TESTING DWP’S ASSESSMENT OF THE IMPACT OF THE SOCIAL RENTED SECTOR SIZE CRITERION ON HOUSING BENEFIT COSTS AND OTHER FACTORS October 2013 Rebecca Tunstall, Centre for Housing Policy, University of York TESTING DWP’S ASSESSMENT OF THE IMPACT OF THE SOCIAL RENTED SECTOR SIZE CRITERION ON HOUSING BENEFIT COSTS AND OTHER FACTORS Contents 1. Summary 2. Introduction: The Social Rented Sector Size Criterion underoccupation penalty 3. DWP’s estimate of the savings from the policy 4. The model the DWP used to estimate the savings from the policy and the risk of higher or lower savings 5. The impact of applying real data in place of assumptions in the DWP model on estimated savings 6. Additional questions for the DWP model 7. Additional factors which affect the overall cost impact of the policy 8. Conclusion 2 1.Summary The Social Rented Sector Size Criterion or under occupation penalty (widely referred to as ‘the bedroom tax’ or ‘end of the spare bedroom subsidy’) reduces the Housing Benefit paid to tenants deemed to be underoccupying their homes. Its main aim is to reduce UK Housing Benefit expenditure. A DWP model predicted savings of £480m in 2013/14, reported in the final official impact assessment of June 2012 (DWP 2012). However, real data available from housing organisations since 1st April 2013 does not match key assumptions about claimant behaviour underlying the DWP’s model. Three of the DWP’s four key assumptions should be re-examined. If we use real data, and take into account regional variations in impact, the total savings the DWP’s model predicts reduces by £160m (33%). Real data also suggests more variation in potential outcomes, and reductions in savings of up to £186m (39%) appear possible. In addition, the DWP model does not contain all of the main factors likely to influence the level of Housing Benefit savings from the policy. Finally, whatever the impact on Housing Benefit cost, research by the National Housing Federation suggests local authorities and third sector organisations are incurring substantial costs from the policy which should be taken into account in an overall assessment. 3 2.Introduction: The Social Rented Sector Size Criterion underoccupation penalty The Social Rented Sector Size Criterion or under occupation penalty (widely referred to as ‘the bedroom tax’ or ‘end of the spare bedroom subsidy’) reduces the Housing Benefit paid to tenants deemed to be underoccupying their homes. It is applied to all social housing tenants in the UK of working age and who are claiming Housing Benefit. Those on full Housing Benefit underoccupying by one bedroom receive Housing benefit that covers 86% of the rent, rather than 100%, as before. Those underoccupying by two rooms receive Housing Benefit that covers 75% of the rent. The policy was introduced on 1st April 2013, when it applied to an estimated 660,000 households (DWP 2012). It has currently been active for almost six months. The underoccupation penalty has three aims: 1. to reduce Housing Benefit costs, 2. to increase mobility amongst social tenants, 3. To reduce underoccupation and overcrowding and to help house people on the waiting list (DWP 2012). The primary goal of the underoccupation penalty appears to be to reduce overall UK Housing Benefit costs (eg DWP 2012). Lord Freud described the cost argument, alongside the desire for parity with the private rented sector in matching Housing Benefit payments to household size, as the “core argumentation” for the policy (House of Lords Debates, 14th February 2012 c706). The aims of this report are to explore: 1) The extent to which key assumptions about claimant behaviour underlying the DWP’s final impact assessment of June 2012 are reflected in real data available from housing organisations since 1st April 2013. 2) The extent to which the DWP model contains all of the main factors likely to influence the level of Housing Benefit savings from the policy. It also discusses whether any other costs should be taken into account in assessing the overall cost impact of the policy. 3.DWP’s estimate of the savings from the policy The DWP published two impact assessments of the underoccupation penalty, in February 2011 (DWP 2011) and June 2012 (DWP 2012). The June 2012 document was published after the Welfare Reform Bill received Royal Assent, in March 2012. It is 4 the most up to date and authoritative assessment. This impact assessment projected that the policy would produce UK Housing Benefit savings of £480m in 2013/14 and £450m in 2014/15, a total of £930m over the first two years of operation (DWP 2012). This estimate of savings was calculated by multiplying the average estimated penalty reduction in Housing Benefit by the estimated number of affected households. For example, for 2013/14, the calculation was £14 a week x 52 weeks x 666,000 affected households = £480m. These figures have been quoted by ministers in debates in Parliament, for example by Lord Freud in the Lords (Wilson 2013). However, there is growing uncertainty about whether projected cost savings can be achieved. The DWP itself stated that there is “uncertainty about likely claimant and landlord responses to the introduction of the social sector size criteria [which] creates uncertainty about the benefit saving likely to be realised” (DWP 2012 p13). If affected tenants respond to the policy by entering work or gaining additional hours of work, or by increasing their income in other ways, the DWP saving is maintained. The saving may even increase, if the total number of tenants on full benefit and/or the size of partial benefit falls. However, if tenants respond by moving home, to homes with fewer rooms and/or different rent, the impact on Housing Benefit savings becomes complicated. Many tenants may try to downsize so that they are no longer underoccupying, and so avoid the penalty. They may take other actions to get out of underoccupation such as taking in a lodger (if their tenancy allows it). Any responses that result in lower numbers and proportions of tenants claiming Housing Benefit being affected by the size criteria will reduce the overall saving resulting from the policy. Some responses could actually add to a household’s Housing Benefit costs, thus also reducing net savings from the policy. For example, some affected tenants might downsize by moving to correctly-sized but higher rent homes in the private rented sector. Some moves within the social rented sector could also have this effect, as many social landlords are now letting homes at ‘Affordable Rents’ which are up to 80% of market value. 5 The DWP’s impact assessment acknowledged, “if a significant number of tenants wished to move, this would reduce direct savings” (DWP 2011a p3). However, the DWP’s savings projection assumed that not one of the affected 660,000 households would respond to the policy by moving to a smaller home. 4.The model the DWP used to estimate the savings from the policy and the risk of higher or lower savings The model the DWP used to calculate likely savings and to carry out sensitivity tests estimating the risks of higher and lower savings has been obtained through a Freedom of Information request (‘VTR1268 SSRS Scenario modelling’). Details of the sensitivity testing of the model (‘VTR1268 Size criteria sensitivity analysis results’) were also obtained. The spreadsheets provided under the Freedom of Information Act revealed the structure and results of the model but contained no formulae. A working version of the model was reconstructed (see Technical Note). The DWP impact assessment stated that the DWP model was based on four key variables relating to tenant behaviour and Housing Benefit: 1) The proportion of those underoccupying by two bedrooms who would move; 2) The proportion of those affected who move out of the social rented sector into the private rented sector (with the assumption that the remainder move within the social rented sector); 3) The proportion of homes vacated by affected movers which would go to households who were formerly overcrowded social rented sector tenants; 4) The proportion of social rented sector tenants who were claiming Housing Benefit (full or partial not specified) (DWP 2012). The variables stated in the impact assessment included only moves by those underoccupying by two bedrooms. However, DWP estimated that those underoccupying by one bedroom made up 80% of those affected by the penalty (DWP 2012). Thus their responses are key to the overall tenant response and to potential savings. The model underlying the figures in the impact assessment did contain an implicit assumption about responses by those underoccupying by one bedroom. The DWP model assumed that the proportion of this group who moved would be inversely 6 related to the number of those underoccupying by two bedrooms who moved. For instance, it assumed that if 10% of those with two spare rooms moved, then 7.5% of those with one spare room would move. However, it assumed that if 30% of those with two spare rooms moved, only 2.7% of those with one would move. It does seem likely that higher proportions of those with two rather than one spare rooms would move, as they face a higher absolute and relative financial penalty from the policy. This is indeed borne out by real data from the four housing organisations (see below). However the idea of an inverse relationship between the two figures seems odd, and has not been borne out by experience. In a later section, this report discusses in more detail whether these four variables were sufficient to capture tenant responses and the impact on projected savings. The DWP model assumed that each of these four variables would fall within a certain range, and tested the effect of varying each in turn on the overall results. The sensitivity tests examine the impact of the 192 possible combinations of the following assumptions: 1) 10%, 20% or 30% of those underoccupying by two bedrooms would move; 2) 15%, 20%, 25% and 30% of those affected who moved would move out of the social rented sector into the private rented sector; 3) 25%, 30%, 35% and 40% of properties vacated by affected movers would go to households who were formerly social rented sector tenants; 4) 65%, 70%, 75% and 80% of social rented sector tenants were claiming Housing Benefit. These tests were intended to explore factors which were uncertain at the time of modelling. However, they included variations in social tenant Housing Benefit claims from 65% to 80%. This is odd, because the DWP knew at the time they were preparing savings projections in mid-2012 that the UK figure was 65%. They also knew that it was unlikely to vary substantially by 2013/14 or 2014/15. In addition, only Housing Benefit increases above 65% were included in sensitivity tests, not falls below 65%. Housing Benefit increases above 65% are likely to result in calculations showing increased savings from the policy. 7 5. The impact of applying real data in place of assumptions in the DWP model on estimated savings and the risk of higher or lower savings The four housing organisations Four housing organisations affected by the policy, Riverside , Affinity Sutton, Gentoo, and Wigan and Leigh, have gathered data on the real impact of the first four months of the policy on their tenants.  Affinity Sutton is a housing association. It owns and manages over 50,000 general needs social rented homes. It is the result of a merger of William Sutton Homes, an early twentieth century housing trust which originally had a concentration of homes in London but later spread nationwide, Downland Housing Association, an association housing association with a concentration of homes in Sussex, and Broomleigh, LB Bromley’s LSVT.  The Riverside Group Ltd. is a housing association group. It owns 34,000 homes for general needs social rent. It developed from a base in Liverpool early in the twentieth century. It still has a concentration of homes in Merseyside, and manages 17% of social rated homes in Liverpool. In addition it now has homes in every region of England, as well as in Scotland, and works in a total of 173 local authorities. This is as a result of growth through development, a merger with ECHG and Irvine, and local authority stock transfers from Hull, Liverpool, Manchester and Carlisle.  Gentoo is a housing association group. It owns and manages 32,000 homes, mainly for general needs with some supported housing concentrated in and around Sunderland, and in central and west Scotland.  Wigan and Leigh is an ALMO (arm’s length management organisation). Since 2002 it has managed the 22,000 homes owned by Wigan Metropolitan Borough Council. It manages 87% of social housing in Wigan, and does not manage homes located outside Wigan. Supported housing accommodation and some homes in Scotland managed by the four organisations were excluded from the analysis. The analysis of real data is based on a total of 127,494 homes in England, 2.6% of the UK national social housing stock. Their homes and tenants are similar to national averages on most relevant dimensions, such as Housing Benefit claim rates, the percentage of tenants affected 8 by the underoccupation penalty, and the extent of underoccupation in individual cases. Three of these four housing organisations previously worked as part of the Housing Futures Network, a group of housing associations which published early research work on the likely impact of the policy which is referenced in the DWP’s second impact assessment. They operate across 225 local authorities in England and Scotland, with particular concentrations in the North West, the North East and London. However, the regional location of their homes differs somewhat from the national pattern for social housing, as they do not have many homes in the East or West Midlands and Scotland, and have none in Wales or Northern Ireland. Real September 2013 data from housing organisations compared to June 2012 DWP assumptions In September 2013, five months into the first year of the underoccupation penalty and the impact assessment period, Riverside, Affinity Sutton, Gentoo, and Wigan and Leigh gathered early real data on the number of affected tenancies by home size, extent of underoccupation, rent and local authority. Some of the data was gathered as part of routine housing management. Some, particularly on the numbers of tenants affected and their moving intentions, had been gathered as part of special research connected with welfare reform. For example, Riverside identified tenants thought to be affected by the underoccupation penalty, and spoke to a total of 5000 households about their options and plans. Local authority specific figures were used where local data was sufficiently robust numbers, while the overall Riverside Group figure was used in local authorities with only small Riverside holdings. Organisations had data on intentions to move and actual moves to date by those affected by the underoccupation penalty. Given intentions to move, and a slow start to moves, it has been assumed that the early actual move rate will be sustained at the same level throughout the remaining months of the first year of the policy. Additional data was gathered on the volume of social housing turnover and rent levels in the private rented sector. This was necessary to calculate the proportion of affected tenants who want to downsize who move into the private rented sector, and the impact on costs. On the basis of the four organisations’ contact with tenants, it was assumed that all tenants who wanted to move would first seek to downsize in 9 the social rented sector. However, on the basis of the four organisations’ actual practice, it was assumed that no more than 33% of empty homes would be made available by landlords to those affected by the underoccupation penalty1. Landlords would have to apply their usual allocations policy and take account of those on the waiting list and other transfer needs, and cannot officially ‘ring-fence’ homes for those affected by the penalty. Because of the structure of the model, those for whom there was no space in the social rented sector were assumed to move into the private rented sector, although in reality there are other possibilities which have not been considered in the DWP’s modelling. Where possible, the organisations active across multiple local authorities gathered specific data for their homes and tenants in each of these areas. For example, the proportion of tenants moving into the private rented sector and the cost of such moves has been based on the local supply of social housing lets and the local cost of the private rented sector. This is in contrast to the DWP figures, which were based on national averages. Data from Riverside and Gentoo on their Scottish homes and tenants was excluded because it was not compatible with other data. The four organisations checked and then pooled their data. Findings from real data Real data suggests that three of the DWP’s four key assumptions should be reexamined. DWP’s savings estimates appear to have been based on: 1. Underestimates of the proportion of those underoccupying by one bedroom who would move; and underestimates of all those underoccupying who would move; 2. Underestimates of the proportion of those affected who would move into the private rented sector, and 3. Underestimates of the proportion of homes vacated by affected tenants which would be let to existing social housing tenants (Table). 1 This figure was estimated as 50% for Gentoo and 33% for Riverside, Affinity Sutton and Wigan & Leigh 10 Table: Real September 2013 data from housing organisations compared to June 2012 DWP assumptions DWP assumption Real data Is real data from four within organisations DWP (means) assumption range? Proportion of those 10%, 20% or 30% 24.7% Yes 7.5% (when above is 10%) 20.1% No 21.0% No 41.5% No 84.8% No 65.2% Yes underoccupying by 2 rooms who move Proportion of those underoccupying by 1 room who move (implicit) Proportion of all those underoccupying who move (implicit) 5.4% (when above is 20%) 2.7% (when above is 30%) 8.1% (when above is 10%) 8.0% (when above is 20%) Proportion of those affected 7.8% (when above is 30%) 15%, 20%, 25% and who move who move to the 30% private rented sector Proportion of homes vacated 25%, 30%, 35% and by affected tenants which are 40% re-let to existing social housing tenants The proportion of social rented sector tenants claiming Housing Benefit 65%, 70%, 75% and 80% 11 It is worth noting that Lord Freud, the Minister for Welfare Reform, rejected the idea that ‘substantial’ numbers of affected tenants might move into the private rented sector: “if that really were the case we would not be implementing the change” (House of Lords Debates 29th February 2012 c1370). However, real data suggests that 41.5% of affected tenants who decide to move may go into the private rented sector, based on moves to date and the likely volume of social rented homes available to move to in the relevant areas. This is higher than the DWP’s 10-30% estimate, and appears a ‘substantial’ figure. The impact of real data on estimates of the likely savings from the underoccupation penalty All the underestimates above are likely to mean that estimates of savings were too optimistic and that the projected £480m savings are unlikely to be achieved. The next part of the analysis inserted real data into the DWP’s model, to examine the effect on expected savings and on sensitivity tests. In summary, using this real data in the DWP’s model suggests that real Housing Benefit savings may be reduced by as much as £125m (26%). Using real data and taking account of regional variations suggests that real Housing Benefit savings may reduce by as much as 33% from what was projected. The DWP model tested different combinations of the figures for the four key variables, to see what impact this might have on projected savings, and to see how confident they could be of savings projections. The spreadsheet obtained under the Freedom of Information Act showed that DWP tests found that 52% of combinations of assumptions would result in reduced savings compared to the estimate of £480m for 2013/14. However, 48% would result in increased savings. 60% of tests carried out resulted in savings within £10m either side of the central estimate, and all had savings within £30m either side (DWP 2012). Thus DWP concluded that despite the question of tenant response, they could be confident of the savings projection of £480m in 2013/14, and that there was little risk of large variations in savings. As noted, these tests included variations in social tenant Housing Benefit claims from 65% to 80%, which was odd. If we use DWP assumptions on tenant moves, but insert the real figure for the four housing organisations, which matches the well12 known UK Housing Benefit figure (65%), 72% of tests result in reduced savings, and just 28% result in increased savings. If we use real data from the four organisations on tenant moves, and carry out DWP tests, 100% of the tests result in reductions in savings. This suggests it is very unlikely that the projected £480m savings from the policy will not be achieved. If we bear in mind that the four organisations house 2.6% of social housing tenants nationwide, and assume that their homes and tenants are typical of those nationwide, the central estimate is a £126m (26%) reduction in savings to the UK Housing Benefit bill. However, the homes and tenants of the four organisations are not typical, and reductions in savings vary by region. For example, the mean net reduction from projected savings suggested by real data is £113 per affected tenant in the East Midlands but £610 in London. To get the best picture of national impact, we should multiple these regional figures by the DWP’s estimated number of affected tenants in each region (DWP 2012). If we use real data from the four organisations, but also take into account regional variations in impact, the total reductions in savings increases further to £160m (33%). In addition to a central estimate that suggests the £480m saving will not be achieved, real data suggests more variation in potential outcomes. With real data, only 29% (not 60%) of sensitivity tests result in savings within £10m of the new central estimate. Reductions in savings of up to £186m in total (39%) appear possible. In summary, applying this real data to the DWP’s model suggests that real Housing Benefit savings may be reduced by as much as £125m (26% of the original projected total). Taking account of regional variations could further increase the reduction in savings to £160m, or around two thirds of the savings projected by DWP. 6.Additional questions for the DWP model Up to this point, we have taken the DWP model at face value, and simply inserted real data in place of assumptions. However, the model has a number of structural features which raise further doubts about the predictions of likely savings.  The model does not allow for all the realistic possibilities in terms of moves. For example, some movers may end up with higher rents and higher total 13 Housing Benefit costs, because of the increasing numbers of housing association homes let at ‘Affordable Rents’ which are close to market levels.  The model assumes that vacated homes are only filled by: a) Existing social rented sector tenants who are overcrowded, but by one bedroom only, or b) Private rented sector tenants. In practice, many vacated homes are likely to be taken by other types of social rented households, or new households, some claiming Housing Benefit for the first time.  In addition, the model does not take into account all the later steps in the housing chains these moves set off. To the extent that chains of moves result in reduced underoccupation in social housing, they will result in reduced savings from the penalty.  The model takes no account of the fact that some tenants affected by the policy might respond to the policy by adding family members or lodgers to their household (where their tenancies allow), so that they are no longer underoccupying and could claim full Housing Benefit once more.  The DWP predicts that annual savings in 2014/15 will be similar to those it predicts for 2013/15 (just 6% lower). However, moves in the first year will leave fewer subject to the penalty at the start of the second year. It seems likely that there will also be numerous moves in the second year of the policy, as some of those who thought they would ‘pay and stay’ decide they need to downsize, and as additional tenants become subject to the policy when household members leave or if they start benefit claims. Moves in the second year of the policy will generate reductions in savings in the second year just like in the first, and so on. In summary, an examination of the DWP’s model and early real data suggest savings from the underoccupation penalty may be substantially lower than projected. 14 7.Additional factors which affect the overall cost impact of the policy Finally, other costs in addition to Housing Benefit should be taken into account in an overall assessment of the impact of the policy on the public sector. These include potential extra net costs to tenants, landlords, local authorities, health authorities and the voluntary sector. The National Housing Federation (NHF) is exploring these costs in more depth. These include:  The £65m increase in budgets such as Discretionary Housing Payment already set aside for 2013/14 to help those facing shortfalls between rent and Housing Benefit. This money should be netted off ‘claimed’ savings immediately;  The additional costs of fitting aids and adaptations in new homes for disabled tenants who chose to move;  The significant additional costs to housing associations facing rising rent arrears, re-let times, rent collection and tenant support costs  The knock-on impact of these costs on the ability of housing associations to build new homes, at a time when Government is trying to encourage increased supply;  Additional indirect costs to other public services coping with the consequences of tenants moving or accumulating debt, for example homelessness, health, education, and social services and advice services. 8.Conclusions At this point, only early data are available, and the four organisations that have provided it do not claim to provide a fully representative sample. However, it appears possible that the underoccupation penalty may be much less able to achieve its main goal of creating UK Housing Benefit savings than projections suggested. In summary, an examination of the DWP’s model and early real data suggest savings from the underoccupation penalty may be substantially lower than projected. . In addition, there appear to be serious shortcomings in the DWP model used to project savings, which excludes a number of significant factors which may further reduce savings. Finally there are a number of additional factors which create 15 additional costs to the public sector and should be taken into account in an assessment of the overall cost impact of the policy. References DWP (2011) Underoccupation of social housing: Impact assessment, February, London: DWP DWP (2012) Housing Benefit: Underoccupation of social housing: Impact assessment Updated June 29 2012 London: DWP Wilson, W (2013) Under-occupation of social housing: Housing benefit entitlement Standard note SN/SP/6272 30th July London: House of Commons Library 16 Technical note The results of the reconstructed model do not exactly match those from the original DWP model, but the figures obtained are sufficiently close that the model can be said to have been successfully reconstructed. 83% of the results of the reconstructed model were within 5% of the DWP results. Explanations for anomalies include a DWP sign error and the effects of DWP rounding some numbers up and down. 17