Upload
others
View
1
Download
0
Embed Size (px)
Citation preview
1
WORKING PAPER
Men, pensions and wellbeing in rural South Africa: Tracking effects through policy shifts Enid Schatz F. Xavier Gómez-Olivé Margaret Ralston Jane Menken Stephen Tollman September 2012 Population Program POP2012-04
_____________________________________________________________________________
Men, pensions and wellbeing in rural South Africa: Tracking effects through policy shifts
Enid Schatz, University of Missouri F. Xavier Gómez-‐Olivé, University of the Witwatersrand Margaret Ralston, University of Missouri Jane Menken, University of Colorado Stephen Tollman, University of the Witwatersrand Contact Information: Enid Schatz 420 Lewis Hall Columbia, MO 65203 Email: [email protected] Paper prepared for the Population Association of America Meeting, Session
187: Public Policy and Families in Developing Countries
Draft, Please do not cite without authors’ permission
Abstract Pensions play an important role in older persons’ wellbeing in rural South Africa, with men and women reporting improved wellbeing in the years directly following pension-‐eligibility. Women’s age-‐eligibility has been set at 60 since the inception of South Africa’s non-‐contributory state-‐funded pension, whereas men until recently have begun receipt at age 65. As of 2010, men’s age of eligibility equalized with that of women. Using two panels of the WHO-‐Study of Global Aging and Adult Health study, collected in 2006 and 2010 in the MRC/Wits Rural Public Health and Health Transitions Research Unit (Agincourt), we examine whether “improvements” in men’s reports of wellbeing mirror men’s age-‐eligibility—65-‐69 in 2006, and 60-‐64 in 2010—thus tracking the pension-‐eligibility policy change. Data from both panels show that pension receipt is an important factor in older persons’ reports of wellbeing. In the 2010 data, those who report not receiving a pension are more likely to report poor wellbeing. In 2010, post policy change, there is a different relationship between pension, gender, and wellbeing than seen in the 2006 data. At the later period, pension-‐eligibility affects men’s and women’s reports of wellbeing similarly for most outcomes, whereas in 2006 we saw differences by age and sex. Finally, the results emphasize the benefits of evaluating both simple and composite indicators, with stronger relationships between pensions and worry and WHOQoL than other outcomes.
Schatz et al. PAA 2012 2
Introduction Pensions play an important role in older persons’ health and wellbeing in
South Africa (Schatz et al. 2011; Ardington et al. 2010; Gómez-‐Olivé et al. 2010). Women’s age eligibility has been set at age 60 since the inception of South Africa’s non-‐contributory state-‐funded pension, whereas men have until recently begun receipt at age 65. As of 2010, men’s age of eligibility equalized with that of women—i.e. men’s age-‐eligibility was lowered from 65 to 60 years. This is a major policy shift, argued on the basis of gender equality (Case number: 32838/05 in the High Court of South Africa). In this paper we investigate whether this policy shift provides evidence of a strong association between pension receipt and reports of improved wellbeing among rural Black South African men.
In Schatz et al. (2011) we examined the relationship between gender, pensions and wellbeing using data from the 2006 WHO-‐Study of Global Aging and Adult Health study (WHO-‐SAGE). We found a striking, but temporary cross-‐sectional improvement in reports of wellbeing upon reaching pension eligibility, with the relationship being stronger for women than for men. Since the downward shift in men’s age-‐eligibility began in 2008, the data offer a baseline prior to the policy change. The shift to younger age-‐eligibility for men, captured in a second wave of WHO-‐SAGE collected in 2010, allows us to assess the impact of this policy change on age patterns in wellbeing. Shifting “improvements” in wellbeing outcomes to younger ages, 60-‐64 in the 2010 data, compared to at 65-‐69 years as seen in the 2006 data, would confirm that the age pattern differences we are finding could be attributed to the pension and its effects. In addition, in 2006, men showed a pre-‐pension eligibility drop in wellbeing; at ages 60-‐64 men reported worse wellbeing than at ages 55-‐59. In this analysis we also will investigate if the pre-‐pension-‐eligibility drop in men’s wellbeing remains, or if shifting to the same age at pension eligibility reduces this pre-‐pension effect for men.
South Africa has high HIV-‐prevalence, as well as high levels of non-‐communicable disease among persons age 50-‐plus (Anderson & Phillips 2006). While HIV/AIDS rates in elders are currently lower than among their younger peers, increasing prevalence stresses families and household systems. Older people, more often women, assume roles as caregivers, particularly in multigenerational households affected by the illness (Munthree & Maharaj 2010; Møller & Devey 2003; Møller 1998). Because of the differentiation in gendered roles surrounding caregiving, much of the research on the effects of the HIV/AIDS on households has focused on the experiences of women.
In order to examine these issues, we examine six measures of wellbeing—four individual indicators based on responses to single questions and two composite measures constructed by WHO from responses to multiple questions. We first show relevant descriptive statistics for men and women in 2006 and in 2010. We then present regression models (including women and men) for each panel to see if similar patterns emerge in the two years. With the 2010 data, we go on to examine how inclusion of a variable measuring pension receipt, as opposed to using age as a proxy, affects the relationship of aging to wellbeing. Background
Schatz et al. PAA 2012 3
Aging and health in rural South Africa has been largely shaped by a burden of disease that includes endemic levels of HIV/AIDS, as well as a growing epidemic of non-‐communicable disease, particularly among those over age 50 (Anderson & Phillips 2006; Kahn et al. 2006). Men’s and women’s mortality profiles are quite different, however. The probability of dying between ages 15 and 60 was 34% for women and 51% for men (Dorrington, Moultrie & Timaeus 2004); men’s excess mortality in this age range largely came from accidents and homicides (Dorrington et al. 2001; Hosegood, Vanneste & Timaeus 2004). Obesity, hypertension, diabetes and stroke are emerging as important illnesses among older persons in South Africa; nearly 70% of those over age 60 report having at least one chronic illness and more than half report having a resulting physical disability (Ferreira 2000). Whereas older women are likely to be widowed and to live alone or with extended family, the majority of older men are married or in a relationship (UN 2010). Women’s longer life expectancy, age differences between spouses with women usually marrying men who are older than them, and men’s greater likelihood to remarry after being widowed or divorced, all lead to older South African men being more likely than older women to live with a spouse (Cohen & Menken 2006; UN 2010). Thus, older men often have more resources and assistance on which they can draw than have older women (Ezeh, Chepngeno, Kasiira, & Woubalem 2006; UN 2010). The social roles of older South African men, and men in rural South Africa more generally, continue to be shaped by apartheid era migration policies which created circular migration flows, particularly of men, from rural to urban areas (Susser 2011). This may mean that older men are more likely than older women to be used to having income and controlling that income. In addition, older men are likely to have spent more of their lives living outside the rural area than in the rural area prior to retirement. South African old-age pension
While the pension is available to all South Africans, it is an important stable economic resource for the majority of black South African households (May 2003; Sanger & Mtati 1999). More than 90% of black older South Africans access the pension (Ferreira 2006; Burns, Keswell, & Leibbrandt 2005), and the cash transfer is as much as twice the median per capita income of the black population (Case & Deaton 1998). Thus, pension receipt may significantly increase the income in black South African households (Barrientos et al. 2003; May 2003; Møller & Devey 2003). In 2005, the monthly pension was SAR780 (approximately USD130) (Samson, MacQuene, & van Niekerk 2006); it increased incrementally annually in relation to inflation and cost of living; in 2010 the grant amount was SAR1080 (approximately USD180)(SASSA 2012).
Many rural households depend on a variety of social grants to sustain them; income-‐pooling of pensions and other social grants provides a reliable and regular safety net for the needs of older persons and their households (May 2003; Sanger & Mtati 1999). Because of older persons’ income–pooling, the pension has also been shown to improve overall food security and wellbeing in older people’s households (Barrientos et al. 2003; Møller & Devey 2003). In many cases pensions are viewed as
Schatz et al. PAA 2012 4
a household resource, covering family members’ health and everyday needs (Case & Deaton 1998; May 2003). The presence of a pensioner, particularly of a woman, enhances the wellbeing of other household members (Ardington et al. 2010; Burns et al. 2005; Duflo 2003). Although there has been a relatively little research conducted on male pensioners’ roles in meeting household needs, limited research does show that men also worry about meeting household needs (Sherr, 2009). For example, Wiener and colleagues (2001) in a study of 31 fathers of HIV-‐positive 6-‐16 years-‐olds reported that over half revealed elevated levels of parenting psychological stress.
Although research on spending habits of pensioners in South Africa is limited, there is evidence of gendered spending habits more generally. The gender of the household head in South Africa significantly influences spending on certain budget items (Maitra & Ray 2003). Namely, women household heads are more likely than their male counterparts to spend on clothing and child-‐care. Posel, Fariburn, and Lund (2006) found that pension income, specifically received by women, was used to support grandchildren. Case and Deaton (1998) find that despite having less income overall, when women control income, they often favor spending money on food over other expenditures. The largest differences, however, were that female household heads spent considerably less than male household heads on alcohol, tobacco and transportation. This is consistent with the finding that female pensioners share more of their income with household members than male pensioners (Posel et al. 2006). This literature suggests that men and women, including pensioners, value household expenses and spend household income in different ways.
While income-‐pooling may decrease the direct benefits of the pension to pensioners, recent research suggests that pension receipt improves health and wellbeing of pensioners, as well. Ardington and colleagues (2010) find that pensions buffer financial and emotional impacts of an adult child’s death and the resulting carework for grandchildren left behind (Ardington et al. 2010). A study from the Eastern Cape shows older South Africans’ (aged 60 plus) perceptions of their ability to provide care for children are primarily dependent on their knowledge about accessing pensions (Boon et al. 2010). Gómez-‐Olivé et al. (2010), using WHO-‐SAGE 2006 data from Agincourt, found that despite having aged, 60-‐69 year olds did not report significantly worse health status or function compared to 50-‐59 year olds. They suggest that the plateau in reported health and wellbeing may be related to receipt of pensions “which allow [pensioners] to have a better life with higher food security and, importantly, with higher capacity to help the children in their households who have also higher food security and higher schooling” (p.32). Schatz et al. (2011) extended the Gómez-‐Olivé et al. (2010) analysis by examining the same outcomes, but in 5-‐year rather than 10-‐year age groups. Across age and sex groups, the findings point to a greater impact of pension receipt on wellbeing for women than men, but with a transitory observed effect for both (Schatz et al. 2011). Women report better wellbeing during a striking “honeymoon” period in the first years of pension-‐eligibility (ages 60-‐64). While, men’s wellbeing drops pre-‐pension eligibility (ages 60-‐64), increases at pension-‐eligibility (65-‐59), then declines (70-‐
Schatz et al. PAA 2012 5
74). Pensions may enhance financial wellbeing, but results from previous papers show their observed effect on social wellbeing is gendered and temporary. Older persons and HIV/AIDS
The focus of research on older persons in HIV/AIDS endemic contexts has largely centered on roles of older women in caregiving and how this might affect their health and wellbeing. Men are often portrayed as not as dynamic a part of family caregiving; however, research from North American and Latin American are affirming that men participate in caregiving in their own ways and that they may participate more than is commonly believed (Bannon & Caorreia 2006). While clearly important, the assumption of men’s lack of participation in caregiving has meant that men’s experiences have often been left out of research (Bannon & Caorreia 2006). Data from Tanzania show that even though women did nearly twice the carework as men, older men and women were equally likely to be caregivers for sick household members (Dayton & Ainsworth 2002). Munthree and Maharaj (2010) found that more South African men than women reported knowing someone who was HIV-‐positive, as well as someone who had died of AIDS. While women reported much more time and energy spent on caregiving due to a clear gendered division of labor within families, “men are also responding to the changing situation by performing roles that extend beyond financial support” (Munthree & Maharaj 2010, p. 169).
Despite contributing fewer hours to caregiving, men may be just as vulnerable to the increased stress caused by the illness. Given their lack of experience with caregiving roles, men may try to relieve unprecedented emotional and physical demands of caregiving through alcohol and other escapist behaviors (Mudege & Ezeh 2009). Thus, stress also could be associated with early deaths among older men (Ice et al. 2010). While much literature to date has focused on women, this paper begins to explore older men’s experiences of living in an AIDS-‐endemic community. In addition, this paper further extends work on the role of pensions in older persons lives to examine how policy changes and longitudinal data can provide additional insight into the relationship between pensions and wellbeing. Data & Methods
Data: The WHO-‐INDEPTH Study of Global Ageing and Adult Health Survey (WHO-‐SAGE) was conducted in 2006 and 2010 in the MRC/Wits Rural Public Health and Health Transitions Research Unit (Agincourt) site as part of a multi-‐country longitudinal project to increase understanding of health and wellbeing of adult populations in developing countries (http://www.who.int/healthinfo/systems/sage/en/index.html). The Short Version of the WHO-‐SAGE questionnaire was incorporated as a census-‐module in the Agincourt Health and socio-‐Demographic Surveillance System (AHDSS) for persons over the age of 50 in 2006 and again in 2010. The module included questions on self-‐reported health, functionality, and wellbeing. The census collected additional demographic and household-‐level data. In 2010, upon the request of this research team, the census included a question about pension receipt. Thus, the data
Schatz et al. PAA 2012 6
presented represent two special surveys plus linked data from the annual census in each survey year.
In 2006, approximately 65% (N=4085) of the target population completed the module. Nearly two-‐thirds of those who completed the module were women due to high and persisting male labor migration even in the older age groups. [For details of the sample and data collection, see Gómez-‐Olivé et al., 2010 and Kowal et al., 2010.] In 2010, 60% of the target population of approximately 10,000 completed the questionnaire with only 0.4% refusing. The rest either were not found at the time of the census (35%), ineligible (4%) or dead (1.6%). The resulting sample is approximately 5,980 individuals age 50 and above, about one-‐quarter of whom were male and three-‐quarters female (see Table 2B).
The main reason for finding many fewer men than women at all adult ages in Agincourt is related to migration. Collinson and colleagues (2007) find a high level of temporary or circular migration between Agincourt and urban areas. These movements lead to strong ties in Agincourt between its rural population and those living in urban areas. Over half the male population between ages 25 to 54 are likely to migrate at some point (Collinson et al. 2007). Males are most likely to take part in circular migration after age 19 and the likelihood remains high until around age 74. Female likelihood of temporary migration is lower then males and they have a greatest likelihood of migrating between the ages 25 to 49 (Collinson, Tollman, & Kahn 2007). Thus, the sample is dominated by women, who are less likely to migrate, and also is likely to be biased toward males who have retired, and those who might be too ill or old to work.
About one-‐third of the Agincourt population is of Mozambican origin. They largely came to Agincourt during the Mozambican civil war from the mid-‐1970s to early 1980s. Previous research from Agincourt has shown that self-‐identified Mozambicans are less well off than the host South African population in terms of education, household assets, and child mortality (Gómez-‐Olivé et al. 2010; Hargreaves, Collinson, Kahn, Clark, & Tollman 2004). Prior to 2006, Mozambican permanent residents were not eligible for South African social grants; however, even before the Constitutional Court ruling, a large number of Mozambicans managed to access pensions (Schatz 2009).
Variables: The basic measures that we use to explore issues of social wellbeing come from two constructs of subjective wellbeing—affect and quality of life. These constructs are often put together, with a number of others, to determine overall quality of life (e.g. as in WHOQoL). We examine four individual domains as dichotomous variables: sad, worry, dissatisfied and unhappy. Respondents were asked the extent to which, in the last 30 days, they felt sad, low, or depressed (sad), felt worry or anxiety (worry), were satisfied with life as a whole (dissatisfied) and felt happy in general (unhappy). In addition, we examine two WHO composite variables: WHOQoL, described above, and WHODASi-‐ which includes a number of measures to assess disability and functionality. All six measures are outlined in more detail in Table 1.
Tables 2A and 2B show demographic and other characteristics of the population by age and sex in 2006 and 2010, respectively. They include household structure, socio-‐economic status (SES), marital status, education, employment
Schatz et al. PAA 2012 7
status, nationality of origin, and self-‐reported health. Household structure includes mean household size and mean percent of household members under age 15. SES makes use of a household asset score derived from 34 variables collected in 2005 for the 2006 models and collected in 2009 for the 2010 models (including information about the type and size of dwelling, access to water and electricity appliances and livestock owned and transport available). The score was derived through principal component factor analysis and then divided into quintiles. Marriage unions in this area may be traditional, civic, or polygamous (a small minority). We dichotomize marital status—currently married or unmarried (those never married, separated, divorced, or widowed). Education is categorized as no formal education or some education. Employment status, collected for 2004 in the 2005 Agincourt census, and again for 2008 in the 2009 census, is coded as currently working or not. The majority of those not working were not looking for work but had retired, having concluded their working career. Employment status focused on those with permanent formal work, so may not capture those doing informal income-‐generating activities. “Percent South African” captures self-‐identification as South African or Mozambican. Self-‐rated health is categorized as “bad” or not. These variables are included in regression models that add covariates to the more basic models focused on age, sex and pension receipt.
Analysis: We first describe characteristics for men by age groups in 2006 and 2010 (Tables 2A & 2B). We then present logistic regression models assessing relationships of wellbeing to age and sex. All outcome measures are calibrated so that, for dichotomous outcomes, 1 indicates that the respondent was worse off -‐ referred to as "poor". We include tables from Schatz et al. 2011 to represent the 2006 data, and present new tables that show similar models for the 2010 data (Tables 3A & 3B). We report age in five-‐year intervals to capture recent pension-‐eligibility. There was no direct measure of pension receipt in 2006, but as shown in Table 4, in 2010 over 80% of age-‐eligible individuals reported pension receipt. In order to compare 2006 to 2010 data in these tables, we continue to use age as a proxy for pension-‐eligibility. We run tests in for these models to see if there is a significant change in each of the six wellbeing outcomes when moving from pre-‐pension to pension-‐eligibility, and from initial pension-‐eligibility to 5-‐years post-‐eligibility. In Table 5, we focus on models using 2010 data that incorporate the direct measure of pension receipt. These models also include tests for significant change in each of the six wellbeing outcomes when moving from pre-‐pension to pension-‐eligibility, and from initial pension-‐eligibility to 5-‐years post-‐eligibility, while controlling for pension receipt. Results
Tables 2A and 2B show the characteristics of men and women in the two panels of data. In each table the age categories that corresponds with pension-‐eligibility are highlighted. The most notable change between tables 2A and 2B is that for men the shift in reported wellbeing at pension-‐eligibility (a smaller percentage report “poor” wellbeing than in the pre-‐pension-‐eligibility age category) mirrors the downward shift in age-‐eligiblity. In other words, in 2006, a smaller percentage of men reported “poor” wellbeing in the 65-‐69 age category than in the 60-‐64 age
Schatz et al. PAA 2012 8
group; where as in 2010, a smaller percentage of men in the 60-‐64 age category reported “poor” wellbeing than in the 55-‐59 age category. Thus, the pattern of changes over age-‐categories in reported wellbeing is more similar among men and women in 2010 than it was in 2006. These differences will be explored further through regression analysis.
Table 3A (taken from Schatz et al. 2011) shows results based on SAGE 2006. In general, the probability of scoring as "poor" the majority of the wellbeing variables increased with age; women were more likely than men to score in the "poor" category. However, there is a significant interaction between age and sex. The tests showed that women 60-‐64 were better off than those younger and older, coinciding with pension eligibility. For men, there appeared to be a slight increase for those 65-‐69 (the age of pension-‐eligibility in 2006). These results held for most of the measures used. They held when only age, sex, and age-‐sex interactions were in the models and also when a variety of social and economic predictors (based on those shown in Table 2A) were introduced (see Schatz et al. 2011).
Table 3B, based on SAGE 2010, replicates the age-‐sex models. As in 2006, women report significantly worse wellbeing than men, controlling for age. However, of the six measures used, only one still had an age-‐sex interaction. This is the first evidence of change over the four-‐year period between the surveys. Since there was no age-‐sex interaction, the tests in Table 3B explore whether there was a significant change for both men and women in wellbeing from age 55-‐59 to 60-‐64 (pre-‐pension to pension-‐eligibility), from 60-‐64 to 65-‐69 (pension-‐eligibility to post-‐pension-‐eligibility), and 65-‐69 to 70+ (post-‐pension to post-‐post-‐pension). The “improvement” in reported wellbeing at pension-‐eligibility found in 2006 was much less common in 2010. Only worry and WHOQoL showed significant “improvement”, with declines in reported “poor” wellbeing. The “honeymoon” effect, a significant increase in “poor” wellbeing from pension-‐eligibility to post-‐pension eligibility disappeared completely. Instead, the expected increase in "poor" wellbeing with age was postponed to older ages (70+).
To understand further the effects of the pension on wellbeing, we requested that a direct question on pension receipt (and all grants) be included in the 2010 census. We therefore had additional leverage—a direct measure of pension receipt rather than having to rely on age as a proxy. Table 4 shows that a small percentage of the population (less than 3%) manages to receive the pension prior to official age eligibility; as one would expect, the percentage of the population covered by the grant increases at the age of eligibility. Over 80% of people over 60 report receiving the grant.
We consider pension receipt directly in Table 5 by adding it to the regression models. In all cases, those who were not receiving the pension were more likely to be in the "poor" category. Pension receipt does not seem to affect gender differences, however; controlling for pension status, women still report significantly poorer wellbeing across all six outcomes. When information on pension receipt is included in the models, there is significant “improvement” in wellbeing at pension eligibility for four of the six outcomes, compared to just two outcomes in Table 3B. Once again, however, there is less evidence of the “honeymoon” effect seen in the 2006 data, with only WHODASi showing a significant decline post-‐pension eligibility
Schatz et al. PAA 2012 9
prior to expected increases in “poor” wellbeing at older ages (70+). Discussion
The analysis above provides new insights into the relationships between gender, pensions and wellbeing. First, women continue to be more likely to report poor wellbeing across a range of indicators. Second, the strong age-‐gender interaction found in the 2006 data does not hold with the 2010 data—showing a less significant difference in patterns of change by age for men and women. Third, the “honeymoon” effect that was so prominent in 2006 with “improved” wellbeing at pension-‐eligibility and then a significant decline post-‐eligibility also is much more muted and stretched out in 2010. Finally, as with Schatz et al. (2011), we note this change is apparent in the specific questions, which indicates that there is value in "unpacking" composite measures.
The continued strong gender differences in reporting of wellbeing need to be explored further. Women fare much worse than men across all wellbeing indicators. It is unclear whether this is an issue of reporting or of true differences in health and wellbeing. The SAGE data include anchoring vignettes. Future work should explore how these can be used to unpack the reasons behind gender differences in wellbeing as reported in both 2006 and 2010, regardless of pension-‐eligibility or receipt.
The age-‐gender interaction, which was so prominent in the 2006 SAGE data disappears with the change in policy, however: when men and women begin receiving pensions at the same age there are fewer sex differences in patterns of reported wellbeing as individuals age. Some of this change may be that when answering questions about wellbeing in 2006, men compared their situation not only with other men in their age group, but also with women of their same age, who at the time were receiving pensions while the men were not. This might have affected the ways in which men reported their own wellbeing, which is 2006 showed declines in the 60-‐64 pre-‐pension age-‐category.
In general, the downward shift in men’s age-‐eligibility by 2010 seems to be associated with reports of better wellbeing upon pension-‐eligibility for both men and women equally—these changes are even more evident in the models where actual pension receipt is measured as opposed to simply using age as a proxy. This means that when pension-‐receipt information is not available, the relationships between pensions and improvements in wellbeing are likely underestimated. The subsequent decline in wellbeing post-‐pension receipt seen in 2006 is not present in the 2010 data. This may be due partly to the equalization of men’s and women’s pension-‐eligibility, as explained above. Future analyses will include social and economic covariates to see if these affect the pattern of reported wellbeing as individuals age.
The analyses presented here are cross-‐sectional. The final step will be to analyze the panel data for those who were respondents in 2006 and again in 2010. The linked dataset is currently being constructed. We will then be able to examine change in wellbeing over time for each respondent and test whether the findings seen in the two cross-‐sections hold at the level of the individual.
Schatz et al. PAA 2012 10
Schatz et al. PAA 2012 11
References
Anderson, B.A. & Phillips, H.E., 2006. Adult mortality (age 15-64) based on death notification data in South Africa: 1997-2004, Pretoria: Statistics South Africa. Available at: internal-‐pdf://Report-‐03-‐09-‐052004-‐2315590144/Report-‐03-‐09-‐052004.pdf.
Ardington, C., Case, A., Islam, M., Lam, D., Leibbrandt, M., Menendez, A., & Olgiati, A. (2010). The impact of AIDS on intergenerational support in South Africa: Evidence from the Cape Area Panel Study. Research on Aging, 32(1), 97 -‐121.
Bannon, I., & Correia, M. (2006). The Other Half of Gender: Men’s Issues in Development. World Bank Publications.
Barrientos, A., Ferreira, M., Gorman, A., Heslop, H., Legido-‐Quigley, P., Lloyd-‐Sherlock, V., & Moller, J. (2003). Non-contributory pensions and poverty prevention: A comparative study of Brazil and South Africa. HelpAge International. Retrieved from http://www.pension-‐watch.net.
Boon, H., James, S., Ruiter, R. A. C., van den Borne, B., Williams, E., & Reddy, P. (2010). Explaining perceived ability among older people to provide care as a result of HIV and AIDS in South Africa. AIDS Care, 22(4), 399-‐408.
Burns, J., Keswell, M., & Leibbrandt, M. (2005). Social assistance, gender, and the aged in South Africa. Feminist Economics, 11(2), 103-‐115.
Case, A., & Deaton, A. (1998). Large cash transfers to the elderly in South Africa. The Economic Journal, 108(September), 1330-‐1361.
Case, A., & Menendez, A. (2007). Does money empower the elderly? Evidence from the Agincourt demographic surveillance site, South Africa. Scandinavian Journal of Public Health, 35(Suppl 69), 157 -‐164.
Collinson, M. A. (2009). Striving against adversity: the dynamics of migration, health and poverty in rural South Africa. Umeå University Medical Dissertations, Epidemiology and Public Health Sciences, Department of Public Health and Clinical Medicine, Umeå University.
Collinson, M. A., Tollman, S. M., & Kahn, K. (2007). Migration, Settlement Change and Health in Post-‐Apartheid South Africa: Triangulating Health and Demographic Surveillance with National Census Data1. Scandinavian Journal of Public Health, 35(69 suppl), 77–84. doi:10.1080/14034950701356401.
High Court of South Africa. Christian Roberts and 4 others vs Minister of Social Development and 3 others, High Court of South Africa, Transvaal Provincial Division, Case no.32838/05.
Duflo, E. (2003). Grandmothers and granddaughters: Old-‐age pension and intra-‐household allocation in South Africa. World Bank Economic Review, 17(1-‐25).
Ezeh, A., Chepngeno, G., Kasiira, A., & Woubalem, Z. (2006). The situation of older people in poor urban settings: The case of Nairobi, Kenya. In B. Cohen and J. Menken (Eds.) Aging in Sub-Saharan Africa: Recommendations for Furthering Research (189-‐213). Washington, D.C.: National Academies Press.
Ferreira M. 2000. Growing old in the new South Africa. Ageing International. 25(4): 32–46.
Schatz et al. PAA 2012 12
Ferreira, M. (2006). The differential impact of social-‐pension income on household poverty alleviation in three South African ethnic groups. Ageing & Society, 26(3), 337-‐354.
Gómez-‐Olivé, X., Thorogood, M., Clark, B. D., Kahn, K., & Tollman, S. M. (2010). Assessing health and well-‐being among older people in rural South Africa. Global Health Action, 3(Suppl 2), 23-‐35.
Ice, Gillian H., Jaja Yogo, Victor Heh, and Elizabeth Juma. (2010). The Impact of Caregiving on the Health and Well-‐being of Kenyan Luo Grandparents. Research on Aging 32:40 -‐66.
Hargreaves, J. R., Collinson, M. A., Kahn, K., Clark, S. J., & Tollman, S. M. (2004). Childhood mortality among former Mozambican refugees and their hosts in rural South Africa. International Journal of Epidemiology, 33(6), 1271 -‐1278.
Kahn, K. Tollman, S. Connor, M. Garenne, M. Collinson, M. & Hundt, G. (2006). Older Adults and the Health Transition in Agincourt, Rural South Africa: New Understanding, Growing Complexity. In B. Cohen and J. Menken (Eds.) Aging in Sub-Saharan Africa: Recommendations for Furthering Research (166-‐188). Washington, D.C.: National Academies Press.
Kahn, K., Tollman, S., Collinson, M., Clark, S., Twine, R., Clark, B., … Garenne, M. (2007). Research into health, population and social transitions in rural South Africa: Data and methods of the Agincourt Health and Demographic Surveillance System. Scandinavian Journal of Public Health 35, 8 -‐20.
Kowal, P., Kahn, K., Ng, K., Naidoo, N., Abdullah, S., Bawah, A., Tollman, S. (2010). Ageing and adult health status in eight lower-‐income countries: the INDEPTH WHO-‐SAGE collaboration. Global Health Action, 3(Suppl 2), 11-‐22.
May, J. (2003). Chronic Poverty and Older People in South Africa. Chronic Poverty Research Centre, Working paper 25.
Møller, V. (1998). “Innovations to promote an intergenerational society for South Africa to promote the well-‐being of the black African elderly.” Society in transition 29:1-‐12.
Møller, V., & Devey, R. (2003). Trends in living conditions and satisfaction among poorer older South Africans: objective and subjective indicators of quality of life in the October Household Survey. Development Southern Africa, 20(4), 457-‐476.
Munthree, C., & Maharaj, P. (2010). Growing Old in the Era of a High Prevalence of HIV/AIDS: The Impact of AIDS on Older Men and Women in KwaZulu-‐Natal, South Africa. Research on Aging, 32(2), 155 -‐174.
Rakodi, C. (2002). A livelihoods approach-‐Conceptual issues and definitions. In C. Rakodi & T. Lloyd (Eds.), Urban Livelihoods: A People-Centred Approach to Reducing Poverty. Earthscan Publications Ltd.
Samson, M., Kenneth M., & van Niekerk, I. (2006). Policy Brief 1: Social grants South Africa. Inter-Regional Inequality Facility. Accessible at: http://www.odi.org.uk/resources/download/1067.pdf
Schatz, Enid, Xavier Gómez-‐Olivé, Margaret Ralston, Jane Menken and Stephen Tollman. (2011). Gender, Pensions, and Social Wellbeing in Rural South Africa. CU Working Paper, POP2011-‐06. http://www.colorado.edu/ibs/pop/pubs/wp.html
Schatz et al. PAA 2012 13
Schatz, E. (2009). Reframing vulnerability: Mozambican refugees’ access to state-‐funded pensions in rural South Africa. Journal of Cross-Cultural Gerontology, 24(3), 241-‐258.
Schatz, E., & Ogunmefun, C. (2007). Caring and contributing: the role of older women in multi-‐generational households in the HIV/AIDS era. World Development, 35(8), 1390-‐1403.
Sherr, L. (2009). Vertical transmission of HIV-‐ Pregnancy and Infant Issues. In P. Rohleder, L. Swartz, S. Kalichman, and L. C. Simbayi (Eds.) HIV/AIDS in South South Africa 25 Years On: Psychosocial Perspectives. Springer.
Susser, I. (2011). AIDS, Sex, and Culture: Global Politics and Survival in Southern Africa. John Wiley & Sons.
Twine, W. & Hunter, L. (2010). Adult Mortality, Food Security and the Use of Wild Natural Resources in a Rural District of South Africa: Exploring the Environmental Dimensions of AID. In A. Niehof, G. Rugalema, & S. Gillespie (Eds.), AIDS and Rural Livelihoods Dynamics and Diversity in sub-Saharan Africa, (chapter 10). London: Earthscan.
Twine, W. & Hunter, L.M., 2011. Adult mortality and household food security in rural South Africa: Does AIDS represent a unique mortality shock? Development Southern Africa, 28(4), pp.431–444.
UN-‐United Nations. 2010. World Population Ageing 2009 Department of Economics and Social Affairs Population Division Report: ST/ESA/SER.A/295 access at: http://www.un.org/esa/population/publications/WPA2009/WPA2009-‐report.pdf
Wiener, L. S., Vasquez, M. J., and Battles H.B. (2001). Brief report: fathering a child living with HIV/AIDS: psychosocial adjustment and parenting stress. Journal of Pediatric Psychology, 26(6), 353-‐358.
Table 1. Wellbeing Outcome Measures Variable Original coding Dichotomous Recoding Sad* None
Mild 0=none/mild
Moderate Severe Extreme
1= at least some sad
Worry None Mild
0=none/mild
Moderate Severe Extreme
1= at least some worry
Unhappy Very happy Happy
0= happy
Neither Unhappy Very Unhappy
1=at least some unhappy
Dissatisfied Very satisfied Satisfied
0= satisfied
Neither Dissatisfied Very dissatisfied
1= at least some dissatisfied
Variable Components Dichotomous recoding of quintiles
WHODASi (Function)
Interpersonal activities, Difficulties in daily living: Standing, Walking, Household duties, Learning, Concentrating, Self-‐care Scale 0 (low ability) to 100 (high ability)
0=60% best WHODASi 1=40% worst WHODASi
WHOQoL (Quality of Life)
Enough energy for daily life, Enough money to meet needs, Satisfaction with: Your health, Yourself, Ability to perform daily activities, Personal relationships, Condition of your living space, Rate your overall quality of life Scale 0 (low QOL) to 100 (high QOL)
0=60% best WHOQoL 1=40% worst WHOQoL
*Questions translated into Shangaan for WHO-‐SAGE: Overall in the last 30 days, how much of a problem did you have… SAD:…with feeling sad, low or depressed?; WORRY: … with worry or anxiety?; DISSATISFIED: Taking all things together, how satisfied are you with your life as a whole these days?; UNHAPPY: Taking all things together, how would you say you are these days?
Table 2A. Background Characteristics and Wellbeing by Sex and 5-‐Year Age Group, Agincourt HDSS and SAGE 2006 Recent pension-‐eligible age highlighted
MEN WOMEN 50-‐54 55-‐59 60-‐64 65-‐69 70-‐74 75+ 50-‐54 55-‐59 60-‐64 65-‐69 70-‐74 75+ Mean household size 6.6 6.3 7.3 7.0 6.9 6.0 8.3 8.0 7.5 7.0 6.5 6.4 Mean percent of household under 15 26.4 20.9 22.3 21.5 21.3 19.0 30.0 26.5 26.0 26.3 25.3 23.9 Socio-‐economic status (quintiles) First (lowest) 13.4 16.0 17.0 9.8 18.3 17.1 13.9 12.8 9.8 13.6 17.4 17.2 Second 21.4 18.4 14.2 10.3 13.3 16.3 15.9 16.4 17.4 16.8 21.4 23.3 Third 16.1 17.6 27.4 20.1 13.3 19.1 20.9 22.4 17.9 23.8 18.9 18.0 Fourth 16.1 18.4 11.3 29.3 23.3 19.1 23.9 19.2 25.8 24.9 22.6 23.4 Fifth (highest) 33.0 29.6 30.2 30.5 31.7 28.5 25.4 29.2 29.1 20.9 19.7 18.2 Percent unmarried 25.9 24.0 18.9 19.5 20.0 23.6 43.9 49.8 53.8 64.4 75.4 84.3 Percent with no formal education 36.6 42.4 44.3 53.5 65.8 70.7 50.4 54.6 57.2 71.5 85.1 84.6 Percent working in 2004 39.3 29.6 36.8 17.8 15.0 5.7 26.1 24.0 16.4 8.6 5.4 2.2 Percent South African 82.3 79.9 83.8 75.4 72.0 68.8 73.1 71.0 75.6 71.4 67.6 71.1 Percent bad self-‐rated health 12.5 16.0 22.6 21.8 10.8 17.5 9.4 13.0 12.2 15.2 18.3 24.7 Wellbeing variables (Percent scored 1)
Sad 39.3 35.2 44.3 35.1 34.2 43.1 45.0 44.8 39.1 47.9 52.3 51.1 Worry 50.9 42.4 50.9 39.1 40.0 49.2 52.6 57.1 44.0 55.8 59.1 60.7 Dissatisfied 30.0 29.0 39.3 32.4 31.6 34.4 32.6 33.7 23.6 29.8 36.8 39.0 Unhappy 34.3 27.7 38.3 27.7 28.7 37.8 33.2 33.4 25.5 32.9 36.3 39.3 Poor WHODASi 28.4 30.9 30.8 29.6 32.2 40.9 30.7 35.1 29.5 41.6 50.8 60.1 Poor WHOQoL 37.9 31.6 41.9 34.7 35.9 41.7 39.4 41.4 30.6 38.3 45.3 50.0 N 112 125 106 174 120 246 460 438 409 382 350 644
Table 2B. Background Characteristics and Wellbeing by 5-‐Year Age Group for Women and Men in 2010, Agincourt HDSS and SAGE Recent pension-‐eligible age highlighted
Men Women 50-‐54 55-‐59 60-‐64 65-‐69 70-‐74 75+ 50-‐54 55-‐59 60-‐64 65-‐69 70-‐74 75+ Mean household size 6.1 6.9 7.4 8.6 6.7 6.0 8.2 8.1 7.7 7.2 6.6 6.4 Mean percent of household under 15
24.4 21.9 20.9 25.9 18.2 17.8 27.5 26.7 25.4 25.6 25.2 22.0
Socio-‐economic status (quintiles)
First (lowest) 19.6 20.0 14.9 11.7 14.4 20.4 15.3 12.2 15.4 13.2 17.0 21.1 Second 16.8 17.5 16.8 15.2 17.3 22.6 19.3 19.7 20.0 21.1 23.3 22.5 Third 20.1 19.6 20.6 23.3 21.4 22.3 23.0 20.9 20.2 23.1 22.4 21.2 Fourth 14.8 22.5 19.5 17.5 16.1 18.3 20.5 20.8 18.2 20.5 19.0 18.2 Fifth (highest) 28.7 20.4 28.2 32.3 30.9 16.4 21.9 26.5 26.2 22.1 18.3 17.0 Percent unmarried 29.2 22.6 22.4 19.3 25.8 31.7 40.5 49.1 58.5 62.3 73.5 82.4 Percent with no formal education
34.6 40.3 49.0 55.7 54.8 71.8 47.5 52.4 55.6 63.9 82.5 88.8
Percent working (2008) 55.1 47.6 40.4 25.9 12.1 6.7 33.1 30.8 20.0 8.9 5.0 3.4 Percent bad self-‐rated health 13.6 11.9 12.1 15.7 14.3 20.5 13.5 15.3 16.4 17.5 22.5 28.7 Percent South African 75.1 73.3 69.8 68.3 71.7 64.8 67.7 73.5 69.9 73.0 64.6 66.4 Wellbeing variables (Percent scored 1)
Sad 33.0 37.5 26.4 27.4 38.7 38.8 39.0 38.1 38.3 38.2 45.5 48.1 Worry 40.5 40.1 32.1 31.7 40.7 41.7 45.1 43.6 39.5 41.1 48.7 51.6 Dissatisfied 20.7 19.0 15.5 19.2 17.8 25.5 19.5 21.2 18.2 21.1 24.0 31.6 Unhappy 25.4 25.1 20.8 17.4 25.8 30.9 27.3 27.0 25.6 28.4 29.1 34.7 Poor WHODASi 26.3 27.6 22.6 26.5 33.6 44.0 29.7 29.5 31.1 36.0 45.2 57.4 Poor WHOQoL 36.2 32.5 24.5 28.7 29.1 41.0 34.6 38.4 30.7 31.8 39.3 50.3 N 213 243 265 230 244 327 807 747 652 589 560 1,100
Table 3A. Relationships of Sex and 5-‐year Age Group to Health and Wellbeing Measures, Agincourt HDSS and SAGE 2006 Logistic regression odds ratios and (95% confidence intervals)1
SAD WORRY UNHAPPY DISSATISFIED POOR WHODASi
POOR WHOQOL
Female 1.53*** (1.31, 1.79)
1.60*** (1.37, 1.86)
1.16 (0.99, 1.37)
1.13 (0.96, 1.33)
1.36** (1.09, 1.70)
1.31*** (1.03, 1.39)
Age Groups 50-‐54 (omitted) 1 1 1 1 1 1
55-‐59 0.96
(0.77, 1.20) 1.09
(0.88, 1.36) 0.95
(0.75, 1.19) 1.03
(0.82, 1.30) 1.21
(0.96, 1.52) 1.01
(0.81, 1.26)
60-‐64 1.51*
(1.00, 2.29) 1.24
(0.82, 1.87) 1.40
(0.91, 2.15) 1.52*
(0.99, 2.32) 1.31
(0.83, 2.08) 1.40
(0.92, 2.12)
65-‐69 1.09
(0.88, 1.36) 1.06
(0.85, 1.31) 0.92
(0.73, 1.16) 0.95
(0.76, 1.20) 1.45** (1.15, 1.82)
0.95 (0.76, 1.19)
70-‐74 1.16
(0.92, 1.46) 1.13
(0.90, 1.43) 1.06
(0.83, 1.35) 1.18
(0.92, 1.49) 1.40
(0.89, 2.21) 1.19
(0.95, 1.50)
75plus 1.30**
(1.07, 1.58) 1.29**
(1.07, 1.58) 1.28*
(1.05, 1.57) 1.30**
(1.06, 1.59) 2.05*** (1.45, 2.88)
1.45*** (1.19, 1.77)
Female*Age 60-‐64 0.50**
(0.32, 0.77) 0.54**
(0.35, 0.84) 0.47**
(0.30, 0.75) 0.42***
(0.27, 0.67) 0.69
(0.42, 1.13) 0.47***
(0.30, 0.73)
Female*Age 75-‐plus -‐-‐ -‐-‐ -‐-‐ -‐-‐ 1.60**
(1.11, 2.29) -‐-‐ Tests run on above models to assess differences between specific age/sex groups Improvement at pension-‐eligibility Women 60-‐64 vs 55-‐59 * *** ** *** * *** Men 65-‐69 vs 60-‐64 NS NS * ** NS * Decline in next age group after pension-‐eligibility Women 65-‐69 vs 60-‐64 ** *** ** ** *** ** Men 70-‐74 vs 65-‐69 NS NS NS * NS ** N 4060 4016 4021 4059 4069 4065
1 Models were first estimated with all sex*age-‐group interactions. Interactions that were jointly non-‐significant were deleted and the models re-‐estimated. NS p>.05, *p<=.05, **p<=.01, ***p<=.001
Table 3B. Relationships of Sex and 5-‐year Age Group to Health and Wellbeing Measures, Agincourt HDSS and SAGE 2010 Logistic regression odds ratios and (95% confidence intervals)1
SAD WORRY UNHAPPY DISSATISFIED POOR
WHODASi POOR
WHOQOL
Female 1.39***
(1.23 -‐ 1.57) 1.35***
(1.20 -‐ 1.53) 1.18*
(1.02 -‐ 1.36) 1.21**
(1.05 -‐ 1.40) 1.47***
(1.29 -‐ 1.67) 1.30***
(1.14 -‐ 1.47) Age Groups 50-‐54 (omitted) -‐ -‐ -‐ -‐ -‐ -‐
55-‐59 1.02
(0.85 -‐ 1.22) 0.96
(0.80 -‐ 1.14) 0.99
(0.81 -‐ 1.21) 1.06
(0.85 -‐ 1.32) 1.01
(0.83 -‐ 1.23) 1.10
(0.92 -‐ 1.33)
60-‐64 0.90
(0.75 -‐ 1.09) 0.77**
(0.64 -‐ 0.93) 0.88
(0.72 -‐ 1.08) 0.87
(0.69 -‐ 1.10) 1.01
(0.83 -‐ 1.23) 0.77**
(0.64 -‐ 0.94)
65-‐69 0.91
(0.75 -‐ 1.11) 0.81*
(0.67 -‐ 0.97) 0.65*
(0.44 -‐ 0.96) 1.07
(0.85 -‐ 1.34) 1.26*
(1.03 -‐ 1.53) 0.85
(0.70 -‐ 1.03)
70-‐74 1.31**
(1.08 -‐ 1.58) 1.12
(0.93 -‐ 1.36) 1.08
(0.88 -‐ 1.33) 1.17
(0.93 -‐ 1.47 1.82***
(1.49 -‐ 2.21) 1.08
(0.89 -‐ 1.32)
75plus 1.41***
(1.20 -‐ 1.67) 1.24**
(1.06 -‐ 1.46) 1.40***
(1.17 -‐ 1.67 1.77***
(1.46 -‐ 2.14) 2.95***
(2.48 -‐ 3.50) 1.74***
(1.48 -‐ 2.06)
Female*Age 65-‐69 -‐ -‐ 1.60*
(1.06 -‐ 2.41) -‐ Tests run on above models to assess differences between specific age/sex groups Improvement at pension-‐eligibility 60-‐64 vs 55-‐59 NS * NS NS NS *** Decline in next age group after pension-‐eligibility 65-‐69 vs 60-‐64 NS NS NS NS * NS 70-‐74 vs 65-‐69 NS *** NS NS *** * N 5,970 5,950 5,977 5,961 5,977 5,977 1 Some models omit certain variables or interaction terms through listwise deletion. Each included variables lists its odds ratios and confidence interval. *p<=.05 **p<=.01, ***p<=.001
Table 4. Percent of pension receipt by Sex and 5-‐Year Age Group, Agincourt HDSS and SAGE 2010 Men Women 50 to 54 3.3 2.2 55 to 59 11.5 10.7 60 to 64 71.7 78.8 65 to 69 84.4 84.6 70 to 74 88.1 86.6 75 plus 83.2 86.1 Total (50+) 59.5 57.1 Pension Eligible (60+) 81.7 84.2 N 1,522 4,455
Table 5. Relationships of Pension receipt Sex and 5-‐year Age Group to Health and Wellbeing Measures, Agincourt HDSS and SAGE 2010 Logistic regression odds ratios and (95% confidence intervals)1
SAD WORRY UNHAPPY DISSATISFIED POOR
WHODASi POOR
WHOQOL
Female 1.39***
(1.23 -‐ 1.58) 1.36***
(1.20 -‐ 1.53) 1.19*
(1.03 -‐ 1.37 1.22**
(1.05 -‐ 1.41) 1.47***
(1.30 -‐ 1.67) 1.31***
(1.15 -‐ 1.48) Age Groups 50-‐54 (omitted) -‐ -‐ -‐ -‐ -‐ -‐
55-‐59 1.02
(0.85 -‐ 1.22) 0.96
(0.80 -‐ 1.14) 0.99
(0.81 -‐ 1.21) 1.06
(0.85 -‐ 1.32) 1.01
(0.83 -‐ 1.23) 1.10
(0.92 -‐ 1.33)
60-‐64 0.88
(0.73 -‐ 1.07) 0.75**
(0.62 -‐ 0.90) 0.79*
(0.64 -‐ 0.97) 0.83
(0.65 -‐ 1.05) 0.97
(0.80 -‐ 1.19) 0.70***
(0.57 -‐ 0.85)
65-‐69 0.90
(0.74 -‐ 1.09) 0.79*
(0.65 -‐ 0.95) 0.61*
(0.41 -‐ 0.90) 1.03
(0.82 -‐ 1.30) 1.22*
(1.00 -‐ 1.50) 0.79*
(0.65 -‐ 0.97)
70-‐74 1.29**
(1.07 -‐ 1.56) 1.10
(0.91 -‐ 1.33) 1.02
(0.82 -‐ 1.25) 1.14
(0.91 -‐ 1.44) 1.78***
(1.46 -‐ 2.16) 1.03
(0.84 -‐ 1.25)
75plus 1.40***
(1.18 -‐ 1.65) 1.21*
(1.03 -‐ 1.43) 1.30**
(1.09 -‐ 1.56) 1.71***
(1.41 -‐ 2.08) 2.88***
(2.42 -‐ 3.42) 1.64***
(1.39 -‐ 1.94)
No Pension 1.10
(0.92 -‐ 1.31) 1.17
(0.98 -‐ 1.39) 1.57***
(1.31 -‐ 1.88) 1.23*
(1.01 -‐ 1.49) 1.18
(0.99 -‐ 1.41) 1.50***
(1.26 -‐ 1.79)
Female*Age 65-‐69 -‐ -‐ 1.59*
(1.05 -‐ 2.40) -‐ -‐ -‐ Tests run on above models to assess differences between specific age/sex groups Improvement at pension-‐eligibility 60-‐64 vs 55-‐59 NS ** * * NS *** Decline in next age group after pension-‐eligibility 65-‐69 vs 60-‐64 NS NS NS NS * NS 70-‐74 vs 65-‐69 *** ** ** NS *** ** N 5,970 5,950 5,977 5,961 5,977
1 Some models omit certain variables or interaction terms through listwise deletion. Each included variables lists its odds ratios and confidence interval. *p<=.05 **p<=.01, ***p<=.001