Wider Bristol Housing Market Area Strategic Housing Assessment 2015: Commentary
Wider Bristol Housing Market Area Strategic Housing Assessment 2015: Commentary
Prepared for Business West
Professor of Urban Studies, Heriot-Watt University, Edinburgh EH14 4AS
This note has been prepared for Business West as an assessment of The Strategic Housing Market Assessment (SHMA) for the Wider Bristol Housing Market Area published in June 2015. It is drawn from a fuller analysis, Planning for Housing in the West of England – the Scale of the Challenge.
The Wider Bristol Strategic Housing Market Assessment (SHMA)
The Wider Bristol SHMA concludes that a total of 85,000 houses need to be built between 2016 and 2036 in order to meet future housing need. This includes the level of need, specifically, for houses which are affordable to those on low incomes. In reaching this figure the SHMA draws on a range of evidence and assumptions including:
- National population projections
- The Government’s household projections
- Migration trends – 10 year trend between 2001 and 2011
- ‘Concealed’ families (e.g. living with parents) and homeless households
- Economic forecasts commissioned from Oxford Economics
- Future employment demand including the target of 95,000 additional jobs between 2010 and 2030 for the WoE set by the West of England LEP
- ‘Market signals’ including the level of house prices and rent and the affordability of housing
- The backlog of housing provision to meet needs up to 2016 which is yet to be delivered
The SHMA was drawn up following Government Guidance on how this should be done. The process is, however, based on a series of assumptions, choices in terms of the evidence used and how this is interpreted.
Business West commissioned Professor Glen Bramley of Heriot-Watt University, Edinburgh, to assess plans for housing delivery as set out in the SHMA as part of a broader study of Planning for Housing in the West of England. This analysis draws on a comprehensive model of local housing market areas covering the country as a whole.
This assessment of the SHMA presented here concludes that the Wider Bristol Housing Market Area needs to increase significantly, its headline figure of planned housing numbers and its delivery of housing if it is to achieve desired outcomes in terms of the criteria for housing and economic performance set out in National Planning Guidance (2014). If it does not do so:
- Overall housing need will deteriorate
- The backlog of unmet housing need will get worse
- House prices will rise relative to incomes and the affordability of home-ownership will get worse both in absolute terms and compared to the national picture
- The affordability of rented housing will also deteriorate significantly over the next twenty years
- The supply of ‘affordable’ housing provided specifically for those on low incomes will fall well below levels of need
- The size of the labour force (population of working age) will fail to keep pace with the potential number of jobs created leading to upwards pressure on labour costs and slower economic growth
The key reasons for this are the assumptions and evidence on which the SHMA relies. These are set out in detail in the longer study but can be summarised here.
- The SHMA is over-reliant on the latest projections of future household numbers produced by the Department of Communities and Local Government, the ‘2012 projections’.
- These are the latest official projections from the Government but are widely believed significantly to understate likely future household growth. This is because they are based to a large extent on the period since the recession from 2008 on, a period when most experts agree that levels of household growth were lower than in more normal times.
- During the recession, young people and couples in particular were forced to postpone forming their own new households – there was higher unemployment, economic uncertainty, there were less houses available on the market and mortgage finance was less readily available. In particular, low levels of housing supply as a ‘one-off’ effect of the biggest post-war recession and global financial crisis suppressed levels of household growth.
- Levels of migration into the area of those taking up employment opportunities were also lower given that fewer jobs were available during the recession, and also that levels of housing output were relatively low in Wider Bristol over the whole period 2001-11. At sub-regional level, migration is strongly influenced by housing supply. Basing expected future household numbers purely on these latest estimates is therefore likely to underestimate future levels of housing need and demand in the housing market. Levels of new household formation and of in-migration will increase as the economy picks up.
- Levels of household growth generated by the model and the headline figure of 85,000 new homes needed by 2036 do not meet the need for employment (based on the size of the population of working age) to meet the likely of future job growth.
- This is reinforced by the fact that the economic forecasts on which the SHMA is based are conservative in terms of likely economic and employment growth, particularly given the past trajectory and future potential of the West of England based on indices of competitiveness.
- Levels of affordable housing provision for those on lower incomes proposed by the SHMA in themselves reasonable. The evidence suggests, however that in order for such levels to be delivered in practice a much higher level of overall housing numbers will be required.
- The level of uplift in overall housing numbers suggested as necessary in response to the overall level of house prices and housing affordability (as required by National Guidance) is a serious underestimate. Affordability in the West of England is 13% worse than the national average. The SHMA, in response, includes an uplift of overall housing numbers by 7.5% to reach the figure of 85,000. Analysis suggests that this underestimates by several orders of magnitude, the level of uplift needed to impact on overall levels of affordability and that the uplift should be at least 35%.
The assessment of the SHMA looks at a range of possible future outcomes based on different assumptions. Overall, however, it concludes that the objectively assessed level of future housing need based on sound evidence and reasonable assumptions including demographic factors and market signals is in the order of 130,000, at least 50% more than the level proposed by the SHMA.
This technical report summarised key elements of the analysis which supports the conclusions set out in the summary above. The full technical analysis is set out in a longer report prepared for Business West: Planning for Housing in the West of England – the Scale of the Challenge.
- The Economic Potential of the West of England
As I outline elsewhere (Housing Challenges Facing the West of England) Greater Bristol (WoE) has relatively high economic growth potential. This matters for housing planning because one of the important criteria for judging how adequate levels of planned housing numbers are is whether they enable the labour force to keep pace with potential employment growth. If they fail to do so, then there are potentially adverse consequences for housing. But there is also a significant risk that economic growth may itself be choked off. In this context, it is significant that Bristol has worse than average home-ownership affordability and this is expected to remain worse into the future.
One immediate conclusion of the analysis is that the current SHMA fails to ground its proposed numbers in an adequately wide range of plausible economic and demographic scenarios, reducing both the credibility of its findings and the ability of the West of England to adequately respond to future growth.
The economic forecasts being used by the current SHMA are still quite conservative, particularly in terms of upside range and employment growth figures, given the economic growth record of the West of England and its economic growth potential.
The employment growth potential of WoE is relatively high, among the highest of the comparable set of city-regions. Taking a longer term perspective, Bristol has one of the highest growth trajectories of key city regions in England. This reflects both a relatively straight forward projection of growth in GVA and productivity, blending past local performance and assumed national trends, and a more qualitative assessment of the factors underpinning growth and competitiveness as well documented by The Centre for Cities.
It might be suggested by some that Bristol’s employment growth has faltered and cannot be expected to recover fully to earlier levels. I think that this is a misreading of the evidence. It is true that Bristol was leading the pack in the early 2000s and then faltered somewhat before dipping sharply in the recession.
But similar faltering and recessionary falls were experienced in nearly all of the comparator areas. In other words, this pessimism is too influenced by the exceptional experience of the recession. Planning should take a longer term perspective, and in that context Bristol can expect significant employment growth.
In the light of this it is surprising that the SHMA does not identify employment prospects as a factor weighing in favour of higher housing plans.
Comparison with a peer group of comparable city regions confirms this picture, and draws attention to the particularly anomalous position with household projections, where WoE has seen an exceptional reduction between 2008-based and 2012-based numbers. Other indicators show WoE as facing relatively high demand. I discuss household projections further in the next section.
- Household Projection Figures
The current SHMA is based on an excessive reliance on flawed household projections that lack credibility – in particular on one set of (2012) figures that are likely to have been strongly conditioned by ‘one off’ effects following the financial crisis and ‘great recession’ of 2009 – 2010.
The approach of the current SHMA also fails to recognise endogeneity and feedback between lower housing supply levels and lower household growth and make sensible upwards adjustments to these.
The reduction in the household projection numbers for WoE between the 2008-based and 2012-based figures, which amounts to a 40% or 56,000 reduction over 20 years, are anomalous and problematic, casting doubt on the value of household projections for housing planning.
The underlying problems here may not be fully appreciated by the lay reader, but they are known by those who study these things closely. In publications going back to Bramley & Watkins (1995) I have drawn attention to the potential ‘circularity’ involved in using household projections in planning for housing growth. Circularity arises when (a) household projections are used as the basis for planning numbers, while (b) actual housebuilding (which is governed by planning numbers) affects the amount of migration to and household formation within an area, and thereby affects observed household growth and the recent trends in those factors, which are then used as a basis for the household projections.
In a recent paper given at the British Society for Population Studies conference in 2014, I made the following comment about the weaknesses of an over reliance on household projection data.
“Against these must be set an awareness of the problems which can result from approaches to household forecasting and planning which ignore key economic effects. One example is the problem of ‘circularity’ (Bramley & Watkins 1995), or the self-fulfilling prophecy, which may lead to a cycle of underprovision in some regions. Another example is the situation whereby projections get out of phase with cycles, because of the convention of using the latest (but significantly lagged) data, ending up changing in the wrong direction relative what is actually going on in the market. There is a significant illustration of this danger in the Interim 2012-based household projections for England, which are heavily affected by the effects of the Great Recession of the period 2008-12. Any set of projections or forecasts which imply persistent discrepancies of households vs dwellings need to be questioned, for reasons set out later*. Belief in the ‘fact’ that a projection says that there will be a certain rate of household growth, regardless of what happens to supply, prices, incomes or employment, betrays a lack of realism about adjustment mechanisms in the market. Last but not least, over-reliance on household projections alone may lead to inappropriate planning between related geographical areas, as in the example of Edinburgh and the Lothians, or London and the South East.”
* this is a reference to the definitional identity relationship between households and dwellings, allowing for vacancies, second homes and sharing. Bramley & Watkins 2014c.
In this paper as well as in a new published peer-reviewed paper in Progress in Planning (Bramley & Watkins 2015), I present further evidence including the use of the SRHMM to suggest that, at sub-regional level and below, if you build more housing in an area then, in the medium-to-longer term you will find more households living there, in a ratio of typically 50-80% and sometimes higher. In Table 3 in that paper it suggests that the ratio of incremental household growth to incremental net additions to housing stock would be 0.74 for WoE, 0.78 for typical South East growth areas, 0.67 for the South East as a whole, but 0.52 for the North West.
The 2012-based household projections imply a growth rate of 4038 extra households per year; as noted this was a 40% reduction on the 2008-based projections figure of 6840 per year. However, the actual growth in households achieved by WoE between the 2001 and 2012 censuses was only 3242.
The main cause of that lower growth was probably the fact that the net additions to housing stock only amounted to 3046 pa over the decade. While the recession clearly reduced supply after 2008, even in the better years (2005-07) completions only totalled 2529 pa. This lower level of supply can reasonably be expected to have impacted on population growth (through net migration) and on household formation, during the decade. Using a combination of actual data and model estimates, it would appear that net migration to WoE was indeed much lower (at 4759 persons pa) in that decade than in our forward forecast.
Furthermore, that positive net migration was more than wholly accounted for by international migration. In other words, Greater Bristol was in a net negative position on domestic migration; it was losing more UK resident population than it was gaining, in spite of being an economically dynamic and attractive city region in the south of England.
This would seem to be strong circumstantial evidence that housing was being underprovided in the sub-region, especially when taken in conjunction with the worsening housing affordability – real house prices rose by 4.6% p.a. in Greater Bristol over the decade 2001-11, a higher figure than all of the high growth comparators examined in Table 1 and markedly above the England average of 2.9%.
Further indirect corroboration is found when we look at household formation, using household ‘headship’ rates (the percentage of people in a particular age group who head, or represent, a separate household).
The following Figure 1 looks at the proportion of 20-29 year olds who head (or ‘represent’) separate households for a number of years over the period 1992-2013 in selected regions, including the South West, based on the Labour Force Survey. Three key features stand out from this. Firstly, southern regions including the South West have seen a generally downward trend in young adult headship, whereas the trend in the North East and East Midlands (the least pressured housing market regions in England) has been close to static.
The decline in the South West started later than in London and the South East, and was not quite as great in 2013 (although it was in 2008). Thirdly, there was a curious upward blip between 2008 and 2010, in all areas, but levels have declined steeply again since then, particularly in the southern areas. It is not quite clear why the blip happened, but I would suggest that this is associated with the sharp growth in private rented lettings associated with buy-to-let (plus owners unable to sell in the market downturn) at this time. A consequence of the blip is that the 2011 Census tends to paint a more optimistic/static picture of recent trends in headship, comparing with 2001, than one gets from seeing this fuller picture of trends over time.
From our model base data and taking account of these evidence sources, including the Census, we estimate that the headship rate for the younger age group (16-24) declined from 0.162 in 1991 to 0.148 in 2001 and 0.108 in 2008, before bouncing back (as part of the ‘blip’) to 0.143 in 2011; however we expect it to fall again, in line with the regional drop shown in Figure 1, and to be at a lower level (0.134) in 2021.
The 2012-based household projections include a table showing the proportion of the increase in households accounted for by population change, household formation or the interaction of the two. For the three WoE authorities, the proportion accounted for by population is 99% (Bristol) and 96-97% (N Soms & S Gloucs).
In other words, household formation as a process is stalled – people’s chances of forming a new household are not improving, and this is now factored into the projections as a ‘permanent’ feature.
However, it is arguable, from a longer term perspective, and having regard to a wider literature on household formation (Bramley et al 1997, Bramley & Watkins 2014c), that there are longer term social changes in society which are likely to lead to further separate household formation, particularly among younger adults, given economic and housing market conditions that permit people to realise their preferences.
There is certainly strong evidence that the recent downturns were very much a reaction to cyclical economic and market conditions. There is obviously room for debate around some aspects of these trends and possibilities for the future, for example the different situation of older people (where more couple relationships may be lasting into later ages), groups who are at ‘saturation level’, the effects of welfare benefit reforms/cuts (which may further squeeze down the already very low headship of under-25s), and possible differences affecting migrant and ethnic communities (although these are likely to gradually assimilate and increasingly follow similar patterns to the host population). However, it is straining credibility to argue that there will be no net household formation in future.
A more realistic approach is in fact to use an economic forecasting model such as my sub-regional housing market model (SRHMM), which takes account of the observed relationships between headship and economic and housing market conditions. Through exploring different scenarios such a tool can give a more ‘evidence-based’ view of the likely range of future outcomes, compared with the essentially arbitrary ‘high’ and ‘low’ scenarios offered alongside official projections.
To sum up this discussion, and to echo comments made in earlier sections, the recent official household projections raise more questions than they answer and serve to undermine confidence in the general approach and they are likely to underestimate future potential household growth significantly for the WoE sub-region, because they are over-influenced by shorter term cyclical market factors in the immediately preceding period.
- Affordable Housing Delivery
The level of affordable housing provision proposed in the SHMA, while justified in needs terms, lacks credibility as something which can be delivered in the actual policy and market situation which now prevails. Realistic affordable housing quotas in plans and s.106 agreements are likely to be lower, which means the only route to delivering this level of affordable housing will be through increasing the total housing numbers substantially.
Problems with delivering new affordable housing (whether social rent or intermediate, for sale or rent) are far from new – they have characterised the situation facing local authorities in their housing strategies for most of my career, particularly since the cuts which commenced in 1976 which ended large scale council housebuilding. The models which have gradually evolved in England have relied largely upon housing associations as developers, utilising a lot of private finance as well as recycled internal surpluses to supplement increasingly low rates and amounts of government grant, together with the use of free/discounted former public land (where still available) and increasingly Section106 planning agreements, based on local policy targets which themselves relate back to the needs assessed in SHMAs.
This system has delivered moderate numbers, which grew in the late 2000s under the previous government, partly through counter-cyclical measures reacting to the early stages of the financial crisis, and then more recently through the current government ‘stretching’ grant more thinly under the Affordable Rent scheme. In Greater Bristol net additions to affordable housing have run at between 500 and 1000 per year recently, of which 75% on average have been social rented housing (including Affordable Rent in last couple of years). However, the trend has been declining from 2008 peak of 1360 to only 570 in in 2013. In our baseline modelling we effectively assume a gradual increase from 500 now to about 790 in 2041. The SHMA has little specific to say about the immediate prospects for delivery of affordable housing, although it mentions the Affordable Rent scheme commitments which equate to 176 units per year up to 2018.
The WoE faces a situation where the need for new affordable housing, looking forward over the plan period, is estimated at 1,455 dwellings per year. A higher figure of 2,562 dwellings per year is derived from estimates of the impact of reduced levels of Welfare Benefit/Housing Allowance support for low income tenants in the private rented sector – however, this figure is not referred to in the main summary document. I have reviewed the method used to derived the former figure, which seems reasonable and defensible although one might query the methodology’s reliance on DWP benefit claim data rather than looking at the overall distribution of household incomes (which would probably reveal more ‘intermediate’ affordable housing need).. There was a similarly relatively large figure which emerged from the previous SHMA which I was associated with.
The obvious and most salient point, strategically, is that this level of affordable housing lacks credibility as something which can be delivered in the actual policy and market situation which now prevails.
Firstly, it is two-and-a-half times higher than current actual delivery in WoE, and higher than delivery in any recent year on record. Secondly, the government’s current headline programme (Affordable Rent) is offering to deliver only 12% of this total over the coming 3 years. Thirdly, the Government is signalling that it is looking to switch support from social rent to low cost home ownership, which alongside other signals suggests that the already meagre subsidy budget will be further cut. One might well add that the government’s currently favoured low cost home ownership product, ‘Starter homes’, is wholly inadequate, as it is barely more affordable than the second-hand market and does not give benefits in perpetuity or recycled subsidy. Fourthly, several post election and post-budget policy changes affecting housing associations, particularly the unexpected reduction in target rents, will reduce association surpluses and reduce their capacity to develop new social or affordable housing. Fifthly, although extensions and revivals of the Right to Buy have promised mechanisms to achieve replacement, it is far from clear that these mechanisms are robust and delivering.
Further to these points, we have to consider the cumulative effect of changes in the Section 106 mechanism which have resulted from a series of policy announcements over the last several years, and in particular the most recent pledge to switch S106 obligations away from affordable rented products to ‘starter homes’ to buy made by the Prime Minister on the 7th of October 2015.
Viability assessments have become important and have been used to get affordable housing requirement watered down or removed in many instances, partly because of the ability of applicants to put forward viability assumptions which slant the picture in their favour. Certain classes of site have been removed from s.106. There is also an inevitable trade-off between getting more affordable housing and getting infrastructure paid for through CIL.
Given the changes to s.106 and the range of sites which will likely evade its reach, and the effects of viability considerations in particular, one has to view with scepticism a needs based target which implies a general ‘quota’ of 34% affordable housing. In the current climate, WoE may be doing well to get much more than half of that. This is particularly reinforced by the scarcity of public subsidy which can be clearly anticipated. Bristol has in the past tried to follow an approach where normal s.106 sites did not collect subsidy and had to break even on cross-subsidy. I would commend this approach, particularly in circumstances where there is not going to be much if any subsidy. But the corollary of this is that the s.106 quotas will have to be lower, and include a high share of LCHO, to pass the viability test.
Adopting a more positive approach to the total housing numbers enables a more virtuous circle to work with respect to affordable housing. A larger total target will lead to more housing supply and improved affordability and need outcomes. But it will also, crucially, open up more sites on which s.106 contributions can be obtained. In this way more of the gap in terms of need can be bridged.
Since of necessity more of the affordable housing will be LCHO, this will also make a significant contribution to bridging the affordability (including the deposit) gap for potential homebuyers, which we have indicated as a significant issue for WoE. However, care is needed in establishing exactly what type of provision counts as affordable. 20% discounts on developers’ new build products may not be more affordable that what is available in the second-hand market.
- Adjustment to Market Signals
SHMA guidance currently requires Local Authorities to have regard to market signals and comparison with benchmarks, but is vague about how to adjust plans where signals are adverse.
The current WoE SHMA draft proposes uplifting numbers by 7.5% on account of affordability being 13% worse than England average, based on one Local Plan Inspector’s report (Eastleigh)
The Sub-Regional Housing Market Model (SRHMM) (which is consistent with other economic models e.g. Professor Geoff Mean’s ‘Reading’ model) demonstrates that the ratio of supply adjustment to affordability impact should be close to the ratio of c.5:1. This would imply raising West of England housing numbers by 64%, not 7.5%.
The trade-off between supply increase and affordability improvements is roughly 5:1 in proportional terms (s.6.4). This means that the SHMA is quite wrong to suggest that a 7.5% uplift in plan numbers is an appropriate response to market signals, when a more appropriate response would be in the range 35-60% (s.8).
2014 Planning Practice Guidance recommends that SHMA’s should consider a range of market signals, comparing their area with other comparable benchmarks and national benchmarks, and looking at trends over time. It tends to focus on the current and recent past and is very cautious about future forecasts, probably having regard to the relatively underdeveloped field of local housing market forecasting, local authorities’ general lack of technical expertise in this area, and the wish to discourage expensive consultancy exercises. It then recommends a weighing of evidence, changing provision numbers in an upward direction if the market signals are generally ‘adverse’, without really giving much of a lead on the magnitude of adjustment. In practice, it is for Planning Inspectors to make rulings on this where there are potential differences of view, but the training and orientation of most PI’s is more geared to legal, procedural and environmental issues, and not usually in economics.
In the WoE SHMA, the consultants have gone through this process in apparent accordance with this guidance. The SHMA report refers to only one recent case (Eastleigh), where the Inspector commented as follows:
5.70 There is no definitive guidance on what level of uplift is appropriate. Nevertheless, the Inspector examining the Eastleigh Local Plan judged 10% to be reasonable given the market signals identified for that HMA:
“It is very difficult to judge the appropriate scale of such an uplift … Exploration of an uplift of, say, 10% would be compatible with the “modest” pressure of market signals recognised in the SHMA itself.”
5.73 Given the relative market signal indicators for the two areas and the views of the Eastleigh Inspector, it would seem to be reasonable to consider an uplift of 10% to be at the upper end of any market signals response for the Wider Bristol HMA. On balance we would recommend that the overall uplift was at least 5% but no more than 10% of the housing need identified based on the household projections. The household projections previously identified an increase of 75,804 households (78,478 dwellings) over the 20-year Plan period 2016-36; so the proposed market signals uplift ranges from 3,924 to 7,848 dwellings over the Plan period. We believe that the mid-point of this range, an uplift of 5,886 dwellings, provides an appropriate response to market signals. This is consistent with the views of the Eastleigh Inspector in the context of the indicators for the two areas.
The report then concludes that overall market signals justify an uplift and on balance recommends an uplift of 7.5% of the housing need identified based on the baseline household projections. This is an uplift of 5,886 dwellings. This incorporates an adjustment in response to the identified unmet need for affordable housing (for concealed families and homeless households) and previous under delivery.
I would argue that this line of reasoning is particularly weak, reflecting both the weakness of the Guidance on this issue and the challenges faced by the Planning Inspectorate (in the case of Eastleigh in this instance) in interpreting conformity with the guidance.
The evidence-based analysis referred to in the longer study prepared for Business West suggests that in round terms there is a five-to-one relationship between the supply change response and the affordability discrepancy. What the Consultants have suggested here is a 7.5% increase in provision in response to affordability and other market signals. Their own figures (Figure 72, p.91) show that their key affordability measure (lower quartile house price to earnings ratio) is 13% above (i.e. worse than) England.
Using the 5-to-1 guideline measure, implies at the need for a supply increase of 5×13=65%. . It might be argued, however, that this increase should be from the base of past average net additions (56,840, based on 1997-2007, before the crash) or previous provision levels (65,500 as at 2011) rather than the demographically-determined need figure in SHMA (79,950). Given that the past output data is rather dated now, and given the intervening recession period, the analysis would suggest that, the most appropriate level of uplift would be in the order of 35%.
This suggests that the consultants and the SHMA are wrong to consider a 7.5% uplift to be an appropriate response to their own evidence on market signals (treating affordability as the key signal, for reasons mentioned above). Applying an evidence-based estimate of sensitivity, which is consistent with the other most reputable model of the English housing market, to the affordability indicator they used, I would argue for an uplift of at least 35%.
However, based on my fuller analysis above, and in my separate paper, I would argue for an uplift of 50-60%, assuming a substantial element of social and affordable housing within the mix.