A report for Friends of the Earth by Duncan McLaren, Olivier Cottray, Mary Taylor, Susan Pipes and Simon Bullock.
This report is based on the data and support provided by Business Geographics, Autodesk and Kingswood Ltd and ESRI.
Friends of the Earth would particularly like to thank Ed Parsons of Autodesk for all his help.
Friends of the Earth would also like to thank Laura Winton, Mike Childs, Zoe Gillard, Chas Linn, Leslie St James, Joe Short and Ian Willmore for all their help, comments and guidance.
Published by Friends of the Earth Trust, April 1999
All across England and Wales the poorest families (reporting average household incomes below £5,000) are twice as likely to have a polluting factory close by than those with average household incomes over £60,000. This is the sharp end of social exclusion - on top of unemployment and crime - these families and communities face the grime of industrial pollution. Here pollution is as far from a middle-class concern as it can get.
Over ninety per-cent of London's most polluting factories are located in communities of below average income. London is just the most extreme example. A similar pattern is found throughout England and Wales. Overall, almost two-thirds of the most polluting industrial facilities are to be found in areas of below average income.
Using the interactive map even at a national scale (England and Wales) the visual correlation between polluting factories and low income areas is clear (although in some areas the density of factories is such that the (red) colour of the area below is masked!).
In the North East areas with the most polluting factories have average household incomes almost £2,000 (or 14%) lower than where there are no such factories, and have over 20% more of the poorest households (with incomes below £5,000 per year) than where there are no such factories. Across England and Wales as a whole the income difference was over £1,500 (or about 9 per cent) and there are over 10% more of the poorest households in areas with the most polluting factories.
The effects are more severe in areas with multiple factories. At the extreme, Seal Sands on Teesside has 17 of the most polluting factories in one small area. The average income here is just £6,200 (just 45% of the regional average income, or 36% of the national average) and over half its households have annual incomes under £5,000.
This is a clear cut issue of 'environmental injustice' in which poorer people are subjected to greater risks and impacts of pollution, and have less control over their environment while the benefits of the industrial activity largely accrue elsewhere. Measures to reduce industrial pollution from these factories would be clearly socially progressive.
This report first sets out the methodology used, then outlines the analysis of the maps and data (using charts and tables), followed by a discussion of the possible relationships. Data interpretation issues are addressed in an Appendix.
This research has linked two data sets in a Geographic Information System and database analysis. The data sets are:
The latter data set effectively covers the 'entire population' - all the potentially most polluting factories in England and Wales - while the former has been weighted according to census data by the data provider, so can be treated as a statistically valid representation of the entire population. Further explanation of the underlying data is given in the Annex.
The maps show the location of IPC sites and the average household incomes of postcode sectors. The visual correlation between IPC sites and lower income areas is striking at all scales from the whole of England and Wales down to specific urban areas.
The data allows us to undertake two main types of analysis:
In both cases we can examine whether the income characteristics of households or areas are different for postcode sectors with IPC sites in them (the 'IPC sectors') than for all sectors, or for those with no IPC sites (the non-IPC sectors)(2).
In this section we present a number of alternative analyses of the data available, beginning with analysis based on sector incomes:
We then turn to two analyses by income bands:
Then we use income band analysis nationally and by region to estimate the size of the income effect:
Finally, we return to the effects of concentrations of IPC sites:
This section answers a series of key questions in turn. First is there a pattern to the data, suggesting a relationship between income and IPC sites, or is it just random? Second, what is the direction of any relationship - are there, as we might predict, more sites in lower income areas? Third, how strong is the relationship? This we look at in two ways, the relative chance of living near an IPC site according to income, and the average effect on local incomes of an IPC site. Finally, we ask whether greater numbers of IPC sites in a sector have a reinforcing effect.
Table 1 below compares the actual distribution of IPC sites according to average income by postcode sector, with the distribution we might expect if the IPC sites were equally distributed according to population. As they are designed to contain a consistent number of addresses, postcode sectors can be assumed to be of the same average population (rather than the same average area). The table also examines the distribution of IPC-authorised processes, of which there may be several at any given site.
Table one [back to text]
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Table 1 (and Graph 1) shows an apparent pattern with a consistent and significant deviation from the 'random' distribution clearly visible across the income scale. In the poorer areas (up to £15,000 average income), there are far more IPC sites than would be predicted on the basis of their total population, whilst in the richer areas (over £20,000) there are many fewer sites than the 'random' distribution would suggest. In the intermediate band (£15-20,000) in which average household income falls (£17,280), there are almost exactly as many sites as the 'random' distribution suggests.
Graph One[back to text]
Graph Two[back to text]
The pattern is similar, albeit even more striking, when we look at the distribution of processes (Table 1 and Graph 2). Quite clearly the IPC sites with more than one authorised process are more concentrated in poorer areas than IPC sites on average. We return to this question later.
By using the interactive mapping system it can be clearly seen that the majority of IPC sites in each region are concentrated in lower income areas, indicated by more intense red background shading. The pattern is most distinct in Yorkshire and Humberside, the North-West, the North-East and the West Midlands where both IPC sites and low income areas tend to be clustered and concentrated in traditional industrial urban areas such as Manchester and Birmingham. On the other hand, the relatively wealthy peri-urban regions of the same cities have relatively few IPC sites. The pattern is least clear - although still visible - in the more generally high-income South East and Eastern regions. We return to regional commentary later.
The above analysis is based on the average income of each post-code sector, which tells us nothing about the distribution of income within each sector. Below we aggregate data from income bands across all sectors, but first we look at another simple indicator of poverty: the proportion of households in the lowest income band (less than £5,000 pa).
On average, this is around 19% of households, but it varies widely - with 0.5% of sectors (42) having 50% or more of their households in Band 1. Those 42 sectors contain 23 IPC sites (1.74% of the total) - or over three times as many as would be expected were the sites distributed randomly. There are 978 sectors (12.3%) with 30% or more of households in Band 1(3) and these contain 261 (19.8%) of IPC sites - or over half as many again as would be expected.
In Table 2 below we look simply at the proportion of IPC sites found in areas of below regional average income - for all the regions of England and Wales. This is a crude but still useful indicator of the extent to which these sites are found primarily in poorer areas(4).
Table 2 shows a strongly consistent pattern, with just under two thirds of IPC sites being located in postcode sectors of below average income. Two regions exhibit starker relationships - almost 80% of the North-East's polluting factories are in communities of below average income and over 90% of the capital's polluting factories are in such areas! This may actually be due to the fact that London has no poor rural communities within its boundaries, and the possible masking effect on the overall correlation that such areas - poor, but with no IPC sites - would have is not present. The area with the least strong pattern is the East Midlands with only just over half of polluting factories in poorer areas. If we look at the maps we can see some areas of severe poverty which would appear to fall in ex-coalfield communities in this area, an effect which may also be having an impact in Wales. In both cases this may contribute to the less strong relationship found.
Table two [back to text]
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Once again, we can also analyse the distribution of IPC authorised processes (rather than simply sites). As we found above, this makes the pattern of discrimination more striking. Almost three-quarters of all processes (72.8%) are located in areas of below average income.
Tables 1 and 2 are based on average incomes for each sector. A more sophisticated comparison can be achieved by examining the distribution of household incomes in postcode sectors with or without IPC sites with the average distribution for all sectors. Such analysis eliminates any distortions that may be introduced by using the average income of the postcode sector as fully representative of its income characteristics and instead takes full account of the distribution of incomes within the relevant sectors.
Table 3 and Graph 3 show a small but significant pattern - lower income band households make up a higher proportion of the total in postcode sectors with IPC sites, whilst higher income band households make up a smaller proportion. The inverse holds true for non-IPC sectors. At the extremes, there are over 10% more Band I households in IPC sectors than in those with no IPC sites, and almost only half as many (48% less) households in band 10 in sectors with one ore more IPC sites, than there are in sectors with none.
In Graph 3 we have normalised the distribution of households by income band for sectors with no IPC sites. This highlights the percentage differences from that distribution that are found in the sectors with IPC sites (where there are more poor households and less rich households).
Table three [back to text]
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Graph three [back to text]
Table 4 below shows broadly the same phenomenon as Table 3, but here we calculate the proportion of households in each income band that share their postcode sector with a IPC site.
More than one in eight of the lowest income households (less than £5,000 pa) lives close to a IPC site (by this definition) whilst for the richest (£60,000 pa plus), only one in sixteen do so. Comparing the poorest fifth of the population with the richest fifth(5) across the whole of England and Wales the proportion found in postcode sectors with IPC sites declines from over one in eight (12.7%) to one in ten (10.0%) respectively - a difference that equates to an additional eighty-four thousand poor households exposed to pollution.
Table four [back to text]
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
As well as quantifying the effect in terms of 'chances' we can quantify it in terms of the size of the income differential. In Table 5 we compare the average income of households in postcode sectors with IPC sites with the average for the whole population, and the average of those not in IPC sites. Throughout the analysis in this section we have rounded average annual incomes to the nearest £10.
Looking at the bottom line of Table 5 we find that in comparison to the overall average, average income is £1,380 pa (or 8.0%) lower in postcode sectors containing IPC sites. Compared with sectors with no IPC sites the contrast is a little starker - household incomes on average being £1,560 (or 8.9%) lower in sectors with IPC sites than in sectors without such polluting factories.
The income variations we have found are of a similar order to those between regions - regional average household incomes vary from £14,750 in Wales to £20,740 in London. Table 5 therefore presents the income analysis at a regional level too, comparing average incomes in sectors with and without IPC sites with regional average household incomes. The table shows quite a variation in this effect, with no immediately clear pattern: the North-West and Yorkshire and Humberside showing relatively little income depression in IPC sectors in general, whilst the North-East has the highest variation outside London.
Table five [back to text]
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
So far we have mainly concentrated on the comparison between postcode sectors with one or more IPC sites and those with none. Similarly in examining the effects by income bands, we have concentrated on the presence or absence of any IPC site within the relevant sectors. However, there are many sectors with two or more sites, and, as previously noted, any given site may have more than one process. Here we examine the relationship between average incomes and the proportion of households in the lowest income band and the number of sites.
Table 6 shows the effect of increasing numbers of IPC sites. As we saw above, the presence of one or more sites is associated with an 8% lower average household income. This effect grows with the number of sites in the sector - those sectors with two or more sites have 10% lower incomes, four or more sites is associated with 15% lower incomes, and seven or more with 20% lower incomes. There are only a handful of sectors with even more sites, but the trend continues: the two sectors with the most IPC sites have an average income almost 50% below the overall average, whilst the sector with the most sites (17 in all) - in Teesside - is one of the poorest in England and Wales with average household incomes 64% below the norm and over 50% of it's population in band 1(with household incomes below £5000pa)(6).
Table six [back to text]
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
The final column of Table 6 shows that there is a similar pattern in the distribution of households in band1, with higher shares in sectors with more IPC sites.
Both measures are highly suggestive of a strong link between more pollution (or at least the potential of more pollution) and more severe poverty. Across England and Wales there are 11% more households in Band 1 (incomes below £5,000) in postcode sectors with IPC sites, than elsewhere. Graph 4, shows the same relationship visually.
Table 7 shows how this effect varies regionally, concentrating on the share of households in band 1. It presents figures for one or more, three or more, and eight or more sites to give a broad picture of the effects.
Rather like Table 5, Table 7 suggests that the pattern is not as consistent within regions as it appears to be across the country as a whole. However, with only one exception (the North West) there are greater percentages of band 1 households in sectors with IPC sites than in those without. In the extreme cases, again, London and the North East, the difference is over 20%. Even in the North West the expected effect emerges for sectors with multiple IPC sites.
Graph four [back
to text]
Before turning to a general discussion, it is worth examining the regional maps to seen if any reasons can be suggested for the different patterns and variations in the relationships found in different regions. Below we comment on each region in turn.
Table seven [back to text]
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
East Midlands: This region stretches across diverse areas from extremely poor ex-coalfields (with no IPC sites, although often severe dereliction) to relatively wealthy parts of the home counties. Although IPC sites do appear to be clustered in poorer urban areas in Derby, Nottingham and Leicester, there are enough sites scattered in less poor areas, that combined with some very poor areas without IPC sites, this region shows only a weak numerical relationship on all the measures we have calculated.
Eastern: This region includes both relatively rich home counties and poor rural areas in East Anglia. The map reveals clustering of IPC sites in poor areas along the north side of the Thames estuary and around Ipswich. Although there also appear to be many sites in the more wealthy south-west of the region, several of these are actually in small poor areas, and, because the south-west is more densely populated, the average income of the region is higher than a cursory glance at the map would suggest. As a result there is a clear relationship (although quite small) between poverty and pollution within this region, seen in all the numerical indicators.
London: This region has the highest average income and a very strong visual clustering of IPC sites into poorer areas in the East Thames Corridor and the Lea Valley. This visual relationship is replicated in the numerical indicators, all of which suggest that the inequalities associated with IPC sites are severe in the London region.
North East: The map of this region suggests very strong clustering of IPC sites in poor urban centres on Teesside and Tyneside. Indeed so strong is this effect, that despite the low average income of this region, all the numerical indicators suggest severe inequalities associated with IPC sites. There are a number of sectors in the region with very high numbers of sites within them (Seal Sands with 17 sites being the most in the country). This will help clarify the apparent relationship.
North West: The North West is an intriguing region. Visually, the map does not look that dissimilar from the North East, with IPC sites clustered in Manchester and on Merseyside but the numerical indicators suggest a much less strong relationship. In fact the region is unique in having less households in band 1 (on average) where IPC sites are present than where they are not (although this inverse relationship does not hold where there are three or more sites in a sector). A closer look at the map suggests that the densely populated areas of severe poverty in Liverpool, where there are no IPC sites could contribute to this effect at a regional level. This would suggest that a finer scale of analysis might well reveal patterns that are partly concealed at the regional level.
South East: The South East is similar to the Eastern region in pattern and indicators, although it has a higher average income and smaller areas of peripheral poverty. Like the Eastern region it has clusters of IPC sites on the Thames Estuary, and also around Portsmouth and Southampton. Combined with the overall relative wealth of the region this tallies well with the small but clear relationship indicated in the tables.
South West: The South West also spans a geographical range of prosperity, from the quite wealthy Western end of the M4 corridor, to the severe poverty of sparsely populated peripheral Cornwall. IPC sites in the region are clustered around Avonmouth and in Bristol, and to a lesser extent, around Plymouth and Bournemouth. As with the Eastern region the map perhaps suggests a lower average income than is the case (as a result of relative population densities), and once again there is a clear although small relationship between poverty and (potential) pollution.
Wales: Wales is another interesting region, with clusters of IPC sites, as expected, around Deeside, Swansea, Port Talbot, Cardiff and Milford Haven. However two factors would appear to be responsible for reducing the numerical relationship revealed in the tables. First, there is severe poverty in the Valleys areas, associated with ex-coal mining settlements, whilst enough of the IPC sites around Cardiff fall into relatively high-income areas to possibly affect the overall picture.
West Midlands: The West Midlands shows a clear relationship, both on the map and in the tables, despite having a relatively high share of IPC sites in areas of above average income. The clustering in the densely populated Birmingham-Wolverhampton conurbation and also in Coventry and Stoke on Trent is quite pronounced, but there are also enough scattered sites elsewhere that the relationship is not as strong as that in the North East.
Yorkshire and Humberside: In many respects, Yorkshire and Humberside is not dissimilar to the North West. Despite an apparent strong clustering of IPC sites in the main industrial and port cities: Leeds, Sheffield, Hull and Grimsby, the numerical indicators suggest at best a weak relationship. This could be partly explained by the relatively widespread scattering of IPC sites over the former metropolitan counties of South and West Yorkshire, many of which are in areas at or near the regional average income, and also partly by some areas of more severe poverty in Hull and in ex-coalfield areas in South Yorkshire, where there are no IPC sites. Unlike the North West, in this region there is not such a clear trend of more sites associated with lower incomes or more households in band 1. Indeed in this region the sector with the most sites (8) is noticeably above the regional average income.
In this commentary we have not been able to do more than begin to tease out different relationships. A more sophisticated approach would examine not only smaller areas, but use a finer scale of income data, and categorise the IPC sites according to their nature and/or scale, so as to capture effects which might arise from the differences between authorised processes.
The above analysis shows that poorer people in England and Wales are more likely to live in close proximity to a potentially polluting factory than richer people. It also shows that the relationship is stronger with increasing numbers of factories in an area.
There are different possible reasons for this. Industrialists might have located their factories in already poor areas, or the proximity of the factory might have degraded the area to the extent that richer people have left (as they have the choice), leaving the poorer people to 'grin and bear it'. What is certain is that, whatever the direction of the link, the injustice is real. Whatever the health impact may be from these factories - and it will vary greatly - this preliminary analysis shows that there is a disproportionate impact on poorer people. This suggests that action to tackle pollution from factories will reduce inequalities.
There are many uncertainties however. For example, it is not possible to assess the relative effects of income, race, housing quality, housing and planning policy, urban redevelopment and differential migration in generating the patterns we find in the maps and tables, although all these factors are interlinked and seem likely to have contributed in some way - albeit not consistently.
This analysis mirrors much more detailed work on “environmental justice” in the United States, where poorer communities - and in particular ethnic communities - bear the greatest burden of toxic pollution. Research undertaken in the USA, by academics, activists and the US Environmental Protection Agency, has shown an apparent causal link in the citing of very polluting factories in poorer areas, especially those dominated by ethnic minorities (Bullard 1996; US EPA, 1992) - so called “environmental racism”. The benefits factory owners have obtained from locating their factories in poor, black communities include not only (in general) less opposition to development but also weaker enforcement of environmental standards and regulation (Bullard, 1999; Lavelle and Coyle, 1992).
Although we might expect to find concentrations of ethnic minorities near polluting factories in some parts of this country, in others race might be almost irrelevant, and, for example, the housing policies of the local authority may be most significant in shaping the pattern of residential location for poorer people (Walker, 1999). The greater degree of residential segregation on ethnic grounds in the USA means that the correlations between more risk and pollution and lower income are generally stronger than those we have identified in this report. However, we can expect that the driving economic forces are little different in the UK.
Any link between the location of factories and poverty is only one of a complex web of issues. The extent to which proximity to polluting factories equates to actual exposure to pollution also depends on a range of factors, whilst directly linking exposure to health or other impacts is also complex and difficult.
Even if the pollutant is emitted to air, the combination of chimney height, wind direction and other atmospheric conditions will determine the shape and direction of the pollutant plume. Nonetheless, proximity remains a significant factor increasing the likelihood of continued or repeated exposure - at least to air pollutants (and to a lesser extent ground- and surface water pollutants). Proximity is also a key factor in the likelihood of exposure in the event of a catastrophic accidental release - which is a real possibility for many of the facilities on the IPC list. It should be noted that the IPC list is not identical to that administered by the Health and Safety Executive (HSE) for the purposes of controlling Major Accident Hazards but the overlap is substantial(7). Thus it is helpful to outline the major emissions to air from IPC sites as a whole (see table 8). Many of these are hazardous and health-threatening.
Table Eight [back to text]
| ||||||||||||||||||||||||||||||||
