Poverty and Social Exclusion...1. Seeking policy Added Value – An overview2 Poverty as a concept...

22
Antigone Lyberaki and Platon Tinios 1 Poverty and Social Exclusion ALMOST FINAL DRAFT 1 Antigone Lyberaki is Professor of Economics at Panteion University, Athens. Platon Tinios teaches at Piraeus University. He was an advisor to the Prime Minister of Greece 1996-2004 and a member of the EU Social Protection Committee 2000-4. Many thanks are due to George Papadoudis and Thomas Georgiadis for valuable help in computation.

Transcript of Poverty and Social Exclusion...1. Seeking policy Added Value – An overview2 Poverty as a concept...

Page 1: Poverty and Social Exclusion...1. Seeking policy Added Value – An overview2 Poverty as a concept has historically played a key role in the shaping of social policy. ‘Poverty’

Antigone Lyberaki and Platon Tinios1

Poverty and Social Exclusion

ALMOST FINAL DRAFT

1 Antigone Lyberaki is Professor of Economics at Panteion University, Athens. Platon Tinios teaches atPiraeus University. He was an advisor to the Prime Minister of Greece 1996-2004 and a member of theEU Social Protection Committee 2000-4. Many thanks are due to George Papadoudis and ThomasGeorgiadis for valuable help in computation.

Page 2: Poverty and Social Exclusion...1. Seeking policy Added Value – An overview2 Poverty as a concept has historically played a key role in the shaping of social policy. ‘Poverty’

February 2006

2

Page 3: Poverty and Social Exclusion...1. Seeking policy Added Value – An overview2 Poverty as a concept has historically played a key role in the shaping of social policy. ‘Poverty’

1. Seeking policy Added Value – An overview2

Poverty as a concept has historically played a key role in the shaping of social policy.‘Poverty’ or ‘social inclusion’ have well-understood, if vague, connotations; theappeal and attraction of the terms are due to their significance to social engagement.We are interested in poverty because we are concerned about the poor. One of thecentral questions in our concerns about ageing is that the nature of poverty maychange with the advent of ageing.

In bringing a new data source to bear on a well-worn subject on which a policy debateis ongoing, we should be careful to slot in the new insights in such a way that theytake into account the current state of policy debate, illuminate concerns and open newavenues.

Poverty is most commonly defined in advanced countries as the situation in which anindividual is unable to participate fully in what is socially accepted as the life of thecommunity. If everything that matters could be obtained in markets, then the idea of‘participating fully’ could be approximated as possessing a minimum level of income.However, crucial goods are provided in ways that bypass the market: Health, socialand care services, are distributed through non-market criteria. Equally, access tosocial networks, the environment and other non-tangibles impinge on social welfare.Most crucially, the instances where the subjective ‘feel’ of poverty transcends thesimple measure of income may be more frequent as age rises.

Nevertheless, there remains a sense in which financial considerations may be accordedprimacy. A ‘pragmatic approach’ has evolved whereby financial poverty isconventionally linked to the shape of the lower end of the income distribution: thus apoverty line is drawn with reference to the income of the median individual (theperson at the middle of the income distribution). Lines of 50% median and 60%median are in common use, while the latter has received most attention at the EUlevel, as the central ‘risk of poverty line’3.

The ‘risk of poverty’ plays a crucial role in EU discussions. The three waves ofNational Action Plans for Inclusion serve as the cornerstones of the open method ofcoordination in the social field. Quantification through the use of indicators is the keyinnovation of this ‘soft law’ approach to social policy, designed to add a Europeanimpetus to an area under exclusive member state jurisdiction. The 2001 EuropeanCouncil at Laeken approved a list of indicators covering dimensions of the ‘risk ofpoverty’ (Eurostat (2004)). It noted that much work still needed to be done, both toimprove statistical infrastructure and to capture the multidimensional nature ofpoverty.

SHARE has the potential to open new roads in this policy discussion. This short paperillustrates that, even at an early stage, SHARE data can illuminate discussions on theextent and characteristics of poverty and begin to ‘flesh out’ our picture of povertyand the poor.

2 This paper updates and in some cases brings forward the work in Lyberaki and Tinios 2005, whichwas based on version 0 of the SHARE data.3 The word ‘risk’ concedes that the demarcation line is ‘fuzzy’ – that there is a range of incomes wherethe risk of poverty is present even if some in this range might not consider themselves as ‘poor’.

3

Page 4: Poverty and Social Exclusion...1. Seeking policy Added Value – An overview2 Poverty as a concept has historically played a key role in the shaping of social policy. ‘Poverty’

2. Is old age poverty more serious than we think?The first step in the analysis is to see how SHARE compares with the ‘stylised facts’of poverty. To do this we must note that the analysis of low income in a survey likeSHARE comes at the end of the data processing phase. Accepting that some resultsmay need to be reviewed later, we need to identify important findings.

The starting point for the analysis is the weighted data for the entire sample of theover 50s in SHARE. The income used was the version 1 estimates of ‘net income’presented in the chapter on Household income. Given the centrality of the income ofthe median individual, infelicities in modelling taxation in the middle of the incomedistribution may well bias the poverty line upwards – more so in the Northerncountries. Total household income, including imputed rent from owner occupation,was attributed in equal part to all household members. Poverty lines are computed onthe basis of the median individual of the SHARE sample of over 50s.

1. Poverty in the SHARE sample - % of the sample

17,5%

21,4% 21,8% 22,4% 23,5%24,4% 24,7% 24,9%

26,3% 27,7%

6%

10%

14%

18%

22%

26%

30%

SE GR DK NL FR IT DE AT ES CH

Poverty line at 50% Poverty line at 60%

Those familiar with the picture of social exclusion from ECHP (e.g. Eurostat 2004)may be surprised by the picture emerging in Figure 1. ‘Poverty’ rates are relativelyhigh – in all but one country more than one in five people are below the poverty line.There is also a smaller dispersion of poverty rates; the country rankings are certainlyunfamiliar. However, the concentration of people between the two alternative povertylines in the Nordic countries is reproduced; the choice of the 50% rather than the 60%line thus leads to a change in country rankings.

To compare our findings with other information, we must allow for the fact that theSHARE sample consists exclusively of individuals over 50, rather than the entirepopulation. The ability to participate fully in the life of society should refer to theentire society: One would need to assume a total breakdown in generationalcommunication to presume that older citizens never compare themselves with thoseunder 50. Thus, the question arises as to how to define such a line, given that detailedinformation only exists for households containing individuals over 50 years of age.

4

Page 5: Poverty and Social Exclusion...1. Seeking policy Added Value – An overview2 Poverty as a concept has historically played a key role in the shaping of social policy. ‘Poverty’

We know (e.g. Joint Pension Report – CEC 2002) that the aged are subject to differentpoverty risks across the EU. In some countries (Greece and Denmark) old age isassociated with greater poverty risk; in those countries the within-generationinequality is relatively smaller to the between-generation inequality. In other countriesthe opposite holds. A possible correction is to attempt to approximate the populationmedian (and hence poverty line) by using outside information to adjust the povertylines upwards in countries where old age incomes are known to be lower and viceversa. In this way age-corrected poverty lines could be computed. To identify thoselines as closer to the ‘true’ population lines would need the additional assumption thatthe extraneous sources employed measure incomes both of the under 50s and the over50s with the same degree of accuracy.

Given the European reference of the data, an obvious candidate for an extraneoussource – for the EU member states - would be the Survey of Income and LivingConditions for 2004, the same reference year as SHARE. However, at the time ofwriting access to the 2004 SILC data was not possible, and hence, as a second-best,the 2001 ECHP data was used instead. The SHARE (over 50) poverty line is correctedaccording to how different the median income of the over 50s is compared to theoverall population in the last available ECHP wave, that of 2001 (Eurostat 2004)

2. Poverty in the SHARE sample - 2001 ECHP age poverty line correction applied

27,7%25,4%24,2%24,1%23,9%23,7%23,4%

21,8%

19,1%

14,7%

6%

10%

14%

18%

22%

26%

30%

SE NL FR DE IT AT ES GR DK CH

Poverty line at 50% Poverty line at 60%

The correction factor appears as column 2 in Table 1. It takes its maximum value inGreece (where SHARE poverty lines are increased by 5.4% and its smallest inSweden and Holland were they are reduced by 7.7%) – the range being 11.1%. Thiscorrection, as would have been expected, introduces some variability and leads tomore familiar patterns emerging. A North-South gradient is complemented with thepresence of Denmark. Sweden is at the one extreme and southern countries at theother. (Swiss data could not be age-corrected as ECHP was not conducted inSwitzeland).

To examine how far SHARE findings approximate those of the ECHP and SILC oneshould examine similar populations. This can be done for subsets of the populationfor whom extraneous estimates can be obtained, viz for those over 65, as well as the

5

Page 6: Poverty and Social Exclusion...1. Seeking policy Added Value – An overview2 Poverty as a concept has historically played a key role in the shaping of social policy. ‘Poverty’

group 50-64. Of these the former may be thought to have more or less severed linkswith the world of work, whereas the latter is still active. Table 1 compares SHAREwith the latest ECHP data, that for 20014.

In this more restricted comparison, poverty rates of the elderly, even after correction,remain larger than in other sources. For the over 65s, in Greece, France, Spain andSweden differences are small. However, in other countries (especially of the North),SHARE appears to lead to poverty more than twice that of the ECHP, though this isoften due to very low values for the ECHP rather than high estimates in SHARE. Forthe group of working age (50-64) differences are larger. Though more investigation ofthis feature is necessary, it is probable that divergences may be related to thecomplexity of income composition. In countries where pensions and labour incomeaccount for larger portions of income, differences are smaller. A further factorpossibly explaining differences is the net income concept employed; if median netincomes are overestimated this would lead to higher computed poverty lines. Finally,serious consideration should be given to the fact that, though strict comparabilityacross countries of the questions posed on income was maintained in SHARE, thesame did not hold for the ECHP; ECHP participants phrased the question used inradically different ways. For instance some countries estimated net income, othersgross income, while the extent of imputations used also varied by country.

Table 1: Poverty rates by large age group: Comparing SHARE with the ECHP

(60% median income poverty line, % of poor)

MEMOCorrection factor to

povertyline (1)

Over 65 poverty 50-64 poverty

Hypotheticalcorrected

poverty lineECHP

Data 2001 income

Hypotheticalcorrected

poverty line

ECHPData – 2001

incomeSE 7,7% 18,4 16 11,2 5DK -5,1% 39,7 24 14,9 5DE 3,6% 22,0 12 24,7 10NL 7,7% 18,3 4 19,7 7FR 2,8% 20,8 19 22,8 13CH *0% *26,0 … *29,2 …AT 1,8% 20,1 24 27,2 9IT 2,2% 22,0 17 25,5 16ES 5,4% 20,1 22 28,4 17GR -5,4% 29,0 33 19,4 21Notes: 1. See text for definition – ECHP2001data. ECHP data from CEC (2002)2. No ECHP data available for SwitzerlandIt is worth dwelling a little in the case of Denmark. As is evident from Diagrams 1and 2, the choice of poverty line makes a large difference: there is a concentration ofincomes between the 50% median and the 60% median line, caused by features of the

4 Eurostat publishes 2003 data, drawing from a variety of sources – SILC for those countries thatconducted pilot surveys in that year and national sources for the rest. The heterogeneity in definitionsand practices is such as to countervail any gain from the more recent time frame. It was thus, and whileSILC data are still expected, to stick to ECHP 2001 data.

6

Page 7: Poverty and Social Exclusion...1. Seeking policy Added Value – An overview2 Poverty as a concept has historically played a key role in the shaping of social policy. ‘Poverty’

Danish social protection system (e.g. the value of the age pension). The finding oflarge poverty rates, especially among the over 65s is corroborated by the ECHP; in the1998 wave the 65+ poverty rate reached 31 per cent. As CEC 2002 makes clear,narrow income-based definitions of poverty ignore features of social protectionsystems such as the provision of benefits in kind; these may increase well-being butare not counted in the income concept used in the ECHP. Given that social protectionafter retirement often involves the substitution of in kind benefits for cash income,such is a potentially serious shortcoming in the study of social problems associatedwith ageing. A full analysis using SHARE data may allow investigation of thishypothesis and the exploration of income concepts that make a fuller allowance forsocial protection systems.

The overall message of SHARE is that spread of poverty may be a more serious issuethan sometimes thought. In some cases the SHARE results appear fully compatiblewith the ECHP, in others considerable divergences arise, whose sources warrantinvestigation before final conclusions are drawn. Understanding the cause ofdifference could leave to symmetric adjustments: it may imply the need for greaterrefinements in SHARE, but it may also conclude that SHARE was more successful inrecording incomes that are particularly relevant for the older population. SHARE maytherefore be able to capture income inequality in old age to a greater extent thanECHP. Whatever the case, apart from noting this intriguing divergence, it is too earlyto settle on firm conclusions on the mechanisms driving it.

Turning to the question of how poverty changes between the group still of workingage (50-65) and those over 65, the countries fall in three groups, regardless of whetheran age correction is applied or not: In one (GR, DK and SE) the group over 65 appearto be at substantially greater poverty risk than the younger group. In the larger, secondgroup, increased age is associated with negligible differences. Finally, in twocountries (AT and ES) poverty in SHARE (though not in the ECHP) is significantlylower in the older group.

Poverty differences by age groups:Differences 50-64 and 65

Significantly larger poverty

Small or no age effect

Significantly smaller poverty

Significance at 95% or greater

7

Page 8: Poverty and Social Exclusion...1. Seeking policy Added Value – An overview2 Poverty as a concept has historically played a key role in the shaping of social policy. ‘Poverty’

This picture hides considerable differences between sexes, (Chart 3). Thedeterioration in poverty risk performance between the younger and older age groups isin all cases except Spain, sharper for women than for men. In some countries, the twosexes are proceeding in opposing directions (DE, FR), while the differences by sex arefar more marked (and approach statistical significance) than those for the entirepopulation.

Increases in poverty risks 50-64 and 65+ by sex

2,3

15,3

-8,9

1,6

-4,8-3,4

-13,6

-6,2 -6,5

5,6

12,5

25,7

2

-1,6

3,4

-3,7 -3,2

-0,2

-9,2

8,2

-15

-10

-5

0

5

10

15

20

25

SE DK DE NL FR CH AT IT ES GR

Difference in %

Men Women

A further question that warrants investigation is the depth of poverty – how far belowthe poverty line is the majority of poor individuals. The measure of the depth ofpoverty favoured in the EU attempts to avoid the problems caused by extreme values(in this cases unrealistically low incomes possibly due to underdeclaration), bycomputing the relative distance between the median income of the poor and thepoverty line as a percentage of the poverty line (the ‘relative median at risk of povertygap). Figure 4 computes this for SHARE data for the entirel population of 50+employing the SHARE-specific poverty line (i.e. data corresponding to Figure 1).Denmark is at one extreme and Switzerland at the other, while the range of variationis considerably greater than the risk factors themselves.

21,8% 22,5%

28,5% 29,6%

35,4% 35,5% 36,3%39,2%

42,7%

46,7%

36,2%

0%

5%

10%

15%

20%

25%

30%

35%

40%

45%

50%

DK SE NL GR DE FR ES AT IT CH Total

Relative median at-risk-of-poverty gap

8

Page 9: Poverty and Social Exclusion...1. Seeking policy Added Value – An overview2 Poverty as a concept has historically played a key role in the shaping of social policy. ‘Poverty’

Disaggregating by age and sex, (Table 2) we see that reaching pension age (65+), inmost cases dramatically reduces the poverty gaps (with the notable exception ofSweden), even in high poverty countries such as Greece. In contrast, sex differencesare generally less marked; where differences exist, males seem to be subject togreater depth of poverty.

Table 2Relative median at-risk-of-poverty gap, by age and gender

Persons50+ (%)

Persons50-64 (%)

Persons65+ (%)

Male(%)

Female(%)

AT 39,2% 48% 30% 48% 38%DE 35,4% 42% 30% 35% 35%SE 22,5% 22% 23% 21% 23%NL 28,5% 33% 20% 25% 29%ES 36,3% 45% 29% 32% 38%IT 42,7% 52% 31% 46% 40%FR 35,5% 39% 30% 35% 36%DK 21,8% 37% 16% 24% 21%GR 29,6% 43% 22% 31% 30%CH 46,7% 63% 25% 58% 39%Total 36,2% 43% 29% 36% 36%

The differences in spread and depth of poverty are due to a complex interplay ofindividual retirement, work and savings decisions, household composition effects andoperation of social protection systems (e.g. the operation of survivors pensions).Equally importantly, cohort effects are conflated with age effects in such a way thatwe cannot express any final opinion on the two key policy questions underlying theanalysis: Do social protection system protect the old, men and women, adequately?And will the coming generation of old be better prepared for old age than the currentgeneration?

3. Living in your own home, poverty and other wealth effects Living in your own home protects you from one of the most socially damaging effectsof income insecurity – the fear of being homeless. As a result an owner-occupier is insignificantly better position than a renter with the same income. Income measures thatignore this are likely to portray the condition of owner-occupiers as appreciablyworse. In cases (common in the South of Europe) where significant numbers of thepoor live in their own house, not allowing for this imputed rent is likely tosignificantly bias computed poverty rates (see, for instance CEC 2004). This effect islikely to be more marked in groups such as the old who are less likely both to betenants, and to have mortgages outstanding.

A full analysis of the effect of housing on poverty should also try to model the effectof social housing and other kinds of subsidised accommodation5. However, at this

5 To the extent that housing benefits are included in the definition of income, that is taken into account.What cannot be taken into account, of course, is the conditional effect of the right to (means tested)social accommodation.

9

Page 10: Poverty and Social Exclusion...1. Seeking policy Added Value – An overview2 Poverty as a concept has historically played a key role in the shaping of social policy. ‘Poverty’

stage of the analysis it is as well to start with the effect of adding imputed rent to thedata of the distribution of Figure 1. As a first approximation imputed rent wasestimated using the short-cut of assuming it is close to 4% of house values as reportedby the respondent. It also reports the absolute effects on the data (a negative figuresignals a fall in poverty rates by the specified amount).

Table 3: SHARE poverty rates – the effect of imputed housing income (totalsample and by large age group)

Poverty rates (%) if housingincome taken into account

Difference attributed to housingincome (percentage points)

TotalSHARE

population

Persons50-65

Persons65+

TotalSHARE

population

Persons50-65

Persons65+

SE 17,5 13,5 21,7 0,2 -0,4 0,9DK 21,8 12,8 34,1 1,6 0,2 3,6DE 24,7 25,7 23,7 -0,9 -1,0 -0,8NL 22,4 22,4 22,4 0,6 -0,4 1,8FR 23,5 23,4 23,5 0,4 -0,2 1,0CH 27,7 29,2 26,0 -2,0 -0,9 -3,3AT 24,9 28,2 21,1 -1,3 -3,0 0,6IT 24,4 25,7 23,1 -3,9 -4,5 -3,4ES 26,3 30,1 22,9 -1,4 -2,6 -0,3GR 21,4 17,8 25,1 -2,7 -3,5 -1,8Notes: See text for definition of housing incomeAdding imputed rent increases incomes of the population. In the actual case, wheremore than 50% of the population are owner-occupiers, it also increases the populationmedian and hence the poverty line6. Thus, it is possible for poverty to increase, evenif all the poor are owner-occupiers, so long as the middle classes choose to live inproportionately better houses – so that poverty lines increases by more than the poor’sincomes. Indeed, this seems to be happening as the proportionate increase in mediansis in many countries sizeable

The data of Table 3, indeed show just such an intricate pattern. There is a definiteNorth-South gradient with the effect of owner occupation significantly reducingpoverty rates in the South, by almost four points in Italy and three in Greece. In somecountries of the North the effect is in the direction of increasing poverty (eg DK, NL).The relative country rankings alter considerably. It appears that the effect isdifferentiated by age: Adding the effect of owner occupation reduces poverty in the50-65 age group in all countries . In the older age group, this observation is reversed:CH, GR, IT show large falls in poverty, other countries increasing – by a maximum of3.6 percentage points (DK). Whatever the precise interpretation, the investigation ofhousing effects on income is a fruitful area of investigation, precisely because itimpinges on poverty differences between age groups.

6 Allowing for imputed rents can be expected to produce greater reduction in poverty for the wholepopulation: if the young (who are excluded in SHARE) are more likely to be renters, the medianincome will increase by less. This would lead to a greater poverty dampening effect among the 50+population.

10

Page 11: Poverty and Social Exclusion...1. Seeking policy Added Value – An overview2 Poverty as a concept has historically played a key role in the shaping of social policy. ‘Poverty’

Table 4 examines the prevalence of owner-occupation (including rent-freeaccommodation) in the SHARE sample. As might be expected, the extent of owneroccupation in the age groups covered by SHARE is very extensive, 75% for the wholesample. What sharply differentiates Northern and Southern Europe is the extent ofhouse ownership among the income-poor population. Though in all cases the poorlive in their own houses to a lesser extent than the richer group, the difference can beas little as 3 percentage points (Greece, over 65s) in the South, and as large as 30points (CH, over 65s).

Table 4 Owner occupation by age group and poverty status

Persons 50+ Persons 50+ Persons 50-64 Persons 65+

Poor (%) Poor (%) Non poor(%) Poor (%) Non poor

(%)AS 69,8 57,2 57,3 75,6 57,0 72,1DE 61,4 43,7 41,2 66,6 46,4 67,3SE 55,3 38,5 47,9 64,0 32,3 52,6NL 59,8 40,0 51,8 75,4 24,3 52,4ES 93,7 87,3 85,1 95,6 90,0 95,8IT 84,5 74,4 73,5 87,3 75,3 87,9FR 80,3 58,4 56,5 86,2 60,5 87,1DK 67,5 33,0 44,7 80,0 27,1 71,0GR 91,0 86,5 80,2 90,2 91,1 94,4CH 58,4 40,9 43,5 62,8 37,4 66,7Total 75,2 60,4 59,7 79,9 61,1 79,6

Note: owner occupation plus rent-free accommodation.

The prevalence of owner occupation means that asset ownership (implying primarilythe family home, though not exclusively7) plays a significant part in the overallfinancial picture of the 50+ population. Table 5 looks at the number of timesestimated net wealth exceeds annual income, distinguishing between age classes andpoverty status. Though that ratio is affected by the low values of annual incomes,wealth in the South amounts to well over 10 years’ average income. Interestingly, thisfigure is far larger in the South and among the poor, indicating possibly housingmarket imperfections that prevent housing consumption to adapt to changedcircumstances.

Table 5 Equivalent net worth / equivalent net income ratio, by age group and poverty status

Country Persons50+

Persons50+ Persons 50-64 Persons 65+

Poor Poor Non poor Poor Non poor AS 4,7 9,7 13,0 4,6 6,0 4,2DE 5,9 9,6 12,0 4,9 6,8 7,0SE 5,3 8,8 10,0 5,4 7,8 4,9NL 5,8 10,1 15,2 6,3 4,6 4,4ES 13,3 24,2 34,3 11,0 14,7 14,4IT 13,7 22,8 35,0 12,4 11,2 14,0FR 9,0 8,9 11,9 9,3 5,8 8,6DK 6,8 4,8 8,0 6,7 3,5 7,4

7 Housing wealth among house owners ranges from 50% to 90% of the value of total wealth, with themodal value being around 70%.

11

Page 12: Poverty and Social Exclusion...1. Seeking policy Added Value – An overview2 Poverty as a concept has historically played a key role in the shaping of social policy. ‘Poverty’

GR 8,1 13,4 20,6 7,1 9,3 8,9CH 7,5 10,7 14,8 7,0 7,3 7,7Total 4,7 9,7 13,0 4,6 6,0 4,2

4. Family solidarity as social protectionHousehold composition and cohabitation with children is probably the oldest socialprotection mechanism for old age. Figure 3 looks at the cases where the respondentlives in the same household, or, quite importantly for social relations, in the samebuilding as offspring. The analysis is confined to the over-65s, as cohabitation with achild in younger age group may probably mean dependency of the child on the parentand not vice versa.

Figure 5 confirms the use of cohabitation as a social protection mechanism in theSouth of Europe: the propensity to live with one’s children is associated with povertystatus. What is less usually appreciated is that this mechanism extends to living in thesame building in separate households; the latter is very important in Germany andAustria. In three countries more than half of the over 65 population live in the samebuilding, while in five it is more than 40%. The full story of poverty cannot be told ifcohabitation is ignored.

Household composition and poverty for people over 65: Proximity to nearest living child (%)

0

10

20

30

40

50

60

p np p np p np p np p np p np p np p np p np p np

SE DK DE NL FR CH AT IT ES GR

np: non poorp: poor Same household Same building

Family solidarity has many more dimensions than cohabitation. Table 6 examines aseries of indicators, focussing on differences between the poor and the non-poor.(These aspects are covered in Attias Donfut et al, 2005). The density of relationsincreases the probability to receive (possibly later on), rather than a simple mechanismproviding aid in cases of low income. Nevertheless, in at least one case (Greece),

12

Page 13: Poverty and Social Exclusion...1. Seeking policy Added Value – An overview2 Poverty as a concept has historically played a key role in the shaping of social policy. ‘Poverty’

transfers as poverty alleviation are visible: 20% of the poor aged receive financialtransfers, while poverty status affects the probability of receiving appreciably.

As regards financial help over 250 euros, the poor elderly are more likely to receive onaverage, this being more pronounced in the 50-65 age group, mainly in Spain, Greece,the Netherlands and France. The reverse pattern characterises Austria, Sweden,Denmark and Switzerland, while no apparent differences are recorded in Italy andGermany. For the older age group (over 65 years of age), the poor are more likely toreceive financial help exceeding 250 euros in Greece (17% versus 10% in non-poor),Germany and Austria, while the reverse is true for Switzerland, Denmark, Spain, Italyand France. As regards financial gifts and inheritances exceeding 5.000 euro, the poorare less likely to get them (13% versus 23% on average), the gap being wider in the50-64 age group. The differences between poor and non-poor are smaller in Austria,Spain, Italy and Greece. This would seem to confirm an interpretation that certainly inGreece and Austria (and possibly elsewhere) informal networks are a significant(though necessarily often ignored) factor in poverty alleviation.

The poor in all countries are significantly less likely to offer non-financial help toother households, though (with the notable exception of Italy and Spain) they are alsomore likely to receive non-financial help (personal care, household help, help withpaperwork). If looking after grandchildren is singled out as a particularly importantinstance of inter-generational transfer, we see that it takes greater values in the agegroups where grandchildren are likely to be youngest (50-64). Nevertheless, in all buttwo cases (GR and CH), the poor are significantly less likely to be helping aftergrandchildren (in particular in Italy, Spain and Denmark).

13

Page 14: Poverty and Social Exclusion...1. Seeking policy Added Value – An overview2 Poverty as a concept has historically played a key role in the shaping of social policy. ‘Poverty’

Table 6 : Aspects of Poverty Experience: Support and contact nexus-Differences between poor and non-poor (A positive sign means that a higher value for poor individuals)

DifferencesPoor –

Non PoorSE DK DE NL FR CH AT IT ES GR

Received (non-financial)help from others (personal,

housework, paperwork)

Total (all 50+) 27,8% 30,2% 29,8% 25,1% 18,4% 20,5% 28,3% 18,4% 15,8% 23,8%65+ 12,7% 7,9% 2,3% -3,2% 2,5% -0,8% 3,9% -7,5% -3,3% 10,3%

50-64 4,3% 9,1% -1,5% 2,4% 3,5% 3,1% -6,2% -2,8% -1,4% 4,3%

Offered (non-financial) helpto others (personal,

housework, paperwork)

Total 40,6% 46,5% 32,1% 40,7% 31,4% 35,0% 24,7% 22,8% 13,4% 19,6%65+ -9,0% -2,2% -5,7% -8,0% -0,2% 5,3% -1,45 -11,6% 1,2% -3,8%

50-64 -9,2% -12,8% -7,3% -5,9% -3,8% -5,1% -1,5% -14,7% -5,7% -3,7%

Look after grandchildren(regularly or occasionally over

past 12 months)

Total 51,9% 60,2% 46,0% 60,6% 55,0% 43,1% 45,5% 47,4% 44,0% 48,3%Diff. 65+ -18,0% -21,7% -13,5% -9,8% -12,2% 6,1% 0,8% -11,2% -7,6% -6,2%

Poor 50-64value 60,5% 66,6% 53,5% 69,0% 60,8% 63,6% 51,4% 49,7% 43,2% 79,3%

Diff. 50-64 -10,7% -15,2% -17,1% -16,2% -13,4% -0,4% -12,9% -14,0% -15,3% 5,4%

Received any gift over 250€ Absolute value 5,6% 6.1% 6.5% 2.7% 3.0% 4.8% 7.3% 3.9% 3.9% 10.3%

50-64 -2.0% -4.5% -0.5% 2.0% 2.2% -2.3% -5.8% -0.1% 5.7% 2.5%65+ -0.6% -2.4% 1.0% 0.3% -0.7% -3.4% 1.1% -1.1% -0.7% 7.0%

Received any gift over 5000€from a family member,

Absolute value 37,0% 36,4% 26,4% 24,1% 24,2% 44,0% 15,2% 14,7% 17,5% 23,3%Total -18,0% -18,2% -13,7% -12,6% -17,0% -19,5% -7,4% -12,0% -6,7% -3,7%50-64 -15,1% -8,6% -14,8% -12,8% -19,2% -22,3% -9,4% -13,8% -10,7% -2,9%65+ -17,3% -22,0% -13,2% -12,3% -14,7% -16,5% -6,0% -10,6% -3,7% -3,9%

Page 15: Poverty and Social Exclusion...1. Seeking policy Added Value – An overview2 Poverty as a concept has historically played a key role in the shaping of social policy. ‘Poverty’

Table 7 Poverty and health: Differences between poor and non-poor: EURO-D caseness, extracted from depression scale EURO-D, Self-perceived health (European version) – cut off point “less than good health”, IADL: Number of

limitations with instrumental activities of daily living – cut off point “one or more limitations”, (A positive sign means that a higher value for poor individuals)

Differences Poor – Non

PoorSE DK DE NL FR CH AT IT ES GR Total

SPH-EU (Less than good)

Absolute value 37,4 30,9 46,8 32,6 37,4 20,1 38,8 49,5 49,6 38,7 43,6Total 14,3 16,4 13,2 9,5 18,9 4,6 2,9 8,4 2,2 12,4 11,450-64 16,1 14,6 13,3 9,3 23,6 9,6 5,6 6,4 0,8 9,1 11,865+ 9,6 11,0 14,6 9,7 13,8 -1,0 4,1 12,1 9,1 8,6 11,7

IADL (one or more limitations)

Absolute value 17,4 17,3 15,2 16,4 17,7 8,5 18,3 14,9 24,1 19,1 17,1Total 18,7 16,3 6,5 4,6 6,3 2,1 3,1 5,3 0,7 7,9 5,650-64 5,9 6,3 5,2 3,6 9,7 6,5 0,6 -2,5 -1,8 2,0 3,265+ 23,5 14,7 8,9 5,9 2,4 -3,0 11,0 14,7 8,8 7,5 8,9

EURO-D caseness

Absolute value 20,8 18,2 20,8 21,2 33,4 18,5 19,8 33,1 36,1 25,0 27,8Total 9,6 9,4 7,1 8,2 8,4 3,3 4,9 6,9 3,3 1,4 6,650-64 7,4 5,2 6,8 12,9 12,8 3,2 5,9 7,1 4,9 0,0 7,765+ 10,2 11,9 8,1 2,0 3,7 3,4 5,2 7,0 5,3 0,2 5,8

Grip strengthAbsolute value 33,2 34,2 34,6 34,1 31,3 33,1 35,3 29,0 27,5 31,4 31,6

Diff. 50-64 -2,0 -1,5 -2,3 -0,9 -0,8 -3,1 -0,8 -3,2 -3,7 -2,4 -2,5Diff. 65+ -5,8 -4,4 -3,5 0,0 -2,5 0,8 -3,1 -3,0 -3,5 -2,9 -3,0

Out-of-pocket payment foroutpatient care (last 12 months, in

euro)

Absolute value 173,5 174,0 65,4 81,5 43,3 223,0 39,0 318,2 58,8 155,3 125,665+ -68,9 -78,9 -26,4 -37,6 -33,2 3,0 0,9 -269,8 -1,0 -9,2 -76,1

50-64 -77,3 4,9 -10,4 -21,4 -2,3 15,4 -6,0 -154,5 -28,3 71,6 -42,2

Forego any type of carebecause of costs (%)

Absolute value 3,3 1,7 5,7 2,3 5,9 3,9 3,0 5,2 3,1 5,7 4,965+ -0,8 1,6 6,3 0,4 4,7 -2,3 1,9 1,0 -2,3 0,3 2,6

50-64 8,3 3,5 1,2 0,4 4,5 0,3 0,8 4,7 -1,0 2,9 2,2

15

Page 16: Poverty and Social Exclusion...1. Seeking policy Added Value – An overview2 Poverty as a concept has historically played a key role in the shaping of social policy. ‘Poverty’

Table 8: Aspects of Poverty Experience: Differences in attitudes between poor and non-poor(A positive sign means that a higher value for poor individuals)

DifferencesPoor –

Non PoorSE DK DE NL FR CH AT IT ES GR

Hopes for the future(%)

65+ -11,5 -8,0 -10,4 -5,6 -6,8 -2,4 -2,7 -0,5 -8,8 -9,250-64 0,3 1,5 -6,0 -4,2 -5,1 -1,9 -3,3 -6,6 -6,6 0,6

No mention of enjoymentfrom any activity (%)

65+ 9,6 5,1 4,6 -0,8 2,5 0,2 0,4 5,2 3,7 -0,950-64 1,8 -2,3 2,1 0,7 6,6 -1,5 2,2 5,0 2,5 -2,7

Grandparents´ duties:financial aspects

Total 48,8 19,4 45,2 31,9 52,8 40,2 39,9 75,6 67,5 66,Diff. Poor-non

poor 0,2 -0,3 0,2 0,1 -0,4 -5,7 7,9 0,4 5,4 6,2

Grandparents´ duties: in-kind help (looking after)

Total 61,1 45,3 80,6 37,7 75,3 60,6 60,8 84,8 77,2 89,1Diff. Poor-non

poor 0,7 -3,6 4,0 1,6 -0,2 -1,3 8,9 -0,5 5,5 1,6Who should assume mainfinancial responsibility forelderly in need: mainly thestate

Total (all 50+) 78,0 88,4 33,5 72,1 51,4 61,3 46,0 38,8 32,6 56,0Diff. Poor-non

poor 1,3 -0,8 1,4 -0,6 5,8 0,8 4,1 3,2 5,7 -12,3

Who should assumeresponsibility for providingcare in kind (h/h chores):mainly the family

Total (all 50+) 11,7 2,9 42,8 12,7 12,1 31,5 34,9 37,5 39,2 52,3

Diff. Poor-nonpoor -3,6 -1,0 1,7 11,7 3,5 4,2 -1,1 -2,4 10,1 4,6

Who should assumeresponsibility for providingcare in kind (nursing,bathing, dressing): mainlythe family

Total (all 50+) 7,7 1,8 26 3,9 7,9 14,4 17,6 29,7 33,9 63,7

Diff. Poor-nonpoor 1,3 0,4 7,2 2,5 -2,4 4,5 -2,4 4,6 7,1 2,7

Retirement perceived as aconcern or as both relief andconcern

Total (50+ all) 19,2 17,9 15,3 11,4 12,4 7,0 7,5 15,0 17,8 21,465+ 2,0 -0,8 -4,9 -5,4 -7,7 0,3 -2,9 -1,8 5,7 -13,6

50-64 10,4 -7,8 -1,2 6,2 -3,7 28,1 9,2 -1,9 7,8 19,4

Page 17: Poverty and Social Exclusion...1. Seeking policy Added Value – An overview2 Poverty as a concept has historically played a key role in the shaping of social policy. ‘Poverty’

5. The multiple dimensions of poverty: Health, Attitudes and EducationThe analysis so far has focused on financial poverty. In doing so, it has ignored thearea where SHARE holds the greatest promise: By combining economic, social andhealth information on the same individual, they allow us to ‘flesh out’ our picture ofpoverty among the older population and enrich it with information on different aspectsof life and perception. Questions such as the link between health and poverty status,or the extent to which informal social or family networks compensate for financialpoverty become approachable. Utilisation of both health care systems and systems ofsocial protection and their effectiveness in alleviating the effects of deprivation can beapproached in more direct ways. In the last section we give a foretaste of the kind ofresults that SHARE can produce in illustrating the multidimensional nature ofpoverty. We focus on differences between the poor and the non-poor (employingSHARE-specific poverty lines) and look at results from three broad areas:

A. The interplay between health issues and poverty, Table 7 gives a taste of somesystematic relationships. It looks at differences between poor and non-poor across anumber dimensions: a measure of subjective evaluation of health (Self PerceivedHealth SPH less than good), two more ‘objective’ measures of health status (severeactivity limitations and grip strength) and two measures of access to the health system(out of pocket expenses and foregoing treatment due to cost). In perceptions of healththere is a very strong link between poverty status, especially in Northern Europe (11percentage points difference in SE and DK), which appears to affect the younger agegroup particularly. Differences are less marked in Southern Europe, as the non-poorare equally likely to complain (with the possible exception of GR). This picture isbroadly confirmed for objective health measures in the North (though the locus ofdifference is moved towards the older group), while in IT, ES and CH there appears anegative correlation between poverty and activity limitation. Poverty status,surprisingly, makes a difference in an ‘objective’ test such as grip strength, a factormore noticeable in the older group. Finally, poverty appears clearly linked to thepropensity for depression as measured by the EURO-D scale, an effect more markedin the North and least in CH, ES and Greece.

Out of pocket payments for outpatient care vary considerably between countries,reflecting systemic differences. In some systems, presumably where access is grantedselectively to lower income groups this may translate to very different impactaccording to the poverty category (e.g IT), whereas in others the impact appears to beindependent of poverty status. Forgoing care due to cost is nowhere widespread(below 6% in all cases), but seems to be focused more in the younger poor.

B. Attitudes. Table 8 attempts to chart differences in perceptions and attitudesbetween the poor and the rest of society on a number of attitudes concerningexpectations and views on social protection and individual services. Expressinghopes for the future and mentioning enjoyment in daily life, predictably mirror the

Page 18: Poverty and Social Exclusion...1. Seeking policy Added Value – An overview2 Poverty as a concept has historically played a key role in the shaping of social policy. ‘Poverty’

depression indicators derived in health. The poverty effect is far more marked in theolder age group (e.g. SE, DE, ES, GR).

The table next reports answers to questions focussing on the respective roles andresponsibilities of the family and the State in a number of relevant circumstances.Differences in those issues are marked between countries, with the South (andGermany and Austria) stating a preference for familial roles for personal services andaccording the State financial responsibilities, while the North (especially inScandinavia) looking more towards the State as a direct provider of services.. Aroundthis general picture, the poor are more likely to adopt a slightly more family attitude.

C. Table 9 shows the expected strong relationship between poverty status andeducational achievement: In all countries the poor have attended education for feweryears than the rest of the population. This relationship is stronger for the older groupand could offer an explanation for their poverty status. Absolute values of years ofeducation are higher and differences between poor and non-poor are smaller for theyounger age group. Given that the position in the income distribution of individualsbetween 50 and 64 is already entrenched could mean that a link between low incomeand education is becoming looser over time.

Table 9 Years of education, by age group and poverty statusCountry Persons 50-64 Persons 65+

Poor Non poor Poor Non poor AS 11,5 12,0 10,3 11,0DE 13,5 14,2 11,5 13,1SE 10,8 11,3 8,0 9,5NL 10,2 12,1 8,8 10,5ES 6,3 7,6 3,7 4,4IT 8,1 9,1 5,4 6,6FR 7,1 10,1 4,5 7,2DK 12,6 13,6 10,4 12,4GR 8,9 10,7 5,0 6,8CH 12,3 13,1 10,1 11,8All 9,7 11,1 7,3 8,8

6. Pensions and povertyIn the case of financial poverty, clearly, the most important social protectionintervention mechanism is that of pensions, whether public (first pillar) or private(second and third pillar). Table 10 examines the coverage of the pension system, byage and poverty status. Rather than using the self description ‘pensioner’, the broaderdefinition of anyone receiving positive amounts in pensions (i.e. including workingpensioners and housewives collecting survivors’ pensions) is used8. Coverage for theover 65 group in most countries is almost complete (around 90%), with the notable8 The difference can be considerable (of the order of 7-8 points) for over-65s in Southern Europe,ES78% as opposed to 56%. IT 85% vs 74%, GR 87% vs 78%.

18

Page 19: Poverty and Social Exclusion...1. Seeking policy Added Value – An overview2 Poverty as a concept has historically played a key role in the shaping of social policy. ‘Poverty’

exception of Spain. In that country (and possibly also in Italy) lack of access to thepension system must be a factor behind old age poverty (64% coverage of elderlypoor). Coverage of the younger group is closely linked to early retirement, with thosecountries where early retirement is commonplace showing a greater prevalence ofpension receipts. Interestingly, in all countries but Denmark, the poor exercise earlyretirement to a considerably smaller extent than the non-poor. The effect isconsiderable (over 10 points difference) everywhere except Germany and Holland.

Table 10 Coverage of the pension systemPersons receiving pensions, by age group and poverty status

Country Persons50+ (%)

Persons50-64 (%)

Persons65+ (%)

Persons 50-64 Persons 65+

Poor (%) Nonpoor (%) Poor (%) Non

poor (%)AT 64,4 43,0 88,8 27,0 49,3 75,4 92,4DE 56,2 23,2 89,6 21,1 23,9 77,7 93,2SE 65,5 35,6 96,7 48,2 33,7 93,6 97,6NL 57,5 28,1 95,8 26,0 28,7 94,2 96,3ES 52,1 22,7 78,5 16,7 25,4 64,4 82,7IT 60,8 36,6 84,9 16,7 43,5 74,5 88,0FR 56,8 25,6 89,5 19,2 27,6 80,7 92,2DK 56,2 28,0 94,7 38,1 26,5 93,8 95,2GR 56,8 27,1 86,9 17,1 29,3 82,3 88,4CH 54,1 19,0 96,0 16,4 20,1 89,2 98,3Total 57,2 27,6 87,5 20,0 30,2 77,6 90,6Note: if pension > 0. See text for details

Table 11 examines the poverty risk of people who collect pensions and reports thepercentage of pensioners who are at poverty risk, using the internal SHARE povertyline (i.e. data comparable to Figure 1). Comparing the data for the population, we seethat the pension system is associated with dampening poverty risk (by up to 6-7percentage points. Disaggregating by age (Figure 6), in the over 65 group the effect isconsistently in the direction of reducing poverty. The situation in the age group 50-65is probably caused by sample selection effects: In the majority of countries earlyretirement is associated with higher incomes (it is an income elastic activity); thosereceiving pensions are far less likely to be poor (eg IT, AT). However, in the twoNordic countries, the effect however is in the opposite direction, that of increasingpoverty, presumably because the poor are more likely to depart the labour forcesooner.

Table 11 Poverty risk of pension recipients, by gender and age group

Country Persons50+ (%)

Persons50-64 (%)

Persons65+ (%)

Persons 50-64 Persons 65+

Male (%) Female(%) Male (%) Female

(%)AT 17,9 17,8 17,9 14,0 20,8 10,0 23,6DE 21,1 23,4 20,5 22,8 23,9 11,3 27,1SE 20,3 18,2 21,0 12,9 22,3 13,8 26,6NL 21,7 20,7 22,0 22,8 18,6 22,7 21,5ES 19,4 22,1 18,7 21,0 23,5 16,7 20,8IT 17,7 11,7 20,2 8,2 15,4 13,8 25,4FR 20,3 17,5 21,2 13,6 22,2 16,6 24,6DK 29,1 17,4 33,8 19,3 16,1 27,5 38,4GR 20,8 11,3 23,8 14,0 9,2 21,4 25,8

Page 20: Poverty and Social Exclusion...1. Seeking policy Added Value – An overview2 Poverty as a concept has historically played a key role in the shaping of social policy. ‘Poverty’

CH 24,3 25,1 24,1 19,0 29,1 20,4 26,7Total 20,1 18,0 20,8 15,7 20,2 15,0 25,3

Poverty risk reduction effect of pensions

-5

-3

-1

1

3

5

7

9

11

13

15

AT DE SE NL ES IT FR DK GR CH Totaldiff

eren

ce in

pov

erty

risk

Persons 50-64 (%) Persons 65+ (%)

Finally, Table 12 examines the extent to which incomes differ by age group (orpossibly cohort). The first column compares the group 65+ with the last group withextensive involvement in the labour market (50-64). The range of results is from100% in Austria to 61% in Denmark. A possible more useful comparison is betweenover 65 pensioners as compared to those working in the age group 50-64. Austrianpensioners are considerably better off than the last age group of workers. The verylow values for Denmark (52%) and Greece (59%) are presumably the reason for thedeterioration of poverty risk in the older age groups which was commented in the firstsection. In both cases, it could be linked to the operation of demogrant-type first pillarpensions for large part of the population. (In Greece, this is due to agriculturalpensions as well as the prevalence of minimum pensions for the private sector).

Table 12 Median incomes ratios by selected age groups

Country 65+ / 50-64 75+ / 50-74 75+ / 50-6465+

(pensioners)/50-64

(working)AS 100,1% 97,0% 97,5% 106,9% DE 79,7% 82,4% 76,4% 69,4%SE 77,1% 66,9% 64,8% 71,1%NL 89,6% 88,4% 85,8% 78,3%ES 95,9% 93,2% 92,4% 82,9%IT 93,5% 87,1% 86,5% 91,2%FR 92,7% 92,6% 90,5% 85,6%DK 60,8% 58,7% 53,7% 51,6%GR 69,2% 73,9% 63,5% 59,3%CH 87,8% 88,9% 83,7% 73,0%Total 86,4% 85,9% 82,3% 79,7%

7. Taking the analysis forward

20

Page 21: Poverty and Social Exclusion...1. Seeking policy Added Value – An overview2 Poverty as a concept has historically played a key role in the shaping of social policy. ‘Poverty’

The analysis of the previous section gave an impression of the kind of insights thatSHARE can bring:

• An intriguing possibility emerges that financial poverty may be more serious,than we thought. The investigation of non-financial dimensions thus acquiresgreater significance.

• Though in the majority of countries there is no significant difference infinancial poverty in the 65+ group, in three cases there was apparentdeterioration (and in two improvement). The gender dimension appears to bedominant in this transformation, though the precise effects differ from countryto country. This may be linked to the substitution of services and goods inkind for monetary income after retirement and the operation of survivors’pensions.

• Imputed rent has a considerable influence in limiting poverty, especially in theSouth, signalling the importance of incorporating housing wealth in theanalysis of ageing.

• Living close to one’s children, in the same household or the same building,remains a very important mechanism of social solidarity with an importantpoverty alleviation role, not only in the South but also in Germany.

Realising the full potential of SHARE implies further work in three directions:

First, data calibration. Once points of contact and divergence between SHARE and thesources of the “stylised facts” are clarified, ongoing policy processes such as socialinclusion and pension strategy can absorb new insights without compromisingexisting understanding. Calibrating SHARE with SILC will obviously havemultiplicative effects on analysins.

Second, SHARE can be used to derive comparable indicators in some of the ‘greyareas’ of the open method of coordination process, fleshing out themultidimensionality of poverty. Access to health services, the problem of take-up ofsocial benefits and the nature of informal social networks are all issues that havereceived much comment in policy discussion but are very imperfectly measured.

The third area presupposes that SHARE follows the example of European sourcessuch as SILC in acquiring a time dimension. Poverty persistence and dynamicmechanisms generating and perpetuating poverty, but also the relationship betweenhealth and poverty are the kind of issues that will benefit most from panel data.

The introduction, posed the central question: will poverty in the future be similar tothat today? SHARE gives us a snapshot and can allow us to sketch possible alternativescenarios. Deciding between those scenarios must await the addition of a timedimension.

References

Attias Donfut, C Ogg, J and F-C Wolff, 2005 ‘Family Support’ and ‘Financial Transfers’ in A.Börsch-Supan, A. Brugiavini, H. Jurges, J.Mackenbach, J.Siegrist and G. Weber, Health Ageing andRetirement in Europe: First Results from the Survey of Health, Ageing and Retirement in Europe,Mannheim, pp 171-185.

Page 22: Poverty and Social Exclusion...1. Seeking policy Added Value – An overview2 Poverty as a concept has historically played a key role in the shaping of social policy. ‘Poverty’

Commission of the European Communities (CEC), 2002, Joint Report by the Commission and Councilon Social Inclusion, Directorate of Employment and Social Affairs.

Commission of the European Communities (CEC), 2004, Joint Report by the Commission and Councilon Social Inclusion, Directorate of Employment and Social Affairs.

Eurostat, 2004, Statistics in Focus: Poverty and Social Exclusion in the EU, Luxembourg.

Lyberaki, A and P.Tinios, 2005, ‘Poverty and Social Exclusion: a new approach to an old issue’ inBörsch-Supan et al, pp302-309

22