Pion Economics (2000) suggest that whilst resource deprivation is likely to differ in nature and severity according to location, it will be present in both urban and rural areas. In contrast, opportunity and mobility deprivation derive specifically from the role of rural geography. Examinations of rural deprivation must, therefore, be sensitive to the particular circumstances facing rural households, both in terms of inequality of outcomes and inequality of opportunities.
Introducing the individual
It follows that objective measures of deprivation alone are insufficient. We also need to consider how individuals actually perceive and experience life in a rural area.
As Copus et al.
(1996) suggest, quantitative research is required to establish how many households could be described as poor, but it requires a qualitative approach to identify what being poor actually means.
There is a considerable literature which does just this (e.g. Bradley et al., 1986; Cloke, 1995a, 1995b, 1996; Shucksmith et al., 1996a; Woodward, 1996; Matthews et al., 2000). There is also a considerable literature that draws on the individual and local to graphic effect often in a specific attempt to influence policy (see for example, Mullins et al., not dated; Derounian, 1993; Newton, 1993; Davis and Ridge, 1997 and Halliday, 1998).
However, the powerful myth of the rural idyll, in which 'rural' and 'exclusion' are often seen as mutually exclusive terms, combines with a lack of anonymity so that ‘those who experience different kinds of deprivation…conceal their condition’ (Scott, 1993:62). In this way serious social problems can be denied, attributed to a point in the past or constructed as a failure of the individual, and there is a similar reluctance to seek assistance. People's subjective assessment of their poverty also tends to contradict objective definitions (Shucksmith et al., 1996b).
Defining rurality
As discussed in section 2 of this report, the definition of rurality is, like the definition of deprivation, far from straightforward.
Martin et al.
(2000) go so far as to argue that the distorted representations of both rural areas and deprivation which are currently available should seriously challenge our existing understanding of some basic research issues concerning rural deprivation and its relationships with other phenomena such as morbidity and mortality. The combination of standard deprivation indicators and standard measures of rurality can thus both contribute to the inadequate measurement and understanding of rural deprivation.
Indicators of disadvantage fail
to show the existence of need in rural areas
The above discussion has shown that the process of identifying urban deprivation is not directly transferable to the identification of rural deprivation. It follows that indicators devised in an urban context should not be expected to perform a similar function in rural areas (see Dunn et al., 1998a; Midwinter and Monaghan, 1990 and Bruce et al., 1995). The reasons for this have been rehearsed many times. Essentially, commonly used indicators (such as unemployment and car ownership) perform in different ways in the urban and rural context, as outlined later in this section, whilst useful indicators (such as ways of capturing social isolation or the cost of delivering services) tend not to be available. People may also respond by moving away in search of jobs or homes, thus removing evidence of the problem. For instance Centrepoint, working in rural Devon, suggest that the reason there are so few young people claiming housing benefit within the South Hams area could be ‘the acute shortage of affordable, accessible private sector accommodation in the area’ (Centrepoint Devon, 1997:111).
Additionally, disadvantaged and poor households in rural areas are unlikely to be spatially concentrated, they tend to live amongst the more affluent and the actual numbers involved tend to be small. Their presence is thus unlikely to make much of a statistical impact on an areal (e.g. a ward or parish) basis. A household rather than area-based approach has therefore been suggested as more appropriate to rural deprivation (see e.g. Hodge et al., 1996).
At the same time the quality of life experienced within rural areas can depend upon conditions at a very local level. For those without access to private transport, gaining access to facilities even a few miles away can represent a significant obstacle. Again, therefore, the definition adopted, the choice of indicators and the way in which location is handled are critical.
Choice of indicators: the meaning attached to indicators in common use
A first problem concerns the ambiguity associated with a number of potential indicators such as car ownership and levels of social housing. An obvious corollary is behavioural differences, reflecting not just (as in these examples) an increased pressure to purchase but, for example, job-search behaviour and the take-up of benefits.
Car ownership
Car ownership, particularly multiple ownership, for instance, tends to be associated with resource availability; money is required to purchase a vehicle and to insure and run it. In rural areas, however, the dispersed nature of employment and services together with the lack of public transport means that car ownership is often considered a necessity and, for household mobility as opposed to mobility for one individual within a household, multiple car ownership will also often be considered a necessity. Pacione, working in rural Scotland, thus found that low-income rural residents were more likely to own cars than their urban counterparts, suggesting an extent of enforced car ownership (Pacione, 1995). For the residual population without a car, or those lacking access to a car within car-owning households, the problems are intensified by minority status.
Unemployment
Unemployment is similarly problematic. The validity of claimant counts has long been questioned as a measure of unemployment in rural areas. People may, for instance, be more likely not to register as unemployed, or to categorise themselves instead as sick, retired or self-employed. This can be in response both to the local labour market (a paucity of local jobs, a restricted range of jobs and seasonal employment, for instance) and to the stigma still seen to be attached to unemployment.
The work of Dunn et al. (see below) attempts to compensate for such problems by following Beatty and Fothergill's approach (1997) taking a broad definition of unemployment. It thus includes not only unemployed claimants, but census self-reported unemployed, those on government schemes, the permanently sick and the early retired. Any excess within the local labour market is calculated relative to a baseline established in the South East
2
, with higher rates taken to represent the effects of fewer employment opportunities and lower levels of labour market participation. However, as the authors admit, this still assumes that behaviour in rural areas does not differ from elsewhere. If, for example, people who retire early or who are permanently sick choose to move to rural areas they would be inappropriately recorded as the 'hidden unemployed' by this method.
Self-employment
In the absence of a census-based question on income, self-employment has been argued by many to be an important indicator of need in rural areas (see for example Peak District Deprivation Forum reported in Hodge et al., 2000). Again, this is a different interpretation to the one that would prevail in urban areas. Pacione (1995) for instance, found a significant correlation between self-employment and households with two or more cars for urban areas. In rural areas, however, there was a strong correlation between self-employment and households with no car. Thus whilst the presence of self-employment in urban areas might be interpreted as indicative of higher income earners the reverse may hold in rural areas.
Indeed, it has also been suggested that high rates of self-employment may depress benefit take-up, because occasional, short-term, seasonal or contract work and irregular agency work mean that people working in this way do not fit easily into the administrative categories used by the tax and benefit authorities (Jordan et al., 1992). Meanwhile, ‘a higher proportion of the labour force in self-employment implies that more people approaching retirement age will not face retirement at a fixed age’ (Phillipson, 1998:81).
Evidence of this kind shows how the same indicator can have very different implications for the measurement of deprivation across the urban/rural spectrum.
3
The requirement for additional indicators
There is also a wide variety of factors potentially associated with rural disadvantage in addition to those typically included in national indices, for example, access to transport, services, and the housing market, together with issues related to gender, culture, isolation and powerlessness. Even within Dunn et al.'s comprehensive approach to describing rural disadvantage notions of economic vulnerability and social isolation appear not to be readily amenable to measurement (see Hodge et al., 2000; also Somerset Health and Social Needs Analysis Group, 1999).
Reliance on census-based data
The reliance of traditional deprivation measures upon census-based data is also partly responsible for their urban bias. Most studies (see below) perceive levels of service provision and accessibility to be important facets of rural deprivation but these are not measured by census-based indicators. Indeed, the fact that the census occurs only every ten years and cannot take account of intervening changes may also be more limiting in the smaller, and hence statistically more volatile, rural areas than in their urban counterparts.
There are of course also a number of more general concerns regarding the several composite indicators based primarily on the census. These include the choice of indicators, the inter-relationships between indicators leading to the possibility of 'double-counting', the statistical methodology by which a composite indicator is calculated, the weighting attached to individual indicators and their spatial bias (Lee et al., 1995; Gordon, 1995; Simpson, 1996; Higgs and White, 2000).
Relationship with policy
Indicators are chosen because they have the potential to track the process of change or the performance of public or private initiatives and the achievement of political goals. There are inevitably problems in ensuring that the chosen indicators effectively capture policy objectives, particularly when policies have multiple goals, some of which are inherently qualitative. The question then becomes whether it is possible to ‘identify quantifiable variables that are in some way causally linked to those qualitative objectives’ (Hodge et al., 2000:1871) or indeed, whether they are effectively linked to policy objectives at all. It is not only that existing indicators ‘may not serve to reveal adequately the nature of rural disadvantage’ (Dunn et al., 1998:22) but that their links to policy aims are unclear.
The Jarman Index or UPA8, for example, consists of eight census variables identified by GPs as contributing most to increased workload/pressure of work, and weighted accordingly. Eight years after its development, however, it was adopted as a means of allocating deprivation payments to GPs (see for example Leavey and Wood, 1985) and has since been used more widely to designate areas as underprivileged or deprived. The move towards evidence-based medicine (both rhetorical and real), it has been suggested, ‘is likely to increase pressure to develop similar indices to implement health policy’ (Taylor, 1998:713). However, necessary precursors include indices that actually follow from policy goals (Carr-Hill and Sheldon, 1991) and the willingness to overcome inertia. Too often ‘the default option is for an index or allocation procedure to be used in perpetuity, even if there are questions about the index or goals of the policy’ (Taylor, 1998:721; see also Farmer et al., 2001).
A range of studies has investigated the extent of rural deprivation within the UK. During the 1980s, McLaughlin (1986) surveyed five rural areas in England. Building on the work of Townsend (1979), he generated an index of poverty and established a deprivation threshold that suggested that, in the five studied areas, the proportion of households in poverty ranged from 21% to 30%.
4
In the 1990s, the Rural Lifestyles research programme provided a snapshot of the incidence and perception of deprivation in rural areas ten years further on (see, for example, Cloke, 1995a, 1997; Woodward, 1996). This sought not only to identify deprivation in an objective sense but also to discover individual perceptions and experiences of rural living. In nine out of the twelve study areas, more than 20% of households were living in, or on, the margins of poverty in the early 1990s (Cloke 1997).
A similar study of disadvantage in rural Scotland examined four case study areas (Harris, Wester Ross, Angus and North Ayrshire) and combined quantitative data on poverty and other aspects of disadvantage with qualitative survey data in order to explore ‘the perceptions and needs of disadvantaged people living in rural Scotland’ (Shucksmith et al, 1994, p.1).
5
All three studies identified similar groups as the most likely to be affected by processes of social exclusion, that is the elderly, the young, women and low income households in general. Low incomes whilst in employment translate into a limited propensity to save and limited or non-existent occupational pensions; they thus affect security across the lifecycle.
The elderly in rural areas are, for instance, vulnerable to low income in combination with problems of isolation and poor access to services. At the time of the 1991 census, for example, there were over 11,000 pensioners living alone in rural Northamptonshire, three quarters of these households had no car and a quarter had no central heating (Northamptonshire County Council, 1997).
Meanwhile, young people experience a combination of problems relating to housing, transport and employment opportunities, and are more likely to move out of remote rural areas than other age groups. A further key group is single person households (for example see Shucksmith and Chapman, 1998).
Certain defining characteristics of deprivation also emerge to a greater or lesser degree from such research into rural areas. Pion Economics (2000) summarises these under six headings: employment, income, housing, transport, education and training, health and social services provision. The latter, given the focus of this report, is explored in the most depth.
Employment
Rural deprivation and poverty tend to be the consequence of low pay, self-employed, part-time and seasonal work rather than long-term unemployment. Those working in the countryside thus ‘survive economically in more diverse ways than city-based wage earners could ever understand’ (Stern and Turbin, 1986).
There are fewer career opportunities in rural areas, firms are smaller and opportunities for graduates are more limited, as are opportunities for training and opportunities for progress within individual firms (see Cartmel and Furlong, 2000). There is also often a limited choice by sector, with primary industries continuing to be the subject of structural decline and alternatives frequently offered only by low wage service occupations such as the tourist sector which is also often only on a seasonal basis.
Chapman et al.
(1998) found that whilst the persistence of low pay is greater for rural than non-rural areas there were few significant differences between the characteristics associated with low pay in the different areas. They suggest that the relatively low escape rate from low pay for individuals employed in small rural workplaces and their dominance in rural employment may be an important explanatory factor. Simply having a job does not in itself prevent social exclusion. Young people in low paid, unskilled and often seasonal employment, for example, have been shown to feel undervalued and marginalised within their community (Pavis et al., 2000). In addition to defining the nature of the problem some studies suggest potential solutions (see Monk et al., 1999, 2000).
Low income
Households with a low income have been shown to be widespread in rural areas and to be no less prevalent than in urban areas. As noted above, low incomes are more typically a consequence of low rates of pay, seasonal employment or low returns from self-employment than a reflection of high rates of unemployment.
Phimister et al.
(2000), for example, draw on the British Household Panel Study to show that one-third of rural households in the persistent low earner category had at least one earner in the rural against only 18% of households in the non-rural case. However, Chapman et al., (1998) similarly analysing rural households in the British Household Panel Study, stress that of those of working age on low incomes in rural Britain, 41% were actually detached from the labour market (e.g. long term sick or family carers).
The limited evidence available suggests this situation is compounded by lower levels of benefit take-up in rural areas as compared with urban areas (see e.g. Bramley et al. 2000). Noble and Wright using grid reference mapping in Shropshire, for instance, found families that claim IS HB/CTB were scattered throughout the district, and that the presence of poverty was not restricted to towns and large villages; there were substantial numbers suffering the same low income as their urban counterparts ‘with the added disadvantage that their sparse distribution, at the very least means that costs of living and access to services may be very much greater.’ (Noble and Wright, 2000:302).
Again the situation reflects a number of variables.
Bramley et al.
(2000) working in Scotland, for instance, found that urban-rural differences were not as important as the ‘marked differences in the extent to which benefits are claimed and the coverage of poverty provided by benefit eligibility between elderly and other households and between areas which are more and less affluent’ (2000:507). The general tendency is thus for eligibility to exceed actual take-up, particularly for elderly households, and for this tendency to be much greater in more affluent areas.
Low incomes and lower levels of benefit uptake are accentuated by generally higher cost of living in rural areas. Transport costs, for instance, are a complex reflection of, inter alia, distances to be travelled, fuel costs at small rural petrol stations, infrequent public transport with few concessions, substitution by more costly private transport or taxi services and the costs of keeping older than average cars on the road.
A report by Cornwall Community Health Council (2000) on transport and access to health services in Cornwall, found that the cost of transport was a real concern, with patients resident throughout Cornwall having to pay their full transport costs, unless they are on benefits or have transport arranged by the ambulance service. For voluntary (independent) car services this averaged 25p–30p per mile and with nearly two-thirds of those sampled at Treliske Hospital having to travel more than ten miles to reach the hospital this is a considerable cost.
In addition a study by Cornwall and Isles of Scilly HAZ found that access to affordable transport limits the ability of young people to participate in a wide range of activities. In Devon, MORI (1998) found that more and cheaper bus and train services emerged as local residents' most common priority in terms of improving their quality of life.
The same concerns are reflected in the cost of a typical food basket. Local shops are more expensive: ‘We've got a little store across the road, which is quite expensive, and the fresh produce is awful, but yes it is there. There's a little Co-op downtown, which you would probably say much the same about. Most of the shops are for tourists’ (Rita, 38, lone parent family, three children, Cornwall, cited in Mullins et al. not dated). Yet the alternatives incur travel costs and opportunity costs. The fact that women are prepared to struggle on public transport with young children and pushchairs rather than shop locally is evidence of the costs associated with shopping locally (see for example Halliday, 1999).
Similar concerns surround the housing market where data from the ONS and Provisor show, for example, a ratio of housing costs to earnings nearly twice as high in the former county of Devon as in the two Devon urban areas of Plymouth and Torbay. Fuel poverty is a further problem which is heightened by the absence of gas in many rural areas. ‘People heating their homes with electric night storage heaters’ it has been suggested ‘are often paying up to £10 a week more than those few families with gas central heating’ (Mullins et al. not dated:37).
The Scottish Poverty Information Unit (1999) reported food prices to be 8% higher in rural areas compared to Aberdeen, and transport costs to be 13% higher. Averaged over all goods and services, the price differential was estimated at 3%. Given that the cost of basic items such as food, fuel and essential transport consume a disproportionate share of household income such price differentials also have a greater significance than a simple average price index implies (Pion, 2000).
Housing
The shortage of low cost housing has been described as the 'principal engine of social change in rural Britain' (Shucksmith, 2000), with house prices inflated by in-migrants and second home owners. Owner occupation rates tend to be higher with opportunities for renting often limited in both the public and private sectors. Insecurity of tenure follows from both tied properties and the holiday market. Given that landlords commonly find tenants by word of mouth, support structures and contacts are vital (Bevan and Sanderling, 1996).
The condition of properties in the private rented sector is also on average relatively poor and if measurable would be a better indicator of rural disadvantage than overcrowding. The limited availability of low cost housing (see Bramley, 1995) in combination with the need to run a car prevents young people from leaving the parental home or leads to migration out of the area. As the RDC (1998) recently emphasised the lack of affordable housing thus ‘not only affects individuals and families, but also undermines the achievement of balanced, sustainable, rural communities.’ Approximately two-thirds of young people surveyed by the University of York, for example, expected to leave the rural area in which they lived, with the availability of suitable housing a key influence (Ford et al., 1997).
Another ramification is the potentially higher than average number of concealed households in rural areas (Lambert et al., 1992). Normalised conceptualisations about rurality and homelessness serve to separate the two concepts and contribute to the assumption that ‘homelessness is an urban phenomenon which is rendered invisible in rural space’ (Cloke, Milbourne and Widdowfield, 2000:715).
Transport deprivation
Lack of public or accessible transport is a key concern and the cause of social exclusion for many rural residents. Hooper (1996) working with rural lone-parents, for example, found social networks to be fairly limited and somewhat fragile. Financial constraints were not only restricting social activity but social contact too because this was mediated by access to either a telephone or a car and two-thirds had no access to the latter.
Particular groups are disproportionately affected, primarily the elderly, many women and young people. Storey and Brannen (2000) for instance, found that, out of a sample of young people (15–24 years) in South West England, over 40% reported that transport issues had influenced their post-16 education decisions. Employment opportunities and social activities are often severely restricted by the availability of transport and there is often little or no provision in the evenings and at weekends. The Countryside Agency (2000) notes that local authority spending on public transport in the rural areas of the South West averaged only £1.40 per head in 1997/8 contrasting with £4.70 per head in the urban areas of the region and an English average of £7.60 per head (although this pre-dated the large increases made possible by the Rural Bus Grant ).
6
The Government has, however, recently emphasised that transport should be considered as a component part of all services. ‘Welfare State provisions were instituted at a time when average weekly mileage per person was about 25 miles. Now it is nearer 130. If a provision has a high transport costs it ceases to be a welfare benefit’ (DETR, 2000a:Executive Summary:1).
Education and training
Due to a recent policy push, there have been some improvements in early years education and childcare facilities. In general, however, the inhabitants of rural areas still suffer from a marked paucity of affordable nursery education and childcare facilities, just as they are typically expected to travel further to access education and further education. Indeed, rural students do not have the same opportunities to live at home while studying, as others are increasingly doing. The problem is compounded by the lack of local opportunities and size and sectoral characteristics of many rural firms means a lack of demand for graduates.
Health status
There are widely, and officially acknowledged, variations in mortality and morbidity within the population (as documented, for example by the DoH 1998 and Acheson, 1998). However, it is argued that there has been a paucity of research into rural health issues (Watt et al., 1994; Higgs, 1999; Pion Economics 2000), perhaps due to the perception that rural areas are healthier places to live than towns and cities. However, most studies which have explored the geographical aspects of this variation draw attention to the contrasts between urban and rural areas and the assumption that a rural lifestyle is inevitably healthier is now regarded as an over-simplification and increasingly open to challenge.
Research tends to reveal a complex picture which in some cases points to a 'healthier' rural environment and in other cases does not. Male suicide rates, for example, have been consistently higher in the Highlands over the last twenty years compared to Scotland as a whole, even excluding non-residents, with farmers the single largest occupational group at risk (Stark et al., 2000). There are also studies which identify other specific problems exhibiting a rural dimension, for example, higher risk of accidents to agricultural workers and higher road traffic accident deaths. Poor diets have also been identified as a particular problem for remote communities such as those in the Western Isles (see McKie, 1996; Clarke et al., 1996). The opportunity to improve diets is hampered by high food prices, low income and (ironically) the limited availability of fresh fruit and vegetables.
In general, however, research tends to suggest that whilst urban areas are characterised by pockets of poor health, rural areas are characterised by low average rates of mortality (see McLoone and Boddy, 1994 and Congdon 1995). This pattern of variation is replicated for all major disease classes, the exceptions being traffic accidents and suicides (Senior et al. 2000).
However, as researchers have become more sophisticated in their appreciation of both indicators of disadvantage and the urban-rural categories used it has become apparent that the observable differences in health outcomes are often explained by social and demographic characteristics rather than locality as such (see for example Verheij, 1996). Again the case is not straightforward. Phillimore and Reading (1992) looking at the SMR for 0–64 year olds in the Northern region of England, for instance, initially found a distinct trend from conurbations (with the highest SMRs) to rural areas. When rural and non-rural areas were then matched by deprivation scores this rural mortality advantage disappeared and little difference remained between the areas. However, when rural areas were further subdivided to include a remote rural component a perceptible difference again became apparent with the remote rural areas having a lower SMR than their more urban counterparts.
It has been recently argued that such findings are a product of the measures of deprivation used. For example, Senior et al. have recently investigated the relationship between premature mortality and material deprivation in urban and rural areas in Wales (Senior et al., 2000). They again initially found inequalities in all-cause premature mortality to be widest in the cities and narrowest in the deeper rural areas but after controlling for socio-economic characteristics (using a range of deprivation measures) the tendency for lower mortality in deeper rural areas was substantially reduced. The residual difference between urban and rural areas was shown to be dependent on the way deprivation is measured and the disease group under study. So for cancers, for example, there were no significant residual mortality differences, whilst for respiratory diseases these were accounted for by employment variables, particularly employment in the coal industry.
Diamond et al.
(1999) similarly emphasise the importance of customised deprivation indices that are specific to the health outcome in urban and rural areas.
Question marks therefore remain about the validity of previous evidence on rural/urban variations in health status. Gregoire and Thornicroft (1998), for instance, suggest that an urban bias both to mental health services and to research has led to an inability to make any definitive statements about rural patterns of disease and mental health and their relationship to rural deprivation.
The previous sections have shown how it has proved difficult to both measure deprivation and to adequately conceptualise and thus measure rurality. In contrast, people within rural areas have been demonstrated to experience poverty combined with a number of dimensions of disadvantage. The next section shows how organisations of all sizes and forms are finding that unless they can make a clear statement of such local need they cannot attract the funding they require to deliver services or to respond to community needs (see Dumfries and Galloway Council, 1995; Powys County Council, 1997; Suffolk County Council, 2000). Critically again, the statistics they have to hand often fail to make this statement.
Requirement for identifying rural poverty for resource allocation procedures
Why it is important to identify rural poverty?
The UK Government allocates money to Local Authorities (LAs) in a variety of ways. The most significant is through the Revenue Support Grant which accounts for around 70% of a LA's income and is calculated using a Standard Spending Assessment (SSA). This has a complex formula with several components, some of which, such as the personal social services sub-block for children's social services, take some account of deprivation/poverty. Section 2.3 above illustrated how this fails to capture the true costs of sparsity.
Since the 1960s there have also been a number of programmes which specifically target resources at regeneration for deprived areas. The Single Regeneration Budget (successor to the Urban Programme) has, for instance, distributed significant money to LAs through an annual competition since 1994, with the Labour government placing an increased emphasis on targeting resources at areas of greatest need. Need for this purpose is often identified using an index of deprivation, currently the Index of Local Deprivation 1998 (ILD) the successor to the Index of Local Conditions (ILC).
7
Both are sub-district level indices derived from the 1991 Census of Population and both have been subject to various criticisms (see Connolly and Chisholm, 1999). The main methodological defects that have been identified are the influence of the scale of the spatial units on the resulting score, the correlations between the variables in the index and the effect of the weights given to the variables.
The Labour government which came to power in 1997 initially accepted the ILC as a persuasive way of identifying such need and the commissioned update was thus ‘limited in scope’ because the Government was satisfied that ‘in broad terms the index is a robust measure of general deprivation’ (DETR 1997). However, they have now accepted that the ILC has limited application to rural areas where deprivation is ‘much more dispersed than it is in urban areas, and tends not to be identified even at ED (enumeration district) level’ (DETR, 1997).
A more fundamental revision has, therefore, recently taken place for both England and Wales, resulting in the Index of Multiple Deprivation (IMD) 2000 (DETR, 2000b). This ward-level index, developed by the Social Disadvantage Research Group (SDRG) at the Department of Social Policy and Social Work at Oxford (see e.g. Noble et al. 1999) is constructed from six sub-indices reflecting different dimensions or domains of deprivation. These are income; employment; health deprivation and disability; education, skills and training; housing; and geographical access to services.
These indices make extensive use of administrative data, especially of claimant counts. They also draw on new estimates for ward-level populations. The result is thus a more up-to-date index. However, it does raise questions about ‘the comparative take-up rates for benefits in rural areas and the interpretation of the take-up of benefits for public policy.’
(Hodge et al., 2000: 1870).
Under a separate contract with the Countryside Agency, the SDRG have also constructed a regional level rural analysis of the indices (Chandola et al., 2000). Our own analysis of the way in which the DETR index depicts deprivation in rural and urban areas is presented in section 4.
Accounting for the cost of rurality
Decisions concerning resource allocation need to take into account not only the presence of poverty but also an accurate assessment of the cost of rurality. In the absence of better indicators allocation models have tended to use existing expenditure as a guide.
However, expenditure on services is the outcome of two quite separate forces. The first relates to demand or need and is driven by factors such as the size of the population and its demographic and socio-economic characteristics, including variations in levels of deprivation. The second relates to the cost of producing and delivering goods and services. This tends to be higher in rural areas because of the absence of economies of scale and the cost of overcoming distance between producers of goods, service providers and consumers.
Pion Economics, describing the operation of the Grant Aided Expenditure (GAE) in Scotland, point out that ‘crucial to the whole structure is the implication that authorities effectively provide the same level of service’ (2000). (GAE are payments above the standard resource allocation for specific services.) As long as this is the case then expenditure will tend to provide a reasonable profile of cost pressures assuming there are no significant variations between authorities in terms of efficiency. However, one response to the existence of higher costs is obviously low and/or declining levels of provision in rural areas. Rural deprivation, as measured by equality of access to services, arises precisely because all households ‘do not have equal access to the same range of services within a stated maximum travelling time/distance’ (Pion, 2000:29).
Indeed, Craig and Manthorpe's (2000) survey of British local authorities found that rural authorities traditionally spent less on social care services and direct provision. They argue that it is no longer sufficient to identify transport difficulties as the main problem for rural areas and suggest that allowance for sparsity in the costs of rural social services is insufficient to cater for the different social care requirements of different types of rural areas. The additional costs of providing accessibility to available services are also more often than not borne by rural residents, rather than suppliers, as evidenced by their greater reliance on private transport even at relatively low income levels.
It may also be argued that rural areas are characterised by suppressed demand for services. Some needs are not registered because of accessibility problems, self-reliance, lower expectations of services or lack of anonymity and are thus not met. The Rural White Paper suggests that ‘information on levels of access will help identify these reasons and (where necessary) develop responses’ (DETR/ MAFF, 2000:18).
The Disadvantage in Scotland Report (Shucksmith, 2000), for example, reported low usage of the welfare and benefits advisory service and suggested that rural people, particularly the elderly, were not well-informed and were reluctant to take up welfare benefits because of what has been termed a dominant rural ideology of self-reliance. In some cases, even calling the doctor is viewed as a last resort. The evidence shows that the elderly, including those in rural areas, are particularly vulnerable in this respect (see Devon Welfare Rights Unit Campaigns, 1998).
There is thus a self-reinforcing cycle of low expectations of provision, low actual levels of provision, and a culture of 'coping', all of which combine to ensure that some needs are not explicitly registered. Indeed, even where area-based initiatives target resources at rural areas there is a suggestion that external agenda, formal requirements for partnership working, competitive bidding regimes, short-term funding and existing power structures limit the effectiveness of rural regeneration initiatives and require new approaches to capacity building in the rural context (Shucksmith, 2000; Edwards et al., 2000). Integrated Area Plans, developed as part of Cornwall's Objective 1 Programme, for instance, propose a delegated fund for capacity building in order to facilitate the necessary community development.
Devising better indicators
A body of research is emerging which addresses the issue of more appropriate indicators of rural disadvantage. This tends to concentrate on factors such as access to employment, services and affordable housing, the quality of employment and low incomes, together with the effects of peripherality and isolation. A first refinement is thus research which introduces such elements explicitly into the equation. These often occur in tandem because, for instance, if one seeks to capture remoteness and the accompanying social or economic fragility it is necessary not only to consider income, employment and industrial structure but also population sparsity, population growth or decline, and accessibility/peripherality (Copus and Crabtree, 1993, 1996).
Introducing service provision and accessibility
Some recent studies perceive levels of service provision and accessibility as the important facets of rural deprivation that are not measured by census-based indicators. A comparison of these indicators with those more traditionally relied upon to detect deprivation shows, in many cases, little correspondence.
Higgs and White draw attention to the need for a research agenda which uses spatial analytical techniques to measure access to key rural services at the community level (1997). Using GIS-based techniques, in conjunction with a database of public services in Wales, they have developed and tested four alternative types of indicators of social disadvantage, each concerned with the provision and/or demand for such key services. The four categories are: levels of service provision; isolation; potential physical accessibility; and public transport dependency (Higgs and White, 2000).
A comparison of these indicators with those more traditionally relied upon to detect deprivation – in this case the Welsh Office Index of Local Conditions, the Index of Local Conditions, Townsend Deprivation score and Breadline Britain Index – showed, in many cases, little correspondence (an outline of the main deprivation indices is given in Appendix 1). The isolation indices, for example, are negatively correlated with the standard deprivation measures (reaching -0.6 in some cases) and the public transport dependency indicator tends to pick out more urban communities. Isolated communities would thus not appear to be highlighted by the usual deprivation measures. Although weak positive correlations of between 0.2 and 0.4 are found between physical accessibility and other deprivation indicators, this still shows communities with limited service accessibility scoring low on the standard indicators. Furthermore, maps of the 50 most deprived communities in Wales as measured by the latter indicator show little correspondence with the accessibility based measures above, indicating how existing measures of deprivation are not effective in detecting areas which experience these aspects of rural deprivation.
A substantial volume of work has also now taken place in both England and Wales to identify and measure access to services as a component of multiple deprivation indices (Social Disadvantage Research Group, 1999). This work has been designed to measure access to services that are generally agreed to be important for everyday living and thus include measures of the distance within each authority to food shops, GP surgeries, post offices, and primary schools.
Pion Economics working with five Scottish Local Authorities also used details on the location of food shops, post offices, GP surgeries, primary schools and petrol stations to construct an average weighted population distance measure for the resident population within each authority, and for residents living both within and outside settlements. They found that non-settlement values were generally substantially in excess of settlement values whilst distances for the statutory service (primary schools) was generally lower than those for other services. Given the constraints of such a ‘rural infrastructure, 'health' players may need to take on extra roles, think laterally and act more pro-actively including the adoption of community development models (Royal College of Physicians, 1998:3).
Dissecting dimensions of deprivation
A second stage has been the recognition that ‘there are uncorrelated dimensions of deprivation and that a measure is needed for each of them’ (Folwell, 1998). This has been taken up nationally in work sponsored by the Rural Development Commission
8
(Dunn et al., 1998b).
This combines related indicators into bundles which together tell a logical story. One bundle, for instance, covers access to employment opportunities and measures not just the registered unemployed but the 'hidden unemployed' together with working age people who have recently moved out of the area (probably to look for work). Another, which addresses access to services, combines indicators on areas with no shop, no regular bus service and no transport to shops elsewhere. Altogether eight such exploratory bundles were defined. In addition to access to employment and access to services, these cover the quality of employment, the vulnerability of employment in the local economy, housing access and affordability, housing quality, low incomes and physical isolation. Each produces an estimate of the number of people experiencing a particular aspect of rural disadvantage. However, as one does not know whether it is the same household facing each different set of circumstances some overlap remains likely and addition does not therefore directly estimate the number of affected people. The output, as the authors stress, ‘remains an indicator, not a direct measure’ (Hodge et al., 2000:1872).
The approach is seen as complementary to the recently developed IMD 2000 (see above). One concern, however, is that ‘this method treats issues of deprivation in rural areas as separate from urban issues of deprivation’ (Noble & Wright, 2000:305). This makes comparison with urban areas difficult and hence requires a parallel study in order to target funds. Noble and Wright suggest this can be achieved by choosing indicators that are relevant to rural as well as urban areas and weighting according to sparsity. Again they suggest that standardised measures of different aspects of disadvantage are required with direct measures of low-income groups being central. The release of Income Support data at ward and then ED level should provide the basis for a weighting formula for sparsity, and with such weightings should be based on empirical research and not on a priori judgements.
Despite this reservation the approach has been tested in both an urban and a rural context, although admittedly within the predominantly rural counties of Lincolnshire, Suffolk and Durham.
9
Here, for example, 30% of the most urban wards fell in the most disadvantaged quartile for the quality of employment bundle. However, just over 20% of the most rural wards were also in this same disadvantaged quartile and there was a higher proportion of rural than urban wards in the next most disadvantaged quartile. Meanwhile, just over half of the most rural wards were found in the two most disadvantaged quartiles, compared to some 40% urban wards. ‘Clearly in neither case is disadvantage shown to be predominantly located in ether urban or rural areas’ (Hodge et al., 2000:1879).
Significantly, the practical application of the bundles also demonstrated how, despite the relatively high correlation between bundles, crosstabulations of the wards by quartile on each dimension found relatively few cases on the diagonal: one third of cases for low income and housing quality, for example, and fewer than 30% for employment by access to services. ‘This difference suggest that the efforts to deal with these two forms of disadvantage should be targeted in different areas’ (Hodge et al., 2000:1881).
Refining the treatment of rural
As noted in section 2, research has begun to be far more sensitive to the differences between rural areas.
Senior et al.
(2000) working in Wales, for instance, distinguished six urban/rural categories based on a combination of land-use characteristics and settlement size.
10
Variations between categories ‘confirmed the highly differentiated nature of rural areas’ which were found to ‘harbour pockets of severe mortality and deprivation’ (Senior et al., 2000; 303).
The bulk of recently published studies focus on the construction of isolation, peripherality and accessibility measures as appropriate indicators of rurality. At the forefront of this work lie indicators based on 'nearest neighbour' concepts and distance to points of service provision such as post offices and GP surgeries.
Martin et al.'s study (2000) of health inequalities in the South West of England investigates three different measures of rurality designed to reveal more about where people live in relation to each other: population density, sparsity and 'nearest neighbour' distances. A comparison of these three measures of rurality showed that only 16% of wards were classed as rural by all three criteria. Two of the measures, population density and nearest neighbour distance, reveal evidence of a U-shaped relationship with deprivation,
11
with deprivation scores at their lowest in the suburbs and rural/urban fringe areas but rising in both the more urban areas and (although less steeply) the more peripheral areas. This pattern is most evident when nearest neighbour distances are used to define rural areas.
When limiting long-term illness (LLTI)
12
is used rather than more general measures of deprivation, this urban/rural U-shaped relationship is even more marked, with wards in the most peripheral county of Cornwall exhibiting the highest rates. The significance of physical isolation suggest that accessibility to public and health services may be an important issue and requires further research (Barnett et al., 2001).
As part of the research for the five Scottish authorities (see above) Pion Economics (2000) developed a nearest neighbour measure for Scotland. The construction of this indicator is very similar to that of the dispersion indicator that currently plays a role in resource allocation. This again demonstrates the much higher isolation that exists in some rural areas and, even within such areas, the extensive isolation that can exist outside settlements.
An alternative approach to peripherality employs principles of the gravity model in which the degree of interaction between two points is a positive function of their economic mass and an inverse function of the distance between them. Economic mass may be measured by variables such as employment or population and distance in either space or cost terms. Communities that are far from major centres are therefore predicted to have a low degree of interaction with such centres and hence a higher peripherality score. Threshold peripherality scores are then selected in order to classify communities as lying in remote, intermediate rural, or urban areas (Copus and Crabtree, 1996).
13
Using direct measures rather than proxies
Another significant development has been the increasing use of direct measures of need rather than proxies. Work conducted at the University of Oxford (see Noble and Smith, 1998) has been key in not only demonstrating the utility of benefits data in describing the distribution of need but also increasing the availability of such administrative data. The methodology has several major advantages, including the fact that the information is up-to-date and can be repeatedly extracted. It is also postcoded, allowing geo-referencing at enumeration district level. It allows the identification, and separate analysis, of various claiming groups such as pensioners and lone parents – a key advantage if we are to understand better the impact of poverty on different sections of the population and tailor services to promote social inclusion.
This work has now been extended to rural areas with Noble and Wright (2000) looking at housing benefit and council tax benefit recipients in the districts of three predominantly rural counties: Dorset, Shropshire and Wiltshire. This, despite acknowledged (but unquantified) problems of under-claiming) reveals ‘there are significant numbers of people living in rural areas who are experiencing poverty’. Over one-quarter of Shropshire's HB/CTB claimants, for instance, live in the least densely populated quintile of English wards.
It also demonstrates how benefit dependent households can be identified at small area levels in rural areas - a potentially useful component of national indices of deprivation for the allocation of central government resources to LAs. Indeed, given the evidence from these three shire counties that existing indices of deprivation do not correlate well with this direct measure of low income in areas of low population density it stresses that the direct measures are ‘inherently more satisfactory measures than the proxies’ (Noble & Wright, 2000; 299).
In a similar study Pion Economics (2000) looked at the claimant rates for housing and council tax benefit in both LA and other sector properties. Like Noble and Wright, they found that the most urban authorities did generally experience the highest claimant rates but that again the rates declined across rural areas in only a moderate manner from that in more urban authorities. The notion of a ‘rural idyll devoid of the problems of deprivation was thus difficult to sustain’ (Pion Economics, 2000).
In recent years there has been considerable progress in developing needs estimates that are both directly tailored for specific service sectors and that are created for functionally meaningful units rather than census administrative areas. Linking individual pupil postcodes with census administrative data at the enumeration district level, Gibson and Asthana (1998) have produced an Index of Educational Disadvantage that can be calculated for individual schools. Used in OLS 'best-fit' prediction models, this index was found to have significantly higher 'explanatory power' in predicting examination performance than the more commonly used proxy of free school meal entitlement (Gibson and Asthana, 2000a; 2000b).
Within the health sector, Gibson et al. (2002) have also developed a method of deriving morbidity-based needs estimates. These have been calculated by linking age/sex/class prevalence tables derived from the Health Survey for England (HSE) with the age/sex/class profiles of general practice and Primary Care Trust populations. When these estimates are compared with those based on standard proxies, a very different picture of variations in health needs between rural and urban areas emerges, depending on the disease/condition examined. Thus, whilst the prevalence of mental disorder is particularly high in urban areas of high social deprivation, rural areas serving demographically older populations have the higher prevalence rates of conditions such as coronary heart disease.
The growing availability of direct needs estimates raises fundamental questions about the purpose of using traditional measures of deprivation. When, as is often the case, deprivation is used as a 'proxy measure' for specific service need, direct methods arguably provide more robust and transparent indicators than general deprivation indicators. The use of the latter should therefore be increasingly confined to work that aims to provide a general (or summary) account of deprivation per se.
Targeting rural disadvantage: places versus people
The work commissioned by the Rural Development Commission has moved some way towards being able to focus on people rather than places. The methodology provides an indication of the numbers of those disadvantaged in each bundle in each ward. It is thus possible, in principle, to identify an 'efficient' pattern of targeting, defined as the designation of the smallest number of wards which will include a given number of people living in disadvantaged circumstances.
Focusing just at the most rural end of the spectrum they firstly calculated the numbers of people within the Rural Development Area and then, for each bundle, defined a list of the most deprived wards (target wards) until the cumulative numbers in disadvantage matched those included within the designated RDA wards. In every bundle the target wards had a higher average percentage of disadvantage ‘suggesting the extent to which it might be possible to direct policy in a more focused way’ (Hodge et al., 2000:1880). For housing accessibility, for example, the methodology identified 49 target wards where some two-fifths of households were disadvantaged as opposed to some 96 RDA wards where approximately one-fifth were disadvantaged.
The Peak District Deprivation Forum also independently tested the bundles and continue to stress that because income disparities occurred on a house-to-house basis there will always be limits to the use of area-based methods in identifying rural deprivation.
Haynes and Gale (2000) make a similar point when they suggest that the differences between urban and rural correlations with poor health are a reflection not of the choice of deprivation indicators or census areas but a product of the greater internal variability, smaller average deprivation range and smaller population size of rural areas. In consequence ‘deprived people with poor health in rural areas are hidden by favourable averages of health and deprivation measures and do not benefit from resource allocations based on area values’ (2000:275).
The Local Government Association also add weight to this argument when they query, from a rural perspective ‘whether the balance is right between...area based policies and people based policies’ (LGA Rural Executive, 2000:2). Similarly, the results of Phimister et al. (2000) looking at the dynamics of low income in rural areas ‘emphasise the need for “client” based measures…addressing the needs of specific groups such as the elderly’ (2000:415).
Conclusion
As the above discussion makes clear, the critique of the use of traditional measures of disadvantage in rural areas is wide-ranging. Concerns have been expressed that indicators which capture deprivation in an urban context (e.g. car ownership, unemployment) should not be expected to perform similarly in rural areas. When more focused studies have been undertaken, significant problems relating to employment, low incomes, housing, transport and education have been revealed, casting doubt on the validity of previous evidence on rural/urban variations in disadvantage.
The assumption that rural environments are inevitably 'healthier' is also open to challenge. Rates of suicide and road traffic accidents have been found to be higher in rural areas. Rural mortality advantages disappear after controlling for socio-economic status and limiting long-term illness appears to be subject to a U-shaped pattern of prevalence, highest rates being observed in the most urban and the most peripheral areas.
Unless rural agencies can make a clear statement of levels of need and of the funding they require to deliver services to their communities, there is a danger that rural areas will suffer from lower levels of service provision relative to their needs than their urban counterparts. As part of this, methods of accurately accounting for the additional costs of providing services in rural areas should be developed. Recent studies that attempt to measure levels of service accessibility go some way towards this. Efforts are also being made to devise more appropriate indicators of disadvantage itself. These include measures that reflect multiple dimensions of deprivation (e.g. the DETR 2000 Index of Multiple Deprivation and the RDC commissioned work on 'bundles' of indicators). Another significant development has been the use of direct measures of service need rather than proxies.
We have argued that, when a measure of a specific service need is required, the potential of using direct estimates rather than deprivation as a proxy should be explored. There remains a need, however, for summary accounts of deprivation in different areas and it is important that these are as effective at capturing disadvantage in rural as in urban areas. In the following section, we explore in more detail the range of different summary measures that are available. To this end, we carry out an empirical analysis, comparing the way in which the DETR's IMD 2000 and more traditional indicators depict deprivation and reflect health status in rural and urban areas.
