Front Cover
 Title Page
 Table of Contents
 List of tables and illustratio...
 HIV seroprevalence and the reported...
 The impact of AIDS on populati...
 AIDS and economic growth linka...
 Looking forward: HIV/AIDS and...
 Appendix 1: Background information...
 Appendix 2: HIV seroprevalance...
 Appendix 3: Description of the...
 Back Cover
 Reprint permission notice

Group Title: Food, agriculture, and the environment discussion paper - International Food Policy Research Institute ; no. 15
Title: The potential impact of AIDS on population and economic growth rates
Full Citation
Permanent Link: http://ufdc.ufl.edu/UF00085352/00001
 Material Information
Title: The potential impact of AIDS on population and economic growth rates
Series Title: Food, agriculture, and the environment discussion paper
Physical Description: vi, 32 p. : ill., maps ; 28 cm.
Language: English
Creator: Brown, Lynn R
International Food Policy Research Institute
Publisher: International Food Policy Research Institute
Place of Publication: Washington D.C
Publication Date: 1996
Subject: Economic development -- Effect of AIDS (Disease) on   ( lcsh )
Population -- Effect of AIDS (Disease) on   ( lcsh )
AIDS (Disease) -- Africa, Sub-Saharan   ( lcsh )
Genre: bibliography   ( marcgt )
non-fiction   ( marcgt )
Bibliography: Includes bibliographical references (p. 29-32).
Statement of Responsibility: Lynn R. Brown.
General Note: "June 1996."
General Note: "2020 vision"--Cover.
 Record Information
Bibliographic ID: UF00085352
Volume ID: VID00001
Source Institution: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
Resource Identifier: oclc - 35618979

Table of Contents
    Front Cover
        Front Cover 1
        Front Cover 2
    Title Page
        Page i
        Page ii
    Table of Contents
        Page iii
    List of tables and illustrations
        Page iv
        Page v
        Page vi
    HIV seroprevalence and the reported incidence of AIDS
        Page 1
        Page 2
        Page 3
    The impact of AIDS on population
        Page 4
        Page 5
        Page 6
        Page 7
        Page 8
        Page 9
        Page 10
        Page 11
    AIDS and economic growth linkages
        Page 12
        Page 13
        Page 14
        Page 15
    Looking forward: HIV/AIDS and food, agriculture, and the environment
        Page 16
        Page 17
        Page 18
        Page 19
    Appendix 1: Background information about HIV/AIDS
        Page 20
        Page 21
    Appendix 2: HIV seroprevalance in Africa
        Page 22
        Page 23
        Page 24
        Page 25
    Appendix 3: Description of the models
        Page 26
        Page 27
        Page 28
        Page 29
        Page 30
        Page 31
        Page 32
    Back Cover
        Page 33
        Page 34
    Reprint permission notice
        Page 35
Full Text

The Potential Impact of
AIDS on Population and
Economic Growth Rates

Lynn R. Brown

2 20

"A 2020 Vision for Food, Agriculture, and the Environment" is an initiative of
the International Food Policy Research Institute (IFPRI) to develop a shared
vision and a consensus for action on how to meet future world food needs
while reducing poverty and protecting the environment. It grew out of a
concern that the international community is setting priorities for addressing
these problems based on incomplete information. Through the 2020 Vision
initiative, IFPRI is bringing together divergent schools of thought on these
issues, generating research, and identifying recommendations.

This discussion paper series presents technical research results that encom-
pass a wide range of subjects drawn from research on policy-relevant
aspects of agriculture, poverty, nutrition, and the environment. The discus-
sion papers contain material that IFPRI believes is of key interest to those
involved in addressing emerging Third World food and development prob-
lems. These discussion papers undergo review but typically do not present
final research results and should be considered as works in progress.

Food, Agriculture, and the Environment Discussion Paper 15

The Potential Impact of

AIDS on Population and

Economic Growth Rates

Lynn R. Brown

International Food Policy Research Institute
1200 Seventeenth Street, N.W.
Washington, D.C. 20036-3006 U.S.A.
June 1996


Foreword v
HIV Seroprevalence and the Reported Incidence of AIDS 1
The Impact of AIDS on Population 4
AIDS and Economic Growth Linkages 12
Looking Forward: HIV/AIDS and Food, Agriculture,
and the Environment 16
Appendix 1: Background Information about HIV/AIDS 20
Appendix 2: HIV Seroprevalence in Africa 22
Appendix 3: Description of the Models 26
References 29


1. Population forecasts, incorporating AIDS-related mortality, to 2020, for
13 African countries 9
2. Population growth rates, based on population forecasts incorporating
AIDS-related mortality, 1990-2020 10
3. Life expectancy at birth, incorporating AIDS-related mortality 11
4. Estimates of the direct cost of AIDS, per case, in selected countries 13
5. Estimates of the indirect costs per case of HIV infection in Zaire and
Tanzania 14
6. Net trade balance for regions of Africa and major food crops, 2020 18


1. HIV 1 seroprevalence among pregnant women in urban and rural areas
of four countries 2
2. HIV 1 seroprevalence among prostitutes in five cities 2
3. African HIV1 seroprevalence for high-risk urban populations 22
4. African HIV1 seroprevalence for low-risk urban populations 23
5. African HIV2 seroprevalence for high-risk urban populations 24
6. African HIV2 seroprevalence for low-risk urban populations 25


IFPRI would be remiss if it did not look at the long-term effects of the AIDS pandemic as part
of its 2020 Vision initiative, which seeks to develop an international consensus on how to
meet future world food needs while reducing poverty and protecting the environment. AIDS
has already become the leading cause of death among people between the ages of 15 and
39 years in at least half-a-dozen Sub-Saharan African countries, and the disease is spreading
rapidly in other parts of the developing world, particularly South and Southeast Asia.
Although AIDS infects the rich and poor alike, the disease is especially devastating to the
poor because they lose their only source of livelihood-their labor-when they become ill.
At the same time, they face rising health care expenses. In rural areas, the disease could affect
farmers' choice of crops: for example, as labor shortages become severe, they may switch
from labor-intensive tradable crops such as maize to lower-value crops such as cassava,
which has repercussions for the country's gross domestic product as well as the nutritional
status of its people.
In this paper, Lynn Brown examines the current status of HIV/AIDS infection, particu-
larly in Sub-Saharan Africa, and reviews existing models that look at the future impact that
the disease is likely to have on population growth, economic growth, and food security,
especially as it spreads to rural areas. Because exploding population growth is considered by
many to be the number one problem facing developing countries in the year 2020, Brown
confronts the question of whether population growth kept in cheek by AIDS mortality might
lead to greater availability of food in 2020 than would be possible in a world without AIDS.
This paper serves as a grim reminder that every effort must be made to slow this
pandemic in its early stages. Its global effects are so profound that we will all feel them by
2020. We cannot afford to leave it to the health sectors of poor countries to combat this
menace on their own. It is a development problem of severe human and economic conse-
quences, requiring national and international attention and resources.

Per Pinstrup-Andersen
Director General, IFPRI


The author would like to thank Peter Way, International Programs Center, U.S. Bureau of the
Census, for his help and encouragement. The author is grateful to him for supplying the U.S.
Bureau of the Census forecasts, as well as for the "special run" of the bureau's model,
implemented at the author's request, which produced the 2020* forecasts. The author would
also like to thank Per Pinstrup-Andersen, Lawrence Haddad, and an anonymous reviewer for
useful comments. Any errors and omissions are the sole responsibility of the author.

By the mid-1990s, acquired immune deficiency
syndrome (AIDS) had surpassed both measles
and malaria to become the second leading cause of
child mortality in Sub-Saharan Africa (United Nations
1994). It is believed that the extensive spread of
human immunodeficiency virus (HIV), the etiologic
agent that causes AIDS, began during the mid- to late
1970s and early 1980s. In July 1994, the World Health
Organization (WHO 1994,1) estimated that the world-
wide cumulative figures for those infected with HIV
were 16 million adults and more than 1 million children,
the majority of whom lived in Sub-Saharan Africa.
The likely result of these infections will be 16 million
adult deaths, manifested in the next 5 to 10 years, and
1 million children who will not see their fifth birthday.
The spread of HIV continues unabated among many
already- infected populations and into areas previously
unaffected. In mid-1993, Asia had just 1 percent of the
AIDS cases worldwide, but just 12 months later it had
6 percent of the global total--an outcome largely driven
by the rapid growth of AIDS in South and East Asia
(U.S. Bureau of the Census 1994b). In almost all cases
those infected by HIV develop AIDS, which is
inevitably fatal. AIDS is believed to be the leading cause
of mortality between the ages of 15 and 39 in Botswana,
Malawi, Uganda, Zambia, and Zimbabwe (World Bank
1995). In Sub-Saharan Africa in 1988, the leading causes
of death in children under five years of age were
diarrhea, measles, malaria, and AIDS.
There is no doubt that HIV and AIDS represent
a human tragedy and a major health problem, but do
they represent more than that? What ramifications
will HIV/AIDS have on the future development path

of many of the world's poorest economies? This
paper concentrates on the macro population and eco-
nomic growth implications of the HIV/AIDS pan-
demic currently devastating several African econo-
mies and seemingly poised to devastate parts of
Asia. The paper first outlines the current state of the
HIV/AIDS pandemic. It then examines the popula-
tion forecasts, incorporating AIDS-related mortality,
of three major agencies: the United Nations, the
World Bank, and the U.S. Bureau of the Census.
Next it explores the cost of AIDS and the potential
effects of AIDS on macroeconomic growth. In con-
clusion, it looks at the implications of HIV and AIDS
for 2020 and attempts to formulate a vision for food,
agriculture, and the environment.

HIV Seroprevalence and the
Reported Incidence of AIDS
To date, the most devastating effects of the HIV/AIDS
pandemic have been concentrated in Sub-Saharan
Africa.' The epidemic began as early as the 1960s
but spread rapidly in the mid- to late 1970s in the
worst-affected countries. HIV was first identified in
1983. Two serotypes of the virus are currently recog-
nized: HIV1 and HIV2. Most HIV transmission in
Sub-Saharan Africa is through heterosexual inter-
course (see Appendix 1). The four maps in Appendix
2, based on data compiled by the U.S. Bureau of the
Census (1994a), show the estimated seroprevalence
levels of HIV and HIV2 for high-risk and low-risk
urban populations in Sub-Saharan Africa.2 Gener-

'Statistics relating to HIV seroprevalence levels reported in this section, unless otherwise stated, are drawn from the database compiled
and maintained by the U.S. Bureau of the Census. Some samples are small, and a variety of testing mechanisms for the presence of
HIV antibodies were used across samples. Where possible, the most valid statistic has been chosen, in terms of sample size or other
2The term "high-risk populations" refers to seroprevalence data from samples of intravenous drug users (IVDUs), prostitutes, sexually
transmitted disease clinic patients, and others with known risk factors, which for some populations may include members of the
military or truck drivers. Low-risk population seroprevalence data arise from samples of pregnant women, blood donors, and general
population groups who are not believed to demonstrate particular risk factors.

ally, the prevalence of HIV1 in high-risk urba
population groups is highest (above 40 percent ofth
adult population) in a belt of countries running
through East and Central Africa-Ethiopia, Ugandc
Kenya, Rwanda, Burundi, Tanzania, Zambia, an
Malawi. In the same high-risk population group,
HIV1 seroprevalence rates above 40 percent are als
found in West Africa in Cote d'Ivoire, Cameroor
and Mali. A slightly different pattern appears among
low-risk urban populations. The highest HIV1 sere
prevalence levels, over 10 percent of the adult popu
lation, are still concentrated in the belt of East an
Central African countries, excluding Ethiopia. I
this continuous belt, Zimbabwe and Botswana d
not demonstrate HIV1 seroprevalence levels abov
40 percent among their high-risk populations bu
have HIV1 seroprevalence levels among low-ris
populations above 10 percent. Figures 1 and 2 shox
figures for HIV1 seroprevalence for selected coun
tries and cities in Sub-Saharan Africa. Although
seroprevalence levels are generally higher in urba
areas, they are not insignificant in rural areas. Eve
where rates of HIV1 seroprevalence are relatively
low, complacency is not in order because rates o
increase in seroprevalence are often high. In Addi

Figure 1-HIV1 seroprevalence among
pregnant women in urban and
rural areas of four countries





0 5 10 15 20
SUrban 0 Rural

Source: U.S. Bureau of the Census 1994d.

25 30


Figure 2-HIV1 seroprevalence among
prostitutes in five cities


Addis Ababa


Dares Salaam


0 20 40 60 80

f Source: U.S. Bureau of the Census 1994d.

Ababa, the HIV seroprevalence level among prosti-
tutes was just 0.6 percent in 1985 but had risen to
54.2 percent in 1990. This rapid growth rate in sero-
prevalence is not just a feature of high-risk popula-
tions. In Blantyre, Malawi, HIV1 seroprevalence
levels among pregnant women rose from 2 percent in
1985 to 33 percent in 1994.
HIV2 infection is largely concentrated in West
Africa. The rates for HIV2 are lower than those for
HIV1; the highest rates are among high-risk urban
populations. More than 12.5 percent of the adult
population in Angola, C8te d'Ivoire, and Mali is
infected. Among urban low-risk populations, only
Guinea Bissau exhibits an HIV2 seroprevalence rate
above 10 percent.
The AIDS epidemic is now beginning to take
hold in Asia. The initial transmission mechanism for
HIV in Asia was through intravenous drug use
(IVDU) but is now largely through heterosexual
intercourse, as in Africa. The epidemic did not begin
to spread dramatically until the mid-1980s, but the
rate of spread is now gaining momentum. Between
5 1994 and 1995, there was an eightfold increase in the
number of AIDS cases, from 30,000 to 250,000, in
South and East Asia. Data for many areas of Asia are
scarce, but experts believe that by the year 2000
most new HIV infections will occur in Asia (Chin



( I I I

1991). This prediction is already being borne out:
new AIDS cases in Asia in 1990 constituted just
0.003 percent of the total new cases reported to
WHO. By 1993, nearly 4 percent of newly reported
AIDS cases were in Asia. These rates are largely being
driven by trends in India, Myanmar, and Thailand.
Newly reported AIDS cases quadrupled in Thailand,
tripled in Myanmar, and almost doubled in India
between 1992 and 1993. In Churachandpur, India,
80 percent of intravenous (IV) drug users are HIV-
positive. HIV seroprevalence among sexually trans-
mitted disease clinic patients in Bombay, India, is
rising alarmingly from virtually nothing in 1987-88
to 24 percent in 1994. HIV seroprevalence levels
among prostitutes in Vellore, India, rose from
0.5 precent in 1986 to 34.5 percent in 1990 (Lal and
Thakur 1995, 834). Many experts believe that if the
Indian epidemic continues on its present course, the
consequences could be disastrous.
Epidemics are becoming apparent in several
other Asian economies, in both high-risk and low-
risk population groups. In 1992, HIV1 seropreva-
lence levels of 9.2 percent were documented among
prostitutes in Cambodia. Increasing HIV seropreva-
lence rates have also been shown among the same
population group in Laos, Malaysia, and the Philip-
pines. Not even China has escaped. In Yunnan Prov-
ince, which borders Myanmar and Laos, HIV1 sero-
prevalence among IV drug users in Ruili and
Longchuan was 82 percent and 45 percent, respec-
tively. Between 1988 and 1991, HIV1 seropreva-
lence rates among drug users in Bangkok rose from
1.2 percent to 45.0 percent.
HIV seroprevalence rates above 4 percent have
been found among pregnant women in the northern
region of Thailand and among women giving birth
but receiving no antenatal care in Bangkok. HIV
seroprevalence rates among blood donors in Gujarat,
India, were nearly 6 percent in 1994. Among blood
donors in Cambodia, HIV seroprevalence rates rose
from 0.1 percent in 1991 to 3.5 percent in 1994.
Initially, HIV transmission in Latin America
occurred primarily through homosexual and bisexual
intercourse and IV drug use-a pattern similar to that
seen in the United States and much of Europe. Brazil
accounts for nearly 11 percent of all AIDS cases
reported in the Americas (55,894 out of 526,682),
including the United States. Mexico follows with

20,077 reported cases, or nearly 4 percent. In 1993 the
number of reported AIDS cases per 1,000 population,
however, was highest in some of the Caribbean coun-
tries: the Bahamas reported 121; Barbados, 33; and
Bermuda, 27. The Turks and Caicos Islands had the
highest rate in the Americas, with 140 AIDS cases
reported per 1,000 population in 1993.3
In Santo Domingo, Dominican Republic, HIV
seroprevalence rates among prostitutes increased
from 1 percent in 1986 to 11 percent in 1993. The
seroprevalence rates among pregnant women rose
from 0.8 percent in 1991 to 1.2 percent in 1993.
Similarly, seroprevalence rates increased 10-fold to
1 percent among blood donors in 12 months from
1992 to 1993. Of concern in Haiti is the young age at
which women become HIV-positive. In a study of
pregnant women, the highest seroprevalence rate was
found to be 13 percent among girls aged 14 to 16.
AIDS is often thought of as an urban-centered
disease. This may be somewhat of a misconception.
In Uganda, AIDS is believed to have started in a
fishing village in 1982. Links to other towns, and
through those towns to Kampala, resulted in the
spread of the disease to the capital (Bold and Vincent
1990). High population densities in urban areas,
often combined with different behavioral and cul-
tural practices, led to a faster spread of HIV there.
The rural nature of many Sub-Saharan African na-
tions, however, means that in absolute numbers
AIDS cases in rural areas predominate. The rate of
the spread of AIDS into rural areas is hard to meas-
ure because of the lack of data on HIV1 seropreva-
lence. This is a result of the lower levels of develop-
ment in the health infrastructure and of the lower
utilization of health facilities of rural populations.
Many of the seroprevalence rates documented for
urban populations are obtained through blood testing
of individuals who contacted health facilities for
other purposes, such as pregnant women at antenatal
clinics and hospitals, patients at sexually transmitted
disease clinics, and blood donors. An absence of
seroprevalence data should not, however, be taken as
indicative of low seroprevalence levels.
Becker (1990) perceived two types of rural areas
with high vulnerability to AIDS. The first and most
vulnerable are rural areas on truck routes, and the
second are rural areas that are sources of migrant
labor to urban areas. While nearly all truck drivers in

3Statistics in this paragraph were obtained from the WHO/Global Program on AIDS World Wide Web site.

Niger had heard of condoms and reported frequent
changes in sexual partners, only 14 percent used
condoms regularly (World Bank 1995). Certainly,
the spread of HIV seroprevalence along trade routes
in Africa has been well documented, a factor likely
to be of significance in the spread of HIV and AIDS
to the agricultural surplus regions of countries. Tra-
ditional rural areas, or subsistence agricultural areas,
have been thought to be less vulnerable. By the
Becker classification, however, there may be few
rural areas that are not at risk, given that most subsis-
tence agricultural regions are also sources of migrant
labor in the agricultural lean season. Available evi-
dence for rural areas shows that HIV seroprevalence
levels are rising.
Zimbabwe, with its well-developed infrastruc-
ture facilitating good urban-rural market links and
population movements, demonstrates how AIDS can
spread into rural areas. A 1991 study, based in the
rural area of Karoi District, found that HIV sero-
prevalence levels among patients with sexually
transmitted diseases were notably high, at 65 percent
for women and 61 percent for men. In 1993, HIV
seroprevalence levels of 1.8 percent were docu-
mented in the general adult population (aged 15 to
49) of six rural areas in Zimbabwe. In two years, the
virus had spread from a high-risk population to the
general population in this rural area.
In Tanzania the HIV seroprevalence rate in the
more rural Bukoba District is higher than that in Dar
es Salaam (Way and Stanecki 1993). Within Bukoba
District, however, the urban seroprevalence rate is
24.2 percent and the rural rate is 10.0 percent (Mtera
1992, 89). In rural areas of Mwanza region in Tanza-
nia, HIV seroprevalence levels in the adult population
(aged 25 to 34) were 5.8 percent for men and 6.3 per-
cent for women. In the Mbeya region of Tanzania,
HIV seroprevalence rates among pregnant women in
rural areas tripled between 1988 and 1992, when they
reached 10.2 percent. In Malawi, HIV seroprevalence
levels among pregnant women attending village clinics
in the Mangochi District were over 5 percent. In two
rural towns in the same district, the HIV seroprevalence
levels among pregnant women attending antenatal clin-
ics exceeded 11 percent. Similarly, HIV seroprevalence
rates of 13.2 percent were documented in rural areas of
Zambia in 1992 among pregnant women (Fylkesnes,
Brunborg, and Msiska 1994, 12).
Many of the statistics cited in the preceding
section may seem insignificant in absolute magni-
tude, but as has been demonstrated in Sub-Saharan
Africa, low recorded HIV seroprevalence levels in-
crease rapidly. Many Sub-Saharan African countries

are already experiencing HIV and AIDS epidemics
on a massive scale, and all signs indicate that several
countries in Asia may be set to follow the same path,
if not a more difficult one, if they fail to address the
problem in the early stages. The increasing integra-
tion of economies and high levels of population
movement, both within and between countries, ensure
that, in time, virtually no country will be untouched
by AIDS. These population movements may be par-
ticularly significant in Asia, where many low-wage
economies, such as the Philippines, export their
labor to high-wage economies, such as Singapore.
Although much is still unknown about how rapidly
HIV spreads through any particular country or at
what level HIV seroprevalence levels will peak,
enough is known to be certain that HIV/AIDS is
more than a public health problem. Even if no new
HIV infections were introduced from today onward,
mortality rates would continue to rise because of the
long incubation period.

The Impact of AIDS
on Population
Recent years have seen a growing awareness of the
magnitude and severity of the AIDS epidemic in
some countries, notably the worst-affected countries
of Sub-Saharan Africa. This recognition has led to
an increasing number of population forecasts that
take AIDS-related mortality into consideration. In
1993 the United Nations published World Popula-
tion Prospects: The 1992 Revisions, the first com-
prehensive set of world population forecasts to in-
corporate AIDS-related mortality. These estimates
covered 15 African countries in which WHO figures
indicated that over 1 percent of the adult population
(those older than 15 years of age) was HIV-positive.
The 1994-95 edition of population forecasts pre-
pared by the World Bank also recognizes the effects
of HIV/AIDS on population growth (Bos et al. 1994).
AIDS-related mortality is incorporated for all coun-
tries where there is a measurable level of infection.
The incorporation of AIDS-related mortality into
population forecasts, however, is no simple task,
given the paucity of nationally representative and
accurate HIV seroprevalence data. In addition, a poor
understanding of the actual amount of time it takes
the HIV infection to progress to AIDS and the paucity
of data relating to behavioral change as a result of HIV
infection increase the difficulty of forecasting the
future spread of HIV infection for many countries.

Population Forecasting:
A Review of the Models
This section considers the three key models that
incorporate AIDS-related mortality into population
forecasts: (1) an Epi model (to predict HIV inci-
dence), used in conjunction with ABACUS (a popu-
lation and demographic forecasting model), by
WHO/United Nations; (2) an HIV/AIDS forecasting
model developed by Bulatao of the World Bank,
used in conjunction with a cohort-component model
of population forecasting; and (3) the IWGAIDS
model developed by the Interagency Working Group
on AIDS Models and Methods of the U.S. Depart-
ment of State and used by the U.S. Bureau of the
Census for forecasting. The first two models follow
a similar process, projecting the path of the
HIV/AIDS epidemic and incorporating AIDS-
related mortality into the life-span tables to make
population projections. The third model adopts a
more complex and integrated approach, using a
deterministic system of differential equations to
model the spread of the epidemic. The WHO/United
Nations projection model requires only three HIV
seroprevalence data points: the current-year estimate
of HIV prevalence, the first year transmission of
HIV became widespread, and the annual rate of pro-
gression from HIV infection to AIDS. The World
Bank and IWGAIDS models are based on equations
that use both data and behavioral assumptions to
simulate the path of the HIV epidemic. Appendix 3
provides further details about the three models.
Many of the underlying behavioral parameters
used in forecasting the path of the AIDS epidemic,
and consequently the mortality pattern due to AIDS,
are weak or unknown. Examples are the number of
sexual partners an individual has had and the preva-
lence of sexually transmitted diseases. Assumptions
often have to be based on small, unrepresentative
population subsamples.
A key assumption for all the models is the point
at which HIV transmission changes trajectory or
reaches its peak. The UN forecasts assume that the
epidemic follows an unabated course such that the
peak will occur 12 years after the start of the epi-
demic, or by the year 2005. It assumes that there will
be no adult transmission of HIV after 2005 but that
perinatal transmission will continue because the

HIV infection will still be widespread in women in
their child-bearing years. The World Bank forecasts
assume HIV transmission will be reduced by 50 per-
cent per year each year after 2005. The IWGAIDS
forecasts assume that every country's epidemic
peaks in 2010, with no further infection after that
year. By 2020 in each of the models, all AIDS-
related deaths will have ceased, implying that mor-
tality rates will have returned to normal, pre-AIDS
levels. The assumption that AIDS will cease to be a
problem by 2020 is purely hypothetical, however,
and does not relate to known information about the
AIDS epidemic or prospects for its cure. In model-
ing, projections become weaker the further into the
future one projects, so various studies have selected
different ending years for the epidemic. At the pres-
ent time, there is no cure or prophylactic for HIV
infection. Even if either were to become available,
the cost of the necessary drugs would probably be
prohibitive to most developing countries. As yet, not
even AZT, a drug commonly used to treat AIDS in
developed nations, is readily available in most Sub-
Saharan African countries. Thus, 2010 is likely to be
overly optimistic as a termination date for new HIV
infections. Therefore, additional unpublished popu-
lation forecasts were provided by the U.S. Bureau of
the Census, at the author's request and specification,
which relax the assumption of no new HIV transmis-
sion after 2010. In this set of forecasts, AIDS-related
mortality declines gradually over the period 2010 to
2050, rather than being reduced to zero by 2020. As
is shown later, this new assumption makes a consid-
erable difference in the forecast of some demo-
graphic indicators, such as population growth rates
in the year 2020.
All models considered in this paper assume that
there will be no direct impact of AIDS on fertility
rates, and thus that fertility will follow the existing
downward trend in the majority of cases. The United
Nations is the only forecasting body considered here
that predicts population growth on the basis of dif-
fering fertility assumptions (representing high-, me-
dium-, and low-fertility transition paths).
These differing fertility assumptions, however,
are independent of any potential direct impact of
HIV.4 The long period between HIV infection, the
development of AIDS, and death means that
women's fertility rates will not necessarily be de-

4An indirect impact of AIDS will be its effect on the age structure of the population caused by changes in mortality patterns as a result
of AIDS. AIDS deaths are concentrated among those under 5 years and those who are sexually active.

creased by shortened reproductive life spans. Behav-
ioral changes induced by higher mortality rates in
both adults and children could, however, change
fertility patterns.
Several arguments support the possibility of in-
creased fertility rates:

Anecdotal evidence suggests that some men
are choosing younger brides to avoid the pos-
sibility of their brides' being HIV-infected.
Women who remain HIV-negative will have
more reproductive years than those who do
not, a circumstance that could lead to higher
fertility rates.
The likelihood that children born to HIV-
positive women will be HIV-positive them-
selves ranges from 15 to 40 percent. Higher
infant mortality rates, as a result of infant
AIDS, may reduce the length of a new
mother's postpartum amenorrhea attributable
to breast-feeding, thus increasing the rate of
childbirth in HIV-positive women.
Couples who are aware of rising mortality
rates among children and younger adults may
believe their own children are at increased risk
of dying in young adulthood. To ensure that
they will have children to care for them in old
age, couples may increase the number of chil-
dren they choose to bear.
Increasing rates of adult mortality owing to
AIDS results in an increasing number of or-
phaned children (Ainsworth and Rwegarulira
1992). In many cases, the absorption of these
children into other families will increase pov-
erty levels and decrease school attendance for
all children in the household, but particularly
the orphans. This is particularly crucial for
girls, given the well-documented negative cor-
relation between increasing education levels
and fertility rates. AIDS could, therefore, slow
down the fertility transition in those countries
most affected by the epidemic.
Other arguments suggest that fertility rates will
Among the main weapons currently used to
prevent the spread of HIV are information and
education programs, particularly the promotion
of condom use. To the degree that these pro-
grams are successful, a by-product would be
falling fertility rates. Evidence on the use of
condoms to prevent HIV transmission, how-
ever, is scant. A hospital-based study in Kin-

shasa, Zaire, tracked the contraceptive use of
238 HIV1 seropositive women and 315 HIV1
seronegative women for three years following
the birth of their last live-bor child. Seventeen
percent of the HIV1-positive women said their
partners used condoms, while only 3.2 percent
of their HIV1-negative counterparts said their
partners did. Condom use also increased with
the progression of HIV1 infection toward
AIDS. Among the HIV1-seropositive women,
34 percent of women with AIDS reported con-
dom use by sexual partners; 17 percent of those
who were experiencing AIDS-related complex
(ARC) and 15 percent of those who were
asymptomatic reported using condoms (Ryder
et al. 1991). It seems therefore that condom use
increases after infection rather than before it. Of
greater concern, however, is that while just
17 percent of HIV1-positive women reported
condom use by partners, 97 percent said they
were reluctant to inform sexual partners of their
HIV1 seropositive status for fear of reprisal,
such as divorce or violence.
Decisionmaking with regard to childbearing
may also be influenced. Partners who believe
that they will not live to an age where they will
need their children to care for them may
choose to increase current consumption rather
than have large families.

All of the forecasting models considered here
assume that there is no link between AIDS-related
mortality and mortality from other causes. Limited
evidence indicates that there may be links, however,
such as the shorter period between HIV infection and
the development of AIDS in African countries than
in the United States and Europe. U.S.-based studies
indicate a median period of 10 years from HIV infec-
tion to the development of AIDS, with one estimat-
ing this transition time to be 7 years (Taylor, Kuo,
and Detels 1991; Bacchetti and Jewell 1991; both
cited in Philipson and Posner 1993). An ongoing
study based in San Francisco produced a similar
result (cited in Ainsworth and Over 1994). A study
based in Kinshasa of HIV-positive but asympto-
matic hospital workers found that a higher percent-
age of these workers developed AIDS, or related
illnesses, in a two-year period than would be pre-
dicted using the results of the San Francisco study
(Ryder and Mgerwa 1994, cited in Ainsworth and
Over 1994). The mean time from HIV infection to
the development of AIDS has been suggested to be
as little as 5 years and as long as 10 years in Sub-

Saharan Africa (Stover 1993a, 1993b, cited in
Ainsworth and Over 1994).
One reason for the shorter time line may be that
the disease burden in the general population of Sub-
Saharan Africa is higher than in the United States.
HIV-positive individuals are exposed to more fre-
quent opportunistic infections, a circumstance that
accelerates the deterioration of their immune sys-
tems. Equally important, the greater susceptibility of
HIV-positive individuals to disease may also in-
crease the incidence of disease in the broader popu-
lation. For example, latent tuberculosis (TB) is often
activated in HIV-infected individuals, which may
then spread to the non-HIV-infected population and
increase the level of TB in the population as a whole.
There was, on average, a 20 percent annual increase
in the absolute numbers of TB cases in Zambia be-
tween 1985 and 1992. All models ignore the possibility
that the vulnerability of HIV-infected individuals to
opportunistic infections increases the disease burden of
the broader non-HIV-infected population.
Increased demand by HIV- and AIDS-infected
individuals on already-overburdened health care
services may reduce the availability of health care to
non-HIV-infected individuals, thereby increasing
the mortality rates for other diseases. In most major
hospitals in Zaire, C6te d'Ivoire, Congo, Rwanda,
Tanzania, Uganda, Zambia, and Malawi, AIDS pa-
tients constituted about 40 percent of in-patients
(Noble 1989; Lewis et al. 1989; UNICEF 1994).
Finally, each model uses different assumptions
and techniques to predict population changes as well
as the path of the HIV pandemic and AIDS-related
mortality. Differences in the final population fore-
casts, therefore, are not just a feature of differing
assumptions used to model the HIV/AIDS pandemic.

Population Forecasting:
A Review ofAfrican Countries
The United Nations has prepared detailed population
forecasts comparing scenarios with and without
AIDS for 15 African countries (Benin, Burkina
Faso, Burundi, Central African Republic, Congo,
C6te d'Ivoire, Kenya, Malawi, Mozambique,
Rwanda, United Republic of Tanzania, Uganda,
Zaire, Zambia, and Zimbabwe) to the year 2005. The
U.S. Bureau of the Census (1994b) did the same for

13 of those countries (excluding Benin and Mozam-
bique) to the year 2020. This section begins with a
discussion of the aggregate findings of the UN fore-
casts before moving on to focus on the population
forecasts by the United Nations, U.S. Bureau of the
Census, and the World Bank for the year 2020 for the
13 countries in all studies. Several factors influenced
the choice of countries:

In 1990, 90 percent of the HIV infections in
Sub-Saharan Africa (4.5 million infected peo-
ple), and more than 50 percent of the world
total, were found in the 15 countries studied by
the United Nations.
The HIV seroprevalence level is above 1 per-
cent in the population aged 15 years and older
in all 15 countries.
Comparative population and demographic data
incorporating the impact of HIV/AIDS are
available for the 13 countries in both studies.
Together, these countries account for 44 per-
cent of the 1990 population of Sub-Saharan
Africa (excluding Nigeria).
With the exception of Zimbabwe and C6te
d'Ivoire, all 15 countries are among the 30
poorest in the world and arguably have the
least resources with which to combat an AIDS

In 1980, the combined population of the 15
countries considered by the United Nations was
138.4 million,5 constituting about 3.1 percent of the
world total. By 2005, without AIDS, their total
population was projected to rise to 310.2 million,
an increase of more than 124 percent. In the presence
of AIDS, the projected increase falls to 115 per-
cent, or 297.9 million people-an absolute differ-
ence of 4 percent. Despite the impact of AIDS, the
world population share of this 15-country aggre-
gate will increase to 4.5 percent. The population
growth rate for the 15-country aggregate from 2000
to 2005 is 2.88 percent, just 0.25 percent lower than
the 3.12 percent that would be forecast in a non-
AIDS scenario. The impact of AIDS-related mor-
tality will reduce the time it takes the population to
double by just 1.1 years. Even though AIDS strikes
adults in their most productive years, the popula-
tion size of most age groups is expected to more

5This section is based on forecasts contained in United Nations 1994. The population and demographic forecasts in this publication
differ in some respects from those published in United Nations 1993.

than double by 2020. The only exception is in the
group aged 50 to 64, whose population will in-
crease only 83 percent. This population subgroup
will be 23 percent smaller than in a non-AIDS
scenario, but the difference will have little bearing
on the overall age structure of the population and
little impact on the overall dependency ratio: 99.6
with AIDS and 98.8 without AIDS. (The depend-
ency ratio is the number of people under 15 years
of age and over 65 per 100 persons between the
ages of 15 and 64 years.)
AIDS has its greatest impact on age-specific
death rates in the 15-country aggregate. Between
1975 and 1980, deaths occurring in the group aged
25 to 49 were just 12.5 percent of all deaths. By
2005, AIDS will be responsible for increasing the
proportion of deaths occurring in the 25-to-49 age
range to 20.3 percent of all deaths as opposed to a
share of 13 percent if AIDS were not a factor.
Despite the human tragedy AIDS represents, it
seems unlikely to have much effect on population
factors at the aggregate level in the 15 worst-affected
countries. At the world level, therefore, AIDS is
unlikely to suppress population size or growth rates,
and the struggle to find sustainable ways to feed the
growing world population continues. The world cur-
rently has the ability to feed all its inhabitants, but
hunger and malnutrition continue because of food
distribution and access problems. Similarly, AIDS
may not be felt as strongly at the aggregate level as
it will be by individual countries. An examination of
forecasts for individual countries confirms this.
Table 1 shows the population forecasts of the
United Nations, World Bank, and the U.S. Bureau of
the Census for 13 Sub-Saharan African countries.
Generally speaking, the base populations in 1990
show fairly close agreement and thus a fairly com-
mon starting point. Perhaps one should not be sur-
prised, given the assumptions of each model and
different modeling techniques for the incorporation
of AIDS-related mortality, that the forecasts of indi-
vidual country populations in 2020 differ quite re-
markably. One might expect that, given that the
United Nations produces three population variants
(using high, medium, and low fertility rates), their
high variant would be the highest of all of the fore-
casts and their low variant, the lowest. This would
constitute a banding, incorporating all the other fore-

casts in 2020. This, however, does not appear to be
the case. Generally speaking, the lowest population
forecasts are produced by the U.S. Bureau of the
Census, using their own assumptions, for 9 of the
13 countries.6 The World Bank produces the lowest
forecast only once, for Rwanda. The United Nations'
low-variant projections are the lowest forecast for
Burkina Faso, C6te d'Ivoire, and Zaire. The U.S.
Bureau of the Census figures, which assume that the
AIDS epidemic does not peak until 2010, later by 5
to 10 years than the other agencies' forecasts, give
some indication of the effects of an ongoing AIDS
pandemic. Given evidence that the HIV virus is ca-
pable of mutation, that more than 8,000 strains of the
virus exist, and that no vaccine for AIDS is currently
on the horizon (or likely to be affordable if and when
it is introduced in most Sub-Saharan African coun-
tries), then it may even be optimistic to predict that
by 2010 there will be no further infections. If one
abandons the assumption that there will be no AIDS-
related deaths after 2020, as the U.S. Bureau of the
Census 2020* forecasts do, population projections
fall still lower. The figures of the U.S. Bureau of the
Census predict that populations will be considerably
smaller as a result of the AIDS pandemic-45 per-
cent smaller in Uganda, 35 percent in Rwanda, and
30 percent in Malawi. Yet, although the population
of Sub-Saharan Africa will be smaller in 2020 in the
face of AIDS, the population of all countries except
Zimbabwe will at least double between 1990 and
2020, and in some cases almost triple, according to
the United Nations and the World Bank. Only the
U.S. Bureau of the Census 2020 forecasts predict
that some countries' populations will increase
slightly less than 100 percent by 2020 as a result of
AIDS-related mortality.
Table 2 shows that predicted population growth
rates differ quite remarkably for 2020, a date by
which all AIDS-related mortality is predicted to
have finished. Underlying differences in growth
rates are driven by changes in mortality and fertility
assumptions and by the differing age structures of
the population forecasts after AIDS. While the U.S.
Bureau of the Census predicts the lowest absolute
populations for Malawi, Tanzania, Uganda, Zambia,
and Zimbabwe in 2020, it predicts the highest popula-
tion growth rates. It appears that post-AIDS population
growth rates will return to formerly high levels

6The figures produced at this author's request are always lower than the U.S. Bureau of the Census's base run. These forecasts are
referred to as 2020*.

Table 1-Population forecasts, incorporating AIDS-related mortality, to 2020, for 13 African countries

United Nations U.S. Census*
2020 World Bank U.S. Bureau of the Census No AIDS
Country 1990 Low Medium High 1990 2020 1990 2020 2020* 2020

Burkina Faso 8,987 17,986 19,213 20,206 9,016 21,336 9,042 18,123 17,010 23,000
Burundi 5,503 11,302 12,103 12,777 5,492 12,233 5,558 10,734 10,210 14,000
Central African Republic 2,927 5,378 5,773 6,092 3,008 6,542 2,866 4,561 4,238 6,000
Congo 2,232 4,684 5,010 5,278 2,276 5,623 2,215 3,775 3,571 4,700
C6te d'Ivoire 11,974 29,512 31,732 33,581 11,980 30,683 12,399 29,705 28,884 33,600
Kenya 23,613 54,705 56,582 58,456 24,160 44,404 24,229 44,240 41,539 56,400
Malawi 9,367 19,168 19,814 20,460 8,507 18,995 9,289 16,697 15,586 22,400
Rwanda 6,986 13,485 14,375 15,067 6,950 12,121 7,415 15,006 13,907 21,400
Tanzania 25,600 54,540 56,347 58,148 24,470 53,438 25,155 48,526 45,498 63,700
Uganda 17,949 39,091 42,435 45,308 16,330 39,273 16,927 29,882 27,495 49,800
Zaire 37,436 85,130 91,752 97,319 37,391 87,007 37,903 92,860 91,003 101,700
Zambia 8,150 16,660 17,238 17,815 7,784 15,834 8,233 15,828 14,562 23,200
Zimbabwe 9,903 17,154 18,271 19,374 9,747 16,567 10,187 14,620 13,624 21,100
Sources: United Nations 1993; Bos et al. 1994; U.S. Bureau of the Census: 2020 figures based on output supplied by Peter Way relating to U.S. Bureau of the Census forecasting model; 2020* based on
output supplied by Peter Way from a modified U.S. Bureau of the Census model removing the assumption of no new infections after 2010.
aSource figures rounded to one decimal place in millions.

Table 2-Population growth rates, based on population forecasts incorporating AIDS-related
mortality, 1990-2020
Average Annual
Growth in GNP
per Capita, United Nations World Bank U.S. Bureau of the Census
Country 1980-92 1990-95 2015-20 1990-95 2015-20 1990 2020 2020*
Burkina Faso 1.0 2.76 2.50 2.93 2.58 2.89 2.92 1.61
Burundi 1.3 3.00 2.38 2.75 2.44 2.60 2.90 1.90
Central African Republic -1.5 2.49 2.09 2.47 2.40 2.30 2.00 0.96
Congo -0.8 2.98 2.62 3.21 2.68 2.40 2.40 1.07
C6te d'Ivoire -4.7 3.48 3.16 3.60 2.53 2.68 2.20 2.10
Kenya 0.2 3.59 2.53 2.77 1.35 3.47 2.15 0.84
Malawi -0.1 3.45 2.56 2.68 2.47 4.25 3.12 1.70
Rwanda -0.6 2.59 2.11 2.29 1.37 3.32 3.27 1.67
Tanzania 0.0 2.96 2.38 3.02 2.08 2.83 2.04 1.71
Uganda ... 3.42 2.70 3.19 2.67 3.03 2.86 1.12
Zaire ... 3.19 2.81 3.00 2.35 3.31 3.05 2.64
Zambia ... 2.97 2.57 3.06 1.86 3.35 3.21 1.46
Zimbabwe -0.9 2.57 1.73 2.61 1.23 2.53 1.93 0.44
Sources: United Nations 1993; Bos et al. 1994; U.S. Bureau of the Census: 2020 figures based on output supplied by Peter Way relating to U.S. Bureau
of the Census forecasting model; 2020* based on output supplied by Peter Way from a modified U.S. Bureau of the Census model removing
the assumption of no new infections after 2010; World Bank 1994.

quickly. Alternatively, the U.S. Bureau of the Cen-
sus 2020* forecast, in which AIDS-related mortality
is assumed to decline gradually from 2020 to 2050,
shows the effect on population growth rates of a
continued AIDS epidemic. The predicted growth
rates are between 1 and 1.75 percentage points lower
than the 2020 predicted growth rates, in many cases
representing a 50 percent decline in population
growth rates. Thus, if the effects of AIDS were to
decline more slowly after 2010, the effect on popu-
lation growth rates would be relatively large. Across
all models, the lowest population growth rates for
2020 are predicted by the World Bank: growth rates
for Kenya, Rwanda, Zambia, and Zimbabwe are pre-
dicted to fall below 2 percent.
The greatest differences in predictions of
AIDS-related mortality are seen in life expectancy
at birth. Table 3 shows life expectancy in 1990, the
minimum predicted life expectancy as a result of
AIDS, and the year in which a person would be
born and be predicted to reach the minimum pre-
dicted life expectancy. The different assumptions
of the models affect the year of birth in which the
minimum predicted life span can be expected. The
impact of AIDS-related mortality is least on pre-
dicted life expectancy in the UN forecasts, which
assume that there will be no new infections after
2005. According to the UN forecasts, AIDS will
have little impact on life expectancy overall: the
maximum decrease in life expectancy from 1990 is

2.9 years for Zimbabwe (this is because Zimbabwe
had one of the highest life expectancies in Sub-
Saharan Africa before the arrival of AIDS). In most
cases, predicted life expectancy is at its lowest for
those born in 1995. By contrast, the U.S. Bureau of
the Census predicts far larger changes in life expec-
tancy. The expected life span in Kenya in 2010 is
just 39.8 years-17 years less than it was in 1990.
Life expectancy in Kenya was one of the highest in
Africa in 1990, and because AIDS-related deaths
occur primarily in young adults and preschool chil-
dren, life expectancy has declined precipitously.
The World Bank forecasts occupy a middle ground,
predicting some shift in life expectancy and the
demographic structure. For example, the mean age
of the working-age population (15 to 64) in Tanza-
nia will fall from 32 in 1985 to 28 in 2020; if the
AIDS epidemic did not exist, the mean working
age would be 31.5 (cited in Cuddington 1993a).
One of the greatest potential tragedies of the
AIDS epidemic will be the reversal in many coun-
tries of the hard-won downward trends in infant and
child mortality rates. In Zambia and Zimbabwe, the
U.S. Bureau of the Census predicts that AIDS will
cause infant mortality rates to almost double and
child mortality rates to almost triple (Sraneck and
way 1994). In Zambia, the infant mortality rate will
climb from 39 to 77 per 1,000 live births and the
child mortality rate from 56 to 160. In Zimbabwe,
the infant mortality rate is expected to rise from 30

Table 3-Life expectancy at birth, incorporating AIDS-related mortality
United Nations World Bank U.S. Bureau of the Census
Minimum Minimum Minimum
Life Birth Life Birth Life Birth
Country 1990 Expectancya Yearb 1990 Expectancy' Yearb 1990 Expectancya Yearb

Burkina Faso 47.4 46.6 2000 48.2 48.2 1990 48.9 35.6 2010
Burundi 50.2 50.2 1990 48.0 47.1 1995 42.2 38.3 2005
Central African Republic 49.4 49.4 1990 47.0 45.7 1995 44.2 32.3 2010
Congo 51.3 49.9 1995 51.5 50.7 1995 49.6 39.5 2010
C6te d'Ivoire 51.0 49.5 1995 56.3 54.7 1995 49.1 48.5 2000
Kenya 55.7 54.2 1995 58.9 58.9 1995 56.8 39.8 2010
Malawi 45.6 44.8 1995 44.2 42.5 1995 42.9 32.7 2010
Rwanda 47.3 46.6 1995 46.2 44.6 1995 44.5 33.6 2010
Tanzania 52.1 51.5 1995 50.8 49.0 1995 46.5 35.3 2010
Uganda 44.9 43.3 1995 43.3 40.0 2000 41.6 31.5 2010
Zaire 52.0 51.9 1995 51.6 51.4 1995 46.8 46.8 1990
Zambia 48.9 46.1 1995 47.8 41.9 2000 50.3 33.4 2010
Zimbabwe 53.7 50.8 1995 59.8 54.7 2000 45.3 38.0 2005
Sources: United Nations 1993; Bos et al. 1994; U.S. Bureau of the Census: 2020 figures based on output supplied by Peter Way relating to the U.S.
Bureau of the Census forecasting model; 2020* based on output supplied by Peter Way from a modified U.S. Bureau of the Census model
removing assumption of new infections after 2010.
"This is the minimum forecast life expectancy for a country.
bIf born in this year, the predicted life expectancy is the lowest. For example, if born in Burkina Faso in the year 2000, life expectancy is just 46.6
years. If born in the same country in 1990, life expectancy is 47.4 years.

to 55 per 1,000 live births, and the child mortality
rate from 38 to 108. It is interesting that the UN
forecasts do not predict a reversal of the trend in
infant and child mortality rates-they continue to
decrease. Yet, in one urban center of Malawi, AIDS
was responsible for increasing the infant mortality
rate by 19 per 1,000 live births, and the mortality rate
for those under age 3 by 37 per 1,000 live births
(Liomba, Kandulu, and Kandiado 1994). Similarly,
in Zambia, the 1992 Demographic and Health Sur-
vey showed a 35 percent and 26 percent increase in
infant and child mortality rates, respectively, between
1980 and 1992 (Gaisie, Cross, and Nsemulkila 1993,
81). AIDS is believed to be a key factor. It is hard to
believe that, in the face of these increases in mortal-
ity rates, fertility behavior will be unchanged.

In general, the later the forecast models predict the
AIDS epidemic to peak, the greater the effects on
population and other demographic factors. At the
present time, there is no end to the AIDS epidemic in
sight, no vaccine available, and little evidence to
indicate that HIV infection is slowing down in the

general populations of the countries most affected.
Thus, it is a matter of conjecture whether the epi-
demic will peak in 2005, 2010, or even later if a
prophylactic is not found. It is unlikely that, for the
13 African countries discussed in the previous sec-
tion, population growth rates will be reduced to zero
before 2020. They are likely, however, to be re-
duced. For many, this may be seen as positive. Im-
provements in welfare, measured by gross domestic
product (GDP) per capital, can be achieved only if
GDP increases faster than population growth. As
Table 2 shows, for many of the countries where
figures are available, population growth rates out-
stripped economic growth rates between 1980 and
1992, resulting in declining welfare levels. In this
light, falling population growth rates could remove
some of the pressure to achieve high levels of eco-
nomic growth just to maintain a welfare status quo.
Therefore, many may argue that, although AIDS
is a tragedy, it is also a Malthusian-type phenomenon
that results in improvements in the ability of the
world to sustain and feed itself. The IFPRI impact
model supports this view in terms of global food
security as a measure of welfare.7 The baseline 2020
forecasts of this model use the UN medium-variant

7The IFPRI impact model is a set of country or regional models linked through trade that determines supply, demand, and prices for
agricultural commodities. For a full description, see Rosegrant, Agcaoili-Sombilla, and Perez 1995.

population forecasts. A simulation using the UN
low-variant population forecasts, which are more
conservative than the U.S. Bureau of the Census
forecasts in terms of the negative impact of AIDS on
population, indicates that global per capital food
availability in 2020 would be 5 percent higher than
in the baseline scenario. This would translate into a
reduction of 33.5 million in the forecast number of
malnourished children in the developing world in
2020. About 31 percent of this reduction would be in
Sub-Saharan Africa. At the same time, net imports
of cereals to the region would be 2.5 million metric
tons less than in the baseline scenario in 2020.
The major caveat to these results, however, is that
the low-population-variant scenario assumes that the
productive capacity and economic structure of the
economy will not change from the baseline scenario.
If AIDS does affect the productive capacity of an
economy, and the effect is greater than the effect on
population growth, then the per capital availability of
food may be less than that of the baseline scenario.

AIDS and Economic
Growth Linkages
The previous section outlined ways in which AIDS
will alter the future demographic structure of popu-
lations. These have consequences for economic
growth from several perspectives.
A critical factor in the development process and in
generating economic growth is investment in human
capital. Rapidly growing populations require in-
creased levels of investment in human capital just to
maintain existing average per capital human capital
stock. A shift in the demographic structure of the
population due to AIDS-a movement toward
younger rather than older populations-will result in
shifts in social service expenditures. If health care
expenditures are higher for the young than for the old,
then higher health care expenditures will be required
regardless of the direct health impact of AIDS. Simi-
larly, although the absolute numbers of the younger
population may be falling as a result of AIDS, the
relative proportion will grow, resulting in higher de-
mand for education expenditures. By shortening life
spans, however, AIDS will lower the returns to public
investment in both health and education.
Similarly, AIDS will affect the private rates of
return to human capital investment, particularly in the
face of the introduction of health and education user
fees as a result of structural adjustment programs. If
parents believe their children are unlikely to survive

long enough to support them in old age, then they may
reduce their investment in preventive health care, such
as immunizations, for their children. Parents are also
faced with a dilemma regarding educational invest-
ment: if AIDS strikes at the more educated, productive
worker, a skill shortage may be created, increasing the
wages of skilled workers and the private rate of return
to education. But households facing labor shortages
may have to choose current child- labor over future
gains from education, offsetting these long-run bene-
fits. AIDS will produce increasing numbers of
orphaned children, who will be cared for by extended
family members or by the state. Evidence from Tan-
zania indicates that maternal orphans have lower
school enrollment rates than their nonorphan counter-
parts (Ainsworth and Rwegarulira 1992). Thus, one
potential outcome of AIDS may be reduced levels of
human capital investment, with consequences for
future economic growth.
AIDS will, however, have a more direct impact
on investment in maintaining human capital. AIDS-
related morbidity will increase demand for curative
health care. The increased demand for health care
services, particularly in urban areas where AIDS is
currently concentrated, is likely to crowd out other
health care demands. In 1988, 50 percent of in-
patients at the Mama Yeo Hospital in Zaire were HIV-
positive (Hassig et al. 1990). Because the hospital
had always operated at full capacity, it must have
had to turn away many non-AIDS patients. Shifts in
the allocation of government health expenditures in
favor of tertiary health care facilities may shift the
current donor-led emphasis of health care expendi-
tures away from primary, particularly rural primary,
health care. Increasingly, resources will be allocated
to health sector expenditures and health care rather
than to a productive investment. The long-run results
will be to lower domestic capital formation, dimin-
ishing the capacity for future economic growth.
Its modes of transmission and its age-group selec-
tivity make AIDS a "clustering" disease. If HIV in-
fection and AIDS are not randomly distributed across
a population, but concentrated in particular population
subgroups that have higher than average economic
productivity, then the effect of AIDS on domestic
capital formation will be magnified, because those
with higher wages are likely also to have higher levels
of saving, both relatively and absolutely.
A 1987 study of pregnant women in urban areas
of Rwanda found their HIV status to be uncorrelated
with their own socioeconomic status but related to
their husbands' or partners' socioeconomic status.
Eighteen percent of pregnant women whose partners

had up to four years of education were HIV-positive,
but 34 percent whose partners had more than eight
years of education were HIV-positive. Similar pat-
terns were found for women whose partners had
higher income levels and worked in higher-paid and
higher-skilled occupations. Pregnant women with
partners who were farmers had an HIV seroprevalence
level of 9 percent. In contrast, pregnant women with
military, private sector, or civil service partners had
HIV levels of 22 percent, 32 percent, and 38 percent,
respectively (Allen et al. 1991, cited in Ainsworth
and Over 1994). An ongoing study in Rakai District,
Uganda, found that HIV seroprevalence rates among
men and women with secondary education (20 per-
cent and 41 percent, respectively) were more than
twice as high as those for men and women with no
education (8 and 14 percent, respectively) (Serwadda
et al. 1992, cited in Ainsworth and Over 1994). In
Kenya, Forsythe et al. (1993) estimated that the
average annual income of a worker with AIDS was
31 percent higher than the average national income.
In 1992, among blood donors in Lusaka, Zambia,
HIV seroprevalence levels were 13 percent for farm
workers and 30 percent for government workers
(Fylkesnes, Brunborg, and Msiska 1994, 18). A fur-
ther Zambian study by Kanyama, Kaona, and Sisiya
(cited in Fylkesnes, Brunborg, and Msiska 1994, 21)

in early 1992 showed that 39 percent of midwives,
44 percent of nurses, and 42 percent of office workers
and teachers were HIV-positive. The high rates among
midwives and nurses are of particular concern. One
assumes that knowledge of AIDS and its transmission
mechanisms is higher among health sector workers
than other socioeconomic groups, and thus one would
expect lower rates ofHIV seroprevalence.
Potentially, AIDS can influence economic
growth in several ways. Researchers have attempted
to assess the effect in a disaggregated sense by esti-
mating its direct and indirect costs, and at an aggre-
gate level using computable general equilibrium
(CGE) and Solow-type growth models.

Direct and Indirect Cost Estimates
of the Impact of AIDS
Direct costs of AIDS are defined as the costs of
treating HIV- and AIDS-infected individuals. Esti-
mates of the direct costs vary tremendously, not
only across countries but within countries, depend-
ing on the socioeconomic characteristics of those
infected and the quality of services to which they
have access. Table 4 shows the diversity of estimates.
What is clear from Table 4 is that in all but the
lowest of country-specific estimates, the direct

Table 4-Estimates of the direct cost of AIDS, per case, in selected countries
Direct Costs* GNP
Country Low Average High Per Capita
(US$) (1992 US$)
Zaire (1987-88)b 132 n.a. 1,585 n.a.
Tanzania (1987-88)b 104 n.a. 631 110
Tanzania (1990) n.a. 290 n.a. 110
South Africa (1991)c 2,857 n.a. 7,143 2,670
Malawi (1989)d n.a. 210 n.a. 210
Kenya (1992)d n.a. 938 n.a. 310
Jamaica (1987)d n.a. 1,807 n.a. 1,340
Zimbabwe (1991) 64 614 2,574 570
Thailand (1993)' 987 n.a. 1,524 1,840
Korea (1993) n.a. 2,010 n.a. 6,790
Malaysia (1993)f n.a. 3,000 n.a. 2,790
Rwanda (1989-90) n.a. 358 n.a. 250
Sources: For Zaire and Tanzania 1987-88, Over et al. 1988; for Tanzania 1990, Pallangyo and Laing, cited in Ainsworth and Over 1994; for South
Africa, Broomberg, Masoe, and Steinberg 1991; for Malawi, Kenya, and Jamaica, Forsythe et al. 1993, Table 7; for Zimbabwe, Whiteside
1991, cited in Ainsworth and Over 1994; for Thailand, Viravaidya, Obremskey, and Meyers 1993; for Korea, Yang 1993; and for Malaysia,
UNDP 1993; for Rwanda, Shephard and Bail 1991, cited in Ainsworth and Over 1994; World Bank 1994.
Note: n.a. means not available.
aEstimates are based on type and quality of treatment sought. If only one figure is cited, it is the average.
b1985 US$.
'Based on an official exchange rate of R2.8 to US$1.
d1991 dollars.
eBased on an 18-month life span with AIDS.
fBased on a 12-month life span with AIDS.

Table 5-Estimates of the indirect costs per case of HIV infection in Zaire and Tanzania
Zaire Tanzania
Average Cost of Average Cost of
Labor Category Annual Income HIV Infection Annual Income HIV Infection
(1985 US$)
Rural 144 890 391 2,425
Primary 287 1,780 626 3,880
Secondary 431 2,669 821 5,093
Source: Compiled from information in Tables 3 and 4 in Over et al. 1988.

costs of AIDS per case exceed per capital gross
national product (GNP). They also exceed average
per capital public expenditures of $5 on health care
in most Sub-Saharan African countries. Ainsworth
and Over (1994, 225) estimated the total costs,
public and private, of treating the estimated num-
ber of AIDS cases in Zimbabwe, Kenya, Malawi,
Tanzania, and Rwanda. These costs ranged from
23 percent of 1990 public health spending in Kenya
to more than 65 percent in Rwanda.
The indirect costs of AIDS are defined as the
discounted healthy years of life lost per individual to
society as a result ofAIDS-related mortality or, when
valued in economic terms, the value of production
forgone as a result of AIDS-related morbidity and
mortality. This approach underestimates the contribu-
tion of the under- and unemployed, assuming they do
not spend all their productive years in this state, and
overestimates the loss of those whose salary is made
up in part of economic rents. Examples of the
approach are provided by Over et al. (1988) for Zaire
and Tanzania, and Forsythe et al. (1993, 16) for Kenya.
In Zaire and Tanzania, preventing one case of HIV
infection saves 8.8 discounted healthy life years,8 a
number that ranks fifth after sickle-cell anemia, neo-
natal tetanus, birth injury, and severe malnutrition in
the number of healthy life years saved.9 It is notable
that the first four diseases are largely diseases one is
born with or that are acquired in early childhood.
AIDS counts for greater savings in discounted healthy
life years than many other endemic adult diseases
such as TB and malaria. In Kenya, Forsythe et al.
(1993, 16) calculated that 10 years of discounted pro-

ductive life years were lost with each new case of
AIDS, or more than three-fifths of an individual's
undiscounted productive years. In Zaire and Tanzania,
Over et al. (1988) used three categories of labor to
arrive at an economic valuation of healthy life years
lost: rural adult, urban adult with primary education,
and urban adult with secondary schooling or more.
The results are shown in Table 5.
Forsythe et al. (1993, 15) valued the discounted
life years on the basis of age, income, and industry
distribution of new AIDS cases and calculated that
the average annual income of a person with AIDS
was US$650. Assuming a loss of 10 discounted pro-
ductive life years per AIDS case, the indirect costs
per case would be US$6,713. Therefore, for these
three countries the indirect costs of each AIDS case
far outweigh both the direct costs and, as a result,
gross domestic product (GDP) per capital (the indi-
rect costs per new adult AIDS case in Kenya were
23 times per capital GDP). The HIV seroprevalence
rate in Kenya in 1991 was estimated at between
3.4 percent and 5.1 percent of the total adult popula-
tion. This represents a considerable potential indirect
cost of AIDS to the Kenyan economy in the period
from 1991 to 2001, dependent on the conversion rate
of HIV into AIDS. By the year 2000, Forsythe et al.
(1993, 17, 24) estimate that the direct cost of AIDS
(that is, the cost of treating AIDS patients) in Kenya
could be as high as 4.7 billion 1991 Kenyan shil-
lings.10 The indirect costs, however, could be as high
as 40 billion Kenyan shillings (in 1991 shillings).
One flaw in using the discounted life years ap-
proach in combination with a valuation of human

8The discount rate used is 5 percent.
9This statement is based on healthy life years saved compiled for Ghana in International Journal of Epidemiology (1981) for other
'oThis figure is based on a calculation of high adult HIV seroprevalence levels of 11 percent in 1995.

capital is that it ignores lost savings. When a worker
becomes too sick to work not only is a productive
lifetime of earnings lost but also the proportion of
earnings that would have been saved. These savings
contribute through investment to capital formation in
the economy. Similarly, the human capital approach
assumes a constant income loss per worker lost. The
standard economic production function approach
generally assumes diminishing marginal productiv-
ity of each additional worker. Thus, as the epidemic
begins to bite, each worker lost is more valuable than
the one before. To overcome some of these difficul-
ties in the human capital approach, various attempts
have been made to use CGE and Solow-type growth
models to forecast the effects of AIDS on various
aggregates in the macroeconomy.

Potential Macroeconomic Impact
An example of the CGE approach, applied to
Cameroon, is provided by Kambou, Devarajan, and
Over (1992). In a dynamic model they portray AIDS
as a shock to the labor market, resulting in the loss of
10,000 workers from each of the three labor catego-
ries (rural, urban unskilled, and urban skilled) in
each year of the simulation between 1987 and 1991.
A savings closure rule is used (that is, foreign savings
are fixed exogenously) such that investment varies
directly with the level of domestic savings. The re-
duction in workers causes the economy to contract
sharply, with GDP growth falling from 4.3 percent
in a no-AIDS economy to 2.4 percent in the AIDS
simulation. The decline in growth is triggered by
falling investment (a result of stagnating private sav-
ings) and falling government savings (a result of
declining trade revenues when exports decline).
To further investigate the selective impact of
AIDS on more productive categories of labor, Kam-
bou, Devarajan, and Over reran the simulations by
reducing 10,000 workers from one labor skill cate-
gory at a time. The removal of 10,000 rural sector
workers each year seems to have little effect on the
economy. The prime reason for this is that 10,000
workers represent just 0.4 percent and 0.3 percent of

the rural and entire labor force, respectively. Be-
cause rural labor supply grows 2 percent per year in
the model, labor force growth in the rural sector is
still positive, despite the impact of AIDS. The result
is similar when the reduction is 10,000 urban un-
skilled workers in each year. The most striking result
occurs with the removal of 10,000 urban skilled
workers in each year of the simulation. Government
savings plummet by 20.6 percent per year on aver-
age, resulting in a fall in fixed investment growth
from 5.1 percent in the base year to 1.7 percent per
year in the simulation. The assumption of this
model-a constant yearly fixed loss of labor-may
not be realistic. What is clear, however, is that an
unequal loss of labor concentrated among the skilled
work force has devastating effects, reducing long-
run economic growth from 4.3 percent to 2.6 percent
per year. In Cameroon, where the average annual
population growth from 1986 to 1995 was 3 percent,
the difference between these two economic growth
rates is the difference between increasing per capital
GDP and declining per capital GDP.
Cuddington (1993a) has used a Solow-type
growth model to explore the macroeconomic impact
of AIDS in Tanzania. He first used a simple neoclas-
sical model in which the economy adjusts to maintain
production with efficient allocation of all resources,
including labor--assumptions that are unrepresenta-
tive of Tanzania, or indeed many African economies.
The model considers only one type of labor and fo-
cuses on the direct effect of increased health care costs
on savings. The model uses the population and AIDS-
prevalence forecasts of Bulatao for Tanzania, in
which AIDS prevalence rises from 0.09 percent of
adults in 1985 to 3.15 percent in 2010. Simulations
were performed using alternative values for the annual
labor productivity lost per AIDS-infected worker
(from zero effect to 2)," and the proportion of AIDS-
related medical expenses financed by reduced savings
(also from zero to 2).12 The simulations show that,
depending on the assumptions made, the Tanzanian
GDP in 2010 may be reduced by 15 to 25 percent from
a no-AIDS scenario. This translates into a per capital
income reduction of zero to 10 percent, depending on
assumptions relating to labor productivity and health

1Zero implies there is no decline in labor productivity; 2 implies the loss of productive labor of both the infected individual and a
12Savings may be reduced not only by the direct cost of health care expenditures to treat AIDS-related morbidity but also by social
sector spending to care for orphans, or by public health programs to prevent the spread of HIV.

care financing.13 Cuddington and Hancock (1994)
applied a similar model to the Malawian economy.
The simulations were limited to a maximum 100 per-
cent reduction in labor productivity of the infected
individual only, and a maximum of 100 percent of
health care costs met from savings (that is, zero to 1 in
both cases). The simulations showed that in 2010,
under a medium forecast for the prevalence of AIDS,
average real GDP growth would be 0.2 to 0.3 percent-
age points lower, and per capital GDP growth 0 to
11 percent lower, than in a no-AIDS scenario.14
One potential criticism of these models is that
the underlying classical economic assumptions are
invalid. Assumptions about the homogeneity of
labor productivity and education in the face of an
epidemic (which has initially hit the most productive
workers disproportionately) and fully flexible labor
markets are likely to result in an understatement of
the true effects of the AIDS epidemic. Cuddington
(1993b) subsequently modified this approach for
Tanzania by relaxing the full-employment assump-
tion and considering a dual-economy labor surplus
situation, a more valid assumption for both the Tan-
zanian and other Sub-Saharan African economies.
This is achieved by introducing a formal sector, with
sticky wages, and an informal sector, with wages
equal to average product, to the model. Results from
the simulations, using the same alternative parame-
ter values for declining labor productivity and reduc-
tions in savings, are similar to the results using the
one-sector model. In a scenario in which wages
adjust slowly, the impact of AIDS is a reduction in
GDP of between 11 and 28 percent, depending on
the assumptions made about the decline in labor
productivity and savings behavior. Similarly, the
effect on per capital GDP ranges from a 3.6 percent
increase in a no-AIDS scenario to a decline of
16.1 percent in the scenario where AIDS reduces the
productivity of a worker and a caregiver by 100 per-
cent, and the decline in savings parameter is set to 2.
Over (1992) takes the modeling one stage fur-
ther by incorporating a high-productivity urban sec-
tor and low-productivity rural sector as well as three
types of workers: uneducated, those with primary
education, and those with more than a primary edu-
cation. Although the model makes no allowance for
reductions in labor productivity as a result of AIDS,
it does allow for assumptions to be made in relation

to risk of infection for different skill categories of
workers. These range from an assumption that the
epidemic disproportionately affects the least-skilled
worker to one in which the worker with more than a
primary education is 16 times more likely to be
infected. Like the models described in the previous
paragraphs, health care costs can be financed to
varying degrees from savings. The Over model is
applied both to an aggregate of 30 Sub-Saharan Af-
rican countries and to a subgroup of the 10 countries
most affected by AIDS. Results indicate that, as the
likelihood of more-educated workers' being infected
increases and the percentage of health care costs
financed from savings increases, the negative impact
on annual GDP per capital growth for both the 30-
and 10-country aggregate increases. In a situation
where the epidemic affects those with no education
most and none of the health care costs are financed
from savings, annual growth rates of GDP per capital
are reduced by 0.56 percentage points. Where those
with more than primary education are 16 times more
likely to be infected and all health care costs are
financed from savings, the negative impact on
annual GDP growth per capital is 1.08 percent. For
the 10 worst-affected countries, the impact on GDP
growth per capital ranges from a decline of 0.73 per-
cent to 1.47 percent. Neither of these extreme sce-
narios is likely, but they reveal an inevitable conclu-
sion-per capital GDP growth rates are likely to be
decreased by at least 0.6 percent and possibly as
much as 1.4 percent. This is a substantial reduction
in relative terms for many Sub-Saharan African
economies, given that the Sub-Saharan African aver-
age annual GDP per capital growth rate was -0.8 per-
cent from 1980 to 1992.

Looking Forward: HIV/AIDS and
Food, Agriculture, and the

IFPRI's 2020 Vision is ultimately about the achieve-
ment of food security for every individual, on a global
level, in a sustainable manner, without irreversible
damage to the environment. Food security is defined
as the availability of food, access to food, and absence

13Population growth is assumed to decline by 0.7 percent, conservative by U.S. Bureau of the Census estimates.
14Population growth is assumed to decline by 1.2 percent, conservative by U.S. Bureau of the Census estimates.

of risk related to either availability or access. As a
concept, it can be applied at the national, household,
or individual level. In this paper, emphasis is on
national food security, a necessary but not sufficient
condition for either household or individual food
security. The concept embraces the availability of
food and, often the more important determinant,
access to food. Food availability at the national level
can be ensured through self-sufficient production or
through international trade, the latter requiring eco-
nomic access. At the national level, economic access
depends on the ability to generate sufficient foreign-
exchange income to purchase required food, assuming
it is available in the world market.
HIV/AIDS is intimately connected to food secu-
rity in a cyclical relationship. It has been suggested
that the consequences of food insecurity, manifest in
protein energy malnutrition and micronutrient defi-
ciencies, render individuals more susceptible to HIV
infection and to a faster progression from HIV infec-
tion to AIDS and, ultimately, to death (Smallman-
Raynor and Cliff 1992; Semba et al. 1994).
HIV/AIDS will, in turn, have an impact on the abil-
ity to achieve national food security through food
supply and demand factors determined by popula-
tion and economic growth.
Many environmentalists argue that the natural
resource base is already dangerously overburdened
by current populations and that future rapid popula-
tion growth can only lead to famine and disaster. In
recent years, population growth rates have outpaced
economic growth rates in many Sub-Saharan Afri-
can countries, resulting in declining levels of wel-
fare. Many may argue that if AIDS reduces popula-
tion size and growth rates it will increase the ability
of the world to feed itself in the year 2020. A recent
study provided a "rule of thumb" approach to esti-
mating the impact of AIDS on population growth:
with a 10-year incubation period from HIV infection
to AIDS death and a 3 percent population growth
rate, it would take a sustained HIV seroprevalence
level of 48 percent in the adult population to reduce
population growth to zero (Stover 1993a, 1993b,
cited in Ainsworth and Over 1994, 24). If, however,
the population growth rate were 2 percent and the
incubation period were five years, a sustained sero-
prevalence rate of 20 percent in the adult population
would reduce population growth rates to zero
(Ainsworth and Over 1994, 24). It is therefore un-
likely that population growth rates in Africa will turn
negative, although the same will not be true for some
Asian economies, notably Thailand (U.S. Bureau of
the Census 1994b).

For most of the African economies considered in
this paper, population growth rates. are unlikely
to fall below 2 percent by 2020, unless the onslaught
of AIDS continues unchecked. This possibility is
supported by the U.S. Bureau of the Census 2020*
figures, which do not assume that there will be no
new HIV infections after 2010 but do assume that all
AIDS-related mortality will cease by 2050. Accord-
ing to UN forecasts, the time needed for the popula-
tion of its 15-country Sub-Saharan aggregate to dou-
ble from its 1980 level will be reduced by just
1.1 years. Even in the four worst-affected coun-
tries-Malawi, Rwanda, Uganda, and Zambia-the
maximum delay in the time it takes the population to
double will be just 2.9 years for Zambia. Even the
U.S. Bureau of the Census 2020* forecasts, which
assume the AIDS epidemic will run beyond 2020,
predict that only the population growth rates in the
Central African Republic and Kenya will be less
than 1 percent in 2020.
Thus, as we look toward 2020, it seems that
AIDS will do little to reduce population growth and
thereby demand for food. The most critical impact of
AIDS is likely to be in the damage inflicted on the
productive capacity of an economy and its potential
to achieve food security through domestic produc-
tion or economic access in world markets.
Many nations, particularly in Asia and Sub-
Saharan Africa, are food insecure-that is, they are
neither domestically self-sufficient in food produc-
tion nor able to purchase sufficient quantities on
world markets, despite historically low food prices.
Some 800 million people do not have access to suf-
ficient food for a healthy and productive life, despite
the fact that the world as a whole produces sufficient
food to feed all of its inhabitants. The existence of
hunger and malnutrition is testimony to the inequita-
ble distribution of food, a factor driven by poverty
and the consequent lack of economic access to food.
At the end of the 1980s, about half of all African
countries were unable to guarantee adequate caloric
intake, about 2,200 calories per day, for all their
citizens, even if, nationally, all of the available food
were distributed equitably (Pinstrup-Andersen 1994).
The selective impact of AIDS, largely on the
population in its most productive years, combined
with an initial disproportionate effect on the most
productive members of society, has the potential not
only to reduce economic growth but also to alter the
economic structure. The direct cost approach to evalu-
ating the impact of AIDS indicates that direct eco-
nomic costs are high. There will be pressure on gov-
ernments, particularly in the earlier stages of the

Table 6-Net trade balance for regions of Africa and major food crops, 2020
Other Roots and
Country/Region Soybeans Meals Wheat Maize Grains Tubers Rice
Nigeria + -
Northern Africa + + +
Central and West Africa 0 + -
Southern Africa +
East Africa + + -
West and North Africa + + -
Source: IFPRI impact model provided by M. Agcaoili-Sombilla.
+ indicates net exports.
- indicates net imports.

epidemic, when the higher-income urban sectors with
most of the political power are most affected, to in-
crease tertiary health care expenditures. It is unlikely
that most governments will be able to meet the de-
mands on their health care systems. Already, patients
with AIDS-related illnesses are crowding non-AIDS
patients out of some urban hospitals, a circumstance
that will result in an increased disease burden within
the whole population. This may be particularly serious
in rural areas. The demand for expensive urban terti-
ary health care to treat the initial AIDS epidemic is
likely to result in lower primary health care expendi-
tures in rural areas where health care is often already
woefully inadequate. Declining rural health will jeop-
ardize agricultural productivity and increase the vul-
nerability of the population to HIV infection. Indeed,
Ruttan (1994) suggests that poor health could become
a major constraint to agricultural production by
the early decades of the twenty-first century. Little
progress has been made in the control of some para-
sitic diseases, and the sustainability of malaria and TB
control is cause for concern.
The greatest costs of AIDS, however, are in for-
gone production-the indirect costs. The Kenyan
economy, for example, will be hard pressed to with-
stand indirect costs 23 times its per capital GDP for
each new AIDS case, in addition to direct costs of
treatment that are more than twice per capital GDP. By
the year 2000, the present value (in 1991 shillings) of
the direct and indirect costs of AIDS to the Kenyan
economy could range from 6 to 15 percent of GDP.
When these costs are fed back into economywide
models, which include the effects of reduced savings,

they show that economic growth will be impaired. As
illustrated by Kambou, Devarajan, and Over (1992),
the impact of AIDS on economic growth is greatest
when the loss of workers is in the productive urban
formal sector. The effects on the economies may be
worse than the economic growth models (which are
based on sectoral aggregates) predict if AIDS hits
selected population subgroups within the urban for-
mal sector. In Zambia, 40 percent of all AIDS cases
are found in Copperbelt, which accounts for 6 percent
of the labor force and 20 percent of GNP. What may
be even more critical in the national food security
debate is that copper accounted for 85 percent of
Zambian exports in 1988; it is a valuable source of
foreign exchange providing economic access to food
on world markets. Africa currently requires 14 million
more metric tons of grain per year than it produces.
Imports of cereals in Sub-Saharan Africa are pro-
jected to triple between 1990 and 2020, with an an-
nualized growth rate of 3.6 percent. The majority of
these imports will be wheat and rice. These imports
will continue to be necessary beyond 2020, although
the needs of regions will differ within the continent
(Rosegrant, Agcaoili-Sombilla, and Perez 1995).
Table 6 shows that East Africa15 and Southern
Africa,16 the African regions currently most affected
by AIDS, will continue to be net importers of virtu-
ally all major staple foods, with the exception of
roots, tubers, and soybeans (East Africa only). One
simulation of the IFPRI impact model considered a
low-investment, low-growth scenario. This has
some similarities to the effects of an AIDS pan-
demic. The scenario incorporated a 25 percent

15East Africa includes Burundi, Kenya, Rwanda, Tanzania, and Uganda.
16Southern Africa includes Angola, Botswana, Djibouti, Lesotho, Madagascar, Malawi, Mauritius, Mozambique, Namibia, Rwanda,
Swaziland, Zambia, and Zimbabwe.

reduction in nonagricultural income growth, the
elimination of public investment by international do-
nors in national agricultural research systems and
extension services in developing countries, along
with funding to international agricultural research
centers, and a reduction in investment in health,
education, and sanitation. As demonstrated, AIDS
will reduce economic growth, and the competition
for both national and international resources to com-
bat AIDS is likely to reduce investment in agricul-
tural research, health (to tackle diseases other than
AIDS), education, and sanitation. This simulation of
the model increases net imports for Sub-Saharan
Africa by 3 million metric tons, or 11 percent, over
the base year. A future simulation is planned to more
closely approximate the AIDS pandemic.
AIDS will not only diminish the possibility that
nations will be able to achieve food security through
economic access, particularly when foreign ex-
change earnings are reduced, but it will also dimin-
ish the potential to increase food security through
domestic food production.
Opportunities to expand domestic agricultural
production by cultivating new land are diminishing,
and much of the required future expansion in agri-
cultural production will need to be generated by
agricultural intensification and adoption of modern
farming technologies. Governments under pressure
to cover rising health care costs are likely to sacrifice
"soft" budget items, such as agricultural research,
which is a key factor in the development of new,
country-appropriate agricultural technology. The
availability of many of the inputs required for agri-
cultural intensification, such as fertilizer, may be
limited, given that they are often imported and con-
sequently at risk in the face of declining export earn-
ings. Many of the new agricultural technologies,
including many hybrid seeds, require higher and
more timely labor inputs to ensure maximum yields.
Equally, environmentally friendly farming tech-
niques, which need to accompany new technology to
ensure sustainability, are labor-intensive.
AIDS is spreading rapidly into many agricultural
surplus rural areas because of good infrastructural
links that facilitate rural urban trade; in its path, the
availability of labor will decline, further jeopardizing
food production. The economic growth model of
Kambou, Devarajan, and Over (1992) indicated that
the loss of rural labor would have little impact on
economic growth as population growth outstripped
the declines in labor. This conclusion is somewhat
misleading, however, in terms of agricultural produc-
tion-currently, many rural communities face labor

bottlenecks, owing to the seasonal nature of agricul-
tural production. In badly affected communities,
farmers are likely to switch to less labor-intensive
crops, often from export and tradable crops, such as
tobacco, maize, and plantains, to less nutritious sub-
sistence crops, such as cassava. Declining nutritional
status will further jeopardize agricultural productivity,
increase the vulnerability of individuals to HIV infec-
tion, speed transition from HIV infection to AIDS,
and ultimately increase poverty. The high incidence of
poverty, particularly in rural areas, is one of the key
factors driving unsustainable agricultural practices as
farmers sacrifice long-term sustainability in favor of
short-term survival.
AIDS is becoming broadly accepted as a disease
of poverty, despite the fact that it may initially spread
fastest among the urban elite (Whiteside 1994). In the
early years of the disease's presence, it is the richer
segments of society who have disposable incomes and
the leisure time in which to engage in behavior condu-
cive to a rapid spread of AIDS. In the longer run,
however, they will also be more receptive to informa-
tion and education campaigns through the mass me-
dia, more able to purchase condoms, and more likely
to live in environments that encourage condom use. It
is the poor, with their poorer health status, inability to
purchase condoms, and lack of access to information
and education campaigns, who will ultimately suffer
the most from AIDS.
Just as AIDS will be concentrated among the
poor, it will exacerbate poverty as the poor lose
access to what is often their only resource--their
own labor. Poverty is the key driver of food insecu-
rity and environmental degradation: as the poor, par-
ticularly in rural areas, struggle to survive and mine
the natural resources around them, they are likely to
put even more pressure on the environment.
HIV/AIDS must be tackled early; like population
growth, once established in a population, it has its
own momentum driven by the lag time between HIV
infection and the development of AIDS. Even at that
point when new HIV infections can be prevented, the
scale of HIV/AIDS devastation will continue for
roughly 10 years. No country can afford to be com-
placent in the face of apparently low HIV seropreva-
lence levels. Many African countries can testify to
the costs of ignoring the spread of HIV infection in
the early stages.
To arrest the AIDS pandemic in Africa, and to
prevent fledgling epidemics in South and East Asia
from becoming pandemics, AIDS must be addressed
by the broader development community, not just the
health sector.

Appendix 1:

Background Information about HIV/AIDS

The human immunodeficiency virus (HIV) is the
etiologic agent of acquired immune deficiency syn-
drome (AIDS). There are currently two recognized
serotypes of HIV: HIV1, which is the most common
form worldwide, and HIV2, which has been found
only in West Africa. On infection, HIV attacks a
type of white blood cell, CD4+, which contains a
specific antigen or disease-fighting agent. Having
bound to the antigen, the HIV virus eventually be-
gins duplicating itself and attacking other CD4+
cells and destroying them. As the CD4+ cell count
falls, the immune system weakens and the body
becomes vulnerable to other infections. This inter-
mediate stage was called AIDS-related complex
(ARC) and gives way to the lethal diseases that
characterize the final stages of AIDS, such as
Kaposi's sarcoma. Given that progression from HIV
to AIDS is a continuous process, the Centers for
Disease Control classifies an individual as having
AIDS when the CD4+ cell count falls below 200, the
normal count being about 1,000. At this stage, a
person may still be asymptomatic.

How Is AIDS Transmitted?
To date, five main transmission mechanisms have
been identified: transfusion of or contact with in-
fected blood, perinatal transmission from mother to
child, heterosexual intercourse, homosexual inter-
course, and intravenous drug use. In general terms,
the virus is not highly virulent. Blood transfusions
using infected blood products have the highest risk
of transmitting the virus. There is some indication
that HIV2 is less virulent than HIV1. HIV2 has been
identified for a longer period in West Africa than
HIV1, yet the infection levels of the latter are already
higher in the region.
Rates of perinatal transmission are in the range
of 15 to 40 percent. Factors that have been shown to

increase the risk of perinatal transmission include
the poor nutritional status of the mother, increasing
vitamin A deficiency in the mother, a generally poor
condition of the fetus, and certain characteristics of
the mother's HIV infection. Although the exact
method of transmission from mother to child is un-
known, it is known that the virus can pass through
the placenta, infecting the fetus in utero or during
birth. HIV2 is less likely than HIVI to be transmitted
from mother to child. Instances of transmission of
the infection have also been documented through
breast-feeding. At this stage, however, WHO still
recommends breast-feeding because the risks of
transmission are outweighed by the risks to infant
health from bottle feeding in areas of poverty, inade-
quate sanitation, and contaminated water supplies.
The risk of transmission from any single act of
sexual intercourse is slight-as low as 0.1 percent.
The risk of transmission is higher from male to
female, possibly 2.9 to 17.5 times higher, than from
female to male (Philipson and Posner 1993, 21). The
likelihood of transmission through sexual inter-
course varies with the nature of sexual activity: male
homosexual intercourse carries a higher risk, par-
ticularly ifcofactors such as genital ulcers and other
sexually transmitted diseases are present. Recent
evidence suggests that the nutritional status of indi-
viduals may also play a role.
The patterns by which HIV is spread vary con-
siderably from region to region. In America, trans-
mission is largely through homosexual or bisexual
intercourse or IV drug use: almost 90 percent of
AIDS patients are men, and 90 percent of those con-
tracted AIDS through homosexual intercourse.
Among women, 50 percent were infected through
intravenous drug use, the principal cause, followed
by 36 percent through heterosexual sex (Philipson
and Posner 1993, 27). In Sub-Saharan Africa, how-
ever, transmission through heterosexual intercourse
is most common: 94 percent of HIV infections are
transmitted through heterosexual intercourse, just
4 percent from blood transfusions, and 1 percent

from homosexual or bisexual intercourse or intrave-
nous drug use (U.S. Bureau of the Census 1994b, 7).
In the United States, the ratio of males to females
infected is 8.5:1; in Europe, 5:1; and in Africa, there is
evidence to suggest that infected women are begin-
ning to outnumber infected men (Mann, Tarantola,
and Netter 1992, cited in United Nations 1994). The
population group at risk in countries where transmis-
sion is mostly through heterosexual intercourse is far
larger than in countries where transmission is mostly
through homosexual intercourse, which makes target-
ing information and education campaigns very

What Is the Time Line
After infection, the body begins to develop antibodies
to the virus. This takes up to two weeks. Current HIV
testing methods actually test for the presence of these

antibodies rather than for the virus itself. The median
time between infection and sufficient antibodies to
result in a positive test is 2.1 months, with 95 percent
of infected persons testing positive within 5.8 months
(Horsburgh 1989). Although the infected person may
experience flulike symptoms on infection, he or she
will recover to an asymptomatic state. The latent stage
of HIV1 infection-the period of time when the
disease is not communicable-is very short but
appears to be longer for HIV2. HIV1 carriers are
capable of transmitting the disease almost from
infection, although the degree of infectivity varies.
The first three weeks after infection are a period of
high infectivity; infectivity then declines until about
60 weeks after infection, when an infected person
becomes highly infective again.
The median time for conversion from infection
to full-blown AIDS is approximately 10 years in the
United States, with about a third converting in
7 years (Taylor, Kuo, and Detels 1991; Bacchetti
and Jewell 1991). Evidence suggests that this time
line may be shorter in Sub-Saharan Africa.

Appendix 2: HIV Seroprevalence in Africa

Figure 3-African HIV1 seroprevalence for high-risk urban populations


/r c

Percent Seropositive

0.0 to 0.2

S0.3 to 2.5

F 2.6 to 10.0

10.1 to 25.0

S25.1 to 40.0

Over 40.0

No Data



Source: U.S. Bureau of the Census 1994a.



i.'- =~=;.
'' ''~''
i ).
;~.~ I~
ft~ ~:

Figure 4-African HIV1 seroprevalence for low-risk urban populations

%. '>% %. % %. \ ,\\ *N. \3
,l' / / / / / // / / ^ /./

Sj / / ^//
'X\X Jf f

Percent Seropositive

[ Less than 0.1


S0.2 to 1.0

S1.1 to 5.0

5.1 to 10.0

Over 10.0

SNo Data

Source: U.S. Bureau of the Census 1994a.

, ,.
A '/

Figure 5-African HIV2 seroprevalence for high-risk urban populations

Percent Seropositive
0.0 to 0.5

] 0.6 to 5.0

S5.1 to 7.5

S7.6 to 10.0

S10.1 to 12.5

Over 12.5

No Data

Source: U.S. Bureau of the Census 1994a.

Figure 6-African HIV2 seroprevalence for low-risk urban populations

Source: U.S. Bureau of the Census 1994a.

Appendix 3: Description of the Models

World Health Organization/
United Nations Forecasting Model,
Incorporating AIDS-Related
The WHO/UN model is the simplest model consid-
ered here; it requires only three data points: the
current-year estimate of HIV prevalence, the first year
transmission of HIV became widespread, and the
annual rate of progression from HIV infection to
AIDS. This Epi model assumes that cumulative HIV
infections follow a gamma curve.18 Knowing the
current year estimates of HIV prevalence and the year
in which HIV transmission became widespread
permits a reconstruction of the HIV epidemic curve
from the initial year. New HIV infections can then be
predicted by extrapolating this function. New adult
AIDS cases will occur in subsequent years in line with
the observed progression rate for HIV infection to
AIDS, and AIDS cases will result in death in line with
the observed progression rate from AIDS to death.
Estimates are also made of pediatric AIDS cases
transmitted from mother to child (vertically). The
number of births to infected adult women is esti-
mated on the basis of age-specific fertility rates for
infected women. The likely number of HIV-infected
births is then estimated by applying the probability
of vertical transmission to the number of births.
The forecasts of the Epi model are then incorpo-
rated into the existing life tables. The resulting modi-
fied life tables, which incorporate the probability of
dying from AIDS, are then used to project the future
population of countries significantly affected by the
HIV/AIDS pandemic.
The accuracy of the forecast provided by the Epi
model is dependent on the correct choice of func-
tional form for growth in prevalence rates of HIV,

the gamma curve, and requires that the underlying
parameters governing the incidence of HIV remain
the same-that is, that people do not change their
sexual practices or behavior. The forecasts assume
that there will be no new adult infections after 2005,
although mother-to-child transmission will continue
after this date. The long period of development from
HIV to AIDS, however, means that deaths as a result
of AIDS will continue for many years after 2005.

World Bank Model Incorporating
AIDS-Related Mortality
The World Bank method of incorporating the effect
of HIV/AIDS on population projections19 also fol-
lows a two-step process, projecting the path of the
AIDS epidemic and using the results to estimate
years of life expectancy lost (expressed as Coale
Demeny model life-table levels). These levels can
then be subtracted from the projected no-AIDS
trends to estimate future population sizes.
The first step in the World Bank methodology is
somewhat more complex than the Epi model used by
WHO. Three processes are modeled-the spread of
HIV, the progression to AIDS among those infected,
and the progression from AIDS to death-for two
population groups: adults over 15 years of age and
children. HIV transmission is assumed to occur sexu-
ally between partners, through blood transfusions,
through use of infected needles, and perinatally. Func-
tional relationships are specified for each process and
for different population subgroups. Then a simulation
model approach is used to map the epidemic. The
epidemiological model and the demographic model
are incorporated and run numerous times, varying the
behavioral parameters and using three different start-
ing dates for the epidemic-1970, 1975, and 1980-

18For further details about the Epi model, see Chin and Lwanga 1991. For details about the population forecasting model, see United
Nations 1993.
19For further details, see Bulatao 1991 and Bos and Bulatao 1992.

to achieve three arbitrarily fixed levels of HIV sero-
prevalence in 1990: 1 percent, 3 percent, and 6 percent
in the general population. By using all combinations
of parameters, a total of 243 simulations of population
projections in five-year intervals to 2025 were made.
Regression analysis, in logarithms, was then used to
establish the relationship between 1990 seropreva-
lence levels and future seroprevalence-level predic-
tions. Three variables were found to be important:
1990 HIV seroprevalence (In), 1990 HIV seropreva-
lence squared (In), and the first year of the epidemic.
A similar regression equation was used to establish the
relationship between adult HIV seroprevalence and
future mortality indicators. The final step was to use
Sub-Saharan African country-specific data with re-
spect to the year the epidemic started and the 1990
HIV seroprevalence levels and, using the regression
relationships estimated previously, forecast future
levels of HIV seroprevalence and the years of life
expectancy lost as a result of AIDS-related mortality.
Final population forecasts incorporating the effects of
HIV/AIDS were then achieved by subtracting the
number of years of life expectancy lost as a result of
AIDS, expressed as Coale Demeny model life-table
levels, in order that these modified life tables could be
subtracted from the projected no-AIDS trends in
World Bank population forecasts include AIDS-
related mortality for countries that have measurable
levels of HIV infection. It is assumed that all of the
Sub-Saharan African country forecasts discussed in
the text incorporate AIDS-related mortality.

The U.S. Bureau of the Census
Model Incorporating
AIDS-Related Mortality
The IWGAIDS model, developed by the Interagency
Working Group on AIDS Models and Methods of
the U.S. Department of State and used by the U.S.
Bureau of the Census to predict HIV/AIDS, is based
on a deterministic system of differential equations
that estimates the level of HIV infections through
blood transfusion, heterosexual sex with a long-term
partner, transmission through casual sex, and peri-
natal transmission.20 Varying the social and biologi-

cal factors that influence the deterministic equations
permits simulation of alternative HIV/AIDS-epidemic
scenarios. Unlike the United Nations and World
Bank approach of forecasting the spread of HIV/
AIDS and then incorporating AIDS-related mortality
into life tables in order to modify population projec-
tions, the IWGAIDS model starts from the demo-
graphic model used to make population projections
and incorporates AIDS-related processes. As a re-
sult, it takes into account changes in sexual behavior,
medical factors, and saturation of population groups.
The IWGAIDS model starts by developing three
alternative scenarios, using the demographic parame-
ters of Sub-Saharan Africa, and varying behavioral
parameters on the basis of various regional studies in
order to produce low-, medium-, and high-AIDS
epidemic paths representing different rates of spread
of the epidemic. Two HIV seroprevalence estimates
from different time periods, which were judged to be
most representative in terms of HIV seroprevalence
levels in low-risk urban populations, were then taken
for each country and used to calculate an annual rate
of growth in HIV infections. The three alternative
epidemic paths mapped initially were then used to
derive an interpolation factor that best matched the
actual growth path of a country-specific epidemic to
the growth path of one of the simulated epidemic
paths in the same period of time. Using this factor
with one of the simulated epidemic paths allowed a
country-specific epidemic path to be mapped. This
establishes the rate of growth of the epidemic in the
urban low-risk population over a period of 50 years
but does not link it to total country prevalence or to a
particular year.
Using both rural and urban data, an estimate was
made of the total adult HIV seroprevalence for each
country. This figure was then compared with the
total seroprevalence figure, which would be inferred
using the appropriate simulated epidemic path (low,
medium, or high) together with the interpolation
factor. Knowing the date of the actual HIV seropreva-
lence level allows an offset period to be calculated-
that is, the number of years that the simulated country-
specific epidemic would have taken to reach that
level of adult HIV seroprevalence.
The final stage of the process was to incorporate
AIDS-related mortality into the projections. This

20For further details, see Stanley et el. 1991.

was done by calculating AIDS-related age- and sex-
specific mortality rates for each of the three simulated
epidemic paths. Using the unique country-specific
interpolation factor with the offset period permitted
estimates of AIDS-related age- and sex-specific

mortality rates to be interpolated for each quinquen-
nium of the forecasting period. These were then
added to non-AIDS-related age- and sex-specific
mortality rates. These combined mortality rates were
then used to produce forecasts to the year 2020.


Ainsworth, M., and M. Over. 1994. AIDS and Afri-
can development. The World Bank Research
Observer 9 (2): 203-240.

Ainsworth, M., and A. A. Rwegarulira. 1992. Cop-
ing with the AIDS epidemic in Tanzania: Survi-
vor assistance. World Bank Africa Technical
Department, Technical Working Paper 6. Wash-
ington, D.C.: World Bank.

Allen, S., C. Lindan, A. Serufilira, P. van de Perre,
A. Rundle, F. Nsengumuremyi, M. Carael, J.
Schwalbe, and S. Hulley. 1991. Human immu-
nodeficiency virus infection in urban Rwanda:
Demographic and behavioral correlates in a rep-
resentative sample of childbearing women.
Journal of the American Medical Association
266 (12): 1657-1663. Cited in M. Ainsworth
and M. Over 1994.
Bacchetti and Jewell. 1991. Nonparametric esti-
mates of the incubation period of AIDS, based
on a prevalent cohort with unknown infection
times. Biometrics 47: 947-960, cited in Philipson
and Posner 1993.

Becker, C. M. 1990. The demo-economic impact of
the AIDS pandemic in Sub-Saharan Africa.
World Development 18 (12): 1599-1619.

Bloom, D. E., and J. V. Lyons. 1993. Economic
implications ofAIDS in Asia. New Delhi: United
Nations Development Program.

Bond, G. and J. Vincent. 1990. Living on the edge:
Structural adjustment in the context of AIDS. In
Uganda: Structural adjustment and change, ed.
H. Harser and M. Twaddle. London: James Currey.

Bos, E., and R. A. Bulatao. 1992. The demographic
impact of AIDS in Sub-Saharan Africa: Short
and long-term projections. International Jour-
nal of Forecasting 8: 367-384.

Bos, E., M. T. Vu, E. Massiah, and R. Bulatao. 1994.
1994 world population projections, 1994-95.
Washington, D.C.: World Bank.

Broomberg, J., P. Masoe, and M. Steinberg. 1991.
The economic impact of the AIDS epidemic in
South Africa. Economic Trends Research Group
Working Paper 5, Development Policy Research
Unit. Capetown: University of Capetown.
Chin, J. 1991. Global estimates of HIV infections
and AIDS cases: 1991. AIDS 5 (Supplement 2):
Chin, J., and S. K. Lwanga. 1991. The World Health
Organization approach: Projections of non-
pediatric HIV infection and AIDS in Pattern II
areas. In The AIDS epidemic and its demographic
consequences. New York: United Nations.
Cuddington, J. T. 1993a. Modeling the macro-
economic effects of AIDS, with an application
to Tanzania. World Bank Economic Review 7
(2): 178-189.
1993b. Further results on the macro-
economic effects of AIDS: The dualistic, labor
surplus economy. WorldBank Economic Review
7 (3): 403-417.
Cuddington, J. T., and J. D. Hancock. 1994. Assess-
ing the impact of AIDS on the growth path of the
Malawian economy. Journal of Development
Economics 43: 363-368.
Devereux, S., and G. Eele. 1991. The social and
economic impact of AIDS in East and Central
Africa. Report prepared for the United Nations
Development Programme, Food Studies Group.
Oxford University.
Forsthye, S., D. Sokal, L. Lux, and T. King. 1993. An
assessment of the economic impact of AIDS in
Kenya. Washington, D.C.: Family Health Inter-
Fylkesnes, K., H. Brunborg, and R. Msiska. 1994.
Zambia: The current HIV/AIDS situation--and
future demographic impact. Background Paper
1, The Socioeconomic Impacts of AIDS.
Lusaka, Zambia: Ministry of Health, Govern-
ment of the Republic of Zambia.

Gaisie, K., A. R. Cross, and K. Nsemulkila. 1993.
Zambia Demographic and Health Survey 1992.
Lusaka, Zambia, and Columbia, Md.: University
of Zambia, Central Statistical Office, and Macro
International Inc.

Ghana Health Assessment Team. 1981. A quantita-
tive method of assessing the health impact of
different diseases in less developed countries.
International Journal ofEpidemiology 10: 73-80.

Hassig, S. E., J. Perrins, E. Baende, M. Kahotwa, K.
Bishagara, N. Kinkela, and B. Kapita. 1990. An
analysis of the economic impact of HIV infec-
tion among patients at Mama Yeo Hospital, Kin-
shasa, Zaire. AIDS 4: 883-887.

Horsburgh, C. R. 1989. Duration of Human Immuno-
deficiency Virus infection before detection of
antibody. The Lancet 2: 637-638.

Kambou, G., S. Devarajan, and M. Over. 1992. The
economic impact of AIDS in an African coun-
try: Simulations with a computable general
equilibrium model of Cameroon. Journal ofAf-
rican Economies 1 (1): 109-130.

Lal, S., and B. B. Thakur. 1995. The problem of HIV
and AIDS in India. Current Science 69 (10):

Lewis, M., G. M. Kenney, A. Dor, and R. Dighe.
1989..AIDS in developing countries. Washing-
ton, D.C.: The Urban Institute.

Liomba, G., M. Kandulu, and V. Kandiado. 1994.
Family and community care situation analysis
report. Malawi: United Nations Children's

Mann, J., J. M. Tarantola, and T. W. Netter, eds.
1992. AIDS in the world. Cambridge, Mass.:
Harvard University Press.

Mtera, K. N. M. 1992. The current AIDS situation in
the Kagera Region. In Behavioural and
epidemiological aspects of AIDS research in
Tanzania. Proceedings from a workshop held in
Dar es Salaam, Tanzania, December 6-8, 1989,
Chapter 4.

Noble, G. R. 1989. Centers for Disease Control's
International AIDS activities. Paper presented at
the African Studies Association Meeting, At-
lanta, Ga., November.

Over, M. 1992. The macroeconomic impact ofAIDS
in Sub-Saharan Africa. Africa Technical De-
partment, Population, Health, and Nutrition Di-
vision Technical Working Paper 3. Washington,
D.C.: World Bank.

Over, M., S. Bertozzi, J. Chin, B. N'Galy, and
K. Nyamuryekunge. 1988. The direct and indi-
rect cost of HIV infection in developing coun-
tries: The cases of Zaire and Tanzania. In The
global impact ofAIDS, ed. A. F. Fleming et al.
New York: Alan R. Liss.

Pallangyo, K., and R. Laing. 1990. Alternative ap-
proaches to managing opportunistic illnesses of
HIV infected persons: Costs and burden on the
Tanzanian health care system. Background pa-
per for Tanzania AIDS assessment. World Bank,
Washington, D.C.

Philipson, T. J., and R. A. Posner. 1993. Private
choices and public health. Cambridge, Mass.:
Harvard University Press.

Pinstrup-Andersen, P. 1994. World food trends and
future food security. Food Policy Report. Wash-
ington, D.C.: International Food Policy Research

Rosegrant, M. W., M. Agcaoili-Sombilla, and N. D.
Perez. 1995. Global food projections to 2020:
Implications for investment. Food, Agriculture,
and the Environment Discussion Paper 5. Wash-
ington, D.C.: International Food Policy Research

Ruttan, V. 1994. Agriculture, environment, and
health: Sustainable development in the 21st cen-
tury. Minneapolis, Minn.: University of Minne-
sota Press.

Ryder, R. W., and R. D. Mgerwa. 1994. The clinical
definition and diagnosis of AIDS in African
adults. In AIDS in Africa, ed. M. Essex,
S. Mboup, P. J. Kanki, and M. R. Kalengayi.
New York: Raven Press. Cited in M. Ainsworth
and M. Over 1994.

Ryder, R. W., V. L. Batter, M. Nsuami, N. Badi,
L. Mundele, B. Matela, M. Utshudi, and
W. L. Heyward. 1991. Fertility rates in 238
HIV-1 seropositive women in Zaire followed for
three years postpartum. AIDS 5: 1521-1527.

Semba, R. D., P. G. Miotti, J. D. Chipwangwi,
A. J. Saah, J. K. Canner, G. A. Dallabetta, and
D. R. Hoover. 1994. Maternal vitamin A defi-
ciency and mother to child transmission of HIV-1.
The Lancet 343 (June 25): 1593-1596.

Serwadda, D., M. J. Waver, S. D. Musgrove, N. K.
Sewankambo, J. E. Kapland, and R. H. Gray.
1992. HIV risk factors in three geographic strata
of rural Rakai District, Uganda. AIDS 6: 983-
989. Cited in M. Ainsworth and M. Over 1994.

Shepard, D. S., and R. N. Bail. 1991. Costs of care
for persons with AIDS in Rwanda. Paper pre-
pared for the World Health Organization, Global
Program on AIDS. Geneva.

Smallman-Raynor, M. R., and A. Cliff. 1992. Sea-
sonality in tropical AIDS: A geographical analy-
sis. International Journal of Epidemiology 21
(3): 547-556.

Sranecki, K. A., and P. O. Way. 1994. Review of
HIV spread in Southern Africa. Report prepared
for the U.S. Agency for International Develop-
ment by the Center for International Research,
U.S. Bureau of the Census, Washington, D.C.

Stanley, E. A., S. T. Seitz, P. O. Way, P. D. Johnson,
and T. F. Curry. 1991. The IWG AIDS model for
the heterosexual spread of HIV and the demo-
graphic impact of the AIDS epidemic. In The
AIDS epidemic and its demographic conse-
quences. New York: United Nations.
Stover, J. 1993a. The impact ofHIV/AIDS on popu-
lation growth in Africa. Paper prepared for the
African Population Advisory Committee. World
Bank, Washington, D.C., May. Processed. Cited
in M. Ainsworth and M. Over 1994.

1993b. The impact of HIV/AIDS on adult
and child mortality in the developing world. Paper
presented at the International Union for the Scien-
tific Study of Population Seminar on AIDS Impact
and Prevention in the Developing World: The
Contribution of Demography and Social Science.
Veyrier-du-Lac, France, December 5-9. Pro-
cessed. Cited in M. Ainsworth and M. Over 1994.

Taylor, J. M. G., J.-M. Kuo, and R. Detels. 1991. Is
the incubation period of AIDS lengthening?
Journal of Acquired Immune Deficiency Syn-
dromes 4: 69-70, cited in T. J. Philipson and
R. A. Posner 1993.

United Nations. 1991. The AIDS epidemic and its
demographic consequences. ST/ESA/SER.A/119.
New York.

1993. World population prospects: The
1992 revision. Geneva.

1994. AIDS and the demography ofAfrica.
ST/ESA/SER.A/137. New York.

UNDP (United Nations Development Programme).
1993. AIDS and Asia: A development crisis.
New York.

U.S. Bureau of the Census. 1994a. HIV/AIDS sur-
veillance database. Population Division, Inter-
national Programs Center, Washington, D.C.
Computer printout.

1994b. The impact ofHIV/AIDS on world
population. Washington, D.C.: U.S. Govern-
ment Printing Office.

1994c. Recent HIV seroprevalence levels
by country: December 1994. Research Note 15.
Washington, D.C.: Health Studies Branch, Inter-
national Programs Center, Population Division.

1994d. Trends and patterns of HIV/AIDS
infection in selected developing countries.
Country Profiles, December 1994. Research
Note 15. Health Studies Branch, Center for
International Research, Washington, D.C.

Viravaidya, M., S. A. Obremskey, and C. Meyers.
1993. The economic impact of AIDS on Thai-
land. In Economic implications ofAIDS in Asia,
ed. D. E. Bloom and J. V. Lyons. New Delhi:
United Nations Development Program.

Way, P. 0., and K. A. Stanecki. 1993. An
epidemiological review of HIV/AIDS in Sub-
Saharan Africa. Washington, D.C.: Center for
International Research, U.S. Bureau of the Census.

Whiteside, A. 1994. AIDS, poverty, and structural
adjustment-Towards a "pro-poor" develop-
ment strategy. AIDSAnalysis Africa 4 (1): 10-11.

n.d. HIV infection and AIDS in Zimbabwe:
An assessment. South African Foundation for
Economic Research, Durban, South Africa.

World Bank. 1994. World development report 1994.
New York: Oxford University Press.

1995. The World Bank and HIV/AIDS: In
partnership to combat an unparalleled threat
to society and development [online] [cited
December 5]. Available: World Wide Web:
WHO (World Health Organization). 1994. The cur-
rent global situation of the HIV/AIDS pandemic.
Global Program on AIDS. Geneva.

1995. WHO/global program on AIDS.
Available on World Wide Web at http://gpawww.

Yang, B.-M. 1993. The economic impact of AIDS
on the Republic of Korea. In Economic impli-
cations of AIDS in Asia, ed. D. E. Bloom and
J. V. Lyons. New Delhi: United Nations Devel-
opment Programme.

1200 SEVENTEENTH STREET, N.W. WASHINGTON, D.C. 20036-3006 U.S.A. 1-202/862-5600
FAX 1-202/467-4439 E-MAIL ifpri@cgnet.com WEB: http://www.cgiar.org/ifpri


Food, Agriculture, and the Environment Discussion Papers

1. Alleviating Poverty, Intensifying Agriculture, and Effectively Managing
Natural Resources, by Per Pinstrup-Andersen and Rajul Pandya-Lorch, 1994

2. Sociopolitical Effects of New Biotechnologies in Developing Countries, by
Klaus M. Leisinger, 1995

3. Africa's Changing Agricultural Development Strategies: Past and Present
Paradigms as a Guide to the Future, by Christopher L. Delgado, 1995

4. A 2020 Vision for Food, Agriculture, and the Environment in Sub-Saharan
Africa, edited by Ousmane Badiane and Christopher L. Delgado, 1995

5. Global Food Projections to 2020: Implications for Investment, by Mark
Rosegrant, Mercedita Agcaoili-Sombilla, and Nicostrato Perez, 1995

6. A 2020 Visionfor Food, Agriculture, and the Environment in Latin America,
edited by James L. Garrett, 1995

7. Agriculture, Trade, and Regionalism in South Asia, by Dean A. DeRosa and
Kumaresan Govindan, 1995

8. Major Natural Resource M.anagement Concerns in South Asia, by Gerard J.
Gill, 1995

9. Agriculture, Technological Change, and the Environment in Latin America:
A 2020 Perspective, by Eduardo J. Trigo, 1995

10. Overcoming Malnutrition: Is There an Ecoregional Dimension? by Manohar
Sharma, Marito Garcia, Aamir Qureshi, and Lynn Brown, 1996

11. Structural Changes in Demandfor Food in Asia. by Jikun Huang and Howard
Bouis, 1996

12. Middle East Water Conflicts and Directions for Conflict Resolution, by
Aaron T. Wolf, 1996

13. The Transition in the Contribution of Living Aquatic Resources to Food
Security, by Meryl Williams, 1996

14. Land Degradation in the Developing World: Implications for Food, Agricul-
ture, and the Environment to 2020, Sara J. Scherr and Satya Yadav

Lynn R. Brown is a research analyst at the International Food Policy Research Institute.

with permission of the
International Food Policy
Research Institute

University of Florida Home Page
© 2004 - 2010 University of Florida George A. Smathers Libraries.
All rights reserved.

Acceptable Use, Copyright, and Disclaimer Statement
Last updated October 10, 2010 - - mvs