LINKS BETWEEN POVERTY AND HIV/AIDS

UNCLAS SECTION 01 OF 02 PRETORIA 004995

SIPDIS

STATE PASS TO AID WASH DC

SIPDIS

DEPT FOR AF/S; AF/EPS; AF/EPS/SDRIANO
DEPT FOR S/OFFICE OF GLOBAL AIDS COORDINATOR
STATE PLEASE PASS TO USAID FOR GLOBAL BUREAU KHILL
USAID ALSO FOR GH/OHA/CCARRINO AND RROGERS, AFR/SD/DOTT
ALSO FOR AA/EGAT SIMMONS, AA/DCHA WINTER
HHS FOR THE OFFICE OF THE SECRETARY/WSTEIGER, NIH/HFRANCIS
CDC FOR SBLOUNT AND DBIRX

E.O. 12958: N/A
TAGS: ECON KHIV SOCI TBIO EAID SF
SUBJECT: LINKS BETWEEN POVERTY AND HIV/AIDS

Summary
--------------


1. Summary. In December 2005, an International Union for the
Scientific Study of Population (IUSSP) conference in Cape Town
presented demographic studies that highlighted the interactions
between poverty and HIV/AIDS in Southern African nations of
Malawi, Zambia, and South Africa. No direct evidence was shown
that poverty causes HIV/AIDS but there were strong correlations
and associations between the two. Because most of the new
information was coming from surveying the same people over time
(panel surveys),much discussion centered around the problems
and interpretations of using this type of data for empirical
investigations. Eight of the 12 studies used Demographic
Health Surveys, household-based national surveys having little
detailed information concerning income. These studies had to
impute assets using either type of flooring, housing or other
asset information having a presumed correlation with income,
making the analysis of the interaction of poverty and HIV/AIDS
subject to possible measurement and specification errors. The
South African studies used local surveys (in Free State and
KZN) trying to determine the impact of socio economic status on
orphans, antiretroviral treatment, and HIV affected households.
Most studies showed relationships between poverty and HIV/AIDS,
but no clear cut causation. End Summary.

Poverty and HIV/AIDS
--------------


2. Prominent African politicians and researchers have long
posited causality between poverty and HIV/AIDS, suggesting that
increased poverty causes high HIV/AIDS prevalence. In
December, The International Union of Scientific Study of
Population (IUSSP) sponsored a conference in Cape Town that
presented 12 papers addressing the impacts of poverty and
HIV/AIDS, focusing on income effects on topics ranging from
orphans to receiving antiretroviral treatment. The conference
organizers hoped to provide empirical evidence on the
hypothesis that HIV/AIDS' main solution lies in eradicating
poverty.


3. The demographic studies focused on Southern African
countries and used a variety of household panel surveys that
were not designed for the specific study of HIV/AIDS
interactions. Several studies used Demographic Health Surveys,
(nationally representative household surveys collected every 5
years in most developing countries),or national and provincial
surveys. Focus countries included Malawi, Zambia, South
Africa, Kenya, Cambodia, Thailand and Tanzania. All studies
used several waves of panel data, trying to discover long run
impacts in order to highlight poverty's impact on various risk
behaviors associated with higher HIV prevalence.


4. Four studies concentrated on South Africa, using provincial
surveys, making generalized national observations difficult.
The South African studies examined the socio-economic impacts
of HIV/AIDS on household in the Free State; impacts of parental
death on school enrollment in KZN; orphans and HIV risk
behaviors among adolescents in KZN; and socioeconomic status as
determinants in treatment outcomes in the Free State. One Free
State study (primary researcher, Sebastian Linnemayr from Ecole
Normale Superieure, using data collected in a USAID-funded
study at the University of the Free State) grouped HIV affected
and non-affected by amount of liquid and illiquid assets and
found the assets to be similar among groups, although since 40%
of people had no income in both groups, one could argue that
poverty impacted the results. The study of the impacts of
parental death on school achievement in KZN (primary
researcher, Anne Case from Princeton University) found that
that there was no link between socio economic status (SES) if
the mother died, and a negative association if the father died;
however children without mothers are behind in school relative
to other children. The study focusing on the orphanhood,
poverty and HIV risk behaviors in KZN (prime researcher Kelly
Hallman from the Population Council) found that orphans did
begin sexual relations earlier than non orphans and differing
income effects by gender, with girls in households with higher
income having lower chances of early sexual debut while boys
had higher chances of multiple partners. Frikkee Booysen's

PRETORIA 00004995 002 OF 002


study (of the University of the Free State) found that only the
well-off felt that they were getting the benefit of anti-
retroviral treatment, having been on treatment and knowing
their status longer.


5. The South African studies used different measures of
poverty. Hallman's study used consumption and household asset-
based measures, Case's study used expenditures, assets, and
access to piped water and electricity as indicators of wealth.
Booysen's and Linnemayr's studies used the same Free State
panel data which collected income, asset and access to services
information. Of the four studies, two found that income did
not explain differences in HIV-impacted households or orphans
and two found income to be an important determinant.


6. Studies that focused on other countries in Africa also
found that income's impacts varied. A Malawian study focused
on the impact of HIV/AIDS on Intra-household Time Allocation
and concluded that HIV/AIDS had no impact on men's allocation
of time and it caused women to diversify income sources. In a
Zambian (focusing on prime age mortality, rather than death by
HIV/AIDS) study, women's prime age mortality was not affected
by income; prime age mortality is more likely to affect wealthy
men.


7. All presented studies were preliminary and discussion
focused on inherent problems using panel data. Panel data
gives temporal explanation but no causality. Selectivity
(caused by attrition in survey respondents) and measurement
error biases are present. Omitted variables and unobserved
fixed effects also presents empirical challenges. One
agreement came from the conference: further research is needed
on the links between poverty and HIV/AIDS.

TEITELBAUM