Pensions, poverty and wellbeing in later life: Comparative research from South Africa and Brazil

Pensions, poverty and wellbeing in later life: Comparative research from South Africa and Brazil

Journal of Aging Studies 26 (2012) 243–252 Contents lists available at SciVerse ScienceDirect Journal of Aging Studies journal homepage: www.elsevie...

226KB Sizes 0 Downloads 30 Views

Journal of Aging Studies 26 (2012) 243–252

Contents lists available at SciVerse ScienceDirect

Journal of Aging Studies journal homepage: www.elsevier.com/locate/jaging

Pensions, poverty and wellbeing in later life: Comparative research from South Africa and Brazil Peter Lloyd-Sherlock a,⁎, Armando Barrientos b, Valerie Moller c, João Saboia d a b c d

School of International Development, University of East Anglia, UK Brooks World Poverty Institute, University of Manchester, UK Rhodes University, South Africa Instituto de Economia, Universidade Federal de Rio de Janeiro, Brazil

a r t i c l e Keywords: Comparative design Pensions Poverty South Africa Brazil

i n f o

a b s t r a c t This paper draws on two linked studies of social policy and wellbeing in later life. The studies make comparisons between distinct groups of older people at the national and sub-national levels, as well as over time. The paper reflects on some of the main challenges for operationalising this complex design, as well as for interpreting findings and identifies lessons for other studies. The first study, conducted in 2002, included a questionnaire survey of around 2000 households containing at least one older person in South Africa and Brazil, supplemented by a set of in-depth qualitative interviews. Intriguingly, these countries had remarkably similar pension programmes, providing the majority of older people a reliable payment of roughly US$3 a day. This offered the prospect of exploring the effects of similar interventions in distinct developmental and cultural settings. In both countries, we found that these pensions had a substantial impact on the prevalence and depth of poverty in the study households, and were usually shared between older people and other family members. The second survey took place in 2008/9 and involved revisiting the households included in the 2002 survey, along with a separate set of in-depth interviews. This provided an opportunity for dynamic analysis of economic and wellbeing effects, against a backdrop of increased divergence in the wider national settings. Among other things, this revealed high and increasing levels of life satisfaction across all the study groups, although the extent to which this was directly related to generous pension provision cannot be ascertained. © 2012 Elsevier Inc. All rights reserved.

Introduction This paper reviews and seeks to interpret selected findings from a comparative study of older people, pensions and wellbeing in South Africa and Brazil. The research design allows for comparisons between two countries, between three distinctive sub-groups of older people in each country, and of change over time. This ambitious approach has the potential to yield important comparative insights, but represents a

⁎ Corresponding author. Tel.: + 44 1603 592327; fax: + 44 1603 45 1999. E-mail addresses: [email protected] (P. Lloyd-Sherlock), [email protected] (A. Barrientos), [email protected] (V. Moller), [email protected] (J. Saboia). 0890-4065/$ – see front matter © 2012 Elsevier Inc. All rights reserved. doi:10.1016/j.jaging.2012.02.003

substantial challenge. Difficulties include the logistical hurdles of designing compatible research tools and minimising attrition over the study period. They also include the challenge of interpreting complex and sometimes unexpected findings. The next section briefly discusses the research design and introduces the study settings. This is followed by a review of findings from the 2002 study, with a particular focus on relationships between pensions and poverty. The paper goes on to examine findings from the follow-on 2008/9 survey, setting out the dynamic policy environment and seeking to interpret how various indicators of subjective wellbeing changed over time. Taken together, the studies show that Brazil and South Africa have both performed relatively well, especially by the standards of middle-income countries, in meeting the

244

P. Lloyd-Sherlock et al. / Journal of Aging Studies 26 (2012) 243–252

Table 1 Selected features of the study countries, circas 2002 and 2008. Sources: UNAIDS (2008) Report on the global HIV/AIDS epidemic 2008, UNAIDS, Geneva http://www.unaids.org/en/KnowledgeCentre/HIVData/GlobalReport/2008/; UNAIDS (2002) Report on the global HIV/AIDS epidemic 2002, UNAIDS, Geneva; http://data.unaids.org/pub/Report/2002/brglobal_aids_report_en_pdf_red_en.pdf; UN Population Division (2009) World population prospects: the 2008 revision. http://esa.un.org/unpp/p2k0data.asp; IPEA www.ipeadata.gov.br.; UNData http://data.un. org/Data.aspx?d=SNAAMA&f=grID%3A101%3BcurrID%3AUSD%3BpcFlag%3A1, Bhorat and Kanbur (2006); Souza (2011). Brazil

Population aged 60 and over (% of total) Per capita GDP US$ current prices Gini index Open unemployment rate (% of all adults 15–64) Aids prevalence (% of adults aged 15–49) Poverty gap ratio at US$1 a day at ppp (%) a

S Africa

Circa 2002

Circa 2008

Circa 2002

Circa 2009

8.1% (2000) US$2825 (2002) 0.594 (2001) 9.0 0.7 (end 2001) 2.2

10.2 (2010)a US$8311 (2008) 0.539 (2009) 8.0 (2007) 0.6 (2007) 1.3 (2007)

5.9% (2000) US$2400 (2002) 0.63 30.0 (2002) 20.1 (end 2001) 8.2 (2000)

7.3% (2010)a US$5566 (2008) 0.68 23.0 (2007) 18.1 (2007) No data

Median variant projection.

economic needs of older people through extensive pension provision. At the same time, older people report high rates of satisfaction with most aspects of their lives. There are, however, important differences across the two countries and the underlying causes for high satisfaction scores are not always evident. Study design and site selection The data discussed in this paper were collected as part of two surveys conducted among selected groups of older people in Brazil and South Africa in 2002 and 2008/9. The main purpose of the original 2002 study was to assess the effects of non-contributory pensions (sometimes referred to as “social pensions”) on the economic situation of relatively deprived older people and their households in developing countries. At the time of the first survey, few low or middle income countries had extensive social pension schemes providing meaningful benefits to substantial numbers of older people. Brazil and South Africa were the two main exceptions, with social pension schemes covering 5.3 and 1.9 million older people respectively. Intriguingly, both countries' programmes offered benefits worth around US$3 a day, offering the prospect of exploring the effects of similar interventions in different national settings. Table 1 provides an overview of each country's wider developmental standing in 2002 and 2008/9. It indicates that Brazil and South Africa shared some common features beyond their extensive social pension schemes. These included similar levels of population ageing and per capita wealth in 2002, as well as levels of income inequality that were very high by international standards. 1 Table 1 also reveals some important areas of divergence, including notably higher rates of unemployment and HIV/AIDS prevalence in South Africa. This combination of similar and distinctive features would both create opportunities and pose challenges for comparative analysis. The pension systems of both countries had evolved over a long period, dating back to the start of the 20th century, and both had also seen a period of particularly rapid expansion from the early 1990s. In Brazil, shortly after the restoration of democracy in 1985, a new constitution was ratified in 1 Brazil's 2002 GDP figure is substantially lower than those for 2001 and 2003, reflecting a sharp, short-term dip in the exchange value of the real. This shows the danger of comparing two time points rather than presenting complete trend data.

1988, setting out an ambitious agenda for upgrading the country's existing welfare programmes. Central to this agenda was a programme of expanding the scope of pension provision for poorer groups (Barreto de Oliveira, Beltrão, Pinheiro, Peyneau, & Mendonça, 2005). Through a complex system which included “contributory” pensions, means-tested benefits for the poor and a dedicated scheme for retired rural workers, Brazil was able to extend pension coverage to the great majority of people aged 65 and over. 2 In 2002 78% of people aged between 60 and 64 received a pension, increasing to 94% for those aged 75 to 79 (Rofman & Lucchetti, 2006). In South Africa there was a similar process of political transition and pension upgrading. Non-contributory pensions were well-established and widely provided across all racial groups during the Apartheid era (Van der Berg, 1997). However, pension values varied markedly across racial groups, until the end of white rule in 1993 when they were equalised. These pensions are notionally means-tested and as of 2002 were available for men aged 65 and over and women aged 60 and over. In 2000 it was calculated that around 80% of eligible older people received a social pension, a very similar take-up rate to that reported for Brazil (Woolard, Harttgen, & Klasen, 2010). In the light of the marked income inequalities in each country, the study was particularly concerned with the impact of the pension system on the most disadvantaged households. This also reflected the priorities of the UK Department for International Development (DFID), the organisation that funded the first round of research. As a bilateral aid agency, DFID had a strong interest in the impact of different public policy interventions on household and individual poverty rates, and their potential transferability to other developing country settings. Lacking the resources to construct nationallyrepresentative samples, the study selected specific disadvantaged groups and locations in order to explore key divergences of experience at the sub-national level. Given Brazil's deeply entrenched geographical inequalities, this country survey focussed on poorer households in two distinctive locations: the municipality of Ilhéus in the north-eastern state of Bahia and a selection of relatively poor districts in the city of Rio de Janeiro. Ilhéus is a region with particularly high levels of socio2 In fact, high rates of evasion and low contribution rates meant that worker contributions to these so-called “contributory” pensions did not cover their full cost and they were heavily subsidised by the state.

P. Lloyd-Sherlock et al. / Journal of Aging Studies 26 (2012) 243–252

economic deprivation due to the collapse of the local cacao industry in the 1990s (Alger & Caldas, 1994). It has not shared in Brazil's recent economic dynamism. There are sharp socioeconomic disparities across the districts that make up Rio de Janeiro: those selected for the survey had comparable rates of poverty to Ilhéus (Saboia, 2009). In South Africa, experiences of later life continue to be strongly framed by the racial and geographical legacies of apartheid (Lloyd-Sherlock, 2010). This country survey sought to capture these effects by focussing on three distinctive groups: Africans living in rural former homeland territories of the Eastern Cape; residents of former African townships in Cape Town and residents of former coloured townships in the same city. 3 As with the Brazil survey, this does not constitute a nationally representative sample, but does provide an indication of the diverse settings in which people experience in later life. In the case of South Africa, most other surveys of older people focus exclusively on Africans, which do not allow for comparison across racial groups. The initial round of data collection in 2002 included a questionnaire survey of approximately 1000 households in each country containing at least one “older person”. 4 A critical difference between the two studies was that the South African team opted to define older people as aged over 55, but the Brazilian team used 60 years as the cut-off point. These local variations in defining the survey parameters created some problems for comparative analysis as discussed below. The questionnaire surveys included separate individual questionnaires for all older household members aged 55 and over in both countries. The composition of these country samples is summarised in Table 2. The questionnaires paid particular attention to the impact of pensions on households, but included a wide range of questions about the livelihoods, health and wellbeing of individual members. The questionnaire survey was supplemented by a round of 20 in-depth interviews in each country. 5 In 2008/9 the same research teams returned to the 2002 survey households. It was possible to trace 64% of the original households in Brazil and 65% in South Africa. To maintain a constant sample size, the 2008/9 survey included a total of 645 substitute households.6 Comparative analysis of the original and substitute households across the questionnaire items reveals no significant discrepancies (Barrientos & Mase, 2011; Saboia, 2009). The 2008/9 questionnaire was supplemented by 51 semi-structured interviews in Brazil and 44 in South Africa,

3 This paper uses the artificial racial categories that have been officially applied in South Africa. These (as all racial categories) are essentially a social construction, but their importance in the country's development has been such that they are essential tools for analysis. The term “White” refers to people with European settler origins; “African” refers to indigenous, darkskinned groups; “Coloureds” refers to people of mixed race. 4 Full details of these surveys' sampling designs and data collection tools are available at Saboia (2009) and Moller and Ferreira (2003). These sought to maximise cross-national comparability. 5 Details of the 2002 qualitative research design and interview summaries are available at http://www.sed.manchester.ac.uk/research/ageingandwellbeing/ ncpps/Papers/APPENDIX%20G.pdf. 6 The “new” households were selected in the same census sectors as the “lost” ones, using the same procedure as in 2002. In a few cases, it was necessary to use a different census sector with similar characteristics. For more details see Saboia (2009) and Moller (2010).

245

Table 2 Structure of quantitative samples in Brazil and South Africa, 2002. Brazil

South Africa

Rio de Janeiro Ilhéus Total Rural Africans Urban Africans Urban Coloureds Total

505 501 1006 379 176 268 1111

with households purposively selected from the quantitative samples.7 Table 3 summarises general socio-economic features of each of the sample groups in 2002. It reveals important differences both between countries and at the sub-national level, and demonstrates the dangers of generalising about the situation of older people and their households. For both Brazil and South Africa, overall levels of access to pension benefits was generally high and in keeping with those reported in other studies (Rofman & Lucchetti, 2006; Woolard et al., 2010). 8 This shows the effectiveness of their pension systems in reaching relatively deprived groups. However, there were notable differences in terms of the types of pensions received by households. In the case of South Africa, these were disproportionately social pensions. In Brazil contributory pensions were the predominant benefit, even for rural households whose older members might be expected to have had less lifetime formal sector employment. The prominence of contributory benefits among deprived Brazilian households was an unexpected finding and indicates that developing countries can take more than one path to achieving broad pension coverage of the poor. To some extent, this challenged the simplicity of our research design, since the nature of the benefits being compared across the two countries was fundamentally different. However, since over 90% of the contributory pensions recorded in the Brazil survey were paid at the same rate as the social pension (US $3 a day), from the beneficiaries' point of view the differences between these schemes were purely bureaucratic. Table 3 reveals a range of other important differences between the study populations. In the case of Brazil, households from Rio de Janeiro reported average per capita incomes to be 38% higher than those in rural Ilhéus, and they were more than three times as likely to have access to piped water. In South Africa, the differences were notably starker. Average per capita income of urban Coloured households was more than three times that of rural Africans, of whom less than one in ten had access to piped water. It should not be assumed that these patterns of inequality are directly representative of national trends: had a different rural setting been selected for Brazil, it is conceivable that levels of relative deprivation could have been closer to those of rural South Africans. Nevertheless, these patterns are broadly consistent with the Gini scores reported in Table 1 and indicate that welfare programmes in 7 Full details about qualitative methods are available in Lloyd-Sherlock (2008). 8 Levels of access for urban Africans stand out as rather lower than other groups studied. In part, this was because the composition of this group was on average younger, and included a smaller share of people over the minimum pension age. By 2008 the ageing of this population and a fall in the pension age saw levels of access to pensions reaching 84% for this group.

246

P. Lloyd-Sherlock et al. / Journal of Aging Studies 26 (2012) 243–252

Table 3 Some basic features of each sub-sample (2002).

% of households with at least one pension % of households with social pensions % of households with contributory pensions Per capita income % with piped water Average household size % with children Household head with completed primary education (%)

Rural Africans

Urban Africans

Urban Coloureds

Rio de Janeiro

Urban Ilhéus

Rural Ilhéus

92.3 91.7 0.9 216 rand 7.2 5.48 80.7 40.5

71.8 69.0 2.8 336 rand 75.3 5.09 78.2 68.4

83.6 71.6 16.4 702 rand 96.1 4.49 53.8 63.5

88.5 24.8 68.5 272 reais 94.1 2.83 31.9 21.6

85.6 21.9 66.8 219 reais 82.9 3.52 46.1 12.9

86.1 31.4 58.4 170 reais 29.7 3.76 50.0 9.8

Findings from the 2002 survey: pensions and poverty

and their households. This responded to the interest of development agencies, including the UK's Department for International Development, in “social pensions” as a potential instrument of poverty reduction for low and middle income countries (DFID, 2005; Kakwani, Son, & Hinz, 2006; UNDP, 2006). To evaluate these effects, it was first necessary to assess the extent to which pension income was being shared across households. Pension sharing would dilute the income gain for individual older people, but would spread the potential economic benefits across a larger population, including young children. The 2002 survey found extensive co-residence in both countries, but particularly in South Africa where households with older people were typically larger (Table 3). Older people lived alone in only 6.8% of households in South Africa, and 22.3% in Brazil, and resided with children in 64.2% of households in the South African sample and 33.4% in Brazil. Table 4 provides data on the extent to which pension income was shared across these households. It shows that the vast majority of pensioners shared all, or most of their benefits with their households. This was an important policy and advocacy consideration, meaning that the potential economic benefits of the pension were felt by a substantially larger section of the population. In the case of South Africa, for example, as many as 20% of the population were living with a pensioner in 2002 and might therefore benefit from household pooling. 9 Among the Brazilian subgroups, rates of pension pooling were broadly consistent, but this was not the case in South Africa, where urban Coloureds were less likely to share their benefits (24% reported they kept the pension entirely for themselves). Lower rates of pension pooling among coloureds reflected this group's smaller household size and higher economic status (thus reducing the need to share benefits). It may also have reflected varying family attitudes between racial groups (Amoateng, 1997). By contrast, the study did not reveal significant discrepancies in rates of pension pooling between older men and women. This is somewhat at odds with the findings of other studies examining gendered patterns of intra-household resource allocation (Young, 1992). The in-depth interviews revealed a small number of cases which contradicted older men's pooling claims. Nevertheless, the overall weight of evidence (including data collected in the follow-on survey) indicates that pooling was the rule rather than the exception. High rates of pooling meant that pension benefits should be understood as a contribution to household income, rather

The main objective of the 2002 survey was to assess the effects of pensions on the economic status of older people

9 This does not include any sharing of pension income with relatives and others beyond the household.

South Africa faced a particularly large challenge to reduce established inequities. Comparing the data in Table 3 across the two study countries reveals a complex picture. Overall levels of access to basic services, such as piped water, were broadly similar across the two countries in 2002, with both displaying large rural–urban inequalities. Access to basic education among household heads (almost always older people) was substantially higher in South Africa, even for rural Africans. This reflected the relatively high level of development of South Africa's education system six or seven decades ago, despite the existence of the apartheid system. In Brazil, average household sizes were smaller than in South Africa and older people were less likely to live with children. Table 3 presents per capita income data in local currency, which permits direct comparison between national sub-groups, but not across the two countries. Obtaining internationally consistent per capita income data is not straightforward. A simple approach would be to factor in international exchange rates. For example, in 2002 the exchange rate between the two currencies was roughly three South African rands to a Brazilian real. As such, the per capita income of the Rio de Janeiro subgroup was equivalent to 816 rand, and the average level of income for the Brazilians was substantially higher than that of the South African sample. However, this crude comparison does not take into account differences in the cost of living which may vary substantially across the two countries, as well as between rural and urban districts. A more sophisticated comparison may reveal a smaller, albeit still substantial, income gap than suggested above. At first sight, the existence of this per capita income gap would seem surprising, given the similarities between each country's pension system in terms of benefit values and coverage. However, since average household sizes in South Africa were substantially larger, pension income was effectively shared by larger numbers of people. More significantly, the Brazilian households contained a much larger proportion of people aged over 65 (37%) than the South African ones did (12%), and as a result were more likely to be in receipt of more than one pension. The higher share of people of pensionable age in Brazil was partly due to the different sample design criteria deployed in each country (see above), as well as a higher prevalence of two and three-generation households in the South African sample.

P. Lloyd-Sherlock et al. / Journal of Aging Studies 26 (2012) 243–252 Table 4 Pension sharing, 2002.

247

Table 6 Self-reported financial situation of households 2002 (% of column).

How much of your pension and your own money can you keep for yourself?

Brazil

South Africa

None A little Some A reasonable amount All

79.4 16.6 2.7 0.7 0.6

65.2 15.9 7.7 2.5 8.7

than to older individuals. Among the poorer households in the sample, pensions accounted for the lion's share of household income. Among the poorest 20% of households (in terms of per capita income), pension benefits represented 100% of total income in Brazil, and 50% in South Africa. Table 5 presents various indicators for the impact of pensions on household poverty across the study groups. Constructing internationally compatible income poverty lines represents a complex challenge. In part, this reflects the difficulties in comparable per capita income, as discussed above. It also reflects a lack of agreement over definitions and thresholds of income poverty in each country. For the purposes of this study, the value of the minimum pension benefit was selected as a proxy poverty threshold. There were several advantages in this approach. In Brazil, the basic pension is equivalent by law to the minimum monthly wage and is widely accepted as representing a minimum living wage. The South African basic pension does not have the same legal status, but is generally regarded as a socially accepted minimum income (Bhorat, Leibbrandt, Maziya, van der Berg, & Woolard, 2001). The similar value of the basic pension across the two countries further enhances the comparative value of this approach. The poverty headcount score shown in Table 5 calculates the proportion of households which would be poor were they not receiving the pension, assuming no off-setting effects. Predictably, these show that household poverty would be more widespread in the absence of pensions, although the effect is rather small, particularly for South Africa. At first sight, this is a surprising result and would appear to contradict claims that pensions can be an effective tool of poverty reduction. The second indicator, the poverty gap, assesses the intensity of poverty by measuring the average distance poor households fell below the poverty line. This shows a larger impact than the simple poverty headcount, particularly in South Africa. Put simply, pensions had an important effect on the intensity of household poverty, but were not always able to bring households out of poverty entirely. One key reason for this in the case of South Africa was the lack of

Table 5 Poverty headcount and gap measures with and without non-contributory pension income (using adult equivalent household income per capita) 2002. Brazil

South Africa

With Without With Without pension pension pension pension Poverty headcount 58.5 Poverty gap as% of poverty line 22.3

63.9 30.0

43.8 20.5

45.7 33.8

Brazil

South Africa

Contributory No NonNo Nonpension contributory pension contributory pension pension pension Very bad Bad Average Good Very good

18.3 32.8 37.4 9.9 1.5

9.0 26.2 49.5 14.3 1.0

7.1 18.8 54.4 18.3 1.4

23.1 32.9 36.6 7.1 0

10.5 56.2 25.7 6.8 0.7

additional income opportunities for younger household members due to unemployment. Per capita income is an important indicator of a household's economic situation, but may exclude other important effects such as the ownership of assets. To take a more complete, albeit subjective approach, the survey included items on household's more general assessment of their financial position. Table 6 examines the effect of receiving a contributory or non-contributory pension on these self-assessments. These are shown to be substantial: in both countries non-pensioner households were more than twice as likely to report that their situation was very bad as pensioner households were. This shows that non-pensioner households had not been able to find alternative sources of income such as paid employment. Relatively few households, particularly in South Africa reported that their financial situation was good/very good regardless of their pension status. This supports the finding reported in Table 5 that pensions reduced the intensity of poverty, but did not entirely eliminate it. A further piece of analysis assesses the extent to which receipt of a regular monthly pension benefit contributed to the financial stability of households. Table 7 provides data on perceived changes to household finances over the preceding three years and shows that pensioner households in Brazil fared much better than non-pensioner ones. In the case of South Africa, the difference is much more marginal. There are numerous possible explanations for the smaller effect here, including larger average household sizes and higher rates of HIV/AIDS (which often has a large impact on household finances through associated illness and mortality). Taken together, the findings of the 2002 survey supported the claim that pensions could have a significant effect on the economic status of poorer households in countries like Brazil and South Africa. The fact that their effect was greater on the poverty gap than on the overall incidence of poverty could be

Table 7 Self-reported change in financial situation from three years before 2002 (% of column). Brazil

South Africa

No Nonpension contributory pension Worse 54.2 Same 37.4 Better 8.4

29.5 58.1 12.4

Contributory pension

No Nonpension contributory pension

35.6 52.5 11.9

69.0 22.4 8.6

61.1 32.5 6.4

248

P. Lloyd-Sherlock et al. / Journal of Aging Studies 26 (2012) 243–252

taken as evidence that benefits reached the very poorest households, not just those whose average incomes were marginally below the poverty line. This finding was widely deployed by DFID, NGOs and other international development agencies as an argument for extending social pensions across other high and middle-income countries (GTZ, 2005; Help Age International, 2003). 10

Findings from the 2008 survey: ageing and wellbeing in a dynamic policy context The project team obtained funding to conduct a follow-on survey of the same households in Brazil and South Africa. In this case, the funder was the UK Joint Research Councils and the grant was awarded as part of a wider research programme on the “New Dynamics of Ageing”. The interests of this academic research programme were wider and more grounded in gerontological science than had been the case with DFID. They included dynamic analytical frameworks to assess changing patterns of wellbeing among older people, including subjective as well as objective components. This called for a broader approach to assessing wellbeing than had been taken in the 2002 survey. At the same time, as part of an increased UK funder focus on non-academic impact, the project was encouraged to build on aspects of the first study that had been most directly influential on policy debates. This allowed for a substantial element of continuity across the two studies. Beyond the logistical challenges of tracing the households and older people interviewed in 2002, it was necessary for the new survey to take account of various changes to each country's wider policy contexts. Politically, both countries were governed by democratic regimes with strong commitments to upgrading welfare programmes and improving the position of more disadvantaged groups. In the case of Brazil, the administration of Luis Inacio da Silva (“Lula”) succeeded in combining rapid economic growth, improved social outcomes and unprecedented political popularity (Bourne, 2008). In South Africa, there was growing disenchantment with the ANC's capacity to deliver on past promises of social reform (Johnson, 2010). Both countries saw significant increases in per capita wealth, although these were notably larger in Brazil than in South Africa (Table 1). In the case of South Africa, growth had led to a fall in official unemployment rates, although the level remained roughly treble that of Brazil. With specific reference to pension policy, both countries saw reductions to the minimum age of eligibility for a social pension. 11 Despite this, the 2008 survey found little overall change in what had already been high rates of access to pension benefits (other than a rise from 72 to 85% of urban African households). Changes in the value of basic pensions were more significant than changes in their reach. For Brazil, these were increased in line with the minimum monthly wage, which rose from roughly US$68 to US$227 a month 10 Whilst the researchers took pains to emphasise that these findings were not nationally representative, this qualification was sometimes lost in the presentation of findings by other organisations. 11 In South Africa, there had been a fall in the minimum pension age for men from 65 to 63 during the study period. In Brazil, the age for the means-tested social pension was cut from 67 to 65.

between 2002 and 2008. 12 Since the great majority of benefits were paid at this basic rate (76% in 2008), this represented a substantial increase in pensioners' personal incomes. When these gains are offset against inflation, which totalled 42% over the six year period, the increase of the real value of the basic benefit was 46%. In South Africa, there was an increase in the basic pension value over the study period; from 620 rand to 960 rand. When offset against inflation, this represented a real increase of 18%. 13 Thus, despite increases to the value of basic pensions in both countries, the value of benefits had diverged, with South Africa falling behind Brazil. Between 2002 and 2008 both countries saw the extension of other cash benefit programmes, including child grants and disability pensions. The South African survey households were larger than the Brazilian ones and more likely to contain children (Table 3). This is reflected in the proportion receiving at least one child grant, which rose from 10.5 to 33.0%. In the case of Brazil, the increase was more modest (4.5 to 6.3%). The higher number of child grants in the South African population may have partly compensated for the relative increase of pension values compared to Brazil. However, the value of these child grants was approximately a quarter of the basic pension, limiting their impact on household finances. There were also changes to a number of other aspects of public policy of relevance to older people in each country. In Brazil, there was a substantial upgrading of the national health care system, which has been credited with improving provision for older people (Cohn, 2009). These were accompanied by policy specific innovations, including a programme to promote access to cheap generic drugs and a national scheme of subsidised credit for pensioners (Lloyd-Sherlock et al., under review). In comparison, there were fewer new developments in South Africa, where health policy remained understandably dominated by the challenges of HIV/AIDS and related conditions such as TB (Kahn et al., 2006). In South Africa a key priority was to improve health services in rural areas, since these had been systematically neglected under apartheid (Mooney & Mcintyre, 2009). Taken together, these shifts in each country's economic development, pension provision and public policy added a further layer of complexity to our comparative analysis. The following section assesses changing patterns of wellbeing for the study populations, starting with a general indicator of life satisfaction, before analysing specific wellbeing domains. This produces a number of useful comparative insights, but also throws up some important unresolved issues, demonstrating both the value and the challenges of this complex research design. Table 8 provides data for general levels of satisfaction with household circumstances, as reported by household heads (in almost all instances older people). The 2002 data 12 In local currency, the apparent increase was rather less: from 200 to 415 reais a month. However, the real appreciated significantly against the US dollar over this period. Historical exchange rates were calculated using IPEADATA from www.ipeadata.gov.br. 13 At the time of the March 2009 survey, the value of basic pensions in South Africa was 960 rand (about 96 US$). In the November 2002 survey the value of the basic South African pension was 620 rand (about US$62). Cumulative inflation (urban CPI) for 2003–8 (inclusive) was 31% (http:// www.statssa.gov.za/keyindicators/CPI/CPIHistory_rebased.pdf). Nominal increase of 54.8%. Real increase of (960/620)/1.31 = 18%.

P. Lloyd-Sherlock et al. / Journal of Aging Studies 26 (2012) 243–252 Table 8 Per cent of households satisfied/very satisfied with general household wellbeing. Country/year

Site

2002

2008/9

Brazil

Rio Ilhéus urban Ilhéus rural Total Rural African Urban African Urban Coloured Total

58.0 69.5 62.1 62.0 9.4 12.1 49.2 25.0

69.7 79.2 77.1 74.0 57.4 32.5 84.5 60.5

South Africa

shows that overall levels of satisfaction were considerably higher in Brazil (62%) than South Africa (25%). This suggests that South Africa's embracing pension system was unable to compensate for a wider set of effects on household wellbeing. The precise nature of these effects is difficult to establish, but is likely to include high unemployment and prevalent HIV/ AIDS. More subtle, but potentially important effects may also include the relatively high level of education among the South African study group (Table 3): other studies have shown that this tends to extend aspirations, thus increasing the threshold for a given level of satisfaction (Sabates & Hammond, 2009). The data for 2008/9 show that levels of satisfaction continued to be higher in Brazil, but that the gap between the two countries had narrowed markedly. The scale of the increase in South Africa over the study period is striking. This period saw increases in the pension value and improvements to basic services, yet similar gains were also made in Brazil. The extension of child grants across the survey households was more rapid in South Africa than in Brazil. However, these grants are worth around 20% of basic pension benefits and therefore any effect on the study households is likely to have been marginal. As such, the apparent scope of the life satisfaction rise in South Africa cannot be fully explained and requires further research. Comparisons across the sub-groups also reveal interesting variations in life satisfaction scores. For Brazil, there is relative consistency across the three sub-groups, whereas in South Africa urban Coloureds reported far higher rates of satisfaction than rural or urban Africans. This happiness discrepancy between racial groupings is consistent and found in national surveys of life satisfaction (Moller, Dickow, & Harris, 1999) and suggests that the social pension has not been able to overcome the heavy legacy of apartheid on older people's wellbeing. By 2008 urban Coloureds reported higher rates of satisfaction than any of the Brazilian groups, showing that simple cross-national comparisons may overlook important patterns of inequality. On a more positive note, there is a striking improvement in reported satisfaction among the poorest sub-group, rural Africans. These apparent increases in reported levels of life of satisfaction may to some extent have been exaggerated by the loss of around a third of households between the 2002 and 2008/9 surveys. It is plausible that those households and older people facing more difficult circumstances in 2002 experienced higher rates of household dissolution and older person mortality between the two survey rounds. If so, this would have led to an “attrition bias” effect, whereby better

249

off households were over-represented in 2008. The research team tested for this effect with reference to household income and spending, and found that it did not lead to a significant bias (Barrientos & Mase, 2011). Nevertheless, comparisons over time should still be interpreted with a measure of care. Overall levels of satisfaction with household wellbeing may be influenced by a very wide range of effects. Interpretation and analysis should also take into account satisfaction with particular wellbeing domains, including financial, health and social wellbeing. Table 9 shows that overall levels of satisfaction with household finances were substantially lower than the general satisfaction scores shown in Table 8. This discrepancy may seem surprising given the generosity of pension programmes, but is in keeping with the findings of most subjective wellbeing surveys (Moller & Saris, 2001). The general patterns of financial satisfaction shown in Table 9 closely parallel those in Table 8, although one notable exception is rural Africans who show a much smaller improvement over time, suggesting that non-financial domains of wellbeing may have contributed proportionately more to this group's improved satisfaction scores, despite this group's increased access to child benefits. Table 10 shows reported satisfaction with health status for older household members. The 2002 data broadly follow the general satisfaction trends shown in Table 8, with higher rates for Brazil, but the urban Coloureds performing best overall. The 2008 data are more challenging to interpret. The Brazilian subgroups show a consistent pattern and a marginal improvement. This suggests that large increases in the pension value and efforts to upgrade health services have produced relatively small effects. However, since the study population had aged six years between the surveys, a fall in health satisfaction might have been expected ceteris paribus. In South Africa, urban Coloureds show a very small decline in reported health satisfaction, whereas the other two groups show substantial improvements. This would appear to show that efforts to reduce inequalities in health service provision have made some headway, especially in rural areas. These apparent trends should be interpreted with a certain measure of caution, however, since self-reported health data should not be read off as a direct indication of clinical health status, particularly for older people (Cockerham, Sharp, & Wilcox, 1983). This can be seen in a separate finding that 24% of older people in the Brazilian survey reported that they never had access to necessary medication, such as antihypertensives, whereas the figure for South Africa was only Table 9 Per cent of households satisfied/very satisfied with household financial situation. Country/year

Site

2002

2008/9

Brazil

Rio Ilhéus urban Ilhéus rural Total Rural African Urban African Urban Coloured Total

15.8 20.1 18.9 17.7 4.0 6.2 11.8 7.5

32.5 33.1 30.1 32.1 7.5 16.0 30.0 17.4

South Africa

250

P. Lloyd-Sherlock et al. / Journal of Aging Studies 26 (2012) 243–252

Table 10 Per cent of older people rating personal health status as good/very good. Country/year

Site

2002

2008/9

Brazil

Rio Ilhéus urban Ilhéus rural Total Rural African Urban African Urban Coloured Total

33.9 33.7 32.6 33.2 8.0 14.9

34.2 36.1 37.2 36.5 23.4 38.5

33.8 20.4

32.6 30.2

South Africa

3%. At face value, this suggests that the South African health system was much more effective at servicing older people than was its Brazilian counterpart. Yet these data depend on older people's awareness of their own health status and their perceptions of what is essential. Data from a separate WHO survey reports that rates of hypertension among over 50 year olds in South Africa are the highest ever recorded for any country in the world and that only 8% of cases were aware of their condition (Lloyd-Sherlock, Beard, Minicuci, Chatterji & Ebrahim, submitted for publication). By contrast, Brazil has invested heavily in a national screening programme, which has enhanced awareness and increased demand for treatment (Farias, Pacheco de Souza, Laurenti, & Martins de Alencar, 2009). This example demonstrates the dangers of drawing quick and easy conclusions from international comparisons without a good knowledge of local contextual factors. Table 11 presents data for two other domains of wellbeing: satisfaction with social relations and satisfaction with living conditions. These findings do not follow the trends for the other wellbeing domains. Overall levels of satisfaction are strikingly high for all subgroups and both countries. These questions were not included in the 2002 and so it is not possible to analyse changes over time. At face value, these data suggest that good social relations may have contributed as much as generous pension provision to high rates of general satisfaction. As with the subjective health data, these findings need to be interpreted with some caution. The high levels of satisfaction with living conditions may reflect aspirations and expectations rather more than objective conditions. For example, only 64% of rural Africans had electricity in 2009, yet 84% expressed satisfaction with where they lived. The qualitative in-depth interviews suggest that there may have been an element of positive bias in the questionnaire survey's treatment of family relations, with several older people who reported high

satisfaction scores revealing much lower levels of satisfaction with family relations on further probing. This included cases of older people who reported that they were coerced into handing over their pension benefits to other family members. Any potential bias may have been due to the presence of other family members during the questionnaire interviews, along with values of stoicism or stigma associated with poor treatment by younger family members (James, 1994). Even factoring in a degree of potential bias, the very high rates of satisfaction with family relations cannot be discounted and provides a very positive picture of later life experiences among the study populations. Conclusions and lessons for comparative international research These studies of older people in Brazil and South Africa demonstrate the dangers of generalising about experiences of later life in low and middle income countries. Among gerontologists, there is still a tendency to make sweeping claims about old age in the developed and developing worlds. Superficially, Brazil and South Africa would appear to have much in common. Both can be designated middle income countries, with democratic systems of governance that are seeking to address legacies of social inequality from lessenlightened regimes. As part of this, both have invested substantial resources into developing embracing welfare programmes, in which old age pensions feature prominently. These included near-universal basic pensions worth US$3 a day in 2002. Such similarities suggest there was more scope to develop and apply a simple comparative research design to explore experiences of later life than would be the case if comparing other countries. However, these similarities belie important and complex patterns of difference. For example, Brazil and South Africa had followed divergent paths to achieving broad pension coverage, with the former relying much more heavily on contributory social insurance mechanisms. Even among the relatively deprived groups included in our survey, a higher share of Brazilian households received contributory pensions than social pensions. This calls into question the exclusive focus on social pensions as a tool of old-age poverty reduction in developing countries. More generally, there were many important aspects of difference across the two countries and between the study populations. Older South Africans were on average better educated but poorer than their Brazilian counterparts. Yet the urban Coloured sub-group had higher living standards than any other included in the study. Large rural/urban and racial

Table 11 Per cent of households reporting satisfaction with social relations and living conditions, 2008/9. Country/year

Site

% satisfied/very satisfied with family relations

% satisfied/very satisfied with respect shown by others

% satisfied/very satisfied with where you live

Brazil

Rio Ilhéus urban Ilhéus rural Total Rural African Urban African Urban Coloured Total

91.2 92.1 94.6 92.4 92.6 92.0 95.5 93.6

90.9 92.7 95.8 92.6 90.8 84.5 97.3 92.0

79.4 92.1 89.9 85.3 84.3 73.6 95.2 86.4

South Africa

P. Lloyd-Sherlock et al. / Journal of Aging Studies 26 (2012) 243–252

disparities demonstrated the danger of generalising about experiences of later life at the national level, let alone across countries. By the time of the follow-up survey, these differences had become more pronounced, driven by shifts in public policy and divergent patterns of economic performance. As such, perhaps the most pertinent insight to emerge from these studies is that cross-country comparative research should be approached with a large measure of caution, if not trepidation. The findings of these country studies cannot be taken as representative of national trends. Indeed, our intentional focus on more deprived groups of older people ensures that they are not. This does not mean, however, that these studies are inherently less valuable than national surveys. By focussing on and becoming very familiar with distinct groups and specific locations, it was possible to situate and contextualise our research findings to an extent that is not possible in national studies. This added depth and credibility to our analysis and facilitated interpretation. For example, our survey found that very few older people in Ilhéus invested pension income in rural businesses. This contrasted with an earlier survey of pensioners in a different region of the country, which had reported a significant share of pension money was spent in this way (Sugamosto & Mohtadi Doustdar, 2000). The main cause of this discrepancy was the distinctive patterns of landholding in the two regions: our rural study site consisted of former cacao plantations with large parcels of land mainly owned by absentee landlords, whereas the rural site included in the earlier study mainly consisted of smallholdings. Since few pensioner household owned land in our site, there was little incentive to invest in it. This regional diversity demonstrates the danger of glossing over divergent sub-national experiences, as well as the value of relating research findings to separate surveys, both national and local. Despite the inevitable messiness of comparative research, it does provide opportunities to generate insights that would not be obtained through simpler research designs. The finding that providing older people with a regular and substantial flow of pension cash has an important effect on household poverty and economic vulnerability is hardly surprising. Nevertheless, its applicability across diverse groups and national settings lends valuable credibility to the claim that basic pensions (be they contributory or social) can reduce poverty in different developmental settings. The extent to which pensions appear to be pooled, and its consistency across sub-groups and countries, is a more significant finding and supports the claims often made by development agencies and NGOs. That variations in reported wellbeing are harder to explain is hardly surprising, given the much wider range of factors that can influence this outcome. The apparently steep rise in wellbeing in South Africa is particularly difficult to explain, although the fact that this rise was reported consistently across the three sub-groups adds credibility to this finding. The lack of a clear association with pension provision or other forms of public policy for older people suggests that the main reason for this rise must lie with other effects. Personal ageing itself may have been a factor: studies from developed countries consistently report that transition from mid-life to later life is associated with increases in subjective wellbeing (Blanchflower & Oswald, 2008). In other words, though important, pensions and cash transfers should not

251

be viewed as the prime determinants of wellbeing in later life for developing countries. The reality is much more complex, and relates to wider processes of development and change, as well as older people's own evaluations and expectations.

References Alger, Keith, & Caldas, Marcellus (1994). The declining cocoa economy and the Atlantic forest of Southern Bahia, Brazil: Conservation attitudes of cocoa planters. The Environmentalist, 14(2), 107–119. Amoateng, A. Y. (1997). The structure of urban black households: New survey evidence from a coloured and an African community on the Cape Flats in the Western Cape of South Africa. African Sociological Review, 1(2). Barreto de Oliveira, F., Beltrão, K., Pinheiro, S., Peyneau, J., & Mendonça, J. (2005). Older persons and social security. In A. Camarano (Ed.), Sixty plus: The elderly Brazilians and their new social roles. Rio de Janeiro: Instituto de Pesquisa Econômica Aplicada. Barrientos, A., & Mase, J. (2011). Poverty transitions among older households in Brazil and South Africa. : Brooks World Poverty Institute http://www. bwpi.manchester.ac.uk/resources/Working-Papers/bwpi-wp-15011.pdf Bhorat, H., Leibbrandt, M., Maziya, M., van der Berg, S., & Woolard, I. (2001). Fighting poverty. Labour markets and inequality in South Africa. Cape Town: UCT Press. Bhorat, H., & Kanbur, R. (2006). Poverty and well-being in post-apartheid South Africa. In H. Bhorat, & R. Kanbur (Eds.), Poverty and policy in post-apartheid South Africa. Johannesburg: HSRC Press. Blanchflower, D., & Oswald, A. (2008). Is well-being U-shaped over the life cycle? Social Science & Medicine, 66(8), 1733–1749. Bourne, R. (2008). Lula of Brazil. The story so far. Berkeley: University of California Press. Cockerham, W., Sharp, K., & Wilcox, J. (1983). Aging and perceived health status. Journal of Gerontology, 38(3), 349–355. Cohn, A. (2009). A reforma sanitaria brasiliera após 20 anos do SUS: Reflexões. Cadernos de Saúde Pública, 25(7), 1614–1619. Deutsche Gesellschaft für Technische Zusammenarbeit GTZ (2005). Social cash transfers — Reaching the poorest. http://www2.gtz.de/dokumente/bib/050542.pdf DFID (2005). Social transfers and chronic poverty: Emerging evidence and the challenge ahead. DFID practice paper. London: DFID. Farias, N., Pacheco de Souza, J., Laurenti, R., & Martins de Alencar, S. (2009). Cardiovascular mortality by gender and age range in the city of São Paulo, Brazil: 1996 to 1998 and 2003 to 2005. Arquivos Brasileiros de Cardiologia, 93(5), 461–467. HelpAge International (2003). Non-contributory pensions and poverty prevention. A comparative study of Brazil and South Africa. London: HelpAge International. James, M. (1994). Abuse and neglect of older people. Family Matters, 37, 94–97. Johnson, R. (2010). South Africa's brave new world. The beloved country since the end of apartheid. London: Penguin. Kahn, K., Tollman, S., Thorogood, M., Connor, M., Garenne, M., Collinson, M., et al. (2006). Older adults and the health transition in Agincourt, rural South Africa: New understanding, growing complexity. In B. Cohen, & J. Menken (Eds.), Aging in sub-Saharan Africa: Recommendations for furthering research. Washington D.C.: National Research Council. Kakwani, N., Son, H. H., & Hinz, R. (2006). Old-age poverty and social pensions in Kenya. Working paper no. 24. Brazil: International Poverty Centre. Lloyd-Sherlock, P. (2008). Ageing, well-being and development: A comparative study of Brazil and South Africa. Designing an effective qualitative strategy. http://www.sed.manchester.ac.uk/research/ageingandwellbeing/index. htm Lloyd-Sherlock, P. (2010). Population ageing and international development: From generalisation to evidence. Bristol: Policy Press. Lloyd-Sherlock, Beard, J., Minicuci, N., Chatterji, S., & Ebrahim, S. (submitted for publication). Hypertension among older adults in developing countries: Prevalence, awareness and control. Lloyd-Sherlock, P., Saboia, J., & Ramirez-Rodriguez B. (forthcoming) “Cash transfers and the wellbeing of older people in Brazil”. Development and Change. Moller, V. (2010). Perceptions of fortune and misfortune in older South African households: Social pensions and the ‘good life’. Institute of Social and Economic Research, research report no. 16. : Rhodes University. Moller, V., Dickow, H., & Harris, M. (1999). South Africa's ‘Rainbow People’, national pride and happiness. Social Indicators Research, 47(3), 245–280. Moller, & Ferreira, M. (2003). Non-contributory pensions and poverty study. South African survey report http://www.sed.manchester.ac.uk/research/ ageingandwellbeing/ncpps/report.htm

252

P. Lloyd-Sherlock et al. / Journal of Aging Studies 26 (2012) 243–252

Moller, V., & Saris, W. (2001). The relationship between subjective wellbeing and domain satisfactions in South Africa. Social Indicators Research, 55, 97–114. Mooney, G., & Mcintyre, D. (2009). South Africa: A 21st century apartheid in health and health care? The Medical Journal of Australia, 189(11/12), 637–640. Rofman, R., & Lucchetti, L. (2006). Pension systems in Latin America: Concepts and measurements of coverage. : World Bank http://siteresources. worldbank.org/EXTGDLNREGIONLAC/Resources/PensionCoverage_LACRofman-Lucchetti_Final.pdf Sabates, R., & Hammond, C. (2009). The impact of lifelong learning on happiness and well-being. London: Institute of Education http://www.niace. org.uk/lifelonglearninginquiry/docs/Ricardo-Wellbeing-evidence.pdf Saboia, J. (2009). Benefícios não-contributivos e combate à pobreza no Brasil 2002-2008. : Instituto de Economia, Universidade Federal de Rio de Janeiro www.sed.manchester.ac.uk/research/ageingandwellbeing/index.htm

Souza, P. (2011). Poverty, inequality and social policies in Brazil, 1995–2005. Mimeo. Brasilia: IPEA. Sugamosto, M., & Mohtadi Doustdar, N. (2000). Impactos da previdência rural na região sul: Énfase nas características mesorregionais. In G. Delgado, & C. CardosoJr. (Eds.), A universalizacão de direitos sociais no Brasil: A previdência rural nos anos 90. Brasília: IPEA. UNDP (2006). Social protection: The role of cash transfers. : International Poverty Centre, UNDP http://www.ipc-undp.org/pub/IPCPovertyInFocus8.pdf Van der Berg, S. (1997). South African social security under apartheid and beyond. Development Southern Africa, 14, 481–503. Woolard, I., Harttgen, K., & Klasen, S. (2010). The evolution and impact of social security in South Africa. : Southern African Labour and Deevelopment Research Unit, University of Cape Town http://erd.eui.eu/media/ BackgroundPapers/Woolard-Harttgen-Klasen.pdf Young, K. (1992). Household resource management. In L. Ostergaard (Ed.), Gender and development. A practical guide. London: Routledge.