Environ Resource Econ (2014) 57:175–196
DOI 10.1007/s10640-013-9664-9
The Amenity Value of English Nature: A Hedonic Price
Approach
Stephen Gibbons · Susana Mourato · Guilherme M. Resende
Accepted: 23 March 2013 / Published online: 27 September 2013
© Springer Science+Business Media Dordrecht 2013
Abstract Using a hedonic property price approach, we estimate the amenity value associated
with proximity to habitats, designated areas, domestic gardens and other natural amenities in
England. There is a long tradition of studies looking at the effect of environmental amenities
and disamenities on property prices. But, to our knowledge, this is the first nationwide study
of the value of proximity to a large number of natural amenities in England. We analysed 1
million housing transactions over 1996–2008 and considered a large number of environmental
characteristics. Results reveal that the effects of many of these environmental variables are
highly statistically significant, and are quite large in economic magnitude. Gardens, green
space and areas of water within the census ward all attract a considerable positive price
premium. There is also a strong positive effect from freshwater and flood plain locations,
broadleaved woodland, coniferous woodland and enclosed farmland. Increasing distance to
natural amenities such as rivers, National Parks and National Trust sites is unambiguously
associated with a fall in house prices. Our preferred regression specifications control for
unobserved labour market and other geographical factors using Travel to Work Area fixed
effects, and the estimates are fairly insensitive to changes in specification and sample. This
provides some reassurance that the hedonic price results provide a useful representation of
the values attached to proximity to environmental amenities in England. Overall, we conclude
that the housing market in England reveals substantial amenity value attached to a number
of habitats, designations, private gardens and local environmental amenities.
Keywords
Amenity value · Hedonic price method (HPM) · Environmental amenities
S. Gibbons · S. Mourato (B) · G. M. Resende
Department of Geography and Environment, London School of Economics and Political Science,
Houghton Street, London WC2A 2AE, UK
e-mail: s.mourato@lse.ac.uk
S. Mourato
Grantham Research Institute on Climate Change and the Environment, London School of Economics and
Political Science, London, UK
G. M. Resende
Institute for Applied Economic Research/Government of Brazil (IPEA), Brasília, Brazil
123
176
S. Gibbons et al.
1 Introduction
Living within or in close proximity to desirable natural areas and environmental resources
such as coastal, river or woodland habitats, managed and protected areas, and urban parks
and gardens is thought to provide a large number of positive welfare benefits to the public.
These include not only numerous opportunities for recreation, leisure and wildlife viewing,
but also the possibility of improved physical health through green exercise, visual amenity,
improved mental or psychological well-being, artistic inspiration, and ecological education.
The Millennium Ecosystem Assessment (2005) refers to these types of amenity benefits as the
‘cultural services’ provided by ecosystems, while the UK National Ecosystem Assessment
(2011) classifies them as the ‘cultural goods or benefits’ provided by environmental settings
and wild species diversity.
Economic valuation methods such as stated and revealed preference techniques have
been widely applied to estimate the cultural ecosystem benefits associated with green areas
and environmental resources (e.g. Garrod and Willis 1999; Tyrvainen and Miettinen 2000;
Earnhart 2001; Poor et al. 2007). In particular, there is a long tradition of hedonic price studies
measuring environmental values by investigating the effect of environmental amenities on
property prices. The first environmental study, Ridker and Henning’s analysis of the effects
of air pollution on house prices, dates back to 1967.
In the 40 years that elapsed since this pioneering contribution, there have been dozens of
studies estimating the price impacts of a wide range of other environmental amenities such as
water quality (Walsh et al. 2011; Leggett and Bockstael 2000; Boyle et al. 1999), preserved
natural areas (Correll et al. 1978; Lee and Linneman 1998), wetlands (Doss and Taff 1996;
Mahan et al. 2000), forests (Garrod and Willis 1992; Tyrvainen and Miettinen 2000; Thorsnes
2002), beaches (Landry and Hindsley 2011), agricultural activities (Le Goffe 2000), nature
views (Benson et al. 1998; Paterson and Boyle 2002; Luttik 2000; Morancho 2003), urban
trees (Anderson and Cordell 1985; Morales 1980; Morales et al. 1983) and open spaces
(Cheshire and Sheppard 1995, 1998; Bolitzer and Netusil 2000; Netusil 2005; McConnell
and Walls 2005). Disamenities such as road noise (Day et al. 2006; Wilhelmsson 2000) have
also been investigated. For the most part, this large body of literature has consistently shown
an observable effect of environmental factors on property prices.
A broad inspection of these previous works shows that environmental hedonic studies
typically focus on a single or a very limited number of environmental attributes, thereby
possibly failing to account for the interplay between multiple environmental amenities and
housing preferences. Examples include recent large studies such as Walsh et al. (2011)
valuing water quality changes in Orange County, Florida, USA and Landry and Hindsley
(2011) valuing beach quality in Tybee Island, Georgia, USA. Garrod and Willis (1992) found
that proximity to hardwood forests had a positive influence on house prices whilst mature
conifers had a negative effect. However, their study does not take account of the influence
of other land cover types. We only found a handful of studies that looked at more than one
environmental amenity. For example, Geoghegan (2002) looked at amenity effects related to
proximity to several types of open space in Howard County, Maryland, and found that only
permanently protected open spaces (preserves, parks, and easements) have a statistically
significant relationship with land prices. Omitting potentially important variables from the
hedonic price model can lead to serious specification bias. By and large, because of lack
of data or small sample sizes, existing studies also fail to control for a wide enough range
of potentially confounding geographical factors and are particularly lacking in location and
neighbourhood characteristics (e.g. school quality, crime rates, job market characteristics,
etc).
123
The Amenity Value of English Nature: A Hedonic Price Approach
177
Furthermore, past hedonic analysis are often applied to narrow geographical locations
(counties, cities or parts of cities) and based on small sample sizes. For example, Cheshire
and Sheppard (2002) used data from a UK city (Reading) to show that the benefits associated
with accessible open space (e.g. parks) considerably exceeded those from more inaccessible
open space (e.g. green belt and farmland). Some of the largest areas and sample sizes we
could find in recent environmental valuation studies were that of Walsh et al. (2011)—who
employ a dataset of 54,000 property sales to investigate the value of surface water quality in
Orange County (covering 2,600 km2 ), Florida—and Netusil et al. (2010)—who use just over
30,000 property sales in a comprehensive second stage hedonic price analysis of the benefits
of tree canopy cover in Portland, USA. Most other recent studies are based on substantially
smaller sample sizes. Pearson et al. (2002) study on the impact of an Australian National Park
on surrounding land values was based on 641 prices for a single year 1999. In 2007, a study
of urban green space in Jinan City in China used a sample 124 property prices for the year of
2004 (Kong et al. 2007). More recently, Yusuf and Resosudarmo (2009) studied the impact of
air pollution on property prices in Jakarta, Indonesia, based on a sample of 470 observations
for 1998, while Landry and Hindsley (2011) valued beach quality in Tybee Island (57 km2 ),
Georgia, USA, using 372 real estate transactions. The representativeness of these small area
studies is open to question, so it is important to know if the link between environmental
characteristics and house prices remains discernible when conducting the analysis over a
much wider geographical area with a greater environmental diversity. Moreover, an analysis
at a wider geographical scale potentially permits the investigation of the value of larger
scale environmental variables, such as different habitats or ecosystems and different types of
protected areas.
In this paper we estimate the amenity value associated with habitats, designated areas,
heritage sites, domestic gardens and other natural amenities in England (and Great Britain to a
lesser extent) using a hedonic price approach (Sheppard 1999; Champ et al. 2003). Our study
adds to the body of evidence on environmental values using this method, by estimating the
value of a wide range of environmental amenities, using a very large and representative data
set of housing transactions over a 13 year period, and a large and diverse geographical study
area (the whole of England and Great Britain). We assemble data on a large number of control
variables (important neighbourhood attributes, transport accessibility and differences in local
labour market opportunities between locations) all of which are potentially highly correlated
with the availability of natural amenities. Our regression specifications control for Travel
to Work Area (labour market) fixed effects, so estimation of the effects of environmental
amenities comes from within-labour market variation. This method controls for earnings and
other labour market differences across space without the need for direct measure of wages and
employment opportunities. To our knowledge, this is the first nationwide study of the value of
such a wide range of natural amenities in England (and Great Britain). The remainder of the
paper is organized as follows. In Sect. 2 we provide further details about our methodological
approach, Sect. 3 presents and discusses our main findings and Sect. 4 offers some summary
conclusions.
2 Methodology
The hedonic price method uses housing market transactions to infer the implicit value of the
house’s underlying characteristics (structural, locational/ accessibility, neighbourhood and
environmental). Rosen (1974) presents the theoretical rationale for this analysis, showing
that the utility benefit of marginal changes in one component of the bundle of attributes in
123
178
S. Gibbons et al.
a composite good like housing can be monetised by measuring the additional expenditure
incurred in equilibrium. These firm foundations in economic theory and observable market
behaviour, rather than on stated preference surveys, make the method desirable from a policy
perspective.
Applied hedonic analysis recovers the marginal valuations or ‘implicit prices’ of the
separate housing attributes from a regression of housing transaction sales prices on the component attributes of the house sold—its structural characteristics, environmental quality,
neighbourhood amenities, labour market opportunities and so on. Hedonic price studies of
environmental quality must therefore link data on housing transaction locations to measures
of environmental quality. In recent years, the use of geographical information systems (GIS)
and the availability of GIS data on environmental quality have increased the detail and flexibility with which these attributes can be linked to house locations, allowing for improved
accuracy in the consideration of proximity to natural features, designated natural areas, and
the amount and topography of the local environmental amenities.
2.1 Data Description
2.1.1 Geographical Area
Whilst most previous analysis using property values for environmental valuation were applied
to relatively restricted geographical areas such as cities or parts of cities, our analysis spans
the whole of England, with some comparisons made with Great Britain (England, Scotland
and Wales). Specifically, our units of analysis are individual houses located across England
(130,395 km2 ), Scotland (78,772 km2 ) and Wales (20,779 km2 ).
2.1.2 House Price Data
We use a very large sample of about 1 million housing transactions in Great Britain, over
1996–2008, with information on location at full postcode level (about 17 houses on average).
The house sales price data is from the Nationwide building society. In this paper, we mainly
make use of house transactions for England as we do not have complete environmental data
for the other regions. However, we present comparison estimates for Great Britain for those
environmental amenities for which this is feasible. Our sample size is the largest we have
found in the environmental hedonic literature.
2.1.3 Environmental Variables
Great Britain is home to a wide range of ecosystems, natural habitats and other green areas
that play an important role in biodiversity conservation. Our analysis considers a large number
of these natural amenities related to land cover, terrain and designated natural areas.
First, we use nine broad habitat categories, which we constructed from the Land Cover
Map 2000 (remote sensed data from the Centre for Ecology and Hydrology) describing the
physical land cover in terms of the proportional share (0–1) of a particular habitat within
the 1km x 1km square in which the property is located: (1) Marine and coastal margins;
(2) Freshwater, wetlands and flood plains; (3) Mountains, moors and heathland; (4) Seminatural grasslands; (5) Enclosed farmland; (6) Coniferous woodland; (7) Broad-leaved/mixed
woodland; (8) Urban; and (9) Inland Bare Ground. The omitted class in this group is ‘Urban’,
so the model coefficients reported in the results section should be interpreted as describing
the effect on prices as the share in a given land cover is increased, whilst decreasing the share
123
The Amenity Value of English Nature: A Hedonic Price Approach
179
of urban land cover. Currently, in Great Britain, farmland occupies the largest area, almost
50 % of the country, followed by semi-natural grasslands and mountains, which together
cover approximately a third of Great Britain, with woodland covering just over 12 % (Fuller
et al. 2002). There are over 5 billion day visits to the English countryside each year (TNS
Travel and Tourism 2004) and about one third of all leisure visits in England were to the
countryside, coast or woodlands (Natural England 2005).
Natural amenities are also provided at a much more localised scale, through urban parks
and other formal and informal urban green spaces such as people’s own domestic gardens.
Mean per capita provision of accessible public green spaces in urban areas of England was
recently calculated at 1.79 ha per 1,000 people (CABE 2010) with just under 50 % of the
population using public urban green spaces at least once a week (Defra 2009) while just
under 90 % said they used their local parks or open spaces regularly (DCLG 2008). Moreover,
approximately 23 million households (87 % of all homes) have access to a private garden.
Domestic gardens in England constitute just over 4 % (564,500 ha) of total land cover with the
majority being located in urban areas and covering an average 13 % of the urban landscape
(Generalised Land Use Database 2005). Despite modern trends, such as the paving over front
gardens, it is increasingly recognized that domestic gardens provide crucial habitats for plant
and animal species (Gaston et al. 2007). Indeed, gardening is thought to be one of the most
commonly practiced type of physical activity in Great Britain (Crespo et al. 1996; Yusuf et al.
1996; Magnus et al. 1979) with British households spending on average 71 h a year gardening
(MINTEL 1997). To try and capture some of these amenities, we also use six land use share
variables taken from the Generalised Land Use Database (CLG 2007). These variables depict
the land use share (0–1), in the Census ward in which a house is located, of the following land
types: (1) Domestic gardens; (2) Green space; (3) Water; (4) Domestic buildings; (5) Nondomestic buildings and (6) ‘Other’. The hedonic model coefficients indicate the association
between increases in the land use share in categories (1)–(5), whilst decreasing the share
in the omitted ‘other’ group. This omitted category incorporates transport infrastructure,
paths and other land uses (Roads; Paths; Rail; Other land uses, (largely hard-standing); and
Unclassified in the source land use classification).
Especially important, rare or threatened natural areas are formally designated under various pieces of national and international legislation to ensure their protection. One of the best
known designations are National Parks, aiming to conserve the natural beauty and cultural
heritage of areas of outstanding landscape value and to provide opportunities for the public
to understand and enjoy these special qualities. There are 10 National Parks in England, 3
in Wales and 2 in Scotland (National Parks 2010). Popular National Parks such as the Peak
District, the Yorkshire Dales and the Lake District, attract in the order of 8–10 million visits
each year (National Parks 2010). Another commonly used designation is the Green Belt,
used in planning policy in Great Britain to avoid excessive urban sprawl by retaining areas
of largely undeveloped, wild, or agricultural land surrounding urban areas. There are around
14 Green Belts throughout England, covering 13 % of land area (CLG 2010), with the largest
being the London Green Belt covering about 486,000 hectares. To capture the value of such
designated areas we created two additional variables depicting designation status: respectively, the proportion (0–1) of Green Belt land and of National Park land in the Census ward
in which a house is located. The model coefficients in the results section show the association
between ward Green Belt designation, National Park designation and house prices.
We also constructed five ‘distance to’ variables describing proximity to various natural
and environmental amenities, namely (1) distance to coastline, (2) distance to rivers, (3)
distance to National Parks (England and Wales), (4) distance to National Nature Reserves
123
180
S. Gibbons et al.
(England and Scotland), and (5) distance to land owned by the National Trust.1 The effects
of these variables are scaled in terms of the distance, in 100s of kilometres, between each
resource and each house identified by its postcode. Distance is measured in a straight line to
the nearest of these features. The inclusion of a variable depicting proximity to National Trust
properties was motivated by the desire to capture the heritage interest or historical importance
sometimes associated with certain natural areas. In Great Britain many of these areas belong
to the National Trust, the country’s leading independent conservation and environmental
organisation, acting as a guardian for the nation in the acquisition and permanent preservation of places of historic interest and natural beauty. The Trust manages around 254,000
hectares (627,000 acres) of countryside moorland, beaches and coastline in England, Wales
and Northern Ireland, 709 miles of coastline (1,141 km), as well as a large number of historic
gardens and nature reserves (National Trust 2010). There are some 14 million yearly visits to
its ‘pay for entry’ properties, and an estimated 50 million visits to its open air properties. We
also included distance to the nearest of the twenty four National Nature Reserves in England
that were established to protect the finest wildlife and geological sites in the country, and are
a selection of the best existing Sites of Special Scientific Interest (Natural England 2011).
Some of our regression specifications include the effect of ‘distance to the nearest church’.
This variable is intended to capture potential amenities associated with the places where
churches are located—i.e. historic locations in town centres, with historical buildings, and
focal points for business and retail—but may arguably also capture to some extent the amenity
value of churches, via their architecture, churchyards, church gardens and cemeteries. This is
only reported for a subset of metropolitan areas in England (spanning London, the North West,
Birmingham and West Midlands) for which the variable was constructed by the researchers
from Ordnance Survey digital map data. The sample is restricted to properties within 2km
of one of the churches in this church dataset.
Table 1 presents summary statistics for the housing transactions data in relation to the
key environmental variables considered. The table contains mean, standard deviation and
maximum of the land area shares (i.e. the proportion of land in a particular use) and distances,
for the housing transactions sample. The figures are thus representative of residential sites
in England, rather than the land area as a whole. Inspection of the table shows that housing
transactions are more prevalent in certain types of land cover. For example, the average house
sale is in a ward in which 20 % of the land use is gardens. The table also indicates that, as
expected, most of the houses are in wards that are urban (i.e. the missing base category among
the land cover variables).
2.1.4 Control Variables
Another distinguishing feature of our analysis is the large number of control variables considered. Along with the house sales price data, we have data on several internal and local
characteristics of the houses. Internal housing characteristics are property type, floor area,
floor area-squared, central heating type (none or full, part, by type of fuel), garage (space,
1 It should be noted that our dataset includes distance to all (916) National Trust properties. Although the
overwhelming majority of these properties contain (or are near) picturesque or important natural environmental amenities, some also contain houses and other built features. For example, NT’s most visited property
Wakehurst Place, the country estate of the Royal Botanic Gardens (Kew), features not only 188 hectares of
ornamental gardens, temperate woodlands and lakes but also an Elizabethan Mansion and Kew’s Millennium
Seed Bank. Hence, the amenity value captured by the ‘distance to land owned by the National Trust’ variable reflects also some elements of built heritage that are impossible to disentangle from surrounding natural
features.
123
The Amenity Value of English Nature: A Hedonic Price Approach
181
Table 1 Summary statistics for the housing transactions data
Mean
Standard deviation
Maximum
Ward share of
Domestic gardens
0.205
0.134
0.629
Green space
0.511
0.267
0.989
Water
0.024
0.067
0.888
Domestic buildings
0.067
0.049
0.311
Other buildings
0.031
0.034
0.496
Green Belt
0.155
0.321
1.000
National Park
0.003
0.049
1.000
Ward area (km2 )
10.385
19.884
462.471
Distance (100 km) to
Coastline
0.276
0.275
1.028
Rivers
0.011
0.012
0.467
National Parks
0.467
0.291
1.669
Nature Reserves
0.130
0.078
0.751
National Trust properties
0.072
0.053
0.459
Marine and coastal margins
0.005
0.036
1.000
Freshwater, wetlands, floodplains
0.006
0.025
0.851
Mountains, moors and heathland
0.029
0.018
0.782
Land in 1 km square
Semi-natural grassland
0.076
0.086
1.000
Enclosed farmland
0.246
0.236
1.000
Coniferous woodland
0.006
0.025
0.943
Broadleaved woodland
0.060
0.077
0.899
Inland bare ground
0.007
0.026
0.895
Altitude (100 m)
0.642
0.484
4.812
Slope (10 s degrees)
0.172
0.161
2.980
East facing slope
0.249
0.432
1.000
South facing slope
0.269
0.443
1.000
West facing slope
0.223
0.321
1.000
Topography
Accessibility and other variables
Distance to station (100 km)
0.028
0.032
0.407
Distance to motorways (100 km)
0.137
0.199
1.695
Distance to primary road (100 km)
0.020
0.024
0.283
Distance to A-road (100 km)
0.013
0.019
0.330
Distance to TTWA centre (100 km)
0.099
0.066
0.449
Population (1,000 s/km2 )
3.205
2.404
17.916
Age 7–11 value added (standardised)
0.000
1.000
4.949
Distance to school (km)
0.843
2.059
85.434
Distance × value-added
0.038
2.456
0.696
123
182
S. Gibbons et al.
Table 1 continued
Mean
Standard deviation
Maximum
Distance to nearest church (km)
0.796
0.461
2.000
Mean purchase price (£, 1996–2008)
135,750
96,230
1,625,000
Ln price
11.608
0.656
16.619
Table reports unweighted means and standard deviations
Sample is Nationwide housing transactions in England, 1996–2008
Sample size is 1,011,831, except distance to church 448,445
single, double, none), tenure, new build, age, age-squared, number of bathrooms (dummies),
number of bedrooms (dummies), year and month dummies.
Hedonic studies that cover multiple labour markets need to take account of variation in
earnings and employment, because amenity differences are potentially compensated through
expected earnings as well as housing prices (Roback 1982; Albouy 2008; Gibbons et al.
2011). Workers will be willing to pay more for housing costs and/or accept lower wages
to live in more desirable places. Consequently, we can only value amenities using housing
costs alone by comparing transactions at places within the same labour market, where the
expected wage is similar in each place. We use Travel to Work Area (TTWA) fixed effects
to control for all labour market variables such as wages and unemployment rates and more
general geographic factors (e.g. climate) that we do not observe. There are 243 TTWAs in
the 2007 definition that is based on 2001 Census data (Coombes and Bond 2008). These
TTWAs are defined as zones where at least 67 % of the resident population work within the
same area, and at least 67 % of the employees in the area live in the area (the means are
around 80 %). Our preferred regression specifications difference all the regression variables
from their TTWA means (the within-groups transformation, equivalent to including TTWA
dummies) and therefore estimate the effects of amenities using variation occurring within
each TTWA (i.e. within each labour market).2
We also constructed a number of other geographic control variables. The first set of these
represent the topography of the site of the house location, derived from digital elevation model
data. These 90 m raster data come from the UK SRTM digital elevation model available
from the ShareGeo service (http://www.sharegeo.ac.uk/handle/10672/5). From these data
we derive the altitude, slope angle, and aspect of the house postcode. Aspect is categorised
into four directions, North (>315◦ or ≤45◦ ), East (>45◦ & ≤135◦ ), South (>135◦ & ≤225◦ )
and West (>225◦ & ≤315◦ ), and dummy variables for the East, South and West directions
are included in the regressions (North being the baseline).
Five variables capture distances to various types of transport infrastructure (stations,
motorways, primary roads, A-roads) and distance to the centre of the local labour market
(Travel to Work Area, 2007 definition). The land area of the ward and the population density
are also included as control variables. Local school quality is often regarded as an important
determinant of housing prices (see for example Gibbons and Machin 2003; Gibbons et al.
2013), so we include variables for the effectiveness of the nearest school in raising pupil
achievement (mean age 7–11 gains in test scores or ‘value-added’), distance to the nearest
school, and interactions between these variables. Summary statistics for housing transactions
in relation to topography, schools, accessibility and other control variables are also contained
in Table 1.
2 In principle, consumer prices are a factor too, but local data on prices is unavailable and goods prices are
unlikely to vary within TTWAs.
123
The Amenity Value of English Nature: A Hedonic Price Approach
183
2.2 Functional Form
The appropriate functional form for the hedonic price regression specification is arguable, but
in our empirical work we follow the standard in recent studies and estimate semi-logarithmic
regression models of the form:3
Ln H Pi jt = α + xit β1i + n it β2i + sit β3i + f j + τt + εit ,
(1)
where the dependent variable (Ln H Pi jt ) is the natural logarithm of the sale price for each
property transaction ‘i’ in labour market j in period t. The environmental variables of interest
are included in vector xit , with control variables for neighbourhood characteristics n it and
structural housing characteristics sit . There are potentially unobserved labour market effects
( f j ), period specific effects (τt ) and other residual unobserved components (εit ). All the
variables are described in detail in Sect. 2.1. Housing market attributes sit include property
type, floor area, floor area-squared, central heating type, garage, tenure, new build, age, agesquared, number of bathrooms, and number of bedrooms. The vector n it includes distances
to various types of transport infrastructure (stations, motorways, primary roads, A-roads),
distance to the centre of the local labour market, topography, land area of the ward, population
density, local school quality, and distance to the nearest school. Labour market fixed effects
( f j ) are controlled for by differencing the data from the TTWA mean (i.e. we use a withingroups fixed effects estimator). Time effects (τt ) are captured by year and month dummy
variables, and serve to deflate and deasonalise the price data.
The environmental characteristics (xit ) that are the focus of our analysis include nine broad
habitat categories, six land use types, proportion of Green Belt land and of National Park
land in the Census ward in which a house is located, nearest distance to coastline, to rivers, to
National Parks, to National Nature Reserves, to land owned by the National Trust and to the
nearest church. Regression estimates of the coefficient vector β1 provide the implicit prices
of the environmental attributes in which we are interested.
2.3 Limitations
Although we have multiple years of transactions in house price data, this is a fundamentally
cross-sectional analysis because the data sources available at the present time offer only
limited information on changes over time in natural amenities and land cover (and we suspect
that the changes would be too small to be useful). There are obvious limitations to this
type of analysis since it is impossible to control for all salient characteristics at the local
neighbourhood level. We do not have data on all potentially relevant factors (e.g. crime rates,
retail accessibility, localised air quality) and if we had the data it would be infeasible to include
everything in the regressions. Our research design must therefore rely on a more restricted
set of control variables (described above), plus TTWA fixed effects, to try to ensure that the
estimated effects of the environmental amenities reflect willingness to pay for these amenities
rather than willingness to pay for omitted characteristics with which they are correlated. Our
representation of the accessibility of amenities is also restricted in that we look only at the land
cover in the vicinity of a property and the distance to the nearest amenity of each type. We do
not, therefore, consider the diversity of land cover or the benefits of accessibility to multiple
instances of a particular amenity (e.g. if households are willing to pay more to have many
National Trust properties close by). Our data also lacks detail on view-sheds and visibility of
3 There is a large body of work investigating different functional forms for the hedonic price equation. Of
note, more recently, several authors have also explored semiparametric and nonparametric specifications (e.g.
Bontemps et al. 2008; Parmeter and Henderson 2007).
123
184
S. Gibbons et al.
environmental amenities, which would be infeasible to construct given the national coverage
of our dataset, although we do include measures of altitude, slope and aspect as discussed
in Sect. 2.1.4. Finally, the main part of our analysis only refers to England for the full set of
environmental variables, as we do not have complete environmental data for the other regions.
Even given these limitations, it turns out that the estimates are fairly insensitive to changes in
specification and sample—once we take proper account of inter-labour market differences.
This provides some reassurance that our regression results provide a useful representation of
the values attached to proximity to environmental amenities in England.
3 Results and Discussion
Table 2 presents the ordinary least squares regression estimates from five hedonic property
value models in which the dependent variable is the natural log of the sales price, and the
explanatory variables are a range of environmental attributes characterising the place in which
the property is located plus a large number of control variables as described in Sects. 2.1.3 and
2.1.4, respectively. Data are taken from the Nationwide transactions database, as explained
in Sect. 2.1.2. The table reports coefficients and standard errors.4
Model 1 (Table 2) is a simple model in which only the environmental attributes (plus
year and month dummies) are included as explanatory variables. Model 2 introduces a set of
structural property characteristics listed in the table notes. Model 3 adds in Travel to Work
Area fixed effects. Finally, Model 4 repeats the analysis of Model 3 for the sub-sample of
metropolitan sales for which we have computed distance to the nearest church and Model 5
provides estimates for England, Scotland and Wales using only those attributes for which we
have complete data for all these countries.
The coefficients report the change in log prices corresponding to a unit change in the
explanatory variables (scaled as indicated in Table 2). The standard errors indicate the precision of the estimates. The asterisks indicate the level of statistical significance, from 1 %
(3 stars) to 10 % (1 star). Note that interpretation of the results requires that we take into
account both the magnitude of the coefficient, and the precision with which it is measured. A
coefficient can be large in magnitude implying potentially large price effects, but be imprecisely measured, and hence statistically insignificantly different from zero. In such cases,
there must remain some uncertainty about whether or not the corresponding characteristic is
economically important.
Looking at the coefficients and standard errors in OLS Model 1 (Table 2) reveals that
many of the land use and land cover variables are highly statistically significant, and represent quite large implied economic effects. For example, in the first row of Model 1, a 1 %
point (0.01) increase in the share of gardens is associated with a 2 % increase in the sales
price. This figure can be calculated by applying the transformation exp(0.01*beta)-1, or, to a
good approximation, by reading off the coefficient beta as the % change in prices in response
to a 0.01 change in the share of gardens. There are similarly large coefficients for other ward
land use shares in Model 1, but no association of prices with Green Belt designation. The
associations with physical land cover types present a mixed picture, with freshwater with
woodland strongly associated with higher prices, semi-natural grassland and bare ground
associated with lower prices, and other land cover types having small associations or associations that are statistically indistinguishable from zero. Some of the coefficients on the
4 Standard errors are clustered at the Travel to Work Area level to allow for heteroscedasticity and spatial
and temporal correlation in the error structure within TTWAs.
123
The Amenity Value of English Nature: A Hedonic Price Approach
185
Table 2 Property prices and environmental amenities (regression estimates)
(1)
OLS
(2)
+ housing
characteristics
(3)
(4)
+ TTWA fixed Metropolitan
effects
areas, with
churches
(5)
All
Great
Britain
2.122***
1.415***
1.016***
1.165***
–
(0.458)
(0.234)
(0.133)
(0.252)
1.837***
1.038***
1.041***
1.184***
(0.269)
(0.129)
(0.076)
(0.146)
Ward share of
Domestic gardens
Green space
Water
Domestic buildings
Other buildings
Green Belt
National Park
Ward area (km2 )
1.363***
0.738***
0.973***
1.088***
(0.285)
(0.144)
(0.080)
(0.152)
3.185***
1.200***
2.177***
2.321***
(0.304)
(0.453)
(0.307)
(0.161)
4.059***
2.952***
2.672***
2.971***
(0.589)
(0.351)
(0.226)
(0.317)
−0.047
−0.023
0.022
0.032*
(0.041)
(0.036)
(0.019)
(0.017)
−0.207**
0.018
0.048
−0.002
(0.096)
(0.051)
(0.039)
(0.043)
0.002***
0.001*
0.001***
0.001**
(0.001)
(0.000)
(0.000)
(0.000)
–
–
–
–
–
–
–
Distance (100 kms) to
Coastline
Rivers
National Parks
Nature Reserves
National Trust properties
Land share in
1km-square
Marine and coastal margins
−0.511*** −0.098
−0.141
−0.620***
−0.204*
(0.074)
(0.091)
(0.124)
(0.227)
(0.117)
0.230
1.269
−0.938
−2.569***
−1.105
(0.910)
(1.055)
(0.819)
(0.718)
(0.718)
–
0.273***
0.158***
−0.240***
−0.407***
(0.090)
(0.058)
(0.088)
(0.137)
−0.473
−0.380*
−0.075
−0.313
(0.306)
(0.193)
(0.241)
(0.538)
−2.083*** −1.744***
−0.695***
−0.320
(0.416)
(0.172)
(0.337)
Semi-natural grassland
–
−0.697*** −0.278**
0.039
−0.112
0.039
(0.238)
(0.034)
(0.105)
(0.041)
Freshwater, wetlands, floodplains 0.901***
Mountains, moors and heathland
(0.242)
–
(0.114)
0.966***
0.357**
0.445***
0.296**
(0.177)
(0.220)
(0.147)
(0.141)
(0.142)
0.113
0.261
0.083
0.012
−0.072
(0.326)
(0.195)
(0.100)
(0.225)
(0.083)
−0.222**
−0.234***
−0.014
−0.029
−0.019
(0.090)
(0.059)
(0.024)
(0.045)
(0.025)
123
186
S. Gibbons et al.
Table 2 continued
(1)
OLS
Enclosed farmland
Coniferous woodland
Broadleaved woodland
Inland bare ground
(2)
+ housing
characteristics
(3)
+ TTWA fixed
effects
(4)
Metropolitan
areas, with
churches
(5)
All
Great
Britain
0.172**
0.081***
0.059***
0.077***
0.088***
(0.065)
(0.030)
(0.012)
(0.025)
(0.017)
0.544*
0.353**
0.119*
0.105
0.147**
(0.307)
(0.151)
(0.062)
(0.126)
(0.068)
0.549***
0.656***
0.193***
0.153***
0.243***
(0.099)
(0.073)
(0.031)
(0.055)
(0.038)
−0.787**
−0.646**
−0.379***
−0.440***
−0.444***
(0.313)
(0.301)
(0.101)
(0.113)
(0.125)
–
−0.052*
0.000
0.045
0.003
(0.028)
(0.023)
(0.044)
(0.018)
−0.048
0.006
−0.001
0.009
(0.032)
(0.015)
(0.026)
(0.018)
0.002
0.006
0.005
0.001
(0.006)
(0.004)
(0.006)
(0.004)
Topography
Altitude (100 m)
Slope (10 s degrees)
East facing slope
South facing slope
West facing slope
–
–
–
–
0.011
0.005
0.004
0.001
(0.008)
(0.005)
(0.009)
(0.004)
−0.004
−0.001
−0.006
−0.001
(0.006)
(0.003)
(0.004)
(0.004)
Accessibility/other
Distance to station
Distance to motorways
Distance to primary road
Distance to A-road
Population (1,000 s/km2 )
Age 7–11 value added (SD)
Distance to school
Distance × value-added
Distance to TTWA centre
123
–
–
–
–
–
–
–
–
–
−1.102***
−0.142
−0.285
0.057
(0.238)
(0.197)
(0.506)
(0.187)
−0.271***
−0.179
−0.415
−0.068
(0.064)
(0.116)
(0.416)
(0.100)
0.687*
−0.177
0.055
0.099
(0.360)
(0.168)
(0.452)
(0.177)
−0.670***
0.159
0.305
0.508**
(0.239)
(0.196)
(0.561)
(0.255)
0.032***
0.002
0.004
0.002
(0.008)
(0.005)
(0.003)
(0.007)
–
0.035***
0.022***
0.032***
(0.006)
(0.004)
(0.004)
−0.002
0.009**
0.045***
(0.003)
(0.003)
(0.013)
−0.003*
−0.002**
−0.011***
(0.001)
(0.001)
(0.003)
–
–
0.984***
−0.603**
−1.105**
−0.598**
(0.138)
(0.270)
(0.499)
(0.266)
The Amenity Value of English Nature: A Hedonic Price Approach
187
Table 2 continued
(1)
OLS
Distance to nearest church
(2)
+ housing
characteristics
(3)
+ TTWA fixed
effects
–
(4)
Metropolitan
areas, with
churches
(5)
All
Great
Britain
−0.042***
–
(0.009)
House characteristics
No
Yes
Yes
Yes
TTWA fixed effects
No
No
Yes
Yes
Yes
Yes
Observations
1,011,831
1,011,831
1,011,831
448,445
1,133,433
R-squared
0.518
0.768
0.866
0.855
0.854
Table reports coefficients and standard errors from OLS regressions of ln house sales prices on environmental
amenities. Standard errors are clustered at Travel To Work Area level (2007 definition)
Ward share coefficients show approximate % change in price for 1 % point increase in share of Census Ward
in land use. Omitted category is ‘other land uses not listed’
1 km2 landcover share coefficients show approximate % change in price for 1 % point increase in share of the
1 km square containing the property (=10,000 m2 within nearest 1 million m2 ). Omitted category is ‘urban’
Distance coefficients show approximate % change in price for 1km increase in distance
Sample is Nationwide housing transactions in England, 1996–2008, except for Model 5, where the sample
refers to Great Britain
Unreported housing characteristics in Models 2–5 are property type, floor area, floor area-squared, central
heating type (none or full, part, by type of fuel), garage (space, single, double, none), tenure, new build, age,
age-squared, number of bathrooms (dummies), number of bedrooms (dummies), year and month dummies
Metropolitan areas in Model 4 include North West, West Midlands and London and is restricted to sales within
2km of nearest church
*** p < 0.01, ** p < 0.05, * p < 0.10
distance to environmental amenities variables in Model 1 (and indeed in Model 2) have
counterintuitive signs, if interpreted as valuations of access to amenities.
The partially counterintuitive pattern in Model 1 is unsurprising, given that there are innumerable price-relevant housing characteristics and geographical attributes that are omitted
from this specification. Many of these are likely to be correlated with the environmental
and land use variables leading to potential omitted variable biases. However, introducing a
set of housing characteristics and measures of transport accessibility as control variables in
Model 2 (Table 2) has surprisingly little effect on the general pattern of results in terms of
coefficient magnitude and statistical significance. There are some changes in the point estimates, and some coefficients become more or less significant, but the general picture is the
same.
Controlling for wage and other inter-labour market differences in Model 3 (Table 2),
our preferred model, provides potentially more credible estimates of the influence of the
environmental amenities on housing prices, and we now discuss these in more detail. The
first column of Table 3 (All England) summarises the estimates of the monetary implicit prices
of environmental amenities in England corresponding to Model 3’s regression coefficients.
Note that these implicit prices are capitalised values i.e. present values, rather than annual
willingness to pay. Long run annualised figures can be obtained by multiplying the present
values by an appropriate discount rate (e.g. 3.5 %).
Domestic gardens, green space and areas of water within the census ward all attract a
similar positive price premium, with a 1 % point increase in one of these land use shares
increasing prices by around 1 % (Model 3, Table 2). Translating these into monetary implicit
prices in column 1 (All England model) on Table 3 indicates capitalised values of around
123
188
S. Gibbons et al.
Table 3 Implicit prices by region (£ capitalised values)
(1)
All England
(2)
London,
South East
and West
(3)
Midlands, East
Midlands and
East
(4)
North, North
West and
Yorkshire
1,982***
1,673***
1,955***
2,515***
Percentage point ward share of
Domestic gardens
Green space
2,031***
2,033***
1,200***
1,804***
Water
1,897***
1,831***
1,180***
1,926***
Domestic buildings
4,271***
4,918***
609
2,329**
Other buildings
5,254***
5,868***
2,858***
4,625***
Green Belt
42
23
81
18
National Park
92
−225**
252***
137
Ward area (km2 )
1.7***
3.2***
1.3**
0.9**
Coastline
−274
−279
−91
−205
Rivers
−1,811
−3,350
−2,684**
−548
National Parks
−465***
−361**
−186
−793***
Nature Reserves
−146
−1,347
632
−397
National Trust properties
−1,344***
−3,545***
−213
−1,118**
Percentage point
in 1 km2
Marine and coastal margins
76
220
49
38
Freshwater, wetlands, floodplains
694**
1,247***
42
169
Mountains, moors and heathland
161
−196
−273*
889***
Semi-natural grassland
−27
−5
−34
−173***
Enclosed farmland
115***
127**
32
73**
Coniferous woodland
232*
281**
296
−159
Broadleaved woodland
376***
433***
405***
237*
Inland bare ground
−733***
−1,024***
−108
−425*
Altitude (100 m)
34
1,1959*
−326
−4,948
Slope (10 s degrees)
1,238
−1,804
3,460
3,697
East
1,231*
3,321***
952
1,133
Distance (1 kms) to
Topography
South
999
3,481***
861
−798
West
−115
374
727
−1,654*
Distance to station (km)
−276
−30
−686*
−236
Distance to motorways (km)
−346
−487
−418
−10
Distance to primary road (km)
−344
−392
221
132
Accessibility/other
Distance to A-road (km)
309
955
−234
−491
Population (1,000 s/km2 )
320
1,250
−3,317***
−1,907**
123
The Amenity Value of English Nature: A Hedonic Price Approach
189
Table 3 continued
(1)
All England
(2)
London,
South East
and West
(3)
Midlands, East
Midlands and
East
(4)
North, North
West and
Yorkshire
Age 7–11 value added (SD)
4,280***
5,644***
3,826***
2,657***
Distance to school (km)
1,656**
3,127***
90
1,494**
Distance × value-added
−399**
−607
−380***
64
Distance to TTWA centre (km)
−1,166**
−1,731*
−516*
−822**
Observations
1,011,831
475,780
341,450
194,601
Mean house price
194,040
243,850
181,058
158,095
Table reports marginal willingness to pay, evaluated at regional mean prices. The All England estimates
correspond to the coefficients in Model 3, Table 2
Distance variables evaluated for 1km change
Land shares evaluated for 1 % point change
School value added evaluated for 1 standard deviation change
*** p < 0.01, ** p < 0.05, * p < 0.10
£2,000 for these land use changes. The share of land use allocated to buildings has a large
positive association with prices. This may, in part, reflect willingness to pay for dense and
non-isolated places where there is other proximate human habitation. However, there is a
potential omitted variables issue here because build density will tend to be higher in places
where land costs are higher, and where land costs are higher due to other amenities that we
do not observe. As such, the coefficients may represent willingness to pay for these omitted
amenities rather than willingness to pay for a more built up environment. Therefore, some
caution is needed in interpretation.
Neither Green Belt nor National Park designation shows a strong statistical association
with prices because the coefficients are not precisely measured. However, the National Park
coefficient indicates the effect of being inside the park relative to just outside it, given that we
control for distance to the National Park boundary (see further discussion below). Despite
this, the magnitudes indicate potentially sizeable willingness to pay simply for National Park
status. National Park designation (i.e. 100 % of the ward in National Park status) appears to
add about 4.8 % to prices, which at the mean transaction price of £194,040 in 2008 was worth
around £9,200 (note that the coefficient in Model 3, Table 2, and respective implicit price in
Table 3 is for an increase of only 1 % point in the share of the ward designated as National
Park).
The results on physical land cover shares (within 1 km2 ) indicate a strong positive effect
from freshwater, wetlands and flood plain locations which is smaller than, though consistent with, the result based on ward shares (i.e. the ward share of water).5 A 1 % point
increase in the share of this land cover attracts a premium of 0.36 % (Model 3, Table 2),
or £694 (All England model, Table 3). There is also a strong and large positive effect from
increases in broadleaved woodland (0.19 % or £376), and a weaker but still sizeable relationship with coniferous woodland (0.12 % or £232, but only marginally significant). Enclosed
farmland attracts a small positive premium (0.06 % or £115). Mountain terrain attracts a
higher premium (0.08 % or £161), but the coefficient is not precisely measured. Proximate
marine and semi-natural grassland land cover does not appear to have much of an effect
5 The ward-based water shares and 1km square freshwater, wetlands and floodplains shares are weakly
correlated with each other which suggests they are measuring different water cover.
123
190
S. Gibbons et al.
on prices, whereas inland bare ground has a strong negative impact, with prices falling by
0.38 % (£733) with each 1 % point increase in the share of bare ground. Given the scaling
of these variables, these implicit prices can also be interpreted as the willingness to pay
for an extra 10,000 m2 of that land use within the 1 million m2 grid in which a house is
located.
The coefficients on the distance variables (Model 3, Table 2) show that increasing distance
to natural amenities is unambiguously associated with a fall in prices. This finding is consistent
with the idea that home buyers are paying for accessibility to these natural features. The
biggest effect in terms of magnitude is related to distance to rivers, with a 1 km increase
in distance to rivers lowering prices by 0.93 % or £1,811 although this coefficient is only
marginally statistically significant (see Tables 2, 3). Smaller but more precisely measured
effects relate to distance from National Parks and National Trust sites. Each 1km increase
in distance to the nearest National Park lowers prices by 0.24 % or £465. This implies that
being inside a National Park (i.e. at zero distance from it), combined with 100 % of the ward
as a National Park, implies a huge £33,686 premium relative to the average house in England
(which is 46.7 km from a National Park). Each 1km increase in distance to the nearest National
Trust owned site is associated with a 0.7 % or £1,350 fall in prices. Distances to coastline
and nature reserves also lowers prices (by about £140–£275 per km), although in these cases
the estimates are not statistically significant.
The accessibility variables at the bottom of Table 2 (and Table 3) are intended as control
variables so we do not discuss these at length. It is worth noting that they generally have
the expected signs when interpreted as measures of the value of transport accessibility, but
are not individually significant. Distance to the TTWA centre reduces housing prices, which
is consistent with the theory in urban economics that lower housing costs compensate for
higher commuting costs as workers live further out from the central business district in
cities. Note also that this coefficient in Model 2 (Table 2) does not have the sign we would
expect from theory, which highlights the importance of controlling effectively for betweenlabour market differences as we do in Model 3. The estimates of the effect of school quality
on house prices in Model 3 (Table 2) is in line with estimates using more sophisticated
‘regression discontinuity’ designs that exploit differences across school admissions district
boundaries (see Black and Machin 2011). The estimate implies that a one standard deviation
increase in nearest primary school value-added raises prices by 2.2 % for houses located
next to the school, which is similar to the figure reported in Gibbons et al. (2013). The
interactions of school quality with distance also work in the directions theory would suggest,
although distance from a school attenuates the quality premium more rapidly than we would
expect, implicitly falling to zero by 110 m from a school and turning negative beyond that
distance.6 Topography variables are generally insignificant across all model specifications in
Table 2.
Restricting the sample to major metropolitan regions in Model 4 (Table 2) leads to a
pattern of coefficients that is broadly similar to those discussed above for Model 3. However, some effects become more significant and the implicit prices larger, particularly those
related to distance to coastline, rivers and National Parks. As might be expected, Green
Belt designation becomes more important when looking at major metropolitan areas. The
results indicate a willingness to pay amounting to around £7,000 for houses in Green
Belt locations, which offer access to cities, coupled with tight restrictions on housing
supply.
6 From the coefficients, the derivative of prices with respect to school quality is obtained as 0.022–0.20 ×
distance (in km).
123
The Amenity Value of English Nature: A Hedonic Price Approach
191
Distance to churches (those classified as having steeples or towers on Ordnance Survey
maps) also comes out as important, with 1 km increase in distance associated with a large
4.2 % fall in prices, worth about £8,150 (Model 4, Table 2). This figure may be best interpreted
as a valuation of the places with which churches are associated—traditional parts of town
centres, focal points for businesses and retail, etc.—rather than a valuation of specifically
church-related amenities and spiritual values. However, the environmental amenities provided
by church grounds and architectural values of traditional churches could arguably also be
relevant factors.
For purposes of comparison, Model 5 in Table 2 extends the analysis to the whole of
Great Britain. The ward land use shares are not available outside of England, and we do
not have data on National Parks in Scotland, Nature Reserves in Wales or National Trust
properties in Scotland, nor any school quality data except in England. These variables are
therefore dropped from the analysis. The patterns amongst the remaining coefficients are
similar to those in the Model 3 regression for England only, providing some reassurance
that the estimates are transferrable to Great Britain as a whole. Indeed, the coefficients on
the 1 km2 land cover variables are generally insensitive to the changes in sample between
Models 3, 4 and 5 in Table 2.
Using the coefficients from Table 2, we can predict the (log) house price differentials
that can be attributed to variations in the level of environment amenities across the country.
We do this using the coefficients from Model 3 (Table 2), and expressing the variation in
environmental quality in terms of deviations around their means, and ignoring the contribution
of housing attributes and the other control variables and TTWA dummies in the regression.
The resulting predictions therefore show the variation in prices around the mean in England,
and are mapped in Fig. 1.
Figure 1 shows the house price variation in 10 categories. The mean house price in 2008
was around £194,000, so, for example, the lightest shaded areas represent the places with the
highest value of environmental amenities, amounting to valuations of £67,900 and above in
present value terms. Annualised over a long time horizon, this is equivalent to a willingness
to pay £2,376 per year at a 3.5 % discount rate. These highest values are seen in areas such
as the Lake District, Northumberland, North York Moors, Pennines, Dartmoor and Exmoor.
The implication is that home buyers are willing to pay this amount per year to gain the
environmental amenities and accessibility of these locations, relative to the average place in
England. Lowest levels of environmental value occur in central England, somewhere in the
vicinity of Northampton. We estimate that people are prepared to pay around £1,765 per year
to avoid the relatively poor accessibility of environmental amenities that characterises these
locations relative to the average in England. Note that from the data underlying this map, we
can estimate that the top 1 % postcode has over 1.7 times as much environmental value as
the bottom 1 % postcode, a difference which is worth around £105,000 (capitalised value) or
£3,700 per year.
As a final step in the analysis, we report separate results for grouped Government Office
Regions in England. Columns 2–4 of Table 3 show the implicit prices (capitalised) for these
groups, derived from separate regressions for each regional group sample and based on
the mean 2008 house price in each sample (reported in the last row of the table). Looking
across these columns, it is evident that there are differences in the capitalised values and
significance of the various environmental amenities according to region, although the results
are qualitatively similar. The ward land use shares of gardens, green space and water have
remarkably similar implicit prices regardless of region. The first notable difference is the
greater importance of National Park designation in the Midlands regions (the Peak District
and Broads National Parks), but lesser importance of National Trust sites. It is also evident
123
192
S. Gibbons et al.
Fig. 1 Geographical distribution of environmental value (predicted price differentials from property value
regressions). % Price differentials are based on log price differentials, and correspond to maximum % differentials relative to the national mean price level
that the value of freshwater, wetlands and floodplain locations is driven predominantly by
London and the south of England. Coniferous woodland attracts value in the regions other
than the north, but broadleaved woodland attracts a positive premium everywhere. Although
mountains, moors and heathland cover had no significant effect on prices in England as a
whole, we see it attracts a substantial positive premium in those locations where this land
cover is predominantly found, i.e. the North, North West and Yorkshire. The topography of
the housing transaction site is also more interesting in London, South East and West, where
we find substantial premia for high ground facing South and East.
123
The Amenity Value of English Nature: A Hedonic Price Approach
193
4 Conclusions
The hedonic price approach was used to estimate the amenity value associated with proximity
to habitats, designated areas, domestic gardens and other natural amenities in England. To
our knowledge, this is the first nationwide study of the value of proximity to such a wide
range of natural amenities in England (and Great Britain). Overall, we conclude that the
house market in England reveals substantial amenity value attached to a number of diverse
natural settings. For convenience, a summary of our key findings for England is presented
in Table 4. Although results are generally similar, for some amenities we found evidence of
significant differences across regions within England. Many of the key results appear to be
broadly transferable to Great Britain.
This article provides new evidence on the benefits of a wide range of environmental amenities within a national setting, using a labour market fixed effects regression design, coupled
with a rich dataset on environmental amenities and other geographical control variables. Our
results are robust to changes in specification and sample. However, our analysis also highlighted a number of limitations in design and data availability for this type of research. First,
Table 4 Implicit prices for key environmental amenities in England (£ capitalised values)
Environmental amenity
% change in house value with
Implicit price in
relation to average
2008 house price
1 percentage point increase in share of land cover
Marine and coastal margins
0.04 % increase in house prices
£76
Freshwater, wetlands, floodplains
0.36 % increase in house prices
£694
Mountains, moors and heathland
0.08 % increase in house prices
£161
Semi-natural grassland
0.01 % decrease in house prices
£-27
***
Enclosed farmland
0.06 % increase in house prices
£115
***
Broadleaved woodland
0.19 % increase in house prices
£376
***
Coniferous woodland
0.12 % increase in house prices
£232
*
Inland bare ground
0.38 % decrease in house prices
£-733***
***
1 percentage point increase in land use share
Domestic gardens
1.02 % increase in house prices
£1,982
***
Green space
1.04 % increase in house prices
£2,031
***
Water
0.97 % increase in house prices
£1,897
***
Being in the Green Belt (major metro. areas)
3.25 % increase in house prices
£6,967
*
Being in a National Park, relative to mean
17.36 % increase in house prices
£33,686
***
Distance to coastline
0.14 % fall in house prices
−£274
Distance to rivers
0.93 % fall in house prices
−£1,811
*
Distance to National Parks
0.24 % fall in house prices
−£465
***
Distance to Nature Reserves
0.08 % fall in house prices
−£146
Distance to National Trust land
0.70 % fall in house prices
−£1,344
Designation
1 km increase in distance
***
Being in a National Park calculation is based on zero distance from National Park and having a ward share of
100 % National Park
The stars indicate statistical significance levels *** p < 0.01, ** p < 0.05, * p < 0.10
123
194
S. Gibbons et al.
control-variable based research designs are always open to criticism since it is infeasible to
include all relevant factors in regression models (for example, we had no data on local crime
rates). Changes in land-cover and environmental amenities (e.g. through erosion, development activities, park designations etc.) offer the potential for more robust quasi-experimental,
repeat-sales based designs. However, instances of these kinds of changes are hard to find,
and good national data is rare. Data limitations (lack of ward level information on land use)
also prevented us from extending the full analysis to the whole of Great Britain. We looked at
a limited set of environmental amenities and have not investigated the effect of disamenities
(proximity to landfill or flood risk), the role of diversity in land cover, the benefits of accessibility to multiple instances of a particular amenity, nor the role of views. There is an inevitable
trade-off between achieving national coverage and representativeness, and providing detail
of amenities at this level.
Overall, the key finding from this work is that environmental amenities are highly valued
by home-owners and have a substantial impact on housing prices. Moving the bottom 1 %
postcode to the best 1 % postcode in England is worth about £105,000 (or £3,700 per year)
in terms of the environmental amenities provided.
Acknowledgments This research was carried out as part of the UK National Ecosystem Assessment (http://
uknea.unep-wcmc.org/). Financial support from UNEP-World Conservation Monitoring Centre (UNEPWCMC) is gratefully acknowledged. We would like to thank Ian Bateman, Carlo Fezzi and David Maddison
for insightful comments on earlier versions of the paper. We are also grateful to Claire Brown and Megan
Tierney from UNEP-WCMC for help in sourcing some of the data used. The authors are responsible for any
errors or omissions.
References
Albouy D (2008) Are big cities really bad places to live? Improving Quality-of-Life Estimates Across Cities,
NBER Working Paper 14472
Anderson LM, Cordell HK (1985) Residential property values improve by landscaping with trees. South J
Appl For 9:162–166
Benson ED, Hansen JL, Schwartz AL Jr, Smersh GT (1998) Pricing residential amenities: the value of a view.
J Real Estate Econ Finance 16:55–73
Black S, Machin S (2011) Housing valuations of school performance. In: Hanushek EA, Machin S, Woessman
L (eds) Handbook of the economics of education, vol 3, chapter 10. Elsevier, Amsterdam
Bolitzer B, Netusil NR (2000) The impact of open space on property values in Portland, Oregon. J Environ
Manag 59(3):185–193
Bontemps C, Simioni M, Surry Y (2008) Semiparametric hedonic price models: assessing the effects of
agricultural nonpoint source pollution. J Appl Econom 23(6):825–842
Boyle KJ, Poor PJ, Taylor LO (1999) Estimating the demand for protecting freshwater lakes from eutrophication. Am J Agric Econ 81(5):1118–1122
CABE (2010) Urban green nation: building the evidence basis. CABE, London
Champ PA, Boyle KJ, Brown TC (eds) (2003) A primer on nonmarket valuation. Kluwer, Boston
Chay KY, Greenstone M (2005) Does air quality matter? Evidence from the housing market. J Polit Econ
113(2):376–424
Cheshire PC, Sheppard S (1995) On the price of land and the value of amenities. Economica 62:247–267
Cheshire PC, Sheppard S (1998) Estimating the demand for housing, land and neighbourhood characteristics.
Oxf Bull Econ Stat 60:357–382
Cheshire PC, Sheppard S (2002) Welfare economics of land use regulation. J Urban Econ 52:242–269
CLG (2007) Generalised land use database statistics for England 2005. Department of Communities and Local
Government, London
CLG (2010) Local planning authority green belt statistics: England 2009/10. Department for Communities
and Local Government, London
Coombes M, Bond S (2008) Travel-to-work areas: the 2007 review. Office for National Statistics, London
123
The Amenity Value of English Nature: A Hedonic Price Approach
195
Correll MR, Lillydahl JH, Singell LD (1978) The effects of greenbelts on residential property values: some
findings on the political economy of open space. Land Econ 54:207–217
Crespo CJ, Keteyian SJ, Heath GW, Sempos CT (1996) Leisure-time physical activity among US adults,
results from the third national health and nutrition examination survey. Arch Intern Med 156:93–98
Davies ZG, Fuller RA, Loram A, Irvine KN, Sims V, Gaston KJ (2009) A national scale inventory of resource
provision for biodiversity within domestic gardens. Biol Conserv 142:761–771
Day B, Bateman I, Lake I (2006) Estimating the demand for peace and quiet using property market data.
CSERGE Working Paper EDM 06–03, University of East Anglia
DCLG (2008) Place survey: England. On-line: http://www.communities.gov.uk/publications/corporate/
statistics/placesurvey2008
Defra (2009) Public attitudes and behaviours towards the environment—tracker study. On-line: http://www.
Defra.gov.uk/evidence/statistics/environment/pubatt/download/reportattitudes-behaviours2009.pdf
Doss CR, Taff SJ (1996) The influence of wetland type and wetland proximity on residential property values.
J Agric Resour Econ 21(1):120–129
Earnhart D (2001) Combining revealed and stated preference methods to value environmental amenities at
residential locations. Land Econ 77(1):12–29
Faber Taylor A, Kuo F, Sullivan W (2001) Coping with ADD: the surprising connection to green play settings.
Environ Behav 33:54–77
Fuller RM, Smith GM, Sanderson JM, Hill RA, Thomson AG (2002) Land cover map 2000 (LCM2000): a
guide to the classification system. Department of the Environment Transport and the Regions
Garrod GD, Willis KG (1992) The environmental economic impact of woodland: a two state hedonic price
model of the amenity value of forestry in britain. Appl Econ 24(7):715–728
Garrod GD, Willis KG (1999) Economic valuation of the environment. Edward Elgar Publishing Ltd, Cheltenham
Gaston KJ, Fuller RA, Loram A, MacDonald C, Power S, Dempsey N (2007) Urban domestic gardens (XI):
variation in urban wildlife gardening in the United Kingdom. Biodiv Conserv 16:3227–3238
Generalised Land Use Database (2005) Generalised land use database statistics for England 2005. Communities and Local Government
Geoghegan J (2002) The value of open spaces in residential land use. Land Use Policy 19(1):91–98
Gibbons S, Machin S (2003) Valuing English primary schools. J Urban Econ 53(2):197–219
Gibbons S, Machin S, Silva O (2013) Valuing school quality using boundary discontinuity regressions. J Urban
Econ 75:15–28
Gibbons S, Overman HG, Resende GM (2011) Real earnings disparities in Britain. Spatial Economics Research
Centre Discussion Paper 0065, London School of Economics and Political Science
Goddard MA, Dougill AJ, Benton TG (2009) Scaling up from gardens: biodiversity conservation in urban
environments. Trends Ecol Evolut 25(2):90–98
Kong F, Yin H, Nakagoshi N (2007) Using GIS and landscape metrics in the hedonic price modeling of the
amenity value of urban green space: a case study in Jinan City, China. Landsc Urban Plan 79:240–252
Landry C, Hindsley P (2011) Valuing beach quality with hedonic property models. Land Econ 87(1):92–108
Lee CM, Linneman P (1998) Dynamics of the greenbelt amenity effect on the land market: the case of Seoul’s
greenbelt. Real Estate Econ 26(1):107–129
Leggett CG, Bockstael NE (2000) Evidence of the effects of water quality on residential land prices. J Environ
Econ Manag 39(2):121–144
Le Goffe P (2000) Hedonic pricing of agriculture and forestry externalities. Environ Resour Econ 15(4):397–
401
Luttik J (2000) The value of trees, water and open space as reflected by house prices in the Netherlands. Landsc
Urban Plan 48(3–4):161–167
Magnus K, Matroos A, Strackee J (1979) Walking, cycling, or gardening, with or without seasonal interruption,
in relation to acute coronary events. Am J Epidemiol 110(6):724–733
Mahan BL, Polasky S, Adams RM (2000) Valuing urban wetlands: a property price approach. Land Econ
76:100–113
McConnell V, Walls M (2005) The value of open space: evidence from studies of nonmarket behaviour.
Resources for the Future, Washington
Millennium Ecosystem Assessment (2005) Ecosystems and human well being: synthesis. Island Press, Washington
MINTEL (1997) Specialist Garden Centres. Market Intelligence
Morales DJ (1980) The contribution of trees to residential property value. J Arboric 7:109–112
Morales DJ, Micha FR, Weber RL (1983) Two methods of valuating trees on residential sites. J Arboric
9:21–24
Morancho AB (2003) A hedonic valuation of urban green areas. Landsc Urban Plan 66(1):5–41
123
196
S. Gibbons et al.
National Ecosystem Assessment (2011) The UK national ecosystem assessment: synthesis of the key findings.
UNEP-WCMC, Cambridge
National Parks (2010) Facts and figures. http://www.nationalparks.gov.uk/press/factsandfigures.htm
National Trust (2010) Facts about the trust. http://www.nationaltrust.org.uk/main/w-trust/w-thecharity/
w-thecharity_our-present/w-what_we_do/w-factsabouttrust.htm
Natural England (2005) English leisure visits survey. Natural England, Cheltenham
Natural England (2010) National parks facts and figures (on-line). http://www.naturalengland.org.uk/ourwork/
conservation/designatedareas/nationalparks/factsfigures.aspx
Natural England (2011) National nature reserves (on-line). http://www.naturalengland.org.uk/ourwork/
conservation/designatedareas/nnr/default.aspx
Netusil NR (2005) The effect of environmental zoning and amenities on property values: Portland, Oregon.
Land Econ 81(2):227–246
Netusil NR, Chattopadhyay S, Kovacs KF (2010) Estimating the demand for tree canopy: a second-stage
hedonic price analysis in Portland, Oregon. Land Econ 86(2):281–293
Office for National Statistics (2007) National park population estimates for England and Wales 2007. http://
www.statistics.gov.uk/statbase/Product.asp?vlnk=15094
Parmeter C, Henderson DJ (2007) Nonparametric estimation of a hedonic price function. J Appl Econ 22:695–
699
Paterson RW, Boyle KJ (2002) Out of sight, out of mind? Using GIS to incorporate visibility in hedonic
property value model. Land Econ 78(3):417–425
Pearson LJ, Tisdell C, Lisle AT (2002) The impact of Noosa National Park on surrounding property values:
an application of the hedonic price method. Econ Anal Policy 32(2):155–171
Poor PJ, Pessagnob KL, Paul RW (2007) Exploring the hedonic value of ambient water quality: a local
watershed-based study. Ecol Econ 60(4):797–807
Ridker RB, Henning JA (1967) The determinants of residential property values with special reference to air
pollution. Rev Econ Stat 49(2):246–257
Roback J (1982) Wages, rents, and the quality of life. J Polit Econ 90(6):1257–1278
Rosen S (1974) Hedonic prices and implicit markets: product differentiation in pure competition. J Polit Econ
82(1):34–55
Sheppard S (1999) Hedonic analysis of housing markets. In: Cheshire PC, Mills ES (eds) Handbook of regional
and urban economics, chapter 41, vol 3, 1st edn. Elsevier, Amsterdam
Smith D (2010) Valuing housing and green spaces: understanding local amenities, the built environment and
house prices in London. GLA Working Paper 42
Thorsnes P (2002) The value of a suburban forest preserve: estimates from sales of vacant residential building
lots. Land Econ 78(3):426–441
TNS Travel and Tourism (2004) GB day visits survey. Edinburgh
Tyrvainen L, Miettinen A (2000) Property prices and urban forest amenities. J Environ Econ Manag 39(2):205–
223
Walsh PJ, Milon JW, Scrogin DO (2011) The spatial extent of water quality benefits in urban housing markets.
Land Econ 87(4):628–644
Wilhelmsson M (2000) The impact of traffic noise on the values of single-family houses. J Environ Plan Manag
43:799–815
Yusuf AA, Resosudarmo BP (2009) Does clean air matter in developing countries’ megacities? A hedonic
price analysis of the Jakarta housing market, Indonesia. Ecol Econ 68:1398–1407
Yusuf HR, Croft JB, Giles WH, Anda RF, Casper ML, Caspersen CJ, Jones DA (1996) Leisure-time physical
activity among older adults, United States, 1990. Archiv Intern Med 156:1321–1326
123
Copyright of Environmental & Resource Economics is the property of Springer Science &
Business Media B.V. and its content may not be copied or emailed to multiple sites or posted
to a listserv without the copyright holder's express written permission. However, users may
print, download, or email articles for individual use.
Purchase answer to see full
attachment