Other Contributors to Review Research Kevin Deardorff, United States Bureau of the Census
Dave Elliot, Methodology Group, Office for National Statistics
Ana Franco, Eurostat
Richard Gianella, Immigration and Nationality Directorate, Home Office
Julie Jefferies, Population and Demography Division, Office for National Statistics
Jonny Johansson, Eurostat
Justine Shelley, Research Development and Statistics Directorate, Home Office
Lars Ostby, Statistics Norway
Aidan Punch, Central Statistical Office, Ireland
Denis Till, Population and Demography Division, Office for National Statistics
Rhian Tyler, Population and Demography Division, Office for National Statistics
Users and Stakeholders See User Consultation Annex
Annex 2 User Consultation
Summary A wide range of users of the international migration outputs were consulted about their needs for migration statistics, their current uses of the statistics, their satisfaction or dissatisfaction with current outputs, and suggestions for continuation and change to the way the outputs are produced and disseminated. Users consulted included central and local government officials and non-government analysts. The most fundamental need for international migration statistics is for the monitoring of changes in the basic demographic characteristics of the UK population: its size, geographic distribution and composition by age and gender. For this purpose, net immigration statistics (immigration minus emigration), at the level of constituent country and government region and local authority, are required. Statistics on the size, composition and duration of stay of immigrants and emigrants are required to address additional key user needs. These include informing policy on the migration of non-citizens subject to immigration control and government and business planning at both the national and local levels. Statistics on both demographic and workforce-related attributes of migrants are needed for these purposes. Statistics needed specifically on persons subject to immigration control include their legal statuses with regard to residence and work, both currently and at the time of their entry to the UK. Users expressed generally high satisfaction with the range and frequency of statistical outputs on international migration. Greater accuracy of the total international migration estimates and more reliable estimates of immigration and emigration at local geographical levels, however, is needed in the years between the censuses. Greater comparability of migration definitions and more immigration-related variables in survey, census and administrative data sources are needed for the reconciliation of estimates across sources and for the reconciliation of migrant stocks and flows. Approximate estimates of the numbers and characteristics of migrants entering or staying illegally are also needed.
National Statistics (NS) outputs of international migration (IM) statistics are chiefly produced by the Home Office (HO) and the Office for National Statistics (ONS). The General Registrar’s Office for Scotland (GROS) and the Northern Ireland Statistical Research Authority (NISRA) estimate IM as a component in their population estimates. Key representatives of each of these organisations were consulted both as users and as indirect sources of information about customer needs. A document detailing the statistical requirements for immigration policy was also produced and circulated around Home Office immigration policy makers for comment.
Other users of IM statistics were also consulted directly and given opportunities to make submissions to the Review Team in a number of ways. First, members of the National Statistics Population and Migration Theme Working Group initiated the development of the Review’s scope and priority classification. The Group made recommendations on the Review Board’s composition to ensure representation of a broad cross-section of users, and generated awareness of the forthcoming Review among them. The summary of user input related to international migration that was elicited as part of the 1998 ONS Review of Population Statistics was also made available to the Review Team.
After agreement by the Board of the Review’s scope, the Project Initiation Document describing this scope was placed on the NS website on 19 August 2002, and user submissions to the review invited by 1 November 2002. Announcements were made to the academic and applied population and demography community at the September 2002 annual meeting of the British Society for Population Studies, and at an October 2002 special meeting of the Population User Group section of the Royal Statistical Society. User needs were communicated directly at these meetings, and subsequently in the form of written submissions from the following organisations: Cabinet Office Strategy Unit; European Commission (Eurostat); Herefordshire Council (Policy and Community Directorate); the Liaison Group on Population Statistics (Local Authority side); Suffolk County (Policy and Regeneration Division); and the Tees Valley Joint Strategy Unit.
Finally, approaches were made to a cross-section of key users, resulting in interviews with representatives of: The Bank of England (Structural Economics Analysis Division); Department of Work and Pensions (Economy and Labour Market Division); Government Actuary’s Department (Demography and Statistics Division); and the Greater London Authority (Data Management and Analysis Group).
Consultations were aimed at eliciting input including descriptions of (1) current uses of IM statistics; (2) types, quality and timing of needs for IM statistics; (3) satisfaction or dissatisfaction with current IM statistics; and (4) suggestions for continuation and change to the way IM statistics are produced and disseminated. IM statistics are used both directly and as inputs to population estimates and projections and users expressed needs for IM statistics relating to both these dimensions.
Uses of international migration (IM) statistics are classified into three categories: to measure population change, both at UK and lower geographic levels, using net IM as one component; to estimate and analyse emigration and immigration and their social and economic impacts; and to monitor the immigration and emigration of persons subject to immigration control.
Measuring National (UK) Population Change
The National Statistics offices of the UK (ONS, GROS and NISRA) measure population change at national and local levels. A major focus of user needs for IM statistics concerned the contribution of IM to the overall measurement of population change. A common theme was the recognition of the much greater uncertainty of the IM component as compared to the births, deaths and internal migration components. At the UK level, births and deaths are the only other components of population change. These events are recorded to a very high degree of completeness. Therefore any error in measuring population change is almost entirely due to error in measuring IM. Users of estimates of overall population change are therefore significantly dependent on the accuracy of measurement of net IM (the difference between immigration and emigration). Users noted the growing importance of IM as a source of population composition and change increases the importance of IM measurement for the quality of estimates of population composition and change.
Population change is measured annually as part of the Population Estimates (PEs). IM statistics, primarily from the International Passenger Survey (IPS), but also with significant adjustments from Home Office administrative data, are used to estimate the net IM component of population change in the inter-censal PEs. Quarterly estimates of population change will be initiated in late 2003. Population change is measured with annual frequency primarily along the dimensions of age and sex. Ethnicity and marital status are secondary dimensions along which population change is measured in inter-censal periods. Every ten years, the Census provides measures of population change on a number of additional dimensions. The Census also provides a decennial check on the measurement of population change in the PEs. PEs for the previous ten years (the inter-censal period) are revised after each Census.
User consultation took place over a period that included the first release of Census 2001 results and the consequent downward revision of PEs over the decade up to the 2001 Census. The comparison of PE and Census results, and IM’s likely role in differences between the two sources, was therefore an issue of great topical interest and importance to many users.
The estimates of population change from the inter-censal PEs are also an integral part of the checks on the quality of Census data before their release. Users also expressed a need for better IM statistics to understand and to independently evaluate the Census, and to understand population change as measured alternately through PEs and the Census.
Large differences between PEs released before the 2001 Census and the revised PEs following the 2001 Census raised concerns with users about the increasing impact of IM statistics on the quality of National Statistics outputs in general, since so many statistics depend on population estimates as a component. One example raised by a key user is the estimation of labour force and employment change. These are based on Labour Force Survey (LFS) estimates. LFS sample weighting is in turn based on PEs. Changes in the size of the working age population are particularly sensitive to IM and, in the short term, insensitive to changes in fertility and mortality. Thus estimates and forecasts of aggregate change in the labour force, a critical element of the macro-economy, are significantly dependent of the accuracy of IM statistics. This was highlighted in the November 2002 re-estimation of recent years’ labour-market statistics following post-Census revision of population estimates.
Another example of the significance of any error in IM statistics for users of PEs relates to the use of PEs as denominators of population rates. Population rates include those for births, deaths and socio-economic and health characteristics. Error in IM estimation at the national level between 1991 and 2001 was perceived in the words of one user to have had “a material effect on estimates of mortality and health”.
Timing of international migration statistics as inputs to population estimates was also mentioned by both users and producer agency representatives. On one hand, some users considered accuracy more important than speed. On the other hand, users of the Labour Force Survey (LFS) expressed concern that the “grossing up” of that survey’s most recent statistics is effectively based on projected rather than estimated international migration. The need for faster release and shorter periodicity of IM estimates is also set to increase with ONS plans for quarterly as well as annual population estimation. Consistency in historical time series of IM are also needed by users of the statistics in aggregate forecasting models.
Users also offered evaluations of, and suggestions for changes to, the data sources and methods currently used to measure the IM component of population change. Limitations on what a survey like the IPS can be expected to produce were expressed. As one person put it, “the biggest problem is knowing who is a migrant”, elaborating that the IPS measures only intention and that expressions of intention depend among other things on the government’s welcome.
One key user expressed concern about “vulnerability to too much reliance on the IPS” to understand an increasingly complex and large set of migration flows. Another suggested “[eventual] full recording of all arrivals and departures” and “in the short term…a ‘super-IPS’…[in which] both intending immigrants and intending emigrants would be over-sampled”. Another user noted that the migration “filters” currently applied only to immigrants, and argued they should also be applied to emigrants.
Some users expressed concern about the estimates of migrants on the boundaries between the IPS and Home Office (HO) data. Asylum applicants are omitted from the IPS if they make their application at the border point, and are subsequently added from HO’s asylum applicant statistics. Concern was expressed, however, about the capturing of applicants’ dependants in the combination of IPS and HO data. Users also expressed dissatisfaction about the level of transparency of the procedures used by ONS to combine the IPS and HO components of total international migration flows.
Cross-checks on IPS estimates through regular and frequent comparisons to UK household survey data and to other countries’ flow data was suggested “to verify the accuracy of the scale of migration and the characteristics of those who migrate”. Cross-checks to administrative records such as those from the health system and electoral registers, school enrolment and child benefit records, National Insurance records and tax authority records, as well as HO records, were all suggested. The use of health record system checks on the continued presence of “inactive” customers was also suggested.
A further user concern expressed was that marriage associated with immigration and short-term absences from the country for the purposes of getting married complicates the estimation of the population by martial status, and that these types of migration associated with marriage are not well picked up by current IM statistics.
Local Area, Regional and Component-Country Population Change
Population Estimates are made at geographical levels including countries of the UK, government regions and local authorities. Estimates of net IM by age and sex are therefore needed at all these levels.
Local area users also expressed needs for IM statistics classified by ethnicity in order to meet legal and policy obligations on reporting on minority ethnicity population inclusion. The need for accurate estimates by age and family status were also expressed for purposes including labour market and housing analysis and planning. The marital status of migrants is also a useful characteristic for users of IM statistics.
Users at the sub-UK level expressed two major types of concern about the way IM estimation is done at these geographic levels. The first is that the small sample size of the IPS is a severe handicap for the purpose of estimating a local geographical dimension to the IM component of population change. Statistical units charged with incorporating international migration into population estimates at the local and component-country level also noted this as an important issue.
The second major user concern at the sub-UK level is that the linking of international immigration with subsequent internal migration may be poor. Internal migration is measured by National Health Service (NHS) systems of patient registration and de-registration, whereas immigration is measured in the IPS by a question about intended address in the UK. A major user concern is that final destination of immigrants is inaccurately represented by the IPS’ question. Local authorities close to major international airports - especially London boroughs - are concerned that immigration to their areas is overestimated, although compensating underestimation at the areas of onward, unrecorded internal migration is also cited by local authority users as a problem. Representatives of the devolved administrations have expressed this concern as major international airports and major IPS collection points are in England, while eventual destination may be Wales, Scotland or Northern Ireland.
Both users and PE producers noted the problem of the mismatch between immigration estimates based on first intended destination from the IPS and internal migration estimates based on moves between first NHS registration area and next NHS registration area. The problem here is of missed moves between a first destination without any NHS registration and a second destination with an NHS registration.
The combined effect of the above two types of threat to the quality of IM estimates at the sub-UK level was perceived by users to be very large. In the words of a local authority user, it results in “widespread and substantial misallocation of resources from Central Government to local authorities”.
Estimating and Analysing Emigration
To estimate the IM component of population change, net IM (the difference between immigration and emigration) only is required. Knowledge of the component gross immigration and emigration flows, however, is also of great interest to users. Understandably, those users most interested in emigration are from regions where volume of emigration is highest and where net IM is negative. Interest at the UK level has been heightened by a recent rise in the outflow of migrants from the UK.
Information including the characteristics of emigrants is needed by users so that analysis and forecasting of outflows can take place. This is needed, for example, to enable accurate assessments to be made of what skills the UK is losing through outward migration and whether any particular policies would encourage them to stay. The IPS collects information on reason for leaving the UK, including work and study. Users also expressed needs for information on the level of skills and occupations held by departing residents.
Users also mentioned the need for knowledge of emigration for research studies on health, mortality and fertility. This is especially the case for studies involving linked records (including those based on data from the ONS Longitudinal Study). In these studies, it may not be known whether a subject of the study has really not experienced an episode of ill health or death, or whether instead that event occurred but did so outside the UK. A similar problem occurs in studies of fertility using linked records. In this case, it is not known if a birth does not occur or whether it occurs outside the country.
Users expressed several concerns about present data sources and methods specifically on emigration. First, users noted that ONS’s 2001 Census Quality Assurance procedures suggested a bias towards underestimation of emigration of UK citizens in estimates from the IPS, and that this bias appears to have been largest for males of working age. This group has important implications for, among other things, labour force and employment.
A second concern was that the statistical variability about estimates of emigration flows, including both their size and composition (age, sex, socio-economic characteristics), is too high.
UK national statistics users were also concerned about the ability of current statistics to capture the changing flows between the UK and Ireland. An additional problem for Northern Ireland is that in distinguishing emigration, often through Dublin, from internal UK migration between Northern Ireland and England.
Estimating and Analysing Immigration
Since the mid-1990s, net immigration has accounted for more than half of the total UK population growth. The effects of such changes in the size and in the characteristics of the UK population are needed for both social and economic policy making (including education, health, employment and monetary policies). As well as demographic details, details of migrants’ skill levels, the knowledge they bring with them to the UK and their participation in the labour force are important.
To be able to assess the social effects of immigration, users expressed a need for statistics on immigrants’ use of public services – housing, education, health, public transport – and for social indicators that make comparisons of immigrants over time and with the general population.
A need for statistics on the immigration and temporary migration of specific populations including students, asylum seekers, overseas armed forces, seasonal agricultural labouring migrants (“itinerant immigrant populations”) and “people from abroad working on short-term contracts” was also expressed by users.
Other academic and policy users expressed a need to understand the long-term demographic impacts of immigration. Statistics on the fertility rates of immigrants in their first and subsequent generations are needed for this understanding. Inter-censal estimation of ethnic group and especially foreign-born fertility rates using PE denominators are heavily dependent on accurate estimation of IM. ONS stopped producing fertility rates for foreign-born populations in the late 1990s because of concern about the dependence of these fertility rates on estimates that are subject to relatively high IPS sampling error. Academics have since used the ONS Longitudinal Study (LS) and Labour Force Survey (LFS) to develop alternate estimates of ethnic and foreign-born fertility rates. Each source has limitations, the LS due to its incomplete recording of inter-censal immigrant arrivals and departures, and the LFS having no direct question on children born to respondents.
The Census is considered by users to be a key source for information on immigrants already in the UK (“immigrant stocks”), as the information available is highly detailed. The Census’ current information includes country of birth and where resident one year before the Census date. A major disadvantage is that the Census is only conducted every ten years. The Labour Force Survey (LFS), with its questions additionally on nationality and year of entry for the foreign-born, supplies two more important elements. The sample sizes of the LFS are large enough to be of considerable use in the inter-censal period. Its annual frequency is also a big advantage. However, users also emphasised the limits inherent in any sample survey for capturing a relatively small component of the total population such as immigrants or, even more, recent immigrants.
Several users expressed needs for IM statistics on immigrants’ lengths of stay in the UK. While some information on length migrants’ stay is available from the “year of entry” question in the LFS, this data source necessarily excludes those migrants who have already left. The IPS asks immigrants entering the country for intended length of stay, including classifications of 1-2 years, 2-4 years and over 4 years. The reliability of these responses, however, is not known.
Users also expressed needs to understand the relationship between international migrant stocks and flows, so that assessments of the economic impacts of migration can be carried out. At present, the various basic definitions and categories for the ‘flow’ of migrants and the ‘stock’ of migrants in the population are not classified as consistently as users need for these purposes. Better matching of ‘flows’ and ‘stocks’ data on outcomes by routes of entry, year of entry and length of stay are needed in particular. One user asked for “consistent definitions of a migrant across government statistical sources and, even better, internationally comparable”. Eurostat asks member countries to supply immigration and emigration statistics in conformity with United Nations recommended definitions. This includes defining “long-term migrants” based on residence for more than 12 months in the receiving country, a practice that is followed in UK IM statistics.
IM statistics are needed also to assess the implications for local and national services of migrants. Users expressed concern about lack of good statistics on the internal migration patterns of immigrants to the UK, and on their demographic composition. Many public services are provided at the local level, so the effects of immigration on local areas where migrants settle and the impact of dispersal on the local services need to be assessed.
Statistics on the characteristics of immigrants are very poor at sub-UK geographic levels according to users. For example, in the words of a local authority user group, the “lack of information on the characteristics of in and out migrants renders serious debate on the actual demand for dwellings and infrastructure all but impossible”.
Immigration and Emigration of Persons Subject to Immigration Control
New immigration policies, particularly arising from the Home Office White Paper “Secure Borders, Safe Haven: Integration with Diversity in Modern Britain”, need to be accompanied by high quality IM statistics. The roles of these statistics include both to evaluate whether immigration policies are working successfully, and to evaluate their economic and social impacts.
A particular feature of the IM statistics needed for immigration policy is that they distinguish persons subject to immigration control (in general, non-EU citizens). Further, among those persons subject to immigration control, IM statistics are needed that identify and categorise immigrants and emigrants by nationality, route of entry and migrant type (asylum, student, entry for work or dependants of these immigrants).
Users in immigration policy work with various Home Office statistics and databases that record immigrant arrivals and in-country applications to remain in the UK. Immigration policy users also need information on longer-term outcomes, including how many persons of each mode of entry or status change are currently in the UK resident population and when they arrived. Estimates of immigrant numbers by age, citizenship,residency and work-permission status are needed in particular, to assess the effectiveness of immigration policy.
Users noted that it is not possible to collect longitudinal information from inflow statistics, such as those collected by the Home Office and in the IPS. Thus they ask that other sources such as the Census, LFS and GHS, longitudinal surveys, and longitudinal administrative databases be developed in ways that address this need. The ability to relate the statistics obtained from administrative databases (such as those belonging to the HO) to those from ONS’s large sample surveys is needed particularly for persons subject to immigration control. In the LFS, ‘country of birth’ and ‘nationality’ details are available. A breakdown between the three sorts of migrants – asylum, family reunion, work – however, is not possible. The IPS similarly lacks information about the formal entry status of immigrants. Nationality-specific statistics are also limited for age and sex or other breakdowns.
Estimates of immigrant numbers by age, citizenship, residency and work-permission status are also needed to calculate take-up rates of government income support programmes and of fraudulent use of these programmes. Businesses and the labour market are effected by the large supply of mainly unskilled/semi-skilled workers and the low wages they are paid, but the extent of it is not known because of the lack of data. Users have expressed concern that “the LFS [Labour Force Survey] data do not appear to be able to supply sufficient citizenship detail”. The LFS provides extensive population data on the labour market and can provide details of the numbers employed, wage outcomes, the take-up of benefits, etc. However, users expressed the need to better identify migrants to enable comparisons of the labour market outcomes of migrants to those of the general population.
Related to these variables is a lack of knowledge about immigrants without legal permission to be resident or to work or draw government benefits (“unauthorised migrants”). While users accept that estimates of unauthorised migrant stocks and flows can only be approximate, need was expressed for “a consistent, cross-government estimate”.
Estimating the size of the illegally-resident population is an essential step in assessing both economic and social impacts, and would also provide information for the allocation of resources. The lack of a complete embarkation control in the UK has the effect that the extent and nature of overstaying cannot be verified, and clandestine entry cannot be captured by any administrative process. Even when an individual is identified as an illegal resident/worker, it is often not known if they are a clandestine entrant, an overstayer or a worker in breach of visa status.
Among the various types of and routes into the status of ‘unauthorised migrant’, the outcomes concerning “failed asylum seekers [who] leave the UK without informing the Home Office” and who are not picked up as emigrants by the IPS is a frequently expressed area of user concern.
Users have also noted that although only limited information on the illegal population can be obtained from migration sources, a number of methods have been successfully used in other countries to estimate, in general terms, the size of the illegally-resident population. They suggest that UK population and migration data may be developed or created to apply such methods in the UK.
It is also important to be able to analyse the illegal flow of migrants across Europe. If improved information bases about illegal residence and illegal working were available (both within the UK and across Europe), then the drivers behind deciding to move to a particular country, the impact of new routes and the success of measures to tackle illegal working could be assessed. Increased liaison is required with source and transit countries to better understand and control these flows. For these source and transit countries to be most usefully identified, more information is required on how migrants travelled from their source countries to the UK and on their ports of entry.
Due to the high numbers of individuals wanting to emigrate to the countries in the EU, and to the high profile that immigration has in a number of EU countries, migration is now increasingly recognised as an EU issue for both judicial and economic reasons. In order for further co-operation to occur across Europe, policies and practices need to be consistent across the European Union (EU). The Treaty of Amsterdam (1997) committed Member States to a broad range of measures designed to establish minimum standards for asylum procedures and policies across the Union as a first step towards a Common European Asylum system. These proposals include the Directive on Refugee Qualification, which will ensure all Member States use the same interpretation of refugee status so that discrepancies in this area do not encourage ‘asylum shopping’. If all Member States conform to international migration definitions, closer co-operation and a more uniform approach can be taken on asylum seekers across the EU and more comparable statistics produced. The need for “internationally comparable data on asylum decisions and appeal outcomes” is key.
Eurostat requires monthly asylum statistics from member countries for its asylum data base (CIREA). They currently receive them from the UK at the end of each quarter. Eurostat also requires enforcement data (refused aliens, apprehensions of illegal residents and removals) for its CIREFI data base. Users noted that enforcement statistics provide the best available source of data for analysis of illegal working.