Session Number: 6A
Session Title: Measures of Poverty and Social Exclusion
Paper Number: 4
Session organiser: Stephen Jenkins, University of Essex, UK
Discussant: Lene Meyer, Eurostat

 
For additional information please contact:
Jonathan Bradshaw
Social Policy Research Unit, University of York, Heslington, York, YO10 5DD, UK.
fax: 00 44 (0)1904 433477
phone: 00 44 (0)1904 433480

This paper is placed on the following websites: www.stat.gov.pl
                                                                      www.econ.nyu.edu/dept/iariw

It has been claimed that social exclusion is different to poverty. Room (1995) for example has claimed that social exclusion

• is a broader concept than poverty;
• is more dynamic;
• moves from the individual, through the family and the household to the neighbourhood;
• is a relational issue – about participation, power and integration rather than merely finance;
• is a catastrophic condition rather than merely a point on a distribution.

Others have sought to hang on to poverty, rather than adopting social exclusion as a new and different distributional notion. This is partly due to the ideological baggage associated with social exclusion – what Levitas (1999) called MUD (the Moral Underclass Discourse) and SID (the Social Integrationist Discourse) rather than RED (the Redistributive Egalitarian Discourse). But it is also because poverty in its broadest conceptualisation and in its empirical operationalisation already incorporated some or all the elements that Room claimed were different. So for example

ˇ poverty had not only been about money, since the deprivation indicator methodology developed by Townsend (1979),

ˇ the dynamics of poverty - spells and episodes are increasingly being researched (Bradbury, Jenkins and Micklewright 2001)

ˇ poverty has always been studied at individual, family and household level as well as spatially (Bradshaw and Sainsbury 2000 a and b)

In fact social exclusion has been contrasted with a parody of the concept of poverty.

But to date the literature on social exclusion has been largely theoretical and there is no real way to resolve the debate as long as social exclusion remains a theoretical construct. However attempts have now begun to be made to operationalise the notion of social exclusion in social research. For example Burchardt et al (1999) have developed a social exclusion index using questions in the British Household Panel Survey. A Dutch team have been working for Eurostat to develop a measure of social exclusion based on the questions in the European Community Household Panel Survey (Dirven et al 2000). Now this paper’s authors have undertaken a national household survey in Britain especially designed explore the interaction of poverty and social exclusion.

We are at an early stage of that exploration and this paper is an initial attempt at the kind of analysis that we might undertake.

This was funded by the Joseph Rowntree Foundation and undertaken by the UK Office for National Statistics in autumn 1999. The sample was a subsample of respondents to the 1998/9 General Household Survey, weighted to boost the bottom three quintiles of the income distribution. The achieved sample (just over 1500) was reweighted to match the known characteristics of the respondents to the GHS. As well as the data derived from the PSE questionnaire itself we also had the data from the General Household Survey itself, including very good income data (which was updated for changes between the GHS and PSE interviews). Further information on the survey can be found on the project website (www.bristol.ac.uk/poverty/pse) and in the first published report of the study (Gordon et al 2000 to be published on September 11 2000). This analysis is undertaken on 1200 households for which we have complete data on the three poverty measures used.

POVERTY

This paper employs three different poverty measures representing three distinct traditions of poverty research.

Income poverty

Those households with net equivalent household income before housing costs less than 60 per cent of the median. The study employed a variety of equivalence scales, including one created especially, based on budget standards research. But for this audience we have used the modified OECD scale. 19 per cent of households were poor using this measure.

This is based on the social indicator methodology pioneered by Townsend (1979) and developed especially by Mack and Lansley (1993) and Gordon and Pantazis (1998). For the PSE survey we developed a new and more elaborate index than previously (including a separate index for children). We established the proportion of the general population who considered an item was a necessity in a survey that preceded the PSE survey. Only items that 50 per cent or more of the general population considered were necessities were included in the index (see Appendix 1). After having done some work on the validity of the index (and excluding some items, which did not contribute significantly) we identified a threshold of lacking two or more items as a threshold of poverty. 26 per cent were poor using this measure.

This follows the tradition pioneered in the Benelux countries of establishing poverty by asking respondents whether they are poor or not. We used three sets of questions including an attempt to operationalise the Absolute and Overall notions of poverty adopted by the UN World Summit on Social Development in Copenhagen in 1995. But this paper uses the results obtained from the following questions.

20 per cent were poor using this measure.

Table 1 summarises the overlap between these poverty measures. Only just over half the income poor are necessities poor and less than half the income poor are subjectively poor. Less than half the necessities poor are income poor. Most overlap is between the necessities poor and the subjectively poor – over 70 per cent of the subjectively poor are necessities poor. Whilst 38 per cent are defined as poor by at least one of the measures, only 7 per cent are poor by all three measures.

This lack of overlap between the poor defined by different measures is striking. There are a host of reasons why there should be some failure of overlap in these measures of poverty but we will avoid a discussion of these because the purpose of this analysis is to observe overlap between these poverty measures and indicators of social exclusion.

Members of the PSE team (Levitas, Pantazis, Patsios and Townsend in Gordon et al 2000) have distinguished between four dimensions of social exclusion: impoverishment or exclusion from adequate income or resources; labour market exclusion; service exclusion; and exclusion from social relations. The first of these aspects - poverty itself - is represented in this paper by the variables outlined above. The other three elements were operationalised using the questions in the PSE survey.

Attachment to the labour market is held to be important for individuals not just because it is seen as a route to an adequate income but because it is an important arena for social contact and social interaction. An individual living in a jobless household may as a result be living in poverty, be service excluded and excluded from social relations. Jobless households are households where there is no-one in employment (or self employment), including both those who are retired, and those of working age. In the PSE survey 21 percent were retired households (11 per cent of who were 55-64), 13 per cent were jobless households and the rest 66 per cent had employed persons in the household. The very high proportion of the population who are inactive should lead us to be cautious about treating labour market inactivity in itself as social exclusion. So for the purposes of this analysis we have not included as the labour market excluded those households with retired persons of pensionable age or student households. The result is that 11 percent of households are labour market excluded.

One aspect of social exclusion is lack of access to basic services, whether in the home (such as power and water supplies) or outside it (such as transport, shopping facilities and financial services). We asked about disconnections of water, gas, electricity and telephone and whether people had restricted their use of these services because of cost. Five per cent had experienced disconnection from one or more services and 11 per cent had used less than they needed because they were unable to afford them. Then respondents were asked about a range of public and private services outside the home and identified whether they did not use them because they were unavailable, they could not afford to or because they did not want to (see Appendix 2). We then counted the number of private and public services that households lacked because they were unaffordable or unavailable and found that 24 per cent lacked two or more and 13 per cent lacked three or more.

It was decided not to include the disconnected and restricted use of utilities indicators on the grounds that the questions asked about whether they had ever done this rather than about now or recently. So one indicator of service exclusion was produced – those lacking three or more services (13 per cent).

A unique feature of the PSE survey is that it seeks direct information about social relations and social participation. In the PSE survey exclusion from social relations has been examined through; non-participation in common social activities generally regarded as socially necessary; isolation; lack of support; disengagement and confinement. The measures for each of these will be discussed in turn.

These are the activities (i.e. not the assets) in the list of socially perceived necessities (see appendix 1). Of these common social activities which people are excluded on grounds of cost (and here we have also included those that less than 50 percent of the population consider necessities), 63 per cent lack none, 11 per cent lack one, 7 per cent lack two and 20 per cent lack three or more. We used three or more.

This measure was derived from questions about the frequency with which respondents spoke to a particular family member outside their household or friend with whom they are in daily contact, including both face to face and telephone contact. As elsewhere there is a judgement to be made about the appropriate threshold for this analysis but we have chosen people who say that they do not have contact with family or friends daily (12 per cent).

One indication of the existence of functioning social relationships and networks is the amount of practical and emotional support potentially available to individuals in times of need. Respondents were asked how much support they would expect to get in seven situations, including support from members of the household, other family and friends and any other means of support. Four items related to practical support; help needed around the home when in bed with flu; help with heavy household or gardening jobs; help with caring responsibilities for children or elderly or disabled adults; someone to look after the home or possessions when away. Three related to emotional support: needing support about important life changes; someone to talk to if depressed; and someone to talk to about problems with spouse partner. Out of the whole sample 54 percent had support in all seven circumstances, 23 per cent lacked support in at least four out of seven areas and nearly 2 per cent lacked support in all areas. We used four or more.

Lack of civic engagement is sometimes deemed to be an important aspect of social exclusion. Respondents were asked which of a list of activities they had done in the last three years and whether they were actively involved in any of a comprehensive range of organisations. We found that 10 per cent were disengaged from all activities and that 28 per cent were disengaged or only voted. We used the totally disengaged.

Participation in social activities and social contact beyond the household depends on being able to get out and about. People who are not able to move freely may be effectively excluded from full social participation. We asked people to identify the factors reducing participation in common social activities. The most important factor was ‘can’t afford to’ (47%), next was ‘not interested’ (44%), then there were the range of reasons summarised in Appendix 3. We excluded those who were ‘not interested’ and identified the rest as confined for reasons outside their control – 29 per cent.

Another form of confinement is personal safety and 30 per cent of the sample report feeling unsafe walking alone after dark. Table 2 summarises the results obtained from these elements of social exclusion

The purpose of the analysis is to explore the interaction between poverty and social exclusion. We start by showing the proportions of the those socially excluded according to each of our indicators of social exclusion that are poor by each of our measures of poverty in table 3. It can be seen that the proportion of the excluded that are also poor varies with the poverty definition. In general the socially excluded are more likely to be necessities poor than income or subjectively poor but this is partly a function of the fact that a greater proportion of the sample are necessities poor. Labour market exclusion, inability to participate in three or more activities and being confined are the elements of social exclusion most associated with the poverty measures. In most elements of social exclusion (and for all measures of poverty) the socially excluded are more likely than average to be poor. The exceptions are the isolated and those who lack support, who are no more likely to be poor by all measures (this is an important finding and needs to be pursued – it may be because paid work is an obstacle to forming social relations).

Table 3: Proportion of socially excluded who are poor

Then table 4 shows the proportion of the poor who are socially excluded. The income poor have the highest proportion of labour market excluded. Over half the subjectively poor are unable to participate in three or more activities and/or confined. Over half the necessities poor are unable to participate in three or more activities (though note that these measures are not independent because they use some of the same data). Also nearly two thirds are confined. Again the poor are no more likely than the rest to be isolated or lack support.

Using these elements of social exclusion we created two indices. The first counts how many classes (labour market/service excluded/social relations excluded) that the respondents experience - maximum possible = 3. It can be seen in Table 5 that there is a clear association between poverty and the number of components and for example over two thirds of the socially excluded on all three components are subjectively poor whereas less than 10 per cent of those not socially excluded are poor - by all measures. 

 Table 6 shows the number of items on which a person is excluded – thus each of the elements of social exclusion in Tables 3 and 4 are cumulated (except that we used disengaged from all activities instead of disengaged from all activities except voting) - maximum score = 8. Again we see that there is a clear association between social exclusion and poverty. The more items that a household is socially excluded from the higher the poverty rate - for all poverty measures. 

 This is explored further in Tables 7 and 8 which compare the intensity of social exclusion with the intensity of poverty. It can be seen in Table 7 that there is a strong association between the intensity of poverty and social exclusion – the more measures of poverty the household is poor on, the more components of social exclusion they are excluded on - the cases tend to concentrate at the top left hand and bottom right hand of the Table. Thus there are only three cases who are poor on all three measures but not socially excluded on any item and six cases which are nor poor on any measure but excluded on all components. 

Table 7: Relationship between the intensity of social exclusion (components) and the intensity of poverty. Numbers

 Table 8 presents the same kind of analysis but comparing the number of items socially excluded against the poverty measures excluded. Again there is only one case excluded on more than five items who is not poor. 

Table 8: Relationship between the intensity of social exclusion (items) and the intensity of poverty. Numbers.

 Finally Table 9 and 10 are a first attempt to answer the question who are the poor and are they the same as the socially excluded? They present the results of a set of logistic regressions of the odds of being poor and or socially excluded by each of the measures that have been used in this analysis. Table 9 presents a bivariate analysis. In general
" females are more likely to be poor and socially excluded – the exception is in respect of labour market exclusion where the difference is not (statistically) significant.
" Older people are less likely to be necessities poor and much more likely to excluded from social relations.
" Non white households are more likely to be poor and socially excluded by all measures except service exclusion.
" Childless couples are less likely to be necessities poor and couples with children are less likely to be socially excluded on some measures. This is true also of married people.
" Lone parents are more likely o to be poor by all measures, labour market excluded and excluded on the composite measures. This is true also for the separated and divorced.
" By far the most consistent picture is for those dependent on social assistance and/or living in social housing –they are much more likely than others to be poor and also more likely to be socially excluded.

The multivariate analysis in Table 10 shows the odds of being poor and/or socially excluded - other factors held constant. Again the most striking results are
" social housing and Income Support receipt – these households are much more likely to be poor and on socially excluded - with the single exception of service excluded for social assistance recipients.
" Females are more likely to be excluded from social relations and service excluded but not (significantly) more likely to be poor – other factors held constant.
" There is no variation with age.
" Non whites are again more likely to be poor and excluded from the labour market and social relations but not service excluded.
" Single parents are more likely to be poor but no more likely to be socially excluded (when social assistance receipt is controlled for).

 Table 10: Logistic regression of the odds of being poor/socially excluded: multivariate analysis

This paper has investigated the relationship between poverty and social exclusion, using data from a (shortly to be published) survey in Britain. We are still at an early stage of the analysis and this paper presents an initial exploration of the data.

Poverty was defined using three of the most common conventional measures – equivalent household income less than 60 per cent of the median; lacking three or more socially perceived necessities; below a subjective poverty line. Social exclusion was operationalised in three different ways – as labour market exclusion; as exclusion from services and as exclusion from social relations. The latter element included social activities, isolation, lack of support, civil disengagement and confinement. 

The extent of overlap between poverty and experiences of social exclusion will depend to some extent on the proportion of the population defined as poor or socially excluded. The greater the proportion, the greater the chances of overlap – thus a quarter of the sample were necessities poor and there tended to be a greater degree of overlap between them and (for example) labour market excluded than the income poor who only included 20 per cent of the sample. Nevertheless there was a stronger association between the necessities poor and all elements of social exclusion than the other poverty measures. Also the subjectively poor were more likely to be socially excluded than the income poor. However there was a strong association between all measures of poverty and most measures of social exclusion. The exceptions were isolation and social support - the poor were not more likely to be socially excluded in this way. One explanation is that older people are more likely to excluded from social relations but they are not significantly more likely to be income poor or subjectively poor and they are less likely to be socially perceived necessities poor. Further, and in contrast, there is evidence that those in the labour market may find it more difficult to maintain relationships with families, friends and caring others. In the PSE study it was found that joblessness does not necessarily increase social isolation and in some respects is associated with less social isolation. 

The other key findings are that the lone parent families, households in social housing and on Income Support are the most likely to be poor and also socially excluded. 

This suggests a line further analysis of the data might take. Our definitions of poverty incorporate between 19 and 26 per cent of all households. Perhaps if we were to take a more stringent definition of poverty we might find a closer association between poverty and social exclusion. In contrast it would be worth identifying the characteristics of those who are poor but not very socially excluded and those that are socially excluded but not very poor- in order to see how they manage to avoid the association. 

There are four (at least) other lines to follow: 

1. It should be acknowledged that the thresholds used in this paper are fairly arbitrarily chosen – an indication of what kind of analysis is possible with this data set - and we will need to evaluate and experiment with different thresholds .
2. The question of whether it is appropriate to combine the dimensions of social exclusion, or even of exclusion from social relations, is an extremely complex one, since they are qualitatively different and do not necessarily co-vary, and there is a great deal more work to be done in exploring their interaction before we can draw firm conclusions.
3. In particular, very different results may emerge from treating labour market exclusion in different ways , either by including all households of any age where no-one is labour-market active, or by focusing on individuals rather than households. If social exclusion is a euphemism for poverty, then joblessness of households is important because it makes people poor (although this is an artefactual outcome of the benefit system, not a natural consequence); this is true of all age groups. If what is at issue is the social interaction which supposedly results from participation in paid work, then what matters is joblessness of individuals – again of all ages. In terms of exclusion from social relations, we need to try to disentangle the effects of poverty and joblessness. The issue is complicated by the fact that there is a political subtext to treating joblessness rather than poverty as a prime cause here. Simply focussing on those in jobless working age households merely buys into the Blairite/Third Way ideology. If we are arguing that non participation in the labour market doesn’t necessarily produce social exclusion, then it makes little sense to just leave out all those whose non participation is seen as politically legitimate. Retired people are labour market excluded, although the consequence of this may (or may not) be different than those of working age. It is possible that poverty and poor health will be much better predictors of social exclusion, even for the retired, than labour-market participation, but this needs testing.

REFERENCES: 

Bradbury, B., Jenkins, S. and Micklewright, J. (2001) The Dynamics of Child Poverty in Industrial Countries, Cambridge University Press. 

Bradshaw, J. and Sainsbury, R. (2000a) Researching Poverty, Aldershot Ashgate 

Bradshaw, J. and Sainsbury, R. (2000b) Experiencing Poverty, Aldershot Ashgate 

Burchardt, T., Le Grand, J. and Piachaud, D. (1999) Social Exclusion in Britain 1991-1995, Social Policy and Administration, 33, 3, 227-244. 

Dirven, H-J. et al (2000) Income Poverty and Social Exclusion in the European Member States, Paper to the Working Group on Statistics on Income, Social Exclusion and Poverty, Eurostat. 

Gordon, D. et al (2000) Poverty and Social Exclusion in Britain, York: Joseph Rowntree Foundation. 

Gordon D. and Pantazis C. (1997), Breadline Britain in the 1990s, Ashgate: Aldershot 

Mack J. and Lansley S. (1985), Poor Britain, London: Allen and Unwin 

Levitas R. (1999), The Inclusive Society, London: Macmillan 

Room G. (Ed.) (1995), Beyond the Threshold: The Measurement and Analysis of Social Exclusion, Bristol: Policy Press 

Townsend P. (1979), Poverty in the United Kingdom, London: Allen Lane and Penguin Books

 Appendix 2: Public and private service exclusion

 Appendix 3: Factors preventing participation in common social activities

Note: Multiple responses allowed