Select Committee on Treasury Written Evidence


Supplementary memorandum from the London Borough of Newham

  Newham would like to take this opportunity to provide a written response to the following questions, taken from the session heard on Wednesday 23 January 2008, to offer suggestions improvement.

Q164 Mr Love:   What I would like you to do is to give us some examples of where you think it is going wrong in your own local experience but perhaps, more importantly, how do we correct those mistakes in a way that will find validity of the ONS?

Q166 Mr Love:   What are the steps that need to be taken to recognise the population increase in Newham [and elsewhere]?

  ONS has identified migration in-flows and out-flows, particularly for international migrants, as an area of primary concern. Migration within the UK is estimated by the movements on GP registers, and this is seen as currently the best available data for this purpose. The international flows, in contrast, use sampled data from the International Passenger Survey (IPS), supplemented by use of the Labour Force Survey (LFS)—another sample survey—to apportion migrants to local authority geographies, and by Census data from 2001.

Quantifying international migration

  The international flows at national level are currently determined by the IPS. There remains some doubt as to whether these accurately reflect international flows because:

    —  It may not adequately cover all migrants as it is a voluntary survey and those with language barriers are less likely to participate.

    —  It is not clear that the coverage operates adequately out-of-hours, thereby missing migrants on cheaper flights that arrive or depart in the early morning or late at night, particularly at smaller ports.

    —  The sampling of each port is believed to be determined by the flows of international flights to and from it—and these smaller (and probably cheaper) ports may attract proportionately larger numbers of aspiring residents than the larger and more expensive ports.

    —  The definition of "migrant", based on intended residency of a year or more, filters out many respondents at the time of questioning who do not plan to stay, but who subsequently become resident.

  If we are to have confidence in the IPS total count, we need to be convinced that the current arrangements adequately reflect migrants to and from the country. Research could be undertaken to link what is gleaned from the IPS with administrative data sources, so that we can have more confidence in the overall counts as well as utilise this data for the more difficult allocation of migrants to local authorities. We need to know what proportion of Flag 4 (foreign patients registering with a GP for the first time) or National Insurance applicants (NINO) records are likely to become "resident".

  Information about foreign students could become part of a more sophisticated model, with some of them becoming resident or remaining resident for a limited time and many leaving. The age group structure of Flag 4 data could be compared to the IPS population as a "reality check" and help to inform the age/gender distribution of international migrants.

  Such research could then be used to assist development of a model for "propensity to leave" of new migrants, identifying age groups of foreign arrivals that tend to leave the country (eg students, or pensioners returning "home") to help us improve our understanding of this aspect of migration.

  E-borders will clearly offer a much richer source of data than that which the IPS provides for in- and out-flows, and should be implemented with high priority to remove a large source of uncertainty in the current methodology. In the absence of E-borders, we will have to accept that the overall national levels of migration are largely correct and seek other sources to help in its use.

Allocating migrants to Local Authorities

  It is the next stage, of allocating international migrants down to local authorities, that is seen as fraught with error. Currently, the IPS data is then modelled to local authorities using data from the Labour Force Survey (LFS), and Census 2001 data on foreign-born residents. This too is seen as flawed because the sampling for the LFS is also small, and the data from the Census does not adequately reflect current migration patterns.

  To address this, it is proposed that administrative data is used to model migration flows to local authorities, using not the GP registration Flag 4 counts, but the proportional flows of international migrants to areas each year, eg:

    Total IPS international migration x (LA Flag 4 counts/national Flag 4 counts)

  The GP register indicates a pattern of settlement. It is assumed that most people register with a GP either when they have a residence in an area and begin making use of community services, or when they are ill. It is not believed that people tend to immediately register in an area only to move on to another area within a year, and it is known that some people do not register for some time after moving into an area. Current methodology using the Census assumes that new migrant arrivals are likely to settle where currently existing communities are, and it would make sense to use Flag 4 GP registration data to model current flows rather than relying on Census data that is years old and may not reflect new migrant communities.

  The disadvantage of this method is that some areas will have low levels of new GP registrants. This can be due to a lack of available GPs (in which case the PCT assigns patients to GPs rather than them being accepted voluntarily) or because NHS services are not required. In this case, these local authorities could be allocated an inadequate number of migrants, but ONS might be able to use disparity between LFS data of new arrivals to an area compared to numbers of Flag 4 registrants to identify where this might be an issue. With the wealth of administrative data that is available, these problematic flows should be able to be identified.

Out-migration

  Out-migration is particularly problematic because of the very small sampling done of those leaving the country. Previously, ONS simply applied a proportion to each local authority's population to determine a count of migrants, but they have now created a "propensity to migrate" model to guide their methodology. This is based on population characteristics to determine what proportion of each local authority's population is likely to emigrate. The premise is that, for example, single young males are more likely to be emigrants than 11-year-old girls, or the families of 11-year-old girls, and that local authorities with large numbers of young adults are more likely to have emigrants than those with large pensioner populations. This makes intuitive sense.

  However, the single characteristic that is likely to impact on whether someone leaves the country is excluded—that of country of origin. ONS seemingly believes that there is an equal likelihood of any particular age or gender leaving the country, without any reference to the ties they may have to another country, and that a former economic migrant from a lesser-developed nation will have the same likelihood of return or departure as someone from Western Europe, North America or Oceania.

  This does not make sense, and it is not understood why ONS has chosen to exclude country of origin as a factor in their analysis. While it is accepted that individual countries of origin may be too complex to model, global regions could provide insight into differing migration patterns, and this should be investigated.

  Administrative data could provide some insight into differential rates of return/emigration if it can be appropriately linked. For example, applicants for National Insurance numbers could be identified and their country of origin noted and this linked to their National Insurance accounts for which no contributions have been made for some time (or possibly tax records), to give an indication of individuals who are likely to have departed from the country. This would help build understanding of demographic characteristics of those who have arrived, stayed for awhile, and then left, to help with the construction of the modelling of out-migration.

Range of error in population statistics

  ONS should be able to guide central government in its use of the statistics for resource allocation by identifying those particular local authority populations that it believes are difficult to accurately estimate. It is not unreasonable to ask for information on margins of error, so that funding arrangements could be considered for those areas with the greatest levels of uncertainty.

  In an age where there is such a mobile population, ONS needs to adopt a less fixed "one-size fits all" approach, and identify the range of confidence it has in its existing data sources and begin to identify alternative ways of painting the picture of population in this country. The continued use of trend-based projection modelling is not seen as viable in areas where there is large-scale development, and ONS needs to be more responsive to local authority knowledge of change and impending change in these areas.

  Administrative data should be able to provide some kind of "reality check" to the estimates produced, so that an increase in new dwellings occupied by more than one adult (according to Council Tax registers) are not accompanied by falls in population. Where there is discrepancy between administrative sources of data and population estimates, a dialogue with local government should be sought to be able to understand failings in the outcome, in order to improve either the model or the local administrative data.

  It would be helpful if a collaborative relationship was developed between ONS and local authorities, with feedback from local authorities about the population statistics being considered as helpful rather than simply a means of obtaining additional funding. We need accurate population statistics not just for our funding bases, but also for our service planning and development to ensure that we are responding accurately to real issues, rather than those created out of flawed statistics.

31 January 2008





 
previous page contents next page

House of Commons home page Parliament home page House of Lords home page search page enquiries index

© Parliamentary copyright 2008
Prepared 22 May 2008