mm" ‘P

‘3‘

 

afﬂ'NQW'"iNIFV'PJW' WNW!

 

 

 

 

.=.

Eng.

, .
93‘

are). t. I
29:...”lﬁuwv"
3IHQ‘..|’I

13:”..0...‘

'1 -‘~ I

3)... v... ..

55.! . . .
333.”; ”nan"... 2:.
(n... .. a? nu...

2:5”! V,

4:9... It...ri.\n\...un..huuxu§.i.v
.33... .i :3: smut»... 2.1
x: . .. 1.

.I‘rts :\
1. ‘3'}. cl

, u .t in!
.0! 5‘

 

 

 

 

I‘hbSlS
3'— Ll ”+3 Di ARY
; o r 9 Michigan State
U: liversaty
This is to certify that the

dissertation entitled

THE POLITICS OF ANTI-RETROVIRAL DRUGS IN AFRICA

presented by

ERIC RYAN LITTLE

has been accepted towards fulﬁllment
of the requirements for the

PhD. degree in Political Science

 

 

MMI—I—o—u

 

Major Professor's Signature
3 I ‘3 I <0

 

Date

MSU is an Afﬁrmative Action/Equal Opportunity Employer

 

0.-.-.-u-o-o-.—v-.-.-.-._v_A_r- ‘

PLACE IN RETURN BOX to remove this checkout from your record.
To AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE

DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

5/08 KIProj/Acd-Pros/CIRC/DateDueJndd

 

THE POLITICS OF ANTI-RETROVIRAL DRUGS IN AFRICA

By

Eric Ryan Little

A DISSERTATION

Submitted to
Michigan State University
in partial fulﬁllment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY
Political Science

2010

ABSTRACT
THE POLITICS OF ANTI-RETROVIRAL DRUGS IN AFRICA
By

Eric Ryan Little

This dissertation investigates the politics surrounding anti-retroviral drugs in Africa.
The ﬁrst article analyzes cross-national variation in ARV coverage. I report that state
capacity is the most important predictor of more pervasive AIDS treatment programs, and
regime type is not empirically important. The second article looks at sub-national
variation in South Africa and Nigeria. I found that higher HIV prevalence and urban
areas are more likely to have ARV resources. Electing members of the ruling party to
executive and legislative positions makes it more likely that sub-national units will have
ARV programs, suggesting that patronage networks are important predictors of ARV
availability. The third article investigates citizen satisfaction with policy performance at
the individual level using Aﬁobarometer data from twenty countries. Round 4 of the
Afrobarometer asked more than 27,000 Aﬁican citizens in 20 countries about their
opinions, and their evaluation of their government’s HIV/AIDS policy performance
allows for the micro-level relationships to be investigated and modeled. Regression
analysis suggests that the six most important predictors of citizens satisfaction with
HIV /AIDS policy performance (in order) are health service satisfaction, perceived
corruption, trust in the state’s institutions, satisfaction with democracy, low HIV

prevalence, and high ARV coverage.

Copyrighted by
ERIC RYAN LITTLE
201 O

ACKNOWLEDGEMENTS

I would like to thank Dr. Michael Bratton, Dr. Sandy Schneider, Dr. Ravi Bhavnani,
and Dr. Michael Rip for their service on my guidance committee. I would also like to
thank my fellow graduate students as well for their assistance, advice, and
recommendations. Finally, and most importantly, I would like to thank my wife Maria

for her constant love and support throughout the dissertation process.

TABLE OF CONTENTS

LIST OF TABLES ................................................................................... vi
LIST OF FIGURES ................................................................................. vii
INTRODUCTION .................................................................................... 1
ARTICLE 1

The Political Determinants of ARV Coverage in Africa: A Cross National Analysis of
States and Regimes ................................................................................. 18
ARTICLE 2

The Sub-National Politics of ARV Provision ................................................... 57
ARTICLE 3

The Micro-Foundations of Public Satisfaction with HIV/AIDS Policy Performance:
Lessons ﬁ'om Aﬁ'obarometer Round 4 ......................................................... 108
CONCLUSION ..................................................................................... 167

LIST OF TABLES

TABLE 1.1
OLS Regression Results for ARV Coverage .................................................... 39

TABLE 2.1
HIV/AIDS Statistics and Features ................................................................ 59

TABLE 2.2
GDP and AIDS Data from South Africa and Nigeria .......................................... 76

TABLE 2.3
OLS Regression Results for Citizens Per Hospital ............................................. 93

TABLE 3.1
Frequency of Citizen Responses Regarding Government HIV/AIDS Policy
Performance ........................................................................................ 137

TABLE 3.2
Descriptive Statistics for Independent Variables .............................................. 146

TABLE 3.3
OLS Regression Results for Citizen Satisfaction with HIV/AIDS Policy Performance.l49

vi

LIST OF FIGURES

FIGURE 1.1

States and Regimes Considered Together ........................................................ 19
FIGURE 1.2

Scatter Plot of Polity Scores and ARV Coverage ............................................... 41
FIGURE 1.3

3-D Scatter Plot of Polity Scores, Stateness Index, and Logged ARV Coverage .......... 43

FIGURE 1.4

Summary of Results and Analysis ................................................................ 46
FIGURE 1.5

Scatter Plot of Unstandardized Residuals and ARV Coverage ............................... 49

FIGURE 2.1

Key Political Similarities and Differences Between South Africa and Nigeria ............ 76
FIGURE 2.2

Map of ARV Programs in Nigeria ................................................................ 90
FIGURE 2.3

Map of ARV Programs in South Africa ......................................................... 91

FIGURE 3.1

Summary of Results and Analysis ............................................................... 157

vii

The Politics of ARVs in Africa: An Introduction

AIDS is undoubtedly one of the most staggering problems facing policy makers
today around the globe. Recent data suggests that over thirty million world citizens are
infected with the disease, of which over twenty two million are Africans (UNAIDS,
2009). The political and economic consequences stemming ﬁom the disease are
staggering and have necessitated the largest international response to any health issue in
the history of the world. This mobilization effort has involved domestic, bilateral and
international organizations working together as part of an unprecedented effort.
Responding to this crisis requires a matrix of policies capable of preventing further
spread of the disease, testing people to ensure people know their status, and treating
those already infected. Various pieces of the prevention response include marketing
campaigns to educate the public about the disease, health education as part of academic
curriculums, and the widespread availability of AIDS testing so that people know their
status. Male circumcision to reduce transmission and provision of condoms are also part
of this template. Even changing laws to decriminalize homosexuality, needle exchange
programs, and blood testing for donors are part of best practices to reduce transmission
for high risk groups. Over time response to the disease is beginning to transition from a
piece meal crisis response to a more stable, robust approach with the goal of ensuring
universal access to all of the primary policy interventions: treatment, prevention,
testing, and care (Universal Access, 2009).

The multiple facets of this problem can be seen from the varying actors and
organizations involved in this crisis: individual citizens, civil society groups,

bureaucracies, state elites, international organizations, foreign donors, political leaders,

and multi-national pharmaceutical manufacturers are all involved in the HIV/AIDS
policy area. From Washington to Geneva and Cairo to Cape Town, AIDS in Africa and
what should be done about it has garnered enormous resources and attention in ways that
many other African issues have been muted. Despite the longstanding prevalence of the
AIDS crisis and its widespread consequences, political scientists have not investigated
the politics surrounding the disease thoroughly. AIDS has an impact on every issue of
African politics from agriculture to tax policy to institutional capacity, yet the literature
is thin and largely undeveloped when one seeks encompassing explanations.

In 2001, Catherine Boone wrote in Africa Today, “Political Science as an academic
discipline has been slow in grappling with the enormous implications of the AIDS crisis
for much of the developing world,” (Boone and Batsell, 2001). Boone presented a series
of AIDS related sub-topics which political scientists could offer valuable research. As
she put it, “Five research agendas for Africa are: variations in state response to the
pandemic; the relationship between governments and NGOs; the AIDS challenge to neo-
liberalism; AIDS and North-South tensions; and connections between AIDS and
international security issues,” (Boone and Batsell, 2001). The research plan below will
focus heavily on the ﬁrst of Boone’s recommendations, focusing on variation in the state
response through ARV treatment at multiple levels. Before discussing the political
puzzle, I will ﬁrst discuss the nature of AIDS treatment, the international response, and
argue for why these resources ﬁt somewhere between public and private goods.

What is AIDS treatment?
AIDS is treated with antiretroviral drugs (ARVs) which are deﬁned as follows in

The Encyclopedia of Medicine:

Antiretroviral drugs inhibit the production of retro-viruses—viruses composed of
RNA rather than DNA. The best known of this group is HIV, the causative agent for
AIDS. Antiretroviral drugs are virustatic agents which block the replication of the virus.
The drugs are not curative; however continued use of these drugs, particularly in multi-
drug regimens, signiﬁcantly slow disease progression, (Antiretroviral, 1).

While these drugs cannot kill HIV, they slow its progression and can allow people to
live for up to twenty years after HIV diagnosis. While ARVs are highly effective, they
are also quite costly. This cost has dropped considerably ﬁ'om over $12,000 per
person/year to between $120 and $500, but is still outside the reach of most individuals,
let alone the poorest of the poor (Parker, 2007). As prices have continued to decrease,
donors, activists, and members of the international community have urged the concept of
universal access. According to Dr. Stephen Morrison from The Center for Strategic and
International Studies, 2003 to 2008 represented a crisis response effort by the
international community, led by the United States which resulted in millions of Aﬁicans
receiving ARVs which are beginning to transition to a more stable and sustainable set of
goals (Morrison, 2009). Millions more have also received treatment via other
institutions including the World Health Organization (WHO) and United Nations AIDS
program (UNAIDS).

ARVs are a considerable expense to whoever is footing the bill; however, a 2004
South Aﬁican report indicated that the cost of ARV therapy was actually cheaper than
not providing such treatment (Badri et. a1, 2006). According to World Health
Organization estimates, only 28% of Aﬁicans in need of these drugs have been able to
access them despite the massive scale up efforts (Parker, 2007). Cross country variation

is rampant as countries like Somalia have less than 1% ARV coverage, while Botswana

has over 75% coverage (WHO, 2008). Comprehensive coverage data for Aﬁican

countries is available ﬁom the World Health Organization. While the wealthy can buy
these drugs on the international market, it is largely the poorest of the poor in Africa
who are disproportionately unable to access these drugs.

Treatment Versus Prevention?

The relationship between treatment and prevention has evolved from one of
contention to mutually beneﬁcial. No one denies that prevention is always preferable to
treatment. In order to successfully reduce the number of people infected that prevention
is the long term answer if prevalence rates are to fall. This is true for individual patients
as well as for policy makers. Treating AIDS patients is extremely costly for developing
countries, and there is no doubt that without international support that treatment would
only be accessible by wealthy individuals. There have, however, been major changes in
how treatment is viewed as it has transitioned from impossibility in developing countries
to an attainable and reachable goal.

According to renowned AIDS expert Paul Farmer, in 2001 an ofﬁcial in the US.
Department of Treasury objected to distributing AIDS treatment in Africa because
Africans would not understand the concept of time and would be unable to take their
treatment at prescribed increments (Farmer, 2001). He put the availability of these
medications in human rights terms when he wrote that, “We should be increasingly
reluctant to reserve these therapies for the afﬂuent, low-incidence regions of the world
where most medical resources are concentrated. Excellence without equity looms as the
chief human-rights dilemma of health care in the 21St century,” (Farmer, 2001).

From such inauspicious beginnings, treatment has provided inarguable results.

The ﬁrst African public clinic providing AIDS medications was built in Gaborone,

Botswana in January of 2002 (Baragona, 1). As of late 2008, 44 percent of Africans in
need of treatment were receiving these medications and an estimated 2.3 million African
years of life had been added due to their availability, according to UNAIDS statistics.
What can explain this transformation? A group of Harvard medical and epidemiological
experts raised the following four points for combining treatment with prevention to
maximize the overall response effectiveness (Adams et a1, 2001):

1. Treatment is essential to the 36 million people already infected with HIV, the vast
majority of

whom will die of AHJS without it. This is the immediate humanitarian rationale for
treatment.

2. Treatment is necessary to optimize prevention eﬁ’orts. When treatment is not
available, less ‘

incentive exists for an individual to take an HIV test, since HIV-positive status not only
rs

associated with social stigmatization but also is tantamount to a death sentence. It is only
when HIV testing is coupled with treatment that people have an incentive to be tested,
thus

enabling a rational response to AIDS: primary prevention for those who are HIV
uninfected,

and antiretroviral treatment for those who are HIV infected. Effective antiretroviral
treatment of HIV-positive people also lowers the viral load within infected individuals,
which in turn has a major effect in reducing the likelihood that they will transmit HIV
infection to others (UN AIDS, 1999, Hart et a1. 1999, Vemazza et a1. 2000). Ultimately,
then, appropriate treatment of infected individuals may become a major tool in AIDS
prevention.

3. Treatment is necessary to save the children -— and fabric -- of societies. Without
treatment,

the number of adult deaths expected from AIDS is so great that the currently
catastrophic

ﬁgure of 13.2 million AIDS orphans will grow into an even more socially devastating
wave

in coming years (U SAID, 2000). Without family support, these children often can not
attend school, suffer from poverty and malnutrition, and become victims of violent and
sexual crimes—all of which places them at high risk for acquiring AIDS and which
threatens to mire them in increasingly desperate conditions. If the current lack of
treatment continues, a demographic shift is predicted in the most severely afﬂicted parts
of Africa such that teenagers will outnumber their elders by 2020 (U .S. Census, 2000).

This demographic shift may contribute directly to increase political instability and
violence.

4. Treatment is necessary for continuing economic development. Without treatment,
millions of

adults in the prime of their working lives will die of AIDS and take with them the skills
and ‘

knowledge base that are necessary for human and economic development (Bonnel,
2000). For example, in Zambia teachers are dying of AIDS almost as quickly as they are
trained (UNICEF, 2000). The loss of skilled workers is a major reason why AIDS
will seriously reduce the rates of future economic growth (Bonnel, 2000). The goal
of simply preventing new HIV infections, without simultaneously offering treatment to
prolong the lives of those already infected, has proved insufﬁcient to appreciably
mitigate these trends. Despite these arguments and despite the proven efﬁcacy of
presently available therapies, antiretroviral drug treatment remains inaccessible to most
of the world’s infected population.

As argued by these scientists, instead of thinking of treatment and prevention as an
either/or proposition, an integrated comprehensive response is necessary as neither
treatment or prevention alone is capable of greatest effectiveness. As Lamptey and
Wilson (2005) state, these treatment and prevention responses are mutually reinforcing
and have positive feedback loops for one another. Increased prevention leads to more
affordable and sustainable treatment, and treatment makes prevention more accessible
and effective as citizens are more likely to get tested with the availability of treatment.
This allows health care workers greater opportunities to discuss prevention messages.
The authors also cite that increased investment in infrastructure associated with
enhancing treatment facilities can also beneﬁt overall health systems. These authors
also state that treatment is responsible for a short-term decline in AIDS deaths which
buys time for prevention efforts to improve in effectiveness before concluding that both
prevention and treatment require greater resources to have greatest impact. Also worthy

of note are funding expansions treating other diseases based on early experiences by aid

workers focused on AIDS. For instance the 2008 reauthorization for the President’s

Emergency Plan for AIDS Relief (PEPFAR) which earmarked $39 billion for AIDS
ﬁmding also included provisions for $5 billion to ﬁght malaria as part of the President’s
Malaria Initiative as well as $4 billion for tuberculosis treatment (PEPFAR, 2009). This
evidence suggests that although AIDS may be more expensive to treat than other
diseases, initial successes and learning by policy practitioners in the ﬁeld have led to
more expansive efforts against other diseases.

Further bolstering the argument made by the Harvard experts is a series of articles
that have investigated the cost effectiveness of AIDS treatment. One of the arguments
against treatment is that it is costly compared to prevention or treating other curable
diseases that are equally deadly in developing countries (Creese et a1 2002). According
to Freedberg (2001) treatment is a cost effective use of resources. Similar research has
shown that AIDS treatment is cost effective as part of the overall response in Cote
d’Ivoire (Goldie et a1, 2006) and South Africa (Badri et al 2006). It is worth
emphasizing that these articles were published in the prestigious publications like The
New England Journal of Medicine, and that the overall trend in reviewed literature
indicated support for treatment as part of a wider effort; these scientiﬁc ﬁndings dovetail
with the international community’s adopting this consensus that treating AIDS patients
is one of the key pillars of their overall effort.

Goals and Justification of the AIDS Response

Based on these ﬁndings that treatment was cost effective, leading international
organizations, working with country partners set some ambitious goals prioritizing
treatment as part of their overall effort. According to Lamptey and Wilson (2005) the

World Health Organization (WHO) set the goal of providing treatment to three million

people in low and middle income countries by 2005. The Global Fund to Prevent AIDS,
Tuberculosis, and Malaria (GFATM) sought to treat 1.6 million people by 2007 in
addition to efforts to provide over ﬁfty million AIDS tests and comprehensive care for
over one million AIDS orphans. Finally, the US. President’s Emergency Plan for
HIV/ADDS Relief (PEPFAR) sought to treat an additional two million people in addition
to efforts to prevent seven million people ﬁ'om becoming infected and providing care for
ten million additional HIV infected patients and orphans. With the ﬁrst treatment clinic
in Africa built in 2002, the speed of this escalation and the size of these goals were
especially grand.

How were these goals justiﬁed by the international community? Stuckler and McKee
(2008) discussed in The Lancet ﬁve key metaphors that apply to the increased focus on
public health, and each applied to AIDS more speciﬁcally. They included
considerations of how the disease threatened western foreign policy interests like
increased trade, economic growth, and the stability of countries where AIDS is
especially prevalent like sub-Saharan Africa. This explains advocacy for these policies
by institutions like the Department of State and USAID among others. Similar to
foreign policy was the notion that AIDS threatened security across the Aﬁican continent
with a common conception that AIDS could undermine African militaries (Whiteside et.
a1. 2006). This resonated particularly strongly with Bush Administration ofﬁcials and
US. Center for Disease Control ofﬁcials who urged treatment support. Another
metaphor for this investment was relief as charity which explains the focus on the issue
by philanthropic groups like the Gates Foundation and other non—govemmental

organizations. Further validating these expenditures were groups advocating a health

response to promote investment. Groups worried that the virus could threaten economic
development in developing states, and organizations like the IMF, World Bank, and the
private sector supported health interventions along these lines according to Stuckler and
McKee (2008). A ﬁnal rationale for these interventions was the provision of global
health as public health which explains the international support by institutions including
the WHO and UN which established UNAIDS to speciﬁcally deal with the disease.

These ﬁve metaphors and the wide cast of advocates for these appropriations go a
long way to explain the speed with which AIDS treatment was justiﬁed in a variety of
venues. These justiﬁcations and initial successes also are dominant in explaining why,
despite the current global ﬁnancial crisis that AIDS treatment coverage continues to
escalate rather than being reduced in scope. This is a sharp break with other past health
interventions which largely disappeared in lean times when only the global health as
charity narrative was used. PEPFAR for instance started off as a $15 billion dollar
program that at the time of its launch seemed like little more than a State of the Union
promise by President Bush. By the time it was reauthorized in 2008, it was expanded to
include $39 billion for AIDS alone, in addition to expanded commitments for malaria
and tuberculosis (PEPFAR, 2009).

ARVs: What Type of Goods Are They?

What is especially helpful ﬁom a social scientiﬁc perspective is the fact that the
international community, through the various international and bi-lateral institutions
offered health care funding support in every viable Aﬁican country, according to WHO
data. This means an inaccessible country like Somalia lacked AIDS treatment, but these

appropriations were made in countries whether they were considered friends or foes of

the western world. With this support, one can make the case that treatment was made
available and Aﬁican governments all had the possibility of accepting this aid. There
were countries that received greater or lesser aid, but it is worth noting that the overall
treatment of international health aid was advocated for all countries, and the availability
of treatment was universally available for individual states to seek through a variety of
mechanisms including international grants and diplomatic efforts to garner bilateral aid.
Also, all countries beneﬁt from the knowledge gained and economies of scale which
have resulted in the prices of the required medications dipping dramatically per capita as
these efforts escalate (Parker, 2007).

Based on these commitments, and the fact that these programs continue to expand
despite the international ﬁnancial crisis, I would argue that international funding for
AIDS is here to stay and that AIDS treatment is no longer exclusively a private good.
Based on the wide variety of narratives used to justify public health and the fact that
international assistance for health care has been accepted by more than forty African
states, I would argue that these resources are available to some degree to every African
nation-state. Still, with no country achieving 100 percent coverage, they cannot be
considered a public good. How can we categorize this good that ﬁts somewhere
between public and private? Kapstein and Busby describe a term known as merit goods
which they describe as, “the transformation of ARVs from private goods, which only a
few victims of AIDS could afford, into merit goods or entitlements, deﬁned as goods
that should be made available to everyone, irrespective of their ability to pay for them,”
(2009, 2). As the Harvard experts advocate, restricting access to only some portion of

patients able to pay for their own care will only allow the disease to spread more widely,

10

making exclusion detrimental to the overall goal of reducing further spread of the
disease. Available data also indicates that despite the global ﬁnancial crisis, resources
have expanded rather than ebbed in ﬂow showing the non-subtractable nature of
resources provided for ﬁghting the AIDS crisis. Chen et. a1. (1999) argue that public
health in the globalized age has become a public good, but ARV availability could only
be qualiﬁed as a public good once coverage rates reach 100 percent. This is
commensurate with Farmer’s advocacy for increased availability of treatment based on a
human rights based approach to this set of issues, which has emerged as'an international
norm as described by Kapstein and Busby (2009). This concept of merit goods will be
used throughout the dissertation.
Domestic Ownership

Equally important is the agency with which countries could accept or reject this aid.
As Jackson and Rosberg (1982) argued, the juridicial characteristics of statehood would
prevent international intervention without at least tacit support from the individual states
in question. As an example of this domestic ownership, despite PEPFAR being a
bilateral, U.S. funded initiative, eighty six percent of the organizations that actually act
on the ground to carry out these supported projects in 2008 were indigenous
organizations (PEPF AR, 2009). South Africa under Mbeki vehemently denied the
linkage between HIV and AIDS, but this did not prevent the country ﬁ'om accepting
international aid, according to Patterson (2006). These statements were likely
responsible for South Africa receiving less aid per capita than more proactive regimes
like Uganda where political leadership was quicker to accept the underlying science of

the disease. This probably explains why Uganda received the largest amount of

11

international assistance per AIDS patient in 2004 among Aﬁican countries (Patterson,
- 166).

While international support by the NGOs dominated the headlines, domestic
governments have also prioritized spending on health. According to WHO data, Aﬁican
states have taken on a large share of the bill for health in their countries, and health
accounts for more that 5% of GDP spending on average by Aﬁican governments (WHO,
2008). The international organizations involved has strongly emphasized that the
overall response effort is nation-state owned and led (UN AIDS, 2009). Based on these
and other examples that will be provided in the dissertation, I argue that despite focus on
the role of foreign assistance, nation-states are in control of their own treatment
programs, and greater or lesser international resource availability is both directly and
indirectly inﬂuenced by the choices made by domestic governments. The South
Africa/Uganda divide is an example of how domestic differences resulted in different
levels of commitment by the international community. The commitments by Aﬁican
regimes to AIDS policy prioritization despite the universal availability of at least some
resources by international institutions is the puzzle to be examined at several levels in
the dissertation. With Aﬁican governments owning responsibility for the AIDS
response, and international institutions making these health resources available in some
degree to all countries, why has there been such dramatic differences in levels of success
scaling up AIDS treatment?

The Dissertation Puzzle
Several of the countries most successful at ﬁghting the disease are among the

weakest in economic and state capacity (Patterson 2006, 28). Why for instance an

12

economically weaker country like Botswana has the most robust treatment of any
country with 76% of needy AIDS patients getting treated according to the WHO’S 2006
data. Economically stronger and with greater health capacity, South Africa lags with
barely more than 20% of AIDS patients having similar access. Explaining these
divergent cross-national outcomes is the goal of the ﬁrst article in this project.

Before proceeding, 1 will ﬁrst articulate my speciﬁc research questions and deﬁne the
part of the wide AIDS ﬁeld that I will investigate. First, this project will explore the
following question: at the country level, what are the political determinants of ARV
coverage in the African context; in other words, why have some countries been able to
treat more patients than others? A second research question disaggregates the analysis
to the sub-national level: what explains the distribution of ARV resources when they are
not available to all patients who require them? Third, and ﬁnally, at the individual level
what explains satisfaction or dissatisfaction with AIDS policy responses? Each of these
questions will be discussed in greater detail below. I should note here that although I
recognize that AIDS treatment is only one part of the overall policy response, the
scientiﬁc consensus indicates that treatment and prevention are mutually beneﬁcial. In
this inquiry I will focus almost exclusively on the treatment portion of this response.
Further research on prevention would be a useful direction for further inquiry.

I would like to quickly summarize the direction of these inquiries before expanding
upon them in greater detail in the individual articles. First, I plan to perform a macro-
level statistical analysis addressing why some states have been more effective in scaling
up ARV programs in Aﬁica than others utilizing OLS regression. I would like to

investigate what political factors lead to the divergent outcomes. For example, is ARV

l3

implementation an outcome of regime type, democratic stability, or economic strength?
This statistical analysis would represent the ﬁrst article for the dissertation. Second,
when countries cannot supply ARVs to all citizens, a common occurrence, what explains
the distribution of these scarce resources? Are these decisions based on economic,
political, or public health considerations? Again I will utilize statistical techniques,
offering a comparative analysis of two countries. Finally, what explains the differences
in citizen attitudes regarding these policy outcomes, either at the national or sub-national
level? In other words, why are some people more or less satisﬁed with their
government’s response to the disease? While this is a very brief account of the research

plan, I will develop these ideas in detail later in the individual articles.

14

Works Cited

Adams, Gregor et a1. (2001). “Consensus Statement on Antiretroviral Treatment for
AIDS in Poor Countries.” Individual Members of the Faculty at Harvard University.

Antiretroviral Drugs. Deﬁnition. Encyclopedia of Medicine.
http://www.enotes.com/medicine-encvclgmdia/antiretroviral-drugs.

 

Badri M, Maartens G, Mandalia S, et al. (2006). “Cost-effectiveness of Highly Active
Antiretroviral Therapy in South Africa.” PLoS Med. 3:e4.

Baragona, Steve. (2005). “Study Examines Lessons Learned at Africa’s First Public
Antiretroviral Treatment Clinic,” Clinical Infectious Diseases.

R. Bonnel (2000). “HIV/AIDS: Does it Increase or Decrease Growth in Aﬁica?”
Mimeo.
World Bank: Washington.

Boone, Catherine and Jake Batsell. (2001). “Politics and AIDS in Africa: Research
Agendas in Political Science and International Relations,” Aﬁica Today.

Chen, Lincoln, Tim Evans, and Richard Cash. (1999). “Health as a Global Public
Good.”
In Global Public Goods. Edited by Inge Paul, Isabelle Grunberg, and Marc Stern.
United Nations Development Program. New York: Oxford University Press.

Creese, Andrew, Katherine Floyd, Anita Alban, and Lorna Gulnness. (2002). “Cost-
Effectiveness of HIV/AIDS Interventions in Aﬁica: A Systematic Review of the
Evidence.” The Lancet. Vol. 359. P. 163 5-1 642.

Farmer, Paul. (2001). “The Major Infectious Diseases in the World —To Treat or Not to
Treat?” The New England Journal of Medicine. Vol. 345. No. 3. July.

Freedberg, Kenneth et. al. (2001). “The Cost Effectiveness of Combination
Antiretroviral
Therapy for HIV Disease.” The New England Journal of Medicine. Vol. 344. No. l 1.
March.

Goldie SJ, Yazdanpanah Y, Losina E, et al. (2006). “Cost-Effectiveness of HIV
treatment
in Resource-Poor Settings: the Case of Cote d’Ivoire.” New England Journal of
Medicine. Vol. 355:1141—1153.

Hart, C. et al. (1999). “Correlation of Human Immunodeﬁciency Virus Type 1 RNA
Levels in Blood and the Female Genital Tract. J Infectious Disease 179:871-882.

15

Hess, Charlotte and Elinor Ostrom. (2007). Understanding Knowledge as a Commons:
From Theory to Practice. Cambridge: MIT Press.

Jackson, Robert and Carl Rosberg. (1982). “Why Africa’s Weak States Persist: The
Empirical and the Juridicial in Statehood. World Politics. Vol. 35. No. 1. October. P.
1-24. Cambridge: Cambridge University Press.

Kapstein, Ethan and Josh Busby. 2009. “Making Markets for Merit Goods: The Political
Economy of Antiretrovirals.” Working Paper No. 179., Center for Global
Development: Washington, DC.

Lamptey, Peter and David Wilson. (2005). “Scaling up AIDS Treatment: What iS the
Potential Impact and What Are the Risks?” PLoS. Vol. 2 Issue 2. February.

Morrison, Stephen. (2009). “Using Measurement to Drive Results: The Case for a New
Strategic Measurement Framework.” Center for Strategic and International Studies
Commission on Smart Global Health Policy.

Parker, Stephanie. (2007). Big Pharma and Big Profits: Denying Access to AIDS
Medication. Aﬁica Action.

Patterson, Amy. (2006). The Politics of AIDS in Africa. Boulder: Lynne Rienner
Publishers.

PEPFAR. (2009). “PEPF AR: A Commitment Renewed.” Updated September.
Washington: PEPFAR. http://www.pepfjar.gov/press‘/ l 07735.htm

Stuckler, David and Martin McKee. (2008). “Five Metaphors About Global Health
Policy.”The Lancet. Vol. 372. July.

UNAIDS. (2009). AIDS Epidemic Update. Geneva: UNAIDS.
hm;://data.unaids.org/pub/Report/2009/2009_epidemic_upd2_1te_en.pdf

UNAIDS (1999). HIV/AIDS Prevention in the Context of New Therapies. Geneva:
UNAIDS.

UNICEF. (2000). “The Progress of Nations.” New York: UNICEF.

“Universal Access to HIV Treatment, Prevention, Care, and Support.” (2009). NA.
UNAIDS. http://www.un_aids.org/en/CountrvReSponseS/UniversalAccess/default.asp

USAID. (2000). “Children on the Brink: Updated Estimates & Recommendations for
Intervention. USAID: http://www.usaid.mv/pressireleasesﬂ000/childrenreport.pdf.

US. Census Bureau. (2000). “Monitoring the AIDS Pandemic Network. The Status and
Trends of the HIV/AIDS Epidemics in the World.”

16

P. Vemazza et al. (2000). Potent Antiretroviral Treatment of HIV infection Results in
Suppression of the Seminal Shedding of HIV: the Swiss Cohort Study. AIDS 28:117-
1 21 .

Whiteside, Alan, Alex de Waal and Tsadkan Gebre-Tensae. (2006). “AIDS, Security,
and
the Military in Africa.” African Aﬂairs. Vol. 105/419. Oxford: Oxford University
Press.

WHO. 2008. “Antiretroviral Therapy Coverage Among People with Advanced HIV

Infections.”Geneva: WHO.
http://www.who.int/whosis/indicators/compendium/2008/3avr

l7

The Political Determinants of ARV Coverage in Africa: A Cross National Analysis
of States and Regimes

Introduction
The complexity of responding to HIV/AIDS in Aﬁica can be seen from the

varying actors and organizations involved in this crisis. Individual citizens, civil society
groups, bureaucrats, state elites, international relief workers, foreign donors, political
leaders, and multi-national pharmaceutical manufacturers are all involved in an
unprecedented international response. From Washington to Geneva and Cairo to Cape
Town, AIDS in Aﬁica has garnered enormous resources and attention. Such a large
scale effort has not occurred in a vacuum, and this article seeks to assess whether
institutional rules or institutional capacity inﬂuence these policy responses, how these
two characteristics inﬂuence one another, and ﬁnally, what components of the regime or
state are most important to ensure policies focused on providing treatment to AIDS
patients. In performing this analysis, focus in on AIDS treatment rather than prevention
or other efforts, though AIDS treatment is representative of an overall policy response.

Before delving into research questions, these important concepts need to be deﬁned
and delineated. Institutional rules make up the regime. As Bratton and Chang (2006)
describe it, the regime is, “the set of political procedures—sometimes called the rules of
the political game—that determine who make decisions and how,” (1060). Regimes
vary from authoritarianism to democracies. Regimes are more transient than states,
which have greater enduring power. Again, quoting Bratton and Chang (2006), the state
is, “the bone structure of the body politic or the set of administrative institutions that

claim a legitimate command over a bounded territory. States vary in their strength from

18

strong to weak, and in this analysis the focus is on formal political institutions. These
two characteristics are illustrated in the diagram below.

Figure 1.1: States and Regimes Considered Together

Strong States (Measured by WBI Stateness Index)
Strong Authoritarianism Strong Democracies

Authoritarian Regimes Democratic Regimes (Measured by
PolitLIV scores)

 

Weak Authoritarianism Weak Democracies
Weak States

 

Testing how state capacity affects human development is an important part of this
paper, but it is not the only part of this project. Whether the relationship between regime
type and state capacity is interactive is another important question. The testable
hypothesis is that more democratic regimes and stronger states will have more robust
ARV coverage. Looking at the ﬁgure above, that would involve a relationship where a
scatter plot would indicate a linear relationship from lower left to upper right. Another
interesting possibility is that strong authoritarian states may be more able to deploy the
necessary resources that allow successful human development than weaker hybrid
regimes. That would be shown if the scatter plot was more ‘V’ shaped with stronger
states having more positive human development outcomes. In order to assess these
relationships a three dimensional scatter plot will be necessary to assess these
relationships, but this ﬁgure demonstrates the importance of testing these two
independent variables side by Side, increasing the dimensionality of governance and

19

providing greater leverage over our research question than bivariate modeling would
provide. Whether or not these two independent variables are additive, reinforcing as an
interaction term, or whether countries can have a stronger authoritarian regime that
provides for greater human development outcomes is worthy of testing.

My research goal is ﬁrstly to explain the political factors that Shape anti-retroviral
drug (ARV) policy outcomes including the scope of such efforts. In other words, what
are the political determinants of ARV coverage, and do democracies and stronger states
provide these resources more robustly than other types of governments? This study
performs a macro-level statistical analysis (OLS regression) addressing why some
countries have been more effective than others in scaling up ARV programs in Aﬁica.

These questions are rooted in a research agenda proposed in 2001. In 2001,
Catherine Boone wrote in Africa Today, “Political Science as an academic discipline has
been slow in grappling with the enormous implications of the AIDS crisis for much of
the developing world,” (Boone and Batsell, 2001). Boone presented a series of AIDS
related sub-topics which political scientists could offer valuable research. As she put it,
“Five research agendas for Aﬁica are: variations in state response to the pandemic; the
relationship between governments and NGOs; the AIDS challenge to neo-liberalism;
AIDS and North-South tensions; and connections between AIDS and international
security issues,” (Boone and Batsell, 2001). Clearly, and as the literature review will
indicate the ﬁeld is wide open for political scientists inquiring about AIDS ﬁ'om
multiple angles aimed at a variety of research questions; the research below will focus

heavily on the ﬁrst of Boone’s recommendations, focusing on variation in the national

20

responses to ARV treatment. I will next review important literature on the nature of
Aﬁican states and regimes before discussing the research design.
Literature Review
The literature review proceeds with a review of relevant literature on the Aﬁican

regime before a discussion of the Aﬁican state follows. After this literature has been
discussed, appropriate hypotheses, measures, and testing are discussed.
Aﬁican Regime Literature

AIDS is obviously an important challenge for the Aﬁican continent, but what is
the political dimension in this project? This set of questions has substantive theoretic
importance for political science scholars as well. One of the great debates within the
ﬁeld is the dialogue between scholars arguing that development promotes democracy
(Lipset 1959) and authors who argue that these traits do not necessarily go together
(Deutsch, 1961). Diamond (1992) retested Lipset’s hypothesis with data showing that
democratic regimes speciﬁcally increase human development. Diamond’s focus on
human development is also limited to cross tabs without a more in depth modeling
component. AIDS treatment is a speciﬁc measure of human development, and it is an
effective indicator that similarly evaluates whether public policy is geared towards
improving public focused policy outcomes. Furthermore this debate is complicated by
those who believe economic development underpins democratic endurance (Przeworski
et a1 2000) and those who have argued that inequality is an intermediary variable
between democracy and development (Boix and Stokes 2003; Acemoglu and Robinson

2006).

21

Less work has been done to look at how regime type structures human development.
Speciﬁcally, although some scholarship suggests a positive relationship (Gerring et a1,
2005, Deacon 2003, Stasavage, 2005, Halperin et. al. 2005) between democratic regimes
and more successful policy outcomes, other work indicates no relationship exists (Ross,
2006). A few authors have targeted the relationship between democracy and health
speciﬁcally with a positive relationship found in several cases. Govindaraj and Rannan-
Eliya (1994) found that when comparing communist and democratic regimes,
democratic regimes had more favorable health outcomes in terms of infant mortality and
life expectancy. Franco et a1 (2004) found similarly that states with greater political
freedom also enjoyed better health outcomes as measured by life expectancy, infant
mortality, and maternal mortality rates. Interestingly, Tsai (2006) found that among
developing countries, democracies out performed non-democracies in health provision,
even when controlling for health spending. As most of the literature suggests, there
appears to be some sort of relationship between democracy and positive health
outcomes.

Diamond and Morlino (2005) state in the introduction of their book that, “We can
analyze democratic quality by what it achieves in terms of government responsiveness to
the expectations, interests, needs and demands of citizens,” (xxix). One of the important
measures of democratic quality as they argue is vertical accountability, linking the fate
of elected officials to how citizens evaluate their political choices (xviii). For a citizen
infected with AIDS, what could be a more important need than life prolonging
treatment? Seeking to explain the relationship between democracy and positive health

outcomes, Vollmer and Ziegler (2009) argue that democratic regimes place higher

22

priority on redistributive policies including health care. They go on to state that
democratic societies are able to overcome inequalities more effectively, leading to
greater provision of these merit goods. In this way, their ﬁndings echo the framework
developed by Diamond and Morlino as public demand for these resources are able to be
met as democratic institutions serve as an interactive highway between regimes and
citizens.

It is often assumed that democratic values will promote greater public focused policy
outcomes. Sen (1981, 1999) argues that democratic regimes offer voters the opportunity
to penalize ineffective leaders and that the ﬁee press allowed under democratic rule
promotes greater information transparency. As Sen argues electoral considerations force
politicians to respond to public demands, necessitating human development investment.
Gerring et al’s (2005) analysis focuses on several possible mechanisms that link
democracy to human development: they considers electoral competition, the importance
of a free press, the likelihood that civil society advocates for human development
investment, the possibility that democratic regimes must provide some measure of
equality, the likelihood that democratic taxing institutions allow for a mechanism of
redistribution, and ﬁnally that democracy provides a more stable environment for the
provision of public goods.

While the idea that more democratic regimes should produce greater human
development outcomes has face validity, the empirical testing to this point does not
necessarily verify this expectation according to Ross (2006). Ross’s (2006) research
argues that this relationship does not hold when one focuses on whether democracies

provide better human development outcomes for the masses. Ross argues that although

23

the commonly accepted wisdom links democracy and human development, there is a
lack of consideration of global health, the lack of focus on country speciﬁc effects, and
missing data for the poorest of countries. With this conﬂicting ﬁnding, it provides an
important opportunity for retesting which I can analyze in my ﬁrst article.

With one of Gerring et al’s (2005) ﬁndings showing that longer lasting democratic
regimes result in greater human development outcomes, it is important to assess regime
stability. This is especially important in Africa given the prevalence of hybrid regimes
that sit between full democracies and authoritarian states as discussed by numerous
authors including Bratton et. al. (2005), Diamond (2002), Schedler, (2002), van de
Walle (2002) and Levitsky and Way (2002).

Gyimah—Boadi (2004) also argues for a connection between democratic transparency
helping Aﬁica’s emerging democracies ﬁght the AIDS epidemic (18-20). He argues
that authoritarianism contributed to the emergence of the crisis, but that transparency has
required states to admit and deal with the disease. Looking at the disease in a different
way, Gyimah—Boadi argues that the disease threatens Aﬁican democracy citing that less
democratic regimes like Uganda have been more effective in AIDS policy than more
democratic Botswana and South Africa. He claims that even weak political
responsiveness and accountability, the key democratic values attributed by Diamond and
Morlino (2005), have crippled the political response. More favorably, he views
democracy as providing some opportunities for successﬁrl AIDS policy, mentioning
democratic openness, civil society strengthening, bolstered rights for women, and
economic liberalization. These characteristics, as he put it, “appears that the prospects

of effective. . .control of the spread of the disease are better now than ever,” (20).

24

Does the act of holding elections guarantee these results? Lindberg (2006) argues
that elections in and of themselves increase the likelihood of liberalization regardless of
ﬂaws and that a three election cycle will increase the quality of democracy. As he
argues, democratic practice enhances democratic values in a society. But is the
procedure of elections enough to ensure public focused policy results? Is it the political
space provided by elections that improves policy or the docs increased competitiveness
between parties that leads to these provisions? Dovetailing with this increased
competition, Wantchekon (2003) indicates that competing parties often turn to
clientelistic messages, but his ﬁndings suggest that women voters are especially swayed
by public goods focused messages that lead one to belief that multi-party competition
may drive up these types of public regarded commitments to a larger extent. Do public
focused outcomes require competition and transitions in power in accordance with
Huntington’s (1991) two turnover test? As Reynolds (1999) argues, democratic
institutions build an inclusionary ethos that can lead to these positive outcomes. As he
states, electoral systems and institutional design are critical components that determine
the success or failure of democratic experiments in Africa as states must overcome the
realities of their multietlrnic societies through successful inclusionary design.

African State Literature

Instead of looking at the regime and the authoritarian/democratic continuum, another
important component of a country is its state structure. While more democratic regimes
either do or do not allow for greater ARV provision, it is likely that the level of
resources and institutional capabilities that a state has will be an important determinant

of ARV provision. State focused scholarship argues that instead of the type of

25

government in place it is the strength of this government that matters most. Englebert
(2000) has made the case that the development of state capacity structures the policy
choices available to elites and, in turn, the quality of governance in respective countries.
In other words, treating AIDS patients would only be possible in states that exceed a
certain strength threshold. Englebert links the importance of capacity and economic
development, one component of human development (2000, 30-3 7).

Villalon and Huxtable (1998) have echoed these sentiments ﬁ'om both Englebert and
Patterson’s work, summarizing them in ﬁve characteristics. Their categories include the
Aﬁican state holding ﬁve key traits, “a client status, a personalized identity, a
centralized or overdeveloped morphology, a prebendal or rentier nature, and an
extractive impulse,” (11). In unpacking these terms, they discuss the Aﬁican state as a
clients, usually ﬁrst as a client of the colonizing nation and then later as a client of either
the United States or Soviet Union during the Cold War. In the post-Cold War era, the
Aﬁican state can no longer count on these relationships to ensure sustainable income
(12). In terms of personalization, they discuss the dominance of an individual leader
(12). In these countries where a single leader dominates the policy process, the lines
between state and regime are often blurred. This relates well to our issue of inquiry,
with Mbeki’s dominance over South African AIDS policy during his regime a
controversial example. They go on to describe the over-centralized and overdeveloped
nature of the Aﬁican state. As they describe it, the state employed too many people with
the power concentrated on the center rather than a multi-layered structure that would
empower local institutions (13). When using the term prebendal or rentier state, they

refer to the state as the key distributional force in which the state plays the role of patron

26

and citizens serve as clients who are rewarded for their loyalty, either through resource
or employment opportunities, thus using the nation’s resources promote political
stability (13). Finally, as they describe, the state serves as an extractive force with the
goal of the state to focus on these activities rather than enhancing the state’s capacity
(14). Boone (2003) also focuses on the state and the variation in the strength of these
institutions as reﬂective of local circumstances, often by rural political elites. Boone’s
research offers a bottom up approach to the state, which is often subject to top-down
discussion.

Providing an example of such a top-down discussion is the work of Jackson and
Rosberg (1982). Rather than focusing on the Weberian deﬁnition of the state which
focuses on the monopoly over the legitimate use of force, with an emphasis on the
nrilitary, police, and courts, these authors focus on two levels to explain why Aﬁica’s
states have survived. They break the state into its empirical and juridical pieces. They
argue that the Aﬁican state exists largely on its juridical, or international law based
status and that Africa’s states are empirically weak where they function at all (4,5).
They argue that the African state generally fails to hold stable Communities where the
nation-state is the dominant institution compared to ethnic groups (Posner, 2005) or
local institutions (5). They go on to discuss the state’s inability to fulﬁll the requirement
of an effective government, capable of penetrating the state and reigning over the
domain of its entire territory. They describe the individuals comprising the state as
under-resourced, both in their level of capacity and their ability to deploy these limited
tools. This ﬁts with the Villalon and Huxtable notion of the state as overdeveloped with

too many people with too centralized a structure; as they discuss, this overdevelopment

27

is further hindered by a lack of investment in resources (8). In summary, the empirical
African state, which is what we are interested in their ability to provide for their
individual citizens can be characterized as weak (12). As they describe, the juridical
elements which can be characterized by international club membership more than any
deﬁnition that beholds citizens to government. In this deﬁnition, boundaries and
international recognition are what characterizes these traits, a minimalist deﬁnition that
does nothing for citizen subject which they discuss in their conclusion (21).

Similarly focused on territorial deﬁnitions of the state, Jeffrey Herbst’s (2000)
research focuses on the lack of penetration of the African state and the inability of these
states to effectively penetrate Aﬁican societies. This project conceptualizes state
penetration through roadways, which Herbst argues are inherently underdeveloped in
most state structures. In his description the territories necessitated colonial institutions
which ruled on the cheap; at independence these international boundaries, the juridical
statehood described by Jackson and Rosberg (1982), were upheld by the international
system and post-colonial state boundaries simply followed this pattern. As Herbst
argues, the deﬁned state often only reigns over the capital city rather than extending its
reign over the full realm of its territory and possibly to a valued region containing an
extractable resource as both Boone (2003) and Villalon and Huxtable (1998) described
in their discussion on the role of states as extractive institutions.

The African state is described and well summed up by van de Walle (2003) as
existing in a state of permanent crisis, and by Chabal and Daloz (1999) as disorderly.
These characteristics well summarize the overall message described by the above state

focused scholars. The relative weakness of the Aﬁican state when compared to similar

28

institutions on other continents is not controversial. What is also clear is that not all
African states are equal. As Patterson (2006) describes in her work on the Aﬁican state
and the AIDS crisis, she ﬁnds there is signiﬁcant variation in terms of different
individual characteristics; in her analysis there was variation in terms of centralization,
neopatrimonialism, capacity, and stability (28). Though she was unable to discern a
clear pattern along these characteristics in terms of how these state level measures
inﬂuenced AIDS policy, what this project reveals is that Aﬁican states, though relatively
weak when compared to others, Show signiﬁcant variability across respective units.

If a relationship is found between state strength and ARV provision, what underlies
such a relationship? Following Bratton and Chang (2006) this analysis will measure
stateness using an index of ﬁve indicators from Kauﬁnann, Kraay, and Mastruzzi (2009)
that focus on governance. They offer six indicators in total: voice and accountability,
political stability, government effectiveness, regulatory quality, rule of law, and control
of corruption. The ﬁrst, voice and accountability, is dropped as this ﬁ'eedom based
measure is more a matter of regime type, our other primary independent variable. To
develop our index of stateness, a simple average of the ﬁve components is taken. Once
an initial relationship is established, models will be run to investigate which elements of
stateness matter most, whether it is stability, governance effectiveness, effective
regulations, the rule of law, or successful control of corruption that makes ARV
provision possible.

As the literature suggests the state and its ability to project power varies across the
continent. The strength of the state (Bratton and Chang, 2006) is as important of an

inquiry in terms of its relationship to ARV provision as regime type, and comparing the

29

importance of these different vantage points systematically is the goal of the research
design below.
Literature Synthesis and Research Design

Capturing the politics of ARV therapy will require investigation into several political
factors and how these help shape this policy ﬁeld. This study will try to explain levels
of ARV coverage across African countries. One characteristic of ARVS is the fact they
are a merit good of mass value to citizens suffering from AIDS. Other things being
equal, a country’s provision of these drugs serves as a proxy of government
responsiveness, which should be enhanced by democratic institutions according to
political science literature (Halperin et al 2005; Goetz and Jenkins 2004; Sen 1999).
The state literature argues that greater state strength Should also boost the likelihood of
citizen focused public policy. As will be seen below many of the hypotheses seek to
investigate whether democratic regimes and stronger states have been more responsive
in providing ARVS than have non-democratic regimes, weaker countries. This is the
goal of the ﬁrst model below, probing whether it is the state, the regime, or some
combination of these institutions that inﬂuence a country’s ability to provide for its
citizens. If regimes are dominant, what characteristics of democracies ensure these
countries are more effective? If it is the state that matters most, what elements of the
state matter most for this provision? Unpacking these concepts and building a research
design capable of testing these questions is the goal of the research design outlined
below.

All of the literature suggests that the country is an appropriate level of analysis for

inquiry regarding ARV provision. Most readings suggest some sort of relationship

30

exists between democratic regimes and positive policy outcomes with all except Ross
(2006) arguing that either human development or health speciﬁcally are improved by
living in democratic regimes. Similarly, as Englebert (2000) argues, stronger states of
greater capability should be more proﬁcient at providing resources than weaker cases.
Getting beyond the democracy/non-democracy regime and strong/weak state typologies,
if relationships are found between states, regime, and the provision AIDS treatment, we
need to ask: what speciﬁcally about these cases enhances these outcomes? In order to
address these questions ﬂre formal political institutions and several characteristics of the
state must be examined to seek out which building blocks are most important for
ensuring drug provision. For instance, are regimes with multiple political parties more
likely to provide ARVS? Is the rule of law the important component of the state that
allows such provision? These hypotheses are listed below.
Dependent Yam

As a dependent variable measure, the World Health Organization has a database with
ARV coverage as a percentage of the total number of advanced stage AIDS patients for
forty two sub-Saharan African countries. Island nations are excluded from the dataset,
but all other sub-Saharan countries are included for N=41. These measures are for 2006
and have been utilized by other researchers who consider them to be the best measure
available of ARV coverage available at the cross national level. All of the variables
included for this analysis are based on 2006 data.
Cross National Hypotheses

Based on the discussed gaps in the literature and the areas of inquiries listed above,

several hypotheses are worthy of investigation. They include the following:

31

H1: Countries with a more democratic regime will be more likely to have a better
developed ARV program than those states with lower quality democracy.

This hypothesis allows for getting below the democracy/non—democracy typology to
investigate deeper into this level of democracy and how it relates to ARVS. This
hypothesis is rooted in Diamond’s (1992) ﬁndings linking greater quality of democracy
with higher levels of human development, and this test will investigate the relationship
between AIDS policy outcomes and quality of democracy similarly. Both democracies
and non-democracies will be considered in the analysis. Freedom House scores were
considered, but due to their strong normative emphasis on democracy, polity IV data
was selected to serve as the measure of regime type.

H 2, Countries with stronger states will be better able to provide ARVS and will have
greater coverage than weaker states.

Focusing on state strength is based on the literature on the Aﬁican state including
Englebert (2000) and others. To measure the strength of a state, an index of stateness
based on Kaufrnann, Kraay, and Mastruzzi (2009) World Bank Indicators is used,
similar to Bratton and Chang (2006).

Underlyinﬂegime Hypotheses

H 3: Countries that have experienced longer periods of democracy will be more likely
to have a better developed ARV program than in countries with shorter durations of
democracy.

This hypothesis will allow for the investigation of whether democracy provides for
greater political responsiveness. Length of uninterrupted democracy is coded in an

available dataset from Lindberg (2006) and regime longevity is coded based on years

32

since a founding election as of 2006. With one of Gerring et al’s (2005) ﬁndings
showing that longer lasting democratic regimes result in greater human development
outcomes, it is important to assess regime stability.

H4: Countries experiencing executive turnover are more likely to have better
developed ARV programs than those without such a history of turnovers.

One controversial aspect of democratic theory is whether two turnovers after a
transition is sufﬁcient in a deﬁnition of a consolidated democracy (Huntington, 1991).
This hypothesis will inquire into whether countries with such turnovers, based in greater
electoral competition make the merit goods, as measured by ARVS more important for
candidates. Turnovers are recorded in the Lindberg dataset, and are added up as of 2006
elections. In the coding for this variable, Lindberg differentiates between turnovers that
involved the election of a new party or candidate versus one in which the candidate was
both new and of a new party. Lindberg refers to the former as half turnovers; in my
coding for this project I elected to only include full turnovers in the data set, meaning
that the transition changed both candidate and party in power rather than one or the other
of these characteristics. This type of turnover represents a full transition rather than a
simple change of administration.

H 5: Countries experiencing more competitive executive elections will be more likely
to commit to ARV therapy than in states without competitive elections.

This is another hypothesis based on the notion that competitive elections will require
greater public responsiveness as assessed by Wantchekon (2003). This hypothesis will
test whether ARVS are one of these policy areas. Electoral results are available in the

Lindberg dataset allowing for competitiveness to be measured. To measure this

33

characteristic, the percentage of the vote garnered by the winning party is coded; thus, a
larger value is consistent with a larger percentage of votes garnered by the leading
political party and a lower level of electoral competitiveness.

H6: Countries with multiple political parties will have more successfully developed
ARV programs than in states that few or dominant parties.

This is another hypothesis interested in the institutional structure and whether
elections change the dynamic and multiparty systems develop more public regarded
policies and stems from Lindberg’s ﬁndings on democratic institutions increasing
competition between parties over time as well as Wantchekon’s (2003) ﬁndings linking
public goods provision to vote choice. Effective number of political parties is available
in a database from Lindberg (2006) and will be used to assess the competitiveness of a
party system. Multi-party states are coded with a dummy variable to delineate single
party from multi-party regimes.

Underlying State Hypotheses

H 7,. Countries with greater political stability will be able to provide ARVS more
effectively than states that are less stable. Stability is one of the aspects of the state
Patterson considers in her analysis of African states and AIDS response, and though her
ﬁndings are mixed, she only looks at four cases, and an expanded analysis that includes
more than forty cases should allow for stronger evidence one way or the other. Stability
is an index component ﬁ'om Kauﬁnann, Kraay, and Mastruzzi (2009).

H3; Countries with greater governmental effectiveness should lead to greater ARV
provision than in states with weaker governance.

This goes back to Herbst’s (2000) notion that states that have effectively penetrated

34

their respective societies will be more able to govern, and these states should be more
responsive to citizen needs. Governmental effectiveness is part of the Kaufrnann,
Kraay, and Mastruzzi (2009) World Bank Indicators Project.

H9; Countries with greater regulatory quality will be more able to provide ARVS than
states with weak regulatory policies.

In their construction of this measure, Kaufrnann, Kraay, and Mastruzzi (2009) focus
on the ability of a state to put in place investment and development ﬁiendly policies, and
it is likely that greater development will yield the resources required for more public
focused policies like ARV provision.

H10; Countries with a stronger rule of law will be more able to provide ARVS for
citizens than states with a weaker rule of law.

Bratton and Chang (2006) ﬁnd that the development of a rule of law is the single
most important building block for a democratic state. This hypothesis tests whether a
strong rule of law also leads to greater ARV provision, the next step from governance to
public friendly policy. This measure is also from Kauﬁnann, Kraay, and Mastruzzi
(2009).

Control Variables:

As control variables, I include ethnic fragmentation to incorporate Lieberman’s
(2009) recent ﬁndings regarding ethnic barriers preventing more effective AIDS policy
responses. I will measure this using the PREG (Posner, 2004) which captures whether
greater numbers of participating politically relevant ethnic groups decrease effective
HIV/AIDS policy response including ARV coverage as suggested by Lieberman (2009).

Posner’s measure is more relevant than the ethno-fractionalization index as it only

35

includes groups that participate in the political process rather than all ethnic
constituencies, making it more politically relevant. I also incorporate data on the source
of this funding to control for international interventions as well as health infrastructure.
This measure controls for the percentage of ﬁrnding coming from international donors. I
can also control for HIV prevalence rates to control for relationships between higher
need states and allocation of these resources; these measures are independent of the
dependent variable measure. One would expect that states facing larger epidemics
would be more likely to prioritize a robust AIDS response.
Statistical Analysis: Measures, Methods, and Models

For the regression analysis, N=40 as numerous island nations did not have ARV
coverage data available from the World Health Organization. This removed four cases
from consideration: Cape Verde, Comoros, Sao Tome and Principe, and the Seychelles.
Additionally Swaziland was found to be an outlier and was subsequently dropped from
the model. The Swaziland case was the only example of a relatively strong non
democratic state which managed to provide ARVS more robustly than many of its
counterparts. Swaziland’s small size and unique standing makes it difﬁcult to compare
to other African cases, though follow up research with a wider data set would be a useful
direction for future research.

Before we analyze the regression results, let us quickly discuss the methodology.
First, the dependent variable is the percentage of AIDS patients receiving ARV
treatment out of the total number of those who require this treatment.

Model 1 compares the importance of state and regime characteristics and their

36

inﬂuence on ARV coverage]. To assess regime type, we utilize polity IV data. The
stateness index discussed below follows Bratton and Chang (2006) and is a mean of ﬁve
components: political stability, governmental effectiveness, regulatory quality, control
of corruption, and rule of law. These measures are established indicators developed by
the World Bank, and they are some of the most important aspects of the state capacity
concept.

Model 2 adds important control variables to assess the robustness of the state/regime

ﬁndingsz. PREG scores includes Lieberman’s ﬁnding that ethnic barriers prevents
effective AIDS policy using a respected measure developed by Posner (2004). I also
control for the source and policy decisions regarding funding for health care funding
with data sourced from the World Health Organization (WHO) for 2006. Health
Spending as a percentage of GDP was the best measure available as it best reﬂected the
decisions available to policymakers in their country’s respective economic contexts.
Overall health spending was also tested but it was even less inﬂuential than health
Spending as a portion of GDP. Finally, we control for the burden of HIV across the

various countries, again with data coming from the WHO.

 

1 It should be noted that although polity IV scores were used to measure regime type and
the World Bank indicators are used to assess state strength, Freedom House scores to
measure regime type and Rotberg’s (2009) Index of African Governance to assess state
strength were also tested. The relationship discussed below in the ﬁrst two models was
robust across these other measures.

2 Numerous other variables were considered as controls and were shown to have no
relationship with ARV coverage. These include GDP per capita, % of the population
that lived in urban areas, controls for both population and size of respective countries,
and measures of the overall health care system quality like doctors per capita, hospital
beds per capita, infant mortality. None of these variables were as important as those
used in the models. Models were also run to seek out interaction terms between
stateness and variables like percentage urban, democracy, ethnic diversity, hospital beds
per capita, proportion of budget from international donors, and total spending. None of
these interaction terms were deemed statistically Si gniﬁcant.

37

Model 3 attempts to get below the aggregated measures of regime and state and
investigates what components of these two concepts are most important for ARV
coverage. Regime data comes from Lindberg (2006). Years of continuous democracy
measures regime longevity. Number of turnovers assesses both regime longevity but
also competitiveness within that life span. Competitiveness is a measure of the
percentage of the party in power’s share from the most recent executive election. Multi-
party is a dummy variable assessing whether a state has multiple political parties in the
regime. The components of the stateness index are the four components that were
combined previously: political stability, governmental effectiveness, regulatory quality,
and rule of law. Again data is taken from Kauffman, Kraay, and Mastruzzi (2009) and
their World Bank Indicator Project.

Model 4 simply offers an attempt at a reduced model with the most empirically
important disaggregated variables included. As will be discussed below, state strength
was consistently the most important predictor of ARV coverage regardless of controls or
level of aggregation. AS was previously noted, this ﬁnding with state strength
dominating over regime type was robust across numerous measures as well. Models 3
and 4 together provide evidence that state capacity components—political stability,
governmental effectiveness, regulatory quality, the rule of law, and control of
corruption—all stand together, both conceptually and in policy practice. In other words,

effective state capacity requires each of these factors rather than a piece meal approach.

38

 

Table 1.1: OLS Regression Results for ARV Coverage

 

Regime

State

Controls

R Squared

Constant

Polity IV Score

Years of Continuous Democracy

Number of Executive Turnovers

Electoral Competitiveness

Multi-Party

Stateness Index

Political Stability

Government Effectiveness

Regulatory Quality

Rule of Law

PREG

Health Spending as a % of GDP

% External Health Spending

Logged HIV Prevalence

Adjusted R Squared

Model 1
33.270***

(4.653)
-0.165
(0.544)

r5.327***
(4.738)

0.256
0.216

Model 2

3185*"
(9.28)
-0232

(0.585)

11.810"
(5.219)

-12.697
(1 1.028)
1.57
(1.346)
-0.224
(0.172)
1.436
(2.2526)
0.379
0.259

Dependent Variable= ARV Coverage

***¢<.01 **=p<.05 *=p<.1

Model 3

19.123
(21.06)

0.568
(0.426)
-3.l69
(4.446)

0.03
(0.126)
3.796
(17.742)

-0.098
(4.923)
9.142
(13.202)
1.547
(9.675)
-0921
(14.346)

0.32
0.145

Standard Errors in Parentheses

Model 4
2288*
(12.983)

0.417
(0.401)

6.719
(5.575)

-14.807
(10.680)
1.445
(1.347)
-O.161
(0.176)
2.023
(2.524)
0.39
0.272

 

The ﬁrst model tests the aggregate relationship between regime type, state strength,

and ARV coverage. As the model indicates, the relationship is dominated by state

 

strength at ﬁrst glance. Stronger states are much more likely to be able to provide drugs
to needy patients than weaker states. The relationship between regime type and ARV
coverage is more complicated than this initial model suggests, however. While the
relationship between regime type and ARV coverage was more U-shaped with
Swaziland in the model, we do not have enough cases to assess whether strong,
authoritarian states are as capable of providing ARVS as strong democratic states, but
the initial evidence suggests this. Further investigation of authoritarianism and ARV
coverage would be useful in a wider analysis if more strong authoritarian countries were
available in a wider data set. If this initial analysis is correct, this can be interpreted to
mean that stronger democratic and authoritarian regimes are more effective at providing
these resources than weaker hybrid regimes. This ﬁnding ﬁts well with the strong
relationship between state strength and seems to indicate that state strength more than

regime type matters for ARV provision, though both variables are important.

40

Figure 1.2: Scatter Plot of Polity Scores and ARV Coverage

 

 

 

 

80.00" Botsw aha
O
Mm'bl
O
60.00- Senegal
0
Rwanda
0
O
53
g 9W5
Gabo
3 40.00- 0 " Benin
0 Swaziland
E 0 Burkina Faso
Uganda Zanbla gen”
. O
. Gunea Malawi oSouth Africa
20.00- Garrbrao 000m d'Noire ° Lesothoo
to' lGu‘ T ogCaneroogAngola Carr ' ”RC °B“"’”di
u re - ..
51.3 r Ineaoog Z'nbabwe 0 . ”Diamemo oil/ozanbique Ghana Mauritius
ErItreao Chad 0 Tanzania 5' L O O O O
COMO (Republic Of) 0 Central African Repu lane: 90'3“” r-
M'aurilania O "138' Issau
ONbdagascar
0.00"
I I I I I
-10.00 5.00 0.00 5.00 10.00
Polity N Score

There is a further interpretation of this scatter plot that is important: while a country

is not completely unable to provide these resources in hybrid regimes, in these types of

countries, the decision whether to provide these resources is more tightly controlled by

political leadership. I would argue that more democratic states provide political space

where social capital and activism has been shown to exert pressure on policymakers to

provide these resources. This is the case both in countries that have many patients

requiring ARVS like Botswana and Namibia, but is also true for states with relatively

low prevalence rates like Mali and Senegal. Madagascar has such low prevalence that

the issue of AIDS treatment is likely farther down the list of policy priorities. Even in a

41

case where political leadership sought to block ARV provision, South Aﬁican activists
were able to use democratic institutions to effect change, despite the policy preferences
of the Mbeki regime. I would argue that it is this type of political activism, only
possible in the conﬁnes of liberalized democratic institutions that ensure ARV provision.
It is these types of democratic cases where future ARV provision is likely to be
prioritized and maintained. This is due to the fact that policy processes in these states
subject policymakers to the vertical accountability described by Diamond and Morlino
(2005) both through policy making processes and through the ballot box.

Model 2 adds robustness checks to investigate whether ethnic heterogeneity prevents
AIDS treatment, how international interventions inﬂuence this relationship, whether
greater investment of scarce state resources inﬂuences ARV coverage, and whether
AIDS prevalence rates inﬂuence policy making decisions. The relationship between
state and regime characteristics holds, even with these control variables added to the
equation. Each of the controls is in the expected direction: greater ethnic diversity
inhibits AIDS treatment as Lieberman (2009) suggests, though this relationship is not
statistically signiﬁcant. Spending by the international community is not an important
driver of ARV coverage. Greater Spending on health as a percentage of GDP increases
ARV coverage, although this is not a statistically Signiﬁcant relationship. Also more
HIV cases make it more likely that countries will develop more robust ARV programs.
This relationship suggests that the worst hit countries have prioritized funding AIDS
policy interventions than states that have lower prevalence rates. This makes sense for
policy makers, both in Aﬁican governments and for international donors who have

prioritized the worst hit states ahead of states with lesser epidemics.

42

Figure 1.3: 3-D Scatter Plot of Polity Scores, Stateness Index, and Logged ARV
Coverage

 

Logged ARV Coverage

 

 

 

 

The 3-D scatter plot above well demonstrates the relationship between state and
regime characteristics upon ARV coverage. Though the models are run with ARV
coverage unlogged, logging ARV coverage provides a clearer visual picture of how state
strength and regime type inﬂuence ARV coverage. Though testing for an interaction
term between these variables was not statistically signiﬁcant, the relationship was
positive as the regression models suggest: greater state strength and more democratic
regimes are best positioned to provide these resources to citizens. As this scatter plot
suggests, the U shaped relationship between regime type and coverage continues, with

increased state strength increasing the likelihood these resources are provided as one

43

moves ﬁ'om weaker to stronger states. Democracy appears to be important, though it is
conditional upon state strength.

Model 3 attempts to disentangle which components of states and regimes are most
important for ensuring ARV coverage. Due to the small number of cases, this model is
limited in the number of variables that could be included. Of the stateness index
components, four of the ﬁve are included with control of corruption absent. Of the ﬁve I
assessed control of corruption to be least important for the provision of ARVS and
corruption was not among the most discussed aspects of the state discussed in the
African state literature. Four of the indicators focus on aspects of regime type, and four
of the ﬁve features of the stateness index are included. This model suggests that the
aggregate measures are largely collinear as none of the individual components of the
stateness index are statistically signiﬁcant or nearly so. This high correlation ﬁnding for
these indicators is similar to that of Bratton and Chang (2006). Substantively, this
reveals that state capacity components tend to hang together, an important ﬁnding in its
own right. Of these measures governmental effectiveness and regulatory quality are the
most important of the state strength measures with governmental effectiveness easily the
most important. Longer lasting regimes are shown to have a positive relationship with
ARV provision, and though the relationship is not statistically signiﬁcant, it is one of the
most important components when the standardized coefﬁcients are evaluated. The
relationship between turnovers and ARV coverage is negative, suggesting political
competition makes it less likely ARVS will be provided. The small number of cases that
have resulted in turnovers and the fact that many of the turnovers that have occurred

happen to have occurred in island nations that lack ARV data make it difﬁcult to assess

this relationship. I would argue that more cases are needed to properly assess whether
Huntington’s hypothesis about two turnovers deﬁning a democratic transition inﬂuences.
ARV coverage. These ﬁndings suggest that although state strength and regime type are
shown to be important predictors of ARV provision, it is the sum of the parts, both state
strength and regime type that matter more than any individual disaggregated
characteristic that is driving these relationships.

Model 4 seeks to provide a reduced model while including robustness checks. The
most important state and regime variables are included with the four control variables.
The reduced model features the most important component of state strength as the most
important political variable: government effectiveness. We also see that governmental
effectiveness is the most important part of stateness to ensure high levels of ARV
provision. Although the control variables are in the expected direction, none of them are
statistically signiﬁcant. The ﬁnding related to governmental effectiveness suggests that
those states with the best civil services that were already efﬁcient in providing other
social services are best able to provide AIDS treatment. The lack of statistical
signiﬁcance, however, sheds light on the fact that effective state capacity building will
require investment to bolster each of the components rather than a focus on any one
component. Governmental effectiveness is important, but it is only one piece of the
puzzle. This relationship also has face validity, as the most effective states like

Botswana and Namibia are also the most effective at providing ARVS.

45

Figure 1.4: Summary of Results and Analysis

 

Expected Actual Statistical

Hypothesis Relationship Relationship Significance
H1: More Democratic Regime ->
Greater ARV Provision Positive Negative No
H2: Stronger States-) Greater ARV
Provision Positive Positive Yes
H3: Greater regime Iongevity->
Greater ARV Provision Positive Positive No
H4: Executive turnovers—) Greater
ARV Provision Positive Negative No
H5: Electoral Competitiveness->
Greater ARV Provision Positive Positive No
H6: Multiple Political Parties-9
Greater ARV Provision Positive Positive No
H7: Greater State Stability—) Greater
ARV Provision Positive Negative No
H8: Greater Governmental
Effectiveness-> Greater ARV
Provision Positive Positive No
H9: Stronger Regulatory Policy-9
Greater ARV Provision Positive Positive - No
H10: Stronger Rule of Law—> Greater
ARV Provision Positive Negative No

 

 

 

The overall goal in this article was to investigate whether democratic regimes and
stronger states would create environments more conducive to the provision of ARVS.
Additionally, it sought which national-level characteristics were most important for this
provision; the literature suggests that democratic regimes would be obligated to pursue
policies that are more public focused as policymakers must respond to public demands
due to vertical accountability as suggested by Diamond and Morlino (2005). In the
most democratic, strongest states, this appears to be exactly what is happening. The

most democratic countries appear to be doing a better job than less democratic states.

46

Based on the number of democracies that ﬁt this deﬁnition, this ﬁnding appears to be
robust. The 3-D scatter plot suggests that Aﬁican states and regimes are tied together,
similar to the ﬁndings of Bratton and Chang (2006). It is in the cases where civil
services are strongest and most capable that social service provision appears to be most
effective.

What does this suggest for those interested in increasing treatment provision for all
citizens inﬂicted with HIV/AIDS? Instead of focusing on piecemeal responses or

viewing African regimes through democratic blinders, one must consider the overall

strength or weakness of these states, as is indicated by the 3-D scatter plot above3. In
other words, and as this graphic indicates, it is not exclusively the most democratic
regimes that make it more likely that citizens who require treatment will be able to
access it; rather, it is the stronger states on both ends of the political Spectrum that are
most effective at ARV provision versus weaker hybrid regimes that are less able to
provide these resources. Examples of states that are less democratic but more effective
in providing ARVS include Swaziland and Rwanda; Swaziland is a monarchy that faces
one of the highest HIV prevalence rates in the world, but it has managed an effective
ARV response to this disease. More democratic regimes that are most successful at
providing ARVS include states like Botswana and Namibia, which have followed the
more traditional model of open political space leading to public demand for these
medications manifesting in governmental policy response. I have suggested above that

such political space is more likely to lead to greater and more robust investments in

 

3 It should be noted that in the regression analysis, statistical investigation did not reveal
an interaction term between state strength and regime type, though the 3-D scatter plot
suggests such a relationship exists, though it is non-linear.

47

ARV provision. Weaker hybrid regimes that have been less successful at such provision
include Liberia and Mozambique among others where the state is simply too weak to
play as signiﬁcant a role in ARV provision. Both of these states are forced to rely on
international funding for their ARV funding to a greater degree than other states with
over forty percent of Liberian health spending and over ﬁfty percent of Mozambique’s
health spending coming from foreign donors, according to World Health Organization
data.

This smoother also explains fairly well why South Africa is often ridiculed for their
ARV provision by the international community: the state has been able to scale up ARV
provision on par with its other Aﬁican counterparts, but it has not been as effective as
one would expect considering its high capacity and high level of political rights. This is
the quintessential case where open political space was used by citizen action groups to
push a hesitant government to engage on the treatment issue, according to Patterson
(2006) among others. Based on this model, one would predict that South Aﬁica would
be best positioned to bolster its ARV coverage than other states on the continent. Post-
Mbeki South Aﬁica appears to be doing just that (Dugger, 2009. The smoother also
reveals states that have been able to do a successful job of providing drugs despite
difﬁcult political circumstances like Uganda and Zambia. As suggested, however, these
types of commitments are likely to wax and wane with the establishment of new
political leadership. In the strong democratic states, however, these commitments are
more likely to be sustained as citizen groups advocate and use democratic institutions to

ensure policies that focus on increasing citizen access to these vital resources.

48

Future Research
Figure 1.5: Scatter Plot of Unstandardized Residuals and ARV Coverage

 

 

 

 

80.00-
c
o
6000-
o
o
O
or
E
g Q:
o 40 oo- o
0
>
E o
CPO o
o
o 0
20.00- 08 6’
o o o
o o o o o cg
° 0
o
o
0.0
l l l |
-20.00000 0.00000 20.00000 40.00000
Unstandardized Residual

There are several research questions not addressed in this analysis, and there is
obviously a great deal of variance left to be explained. Above is a residual plot from
model 4, which suggests that even the model with the greatest explanatory power is
underspeciﬁed. What other factors could be driving ARV coverage? I would speculate
that political leadership, a difficult concept to measure, remains a consistent important
factor, regardless of regime type and state strength. As has been discussed, we see cases
where lower capacity hybrid regimes have made the decision to more effectively tackle
the disease like Uganda. This is an important as it points to the agency that African

policymakers hold regardless of their economic means. It is not a coincidence that the

49

most democratic countries consistently opted to spend more of their resources on health
care and public regarded outcomes than hybrid regimes, especially when HIV is a larger
threat. What is evident from the dependent variable measure is that no Aﬁican state at
this point provides ARVS for every citizen who needs them. How these precious
resources are distributed at the sub-national level is the subject of the next article.
Another piece of the puzzle that is not considered in this analysis is the role of political
leadership. Individual leaders can clearly Shape public policies, especially in Africa’s
largely centralized state structures. We see this in over-performing examples like
. Uganda under Museveni’s leadership as well as South Aﬁica’s relative
underperformance while Mbeki was president. I have made the argument that political
leadership is likely to have less inﬂuence in strong democracies where open political
institutions create political space for groups seeking greater social services like ARVS.
Understanding the circumstances that direct leaders to prioritize treatment and public-
regarded policies more broadly would be a useful direction for future investigation. In
order to situation leadership, however, one must also investigate the role of citizens and
how the public views these policy responses. That will be the subject of the third article
in this project.
Conclusion

To answer the title of this article, Africa’s strongest states and most democratic
regimes are more effective at providing ARVS than their peers. What this analysis
suggests is that in order to most effectively improve the lives of African citizens over the
long term, one must ﬁrst build up the state by investment in state institutions if Aﬁican

countries are to eventually manage the AIDS issue with reduced reliance on the

50

international community. Instead of focusing exclusively on treating the symptoms
through exclusively directing resources towards individual issues like AIDS treatment,
and despite the noblest of intentions, taking a wider perspective and considering
investment in enhancing Aﬁican state capacity would appear to be a wise investment for
the future of Africa’s citizens.

What is also interesting is that while much literature has focused on the connection
between elections and democratization moving towards consolidation, there has been
less focus on the importance of state capacity as part of this process (Englebert, 2000).
If one is interested in Aﬁican states taking the next step, from focusing exclusively on
democratization to actually providing merit goods and enhancing human security for a
state’s respective citizens through public-focused policy, the emphasis must broaden
from an exclusively electoral focus to building the state’s capacity, whether we are
interested in the provision of ARVS or other necessary governmental interventions.

How can such capacity be built? Speciﬁcally in health care, this would involve
reducing incentives for Aﬁican health care workers to leave their countries once they are
trained for better pay in other countries. Levy and Kpundeh (2004) writing on
enhancing state capacity in African states discuss the importance of changes like
administrative reform, budget transparency, enhancing the role of parliament as an
oversight institution, strengthening anti-corruption institutions, and decentralization of
services like education. As they argued in a 2004 presentation at the World Bank, the
equation for successful Aﬁican policy performance is as follows: Stabilization +
Democracy + Capacity —+ Performance. As it relates to ARV availability, democracy

and state strength have been shown to strongly inﬂuence policy performance, though

51

stability may be a pre-condition. Stabilized democratic regimes appear to be the best
vehicle for such provision, but high levels of state capacity appear to be the most
important individual component for Aﬁican citizens to access these life prolonging

medications.

52

Works Cited

Acemoglu, Daron and James Robinson. 2006. The Economic Origins of Dictatorship
and
Democracy. New York: Cambridge University Press.

Boix, Carles and Susan Stokes. 2003. “Endogenous Democratization,” World Politics,
55, 4: 517-549.

Boone, Catherine and Jake Batsell. (2001). “Politics and AIDS in Africa: Research
Agendas in Political Science and International Relations,” Africa Today.

Boone, Catherine. 2003. Political T opographies of the African State: Territorial
Authority and Institutional Choice. New York: Cambridge University Press.

Bratton, M., E. Gyimah-Boadi, and R. Mattes. 2005. Public Opinion, Democracy and
Market Reform in Africa. Cambridge: Cambridge University Press.

Bratton, Michael and Eric Chang. 2006. “State Building and Democratization in Sub-
Saharan Africa: Forwards, Backwards, or Together?” Comparative Political Studies,
39, 9: 1059-1083.

Chabal, P. and J. Daloz, 1999. Africa Works: Disorder as a Political Instrument
(Oxford:
James Curry Press.

Diamond, Larry. 1992. “Economic Development and Democracy Reconsidered,”
American Behavioral Scientist 35, 4/5: 450—499.

Diamond, Larry. 2002. “Thinking about Hybrid Regimes,” Journal of Democracy. Vol.
13 No. 2.

Diamond, Larry and Leonardo Morlino. (2005). Assessing the Quality of Democracy.
Baltimore: John Hopkins University Press.

Deacon, Robert. (2003). “Dictatorship, Democracy, and the Provision of Public Goods.”
Departmental Working Paper, University of California at Santa Barbara
http://repositories.cdlib.org/ucsbecon/dwp/20-03.

Deutsch, Karl. 1961. “Social Mobilization and Political Development,” American
Political Science Review 55, 3: 493-514.

Dugger, Cylia. “Breaking With the Past, South Aﬁica Issues Broad AIDS Policy,” New
York Times. 1 Dec. 2009. English Edition. Print.

53

Englebert, Pierre. 2000. State Legitimacy and Development in Africa. Boulder: Lynne
Rienner.

Franco, Alvaro, Carlos Alvarez-Dardet, and Maria Teresa Ruiz. (2004). “Politics of
Democracy on Health: An Ecological Study. British Medical Journal. Vol. 329. p.18-
25.

Gerring, John, Strorn Thacker, and Rodrigo Alfaro. Democracy and Human
Development. Presented at the American Political Science Association Meetings,
2005. Washington, DC.

Goetz, Anne and Robert Jenkins. (2004). Reinventing Accountability: Making
Democracy
Work for Human Development. London: Palgrave McMillan.

Govindaraj, Rarnesh and Ravindra Rannan-Eliya. (1994). Democracy, Communism, and
Health Status: A Cross National Study. Data for Decision Making Project, Harvard
School of Public Health.

Gyimah—Boadi, E. (2004). Democratic Reform in Africa: The Quality of Progress.
Boulder: Lynne Rienner.

Halperin, Morton, Joseph Siegle, and Michael Weinstein. (2005). The Democracy
Advantage: How Democracies Promote Prosperity and Peace. New York:
Routledge.

Herbst, J efﬁey. 2000. State and Power in Africa: Comparative Lessons in Authority and
Control. Princeton: Princeton University Press.

Huntington, Samuel. 1991. The Third Wave: Democratization in the Late Twentieth
Century, Norman: University of Oklahoma Press.

Jackson, Robert and car] Rosberg. “Why Aﬁica’s Weak States Persist,” World Politics.
Vol 35 p. 1-24.

Kaufrnann, Daniel, Aart Kraay and Massimo Mastruzzi. 2009. "Governance Matters
VIII: Governance Indicators for 1996-2008". World Bank Policy Research June.
Policy Working Paper 4978.

Levitsky, Steven and Lucan Way. 2002. “The Rise of Competitive Authoritarianism,”
Journal of Democracy. Vol. 13. No. 2.

Levy, Brian and Sahr Kpundeh. 2004. Building State Capacity in Africa. Washington,
DC: World Bank Publications.

54

Lieberman, Evan. (2009). Boundaries of Contagion: How Ethnic Politics Have Shaped
Government Responses to AIDS. Princeton: Princeton University Press.

Lindberg, Staffan. (2006). Democracy and Elections in Africa. Baltimore: John Hopkins
University Press.

Lipset, Seymour Martin. 1959. “Some Social Requisites of Democracy: Economic
Development and Political Legitimacy,” American Political Science Review, 53, 1:
69-105.

Patterson, Amy. (ed.) (2005). The African State and the AIDS Crisis. Burlington:
Ashgate.

Patterson, Amy. (2006). The Politics of AIDS in Africa. Boulder: Lynne Rienner
Publishers.

Posner, Daniel. (2004). “Measuring Ethnic Fractionalization in Aﬁica,” American
Journal Of Political Science. Vol. 48, Issue 4. p. 849-863.

Posner, Daniel. 2005. Institutions and Ethnic Politics in Africa. Cambridge: Cambridge
University Press.

Przeworksi, Adam, Michael Alvarez, Jose Antonio Chiebub, and Fernando Limongi.
2000. Democracy and Development: Political Institutions and Well-Being in the
World. New York: Cambridge University Press.

Reynolds, Andrew. (1999). Electoral Systems and Democratization in Southern Africa.
Oxford: Oxford University Press.

Ross, Michael. 2006.“Is Democracy Good for the Poor?” American Journal of Political
Science, 50:4, October.

Schedler, Andrea. 2002. “Elections Without Democracy: The Menu of Manipulation,”
Journal of Democracy. Vol. 13 No. 2.

Sen, Amartya. 1981. Poverty and F amines: An Essay on Entitlement and Deprivation:
New York: Oxford University Press.

Sen, Amartya. 1999. Development as Freedom. New York: Alfred A. Knopf.

Stasavage, David. 2005. “Democracy and Education Spending in Aﬁica,” American
Journal of Political Science. Vol. 49. No. 2. 343-358.

Tsai, Ming-Chang. (2006). “Does Political Democracy Enhance Human Development in
Developing Countries?” American Journal of Economics and Sociology. Vol 65. No
2. P. 233-268.

55

Van de Walle, Nicholas. 2001. African Economies and the Politics of Permanent Crisis,
1979-1999. Cambridge: Cambridge University Press.

Van de Walle, Nicholas. 2002. “Aﬁ'ica’s Range of Regimes,” Journal of Democracy.
Vol. 13 No. 2.

Villalon, Leonardo and Phillip Huxtable. 1997. “The Aﬁ'ican State at a Critical Juncture:
Between Disintegration and Reconﬁguration. London: Lynne Rienner Publishing.

Vollmer, Sebastian and Maria Ziegler. (2009). “Political Institutions and Human
Development.” The World Bank, Policy Research Working Paper 4818.

Wantchekon, Leonard. (2003). “Clientelism and Voting Behavior: Evidence from a

Field
Experiment in Benin,” World Politics Vol 55. No. 3.

56

The Sub-National Politics of ARV Provision
Abstract

This paper is an analysis of the location of hospitals that provide antiretroviral drugs
(henceforth ARVS) in South Africa and Nigeria. I review literature on distributive
politics in Aﬁica as well as public policy research on policy inequity and sub-national
healthcare policy choices. As explanatory factors, I consider the various motivations of
policy makers: political, economic, and public health. This paper also challenges the
practice of researchers who rely exclusively on aggregate national statistics in
comparative politics research, thus overlooking sub-national variation. By focusing
analysis on administrative sub-units in two countries, I seek to show that institutions
below the national level are critical in the ﬁght against HIV /AIDS. Not surprisingly,
initial results show that ARV provision is more likely where AIDS prevalence rates are
higher. More interestingly, hospital provision of ARVS also appears to increase in states
or provinces where citizens are represented by leaders who are members of the party in
power. The latter ﬁnding suggests that African governments distribute public health
resources along political lines, thus raising normative concerns and demonstrating the
need for further political science research. Against expectations, the availability of oil
revenues makes hospital provision of ARVS less likely; this implies that the “resource
curse” may undermine the equitable provision of social services. The paper concludes
with a discussion of steps necessary to improve the measurement of indicators and to

ﬁtrther analyze the mix of inﬂuences on ARV provision in future research efforts.

57

Introduction

This paper evaluates why anti-retroviral drug supplying hospitals are located in
certain sub-national regions and not in others. In doing so, I disaggregate the data into
the provinces of South Aﬁica and states of Nigeria to investigate the domestic
distribution of ARVS. When discussed together sub-national territories will be termed
as sub-units or sub-national units. Are hospital resources distributed equitably or are
there political determinants to who gains access to ARVS? This is an important
question for donors, governmental agencies, non-governmental organizations, and
citizens; if these drugs are distributed along political lines, this would raise signiﬁcant
normative questions about ARV policy equity. I will build a model testing whether
political, economic, or public health considerations are the factors driving policy
performance. To build the model I address two ﬁelds of work: the literature on
clientelism and neopatrimonialism from the Aﬁican politics ﬁeld and policy inequity
research drawn mostly ﬁ'om the American state politics literature.

The sub—national unit (provinces in South Aﬁica and states in Nigeria) is the unit of
analysis. This is important for several reasons. First, sub-national analysis establishes
the existence of variance below the national level, which has been established as
important for explaining AIDS expenditures across all Indian states (Lieberman, 2008).
If signiﬁcant variance exists, it would indicate that the current trend in comparative
politics of working exclusively at the national level on AIDS policy issues is missing a
large part of the story and that future work will need to take a more encompassing
approach. There are nine provinces in South Aﬁ‘ica and thirty-seven states in Nigeria

for a total of forty-six sub-units. Increasing the N ﬁom two to forty—six increases

58

leverage over the research question (King, Keohane, and Verba, 1994). Forty six cases
also make quantitative analysis possible across the sub-units.

Second, this paper also offers the opportunity to look comparatively at the
determining factors and mechanisms behind policy implementation and the relative
importance of these factors. Table 1 below shows that 25 million sub-Saharan Aﬁican
citizens are infected with HIV/AIDS it is vital for the future of the African continent that
these people gain access to life sustaining ARVS as a short term policy solution while
longer term plans are developed and prevention efforts expanded. Of these citizens
nearly 9 million live in the two nations discussed below. These two countries account
for about forty percent of African HIV/AIDS cases and nearly twenty ﬁve percent of
world-wide cases.

Table 2.1: HIV/AIDS Statistics and Features

 

 

 

 

 

 

Adults & children
Level of living with Adult Adult & Child ARV Therapy
Analysis HIV/AIDS , prevalence deaths due to Coverage %
rate % AIDS
(prevalence)
World “”314 37.8 million 1.11 2.9 million 42
Sub-Saharan 25 million 7.5 2.2 million 22
Africa

Nigeria 3.6 million 3.1 310,000 13

South Africa 5.3 million 10.9 250,000 21

 

 

 

 

 

 

From a donor institution’s perspective, it is important that access to HIV/AIDS
treatment is as equitable as possible. Equitable delivery would mean that each

individual patient would be equally likely to receive ARV therapy. This treatment could

 

4 World Total and Sub-Saharan Aﬁican Figures are from 2004 from CLA World Fact
Book and the World Health Organization. Nigerian and South Aﬁican statistics are
from 2008 using CIA World Fact Book and the World Health Organization’s statistics
database http://appswho.int/whosis/data/Search.j sp. ARV Coverage statistics are from
the World Health Organization from 2008.

59

extend the life of a recipient patient by an average of 6.7 years, according to Chigwedere
et a1 (2008). If political elites use ARV therapy as a form of patronage, that would raise
important normative questions about whether these resources are distributed equitably.
These questions stem from the Hobbesian ideal that all men are created equal (Hobbes,
1924). They are ﬁrrther codiﬁed by UNAIDS who codify that all patients should be
treated equally, commensurate with a human rights based approach rooted in the idea
that all citizens should have equal access to ARV treatment (Human Rights, 2010).
Contrasting these values, one can imagine political patrons distributing these
resources along political lines. If distribution occurs along political lines, the ideals of
human rights guiding ARV treatment would be undermined. This could necessitate
greater oversight by donor institutions and civil society groups. AIDS is a transcendent
and potentially crippling threat to ﬁrture Aﬁican political and economic development,
and ARVS offer the possibility of slowing the worst effects from this threat while more
effective long term solutions are developed.
Understanding how policies are being implemented is an important endeavor, and this
paper attempts to model these phenomena. Methodologically, the sub-national unit as
the level of analysis allows one to focus upon two countries while providing a large
number of comparable units. In this case, looking at South Aﬁica and Nigeria, we
increase our N size ﬁom two to forty-six, providing greater leverage over our research
question. As will be demonstrated by two diverse literatures focusing on distributive
politics in Africa and public policy equity perspectives, established theory underpins
modeling these concepts, and the available data make the development of a model

possible. The relevant literatures will next be discussed.

60

Literature Review
Why Sub-National Units?

One of the key priorities for the World Health Organization is the strengthening of
health systems within developing countries. In order to most effectively do this, the
WHO has established a district and provincial health system as the most effective means
to bolster capacity (Cassels, 1995). As Cassels argues, sub-national organization
empowers team-oriented responses, promotes local ownership, and develops institutions
that can respond more quickly than a purely nationally based structure (7). Citing
numerous examples ﬁom around the world, Ashford et a1 (2004) make the case that the
largest gap in terms of health care provision is between the rich and poor as they
comparing across wealth quintiles within countries. As Ashford et al. argue, one of the
most effective means for combating such variation is the establishment of more locally
controlled institutions. They argue that such policies link health care providers to
communities and promote implementation adapted to local circumstances.

Braveman and Tarimo (2002) establish that, although measures of sub-national
variation are more limited than at the cross-national level, health inequity is rampant at
the sub-national level. They build a case that income, gender, and ethnicity serve as
effective proxies for individual access and that there are widening gaps of inequity
between groups at the sub-national level, regardless of national level economic
indicators. As Braveman and Tarimo suggest, growing inequality in access to health
care at the national level puts greater pressure on governments competing for
international business to cut spending on health care, widening sub-national inequality

levels and promoting greater levels of health care inequality within countries (2002).

61

More provocatively, Pradhan et al (2001) makes the case that sub-national health care
inequity exceeds national level variation. Pradhan et al. use height of individuals as a
proxy for overall health of citizens, and they establish that sub-national income
inequality is a better predictor of overall health outcomes than cross-national inequality.
Building on sub-national inequality ﬁndings, Gwatkin et a1. (2007) divide health
outcomes by quintiles based on economic standing. As Gwatkin et al. demonstrate,
wealthier groups are much more likely to beneﬁt from greater policy commitments,
leading to greater policy performance for wealthier groups. Their ﬁndings are robust
across regions and among numerous health outcomes, from childhood illness to attaining
a health height. Also analyzing economic standing and based on household level data,
Makinen et al. (2000) ﬁnds that wealthier patients are more able to access care and are
more likely to receive medicines than poorer patients. Their results are based on survey
research within eight developing countries. In their conclusion Makinen et al. call for an
establishment of research indicators that most effectively capture this sub-national
variation so that it may be more robustly cataloged and measured more effectively.
Furthermore, much of the work on HIV/AIDS has discussed how the disease would
threaten the political fabric of the continent with almost exclusive focus on national
level outcomes (de Waal, 2003, Peterson and Shellman, 2005, Singer, 2002). Political
scientists fear that AIDS would weaken nation states to the point that increased
international conﬂict would be inevitable.
While worst-case scenarios have not materialized, the weight of the disease appears to
weigh most heavily at the sub-national level, according to work from Jeremy Weinstein

(2006). Weinstein’s research serves as an example of how the social fabric of African

62

societies—one of the primary assets of Aﬁican countries according to Hyden (2006)—is
undermined by the disease. Weinstein’s study points to how HIV weakens both formal
political institutions and informal societal values. Weinstein’s study looks at subunits in
Uganda and analyzes data on crime and HIV prevalence. This study ﬁnds that subunits
with greater disease burden also suffer from larger crime rates. As Weinstein argues,
this relationship suggests the disease weakens the social fabric of societies at the local
level. This ﬁnding contrasts the expectations of scholars who anticipated macro-level,
nation state conﬂict based on HIV/AIDS weakening national institutions. Such a macro-
level relationship has failed to materialize, though Weinstein’s ﬁndings suggest that
HIV/AIDS resonates more saliently at local levels. Weinstein’s work represents a new
more complex view of the disease and merits further sub-national inquiry.

Further justiﬁcation of sub-national inquiry is provided by state policy literature from
American politics which uses state level analysis to understand policy development. In
analyzing policy outputs across the American states, Erikson, Wright, and Mclver ﬁnd
signiﬁcant variation in terms of policy outcomes and public preferences across the
American states (1993). These authors ﬁnd that state level political ideology is a strong
predictor of policy outcomes. This serves as example an indicating that sub-national
research offers valuable conclusions for other levels of analysis. Ringquist and Garand
(1999) suggest that state policy diversity is a function of the state level differences
discussed by Erikson, Wright, and Mclver as well as external forces like national level
politics and policy learning from other states. Ringquist and Garand also ﬁnd policy
speciﬁc factors inﬂuence state policy change including focusing events and issue

redeﬁnition. Finally, Ringquist and Garand suggest there is variation in terms of how

63

innovative states are willing to be, largely based on state ideology. The work of
Erikson, Wright, and Mclver as well as the paper from Ringquist and Garand suggest
empirical variation at the state level as well as the important lessons that sub-national
research can provide for overall understanding of policy analysis.

The American politics literature provides some important evidence in terms of health
policy at the state level. Barrilleaux (1998) ﬁnds that states expend varying amounts of
resources on health across the states with political ideology, income, and state capacity
as important predictor variables of health care spending. Hero (1998) ﬁnds that racial
and ethnic diversity negatively inﬂuences not only health care spending, but health care
outcomes as measured by infant mortality rates across the ﬁfty states. Cantor (2006)
also suggests American state health outcomes vary as a function of state capacity and
income per capita at the state level. Cantor measures state capacity looking at hospitals
and health care systems, concluding that signiﬁcant variation in terms of both capacity
and outcomes exist at the state level. Clearly, the work of Barrilleaux, Hero, and Cantor
suggests that the state is an appropriate unit of analysis for health care inquiry.

The previous discussion of scholarship focused on sub-national variation provides
evidence that when it comes to health care the inequality, the story extends into the sub-
national context. The sub-national unit is not only relevant based on this inequality, but
as Cassels suggests, sub-national empowerment is seen by international donors as a
means to decrease inequitable trends. Furthermore, Weinstein’s argument implies that
the negative consequences of HIV/AIDS resonate differently within sub-national units.
Finally, the American state politics literature suggests that political factors are important

predictors of health care outcomes which vary across the American states. As will be

64

seen below, indicators for measuring sub-national health care responses are limited as
Makanin et al. (2000) discuss, but variation nonetheless exists and is worthy of further
empirical consideration. I will next discuss how sub-national variation in health care
policy performance ﬁts into the Aﬁican context with reference to literature on
clientelism within Aﬁican states before an expanded discussion on policy inequity more
generally.
African Politics and the Role of Clientelism

One would be remiss in discussing the distribution of highly valued, scarce resources
in Africa without brieﬂy summarizing the importance of patronage networks and the
response of clients. The literature on the subject is pervasive as authors have cited the
importance of neopatrimonial relations between the state and citizens (Bratton and van
de Walle, 1997, Villalon 1998).

Roniger (2004) provides a deﬁnition suggesting that clientelism,
implies mediated and selective access to resources and markets from which others are
normally excluded. This access is conditioned on subordination, compliance or
dependence on the good will of others. Those in control—patrons, subpatrons, and
brokers—provide selective access to goods and opportunities and place themselves or
their supporters in positions from which they can divert resources and services of favor.
Their partners—clients—are expected to return their benefactors’ help, politically and
otherwise, by working for them at election times or boosting their patron’s prestige and
reputation,” (353, 354).

Harold Lasswell described politics as who gets what, when, and how (Lasswell,
1936). Roniger’s deﬁnition of clientelism goes a long way to describe how power
relationships function across Aﬁica. Citizens subjugate themselves to authority in order

to access political goods. In other words, elites use their resources and often the

resources of the state in order to legitimize themselves and their regimes. This type of

65

distribution leads to some citizens within a respective country included in the
distribution of resources while others are not.

Bratton and van de Walle (1997) discuss how clientelism manifests in African
politics. As they discuss, African’s political power is largely concentrated in the hands
of individual leaders, a concept known as presidentialism (63). Leaders use patronage in
the form of state resources to legitimate themselves and their regimes (66, 67). When
concentrated presidential power is combined with clientelism and the distribution of
state resources, these three features are known together as neopatrimonialism (63-68).
Both the Bratton and van de Walle and Hyden descriptions refer to neopatrimonialism as
one of the preeminent informal institutions of African politics.

As Bratton and van de Walle argue, however, “when patrimonial logic is internalized
in the formal institution of neopatrimonial regimes, it provides essential operating codes
for politics that are valued, recurring, and reproduced over time,” (63). As Bratton and
van de Walle detail, informal institutions resonate and become a dominant feature of
formal institutions. Bratton and van de Walle also discuss the fact that although
clientelism is controlled by the political center, it is not only a feature of the top of
political regimes. Rather, “this [clientelism] happened at every level; at the top, the
ruler’s faithful political aristocracy was rewarded with prebendal control of public
ofﬁces, monopoly rents, and the possibility of creating its own clientelist
networks. . .Nor were patronage and clientelist beneﬁts limited to the political
aristocracy,” (66). Instead of thinking of patronage networks exclusively in terms of
relationships between individual rulers and citizens, multiple levels of governance exist

within the formal institutions. This allows for potential patronage distribution not just

66

directly from top to bottom but throughout the multiple formal institutions of
government. Neopatrimonial relations may be just as prominent in local and sub-
national governance as at the national level.

Moreover, Villalon (1998) also mentions clientelism as one of the premiere
features of the modern Aﬁican state. As he describes it, patrons reward clients that help
empowered regimes to consolidate control and promote political stability (13). Political
goods are allocated with preferential treatment to loyal groups (13). As Wantchekon
(2003) argues, clientelistic messages are especially salient at election time. Candidates
seek electoral support, and clientelistic messages are often effective in increasing vote
share, especially for the ruling party as individuals seek access to the state’s resources.
The concept of regimes utilizing state resources particularly at election time to sway
voters towards the ruling regime is known as the political business cycle (N ordhaus et
al. 1989). Through elections formal democratic institutions serve as a process through
which Aﬁican leaders legitimize themselves, with patron-client networks utilizing state
resources as patronage. This phenomenon serves as yet another example of the informal
institution known as clientelism subjugating formal elections.

As the political business cycle demonstrates, and Bratton and van de Walle discuss,
neopatrimonialism can resonate in regimes, with or without elections. Bratton and van
de Walle discuss the idea that clientelistic networks exist across regime types (7 7-82).
Both Bratton and van de Walle and Roniger (2004) discuss the idea that increased levels
of clientelism are negatively correlated with the development of democratic institutions.
In other words, in order for countries to transition and become more democratic,

clientelism and access to the state through patronage networks must be broken.

67

What does clientelism, such a key feature of Aﬁican regimes, mean for Aﬁican
policy making processes? Hyden summarizes rather poignantly how clientelistic
patronage networks have subjugated aggregate public policy in Aﬁican countries:

Aﬁican countries do not have policy governments, but public institutions operating
on the basis of patronage. These governments conduct their business not with a view to
implementing ofﬁcially agreed-upon policies, but look to rewarding individuals and
groups that have shown exemplary loyalty or contributed to the political success of a
government leader. In short, resources ﬂow along very different paths than those that
are identiﬁed in ofﬁcial statements, be that a policy announcement or the national
budget. The result is that Aﬁ'ican governments tend to look to the past rather than to the
future. To the extent that policies feature in politics, they are more often for window-
dressing purposes than for real implementation, (229,230).

As Hyden argues, western notions of the policy making process apply less well in
Africa given the predominance of clientelism. Rather than a focus on policy for
development’s sake, implementation more often follows the informal processes related
to clientelism. Patron-client networks give rise to factionalism with groups loyal to the
regime considered insiders while those who compete with the empowered regime left
outside of patronage networks. This notion of political insiders and outsiders is
described by Bratton and van de Walle with transitions away ﬁ'om neopatrimonialism
towards liberalization and democracy often occurring as access to patronage wanes and
factionalism can no longer be contained within the established networks (84).

Clientelism is antithetical to enhanced governance and democratization. As Hyden
(2006) puts it, “Political rulers treat the exercise of power as an extension of their private
realm. ...Clientelism is deemed problematic, especially in circles that are concerned with
improving governance in African countries. It keeps African countries barely aﬂoat, but

it does not help them swim forward,” (79).

In the political science literature on clientelism in Africa much of the focus has been

68

on resources like jobs, infrastructural projects like roads. But given the importance of
ARV medications to infected citizens, I will investigate whether patron-client relations
inﬂuence the distribution of AIDS treatment resources. With donor institutions focused
on AIDS as an aid issue it is important to investigate whether Aﬁican leaders have
adapted ARV distribution programs to ﬁt into their patronage needs. I investigate
whether the resources used to ﬁght HIV /AIDS have been distributed along political
patronage lines. Speciﬁcally, have ARV programs been located in areas that are
supportive of respective ruling regimes in sub-national units within South Aﬁica and
Nigeria? If so, how do patron-client networks function? As will be seen below, the sub-
national unit is a salient unit of analysis, not only as was described above in terms of its
variation, but also in terms of its function as a key layer of AIDS policy implementation.
Whether ARV resources are distributed to actually ﬁght the disease’s effects, or whether
medications are doled out as part of patron-client relations, usurping their intended
purpose is to be determined empirically below.
Public Policy Research on Policy Inequity and Sub-National Politics from the American
Perspective

The uneven distribution of political goods is not exclusively an African issue. There
exists a large body of literature from American politics and comparative public policy
that has looked at the issue of inequity in goods provision. Much of this literature has
looked at health policy, but relevant work from other areas is also illuminating. I will
brieﬂy review some of the most pertinent literature before I integrate this with the AIDS

literature and discuss the model.

69

Some of this literature looks at the US. from a cross-national perspective, focusing
on why the US. has done so poorly relative to other developed countries in terms of
health care outcomes (Raphael, 2007). Raphael analyzes what factors lead to poor
public health outcomes and inequity across different segments of American society
(2007). Raphael links racial and ethnic minorities and low income to poor health
outcomes across a group of measures. Indicators include life expectancy and infant
mortality rates among others. Raphael argues that public policy intervention which
prioritizes equitable distribution of health resources to minority and impoverished
groups are critical determinants of public health outcomes in the United States. As
Raphael argues, “It is increasingly recognized that the quality of numerous social
determinants of health are shaped by decisions governments make — that is, public
policy— on how to allocate resources among the population,” (Raphael 2003). As
Raphael details, the US. has targeted minority and impoverished groups and the
symptoms of health problems rather than the structural determinants and causes of
inequality. Raphael’s analysis suggests several important conclusions. First, public
policy interventions are highly inﬂuential in determining social outcomes. Second,
cleavages exist between groups with economic and minority status linked to whether
‘ citizens can access health interventions.

Rom (2008) also looks at the American states. He measures policy determinants
according to political, economic, and demographic factors (324). Rom’s categories were
previously developed in Bailey and Rom (2004) to explain how states compete for
redistributive resources (344, 345). Politically, Rom’s analysis considered political

culture, ideology, institutions, and public opinion ﬁnding that more conservative states

70

spend less on health than more liberal states. Economically, the model considers both
cyclical and structural considerations ﬁnding that states with larger economies have
more developed health care institutions. When economies perform better, states are
more likely to expand health care institutions (Rom, 2008; p. 323). Demographically,
Rom focused on state poverty rates with poorer people relying more heavily on the state
for health care beneﬁts. Overall state wealth is also important as wealthy states are more
likely to offer larger payments and greater services (323). He also ﬁnds signiﬁcant
variation in terms of state spending per citizen on health policy implementation across
the American states (328). Rom also ﬁnds that, in the United States, political disputes
often exist between the states and the federal government over what must be provided
and who must fund expenditures. As Rom argues, there will be questions on which the
state and federal governments agree and others in which they conﬂict (321). Rom’s
ﬁndings suggest that health care outcomes are determined not only by social factors like
those described by Raphael. Rather political, economic, and demographic variables
together inﬂuence outcomes.

Schoen and Doty (2003) focus on ﬁve developed nations and public opinion survey
responses by citizens about their health care experiences. Their project also indicates
the existence of signiﬁcant cross national variance in health care system design, even
among the most developed nation. Their ﬁndings suggest that countries that provide
universal health care have more equitable distribution of access to health resources than
in countries like the US. where health care is largely an individual responsibility.
Schoen and Doty ﬁnd that countries that have developed public programs that prioritize

equitable health care distribution like Canada and New Zealand have led to more

71

equitable outcomes than in countries like the US. that have not established equitable
programs. When controlling for insurance access, they found that income was positively
related to health care outcomes more strongly in the US. than in countries with
universal health care. In universal health care cases the relationship between income
and health care satisfaction were insigniﬁcant. This serves as yet another example of
how government interventions can inﬂuence health care outcomes.

Evan Ringquist’s work on environmental policy is a good example of the local
dynamics involved with public policy implementation (Ringquist, 2005). He argues that
racially underprivileged areas are more likely than other groups to have environmentally
hazardous facilities built near them. His work also ﬁnds support for socioeconomically
disadvantaged citizens more likely to have environmentally damaging institutions built
near them. Ringquist ﬁnds that racial minority status is the most inﬂuential variable as it
is more likely that environmentally damaging facilities will be built in areas inhabited by
minorities. This paper makes it clear that racial and economic variables are important
inﬂuencing forces behind policy equitability. Although environmental cases represent
negative cases where location sites were chosen in impoverished and minority inhabited
areas, the same logic could apply to more desirable institutions. This would mean that

valuable resources like hospitals, police stations, and schools would be more likely to be

pnontrzed 1n areas controlled by affluent and racral majorrty crtrzens .

 

5 Gamble and Stone (2006) similarly looks at the role of racial and ethnic disparities in
public health across the American states. While their focus is on issue framing using a
qualitative research design, they make it clear that similar to Ringquist, issues of race
and ethnicity are equally salient in public health policy debate. Their work is more
historically and contextually focused, but their similar ﬁndings indicate that race and
ethnicity are important issues for public health considerations, and I will attempt to build
these issues into the model.

72

So what can be taken systematically from this unique and diverse set of public policy
investigations? It is clear that racial, ethnic, and socio-economic considerations are
important considerations that determine who has access to public programs. Variables
that effectively measure these concepts will be necessary to model the placement of
ARV hospitals. While there is less consistent support, Rom, Ringquist, and the Doty
and Schoen papers also indicate that economic elements should also be taken into be
tested. As these authors suggest it is a combination of political, econonric, and
demographic inﬂuences that determine who wins and who loses when goods are
distributed.

Synthesizing the Literature

When one takes into account both the literature on clientelism and that on public
policy inequity, several trends are clear. As both of these literatures discuss, not all
citizens have equal access to political goods. The literature on clientelism suggests that
maintaining access to the regime is a signiﬁcant determining factor in one’s ability to
access political goods meaning that political variables are highly inﬂuential. Succinctly
put, political insiders and those loyal to the regime will be more likely to gain political
goods than those outside patronage networks. Hyden goes so far as to suggest that the
policies themselves are constructed to serve the regime’s needs rather than development
based decision making.

The inequity literature also suggests that other cleavages exist that are inﬂuential in
determining who is able to access goods. When considered together, the inequity
literatures suggest that racial and ethnic minorities and economically disadvantaged

citizens are less likely to be prioritized for distribution by their respective governments.

73

Rom (2008) helps to mesh these literatures in his division of variables into political,
economic, and demographic categories.

One author has synthesized these questions to look at the effectiveness of AIDS
policy at the sub-national level. Evan Lieberman looks at the effectiveness and diversity
of policy outcomes across the states of India (2008). Similar to the literature on policy
inequity, Lieberman (2008) conﬁrms the importance of ethnicity in AIDS treatment
availability. His work suggests that ethnically heterogeneous areas are less effective at
developing effective AIDS policies. Though his theory is developed to explain cross-
national variation in AIDS policy responses, he ﬁnds that less diverse areas were faster
in developing HIV/AIDS policies. His work serves as yet another example of the sub-
national variation in public policy, similar to the sub-national ﬁndings described by the
literature that suggested sub-national variation in health care access on pages four
through six (Cassels, 1995; Pradhan, 2001; Braveman and Tarimo, 2002; Ashford et a1.
2004; and Gwatkin et al 2007).

More importantly, Lieberman describes a mechanism in which ethnically divided
groups are unable to solve commitment problems. Rather, as Lieberman describes,
political debates over responding to the disease break down into a blame game with the
focus on how the disease evolved rather than on policy solutions (208). The
development of policy programs to test for and treat AIDS patients would be
synonymous with admission of the disease’s existence. This type of stigma provides a
barrier to effective policymaking. Caught in the middle of these debates, rational
policymakers have sought to limit attention to the disease and blame other areas for the

cause rather than delving into these controversial issues (216-220). Lieberman tests

74

these theories using both quantitative and qualitative analysis. His results suggest that
areas with greater ethnic diversity and ﬁactionalization spend less per capita on
HIV/AIDS policy responses than areas that are less ﬁ'actionalized (227, 228).
Lieberman’s analysis puts together some of the key building blocks that have been
discussed so far in this paper. Namely, there exists signiﬁcant variation in terms of
HIV/AIDS policy responses. Variation is not limited to the national level. Rather, sub-
national variation is rampant as well. Lieberman and the literature on policy inequity
suggest that ethnic and socio-economic statuses are largely determinant of individual
access to these goods. Rather than focusing exclusively on inequity variables, the
clientelism literature suggests that it access to patronage networks that is determinant of
effective policy responses. When these ideas are combined, one can test what
combination of these variables drives ARV provision at the sub-national level.

As WHO data suggests, no Aﬁican country has managed to supply every citizen
requiring ARVS with the necessary drugs. When this is the case, how are these scarce
resources allocated? I intend to address this research question in by looking at sub-
national ARV distribution across South Africa and Nigeria, countries selected for their
critical gee-political importance on the continent. In the next section, the cases will be
discussed.

Case Selection, Justification, and Discussion

First, from a gee-political perspective, South Africa and Nigeria are the largest and
most dominant economies on the Aﬁican continent as Figure 2 indicates. These two
economies vastly outperform the economies of other African nations, and, as regional

powers, their performance is critical to the economic welfare of the other nations in

75

southern and western Aﬁica, respectively. Also, South Africa and Nigeria are two of
the countries with the largest number of AIDS cases in the world.

Table 2.2: GDP and AIDS Data from South Africa and Nigeria

 

GDP Adjusted Sub-Saharan World

Power Pari Africa Ranking

Country for Purchase AIDS Cases

 

 

For this analysis I will look at all of the sub-national units within South Africa and
Nigeria which combine for a total of forty six cases: thirty-seven states in Nigeria and
nine provinces in South Africa.

Figure 2.1: Key Political Similarities and Differences Between South Africa and
Nigeria

 

Table three reveals some important cross-national similarities and differences

between these sets of cases. Underpinning the importance of sub-national politics is the

 

6 Data from the CLA World Fact Book; Data are from South Aﬁica and Nigeria.
76

fact that both South Africa and Nigeria are among the few federal governments in
Aﬁica. Both countries have salient historical cleavages with South Aﬁica’s history
dominated by racial strife. Nigeria also has a history of ethically based political parties,
according to Aborisade and Mundt (1998). Both have multiple political parties, and
histories marred by political violence. South Aﬁica and Nigeria have both
demographically heterogeneous and historically relevant ethnic and racial divisions.
Each also has signiﬁcant variance among sub-national units in terms of HIV prevalence
rates. Finally, with both countries relying economically on natural resources (oil in
Nigeria and gold and platinum in South Africa) these similarities make for an interesting
opportunity to analyze the role such natural resources play in AIDS policy
implementation.

There are a few limitations to pairing these two countries. First, South Africa has a
proportional representation electoral system, while Nigeria utilizes ﬁrst past the post.
South Aﬁica has an independent judiciary, an institution that played a key role in ARV
provision in that country as will be detailed below. Nigeria’s judiciary is not
independent. South Africa is considered by Freedom House to be a democracy, while
Nigeria is considered to only be partially free. South Aﬁica emerged as a democracy in
1994 following a long protracted elite pact process and has remained democratic.
Nigeria’s political history has involved a series of military coups, elite struggle, and a
history of conﬂict between military and civilian leadership. While South Africa’s
political history has been ﬁlled with conﬂict as the country transitioned from white rule
to a more representative regime, Nigeria’s history has been through more ﬁts and starts.

Since the 1994 transition, South Aﬁica has been more consolidated than that of Nigeria. -

77

South Aﬁica is also more economically developed and diversiﬁed than Nigeria. South
Aﬁica also has a more dominant single party than Nigeria. While HIV/AIDS is a
signiﬁcant problem in both countries and varies in prevalence across the country, it is
more pervasive in South Aﬁica than in Nigeria.

The HIV/AIDS Treatment Story in Nigeria and South Africa

The respective sub-national units play similar roles in both countries as will be
discussed next. Both have national level committees that lead the Nigerian and South
Aﬁ'ican AIDS responses, but the sub-national unit is the primary level of
implementation of these national policy decisions, according to the World Health
Organization (WHO, 2005, Progress Report, 2006).

In Nigeria, the state is the primary irnplementer of AIDS policy, and the various
states of Nigeria suffer from an overall lack of capacity (WHO, 2005). The Nigerian
states along with the federal government and local government areas share in the
responsibility for health care service (Johnson, 2000). V0 gel describes these
responsibilities as follows:

Health care provision in Nigeria is a concurrent responsibility of the three tiers of
government in the country. However, because Nigeria operates a mixed economy,
private providers of health care have a visible role to play in health care delivery. The
federal government’s role is mostly limited to coordinating the affairs of the university
teaching hospitals, while the state government manages the various general hospitals and
the local government focus on dispensaries, which are regulated by the federal
government (Vogel, 1993, p. 18).

The federal government serves to provide overall policy goals and technical expertise,
but actual ﬁnancing and implementation of health care goals including HIV /AIDS

policy is the responsibility of the individual states which leads to divergent outcomes,

according to the same report. This report also states that hospitals are funding by a

78

combination of state sources, by for proﬁt services, and through foreign donations with
limited technical assistance from the federal government. According to their 2007
report to the UN, Nigeria’s overall government expenditure on ﬁghting AIDS was 7.4
billion naira with foreign donors spending 5.9 billion naira indicating that the
government is the primary source of AIDS funding in the country (UN GASS, 2007).
This means that over 55% of AIDS funding was from national sources. UNAIDS has
stated that it is their goal to assist states in developing individual state agencies as they
view states as the most effective unit for scaling up treatment efforts (UN AIDS, 2008).
The vast majority of Nigerian state revenues come from the federal government with oil
revenue accounting for the vast majority of this budgetary support (Aborisade and
Mundt, 1998). Initially, ARVS were only available to those wealthy enough to
purchase them privately, and the ﬁrst government sanctioned ARV program did not
begin until 2002 (Odutolu et a1, 2006). Initial demand for drugs vastly exceeded
supplies, and ARVs were distributed ﬁom only twenty ﬁve facilities nationwide. Most
of these facilities were in either Abuja or Lagos. In 2006, Nigeria opened an additional
41 ARV facilities (Nigeria Opens, 2006). The country has set numerous targets for its
AHDS treatment programs over the years, most notably that they strive to reach eighty
percent treatment coverage to AIDS patients by 2010, a goal they are unlikely to reach
(WHO, UNAIDS, & UNICEF, 2008). The country has expanded AIDS treatment
facilities to 197 clinics, according to a list available from the National Agency for the
Control of AIDS website.

In South Aﬁica similarly the national government sets the national policy for ﬁghting

AIDS, and the provinces along with municipalities, the private sector, and community

79

organizations are responsible for the actual implementation of these policies, according
to a 2006 governmental progress report to the UN General Assembly (2006). The South

African government lists the following facts about the country’s health care system:

Public health consumes around 11% of the government's total budget, which is allocated
and spent by the nine provinces. How these resources are allocated, and the standard of
health care delivered, varies from province to province. With less resources and more
poor people, cash-strapped provinces like the Eastern Cape face greater health
challenges than wealthier provinces like Gauteng and the Western Cape (Health Care,
2009)

As can be seen here, the South African government describes the variation in health
care availability across the provinces of the country. The South Aﬁican government
also lists the provinces as largely responsible for health care, as is described below. The
following is taken from South Aﬁica’s Department of Information, describing provincial

responsibilities for health care:

Provincial health departments provide and manage comprehensive health services at all
levels of care. The basis for these services is a district based PHC [Primary Health Care]
model. The major emphasis in developing health services in South Africa at provincial
level has been the shift ﬁ'om curative hospital-based healthcare to that provided in an
integrated community-based manner.

The provincial health departments are responsible for:

providing and/or rendering health services

formulating and implementing provincial health policy, standards and legislation
planning and managing a provincial health information system

researching health services to ensure efficiency and quality

controlling quality of health services and facilities

screening applications for licensing and inspecting private health facilities
coordinating the funding and ﬁnancial management of district health authorities
consulting effectively on health matters at community level

ensuring that delegated functions are performed.

The national department assists provincial health departments to develop service-
transformation plans to reshape and resize the health services and develop appropriate,

80

adequately resourced and sustainable health service-delivery platforms, which are
responsive to needs (Provincial Health, 2010).

As can be seen from the last part of this statement, the national department of health
coordinates priorities across the country, but it is the provinces that are responsible for
the day to day management of health care availability for citizens.

The history of ARV provision in South Aﬁica is controversial and involved a lengthy
history which will be brieﬂy summarized. African National Congress leadership was
initially unwilling to agree to provide ARVS citing cost. Despite their preferences the
only province not under ANC control, Western Cape began a pilot ARV program in
1999 (Viall, 1999). A citizen action group known as the Treatment Action Campaign
(TAC) ﬁles a Constitutional Court case against President Thabo Mbeki’s government in
2001, claiming that drugs should be provided to HIV positive mothers to prevent disease
transmission to unborn children (AVERT, 2010). The courts sided with TAC both
initially and after a government appeal. This set of events serves as a key example of
how democratic institutions paved the way towards ARV provision, despite the
government’s preferences. 2002 and 2003 saw the government block the
implementation of a drafted national treatment plan (AVERT, 2010). TAC staged
numerous demonstrations to protest these decisions, and it was not until 2005 that
manufacturers were selected to provide ARVS as part of a comprehensive rollout
operation (AVERT, 2010). By 2007, the government launched a new initiative with the
goal of providing ARVS to eighty percent of those who needed them by the end of 2011
(Department of Health, 2007). Similar to Nigeria, the country has struggled to reach
targeted goals for numbers of patients treated as established in this plan.

Similar to Nigeria as well, in South Aﬁica there is signiﬁcant variation in the level of

81

AIDS cases across the provinces of South Africa which necessitate different types of
responses. While the UN stated that enhancing capacity was a key challenge to Nigerian
scaling up of treatment efforts, a 2008 report suggested that bolstering provincial AIDS
councils was among their goals to best enhance treatment efforts. South Aﬁica was able
to ﬁnance almost 70% of their own AIDS funding with donors accounting for about
30% of 2006 AIDS spending, according to the 2006 progress report. This is about 15%
more than in Nigeria where donors account for a larger proportion of spending, however
the key points are as follows: there exists key variation both in terms of burden of
disease and effectiveness of policy implementation across the sub-national units in these
two respective countries, and the sub-national unit is the key implementer of national
level HIV/AIDS policy in both countries. I will next lay out my hypotheses, variables,
and issues of measurement.

Variables, Measurement, and Hypotheses

Dependent Valiablg I want to explain why ARV programs are available in certain sub-
national units and not in others. It is important to note that the majority of the facilities
discussed are not new institutions. Rather, ARV programs have been established in a
small percentage of hospitals across these two countries. For example, there are over
23,000 total medical facilities across Nigeria (National Bureau, 2005). Of these, slightly
less than 200 provide ARVS to the public, according to their national HIV /AIDS Agency
(N ACA, 2008). While the focus is on hospitals, it should be stressed that it is the ARV
program that is the new institution. The policies calling for their distribution and the
implementation did not occur until 2002 at the earliest in Nigeria.

My dependent variable indicator is the sub-unit population divided by the number of

82

ARV providing hospitals in each sub-national unit. This indicator cuts through the
epidenriological complexity by focusing upon public policy choice based on the primary
delivery mechanism for ARVs: the hospital. Lists of medical facilities are the best

available source to assess the delivery of ARVS as these facilities are the institution that

provides these medications]. The difﬁculty with the indicator is that it does not
measure the variance in hospital capabilities or differentiate for size of these institutions.
Still, I believe this indicator is the best measure that gets at the equity or inequity of
citizen access to ARVS. I also recognize that it does not capture the full scope of what is
required to provide ARVS, but this is a measure of institutional infrastructure which is
what this paper attempts to model. In the regression analyses below, a linear
transformation was performed so that citizens per hospital are in terms of hundreds of
thousands of citizens per hospital. This was done to make interpretation of results easier
as initial numbers had extremely large coefﬁcient values.

Independent Variables: Independent variables are divided into several categories:
political variables, economic variables, and public health variables. These variable
measures are theoretically based on Rom’s (2008) typology and also following
Lieberman’s (2008) conceptualization of in group/out group relations.

Political Variables

 

7 A list of ARV facilities in Nigeria was taken ﬁ'om the National Agency for the Control
of AIDS website. It was updated as of 2008, and I would consider it to be the most
reliable source available. A similar list of facilities was taken ﬁ'om the South Aﬁican
Joint Civil Society Monitoring Forum’s website. This list was updated as of August
2007, so the lists are based on similar time frames. This institution is a combination of
several civil society organizations which claims to be dedicatedto monitoring the
government’s implementation of the National Operational Plan.

83

To capture political dimensions across units, I code whether the chief executive in
each sub-unit is of the same party as the party in power at the national level. These
positions are ﬁlled by governors in Nigeria and premiers in South Africa. In both
countries, these executives are responsible for signing state level bills into law,
establishing sub-unit level cabinets, and implementing legislation passed by each sub—
unit’s legislative body. Lists of these individuals are readily available from numerous
sources. I also measure the percentage of national legislators from each sub-unit who
are members of the party in power. Coding for both sub-unit executives and legislators
can be considered a proxy for sub-unit relations with the overall national government as
one would expect better relations when all members are of the same political party.
Both of these measures are coded as dummy variables, coded l for members of the
ruling party and 0 for members of other political parties. This is based on the
conceptualization of political in groups ﬁom Bratton and van de Walle and serves to

capture elite level relations between these individual actors (1997).

Finally, I measure the percentage of provincial vote in the last presidential election8
for the ruling party to see whether loyal sub-units are able to secure more hospitals for
their units to ﬁght AIDS. This assesses direct relations between the national government
and citizens. When the presidential vote measure is combined with the other executive

and legislative measures, this comprehensively captures relations between the national

 

8 Because of the controversial nature of the 2007 Nigerian election, the results from the
2003 results were used. Final results from the 2007 election have not been accepted as
the election was not considered free and fair. One limitation of the data available was
that only votes from the People’s Democratic Party and All Nigeria People’s Party were
available for coding. These parties received over 94% of votes in that election, but if
other parties ran particularly strong in certain states, this variation is missing in my
measures. This was a limitation in the data available. Raw data were taken ﬁ'om
Ojarneruaye (2003), but other parties involved in the contest were not coded.

84

government and the sub-units. Additionally, these political measures measure both
which political players are salient and whether politically motivated patronage networks
resonate in ARV distribution as the literature suggests.
Economic Variables

For a proxy of state economic interests inﬂuencing ARV provision, I use dummy
variables to code for sub-units with signiﬁcant natural resource revenues, namely those
containing gold or platinum in South Aﬁica and oil in Nigeria. The idea here is that
sub-units with signiﬁcant natural resource deposits will prioritize keeping workers
healthy to ensure a steady ﬂow of state revenue from these resources. For example,
South Aﬁican gold mining ﬁrms were among the ﬁrst private sector (Paris, 2004). In

states lacking these resources I would expect sub-units to be less focused on ARV

provision. I also measure individual unit total budgetary spending9 per capita. The
budgetary data allows an assessment based on total spending on individual citizens. This
measure captures the agency of policy makers and also helps to control for the
availability of resources in given states.
Demographic and Public Health Variables

To measure the demographics of ethnicity, I use Demographics and Health (DHS)
survey data ﬁ'om both Nigeria and South Aﬁica to assess ethnic heterogeneity across the
individual sub-units. Numerous measures were considered including census data as well
as data ﬁom the Aﬁ'obarometer. The DHS surveys offered the largest data sets with

over 7700 cases in the Nigerian data set and over 9000 in the South African set.

 

9 Data were taken from the South Aﬁican Treasury website and ﬁ'om the Business
Environment and Competitiveness Across the Nigerian States Initiative, a project
launched by the Aﬁican Institute for Applied Economics since 2003. More information
is available in the data source section of the paper’s concluding references.

85

Although the DHS sub-unit samples are small (averaging around 200), they are larger
than any other survey set available and were of superior quality to census data. The
most recent census in Nigeria that actually released ethnic demographic data at the sub-
unit level was from 1963. This was assessed to be out of date and likely politicized, and
though I recognize the DHS data is imperfect, it was deemed the best measure of
ethnicity available at the sub-national level. To measure ethnicity, the size of the
largest ethnic group in each sub-unit is used. I also coded the size of second largest sub-
unit, but exploratory analysis indicated that the size of the largest group was a more

effective proxy. To measure a second public health variable, I include the HIV

prevalence rates in each unit"). This measure of prevalence captures whether areas of
greater disease burden are receiving greater institutional assistance.

Control Variables: I control for population density, dividing the population of individual
units by the total land mass of respective areas. This was deemed necessary to account
for differences between urban and rural access to hospitals, a challenge across the
developing world.

These measurement choices are not without problems; some have been brieﬂy
mentioned, and I reemphasize thenr here. First, the indicator of the dependent variable
takes into account hospital numbers, but does not capture the reach of hospital activities
or the other services they provide which are critical for the success of ARV success.
Also, the dummy variables for natural resource deposits are not perfectly speciﬁed, but

they are theoretically based and represent the type of justiﬁable ignorance that dummy

 

'0 Sub-unit prevalence data was taken for South Aﬁica from a 2005 Treatment Action
Campaign document and from Nigeria from a 2005 Nigerian Federal Ministry of Health
document. Both are included in the data reference section at the conclusion of the paper.

86

taria

Vida

the

El

variables help capture (see for example Carsey and Wright, 1998). But each of these
variables is measured in currently available datasets and these measures will be included
in the bibliography.

To test hypotheses, I run OLS regression models. While the greatest contribution of
this model is to offer an explanation of sub-national variation, it also helps us compare
the independent variables systematically. For instance, if ethnicity is the main
explanatory factor, Lieberman’s (2008) conclusions about ethnic heterogeneity
preventing effective policy implementation will be conﬁrmed. But if political variables
dominate, the clientelism literature will be demonstrated to have a better comparative ﬁt.
Again, to emphasize the point, this is an exploratory analysis to establish sub-national
policy implementation variation. 1 will ﬁrst discuss my hypotheses before discussing
the regression results.

Hymtheses
Political Hypotheses

H1: Sub-national units represented by executives of the same political party as the
president will be more likely to have greater number of ARV providing hospitals than
areas with governors from other parties.

H2; Units represented by legislators that are of the same political party as the
president will be more likely to have greater numbers of ARV providing hospitals than
units that are represented by members of other political parties.

H3; Citizens living in units which voted for the ruling regime in presidential elections
will be more likely to receive ARV providing hospitals than in states that supported

other candidates.

87

The political hypotheses are based on the tradition of patronage networks in African
politics (Bratton and van de Walle, 1997, Wantchekon, 2004). Aﬁican politics are often
characterized by centralized authority with resources largely distributed along
clientelistic relationships (Villalon and Huxtable, 1997, Hyden 2006). This hypothesis
tests whether such patronage networks exist in ARV treatment policy and whether these
resources are distributed along politically-motivated lines, and if so which political
actors are most salient in the distribution of these valuable resources. To assess these
variables, party membership of executive leadership in each sub-unit was coded as either
a member of the ruling regime or as a non-member. Legislative delegations were coded
based on the percentage of legislators ﬁ'om a given unit and whether these members
were of the same party as the ruling regime. To assess whether presidential election
voting mattered data from the 2003 Nigerian presidential election and the 2004 South
African presidential election are differentiated to see whether units whose majorities
supported the ruling regime were more likely to receive ARVS.

Economic Hypotheses

H4: Units where oil, gold, or platinum are mined will be more likely to have ARV
hospitals than units that do not have these resources.

H5: Units with higher overall spending per capita will have more ARV providing
hospitals. .

Oil in Nigeria and gold and platinmn in South Aﬁica are the most valuable resources
in these two countries which provide valuable revenue to the sub—national governments
and the national governments alike. Budgetary spending per capita was available for

each sub-unit, although it is not exclusively health spending. This does provide a

88

measure of the available resources and proxies the general administrative capacity for
individual sub-units. From a political economy perspective, it stands to reason that units
with natural resources will be more likely to have more ARV providing hospitals than
units without these resource deposits as governments have the incentive to keep these
revenue generating citizens healthy to maximize production. For instance, in South
Africa some of the ﬁrst actors who supported ARV provision were mining companies
which have gone to great lengths to keep their workers healthy and productive.
Demographic and Public Health Hypotheses

H6: Sub-national units that have greater HIV prevalence rates are more likely to have
more ARV providing hospitals than states with lower HIV prevalence rates.

H7: Sub-units with more dominant ethnic groups are more likely to have more ARV
hospitals than units with more heterogeneous ethnic composition.

This is based on the assumption that policy choices will be made to prioritize ARV
assistance in areas that are heaviest hit with AIDS patients. H6 stems from the desire,
both by Aﬁican elites and by international donors (that provide thirty and forty-ﬁve
Percent of the AIDS ﬁrnding in these two respective countries) are going to require that
diStribution get to areas with the highest burden of disease, i.e. areas in greatest need of
interventions. H7 stems from Lieberman’s (2008) analysis where he ﬁnds that ethnic
heterogeneity creates in-group/out-group cleavages that limit AlDS treatment policy
incentives. Dominant ethnic groups should make it easier for respective sub-units to
solve these policy dilemmas and make provision of these resources possible to a larger

degree than in more fractionalized regions. Although I hypothesize that patronage plays

89

a role in the allocation of these supplies, monitoring and evaluation by the international

community should ensure that these areas receive care to some degree.

Establishing Sub-National Variation

Figure 2.2: Map of ARV Programs in Nigeria

 

90

Figure 2.3: Map of ARV Programs in South Africa

Western
Cape

 

As Maps 1 and 2 shows the distribution of hospitals with ARV programs across units
and illustrates substantial variation within the two countries. It bears repeating that
although the overall hospitals may have been built prior to democratization, ARV
programs have been allocated within the last few years. ARV programs exist within the
preexisting hospital structure, but they are only located in a small portion of the total
number of medical facilities. Sub-national variation within each country is evident.
South Aﬁican provinces have more hospitals than Nigerian states, but this is not

surprising given their higher HIV prevalence rates and more developed economy.

Regression Results

91

One point should be made before we get into the regression analysis. The signs of
the coefﬁcients make for confusing relationships. Negative coefﬁcients indicate that
fewer citizens are being served by a given hospital. These negative relationships should
be viewed as normatively beneﬁcial. Fewer citizens per hospital should make it more
likely that a given individual will be able to access necessary ARV resources when they
are needed. A smaller population per hospital results in less pressure on the health care
system, and should make for greater beneﬁt and likelihood that needy citizens will gain
access to treatment.

Model 1 establishes a baseline model with only the political variables included. It
illustrates the fact that political explanations alone can account for a substantial portion
of the variance across the sub-units. We can infer from the strong and signiﬁcant
relationship on two of the three political variables that patronage is the driving
explanation for the distribution of ARV resources. More precisely, the election and
representation of intermediate (governors or MP8) political elites who are members of
the ruling regime explains how ARV programs are allocated rather than votes for the

president’s political party in presidential elections as part of a more direct relationship.

92

Table 2.3: OLS Regression Results for Citizens Per Hospital

 

Political

Economic

Public Health

Demographic

Control

R Squared
Adj. R Squared

 

Constant

Governor's Party Same as President

% of Leg. Delegation Same Party as
Pres.

Unit Vote Last Pres. Election

Oil Producing State

Gold Producing State

Platinum Producing State

LN Total State Spending Per Capita

HIV Prevalence Rate

Size of Largest Ethnic Group

Population Density

***=p<.01 **=p<.05 *=p<.l

Model 1
31.47***
(4.33)
-8.422**
(3.348)
-0.135***
(0.04)
-0.071
(0.051)

0.394
0.350

Model 2
41.709***
(8.799)
-7.88**
(3.553)
-O. 107 ***
(0.037)
-0.091
(0.06)
5.912*
(3.24)
6.495
(5.786)
4.267
(6.166)
-1.821
(1.847)
-109.397*
(60.177)
0.064
(0.049)
-0.008**
-0.003
0.645
0.543

Standard Errors in Parentheses

Model 3
3587*"
(3.768)

-10.625***

(3.173)
-O.116***
(0.036)

-113.636***

(36.453)

-0.008***
(0.003)
0.555
0.512

Dependent Variable= Citizens Per Hospital (in Hundred Thousands of Citizens)

 

Model 2 tests all hypotheses simultaneously. As we can see, the two political

variables that were signiﬁcant in the initial model remain the key drivers: having a

governor and legislators who are members of the ruling party make it more likely that

these resources will be allocated to a respective unit. Voting remains insigniﬁcant.

 

11 To ensure cross-national inﬂuences were not driving the model, a dummy variable
was used to compare the South Aﬁican and Nigerian sub-units. This dummy variable
was found to be statistically insigniﬁcant and is not included in the models. Residual
plots were run to check for the inﬂuence of outliers, and none were found to inﬂuence

results.

93

 

Turning to the economic predictors, there is evidence that spending on citizens does
seem to have an effect on the availability of hospitals. The only apparent inﬂuential
economic variable is the availability of oil, but the relationship runs in the opposite
direction of what was hypothesized with oil resources making it less likely that hospitals
will be developed. Neither gold nor platinum, resources found in South Aﬁican
provinces, seem to have an inﬂuence on ARV provision. Similarly we see that only one
of the two public health variables has a statistically signiﬁcant relationship with hospital
allocation. Ethnic diversity does not seem to have much inﬂuence on whether or not
hospitals are allocated to respective regions. It can be reported, however, that, as
expected, states with greater burdens of disease are more likely to receive these
resources, suggesting that some resources are being distributed to meet health needs.
One also sees that population density is statistically signiﬁcant, accounting for the fact
that most of the hospitals available are in large population centers, a ﬁnding that
Lieberman (2008) also found to be the case. Both projects reveal the shortage in
treatment resources available in rural areas, a theme found in the health literature
previously discussed.

Model 3 provides a reduced, parsimonious model that includes only the most
inﬂuential variables. As we see the relationship is dominated by the two political
variables, with more urban areas and areas with more HIV patients prioritized. The oil
measure lost its statistical signiﬁcance once other variables were dropped, so it is not
included in the parsimonious model. These ﬁndings again suggest that patronage at the
sub-national level is based largely on the election of intermediate political elites rather

than a direct relationship between citizens and the political center. This model suggests

94

that although a response based on public health is salient that patronage and distributing
resources based on membership in the ruling party plays a signiﬁcant role in
determining who gains access.

What isespecially interesting from these ﬁndings is that variables from each of our
major theoretical categories are statistically signiﬁcant, according to model 2. Aﬁican
politics scholars would not be surprised that among political variables, electing members
of the ruling party is one of the best strategies to ensure ARV hospital provision. These
ﬁndings lend support to Patterson’s (2005) argument about the importance of executive
leadership in AIDS policy treatment; we see that although policies in both of these
countries are set at the national level, having implementing authorities at the sub-unit
level makes a difference. One of the challenges to the overall AIDS response is
ensuring not only that sub-national units have the capacity to perform to the
speciﬁcations of national level plans, but also that effective working relationships exist
between the multiple tiers of governance. The ﬁndings suggest that intermediate elites
play a key role as mid-level patrons distributing resources on behalf of the regime in
power. As this model suggests, these relationships are important to ensure that sub-
national units gain access to these important resources, translating ﬁ'om policy
development to actual local level implementation. These ﬁndings also point to the role
of legislative and sub-national unit executives in the policy process: as Hyden’s
discussion of patronage substituting for an effective policy process suggests, legislators
and sub-unit executives play a role, not as policy developers but rather as resource
seekers on behalf of their constituencies. Rather than the patronage process being

exclusively dominated by one big man at the top, there exists an entire class of big men,

95

all seeking their portion of these resources, both as patron to their constituencies, but
also as clients of national leadership. In American politics these resources might be
considered pork barrel projects, but as Hyden discusses, the line between state and
regime is often even more blurred. In the Aﬁican case, the informal institutions of
clientelism appear to have overtaken the formal institutions. This is especially true in
the Nigerian case, although it is as well to an increasing degree in South Aﬁica.

Units with more AIDS patients receive greater ARV provision. Surprisingly, the
size of the largest ethnic group is not statistically signiﬁcant, but it would be important
in future research to develop improved measures of ethnic representation and saliency as
the limitations of the measures may be obscuring the importance of these results.
Political and ethnic lines are also not independent features, and I suspect that ethnic and
political party in and out groups may be largely entangled in ways that are not included
in Lieberman’s (2008) analysis.

Lastly, in terms of the economic hypotheses, the model provides some interesting and
surprising ﬁndings. First, budgetary spending is not statistically signiﬁcant which is
unexpected. Furthermore, oil is statistically signiﬁcant, although contrary to the
hypothesized relationship. My hypothesis was based on the idea that units would likely
develop these institutions to protect the citizens that provided the economic fuel for
these units to maximize production. This was a relatively confounding ﬁnding until one
' considers the ﬁndings on revenue and natural resources by Snyder and Bhavnani (2005).
As these authors argue, when units can garner large sums of funding from natural
resources, especially those that require interaction between the sub-national unit

government and multi-national corporations rather than more lootable, alluvial

96

resources, units lack the incentive to build institutions that provide tax revenue, thus
bypassing need to build social services (ibid). When the resource is both non-lootable,
and is extracted using industrial extraction, these authors argue that low risk to the state.
Furthermore, with the incentive to tax removed, these units lack the incentive to build
either coercive or social welfare institutions (ibid). This helps to explain why ARV
hospitals are less available in the Niger Delta region; if oil required large numbers of
local workers the incentive structure would likely be different and would likely behave
more like gold mining in South Aﬁica. This suggests that the lack of these resources is
another data point of the limited availability of state resources being provided in this
conﬂict-ridden region.

Discussion, Implications, and Future Research

These results yield both good and bad news for interested public policy ofﬁcials. The
good news is that ARV providing hospitals are more likely to be built where there are
citizens who need them. Nigerian and South Aﬁican ofﬁcials can proudly show these
results to potential donors and their own citizens. Although this result was expected, it
is reassuring that pressing social need is among the key explanatory variables.

The bad news is that, unfortunately, these same decisions are being made with
hospital resources distributed along patronage lines. While this type of distribution is
not unique to ARVS, it is a concern for donors interested in maximizing their
contributions for public health purposes. This ﬁnding also shows that although
patronage is often viewed as rewarding support between the big man and his subjects,
numerous mid-level big men, represented by sub-unit executives and legislators play an

intermediary role, striving to bring home these assets to their constituents as they seek to

97

consolidate their power. Although one would want to analyze a larger data set before
reaching deﬁnitive conclusions, it appears likely that patronage networks, an important
feature in Aﬁican politics, extends prominently into ARV distribution. This is
extremely problematic for donors, and is worthy of further future inquiry. It also points
to the importance of the monitoring and evaluation processes as international donors
should place emphasis when dealing with recipient governments on equitable
distribution of resources. Some emphasis has been placed on this type of equitable
distribution, but most emphasis has been put on gender. Gender an important
component, but these preliminary ﬁndings suggest that areas that do not elect members
of the ruling party may be disadvantaged when allocation decisions are made. If this
relationship is further veriﬁed, donors should add the inclusion of these political
minorities in the distribution coordinating mechanisms and include political
considerations in their monitoring and evaluation of these ARV programs.

Of most concerning is the role of oil in this relationship. This relationship shows
that hospital treatment is lacking especially in oil producing areas. Such ﬁndings seem
to illustrate feedback loops existing in these oil producing areas where oil is part of a
vicious cycle between nonresponsive governments and needy AIDS infected citizens.
This outcome may be an element of the resource curse that is less developed and worthy
of further inquiry. These ﬁndings demonstrate that each of the model’s categories—
politics, economics, and public health—are all important in modeling sub-national AIDS
policy implementation. Although this is not a complete model, these initial results
indicate that this ﬁ'arnework is an effective starting point.

Although the ﬁndings of this research yields some interesting results, forty-ﬁve

98

percent of variation remains unexplained in the most parsimonious model. Other
variables will need to be tested to increase the utility of this model. Similarly, ethnic
variables were not signiﬁcant, likely due to the weakness of the measures used. Future
work will need to incorporate new and better measures of ethnicity at the sub-national
level.

The ﬁndings also will require more investigation into whether the source of funding
matters. Even relatively wealthy African countries like South Aﬁica and Nigeria that
provide a majority of their own funding get a portion of their ﬁrnding ﬁom foreign

donors, and other African countries garner an even larger portion of their AIDS
'spending ﬁom foreign donor sources. It would be interesting to see whether the source
of funding determines how it is spent at the sub-national level. To do this, more
budgetary data will be needed, and greater transparency by not only national, but sub-
national governments will be important.

To improve the quality of the model more units in other countries need to be
considered, and mechanisms underlying how these decision making processes are made,
from national level plans to policy implementation will be needed in future research. It
is impossible to generalize to the continent based on such a limited data set, and tracking
dollars and decision making processes would make for more robust conclusions.
Finally, it is important to acknowledge that hospitals are but one distribution mechanism
of importance in ﬁghting AIDS. As other scholars have discussed, there are a great
many other important institutions including personnel, drug provision, and reducing
costs that are critical to the ﬁrture requirement of mass AIDS treatment in Aﬁica and

across the globe. These other institutions would need to be considered in developing a

99

more expansive model of AIDS policy (Kombe et al 2004, Oloriegbe, 2007).
Conclusion

This paper set out to consider whether political, economic, or public health variables
drove ARV hospital provision for AIDS patients in Nigeria and South Africa. This
model showed that aspects of each of these categories play a role in determining where
ARVS are distributed. Most importantly, this paper establishes the existence of sub-
national variation. Looking back at our hypotheses, we ﬁnd some support for governors
and legislative representation playing a role in ARV policy choice (H1 and H2). Voting
for the president does not appear to be an inﬂuential factor driving resource allocation
(H3). We do not see budgetary spending has an inﬂuential relationship on resource
distribution (H5).

Oil (H4) is the only natural resource that appears to play a substantial role in the
distribution of these resources, although it hampers availability rather than ensuring it.
These ﬁndings point to the dyadic relationship that oil creates between governments and
multi-nationals. With large sums from oil revenue, governments do not have the
incentive to provide anti-AIDS resources to citizens. This, along with the concerns from
the resource curse literature paints a very bleak picture for oil producing countries in
ﬁghting AIDS.

Greater burden of disease (H6) appears to be a key driver for the allocation of greater
hospital resources. It also is clear that city dwellers are more likely to gain access to
ARVS than rural citizens. Ethnicity (H7) does not appear to be an inﬂuential factor,
according to the models. As was discussed this may be due to the limited measures

available as well as potential relationships between ethnic and political lines.

100

Based on these ﬁndings, where will we see people receiving the most robust ARV
treatment programs? Areas are more likely to successﬁrlly develop these programs
when their political representation is able to bring them more patronage. There are
exceptions with areas with greatest HIV prevalence and urban areas gaining resources as
well. However, it should be noted that patronage networks discussed by Aﬁican politics
scholars are not irrrrnune ﬁom involving themselves when these resources are
distributed. While severe policy inequity at the national level exists, the sub-national
level appears to have even greater disparities, which supports Pradhan’s (2001) ﬁndings.
As Pradhan and the other public health authors suggest is that wealthy citizens will ﬁnd
a way to access these resources, but the poor are particularly vulnerable to these
practices. What this analysis suggests is that being poor and outside of the patron-client
networks may be the worst place of all. Also, these ﬁndings add to Lieberman’s
ﬁndings that such in and out group cleavages matter and extend beyond ethnic lines with
access to the ruling regime playing an important role in who gains access to these
provisions.

While there remains more work to be done, these ﬁndings indicate large inequities
in terms of ARV access at the sub-national level. Most disturbing is that political
patronage appears to be playing into this discussion as resources are being distributed to
those who support the party in power with legislators and governors serving as clients of
the central governments. What resonates most clearly from this work is that political
science has a great deal to say about ARV policy equity and policy choice, and that
more work is needed to pin down the speciﬁc mechanisms that underpin these

relationships.

101

Data Sources:

South African Electoral Commission Website with electoral breakdown by region:
http://www.elections.orgza/Results/natperprov.asp

South Aﬁican data on ARV hospitals: Martin, Gayle. 2003. “A Comparative Analysis of
the Financing of HIV/AIDS Programs.”

South African antiretroviral drug hospital availability by province: Western Cape
Department of Health: December 2004:
114th)://www.tac.org.za/Documents/ARVRollout/arvstats.htrn

South African ARV Hospital List by Province: Published on Joint Civil Society
Monitoring Forum (http://www.icsmforgza) http://wwwicsmforgza/node/Z1.

South African natural resource distribution: South Aﬁican Department of Energy. N.A.
Operating Gold Mines and Recovery Plants in South Aﬁica. 2006.

Nigerian ARV hospitals: Oloriegbe, Dr. Ibrahim. “Impact, Challenges, and Long-Tenn
Implications of Antiretroviral Therapy Program in Nigeria.” Health Reform
Foundation of Nigeria. August, 2007.

hnp://www.n_aca.gov.ng/index.th?option=com_content&ta_sl_(=view&id=l 57&Itemid= 1

91

Nigerian Total List of Health Facilities: National Bureau of Statistics. Social Statistics in
Nigeria, 2005.

Nigerian Electoral Data:
http://aﬁicanelections.tripod.com/g.html#2003_Presidential_Election

Nigerian Oil Distribution: United Nations Development Report. NA. 2006. Abuja,
Nigeria.

Nigerian State Economic Data: http://www.aiae-nigeria.orgbecgs/berrenortshtm
Nigerian Electoral Data ﬁom 2003: Ojarneruaye, Emmanuel. “Some Signiﬁcant
Outcomes of the 2003

Nigerian Presidential Elections.” International Federation for Education and Self Help:

Phoenix. http://www.waado.org/NigerDelta/Politics/2003Elections/President-
election/NigerDelta-OjameruayeII.htm1

102

Works Cited:

Aborisade, Oladimeji and Robert Mundt. 1998. Politics of Nigeria. Redwood City, CA:
Addison Wesley.

Ashford, Lori, Davidson Gwatkin, and Abdo Yazbeck. 2004. “Designing Health and
Population Programs to Reach the Poor,” Washington, DC: Population Reference
Bureau.

AVERT. “HIV and AIDS in South Aﬁica.” 2010. Online.
Ilium/WWW.avert.org/aidssouthaﬁica.htm.

Barilleaux, Charles. 1998. “The Politics of State Health and Welfare Reforms.” In
Governing Partners, ed. Russell L. Hanson. Boulder, CO: Westview Press.

Bratton, Michael and Nicholas van de Walle. 1997. Democratic Experiments in Africa:
Regime Transitions in Comparative Perspective. Cambridge, MA: Cambridge
University Press.

Braveman, Paula, and Eleuther Tarimo. 2002. “Social Inequalities in Health Within
Countries: Not Only an Issue for Affluent Nations,” Social Science and Medicine.
Vol. 54: 1621-1635.

Cantor, Joel. 2006. “State Health Policy.” In The State of the States, ed. Carl E. Van
Horn. Washington, DC: Congressional Quarterly Press.

Carsey, Thomas and Gerald Wright. 1998. “State and National Factors in Gubernatorial
Elections.” American Journal of Political Science. Vol. 42 No. 3: 994-1002.

Cassels, Andrew and Katja J anovsky. 1995. “Strengthening Health Management in
Districts and Provinces,” Geneva: World Health Organization.

Chigwedere, Pride et al. 2008. “Estimating the Lost Beneﬁts on Antiretroviral Drug Use
in South Aﬁica.” Journal of Acquired Immune Deﬁciency Syndrome. Vol. 49 No. 4:
41 0-41 5 .

De Waal, Alex. 2003. “How will HIV/AIDS Transform African Governance?” African
Aﬂairs Vol. 102. P. 1-23.

Erikson, Robert S., Gerald C. Wright, and John P. Mclver. 1993. Statehouse
Democracy.

York: Cambridge University Press.

Faris, Stephan. 2004. “Deadly Denial.” Time Magazine Europe. Accessed Online 4 May
2010. http://www.time.com/time/europe/html/O40419/aids.html

103

Gamble, Vanessa and Deborah Stone. 2006. “US. Policy on Health Inequities: The
Interplay of Politics and Research.” Journal of Health Politics, Policy and Law. Vol.
31, No. 1: 93-126.

Gwatkin, Davidson, et al. 2007. “Socio—Economic Differences in Health, Nutrition, and
Population Within Developing Countries,” The World Bank: Health, Nutrition, and
Population.

“Health Care in South Africa.” 2009. Accessed Online 19 Feb. 2010.
http://www.southﬂica.info/about/health/health.htm

Hero, Rodney. E. 1998. Faces of Inequality: Social Diversity in American Politics, New
York: Oxford University Press.

Hobbes, Thomas. 1924. The Leviathan. New York: P.F. Collier & Son

“Human Rights and HIV.” UNAIDS. ND. NA. Online.
http://www.un_aids.org/en/PolichndPractice/HumanRights/defaultasp

Hyden, Goran. 2006. African Politics in Comparative Perspective. New York:
Cambridge University Press.

Johnson, David. 2000. “Nigeria: Health Brieﬁng Paper,” DFID’s Health Systems
Resource Center for the UK Department for International Development

King, Gary. Robert O. Keohane, and Sidney Verba. 1994. Designing Social Inquiry:
Scientific Inference in Qualitative Research. Princeton: Princeton University Press.

Kombe, Gilbert, David Galaty, and Chizoba Nwagbara. 2004. “Scaling up Antiretroviral
Treatment in the Public Sector in Nigeria: A Comprehensive AnalySis of Resource
Requirements.” Bethesda, MD: Partners for Health Reform.

Lasswell, Harold. (1936) Politics: Who gets What, When, and How. McGraw-Hill, New
York.

Lieberman, Evan. 2009. Boundaries of Contagion: How Ethnic Politics Have
Shaped Government Responses to AIDS. Princeton, NJ: Princeton University Press.

Makinen, M. et al. 2000. “Inequalities in Health Care Use and Expenditures: Empirical
Data from Eight Developing Countries and Countries in Transition,” Bulletin of the
World Health Organization. Vol. 78. No. 1. Geneva: World Health Organization.

Nordhaus, William, Alberto Alesina, and Charles L. Schultze. (1989). “Alternative
Approaches to the Political Business Cycle,” Brookings Papers on Economic

104

Activity,
Vol. 1989, No.2, pp. 1-68. The Brookings Institution.

Oloriegbe, Ibrahim. 2007. “Impact, Challenges, and Long Term Implications of
Antiretroviral Therapy Programs in Nigeria.” Abuja, Nigeria: Health Reform
Foundation of Nigeria.

Patterson, Amy, ed. 2005. The African State and the AIDS Crisis. Burlington, VT:
Ashgate.

Patterson, Amy. 2006. The Politics of AIDS in Africa. Boulder, CO: Lynne Rienner
Publishers.

Peterson, S. and S. Shellman. 2005. “AIDS and Violent Conﬂict: The Indirect Effects of
Disease on National Security. Manuscript, College of William and Mary.

Pradhan, Menno, David Sahn, and Stephen Younger. 2001. “Decomposing World
Health
Inequality. Tinbergen Institute.

Progress Report on Declaration of Commitment on HIV and AIDS. 2006. NA. Prepared

for the UN Special Session on HIV and AHDS.
http://dataunaids.org/pub/Report/2008/south_africa 2008 counﬂ progess report en.
pit“

“Provincial Health Departments.” 2010. South African Government Information.
http://www.info.gov.za/aboutsa/health.htm#Provincial.

“Public Opinion and HIV/AIDS: Facing up to the Future.” NA. 2004. Afrobarometer
Brieﬁng Paper Number 12. Michigan State University.
<htLp://afrobarometer.org/papers/AfrobriefNo1 2.pdf.>

Odutolu, O, Ahonsi, B.A, Gboun, M & Jolayenri, OM (2006) ‘AIDS in Nigeria: A
nation
on the threshold’. Chapter 11: The National Response to HIV/AIDS. Harvard Center
for Population and Development Studies.

Raphael, Dennis. 2003. “Towards the Future: Policy and Community Actions to
Promote
Population Health.” In Health and social justice: a reader on Politics, Ideology, and
Inequity in the Distribution of Disease. Ed. Richard Hofrichter. San Francisco, CA:
Jossey Bass. Pp. 453—68.

Raphael, Dennis. 2007. “Public Policies and the Problematic USA Population Health
Proﬁle.” Health Policy Vol. 84: 101—1 11.

105

Roniger, Luis. 2004. “Political Clientelism, Democracy, and Market Economy.”
Comparative Politics, Vol. 36, No. 3 (Apr., 2004), pp. 353-375.

“Nigeria Opens 41 New AIDS Treatment Centers.” Reuters News Service. 14 March
2006. Online. ﬁnd/WWW.aegis.com/news/re/2006/RE060316.htrnl

Ringquist, Evan J ., and James C. Garand. 1999. “Policy Change in the American
States.”
In American State and Local Politics, ed. Ronald E. Weber and Paul Brace. New
York: Chatham House.

Ringquist, Evan. 2005. “Assessing Evidence of Environmental Inequities: A Meta-
Analysis.” Journal of Policy Analysis and Management. Vol. 24, No. 2: 223—247.

Rom, Mark. 2007. “State Health and Welfare Programs.” In Politics In the American
States: A Comparative Analysis. Eds. Virginia Gray and Russell Hanson.
Washington, D.C.: CQ Press. (9“1 Edition).

Schoen, Cathy and Michelle Doty. 2004. “Inequities in Access to Medical Care in Five
Countries: Findings for the 2001 Commonwealth Fund International Health Policy
Survey.” Health Policy. Vol. 67: 309-322.

Singer, PW. 2002. “AIDS and International Security. Survival, 44(1), 145-158.

Snyder, Richard and Ravi Bhavnani. 2005. “Diamonds, Blood, and Taxes: A Revenue-
Centered Framework for Explaining Political Order.” Journal of Conﬂict Resolution.
Vol. 49. No. 4. Pp. 563-597.

UNAIDS. 2008. Nigeria: Country Situation. N.A.
http://data.unaids.(ﬂ/pub/FactSheet/2008/sa08fnie_en4)df

UNAIDS. 2008. South Aﬁica: Country Situation.
http://data.unaids.org/pub/FactSheet/Z008/sa08_soa_en.pdf

UNGASS. 2007. “United Nations General Assembly Special Session Report on
Nigeria.”

http://cfs.unaids.org/country_factsheet.aspx?ISO=NIE
Viall, Jenny and Joyful Mlamli. 1999. “Cape Will Give Pregnant Women AZT.”
Aﬁican

News Service. Online. http://www.aegis.com/news/ads/1999/AD990023.html.

Villalon, Leonardo and Phillip Huxtable. 1997. “The African State at a Critical Juncture:
Between Disintegration and Reconﬁguration. London: Lynne Rienner Publishing.

106

Vogel, Ronald. 1993. “Financing Health Care in Sub-Saharan Aﬁica,” N.C.:

Greenwood
Press.

Wantchekon, Leonard. 2003. “Clientelism and Voting Behavior: Evidence from a Field
Experiment in Benin,” World Politics Vol 55. No. 3.

Weinstein, Jeremy. 2006. “AIDS Security and Social Stability,” Manuscript, Stanford
University. http://www.stanford.edu/~jweinst/ﬁles/Crime_DEC06_FINAL.pdf

World Health Organization. 2005. NA. “Nigeria: Summary Country Proﬁle for
HIV/AIDS Treatment Scale-Up.” Geneva: WHO.

WHO, UNAIDS & UNICEF (2008) ‘Towards universal access: scaling up priority

HIV/AIDS interventions in the health sector’.

107

The Micro-Foundations of Public Satisfaction with HIV/AIDS Policy
Performance: Lessons from Afrobarometer Round 4
Abstract:

This paper investigates what factors lead to increased citizen satisfaction with their
government’s response to HIV/AIDS? Recent investigation of ARV policy across
Aﬁ'ica revealed the strength of states more than regime type that inﬂuences the level of
ARV coverage across African countries. Further modeling revealed that although ARV
hospitals are distributed to a larger degree in urban and areas with high levels of
HIV/AIDS, patronage networks persist with those areas represented by elected members
of the party in power more likely to receive ARV facilities. Is the availability of ARVS
an inﬂuential predictor of citizen satisfaction? Round 4 of the Afrobarometer asked
more than 27,000 African citizens in 20 countries about their opinions, and their
evaluation of their government’s HIV/AIDS policy performance allows for the rnicro-
Ievel relationships to be investigated and modeled. Regression analysis suggests that the
six most important predictors of citizens satisfaction with HIV /AIDS policy
performance (in order) are health service satisfaction, perceived corruption, trust in the

State’s institutions, satisfaction with democracy, low HIV prevalence, and high ARV

coverage.

108

Introduction

This paper investigates why some Aﬁican citizens are more satisﬁed with their
respective government’s response to the HIV/AIDS crisis than others. This research
follows cross-national and sub-national investigation and provides an opportunity to re-
test previously established relationships with disaggregated, individual level data.

Before ernbarking on new research, it is important to quickly revisit the earlier macro-

Ievel results on which this study is based.

Previous investigation established that at the cross-national level, ARV coverage was
dependent on the strength of a respective state rather than regime type. In other words, a
state with an effective bureaucracy, established infrastructure, and the capability to
provide other basic social services was more likely to scale up ARV distribution to a
larger degree than weaker states. This state strength or stateness was measured using the
Kauffman, Kraay, and Mastruzzi World Bank Institute indicators (2009). Several of
their indicators—political stability, governmental effectiveness, regulatory quality, rule
of law, and control of corruption were combined into an index of stateness which was
previously used by Bratton and Chang (2006). This stateness index was found to be a

Strong positive predictor of ARV coverage while regime type was not found to have a

Strong relationship with coverage.

At the sub-national level, investigation revealed that sub-national units represented
by executives and legislators of the party in power are more likely to receive ARV
resOtlrces than areas represented by other political parties. Additionally, areas with a

gr eater burden of disease are prioritized as HIV prevalence is positively related to sub-

109

national ARV provision. Lastly, a positive relationship was found between greater
population density and ARV programs, suggesting that more urban areas have been
prioritized.

Taken together the cross-national and sub-national ﬁndings provide some insight into
why some Aﬁicans are more likely to receive ARVs than others. Citizens living in
stronger states who have access to patronage resources are more likely to receive
treatment than those residing in weaker countries and those who do not support the party
in power. A weakness of the past inquiries is the fact that statistical analyses were based
on relatively small samples with N sizes between forty and ﬁfty. Would these macro-
level relationships hold if one had larger data sets to draw upon? An additional question
is whether state strength or patronage access is the more important predictor of ARV
coverage. Fortunately, the Afrobarometer provides a data set allowing for micro-level
inquiry. Round 4 data include over 27,000 citizen surveys taken ﬁ'om twenty countries
across Aﬁica. This large number allows for disaggregated investigation at the
individual level. An additional strength of Afrobarometer data is that it allows for the
inclusion of more variables and many more degrees of freedom than previous research.
Previous ﬁndings were based on macro and meso-level units of analysis, and
Aﬁ'obarometer data allows for micro-level testing of previously articulated relationships.

Previous research focused on ARV coverage as a dependent variable to answer the
question: what political components lead to scaled up AIDS treatment resources? As
will be detailed later, this paper is based on the idea that citizens will evaluate their
government’s policy performance more favorably when they see anti-AIDS policies

implemented more effectively. Namely, individuals who interact with anti-AIDS

110

institutions should be more satisﬁed than those who do not. One of the goals of this
paper is to see whether (and to what degree) the availability of ARVS drives citizen
satisfaction with their government’s handling of the HIV/AIDS crisis. In the
introductory chapter of the dissertation, the argument was made that AIDS treatment
serves as a proxy for an overall response to the disease. In other words, countries that
have effectively scaled up treatment have also expanded prevention, awareness, and
testing services in accordance with international best practices. To what degree do
Aﬁicans interact with these efforts, and does this interaction with anti-AIDS responses
shape whether citizens are satisﬁed with the performance of these policies at the
individual level? Are countries that have invested in ﬁghting HIV /AIDS rewarded with
greater citizen satisfaction?

In order to answer these questions, individual level data will be used. One example
of this type of individual level interaction is the story of Ms. Trisca Mkanda, a primary
school teacher in Malawi. Her story is not unusual and provides an individual level
account of how anti-AIDS policies are beneﬁcial. Here is how she describes her
experience:

I got very sick in 2004 after the birth of my second child. I lost a lot of weight and
could only walk with the support of a stick. When my HIV test revealed I was HIV
positive and had an AIDS related condition, the hospital started my treatment on ARVS.
I am now better and have resumed my teaching job. No-one would suspect I have HIV if
they see how healthy I look. My child is also in good health since she started taking the
ARVS, (DFID, 1).

Trisca Mkanda is one of over 70,000 Malawians who have received ARVS from over

440,000 who suffer from HIV. Her example provides a snapshot of not only the

individual level beneﬁts of effective anti-AIDS policies, but the positive extemalities

111

which extend to both her family and community through her increased productivity as a
teacher. Trisca is a member of an AIDS support group which is funded through
Malawi’s National AIDS Commission.

Gertrude Tandandi, who runs the Tikondane AIDS Support Group provides a broader

account of how these groups make a difference:

ARVS are ﬁee of charge in hospitals. However, our group visits a lot of sick and poor
people who can’t afford the bus fare to go to the district hospital. Bringing ARVs to the
rural community where the majority of Malawians live would save a lot of lives. We see
a lot of helpless orphans in the villages that either live with their grandparents or siblings
after the death of their parents. More parents and guardians would be able to be more
productive and take care of their children if these life saving drugs became easily
accessible, (DFID, 1).

This group is one of hundred across the continent working to prevent the worst
effects of HIV/AIDS, and provides just one example of effective anti-AIDS policy in
action. Effective policy requires a host of interventions, and ARVS are only one tool
available. Policies including prevention campaigns through education, condom
PTOVision, and male circumcision which decrease disease spread. Testing and
cOunseling so individuals know their HIV status is another important component. The
1218t line of defense is ARV treatment. The goal of this project is to understand what
faetors lead citizens to evaluate their governments’ HIV/AIDS policies as effective.
Before specifying a research design, literature on public opinion and satisfaction with
so<=ial service provision will be reviewed. The overall goal of this paper is to answer
What factors drive citizen satisfaction with their government’s response to HIV/AIDS.
Literature Review

Three literatures are brought to bear on our subject matter. First, public opinion

literature relating to social service provision in Africa is discussed. Second, literature

discussing both regime type and state strength is brought to bear. Third, literature on

112

political cleavages in Africa is also discussed. A brief synthesis of these literatures then
follows.
Public Opinion and Social Service Provision

The literature review is organized with social service provision discussed more
generally before focusing on health and HIV /AIDS policy speciﬁcally. Bratton, Mattes,
and Gyimah-Boadi (2005) provide valuable context for how social service provision and
public satisfaction with policy performance ﬁt together. Firstly, from a demand

perspective, Aﬁican citizens view economic problems as the most important problem

facing their countries that they want their governments to address12 (99). Immediately
following economic problems is the provision of social services like health services,
education, the rule of law, clean water, and AIDS treatment (99). As Bratton et al.
report, there is a great deal of variation in terms of what problems citizens list among the
three most important problems they would like to see solved. They found that only
citizens in Botswana mentioned HIV /AIDS as one of the top ten problems for
governments (102). The authors also reported that public demand for health services
exceeded the importance placed on health provision by donor institutions (103). As
should be clear from these results, demand for health care and social services more
generally is quite high across Africa.

While Aﬁican citizens view health as a key development challenge, Bratton et al.
also ﬁnd that sixty-ﬁve percent of Aﬁicans would like their governments to be the
supplier of health care resources rather than private or faith based institutions (109, 110).

Citizens were also asked where they turn for healthcare when they encounter shortages

 

'2 These ﬁndings were based on survey data collected in 12 countries from 28,795
citizens from Aﬁobarometer Round 1 surveys.

113

(138). Exposing citizen vulnerability when healthcare is unavailable, more than thirty
percent of citizens had no other source when healthcare was needed. Nearly forty
percent turned to their families to help provide in times of need, the most ﬁequent
response (138). Leaning on one’s family is an understandable short-terrn response, but
when dealing with a chronic disease like HIV/AIDS, leaning on family is a sub-optimal
strategy. These individuals cannot provide the type of medical expertise or treatment
required, though this is a rational response. Another interesting ﬁnding is that despite
citizens indicating their preference for governmental provision of health care services,
people were more likely to seek healthcare ﬁom markets than they were from
governmental institutions (138). There appears to be signiﬁcant variation across
countries as only a few countries like South Aﬁica have robust market-based health
institutions. Private health care facilities are generally restricted to those who either can
afford to purchase insurance or are based on cash for services arrangements largely
unavailable to the poor. People want their governments to provide healthcare and
simultaneously seek non-governmental answers in times of shortage. This trend points
to a striking pattern: the supply of healthcare by African governments is not meeting
public demand for these resources. In other words, although Africans want their
governments to provide, these institutions have thus far been unable to meet healthcare
needs.

In terms of policy performance, Bratton et al. ﬁnd that Aﬁicans are more satisﬁed
with social service provision than they are with economic reforms (238). Sixty-two
percent of citizens claim their governments are doing “fairly well” or “very well” at

preventing AIDS, the largest rate of satisfaction with any individual policy area (239).

114

Fifty-four percent report similar satisfaction with health service provision (23 9). As
Bratton et al. analyze their data, they suggest that these ﬁndings are based on low public
expectations and marginal improvements. The ﬁnding that in times of shortage only ten
percent of Africans turn to governments for health care indicates individuals are not
especially satisﬁed with their healthcare systems, providing ﬁrrther evidence that supply
is not meeting demand. Bratton et al. (2005) provides robust evidence of limited
satisfaction with health care availability at the individual level.

What accounts for the variance in satisfaction across individuals? Bratton (2007)
explores what determines public satisfaction with both health and educational services
utilizing Afrobarometer Round III data. He ﬁnds that ease of access is a critical
component leading to positive citizen evaluation. High fees detract ﬁ'om citizen
satisfaction. Corruption also detracts from satisfaction, although paying a bribe actually
leads to improved satisfaction. Of most importance is the ﬁnding that citizens link their
satisfaction with health and educational services to their overall evaluation of
democracy. This linkage of social services and overall evaluation of democracy points
to the critical nature of these resources to citizens.

Bratton found that the availability of health care and increased quality of these
services was the most important determinant of satisfaction with healthcare services
(22). Not surprisingly, satisfaction was also determined by social structure (24).
Speciﬁcally, greater levels of poverty detract ﬁ'om satisfaction with healthcare services.
The negative relationship between poverty and satisfaction was the second most
important predictor, behind only the positive relationship between ease of access and

increased levels of satisfaction (24). Bratton assesses perceived corruption to negatively

115

 

inﬂuence one’s satisfaction with healthcare availability (24). The high cost of healthcare
was the next most important concern as increased perception of costs led to decreased
satisfaction with healthcare (24).

Bratton goes on to link the importance of positive provision of these public services
to increased satisfaction with democracy—a measure he suggests may indicate citizens
evaluate service provision and democracy using similar reasoning (26, 27). As Bratton
puts it, “satisfaction with basic social services also contributes to building a mass
constituency for democracy,” (28). Citizens want to see an overall increase in the
quality of these services with a focus on responsiveness in particular (29). As Bratton
suggests, people are willing to pay for services as long as quality continues to improve.
The ﬁndings ﬁ'om Bratton (2007) provide evidence supporting those discussed from
Bratton et al. (2005).

Though Bratton et a1. (2005) and Bratton (2007) make references to HIV/AIDS
several authors have speciﬁcally analyzed citizen satisfaction with HIV /AIDS policy
speciﬁcally. Afrobarometer Brieﬁng Paper Number Twelve provides some descriptive
statistics related to citizen experience with the disease utilizing data ﬁ'om Aﬁ'obarometer
Round 11 (2004). The data suggest tremendous variance across countries, similar to
Round I ﬁndings. For instance, eighty-ﬁve percent of Ugandans indicate they know
someone who died of the AIDS virus (1). Of the ﬁfteen countries included in this round
of surveys, six countries had less than twenty percent of citizens indicate that they knew
someone who died of the disease (1). This included high HIV prevalence countries like
South Africa and Lesotho. These data suggest that stigma related to AIDS varies across

countries, and stigma is likely responsible for gap between positive response and high

116

HIV prevalence statistics. When stigma is less pervasive, citizens appear more willing
to answer positively indicating their knowledge of people who died of the disease. The
paper suggests that higher levels of educational attainment and people exposed to
technology were more likely to report knowing someone who had died of the disease
(2).

In terms of how HIV/AIDS is viewed as a policy priority, there is again tremendous
variance across countries. Thirty percent of Botswana citizens and twenty-eight percent
of Namibians consider AIDS a policy priority when asked to mention the three most
important problems facing their country (4). This high rate is in sharp contrast to seven
countries where less than ﬁve percent of citizens prioritize the disease (4). Again, this
pattern includes both high and low prevalence states. When asked whether greater
resources should be allocated towards AIDS versus focusing on other problems, citizens
were divided (6). In three countries, citizens wanted more resources for AIDS while in
four others citizens wanted less resources allocated to AIDS compared to other issues.
The most common pattern was similarly divided values with near equal numbers of
citizens wanting more and fewer resources allocated towards ﬁghting AIDS. This
pattern was evident in eight cases. What is most clear ﬁ'om this article is the cross-
national variation both in terms of citizen experience with the disease and in terms of
their comparative demand for governmental intervention.

eriteside et a1. (2002) also analyze the Afrobarometer data on HIV /AIDS. These
authors ﬁnd that the Afrobarometer questions are valid and reliable when compared to
epidemiological data like HIV prevalence rates. The authors ﬁnd high correlations

between people reporting knowing someone who died of the disease and HIV

117

prevalence. This paper also indicates that although people report widespread experience
with the disease, they do not list it highly as a policy priority compared to other concerns
like increasing employment opportunities and improving the economy. The authors
speculate on the causes of this cross-national trend, arguing that citizens may view AIDS
as family problem. Another possible cause is discussed. The authors hypothesize that
citizens are rationally listing AIDS below other policy issues that are more immediate
like jobs. eriteside et al. also report that a high number of Aﬁicans list health rather
than AIDS as a policy priority. These health responses correlate with those who claimed
knowing someone who had died of AIDS, suggesting that many of these individuals are
equating health with AIDS. Overall, Whiteside et al. suggests that the Afrobarometer
questions on AIDS are both valid and reliable when compared to the epidemiological
data. It also suggests that signiﬁcant variation exits across countries in terms of the
demand for greater governmental AIDS response.

Lieberman (2009) models whether citizens think more resources should be utilized to
combat the AIDS virus even if it means less money on other projects. This research
suggests that greater ethnic ﬁ'actionalization makes it less likely for people to prefer
greater funding spent on HIV/AIDS (276, 277). He also ﬁnds that greater GDP per
capita has a negative relationship with demand for increased AIDS funding (277). As
Lieberman summarizes his key ﬁnding on ethnic boundaries he writes that, “Ethnic
political competition—a standard axis for conﬂict over public policy—has mediated
demands for, and in turn the supply of, AIDS policies across countries,” (276).

Youde (2009) performs statistical analysis investigates what factors lead to high

levels of satisfaction with HIV /AIDS policy performance across the Aﬁobarometer

118

countries. This was measured using an Afrobarometer question which asked citizens
how satisﬁed they were with the government’s performance related to HIV /AIDS
policy. Youde ﬁnds that people who are more informed about the news and have an
interest in public affairs are more likely to support governmental efforts. Individuals
who know someone who has died of the disease also increases the likelihood of
someone supporting governmental intervention. People who assess the government as
being more democratic are similarly more likely to applaud AIDS efforts. The strongest
relationship Youde ﬁnds relates to other health services: if one is satisﬁed with the
government’s overall effort on other health services, they are much more likely to
approve of a government’s AIDS response effort. Youde includes dummy variables for
each country in subsequent models which are consistently positive and statistically
signiﬁcant. This ﬁnding suggests high levels of cross-national variation. As Youde
summarizes his ﬁndings he writes that, “We see that Africans generally think their
governments are doing a good job combating the AIDS epidemic, but that these
evaluations vary along informational and experiential lines. More information leads to
more support. More experience with democracy and AIDS leads to more support,”
(232). A linkage between how citizens evaluate social services and democracy is
consistent with Bratton’s (2007) assessment. Youde also ﬁnds that support for AIDS
policies is linked to support to other effective policy responses like health care more
generally as well as positive evaluation of economic reforms. He also suggests that
Aﬁicans evaluate AIDS in the same manner as they would assess other policy
performance. The analysis suggests that AIDS is not unique with citizens willing to

reveal their true feelings. Youde’s analysis suggests citizens assess their government’s

119

handling of AIDS similar to other political issues, indicating that HIV/AIDS has entered,
“the realm of normal politics,” (232). This ﬁnding conﬂicts with Lieberman’s argument
which argues that ethnic boundaries resonate on HIV /AIDS issues to a larger degree
than on other policy issues. Contrastingly, Youde makes the case that citizens can
assess policy performance related to the disease similarly to other political issues.

What key points can be taken from the public opinion and AIDS literature? First, the
authors agree that the Aﬁobarometer data provides reliable and valid questions that
African citizens were willing to answer truthﬁrlly. Second, a great deal of variation
exists at the cross-national level in terms of satisfaction with HIV/AIDS policy. Both
Bratton (2007) and Youde (2009) point to the importance of an effective AIDS response
as it relates to citizens’ evaluation of their countries as democracies: citizens link
effective AIDS response with effective governance. This suggests the high stakes for
governments in their management of the HIV/AIDS crisis. Whiteside (2002), Bratton
(2007) and Youde (2009) all agree that AIDS responses are integrally tied to the overall
evaluation of health care service provision. Whiteside suggests that citizens consider
AIDS and health similarly and mix the two terms together when revealing the most
important challenges facing their countries (2002). Both Bratton (2007) and Youde
(2009) indicate that in evaluating their satisfaction with their government’s response to
the disease, citizens who assess their health care system more favorably also are satisﬁed
with AIDS responses speciﬁcally.

Before moving on to discuss methods and testing, two other literatures need to be
considered: those covering the role of regimes and states in Aﬁica, and another which

focuses on political cleavages on the continent. State and regime literature will next be

120

discusses.
Literature on Political Regimes

Political science research has been conducted investigating how regime type
structures human development. Speciﬁcally, some scholarship suggests a positive
relationship (Gerring et a1, 2005, Deacon 2003, Stasavage, 2005, Halperin et. al. 2005)
between democratic regimes and more successﬁil policy outcomes, other work indicates
no relationship exists (Ross, 2006). A few authors have targeted the relationship
between democracy and health speciﬁcally with a positive relationship found in several
cases. Govindaraj and Rannan-Eliya (1994) found that when comparing communist and
democratic regimes, democratic regimes had more favorable health outcomes in terms of
infant mortality and life expectancy. Franco et a1 (2004) found similarly that states with
greater political freedom also enjoyed better health outcomes as measured by life
expectancy, infant mortality, and maternal mortality rates. Interestingly, Tsai (2006)
found that among developing countries, democracies out performed non-democracies in
health provision, even when controlling for health spending. As most of the literature
suggests, there appears to be some sort of relationship between democracy and positive
health outcomes.

Diamond and Morlino (2005) state in the introduction of their book that, “We can
analyze democratic quality by what it achieves in terms of government responsiveness to
the expectations, interests, needs and demands of citizens,” (xxix). One of the important
measures of democratic quality as they argue is vertical accountability, linking the fate
of elected officials to how citizens evaluate their political choices (xviii). For a citizen

infected with AIDS, what could be a more important need than life prolonging

121

treatment? Seeking to explain the relationship between democracy and positive health
outcomes, Vollmer and Ziegler (2009) argue that democratic regimes place higher
priority on redistributive policies including health care. They go on to state that
democratic societies are able to overcome inequalities more effectively, leading to
greater provision of public goods. In this way, their ﬁndings echo the framework
developed by Diamond and Morlino as public demand for these resources are able to be
met as democratic institutions serve as an interactive highway between regimes and
citizens.

It is often assumed that democratic values will promote greater public focused policy
outcomes. Sen (1981, 1999) argues that democratic regimes offer voters the opportunity
to penalize ineffective leaders and that the free press allowed under democratic rule
promotes greater information transparency. Gerring et al’s (2005) analysis focuses on
several possible mechanisms that link democracy to human development: they considers
electoral competition, the importance of a free press, the likelihood that civil society
advocates for human development investment, the possibility that democratic regimes
must provide some measure of equality, the likelihood that democratic taxing
institutions allow for a mechanism of redistribution, and ﬁnally that democracy provides
a more stable enviromnent for the provision of public goods.

While the idea that more democratic regimes should produce greater human
development outcomes has face validity, the empirical testing to this point does not
necessarily verify this expectation according to Ross (2006). Ross’s (2006) research
argues that this relationship does not hold when one focuses on whether democracies

provide better human development outcomes for the masses. Ross argues that although

122

 

the commonly accepted wisdom links democracy and human development, there is a
lack of consideration of global health, the lack of focus on country speciﬁc effects, and
nrissing data for the poorest of countries. With this conﬂicting ﬁnding, it provides an
important opportunity for retesting which I can analyze in my ﬁrst article.

With one of Gerring et al’s (2005) ﬁndings showing that longer lasting democratic
regimes result in greater human development outcomes, it is important to assess regime
stability. This is especially important in Aﬁica given the prevalence of hybrid regimes
that sit between ﬁrll democracies and authoritarian states as discussed by numerous
authors including Bratton et al. (2005), Diamond (2002), Schedler, (2002), van de Walle
(2002) and Levitsky and Way (2002).

Gyimah-Boadi (2004) also argues for a connection between democratic transparency
helping Aﬁica’s emerging democracies ﬁght the AIDS epidemic (18-20). He argues
that authoritarianism contributed to the emergence of the crisis, but that transparency has
required states to admit and deal with the disease. Looking at the disease in a different
way, Gyimah-Boadi argues that the disease threatens Aﬁican democracy citing that less
democratic regimes like Uganda have been more effective in AIDS policy than more
democratic Botswana and South Aﬁica. He claims that even weak political
responsiveness and accountability, the key democratic values attributed by Diamond and
Morlino (2005), have crippled the political response. More favorably, he views
democracy as providing some opportunities for successful AIDS policy, mentioning
democratic openness, civil society strengthening, and enhanced rights for women among
important developments.

The preponderance of evidence from the regime literature tends to suggest that

123

 

Aﬁicans living in democracies should beneﬁt ﬁ'om more robust and effective AIDS
responses. As Bratton et al. (2005) suggest in their evaluation of public opinion data,
citizens can effectively evaluate the performance of several regime characteristics (247).
This ﬁnding suggests that overall evaluation and satisfaction with a respective regime
can be measured at the individual level, a point that will be discussed in greater detail

later. The literature on the modern Aﬁican state will next be considered.

A ﬁican State Literature

Instead of looking at the regime and the authoritarian/democratic continuum, another
important component of a country is its state structure. While more democratic regimes
either do or do not allow for more robust AIDS response, it is likely that the level of
resources and institutional capabilities that a state has will be an important determinant
of a government’s response. State focused scholarship argues that instead of the type of
government in place it is the strength of this government that matters most. Englebert
(2000) has made the case that the development of state capacity structures the policy
choices available to elites and, in turn, the quality of governance in respective countries.
In other words, effective AIDS intervention would only be possible in states that exceed
a certain strength threshold. Englebert links the importance of capacity and economic
development, one component of human development (2000, 30-3 7).

Villalon and Huxtable (1998) have echoed these sentiments, summarizing the modern
Aﬁican state in ﬁve characteristics. Their categories include the Aﬁican state holding
ﬁve key traits, “a client status, a personalized identity, a centralized or overdeveloped

morphology, a prebendal or rentier nature, and an extractive impulse,” (11). In

124

unpacking these terms, they discuss the Aﬁican state as a clients, usually ﬁrst as a client
of the colonizing nation and then later as a client of either the United States or Soviet
Union during the Cold War. In the post-Cold War era, the Aﬁican state can no longer
count on these relationships to ensure sustainable income (12). In terms of
personalization, they discuss the dominance of an individual leader (12). In these
countries where a single leader dominates the policy process, the lines between state and
regime are often blurred. This relates well to our issue of inquiry, with Mbeki’s
dominance over South Aﬁican AIDS policy during his regime a controversial example.
They go on to describe the over-centralized and overdeveloped nature of the Aﬁican
state. As they describe it, the state employed too many people with the power
concentrated on the center rather than a multi-layered structure that would empower
local institutions (13). When using the term prebendal or rentier state, they refer to the
state as the key distributional force in which the state plays the role of patron and
citizens serve as clients who are rewarded for their loyalty, either through resource or
employment opportunities, thus using the nation’s resources promote political stability
(13). Finally, as they describe, the state serves as an extractive force with the goal of the
state to focus on these activities rather than enhancing the state’s capacity (14). Boone
(2003) also focuses on the state and the variation in the strength of these institutions as
reﬂective of local circumstances, often by rural political elites. Boone’s research offers
a bottom up approach to the state, which is often subject to top-down discussion.
Providing an example of such a top-down discussion is the work of Jackson and
Rosberg (1982). Rather than focusing on the Weberian deﬁnition of the state which

focuses on the monopoly over the legitimate use of force, with an emphasis on the

125

 

military, police, and courts, these authors focus on two levels to explain why Aﬁica’s
states have survived. They break the state into its empirical and juridical pieces. The
authors argue that the Aﬁican state exists largely on its juridical, or international law
based status and that Aﬁica’s states are empirically weak where they function at all
(4,5). Jackson and Rosberg argue that the Aﬁican state generally fails to hold stable
communities where the nation-state is the dominant institution compared to ethnic
groups (Posner, 2005) or local institutions (5). They go on to discuss the state’s inability
to fulﬁll the requirement of an effective government, capable of penetrating the state and
reigning over the domain of its entire territory. They describe the individuals
comprising the state as under-resourced, both in their level of capacity and their ability
to deploy these limited tools. This ﬁts with the Villalon and Huxtable notion of the state
as overdeveloped with too many people with too centralized a structure; as they discuss,
this overdevelopment is ﬁrrther hindered by a lack of investment in resources (8). In
summary, the empirical African state, which is what we are interested in their ability to
provide for their individual citizens can be characterized as weak (12). As they describe,
the juridical elements which can be characterized by international club membership
more than any deﬁnition that beholds citizens to government. In this deﬁnition,
boundaries and international recognition are what characterizes these traits, a minimalist
deﬁnition that does notlringfor citizen subject which they discuss in their conclusion
(21).

Similarly focused on territorial deﬁnitions of the state, Jefﬁey Herbst’s (2000)
research focuses on the lack of penetration of the Aﬁican state and the inability of these

states to effectively penetrate Aﬁican societies. This project conceptualizes state

126

penetration through roadways, which Herbst argues are inherently underdeveloped in
most state structures. In his description the territories necessitated colonial institutions
which ruled on the cheap; at independence these international boundaries, the juridical
statehood described by Jackson and Rosberg (1982), were upheld by the international
system and post-colonial state boundaries simply followed this pattern. As Herbst
argues, the deﬁned state often only reigns over the capital city rather than extending its
reign over the full realm of its territory and possibly to a valued region containing an
extractable resource as both Boone (2003) and Villalon and Huxtable (1998) described
in their discussion on the role of states as extractive institutions.

As Patterson (2006) describes in her work on the African state and the AIDS crisis,
she ﬁnds there is signiﬁcant variation in terms of different individual characteristics; in
her analysis there was variation in terms of centralization, neopatrimonialism, capacity,
and stability (28). Though she was unable to discern a clear pattern along these
characteristics in terms of how these state level measures inﬂuenced AIDS policy, what
this project reveals is that Aﬁican states, though relatively weak when compared to
others, show signiﬁcant variability across respective units.

If a relationship is found between state strength and AIDS response, what underlies
such a relationship? Bratton et al. (2005) suggest that an assessment of state
effectiveness can be measured at the individual level. Though I will not get into speciﬁc
measures in this section, it should sufﬁce to say that individual Africans are able to
assess the effectiveness of their states at the individual level. As the literature suggests
the state and its ability to project power varies across the continent. The strength of the

state (Bratton and Chang, 2006) is as important of an inquiry in terms of its relationship

127

to AIDS responses as regime type, and comparing the importance of these different
vantage points at the individual level is an important part of this project.

One ﬁnal literature must be brought to bear before one can encapsulate how these
pieces ﬁt together. The next section will detail the literature on clientelism in Aﬁica
before the literature in synthesized.
the Role of Political and Ethnic Cleavages

One would be remiss in discussing the distribution of highly valued, scarce resources
in Africa without brieﬂy summarizing the importance of patronage networks and the
response of clients. The literature on the subject is pervasive as authors have cited the
importance of neopatrimonial relations between the state and citizens (Bratton and van
de Walle, 1997, Villalon 1998).

(2004) provides a deﬁnition suggesting that clientelism,
implies mediated and selective access to resources and markets from which others are
normally excluded. This access is conditioned on subordination, compliance or
dependence on the good will of others. Those in control—patrons, subpatrons, and
brokers—provide selective access to goods and opportunities and place themselves or
their supporters in positions from which they can divert resources and services of favor.
Their partners—clients—are expected to return their benefactors’ help, politically and
otherwise, by working for them at election times or boosting their patron’s prestige and
reputation,” (353, 354).

Harold Lasswell described politics as who gets what, when, and how (Lasswell,
1936). Roniger’s deﬁnition of clientelism goes a long way to describe how power
relationships function across Aﬁica. Citizens subjugate themselves to authority in order
to access public goods. In other words, elites use their resources and often the resources

of the state in order to legitimize themselves and their regimes. This type of distribution

leads to some citizens within a respective country included in the distribution of

128

resources while others are not.

Bratton and van de Walle (1997) discuss how clientelism manifests in Aﬁican
politics. As they discuss, African’s political power is largely concentrated in the hands
of individual leaders, a concept known as presidentialism (63). Leaders use patronage in
the form of state resources to legitimate themselves and their regimes (66, 67). When
concentrated presidential power is combined with clientelism and the distribution of
state resources, these three features are known together as neopatrimonialism (63-68).
Both the Bratton and van de Walle and Hyden descriptions refer to neopatrimonialism as
one of the preeminent informal institutions of Aﬁican politics.

As Bratton and van de Walle argue, however, “when patrimonial logic is internalized
in the formal institution of neopatrimonial regimes, it provides essential operating codes
for politics that are valued, recurring, and reproduced over time,” (63). As Bratton and
van de Walle detail, informal institutions resonate and become a dominant feature of
formal institutions. Bratton and van de Walle also discuss the fact that although
clientelism is controlled by the political center, it is not only a feature of the top of
political regimes. Rather, “this [clientelism] happened at every level; at the top, the
ruler’s faithful political aristocracy was rewarded with prebendal control of public
ofﬁces, monopoly rents, and the possibility of creating its own clientelist
networks. . .Nor were patronage and clientelist beneﬁts limited to the political
aristocracy,” (66). Instead of thinking of patronage networks exclusively in terms of
relationships between individual rulers and citizens, multiple levels of governance exist
within the formal institutions. This allows for potential patronage distribution not just

directly ﬁ'om top to bottom but throughout the multiple formal institutions of

129

government. Neopatrimonial relations may be just as prominent in local and sub-
national governance as at the national level.

Moreover, Villalon (1998) also mentions clientelism as one of the premiere
features of the modern Aﬁican state. As he describes it, patrons reward clients that help
empowered regimes to consolidate control and promote political stability (13). Public
goods are allocated with preferential treatment to loyal groups (13). As Wantchekon
(2003) argues, clientelistic messages are especially salient at election time. Candidates
seek electoral support, and clientelistic messages are often effective in increasing vote
share, especially for the ruling party as individuals seek access to the state’s resources.
The concept of regimes utilizing state resources particularly at election time to sway
voters towards the ruling regime is known as the political business cycle (N ordhaus et
al. 1989). Through elections formal democratic institutions serve as a process through
which Aﬁican leaders legitimize themselves, with patron-client networks utilizing state
resources as patronage. This phenomenon serves as yet another example of the informal
institution known as clientelism subjugating formal elections.

As the political business cycle demonstrates, and Bratton and van de Walle discuss,
neopatrimonialism can resonate in regimes, with or without elections. Bratton and van
de Walle discuss the idea that clientelistic networks exist across regime types (77-82).
Both Bratton and van de Walle and Roniger (2004) discuss the idea that increased levels
of clientelism are negatively correlated with the development of democratic institutions.
In other words, in order for countries to transition and become more democratic,
clientelism and access to the state through patronage networks must be broken.

What does clientelism, such a key feature of Aﬁican regimes, mean for African

130

policy making processes? Hyden summarizes rather poignantly how clientelistic
patronage networks have subjugated aggregate public policy in Aﬁican countries:

Aﬁican countries do not have policy governments, but public institutions operating
on the basis of patronage. These governments conduct their business not with a view to
implementing ofﬁcially agreed-upon policies, but look to rewarding individuals and
groups that have shown exemplary loyalty or contributed to the political success of a
government leader. In short, resources ﬂow along very different paths than those that
are identiﬁed in ofﬁcial statements, be that a policy announcement or the national
budget. The result is that Aﬁican governments tend to look to the past rather than to the
future. To the extent that policies feature in politics, they are more often for window-
dressing purposes than for real implementation, (229,230).

As Hyden argues, western notions of the policy making process apply less well in
Africa given the predominance of clientelism. Rather than a focus on policy for
development’s sake, implementation more often follows the informal processes related
to clientelism. Patron-client networks give rise to factionalism with groups loyal to the
regime considered insiders while those who compete with the empowered regime left
outside of patronage networks. This notion of political insiders and outsiders is
described by Bratton and van de Walle with transitions away ﬁ'om neopatrimonialism
towards liberalization and democracy often occurring as access to patronage wanes and
factionalism can no longer be contained within the established networks (84).

Clientelism is antithetical to enhanced governance and democratization. As Hyden
(2006) puts it, “Political rulers treat the exercise of power as an extension of their private
realm. ...Clientelism is deemed problematic, especially in circles that are concerned with
improving governance in Aﬁican countries. It keeps Aﬁican countries barely aﬂoat, but
it does not help them swim forward,” (79).

Anderson and Guillory (1997) suggest that political winners and losers can be

assessed at the individual level, similar to how Bratton and van de Walle (1997) describe

131

access to patronage networks. Anderson and Guillory code individuals as political
minority or majority members based on whom they supported in the last election.
Bratton et al. (2005) suggests that similar assessments can be made based on whether an
individual citizen indicates they are close to the political party that won the last national
election (260). The concept of political cleavage groups allows for an individual level
assessment of access to patronage networks. Such a measure is based on the notion of
clientelism as playing a key role in who gains access to resources and who is restricted
ﬁom such access based on the assumption that ruling party support will be rewarded
through access to political patronage.

Literature Synthesis

The literature described above comes ﬁ'om a variety of sources: literature on
individual policy performance of social service provision, literature on the role of the
state and regime in Africa, and the role of clientelism have all been discussed.

First, the public opinion literature suggests that individual level respondents will
honestly assess AIDS policy performance. Bratton (2007) and Youde (2009) suggest
that Africans assess AIDS policy performance similar to how they evaluate other
policies. It is clear from the public opinion research that AIDS and health more
generally are important to citizens, with demand for effective responses high and
satisfaction on HIV/AIDS responses higher than for other policy issues according to
Bratton et al. (2005). The literature on the role of the regime and the state in Aﬁica
suggests that when more democratic, stronger countries should be more effective at
provision of these types of political goods than weaker, non-democratic countries.

Bratton et al. (2005) further indicates satisfaction with state and regime characteristics

132

can be assessed at the individual level (247).

Clientelism also plays a key role in Aﬁican politics. As Bratton and van de Walle
assess, access to patronage networks is often based on one’s access to the regime.
Previous inquiry makes the case that access to locally based patrons is an effective
predictor of one’s likelihood of AIDS resource distribution. Anderson and Guillory
(1997) suggest that individuals can be divided into political majority and minority
statuses on their support of the regime during the most recent election.

The Afrobarometer data allow for these various considerations to be assessed at the
individual level. Youde (2009) provides an initial model of individual evaluation of
policy performance. Although this is an effective ﬁrst cut, Bratton et a1 (2005) suggest
that state and regime considerations can be measured reliably at the individual level.
Based on the ﬁndings from earlier dissertation work and Strand, Mattes, and Kinney
(2008) the effectiveness of the state and regime should be revisited with individual level
Afrobarometer data. Furthermore, Anderson and Guillory provide a method for
assessing the role of clientelism in determining who can access resources. Youde
included HIV prevalence rates in his analysis, and he also included dummy variables at
the country level which were all strongly signiﬁcant.

Youde’s analysis provides an effective ﬁrst attempt, but as the literature review
suggests, other variables should be included to test several potentially important factors:
ﬁrst, the regime and state literature suggests that individual interaction with the regime
and state may inﬂuence their satisfaction with HIV/AIDS responses. Second, one’s
political status as a member of the majority or not is an important consideration for

determining one’s satisfaction with AIDS policy. Additionally, improved data allows

133

 

for greater speciﬁcation than the country level dummy variables he utilizes. Though
including HIV prevalence rates is important, the inclusion of ARV coverage, a more
robust measure of government response to the disease may improve speciﬁcity. HIV
prevalence may capture one’s lived experience with the disease, but the availability of
ARVs would demonstrate tangible distribution of goods. Individual satisfaction with
AIDS policy would likely increase if ARVS were available.

What follows is an attempt to put these concepts together. Do individual assessments
of the state and regime follow the cross-national ﬁndings suggested previously in cross-
national analysis? Do citizens who supported the regime in power have greater
satisfaction with their government’s AIDS response than those ﬁom political minorities,
similar to sub-national ﬁndings? Can a model be developed that encapsulates these
research questions and provide greater explanatory power than Youde’s model (2009)?
The next section will detail the variables used, how these variables will be measured,
and a set of hypotheses aimed at achieving this goal.

Data, Variables, Measures, and Hypotheses
Data

The section below provides information on data sources. The vast majority of data
for this project comes from surveys in Aﬁobarometer Round 4. Surveys were conducted
ﬁom 2008 to 2009. Data is available for twenty Aﬁican countries, the largest number
included to date by the Aﬁobarometer with over 27,000 citizens surveyed in total. Most
previous literature utilizes data ﬁom Rounds 1 and 2, though Youde’s model is based on
round 3 data. This project will be among the ﬁrst to utilize the expanded twenty country

data set from round 4 as it relates to satisfaction with HIV/AIDS interventions.

134

 

Survey questions allow assessment of citizen attitudes in a number of areas including
demographic inﬂuences, service accessibility, knowing someone who has died of the
disease, and opinions on health service provision overall. Data related to these questions
allow for combining some of the most important predictors from Bratton’s (2007) and
Youde’s (2009) empirical models.

There are several additional questions available that can bring even greater leverage
over our research question. Bratton et al. (2005) suggests that individual assessments of
states, regimes, and governments can be effective predictors for policy performance
satisfaction (247). In this analysis, dependent variables include commitment to
democracy, perception of democracy, support for economic reforms, and satisfaction
with structural adjustment (247). Individual measure assessments of state, regime, and
governmental effectiveness can just as easily be used to consider whether citizens’
evaluations of these institutions inﬂuence their evaluation of HIV/AIDS responses. By
utilizing measures of individual assessments of these individual institutions, it will be
possible to incorporate concepts that inﬂuenced ARV coverage at the cross-national
level. In changing levels of analysis, I am assuming that Aﬁican citizens should view
policy performance more favorably if their governments have been more successful in
combating the AIDS crisis. Citizen evaluation of policy performance is assumed to be
at least partially based on whether citizens see anti-AIDS resources as work. The
individual level evaluations of state and regime allow for integrating the concepts as
discussed in the state and regime literature above.

Anderson and Guillory (1997) provide a useful measure of political access to the

ruling regime. They divide citizens into whether they voted for the ruling party in power

135

during the most recent election provides an example of how political winner and loser
cleavages can be delineated. Individuals are coded in terms of their support for the
ruling party in the most recent elections. This measure takes into account previous
ﬁndings from previous sub-national research based in the literature related to clientelism
in Aﬁica.

In addition to individual level analysis, country ﬁxed effects taking into account
ARV coverage and HIV prevalence rates will be included. Governmental responses to
HIV/AIDS have increased dramatically in the last few years, at least in terms of ARV
coverage.

Forty-four percent of Aﬁicans who need AIDS treatment have access as of 2009
(UNAIDS, 2009). Adding ARV coverage should provide improved speciﬁcity to the
dummy variables in Youde’s (2009) model, the vast majority of which were found to be
statistically signiﬁcant.

Measures and Variables

Variables are broken into several categories. The dependent variable will be
discussed ﬁrst before independent variables are unpacked. Independent variables are
broken down into several categories: social structural variables, state capacity, regime
type, political cleavage measures, and variables that get assess individual experience
with HIV/AIDS and health services more broadly. Once the variables are parsed into
categories and justiﬁed, hypotheses will be developed in the next section of the paper.
Dependent Variable

In this analysis, we seek to explain citizen satisfaction with their government’s policy

performance related to HIV /AIDS. Rather than seeking to proxy ARV coverage, the

136

 

goal of this inquiry is evaluation of what drives citizen satisfaction with policy
performance. Capturing citizen satisfaction with HIV/AIDS policy performance allows
us to investigate whether ARV availability or other variables inﬂuence how citizens
evaluate HIV/AIDS interventions. The dependent variable for this study asks the
following: How well or badly would you say the current government is handling the
following matters, or haven’t you heard enough to say: Combating HIV /AIDS?
Responses range on a four point scale from very well to very badly. Below is a

distribution of this variable for 20 Aﬁican countries in 2008 and 2009.

Table 3.1: Frequency of Citizen Responses Regarding Government HIV/AIDS
Policy Performance

 

 

 

 

 

 

Cumulative
Frequency Percent Percent
Valid Missing 26 .1 .1

Very Badly 2927 10.6 10.7
Fairly Badly 3702 13.4 24.0
Fairly Well 11219 40.5 64.5
Very Well 7776 28.1 92.6
3:; :gjgamt , 2063 7.4 100.0
Total 27713 100.0

 

 

As we can see, most African citizens, nearly seventy percent, assess their
governments to be doing a good job handling the disease. This ﬁnding dovetails with
past Afrobarometer ﬁndings which suggest that citizens are more satisﬁed with their
governments on their response to HIV /AIDS than on other political issues (Bratton et al
2005). Satisfaction is similar to Round three ﬁndings in 2005 which found similarly that
about seventy percent of citizens across the continent were satisﬁed with their
governments HIV/AIDS policies (Bratton, 2007). This represents a small increase since

2000 where sixty-two percent of citizens were satisﬁed with AIDS responses (Bratton et
137

al. 2005). This table also indicates an N exceeding 27,000, meaning we have the
degrees of freedom required to include numerous variables. This allows for
investigation into a wider variety of measures than in previous cross-national or sub-
national inquiries.

Social Structural Variables

Demographic variables are selected from Bratton’s model of satisfaction with health
care policy performance (2007). Bratton includes gender, residential location,
educational attainment, and poverty as social structural variables. His empirical model
reveals that women are less likely to be satisﬁed with health services. This is likely to
be especially true as HIV/AIDS disproportionately impacts women, and women’s health
services are especially lacking across the continent. Rural citizens are more satisﬁed as
well. Urbanites are more likely to have increased expectations of social service delivery
and are likely to be less satisﬁed as these expectations go unﬁrlﬁlled. Bratton also ﬁnds
that more educated citizens and people living in poverty are less satisﬁed with health
care services. As AIDS is a speciﬁc health care intervention, my expectation is that
similar relationships will occur in terms of HIV.

Gender, education, and rural dwelling are all coded in the Afrobarometer data. To
assess poverty, I will use a question asking citizens to assess their present living
conditions. This question is coded into ﬁve parts ranging ﬁ'om very good to very bad.
Youde used the same question in his empirical model and found a moderately strong
relationship suggesting that increased levels of poverty decrease satisfaction with
government performance (2009). This ﬁnding is similar to Bratton who also found a

strong relationship between higher levels of poverty and dissatisfaction with the

138

‘ _

r.

government on health service provision.
Individual Assessments of the State

At the individual level the goal is to seek out whether an individual’s interaction with
state agencies and their trust of the state institutions generally inﬂuence satisfaction with
HIV /AIDS interventions speciﬁcally. In order to assess institutional trust, an index is
constructed from questions on trust in the court system, the police, and their national
electoral system. Taken together, these questions will be used to measure whether
citizens trust their respective states.

A second component of the state that may inﬂuence citizen assessment of their
respective state’s institutions is corruption. Do citizens assess the members of
governmental institutions as corrupt, and if so, does a perception of corruption inﬂuence
satisfaction with policy performance? The Afrobarometer includes questions asking
citizens to assess the proportion of ofﬁcials from several institutions who are corrupt,
ranging ﬁ'om none to all. Institutions include the ofﬁce of the presidency, parliament,
tax ofﬁcials, and judges, among others. This measure of perceived corruption is based
on the work of Bratton et al. (2005) who used a similar index to capture the inﬂuence of
corruption on commitment to democracy (247).

Taken together these measures of institutional trust and perceived corruption are used
to measure how individuals perceive their governments. These measures seek to capture
whether individuals perceive their state institutions as both trustworthy and honest.
Individual Assessments of the Regime

In assessing the regime, the goal is to answer whether individual assessment of the

regime inﬂuences their satisfaction with policy performance. At the aggregate country

139

level, a positive relationship existed between more democratic regimes and greater ARV
provision, although this relationship was not statistically signiﬁcant. Do these
relationships exist at the individual level? Speciﬁcally, do individuals assess democratic
regimes as prone to providing more satisfactory social services? The public health
literature suggests that at the country level, democratic regimes lead to greater service
provision. Here we seek whether such patterns exist at the individual level.

To assess citizen-regime interaction, several measures are included. First, an index
of political rights is constructed. This index is based on questions which ask individuals
whether they have freedom to say what they think, freedom to join the political group of
their choosing, and freedom to select their own candidate at election time. Taken
together these questions can be used to measure whether citizens believe political rights
are being delivered by the regime in power. This index is used by Bratton et al. (2005)
in a similarly constructed fashion.

Another question used to assess citizen-regime relations is whether citizens view
their electoral processes as free and fair. This incorporates whether citizens view
elections as providing the type of vertical accountability described by Diamond and
Morlino (2005). Another component of citizen-regime relations is whether citizens
believe they have to be careﬁrl about they say. In other words, do citizens believe they
have ﬁ'eedom of speech? This question speciﬁcally asks about political speech and
provides a more speciﬁc measure than the political rights question used in the index
described above.

Lastly, a measure is included to assess citizen satisfaction with democracy. A

question asks citizens whether satisﬁed they are with how democracy works in their

140

country. This question seeks to answer whether citizens are satisﬁed with democracy,
and responses range on a four point scale from very unsatisﬁed to very satisﬁed with
democracy.
The Role of Political Cleavages

This set of variables seeks to assess whether political cleavages inﬂuence citizen
satisfaction with HIV /AIDS performance. Political cleavages here refer to ethnic or
political afﬁliation determining who accesses these resources. Measures of political
cleavages seek to include two divergent concepts: political clientelism in Aﬁica and
Lieberman’s ﬁndings that ethnic heterogeneity prevents effective policy making. Taken
together, included measures seek whether political clientelism based on access to the
regime in power inﬂuences satisfaction with HIV /AIDS interventions or whether ethnic
considerations are more salient. To measure political clientelism, citizens were asked
which party they would vote for in a hypothetical future election. This variable is coded
into those who support the party in power, similar to the political winner/loser
dimension developed by Anderson and Guillory (1997). Previously reviewed literature
on clientelism suggests that distribution of state resources through neopatrimonial
networks is a positive predictor of loyalty to the party in power. This winner/loser

measure seeks to capture whether clientelism leads to a gap between those who are and

13 .
who are not able to access state resources . Bratton et al. (2005) suggest there IS
signiﬁcant variation in citizen satisfaction with a variety of institutions based on their
political winner/loser status (260). Previous research indicated that political clientelism

at the sub-national level was a strong predictor of ARV program distribution. Areas

 

'3 In future research it would be useful to control for the ethnic group in power to
investigate whether this inﬂuences citizen satisfaction.

14]

inhabited by citizens that elected executives and parliamentarians of the party in power
were more likely to receive ARV programs than areas that elected opposition parties.

The other important political cleavage measure seeks to include Lieberman’s
argument that ethnic politics dominate effective policy making. Lieberman argues that
ethnic heterogeneity prevents effective policy making. First, citizens are asked which
ethnic group they associate with. Subsequent Afrobarometer questions ask citizens to
compare their ethnic group’s economic position to others, their relative political power
compared to other ethnic groups, and whether their group is treated fairly by the
government. These three questions are combined into an index of ethnic treatment
which scales together. Previous inquiry did not reveal the type of ethnic inﬂuences
found by Lieberman at the country or sub-national levels. Including ethnicity seeks
whether group membership is inﬂuential in policy satisfaction.
Health Experiences
‘ No Aﬁican country has managed to provide an effective AIDS response that provides
resources to every citizen. This set of variables seeks to answer whether empowered
governments allocate these scarce, highly-valued commodities to the beneﬁt or
detriment of certain segments of society, leading to increased policy satisfaction.

At the Individual Level

This set of variables seeks to assess whether individual experience with the overall
health care system and with HIV/AIDS speciﬁcally inﬂuence satisfaction with
HIV /A1DS responses. First, a measure is included that asks how satisﬁed citizens are
with the provision of basic health services. One would expect that greater satisfaction

with health services would correlate with greater satisfaction with HIV /AIDS policy

142

speciﬁcally as most countries developed their HIV/AIDS responses utilizing long-
standing health care institutions. Both Youde (2009) and Bratton (2007) ﬁnd that
overall satisfaction with health services is a strong predictor of HIV/AIDS satisfaction.
Additionally, an Afrobarometer question asks citizens whether they know anyone who
has died of HIV/AIDS. Youde (2009) ﬁnds that knowledge of someone who died of
HIV/AIDS makes citizens more likely to look favorably on their government’s effort to
combat the disease.

At the Area Level

To assess a citizen’s access to health care resources, Afrobarometer interviewers
coded whether a health clinic existed within each research unit. Although we cannot
control for the quality of these facilities or whether they have ARV resources, Bratton
(2007) ﬁnds that the availability of a clinic has a positive inﬂuence on satisfaction with
health services (24). Including this variable investigates whether clinics are considered
by individuals when they are asked to assess HIV/AIDS interventions.

At the Country Level

Two measures are included to assess whether individual satisfaction with their
government’s HIV/AIDS response is inﬂuenced by national level inﬂuences. First, HIV
prevalence rates are included. Youde (2009) ﬁnds a negative relationship between
national level HIV prevalence and individual satisfaction with government responses
(229). This ﬁnding provides evidence that citizens are able to assess the role of
HIV/AIDS in their lives. Second, based on Youde’s evidence that national level
measures can inﬂuence citizen satisfaction, ARV coverage is also included. Both of

these variables are included as country level ﬁxed effects.

143

Hypotheses

In this section the variables will be put into context. Hypotheses are categorized in
the following groups: social structure, attitudes toward the state, attitudes toward the
regime, political cleavages, individual attitudes about health services, and country ﬁxed
effects.

Social Structural Hypotheses

H1: Individuals who assess their living conditions positively will be more likely to
view governmental performance at combating HIV/AIDS favorably.

H2: Women will be less likely to approve of governmental HIV/AIDS performance
than men.

H3: Education will be negatively related to approval of governmental HIV/AIDS
performance.

H4: Rural citizens will be more satisﬁed with governmental HIV/AIDS interventions

than city dwellers.

State Based Hypotheses

H5: Individuals who trust state institutions are more likely to be satisﬁed with their
government’s HIV/AIDS performance.

H6: Individuals perceptions of ofﬁcial corruption will undermine satisfaction with the

government’s HIV/AIDS performance.

Regime Based Hypotheses

H7: Individuals who believe regimes are providing political rights will be more likely

144

 

to view governmental HIV/AIDS performance favorably.

Hg: Individuals who believe elections are free and fair will be more likely to view

governmental HIV/AIDS performance favorably.

H9: Citizens who believe they can speak ﬁeely will assess governmental HIV/AIDS

performance more favorably.

H10: Citizens who support democracy will be more likely to be satisﬁed with

governmental HIV/AIDS performance.

Political Cleavage Hypotheses

H11: Political winners will be more likely to view governmental HIV/AIDS

performance favorably than political losers.

H12: Citizens who identify their ethnic group as being treated better than or equal

will view governmental HIV/AIDS interventions more favorably than those who claim
their ethnic group is not treated equally.

Citizen Level Health and AIDS Experience

H13: Citizens with favorable evaluation of health care services generally will also

have more favorable evaluations of governmental HIV/AIDS performance than those
who evaluate health care less favorably.

H14: Citizens who know people who have died of HIV/AIDS will look more
favorably on governmental HIV/AIDS performance than those who do not claim to

know anyone who died of the disease.

Area Level Health Experience

145

 

H15: Citizens living in areas that have a health clinic will be more likely to have

favorable evaluations of governmental HIV/AIDS performance than those who live in
areas without hospital services.

National Level Health Care Hypotheses

H16: Citizens living in countries with higher HIV prevalence rates will have less

favorable view of governmental HIV /AIDS performance than citizens who live in

countries with lower HIV prevalence rates.

H17: Citizens living in countries with higher ARV coverage rates will have more

favorable evaluations of governmental HIV/AIDS performance than citizens who live in

countries lower ARV coverage rates.

146

Descriptive Statistics

Table 3.2 Descriptive Statistics for Independent Variables

 

 

 

N Range Mean Dersritgion Variance
Q4b. Your present living conditions 27713 10 2.65 1.250 1.562
Q101. Gender of respondent 27713 1 1.50 .500 .250
Q89. Education of respondent 27713 100 3.27 3.916 15.334
Urban or Rural Primary Sampling Unit 27713 1 1.62 .485 .236
Institutional Trust Index 49C 49G 49H 27713 10.00 2.0830 1.64591 2.709
Perceived Corruption Index 50 A-G 27713 10.00 2.7533 2.57254 6.618
Perceived Political Rights Index 15 A-C 27713 10.00 3.6254 1.06190 1.128
Q71. Elections free and fair 27713 10 3.45 1.971 3.886
Q46. How often careful what you say 27713 10 2.07 1.853 3.433
Q43. Satisfaction with democracy 27713 10 3.05 2.074 4.303
POLWINNER 18756 1.00 .5940 .49109 .241
Ethnic Treatment Index 80 81 82 27713 10.00 3.0439 1.68045 2.824
Q57g. Handling improving basic health 27713 10 2.77 1.412 1.993
services
EA-FAC-D. Health Clinic in the PSU/EA 27713 9 .72 .977 .954
Q95. Know died of AIDS 27713 10 .66 1.622 2.632
HIV Prevalence 26449 23.80 8.7525 7.55935 57.144
ARV Coverai 26449 73.00 27.0557 18.82732 354.468

 

 

Before running the empirical models, it is useful to report descriptive statistics of the

indexes for citizens across twenty Aﬁican countries. See table 3.3. Variables are listed

in the previously developed categories. First, we see that the average Aﬁican assesses

his or herself as living in a little less than average living conditions. As many men as

women were included in the sample. The average person completed primary education.

More rural than urban respondents were included in the samples, reﬂecting the overall

total population within these countries.

When asked about their level of trust of three political institutions—the national

electoral commission, the court system, and the police——the most frequent response can

147

‘.

be described as lukewarm. Institutional trust responses range ﬂom “not at all” to “a lot”
along a four point scale. The average citizen trusts institutions somewhat. When asked
about their perception of corruption across a variety of government ofﬁcials, the average
African citizen believes that some to most government ofﬁcials are corrupt. When
institutional trust is compared to corruption, it appears that Aﬁicans trust institutions
more than they trust those who run them.

When asked about political rights, Aﬁican citizens have a somewhat more positive
outlook. Perceived political rights are scaled ﬂom one to four with four indicating
citizens believe they are completely ﬂee. As one can see, the average is over 3.6,
suggesting that the average citizen believes they are quite ﬂee. A similar average of
3.45 reﬂects that Aﬁicans believe their election processes are largely ﬂee and fair. Free
speech is less apparent with the average response suggesting that Africans must often be
careful what they say. Overall, the average African reports fair satisfaction with
democracy, which is reﬂective of the individual questions about different democratic
components.

In terms of ethnic group treatment, most Africans feel their ethnic group is treated
similarly to others. Ethnic treatment ranges ﬂom one to ﬁve with a one indicating their
group is treated better than others. Interestingly, more citizens believe their group is
advantaged compared to others rather than disadvantaged. Citizens were also asked
which political party they would vote for if an election was held tomorrow for president.
This variable was coded one for citizens who selected the ruling party and zero for other
political parties. As we can see, nearly sixty percent of African citizens polled

suggested they would support the ruling party in their respective countries.

148

 

Several questions asked citizens about their interaction with health services. When
asked about their government’s performance on providing basic health services, the
average score was between fairly badly and fairly well, which I argue reﬂects an average
performance evaluation. More than sixty percent of survey participants lived in areas
with health clinics within the survey area. Nearly thirty ﬁve percent of those surveyed
said they knew someone who had died of HIV/AIDS, a ﬁgure that reﬂects a willingness
to discuss this difﬁcult topic. Country level statistics reﬂect that the average citizen
lived in a country with over eight percent HIV prevalence with just more than one
quarter of those who needed ARVS receiving these medications.

In the next section regression models are listed followed by a brief summary of
results. Analysis and unpacking these relationships follows thereaﬁer. Bold numbers
reﬂect the rank order of individual variable’s explanatory power based on standardized
coefﬁcients which are not shown. Results are brieﬂy summarized and analysis follows

thereafter.

149

Table 3.3: OLS Regression Results for Citizen Satisfaction with HIV/AIDS Policy Performance

 

 

Model 1 Model 2
Constant 1 .379 1.427
(0.055)*** (0.077)***
Social/ Structural Present Living Conditions -0.013 -
(0.009) -
Women -0.005 -
(0.022) -
Education -0.007** -0.009
(0.003) (0.007)
Rural Citizens 0.035 -
(0.023) -
State Capacity Institutional Trust Index 0095*" (3) 0097*" (3)
(0.009) (0.012)
Perceived Corruption 0062*" (2) 0063*” (2)
(0.005) (0.007)
Regime Type Perceived Political Rights 0077*” 0078*"
(0.012) (0.017)
Elections Free and Fair 0020*“ 0.020"
(0.007) (0.008)
Freedom of Speech 0027*" 0028*"
(0.007) (0.008)
Satisfaction with Democracy 0055*" (5) 0.055*** (5)
(0.006) (0.009)
Political Cleavages Political Winners -0.046* -0.048*
(0.024) (0.025)
Ethnic Treatment Index 0.012 -
(0.008) -
Citizen Level Health Eval. Basic Health Services 0364*" (1) 0362*" (1)
(0.009) (0.014)
Know Person Died of HIV/AIDS -0.015** -0.014*
(0.007) (0.009)
Area Level Health Health Clinic in the Area -0.009 -
(0.009) -
National Level
Health HIV Prevalence -0.015*** (4) -0.015*** (4)
(0.002) (0.002)
ARV Coverage 0005*” (6) 0005*" (6)
(0.001) (0.001)
R Squared 0.152 0.152
Adjusted R Squared 0.151 0.151

 

150

 

Model 1 provides a baseline model that includes all of the previously listed measures

tested simultaneously”. As one can see, this model explains about ﬁfteen percent of
variance. Of the social structural variables, only education has a statistically signiﬁcant
inﬂuence on citizen evaluation of HIV/AIDS performance. As hypothesized, more
educated citizens are less satisﬁed with their governments. None of the other measures
ﬂom this category of independent variables was statistically signiﬁcant. We see that the
state capacity variables are among the strongest predictors of citizen satisfaction with
perceived corruption and increased institutional trust both leading to greater satisfaction.
The surprising positive relationship between perceived corruption and increasingly
positive performance evaluation will be unpacked below. We see that each of the
regime type variables has a positive and statistically signiﬁcant relationship with more
favorable performance evaluation. Political winners are less satisﬁed, and no
discernable relationship exists between ethnic group afﬁliation and HIV/AIDS policy
evaluation. Among the health variables we see that the strongest predictor of citizen
satisfaction is individual evaluations of overall health service delivery. This ﬁnding
suggests that individuals may nest their evaluation of HIV/AIDS policy within overall
health service performance evaluation. Individuals that know someone who had died of
the disease are less likely to assess their country’s HIV/AIDS performance as effective.
Health clinics did not play a role in how individuals evaluated AIDS policy
performance. Both national level variables were important predictors with both HIV
prevalence and ARV coverage among the most signiﬁcant predictors across individuals,

reﬂecting that citizens do perceive the effectiveness of HIV/AIDS policies when they

 

'4 Bold numbers in each model reﬂect the rank order of the strength of relationships
based on standardized coefﬁcients (not shown).

151

 

see them in action.

In model 2 I remove the variables that were not statistically signiﬁcant and provide a
more parsimonious model of AIDS policy satisfaction at the individual level. Removing
insigniﬁcant variables does not reduce explanatory power signiﬁcantly. Education is
kept in the model as it is nearly statistically signiﬁcant. The same variables remain
inﬂuential predictors, and the same six variables continue to drive the model. Based on
standardized coefﬁcients (not shown) satisfaction with basic health care services
remains the most important predictor, followed by citizen interaction with state
institutions. Regime level indicators remain signiﬁcant with individuals who believe
they live in under a more democratic regime more satisﬁed with AIDS policy
performance. We also see that country level measures indicate that citizens are more
likely to be satisﬁed when HIV prevalence is lower and ARV coverage is higher. Model
2 was slightly heteroskedastic, so robust standard errors are reported in this model to
correct for heteroskedasticity. Now that models have been speciﬁed, the next section
provides further analysis of these results.

Discussion and Analysis

In terms of the state capacity variables, one can see that both institutional trust and
perceived corruption are among the most important variables in the model. Greater
institutional trust and increased perception of corruption are important predictors of
satisfaction with governmental HIV/AIDS responses. In other words, citizens who trust
their states are most satisﬁed with how these institutions have handled the AIDS crisis.
The increased perceived corruption ﬁnding is surprising. I interpret this to be based on

the anomalously high levels of citizen satisfaction with HIV /AIDS policy performance

152

 

alongside high levels of perceived corruption. When they are asked about satisfaction
with policy performance on other policy areas, HIV/AIDS policy performance is
consistently higher than is support for other policy areas. Further evaluation reveals that
in countries where HIV/AIDS is less salient due to lower HIV prevalence rates, like in
parts of West Africa and in Madagascar, citizens are more likely to state that HIV/AIDS
policy is more effective than in high salience, high prevalence countries. In cases like
Uganda, and the countries in southern Aﬂica where HIV/AIDS policy is of higher
salience as part of daily life, respondents report greater variance within these settings. In
low salience countries, individuals may be evaluating these policies more favorably due
to a lack of information or lack of preference while simultaneously wanting to provide
the least offensive response.

The other possibility is that HIV/AIDS policy is rife with corruption. When I
included a measure to test experienced corruption, a weak negative relationship exists,
but this does not seem to drive HIV/AIDS policy performance evaluation. Anecdotal
evidence supports that HIV/AIDS service delivery is fairly corrupt with examples
ranging ﬂom stolen funds and medications to ineffective counterfeit drugs (Jones, 2005,
Tayler and Dickenson, 2005). Though corruption is a barrier to effective policy
performance, this peculiar ﬁnding is due in large part to high levels of HIV /AIDS policy
support rather than high levels of corruption driving the model. It also may suggest that
citizens are willing to tolerate a sizeable amount of corruption as long as they see their
governments functioning effectively. When they see their governments working,
citizens may view these types of relationships at clientelism; it is only when service

delivery breaks down that they view these processes as corrupt. The strength of the

153

 

relationship between institutional trust and high levels of policy performance satisfaction
also bolsters previous work which found similarly that the strength of state institutions
was a critical predictor of country-level ARV coverage.

Among the regime type measures, all of these variables are statistically signiﬁcant,
though satisfaction with democracy is the only one that is among the top variables in
terms of explanatory power. All four indicators similarly indicate that citizens who
assess their regimes as more democratic also believe their respective regimes are
handing the AIDS crisis more effectively. People who believe they have greater
political rights, those who assess elections as ﬂee and fair, individuals who believe they
can speak ﬂeely, and citizens who are more satisﬁed with democracy are also more
satisﬁed with their government’s efforts related to HIV/AIDS. Taken together regime
type ﬁndings indicate that African citizens evaluate health services as an important
component of governmental service delivery, suggesting that governments that
effectively allocate resources will be rewarded with more loyal, satisﬁed citizens.

When one looks at the political cleavage measures, we see that an individual’s
political status (as measured by expressed closeness to the ruling party) is more
important than individual assessment of how one’s ethnic group is treated. Although
neither of these variables is an especially strong of predictor compared to the other
categories, these ﬁndings call to question Lieberman’s ﬁndings related to ethnic
boundaries. At the individual level, political afﬁliation with the ruling party is a better
predictor of HIV/AIDS policy satisfaction than one’s assessment of how their ethnic
group is treated. Surprisingly, individuals who support the ruling party are less satisﬁed

with HIV /AIDS policy. This unexpected ﬁnding suggests that like greater educational

154

 

attainment, supporters of the ruling party place increased expectations on what their
governments can achieve. When these govemments are unable to deliver services to
these individuals, a gap develops as supply fails to meet demand, leading to decreased
policy satisfaction among ruling party supporters. The implication ﬂom this ﬁnding
indicates that policy performance trumps loyalty when such a gap develops. loyalty to
the ruling party requires continued effective policy performance, or citizens will
reconsider their political allegiance.

In terms of citizen level interaction with health care institutions, citizen satisfaction
with the provision of basic health services is the strongest predictor of citizen
satisfaction with HIV/AIDS interventions. This relationship conﬁrms that citizens think
of HIV/AIDS and health somewhat interchangeably. The strong relationship between
health and HIV/AIDS performance also suggests that better health care institutions lead
to better AIDS policy effectiveness as well. This interpretation is based on the idea that
citizen satisfaction is most tightly tied to overall evaluation of the health care system at
the individual level, and overall ARV coverage at the national level is similarly tied to
state institutional strength. Taken together, these results indicate that HIV /AIDS policy
performance is nested within the overall health system at the country level and within
one’s interaction with these institutions at the individual level. We also see that
knowing someone who died of AIDS negatively inﬂuences a citizen’s view of their
government’s performance, contrasting Youde’s earlier ﬁndings. This ﬁnding makes
sense as citizens would be less satisﬁed with health policies when those close to them
die of HIV /AIDS—losing someone to the disease would increase saliency of the disease

and provide a concrete negative experience for a person to judge policy performance.

155

 

The existence of a health care facility within the surrounding area is not a signiﬁcant
predictor of policy satisfaction. The lack of a relationship between the proximity of
clinics and policy performance was a surprising ﬁnding. This ﬁnding indicates that
when citizens think about HIV/AIDS, they consider their government’s policy
performance along more expansive lines than just their immediate local areas. The lack
of relationship between clinics and HIV /AIDS policy evaluation may also indicate that
clinics do not enter a citizen’s mind when asked about HIV/AIDS policy, potentially due
to citizens evaluating these clinics as ineffective.

Country level ﬁxed effects were among the most important predictors of HIV/AIDS
policy satisfaction. Citizens who live in countries with higher HIV prevalence rates are
less satisﬁed than citizens who live in, countries with lower prevalence. One can think of
this indicator of an individual’s day to day experience with the disease: where HIV
prevalence is higher one would expect citizens to experience HIV/AIDS more readily as
part of their day to day lives. Similarly, greater availability of ARVS is also among the
most important explanatory variables. Living in countries that have prioritized AIDS
treatment makes citizens more likely to be satisﬁed with governmental efforts related to
the disease. These ﬁndings remained statistically signiﬁcant even when country was
included in robustness testing, which suggests that policy output resonates with
individuals when they consider policy performance.

When citizens think about HIV /AIDS policy, health generally has a strong inﬂuence
upon their opinions. Where healthcare is more readily available, citizens tend to be
more satisﬁed. These ﬁndings give credence to the argument that stronger, higher

capacity states are better able to respond to the AIDS issue leading to greater citizen

156

 

satisfaction. It is not surprising that Botswana is the country with the highest levels of
citizen satisfaction. Botswana also was the most effective country at ARV coverage
during cross-national analysis and has set the Aﬁican standard for an effective policy
response. This is contrasted by Nigeria where citizens are less satisﬁed, both with
HIV/AIDS policy speciﬁcally and health service delivery more generally. At the
country level, Nigeria has only allocated ARVS to 13 percent of those who need them;
contrastingly, Botswana has managed nearly eighty percent coverage. These national
level output measures appear to resonate at the individual level as well.

When citizens evaluate HIV/AIDS policy, country level measures of HIV prevalence
and ARV coverage are two of the strongest predictors of citizen satisfaction. Citizens
living in low HIV prevalence and higher ARV coverage countries are more satisﬁed
than others. Such low prevalence and expanded ARV programs are the direct output
measures of governmental action. The fact that these variables are among the most
salient predictors make it clear that citizens know effective policy when they see it,
whether that is through effective AIDS prevention campaigns, availability of condoms,
or through greater testing and ARV treatment efforts. When citizens see these
programs, they not only beneﬁt ﬂom a lessened burden of disease and effective
treatment efforts, but they also are willing to credit their governments for improved

policy performance.

157

 

Figure 3.1: Summary of Results and Analysis

 

 

Hypothesis
H1: Wealthier Self Assessment —+
Greater HIV /AIDS Policy Performance
H2: Women -» Less Satisfaction with
HIV/AIDS Policy Performance
H3: Greater Education —> Less
Satisfaction with HIV/AIDS Policy
Performance
H4:Rural Citizens —> Greater
Satisfaction with HIV/AIDS Policy
Performance
H5: Greater Institutional Trust—>
Greater Satisfaction with HIV/AIDS
Policy Performance
H6: Less Perceived Corruption—r
Greater Satisfaction with HIV/AIDS
Policy Performance
H7: Greater Political Rights—* Greater
Satisfaction with HIV/AIDS Policy
Performance
H8:Free and Fair Elections —-v Greater
Satisfaction with HIV/AIDS Policy
Performance
H9: Free Speech —> Greater
Satisfaction with HIV/AIDS Policy
Performance
H10: Greater Support for Democracy
--3 Greater Satisfaction with HIV/AIDS
Policy Performance
H11: Support for Ruling Party ->
Greater Satisfaction with HIV/AIDS
Policy Performance
H12: Greater Ethnic Self Assessment
—» Greater Satisfaction with HIV/AIDS
Policy Performance
H13: Greater Health Provision
Satisfaction —» Greater Satisfaction
with HIV/AIDS Policy Performance
H14: Knowing AIDS Victim —> Greater
Satisfaction with HIV/AIDS Policy
Performance
H15: Living Near 3 Clinic —> Greater
Satisfaction with HIV/AIDS Policy
Performance
H16: Greater HIV Prevalence in
Country —v Less Satisfaction with
HIV/AIDS Policy Performance
H17: Greater ARV Coverage -->
Greater Satisfaction with HIV/AIDS
Policy Performance

Expected

Positive

Negative

Negative

Positive

Positive

Negative

Positive

Positive

Positive

Positive

Positive

Positive

Positive

Positive

Positive

Negative

Positive

Actual

Negative

Negative

Negative

Positive

Positive

Positive

Positive

Positive

Positive

Positive

Negative

Positive

Positive

Negative

Positive

Negative

Positive

Statistical

No

No

Yes

Yes

Yes

Yes

Yes

Yes

Yes

No

Yes

Yes

Yes

Yes

Strongest

Relationship Relationship Signiﬁcance Predictors

 

158

 

The table above provides a summary of the relationships ﬂom model one. The model
revealed a couple surprising ﬁndings that conﬂict with the hypothesized relationships.
The surprising ﬁnding related to corruption has been previously discussed. Another
surprised was that support for the ruling party does not lead individuals to be more
satisﬁed with HIV /AIDS policy once other variables are considered. Ruling party
support actually decreases citizen satisfaction. This is assessed similar to education and
urbanization which leads citizens to increase their expectations which may lead to
decreased satisfaction. This set of variables gives further credence to the discussion
related to high levels of support for HIV/AIDS policy simultaneously existing with high
levels of perceived corruption: in low prevalence countries where people lack education
and live in rural areas, HIV/AIDS is of lower salience compared to other more
immediately pressing economic concerns like increasing job opportunities, narrowing
income gaps, and keeping prices low. When citizens’ basic needs are more effectively
met and they gain education and move to cities, their expectations for other services
increase. This leads to greater variance as it relates to HIV/AIDS policy performance
evaluation which becomes more a product of policy effectiveness.

The other ﬁnding that deﬁed expectations was the negative relationship between
knowing someone who died of HIV/AIDS and policy assessment. Youde (2009)
reported a positive relationship between these variables, which he assessed to indicate
citizen willingness to move beyond AIDS stigma. Contrasting this ﬁnding, and more
intuitively, the above analysis indicates that citizens who know someone who died of the

disease look less favorably on their government’s response to the disease.

159

 

Although two relationships were found that contrasted with hypothesized
expectations, the model does tell a compelling story. First, citizen interaction with the
state is critical to determining their satisfaction with policy performance. This
satisfaction is rooted in a person’s interaction with health care facilities, but it extends
beyond health care to include state institutions writ large. One can also see that
individuals who believe they are being supplied with more democratic regimes have a
more favorable view of HIV/AIDS policy. This clariﬁes ﬁndings ﬂom earlier cross-
national analysis which revealed that the relationship between regime type and AIDS
policy was more muddled. At the individual level, individuals link effective HIV/AIDS
policy as a speciﬁc example of service delivery, which they evaluate along similar lines
as they evaluate the supply of democracy more broadly.

In terms of political cleavages, we see that political party membership rather than
ethnic afﬁliation is the key variable, contrasting Lieberman’s ﬁndings related to ethnic
barriers preventing effective policy making. The models suggest that Aﬁican citizens
view AIDS policy similarly to other policy issues and are as willing to assess their
governments on this issue as they are others. On whether citizens evaluate HIV/AIDS
similar to other issues, this analysis agrees with Youde (2009).

In terms of the most important factors that inﬂuence citizen satisfaction with
HIV/AIDS policy performance, the most important factor is building an effective health
care system. Citizens who are satisﬁed with health care are more satisﬁed with
HIV/AIDS policy efforts. Strong health care institutions may be better able to adapt to
emerging crises like HIV /AIDS than weak health care institutions. While health care

institutions speciﬁcally are the most important individual factor, a stronger state

160

 

 

generally is also important. An effective crisis response will require more than
additional clinics—stafﬁng, equipping, and long term operational planning are all also
required. Greater state institutional trust and decreased experienced corruption lead to
greater satisfaction with policy performance. Citizens who believe they live under a
more democratic regime are also more satisﬁed with policy performance. Finally,
citizens are more satisﬁed when they see live in countries that have effectively managed
the disease, both in terms of reduced burden of disease and in terms of ARV provision.
This suggests that citizen policy performance evaluation is largely reﬂective of policy
effectiveness. Rather than focus on ethnic barriers, citizens are able to assess the
effectiveness of policy performance and evaluate policy effectiveness as a reﬂection of
its outputs. This is based on the results which suggest that country level policy efforts
which lead to lower HIV prevalence and increased ARV coverage are among the most
important predictors of high levels of citizen satisfaction.

Micro level analysis conﬁrms many of the ﬁndings ﬂom cross-national and sub-
national inquiry. The capacity of state institutions has greater explanatory power than
regime characteristics. This is true both at the national level as well as the individual
level of analysis. Sub-national distribution of ARVs as political patronage appears to
raise expectations among citizens as those who support the ruling party are less satisﬁed
with HIV/AIDS interventions than others. Another key ﬁnding is the importance of
acquiring knowledge and experience leading to increased citizen demand for public
attentive policy: those who are more educated, wealthier, and those who live in urban
areas are more likely to be dissatisﬁed with their governments. This trend is likely due

to their increased expectations once they engage with their governments in a more

161

 

connected manner.

In summary, when asked about HIV/AIDS policy, citizens consider this issue as a
speciﬁc component of health service delivery. Their evaluation is based largely on their
interaction with state institutions. Where they evaluate these institutions as trustworthy
and effective, both in terms of their interactions and in terms of the outputs provided,
they are more likely to evaluate policy performance favorably. How citizens evaluate
HIV/AIDS policy performance is especially important because individuals consider this
and health more generally using the same type of thinking as they evaluate democracy.
Where they see more effective healthcare, they are more satisﬁed with democracy with
their interaction with institutions linking HIV/AIDS and democracy policy evaluation.
Conclusion

What do these ﬁndings reveal about the relationship between citizens and Aﬁican
governments as it relates to HIV/AIDS policy? The most important linkage appears to
be between citizens and the state—both in terms of health care service quality
speciﬁcally and trust in state institutions more broadly. Citizens are satisﬁed with
HIV/AIDS policy performance where the state is strong, effective, and citizens trust that
state actors as trustworthy. Strong, responsive institutions paired with democratic
regimes lead to more robust policy responses. Botswana provides a concrete example
where these important pillars have led to the most effective HIV/AIDS response in
Africa. Under these circumstances, when HIV prevalence is lowered through effective
prevention programs and those who contract the disease can access treatment, citizens
are more satisﬁed, largely due to the effectiveness of the policy outputs themselves.

Effective policy provision provides governments with satisﬁed citizens, and citizens

162

 

 

with the resources they need. The story of Trisca Mkanda provides one individual level
example of effective treatment can lead to positive extemalities for the individual, the
government, and society as a whole.

ARVs are a speciﬁc type of good. ARV resources offer a highly valuable commodity
capable of lengthening the lives of those infected by the disease. Especially in areas

where citizens do not interact with the state on a regular basis, the distribution of these

 

resources offers governments a unique opportunity to provide for citizens in a new and
meaningful way. Such provision could offer states the chance to bridge the gap between
individuals and states—leading to increased political consolidation. Fifteen years ago,
ARV provision in Africa seemed an impossible goal. The next step is to build on the
progress already made, maximize efﬁciency of every dollar spent, building and linking

institutions between HIV/AIDS and broader health systems to avoid siloed responses,

 

and moving from a short term crisis response to an established, forward looking long
term model. In order to achieve these goals, the ﬁrst and most meaningful goal must be

to build more responsive, higher capacity states across the continent of Africa.

163

Works Cited
Aﬂobarometer Brieﬁng Paper No. 12. NA. “Public Opinion and HIV/AIDS: Facing Up
to the Future.” httn://afrobarometer.org/peyers/AﬂobriefNo12.pdf

Anderson, Christopher and Christine Guillory. 1997. “Political Institutions and
Satisfaction with Democracy: A Cross National Analysis of Consensus and

Majoritarian Systems.” The American Political Science Review, Vol. 91, No. 1: 66-
81.

 

Boone, Catherine and Jake Batsell. 2001. “Politics and AIDS in Africa: Research
Agendas in Political Science and International Relations,” Africa Today.

Boone, Catherine. 2003. Political T opographies of the Aﬁ‘ican State: Territorial
Authority and Institutional Choice. New York: Cambridge University Press.

Bratton, Michael and Nicholas van de Walle. 1997. Democratic Experiments in Africa:
Regime Transitions in Comparative Perspective. Cambridge, MA: Cambridge
University Press.

Bratton, M., E. Gyimah-Boadi, and R. Mattes. 2005. Public Opinion, Democracy and
Market Reform in Africa. Cambridge: Cambridge University Press.

Bratton, Michael and Eric Chang. 2006. “State Building and Democratization in Sub-

Saharan Aﬂica: Forwards, Backwards, or Together?” Comparative Political Studies,
39, 9: 1059-1083.

Bratton, Michael. 2007. “Are You Being Served? Popular Satisfaction with Health and
Education Services in Africa.” In Democratic Deﬁcits: Addressing Challenges to
Sustainability and Consolidation Around the World. Gary Bland and Cynthia Arnson,
editors. Woodrow Wilson International Center for Scholars: Washington DC. p. 37-
66.

Diamond, Larry. 1992. “Economic Development and Democracy Reconsidered,”
American Behavioral Scientist 35, 4/5: 450-499.

Diamond, Larry. 2002. “Thinking about Hybrid Regimes,” Journal of Democracy. Vol.
13 No. 2.

Diamond, Larry and Leonardo Morlino. 2005. Assessing the Quality of Democracy.
Baltimore: John Hopkins University Press.

Deacon, Robert. 2003. “Dictatorship, Democracy, and the Provision of Public Goods.”
Departmental Working Paper, University of California at Santa Barbara
http://repositories.cdlib.org/ucsbecon/dwp/20-03.

Deutsch, Karl. 1961. “Social Mobilization and Political Development,” American
Political Science Review 55, 3: 493-514.

164

Englebert, Pierre. 2000. State Legitimacy and Development in Africa. Boulder: Lynne
Rienner.

Franco, Alvaro, Carlos Alvarez-Dardet, and Maria Teresa Ruiz. 2004. “Politics of
Democracy on Health: An Ecological Study. British Medical Journal. Vol. 329. p. l 8-
25.

Gerring, John, Strom Thacker, and Rodrigo Alfaro. 2005. Democracy and Human
Development. Presented at the American Political Science Association Meetings.
Washington, DC.

Govindaraj, Ramesh and Ravindra Rannan-Eliya. 1994. Democracy, Communism, and
Health Status: A Cross National Study. Data for Decision Making Project, Harvard
School of Public Health.

Gyimah-Boadi, E. 2004. Democratic Reform in Africa: The Quality of Progress.
Boulder: Lynne Rienner.

Halperin, Morton, Joseph Siegle, and Michael Weinstein. 2005. The Democracy
Advantage: How Democracies Promote Prosperity and Peace. New York:
Routledge.

Herbst, Jeffrey. 2000. State and Power in Aﬁica: Comparative Lessons in Authority and
Control. Princeton: Princeton University Press.

Hyden, Goran. 2006. African Politics in Comparative Perspective. New York:
Cambridge University Press.

Jackson, Robert and carl Rosberg. “Why Africa’s Weak States Persist,” World Politics.
V0135 1-24.

Jones, Peris. 2005. “On a Never Ending Waiting List: Toward Equitable Access to Anti-

Retroviral Treatment? Experiences ﬂom Zambia.” Health and Human Rights. Vol 8.
No.2.

Kaufmann, Daniel, Aart Kraay and Massimo Mastruzzi. 2009. "Governance Matters
VIII: Governance Indicators for 1996-2008". World Bank Policy Research June.
Policy Working Paper 4978.

Lasswell, Harold. (1936) Politics: Who gets What, When, and How. McGraw-Hill, New
York.

Levitsky, Steven and Lucan Way. 2002. “The Rise of Competitive Authoritarianism,”
Journal of Democracy. Vol. 13. No. 2.

165

 

Levy, Brian and Sahr Kpundeh. 2004. Building State Capacity in Africa. Washington,
DC: World Bank Publications.

Lieberman, Evan. 2009. Boundaries of Contagion: How Ethnic Politics Have Shaped
Government Responses to AIDS. Princeton: Princeton University Press.

Lipset, Seymour Martin. 1959. “Some Social Requisites of Democracy: Economic
Development and Political Legitimacy,” American Political Science Review, 53, 1:
69-105.

Nordhaus, William, Alberto Alesina, and Charles L. Schultze. (1989). “Alternative
Approaches to the Political Business Cycle,” Brookings Papers on Economic
Activity,
Vol. 1989, No.2, pp. 1-68. The Brookings Institution.

Patterson, Amy. 2005. The Aﬁican State and the AIDS Crisis. Burlington:
Ashgate.

Patterson, Amy. 2006. The Politics of AIDS in Africa. Boulder: Lynne Rienner
Publishers.

Posner, Daniel. 2004. “Measuring Ethnic Fractionalization in Africa,” American Journal
Of Political Science. Vol. 48, Issue 4. p. 849-863.

Posner, Daniel. 2005. Institutions and Ethnic Politics in Africa. Cambridge: Cambridge
University Press.

Roniger, Luis. 2004. “Political Clientelism, Democracy, and Market Economy.”
Comparative Politics, Vol. 36, No. 3 (Apr., 2004), pp. 353-375.

Ross, Michael. 2006.“Is Democracy Good for the Poor?” American Journal of Political
Science, 50:4, October.

Schedler, Andrea. 2002. “Elections Without Democracy: The Menu of Manipulation,”
Journal of Democracy. Vol. 13 No. 2.

Sen, Amartya. 1981. Poverty and Famines: An Essay on Entitlement and Deprivation:
New York: Oxford University Press.

Sen, Amartya. 1999. Development as Freedom. New York: Alfred A. Knopf.

Stasavage, David. 2005. “Democracy and Education Spending in Africa,” American
Journal of Political Science. Vol. 49. No. 2. 343-358.

Strand, Per, Mary Kinney, and Robert Mattes. 2008. “Politics and Political Outcomes on
Children Affected by HIV/AIDS in Africa.” Institute of Development Studies

166

 

 

Bulletin. Vol. 39 No. 5.

Taylor, Elizabeth and Clare Dickenson. “The Link Between Corruption and
HIV/AIDS.”

Global Corruption Report 2006. Transparency International: Washington, DC.

Tsai, Ming-Chang. 2006. “Does Political Democracy Enhance Human Development in
Developing Countries?” American Journal of Economics and Sociology. Vol 65. No
2. P. 233-268.

UNAIDS. 2009. “Towards Universal Access: Scaling up Priority HIV/AIDS
Interventions in the Health Sector.” Progress Report 2009. World Health
Organization, UNAIDS, UNICEF.

Van de Walle, Nicholas. 2001. African Economics and the Politics of Permanent Crisis,
1979-1999. Cambridge: Cambridge University Press.

Van de Walle, Nicholas. 2002. “Africa’s Range of Regimes,” Journal of Democracy.
Vol. 13 No. 2.

Villalon, Leonardo and Phillip Huxtable. 1997. “The Aﬁican State at a Critical Juncture:
Between Disintegration and Reconﬁguration. London: Lynne Rienner Publishing.

Vollmer, Sebastian and Maria Ziegler. 2009. “Political Institutions and Human
Development.” The World Bank, Policy Research Working Paper 4818.

Wantchekon, Leonard. 2003. “Clientelism and Voting Behavior: Evidence ﬂom a Field
Experiment in Benin,” World Politics Vol 55. No. 3.

Whiteside, Alan, Robert Mattes, and Ryan Manning. 2002. “Examining HIV/AIDS in
Southern Aﬂica Through the Eyes of Ordinary Southern Africans.” Afrobarometer
Working Paper No.12. litp://afrobarometer.org/papers/AfropaperNo2l .pdf

Youde, Jeremy. 2009. 'Govemment AIDS Policies and Public Opinion in Africa',
Politikon, 36: 2, 219 —235.

167

 

 

The Politics of ARVS in Africa: Concluding Remarks

The concluding section of this paper seeks to meet several goals. First, I aim to
brieﬂy summarize the overall project and its results. In doing so I discuss the relevance
of results both theoretically and substantively to contextualize where this project ﬁts into
the overall literature. Finally, I will offer some comments on the project’s limitations
and potential future direction.

Project Summary
When I began thinking about this project my goals were to answer several
substantive questions:
1. What explained cross national variance in anti-retroviral drug provision?
2. What determined distribution of ARV programs at the sub-national level?
3. What explained citizen satisfaction with HIV/AIDS policy performance, and to

what degree were ARVS part of this story?

This research agenda had been discussed and advanced within the health economics
literature, and some effort had been made within the ﬁeld of political science. The only
large scale empirical work that I was aware of was Evan Lieberman’s work. Very little
work had focused on ARVS speciﬁcally as most work was focused on the HIV/AIDS
efforts more generally. Much of this work had focused on the effects of the disease
rather than explaining policy response variance. Although a great deal remains
unexplained, I believe this project has advanced an empirically interesting and
substantively important line of research. This effort should be useful not only for those
interested in the politics of ARV distribution, but also for those interested in public

service delivery in developing countries most generally.

168

In answering the ﬁrst research question, I found that state capacity more so than
regime type inﬂuences an Aﬂican country’s ability to provide ARVs. This ﬁnding
suggests that stronger,lhigher capacity states yield a more effective HIV/AIDS policy
response. This ﬁnding ﬁts in with Pierre Englebert’s results which suggest that
enhanced state capacity provides a larger menu of policy options available to
policymakers in developing states (2000). Substantively, the state capacity ﬁnding
provides evidence that building state capacity offers a long term goal which would help
to combat not only HIV/AIDS but the other governance challenges facing the
developing world. As we know from recent historical events, state building is not easy.
However, as states are able to take on greater responsibilities, the costs for donors begin
to decrease as domestic governments are able to shoulder the burden of governance.

As an anecdotal account of the research project, I should note that the ﬁnding on the
importance of state capacity grew out of what was expected to be a control variable.
When governmental effectiveness proved to not only be statistically important but the
driving force in my initial cross-national modeling, it was clear that considering state
capacity would be an important part of the larger project moving forward. This variable
turned out to be consistently important at both the cross-national and individual levels of
analysis. In my opinion, this is the most important substantive result to come out of this
dissertation.

The second article sought to explain how resources are allocated within countries,
given the fact that no African country has been able to provide ARVS to every citizen
that requires these medications. In order to answer this question, sub-national units in

South Africa and Nigeria were considered. The results were much stronger than in the

169

cross-national section of the paper, and suggested that ARV programs were allocated to
urban areas and areas with higher HIV prevalence rates. More interestingly, regions that
were represented by members of the ruling party were more likely to have ARV
programs than regions which were represented by minority parties. This political
representation ﬁnding suggests that patronage networks are important in ARV
distribution, at least at the elite level when resource allocation decisions are made. This
ﬁnding ﬁt with Goran Hyden’s characterization of the policy process in Africa as more
focused on the distribution of resources along patronage lines rather than to solve policy
problems in pursuit of development (2006). Substantively, the existence of elite level
patronage decisions inﬂuencing ARV distribution provides evidence that international
donor institutions would beneﬁt if they consider political representation in their
monitoring and evaluation practices. Current practices pay similar attention to gender
and income, but if ARVs are distributed through patronage networks, it is important to
consider this phenomenon when considering equitable treatment access. If developing
equitable resource distribution is important to these organizations, political access to
patronage networks appears salient.

The third paper focused on public opinion and sought to explain individual
satisfaction with HIV/AIDS policy performance. The goal here was to see to what
degree the availability of ARVS inﬂuenced individual perceptions. Were citizens more
satisﬁed when they saw ARV programs? Results indicated that ARV coverage and HIV
prevalence rates, measures I argue measure the effectiveness of respective policy
responses, are important predictors of HIV/AIDS policy performance satisfaction. This

bolsters the case for ARVS as it suggests the tangible political rewards available for

170

leaders willing to invest in these efforts. Further evidence suggested that state
characteristics like trust in institutions are important predictors of policy satisfaction.
The relationship between institutional trust and policy performance satisfaction ﬁts with
the cross-national relationship between state capacity and ARV coverage. Together
these ﬁndings provide strong evidence for the importance of strong, effective state
institutions at both the micro and macro levels of analysis. When such institutions exist
and function effectively, policies yield more effective outputs which leads to greater

citizen satisfaction. In other words, effective institutions provide the highways that

 

make effective policy outputs possible. Greater citizen satisfaction is especially
important as results suggest that citizens evaluate HIV/AIDS policy along the same lines
as they evaluate the supply of democracy: when they are satisﬁed with HIV/AIDS and
health policy, they are more likely to be satisﬁed with the effectiveness of democracy.
Another interesting ﬁnding that contradicted expectations was the negative
relationship between citizens who said they supported the ruling party and HIV/AIDS
policy performance satisfaction. Originally I considered that this ﬁnding contradicted
with the sub-national analysis which revealed the existence of patronage networks.
Rather, I would argue that it suggests that such patronage related to ARVs exists at the
elite level where allocation decisions are made. The lack of mass level patronage
suggests that policy performance trumps political loyalty, an important ﬁnding in its

own right. Further investigation into political patronage at the local level would be a

 

useful future research direction, though such an effort would require ﬁeld level
observation. Answering exactly how individuals make their way ﬂom a positive HIV

test to accessing ARVS is not a question I am prepared to answer at this time. Those

171

types of individual level decision processes would require ﬁeldwork to investigate
whether mass level patronage or bribery are part of the story. Anecdotal evidence from
Afrobarometer Round 3 does suggest that more than twenty ﬁve percent of people are
asked to pay bribes when they interact with health institutions (Bratton, 2007). Based on
this dissertation research, however, I can only suggest that elite level patronage exists
and merely speculate on mass level patronage relations.

Researching citizen policy performance satisfaction also suggests people are more
critical of governmental efforts when policies are most salient for them. The modal
response was that governments were doing pretty well combating HIV/AIDS, but
greater support or dissatisfaction was evident in countries where HIV/AIDS was of
greatest salience due to higher HIV prevalence rates. In low prevalence countries,
citizens who are less affected by the disease have no reason for dissatisfaction. This
partially explains why support for HIV/AIDS policies exceeds support for other policy
efforts for Africans. As salience increases, people develop stronger opinions, and the
general trend is for greater knowledge to lead to increasing dissatisfaction. I would
argue that this helps to explain the negative relationship between educational attainment,
urbanization, and even political support for the ruling party. As people gain education,
live in cities, and support the ruling party, they sharpen their opinions about policy
performance, and they generally are less satisﬁed as their expectations rise. Now that
results have been summarized I will discuss some of the limitations of the project and
some future directions for research.

Limitations and Future Research

One of the issues I struggled with in working through this project was where to place

172

 

 

political leadership in the story. It is strikingly absent from the statistical analysis. We
know that individual leaders and their preferences matter to a larger degree in Aﬁica
than in other contexts. However, I would argue that individual leaders can only enhance
or prevent policy responses so much and for so long. South Aﬂica’s history provides a
good example. Mbeki’s unwillingness to engage HIV/AIDS issues might have
prevented an effective policy response in the short term; however, strong independent
institutions provided a venue through which to overturn the regime’s preferences as civil
society utilized the court system. Once Mbeki left ofﬁce, South Aﬁica has launched a
more robust HIV/AIDS policy than previously existed. This is as much in spite of
President Zuma’s leadership as it is because of it which provides an example of
institutions being more important than leadership. South Aﬂica provides a case where
political leadership’s unwillingness to engage HIV/AIDS was trumped by strong,
independent institutions and civil society engagement, as exempliﬁed by the Treatment
Action Campaign.

In other contexts we have seen examples where effective leadership can make a
difference, but even Uganda, which has been noted for its effective HIV/AIDS response
under Museveni’s leadership, has only been able to achieve so much given its
institutional weakness. 1 would argue that building strong institutions provides
leadership a wider array of choices and broader engagement opportunities as Englebert
argues. Botswana provides the quintessential example where the institutional capability
led to increasingly effective policy. Leadership matters, but I argue the institutions are
more important, and enhance a respective leader’s toolbox. Finding new and innovative

methods for measuring how leadership and institutions interact would be a useful

173

 

direction for future research.

Data limitations were a constant consideration working through the dissertation. In a
perfect world, I would have ARV availability data at the country, sub-national, and
individual levels measured using the same unit which would allow for multi-level
modeling to be developed. Although signiﬁcant variance exists at all of the levels of
analysis observed, it would be interesting to know where the greatest amount of variance
lies. My hunch would be that sub-national variance exceeds cross national in terms of
ARV availability as elites can access ARVS on international markets. For example, one
would expect an HIV-infected wealthy citizen to be as able to purchase ARVS on the
open market, whereas impoverished individuals lack these options.

In order to signiﬁcantly enhance my dissertation, two immediate directions jump to
mind. The ﬁrst would be to spend several months observing ARV programs in the ﬁeld
to understand the relationship and decision making processes at the local and national
levels. I have made the case that African countries have signiﬁcant power in these
decision making dynamics, but situating how international organizations and non-
governmental organizations function on the ground cannot be veriﬁed without traveling
to these countries. Field observation would allow for interviewing ARV personnel,
country level HIV/AIDS committee members, and non-governmental workers.
Observation would be especially useful to understand the linkages between prevention
and treatment as prevention campaign efforts must take on the culturally sensitive taboo
issues like sex, condoms, and the role of women in society. Observation would also
allow investigation of how political elites view ARV medications and how individuals

access these programs. Such work would provide greater context and would allow for

174

 

more individual level examples. A case study chapter based on observing in several
clinics would greatly enhance this project, a limitation that I readily admit.

The second course of action would be to spend some time building up some new
methodological tools. The ﬁrst cross-national article could be improved with some non-
linear regression techniques. I am not completely convinced that regime type is as
insigniﬁcant at the country level as the paper suggests. The individual level model
suggests that individuals who believe they live under more democratic regimes are more
satisﬁed with HIV/AIDS policy performance, though this relationship is not all that
inﬂuential. If I had six more months to pursue this project and the funding available, I
believe the ﬁeld based course of action would be more beneﬁcial. Critiquing my own
work, I believe the limited context that can be drawn ﬂom books and database searches
is a greater weakness of this project than the lack of empirical speciﬁcity.

There are other directions future research could pursue as well. It would be valuable
to contextualize how HIV/AIDS efforts have beneﬁtted or hindered other health care
policies. One would expect that political learning must be occurring both across
countries and in generating knowledge of value to for other health care interventions.
The scope of the HIV/AIDS response is truly without precedent, and trying to
understand how these efforts have helped or hindered other health and development
programs would be useful. For instance, efforts to combat malaria and drug resistant
tuberculosis have grown out of initial AIDS efforts. Seeking whether AIDS efforts have
limited the scope of other development programs would also be useful.

Another assumption that would beneﬁt ﬂom further research is providing greater

context regarding the international pharmaceutical companies and the role of the private

175

 

sector in anti-AIDS policy. Anecdotal evidence suggests that the private sector is
engaged to a much larger degree in some of these countries than in others. For instance,
Anglo-American Gold in South Aﬂica has one of the most expansive ARV treatment
programs on the continent which provides medications not only to mine workers and
their families. These efforts also provide testing, counseling, and treatment services for
sex workers who reside in areas around the mine areas. These types of efforts are
important; however, they are not the norm across the continent. In the sub-national
chapter, I did not ﬁnd evidence of similar efforts by oil companies operating in the Niger
Delta of Nigeria. Greater private sector engagement would beneﬁt large segments of
Society, the companies themselves, and would likely bear lessons that could enhance
public sector efﬁciency. Developing research that could take into account why these
efforts have occurred so irregularly across the continent would be useful.
Concluding Remarks

We live in a globalized society where HIV/AIDS in Africa is important for
international markets, security, and governance. This issue has received unprecedented
attention by the world community and is beginning to receive similar attention by social
scientists. Just in the course of researching this topic over the last couple of years, many
more books and better research on the social science surrounding the disease has begun
to emerge. Fighting HIV/AIDS will continue to require new, innovative, and increasing
interdisciplinary efforts. The task at hand is transitioning ﬂom crisis response to a long
term, entrenched, and forward looking plan to attack HIV/AIDS. This must occur both
on the ground in Africa and in our research agendas which can potentially yield

important actionable lessons. It is only through these efforts that we can expect

176

 

improved policy performance and less people to suffer from this awful, dreaded disease.
HIV/AIDS is likely to remain one of the leading development challenges for Aﬂica for
the foreseeable future. Every step that enhances our knowledge of the disease provides
one more tool in the scientiﬁc toolbox which will enhance the ability to combat this
disease. The people of Africa ﬁghting AIDS require such an effort, and as scientists we
should continue to strive towards advancing an agenda surrounding the politics of the

AIDS response.

177

Works Cited

Bratton, Michael. 2007. “Are You Being Served? Popular Satisfaction with Health and
Education Services in Aﬁica.” In Democratic Deﬁcits: Addressing Challenges to
Sustainability and Consolidation Around the World. Gary Bland and Cynthia Arnson,

editors. Woodrow Wilson International Center for Scholars: Washington DC. p. 37-
66.

Englebert, Pierre. 2000. State Legitimacy and Development in Aﬂica. Boulder: Lynne
Rienner.

Hyden, Goran. 2006. Aﬁican Politics in Comparative Perspective. New York:
Cambridge University Press.

178

 

  
 
          

 

 

             

VERSITY LI

nmmwwﬁllll“

063 5852

mllllllllllllllllllm
3 1293 03