Annex B - Uses, Limitations and Consequences of Projections

1. Uses

The main use of the household projections is as part of the process of assessing future housing demand. Specifically, local authorities make widespread use of the household projections in housing plans to assess future housing need, in the structure plan process and as a context for planning approval decisions. Uses of household projections are partly determined by central government guidance (PAN 38 and NPPG3).

Projections are (or should be) only one element in assessing future housing need - in recent years, the analysis of housing market areas has become increasingly sophisticated, and other factors affecting demand are taken into account.

2. Limitations

The limitations of these household projections must be fully recognised. A projection is a calculation showing what happens if particular assumptions are made. The household projections are trend-based, and are not, therefore, policy-based forecasts of what the Government expects to happen. Many social and economic factors influence the formation of households, including policies adopted by both central and local government. It should be remembered that new local planning policies are often intended to modify past trends. Structure plans may be based on reasoned and agreed departures from the projections that seem better able to fit particular local circumstances.

2.1. National Records of Scotland Population Projections

The household projections are based on the population projections and, as a result, assumptions used for the population projections, such as future migration, fertility and mortality, will have an effect on the household projections. Since smaller areas tend to be affected more by migration, projections for these areas tend to be less reliable than those for areas with large populations, especially the further into the future the projections are taken.

2.2. The Use of Census Information

The second main inputs to the household projections are the headship rates from census information projected forwards to the end of the projection period. Since the 2001 census information will not be available until the next set of projections (expected to be published in 2004), the latest available set of figures must come from the 1991 census. As a result, the number of years between the latest measured data point (the census) and the base year for these 2000-based projections is at its greatest. The base year uses headship rate projections that are 9 years ahead of the last known data point. The end of the projection period (2014) being 23 years ahead of the last measured data point. This represents a serious limitation and may explain the increased adjustment for the minimum number of adults that was required in the 2000-based projections.

The Household Analyses Review Group looked into the possibility of using other, more recent sources of information to project headship rates. It is acknowledged that there may be more recent data available for some local areas such as voluntary population surveys. However such surveys are not available or consistent for all areas within Scotland. The new Scottish Household Survey, which began in 1998, has the potential to provide information on more recent trends in household composition. Once the results of the 2001 Census become available the Household Analyses Review Group will consider how the survey data might be incorporated into the projections methodology.

It is accepted that since the methodology is designed to project consistently across all Local Authority areas and give a national total, figures for individual authorities may be less reliable than fuller local analyses.

An example of this problem can be seen in the Glasgow and Clyde Valley Structure Plan area. The year 2000 structure plan document contained household projections which used 1971 to 1991 trends in household formation. Justification for using these trends, rather than the 1981 to 1991 trends incorporated in the Scottish Government 1998-based household projection, was given by reference to the lower rate of household growth observed in the 1990s. Additionally, there was a discrepancy between projected rate of change in number of single parent households and observed rate of change from the Voluntary Population Survey, particularly for Glasgow City. It was accepted that the rate of household increase in Glasgow in the 1980's was particularly high because of various social and economic factors, as well as a one off change in letting policy for local authority houses. As a result the use of the 1981 and 1991 Census data for the Scottish Government projections was perhaps less appropriate for these particular areas than the 1971 and 1991 Census data which would tend to reduce the effect of the household formation seen in the 1980s. This problem will continue with these 2000-based projections since the same 1981 and 1991 Census data have been used. As mentioned in Paragraph A.7, this was agreed by the Household Analyses Review Group to minimise the adjustment required across all Local Authority areas. This was to ensure that the number of adults living in private households (estimated from the National Records of Scotland population projection) are at least equal to the number of adults implied by the household projections.

Similarly, particular care is needed in interpreting the projections for Aberdeen City. This is due to the very large adjustment that has been required for all projection years (both in the 1998-based and current 2000-based projections) to satisfy the minimum adult requirement (refer to paragraph A.7.2). It is possible that this large adjustment (to reduce the minimum adults required to fill the projected household by 11,430) reflects an over projection of the total number of households in Aberdeen City over the whole projection period. The following table shows the changes made by household type. Please note that the total number of households across all household types is kept constant.

Aberdeen City: Minimum Adult Adjustments

2.3. Rounding

After consultation with users through the Household Analyses Review Group, it was decided that from 2000, household projections would be published to the nearest 10. The main reason for this is to help distinguish trends within the smaller Local Authorities and / or within household types or age bands with fewer households.

In the past figures were rounded to the nearest 100 to demonstrate the imprecise nature of the assumptions used in the projection methodology. This rounding does not demonstrate any measurable confidence limit within the assumptions used, but is simply a way of indicating that they are based on many assumptions and are not precise estimates. It is possible to produce figures to the nearest 10 (or even to the nearest integer) and this is how they are presented in this publication. However, it is important to note that this does not mean that the projection is any more precise than in previous years.

3. Consequences [Footnote 1]

It should be recognised that household projections, like other projections, may indicate that existing trends and policies are likely to lead to situations which are judged undesirable. If new policies are then introduced they may result in the original projections not being realised. However, this means the projections will have fulfilled one of their prime functions; to show the consequences of present trends with sufficient notice for any necessary action to be taken.

Footnotes

1. This note is partly based on guidance provided on the uses and limitations of projections from 1993-based Subnational population projections; OPCS series PP3 no.9 (HMSO).