as set out in the current legislation, Eurostat’s demographic and international migration data collections are mostly annual and available in full detail 12 months after the reference year. This was insufficient in relation to the COVID-19 pandemic, for instance, when timelier data were needed. Eurostat currently has no data available to answer user questions about EU citizens returning to their home countries as a result of Brexit or COVID-19 until the annual data become available. In addition, a very recent ad hoc collection of statistics on Ukrainian residents in the EU directly before Russia’s unprovoked invasion of Ukraine showed that a majority of, but by far not all, Member States could respond with the necessary details on time. Furthermore, EU-level statistics on population and housing censuses are made available only 27 months after the reference year. This delay is considered to be too long by key policy users, even while they acknowledge the complexity of the process in some Member States. There are time lags between the releases of national census results and EU census outputs as well as across Member States.

While measures were taken to align concepts and definitions in the current legal framework, the population base is defined in each regulation with some statistically important differences in the formulation. Moreover, most of the legal definitions currently in place offer default exceptions for Member States to resort to national population definitions where the harmonised concept of usual residence based on 12 months’ presence is difficult to determine.

The evaluation has shown that this lack of harmonisation reduces the relevance, coherence, consistency and comparability of statistics across Member States, with negative impacts for decision-making based on them. There is therefore the need for a new legal basis that can provide a long-term framework for necessary developments towards further convergence of population statistics. This should include real progress on harmonisation and sufficient flexibility to adapt better to evolving policy needs and to make the most of opportunities emerging with new data sources. Since there are also potential opportunities for administrative simplification and process integration compared to the status quo under three non-aligned legal bases, this initiative is included in the 2022 Commission work programme as a REFIT initiative.

2.2.    Problem definition

2.1.2.1.    What are the problems?

The evaluation identified issues structured around the following four main problems. The current legal framework:

·does not ensure sufficiently complete, coherent, and comparable statistics, which may lead to suboptimal decision-making (Problem 1);

·does not ensure sufficient availability of population data in terms of frequency and deadlines for data transmission (Problem 2);

·fails to capture characteristics and details of politically relevant topics or groups, e.g. the data provided to users, and the way it is presented, is not detailed enough in terms of statistical topics, characteristics, and breakdowns for the population in general and relevant specific groups of interest (Problem 3);

·is not flexible enough to adapt to evolving policy needs and to enable using data from administrative and other new sources in Member States and at EU level (Problem 4).

The next section explains these problems in turn and how significant they are in terms of scale and impact. It also examines the main drivers leading to the problems and reflects on how these problems are likely to evolve in the absence of targeted action.

2.2.2.2.    What are the problem drivers?

2.2.1.2.2.1.    Problem 1: The current legal framework does not ensure sufficiently complete, coherent, and comparable statistics, which may lead to suboptimal decision-making

The most significant quality gap identified by the evaluation in the current legal framework is the lack of harmonisation of the population base. More precisely, three conceptually different definitions for resident population are currently allowed in varying formulations across the three base acts: either strict usual residence based on 12 months’ presence, or in default registered residence, or legal residence where strict usual residence cannot be determined. The default options were originally introduced in the legislation to accommodate specific constraints of a small number of Member States. However, as more and more countries are relying on administrative data sources, such non-harmonised default definitions are becoming an increasingly common practice across Member States, sometimes even using different definitions for different datasets. Figure 2 illustrates this fragmented landscape across the ESS. Moreover, the legislation is not detailed enough to set out exhaustively what is included in (and excluded from) the population.

An entailed gradual loss of relevance had already been accepted when the current legislation was adopted, since many policy and democratic representation considerations, notably at EU level, require a population base definition (who is counted among the population and who is not) that reflects the actual population present. The compromises detailed above have generally led to a barely sufficient situation in this respect. The Demography Regulation only requires national population figures based on strict usual residence for Council voting weights. However, these figures must be of the highest possible quality 24 but the factual accuracy or comparability achieved is currently not quantified, and the evaluation indicates that the strict definition for this purpose is not implemented consistently in all Member States. (This touches on another issue of reduced completeness, namely insufficient metadata being published along with the statistics, which was also raised by users during consultation surveys and interviews).

The harmonisation gap also entails a situation where the vast body of demographic and migration statistics cannot reach its potential in terms of comparability between Member States and consistency between datasets due to differences in the definitions applied. For instance, different population bases are often used for population stocks and migration flows, which leads to inconsistencies between stock differences and demographic changes (demographic balance) between reference years. During the current legislation’s implementation, only between 8 and 14 ESS members have been reporting consistent demographic balances. Regarding comparability, the differences between total populations reported for Council voting weights (strict usual residence nominally enforced) and in other annual datasets (no common population base enforced) may be used as a rough proxy indicator for the remaining level of comparability limitation across Member States. From 2014 to 2020, relative differences over the years were on average smaller than 1% (for 23 of 27 Member States), and the largest relative differences encountered for single Member States were below 5% for any reference year 25 .

The open public consultation 26 (OPC) has revealed a complex picture around this harmonisation gap and its relation to coherence and comparability. When asked whether the current statistics are sufficiently harmonised, comparable and/or coherent overall, OPC respondents across all key stakeholder groups did not have a strong tendency to either agree or disagree 27 . Across all stakeholder groups, however, there is a lack of knowledge about the current harmonisation gap of the population base definition. Only among researchers, the majority of respondents is aware, but for example in the institutional user group, just about half of the respondents know about this gap. Then, contrary to generic opinions above, once respondents have been made aware of the current harmonisation gap, more than 50% across all stakeholder groups strongly agree that harmonisation would be important 28 .

The survey of national statistical institutes (NSIs) 29 and in-depth interviews highlighted that NSIs consider their national definitions to be adapted to the national context and that the benefits of the current use of national definitions centre around meeting current national requirements. Some NSIs asserted that this can lead to not entirely comparable statistics at EU level, which can be an issue for some data users with more advanced needs for comparable population statistics (those for whom precision at a granular level is required). The use of national definitions can also lead to double counting when people move between Member States, which may result in discrepancies in European population statistics. Data users and international partners of Eurostat identified similar problems, for instance when counting people who migrate between Member States to study or work, or where people have second homes and spend parts of the year living in two or more places. Figure 3 illustrates NSIs’ (i.e. statistics producers) opinion that the status quo means accepting a trade-off between certain drawbacks for the quality and use of the data at EU level (considered mostly minor impacts) and advantages in terms of reduced production costs of these statistics (considered mostly a major impact). 

The lack of rules on time series revision is another gap leading to patchy revision practices across Member States and thus to reduced comparability over time, in particular for Member States relying on more traditional data collection approaches (as opposed to administrative source-based approaches). This is the case for instance when annual data is computed as incremental changes (vital and migratory events accumulating over time) since the last census year and thus suffer from accuracy decreasing with the time gap after the last census. However, the current legal framework does not contain any rules on time series revisions of annual data. The 2011 EU census experience illustrates some consequences, where, on Eurostat’s initiative, 18 countries voluntarily revised parts of their annual time series with average correction magnitudes at total population level of around 1.4% but up to 7.5% in individual cases 30 . However, two of these countries only revised back to the reference year of 2011 (Germany; avg. -1.87%) or 2010 (Poland; avg. -0.81%). This meant that breaks in their time series remained around the census year 2011, with the German case extending to a visible time series break of the entire EU population due to the population size of the country. Consequently, most OPC respondents across almost all key stakeholder groups give high priority to potential future improvements leading to better revision rules 31 .

Finally, continuing large parts of the statistical outputs under voluntary collections leads to various gaps in quality and other related gaps documented in the evaluation. In particular, a key consequence is the lack of completeness of voluntary statistics across all Member States ( Figure 4 ). This limits the utility for EU-level analysis or publication of EU aggregates, and thus reduces the EU added value significantly. A related issue is the reduced efficiency of this practice: the overall baseline production costs across Member States delivering voluntary statistics fail to deliver significant benefits at EU level. Even if more Member States start producing a given voluntary dataset, the practice will not be cost effective as long as there are other Member States that continue not to produce this dataset. These findings are in line with the opinion of OPC respondents across all stakeholder groups (except statistics producers) that potential future improvements should include measures to regulate data that is currently voluntary 32 . Current completeness and comparability gaps are seen as most critical in the datasets on international migration – from/to and within the EU.

2.2.2.2.2.2.    Problem 2: The current legal framework does not ensure sufficient availability of population data in terms of frequency and deadlines for data transmissions

Currently, mandatory demographic and international migration statistics are published with an annual frequency and a default deadline of 12 months after the reference period (timeliness) for the full details. These annual data are complemented by detailed decennial EU outputs from population and housing censuses that are published based on national schedules within a deadline of 27 months from the end of the census year (e.g. by March 2024 for 2021 EU census outputs) 33 .

The stakeholder consultation has revealed that the current frequency and timeliness of the publication of statistics remains below user expectations. There were no strong opinions from OPC respondents across all stakeholder groups (and only a minority of respondents among institutional users found that the current statistics were generally not timely enough). However, population and housing census outputs were mentioned most often as most critically affected by insufficient frequency and especially timeliness 34 . The targeted consultation of Commission departments (representing EU-level institutional users) also indicated critical gaps. In particular, regarding frequency, various EU policy areas (e.g. rural integration, seasonal movements between rural and urban territories including workers) have been identified as requiring at least quarterly data on population stocks and demographics, including migratory events. Moreover, monitoring the European Green Deal requires housing data more frequently than every 10 years, including data on energy characteristics of housing 35 .

Similarly, responding effectively to disruptive events or crises – requiring effective measures for quick and very frequent (e.g. monthly or weekly) ad hoc data collections – has become more important during the past decade. For instance, Eurostat has not been able to sufficiently answer user questions on EU citizens returning to their home countries following Brexit or the onset of the COVID-19 pandemic until the standard annual data became available, and provided that Member States sent voluntary breakdowns. An ad hoc data collection on weekly deaths was initiated to address COVID-19 information needs, but OPC responses indicate that generally such (voluntary – see Section 2.2.1) ad hoc measures may not be sufficient to meet such highly dynamic crisis response needs 36 . These issues are also linked to the lack of flexibility in the legal framework (see Section 2.2.4).

Finally, from the statistics producers’ perspective, the frequency with which NSIs publish population statistics nationally varies significantly 37 . However,  Figure 5  shows that many of them already publish some statistics topics at national level more frequently and/or more quickly than Eurostat does at EU level. For instance, 22 ESS members publish national population stock data within 6 months of the reference date, which is 6 months earlier than the 12-months legal deadline, and at least 17 ESS members manage to do so for data on vital events and on migration as well. The use of administrative data sources in particular makes more frequent/timely publication less costly and more efficient. In a poll carried out during a consultation workshop with directors of NSIs 38 , 64% of respondents indicated improved timeliness as the biggest priority for a future EU initiative. However, increasing the EU-level frequency and timeliness of data collections would likely have varying impacts across Member States, including increased administrative burden (and costs) on some NSIs, with production processes relying less on administrative sources. Nevertheless, there is a similar overall cost-effectiveness argument at EU level as regards completeness (see Section 2.2.1) – understanding that the status quo gets less efficient the more Member States use baseline resources to attain a certain national standard without reaping the EU added value. 

2.2.3.2.2.3.    Problem 3: The current legal framework fails to capture characteristics and details of politically relevant groups or topics

The evaluation has identified various gaps in detail that have led to a significantly reduced relevance of the current legal framework over time from a policy perspective. In addition, a majority of OPC respondents from professional user groups were of the opinion that, generally speaking, the statistics do not provide enough detail 39 . The most crucial gaps identified revolve around characteristics of politically relevant topics and groups (migrants, migration and mobility within the EU; urban/rural population; vulnerable groups for equality, non-discrimination and fundamental rights policies; energy efficiency of housing for the European Green Deal), as well as around an insufficient geographic detail of the statistics (especially functional typologies and georeferenced data for urban/rural mobility and cross-border analysis). These gaps were confirmed by the targeted consultation with Commission departments, as well as by a majority of OPC respondents across almost all key stakeholder groups – for instance professional users’ views shown in Figure 6 . Each topic will now be addressed briefly in turn.

Migration and mobility to, from and within the EU

The relevant concept of migration used in ESOP-related statistics is any change of residence at a given geographic level of detail, where the concept used for ‘residence’ links migration directly to the population base, or any change thereof, at the same geographic level 40 . Therefore, changes of residence at any geographic level across the EU are in principle considered migration flows within the scope of ESOP. This includes international migration (between countries) both from/to the EU and within the EU, as well as internal migration (between regions or other geographic units inside the same country). Moreover, migrant stocks relate to population subgroups with an immigrant background (typically identified by a country of birth and/or citizenship different from the country of residence).

Currently, Eurostat publishes some mandatory data on international migration flows and migrant stocks under the Migration Regulation (flows and stocks by core demographic characteristics and broad groups of countries of origin/destination), complemented by a variety of voluntary data breakdowns of increasing policy relevance, including migration patterns at subnational level, more detailed and ideally single-country information on origin or destination, and socio-economic characteristics. However, as the evaluation has shown, most of these voluntary data are affected by a lack of EU completeness and comparability. Furthermore, there are no European statistics yet for other specific migratory movements (like short-term/seasonal movements or return migration) that receive increasing policy attention but are even harder to capture.

Consequently, OPC responses across user-type stakeholder groups indicate an insufficient status quo; for instance, EU and subnational mobility rank among the most insufficient in professional users’ views, as seen in Figure 6 . Moreover, out of 92 OPC respondents overall that noted any inadequacies in the current data, the majority identified migration within the EU (66%) directly followed by migration from/to the EU (62%) as high priorities to improve by 2030 41 . This is substantiated by replies to the targeted consultation of Commission departments, where various policy areas signalled the need for more detailed migration data – including flows and stocks by socio-economic characteristics and by single country of origin/destination (flows) or of birth/citizenship (stocks) – to monitor the integration of immigrants in labour markets and society.

Migration data also appears to be one of the datasets most requested by national data users, as 79% of NSIs have indicated migration flows within the EU and from/to non-EU countries among the most commonly requested topics 42 . At the consultation workshop with NSIs (footnote 38), migration within and from/to the EU was also ranked among the top priorities to be improved by 2030, where migration within the EU in particular poses increased challenges due to freedom of movement entailing reduced availability of administrative sources. Moreover, the quality of emigration data may be limited and may underestimate the true scale of the phenomenon as existing data collection processes may not be sufficiently sensitive to disaggregate emigration by detailed characteristics.

Specific population groups

The term ‘equality data’ refers to data on population subgroups characterised by attributes that make them more vulnerable to inequality or discrimination. Articles 10 and 19 of the TFEU list six grounds for discrimination (variables) that are relevant for EU non-discrimination policies: sex, racial or ethnic origin, religion or belief, disability, age and sexual orientation. To date, European equality statistics in the area of population statistics exist for the basic demographic characteristics, sex 43 and age 44 . More detailed information is available in other areas of European statistics on disability, but notable gaps persist on race/ethnicity, religious affiliation, gender identity and sexual orientation. Therefore, the targeted consultation of Commission departments confirmed that European equality data are greatly needed across all grounds for discrimination for EU policies on fundamental rights and non-discrimination, especially in the context of the ‘Union of equality’ 45 . Moreover, OPC respondents from professional user groups believe that equality is the topic with the most insufficient available data (see Figure 6 ; only 18% of the group that expressed an opinion finds the current data sufficient). Most recently, the final proposals of the Conference on the Future of Europe have also noted a need for further EU efforts to collect such data (footnote 23).

At national level in the ESS, the coverage of equality data is patchy as well. For instance, according to a recent Eurostat survey on implementation of the 2021 census round, only 19 ESS members published any variables on population subgroups at risk of inequality or discrimination as part of their national 2021 census outputs 46 .  Figure 7  shows the number of ESS members publishing specific equality variables. The most prevalent reasons for not providing certain equality topics across Member States seem to be historical or cultural perceptions of such data as very sensitive or not to be collected by public authorities. Moreover, a majority of NSI respondents saw implementation costs (10 Member States) and national census modes (9 Member States) as a huge barrier to including self-declared information in their 2031 censuses 47 . National legal barriers are often brought up as an argument as well, but recent Commission studies have found that such barriers do not factually exist in any Member State on race/ethnicity and gender 48 .

National contexts aside, there are technical and feasibility challenges for the collection of such data. For instance, there is a generally accepted principle that these characteristics should be self-identified 49 , which would require primary data collection (e.g. based on surveys) rather than extraction from administrative sources (let alone the fact that such characteristics are often not recorded at all in public administration registers). Moreover, the quality of results from sample surveys depends on an accurate sampling of the target population. This is another key challenge because the best sampling frames available are often based (again) on administrative registers 50 . However, these registers have notorious under-coverage issues with certain vulnerable groups – like Roma or migrants with unclear or irregular status – due to them not being registered in the national systems.

Housing data for the European Green Deal

Only the censuses currently provide decennial outputs on housing arrangements and homelessness, including data on housing characteristics such as the type of facilities and heating systems. However, the available variables have become less and less relevant to professional users, including policy users. For instance, housing is the second-lowest-rated data topic (after equality data) according to professional users among OPC respondents (see Figure 6 ; only a quarter of the respondents that expressed an opinion finds the current data sufficient). The consultation of Commission departments also identified a need for more frequent and timelier housing data, including appropriate variables to capture energy efficiency of housing and other aspects relevant to Green Deal monitoring. Relevant European statistics serving these needs will be important to measure whether the EU remains on the right track towards its ambitious greenhouse gas reduction goals over the next decades.

During technical consultations with Commission informal expert groups, Eurostat collected several opinions from NSI experts indicating that housing data are indeed often available from administrative sources within the countries. In principle, this would lead to cost-efficient production of more frequent and more relevant housing data, at least in some Member States. The Census Regulation covers the need for data supporting ‘the protection of the environment and the promotion of energy efficiency’ in recital 2. However, its legal structure is not flexible enough to change or add variables in response to such needs (see problem 4). Moreover, the intention to address this under the present initiative has been subject to more critical opinions at higher levels of the ESS governance, where the main concern was rather whether these data should remain within the scope of ESOP, especially when buildings in general (beyond housing) are addressed.

Geographic granularity

Currently, mandatory annual statistics on demography and international migration are mostly only collected at national level, with some less detailed demographic breakdowns at regional level (NUTS 2, 3) 51 . More detailed regional breakdowns of annual statistics are currently only provided through voluntary data collections, with the typical consequences outlined in Section 2.2.1. This annual picture is complemented by mandatory decennial population and housing census outputs with very detailed breakdowns, including at local administrative unit (LAU) level and – for the 2021 round – some key population indicators georeferenced to a pan-European 1 km square grid.

The evaluation and stakeholder consultation found that most key users of the data believe the current situation is inadequate, due to a complete lack of subnational data on migration (stocks and flows). For instance, insufficient geographic detail was the second most frequent critical gap (after insufficient detail of characteristics) noted for any statistical topic by OPC respondents across all stakeholder groups 52 . Consequently, most OPC respondents across almost all stakeholder groups (except statistics producers) also put a high priority for a future initiative on adding (mandatory) regional detail to annual demographic and migration statistics, and on providing functional typologies (DEGURBA, cities, FUAs) and 1 km square grid data annually 53 . This is in line with findings from the targeted consultation of Commission departments identifying policy needs for more annual NUTS 3 data, functional typologies, including city and FUA data) and prominently 1 km square grid data 54 .

The 1 km square grid data play key roles, both technically and in terms of policy. From a technical view, the total population at 1 km square grid level is needed to construct the functional typologies (DEGURBA, cities, FUAs), which Regulation (EC) No 1059/2003 defines as clusters of LAUs according to certain population density and movement patterns based on the grid. This means that even if Eurostat does not publish European grid data, NSIs need grid information at national level to determine and maintain the functional typologies. In terms of policy, only 1 km square (or more detailed) data allow accurate analysis on access to services (e.g. driving time to next hospital), local crisis management (e.g. locally contained natural disasters like flooding or wildfire) or cross-border patterns (e.g. functional urban areas clustering in a river basin, where the river happens to mark a national border). Finally, the baseline includes substantial investment across the ESS to create infrastructure for the first mandatory grid data collection from the 2021 EU census round 55 .

Urban/rural integration

Even though increased availability of regional data at NUTS 3 is widely acknowledged by users, the angle from an urban/rural dichotomy is often underexposed. This is because NUTS 3 regions are often still too coarse to capture urban/rural patterns according to functional characteristics accurately (mostly related to local population density). Therefore, during the targeted consultation with Commission departments, policy needs were raised for cohesion and the integration of urban and rural areas, notably for the Commission’s long-term vision for rural areas 56 . These policy areas require any relevant population data cross-tabulated against the degree of urbanisation (DEGURBA) classification defined in Regulation (EC) No 1059/2003 based on clusters of local administrative units (LAUs). In addition, population data on cities and their functional urban areas (FUA) are needed to shed further light on urban/rural patterns.

While European statistics cross-tabulated against DEGURBA have become more and more prevalent over the past years, no statistics published under the current legal framework have provided DEGURBA so far. Moreover, annual statistics on the population of cities and FUAs are currently published regularly on a voluntary basis and outside the ESOP data collections. Therefore, these data suffer from a lack of harmonisation with ESOP data, in addition to the identified drawbacks of voluntary collections (Section 2.2.1). Consequently, Figure 6 shows that urban (47%) and rural (44%) populations also rank among the lowest scoring data topics among OPC respondents from professional user groups when asked about their agreement on whether the current data are sufficient.

2.2.4.2.2.4.    Problem 4: The current legal framework is not flexible enough to adapt to evolving policy needs and to enable using data from administrative and other new sources in Member States and at EU level

Currently, with regard to the output of the statistical production process, the legal framework sets out a fixed set of statistical units, variables/breakdowns and cross-tabulations to be produced regularly, without providing for specific mechanisms to update this statistical content efficiently. The evaluation has shown that this current static framework has been losing relevance rather quickly, starting during its implementation period and continuing to the present, due to a framework that is too rigid and lacking flexibility to adapt data collections to evolving needs. This is confirmed by the OPC, where a majority of respondents in all stakeholder groups (except statistics producers) agree only ‘somewhat’ that the legislation is fit for purpose. The Census Regulation represents a minor exception, as it leaves some room to specify statistical needs for each EU census round before the census year, thus maintaining a higher relevance of census outputs over time. However, the 2021 round has shown that this flexibility was not sufficient in terms of introducing georeferenced/grid data, which necessitated an ad hoc legal act 57 to ensure EU-level completeness, comparability and coherence.

With regard to input, all three base regulations contain enabling provisions allowing reporting countries to choose appropriate sources for the statistics, with the Census and Migration Regulations mentioning administrative sources or registers explicitly. This is in line with a general principle that has been prevalent in the ESS of ensuring that European statistics are ‘output oriented’, i.e. minimising legal constraints on inputs and processing of the statistical production chain 58 . Currently, there is a strong ongoing trend towards register-based production systems drawing from administrative sources. The evaluation has clearly shown that this leads to increased data availability at significantly reduced production costs. For example, according to the baseline cost assessment, the median cost of a traditional census across EU-27 countries was roughly 20 times the median cost of a register-based census both in the 2001 and the 2011 census rounds. The single most significant cost driver in this context for NSIs conducting traditional censuses is the full field enumeration of the entire population, which entails a complex organisation and coordination at national level, extensive procedures, and repetitive training and temporary employment of large workforces of enumerators. However, NSIs that base their production system on administrative information, for instance from administrative population registers, can usually reduce costs and avoid such difficulties. Nevertheless, this creates new constraints, as the efficiency of such systems then depends on the information already available from such sources. If other information is needed, it must then be modelled (a typical concern being quality) or collected again from field surveys (which means costs scale up again quickly).

The benefits of moving towards using more administrative and other sources are most pronounced in combination with a statistical population register as a central processing element of an integrated multisource statistical production system. According to the targeted NSI survey, 12 Member States (Austria, Belgium, Bulgaria, Denmark, Estonia, Finland, Italy, Latvia, Netherlands, Slovakia, Spain, Sweden) already have a statistical population register and 8 more are planning one (Croatia, Cyprus, France, Greece, Hungary, Lithuania, Malta, Portugal). The main reasons for not having one, or not having one yet, indicated by the 17 NSIs concerned are the current national legal framework (mentioned seven times) followed by historical reasons (six) 59 . A collaborative network of ESS experts on population and household frames for social statistics also acknowledged the significant potential added value of statistical population registers, especially when linked to rich information from various sources 60 .

However, the current statistical legislation on data sources does not allow statistics producers to easily access appropriate sources held by other owners, including administrative sources. General access enabling legislation exists in Article 17a of Regulation (EC) No 223/2009 and also very often in national legislation, as 26 of 29 NSIs confirmed in the survey 61 . Such enabling legislation is generally covered by the GDPR 62 and EUDPR 63 providing for the processing of personal data for statistical purposes. However, legal access problems often persist, mainly where specific legislation on a given source database has more restrictive access limitations. Eurostat is experiencing such problems with administrative sources held by eu-LISA 64 , and 15 NSIs indicate similar issues by stating that additional legal acts are needed to access some or all specific sources relevant for population statistics 65 . This suggests that the situation could be improved by more explicit sector-specific rules granting access to statistical offices as needed to all relevant sources for population statistics. Nevertheless, the evaluation also showed that the effectiveness and efficiency of practical cooperation agreements between NSIs and administrative source owners are as important as legal enablers.

Finally, the current legal framework does not encourage the use of new sources, such as privately held data, including, for instance, geospatial systems or mobile operator data. While the Demography and Migration Regulations at least allow the use of such sources in principle, the Census Regulation contains a closed list of eligible source types, excluding any new sources 66 . On top of the issues around access enabling legislation outlined in the previous paragraph, this puts a strict legal barrier on the use of such new sources for European population statistics. Therefore, the status quo falls critically short of the state of the art both methodologically 67 and legally. Regarding the legal perspective, the recent Commission proposal on a Data Act (footnote 21) provides enablers for B2G data sharing, including for official statistics. In this context, the previous paragraph pointed out that explicit sector-specific enablers (aligned with the Data Act proposal) on relevant new sources for population statistics may help improve new ESOP legislation in terms of continued relevance, effectiveness and efficiency.

2.3.2.3.    How likely are the problems to persist?

The evaluation has identified the following four key legislative drivers for the problems described in Section 2.2.

·Only mandatory data collections with common rules can ensure completeness and timeliness of statistics at EU level; regulating voluntary data collections that already have high completeness may lead to significant effectiveness and efficiency gains, as considerable EU added value can be generated at limited incremental costs (linked to problems 1, 2, 3).

·Voluntary data collections are appropriate instruments to pilot the production of new topics or characteristics, and to foster the incremental capability of national statistical systems to provide such new data. However, they tend to become inefficient over time, as recurrent production costs eventually fail to generate substantial EU added value in terms of completeness and timeliness for all Member States (linked to problems 1, 2).

·Loose legal definitions of statistical concepts lead to a gradual loss of control over conceptual harmonisation, and thus ultimately over coherence and comparability, over time. The population base example shows how a default clause originally introduced as an exception with limited scope turned into a new standard that is detrimental to European statistics. However, a stronger stance on definitions also requires closer guidance and monitoring of implementation to ensure real harmonisation (linked to problem 1).

·A legal framework that is too rigid prevents it remaining relevant over time. The current framework has been losing relevance rather quickly, starting during its implementation and continuing to the present day. This is due to a lack of flexibility mechanisms for adapting data collections to evolving needs or for benefiting from opportunities driven by the availability of new data sources (linked to problems 1-4).

The baseline assumes continuing the status quo regarding the availability of European population data (see section 5.1). The insight set out above – that the problems emerge directly from legal constraints enshrined in the current framework – leads to the conservative scenario in which the current framework will continue losing relevance as observed in the evaluation. This typically leads to professional users turning to other national or unofficial sources of statistics, both affected by reduced quality (measured from an EU perspective) and thus leading to poorer policy decisions. A more optimistic scenario would have to operate on the assumption that evolving policy needs for data would be sufficiently addressed in the future by voluntary data collections. However, the experience over the past two decades does not leave much room for such optimism.

3.3.    Why should the EU act?

3.1.3.1.    Legal basis

The legal basis for EU intervention in the area of European statistics is Article 338 of the Treaty on the Functioning of the European Union (TFEU), which confers on the European Parliament and the Council the power to adopt measures for the production of statistics where necessary for the performance of the activities of the EU, in line with the ordinary legislative procedure. Article 338 of the TFEU sets out the requirements for producing European statistics, stating that they must conform to standards of impartiality, reliability, objectivity, scientific independence, cost-effectiveness and statistical confidentiality. This section assesses whether a revision of the current legal framework governing European statistics on population is appropriate and justified, in view of its purpose to ensure high-quality EU population data in line with the statistical principles and quality criteria applicable to European statistics.

Population data are currently collected under three separate Regulations that neither allow flexibility nor respond sufficiently to new and emerging statistical needs of users and a recognised necessity for modernisation (Section 2.3). A partial revision of the legal framework on population statistics is inevitable for current Regulation (EU) No 1260/2013 on European demographic statistics, as, under Article 12, it will cease to apply in 2028. A minimal revision is therefore necessary even if the status quo is to be maintained. However, this also generates an opportunity for a wider review to modernise the broader legal framework for population statistics, which also includes Regulation (EC) No 763/2008 on population and housing censuses as well as Article 3 of Regulation (EC) No 862/2007 on statistics on international migration. This would also enable maximising the added value of capacity already developed for the one-off Commission Implementing Regulation (EU) 2018/1799 on georeferenced census 2021 outputs, and fully accounting for the integrated approach to statistical quality set out in Regulation (EC) No 223/2009 on European Statistics.

Article 5 of the Treaty on European Union (TEU) enshrines three principles: conferral, proportionality and subsidiarity. As an expression of the principle of conferral, Article 338 of the TFEU empowers the EU to set up European statistics. Furthermore, Article 338(2) of the TFEU mentions six general principles that EU statistics must follow, namely impartiality, reliability, objectivity, professional independence, cost-effectiveness, and statistical confidentiality. These principles are specified in Article 2(a) to (f) of Regulation (EC) No 223/2009 (footnote 2) and further specified in the European statistics Code of Practice (footnote 3) maintained by the ESS Committee in line with Article 11 of the same regulation. Article 338(2) of the TFEU also stipulates that EU statistics must not entail excessive burdens on businesses. This provision, together with Article 338(1) of the TFEU, according to which statistics must only be produced where necessary, reflect an expression of the principle of proportionality. Finally, legislative action on European population statistics falls under supporting competences, where the principle of subsidiarity authorises an EU intervention only if a specific issue cannot be addressed by the individual action of Member States (necessity test, Section 3.2) and provided the objective can be better achieved ‘by reason of the scale and effects of the proposed action’ at EU level (EU added value, Section 3.3).

3.2.3.2.    Subsidiarity: Necessity of EU action

The ESS provides an infrastructure for statistical information. The system is designed to meet the needs of multiple users for decision-making in democratic societies. The proposal for this regulation has been drafted to protect core activities of ESS partners while better ensuring the quality and comparability of statistics. One of the main criteria that statistical data must meet is to be consistent and comparable. Member States cannot achieve the necessary consistency and comparability without a clear European framework, that is to say without EU legislation laying down the common statistical concepts.

The principle of proportionality aims at identifying the best level of governance to ensure that decisions meet the public’s needs to the greatest extent possible. Together with the principle of subsidiarity, the principle of proportionality regulates how the EU exercises the powers conferred by the Member States on the EU. According to the principle of proportionality, EU action must be limited to what is necessary to achieve the objectives of the Treaties. It requires the EU to only take action that is deemed essential to achieve the aim pursued. To ensure compliance with the principle of proportionality, any EU intervention must ensure a higher level of coherence and comparability of population statistics across Member States. Further action at EU level is justified in light of the variety of measures adopted at national level and given that a robust legal framework for the collection of population statistics is essential for maintaining relevant and comparable statistics at EU level based on harmonised concepts and approaches to methodology.

Proportionality also requires that the intervention matches the size and nature of the EU-related problems identified, including the right choice of policy instruments to address the problems. From this perspective, a legislative solution would be necessary and proportionate, as the problems identified have a genuine EU-wide scope clearly linked to gaps in the current EU legislation (Section 2). Without further EU legislative action, these problems will persist or worsen. The current EU legislation will likely continue to become less effective and efficient in achieving its objectives as many Member States continue to modernise nationally by setting up statistical population registers and harnessing new data sources. Relevance will also likely decrease further, as the EU-level statistics are expected to diverge further from users’ needs in terms of content, desired frequency or timeliness. Without EU legislative action, national approaches will diverge further, leading to less comparable statistics, which in turn risks compromising policymaking at EU level.

3.3.3.3.    Subsidiarity: Added value of EU action

The added value of complete and comparable population and demographic statistics at EU level lies primarily in their significant contribution to various institutional needs and policy areas of the EU that are highly relevant for many political priorities of the Commission (i.e. An economy that works for people, Promoting our European way of life, A new push for European democracy). Population and demographic statistics are also needed to feed into various EU institutional tasks and procedures laid down by the Treaties, such as national population weights to determine the 65% EU population quota for qualified majority voting of the Council (Article 16 of the TEU), EU long-term economic and budgetary projections within the European Semester (Article 121(6) of the TFEU detailed in Regulation (EU) No 1175/2011), and monitoring of the annual EU demographic situation (Article 159 of the TFEU).

These data inform EU policies that fall under shared competences (e.g. social policy; economic, social and territorial cohesion; and the area of freedom, security and justice) and supporting competences (e.g. health, youth, civil protection and administrative cooperation). Population statistics are the backbone of other European statistics (sample surveys, national accounts) and used to calculate per capita indicators. Finally, population and demographic statistics are also designed to meet the needs of multiple users, for decision-making at all levels in the EU, as well as research and informing the general public. The EU may therefore adopt measures in this area in line with the principle of subsidiarity under Article 5 of the TEU.

4.4.    Objectives: What is to be achieved?

Figure 8 presents a problem tree showing the logical links between problem drivers and general and specific problems identified during the evaluation and the consequences of these problems for policymakers and other data users. As shown in the figure, the regulatory problem drivers reveal the limitations in the current legal framework, which is not harmonised and lacking in coherence and flexibility, while at the same time practices in Member States are evolving and societal trends changing. This leads to specific problems in comparability of population data, data not being as timely and frequently produced as possible, limitations in the depth and comprehensiveness of data topics, and quality-related problems. Combined, these problems lead to content and quality gaps compared to policymakers and users’ evolving needs regarding demographic and societal challenges.

The general and specific objectives for an EU intervention to modernise European population statistics have been linked to the problems and their drivers. These are shown in the lower half of Figure 8  and detailed below.

4.1.4.1.    General objective

As shown, the general objective of EU intervention in this case is to better respond to users’ needs over time and to modernise and enhance the relevance, harmonisation and coherence of European population statistics.

4.2.4.2.    Specific objectives

Statistics are assessed in the context of the ESS framework for statistical quality 68 , and the key dimensions for this initiative are ‘relevance’, including ‘completeness’, ‘coherence’, ‘comparability’, ‘timeliness’ and ‘frequency’ (or ‘periodicity’). Therefore, the general objective can be broken down into four specific objectives (SOs) along statistical quality dimensions.

·SO1. Ensure complete, coherent and comparable European population statistics. This means all European statistics should be available from all Member States and include EU-level aggregate information. These statistics should follow the same underlying concepts operationalised in a comparable way, so that the information is comparable across Member States and coherent across statistical products.

·SO2. Ensure availability of timely and frequent population statistics to meet users’ needs. This means all European statistics should be provided by all Member States and ready for publication according to a strict agreed deadline that takes into account user needs for the time span between reference period and publication of the statistics (timeliness). Statistical products should also be published with a sufficient granularity of publication dates considering current user needs (frequency).

·SO3. Provide statistics that are sufficiently comprehensive in terms of relevant topics and sufficiently detailed in terms of characteristics and breakdowns. This means the European statistics should address the topical and detail needs of key statistics users and give information that benefits society at large. Statistics must provide the data evidence needed for better policymaking in priority areas of the EU as well as support decision-making at other governance levels or provide the data needed for relevant research or for public debates.

·SO4. Promote legal and data collection frameworks that are sufficiently flexible to adapt datasets to evolving policy needs and to opportunities emerging from new sources. This means the EU statistical legislation should be better able to maintain the relevance of statistical products over time, by enabling the use of more diverse sources and by offering mechanisms to update the statistical products. As worked out in Section 2.3, a key issue of the current legislation is its inability to adapt to evolving input and output contexts. The evaluation measured this in terms of the gradual loss of relevance of the statistical products.

These specific objectives are SMART 69 in that they directly link to the four gaps quantified in the evaluation (footnote 7). The evaluation provides quantitative indicators on each specific objective to follow up in the future, for instance over the first one or two production cycles (including censuses) after adoption of a new legal framework. This completes the pathway from the problem drivers to the proposed solutions and builds the foundation for developing the policy measures and options outlined in Section 5.

Figure 8 – Problem tree (contextual factors)

In general, as explained above, the specific objectives are clearly problem-driven as they match the specific problems worked out in detail in the evaluation (footnote 7) and described in Section 2. From this perspective, the specific objectives have clear support from essentially all stakeholder groups because the underlying problems are flagged by those groups. For instance, in the OPC 70 , completeness and comparability (SO1) and insufficient detail (SO3) were mentioned as critical gaps in the current statistics by significant shares of respondents across all stakeholder groups for all statistical domains (except marriages and divorces), ranging from 53 respondents on acquisition or loss of citizenship to 94 on international migration (out of 172 respondents overall). Similarly, a lack of timeliness and frequency (SO2) were noted as critical gaps by between 19 respondents (acquisition or loss of citizenship) and 43 respondents (population census outputs). These current gaps and their relevance to EU policies were expressed even more strongly in the targeted survey of Commission departments. NSIs generally also acknowledged these gaps across the consultation activities (OPC, targeted survey and workshops), but typically saw them as less critical than did stakeholder groups representing statistics users. SO4 emerges as a foresight requirement for a new legal framework to minimise the risk of accumulating such critical gaps again quickly in the future.

Furthermore, during interviews conducted in the external support study 71 , many of the stakeholders were asked to provide their views on the specific and operational objectives proposed for the ESOP initiative. Data users and international partners of Eurostat largely supported the policy objectives shown in Figure 8 , though different stakeholders put greater or lesser emphasis or priority on different aspects. Only a couple of the NSIs commented on the policy objectives and highlighted some of the practical difficulties of addressing them. Stakeholders interviewed from NSIs and international organisations also pointed to some of the trade-offs between the policy objectives. For example, there is a trade-off between ensuring the timeliness and the quality/accuracy of the data. There may also be a trade-off between adapting the current system and ensuring consistency in the time series (though some international partners indicated that it can be relatively easy to retrospectively adjust historic time series). Some NSI stakeholders further mentioned potential challenges related to personal data protection, particularly if data are to be disaggregated by very small areas and/or specific population groups, though it is also understood that methodologies are available or under development that could help to overcome these challenges.

5.5.    What are the available policy options?

5.1.5.1.    What is the baseline from which options are assessed?

Under the baseline scenario as analysed in depth in the evaluation, each statistical domain will continue to be implemented based on the three separate legal acts currently in force (Census, Demography and Migration Regulations). One issue is that the Demography Regulation contains a ‘sunset clause’ stating that it will cease to apply on 31 August 2028. However, mandatory data collections under that Regulation are currently generally running well with no reason to assume that this status quo would be seriously questioned by relevant stakeholder groups (including statistics producers) in the baseline scenario. It is therefore assumed at the baseline that the Demography Regulation would simply be amended to extend its application (see Section 5.3). In summary, the baseline has the characteristics described below.

There are no changes regarding harmonisation of statistics across Member States. The limited harmonisation of the population base definitions remains, which would likely mean that the EU-level comparability of national definitions will continue to diverge further, as observed in the evaluation, in the wake of an ongoing and still increasing transition to national register-based systems in most Member States.

Demography, international migration, and population and housing census statistics are implemented as separate statistical processes. Member States may or may not update, streamline or integrate their national processes in future, according to their own considerations. Eurostat keeps the existing data flows from Member States unchanged. This means possible efficiency gains by embracing administrative and other new sources remain uncoordinated at EU level and up to Member States.

There are no new or more detailed statistical outputs adding EU value, as the mandatory data requirements are not updated. Voluntary data collections when agreed with Member States lead to gaps in terms of data completeness and timeliness and cannot sufficiently deliver EU value, as shown in the evaluation.

The legal framework is not flexible enough to adapt the mandatory data collections to evolving policy or data environments. This means that, as needs evolve, even if new voluntary collections should perform better than the baseline assumes, there are no mechanisms to reap the full benefits at EU level and in terms of the efficiency of such achievements. With regard to evolving data environments, the current legislation generally does not sufficiently enable the use of new sources (e.g. B2G), and the Census Regulation currently even excludes such uses.

5.2.5.2.    Description of the policy options

The aim of any policy option beyond the baseline is to deliver on each of the specific objectives set out in Section 4.2. To achieve this, more detailed policy measures were identified under 17 topics, where each topic acts on exactly one specific objective. More precisely, each topic provides a set of different policy measures typically relating to varying levels of ambition for that topic (see  Table  2 ). Each overarching policy option is then constructed bottom-up as a combination of policy measures from all 17 topics, thus acting with varying ambition on all specific objectives.

Before grouping the various measures into policy options, however, a first feasibility assessment of each individual topic and measure was carried out. This process involved assessing how the topics and measures relate to one another (mutually exclusive or complementary), and carrying out an initial screening of their likely effectiveness, as well as their technical, operational, economic and legal feasibility. This initial screening had the results described below.

·A topic on the census frequency (original measures to maintain status quo, split the census into decennial and annual outputs depending on the relevance of the data, and annual census) has been discarded because combinations of other measures factually imply that the frequency of (parts of) the census changes by default.

·Two of the measures relating to the topic on output flexibility of the legal framework have also been discarded. One measure ‘Include limited flexibility to adapt statistics; use voluntary data collections to serve emerging needs’ was discarded because it was assessed as being very similar to the status quo. The other measure ‘Include effective mechanisms to adapt statistics more quickly and efficiently when there is an added value for the EU, even if the administrative cost and burden are high’ was discarded as unrealistic.

The remaining list of the most feasible individual measures, which have been retained for further analysis, is presented in  Table 2 .

Policy options were then constructed by grouping measures based on four main characteristics:

·harmonisation of statistics where the main focus is on the population base definition (status quo, harmonisation with justified exceptions, strict harmonisation);

·integration of statistical processes (status quo, improved statistical processes, integrated statistical processes);

·statistical outputs (status quo, limited upgrade, more expansive upgrade, major upgrade);

·framework flexibility (status quo, limited flexibility, effective mechanisms).

Table 2  lists all policy topics and potential measures with a short description of each. Figure 9 shows the resulting grouping of the measures into seven policy options:

A – Baseline as described in Section 5.1.

Policy options B.1 and B.2 – For these options, there is some simplification of EU-level data collections, but the statistical production processes remain separate and somewhat improved depending on plans of Member States, and the harmonisation of the definition of population remains limited. However, these options introduce two incremental levels of ambition in terms of statistical outputs (limited upgrade for B.1, more expansive upgrade as well as improved flexibility for B.2).

Policy options C.1 and C.2 – In these options, EU data collections are simplified and the statistical production processes remain separate and somewhat improved depending on plans of Member States, but the harmonisation of the population base definition is improved relative to policy options A and B. Policy options C.1 and C.2 introduce the same two incremental levels of ambition in terms of statistical outputs as policy options B.1 and B.2 described above (limited upgrade for C.1, more expansive upgrade as well as improved flexibility for C.2).

Policy options D.1 and D.2 – These are the most ambitious options, introducing strict harmonisation of the population base definition as well as a major upgrade of statistical outputs. These options also include effective flexibility mechanisms to adapt statistics more quickly and efficiently to meet emerging user needs and exploit new data sources or methods. While option D.1 leaves statistical production processes within the competence of Member States, option D.2 requires all Member States to set up statistical population registers, thus initiating the redevelopment and integration of statistical processes at national and EU levels.

From the perspective of the four main characteristics introduced above, these policy options thus implement incremental ambitions to improve as follows (see also Table 1 ).

With regard to the harmonisation of population base definitions, the status quo is maintained in policy options B (both B.1 and B.2). Incremental levels of harmonisation would be introduced for policy options C and D as follows. In policy option C (both C.1 and C.2), a harmonised population definition would be introduced for all datasets, but Member States would still be able to use national population definitions in justified exceptions with limited impact on comparability of statistics across Member States. Policy option D (both D.1 and D.2) introduces a strictly harmonised population definition whereby Member States would be required to use the strict usual residence concept including the 12-months rule to define their populations for all datasets for European statistics purposes.

On statistical processes, all options will revise the legal framework and thus offer an opportunity to streamline it, which would lead to simplified data collection procedures between NSIs and Eurostat. In the most ambitious option (D.2), all Member States would additionally be required to set up a statistical population register. This would align with earlier ESS expert recommendations in the context of population and household frames for social statistics (footnote 60).

In terms of statistical outputs, three incremental levels of ambition are introduced across the options as follows. In policy options B.1 and C.1, the annual statistical outputs for demography and migration would become more detailed to a limited extent, either in terms of characteristics or geographical disaggregation. In policy options B.2 and C.2, statistical outputs on demography and migration would become even more detailed compared to policy options B.1 and C.1. Some currently voluntary datasets and breakdowns, as well as proportionate time series updates, would become mandatory. In all policy options B and C, the timeliness would be improved. In policy option D (both D.1 and D.2), the required statistical outputs on migration and demography as well as regulation of currently voluntary datasets and breakdowns go further than in policy options B.2 and C.2. Moreover, timeliness would be improved and the NSIs would be required to coordinate their census data releases. LAU-level data would become annual, and time series would be updated entirely.

On the flexibility of the legal framework governing European statistics on population, the policy options B-C introduce limited flexibility to adapt statistics to emerging needs by piloting voluntary collections, combined with a legal mechanism to regulate successful pilots and strengthen the legal base to benefit from administrative and other new data sources. Moreover, policy options B.2 and C.2 introduce basic interoperability of statistical population registers. In policy options D.1 and D.2, the flexibility of the legal framework would be substantially improved by setting up full EU-wide interoperability of statistical population registers as well as an effective mechanism to adapt statistics more quickly and efficiently, as proportionate, to cover more general emerging needs.

Table 2 – Detailed overview of policy topics and their policy measures and sub-measures

Figure 9 – Mapping of policy options to topic measures (the cell shading represents the different levels of ambition of each sub-measure: the darker, the more ambitious)

5.3.5.3.    Options discarded at an early stage

Another option has been taken into consideration, in which demography statistics would become deregulated because a sunset clause is included in the Demography Regulation. However, this option has not been included in the shortlist of options for further analysis as it was too similar to the baseline scenario. The rationale is that this sunset clause states that the Demography Regulation will cease to apply on 31 August 2028. This impact assessment looks at the impacts over a 10-year period, until 2031. Therefore, the differences between the baseline scenario and this option would only apply to the 3 last years of the time period assessed. Moreover, the evaluation has shown that the currently regulated data collections are mostly running well, so it can be assumed that an amendment of the Demography Regulation could be put in place simply by extending its application.

6.6.    What are the impacts of the policy options?

This section provides a partial quantitative assessment of the efficiency of the policy options, with impacts monetised to the extent possible based on the available evidence. It also contains a qualitative assessment for each of the policy options covering the following criteria:

·effectiveness, including impacts on fundamental rights and wider environmental and social impacts;

·coherence with the EU policy objectives, including consistency with the proportionality principle;

·efficiency.

A detailed analysis of impacts is provided in Annex 2 based on the external support study (footnote 71).

6.1.6.1.    Quantitative analysis of the options

Quantitative analyses of costs have been carried out for the baseline and each of the six options, over a 10-year period (2022-2031) 72 . The costs associated with the baseline represent an estimation of the current costs, assuming these costs would remain constant over 10 years. The costs assessed for policy options B.1, B.2, C.1, C.2, D.1 and D.2 are incremental relative to the baseline (denoted as A). These only estimate the added costs associated with the implementation of each of the measures/action required by each option, over and above the baseline situation, i.e. costs that would be incurred in any scenario. As such, total costs associated with each option should be interpreted as being equal to the costs reported for the baseline in addition to those reported for every other relevant option. A specific cost itemisation was made and the different ‘cost items’ were then matched to the generic cost itemisation in the Better Regulation guidelines (BRG). Table 13  in Annex 2 (part 1.1) presents the cost types considered in the quantitative assessment of all options and how these relate to the specific cost items.

Most benefits associated with each option, except for some potential cost savings related to reduced administrative burden, have not been quantified and are therefore assessed qualitatively in Section 6.2. A more detailed costs-and-benefits assessment of each option is provided in Annex 2 .

6.1.1.6.1.1.    A – Baseline

In the baseline scenario A, the harmonisation of population base definitions will remain limited by default provisions under the current legislation, statistical production processes will also remain up to Member States and the current requirements in terms of statistical outputs will continue to be the same. Finally, the current lack of flexibility to address evolving statistical needs as well as to benefit from opportunities from new data sources will persist.

The cost assessment of the baseline scenario was carried out based on the evaluation results. For all policy measures included in the baseline, and as always in a baseline scenario, the status quo will remain. It should be noted that even though there is a sunset clause in the Demography Regulation, it was assumed in the baseline that a minimal amendment would be adopted to extend its application (see Section 5.3).

All recurrent costs incurred by the Census, Demography, and Migration Regulations were estimated to remain constant and were calculated over a 10-year period, for both the Commission (Eurostat) and the 27 Member States and their respective NSIs. However, the baseline scenario is not one of ‘no change’. The baseline is dynamic in the sense that it takes into account ongoing and/or planned activities and takes into account the effects of national action (both legislative and voluntary) that can be expected to be taken regardless of any future EU initiative. Therefore, costs and cost savings associated with maintaining statistical population registers (for those Member States that have implemented or are planning to implement one regardless of this initiative), implementing a strict usual residence definition of the population (for those Member States that currently adopt this definition) or providing voluntary statistics are also included in the dynamic baseline. The rationale for the including these elements in the baseline is as follows.

·In terms of statistical processes, Member States that already have a statistical population register 73  are assumed to continue to incur costs to keep these registers up to date. These costs are therefore included in the baseline. Similarly, some NSIs plan to set up a statistical population register in the near future 74 and will therefore incur implementation costs as well as cost savings related to subsequent efficiency gains, regardless of a potential new EU legal base. These upcoming effects cannot be attributed to a potential new EU requirement on NSIs to set up and maintain national statistical registers (see measure 1.2), and are therefore also included in the baseline 75 .

·On the harmonisation of population definitions, some Member States already use a population base definition based on the strict usual residence concept 76 . Following the same reasoning as above, those Member States already incur related costs and will continue to incur them. These costs cannot be attributed to a potential new EU requirement to use that same definition (measures 2.2 or 2.3) and are thus included in the baseline.

·Similarly, some Member States already provide datasets voluntarily. Data currently collected on a voluntary basis incur costs for both the Commission (Eurostat) and NSIs. These costs are likely to remain whether a new regulation makes this voluntary data collection mandatory or not. These costs associated with voluntary data are therefore also included in the baseline.

·

Table 3 – Overview of baseline total and incremental costs in million 2021 EUR by policy option, and split by one-off and recurrent per 10 years on Eurostat and on all aggregated 27 Member States.

Table 4 – Overview of median costs per capita for the three Member State groups (see text) in 2021 EUR per capita.

Table 5 – Overview of average costs per capita in the three Member State groups (see text) in 2021 EUR per capita.

·

In summary, it is thus important to note that the baseline costs estimated as part of this impact assessment go beyond the costs associated with the current legal framework, in that they also include current costs incurred on aspects affected by some policy measures and options. As described in Section 2.2.4 and substantiated by the evaluation findings, there is a clear pattern expected in the overall baseline costs depending on whether NSIs can use administrative and other appropriate sources efficiently or not. While it is not straightforward to categorise Member States under this criterion, the existence (or not) of a statistical population register is a good proxy. Therefore, Member States are split broadly into three groups as follows:

·group 1 – already have a statistical population register 77 ;

·group 2 – currently do not have a statistical population register but are planning to set up one in the near future regardless of any EU intervention 78 ;

·group 3 – do not have and are not planning a statistical population register 79 .

Indeed, the contract support study (footnote 71) has confirmed the expected pattern of higher per capita costs in groups 2 and 3 than in group 1, both in medians ( Table 4 ) and in averages ( Table 5 ) across members of each group. Note that the pattern holds despite cost savings included in the baseline for group 2, likely due to the very conservative approach taken in the estimation model (footnote 75).

6.1.2.6.1.2.    Incremental costs incurred by policy options B-D

A comprehensive and detailed quantitative assessment of incremental costs incurred by policy options B-D is provided in Annex 2 (part 1.1). Table 3 shows the aggregated baseline and incremental costs in 2021 EUR by policy option. The cost model is aligned to the basic characteristics of the policy options in terms of harmonisation – process integration – upgrade of statistical outputs (see Section 4.2 and Figure 9 for context). Incremental costs related to flexibility characteristics are considered indirect and largely depend on unknown context factors (e.g. specific new data needs or data source environments in the longer-term) so they could not be quantified.

Table 3  also shows that the largest part of both baseline and incremental costs (97.7% to 98.5% depending on the policy option) are, or would be, incurred by Member States, more precisely by their national statistical production systems coordinated by the respective NSIs. This reflects the structure of the ESS governed under Regulation (EC) No 223/2009 (footnote 2), where Member States produce statistics under the coordination of Eurostat and governance of the ESS Committee (ESSC). Key features of each policy option regarding incremental costs are summarised below.

B.1 causes incremental costs compared to the baseline solely through the limited upgrade of statistical outputs, including for instance improved detail and timeliness of statistics and decennial population grids.

B.2 causes higher incremental costs than B.1 solely through a more expansive upgrade of statistical outputs, including improved frequency and timeliness, regulating some of the currently voluntary datasets and introducing annual population and housing grids as well as basic equality data.

C.1 and C.2 cause higher incremental costs than B.1 and B.2 respectively solely through the improved harmonisation of the population base definition based on the strict usual residence concept, subject only to justified exceptions (measure 2.3).

D.1 causes higher incremental costs than C.2 by fully harmonising the population base definition without exceptions and through a major upgrade of statistical outputs, including monthly demographic data, annual data at LAU level and more detailed equality data.

D.2 causes higher incremental costs than D.1 solely through the obligation on Member States in group 3 to set up a statistical population register. However, this option also generates long-term recurrent cost savings for these Member States as described in Section 6.1.3.

Regarding quantitative incremental cost estimates by policy option and Member State group, Table 4 and  Table 5  show that average incremental costs per capita are generally increasing from group 1 to group 3, while median incremental costs per capita follow the opposite pattern (decreasing from group 1 to group 3). This is in contrast to the baseline assessment in Section 6.1.1 (where per capita cost patterns are consistent for averages and medians over group members), and it suggests that the cost model provided in the contract support study (footnote 71) was not detailed enough to allow reliable quantitative estimates broken down by specific Member States or Member State groups.

Nonetheless, the stakeholder consultation and regular expert consultations with NSIs provide three broad qualitative patterns described below of how adaptation or modernisation challenges vary across Member States.

1.While upgrading statistical outputs does depend a lot on the specific details of the upgrade (e.g. whether only timeliness or frequency is improved or whether new variables or breakdowns are to be added that may or may not be available from sources already in use), it is generally acknowledged that an integrated statistical production system in place, with a statistical population register as a backbone, does reduce the costs of such upgrades. This means that Member States in group 3 (and in group 2 while national efforts to set up a statistical population register are ongoing) will likely face more severe challenges than group 1 in adapting to any of the policy options. The same pattern is also relevant when it comes to preparing for additional flexibility (either to address changing policy needs or to set up data-sharing systems with new sources).

2.Most of the modernisation activities covered to a varying degree in the policy options would entail upfront investment (adaptation costs) that are not necessarily proportional to country size in terms of population or gross domestic product. For instance, upgrading IT infrastructure or introducing new data products (e.g. infra-annual publications) typically entails incremental financial and human resources that are at least partly comparable in volume across countries, whereas smaller countries typically have fewer staff and smaller budgets for official population statistics. This means that these countries are generally expected to face relatively more severe challenges than larger countries in adapting to any of the policy options.

3.The correct implementation of a harmonised population definition based on a strict usual residence concept as a basis for annual and more frequent statistics is challenging for all Member States, in particular in combination with the expected upgrading of statistical outputs. It is generally acknowledged that an integrated statistical production system based on administrative and other readily available sources is needed to produce detailed statistical outputs frequently and with shorter deadlines. However, multisource production systems ultimately based on administrative population registers have well-known coverage issues with respect to the usually resident population. While efficient methods like signs of life are available in principle to control over-coverage (i.e. to identify persons not factually present at a given address), ESS 80 and international 81  guidance generally acknowledges that the only effective method currently available to check for systematic under-coverage of the registers against the usually resident population is to carry out dedicated field surveys. The costs of such field surveys scale with their frequency and geographic area covered and would be incurred by all Member States using (or transitioning towards) integrated register-based production systems with a proper implementation of the usual residence concept.

From these points, in particular 1 and 2, it may be roughly summarised that the smallest Member States in groups 2 and 3 are likely to face relatively the most severe challenges in adapting to any of the policy options.

6.1.3.6.1.3.    Incremental cost savings (benefits) incurred by policy option D.2

While incremental quantitative costs could not be broken down by specific Member States or Member State groups due to a lack of granular information, the quantitative evidence for baseline costs by Member State is more robust and makes it possible in principle to model some specific cost savings. This is particularly true for census costs, for which there is comparably rich granular information from the 2001 and 2011 rounds that may be used to model efficiency differences between production systems, either based on statistical population registers or having to cope without them. This is used here as a rough proxy to estimate potential cost savings expected specifically from policy option D.2 that would introduce such statistical registers for all Member States.

The detailed baseline cost assessment by Member State has shown that there is a significant gap between Member States producing their annual and census statistics from an up-to-date statistical population register (group 1 in Section 6.1.1) and Member States currently coping without such a register (groups 2 and 3). According to the baseline cost assessment, the median cost of a traditional census across EU-27 countries was roughly 20 times the median cost of a register-based census both in the 2001 and the 2011 census round. A rough estimation of respective potential cost savings in the future, accounting for scale effects from changing population sizes, amounts conservatively to EUR 2 (EUR 1-4 with sensitivity margin of 50-200% around the best estimate) per capita less in a census round 82 , incurred only by six Member States not currently planning to set up a statistical population register 83  ( Annex 3 ). The total potential cost savings at EU level per future census round could thus amount to EUR 281 million (between EUR  141 million and EUR  563 million according to sensitivity margin). This is roughly between two and eight times the total combined one-off and recurrent (over 10 years) incremental cost estimate for all affected Member States to set up and maintain such registers (see Annex 2 part 1.1.7).

Similarly, recurrent cost savings can also be expected for the production of annual and infra-annual data, even at the highest demographic and geographic details and improved timeliness, due to the continuous availability of a single statistical source at microdata (individual person record) level from which the tabulations can be extracted. This would incur a substantial discount on the incremental costs related to the major upgrade of statistical outputs in policy option D.2, but such a discount was impossible to quantify reliably on top of the estimate of these costs.

6.2.6.2.    Qualitative analysis of the options

Most benefits associated with each policy option, except for some potential cost savings related to reduced administrative burden (see Sections 6.1.3 and 8.2), have not been quantified. This is in part due to a lack of available data on the benefits associated with population statistics, and in part due to the nature of the benefits that are likely to be generated by this initiative. For many of the benefits considered, their effects are more indirect and variable across Member States and stakeholder groups, which make them challenging to quantify or monetise. For example, the benefits to data users from increased access to high-quality European statistics on population would depend on several additional factors, such as how these data would be used or the cost of accessing data through alternative sources. In addition, benefits to policymaking at EU or other governance levels, from improved access to reliable, comparable population data across the EU, would be difficult to monetise since these would depend on several contextual factors, such as the types of policies at each governance level that rely on population data or what the impact of evidence quality would be.

For these reasons, respective benefits have been considered qualitatively; and Table 6  shows to what extent each of the policy options B-D achieve incremental benefits compared to the baseline, by beneficiary groups and specific benefit items. The tables in Annex 2 (part 1.2) provide a detailed and systematic qualitative assessment of the proportionality and incremental impacts by option, based on the policy measures included in each option (see  Figure 10 ) and how these affect the effectiveness, coherence with wider EU objectives and efficiency; and applying the qualitative scoring method introduced at the beginning of Section 6.

Key features of each policy option regarding incremental policy measures and their benefits on stakeholder groups are summarised below. Stakeholder groups’ views on each policy option were collected at three dedicated option validation workshops: one with participants from NSIs, one with representatives of Commission departments and one with other professional users (institutional users, researchers, NGOs). Earlier consultation activities could not refer directly to the policy options as these were not yet finalised at the time, but the OPC did ask for respondents’ priorities among the various potential measures considered under each of the policy topics (see  Table 2 ). Mapping their answers to policy options ( Figure 9 ) makes it possible to calculate an OPC-based relative preference for the policy options by stakeholder group. The result, shown in  Figure 10 , corroborates the following summaries by option.

B.1 causes modest incremental benefits from better data evidence for various professional users, including policymakers (measures 3.2, 4.2, 6.2, 8.2, 9.2, 10.2, 11.2) but fails to deliver on key aspects like geographic granularity or specific topics of policy interest. Statistics producers, including Eurostat, benefit moderately from simplified procedures due to a streamlined legal framework for all datasets and lighter adaptation processes to evolving data needs (measure 15.2). Source data providers benefit from simplified procedures due to a clearer legal base for sharing relevant data (measures 17.2 and 17.3). During the option validation workshops, participants from NSIs and professional users (Commission departments, other institutional users and researchers) identified policy option B.1 as realistic, i.e. feasible to implement 84 .

B.2 causes larger incremental benefits for professional users than B.1 in terms of further improved data evidence for various policy and institutional purposes (additional measures 3.4, 4.3, 6.3, 7.3, 8.3, 10.3, 11.3, 12.2, 13.2, 14.2, 16.2). Statistics producers additionally benefit from the interoperability of statistical population registers for simplified sharing of relevant data (measure 17.4). The feedback from the option validation workshops was mixed: a majority of NSI participants stated that the measures under policy option B.1 would be unrealistic due to the increased burden involved, while the opposite view was broadly expressed among professional users (see footnote 84).

C.1 (C.2) causes larger incremental benefits relative to B.1 (B.2) solely through improved EU-level comparability across Member States, stemming from the harmonised population base (measure 2.3). Notably, large majorities of OPC respondents in all stakeholder groups (from 73%, or 8 out of 11 respondents among statistics producers, up to 93% or 39 out of 42 respondents among professional users) put a high priority on introducing a strictly harmonised population definition (either for all datasets or with justified exceptions) 85 . In the targeted survey with NSIs, maintaining the status quo of non-harmonised population definitions was considered to have minor or moderate negative impacts by a majority of the 29 respondents (69% on usability of European data for policymaking, 79% each on comparability of national data and on quality of European data), while production costs was the only area where a majority (55%) of respondents saw a positive impact (38% ‘major’, 17% ‘moderate’ or ‘minor’) 86 . During the option validation workshops, most participants from NSIs as well as from among professional users found option C.1 feasible to implement, whereas the views on C.2 were split: not realistic according to a majority of NSI participants, realistic according to professional users (see footnote 84).

D.1 brings larger incremental benefits for users than C.2, with even better EU-level comparability through full harmonisation (measure 2.2), and a major improvement of data evidence (additional measures 5.2, 7.2, 9.3, 14.3, 16.3). Statistics producers and Eurostat both benefit from further simplified processes due to effective adaptation mechanisms to evolving data needs (measure 15.3) and from full EU-level interoperability of statistical population registers (measure 17.5). In the option validation workshops, participants from NSIs unanimously found D.1 unrealistic, driven mostly by expectations of excessive additional burden on NSIs, while participants among professional users generally stated that the option is realistic (see footnote 84).

D.2 causes larger incremental benefits than D.1 solely through the introduction of statistical population registers across all Member States (measure 1.2). This benefits users with higher data quality in general and quicker adaptation to their evolving needs. Statistics producers also benefit substantially over time thanks to the integrated statistical production processes. In the option validation workshops, NSI participants unanimously agreed that D.2 is unrealistic, largely due to the costs involved in transitioning to statistical population registers and due to the subsidiarity and proportionality concerns around them, while this option was largely viewed positively by professional users (see footnote 84). In the OPC, a large majority (84%, or 27 out of 32 respondents) among professional user groups also prioritised the introduction of statistical population registers, while the opinions of statistics producers were split (50%, or 5 out of 10 respondents prioritised maintaining the status quo) 87 .

Table 6 – Beneficiaries and qualitative level of incremental benefits expected by policy options compared to the baseline (scale: no increment, ‘+’, ‘++’ or ‘+++’).

Table 7 – Comparative overview policy option assessments (see Annex 2 part 1.1 for detailed quantitative cost estimates and part 1.2 for detailed qualitative assessment scores)

7.7.    How do the options compare?

In line with the Better Regulation guidelines (BRG) 91 , a multi-criteria decision analysis (MCDA) should be used to compare the different options in terms of compliance with the proportionality principle, effectiveness against specific objectives and wider impacts (fundamental rights, social, environmental), coherence with EU objectives, and efficiency. Table 7 summarises the assessment of all policy options for each of these criteria, by also including views of key stakeholder groups. A detailed reasoning for each of the assessments, presented by policy option, is provided in Annex 2 . A comparison of policy options synthesising these findings is presented in this section.

On proportionality, four questions were answered (see also Section 4.6.1 of the external support study, footnote 71):

1.Does the option go beyond what is necessary to satisfactorily achieve the objectives?

All options pass, as all policy measures considered are clearly linked (albeit with varying ambitions) to specific problems with the current EU legal framework, as established by the evaluation.

2.Is the scope of the option limited to those aspects that Member States cannot achieve satisfactorily on their own and where the EU can do better?

Member States alone clearly cannot provide European statistics fulfilling required common quality standards, especially on completeness, comparability, frequency, timeliness and geographic and topical detail, and thus relevance for EU policy needs.

Policy options B.1 through D.1 pass, as all their measures are clearly linked to creating or improving such common standards. The status of option D.2 is uncertain due to measure 1.2 requiring statistical population registers in all Member States.

While the goals of improving efficiency, flexibility and comparability across all Member States are proportionate, some NSI experts have expressed concerns that national systems would lose flexibility to find their own solutions 92 . However, it is the freedom of movement in the EU internal market under EU law that presents characteristic challenges for statistics, such as highly dynamic and further increasing mobility, cross-border work/family patterns and several places of residence split across regions and countries becoming common practice.

Experience over the past years suggests that national approaches are ineffective to produce genuine European statistics 93 , so that EU action would be proportionate in the form of (ideally interoperable) national population registers solely for statistical purposes.

3.Are costs for the EU, national governments, regional or local authorities, economic operators or the public commensurate with the objectives of the initiative?

Based on the quantitative analysis of the costs (Section 6.1), all policy options are judged as passing this question. This includes the costliest option, policy option D.2, because a large part of these costs (around the statistical population registers) would be an investment in the capacities of a few Member States (group 3 in Section 6.1) that is expected to be offset by long-term savings and wider benefits at EU level.

4.Is the form of action (choice of instrument) as simple as possible and coherent with the satisfactory achievement of the objective and effective enforcement?

The choice of legal instrument (one new regulation or an update of existing legislation) to implement the policy options has not been explored in depth in this study. Any of the measures entailed can, in principle, be achieved by each approach, whereas integrating the current three basic acts into one would have natural benefits for streamlining all related processes. Thus options B.1 through C.2 pass this question, while whether D.1 and D.2 pass remains uncertain: D.1 is so ambitious on statistical outputs that there are increased risks of significant non-compliance across Member States. The statistical population registers required under D.2 can be seen as facilitating the statistical infrastructure in this respect. It is known from experience that statistical population registers are not simple to implement and may compete with other measures that address comparability issues at EU level (e.g. harmonised population definition, internal migration flows) 94 .

Regarding effectiveness in achieving the specific objectives (SO), all options were scored against the relevant policy measures included (see  Figure 9 ) for each SO.

·SO1 (completeness, coherence, comparability) took into account policy measures to harmonise the population definition (policy topic 2), ensure completeness of voluntary statistics (topics 12, 13) and ensure a consistent time series (topic 16). Moreover, statistical population registers (topic 1) were considered to reduce implementation risks of a harmonised population definition. Therefore, option D.2 scored highest (+4), followed by D.1 (+3) and C.2 (+2.5).

·SO2 (timeliness and frequency) took into account measures to improve the publication frequencies (topics 5, 7) and the time between reference and publication dates (topics 8, 9) of various statistical products. Statistical population registers (topic 1) were considered to improve production efficiency and thus reduce compliance risks. Option D.2 scored highest (+4), followed by D.1 (+3) and by B.2 and C.2 (+2).

·SO3 (geographic and topical detail) took into account measures to improve the geographic detail of existing statistics (topics 3, 4, 6) and introduce new statistics or more details on topics of key policy interest (topics 10, 11, 14). Statistical population registers (topic 1) would allow georeferencing of the whole population and also improve the availability of statistical detail and thus reduce compliance risks. Option D.2 scored highest (+4), followed by D.1 (+3.5) and B.2 resp. C.2 (+3).

·SO4 (input and output flexibility) took into account measures to adapt the statistical products more quickly and efficiently to evolving policy needs (topic 15) and enable better use of existing or new data sources that may become available in the future (topic 17). Statistical population registers (topic 1) would streamline the integration of data sources and also be more flexible to extract new statistical outputs, and thus reduce compliance risks. Option D.2 scored highest (+4), followed by D.1 (+3.5) and B.2 resp. C.2 (+2).

Options were also scored against wider impacts on fundamental rights, society and environment. While the scores generally followed the patterns of SO scores, the impact of statistical population registers in option D.2 were particular, in that some concerns were expressed about certain risks of negative impacts on fundamental rights (related to personal data protection) and social impacts (doubts around the social licence). However, they would have specific positive environmental impacts by embracing digitalisation and efficient production systems, thus reducing the carbon footprint of statistics production, especially when full field enumerations for traditional censuses become obsolete. Overall, on effectiveness, option D.2 scored highest (+3.71) followed by D.1 (+3.14) and C.2 (+2.21).

Regarding coherence with EU objectives, all options were scored against the relevant policy measures on how ambitiously statistics move towards EU-level comparability of population definitions (topic 2), deliver on the data needs for current EU priorities (topics 3, 6 as regards urban/rural integration and the Green Deal; topics 4, 10, 11 for migrant integration; topic 14 for equality) and improve flexibility to adapt to evolving policy needs and data environments (topics 1, 15, 17 for enabling new data sources and interoperability of ESS production systems). Overall, on coherence options D.1 and D.2 scored the highest (+4) followed by C.2 (+3) and B.2 resp. C.1 (+2).

Regarding efficiency, baseline and incremental costs by policy option have been quantified in Section 6.1. The baseline (A) sets zero incremental costs or benefits but is considered dynamic in that key impacts of modernisation already planned in some Member States (group 2 of Section 6.1) are factored in. However, most of the benefits could not be quantified, so incremental costs and benefits by option can only be confronted qualitatively. Nevertheless, three broad qualitative trends can be identified.

1.Incremental effectiveness grows steadily with incremental costs. This means that a simple retail analogy applies: the more expensive option buys more or bigger benefits – in increasing order, this is B.1, C.1, B.2, C.2, D.1 and D.2. However, a key aspect to note is that most costs are incurred by statistics producers, while the most relevant direct beneficiaries are statistics users (see  Table 6 ). Nonetheless, at a higher level, both costs and direct benefits are incurred by the same actors, namely public administrations across the EU. Researchers and other professional users also benefit directly from better statistics availability, while society as a whole benefits from key indirect benefits, mostly through better policymaking.

2.Introducing a harmonised population definition based on a strict usual residence concept (going from option B.1 to C.1 or from B.2 to C.2) is estimated to be almost as expensive as all the significant improvements on statistical output and flexibility together (going from B.1 to B.2 or from C.1 to C.2). This is because, even in advanced integrated statistical production systems based on administrative registers, complex specific efforts are needed to capture the actual presence of persons at a registered place of residence (i.e. control over-coverage) and the size and characteristics of population groups actually present on a given territory but not registered there (i.e. control under-coverage). Nonetheless, improving on this aspect has a leverage effect on higher quality, not only of population outputs from this initiative, but of all European social statistics (since accurate enumeration of the whole population is the backbone of other statistical tools such as collecting data from samples).