Annex Report 2 - Technical

advertisement
Sheffield Strategic
Housing Market Assessment
November 2013
Annex Report 2: Technical Annex
Contents
Contents ............................................................................................................ i
1
Introduction ............................................................................................ 1
1.1
Structure of this annex ............................................................................................. 1
2
Survey deployment and response............................................................ 1
2.1
2.2
2.3
2.4
Introduction ............................................................................................................... 1
Sample frame ............................................................................................................. 1
Sampling ..................................................................................................................... 1
Survey deployment .................................................................................................... 4
2.4.1
2.4.2
2.4.3
2.4.4
2.4.5
Wave A
Wave B
Wave C
Wave D
Response rate
4
4
5
5
6
2.5
2.6
Sample Error .............................................................................................................. 6
Weighting and Grossing ........................................................................................... 6
3
Estimate of housing need ....................................................................... 9
3.1
Estimate of the backlog of housing need ............................................................ 10
3.1.1
3.1.2
3.1.3
3.1.4
3.1.5
3.2
Households in unsuitable housing
Sub Total: All households in unsuitable housing and need to move
Percentage unable to afford to move
Homeless households
Annual total backlog of existing housing need
Estimate of newly arising housing need............................................................... 16
3.2.1 New household formation
3.2.2 Percentage unable to buy or rent in the market
3.2.3 Existing households falling into priority need
3.3
16
18
19
Estimate of the supply of affordable housing ..................................................... 20
3.3.1
3.3.2
3.3.3
3.3.4
3.4
10
11
13
14
15
Supply of social relets
Supply of Shared Ownership re-sales
Units taken out of supply of affordable housing
Committed units of new affordable supply
20
22
22
23
Overall annual shortfall calculation ...................................................................... 23
References ...................................................................................................... 24
i
Sheffield SHMA Annex Report 2: Technical Annex
ii
1
Introduction
This technical annex is intended to act as a reference document for the Sheffield
Housing Survey – a major household survey – and the Housing Needs Model
produced by the partnership for Sheffield’s 2013 Strategic Housing Market
Assessment (SHMA). The results of the survey are presented throughout the
SHMA main report and are used to inform the Housing Needs Model. The main
outputs of the Housing Needs Model are summarised in Chapter 6 of the SHMA
main report.
A multi-method approach was used to provide a robust assessment, in line with
Strategic Housing Market Assessments Practice Guidance. 1 The study employs a
blend of secondary data, primary survey data, and qualitative insights. Secondary
data sources, such as those drawn from the Census and other official data sources,
are used where appropriate – as encouraged in the practice guidance – although in
many places this is augmented with analysis of the 2013 Sheffield Housing Survey.
Appendix 1 of the main report provides a copy of the print questionnaire for this
survey. A separate annex to the main report, Annex Report 1: Home Truths II,
contains an analysis of a programme of qualitative work, which is used to highlight
Sheffield resident’s perceptions of housing need, the housing market, quality of
housing and neighbourhoods. The findings of Home Truths II triangulates the
evidence from the housing survey and confirms the robustness of key quantitative
findings from the survey.
1.1
STRUCTURE OF THIS ANNEX
This annex is structured into two main chapters (not including this introduction).
Chapter 2 provides technical details of the design and deployment of the Sheffield
Housing Survey, including details of the sampling and weighting techniques used.
Chapter 3 provides further technical information on the derivation of key inputs to
the housing needs model presented in the main SHMA report (Chapter 6).
1
DCLG (2007).
1
2
Survey deployment and response
2.1
INTRODUCTION
This section of the Annex provides key details of the selection of the sample for the
survey, the deployment of the survey instrument (postal and online questionnaires),
and the response rates to several waves of the survey.
A copy of the questionnaire is provided in the main report (Appendix 1).
2.2
SAMPLE FRAME
The sample frame for the Sheffield Housing Survey was all households within the
city of Sheffield, excluding those living in the Peak District National Park (although
note that we have defined a Peak District housing market area [HMA] comprising
neighbourhoods that are outside the park’s legal boundary but within Sheffield. See
Chapter 2 of the main report for further details.)
The sample frame was drawn from an extract of the Council Tax database drawn
on 31 January 2013. The extract included properties for which there was an
exemption (e.g. student properties) but excluded those properties registered as
empty. The full file contained 233,991 records.
These were geocoded against the August 2012 version of the Office for National
Statistics Postcode Directory (ONSPD) using the full postcode from the Council
Tax data. 233,956 records (99.985%) were successfully matched and assigned a
map location to the nearest 100m. Of the 35 records not matched the majority were
clearly erroneous, relating to deleted properties that had not been removed from
the database and which had null postcodes.
The geocoded records were then assigned one of Sheffield City Council’s 100
defined neighbourhoods on the basis of point-in-polygon membership.
Neighbourhood labels were attached successfully to 233,757 of 233,956 (99.915%)
geocoded addresses (199 addresses have no neighbourhood, because their postcode
centroid positioned them outside the city boundary).
2.3
SAMPLING
A random sample of approximately 18,000 addresses was drawn from the sample
frame using the random selection algorithm in SPSS version 19, stratified by
neighbourhood using the syntax shown in Box 2.1.
1
Sheffield SHMA Annex Report 2: Technical Annex
Box 2.1. Stratified random sampling syntax.
COMPUTE ran1 = uniform(1).
RANK
VARIABLES=ran1 (A) BY r_neigh
/PERCENT into neighpct
/PRINT=NO .
SELECT IF (neighpct <= 7.7003).
EXECUTE .
The total selection made using this technique resulted in a sample of 17,951 cases
(7.7003% of the cleaned sample frame, or 1 in 13 addresses).
The household survey was completed by a total of 3,363 households from all parts
of the city. The survey was deployed in four waves. The first, wave A, was a
random sample of addresses from across the city. Subsequent waves were used to
boost underrepresented areas of the city or groups of the population.
The sample was sorted by the Council Tax reference. Each record was then
assigned a unique survey code which comprised of a two digit neighbourhood
identifier, a five digit serial number, and a one digit checksum. The serial number
was designed to reduce the likelihood of consecutive numbers appearing (which
might be easily guessable) and to permit easy identification by the data entry team
of the responding neighbourhood.
Table 2.1 shows the final profile of the sample by neighbourhood.
Finally, six duplicate or erroneous records were discarded at the mailing stage
because they were found to have unparseable addresses despite having valid
postcodes. The total number of mailings for the main survey (‘wave A’) was 17,944.
2
Survey deployment and response
47
152
50
260
145
162
155
274
162
96
54
369
151
442
136
329
35
203
254
195
287
146
162
200
169
63
33
100
194
186
207
104
182
114
189
n
.3
.8
.3
1.4
.8
.9
.9
1.5
.9
.5
.3
2.1
.8
2.5
.8
1.8
.2
1.1
1.4
1.1
1.6
.8
.9
1.1
.9
.4
.2
.6
1.1
1.0
1.2
.6
1.0
.6
1.1
%
4.7
5.5
5.8
7.3
8.1
9.0
9.8
11.4
12.3
12.8
13.1
15.2
16.0
18.5
19.2
21.1
21.3
22.4
23.8
24.9
26.5
27.3
28.2
29.3
30.3
30.6
30.8
31.3
32.4
33.5
Cum.
%
1.2
1.7
2.7
3.4
4.4
Table 2.1. Neighbourhood profile of random
sample stratified by neighbourhood.
Neighbourhood
Abbeyfield
Acres Hill
Arbourthorne
Base Green
Batemoor /
Jordanthorpe
Beauchief
Beighton
Bents Green
Birley
Bradway
Brightside
Brincliffe
Broomhall
Broomhill
Burncross
Burngreave
Chapeltown
Charnock
City Centre
Colley
Crookes
Crookesmoor
Crosspool
Darnall
Deepcar
Dore
Ecclesall
Ecclesfield
Endcliffe
Fir Vale
Firshill
Firth Park
Flower
Fox Hill
Fulwood
Neighbourhood
Gleadless
Gleadless Valley
Granville
Greenhill
Grenoside
Greystones
Hackenthorpe
Halfway
Handsworth
Heeley
Hemsworth
High Green
Highfield
Hillsborough
Hollins End
Housteads
Langsett
Lodge Moor
Longley
Lowedges
Loxley
Manor
Meersbrook
Middlewood
Millhouses
Mosborough
Nether Edge
Netherthorpe
New Parson Cross
Norfolk Park
Norton
Old Parson Cross
Oughtibridge
Owlthorpe
Park Hill
Ranmoor
Richmond
Rural Area
Sharrow
Shirecliffe
Shiregreen
Sothall
218
228
100
266
150
233
207
175
379
269
102
305
218
263
204
97
126
134
225
173
50
322
297
204
151
166
391
144
171
250
154
372
66
102
64
160
166
98
251
100
336
166
n
1.2
1.3
.6
1.5
.8
1.3
1.2
1.0
2.1
1.5
.6
1.7
1.2
1.5
1.1
.5
.7
.7
1.3
1.0
.3
1.8
1.7
1.1
.8
.9
2.2
.8
1.0
1.4
.9
2.1
.4
.6
.4
.9
.9
.5
1.4
.6
1.9
.9
%
Cum.
%
34.7
35.9
36.5
38.0
38.8
40.1
41.3
42.2
44.4
45.9
46.4
48.1
49.3
50.8
51.9
52.5
53.2
53.9
55.2
56.1
56.4
58.2
59.9
61.0
61.9
62.8
65.0
65.8
66.7
68.1
69.0
71.0
71.4
72.0
72.3
73.2
74.1
74.7
76.1
76.6
78.5
79.4
Neighbourhood
Southey Green
Stannington
Stocksbridge
Stubbin / Brushes
Tinsley
Totley
Upperthorpe
Wadsley
Walkley
Walkley Bank
Waterthorpe
Westfield
Wharncliffe Side
Whirlow / Abbeydale
Wincobank
Wisewood
Woodhouse
Woodland View
Woodseats
Woodside
Woodthorpe
Worrall
Wybourn
Total
n
%
159
.9
151
.8
259
1.4
162
.9
118
.7
115
.6
106
.6
145
.8
232
1.3
104
.6
202
1.1
116
.6
45
.3
49
.3
148
.8
161
.9
336
1.9
229
1.3
269
1.5
170
.9
151
.8
50
.3
214
1.2
17951 100.0
Cum.
%
80.3
81.2
82.6
83.5
84.2
84.8
85.4
86.2
87.5
88.1
89.2
89.9
90.1
90.4
91.2
92.1
94.0
95.2
96.7
97.7
98.5
98.8
100.0
3
Sheffield SHMA Annex Report 2: Technical Annex
2.4
SURVEY DEPLOYMENT
The survey was deployed in four waves, A-D. Table 2.2 summarises the survey
waves and responses.
Table 2.1 Survey distribution.
Survey wave
A
B
(original (community
mailout) email lists)
Number of survey
mailings
17,944
Response
2,241
Response rate %
12%
Source: UoS and Sheffield City Council
2.4.1
3,315
312
9%
C
(booster
mailout)
1,500
134
9%
D
Total
(students) (excluding
D)
Approx.
30,000
22,759
676
2,687
n/a
12%
Total
Over
52,000
3,363
n/a
Wave A
Wave A was the main, stratified random sample mail out with a closing date of 27
March 2013. Responses were incentivised with a £100 High Street voucher, which
was sent to a random selected respondent who had given their contact details for
the purpose.
Each respondent could elect to return the completed printed survey in a supplied
‘response paid’ envelope, or they could choose to complete an online version of the
survey. The online version was only available to those invited to participate and
users had to enter their unique survey serial number to complete the online survey.
The online survey was available at http://sheffieldhousingsurvey.org.uk/.
The majority of respondents elected to return the postal version of the
questionnaire, although 151 respondents in Wave A completed the online version.
A helpline (telephone and email) was available to help respondents complete the
survey. Approximately 60 individuals contacted the helpline.
A check was made of the final data file to ensure that multiple entries with duplicate
serial numbers were not received, although none were found.
2.4.2
Wave B
Following monitoring of Wave A’s response rate and in order to boost response
from some underrepresented groups, especially city centre residents, an online-only
version of the survey was distributed to members of selected community email lists
maintained by Sheffield City Council for the purposes of community consultation.
These lists had 3,315 members and attracted 312 responses (9% response rate), all
of them online except in a small number of cases where a paper copy was
4
Survey deployment and response
specifically requested. Respondents of Wave B were also eligible for the incentive
draw and the helpline was also available to them.
2.4.3
Wave C
Again following monitoring of Wave B’s response rate and in order to boost
response from selected Housing Market Areas (see Chapter 2 of the main report), a
booster sample of printed questionnaires was sent to an additional random sample
of addresses from seven selected HMAs:







Chapeltown/Ecclesfield
City Centre
East
North West
Rural Upper Don Valley
Manor/Arbourthorne/Gleadless
Stocksbridge and Deepcar
The random sample was again stratified by neighbourhood using a variant of the
syntax in Box 2.1. 1,500 additional addresses were sampled, as shown in Table 2.2.
Table 2.2. Wave C sample by HMA.
HMA
Chapeltown/Ecclesfield
City Centre
East
North West
Rural Upper Don Valley
Manor/Arbourthorne/Gleadless
Stocksbridge and Deepcar
Total
Sample
237
148
270
326
50
355
110
1,500
This booster sample attracted 134 responses (9% response rate). 125 were postal
responses and 9 were made online. Respondents of Wave C were also eligible for
the incentive draw and the helpline was also available to them.
2.4.4
Wave D
Waves A-C under-represented full time students (because the Council Tax database
did not generally include first year residents of University-owned and managed
purpose-built student accommodation). For this reason, a final survey wave was
deployed using direct email to students of the University of Sheffield, and using
social media and a link from Sheffield Hallam University student website portal.
The link was to a special, expanded online version of the survey aimed at students.
It included several additional questions aimed at eliciting basic details on term and
vacation time addresses, and student status.
5
Sheffield SHMA Annex Report 2: Technical Annex
Owing to differences in the two University’s email list management and policies it
was not possible to adopt the same approach. Consequently, it is not possible to
ascertain the extent to which the link was propagated, and the final response rate is
not known although the method yielded an additional 2,687 responses from
students, all of them online.
2.4.5
Response rate
The overall response rate (excluding wave D) was 11.8%. Wave D was excluded
from this calculation because the sampling method used was purposive and it was
not possible to know how far the survey link had been propagated. The additional
2,687 responses from students were used to weight the final sample data file so that
it better reflected the household structure of the city.
2.5
SAMPLE ERROR
It was calculated that using standard assumptions about variability of 50% of any
variable, that the confidence limit was +/- 2.21% at the 99% level, and +/- 1.68%
at the 95% level. This means that in general we expect the ‘true’ answer to lie within
a range of 2.21% of the given answer 99% of the time, and within 1.68% of the
given answer 95% of the time. These calculations were based on the formula given
in Dixon and Leach (1977), applying a population of 223,906 (the number of
occupied household spaces in Sheffield according to the 2011 Census) and a sample
size of 3,363 (the total number of responses).
2.6
WEIGHTING AND GROSSING
The survey data file was weighted using post-stratification to better reflect the
underlying tenure composition of the population of households and their
geographic distribution. This was done by producing a cross tabulation of tenure by
HMA from both the survey and 2011 census (at the output area level) and
producing weights that compensated for the cell-wise differences between the two
tables. These weights were then applied to the survey data file in SPSS. Several key
variables were then tested against the 2011 Census to validate the weights. The
impact of the final weighing is shown in in Table 2.3 for the tenure profile of the
final weighted dataset.
The weights were also adjusted so that all analysis was grossed up to the number of
households in Sheffield.
6
Survey deployment and response
Table 2.3 Outcome of post-stratification household weighting by tenure.
Tenure
Owned: Owned with a Mortgage or Loan
Owned: Owned Outright
Shared Ownership: Part Owned and Part Rented
Social Rented: Rented from Council
Social Rented: Other
Private Rented: Private Landlord or Letting Agency
Private Rented: Other
Living Rent Free
Source: Sheffield Housing Survey, 2013 and Census, 2011
Census
proportion
%
31.23
27.11
0.38
17.71
7.04
14.18
1.37
0.98
Proportion of
weighted data
file %
30.90
31.70
0.30
15.90
5.90
13.20
0.80
0.20
7
3
Estimate of housing need
The Housing Needs Model was prepared in accordance with the DCLG guidance
to assess the number of dwellings needed per annum to enable households who
cannot access market housing (at reasonable expense) to be housed.
“[H]ousing need is defined as ‘the quantity of housing required for households who
are unable to access suitable housing without financial assistance’. For the
purposes of assessment, this means partnerships need to estimate
the number of households who lack their own housing or live in
unsuitable housing and who cannot afford to meet their housing
needs in the market.” (DCLG, 2007, P.41, emphasis in original)
The evidence in the SHMA is intended to produce a single estimate of the number
of dwellings required to meet housing need. The numerical estimates are a key
component of the overall report and DCLG sets out guidelines for the process for
combining evidence for the Housing Need Model, which is outlined in Chapter 6 of
the SHMA main report.
The DCLG guidance sets out five stages to estimate the level of housing need. The
stages are:





Current housing need,
Future housing need,
Affordable housing supply,
Housing requirements of households in need, and
Bringing the evidence together
The analyses of these stages are undertaken in Chapters 5 and 6 of the main report.
The needs model relates directly to these stages, but uses a four-stage process. This
technical appendix draws out the calculations used to provide evidence for each of
the values in the need model, and each sub-stage is indicated within the four overall
stages:




Estimate of the backlog of housing need (sub-stages 1-5)
Estimate of newly arising need (6-8)
Estimate of the supply of affordable housing (9-11)
Bringing the evidence together in the overall annual shortfall (12)
These stages and sub-stages correspond to sections and lines of the model summary,
which is presented as Table 6.1 of the main report. The remainder of the technical
appendix is structured according to the four stages.
9
Sheffield SHMA Annex Report 2: Technical Annex
3.1
ESTIMATE OF THE BACKLOG OF HOUSING NEED
DCLG guidance recommends using survey-based data for addressing the backlog
of housing need as secondary data is unlikely to adequately address the issues
related to housing need.
“Partnerships need a good understanding of the scale of current housing need,
including any backlog, at the local authority level. However, it can be difficult to
obtain a complete and robust estimate of backlog due to data limitations.
Traditionally, local surveys have been used although these can be costly to administer
and difficult to interpret.” (DCLG, 2007, P.43)
The recent census of population (2011 Census) provides extensive background
information about the population in Sheffield, and some relevant information about
housing within the market area. The limited time lag between 2011 and 2013
enables greater use of the census as a credible source of information than many
other SHMAs have been able to. The census is used where appropriate, and has
provided a source for triangulation with other information sources, but the census
does not adequately address all of the issues required in a calculation of housing
need, aspirations and preferences. Therefore, for the calculation of the backlog of
housing need, the Sheffield Housing Survey data is used as the primary source to
provide the robust and consistent approach required.
The total annual number of dwellings to reduce the backlog of housing need
calculation is divided into five lines in the housing needs model. The first line
calculates the overall number of households in unsuitable housing. The second line
removes households whose housing need could be met without the need to create
additional affordable housing opportunities, whether that is by remaining in situ,
movements within the social rental stock, or from out-migration from the overall
total. The third line deflates the sub total to the number of households that are
financially unable to meet their housing need through the market (whether rental or
owner occupied). The fourth line estimates the number of existing homeless
households requiring housing. The fifth line distributes the total backlog of need
over five years to provide the total annual need to reduce the backlog of housing
need. Each line is now described further in turn.
3.1.1
Households in unsuitable housing
The number of households in unsuitable housing is calculated using the data from
questions A7 and A8 in the survey (the full survey can be found in the appendix of
the main SHMA report).
Households in unsuitable housing include households in all tenures where the
current situation of the household is inadequate. The reasons for inadequacy were
surveyed in question A8. The options include aspects related to the household (e.g.
financial) and the physical elements of the dwelling they reside in. In addition to the
responses provided by the household about question A8, other questions in the
survey enable a calculation of technical overcrowding (e.g. A4 and A9). This
possibility enables a refined approach to analyse the overall number of households
in unsuitable housing, which takes account of overcrowding. This approach divides
the total population of households in unsuitable housing into three categories:
10
Estimate of housing need
 Category 1: households that identify themselves as in inadequate housing for
any reason other than the dwelling being too small (e.g. access, maintenance,
disrepair, unhygienic conditions, expense, insecurity, harassment, inadequate
services, and so on);
 Category 2: households that identify themselves as in inadequate housing
because the dwelling is too small, and these households are technically
overcrowded according to our own assessment (see below); and
 Category 3: households that do not identify themselves as in inadequate
housing, but are technically overcrowded.
The computation for technical overcrowding uses the Sheffield Allocations Policy’s
bedroom standard and applies this to the survey responses for household size and
number of bedrooms. 2
The results are outlined in Table 3.1 below. Combining the three categories, 27,857
households in Sheffield are in inadequate housing.
Table 3.1. Households in unsuitable housing (A7, A8, A4, A9)
Household in unsuitable housing
Households in unsuitable housing (sum cat 1, 2, 3)
cat 1: hhld says unsuitable (any reason other than too small only)
cat 2: hhld says unsuitable (too small only) and is technically overcrowded
cat 3: hhld says adequate, but technically overcrowded
Source: Sheffield Housing Survey, 2013
No.
27857
23789
1324
2744
% of all
households
12.1%
10.3%
0.6%
1.2%
Whilst slightly different to the steps implied by the DCLG guidance, this
explanation of the current level of housing need provides a logical and transparent
framework, and avoids the challenge of double-counting as outlined in the
guidance.
3.1.2
Sub Total: All households in unsuitable housing and need to move
The gross figure of existing households in unsuitable housing includes many
households who do not require new affordable housing. Households may be
included in line 1 but do not count towards the overall number in housing need for
an array of reasons. Having the financial ability to afford market housing (whether
private rental or owner occupation) means that a household does not need
affordable housing to meet their household’s need. This is a significant element and
is applied in the following line (3). Prior to removing households who can afford
2
This assessment applied the bedroom standard implied by the Sheffield City Council allocations
policy to the household composition of survey respondents. For example, non-related adults
require a separate room, as do children of the opposite sex. See page 11 of the Sheffield
Allocations Policy issue 6.
11
Sheffield SHMA Annex Report 2: Technical Annex
market housing a number of categories need to be removed from the overall
number in unsuitable housing. These categories are:
 households where technical overcrowding can be removed by a concealed
household moving out;
 households who currently reside in the social rented stock and any movement
would have a zero net impact upon the level of housing need (a movement
would require, but at the same time also release an existing affordable home);
and
 households whose housing need is met by moving out of Sheffield (i.e., outmigration).
The combination of these categories is removed from the total number of
households in unsuitable housing to provide the sub total of all households in
unsuitable housing and who need to move.
The first calculation is the number of households in each category that can be
removed by resolving the inadequacy ‘in situ’. No households from category 1 are
removed as the issues relate to problems that cannot be resolved through a
concealed housing moving out, e.g. suffering harassment from a landlord. For
categories 2 and 3 some households’ problem with housing can be resolved through
a concealed household moving out, and these are removed using answers in Section
E of the survey. Whilst these ‘concealed households’ may contribute to the overall
level of need, they are counted instead in line 6 of the needs model and are
removed from this section to avoid double counting.
The removal of households in the social rented sector (SRS) requires a note. The
assumption that household movements within the SRS stock should be removed
from the overall level of households in unsuitable housing is based on a matching
assumption, whereby the existing stock is assumed to be able to adequately support
a redistribution of households on a one-for-one basis. Line 2 therefore includes an
adjustment to the unsuitable housing resolved within SRS stock to take account for the
reality of an imperfect distribution of available property sizes to cater for the
household sizes of those requiring housing. This was achieved by comparing the
size (bedroom) profile of houses being released by vacating tenants with those
required by households. This was subject to a matching constraint: households
needing a smaller property could in theory be allocated a larger property (although
we do recognise that is rarely done in practice, and the removal of the spare room
subsidy further diminishes this likelihood), while those needing a large property
would not be housed in a smaller property. Hence, there was a matching deficit of
large family homes and we found that a 1-for-1 match was not possible, but that
there would be an inefficiency in matching that accounts for approximately 6% of
the stock, and can be seen in Table 3.2. The percentage is derived from the
mismatch between the stock of households who will be moving out of their current
SRS property and their preferences for larger properties (we have used 3 bedrooms
and larger) based upon the ability of households to be allocated larger homes than
their needs, but not smaller.
12
Estimate of housing need
Table 3.2 Mismatch between size of properties and housing preferences
Implied no. of
No. of households
properties
with needs for
Cumulative mismatch (preferences to
Bedrooms
released
property of this size
size, including larger property sizes)
5+
105
15
105 – 15 = 90
4
214
867
214 – 867 + (90) = –563
3
2941
3363
2941 – 3363 + (–563) = –985
2
4468
4622
4468 – 4622 + (–985) = –1139
1
2786
1646
2786 – 1646 + (–1139) = 1
Source: Sheffield Housing Survey. Note that the cumulative mismatch between implied supply and
demand mismatch for properties of 3 bedrooms and greater is 985. This represents approximately
6% of the 15,033 households expected to resolve their housing needs within the SRS stock.
Households whose housing needs will be met outside of Sheffield, through outmigration, are discounted from the total. This can be calculated in a number of
ways including using historic migration trends, migration forecasts and survey
estimates. This model uses projections based on the survey, including answers to
Section D of the household survey about households’ future housing intentions and
the locations they expect to move to. This accounts for 484 households (see Table
3.3).
Table 3.3 Households in unsuitable housing where need is met ‘in situ’
MINUS 1. of which overcrowded is 'resolved in situ'
cat 1: hhld says unsuitable (any reason other than too small only*)
cat 2: hhld says unsuitable (too small only*) and technically overcrowded
MINUS - unsuitable housing resolved within SRS stock
Adjustment to account for bed size mismatch in SRS turnover
Subtotal MINUS - unsuitable housing resolved within SRS stock
MINUS - unsuitable housing resolved by out migration
Source: Authors’ calculations based on Sheffield Housing Survey.
3.1.3
Num.
-658
-1254
-(15033)
6%
-14131
-484
% of all
households
-0.3%
-0.5%
-6.5%
-6.1%
-0.2%
Percentage unable to afford to move
Of the households that are in unsuitable housing and need to move, some can
afford to meet their need within the market. The percentage unable to afford to
rent or buy only is included.
The percentage applied is the proportion of households indicated in the survey who
have been wanting to move but not able to do so (question D3) and could not
move for financial reasons (question D4, response categories a, b, c, g, h, i and j).
The resulting proportion of 67.1% of households is further validated by
comparison with the 67% of households found in Sheffield City Council’s own
13
Sheffield SHMA Annex Report 2: Technical Annex
model of the affordability of one bedroomed properties to newly-forming
households.
Table 3.4 Reasons given why household is unable to move (but would like to)
Reason
a
Cannot afford the monthly cost of a mortgage
b
Cannot afford the deposit on a house
c
Cannot afford moving costs
d
Local education choices
e
Family reasons
f
Location of employment
g
Lack of affordable rented housing
h
Rent/mortgage arrears
i
Unable to sell
j
Negative equity
k
Need advice/support to move
Percentage of households who would like to move, but
identify as unable to move for financial reasons
(respondents choosing any one or more of a,b,c,g,h,i,j)
Source: Sheffield Housing Survey.
3.1.4
Percentage of Responses
25.4%
43.3%
29.2%
4.6%
14.6%
5.4%
39.8%
9.9%
11.2%
7.0%
16.3%
67.1%
Homeless households
Line 4 of the housing need model includes households who require housing
because of homelessness. The backlog of homeless households is taken from the
number of eligible, unintentionally homeless and in priority need individuals (290)
in 2013, as accepted by Sheffield City Council. This is derived from the P1E returns
made by SCC (see Table 3.5). The P1E form is part of English local housing
authorities’ statutory obligation under homelessness legislation and provides an
accurate and up to date summary of the authority’s records of homeless
presentations.
14
Estimate of housing need
Table 3.5 Homeless numbers, 2013
Other
Ethnicity
Not
Stated
All
Ethnicity
Groups
17
15
20
290
2
1
3
3
35
49
23
16
30
21
338
95
17
7
8
14
10
151
4
3
0
0
5
5
17
505
101
57
42
67
59
831
White
Black
Asian
183
30
25
24
2
199
Eligible, but not homeless
Ineligible
Total decisions (sum of rows 1 to 5
above)
Source: P1E, January-March 2013
Eligible, unintentionally homeless
and in priority need
Eligible, homeless and in priority
need, but intentionally so
Eligible, homeless but not in
priority need
3.1.5
Mixed
Annual total backlog of existing housing need
Standard SHMA practice reduces the existing backlog of need within the area to an
annualised target spread over 5 years, hence the reduction to 20% of overall
backlog. The precise time period to meet the target can be altered if the Local
Authority sets different time period to meet its objectives as per the SHMA
guideline set out by DCLG, which is quoted extensively to provide the evidence for
a 5-year period.
“The quota should be based upon meeting need over a period of five years, although
longer timescales can be used. In particular, there may be merit in linking quotas to
the remaining time period of adopted housing policies in plans. For the component of
need derived from existing affordable housing tenants, partnerships could estimate the
proportion that are expected to be re- housed based on previous allocations. Whilst
the decision is the responsibility of individual local authorities, partnerships should
bear in mind the need for comparability. Partnerships should avoid using a period of
less than five years in which to meet unmet current need.
If a five-year period is used, this means that 20 per cent of current unmet need should
be addressed each year. The output of this should be an annual quota of households
who should have their needs addressed.” (DCLG, 2007, P.52)
For illustrative purposes only the Total Backlog of Need is distributed over a range
of time periods, from 3 years to 7 years. As indicated above, the normal 5-year
distribution is used in the Housing Needs Model (see Table 3.6).
15
Sheffield SHMA Annex Report 2: Technical Annex
Table 3.6 The total annual backlog through
scenarios of 3-7 year progressive targets for
reduction.
TOTAL BACKLOG
7892
Annualised over 3 years
2631
Annualised over 4 years
1973
Annualised over 5 years
1578
Annualised over 6 years
1315
Annualised over 7 years
1127
Source: Composite of Backlog of Housing
Need using Sheffield Housing Survey and
P1E sources
3.2
ESTIMATE OF NEWLY ARISING HOUSING NEED
An estimate of newly arising housing need is made in addition to the estimate of the
backlog of housing need. DCLG guidelines (2007) indicate that there are numerous
possibilities for forecasting newly arising need and new household formation. We
have used the Sheffield Housing Survey as the primary source of information for
this stage, but we also draw upon Sheffield City Council’s records, for example on
estimated average rents and homeless presentations.
3.2.1
New household formation
The SHMA outlines two possible ways of calculating new household formation,
and is reprinted here for clarity:
“It is first possible to use population and household projections to predict levels of new
household formation (the net change between periods). This technique, however, is
prone to the 'circularity' problem in that projections are essentially based on historical
trends that are themselves a result of past market conditions and constraints. They
reflect effective rather than notional demand. For this reason, a second method is
often employed: the use of survey data on moving needs and expectations. In the
household survey we asked respondents about the expected movement of their
household and any concealed households. One major problem with using survey data
in this way is that it has been shown in various studies that households tend to
systematically over-predict their likelihood to move or to form (Watkins et al.,
2012). Studies in a range of countries including the UK, the US and the
Netherlands all support the need to adopt a deflator to household survey predictions
of mobility. Once we have adjusted for over-prediction, we estimate that new
household formation will lead to 2,269 new households in the city per year over the
next 5 years (line 6 in Table 7.1). Once adjusted in this way, the figure is broadly
compatible with that suggested by the ONS subnational population projections for
households over the period 2013-2018.” (SHMA main report, p. 113.)
16
Estimate of housing need
Whilst secondary data can be used to project household formation the guidelines
recognise that there are problems with essential data at the local level. Our
approach has been to start with estimates of the level of new household formation
that are implied by several household projections including the SNPP-based
projections (see main report) but also other projections that we have reviewed. We
then test these against different methods of deriving new household estimates from
the household survey.
First, we looked at new household formation as implied from simply looking at the
number of households likely to move to a different home. Underpinning this is a
basic assumption that the majority of these households would prefer a home in the
city (i.e. undertaking an adjustment move to cater for changed circumstances or
needs), but that the houses they release might be taken up disproportionately by
those moving into the city (given increases in population and household
projections).
A basic analysis of the survey shows an annualised household formation level of
13,351 (see Table 3.7a). As noted earlier, we know that survey respondents tend to
overestimate the actual household formation rate significantly, and so a reduction to
17% of this value is applied in line with previous research (Watkins, 2012). This
gives a figure of 2269. The level of new household formation implied by this
calculation is consistent with the household projections we looked at (chapter 5 of
the main report).
Table 3.7a D2 Do you think you are likely to move to a different home in the future?
Answer
Yes, as soon as possible (e.g. 1 month)
Yes, within a year
Yes, in 1 to 2 years
Yes, in 3 to 5 years
No, don't want to
No, but would like to
Don't know
Total
Yes Total
Yes Total annualised (at 5yrs)
Deflated to 17% of projection
Source: Sheffield Housing Survey, 2013
Percentage
3.0%
7.6%
7.0%
11.3%
32.5%
7.1%
31.4%
100.0%
Implied No.
6976
17618
16187
25975
74767
16418
72431
230372
29.0%
n/a
n/a
66756
13351
2269
An alternative approach is to estimate new household formation as arising from
concealed households that might form (drawing on questions E1 and E2 from the
survey). These questions ask whether there are current household members (and if
so, who) who are looking to form their own household in the next three years.
Information is sought on up to two such household members (and so calculations
are likely to be an underestimate of the true number). Table 3.7b summarises the
survey response to these questions.
17
Sheffield SHMA Annex Report 2: Technical Annex
Table 3.7b Questions E1/E2: Are any existing members of your household
looking to move into their own accommodation in the next three years? [If so],
who is looking or likely to look for accommodation in the next three years?
Answer
Person 1
Parent / Grandparent
127
Child aged 16 or over
17502
Partner/Spouse
887
Lodger
659
Friend
895
Other
1075
Total
21145
Total (person 1 and 2)
Annualised over 3 years
Of which 17% (to adjust for unrealised expectations)
Source: Sheffield Housing Survey, and authors’ calculations.
Person 2
136
6157
207
62
1396
200
8158
29303
9768
1660
This method results in an estimate of around 1,660 households per annum,
although it must be noted that this is likely to be an under estimate for the reasons
given above.
On balance, given that we find that the various methods result in broadly consistent
estimates of new household formation we adopt the figure of 2,269 as suggested by
both the survey and household projections.
3.2.2
Percentage unable to buy or rent in the market
Of the total new household formation, some households will be able to meet their
housing requirements within the market at an affordable rate for that household.
The percentage of newly forming households unable to afford to buy or rent in the
market is therefore applied to the overall newly arising need. We calculate the
percentage using the projected incomes of newly forming households in question
E12 of the survey and applying this to average rental levels and lower quartile house
prices provided by Sheffield City Council (using their mix-adjusted method for
house prices).
In line with national research (e.g. the English Housing Survey) the percentage of
newly forming households unable to afford to rent or buy is higher than the
percentage of existing households unable to afford. Comparing 80% market rent
for a one or two bedroom property in the city (as the lowest cost option in
comparison to buying) at 30% of household income, 75% of households could not
afford to rent or buy within the market (Table 3.8).
18
Estimate of housing need
Table 3.8 Average market rent, affordability assumption and percentage of newly forming
households unable to afford market provision
80% of Average Market Rent (monthly) for a two bedroom dwelling
80% of Average Market Rent (annual) for a two bedroom dwelling
Assume housing costs of 30% income is affordable
% of households unable to afford £18k
Source: Sheffield City Council and Sheffield Housing Survey, 2013
£450
£5,400
£18,002
75%
The distribution of incomes of newly forming households is outlined in Table 3.9
below. The 75% unaffordability figure is based on the cumulative percentage of
households earning less than £14,999 (i.e. 65%) and an assumption of even
distribution of incomes in the £15,000-£19,9999 bracket deflated to 0.6 of the overall
percentage (i.e. 10%) to account for the approximate midpoint of that bracket.
Table 3.9 Question E12 Distribution of
income of newly forming household
Up to £4,999
26%
£5,000 - £9,999
16%
£10,000 - £14,999
23%
£15,000 - £19,999
17%
£20,000 - £24,999
14%
£25,000 - £29,999
2%
£30,000 - £34,999
2%
£35,000 - £39,999
0%
Total
100%
Source: Sheffield Housing Survey, 2013
3.2.3
Existing households falling into priority need
Existing households falling into priority need are added to the newly forming
households unable to buy or rent in Sheffield. These households are by definition
unable to find suitable market housing. The P1E returns provide annual data on
homeless presentations and acceptances, the average for 2008-12 is 3,127
presentations and 1,326 acceptances (see Figure 3.1).
19
Sheffield SHMA Annex Report 2: Technical Annex
Figure 3.1 Presentations and acceptances of homelessness
5000
4000
3000
2000
1000
0
2007
2008
2009
Presentations
2010
2011
2012
Acceptances
Source: P1E, 2013
3.3
ESTIMATE OF THE SUPPLY OF AFFORDABLE HOUSING
The total number of affordable dwellings added to the existing stock (where
positive) will alleviate some of the pressure on the total level of housing need. The
supply of affordable housing includes the supply of social relets, the number of
shared ownership re-sales and the number of new affordable homes built, minus
the number of affordable houses removed from the stock. The total affordable
supply (per annum) is estimated in stage 3 of the housing need model at lines 9, 10
and 11.
3.3.1
Supply of social relets
The supply of affordable housing in line 9 of the need model is derived from the
supply of social relets and the level of shared ownership re-sales.
DCLG practice guidance outlines that the estimate of supply of social relets can be
attained using historic data trends using secondary data, primarily HSSA 3 and
CORE4 data. While guidance suggests that it is normal to use a three-year average, it
also outlines possible deviations from this practice, for example because of a change
in the stock base predicted. Figure 2.2 reveals that the level of relets amongst the
Housing Association stock is steady over the period 2008-12, whilst the relet figure
fluctuates extensively within the Council stock over the same period (Figure 2.2.).
The variation in historic relets reveals the necessity for lengthy averages and the
likely volatility in relet levels in the future projections. A three-year average would
not include the significant decline in 2009, and would over inflate the number of
relets, and as such we use a five year average. Ongoing assessment of these levels is
necessary for understanding whether future supply is meeting projected future need.
3
4
20
Housing Strategy Statistical Appendix, now Local Authority Housing Statistics (LAHS).
Continuous Recording of Social Housing lettings and sales, collected by the Homes and
Communities Agency.
Estimate of housing need
The precise levels historically are broken down by category in the table 2.10. Using
the HSSA and CORE data the average for relets during 2008-12 is 3728.
Fig 2.2 Number of social relets per annum for Sheffield Homes and Housing Association stock 200812
6000
5000
4000
3000
2000
1000
0
2008
2009
AllCORE
2010
2011
Sheffield City Council stock
2012
Housing Association stock
Source: HSSA and CORE, 2013
Table 2.10 Level of social rental sector relets in Sheffield 2008-12
LA lets to tenants new to the SRS
General Needs lets to tenants from
Private Sector
Tied/rented with job
Private rented sector
Owner occupation
Mobile home/caravan
Living with friends/family
Council lets to tenants from Private
Sector
Tied/rented with job
Private rented sector
Owner occupation
Mobile home/caravan
Living with friends/family
Relets to tenants new to SRS
Source: HSSA and CORE, 2013
2008
2400
2009
2644
2010
2608
2011
2658
2012
2610
5 yr ave.
2584
522
2
110
45
1
364
715
4
136
54
2
519
532
4
152
67
1
308
657
5
173
64
0
415
344
0
109
24
0
211
554
3
136
51
1
363
571
4
41
45
1
480
3493
630
3
54
22
2
549
3989
655
1
70
45
0
539
3795
654
2
79
30
0
543
3969
442
1
49
23
0
369
3396
590
2
59
33
1
496
3728
21
Sheffield SHMA Annex Report 2: Technical Annex
3.3.2
Supply of Shared Ownership re-sales
The number of Shared Ownership properties resold per annum is calculated using a
regional average of resale for Yorkshire and Humber (2%) applied to the overall
number of Shared Ownership properties in the city. The average level of resale
fluctuates between 2.5% and 1.6% in Yorkshire & Humberside over the period
2004-11.
Table 2.11. Average resale percentage of Shared Ownership for Yorkshire and Humberside 2004-11
Yorkshire & Humberside 04/05 05/06 06/07 07/08
Percentage
1.7%
1.6%
2.5%
2.9%
Implied number
16
15
23
27
Source: CORE and RSR as quoted in Clarke and Heywood, 2012
08/09
1.9%
18
09/10
1.6%
15
10/11
2.0%
18
Ave
2.0%
18
The overall effect on the model is very small on account of the small size of the
Shared Ownership sector in the city. The survey reveals there are 921 properties in
the city currently in Shared Ownership. When these are multiplied by 2% we
estimate that 18 Shared Ownership properties will be available for resale per
annum. Whilst 18 is used in the housing needs model, Table 2.11 reveals that a
variation of less than ten properties per annum over the period 2004-11 and hence
at any annual level is unlikely to have a significant impact on the overall average
estimate of affordable housing shortfall.
3.3.3
Units taken out of supply of affordable housing
The next step is to subtract vacancies (if applicable) and units taken out of
management (including Right to Buy, RTB).
The total number of properties indicated by SCC that are likely to be demolished
between 2014 and 2018 (six years) was 317. The precise rate of demolition over
those years is likely to vary considerably, but computing an average annual rate of
demolition over these six years, which is 53 properties, is acceptable.
Forecasting the average number of RTB units removed from the social stock where
households’ properties are currently inadequate need is more complex. The
calculation utilised uses the average number of RTB transactions over the period
2010-13 (114 per annum) and then deflates it by the average relet rate of council
stock (10.8% for the period 2009-12). This is done to avoid overestimating the
impact of RTBs on households in need. It should be recalled that an RTB does not
generally remove an affordable housing unit (in the short term at least) but simply
transfers its tenure.
We estimate, then, that the average number of RTBs which might otherwise house
a household in need is 12 per annum. When combined with the forecast number of
demolitions per annum, 53, this gives a total average of 65 properties per annum
taken out of the supply affordable housing.
22
Estimate of housing need
3.3.4
Committed units of new affordable supply
The DCLG Practice guidance suggests that new affordable supply should be
estimated based on committed units of affordable supply. However, in reality this is
not certain and, in line with Sheffield’s previous SHMAs , we have taken an average
based on recent delivery. This produces an estimate of approximately 201 new
affordable dwellings based on the period 2008-13. We have rounded this to 200 in
the needs model for simplicity.
Table 2.12. The number of affordable units delivered each year, 2007-13
Affordable housing delivery
07/08
HCA Funded
294
Section 106
11
TOTAL
305
Source: Sheffield City Council, 2013
08/09
178
14
192
09/10
290
7
297
10/11
168
0
168
11/12
270
0
270
12/13
76
0
76
Ave. 08-13
196
4
201
This estimated figure is unlikely to match the actual figures for each year in the near
future, as the actual delivery of new affordable dwellings will fluctuate depending
upon wider economic circumstances, market conditions and wider development
processes (for example the outcome of renegotiations of S106 agreements). The
significant fluctuation between 2011/12 (270) and 2012/13 (76) highlights this
potential fluctuation in future delivery.
3.4
OVERALL ANNUAL SHORTFALL CALCULATION
The overall annual shortfall calculation draws together the summary of evidence
provided in lines 1-11 of the needs model to estimate the required number of
dwellings per annum to meet existing and projected future need taking into account
the projected level of affordable supply. This projection of overall annual shortfall
is determined by combining the total annual need to reduce the backlog (stage 1)
with the total newly arising need (stage 2), and the subtracting the total annual
supply of affordable housing (stage 3). This is summarised in the SHMA report
thus:
Requirementt+5 = (NeedBacklog,t + NeedArising,t:t+5) – Supplyt:t+5.
where t is the current time period (2013) and t+5 is the horizon period of the SHMA in years.
23
Sheffield SHMA Annex Report 2: Technical Annex
References
Clarke, A. and Heywood, A. (2012) Understanding the second-hand market for
shared ownership properties, Cambridge Centre for Housing and Planning
Research, University of Cambridge.
DCLG (2007) Strategic Housing Market Assessment practice guidance Version 2,
Department for Communities and Local Government.
24
Download