National Records of Scotland

Preserving the past, Recording the present, Informing the future

Background Information: Methodology and Sources Used

Background Information: Methodology and Sources Used

5.1 Previous Household Estimates Based on Changes in Housing Stock

In the past, household estimates were based on changes in the housing stock since the last Census.  Previous household estimates, based on this approach, are available from this website, along with a description of the methodology used.

This approach involved annually updating the number of households identified at the last Census on the basis of changes in the housing stock (new builds, demolitions and conversions). The figures were adjusted to account for changes in the number of vacant properties in public sector housing, and assumptions are made about trends in the number of vacant properties in other housing sectors.  However, the time taken to assemble information from each local authority on changes in the housing stock meant that there was a time lag in producing the household estimates.  In addition, this approach required an assumption about the level of vacancies in private sector housing, which is the largest component of the housing stock.

A review of the household estimates methodology was carried out in 2004-05, with local authorities and other data users, to agree which approach should be used for producing household estimates in the future.  The conclusion of this was that future household estimates should be based on Council Tax figures.  This has the advantage that the figures will be more timely, and they are less likely to ‘drift’ between Censuses, as information about vacancy levels is updated annually, rather than relying on assumptions based on the last Census.  In addition, the National Records of Scotland is investigating the potential to develop small area estimates of occupied and vacant dwellings from Council Tax systems.  These figures would be published for each ‘data zone’ - a standard geography used by the Scottish Neighbourhood Statistics project, containing an average of 750 people.  These figures will provide far more detailed, local information about occupied and vacant dwellings, which will enable estimates to be produced for different geographies, and the figures will also tie in with the approach used here.

5.2 Council Tax Based Household Estimates (2001-2004)

Following the review described above, this publication contains the first set of household estimates for Scotland that are based on Council Tax figures.  Each council maintains a register of all domestic dwellings for Council Tax purposes, which records whether they are entitled to any discounts or exemptions.  Certain discounts are available for dwellings that are empty or unoccupied, which can be used to estimate the number of vacant and occupied dwellings, as follows:

  • Unoccupied dwellings that are exempt from Council Tax
  • Dwellings subject to second home or long-term empty property discount (including holiday homes and self-catering accommodation available to let for less than 140 days per year)

This information is provided by each council to the Scottish Executive’s Local Government Finance Statistics Branch each September, via the form CTAXBASE (which can be found on the Scottish Executive's website, on the section corresponding to these statistics, in the Data Supplier Area).  The figures contained in Tables 1-4 are based on these figures. 

For each council, information is available on the total number of dwellings on the valuation list (Table 2).  Free-standing lock-ups are excluded, as they do not provide accommodation so are not considered to be dwellings.  The number of vacant dwellings (dwellings entitled to unoccupied exemptions or second home and long-term empty property discounts) is shown in Table 3, and the difference between these two figures is the number of occupied dwellings, shown in Table 4.

5.2.1 Differences Between Council Tax Figures and the 2001 Census

The number of occupied dwellings, based on Council Tax figures (Table 4) does not exactly match the number of households recorded in the 2001 Census.  There are a number of reasons for this, for example:

  • Shared dwellings: The Council Tax figures record the number of occupied dwellings, not the number of households.  A single occupied dwelling, as recorded in the Council Tax return, could contain more than one household, as defined for the Census.
  • Communal establishments: The Council Tax figures on occupied dwellings (Table 4) include dwellings classified as ‘occupied exemptions’ – dwellings that are occupied, but exempt from paying Council Tax, such as all-student households.  This category can also include some communal establishments, such as halls of residence, barracks and prisons, which for the Census would be classified as communal establishments, so would not be included in the total number of households.
  • Self-catering holiday accommodation: Self-catering holiday accommodation which is available to let for less than 140 days per year is entitled to a second home or long-term empty property discount.  These dwellings are included in Table 3 (empty/vacant dwellings) and excluded from the total number of occupied dwellings (Table 4).  If they are available to let for more than 140 days per year, they will be included on the non-domestic rates valuation roll instead.  This can lead to a difference between the Census and Council Tax-based figures, and can also lead to fluctuations in the number of vacant and occupied dwellings between years, if properties yo-yo back and forth between the different forms of occupation.  Such fluctuations will be highest in areas with relatively high levels of self-catering holiday accommodation, such as Argyll & Bute, Highland and the island authorities.
  • Lags in Council Tax systems being updated: There can be lags in Council Tax billing systems being updated, particularly the information on any discounts that apply to the dwelling, such as whether it is empty or vacant.  This can lead to lags in the figures being updated, particularly the figures in Table 3 (empty/vacant dwellings) and Table 4 (occupied dwellings).  These lags may vary between areas, and are likely to be greatest in areas of highest turnover, such as areas with a high proportion of private renting, or areas with a large number of vacant dwellings or holiday accommodation.
  • Persons disregarded for Council Tax purposes:  A small number of dwellings which are classified as vacant according to Council Tax information could be occupied, if they consist entirely of people who are ‘disregarded’ for Council Tax purposes.

To account for these differences, the figures are anchored to the 2001 Census: the difference between the 2001 Census and the number of occupied dwellings in 2001 is subtracted from each year’s figures.

5.3 Population-based Household Estimates (1991-2001)

Council Tax was introduced in the early 1990s, so a different source must be used to provide household estimates going back to 1991.  In order to provide an estimate of the number of households each year between 1991 and 2001, information from the 1991 and 2001 Censuses has been used, along with each year’s mid-year population estimate.  The method used is described below. 

5.3.1 Private Household Population

The population aged 16 and over is taken from each year’s mid-year population estimate, produced by the National Records of Scotland.  The number of people living in communal establishments is then subtracted from this figure in order to estimate the population living in private households.  The communal establishment population is estimated by taking the proportion of the population living in communal establishments, by age group and sex, in the 1991 and 2001 Censuses, and using these figures to estimate the proportion in each of the intervening ten years.  This is done using the 2-point exponential function described in section 5.3.2 below.  The proportion of the population assumed to be living in communal establishments is then multiplied by the mid-year population estimate (for each age group and sex), and subtracted from the overall population to estimate the number of people living in private households.

5.3.2 Household Composition

The proportion of households in each local authority area, by household type and the age group of the head of household, is known from the 1991 and 2001 Censuses.  This is used to estimate the household composition in the intervening years.

The number of people who head a household will be equal to the number of households.  The proportion of people heading households is known as the ‘headship rate’, and a modified 2-point exponential function is used to interpolate both the ‘headship’ and ‘non-headship’ rates between 1991 and 2001.  The formula for this is as follows:

yi = k + abxi

                               

 where i = the year, from 1991 to 2001
          yi = headship rate in year i
          k = 1 if y2001 ≥ y1991
                0 if y2001 < y1991
          a = y1991 – k
          b = (y2001 – k)/(y1991 – k)
          xi = (i -1991)/(2001-1991)

The calculated headship rates are constrained in two ways – they cannot go above 1 or below 0, and the headship and non-headship rates must sum to 1 within an area and household type.

The household estimates for each year are then calculated by applying each year’s headship rates to the population projections.

The household estimates for 1991 and 2001 do not exactly equal the results of the 1991 and 2001 Censuses.  This is because the population figures used are the mid-year population estimates, from the end of June, whereas the Census was carried out in April.  In addition, the mid-year population estimates for 1991 to 2000 were revised in light of the results of the 2001 Census, and these revised figures are used here, as they are more accurate than the original population estimates published.

5.4 Comparisons Between Household Estimates Based on Council Tax and Population Figures

As described above, the household estimates shown in Table 1 are based on population figures for 1991-2001, and based on Council Tax figures for 2001-2004.  As Council Tax figures are collected in September rather than June, they have been adjusted to June of each year in Table 1, to provide comparable figures for each year.

5.5 Scottish Household Survey Results on Household Type

Household estimates based on Council Tax figures do not provide much information about household composition.  Instead, the Scottish Household Survey (SHS) is used to estimate the proportion of people living in different households types, across Scotland (Table 5).  The SHS is a sample survey of around 15,000 households per year.  The sample size is not large enough to provide a reliable measure of the change in each household type, in each local authority area, each year.  However, it does provide reliable figures for Scotland as a whole. 

Unlike the Census, the SHS is not compulsory.  When  the results for 2001 are compared to the 2001 Census, the SHS slightly under-estimated the proportion of the population living in certain household types (such as single men), and over-estimated the proportion living in other household types.  To account for this, the results shown in Table 5 are adjusted to account for the difference between the SHS and the Census in 2001.  The figures are also adjusted to equal the total number of households in each year, as shown in Table 1.