17 May, 2016
This paper extends the original research by Christian Dustmann and Tommaso Frattini (Dustmann and Frattini 2014) on the fiscal impact of immigration to the UK in individual years from 1995 to 2011 by applying the same methodological principles to the most recent year for which equivalent data is available - fiscal 2014/15.
The broad findings are that the overall fiscal effect in 2014/15 of the immigrant population in the UK was negative. This was so for post-2000 arrivals too using Dustmann and Frattini's categorisation of sub-population groups, with negative contributions by immigrants from the EU A10 group of countries and countries outside the EEA outweighing a positive contribution by immigrants from the EU15/other EEA countries. This is the same result as obtained by Dustmann and Frattini in the final year of the period they observed.
Some slight differences in assumptions have been made in the interests of greater accuracy, in part because of the possibility of more precise allocations of some particular heads of expenditure or changes in the availability of data. Apart from these, some calculations have been carried out differently where it seems that this will give more accurate results, and it is hoped that the detailing of the reasons for these differences will be helpful to other researchers in this area.
Comments are welcomed by email to firstname.lastname@example.org
Dustmann and Frattini's findings indicated that, when considering the resident immigrant population in each year from 1995 to 2011 (fiscal years 1995/6 to 2011/12), immigrants from the EEA had made a positive fiscal contribution even during periods when the UK was running budget deficits, while non-EEA immigrants had made a negative contribution. For immigrants that had arrived since 2000, contributions had been positive throughout, and particularly so for immigrants from EEA countries. The authors highlighted their finding of a strong positive contribution made by immigrants from countries that joined the EU in 2004.
However, Dustmann and Frattini observed a general downward trend over time in net fiscal contributions and in particular after 2008 they found contributions even by recent immigrants from the A10 and non-EEA countries turned negative. Bearing in mind that their observations ended in fiscal year 2011/12 as the UK economy began to emerge from recession it is of interest to make a further observation now that on a number of key measures the UK economy has recovered from recession.
This paper performs two types of analysis. Firstly, it considers the total population of immigrants residing in the UK in fiscal year 2014/15, distinguishing between immigrants from countries that are not part of the European Economic Area (EEA) and immigrants from EEA countries, and computes their net fiscal contribution. Secondly, the same analysis is carried out for the immigrant population who arrived in the UK in 2001 and subsequent years, thus distinguishing more recent arrivals on the same basis as Dustmann and Frattini. The analysis further breaks down immigrants into those who arrived from A10 countries and those who came from the rest of the EEA (EU15 countries together with Iceland, Norway and Switzerland).
The net fiscal contribution of different population groups is estimated by assigning individuals their share of cost for each item of government expenditure and identifying their contribution to each source of government revenues. This procedure allows estimation for the year of both the overall expenditure on the respective immigrant populations and the revenues they have produced in comparison to people born in the UK. Although such an approach is ‘static’ in the sense that it does not compute the hypothetical life cycle contributions for each immigrant at one point in time (see Preston (2014) on the advantages and disadvantages of such an approach), the Dustmann and Frattini paper was also ‘dynamic’ in that it provided a picture of the UK immigrant populations’ net contribution to the tax and benefit system over a longer period by reporting results for a series of individual years; this paper provides a further addition to that series.
This analysis goes beyond the previous study by reporting in more detail the composition of government expenditures and revenues and the relative contributions made and costs incurred by the immigrant groupings observed in relation to these heads of expenditure and revenue, and so allows greater explanation for the differences between sub-population groups.
The main findings can be summarised as follows. Whereas the fiscal cost of the UK-born population amounted to nearly £88bn in 2014/15, EEA immigrants cost £1.2bn and non-EEA immigrants cost £15.6bn, a net cost of £16.8bn. Looking at recent immigrants, i.e. those who came during 2001 and subsequently, the net fiscal cost was £6.2bn, comprised of recent A10 immigrants costing £2.8bn, other recently arrived European immigrants contributing £2.8bn and recent non-European immigrants costing £6.2bn.
Overall, this extension to 2014/15 suggests that the downward trend in the fiscal contribution of immigrants observed by Dustmann and Frattini has continued, and that the only sub-population group making any positive contribution in the year is composed of people from the 'old' EU countries together with people from Norway, Iceland and Switzerland. While it could be said that the recent A10 and recent non-EEA groups are making a less negative contribution than the UK-born, it should be noteded that the UK-born 'deficit' is more than entirely accounted for by the cost of spending on pensions, without which the UK-born group would make a positive fiscal contribution. In contrast, the immigrant group is in 'deficit' even without bearing pension costs.
The remainder of this paper unfolds as follows. Section 1 provides a discussion of the conceptual framework and Section 2 the assumptions made and the measures for the expenditures and revenues that underlie the analysis. Section 3 outlines the data sources, Section 4 observes some characteristics of immigrants in the UK, Section 5 reports the results and Section 6 concludes.
1.1. Conceptual Issues
The assessment of the fiscal contribution of immigrants typically assigns to each individual his or her estimated tax contributions and the expenditures in benefit payments and public services received. Doing so accurately, however, requires detailed data on the various items of government revenues and expenditures, data that are not always available. It also requires that the researcher estimates the amount attributable to each individual or group of individuals for all items. This estimation exercise requires rich survey data on the population of interest, complemented by administrative data sources. The next Section, therefore, describes how these numbers are computed and how incomplete information is dealt with.
1.1.1. Second generation immigrants
Dustmann and Frattini note that limitations in the primary source of data prevent the identification of adult second-generation immigrants because the UK Labour Force Survey has no information on parents’ country of birth for individuals who live outside their parents’ household. Hence, they concluded that second generation immigrants can only be identified while they are children (i.e. while under 16 and living in their parents’ households). This is relevant to an assessment of the cost of education services, as adults who arrived in the UK as children and benefited from education in the UK cannot be identified and have the costs of this education attributed to them. Nor when grown up, working and paying taxes, and making fiscal contributions, can they be fully identified in the survey data available. Although theoretically some inferences could be drawn based on ethnicity data, this would necessarily be only partial, and could bias the results if there are considerable differences in outcome between different ethnic groups. Thus, while assigning to immigrants the cost of educating their UK-born children, Dustmann and Frattini were unable to assign to them the benefits that their children might bring after leaving the education system and entering the labour market. On this basis there could be some underestimation of immigrants’ net fiscal contribution over the period observed, but this would rely on an overall positive contribution by the UK-born children of immigrants which is not, as they point out, observable from the data. Finally on this point, as legislation now requires increasing numbers of people to stay in secondary education for longer it might be more appropriate to extend the immigrant classification to second generation immigrants who are under the age of 18, or indeed to all still in secondary education living with their parents regardless of age. As these would necessarily consume public services to a greater extent than they contributed in taxes, this would increase the fiscal costs of the immigrant population. However, to follow as closely as possible the original methodology, this paper counts as second-generation immigrants only those under the age of 16.
1.1.2. Average population versus arrival cohorts
Below, two sets of calculations are provided: the estimated fiscal contribution of the population of immigrants resident in the UK during 2014-2015 in that year and the estimated fiscal contribution for the subset of immigrants who arrived during 2001 and subsequently.
Dustmann and Frattini noted that their two sets of estimates had different interpretations. One was said to answer the well-defined question ‘What is the net fiscal contribution of immigrants who arrived in the UK since 2000?’, thereby providing a 'clean' description of individuals currently in the UK from the start of residency onwards. This paper asks similarly 'what was the net fiscal contribution of immigrants in the UK in 2014-15 who arrived after 2000?'.
Their other question focused on all immigrants who resided in the UK in each year between 1995 and 2011 regardless of arrival date and thus depended on past immigration dynamics, which in turn determined the demographic composition of the immigrant population at any point in time. To illustrate, they gave a hypothetical cohort of immigrants who arrived in 1950 and were observed in 1995, the 1950 arrival cohort having been in the UK for 45 years. The net fiscal contribution of that cohort for the years after 1995 was stated not to be informative about its overall net fiscal contribution because its net fiscal contributions for the first 45 years after arrival were not observed. They proposed that as this cohort was now older, it was likely to have high rates of welfare dependency and low labour force participation that did not reflect its overall contributions since arrival. It was suggested moreover, that a substantial fraction of the immigrants who originally came in 1950 may have returned to their home countries by 1995, after spending their most productive years in the UK. Hence, they asserted that the net fiscal contribution of immigrant populations at a particular point in time varies according to demographic composition, which in turn depends on historic arrival intensities and return migration pattern. Therefore, although they reported these figures for completeness, they felt such figures were 'difficult to interpret', which was one rationale for focusing their discussion on arrival cohorts since 2000.
While such figures might be difficult to interpret in terms of lifetime costs or contributions, they should not be overlooked, as the continued presence in the UK of earlier arrival cohorts is precisely the long-term result of previous flows into and from the UK. While it might be that elderly people who migrated to the UK many years ago might now be a fiscal cost (though forming a small part of that age cohort in the UK population), so are the UK-born of the same age, and so today's working-age contributors will be in future years if they remain in the longer term. As, broadly speaking, government expenditures are funded out of current revenues rather than past contributions, it is entirely conventional to weigh today's contributions against today's cost. In general calculations and forecasts of the fiscal balance, the present cost of healthcare and pensions for the elderly is not ignored because of past contributions they had made, nor are the costs of education for the young ignored because they will pay taxes in the future. As Rowthorn observes in relation to population growth through immigration to constrain the old/young dependency ratio "Rejuvenation through immigration is an endless treadmill" (Rowthorn, 2015).
1.1.3. Net fiscal contributions and the deficit
As detailed below, the approach followed allocates all revenues and all expenditures for the fiscal year to different populations. First, the sum of all revenues and expenditures approximates to the deficit that the UK is running. Thus, for an average individual, the net fiscal contribution must be negative in the year. While the relative contribution of different groups in comparison to other population groups provides an interesting insight, it should be noted that a less negative contribution is still a negative contribution and that adding to the population anyone from outside the UK who does not make a positive contribution will increase the deficit rather than mitigate it. The subsequent discussion identifies the net fiscal contributions of different immigrant groups to the UK-born (the reference population), observes the differences between them and provides some explanation for these differences.
1.1.4. Public goods
Answering the question of what is the cost of providing public goods to immigrants is critical because public goods and services represent a substantial part of government expenditures. Thus, the choice of the apportioning coefficient for public goods plays an important role in determining the overall result of any analysis of the type reported here. Dustmann and Frattini distinguished between two types of public goods and services: ‘pure’ and ‘congestible’.. ‘Pure’ public goods and services are not rival in consumption and the in-year marginal cost of providing some of them to immigrants may be estimated as likely to be zero. For example, the expenditure for defence or for running executive and legislative organs is largely independent of population size. ‘Congestible’ public goods and services are in contrast likely to be rival in consumption, as for example a larger population may require the building of more schools or roads.
If pure public goods can be provided in the same amount and at the same cost regardless of population size, the marginal cost of providing them to an immigrant would be zero. Hence, if it is the average cost of that public good that is assigned to the immigrant, immigration simply allows the cost to be shared among a larger number of individuals and thus represents a form of implicit saving for natives. Conversely, immigration can be expected to increase the expenditure for congestible public goods and services.
Obviously, the ideal would be to measure the marginal cost of providing each public good and assign it to every new immigrant. Unfortunately, however, no data are available on the actual marginal cost of public good provisions, so all that is certainly known is the average cost (the ratio of total expenditure for the good to the total population). In addition, although it is possible that the marginal cost of public goods provision to immigrants is lower than its average cost, in the long term it is likely that is will actually be higher in circumstances where choices are driven by population growth that is driven mainly or exclusively by immigration. The latest population projections by the Office for National Statistics are that immigration will account for 66% of population growth over the period 2014 to 2039. As Dustmann and Frattini assigned to immigrants the average cost of publicly provided goods and services in their central reported findings, that is the approach followed in this paper, but for completeness a marginal cost estimate is provided in Appendix B.
Table A1 (below) provides a detailed list of all goods and services in each category.
2.1. Benefits , transfers and social housing
The first part of the analysis carried out by Dustmann and Frattini assessed the degree to which immigrants drew on benefits and tax credits or lived in social housing in comparison to the UK-born. This paper does not reproduce that part of their analysis because it did not observe or establish any amount of fiscal expenditure, and so is not as such relevant to an estimation of the fiscal impact of immigration. Simple likelihood of claiming any benefit is not particularly meaningful bearing in the mind the very different costs of different benefits (i.e. the amount that can be claimed by an individual) and their different prevalence (i.e. some benefits are paid to very large numbers of people, some to much smaller numbers). Further, the fact that some benefits that make up a large part of cost of welfare expenditures are paid in amounts that depend on the income of the claimant means that even identifying two people as in receipt of the same benefit does not necessarily mean that government incurs the same fiscal cost in relation to each, any more than the likelihood of paying taxes says anything about the actual amounts of tax paid by each taxpayer.
2.2. Fiscal cost and benefit To compute estimates on immigrants’ net fiscal contributions, this paper follows the approach first used by Dustmann et al. (2010) to construct quantitative measures of immigrants’ cost and tax receipts and subsequently adapted by Dustmann and Frattini to examine the net fiscal contributions of immigrants from different areas of origin (EEA and non-EEA), further broken down into immigrants from one of the 10 Central and Eastern European countries that joined the EU in 2004 or in 2007 (A10 countries), immigrants from other EEA countries (the 'old' EU together with Iceland, Norway and Switzerland) and non-EEA immigrants from the rest of the world.
Because in every year, the government surplus or deficit amounts to the difference between receipts and expenditures, total receipts can be decomposed into the revenue from each tax and duty levied by the government, plus interest and dividends and gross operating surplus and rents. Similarly, total expenditures in every year can be decomposed into expenditure for different services. Each revenue and expenditure item can then be further decomposed into the amount paid or received by the UK-born population versus immigrants (distinguishing between the UK-born and immigrants from different origins as explained above).
Revenues and expenditures are directly observable in the data (see Section 2), but apportioning coefficients need to be estimated which can then be used to estimate the total revenues and expenditures for each group. These calculations allow computation of the net fiscal contribution of every group in the year. Because the amount of the net fiscal contribution of each group obviously depends on group size, an interesting metric for assessing the relative size of the net fiscal contribution made by each population group is the ratio of revenues to expenditures paid in every year. As discussed above, the net fiscal contribution of any population group in any fiscal year is determined by whether the government runs a deficit in that particular year. Therefore, to assess the relative contribution of a particular group, the revenues/expenditure ratio is calculated for each particular immigrant group and the native population.
To measure the net fiscal contribution of immigrants and natives, government receipts are first grouped into the eight categories listed below, each of which has computed for it a different apportioning coefficient (see Section 3 for the data sources, Appendix A for details on constructing the apportioning coefficients, and Appendix Tables A1 and A2 for details of items included in each revenue and expenditures category respectively):
(1) For income tax and National Insurance Contributions each group’s share of total payments is estimated by applying year-specific National Insurance Contribution (NIC) and income tax rates and allowances to individual wage data in the LFS. Because the LFS has no information on self-employment income or on other incomes beside employee’s earnings, the overall revenues from income taxes and NIC are apportioned using the share of payments estimated on the sample of employee income only as a proxy for all earnings by non-retired individuals, weighted according to their share among all those in employment, with an allowance made for income tax paid by those who are retired.
(2) For VAT and other indirect taxes each group’s share of payments for each of these consumption taxes were estimated by Dustmann and Frattini by applying calculated effective tax-specific rates by decile of household disposable income to gross individual incomes from the LFS. As for income tax and NICs, they used the information on employee wages that is directly available from the LFS to compute payments of VAT and other indirect taxes. As they noted, however, this results in no estimated indirect tax payments for individuals with no employee income and an underestimation of total household income for households whose total income also includes welfare transfers and self-employment income, and necessarily biases the estimation of payment of these indirect taxes towards low-earning employees. This paper refines that approach to reduce the bias inherent in this method as described below in Appendix A.
However, they also noted that some studies show that immigrants, especially those who have recently arrived or intend to return to their country of origin, may have lower consumption rates than natives with similar income. This difference in behaviour may arise partly due to remittances sent to the home country and partly due to higher labour market uncertainty leading to a higher level of precautionary savings (Dustmann, 1997; Carroll et al. , 1999 for the US; Piracha and Zhu, 2012 for Germany; and Dustmann et al. , 2014 for Italy). Because there is no direct source of information on immigrants’ consumption patterns in the UK, Dustmann and Frattini construct an alternative scenario in which they assume that the consumption of recent arrivals is only 80% that of the UK-born with a similar income. More recently, the Migration Observatory at the University of Oxford has reviewed the available evidence and concluded that in 2014 remittances might fall in a range between £1.5bn and £16.5bn (Migration Observatory 2016). In particular they note evidence that some of the poorest groups remit the most, for example £3,500 by Romanian men and £2,500 by Bangladeshi women. However, the magnitude of the range of estimates based on very different data sources make any estimation of uncertain reliability. The primary impact of remittances is on consumption in the UK and thus in fiscal terms on indirect taxes like VAT. The lower £1.5bn end of the range would have an impact of £300m which in terms of an overall fiscal cost of £13bn is not a particularly significant amount. On the other hand at the upper end of the range, £16.5bn lower consumption in the UK could result in payment of up to £3.3bn less in indirect taxes. Because of the limited information of the incidence of remittance payments by origin combined with the wide range of estimates, calculation of a variant assumption might give a spurious impression of a reliability by sub-population group that the data will not bear, and so this has not been calculated, but in interpretation of the overall results it should be borne in mind that they may underestimate the total fiscal cost of all immigrants in the UK by anything from £300m to as much as £3bn.
(3) ‘Company and capital taxes’: The allocation of corporate taxation raises complicated questions of incidence that are the subject of several studies in the specialised literature. However, although there is consensus that some of the corporate income tax burden is shifted away from shareholders, there is no agreement about how it is shared between shareholders, workers and consumers. This analysis takes no stance on this debate and apportions these tax payments, net of the percentage likely to be paid by foreign shareholders on a per capita basis among the adult population. This implicitly assumes that the benefit of company ownership (i.e. share ownership) is similarly distributed among the native and immigrant population. Alternatively, it could be seen as capturing the fact that the real burden of corporate taxes may fall ultimately on consumers and must thus be shared on a per capita basis. However, as the former case relies on the extreme assumption that even immigrants who have only just arrived in the UK have effective beneficial ownership in shares of British companies to the same extent as the UK-born and long-term residents, and that, accordingly, they bear just the same burden of corporate taxation, in an alternative scenario, therefore, company and capital tax payments, net of the share likely to paid by foreign shareholders, are allocated primarily to the population that first arrived in the UK before 2001 (i.e. natives and longer-residing immigrants). Taking account of the data that the more recently arriving EEA other group has income greater than the general population, however, it is assumed that they will in fact be accruing assets even in earlier years of residence and so allocates these taxes at half the rate of allocation to natives. Rest of the world ownership stood at an estimated 54% of the value of the UK stock market at the end of 2014. This was up from 31% in 1998.
(4) ‘Council tax’ payments are levied on domestic residences by individual local authorities dependent on estimated value-bands of property. Because there is no detailed information on individual housing value or local tax levels, each group’s share of council taxes are estimated as proportional to the number of households headed by an individual in each sub-population group.
(5) ‘Business rates’ is a tax on non-domestic property typically paid by businesses and other organisations that occupy non-domestic premises. These payments are treated similarly to company and capital tax payments and so allocated primarily to UK-born and longer-residing immigrants (i.e. those first arriving the UK before 2001), and alternatively apportioned proportionately to the adult population.
(6) ‘Gross operating surplus and rents and interests and dividends’ are allocated proportionately to the share of each group in the 16+ population.
(7) ‘Inheritance tax’ payments. Home ownership (from the LFS) is used as a proxy for asset ownership and inheritance tax allocated proportionately to the share of UK-born and immigrants over the age of 70 in the home-owning population.
(8) All remaining tax payments including landfill tax, climate change levy, aggregates levy, other taxes and royalties and other receipts, are apportioned according to the share of each group in the 16+ population.
Government expenditures are apportioned in a similar way, estimating a different apportioning coefficient for each of the following eight categories (see Appendix A for details):
(I) and (II) ‘Pure public goods and services’ and ‘Congestible public goods and services’ include all public goods and services whether they are typically rival or non-rival in consumption. The cost of providing these goods is allocated proportionately to the share of each group in the adult (16+) population.
(III) ‘Medical and other health services’ are allocated to groups based on the age structure of the population on the evidential basis that age is a primary determinant of health status and of the demand for GP and hospital visits in the UK, and the assumption that there are no large differences in health service use between immigrants and natives in similar age groups.
(IV) ‘Education’: for compulsory education, Dustmann and Frattini estimated each group’s apportioning coefficient based on the share of its children in the age bracket 0– 4 years for pre-primary, and 5– 15 for primary and secondary. However, as expenditure on primary and secondary expenditure is separately identified in government accounts, this paper allocates primary education expenditure to 5-10 year olds, and secondary education to 11-18 year olds for greater precision and to reflects changes in school leaving age since the 1995-2011 period. For post-secondary education, the share of the population in education for each group uses direct information from the LFS on type of establishment attended by those still in the education system.
(V) ‘Social protection’ includes expenditure for sickness and disability, old age, family and children, unemployment, housing and social exclusion and is broadly what is usually described as 'welfare spending'. Since the great majority of these expenditures are allocated in terms of cash benefits, LFS information on the receipt of different types of benefits are used to compute for each group the share among the total recipients of each type of benefit. Because the LFS contains no information on the amount of benefits received, Dustmann and Frattini assumed that every recipient received the same amount. However, since the average amount of housing benefits receipts varies substantially across regions, when apportioning housing benefits, they first allocated housing benefits expenditures across regions and then assumed that within the same region everyone received the same amount. It is quite possible to develop more sophisticated analysis of benefit receipt even using only LFS data. In principle, if the limited income data in the LFS is sufficient for estimation of direct and indirect tax payments, then it will on the same basis be sufficient for modelling means-tested benefit receipt. However, as the purpose of this paper is the continuation of the series of observations in earlier years, rather than a re-estimation using different assumptions that might not be at all comparable, the Dustmann and Frattini approach of assuming all recipients of particular benefits receive the same is continued, and allocations are made pro rata by allocating amounts of specific benefit expenditures simply on the basis of share of claimants identified in the LFS as detailed in Annex A, and with amounts of 'personal social services' within each category of social protection expenditure allocated on the same basis.
Dustmann and Frattini's 'robustness check' of one aspect of benefit receipt to take account of the higher receipt of child benefits in families with more children by making a different apportionment on the basis of proportion of recipients of income support or 'other family-related' benefits is not replicated, because the updated LFS variables for benefit claims no longer include the category of 'other family-related' benefits.
(VI) ‘Prisons and law courts’ were separately distinguished by Dustmann and Frattini who used information on the nationality of prison inmates from the Ministry of Justice’s offender management statistics to apportion costs of the law courts and prisons proportionately to the size of group in the prison population, which is followed despite the fact that the cost of the law courts is not comprised only of the costs of criminal justice leading to custodial sentences. Annual data for the year to June 2015 are used. The prison population data do not allow separate identification of recent arrivals, so the cost is allocated pro rata to the share of recent arrivals in each sub-population group.
(VII) ‘Housing development’: Dustmann and Frattini estimated each group’s share of total costs based on its share of social housing tenants reported in the LFS. This paper allocates the separately identified costs of local authority and housing association housing to shares of occupants of each as the LFS now allows for this slightly more detailed examination.
(VIII) ‘Police services’ will generally be used by immigrants and UK-born to the same extent and so attribution is on a per capita basis, but since ‘immigration-related police services’ are separately reported as a sub-item of expenditure for police services in PESA, their cost is attributed only to immigrants, as while the expenditure might be for the benefit of the resident population, it is necessarily incurred as a result of immigration.
The primary data source for population characteristics, the UK Labour Force Survey (LFS), is a quarterly representative survey of about 60,000 households in the UK or about 0.2% of the UK population. This survey records respondents’ labour market status (and the wages of a sub-sample), as well as their personal and household circumstances, including country of birth and year of arrival in the UK if applicable, although not parental country of birth (see above). It also gathers self-reported information on any type of state benefit or tax credit received as well as type of accommodation, which is used to identify individuals living in particular kinds of housing by tenure. The earlier years in both analyses carried out by Dustmann and Frattini will have included some very small samples of particular groups, and they maximised sample sizes by 'pooling the four quarterly waves in every fiscal year'. It is not entirely clear whether this involved pooling the five waves in each of the four quarters, as this will have resulted in an element of double counting which might have been distortive of some results. Where relevant, this paper ensures that only unique observations are included.
The LFS is used as the main source of information on UK-born and immigrant population characteristics and as the basis for many of the apportioning coefficients in the fiscal cost and benefits analysis. For the fiscal analysis, this paper uses official data on government receipts and expenditures. Information on government revenues are taken primarily from the Office of Budget Responsibility's Economic and Fiscal Outlook, which brings together information on receipts by HM Revenue and Customs, taxes collected by local authorities, and other government income. Information on public expenditure is taken from the ‘Total Expenditure on Services by Sub-function’ Table 5.2 of the annual Public Expenditure Statistical Analyses (PESA) published by HM Treasury. This latter reports expenditures for different items classified according to the United Nations Classification of the Functions of Government (UN COFOG) definitions at level 2.
Immigration to the UK has seen considerable increases over the past two decades, and this has been reflected in increasing immigrant participation in the UK labour market, and an increasing proportion of children born to immigrant parents. The pace of these increases, the changing mix of countries of origin, and differences in labour market outcomes and earnings between different groups means that regular observation is necessary to estimate what is the current and evolving fiscal impact of immigration to the UK. The composition and various characteristics of migrants in the UK are regularly reported in official statistics, but critically there are few direct sources of information either on the fiscal costs or benefits deriving from them. In the absence of any routine publication of administrative data held by government, any analysis can necessarily give only illustrative or tentative results.
5.1. Expenditures and Revenues
This section reports estimates of the net fiscal contribution for the different immigrant groups defined in subsection 1.1.2. The basic specification allocates all public goods to immigrants according to their average cost, following Dustmann and Frattini reporting this specification as their central scenario.
5.2 Sub-population groups
Estimates for the whole immigrant population residing in the UK in the fiscal year to end-March 2015 are that immigrants from EEA countries made a negative contribution of £1.2 billion in the year while those from non-EEA countries made a negative contribution of £15.6 billion, compared to an overall negative fiscal contribution by the UK-born population of £88 billion.
Table 1 reports the revenues and expenditures for each sub-population group distinguishing between recent and earlier arrivals, the ratio of revenues to expenditures, and the relative ratio of revenues to expenditures defined as the revenues/expenditures ratio for each sub-population group divided by the revenues/expenditures ratio for natives.
Table 1 Overall fiscal impact (£ million)
|Expenditures £m||Revenues £m||Ratio||Relative ratio||Net fiscal effect £m|
|Other EEA pre-2001||11,692||10,955||0.94||1.09||-737|
|recent EEA other||4,876||7,695||1.58||1.84||2,819|
These estimates show an excess of expenditure over revenues for each sub-population group with the exception of recent arrivals from the other EEA group (EU15 countries together with Iceland, Norway and Switzerland). Apart from this group the relative ratio of expenditures to revenues is no better for immigrants overall than for natives, and the total expenditures on immigrants exceeds revenues by £16.8 bn. On this basis it is hard to see much support for the contention that immigrants - or even recent immigrants - as an undifferentiated group are helping to reduce the fiscal burden for native workers or contributing to reducing the UK's fiscal deficit. Notably both revenues and expenditures for the earlier-arriving other EEA group are considerably lower than in the final year observed by Dustmann and Frattini's work. This results from a diminution in the size of the group which is likely to be caused primarily by out-migration, and thus the change in fiscal impact not necessarily a reflection of improving fiscal position.
Table 2 reports the estimated amount of government expenditures and revenues attributable to the native and immigrant sub-populations by country of birth and time of first arrival by broad groupings of expenditure and revenue. This breakdown by broad headings of interest allows the observation that the native fiscal cost may be accounted for entirely by the cost of the elderly. Notably, the fiscal cost for the whole immigrant population and for even for recent arrivals is negative even without such a burden to support.
Table 2 Grouped Expenditures and Revenues (£ million)
|Native||A10 pre-2001||other EEA pre-2001||non-EEA pre-2001||recent A10||recent EEA other||recent non-EEA|
Table 3 shows the variant assumption in which corporate taxes are simply allocated on the basis of 16+ population. This illustrates the sensitivity of the results for sub-population groups to this assumption, though shifting the burden of these taxes from more recent arrivals to the UK-born and longer-term residents still results in an overall fiscal cost for immigrants of some £13bn.
Table 3. Variant assumption for corporate taxes
|Expenditures £m||Revenues £m||Ratio||Relative ratio||Net fiscal effect £m|
|Other EEA pre-2001||11,692||10,882||0.93||1.09||-810|
|recent EEA other||4,876||7,947||1.63||1.91||3,071|
5.3. Consistency and robustness of results
Dustmann and Frattini reported that although the estimates of immigrants’ and natives’ contributions fluctuate across the different scenarios, their qualitative results remained unaffected: the fiscal contributions of EEA immigrants and recent immigrants from all areas of origin were higher, in relative terms, than those of natives in all scenarios. They noted that even in what was described as the most restrictive scenario in which it was assumed that recent immigrants do not contribute to revenues from company and capital taxes and business rates, EEA immigrants who arrived since 2000, both from the A10 and other EEA countries, had all made overall positive fiscal contributions.
However, more restrictive scenarios exist than their most restrictive scenario if more than one of the modifications are applied at the same time. It is theoretically quite possible for the baseline scenario to overestimate the effective indirect tax rates, and that at the same time immigrants do have a lower rate of consumption so that the tax base to which these rates should be applied is a lesser amount, and that at the same time recent arrivals have less financial investments in the UK than the UK-born population. The modifications examined by Dustmann and Frattini were not alternatives that are each mutually exclusive, and so a simple application of each modification separately does not as such demonstrate the robustness of the qualitative result.
Nonetheless, where Dustmann and Frattini's overall reported results and modified scenarios resulted in positive contributions for recent arrivals over the period 2001-2011, it is clear that these derived from positive contributions made in the earlier part of the period which were not maintained throughout. Indeed, by the final years of the period only the 'other EEA' sub-population group was making a positive contribution. That is also the result of the present analysis.
The previous Migration Watch assessment (Migration Watch 2014) included an annex noting that some of the sample sizes used by Dustmann and Frattini were quite small, especially in relation to the earlier years of recent arrivals. Over time, as the number of people in the UK who have arrived since 2000 has increased considerably, this is unlikely to be the issue that it might once have been in earlier years, especially for the key expenditure variables. However, the majority of revenues are allocated using formulae that are based on LFS income data, and as noted above, this information is only collected for a sub-sample of LFS respondents. This means that a large edifice has to be erected on quite shallow foundations, and for example the finding of £8.5bn revenues raised from the recent EEA other group is derived from income data in fewer than 500 cases within the survey. For comparison, the recent A10 revenues were derived from around 1200 cases, recent non-EEA from around 1750 cases, and UK-born revenues from over 38,000 cases. This a necessary consequence of limitations in the primary data source but should be borne in mind in any interpretation of the results.
Finally, while the Labour Force Survey is regarded as the 'best available' data source it is notable that when compared with data from the 2011 Census, the ONS discovered that the LFS appeared to have undercounted the immigrant population over number of years, and adjusted the weightings in the LFS and the data in a range of published material to take account of this (ONS 2014). This is a further illustration of a degree of uncertainty and need for caution, at least in the absence of administrative data.
The fiscal contribution of immigrants has emerged as a key issue of concern in the public debate on immigration but little evidence for the UK was available before the publication of Dustmann and Frattini's detailed discussion paper in 2013 and final publication in 2014
This paper has applied a similar methodology to the immigrant population resident in the UK in 2014/15. It suggests that the net fiscal impact differs significantly between different sub-population groups and that recent arrival is no assurance of a positive contribution.. This paper additionally observes the key reasons for these differences by breaking down contributions to revenues between direct and indirect taxes, and identifying expenditures under the broad headings of old-age benefits, working-age benefits, health and education.
The Dustmann and Frattini finding that recent immigrants have made substantial net contributions to public finances, appears to have resulted from the strong economic growth in the UK leading up to the recession in 2008. These net contributions were not maintained in the subsequent years. Furthermore, the present analysis of the contribution made in 2014/15 suggests that the subsequent recovery from recession has not resulted in a return to a positive contribution. It is true that the native population is also in fiscal deficit but this may be said to be entirely accounted for by spending on pensions. Thus - in broad terms - the working-age native population is in fiscal credit while the equivalent immigrant population is in fiscal deficit.
This Appendix details the procedure and data sources used to construct the apportioning coefficients for each item of government revenues or expenditures outlined in subsection 2.3.
Each group’s share of total payments are calculated firstly by applying NIC and income tax rates and allowances for 2014-15 to individual wages based on the LFS variable grsswk (gross weekly pay) and grsswk2 (gross weekly pay in second job). These weekly wages are aggregated for individuals where variable grsswk is positive, and not in those cases where grsswk is zero. If the main job is self-employment, no income from this job is recorded by the Labour Force Survey, and attributing only the income from work in a second job as an employee to the individual will understate total income and distort calculations of tax and of average income for the group.
Secondly, income tax is also paid by retired households on their income. Dustmann and Frattini note that the Basic State Pension is taxable and so in a robustness check imputed state pension income to pension claimants assuming that they all claimed the full amount of Category A pension, computed the amount of income tax paid accounting for imputed pension income and using these payments to estimate their apportioning coefficients. However, the income of retired households includes both state and private pensions as well as property and investment income. The ONS 'Effect of Taxes and Benefits on Household Income' publication shows that less than half of the income of retired households is derived from the state pension, and HMRC taxpayer statistics record that nearly one in five taxpayers is aged 65 or over. To take account of this, the share of tax paid by retired and non-retired households is calculated from the ONS 'Effect of Taxes and Benefits on Household Income'.
Then, for income tax, the share estimated to be paid by the non-retired in total receipts is first attributed to sub-population groups on the basis of their share in total payments calculated from their employee earnings data in the LFS. The share estimated to be paid by the retired in total receipts is attributed directly on the basis of share of each sub-population group in the retired population as it is not possible to estimate the unearned income (including pensions) of retired households from the Labour Force Survey. This will necessarily provide a more accurate estimation than both Dustmann and Frattini's central scenario and their robustness check.
To summarise, income tax is shared between retired and non-retired on the basis of the ONS 'Effect of Taxes and Benefits on Household Income'. The share of income tax paid by sub-population groups among those who are working is calculated directly from their employee earnings in the LFS. The share of income tax paid by sub-population groups among those who are retired is calculated from their share in the retired population.
Pension contributions are taken into account in calculating the taxable element of gross employee earnings by applying factors for pension prevalence and level of contribution by income band from Tables P3.1 and P7.1 in the ONS Annual Survey of Hours and Earnings for 2014. These are deducted, together with the personal allowance, from gross annual earnings to obtain a measure of taxable income and income taxes calculated by applying the appropriate rates of income tax to this measure of taxable earnings.
Because the LFS has no information on self-employment income, in Dustmann and Frattini's central scenario, they apportioned the overall revenues from income taxes and NIC using the share of payments estimated for the sample of employees only. As a robustness check they also imputed self-employment income based on mean self-employment income by sector of activity constructed from the Survey of Personal Incomes (SPI) conducted by HMRC. However, the HMRC data does not seem suitable for the purpose of enabling calculation of income per person in self-employment by sector as an individual with two or more sources of self employment income is counted more than once, according to the industry group and profit for each source. Further, these 'sources' will include the self-employment income of those employees for whom self-employment is only a second job. Thus the number of 'sources' of self-employment income considerably exceeds the numbers of people who are self-employed in their main job, and as a result the mean of these 'sources' of income must considerably understate the mean earnings of those who are self-employed in their main job. For this reason that 'robustness check' has not been reproduced.
However, if reliance is placed only on tax share derived from employee income, some allowance must be made for the compositional effects resulting from the different proportions of employees and self-employed in each sub-population group. Imagine one group comprising 20,000 employees and 5,000 self-employed, and another group comprising 20,000 employees and 10,000 self-employed. If total income tax share is calculated for each group on the basis of the share of income tax paid by employees only, then all other things being equal, this will overstate the share of tax paid by the first group and understate the share paid by the second group. So a weighting needs to applied to each 'employee share' of income tax to reflect the proportion they form of all of those in work. The implicit assumption in this is merely that differences in employee income between sub-population groups are reflected in similar differences in self-employment income.
As a final cautionary summary, while it is entirely possible to estimate the annual payment of direct taxes by individuals and indeed households surveyed in the LFS whose income derives wholly or primarily from employee income, there are shortcomings in extrapolating from these results to estimations of the share of actual tax receipts attributable to any part of the whole population defined by particular characteristics because of the absence of information in the LFS on any other kind of income. Firstly, as is well known, no income data is recorded for self-employment, although it is intended that the Survey will include such questions in the future. Secondly, no income data is recorded for any other income on which income tax might also be payable. In addition, a large amount of income tax is paid by the very highest earners whose numbers in the whole population are so small that they might not be picked up sufficiently in the LFS to provide a representative sample by a characteristic such as country of birth. The conceptual difficulties are compounded by the fact that only a part of the LFS sample is asked about earnings, and no account is taken in this paper (nor in Dustmann and Frattini's) that the confidence intervals for the earnings of smaller sub-population groups become quite large.
Each group’s share of payments for each of these taxes is estimated using a multi-step procedure following Dustmann and Frattini's approach which correctly recognises that payment of consumption taxes is a function of household income. The approach in their central scenario is to calculate for each indirect tax and based on the ONS publication ‘The Effect of Taxes and Benefits on Household Income’ the ratio of tax payments to original income for households in each decile of the distribution of households’ disposable income to obtain an ‘effective rate’ individuals in different parts of the income distribution should pay on their gross earnings in order to achieve the amount of payments made on average by individuals in that income decile. Then they sum all individual disposable earnings (i.e. gross earnings minus income tax and NIC payments) within each household in the LFS and determine the household position in the sample distribution of disposable earnings. Since there is no information on other type of incomes in the LFS they assume that households are in the same decile of the distribution of disposable income as of disposable earnings, and then apply the estimated decile-specific ‘effective rate’ to gross individual earnings from the LFS. Then apportioning shares are computed for each indirect tax by summing total estimated payments within the sample and estimating the share of each sub-population group (natives, EEA immigrants, non-EEA immigrants) in total payments.
The implicit assumption with this strategy was that immigrants and natives with similar incomes have the same consumption patterns. However, they pointed out that immigrants, especially those who have arrived more recently, may have lower consumption than natives. For this reason, they also computed an alternative set of apportioning coefficients where they assumed that the consumption of recent immigrants (those arrived since 2000) is only 80% that of natives with a similar income. In practice, under this assumption the ‘effective tax rate’ was applied to 80% of immigrants’ income, the remaining 20% being remitted abroad. The extent of remittances internationally has been the subject of considerable study but little hard evidence exists to enable a robust estimation of remittances from the UK. The Office for National Statistics includes an amount for remittances in its calculations for the UK National Accounts, but bundled up with payments made abroad by charities so that they cannot be separately distinguished. This has been stated to be a deliberate choice due to uncertainty about the accuracy of the amount of remittances. Some data is available from other sources. For example the central bank of Bangladesh regularly reports on remittances received from workers abroad. Putting their reported data for remittances from the UK together with information from the LFS (and indeed the Family Resources Survey) suggests a figure of much greater than 20% for this particular group. On the other hand, this rate of remittance might be quite atypical, and it is beyond the intended scope of this paper to explore this further.
As discussed above, the LFS does not include self-employment income and income originating from benefits and transfers. In their final published paper, Dustmann and Frattini noted that the lack of this information may generate two sources of bias in estimating payments of indirect taxes using their methodology. First, a household that receives part of its income from self-employment or welfare benefits will be ranked lower than a household with the same total income coming entirely from labour earnings and since the ‘effective rates’ computed for indirect taxes are higher for the bottom income deciles, this will bias upwards the estimated ‘effective rate’ of households that receive self-employment income and welfare. Second, for these households total income calculated would be lower than actual income which would bias downwards the estimated tax payments. They noted that overall there would thus be two potential sources of bias, going in opposite directions and potentially partly cancelling each other out. They stated that to try and correct for these biases, they carried out robustness checks imputing self-employed income and pension transfers on the same basis as the similar checks carried out for direct taxes as described above.
However, the methodology for calculating the effective rate of tax appears flawed in its basic assumption that these taxes are paid only out of earnings rather than total income. Even making some allowance for self-employment and state pensions in a robustness check will only go some way to correct this, because it does not take into account the large extent to which income in the lowest deciles is comprised of working-age benefits, nor that only half of the income of retired households derives from state pensions. The actual effective rate of these taxes is the amount of indirect tax paid by households at a given point of the income distribution divided by disposable income. The rate of indirect tax paid on money spent out of the employee earnings in a household is necessarily the same as the rate paid on money spent out of any other component of household income, as different income sources are not in practice 'hypothecated' and allocated to different kinds of expenditure. The key exception to this is housing benefit as it is generally not available for adding to the household pot for spending, and no tax is likely to be borne on the expenditure of this income. For this reason, a slightly modified calculation is carried out while following the same approach in principle.
Firstly, as the 'Effect of Taxes and Benefits on Household Income' makes clear that retired households have a very different mix of income sources and it is observable from the LFS data that the proportions of sub-population groups are very different within retired and non-retired households, indirect tax receipts are apportioned initially between retired and non-retired households - as with income tax above.
Then non-retired households are ranked on their LFS wage earnings and placed in the matching decile position of disposable income, following Dustmann and Frattini's general approach. But the actual effective rate of indirect tax is then applied to these earnings and a share of tax on this basis calculated for each sub-population group.
Apportioning shares are then computed for the non-retired households by summing total estimated payments within the decile and thus estimating the share of each sub-population group in total payments. The effect is that the share of the relevant tax paid by these households in each income decile is allocated to sub-population groups according to the proportion the wage earnings of their households form of all wage earnings within households in that decile. For example, if 6% of VAT is paid by households in the 3rd income decile, and 10% of wages in the households allocated to this decile are allocated to immigrant groups, then 0.6% of the amount of total VAT is attributed to the immigrant group, and 5.4% to the native group. This adopts - for non-retired households - the implicit assumption in Dustmann and Frattini that households within each income decile have a similar mix of income sources that does not vary materially across sub-populations, while taking account of the fact that the mix of income sources may vary across income deciles.
For retired households, no income data is available because by definition they are not working and the LFS only has data on employee earnings, and not on income from property, interest, investments or pensions. The share of tax for retired households is thus apportioned on the assumption that there is no difference in incomes across sub-population groups in these households. Intuitively, this is likely to over-estimate the contribution by migrant sub-population groups, but on the other hand they form only a small proportion of the elderly.
Dustmann and Frattini noted that the allocation of corporate taxation raises complicated questions of incidence which are the subject of several studies in the specialised literature and that the actual burden of corporate taxation does not fall on business owners only but falls also upon workers and customers. However, as they felt there is no consensus on the exact allocation of this burden, in their central scenario, they apportioned these tax payments, net of the percentage likely to be paid by foreign shareholders on a per capita basis among the adult population (The share of foreign ownership in UK companies is available from the annual ONS ‘Share Ownership’). This was on the basis that the allocation criterion can be interpreted in two alternative ways that lead to essentially the same conclusion. Firstly, as capturing that the real burden of corporate taxes may fall ultimately on consumers, thus it has to be shared on a per capita basis, and alternatively as the consequence of an implicit assumption that share ownership is shared equally among the whole resident population. The latter seems very unlikely to be the case, particularly bearing in mind the extent to which recent inflows to the UK have been composed of people coming for lower-earning work. Dustmann and Frattini used as a sensitivity check on their results an alternative computation that apportioned shares under the assumption that the burden of company and capital taxes falls on their owners and that ownership is equally shared only among long-term residents in the UK (i.e. people born in the UK and those born abroad who had been in the UK for more than 10 years). Their results for recent arrivals from the A10 group of countries in particular were highly sensitive to the attribution of taxes on business (corporation tax and business rates), as their alternative scenario showed that virtually the entire fiscal contribution of this sub-population group was attributable to the payment of such taxes.
Council tax is levied on domestic residences by individual local authorities dependent on value bands assigned to properties when the tax was introduce. Council tax rates vary substantially across local authorities and do not reflect the average value of housing between local authorities. While within each local authority higher value houses are subject to a higher tax, it is not true that areas with the highest average housing value have higher council taxes. For instance, Dustmann and Frattini noted that the band D tax in the wealthy London council of Westminster was £676.74 in 2013/14, while a band D property in the poorer Hackney council was charged a council tax of £1,301.45. However, as they felt there was an absence of precise information on the local authority and on the type of housing where immigrants and natives live, they apportioned council tax receipts at the UK level simply to each group in proportion to the number of households in the group. They felt that allocation based on number of households, rather than individuals, accounted for potentially different household sizes across population groups. This approach has been followed, allocating on the share of sub-population groups among heads of household, although in principle a more sophisticated approach would be possible using a regional analysis.
These are treated similarly to company and capital tax payments. The correct attribution is conceptually slightly different, as literature on the incidence of business rates considers whether the tax actually falls on landlords or their business tenants. If the latter, then as a business overhead they might fall to be attributed in the same way as company taxes treated as falling on the owners. If the former, then a different method might be more appropriate, but appropriate data as to the ownership of non-domestic property as such (rather than businesses) do not appear to be available. It is notable though that the tax base - which ultimately is the stock of commercial property in the UK - appears to be largely uncorrelated to changes in population and has barely changed over a decade. Whereas DCLG data show the number of dwellings in England and Wales increased by 7.4% between 2003 and 2012, commercial property floorspace stock only increased by 1.1% over the same period (IPF 2014). This suggests that the attribution of a share of business rates to recent arrivals pro rata to their population share is in any event inappropriate as the cost is essentially fixed and unaffected by their arrival.
These are apportioned proportionately to the share of each group in the 16+ population.
House ownership from the LFS is used as a proxy for asset ownership and inheritance tax allocated proportionately to the share of UK-born and immigrants above the age of 70 in the home-owner population.
All remaining tax payments including landfill tax, climate change levy, aggregates levy, other taxes and royalties, and other receipts, are apportioned according to the share of each group in the 16+ population.
This category includes all public goods and services that are typically non-rival in consumption and are apportioned on the basis of the share of each group in the 16+ population.
These costs are apportioned on the basis of the share of each group in the 16+ population.
The proportion of health services expenditure is attributed to each sub-population group based on the group’s age structure. Dustmann and Frattini used the distribution of health costs by age group, as reported in Figure 6.2 of the Department for Health Departmental Report 2006, computing the estimated amount spent for health services on each age group, and then apportioning the costs of each age group proportionately to the share of immigrants and natives in each age band. The data underlying the Department of Health report was from 2003/4 and it seems possible that the distribution of health costs by age might well have changed over the intervening eleven years, but no more recent data appear to be available.
The Office for National Statistics still uses this distribution of health costs by age for the estimation of the costs of healthcare as a benefit in kind in the "Effects of Taxes and Benefits on Household Income" publication. However, an updated methodology is expected to be published in 2016. Dustmann and Frattini also made the implicit assumption that immigrants and natives with the same age make the same use of health services. Again, this might well have changed over time as for example the increasing proportion of births to mothers born outside the UK means that the cost of maternity services and healthcare for children more generally should be biased more towards immigrant groups. The ONS in fact splits out maternity services from their generally age-based distribution and allocate these to households with children under the age of one year. On the other hand the ONS allocate prescription costs on a simple per capita basis to each individual regardless of age. However, without more data to provide a sound basis for updating that might capture more accurately the distribution of health costs among the relevant sub-populations, the original methodology is followed.
For compulsory education, each group’s apportioning coefficient is based on the share of children in the relevant age bracket for each school level (0–4 for preprimary, 5–10 for primary and 11-18 for secondary). As Dustmann and Frattini noted, there are factors pulling in different directions in this area of expenditure, as the cost of educating an immigrant child or a child of immigrant parents may be higher or lower than average, and in the absence of clear evidence either way, assume the average cost of their education is the same.
Expenditures for social protection include expenditure for sickness and disability, old age, family and children, unemployment, housing and social exclusion. Since about 85% of these expenditures are allocated in terms of cash benefits, LFS information on the receipt of different types of benefits is used and for each group the share among the total recipients of each type of benefit is computed. The LFS variables for benefit receipt have changed since the periods observed by Dustmann and Frattini in some respects, and so exact replication of their methodology is not possible. However, the newer variables allow for a more direct attribution of the different benefits to sub-population groups and expenditures identifiable as comprising wholly or mainly benefits with a matching LFS variable have been allocated directly on the basis of shares of claimants by sub-population group in the LFS. This should itself give a more accurate picture. Dustmann and Frattini also assumed that - with the exception of housing benefit which they allocated on a regional basis to reflect differences in claim amounts across regions - claimants of particular benefits all received the same amount of the benefit in question. Conceptually, it would be possible to model the amount of benefits received on the basis of household incomes in the same way as income tax and NICS payments are modeled. However, in practice, the shortcomings in the LFS data outlined above in relation to the calculation of direct and indirect taxes are compounded by the fact that because only a minority of those surveyed are asked questions about their earnings, cross-tabulating earnings with benefit receipt for unique households results in small sample sizes, which when further account is taken of the detailed characteristics that determine entitlement, mean that little assurance could be given that the results would be significantly more reliable for these purposes than using average amounts. For this reason work undertaken along these lines are not reported in this paper. Finally, some benefits are paid on an individual basis - like Jobseeker's Allowance and ESA, and some on a 'family' or 'benefit unit' basis - like tax credits and housing benefit. In relation to benefits of the latter type it is clear that respondents to the LFS are not entirely consistent in their responses: in some couple households only one partner is recorded as claiming the benefit, in others both partners are recorded as claiming the benefit. So the implicit assumption is that this response error does not vary significantly across sub-population groups.
Dustmann and Frattini apportioned all of these costs on the basis of nationality of prison inmates from the Ministry of Justice’s Offender Management Caseload Statistics proportionately to the size of each group in the prison population as these statistics do not record country of birth. Using nationality as an indicator is likely to underestimate the cost attributable to immigrant groups because of the number of immigrants who will have obtained UK nationality since arrival in the UK. This is supported by the observation that by nationality A10 prisoners form a higher proportion of non-UK prisoners than A10 immigrants by birth do in the population of non-UK born. The apportionment is followed although while this might be a reasonable basis for apportioning prison costs specifically, the costs of operating of the wider court system, which encompasses all sorts of non-criminal proceedings and civil litigation might well be different. Also in relation to court costs, it does not take account of any different likelihood of members of sub-population groups to receive custodial sentences.
Housing development comprises of expenditures for social and local authority housing. These are directly allocated according to shares of housing association tenants and local authority tenants respectively as this distinction is now made in the LFS.
Expenditure on ‘Immigration and citizenship-related police services’ are apportioned only to immigrants, proportionately to the share of each group in the total immigrant population. Expenditure on ‘Other police services’ that encompasses general policing is attributed on a per capita basis among the whole population.
Total Managed Expenditure (TME) is an aggregate drawn from National Accounts, covering the current and capital expenditure of the public sector, net of some receipts. It therefore includes expenditure of central and local government and also the capital expenditure of public corporations. TME excludes grants and interest payments between parts of the public sector (as it is a consolidated measure) as well as all financial transactions (such as lending). Accounting adjustments are included in PESA to reconcile total expenditure by sub-function with TME. It appears that Dustmann and Frattini might not have included this component of TME in their calculation. Because TME is the expenditure side of the equation that gives public sector net borrowing as a measure of the fiscal stance, it seems preferable to include it, and the amounts are allocated in line with shares of total population.
1. Dustmann and Frattini in their central scenario allocated 'pure public goods' expenditures on the basis of 16+ population share. These expenditures cannot be directly or easily attributed on the basis of individual characteristics, and so they are assumed to have been spent to the equal benefit of everyone in the adult population. This is the average cost method.
2. However, where these expenditures will not vary directly in proportion to the size of the population, a different approach is to say that the cost to be allocated to migrants should be reduced to take account of the fact that the increase in population resulting from migration does not increase these costs to the same extent. This is the marginal cost method. Dustmann and Frattini calculated an alternative scenario of the fiscal impact of immigration assuming that the marginal cost for all ‘pure public goods’ was zero.
3. For some of these expenditures the marginal cost might in fact be zero, but it is not necessarily that case that they will be, as Williams points out in his commentary on Dustmann and Frattini (Williams 2013).
4. An alternative scenario has been calculated which allocates a zero cost to migrant sub-population groups for the PESA components comprising Public and Common Services and International Services, and for all Defence. Public sector debt interest is excluded because to assume a zero marginal cost would be to ignore the extent to which debt is incurred to avoid taxing the total population more heavily or cutting public services that are used by the total population.
5. While Dustmann and Frattini also allocated a zero cost to the migrant sub-population groups for expenditures on transport, energy, communication and construction, it seems more plausible to assume that these are will in reality be influenced by population size - as having overall a marginal cost lower than the average cost, bearing in mind that the marginal cost will in fact in some cases be higher than the average cost.
6. Unsurprisingly, this shifts more of the burden of expenditures onto the native population, but does not affect the overall balance sufficiently to push the migrant population overall into a positive contribution, or change the balance for any sub-population group from negative to positive other than for the earlier-arriving Other EEA group.
|Expenditures £m||Revenues £m||Ratio||Relative ratio||Net fiscal effect £m|
|Other EEA pre-2001||10,759||10,911||1.01||1.20||152|
|recent EEA other||4,341||8,500||1.96||2.32||4,159|
Expenditures Allocation Criteria categories from PESA 2015 Table 5.2 Public sector expenditure on services by sub-function
|1. General public services||Share of 16+ population|
|2. Defence||Share of 16+ population|
|3. Public order and safety|
|3.1 Police services|
|of which: immigration and citizenship||All to non-UK born 16+ pro rata|
|of which: other police services||Share of total population|
|3.2 Fire-protection services||Share of 16+ population|
|3.3 Law courts||Nationality of prison population|
|3.4 Prisons||Nationality of prison population|
|3.5 R&D public order and safety||Share of 16+ population|
|3.6 Public order and safety n.e.c.||Share of 16+ population|
|4. Economic affairs||Share of 16+ population|
|5. Environment protection||Share of 16+ population|
|6. Housing and community amenities|
|6.1 Housing development|
|of which: local authority housing||Share of LA tenants|
|of which: other social housing||Share of HA tenants|
|6.2 Community development||Share of 16+ population|
|6.3 Water supply||Share of 16+ population|
|6.4 Street lighting||Share of 16+ population|
|6.5 R&D housing and community amenities||Share of 16+ population|
|6.6 Housing and community amenities n.e.c.||Share of 16+ population|
|Medical services||2006-reported distribution of health costs by age group|
|Medical research||Share of 16+ population|
|Central and other health services||2006-reported distribution of health costs by age group|
|8. Recreation, culture and religion||Share of 16+ population|
|9.1 Pre-primary and primary education|
|of which: under fives||Share of 0-4 population|
|of which: primary education||Share of 5-10 population|
|9.2 Secondary education||Share of 11-18 population|
|9.3 Post-secondary non-tertiary education||Share of 16+ population|
|9.4 Tertiary education||Share of students at tertiary institutions|
|9.5 Education not definable by level||Share of 16+ population|
|9.6 Subsidiary services to education||Share of 16+ population|
|9.7 R&D education||Share of 16+ population|
|9.8 Education n.e.c.||Share of 16+ population|
|10. Social protection|
|10.1 Sickness and disability||Share of sickness and disability benefit claimants|
|10.2 Old age||Share of pension claimants|
|10.3 Survivors||Share of pension claimants|
|10.4 Family and children|
|of which: personal social services||Share of income support and child benefit claimants|
|Income Support||Share of income support claimants|
|Child benefit||Share of child benefit claimants|
|10.5 Unemployment||Share of unemployment benefit claimants|
|10.6 Housing||Share of housing benefits claimants|
|10.7 Social exclusion n.e.c||Share of tax credits claimants|
|10.9 Social protection n.e.c.||Share of 16+ population|
|EU transactions and accounting adjustments||Share of total population|
Receipts Allocation Criteria categories from OBR EFO December 2015
|Income tax||Direct attribution/retired households|
|National insurance contributions||Direct attribution|
|Value added tax||By income decile position|
|Corporation tax||After allowance for foreign shareholdings, Long-term resident/all16+ population|
|Petroleum revenue tax||16+ population|
|Fuel duties||By income decile position|
|Business rates||Long-term resident/all16+ population|
|Council tax||By heads of household|
|VAT refunds||By income decile position|
|Capital gains tax||Long-term resident/all16+ population|
|Inheritance tax||By homeowners 70+|
|Stamp duty land tax||By income decile position|
|Stamp taxes on shares||By income decile position|
|Tobacco duties||By income decile position|
|Spirits duties||By income decile position|
|Wine duties||By income decile position|
|Beer and cider duties||By income decile position|
|Air passenger duty||By income decile position|
|Insurance premium tax||By income decile position|
|Climate change levy||16+ population|
|Other HMRC taxes||16+ population|
|Vehicle excise duties||By income decile position|
|Bank levy||Long-term resident/all16+ population|
|Apprenticeship levy||16+ population|
|Licence fee receipts||16+ population|
|Environmental levies||16+ population|
|EU ETS auction receipts||16+ population|
|Scottish taxes||16+ population|
|Diverted profits tax||16+ population|
|Other taxes||16+ population|
|Less own resources contribution to EU||16+ population|
|Interest and dividends||16+ population|
|Gross operating surplus||16+ population|
|Other receipts||16+ population|
Altonji, J.G. and Card, D. (1991), ‘The effects of immigration on the labor market outcomes of less-skilled natives’, in (J.M. Abowd and R.B. Freeman, eds.), Immigration, Trade and Labor, pp. 201–34, Chicago, IL: University of Chicago Press.
Borjas, G.J. (2003). ‘The labor demand curve is downward sloping: reexamining the impact of immigration on the labor market’, Quarterly Journal of Economics, vol. 118(4), pp. 1335–74.
Card, D. (1990). ‘The impact of the Mariel Boatlift on the Miami labor market’, Industrial and Labor Relations Review, vol. 43(2), pp. 245–57.
Card, D. (2001). ‘Immigrant inflows, native outflows, and the local labor market impacts of higher immigration’, Journal of Labor Economics, vol. 19(1), pp. 22–64.
Devlin et al (2014). 'Impacts of migration on UK native employment: An analytical review of the evidence' , Home Office/BIS Occasional Paper 109. At https://www.gov.uk/government/uploads/system/uploads/attachment_data/file/287287/occ109.pdf
Drinkwater, S. and Robinson, C. (2013). ‘Welfare participation by immigrants in the UK’, International Journal of Manpower, vol. 34(2), pp. 100–12.
Dustmann, C. and Fabbri, F. (2005). ‘Immigrants in the British labour market’, Fiscal Studies, vol. 26(4), pp. 423–70.
Dustmann, C., Fabbri, F. and Preston, I. (2005). ‘The impact of immigration on the British labour market’, The Economic Journal, vol. 115(507), pp. F324–41.
Dustmann, C., Frattini, T. and Halls, C. (2010). ‘Assessing the fiscal costs and benefits of A8 migration to the UK’, Fiscal Studies, vol. 31(1), pp. 1–41.
Dustmann, C., Frattini, T. and Preston, I. (2013). ‘The effect of immigration along the distribution of wages’, Review of Economic Studies, vol. 80(1), pp. 145–73.
Dustmann, C. and Frattini T. (2014) 'The Fiscal Impact of Immigration to the UK', The Economic Journal, vol. 124(580), pp. F593-F643 November 2014
Dustmann, C. and Theodoropoulos, N. (2010). ‘Ethnic minority immigrants and their children in Britain’, Oxford Economic Papers, vol. 62(2), pp. 209–33.
Dustmann, C. and Weiss, Y. (2007). ‘Return migration: theory and empirical evidence for the UK’, British Journal of Industrial Relations, vol. 45(2), pp. 236–56.
Gott, C. and Johnston, K. (2002), ‘The migrant population in the UK: fiscal effects’, Home Office Research, Development and Statistics Directorate Occasional Paper No.77, London.
Investment Property Forum (2014), 'The Size and Structure of the UK Property Market 2013: A Decade of Change', available at http://www.indirex.com/uploads/Size_and_Structure_of_UK_Property_Market_2013_-_A_Decade_of_Change_Summary_Report.pdf
Lemos, Sara, Mind the Gap: A Detailed Picture of the Immigrant-Native Earnings Gap in the UK Using Longitudinal Data between 1978 and 2006. IZA Discussion Paper No. 6058. Available at SSRN: http://ssrn.com/abstract=1955398
Liebig, T. and Mo, J. (2013), ‘The fiscal impact of immigration in OECD countries’, chapter 3, in (OECD, ed.), International Migration Outlook 2013, pp. 125–89, Paris: OECD Publishing.
Manacorda, M., Manning, A. and Wadsworth, J. (2012). ‘The impact of immigration on the structure of wages: theory and evidence from Britain’, Journal of the European Economic Association, vol. 10(1), pp. 120–51.
McInnes, R., (2014) 'Statistics on Migrants and Benefits', House of Commons Library Commons Briefing papers SN06955 available at http://researchbriefings.parliament.uk/ResearchBriefing/Summary/SN06955
Migration Observatory, (2016) 'Migrant Remittances to and from the UK', available at http://www.migrationobservatory.ox.ac.uk/briefings/migrant-remittances-and-uk
Migration Watch UK (2014), ‘An Assessment of the Fiscal Effects of Immigration to the UK’, Available at http://www.migrationwatchuk.org/briefing-paper/1.37
OECD (2013) 'International Migration Outlook 2013', available at http://www.oecd-ilibrary.org/social-issues-migration-health/international-migration-outlook-2013_migr_outlook-2013-en
ONS (2014), 'Revisions to Labour Force Survey estimates due to re-weighting to the Census 2011 population' available at https://www.gov.uk/government/statistics/revisions-to-labour-force-survey-estimates-due-to-re-weighting-to-the-census-2011-population
Ottaviano, G.I.P. and Peri, G. (2012). ‘Rethinking the effect of immigration on wages’, Journal of the European Economic Association, vol. 10(1), pp. 152–97.
Rowthorn, R. (2008). ‘The fiscal impact of immigration on the advanced economies’, Oxford Review of Economic Policy, vol. 24(3), pp. 560–80.
Rowthorn, R. (2014), ‘A note on Dustmann and Frattini’s “Estimates of the fiscal impact of UK immigration”’, available at http://www.civitas.org.uk/pdf/rowthorndustmannfrattini.pdf
Rowthorn R. (2015), 'The Costs and Benefits of Large-Scale Immigration', Available at http://www.civitas.org.uk/pdf/largescaleimmigration.pdf
Ruist, Joakim (2014), 'The fiscal consequences of unrestricted immigration from Romania and Bulgaria', Discussion Paper Series, Centre for Research and Analysis of Migration, CDP04/14
Sriskandarajah, D., Cooley, L. and Reed, H. (2005). ‘Paying their way: the fiscal contribution of immigrants in the UK’, research report, Institute for Public Policy Research, London.
Wadsworth, J. (2013). ‘Musn’t grumble: immigration, health and health service use in the UK and Germany’, Fiscal Studies, vol. 34(1), pp. 55–82.
Wadsworth, J. (2015) 'Immigration and the UK Labour Market' Paper No. CEPEA019 Centre for Economic Performance (LSE) available at http://cep.lse.ac.uk/pubs/download/EA019.pdf
Williams N (2013) “Responding to ‘the Fiscal Effects of Immigration to the UK’”, Civitas, London 2013