Foreign Workers in the UK - Statistics Commission Briefing Note (December 2007)

1. In October and November 2007, a series of parliamentary questions asked about statistics on the number of foreign workers in employment in the UK. The departments to which the questions were directed (primarily the Department for Work and Pensions and Treasury/Office for National Statistics, but also the Department of Business, Enterprise and Regulatory Reform and Home Office) all drew on Labour Force Survey (LFS) data to provide answers.   However, due partly to error (already acknowledged publicly), partly to matters of definition, and partly to complexities in scaling up LFS survey results to give national figures, there was confusion over the most appropriate figures to use in public debate. 

2. T his note by the Statistics Commission reviews the position and seeks to clarify for the non-expert the relationship between the various published estimates. It does not introduce any new estimates other than to offer some more variants, derived from information provided directly to us by ONS, of the figures already in the public domain.

3.  Although our approach is non-technical it is still a complex story when told in full.  Annex 1 summarises the valid published statistics and a summary is as follows:

Using the best available population estimates, the number of people in employment increased between 1997 and 2007 by 2.7 million (counting all over 16s).  Due to technical weaknesses discussed below, this figure is estimated as 2.1 million [1] when working from the detailed LFS data.  Whichever figure is used, more than half of the increase was accounted for by foreign/migrant workers.  The actual proportion of the employment increase accounted for by foreigners/migrants ranges from just over 50% when looking at foreign nationals and the 16+ age group to just over 80% when looking at country of birth and excluding workers who are over state pension age.

The Labour Force Survey as a source

4. There is presently no source other than the LFS suitable for answering the questions. The LFS also has the virtue of being available for public analysis; so that, at least in principle, any analyst can reproduce the official estimates. However, the LFS is neither an ideal source nor easy to use in this context. It contains questions about nationality (two questions in different parts of the survey interview), country of birth and date of arrival in the UK. So it offers the analyst a range of options in relation to how to estimate the level, and changes over time, of employment of foreign or foreign-born workers.   

5. It is also important to bear in mind that the LFS is a survey of private households and excludes people in communal establishments (eg hostels, students living in halls of residence) and people who have lived in the UK for less than six months. Arguably the excluded groups are likely to include disproportionately large numbers of migrants. It would be reasonable to infer from this that estimates of migrant/foreign worker numbers based on the LFS may be a little lower than they should be.

6. And, as for all household surveys, the data relate only to what the survey respondent says to the interviewer.  No formal documentation or official record is involved. On matters of immigration there may be both language barriers to, and reticence about, providing full information. The LFS user must ultimately decide whether to assume that, for example, responses on questions about nationality will be more or less reliable than responses on country of birth.

The PQ answers

7.  In responding to similar, but not identical, parliamentary questions, DWP and ONS provided different figures. One reason for this was that different technical definitions of “foreign” were adopted – DWP used a definition based on the nationality questions in the LFS, while ONS relied on whether people said they were born overseas. In the context of employment in the UK, these two definitions give substantially different results. Annex 2 (tables 4 and 6) give the estimates published by the two departments, which both relate to the second quarter of 2007 - 2.1 million foreign nationals in employment (DWP) as against 3.3 million people in employment who were born abroad (ONS). 

8.  Another difference was in the age group selected – DWP opted for all people aged 16 or over while ONS restricted the population to people of working age (from 16 to 59 or 64).   There is a difference of over 1 million between the 2007 estimates of total employment (used by DWP) and working age employment (used by ONS).

9.  The reasons for choosing different definitions were at least partly that both departments were seeking to answer the specific, slightly different questions they had been asked.

10.  A further source of confusion was that neither of the two sets of statistics is consistent with the estimates of total employment and working age employment in the monthly Labour Market Statistics First Release.   The reason for this is that the LFS results (derived from a relatively small sample survey of households) have to be scaled up by a factor of several hundred to tally with estimates of the relevant total population figures.  For technical reasons, the detailed LFS data used in one-off analyses (such as estimation of employment of foreign/migrant workers) are derived using 2003-based population projections for 2007, whilst estimates of total employment and working age employment in the monthly labour market statistics (also derived from the LFS) are scaled up to be consistent with the most recent population estimates (2005-based at the time the questions were answered;  2006-based estimates are now available). We understand that the technical problems behind this inconsistency will be overcome in time by the ONS modernisation programme. But at the moment, any ad hoc analysis using the detailed LFS data is likely to be affected by the problem , resulting in estimates being lower than they should be. In this case, the estimated increases between 1997 and 2007 may be under-estimated by as much as one-quarter.

11.  The text of all the parliamentary questions and answers are set out in Annex 3. The initial one, on 1 October 2007 (PQ152969), was addressed to the Home Office and was about work permits. The answer referred to an academic report that included work permit information as well as an estimate of 1.5 million foreign nationals working in the UK in 2005 (although this figure was not quoted directly in the reply). This is not an official estimate and is lower than the estimate of 2.1 million quoted by DWP (see Annexes 2 and 4) – although the latter relates to 2007 rather than 2005.

12.  Responding to a claim that 2 million people had come to the UK to work in the past three years, DWP stated in an oral answer on 8 October 2007 (see Annex 3), that there were “2.7 million extra jobs in the past 10 years, about 800,000 of which involve those who have come from outside the country to work in Britain and are contributing to our economy.”  

13.  The 2.7 million figure was a valid estimate, based on aggregate labour market statistics, although it is worth noting that strictly the reference should have been to “2.7 million more people in employment” rather than “extra jobs”. There are more jobs than people who have jobs but in the context of an oral answer this is a small point. However the 800,000 figure was an error, for which DWP has apologised. The details of the calculation that led to the figure of 800,000 have been explained to us. This error was partly due to the complexity of the LFS dataset. 

14.  The same calculations were used to answer another PQ (PQ154368 in Annex 3) a few days later, which asked DWP for the number of jobs created since 1997 that had been filled by immigrants, asylum seekers and foreign nationals who had arrived in the UK since 1997. DWP responded that there was a total of 1.9 millionforeign nationals in employment in UK in 2007, a rise of 0.8 million since 1997 (both figures were wrong and have been amended by DWP). 

15.  These answers prompted four separate follow-up questions – (PQ157946, PQ157708, PQ155993, and PQ158012 in Annex 3). On 29 October, DWP responded separately to each of the questions, revising the 800,000 estimate to 1.1 million [a valid estimate on the definition they were using].   It was explained that the reason for the change from 0.8m to 1.1m was that DWP analysts had not identified all foreign nationals from the LFS dataset, and not compared the same time periods in 2007 as 1997. An explanatory note (attached at Annex 4) was placed in the House of Commons Library.

Parliamentary questions answered by ONS

16.  ONS replied to two written parliamentary questions on the topic. On 18 July 2007, a PQ (PQ150245) asked how many people who have entered the UK in the last ten years are included in the figures for employment. The National Statistician replied on the basis of country of birth – “there were 1.5 million overseas born people in employment who had entered the UK in the last ten years” (see Annex 2, Table 5 and Annex 3). This answer was produced on a different basis – the number of people in 2007 who had indicated in response to LFS questions that they arrived in the last ten years. It is therefore not comparable with the estimates based on changes between 1997 and 2007 in the ‘stock’ of foreign workers at each date. Rather, it should be compared with the estimate of the stock of overseas-born workers in 2007 (3.3 million) subsequently published by ONS in October (see paragraph below and Annex 2, Table 6) – although the two numbers relate to different quarters of 2007.  This comparison suggests that that no more than half of the 3.3 million ‘overseas-born’ in UK employment in 2007 were recent migrants who had arrived in the past 10 years.  

17.  On 19 October 2007, the National Statistician provided a written response to a PQ (PQ158565) asking for the number of migrant workers in the UK each year 1997-2012. The statistics provided in response were based on country of birth and showed the number of overseas-born workers for each year. They also covered people of working age only, in contrast to those provided by DWP which were for total employment (people aged 16 and over). Between 1997 and 2007, the estimated number of workers born abroad increased by 1.4 million (from 1.9m to 3.3m) and the workers born in UK increased by 0.3m (from 23.6m to 23.9m). See Annex 2, Table 6.

18.  A PQ on 29 October (PQ160710) asked how many overseas workers worked in the UK in 2006. The Department for Business, Enterprise & Regulatory Reform (DBERR) replied that “according to the Labour Force Survey in 2006, 11.3 per cent of people employed in the UK were born overseas” (see Table 7 in Annex 2). These figures are consistent with (and we assume are derived from) those provided by ONS to the House of Lords Select Committee on Economic Affairs in October 2007,[2] and on the same definitions as the 19 October PQ answer (see preceding paragraph). 

Conclusions on the statistics

19.  The parliamentary questions discussed in the preceding sections sought information about foreign/migrant workers and their contribution to UK employment. The LFS is the obvious, and only, statistical source for answering such questions. However the LFS does not provide a unique answer, but rather a number of possible answers, depending on the definitions selected. There are key choices to be made on definition of foreign/migrant workers – foreign nationals or foreign-born – and on population coverage – all over 16s or working age only.

20.  The effects of these choices are illustrated in Annex 1, Tables 1 and 2, which set out four different LFS-derived estimates of the change in employment of foreign/migrant workers between the second quarter of 1997 and the second quarter of 2007. The key figures are summarised below:

Estimated numbers of foreign/migrant workers in employment

 

21.  The figures above and in Annex 1, Tables 1 and 2 should be seen only as estimates based on the differences between sample survey results – and therefore liable to more sampling variation than any one set of survey results. Nevertheless they tell a clear story. Of a total increase in employment between 1997 and 2007 of about 2.1 million (counting all over 16s), more than half was accounted for by foreign/migrant workers. But the actual proportion of the employment increase accounted for by foreigners/migrants ranges from just over 50% to just over 80%, depending on the definition of migrant and the population coverage used.

22.  ONS have consistently used the LFS country of birth variable to answer questions about migrant workers in employment. When the central issue is migration, this approach probably gives the clearest picture, for two reasons. Firstly, information about country of birth would seem to be more relevant to questions about migration – people who have come to this country from abroad – than information about nationality, which can change over time. Secondly, we feel that survey answers to questions about country of birth may be more reliable than answers to questions about nationality.

23.  There are however some important caveats to make about the use of country of birth data in this context. In particular:

• ‘migrant’ workers and ‘foreign’ workers are not the same thing – over one third of those born abroad and in UK employment in 2007 were UK nationals rather than foreign nationals;
• most ’migrant’ workers are not recent arrivals – no more than half of those born abroad, and in UK employment in 2007, had arrived in the UK in the past 10 years. (See paragraph 16 above).

24.  Where the main interest is in workers of foreign nationality, rather than in migrants, it may be more appropriate to base estimates on nationality data from the LFS rather than on country of birth data. This was the approach followed by DWP in their PQ answers. The number of foreign nationals in UK employment is substantially below the number of people born abroad, and this approach therefore produces lower numbers for both levels and changes than the use of country of birth. But qualitatively the story is similar – substantial increases over the past 10 years in both the absolute number of foreign workers and in the proportion of total employment that they account for.  

25.  The details of the story on foreign/migrant workers are also affected by the population coverage selected – whether numbers in employment of working age, as adopted by ONS, or total employment (all aged over 16), used by DWP.   There has been a substantial increase in above ‘working age’ employment in recent years, which means that the change in employment of UK-born people since 1997 is substantially higher when measured for total employment – an increase of 0.7 million, rather than 0.3 million (see Annex 1, Tables 1a and 2a). On this basis the proportion of the increase in employment accounted for by those born abroad is reduced from over 80% to under 70%.  The figures for employment of UK nationals paint a similar picture – an increase since 1997 of 1.0 million, rather than 0.6 million, with the proportion of the increase in employment accounted for by foreign nationals reduced from around 65% to under 55%. (see Annex 1, Tables 1b and 2b).

26.  In summary, there are different ways of using LFS data to measure the contribution of foreign/migrant workers to the overall change in employment over recent years. The tables in Annex 1 to this note show four such measures, two of which have been used in recent PQ answers. In broad qualitative terms, all the measures tell a similar story, but at a detailed level there are significant differences between the numbers.  All the measures shown are consistent with each other and a valid representation of the data. Nonetheless, government departments might need to agree on a common approach to answering these questions so as to avoid continuing confusion. We understand that discussions to this end are taking place.

December 2007

Annex 1         Labour Force Survey data on foreign workers

 

This annex is only available in the pdf version of the briefing note.

Annex 3. Parliamentary questions

 

 

Annex 4. DWP explanatory note, 29 Oct 2007 (made available alongside the PQ answers on that date)

Employment growth since 1997 by nationality in answer to PQ/07/ 157946

1.   This note seeks to clarify employment growth since 1997 by nationality and should be considered alongside answers to PQ/07/158012, PQ/07/155993, PQ/07/157708 and PQ/07/157946. Earlier analysis by DWP indicated that the increase in the number of foreign nationals in employment since 1997 was 0.8 million. However it has become clear that this figure is not as comprehensive as it should be. Therefore Ministers requested detailed analysis to ensure that the Department could provide the most robust estimates available as soon as possible.

2.  Further analysis of the Labour Force Survey published by the Office for National Statistics shows that there are, in total, an extra 1.1 million foreign nationals in employment in the United Kingdom since 1997. This estimate is 0.3 million higher than estimates which the Government have previously made.  

3.  The reason for the increase is that we have applied a more comprehensive approach to identifying a person’s nationality from the answers they give to questions in the Labour Force Survey. We have used answers to two questions and cross-referenced them to obtain a more robust view of a person’s nationality. For the O.8 million estimate the answer to only one question was used to determine a person’s nationality. The revised figures are therefore more robust.

5. To calculate employment levels from the responses to the Labour Force Survey it is necessary to use population estimates to scale up the sample size (60,000 households) to give numbers at a national level. The 1.1 million increase in foreign nationals in employment is not directly comparable with the 2.7 million increase in the employment level, which can be derived from the most recent Labour Market Statistics release. The 1.1 million figure is calculated by using estimates made in 2003 of what the population would be in 2007. These are the most recent estimates which can be used in respect of particular categories of the labour force such as migrants. However, when calculating the increase in the employment level since 1997 (ie the 2.7 million) the population estimates used are those made in 2005 of what the population would be in 2007. In 2005 it was concluded that the projections of future population levels made in 2003 understated the true level of the population. As a result, the revised population estimates made in 2005 substituted higher levels of population for future years (including 2007) than those which had been projected in 2003.

6.  These figures are subject to seasonal variation and fluctuate to some extent from quarter to quarter. As these figures are b ased on a sample survey they are also subject to sampling variability. Further details of this are given in the Labour Market Statistics release on page 8, found at: http://www.statistics.gov.uk/pdfdir/lmsuk1007.pdf

What this means

7.   The following can be stated:

 

• The employment level in the UK, using the latest population estimates available (those made in 2005), is 29.1 million, that is 2.7 million higher than in 1997.

 

• If 2003 estimates of the population in 2007 are used, purely for the purposes of comparing how the increase in employment level breaks down into UK nationals and foreign nationals, the increase in employment level becomes 2.1 million

 

• Of that 2.1 million increase, about 1.0 million has been accounted for by UK nationals and about 1.1 million has been accounted for by foreign nationals (of which about 0.5 million relates to EU nationals and 0.6 million to non EU nationals)

 

• The employment rate of UK nationals is higher today than it was in 1997 (an increase from 73.2% to 74.8% when using 2003 population estimates).

 

• The employment rate of foreign nationals is also higher today than it was in 1997 (an increase from 60.6% to 67.6% when using 2003 population estimates). This shows that foreign nationals coming to the United Kingdom are increasingly making an important economic contribution to this country.

 

8.  The revised estimates of the increase in foreign nationals from 0.8 million to 1.1 million do not affect the evidence the Government recently provided to the House of Lords Select Committee on Economic Affairs (“The economic and fiscal impacts of migration”). That evidence included reference to increases in the number of migrants not preventing British people obtaining employment. With a high level of vacancies of around two thirds of a million existing in the UK economy at any one time there is the opportunity to get more people living in the United Kingdom off benefits and into work.

9.  The Government’s recent green paper 'In work better off' makes clear that the next stage of welfare reform is to help those people who have not been able to take advantage of the opportunities within the job market over the last decade. We have announced a new initiative, Local Employment Partnerships with some of the country's leading employers to help give 250,000 opportunities to those people who have had difficulties getting work and have been on benefits for a pro-longed period of time.

Data source and methodology

10.  The data used is Labour Force Survey (LFS) microdata from Quarter 2 in 1997 and 2007. The Labour Force Survey (LFS) is a quarterly survey of 60,000 households living at private addresses in Great Britain. Its purpose is to provide information on the UK labour market that can then be used to develop, manage, evaluate and report on labour market policies. It is conducted by the Office for National Statistics. Data collected separately for Northern Ireland is combined with Great Britain data to give a UK dataset. A website with further information about the LFS can be found at: http://www.esds.ac.uk/government/lfs/. Quarter 2 from 1997 and 2007 are used as Quarter 2 2007 is the latest data available; as this data is not seasonally adjusted it is necessary to compare like quarter with like quarter to take account of seasonal variation.

11. The variables used in this analysis from Q2 1997 are: ‘cry’ and ‘cryo’ to create a variable for country of birth; ‘nation’ and ‘nato’ to create a variable for nationality; ‘grwght’ to weight the sample and provide total population estimates; ‘ilodefr’ to identify economic activity to ILO definitions.

12. The variables used in this analysis from Q2 2007 are: ‘cry01’ and ‘cry07’ to create a variable for country of birth; ‘ntnlty’ and ‘nat07’ to create a variable for nationality; ‘grwght’ to weight the sample and provide total population estimates; ‘ilodefr’ to identify economic activity to ILO definitions; ‘cameyr’ to identify the year of arrival for those born outside of the UK.

13.  The use of these variables has been revised since answering PQ 154368. In PQ 154368 in creating a variable for nationality only ‘ntnlty’ in the 2007 dataset and ‘nation’ in the 1997 dataset were used. Analysis for the PQs that have followed has used two variables that are cross referenced, as detailed in paragraphs 6 and 7, to create a ‘nationality’ variable for both periods. This improvement to the methodology has increased the robustness and accuracy of the results given.