Response to the ONS Consultation on the Household Projections for England

Response to the ONS consultation on the household projections for England

Response to the ONS consultation on the household projections for England

30th June 2017

Introduction

This document summarises the responses received to the consultation on the proposed changes to household projections and what actions the Office for National Statistics (ONS) plan to take.

The consultation ran from 31 January 2017 to 13 March 2017 and gave users an opportunity to provide feedback on the proposed changes to the household projections methodology and on what outputs they need from the household projections. The consultation document is available to download from the ONS website.

The consultation was conducted in line with the Code of Practice for Official Statistics.

Background

Household projections project the number of households in England and its local authority districts for 25 years ahead from the base year and are produced every two years. They are used as part of the assessment of future housing need and demand. The most recent set are the 2014-based household projections published in July 2016.

The Office for National Statistics (ONS) recently took over the responsibility for producing household projections from the Department for Communities and Local Government (DCLG). The transfer of responsibility should help to further improve the consistency between the household projections and the national and subnational population projections and also allow us to make some efficiencies in their production.

There were two objectives to the consultation. The first objective was to get your views on the changes that we propose to make to the household projections methodology following the transfer of responsibility to us. The second objective was to improve our understanding of what you needed from the household projections and related outputs.

We are grateful to everyone who took time to respond to the consultation. We had a good response to the consultation and received many detailed replies. This feedback from users with experience of using household projections is invaluable to us and will help us to develop the household projections in a way that will continue to meet your needs.

Summary of responses and our actions

In total, 42 consultation responses were received from a variety of organisations including local authorities, county councils, central government, house building and research groups. We also received responses from a number of academics and private individuals. A full list of organisations that responded can be found in Annex A.

The key points to note from the responses to the consultation are:

·  Users welcomed the transfer of responsibility for household projections to ONS which would help develop their consistency with the national and subnational population projections.

·  Users felt that household projections should continue to be regarded as a robust starting point for assessing future housing need.

·  Some users were concerned about the proposal to project the household representative rates using only the 2001 and 2011 Censuses and asked ONS to consider and report on a number of alternative options.

·  Issues surrounding other aspects of the methodology were raised including concern about the proposal for removing gender and marital status from the projection model and the age groups currently used in the model.

·  Users asked for variant household projections at local authority level.

·  Users asked ONS to publish the household projections at the level of detail previously published by DCLG.

·  There was an interest in what household types would be available in the outputs.

As a result of this consultation, we will take the following actions:

·  We will move to using the standard 2011 Census definition for Household Reference Person as soon as possible.

·  We will set up a programme of research to look at how the methodology can be improved in the light of the feedback from this consultation. Topics to be covered in this research are described in the Detailed responses and our actions section below.

·  We will establish a Household Projections Collaborative Group, including experts from within and outside ONS, to advise and work with us on this research and the longer-term development of the household projections.

·  We will use the current methods as a starting point for the next release of the projections in Summer 2018. The programme of research, and the Collaborative Group, will help us to identify improvements to the methods that we could make in time for that release.

·  We will retain the current level of detail in published outputs and seek further evidence of requirements for variant projections.

It is important that we are able to research the impact of any changes that are made to the household projections methodology before they are implemented. The longer term programme of research will give us the time to fully investigate what improvements can be made to the household projections methodology.

The following sections consider in more detail how users responded to the key issues and how we intend to act on this information.

Detailed responses and our actions

Use of the 2011 Census definition for household reference person projections

Users were asked for their views on the proposal to use the 2011 Census definition for household reference person, which is broadly the oldest economically active person. Their views included:

·  A number of respondents (11) said that they were happy with the proposal to use the 2011 Census definition of household reference person.

·  A small number of respondents (6) said that the definition for household reference person should be consistent with the definition to be used in the 2021 census. Some suggested that the household reference person should be the oldest person in the household to address any possible future inconsistencies. There was concern that use of the 2011 Census definition would result in incompatibility with earlier censuses.

·  A small number of respondents (5) acknowledged the work being taken forward by the Administrative Data Sources Division to look at the feasibility of producing estimates of occupied households using council tax data and said that the definitions used in the projections should be consistent with that work.

Our response

We will move to using the standard 2011 Census definition for Household Reference Person (as shown in the Census Glossary), as soon as this is consistent with the rest of the projections methodology. This will improve the transparency of the projections and increase coherence with other standard population statistics. We have no current plans to change the definition for Household Reference Person in the 2021 Census from that which was used in the 2011 Census.

Use of an age only projection model

Users were asked for their views on our proposal to remove the split by gender and marital status from the household representative rate and to project the rate for quinary age groups only. Their views included:

·  A number of respondents (7) said that they were content with the proposal to use an age-only model.

·  A number of respondents (12) were unhappy with the proposal to remove gender from the projected household representative rates and asked to see what the implications would be of removing it from the model. There was concern that losing both gender and marital status would mean that information on how these two factors drive household formation would be lost.

·  A number of respondents (9) supported the proposal to remove marital status from the projected household representative rates either because they felt it to be a poor indicator of household formation or because of the unreliability of the marital status projections which have not been updated since 2010. A very small number of respondents (2) felt that marital status was a good indicator of household formation and wanted to see it retained in the model.

·  A number of respondents (9) said that the quinary age groups used in the model were not appropriate for young adults, students and the elderly population where they felt there was greater variation in how households were formed. Suggestions included changing the 15-19 group to 16-19 and expanding the 85+ group to include 85-89 and 90+.

Our response

We will conduct further research, in conjunction with the Household Projections Collaborative Group, into the proposal to remove marital status and, in particular, gender from modelled Household Representative Rates to ensure that such a change would improve, rather than damage, the quality of the projections.

We will also consider some changes in the age groups used in the model: specifically, whether the 15-19 group should be changed to 16-19 and whether the highest age group should be changed to 90+ (improving coherence with the population estimates and subnational population projections).

Move to a one stage production approach using Census data from 2001 and 2011 only

Users were asked for their views on our proposal to move to a one stage production approach using only the 2001 and 2011 Censuses to project the household representative rates. Previously Census data from 1971 to 2011 was used to project household numbers in stage one and Census data from 2001 and 2011 was used to project household numbers by household type in stage two. The projected household numbers by household type produced in stage two were then constrained to the projected household numbers in stage one. The new one stage approach would use only Census data from 2001 and 2011 to project the total number of households and by household type. User views included:

·  A number of respondents (15) considered using data from only two Censuses to be insufficient. Some respondents said that a longer trend was needed to smooth out events such as the recent recession, the expansion of the European Union and also any inaccuracies that may have occurred in the 2001 and 2011 Censuses. There was a view that only using the 2001 and 2011 Censuses would result in a downward trend in household formation for the younger age groups, which in turn would downplay the need for housing for younger people.

·  A number of respondents (11) asked for further work to understand the impact of any new methods. Some (5) also asked to have the opportunity to comment on a number of alternative options for projecting the household representative rate using census data.

·  Some respondents (7) suggested that using data from the 1991, 2001 and 2011 Censuses, supplemented by survey data, would be a better way of projecting the household representative rate.

·  A small number of respondents (4) suggested that more recent data is used. Suggestions included giving greater weight to the Labour Force Survey or using a shorter period of recent data to match the methodology used to produce the subnational population projections, which uses a shorter time period to project trends.

·  A number of respondents (6) asked to see the research note mentioned in the consultation document that would show whether the move to a one stage approach using a shorter period of Census data would be reliable. Respondents also asked for ONS to publish some research on the factors that drive household formation. They also asked for research on the methods used by other organisations to produce household projections.

·  Some respondents (10) asked for a review of the methods used by Wales, Scotland and Northern Ireland to produce household projections. They said that adoption of similar methods would develop consistency and enable comparisons to be made.

·  A number of respondents (5) felt that a simplification of the methodology would make it easier to explain to others. A small number (4) felt that the proposed methodology was oversimplified or too mechanistic, or that they preferred the current two stage methodology. The methodology used to constrain and its impact in some areas was also raised.

Our response

We will consider, with the help of the Household Projections Collaborative Group, a number of options for using Census data in projecting household representative rates including the use of Census data from 1991, 2001 and 2011. We will also consult with the devolved administrations on how household projections are produced for Wales, Scotland and Northern Ireland and consider whether we should adopt any aspects of their methodologies, recognising the advantages of consistent approaches across the countries of the UK.

We will investigate whether the preferred model allows a one-stage production approach, recognising that such an approach could improve the transparency of the projections.

Household projections and your needs

Users were asked for their views on the household projection outputs that we propose to publish. Their views included:

·  A number of respondents (8) asked for a methodology document with sufficient detail to enable users to replicate the household projections. They asked that this document is then updated to reflect any subsequent changes to the methodology.

·  A number of respondents (9) said that they found the Quality and Methodology Information (QMI) document and the statistical release useful.

·  A number of respondents (10) said that they wanted to continue to access the detailed unrounded data previously published by DCLG for use in their own analysis. Some users need this data for their own models using Chelmer or Popgroup. A number of users asked for aggregated totals by region.