A Report Prepared For
Defra
Land Use Change in Protected Landscapes (AONBs and National Parks): A Guide to the Tabulations
(LAND USE CHANGE INDICATORS FOR PROTECTED AREAS)
February 2014
Department of Town and Regional Planning
University of Sheffield
Land Use Change in Protected Landscapes (AONBs and National Parks): A Guide to the Tabulations
1 Introduction
1.1 This document stands alongside a series of tabulations (University of Sheffield, 2013) that are designed to capture aspects of physical development in protected landscapes in the period since 1985. The present document provides a commentary on those Tabulations, exploring the relationships between the outcomes which they summarise and the operation of the planning system. Some supplementary tables are included to assist in this, which are styled Table S1, S2 etc to distinguish them from those in the principal document (Table 1, Table 2 which are also referred to here)
1.2 The tabulations themselves rest on two principal sources. The first is Ordnance Survey's Land Use Change Statistics (LUCS), which provides a very fine-grained view of land use changes observed in the course of updating the national survey base. The second is Royal Mail's Post Code address file (PAF) which is used to provide supplementary information about the changing numbers of properties. Together they provide a highly detailed picture of the changing pattern of built development in England's protected landscapes and elsewhere so we can track both the areal extent of change, and the density of development.
1.3 To appreciate both the strengths and limitations of the tabulations, it is important to understand the nature of the Land Use Change Statistics. They have been collected by OS since 1985 as an adjunct to updating basic scale maps (usually 1:1250, though sometimes 1:2500 or 1:10,000 in mountain and moorland areas). In this process, topographic changes are made (that is changes to the linework or annotation of maps at the largest possible scale). Where there is a corresponding land-use change a LUCS record is set up. These indicate
location of change (10 metre reference) estimate of the area affected estimated year of change previous use and final use (24 categories)
1.4 Post Code Address File (PAF) data are assembled by Royal Mail to assist in the delivery of mail. It is used here to the stock of dwellings- residential property to which mail is delivered -for very small and irregular areas. In the tabulations in this document, dwelling stock change is measured by comparing estimated stocks from PAF at the time of the 2001 and 2011 Censuses. PAF is continuously updated and when combined with the Office of National Statistics (ONS) Postcode directory provides among other things: a structured address (including name or number of each sub-building, building, thoroughfare, locality and postcode)
an occupier name in the case of non-residential property grid reference (100 metres) ward and other indicators, assessment of the number of dwellings at each full postcode
1.5 Together LUCS and PAF provide indicators of the changing pattern of built development which can be helpful in understanding both market outcomes and the operation of the planning system. Moreover, as they refer to individual properties and land use change events they involve no issues of confidentiality. Thus it is always possible to move back from the tabulations and identify the underlying individual sites. Together these two sources are particularly useful in monitoring the incidence of residential development. Their value in tracking non-residential development and changes between extensive uses between agriculture and woodland is more limited (see Section 4).
1.6 As the tabulations are intended to aid appreciation of the influence of the statutory planning system, it may seem surprising that statistics generated by the planning system are not included. Although the detailed information at property or land parcel used here to provide information about development outcomes might be ideally be combined with similarly fine- grained information about the passage of proposals through the planning system, this is not possible in practical terms. Setting aside the difficulties of extracting information for AONBs which are not in themselves local planning authorities (LPAs), the monitoring of development on particular sites is extremely costly and may be complex. Thus although LPAs hold detailed information on the individual applications which they receive and on decisions made, they rarely have detailed and comprehensive information about actual subsequent development.
1.7 Several considerations militate against the monitoring of planning consents and against the exploitation of monitoring information that is produced. Many sites may have multiple outstanding permissions, and as many consents granted are never implemented, monitoring requires repeated site inspections, or at least telephone calls to developers. The portfolio of sites to be monitored is vast, and as nine out of ten residential planning consents allow the construction of less than ten units (1), where detailed tracking is undertaken, it tends to be focused on larger sites. LPAs might use a cut off of 10 or five dwellings. Further difficulties arise, moreover, in assembling materials for use in a document such as this from a partial patchwork using different cut-offs and definitions. Glenigan, a private sector organization, maintains a database tracking development on more substantial residential planning sites which forms the foundation for monitoring reports for organizations including the Home Builders Federation (2012) and the Local Government Association (2012,2013). Only 7% of consents within its scope, however, refer to sites with less than 10 units (LGA 2012). Crucially – as discussed in Section 2 (Table S2a), sites with 10 or more units are rarely found in protected landscapes.
2 Interpreting the Tabulations: The Residential Development Pipeline
2.1 To understand the patterns of land-use change represented by the tables, it is helpful to focus on the flow of residential development and the contribution made by protected landscapes to its accommodation. This is often likened to a 'pipeline' through which development schemes move from conception to completion (See Figure 1). The schemes in question constitute the portfolio of projects which are so varied and so difficult to monitor. The notion of a 'pipeline' is important to volume housebuilders. Their business strategies typically involve securing a degree of control over housing land supply through the maintenance of land banks including what are termed 'oven-ready' sites ready for immediate development. (See e.g. Barker 2003). For such developers, it is the larger sites that are of concern and HBF Pipeline reports based on Glenigan's database are valuable from their perspective.
2.2 Figure 1 may be understood working from right to left starting with the number of dwellings that are completed, say at the end of a particular year. The tabulations focus on the volume of development emerging from the pipeline and some aspects of its character, while the discussion that follows attempts to explain how statutory planning procedures over the preceding period, stretching leftwards across Figure 1, have influenced and tend to influence that flow. Maintenance of the flow of completions depends upon having sufficient land available with full planning permission ready for construction work to start. Construction typically takes 18 months but this period varies as builders respond to economic conditions. Although terms are not used consistently, the stock of land with outstanding planning permission is usually taken to be that on which development has not started. It increases as new permissions are secured, but also increases when there is little desire to build and the number of starts contracts.
2.3 As the LPA seeks to control the pattern of new development, the flow through the pipeline depends in part on the allocation of sites in development plans. Consistent with Figure 1, plan preparation activity more than five years ago (and often much earlier) would influence the current rate of completions. Preparation of development plans takes place over an attenuated period with extensive public consultation opportunities including examination in public, in which landowners and development interests tend to engage. Other groups may find it difficult to mobilise in response to vague possibilities at these early stages. Only very rarely will there be proposals to allocate major sites in protected landscapes, though this happened at Hawkinge airfield (Kent Downs) as discussed below.
2.4 Although the pipeline metaphor is helpful and much used, it will become clear that some care is required in its application. At the scale of an LPA, or an AONB, the „flow‟ may be far from smooth, and the sudden release of land for residential development through site allocation or following a planning appeal may lead to a jerkiness which becomes more significant the smaller the area considered. Statistical indicators of development for areas such as LPAs or AONBs are thus often considered „volatile‟. Figure 1: The Residential Development Pipeline
Source: Bibby and Shepherd 1993 2.5 The quantum of prime concern is the number of new dwellings built within each protected landscape over a period. Column 1 of Table 1 uses LUCS to estimate this for the period 2001 to 2011, with the corresponding measure for the entire period 1985 to 2011 appearing in Column 1 of Table 6a. By considering these long-term measures it is easier to abstract from issues of volatility and observe more enduring place-to-place differences. All the figures provided are aggregations of the individual land parcel estimates made during map revision and so fit the protected landscapes almost exactly. Because of the lags that occur between construction of new property and survey, this total is an under-estimate. No attempt has been made to adjust the value, but as explained in the notes to the Table, the corresponding building rate shown in column (3) has been adjusted. As survey is inevitably retrospective, there may also be uncertainty about when a development in fact took place.
2.6 In the absence of specific monitoring of the implementation of planning consents or of housing completions, LUCS housebuilding figures are the only ones available at such a fine scale. Housing starts and completions estimates are available for local authority districts as a whole and published by DCLG, though their reliability varies from place to place. It is not generally possible to provide such figures for AONBs. Where comparison is possible, it is found that LUCS estimates tend to track DCLG housing starts and completion fairly well (see Bibby and Brindley 2006). They might be more generally used in planning for protected landscapes and might have been used, for example, in estimating historic dwelling completions information for the newly created South Downs National Park.
2.7 The rate at which new houses are built in principle provides a useful means of monitoring variation in development over time and between localities. A rate represents an amount relative to stock per unit time. Over the inter-censal decade, the building rate across England as a whole averaged 0.7% per annum (as shown in Table S1). Column 3 of Table 1 shows how this varied between the protected landscapes over that period, while Figure 2 shows how it varied more generally from place to place. A rate of more than 1% pa should be considered high. Equally important, it should be appreciated that such high rates are only found in practice where the existing stock of dwellings is very low, and that a high rate does not of itself imply a large volume of development, not least in sparsely populated rural areas common to protected landscapes. This is very starkly illustrated by comparison of Figures 2 and 3. Figure 2 shows the absolute number of dwellings built over the decade within 10kms of any point- highlighting the role of the cities where in absolute terms three fifths of new housebuilding was concentrated. Figure 3, on the other hand highlights the higher rate of development in a belt stretching from the South West peninsula to the Wash and northwards to the Humber. Figure 2 Absolute Growth in Dwelling Stock; England and Wales; 2001-2011 (10km Moving Average)
Focus on growth rates can lead to underestimating the role of major cities and particularly the northern conurbations in accommodating additional dwellings. At this scale, urban intensification and urban expansion cannot be distinguished. Figure 3 Building Rate; England and Wales; 2001-2011 (%) (10km Moving Average)
Areas shown in deep reds show the highest rates of increase in dwellings relative to stock within 10km of any point. The rate at which London absorbed additional dwellings was high relative to stock in contrast to cities such as Birmingham Liverpool, Hull and the Tyneside and Teesside conurbations. Growth rates in Manchester and Leeds were near the overall average. High rates of growth are evident across many rural tracts. 2.8 While variation in the rate of development is important, physical development proceeds by accretion of new property around existing settlements and so the principal influence on the volume of new development within 10km of any point is the size of the existing stock (see Bibby and Brindley 2006). If the AONBs in England accommodate 1.9 % of England‟s population (as suggested by the estimates used in Table S2) there might have been some expectation that would accommodate a similar proportion of new housebuilding over the decade - which would have amounted to around 26,000 units. Given the emphasis on landscape protection, it is not surprising that the actual proportion is only two-thirds of this (though this is partly offset by other adjustments as discussed in Section 3).In other words, the AONBs frequently have higher rates of development relative to their dwelling stocks than is typical of the country as a whole, but have markedly lower rates of development than those of comparable areas without protection as illustrated in Tables 8a and 8b of the tabulations.
2.9 The rate at which schemes pass through the pipeline might be thought of a market outcome mediated by the planning system. The paragraphs that follow attempt to explore some of the ways in which the planning system affects this rate, and other aspects of development outcomes. There are various points where influences can be measured or observed (bearing in mind the strictures of the opening section). Legally, there is an expectation that individuals intending to develop will apply for planning permission, and that such applications will be assessed by local planning authorities having regard to the development plan in force (and other material considerations). Moreover, would-be developers aggrieved by an LPA‟s refusal of permission may appeal to the Secretary of State (on grounds of policy) and beyond that to the courts grounds that the decision-maker has erred in law.
2.10 It is important to appreciate that both the overall rate of housing output and the relationship between the residential development pipeline and planning system practices have been matters of national policy concern. While much popular discussion tends to presume that volumes and rates of residential development are high, Figure 4, based on the Housing Completion statistics discussed above, indicates that in broad historical terms they are low. This is not simply a matter of low housing output in the period since 2008, although this is important (as reflected in planning applications statistics). Various commentators (relying on Holmans, 2005) have pointed out that housing output reached its lowest peacetime level since 1924 and this reflects combination of recession with a long-term tendency.
Figure 4: Housing Output England 1948-2011
400,000 Units Built 300,000
200,000
100,000
0 1940 1950 1960 1970 1980 1990 2000 2010 2020
2.11 It is this long-term characteristic that lay behind the Treasury's concern at the beginning of the last decade that the house building sector was unable to expand output sufficiently, leading to increasing house prices, and consequent increases in indebtedness with house price volatility leading to macroeconomic instability. This motivated a review of housing supply by the Treasury economist Kate Barker (2003, 2004) identifying the scale of additional house building required to moderate house price growth. It identified that construction of an additional 70,000 houses each year in England might reduce the annual trend of real house price growth to 1.8 per cent.
2.12 Although from the perspective of housebuilders, schemes originate in their responses to the market, and while there is a national policy concern to increase the flow in aggregate; from the perspective of an LPA within the statutory planning system, the most important characteristic of the pipeline is the annual flow implied by the housebuilding target which it has set (or inherited). This is expressed as an annual average number of units over the period of its Local Plan. Targets having been set, the amount of land known to be available for residential development can be expressed as a particular number of years supply at the target rate. NPPF (para 47) reiterates commitment for LPAs to maintain a five year supply of housing land available at its target level.
2.13 To monitor the influence of the planning system on residential development effectively, it would be necessary to augment the tabulations by adding information about the local housebuilding target and the extent of the housing land supply on this particular definition. For those National Parks and the Broads Authority with appropriate plans this would clearly be possible in principle. As local planning authorities themselves, they are required within their local plans to set targets for the plan period, to commute these to annual targets, and then to identify or 'allocate' sites for development. Thus, for example, Yorkshire Dales NPA (2012) allocates land for 231 units over the period 2012-2025 implying a (very low) build rate of 0.17%, which might be compared with the rate of 0.55% realised in the inter-censal decade (Table 1 Column C). For each of these authorities it would therefore be possible to tabulate the number of years supply of land alongside the type of output information in Table 1. For the AONBs this is not possible, even in principle, as they have no targets of their own; instead planning proceeds by reference to the targets for the various LPAs in which they sit.
2.14 From the pipeline perspective, these considerations open up two problems of practical importance. First, it must be recognized that many units actually built, often the majority of such units, are not in fact accommodated on allocated sites but on „windfall‟ or „adventitious‟ sites. One implication of this is that number of dwellings actually constructed may be substantially greater than targets, and another is that prudence (represented in the guidance on making windfall allowances within NPPF) may dictate that land is released by an LPA for housing development although it may become clear subsequently post-hoc that this was not required. LUCS is potentially very useful in monitoring this situation.
2.15 The second problem - and one of great practical significance at present- is that where an LPA cannot identify a sufficient supply of housing land, this may be used to justify housing development on appeal, and this might include land within or adjoining an AONB. Viewed from this perspective, monitoring appeal decisions becomes important, first because it reveals the changing way in which potentially conflicting policy objectives are reconciled (which must guide the future practices of LPA). Second, it shows where fundamentally irreversible changes in the incidence of development occur. In the most straightforward cases, the principle at stake on appeal may simply be that of allowing residential development on sites already included within a development plan. Thus, permission was granted on appeal for housing at a site at Commonhead (Swindon) already allocated as a part of an urban expansion scheme (although its limits came within 100 metres of the boundary of the North Wessex Downs AONB). The key principle embodied in upstream decisions was merely reaffirmed.
2.16 Far more frequently, the matters of principle are significant policy objectives which in particular circumstances conflict. Most usually, these are the need to provide more housing where the supply of available land is inadequate relative to target; and the need to avoid harm to a protected landscape (or more usually) to its setting -which may be the subject of specific policies in some development plans). Sometimes, as in a recent case at Chipping Camden in the Cotswolds AONB, the principle of protecting the AONB (in accordance with para 115 of NPPF) has prevailed. More frequently, however, recent decisions have accorded greater emphasis to shortfalls in the five supply of housing land (in accordance with para 47 of NPPF). Thus consent was granted on appeal to Redrow Homes for 120 dwellings outside the settlement boundary of Winchcombe, the inspector responding to „serious short-term housing land supply deficit‟ in Tewkesbury district with only 2.7 years supply, while acknowledging the prospect of „significant harm to the setting of the [same] AONB‟, „contrary to development plan policy‟.
2.17 Similarly, in February 2013 at Tetbury in Cotswold District, a rural town washed over by AONB designation, two separate housing schemes of 250 and 39 units were allowed on appeal, primacy again being given to the inadequacy of the five-year land supply although the inspector and the Secretary of State agreed that the proposed development would fundamentally conflict with the adopted development plan. Again, the critical issue in determining the appeal was the adequacy of the five-year land supply across the local planning authority’s jurisdiction.
The Character of Housing Development in Protected Landscapes
2.18 Appeal decisions such as the Tetbury case are not only newsworthy, but are important both through their influence upon subsequent planning decisions and because they may in themselves have significant and irreversible implications for a particular protected landscape. Nevertheless, it is important to place such decisions in the context of the general pattern of development within protected landscapes, and to recognise that they are exceptional. This is not to diminish their significance, or to suggest that AONB partnerships and others should not engage with planning authorities to temper development of this kind.
2.19 Given that popular perceptions of the manner in which new housebuilding is accommodated tend to focus on the expansion of the urban area through large scale development at its fringe, it is important to attempt to understand its real significance and to weigh it alongside other forms of housing development which if less contentious may in aggregate be more important. The popular “urban expansion” stereotype does not fit well with the evidence shown in the tabulations for two reasons. First, over the last decade the rate of development in England as a whole has not only been historically low, but three-fifths of new housing development has been concentrated within urban areas (with more than 10,000 residents on their 2001 footprint-see Table S1). With a few notable exceptions such as Swindon, expansion of urban areas has been very modest. The second is that over the entire period since 1985 development either on larger sites and / or at the urban fringe has very much been the exception in protected landscapes.
2.20 Tables 2 and 3 (in the main report) attempt to distinguish those elements of change within protected landscapes involving growth at urban fringes from intensification of urban areas and from those which thicken the pattern of building dispersed more widely through the landscape. Obviously, because of their timing, the effect of the types of decision made at Tetbury are not reflected in the Tables, but because the developments actually fall within the AONB they would subsequently be reflected in such tables (This would not be true of the development in the pipeline at Commonhead (Swindon) or Winchcombe which affects only the setting).
2.21 The generally limited extent of urban fringe development is suggested in Table 3. This is of course to be expected simply because across most of the protected landscapes, the level of urbanization is very low (defined here as the extent of the urbanized area relative to the entire designated area and estimated in Table 3 by the generalized urban area (2)). Moreover there has generally been little urban fringe development. Table 3 shows for each protected landscape both the absolute increase in urban extent, and an annual rate of increase. Attention should be given to both because extreme rates of growth may be found where stocks are very small and these may mislead. Where no part of a protected landscape met the generalised urban area criterion in 2001, but some patch of it came to meet that criterion over the decade, arithmetically the rate would be infinite (indicated by ! in the Table) however small that patch might be. The rate would be grossly misleading as in indicator of the substantive importance of the volume of development. In the Peak District NP there was a very modest increase in urban extent, which implies arithmetically an extreme rate given the minimal level of urbanization in 2001.
2.22 Figure 5 shows absolute and relative measures of urban expansion from Table 3 together (excluding the Peak District and those protected landscapes with infinite rates of urban growth). It is not intended to represent a systematic relationship comparable to that between, say, rainfall and a measure of vegetational diversity such as NDVI or between the demand for a good and its price, but might best be thought of as a 'property space' diagram within which groups of similar or dissimilar protected landscapes might be identified. Most protected landscapes 'cluster' around A, showing minimal urban expansion in absolute or relative terms, while no instances are found near B. It is perhaps more difficult to assess the implications of the values associated with C or D. Table S1: Construction and Net Change in Dwelling Stock by Settlement Type; England, 2001-2011
Share Build Net of Share Locale Stock Stock Net LUCS LUCS /Net Build Change LUCS of Net Type 2001 2011 Change Built Adjusted Change Rate Rate Build Change 000s 000s 000s 000s 000s %pa %pa % % (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11)
Urban 16,398 17,506 1,108 849 936 0.8 0.5 0.6 62.3 60.0 Town 1,597 1,792 196 156 172 0.9 1.0 1.1 11.5 10.6 Fringe 783 902 119 96 106 0.9 1.2 1.4 7.0 6.4 Periurban 363 538 175 132 146 0.8 3.7 4.4 9.7 9.5 Village 1,420 1,547 127 92 102 0.8 0.7 0.8 6.8 6.9 Dispersed 407 530 123 36 40 0.3 0.9 2.7 2.7 6.6
Total 20,967 22,814 1,847 1,363 1,502 0.8 0.7 0.8 100.0 100.0 Notes 1 Urban- Urban areas with population>=10,000; Town, Fringe, Periurban as per RUC2001, Village includes Village Envelope; dispersed includes Isolated farms, Hamlets and all other locales 2 Estimated as residential delivery points from PAF QII 2001 3 Estimated as SCUOS from PAF QII 2011 4 Column 3-Column2 5 This column provides an estimate of new residential construction implied directly by LUCS. (See Explanatory Note) 6 This inflates Col 5 by 10.2% to bring it into line with DCLG completion statistics, compensating for recording lags 7 This column expresses new residential construction as a proportion of net change in dwellings (Col 6/ Col 4) 8 Straight line building rate ie the rate at which dwellings were built per annum as a percentage of stock (0.7%). 9 Straight line rate of net change in stock ie overall change per annum as a percentage of stock (taking account not only of building but new units generated through conversion and intensification net of losses through demolition etc
Very close to C where rates are high and volumes tiny, if there is cause for concern this must lie in the nature of the character and setting of the individual developments. At D where rates are relatively high and absolute volumes more substantial there may be more cause for concern.
Figure 5: Urban Expansion in Protected landscapes: Absolute and Relative; 2001-2011
C B
D
A
2.23 When both absolute and relative values are considered however, attention is drawn to those areas to the right of C. The Kent Downs AONB stands out as an area which has experienced major, but expected, change, showing a rate of urban growth of almost 10%.pa, with the urban area spreading by 14 ha over the decade (see Column 5). Column 4 of Table 3 also estimates that the urban area within this AONB increased by 112ha overall through that period (i.e. 1.12 square kms). The reason for that increase is that during the inter-censal decade the urban threshold (on this definition) was crossed at Hawkinge, and hence a figure of 98ha is recorded in column 6 of Table 3. The spread component thus represents incremental growth in Hawkinge over the decade (anticipated in the Kent and Medway Structure Plan; Future Policy Directions; March of 2002 which described it as an area for strategic expansion ) , while the threshold component reflects expansion in the period since 1990 when it was identified as a growth point). Hawkinge is the only locality within the protected landscapes considered where the development pattern approximates the popular stereotype. In contrast the high growth rate of the generalized urban area within Cornwall AONB (largely abutting St Ives) must be understood relative to the extremely low base, and assessed with the small absolute area of growth in mind.
2.24 The Cotswolds AONB plots below point D and merits attention because it has the largest volume of additional urbanization in absolute terms, and the rate is on a par with that typical over the decade of many small towns away from protected landscapes, and rather higher than that of the South Downs NP. Over the period the Cotswolds AONB showed some urban fringe development weakly paralleling the popular stereotype, with further development clustered on the on the fringe of Bourton-on-the-Water (which remained below the urban threshold throughout the decade). This latter development appears consistent with the Cotswold District Local Plan 2001-2011 (Chapter 8 The Most Sustainable Principal Settlements. Adopted April 2006).
2.25 A closely-related aspect of the stereotypical image of new housebuilding which proves markedly at variance with recent experience is the assumption that it typically involves conversion of greenfield sites. One early understanding allowed by introduction of LUCS was that towards the end of the 1980s just under half of all new dwellings were being built on previously developed sites. With an increasing „urban emphasis‟ in planning policy, this proportion increased through the 1990s. Despite widespread belief that it was unrealistic to imagine that by 2006 60% of all new dwellings could be built on brownfield sites, analyses of LUCS suggested that at national level this should easily achievable (Bibby and Shepherd, 1999). Generally, the flow of land coming forward may be highly uneven over time and space, responding to shifts in the nature of economic activity as particular uses are abandoned, producing reserves of land available for re-use. The volume of brownfield land being brought forward for re-development depends to a considerable degree on the scale of the existing building stock, implying an obvious but overlooked relationship between the proportion of new dwellings accommodated on brownfield land and variations in settlement structure. This general pattern is clear in Figure 6a, and implies that outcome proportions, between greenfield and brownfield, should not be treated in any simple way as a measure of the performance of planning authorities.
2.26 An important implication of this type of analysis is that even if such a target is achieved nationally, a similar figure should not be expected in the protected landscapes, because of the limited building stock. Nevertheless, if demand is high and costs of redevelopment modest (relative to the value of assets created), there will be a flow of previously developed land albeit vacant land stocks will be low. Throughout the period since 1985 roughly half the houses built in the protected landscapes have occupied previously developed sites even though the stock of previously developed land within the National Parks available for development amounted to only 161 hectares, or one hectare for every 100 square kilometres (10,000ha) (NLUD-PDL 2004). As central government pressure to conserve greenfield land intensified across the country, concern to avoid development on greenfield sites in protected landscapes was sustained. Between 2000 and 2010, the proportion of new dwellings built on previously developed land within protected landscapes increased only very marginally (reaching 52.3% over the decade as opposed to 49.7% over the entire period).
2.27 In considering the role of brownfield land, it is also important to distinguish the percentage of all new dwellings built on previously developed sites (a quantity carefully measured for much of the decade), from the proportion of all land used for housebuilding that had been developed previously (a quantity rarely discussed). The difference depends on the actual pattern of development achieved, with dwelling density the key mediating variable. An emphasis on building smaller units within urban areas has driven up both the density of new construction and the proportion of dwellings accommodated on brownfield sites, reflected in the parallels between Figure 6a and 6b. The effect of the emphasis on urban intensification in central government planning policy was relatively limited away from the major urban areas. Over the entire period from 1985-2011 63.3% of the land used to accommodate new dwellings within protected landscapes had not previously been developed and this proportion remained at 61.9% in the final decade. While comparison of Table 6 a and b indicates a fall in the rate at which greenfield land was converted to housing purposes in the protected landscapes after 2000, this was almost entirely attributable to a fall in housing output rather than to a greater use of previously developed sites.
2.28 Consideration of the actual and potential role of „brownfield‟ sites must also take account of the very different scale and character of the sites vacated as uses are abandoned. In addition to the relatively small developed sites implicit in the discussion above, there are within the protected landscapes very extensive sites such as former military airfields which are typically relatively easy to develop. The future use of reserves of previously developed land of this type is above all else a matter of policy (Hawkinge in the North Kent Downs AONB provides the best example of a circumstance in which policy has favoured re-development; Rendlesham (former RAF Bentwaters) providing an example of where policy is currently doing so). It is also necessary to take account of classes of previously developed land including sites such as former minerals operations which are both extensive and extremely difficult to redevelop. Within the PLs these include Blackaller Quarry in the Dartmoor NP, the Devon Great Consols Mining Complex (with arsenic contamination) in the Tamar Valley AONB and Shoreham cement works at Upper Beeding, in the South Downs NP.
2.29 A final misleading aspect of the popular urban expansion stereotype when applied to the inter-censal decade is its emphasis on large sites. Across England generally, the overwhelming majority (90%) of housing schemes involve building less than ten houses. Admittedly, this does not of itself necessarily imply that the few larger schemes might not still accommodate a markedly disproportionate share of all dwellings built. In many circumstances there is merit in the Pareto principle (or 80:20 rule) which in this particular case might posit that 20% of the sites might be expected to accommodate 80% of all the houses built and hence justify attention on the largest schemes. Indeed, the pattern of development across England as whole provides some weak justification for this type of view (Table S2 indicates that although half of all English housing sites had 2 units or less, half of Over the entire period from 1985-2011 63.3% of the land used to accommodate new all dwellings were built on sites of 22 units or more).
2.30 Crucially, however, within protected landscapes, not only is it the case that the overwhelming majority of sites are small; it is also true that the overwhelming majority of new dwellings are built on small sites. LUCS indicates that over the inter-censal decade 14,270 dwellings were built in protected landscapes- spread across 6,623 sites, i.e. with an average of 2.2 units per site. No LUCS land parcel within an AONB accommodated more than 9 units (3). Decisions
Figure 6a Percentage if New Dwellings Built on Figure 6b Density of New Residential Development Brownfield Sites, England 2001 -2011 England 2001-2011 (Dwellings per Hectare)
England 2001-2011 Table S2a: Residential Site Sizes 2001-2011; Unweighted
England PLs Hectares Units Hectares Units Mean 0.3 9.3 0.2 2.2 Median 0.1 4 0.1 1 Mode 0.1 1 0.1 1
Percentiles 10 0.1 1 0.1 1 20 0.1 1 0.1 1 30 0.1 1 0.1 1 40 0.1 1 0.1 1 50 0.1 2 0.1 1 60 0.2 3 0.2 1 70 0.2 5 0.2 2 80 0.3 8 0.2 3 90 0.6 16 0.4 5
Table S2b: Residential Site Sizes 2001-2011; Weighted by Units
England PLs Hectares Units Hectares Units Mean 0.6 44.8 0.2 3.9 Median 0.3 22 0.1 3 Mode 0.1 1 0.1 1
Percentiles 10 0.1 4 0.1 1 20 0.1 8 0.1 1 30 0.2 12 0.1 2 40 0.3 17 0.1 2 50 0.3 22 0.1 3 60 0.5 30 0.2 4 70 0.6 40 0.2 6 80 0.8 58 0.3 7 90 1.3 100 0.4 8
A percentile is the value below which a given percentage of instances fall. Thus in Table S2a 50% of all sites across England as a whole accommodated 2 dwellings or less and 70% of all sites accommodated 5 dwellings or less. Weighting as in Table S2b shifts the focus from sites to dwellings. In the calculation, each site is weighted by the number of dwellings on it. It shows that across England as a whole, 50% of dwellings were built on sites accommodating 22 houses or less, and that 70% of dwellings were built on sites with 40 houses or less. While the number of units on the 'average site' was 9.3, the 'average unit' was built on a site with 44.8 units. on such schemes (along with 90 % of planning decisions generally (4)) are delegated to officers, consistent with central government‟s guidance. In National Parks the proportion of decisions delegated at 84% overall (5) is slightly lower. In AONBs delegation levels also appear to be broadly similar, where statistics have been generated e.g. Chilterns, South Devon. Thus it is clear that not only housing classed as major development (sensu NPPF para 116) is very rare but high levels of delegation imply that most housing developments are not contentious.
2.31 Reflection of the inapplicability of the popular stereotype is important not so much because it might play down the importance of newsworthy cases such as the Tetbury appeals , but because there may be a tendency not to give much consideration to the cumulative effect of the flow of typically uncontentious minor developments. Tables 6a and 6b provide indications of characteristics of housing sites in protected landscapes other than their size. The density at which new dwellings are built- referred to as the plot density on the tabulations- is an important characteristic. Following revision of PPG3 in 2000 government sought to secure development at densities of 30 dwellings or more to the hectare, and across England as a whole LUCS shows that an average of 35.02 dph was achieved in the inter- censal decade. The development pattern in the protected landscapes was entirely different (with an overall mean development density of 11.13 dph over the decade). In part this reflects a general differential between urban and rural areas that persisted even as policy changed (see Bibby and Brindley 2007). The lower densities should be understood not only in relation to the larger dwellings typical of protected landscapes; but also to the larger curtilages formed where sites for single dwellings are carved out of agricultural parcels. The combination of low density and small sites typifies forms of development quite unlike those associated with volume housebuilders‟ agricultural parcels for development. Within the protected landscapes, the volume housebuilders,who derive economies of scale from building to a standard plan, are under-represented. Instead it is local companies building to order or producing more expensive bespoke houses (cf Satsangi et al, 2010) are typical ; this is further picked up by ONS returns on house prices as part of an analysis of affordability undertaken at district level by DCLG. In simple terms it might also be inferred that demand will also stimulate relatively higher quality development as this will bring in greater returns in a buoyant market. Other planning factors such as provision of affordable homes, subsidised by market housing, will have an effect on supply especially on larger developments.
2.32 The pattern of variation in average plot densities between the various protected landscapes shown in Table 6b should be understood primarily in relation to differences in the mix of types of site. Across England as a whole plot densities vary in accordance with settlement structure and subregional price (see Bibby and Brindley 2006). While the protected landscapes as a whole show average densities markedly lower than the national average there is a mix of densities within each, and variations between PLs reflect difference in the mix of development forms found. Thus, Column 6 of Table 6b shows higher mean densities in several of the coastal protected landscapes (Cornwall, Dorset, Norfolk Coast, North Devon, Northumberland Coast and South Devon AONBs) which appears consistent with a larger proportion of development being devoted to holiday accommodation. This appears to be corroborated by Table S3 which indicates that within several coastal PLs the increase in habitable units is markedly higher than the growth in permanent households.
Figure 6a Percentage if New Dwellings Built on Figure 6b Density of New Residential Development Brownfield Sites, England 2001 -2011 England 2001-2011 (Dwellings per Hectare)
England 2001-2011 2.33 The planning issues which typically arise in bringing forward bespoke development on small sites typically differ from those on sites which fit the popular stereotype. Before the application stage, landowners may identify opportunities for development of „rural exception sites'. Planning appeals tend to be resolved by written representations and are unlikely to involve the clear conflicts of principle discussed above. Where new development is concerned, applications - most usually not on sites allocated within local plans- may be supported by claims that an additional dwelling would be critical to the success of a rural enterprise. These need not be limited to dwellings said to be essential to agriculture or forestry. Other classes of appeal frequently found in protected landscapes are against the LPAs refusal to remove an agricultural occupancy condition, or refusal to grant a certificate of lawful development for a structure already built and with a view to being able to bring a potential residential unit into the general market.
2.34 Crucially, in contrast to cases such as Tetbury discussed above, the outcome of appeals on sites typical of protected landscapes is unlikely to alter the pattern of development fundamentally. The grant of planning permission for new construction on the sites typical of protected landscapes- on appeal or otherwise- tends to thicken the scatter of buildings dispersed through the landscape rather than to expand significant settlements or involve development at entirely new locations. The flow of new units created is best understood alongside construction achieved through the exercise of permitted development rights together with conversion and intensification.
3 Interpreting the Tabulations: Extra Housing Units Without New Construction
3.1 The flow of dwellings arising from new construction, although offset to some degree by demolitions may be potentially augmented by a flow of units arising from intensification of use and conversion of existing property. In the tabulations the net effect of these adjustments is gauged simply by reference to difference between the number of dwellings built over a period and the net change in dwellings over the same period (see Table 1 in which the overall effect over the inter-censal decade is evident in Column 6; expressed as a rate relative to the 2001 stock in Column 7). Nationally the key figures are 0.7% per annum, with a net (adjusted) rate of 0.88% per annum.
3.2 The net adjustment shown in Column 6 is the outcome of a series of a myriad of minor actions. Non-residential property such as farm buildings may be converted to residential use, but of course dwellings are sometimes converted to non-residential use. Moreover residential property within the same areas may be subdivided creating additional units or knocked together to create larger properties. This continuous modification of the housing stock is highly adaptive and is modulated by the actual pattern of demand. Intensification – that is the creation of an increased number of dwellings within buildings already within the landscape- obviously reflects demand. Disintensification may also reflect demand, but demand for larger units. In areas of low demand, however, housing may be converted simply for storage, In the protected landscapes intensification of this type with or without extension of existing buildings provides a substantial addition to the flow of additional units. Thus although the share of new housebuilding accommodated within England‟s AONBs is lower than its share of population, its share of the net increase in dwelling units in the inter-censal decade appeared slightly higher than average. This can be appreciated by comparing the rates of new build and of net change in various settlement types shown in Table S2 with those for protected landscapes shown in Table 1 and Table S3.
3.3 It is also clear that within protected landscapes the aggregate contribution of these sources of additional units may exceed the flow of newly-built dwellings. Hence while the policy target rate for the current plan period for the Yorkshire Dales is much lower than that for the 10 years leading up to the 2011 Census, 1300 units were added by net change (See Table 1) i.e. largely due to intensified use of existing buildings. This relationship between a low volume of housebuilding and a larger flow of new housing units arising from other sources is not unusual in the protected landscapes. In several national parks all characterized by low rates of housebuilding (Exmoor, the Lake District, The North York Moors, Northumberland and the Peak District) these “intensification” adjustments together accounted for the stock of units provided within existing buildings to come forward at a rate of more than 1% per annum. The viability of conversion and intensification schemes depends on the relationship between the cost of a potential scheme and the value of the property created. This value depends on reigning house prices and so is greater in areas of high residential desirability and should be expected to be higher where new building is restricted, not just in protected landscapes but in rural areas generally. The character of the inherited building stock itself influences the scale of the flow (the farm plan forms of the Forest of Bowland for example being particularly amenable to conversion). Factors such as these help to explain the particular relation between housebuilding and net change characteristic of the Forest of Bowland and Nidderdale AONBs.
3.4 The planning system obviously influences the nature of these adjustments, although monitoring is difficult and there is little systematic analysis or discussion of its effect on the flow of additional units. Owner occupiers of residential property and agricultural holdings may seek to bring forward a flow of additional habitable units by adapting, extending and subdividing the stock of buildings dispersed through the landscape, augmenting it through the addition of new units required to support rural enterprise, rural exception sites, and construction sanctioned by permitted development (PD) rights to serve an agricultural purpose. Their comparatively long-term perspective allows them to pursue creative sequences of development. Although subdivision of dwellings explicitly requires planning consent under statute, they may choose to defer applications to appropriate times, to exploit current PD rights creatively (6), to make repeated applications, to rely on the possibility of securing post-hoc certificates of lawful development (7), and to accept restrictive occupancy conditions which might subsequently be removed. All ultimately contribute to the same purpose of creating a supply of units attuned to market demand.
3.5 Corresponding to these possibilities comes a constellation of planning appeal decisions and even case law. Examination of cases reveals the flexibility with which the building stock may be used, the elasticity of distinctions between conversion and new building and ,by implication, the limitations of planning controls. Appeals regarding use of existing buildings might concern the proportion of a building retained in a 'conversion' or whether the removal of agricultural occupancy conditions would be justified. Documents relating to appeals against enforcement action against unauthorized subdivision in the Yorkshire Dales NP reveal the difficulty of counting dwellings where a property is marketed as available for rent either as one 7 bedroomed dwelling, or as two separate 3 and 4 bedroomed properties, or when the same building is occupied as differing numbers of dwellings in different seasons.(8). 3.6 In the period covered by the tabulations, intensifications and conversions across the protected landscapes appears to have exceeded the number of newly-built properties. The viability of conversion has apparently been supported by the paucity of new building. Planning policy may thus be considered to have a direct impact restricting new construction, an indirect impact stimulating adaptations, and a more subtle indirect impact on the pattern of development. Table 1 of the main report, as discussed more fully in 5.4 below, suggests the scale of the first two effects. Anecdotal evidence might suggests the variety of routes that might be used to bring forward additional residential units, and for those unfamiliar with such material the history of a barn proposal at Blubberhouses in the Nidderdale AONB may provide insight (9).
3.7 Overall, the evidence of the tabulations suggests the aggregate significance of the range of practices which are the subject of anecdotal discussion and which can be illuminated by reference to appeal materials. In contrast to substantial developments at the urban fringe where strong control is exercised by the planning system (albeit potentially subject to catastrophic effects on appeal where goals conflict), it seems that the planning system limits the flow of additional units brought forward in a particular period but with limited effects on its spatial pattern.
Other Adjustment: Implications
3.8 The environmental significance of this little-discussed pattern of adaptation within protected landscapes is not clear. Apart from the effect on existing buildings themselves, residential conversion and subdivision intensifies the use of land by increasing the number of households present, which is likely to imply physical change - modification of curtilages, provision of garden space, bin storage, drying space, and additional parking space and to impose further pressure on such infrastructure as exists. Beyond these obvious points, such adaptations allow the role of place to change as the balance of temporary and permanent occupation shifts, and commuting patterns adjust. Moreover, land management practices change as the existing building stock is severed from its setting to a degree that depends in part on the settlement context of the new units and on seasonal pattern of occupancy.
3.9 As a prelude to such a discussion it may be useful to move beyond the principal tabulations (which cannot distinguish holiday lets from other units for example) to consider the numbers of people and households associated with the changes reported for protected landscapes in England. Table S3a brings together, for the first time, information from the main tabulations and published material from the 2011 Census for the National Parks. Similar Census tabulations for the AONBs are not available and so Table S3b draws on information from the ONS „Headcounts and Household Estimates for Postcodes in England and Wales‟ from the 2011 Census (and its 2001 counterpart) for this purpose. These estimates fit the protected landscapes exactly but provide only the most basic information (total males, total females, total households) for every unit (full) postcode with „usual residents‟ on Census night.
3.10 Tables S3 a and b suggest how the relationship between population change and the physical changes reported in the principal tabulations varied between different protected landscapes in England. The simplest relationship is found in the Kent Downs where the physical indicators in combination highlight the significance of settlement expansion at Hawkinge, reflected here in a high rate of population growth. In contrast to the other areas, the rate of population growth approaches the rate of household growth The Kent Downs figures reflect not only new construction over the decade but the maturation of families settling since implementation of the growth strategy.
3,11 More generally across England, numbers of households have grown as household size has fallen. Household size has fallen gradually as people have sought more living space and more independence (demanding of course, gradual but sustained growth in real income). It is household growth of this type rather than population growth per se that exerts a very important influence on housing demand. Furthermore it is clear from ONS data that the demographics of protected landscapes point to relatively older and wealthier communities, typified by small households. Most of the protected landscapes have therefore experienced household growth. The particular growth profile of the Kent Downs AONB is atypical of protected landscapes. It thus contrasts with that of the Shropshire Hills where population growth over the decade was relatively high, but much lower than household growth.
3.12 The Shropshire Hills AONB exemplifies a distinctive pattern. Although there was some growth at the urban fringe (see Table 3 of the Tabulations), taken together, the tabulations point to expansion of the dwelling stock in the Shropshire Hills AONB achieved not primarily through the expansion of substantial settlements by way of sustained housebuilding, but largely through conversion and intensification of buildings already dispersed within the landscape for occupation predominantly by permanent residents. There are similarities with the patterns of change found in the Forest of Bowland, the High Weald and Nidderdale AONBs. Nidderdale is distinctive in its high rate of growth of dwelling stock, households and population. In these contexts, the relation of outcomes to planning strategy is far less obvious, but they should perhaps be understood in terms of the interaction between high demand and policy restraint
3.13 Overall, therefore the basic tabulations point towards a distinction between two patterns of change incipient in the foregoing discussion, one associated with settlement expansion through volume housebuilding and the other shaped by diverse marginal changes and additions to the existing building stock. Table 1 draws attention to the gap between housebuilding and net additions to the dwelling stock, while Tables 2 and 3 demonstrate the limited significance within protected landscapes of residential development at the urban fringe. Moreover, while there was continued pressure to conserve greenfield land, the specific forms and circumstances of development captured by Table 6 - with rather lower proportions of new units accommodated on greenfield sites than applies nationally, and much lower development densities. These differences are closely linked to the particular size distribution of sites coming forward for development in protected landscapes illustrated in Table S2. It is, however, not sufficient to focus on these differences in development patterns alone. Consideration of Tables S3a and b suggests the importance of paying great attention to the demand for dwelling space for seasonal, weekend or occasional use by owner occupiers or short-term tenants.
3.14 Demand for property for second homes or for holiday letting underlies a marked difference in the relationships between numbers of households and of residential units found in the protected landscapes and those typical of England more generally. The coincidence within the Cornwall AONB over the inter-censal decade of substantial expansion of the stock of residential units (secured primarily through adaptation) and falls both in population and.
Households per Table S3a Stock Change Households Change Population Change Dwelling National Park 2001 2011 % 2001 2011 % 2001 2011 % 2001 2011
Dartmoor 13,950 15,339 1.00 13,757 14,245 0.35 33,552 33,596 0.01 0.99 0.93
Exmoor 4,833 5,502 1.38 4,896 4,678 -0.45 10,873 10,273 -0.55 1.01 0.85
Lake District 20,621 23,294 1.30 17,937 18,112 0.10 41,831 40,770 -0.25 0.87 0.78
New Forest 14,538 15,347 0.56 14,233 14,536 0.21 34,000 34,922 0.27 0.98 0.95
North York Moors 10,310 11,766 1.41 10,084 10,342 0.26 23,939 23,380 -0.23 0.98 0.88
Northumberland 905 1,028 1.36 799 871 0.90 1,936 1,993 0.29 0.88 0.85
Peak District 15,342 17,769 1.58 15,949 16,461 0.32 37,937 37,905 -0.01 1.04 0.93
South Downs 47,575 50,988 0.72 46,215 47,273 0.23 105,200 112,343 0.68 0.97 0.93
The Broads Authority 3,059 3,436 1.23 2,624 2,931 1.17 5,876 6,271 0.67 0.86 0.85
Yorkshire Dales 9,254 10,554 1.40 8,363 8,804 0.53 19,654 19,761 0.05 0.90 0.83
All NPs England 140,387 155,023 1.04 134,857 138,253 0.25 314,798 321,214 0.20 0.96 0.89
England 20,967,000 22,814,000 0.88 20,451,427 22,063,368 0.79 48,248,150 52,059,931 0.79 0.98 0.97
Stock Households Change Population Change Households Per Dwelling Table S3b Change AONB 2001 2011 %pa 2001 2011 %pa 2001 2011 %pa 2001 2011 Arnside & Silverdale 3642 3873 0.63 2555 2558 0.01 7610 7680 0.09 0.70 0.66 Blackdown Hills 4899 6018 2.28 4408 5047 1.45 12906 13374 0.36 0.90 0.84 Cannock Chase 2401 2515 0.47 1686 1711 0.15 5994 6110 0.19 0.70 0.68 Chichester Harbour 4126 4198 0.17 2437 2498 0.25 8120 8128 0.01 0.59 0.60 Chilterns 32445 33845 0.43 22958 24688 0.75 81756 82606 0.10 0.71 0.73 Cornwall 26108 29075 1.14 16724 16500 -0.13 51006 49861 -0.22 0.64 0.57 Cotswolds 61211 65588 0.72 40335 43362 0.75 139714 142590 0.21 0.66 0.66 Cranborne Chase & West Wiltshire 13430 14848 1.06 9780 10726 0.97 32778 33679 0.27 0.73 0.72 Dedham Vale 2826 3004 0.63 2173 2327 0.71 6550 6686 0.21 0.77 0.77 Dorset 33593 37661 1.21 18954 20347 0.73 69744 71973 0.32 0.56 0.54 East Devon 9234 10246 1.10 5947 6464 0.87 19318 20479 0.60 0.64 0.63 Forest of Bowland 5689 6920 2.16 4477 5458 2.19 15599 16244 0.41 0.79 0.79 High Weald 47043 51634 0.98 33850 38595 1.40 118074 125780 0.65 0.72 0.75 Howardian Hills 2013 2217 1.01 1631 1687 0.34 5465 5474 0.02 0.81 0.76 Isle Of Wight 5003 5393 0.78 3575 3622 0.13 10715 10444 -0.25 0.71 0.67 Kent Downs 27733 30702 1.07 19660 22133 1.26 66534 73308 1.02 0.71 0.72 Lincolnshire Wolds 3994 4569 1.44 3392 3679 0.85 9871 10089 0.22 0.85 0.81 Malvern Hills 4647 4882 0.51 3275 3723 1.37 11011 10942 -0.06 0.70 0.76 Mendip Hills 3009 3358 1.16 2501 2752 1.00 8114 8095 -0.02 0.83 0.82 Nidderdale 4261 5380 2.63 3204 3948 2.32 11525 12274 0.65 0.75 0.73 Norfolk Coast 10658 11652 0.93 6652 6877 0.34 18779 18419 -0.19 0.62 0.59 North Devon 4593 5609 2.21 2766 2884 0.43 9561 9986 0.44 0.60 0.51 North Pennines 5176 6048 1.68 3926 4268 0.87 11973 12099 0.11 0.76 0.71 North Wessex Downs 38017 41609 0.94 26483 29004 0.95 93657 98889 0.56 0.70 0.70 Northumberland Coast 3346 3992 1.93 1457 1597 0.96 5263 5496 0.44 0.44 0.40 Quantock Hills 1161 1352 1.65 1103 1173 0.63 2927 2874 -0.18 0.95 0.87 Shropshire Hills 8036 9169 1.41 5883 7264 2.35 18381 19968 0.86 0.73 0.79 Solway Coast 1404 1558 1.10 908 907 -0.01 3199 3040 -0.50 0.65 0.58 South Devon 17185 19034 1.08 8404 8779 0.45 30877 31274 0.13 0.49 0.46 Suffolk Coast & Heaths 10871 11678 0.74 6107 6228 0.20 20353 20154 -0.10 0.56 0.53 Surrey Hills 15297 16135 0.55 11715 12908 1.02 39760 40764 0.25 0.77 0.80 Tamar Valley 5415 5901 0.90 3559 3624 0.18 12750 12744 0.00 0.66 0.61 Wye Valley 10859 11993 1.04 7172 8420 1.74 24998 26678 0.67 0.66 0.70 All AONB's England 429325 471656 0.99 289,657 315758 0.90 984,882 1,018,201 0.34 0.67 0.67
England 20,967,000 22,814,000 0.88 20,451,427 22,063,368 0.79 48,248,15 52,059,931 0.79 0.98 0.97
resident households is particularly marked implying an increase in holiday lets, second homes or similar accommodation. The combination of low household growth and fall in population is also found in the Isle of Wight, the Solway Coast and less markedly the Norfolk Coast and the Suffolk Coast and Heaths AONBs. More generally, within the AONBs and particularly the coastal ones, there appears to have been a tendency for the dwelling stock to have grown more rapidly than the number of households. More obvious cases aside, it appears that in the Quantocks - though the scale of activity was small- development provided opportunities for non-permanent residents
3.13 These preliminary remarks only begin to open up the area of discussion, but might be usefully extended by developing the logic suggested here to improve understanding of the incidence of second homes, and particularly to improve assessments of the numbers within protected landscapes. The Census - in part because of its overall design- only tabulates second home locations at local authority district level(10). Use of the „Headcounts and Household Estimates for Postcodes‟ together with information from LUCS and PAF might potentially allow more precise examination of the likely incidence of second homes within protected landscapes.
4 Interpreting the Tables: Non Residential Development
4.1 Previous sections demonstrate that bringing the pipeline metaphor to bear on LUCS data (supported by PAF) provides a useful approach to monitoring residential development. It might hardly seem necessary to point out that this depends on there being value in aggregating numbers of units built on specific parcels, in aggregating their areas to provide measures both of land developed for housing and of other uses relinquished and also in visualizing these various measures as flows. Treating a large number of relatively small and relatively homogenous housing developments as a flow provides a workable method of tracking matters of particular concern.
4.2 When attention shifts to rural non-residential development, however, this type of approach proves far less likely to be illuminating. The visual impacts of greatest concern frequently cannot be satisfactorily captured by LUCS, the heterogeneity of the developments make simple aggregation far less useful, and reduction to flows produces far less meaningful comparisons between places or over time. This section provides an overview of forms of non- residential development that LUCS cannot adequately measure, and of forms that can be measured but where interpretation of indicators is troublesome, before considering contexts in which LUCS proves more useful.
Non-Residential Development Which LUCS Cannot Measure
4.3 There are at least three types of circumstance where development which has marked visual impact is not adequately captured by LUCS. These occur where
the development does not generate a topographic change, the visual impact of developments is disproportionate to their aggregate area, or the visual impact is obscured when areas are aggregated
4.4 LUCS records are only created where there is a „topographic change‟ in a specific sense used by Ordnance Survey - that is to say a change in linework or annotation on survey documents at the largest possible scale (usually 1:1250). Some visually intrusive features are not mapped, hence are not recorded on LUCS. As OS survey practice records „buildings‟ only where it is 'reasonable to assume' they 'will remain in position for at least 10 years' the development of ,say, 'polytunnels' will not lead to creation of LUCS records even where they cover extensive areas. Thus a class of development which has been highly contentious in some locations cannot be recognized within LUCS.
4.5 Reflection on the origins of LUCS as an adjunct to updating the national survey base suggests other forms of development which are likely not to be recorded on LUCS, given the exigencies of survey. Particular forms of extensive non-residential development (in the sense of planning law) may involve alteration of levels, or provision of all-weather exercise areas, or maneges which although having a visual impact on the character of a protected landscape will not necessarily generate a topographic change. Proposals for equestrian development frequently lead to contentious planning applications in protected landscapes in the South East.
4.6 The most obvious example of features which are mapped and do imply topographic changes, but where visual impacts are vastly disproportionate to area are wind turbines. Telecommunication masts and outdoor lighting (of equestrian facilities for example) have similar characteristics. The height of wind turbines rather than their area provides a pointer to the dimension of their intrusiveness, and of course implies concern not only with their location within protected landscapes but their visibility from them. Exploratory work has demonstrated the difficulty of identifying ancillary features associated with existing wind farms recorded in the UK Wind Energy Database within LUCS. Clearly, wind turbines and similar features are classes of development which cannot be monitored effectively using LUCS, in stark contrast to developments such as solar farms, which may be tracked.
4.7 Where concern is with visual impact of non-residential development, the limitations of LUCS do not arise solely because of the absolute scale of the footprint of wind turbines and similar features. The character of a protected landscape may be adversely affected by combination of relatively small developments such as cabins, the aggregated area failing to stand as a proxy for the impact across that tract. Moreover it would be intensely difficult to begin to exploit LUCS data to illuminate impacts of that type. Construction of fisherman‟s cabin at Wylands International Angling Centre in the High Weald AONB serves as an example of development of this sort which has proved highly contentious (FN).
4.8 Finally, it should be appreciated that landscape quality may be adversely affected in circumstances where a change record appears on LUCS but there is no change in the aggregate area. Hence between 1985 and 2011, LUCS records 15,855ha as having changed from A (agriculture) to A (agriculture) - that is almost 160 sq kms involving more than 32,000 separate parcels. These do not involve readily interpretable land-use changes. This may involve fragmentation of fields to form paddocks producing settings more typical of the urban fringe. Concerns that this sort of development is eroding development the landscape pattern of the locality has motivated a Supplementary Planning Document by the New Forest National Park Authority. What LUCS Can Measure, but which is not Locally Determined: Transport Infrastructure, Minerals and Waste
4.9 Other categories of development do generate topographic change, and allow interpretable areal aggregations which are likely to be broadly in step with visual impact. Development of highways and similar infrastructure provide good examples. With caveats, this also applies to minerals-related development. The role of LUCS in monitoring subsurface mineral working – most topically fracking - is limited to visual intrusion into the local setting and the wider landscape caused by any the placement of any building or structure within the application site area (DCLG 2013 para 30). LUCS provides coverage (although it may be subject to long recording lags) of the development and restoration of opencast sites. Over the period since 1985 , 1906ha of land have been converted to minerals uses within the protected landscapes, and 1203ha recovered from such uses (with opencast coal extraction near Telford affecting the Shropshire Hills AONB)
4.10 While LUCS can provide broadly satisfactory measures of these types of development, it is far more difficult to use indicators constructed using these data to monitor the operation of local planning regimes. This is simply because the balance between local control and national policy differs from that which shapes residential development. In the case of major road investment, decisions on priorities and alignments are not made at the level of LPAs, and in response to patterns of travel demand which have to be understood in relation to much broader areas than the protected landscapes. Nevertheless, the concomitant land conversion may account for a substantial share of the area of non-residential development within a protected landscape (underlying high volumes per household in the Northumberland National Park which appear in Table 7a). Over the long period highway or transport development accounted for 18.7% of the area changing to or between developed uses across all the protected landscapes, but stood at 69% in the Northumberland National Park, largely down to MoD developments in the Otterburn Training Area.
4.11 The fundamental premise of minerals planning policy touched on in NPPF para 142 and restated most recently in DCLG (2013 para 6) is that „minerals can only be worked (i.e. extracted) where they naturally occur, so location options for the economically viable and environmentally acceptable extraction of minerals may be limited‟. Although in terms of planning law exploitation of minerals involves development, minerals exploitation is not a developed use and change to minerals uses is not included within the tabulations
Non-Residential Buildings
4.12 LUCS can serve to monitor construction of new non-residential buildings, although heterogeneity of building form and function is far greater than what is possible with residential development. Most critically, there is no quantum analogous to numbers of dwellings, and it can be difficult to relate hectares of development recorded on LUCS to the more familiar measures of built floorspace in square metres. (This is not simply a matter of estimating plot ratios as areas recorded in LUCS as serving a particular use may or may not include building. Thus 0.2ha of „S‟ (storage) might refer to open storage, to a warehouse or even a glider hangar. 4.13 Although beyond the scope of work undertaken to produce the Tabulations, weaknesses of LUCS as a source for monitoring non-residential development can be mitigated, and richer evidence provided by combining it with other sources such as PAF and the Valuation Office Agency‟s Non-Domestic Rating list at the individual parcel level. Analyses of this form undertaken for the former Commission for Rural Communities help to understand the distribution of non-residential development between different types of locale as shown in Table S4. This table strongly suggests that much of the rural domain, and in particular the locale types characteristically found within protected landscapes, are not particularly attractive for new industrial and commercial development. It suggests that while 22.7% of households were to be found in the rural domain, using capital service in buildings (CSB) as a proxy, it accommodated only 13.5% of the capital stock.(See Note 1 to Table S4) Moreover, the part of the rural domain favoured for development is the periurban (where there might be a desire to site business or retail parks) rather than villages or areas of dispersed settlement.
Table S4: Incidence of New Non-Residential Development 2000-2006
Shares (%) Land Developed IVCSB Households CSB (I & C) IVCSB per ha Urban 77.3 86.5 45.2 82 506 Rural 22.7 13.5 54.8 18 90
Fringe 3.9 1.3 3.0 1.4 126 Periurban 2.1 4.7 27.0 9.6 99 Other 1.1 2.3 16.8 2.9 49
Town 7.8 2.7 4.2 2.5 164 Village and Dispersed 7.9 2.5 3.8 1.4 103
Total 100 100 100 100 278 Notes
1. CSB. An area might be thought of as embodying an amount of productive capacity represented by physical capital ie a stock of fixed assets such as buildings and machinery. This stock yields up its value over time and might be thought of as providing each year an input into the production of goods and services in the area with a particular money value. The specific contribution of to this flow might be thought of as the value of capital service in buildings. Although it might seem more natural to consider the capital value of the commercial building stock, this is less relevant to consideration of productivity, and involves valuations that are not available. CSB can, however, be proxied as here by the aggregated annual rateable value in the area.
2 IVCSB. New productive investment entails the construction of factories, warehouses and the range of other commercial facilities which leads to an increase in the annual value of capital service supplied by buildings in an area. This is referred to here as the increase in value of capital service in buildings (IVCSB). It is estimated here by the net increase in aggregate annual rateable value in the area over the period 2000-2006. Just as CSB serves as a flow measure which can take the place of an estimate of the capital value of non-residential buildings in an area, IVCSB measures the increase in that flow and can take the place of an estimate of the capital value of new non-residential buildings. 4.14 While those tendencies undoubtedly seem to suggest less pressure on the protected landscapes, one caveat is necessary. It is also evident from Table S4 that LUCS shows that the area of land developed for industrial and commercial purposes in the rural domain is disproportionate to its share of capital stock (CSB) or investment (as proxied by the increased value of capital service in buildings IVCSB- see Note 2 to Table S4). Although land development is concentrated in the periurban area (which between 2000 and 2006 accommodated more than a quarter of all industrial and commercial development) across the rural domain as a whole more land was developed for any given amount of new non- residential building (as gauged by IVCSB).
4.15 Given the overall pattern indicated in Table S4, the volume of industrial and commercial development to be found in the protected landscapes should be expected to be limited. Table S5 shows summary quanta for industrial and commercial development, other forms of non- residential development, and in addition shows agricultural building.
4.16 Overall, the area being developed in protected landscapes over the long period for those non- residential developed uses shown in Table S5 (C,I,J,K,L,S,T, and U) amounts to about 0.7ha per 1,000 households, per annum (cf household stocks in Table S3). This might be visualised as equivalent to two square metres of floorspace constructed for each household every year. Table S5 provides an indication of the balance between these different forms of non- residential development, insofar as that balance is reflected in aggregate measures of land converted. Areas converted to agricultural buildings (which would not usually be considered a non-residential use) are included to allow comparison. Since 1985 across the protected landscapes taken together 139ha have been developed, or redeveloped, for the non-residential uses shown for every 100ha devoted to agricultural building. While agricultural buildings on this definition includes glasshouses (and related structures associated with garden centres) this should provide some perspective.
Table S5: Non Residential Development: Protected Landscapes, England, 1985-2011 National Parks AONBs Protected Hectares % Hectares % Hectares %
I Industrial 1215 53.3 2574 44.7 3790 47.1 J Commercial 146 6.4 436 7.6 582 7.2 K Retail 24 1.0 122 2.1 146 1.8 L Leisure 89 3.9 244 4.2 334 4.2 S Storage 339 14.9 811 14.1 1151 14.3
H Highways 189 8.3 482 8.4 672 8.4 C Community Buildings 40 1.8 208 3.6 248 3.1 T Transport 117 5.1 508 8.8 624 7.8 U Utilities 118 5.2 377 6.5 495 6.2 2279 100.0 5763 100.0 8041 100.0
B Agricultural Buildings 1665 4117 5782 Ratio NR: Agricultural Building 1.37 1.40 1.39
4.17 Table S5 indicates that three quarters of the land developed for non-residential uses tends to serve industrial or commercial purposes directly. Areas for individual protected landscapes appear in Table 6, appear in Table 6 but variation in the pace of development is most readily apprehended by reference to Table 7 which expresses rates of development relative to households. Only in two protected landscapes did this exceed 1 hectare per 1000 households per annum (Northumberland National Park and the Broads), and here non-residential development accounted for more than a half of all development by area. The overall values for the Northumberland National Park which do not reflect conditions over the inter-censal decade result from highway development consistent with the discussion in para 4.10
Understanding the Tables: What Values Might Be Expected?
5.1 Hopefully, the foregoing sections serve to suggest some of the influences on the way in which the market and the planning system interact to produce a particular flow through the residential development pipeline, part of which involves conversion of previously undeveloped land to urban use. It has also suggested how this interaction also stimulates a further flow of units secured through other adjustments to the stock of buildings already within the landscape. In so doing it has provided some illustrations of the general pattern of development repeatedly demonstrated by detailed investigation, but which frequently goes unrecognised in popular discussion. The remainder of this section tries to take this a little further by suggesting how these general principles might be applied to consider the volume and pattern of development that might be expected.
5.2 On the basis of the discussion provided below, „expected‟ outcomes for various land-use change indicators are included as Tables 9a and b. These expectations are intended as a „counterfactual,‟ a statement of the outcomes that might be expected in the absence of specific policy-based actions within the protected landscapes. Conceptually, the starting point is with the demand for additional household units. Most new households are formed by „fission‟ from households already present, as individuals set up homes of their own, though the scale of the overall stock of households in the country also depends on international migration balances. Moreover, as household movement is predominantly over short distances, demand the geography of demand for new dwellings is very closely related to the existing distribution of households and population (see Bibby and Brindley 2006). These tendencies are the principal influences on the geography of new housebuilding shown in Figure 2, which shows the absolute number of new dwellings constructed over the period 2001-2011 within 10kms of any point.
5.3 The number of households within 10km of a point exerts a major influence on housing demand. The geographic variation in this quantum- which stands as a proxy for „economic mass (as discussed in Bibby, 2009) is shown in Figure 7a, and average values for each protected landscape are provided in Column 13 of Tables 6a and b. It measures the average density of households within a circle of 10km radius centred on any point. Given that such circles are of identical area, the variation displayed exactly tracks that of the absolute number of households. In addition to providing a pointer to housing demand it is also useful in considering matters such as retail potential. Other things being equal, lower levels of housing demand should be anticipated in those protected landscapes with lower economic mass. From the perspective of protected landscape professionals there may also be merit in considering demand pressure based associated with households across a wider area as set out in Column 14 of Table 6a and illustrated in Figure 7b which given presumptions of some protected landscape professionals and behavioral patterns evident in Natural England‟s MENE dataset might serve as a proxy for the likely flow of day visitors. Statistical investigation also shows that residential desirability of AONBs stimulates longer commuting distances than might be expected on the basis of spatial structure alone (i.e. economic mass at different scales) (Bibby 2009).
5.4 Apart from responding to geographic variation in housing demand, the tabulated values reflect geographic variation in housing land supply, which is of course mediated by the planning system. In some circumstances supply may be constrained. More generally the geographic configuration of housing land supply- as modified by the pattern of land release secured through the planning system - influences the detailed pattern of residential development. This goes much of the way to accounting for the differential incidence of housing development in different types of locale evident in Table S1. As discussed above (2.31), constraining new building may exert upward pressure on prices, thereby increasing the viability of conversion and adaptation schemes. The cost of the work involved in a building extension or adaptation varies relatively little from place to place, but the value of the asset created as indicated by house-price differentials (and also of any associated stream of rental income) varies markedly. Restricting new construction in areas of high demand thus stimulates a supply of additional units though alterations and adaptation which allow intensification of existing property. Hence in Tables 1 and 2 divergence between the rate of new construction (column 3) and the rate of net change (column 5) – quite marked in many protected landscapes -tracks the manner in which constraint stimulates adaptation i.e. the higher the differential, the greater the tendency for new units to be brought forward through adaptation of the existing stock.
5.5 The immediate effect of the planning system on the pattern of land release is suggested by comparison of Tables 1, 2 and 3 which illustrate the significance of the urban fringe development in different protected landscapes and by the varying contribution of previously developed sites to the housing land supply. Divergence between column 3 in Table 1 and its counterpart in Table 2 indicates the importance of development at the urban fringe and urban infill development. In nearly all of the protected landscapes, urban expansion is minimal and there is very limited divergence. Consistent with the discussion above, urban fringe development in the protected landscapes is very modest, and divergence highlights the Kent Downs AONB (effect of development on the urban margin of Hawkinge on) –evident also in Table 3. As indicated above, the Shropshire Hills AONB provides another instance of an area where some urban fringe development has occurred (albeit where rates are lower). Figure 7: Economic Mass Calculated at 10km and 30km; 2011
Dwellings per hectare
a) 10km b) 30km 5.6 The interaction of housing demand and the supply as mediated by the planning system generates both a geographically-differentiated set of prices for property of different characteristics and a set of development outcomes: numbers of units of varying sizes and types at specific locations and sites of different sorts. The rates of development actually achieved and the actual rate net change in the dwelling stock summarize particular very important aspects of this pattern, while the proportion of dwellings built on brownfield sites and the average plot density of newly-built dwellings summarise others. Apart from the actual outcomes, the prime concern is with how this pattern of outcomes differs from that which might have occurred without protected land policy.
5.7 For the purposes of these tabulations, no attempt is made to model the effect of protected status on key development indicators. Instead the particular values of these specific outcome indicators found in each protected landscape are simply compared with those which pertain in an area within 10kms, but without similar protection, adjusting for settlement structure. The counterfactual expectation for each protected landscape is based on outcomes observed within each separate locale-type in the surrounding policy-off areas, but these are re-weighted in accordance with the mixture of locale types present within the specific protected landscape. In this way variations in demand aligned with regional or subregional location are substantially controlled for, while the type of variation between locale types evident in Table S1 is also accommodated.
5.8 The first critical assumption is that conditions of demand in the policy-on and policy-off areas will be broadly comparable. That is to say that the number of potential households willing to purchase a house of identical specification and character in the same type of locale (village, hamlet, town etc) at any particular price within the protected landscape and the comparator area would be broadly the same. A similar assumption is made regarding conditions of supply of housing and of housing land (setting aside the effect of the planning system). In other words we assume that the number of identical dwellings that developers would be willing to construct on „identical‟ sites in locales of the same type would be roughly equal in the policy-on and policy-off areas, were they to be sold at an identical price. Builders‟ capacity to do this would depend upon landowners also being equally willing to bring forward sites of identical size and quality in locales of the same type in the protected landscapes and the comparator areas (given their own perceptions of the housing market). To the degree that these assumptions do not hold, different outcome measures should be expected for the protected landscapes and the comparator areas, even if the planning regimes operate in an identical manner. In comparing the outcomes in Tables 8a and 8b with expectations based on outcomes in the policy-off areas, the implicit assumption is that it is differences in the operation of planning control that exert the most substantial effect on outcomes.
5.9 It is important to appreciate the significance of variation in locale types and the manner in which they are accounted for when gauging differences between protected landscapes and their comparators. Different types of locale have typically different mixes of dwelling types with typically different plot sizes. Moreover, national and local planning policy have tended to regard different settlement types as differentially sustainable, hence influencing the flow of development land coming forward. The mix of settlement types in each protected landscape and comparator area have been assessed on the basis of hectare grids (the construction of which is described in Bibby and Shepherd 2004) characterizing the locale type of each cell. The mix for the protected areas themselves is provided in Table 5. Having drawn from the full set of indicators a summary set standardized by households (provided in Table 7a and 7b), further standardization for variation in the mix of locale types between each protected landscape and its comparator allows for the construction of the policy-off areas (set out in Tables 9a and 9b).
5.10 When actual outcomes are compared with policy-off expectations, aspects of the effect of designation emerge. Overall, the succession of minus signs in Column 1 of Table 8a shows that the rate at which new dwellings have been built per annum and relative to stock has been lower in all the National Parks than in their corresponding policy-off areas. The results for AONBs are for the most part similar, the exceptions being the Blackdown Hills, Dorset and the Northumberland Coast AONBs. The rates of greenfield land conversion associated with this housebuilding in the protected landscapes likewise prove lower than in the respective policy-off areas- once again with the three exceptions. Focusing on the period between 2001 and 2011, although the same marked tendency to lower building rates and lower rates of associated greenfield land conversion are found, the distinctions are less clear. Three coastal AONBs, Northumberland Coast, Norfolk Coast and Dorset had both higher building rates than the corresponding policy-off areas over that decade and higher rates of associated greenfield land conversion.
5.11 Overall, therefore, Table 8 points to a series of outcomes within which each protected landscape area tends to show lower building rates and lower rates of greenfield land conversion than its policy-off comparator. As residential development forms a very substantial part of all built development, there is a tendency for the rate of housebuilding and the sourcing of housing land to shape outcomes substantially. This pattern appears to be replicated in virtually all of the individual protected landscapes.
References:
Barker, K, 2003, Barker Review of Housing Supply: Securing Our Future Housing Needs; Interim report, scope of review, consultation and commissioned research Available at http://webarchive.nationalarchives.gov.uk/+/http:/www.hm- treasury.gov.uk/consultations_and_legislation/barker/consult_barker_background.cfm
Bibby, P. (2009) „ Productivity, Peripherality and Place', Report to Local Government Association, London
Bibby, P. and Brindley, P. (2006) Residential Development since 2000, CLG.
Bibby, P. and Brindley, P. (2013) Urban and Rural Area Definitions for Policy Purposes in England and Wales: Methodology (v1.0) Defra, DCLG, ONS, Welsh Government, https://www.gov.uk/government/uploads/system/uploads/attachment_data/file/239477/RUC11met hodologypaperaug_28_Aug.pdf
Bibby, P. and Shepherd, J. (1999) Re-focusing National Brownfield Housing Targets. Town and Country Planning 68 (10): 302-305.
Bibby, P. and Shepherd, J. (2002) Re-focusing National Brownfield Housing Targets: Brining in the Settlement Pattern. Town and Country Planning 68 (5).
University of Sheffield, 2007, „Land Use Change Indicators For Protected Areas: Tabulations‟ Report Prepared for Natural England
University of Sheffield, 2013, „Land Use Change Indicators For Protected Landscapes: Tabulations‟ Report Prepared for Defra Notes
1. See Planning Application Statistics Table P135. See https://www.gov.uk/government/uploads/system/uploads/attachment_data/file/254667/Table P135.xls
2. The generalised urban area is defined as that area where residential densities of 8 units to the hectare pertain at the 800m scale (see Bibby and Brindley 2013)
3. The size of „sites‟ included in LUCS depend on the linework on OS basic scale maps. A „site‟ purchased by a housebuilding may embrace more than one such parcel. LUCS in this sense tends to underestimate the size of sites (but this applies equally in AONBs and elsewhere).
4. See Planning Application Statistics Table P134 https://www.gov.uk/government/uploads/system/uploads/attachment_data/file/254666/Table P134.xls
5. See Planning Application Statistics Table P134 see https://www.gov.uk/government/uploads/system/uploads/attachment_data/file/254666/Table P134.xls
6. See the discussion of „front extensions‟ in http://planningjungle.com/wp- content/uploads/Part-1-of-the-GPDO-The-10-Worst-Permitted-Development-Loopholes.pdf
7. The Planning and Compensation Act 1991 introduced rolling time limits within which LPAs can take planning enforcement action. Most development, including change of use of a building, or part of a building, to use as a single dwelling becomes immune from enforcement action four years after it is substantially completed. On lawful development certificates see https://www.gov.uk/government/uploads/system/uploads/attachment_data/file/11497/develop mentcertificates.pdf
8 . The difficulty of holiday cottages and numbers of units become clear in documentation related to enforcement against unauthorized subdivision of property at Prior Hall (Farm), Malham and Penyghent Cottage, Selside and both in the Yorkshire Dales National Park. Prior Hall is a farmhouse with attached barn conversion. This property is marketed as available for rent either as one 7 bed roomed dwelling or as two separate 3 and 4 bed roomed properties. A recent appeal regarding against an enforcement notice at Penyghent Cottage, Selside was dismissed, the inspector commenting that 'notwithstanding Penyghent‟s single electricity supply, single central heating system and single water supply the inclusion of 'whether permanently or temporarily‟ in the definition of a dwelling house in .. two circulars cannot be ignored‟ See http://wam.wychavon.gov.uk/WAM/doc/BackGround%20Papers- 818530.pdf?extension=.pdf&id=818530&location=volume2&appid=1001&contentType=app lication/pdf&pageCount=1
9. This culminated in the legal case of Harrogate Borough Council v Crossland & Anor [2012] EWHC 3260 (QB) (20 November 2012) reported at http://www.bailii.org/ew/cases/EWHC/QB/2012/3260.html 10. 2011 Census: Number of people with second addresses in local authorities in England and Wales, March 2011 http://www.ons.gov.uk/ons/rel/census/2011-census/second-address-estimates-for-local- authorities-in-england-and-wales/stb-census-2011-second-addresses-in-e-w.html