Evaluation, Revision and Extension of Ethnic Population Projections - NewETHPOP

Summary
This project aims to understand and to forecast the ethnic transition in the United Kingdom's population at national and subnational levels. The ethnic transition is the change in population composition from one dominated by the White British to much greater diversity. In the decade 2001-2011 the UK population grew strongly as a result of high immigration, increased fertility and reduced mortality. Both the Office for National Statistics (ONS) and Leeds University estimated the growth or decline in the sixteen ethnic groups making up the UK's population in 2001. The 2011 Census results revealed that both teams had over-estimated the growth of the White British population and under-estimated the growth of the ethnic minority populations. The wide variation between our local authority projected populations in 2011 and the Census suggested inaccurate forecasting of internal migration. We propose to develop, working closely with ONS as our first external partner, fresh estimates of mid-year ethnic populations and their components of change using new data on the later years of the decade and new methods to ensure the estimates agree in 2011 with the Census. This will involve using population accounting theory and an adjustment technique known as iterative proportional fitting to generate a fully consistent set of ethnic population estimates between 2001 and 2011.

We will study, at national and local scales, the development of demographic rates for ethnic group populations (fertility, mortality, internal migration and international migration). The ten year time series of component summary indicators and age-specific rates will provide a basis for modelling future assumptions for projections. We will, in our main projection, align the assumptions to the ONS 2012-based principal projection. The national assumptions will need conversion to ethnic groups and to local scale. The ten years of revised ethnic-specific component rates will enable us to study the relationships between national and local demographic trends. In addition, we will analyse a consistent time series of local authority internal migration. We cannot be sure, at this stage, how the national-local relationships for each ethnic group will be modelled but we will be able to test our models using the time series.

Of course, all future projections of the population are uncertain. We will therefore work to measure the uncertainty of component rates. The error distributions can be used to construct probability distributions of future populations via stochastic projections so that we can define confidence intervals around our projections. Users of projections are always interested in the impact of the component assumptions on future populations. We will run a set of reference projections to estimate the magnitude and direction of impact of international migrations assumptions (net effect of immigration less emigration), of internal migration assumptions (the net effect of in-migration less out-migration), of fertility assumptions compared with replacement level, of mortality assumptions compared with no change and finally the effect of the initial age distribution (i.e. demographic potential).

The outputs from the project will be a set of technical reports on each aspect of the research, journal papers submitted for peer review and a database of projection inputs and outputs available to users via the web. The demographic inputs will be subject to quality assurance by Edge Analytics, our second external partner. They will also help in disseminating these inputs to local government users who want to use them in their own ethnic projections. In sum, the project will show how a wide range of secondary data sources can be used in theoretically refined demographic models to provide us with a more reliable picture of how the UK population is going to change in ethnic composition.

Impact summary
The project will develop new estimates of local ethnic population estimates and components for the 2001-2011 decade and use them to produce new projections of the UK's ethnic transition. To do this we will need to apply demographic methods in innovative ways and to develop a new projection model compatible with ONS's Conceptual Framework for Population and Migration Statistics.

The Office of National Statistics (ONS) will benefit from co-production of revised estimates of ethnic population change. The ONS team and the project team will discuss their respective methods and data sets for revising the 2001-2011 ethnic estimates. Although the ONS revised PEEG estimates and the Leeds/Newcastle NewETHPOP estimates will be separate products, the collaboration will produce aligned data sets. The project may help in moving the PEEG estimates from the status of experimental statistics to official statistics status.

ONS will benefit from use of a full set of estimated migration flows between local authorities in and between all home countries that make up the UK (based on the work of Lomax et al. 2013 and Lomax 2013) and the total out- and in-migration classified by ethnicity which we develop in the project. We note that Eurostat require from the UK statistical agencies information on migration flows between local authorities within the whole country and their aggregations to regions defined in the Nomenclature des unites territoriales statistiques (NUTS). In a previous project Dennett and Rees (2010) supplied migration flow estimates for NUTS 2 regions to Eurostat on behalf of ONS.

ONS have a goal to produce confidence limits in future versions the National Population Projections based on probabilistic projections. We anticipate that our experience in producing probabilistic projections for local ethnic group populations will be helpful for this programme.

We will inform the statistical agencies of the devolved administrations (the Welsh, Scottish and Northern Ireland Governments) of the progress of our work, which will be of considerable interest. The 2011 Census provides ethnic classifications much better harmonised than in the 2001 Census, so we will be able to produce and publish local authority estimates and projections for these home countries.

Working with non-academic partner Edge Analytics, we will disseminate our new estimate and projection data to Local authorities (LAs). They will have access to a new updated set of ethnic population projections via the web interface and database. They will have access to our ethnic component estimates and assumptions for use in their own projections, based on the POPGROUP software. We will consult with colleagues at the Greater London Authority (GLA) and keep them informed of our work. Both Edge Analytics and the Leeds/Newcastle team have carried out work for the GLA. The GLA produce the most comprehensive set of local authority ethnic projections in the UK.

We will also aim to reach a wide variety of other users. The ETHPOP user list has currently ~200 registered users The largest group are academic users (58), NHS users (36) and a government users (36). There are also researchers who use a private account and from organisations such as diabetes.org, mariecurie.org, havenhouse.org, bmecancer and demelza.org. A few subscribers are from newspapers, consultancies and other private sector companies.