ACCOUNTABILITY THROUGH PROGRAMMATIC GOODS DISTRIBUTION: THE GHANAIAN NATIONAL HEALTH INSURANCE SCHEME By Stephen Edward Anderson A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of Political Science—Doctor of Philosophy 2022 ABSTRACT ACCOUNTABILITY THROUGH PROGRAMMATIC GOODS DISTRIBUTION: THE GHANAIAN NATIONAL HEALTH INSURANCE SCHEME By Stephen Edward Anderson One way in which political leaders gain support in democratic systems is by distributing goods to citizens. Theories of accountability suggest that when political authorities provide these goods to citizens, they will be rewarded. Yet, incumbents have at their disposal multiple mechanisms through which they can distribute goods to citizens. Generally, goods are either distributed through programmatic or non-programmatic means. This dissertation evaluates how the mechanism through which goods are distributed influences the ways in which citizens evaluate goods performance and political leaders. I address this question by focusing on the distribution of healthcare in the West African country of Ghana. I propose a theory to explain the process by which individuals reward or punish incumbent leaders for healthcare provision distributed via a national health insurance scheme (programmatic distribution) and targeted monetary transfers (non-programmatic distribution). The central insight proposed by the theory is that the characteristics of healthcare distribution via a national insurance scheme lead individuals to evaluate incumbent leaders based on the quality of goods they receive, while the nature of healthcare distribution through targeted transfers leads individuals to evaluate incumbents based not on the quality of goods they acquire, but on their receipt of a transfer. Interview and survey research accord with these propositions. I find that individuals are more likely to sanction incumbents for poor performance when low-quality healthcare goods are distributed through a national insurance scheme. When these same low-quality goods are acquired via monetary handouts, individuals are less likely to sanction incumbents for the quality of said goods. In the latter case, poor performance is not associated with the quality of healthcare goods acquired by citizens, but whether or not the necessary resources for the acquisition of these goods were dispensed by incumbents. I contribute to the literature on goods distribution and accountability and provide novel theory and evidence on the impact of programmatic distribution and incumbent support. Copyright by STEPHEN EDWARD ANDERSON 2022 ACKNOWLEDGEMENTS While conducting research and drafting this dissertation, many individuals have provided me with invaluable assistance, support, friendship, and love. During what often seemed an insurmountable task, I would not have been able to complete this journey without all of you. First and foremost, I owe the largest debt to Elisabeth for her unwavering support and love. You put a number of your goals and aspirations on hold so I could achieve mine. You took the lead in supporting our growing family and kept me sane during my graduate studies. I will never forget what you have done for me and our family. I cannot wait to see what the future holds for us. During the course of my research, we welcomed our little Elodie to the world. Elodie, your love and light brought levity when I needed it most and you make coming home every day all the more sweet. To my future daughter, I cannot wait to meet you. Unbeknownst to you, you have provided me with the final motivation to push and complete this project. I will see you soon. Second, I thank my parents and grandparents for instilling in me a desire to learn and encouraging me in my pursuits. I would not be who I am today without your support. I also thank Patty and Luke for the support they provided my family and I during my graduate studies. Having you in my life during this period has truly been a blessing. To my brothers, Eric and Ted, thank you for providing me some respite, even if it was only to play video games or share a meme. Many of my colleagues have helped me over the course of my studies. I thank Jonathan King, Tara Iseneker, Caleb Lucas, Alex Mardon, N. Charles Smith, Crystal Williams, v and Chrissy Scheller for their friendship and assistance. I owe a special thanks to my two closest friends Nicolas Bichay and Marcella McCalmont. Nico, you have always been there to let me know that I am using an incorrect statistical model, and your humor made so many long days bearable. Marcie, you often provided a willing ear when I needed to complain, and your friendship has helped me weather many a storm. I also owe a debt of gratitude to my dissertation committee. Christian Houle, Erica Frantz, and Michael Wahman have been generous with their time and assistance. I thank you for your helpful comments and guidance. A special thanks goes to my committee chair Jeffrey Conroy-Krutz. It has been a long process, and you have been the perfect guide. Thank you for providing inspiration and insight, reading dozens of drafts, dealing with my use of passive voice, and always providing a kind ear. Beyond my committee, I would like to thank Michael Bratton for his wisdom and expertise, as well as his support for my field research. I also thank Eric Chang for his kindness and general encouragement during my first year of graduate school, you are truly a credit to the Department of Political Science at Michigan State University. Many assisted me during my fieldwork in Accra, Ghana. I thank Lauren MacLean for providing contacts and guidance upon receiving a random email from me. At the University of Ghana, I thank Hassan Wahab for his kindness and assistance throughout my stay in Accra. A special thanks goes to Muhammad Nuhu Adam. Without your companionship and research assistance, this project would not be complete. I have found a lifelong friend in you, and I wish you all the best in your own academic journey. Lastly, I thank all the Ghanaians that participated in this study, without you none of this would be possible. vi TABLE OF CONTENTS LIST OF TABLES .....................................................................................................................................................ix LIST OF FIGURES .................................................................................................................................................... x INTRODUCTION ......................................................................................................................................................1 CHAPTER 1: HOW THE DISTRIBUTION OF GOODS INFLUENCES ACCOUNTABILITY IN SUB- SAHARAN AFRICA ............................................................................................................................................... 10 1.1 STATE POLICY AND POLITICAL BEHAVIOR IN SUB-SAHARAN AFRICA ................ 12 1.2 GOODS AND ELECTORAL ACCOUNTABILITY IN SUB-SAHARAN AFRICA.............. 15 1.3 HEALTHCARE POLICY IN SUB-SAHARAN AFRICA........................................................... 19 1.4 THE PROGRAMATIC DISTRIBUTION OF GOODS THROUGH A NATIONAL INSURANCE SCHEME .......................................................................................................................... 24 1.5 NATIONAL HEALTH INSURANCE SCHEMES: PERFORMANCE, AND ACCOUNTABILITY ................................................................................................................................ 31 1.6 NON-PROGRAMMATIC DISTRIBUTION OF HEALTHCARE: PERFORMANCE, AND ACCOUNTABILITY ................................................................................................................................ 36 CHAPTER 2: A NOVEL FORM OF GOVERNMENT HEALTHCARE PROVISION - THE GHANAIAN NATIONAL HEALTH INSURANCE SCHEME ...................................................................... 43 2.1 THE EXPANSION OF SOCIAL POLICY IN THE DEVELOPED WORLD AND SUB- SAHARAN AFRICA................................................................................................................................. 44 2.2 THE IMPLEMENTATION OF THE GHANAIAN NHIS ........................................................ 52 2.3 THE DESIGN OF THE GHANAIAN NHIS ................................................................................ 54 2.4 NHIS ENROLLMENT ..................................................................................................................... 58 2.5 EFFECTIVENESS OF THE NHIS ................................................................................................ 60 2.6 THE NHIS IN CONTEXT ............................................................................................................... 63 2.7 THE NHIS AS A FORM OF PROGRAMMATIC DISTRIBUTION....................................... 65 2.8 CONCLUSION ................................................................................................................................... 69 CHAPTER 3: CITIZEN PERCEPTIONS OF THE NHIS .............................................................................. 71 3.1 RESEARCH DESIGN ....................................................................................................................... 73 3.2 DETERMINANTS OF ENROLLMENT ...................................................................................... 77 3.3 THE NHIS – PERCEPTIONS AND EXPERIENCES ............................................................... 86 CHAPTER 4: THE POLITICAL IMPACTS OF PROGRAMMATIC AND NON-PROGRAMMATIC DISTRIBUTION: HEALTHCARE ALLOCATION AND ACCOUNTABILITY..................................... 101 4.1 DISTRIBUTION AND PERFORMANCE-BASED ACCOUNTABILITY.......................... 102 4.2 METHODOLOGY AND EXPERIMENTAL VIGNETTE ...................................................... 105 4.3 RESULTS ......................................................................................................................................... 109 4.4 DISCUSSION .................................................................................................................................. 117 vii CONCLUSION ...................................................................................................................................................... 119 APPENDIX ............................................................................................................................................................ 124 BIBLIOGRAPHY ................................................................................................................................................. 137 viii LIST OF TABLES TABLE 2.1 NHIS INFORMAL SECTOR CLASSIFICATION...................................................................... 56 TABLE 2.2 NHIS ACTIVE ENROLLMENT: 2005-2009........................................................................... 59 TABLE 2.3 NHIS ACTIVE ENROLLMENT: 2010-2014........................................................................... 59 TABLE 3.1 DEMOGRAPHIC PROFILE OF STUDY PARTICIPANTS .................................................... 75 TABLE 3.2 LIVED POVERTY INDEX.............................................................................................................. 76 TABLE 3.3 NHIS ACTIVE ENROLLEES......................................................................................................... 77 TABLE 3.4 DETERMINANTS OF NHIS ENROLLMENT .......................................................................... 80 TABLE 3.5 DETERMINANTS OF HEALTHCARE UTILIZATION ......................................................... 85 TABLE 4.1 EXPERIMENTAL VIGNETTES ................................................................................................ 107 TABLE 4.2 VIGNETTE RESULTS: OPPOSITION MP ............................................................................. 109 TABLE 4.3 TREATMENT EFFECTS: ORDERED LOGIT MODELS .................................................... 111 TABLE 4.4 VIGNETTE RESULTS: PRESIDENTIAL ELECTION ......................................................... 113 ix LIST OF FIGURES FIGURE 1.1 EXPECTATIONS AND THE QUALITY OF GOODS PRODUCED THROUGH A NATIONAL HEALTH INSURANCE SCHEME .............................................................................................. 34 FIGURE 1.2 EXPECTATIONS AND THE QUALITY OF GOODS: INDIVIDUALLY TARGETED, NON-PROGRAMMATIC HEALTHCARE DISTRIBUTION ....................................................................... 41 FIGURE 2.1 HEALTHCARE UTILIZATION IN GHANA 2005-2013 (MILLIONS OF VISITS) ..... 61 x INTRODUCTION One of the ways in which political leaders elicit support is through the distribution of public goods to citizens. Theories of democratic accountability suggest that when political authorities adequately provide public goods to citizens, they will be rewarded at the ballot box. When authorities fail to provide these goods, they will be sanctioned accordingly. Political leaders have at their disposal a variety of ways in which to distribute goods to citizens. Broadly, modes of goods distribution are either programmatic or non- programmatic in nature (Stokes et al. 2013). Goods distributed through programmatic means are available only to those within a defined category according to formal rules. Moreover, the receipt of goods by eligible recipients is not conditional on political support. An eligible recipient can benefit from programmatic distribution regardless of the party or individual in power. Programmatic goods are often distributed through a specific social program or policy to eligible recipients and include conditional cash transfer programs (Diaz-Cayeros, Estevez, and Magaloni 2009; Zucco 2013), nutritional assistance (Weitz- Shapiro 2014), and educational vouchers (Pop-Eleches and Pop-Eleches 2012). Non- programmatic distribution contravenes these tenets. Goods distributed via non- programmatic means are often targeted to individuals according to informal or non-public criteria, rather than the formal rules dictating the receipt of goods via programmatic distribution. Furthermore, one’s eligibility to acquire non-programmatic goods may be conditional on political support – if support is retracted, access to these goods can be cutoff. One of the more common forms of non-programmatic distribution involves the distribution of goods targeted to individuals in exchange for political support, or clientelism (Stokes 2007). 1 Existing research suggests that voters condition their support for individual politicians in part based on those politicians’ performance in delivering goods via both programmatic and non-programmatic means (Golden and Min 2013, Hicken 2007). Performance-based voting is often said to lead to better governance; when voters condition their support on goods production, political leaders should be motivated to produce more goods. Yet, in many parts of the developing world, public goods are generally underprovided and non-programmatic distribution, such as clientelism, often the norm. When it comes to sub-Saharan Africa in particular, a vast literature describes the prevalence and effectiveness of targeted distribution in exchange for political support (van de Walle 2003; Wantchekon 2003; Kitschelt and Wilkinson 2007; Bratton 2008). This literature often describes how under these distributive conditions, political leaders are not evaluated based on the production of collective goods, but rather on their ability to target individual goods to citizens. Thus, leaders who fail to produce more universal goods often retain office. While this robust literature has provided important insights into how African citizens respond to a specific form of non-programmatic goods distribution, it remains to be seen how citizens evaluate politicians in relation to goods distribution via programmatic means. Though schemes for programmatic goods distribution do exist in many states within sub-Saharan Africa, the extant research is in its infancy. This research program seeks to evaluate a broad question: How does the mechanism through which goods are distributed (programmatic or non-programmatic) influence how citizens evaluate goods performance and political leaders in the developing world? I evaluate this question in the context of Ghana, an established democracy in West Africa with a history of both programmatic and non-programmatic goods distribution. 2 Utilizing original survey data collected during field research in the country, as well as interviews with public officials and citizens, I describe how Ghanaian citizens evaluate the performance of political leaders when goods are distributed through the National Health Insurance Scheme (NHIS), a social policy responsible for the programmatic distribution of healthcare to over 12 million Ghanaians in 2019. In comparison, I also describe how Ghanaian citizens evaluate goods performance when healthcare is acquired via non- programmatic distribution, whereby political leaders provide individual citizens with resources through which to secure healthcare. Programs like the NHIS, in which citizens enroll in a government insurance scheme, pay some form of contribution, and then are given subsidized care when they require medical treatment, are a relatively established phenomenon in the developed world, but in sub-Saharan Africa, these programs exemplify a relatively new form of programmatic goods distribution through which governments can provide healthcare to their citizens. As with any form of goods distribution associated with government provision, national health insurance schemes and the governments and political leaders that administer them are amenable to performance evaluations on behalf of citizens. In terms of goods production, these performance evaluations are often linked to notions of accountability – when incumbent political leaders are perceived of by citizens as not adequately performing in terms of goods production, they are liable to be punished for this poor performance in democratic contexts. While the political ramifications of non- programmatic modes of goods distribution in sub-Saharan Africa has received significant attention in the academic literature focusing on clientelism and ethnic politics, a national health insurance program is novel in the ways in which it provides goods to citizens, the 3 means through which goods are accessed, and how this form of distribution impacts the relationship between citizens and political leaders. Overall, I present multiple aspects of related research that describe the characteristics of programmatic distribution through a national health insurance scheme and the implications of these characteristics for performance-based evaluations, Generalizable arguments are made regarding: 1) How goods distribution through a national health insurance scheme differs from that of non-programmatic distribution in sub-Saharan Africa, and 2) How the design features of national insurance schemes may engender expectations among beneficiaries that form the basis of performance evaluations associated with the goods distributed and how this compares with performance evaluations related to non-programmatic distribution. I then apply these arguments to the Ghanaian context to determine 3) The expectations citizens have regarding the goods produced via the NHIS and how these expectations inform their evaluations of the program and political leaders. In a final analysis, I compare 4) How the programmatic distribution of goods through the NHIS impacts performance-based voting intentions relative to a non- programmatic distributive context. I begin by analyzing what differentiates programmatic goods distribution through a national health insurance scheme from non-programmatic forms of distribution. A national health insurance scheme differs from non-programmatic means of goods allocation in at least three ways. First, the goods derived from a national health insurance scheme are not politically conditional in nature – the distribution of these goods to eligible recipients is not contingent on political support or ties to a specific politician. Second, the goods that a national health insurance scheme proffers are associated with extended time horizons – 4 citizens enroll in a specific program to acquire specific goods at a future date on an as needed basis; these goods are deferred. Lastly, the accrual of goods though a national health insurance scheme often requires direct monetary contributions from citizens to the state in the form of premium payments and registration fees and/or specific taxes. I argue that these three features of programmatic distribution via a national health insurance scheme increase the likelihood of performance-based evaluations concerning goods distributed through such a policy. With these characteristics in mind, I then proceed to describe the process by which these traits engender expectations and beliefs, which inform performance evaluations concerning the distribution of healthcare through a national insurance scheme. The deferred nature of the goods distributed through insurance schemes enable citizens to develop expectations about the type and quality of future goods to be distributed. Expectations are reified by the fact that members of a national insurance scheme are often required to contribute economic resources in the form of payments which are transferred to the state and/or specific taxes. These payments are for a specific service, one in which poor performance is harmful to a citizen’s wellbeing. Whether or not citizens’ expectations are in-line with what a national health insurance scheme provides in terms of coverage is less important than what they believe it provides. If these expectations are not met, citizens perceive poor policy performance. These notions lie in stark contrast to how goods are evaluated when distribution occurs via a non-programmatic pathway, in which healthcare is acquired through a monetary handout from an incumbent politician. The main difference between the programmatic distribution of healthcare through a national insurance scheme and the non- 5 programmatic distribution of healthcare via a cash handout relates to the factor around which citizens develop expectations and evaluate performance. The characteristics of a national health insurance scheme lead citizens to form expectations concerning the actual healthcare goods to be acquired through such a program. Upon the receiving these goods, citizens evaluate them relative to their prior expectations and form performance preferences accordingly. For this type of distribution, the evaluation of goods is the source of incumbent support or opposition. The source of performance evaluations associated with non-programmatic healthcare distribution schemes, whereby political leaders distribute monetary resources to individual citizens for the purpose of acquiring healthcare is whether or not monetary resources were collected from incumbents. In this case, poor performance is not dictated by the quality of healthcare goods a citizen eventually acquires, but whether or not the resources required for the acquisition of these goods were distributed by incumbents. If an incumbent provides monetary resources in this manner, that same leader has performed adequately (inadequately) according to this distributive schema and will be evaluated positively (negatively). Thereafter, I specifically evaluate the Ghanaian NHIS relative to the expectations described above. Utilizing observational and experimental data, I assess the expectations of Ghanaian citizens regarding goods distributed via the NHIS, and how these expectations relate to their experiences utilizing the program and their perceptions of its performance. The citizens included in this study have exceedingly high (and often unrealistic) expectations regarding the goods the NHIS offers, which are rarely met in practice. This notion gives citizens the impression that the NHIS is performing poorly. Despite the fact 6 that the NHIS involves a decentralized structure of responsibility – doctors, nurses, and hospital administrators in both the public and private sectors are involved in the final procurement of the goods the NHIS distributes (these actors are also often the sources of misgivings with the program) – the contribution structure of the NHIS in the form of premium payments and a specific tax informs enrollees that the incumbent government is to blame for the perceived failings of the NHIS. Lastly, I also test these theoretical expectations through a vignette experiment, in which I compare the relationship between incumbent support and healthcare distributed through the NHIS (programmatic distribution) and individually targeted monetary transfers (non-programmatic distribution). My findings suggest that the non-conditional nature of the NHIS enables citizens to engage in performance-based voting. When healthcare is distributed programmatically and citizens are not dependent on individual politicians for targeted goods, they sanction incumbents for poor healthcare provision. When citizens acquire healthcare through targeted monetary transfers, they are unlikely to sanction poor healthcare provision. I make multiple contributions to the literature on goods distribution, accountability, and social policy in sub-Saharan Africa. First, I expand upon the goods distribution literature by moving beyond conventional forms of goods allocation and describing the political relevance of programmatic distribution in the region. Despite programmatic distribution playing a central role in determining “who gets what,” research has only recently begun focusing on this type of distribution in sub-Saharan Africa. Pieces that do evaluate programmatic distribution focus almost solely on conditional cash-transfer programs (Lavers 2019; Hickey et al. 2019), Additionally, these studies most often take a 7 top-down approach, focusing more on the implementation of such policies and the political incentives leaders face in doing so. My research describes the features associated with a programmatic distributive policy which are influential in determining how citizens evaluate goods allocated via this mechanism and highlights the political ramifications of distribution for the ultimate recipients of programmatic goods - citizens. I also add nuance to the literature focusing on voter preferences in sub-Saharan Africa. Whereas many studies have sought to explain voting intentions as resulting primarily from vote buying, clientelistic linkages, and ethnic kinship, I introduce the production of goods via programmatic policy as an influential force that plays a role in how citizens evaluate incumbent performance and form preferences related to perceptions of incumbent performance. Third, my research contributes to the growing literature on social policy in sub-Saharan Africa, and healthcare policy specifically. Beyond Ghana, several states in the region including, Rwanda, Nigeria, Burundi, and Kenya, have established national health insurance schemes as a means with which to increase access to healthcare services. Other states in the region that are seeking to enact this form of social insurance are in the pre-implementation and piloting phases, including South Africa, Tanzania, and Senegal. While this fact has drawn the attention of scholars associated with healthcare financing,public health, and development economics, the political science literature rarely evaluates this type of policy in the developing world. Finally, by highlighting the relationship between both programmatic, and non- programmatic distributive mechanisms, performance evaluations, and incumbent support, I provide additional insights into how “messy” distributive politics is in Ghana. In developing contexts, targeted distribution and clientelist relationships may exist alongside 8 more conventional forms of programmatic distribution. Indeed, some goods may be distributed programmatically, while others are specifically linked to individual politicians. My research highlights this variation and provides a valid account of this reality. Thus, this research program describes national health insurance schemes, and the NHIS in particular as distinctly programmatic in nature. The characteristics that make this type of distribution scheme unique – its non-conditional nature, deferred benefits, and association with direct citizen contributions - engender beliefs that structure performance evaluations and inclinations toward accountability. This notion lies in contrast to non- programmatic distribution, where the link between goods performance and accountability is blurred. Taken together, these analyses provide important insights associated with this novel form of distribution, including how distribution through a national insurance scheme differs from targeted and other non-programmatic modes in sub-Saharan Africa, as well as how and why the characteristics of these schemes relate to broader political attitudes regarding the provision of public goods. 9 CHAPTER 1: HOW THE DISTRIBUTION OF GOODS INFLUENCES ACCOUNTABILITY IN SUB- SAHARAN AFRICA This chapter describes the literature on public goods distribution and accountability in sub-Saharan Africa. As this dissertation focuses on the Ghanaian National Health Insurance Scheme (NHIS), a programmatic state healthcare policy, a brief review of the literature highlighting the political impacts of policy outputs is also included to provide broader context, as well as a historical review of state healthcare policy in sub-Saharan Africa. Following the literature review, theoretical expectations are developed that explain how the characteristics of programmatic healthcare distribution through a national health insurance scheme allow eligible recipients to formulate expectations with which they evaluate goods performance and translate these evaluations into incumbent support or opposition. Additional theoretical expectations are developed regarding how citizens evaluate incumbent performance when healthcare is derived through non-programmatic means, whereby resources for healthcare provision are individually targeted to citizens at the discretion of political leaders. Whereas myriad studies have examined the relationship between public goods, the provision of public goods, and accountability – the notion that citizens use their votes to influence government behavior – researchers broadly focus on a narrow set of “goods” distributed in familiar ways. Governments building schools, hospitals, and other infrastructure projects provides citizens with concrete displays of performance on which they can evaluate political leaders. Yet, governments have at their disposal various ways with which goods can distributed that may alter the relationship between goods provision and accountability. 10 In light of these various means of distribution, the theoretical argument made in this chapter focuses on how goods distributed programmatically, through a national health insurance scheme can influence the relationship between citizens and their political leaders. Under a national health insurance scheme, the state covers or subsidizes healthcare costs for members, who are charged or taxed at recurring intervals of time (Gros 2016). A national health insurance scheme is a form of programmatic policy. Goods (healthcare) are distributed to eligible individuals (health insurance enrollees) according to a formal set of rules (enrollees must register and contribute to the program to be considered eligible for benefits). In terms of distribution, a national health insurance scheme differs from other modes of goods distribution in that the goods derived from such a scheme are not conditional on support, nor tied to individual political authorities. Moreover, the goods this type of policy proffers are associated with extended time horizons. Citizens enroll in a specific programmatic distribution schemes for the purpose of accruing a particular good (healthcare), which is distributed at a later future date. The goods associated with an insurance scheme are deferred. This notion enables citizens to develop expectations about the type and quality of future goods to be distributed. Expectations are solidified by the fact that members of a national insurance scheme are often required to directly contribute economic resources (payments – premiums and registration fees, or specific taxes) for this specific type of good. Citizens are invested in these programs and have expectations regarding how their resource contribution will be utilized. Research on public goods and accountability behavior in sub-Saharan Africa suggests that expectations are crucial in influencing the likelihood that citizens hold leaders accountable for service provision (Gottlieb 2016). Whether or not citizens’ 11 expectations are in-line with what a national health insurance scheme actually provides in terms of coverage is less important than what they believe it provides. If these expectations are not met, citizens will react to what they perceive of as poor policy performance. Owing to the non-conditional nature of programmatic distribution, citizens are more likely than those acquiring healthcare through individually targeted non-programmatic means to convert these performance evaluations into voting preferences as their receipt of goods is not bound to individual politicians. These outcomes are then compared to theoretical expectations associated with healthcare distribution via non-programmatic means. In contexts where individuals rely on political leaders to distribute individually targeted monetary resources for healthcare, I argue that performance is judged not according to the quality of healthcare goods received, but relative to the receipt of monetary assistance. When political leaders provide resources, they are rewarded regardless of the actual quality of goods procured with those same resources. When leaders fail to provide these resources, they are sanctioned accordingly. This chapter begins by surveying the literature on state policy, goods distribution, and accountability in sub-Saharan Africa, with an emphasis on state policy outputs (public goods and services). Thereafter, the discussion moves to the relationship between accountability and public goods. Finally, the theoretical argument is described in relation to goods distributed through a national health insurance scheme. 1.1 STATE POLICY AND POLITICAL BEHAVIOR IN SUB-SAHARAN AFRICA How does goods provision impact accountability linkages between citizens and political leaders in developing contexts? A popular image of African elections describes 12 them as more strongly related to the acquisition of direct particularistic benefits on behalf of voters than as tools through which to hold politicians accountable for service delivery. Numerous studies have highlighted how widespread the transfer of private goods in exchange for political support is on the continent (van de Walle 2003; Bratton 2008; Bratton and Logan 2006), as well as the preferences of voters in sub-Saharan Africa for clientelist political appeals over programmatic (Wantchekon 2003). Yet, despite this image, a growing body of work has suggested that African voters are also influenced by public goods provision, economic factors, and performance evaluations (Harding 2015; Posner 2005; Lindberg and Morrison 2008). Moreover, recent studies have demonstrated the potential for social policy to influence political behavior on the continent through a policy feedback mechanism (Bleck 2015; MacClean 2011; Hern 2017). These studies focuses on the influence of policy outputs as important factors in influencing the ways and likelihood with which citizens interact with the state and political leaders (Skocpol 1992; Soss and Schram 2007; Soss 1999; Mettler and Stonecash 2008). While there is scant empirical research on state policy outputs and citizens’ political behavior in sub-Saharan Africa - policy studies focusing on Africa in general are lacking, with Gros (2016) noting that “Public policy remains a black hole in African studies” (pg. 3) - the scholars who have evaluated policy outputs as influential determinants often suggest that basic service provision provides a means through which political leaders can connect with their constituencies (Bleck 2015; Harding and Stasavage 2014). In this context, basic service provision provides citizens with visible evidence that their government is capable, responsive, and legitimate, as well as willing and able to provide the rights of citizenship. Provided with this evidence, citizens are encouraged to engage with the state through 13 existing channels and institutions, enhancing participation and accountability links between citizens and governments (Bleck 2013). It is those policy outputs that are visible and readily attributable to government action that demonstrate responsive government to citizens. McLoughlin (2015) notes that the provision of vital services - and healthcare services in particular - are key to building legitimacy in developing states, as these basic services “represent a material expression of reciprocal state-society relations” (pg. 76). Researchers in the development literature often highlight that the provision of these social services is especially likely to enhance legitimacy when citizens believe the state should be providing them, as government provision provides citizens with a signal that the state’s values and priorities lay with them (Bellina et al. 2009; Corbridge 2005). When citizens perceive the state as being responsive in this way, it is a signal that the state is upholding its “end of the bargain” and the political system is worthy of the tax funding, political support, and the participation of the masses (McLoughlin 2015). Indeed, in sub- Saharan Africa, several leaders of developing countries such as Rwanda and Kenya have focused explicitly on the provision of public healthcare to engender legitimation and credibility (Chemouni 2018). Researchers often single-out healthcare policy in particular, as instrumental in engendering political trust, a vital aspect of governmental legitimacy (Dionne and Grepin 2013; Kumlin and Rothstein 2005). The studies that evaluate the influence of policy outputs on citizens’ political attitudes and behaviors in sub-Saharan Africa often conceptualize citizens’ experience with state policy outputs in terms of access to state goods and public facilities such as public schools and health clinics (Hern 2017; Bleck 2015; MacLean 2011). These authors suggest that in low-capacity and emerging democracies, the simple provision of public goods by the 14 state, and citizens’ experiences with these goods and the institutions that provide them, shape political participation and engagement in politics. For example, Bleck (2015) evaluates the impact of publicly provided education on citizens’ political and civic engagement in Mali, finding that parents who sent their children to public schools were more likely to vote, campaign for political parties, and be registered to vote, whereas parents who enrolled their children in non-state schools (madrassas) were less likely to engage with the state. Bleck argues that this relationship exists due to the fac that citizens’ exposure to public schools demonstrates evidence of the state’s capacity, eroding their skepticism of democracy and engendering political efficacy and enhanced political interest. Hern (2017) and MacLean (2011) come to similar conclusions regarding sub-Saharan Africa, specifically, whereby access to public facilities enhances electoral (voter turnout, voter registration) and non-electoral (attending community meetings, contacting officials) participation. MacLean suggests this enhanced participation stems from a mobilized reaction to the retrenchment and decline of state services in sub-Saharan Africa – when citizens utilize poor quality public services, which are perceived of as universal entitlements, they are mobilized to participate in politics. Hern posits that Zambians who utilize public services are more likely to participate in politics as they see the provision of these services as an attempt by the state to meet their needs despite its meagre resources, and thus perceive the state to be responsive and worthy of engagement. 1.2 GOODS AND ELECTORAL ACCOUNTABILITY IN SUB-SAHARAN AFRICA While the latter literature focuses on African citizens’ experiences with public policy outputs - usually conceptualized as experiences with public health clinics and schools - in 15 relation to political participation and behavior, the concept of political accountability in relation to social policy has rarely been evaluated. This distinction is important, for if citizens hold their representatives accountable in a manner that is responsive to state policies and services, rather than to private goods, then these same policies may bolster the developmental benefits of democracy. The notion of electoral accountability assumes that politicians are concerned with citizens’ demands and how they vote, and that citizens will utilize elections to punish or reward politicians based on their performance (Fearon 1999). Do citizens behave accordingly? This question is complicated by the fact that the extent to which democratic elections provide avenues for accountability varies by context. There is a broad literature regarding the structural and institutional constraints impacting the ability of citizens to hold politicians accountable and act as effective principals in developing countries (de Kadt and Lieberman 2017; Keefer 2007; Gottlieb 2016). In the context of sub-Saharan Africa, this literature often highlights targeted redistribution and clientelistic exchanges, as well as ethnicized politics as factors mitigating the link between goods provision and account (Wantchekon 2003; Kramon and Posner 2016; Briggs 2012). The literature highlighting the accountability impacts of basic service provision in developing contexts does provide some important insights into whether citizens reward or punish politicians on the basis of public goods and service provision. For policy outcomes and public goods provision to influence accountability relationships between politicians and voters, high-quality information regarding government performance must be accessible to voters (Mani and Mukand 2007). Contexts associated with “informational asymmetries” – instances where voters lack information regarding government 16 performance - can often lead to a lack of electoral accountability on behalf of voters (Besley and Burgess 2002). In some contexts in sub-Saharan Africa, the mere construction or presence of public buildings (clinics and schools) can provide valuable political information to voters regarding government presence and legitimacy (Bleck 2015; Hern 2017). When voters are more informed regarding the activities of governments, they are better able to evaluate performance and vote accordingly (Keefer and Vlaicu 2008). In addition to information, the attributability of a particular policy outcome to political action often reigns paramount in influencing the likelihood of accountability. Harding (2015) describes how improvement in the quality of a non-clientelistic public good – roads—influences incumbent vote share in Ghana. The author argues that public goods that are attributable to political action (such as road conditions) are critical in engendering this relationship, whereas policy outcomes that are not similarly attributable do not exhibit the same influence on accountability behavior. In contrast, Harding details a variety of education inputs, which vary in the extent to which they are directly attributable to government action. He argues that teacher supply and the building of new schools, the funding of which falls under the direct purview of the executive, are directly attributable to executive action and thus most likely to influence incumbent vote shares. Thus, the attributability of policy to political action can vary across policy type. The notion of attributability is further echoed by Harding and Stasavage (2014), who argue that the removal of primary school fees can be directly attributed to executive action and thus impacts voting intentions. Whereas the aforementioned research emphasizes the accountability linkages present when citizens have direct, attributable, and concrete evidence of performance or 17 accrue targeted goods, a national insurance scheme provides a different distributional context altogether, a notion which is elaborated in a later section of this chapter. In terms of provision, the characteristics of national insurance schemes differ in multiple ways from other modes of distribution. While the construction of roads and schools presents can present citizens with objective evidence of government performance, the distribution of goods through a national health insurance program presents us with a different context altogether. Goods distributed through a national health insurance scheme are allocated in manner that is not conditional on political support or linked to any individual politician– all individuals who are enrolled in said scheme have the potential to benefit from the policy, regardless of which party or politicians they support. This notion lays in stark contrast to benefits derived from non-programmatic modes of distribution such as clientelism and patronage, whereby access to benefits is contingent on political support (Kitschelt and Wilkinson 2007). Beyond the non-conditional distributional aspect of these policies, citizens that enroll in national health insurance schemes do so to acquire a specific good as needed at some point in the future. Lastly, the payment structures associated with national insurance schemes often entail direct contributions from enrollees which differ from general taxation which is often used for the procurement of other public goods. As will be described in the theoretical portion of this chapter, these characteristics form the basis of the relationship between this particular mode of distribution and performance evaluations. The next section provides a brief introduction as to the history of healthcare policy and distribution in sub-Saharan Africa. In doing so, this section highlights the particular significance African citizens place on state-provided healthcare services. The following also 18 establishes the basis for the implementation of national insurance schemes on the continent. 1.3 HEALTHCARE POLICY IN SUB-SAHARAN AFRICA Healthcare policy, as it is referred to in this research program, constitutes the actions and decisions made by political leaders regarding the health of citizens (Gros 2016). As such, the general goal of state healthcare policy is to improve population health (de Leeuw et al. 2014). Healthcare policy in sub-Saharan Africa has significantly evolved from the colonial era to the present, with wide variation seen in the degree of state involvement in the healthcare sector, the extent to which non-state actors are involved in public health policy, health challenges faced by governments, and health outcomes experienced by citizens as a result of state policy. Herein, a brief introduction is given concerning healthcare policy in the sub-Saharan region from the colonial era to the present. The remainder of the chapter highlights how, beyond impacting the physical health of African citizens, state healthcare policy can also influence citizens in the political arena. As with most aspects of governance in sub-Saharan Africa, an analysis of state health care policy is incomplete without an understanding of its origins and the influence of colonization. Colonial healthcare policy broadly consisted of curative care and the study of disease and disease prevention (Vaughan 1991). In this period, healthcare policy broadly reflected the views of European colonizers and missionaries, who viewed healthcare in the region as a political project that sought to bring “progress” to the region and serve colonizers and their African laborers in pursuit of capital accumulation (Gros 2016). While practitioners of traditional medicine, African elites, and other local actors may have had 19 some informal influence on healthcare policy at the time, the majority of influence laid with European colonizers. Though healthcare policy was immensely fractured and varied among the multiple colonizers, in general, as the colonial era advanced, it became increasingly state-centric as colonizers were increasingly torn between the pursuit of capital and ensuring they had a ready supply of healthy laborers to accrue it. While public health services were rarely free and as a result often unutilized, non-state actors (missionaries) played a large role in procuring healthcare for Africans during this period (Gros 2016). Healthcare policy in the immediate post-colonial period became a focal point for governments of newly independent states. Nearly all post-colonial states committed to expanding healthcare coverage for African citizens. From 1957 to the late 1970s, healthcare policy in independent states was decidedly state-centric and sheltered from the influence of the policy preferences of external actors (Gros 2016). During this period, healthcare was broadly defined by states as a social and human right, and governments sought to eliminate financial obstacles to ensure universal access to government facilities – often with little regard for cost recovery. African states largely pursued a national health service model of healthcare systems, whereby services at public clinics and hospitals were nominally free for citizens, and funds for healthcare systems were derived from general government revenue and distributed by ministries of health. Public health facilities were broadly confined to urban areas during this period (Bates 1985). While states sought to significantly broaden social welfare, a lack of administrative capacity, underinvestment, and high demand crippled public health facilities. Any advances that were made in access to care and public health provision during this period were relegated by the late 1970s, as 20 economic recession, indebtedness, and political instability gripped the continent (Streefland 2005). By the early 1980s, adult life expectancy on the continent was nearly 25 years lower than the average in developing countries, and healthcare systems were experiencing fiscal crisis (NEPAD 2014). To address shortfalls in funding, many states turned to the International Monetary Fund (IMF) and World Bank for relief. These institutions instructed governments to emphasize cost recovery, decentralization, and privatization in their healthcare systems in exchange for much needed financial assistance in the form of structural adjustment programs (SAPs). As a result, fee-for-service (aka “cash and carry”) became state policy, with many citizens being excluded from state-provided healthcare (Gros 2016). External actors and the international community have since become prominent players in shaping domestic healthcare policy in sub-Saharan Africa. Through the 1990s and into the new millennium, numerous health care initiatives focusing on sub-Saharan Africa have been advanced by the international community in an effort to improve access to quality care through state policy. Perhaps the most impactful among these initiatives is the United Nations Millennium Development Goals (MDGs), which outline specific development policy goals to be pursued and/or supported by signees, with financing provided by the wealthiest countries and intergovernmental organizations (IGOs) (UN 2015). In 2000, all 189 UN member states and numerous IGOs (including the IMF, World Bank, and African Development Bank) supported the initiative. The MDGs’ emphasis on healthcare systems is evinced in the fact that three of the eight MDGs focus specifically on healthcare: The reduction of infant mortality, combating HIV/AIDS and malaria, and 21 improving maternal health (WHO 2018). With all African countries signing the declaration, the MDGs were incorporated into existing health policies. While MDG signees demonstrated a strong commitment to the healthcare policy goals described in the declaration, the achievement of these goals has proven difficult for most countries in sub- Saharan Africa (UN 2016). At independence, both citizens and their governments shared the perception of healthcare as a state-provided good. This perception has persisted. In most African countries, access to public clinics has been perceived of as a right of citizenship since independence (MacLean 2010). Bratton and Chang (2006) note that Africans largely rely on non-state services when they encounter shortages of basic needs, with medical care a glaring exception, suggesting the state is a key provider of this particular public good. Bratton (2007) further notes that African citizens largely recognize healthcare provision as the responsibility of the state and perceive healthcare and education as an entitlement of citizenship. Additionally, Bratton (2007) notes that citizens are also relatively well- informed regarding government healthcare policies such as the promise of free publicly provided care. In terms of usage, surveys on the continent describe relatively frequent experiences with public health clinics, with the Afrobarometer Round 6 survey indicating that 64% of 47,936 respondents in 32 countries in sub-Saharan Africa had utilized public clinics or hospitals in the previous 12 months. Moreover, the survey indicates relatively similar degrees of contact among urban (61% reported visiting a public clinic or hospital) and rural (65% reported contact) dwellers. Finally, the same survey reports that respondents (29% of total) deemed healthcare as the most important problem that their government should address, second only to unemployment; this finding mirrors that of 22 Bratton (2007), who utilized an earlier dataset. These notions suggest that healthcare access and government health policy are particularly salient among citizens in sub-Saharan Africa. Owing to this notion, beyond being a requisite for good health, and likely because if it, it is not surprising that healthcare policy is also politically important for African citizens. Recently, decentralization and healthcare system financing have been emphasized by international organizations and donor institutions as a means with which to increase access to care while also ensuring a degree cost recovery. In pursuit of these goals, multiple states have enacted public programs (e.g. state health insurance schemes) to increase access through the reduction of out-of-pocket costs with varying degrees of success. National insurance schemes provide a novel means with which to distribute the good of healthcare to citizens and far removed from traditional channels of healthcare distribution – direct payment for services or nominally free care at public clinics. National insurance schemes are a form of programmatic distribution; goods (healthcare and healthcare services) are provided to eligible recipients according to a formal set of rules. This type of state healthcare policy, whereby citizens enroll in a government insurance scheme, pay some form of contribution, and then are given subsidized care when they require medical treatment, provides a novel context in which to analyze citizens’ political responses vis-à-vis state healthcare policy. Social insurance schemes are programmatic policies, rather than tangible goods (such as a new clinic or pharmacy), and involve varied time-horizons and contributory funding methods which structure expectations in relation to the conveyance of the public good they proffer. Evaluating the political outcomes associated with citizens who have utilized a healthcare policy output is a blunt instrument, and it is presumptuous to assume that African citizens 23 conceptualize healthcare policy as merely the construction and administration of public clinics and hospitals. The next section provides the theoretical bases for the argument linking goods distributed through a national insurance scheme, the expectations they instill, and performance-based accountability. The unique characteristics of national insurance schemes which engender this relationship illustrate this argument. 1.4 THE PROGRAMATIC DISTRIBUTION OF GOODS THROUGH A NATIONAL INSURANCE SCHEME What differentiates the distribution of goods through a national health insurance scheme from other forms of distribution? How do the characteristics of this form of distribution influence performance-based accountability? A key aspect of this relationship lies in the ways in which goods are distributed through an insurance scheme. As described in the previous section, several African states, in an effort to increase access to care for citizens, have enacted social insurance programs as a means with which to increase access through the removal of financial impediments. In this section, I highlight national health insurance schemes as distinct forms of programmatic distribution that are significantly different from those previously studied in the political science literature. In particular, the timing, funding of, and the means through which goods are derived from programmatic policies such as national health insurance programs mean that these programs differ from more common, non-programmatic modes of goods provision in sub-Saharan Africa in three ways. First, as a programmatic policy, the goods distributed by national health insurance schemes are non-conditional in nature. They provide benefits to all those in a defined category, regardless of political affiliation or access to individual politicians (Stokes 2007). 24 All individuals enrolled in the scheme have the potential to benefit from the policy, regardless of which party or politicians they support. Second, the benefits associated with national health insurance schemes are deferred. For enrollees, the public good that these schemes offer (healthcare) is one based on need and is to be acquired in the future. For that matter, the final good associated with these policies may not be acquired at all if an individual is not in need. Third, national health insurance schemes differ from other forms of goods distribution in that many enrollees directly pay government officials for future goods in the form of premiums, and enrollment fees. In addition to direct payments, specific taxes are often levied to supplement such schemes. These traits make the national insurance schemes distinct from other forms of healthcare distribution. This section focuses on the nature of these characteristics. This type of state policy, whereby citizens enroll in a government insurance scheme, pay some form of direct and/or specific contribution, and then are given subsidized care when they require medical treatment, provides a novel context with which to evaluate programmatic goods distribution. As noted in the previous section, scholars often describe how public goods provision provides a means through which political leaders can connect with their constituencies (Bleck 2015; Harding and Stasavage 2014). Goods provision provides citizens with visible evidence that their government is responsive, capable, and willing to supply citizens with the goods that they believe they are entitled to. In sub-Saharan Africa, the mere construction or presence of public buildings (clinics and schools) can provide valuable political information to voters regarding government presence and legitimacy (Bleck 2015; Hern 2017). Yet there are different modes of distribution through which governments and political leaders provide these goods to their citizens. In many instances, politicians are 25 rewarded when they distribute goods to supporters, while excluding their detractors from access to these goods (Golden and Min 2013). Non-programmatic distribution often favors certain groups or individuals in a society. One form of non-programmatic distribution is what is commonly known as pork barrel politics, whereby certain geographic constituencies are targeted for distribution, the costs of which are shared by all districts (Stokes 2007). In terms of sub-Saharan Africa, the targeting of goods to constituencies and groups through non-programmatic means has been described in relation to building and infrastructure projects, aid funding, and local public goods among others (Barkan and Chege 1989; Koter 2013; Carlson 2015; Briggs 2012). Other forms of non-programmatic distribution are contingent on political support. In contrast to programmatic distribution, whereby goods can be gained regardless of political support, the accrual of goods through these non-programmatic means often entails conditionalities for recipients; the receipt of goods is conditional on political support. Conditional exchanges are associated with the practice of clientelism, where particularistic benefits are targeted to individuals in exchange for political support. Clientelistic exchanges involving the transfer of material benefits to individuals have been well- documented in sub-Saharan Africa (van de Walle 2003; Bratton 2008; Bratton and Logan 2006; Wantchekon 2003; Lindberg and Morrison 2008). Non-programmatic distribution in the form of clientelism has the potential to pervert the accountability relationships that are the basis of democratic theory. When clientelistic linkages exist between citizens and politicians, the act of voting denotes a demonstration of loyalty in pursuit of particularist goods. In this context, voters may lose the ability to effectively hold politicians accountable for their performance in office (Hicken 26 2011). Under clientelism, accountability does not depend on how successful leaders are in terms of the production of collective goods nor the realization of the broader redistributive preferences of citizens, but on the exchange of access to goods and services in return for support (Kitschelt and Wilkinson 2007). This context leads to the overproduction of goods and services targeted to smaller constituencies and the underproduction of broader public goods (Keefer 2007). Myriad research has been done on this type of distributive politics (see Hicken 2011 for a review of this literature). However, if the mode of goods distribution is programmatic, rather than targeted, both political supporters and opponents have access. A policy is programmatic if its goods distribution is formalized, according to a certain set of rules for those in a defined category (Stokes 2007).When a policy is programmatic in nature, incumbents have little or no say in the delivery of benefits, as citizens receive these benefits based on objective, publicly stated criteria (Kitschelt and Wilkinson 2007); programmatic policies bestow goods that cannot be withdrawn if an individual does not support a particular political leader - they are non-conditional in the sense that the receipt of benefits is not tied to any individual politician. Programmatic distribution benefits citizens indirectly rather than with selective incentives (Kitschelt 2000). Voters receive goods through programmatic redistribution regardless of whether or not they voted for the party in power. Competitive elections incentivize politicians to expand the provision of public goods because they enable voters to hold politicians accountable. The broad dispensation of goods across society which is derived from programmatic distribution lays in stark contrast to the targeted means of distribution involved in the non-programmatic delivery of goods to individuals, groups, and 27 constituencies. Moreover, goods derived through programmatic means cannot be as easily withdrawn as those accrued through quid-pro-quo arrangements and constituency targeting. Programmatic policy is significantly under-studied in the literature on goods provision in the developing world (Mares and Carnes 2009), and in sub-Saharan Africa in particular (Gros 2016). Broadly, programmatic policies have been described as influencing electoral politics by increasing turnout (Layton and Smith 2015) and incumbent vote share (Golden and Min 2013), though some researchers dispute the latter point (Imai et al. 2020). Stokes et al. (2013) cite two criteria which must be fulfilled for a distributive strategy to be considered programmatic: The criteria of distribution must be formalized and public, and the formal criteria of distribution must dictate how resources are actually distributed. In the case of a national health insurance scheme, a policy distributes goods (healthcare) to those in a defined category (enrollees) based on a certain set of rules (registration and premium payment), regardless of their political affiliation. Studies highlighting programmatic policies and their impacts related to accountability and incumbent support in the developing world often focus on conditional cash transfers (CCTs) (Zucco 2013). CCT programs are a type of programmatic policy which distribute targeted transfers to assist low-income individuals and families most often in developing countries. Distribution in these programs is targeted to certain identified individuals – transfers are restricted to individuals within these categories. Studies focusing on CCTs broadly find that voters often reward incumbents for the particularistic benefits associated with this form of programmatic distribution (Ortega and Penfold- Becerra 2008; De La O 2012; Zucco 2013). Though the benefits from CCT transfers include 28 cash payments from incumbent governments which are targeted to individuals, formal rules exist designating which individuals qualify for these benefits and these exchanges are not contingent on political support as is the case with clientelism – although the potential exists for such programs to foster clientelistic exchanges (Weitz-Shapiro 2012). While both CCT programs and national health insurance schemes are both programmatic social policies, significant differences exist between these two types of programs. CCT benefits are only accessible to certain segments of society. Individuals who do not qualify for targeted transfers are unable to acquire them. While the benefits derived from a national health insurance scheme are particular to enrollees, these programs do not possess this same exclusionary principle in relation to eligibility. Moreover, the actual benefits one is able to derive from a national health insurance scheme (healthcare) are far removed from direct monetary transfers. Beyond the notion that the goods produced by national health insurance programs are not conditioned on political support or access to particular politicians, the goods acquired through a national insurance scheme differ from those acquired through non- programmatic means in relation to time horizons. Goods provided through an insurance scheme are deferred; citizens take an initiative to enroll in a program and expect that benefits will be provided as needed in the future. Indeed, benefits derived through an insurance scheme might not be utilized at all, if a citizen so chooses or does not have the need. For policies such as these, access to goods is based on the ability of a government to provide them and the commitment that a government has in keeping its end of the bargain – that goods will be available to citizens when they need them. Under these conditions, the state has made a commitment to provide the good of healthcare, and citizens must trust 29 that it will be provided if needed. This notion demonstrates the ties that bind insurance enrollees to their government. Whereas other modes of distribution provide concrete or immediate evidence of procurement, in the case of goods distribution through a national health insurance scheme, enrollees have registered for a specific service which is to be utilized in the future. National health insurance schemes also differ from other modes of goods distribution in that enrollees often directly pre-pay for a specific good (healthcare). National health insurance schemes are a form of social insurance - they are managed by a public organization, and their funding is generated through general taxation and mandatory contributions from certain groups. Direct payments are often made through premiums and enrollment fees. Additionally, taxes may be levied specifically to fund a national health insurance scheme. As regards pre-payment through direct fees – these types of transactions are different than governments taxing citizens and then using these funds for various services throughout the country or a particular region. When citizens directly pay premiums and enrollment fees, they are pre-paying for their own personal services – in this case, healthcare. A citizen is that is generally taxed under normal circumstances does not necessarily know where those resources will eventually be used. In contrast, direct payments paid to a national health insurance scheme are pre-payment for specific individual services. The above characteristics differentiate national health insurance schemes from other forms of goods distribution. The elements that make this form of goods distribution unique are critical in engendering expectations related to the quality of goods that will 30 eventually be provided. The next section describes the link between these characteristics, expectations, and government performance. 1.5 NATIONAL HEALTH INSURANCE SCHEMES: PERFORMANCE, AND ACCOUNTABILITY How do the characteristics of this form of distribution influence performance-based accountability? Citizens’ beliefs and expectations are central to the relationship between a national health insurance program and how its performance is evaluated. Researchers have suggested that actual levels of performance are less important for accountability than performance relative to citizens’ expectations (Gottlieb 2016); citizens are likely to hold governments accountable when they have high expectations and these expectations are not met. In the case of the national insurance schemes, expectations are informed by the characteristics of their design described above: The deferred nature of the benefits provided, and pre-payment in the form of annual premiums and registration fees, and/or taxation. The extended time horizon associated with insurance programs ensures that individuals who enroll in the program are able to form expectations regarding the quality of promised goods. These expectations may or may not be in-line with what a given insurance scheme actually covers depending on the information available to a given enrollee, but they inform beliefs regarding the quality of expected goods. In developing and low-information contexts, citizens may be relatively uninformed regarding specific medications or procedures that are covered by an insurance scheme. Individuals within these contexts may form lofty expectations concerning what a national health insurance scheme covers. If these expectations are not met, these same individuals are likely to 31 believe the program is not being adequately administered by government officials or incumbent leaders. Research has also demonstrated the significance of ownership – the perception that a resource “belongs to” an individual (Pierce et al. 2001) - in conferring high expectations. Taxation has been suggested to increase accountability pressures by engendering perceptions of ownership over budgets (Paler 2013; de la Questa et al. 2021). In terms of national health insurance schemes, funding sources often include direct payments in the form of premiums, and/or specific taxes. These taxes and premiums are a form of social contract, whereby a government provides/subsidizes goods to citizens for payment. Tax payments are often said to inspire political action because they require a contribution from citizens – they are “generally conceived of as a cost that individuals must pay out of their income, and this incentivizes individuals to hold governments accountable for how they spend their money,” (Sandbu 2006). Citizens expect to retain their income and taxation forces a loss of earnings that they are eager to regain through government spending. If a citizen’s willingness to sanction a government for poor performance is linked to the extent that poor performance hurts them personally, taxation will increase citizens demands for accountability. (Martin 2016). However, some taxes are more amenable to increasing the likelihood of performance-based accountability than others. Visible taxes, those which are directly linked to specific programs or levied on individuals, are more likely to be associated with sanctioning poor performance than indirect taxes (De La Cuesta et al. 2021). When it comes to premiums and registration fees, these contributions are similar to direct taxes in that they are levied on individuals and transferred to the government. Direct payments from 32 citizens paid through premiums and registration fees are likely to heighten expectations regarding future goods because these fees move directly from the citizen to government officials associated with the insurance scheme. Moreover, these payments are individual payments that are for a specific service (healthcare), a service for which poor performance has significant negative ramifications for individual livelihoods. Notions of direct payment are particularly important in developing contexts as low-income citizens may be required to contribute scarce resources to participate in these programs. The deferred nature of the benefits and fee structures associated with goods distribution via a national health insurance scheme engender expectations about the quality of goods that will be received. The fact that the goods associated with a national health insurance scheme are not immediately available to citizens and are goods for which they have pre-paid, both through direct contributions via premiums and/or taxation, gives citizens the perception that they have done their part, and the government must now fulfill its promise. Upon the acquisition of goods distributed through an insurance scheme, recipients evaluate the goods they receive relevant to their expectations and form perceptions regarding performance. It is the non-conditional aspect of programmatic distribution that allows beneficiaries to convert these evaluations into voting intentions. If an individual’s wellbeing is not linked to support for, or association with an individual incumbent, as is often the case with non-programmatic distribution of healthcare, citizens are more likely to engage in performance-based voting (Carlson 2021). Under these conditions, citizens who have high expectations regarding future benefits and are monetarily invested in a government distributive program are likely to hold political leaders accountable when they deem the goods delivered inadequate. 33 FIGURE 1.1 EXPECTATIONS AND THE QUALITY OF GOODS PRODUCED THROUGH A NATIONAL HEALTH INSURANCE SCHEME Expectations Reward met – Strong incumbent performance Pre- payment for goods: Premiums Non-conditional distribution Expectation Actual s regarding Quality of Deferred the quality goods goods of future received goods Pre- Expectations Punish payment for not met – Poor incumbent goods: performance Taxation Non-conditional distribution 34 As Figure 1.1 shows, the characteristics of a national health insurance policy – the deferred goods, and pre-payment for expected goods in form of premiums and specific taxes make goods distributed through this type of policy especially amenable to performance-based evaluations. These traits structure beliefs about the quality of yet-to- be-received goods. Upon the receipt of the actual goods that are distributed through the insurance program, citizens evaluate the quality and quantity of the good received in light of their expectations. If the good received does not meet expectations, the non-conditional nature of programmatic distribution allows them to punish/reward incumbents without sacrificing their wellbeing. These notions raise the question of why citizens would punish an incumbent based on the performance of a programmatic policy when those incumbents might not have any control over its administration and functioning? This is a valid concern given that national health insurance schemes; though they may have been created and implemented by a particular leader, they are administered by bureaucrats and administrative institutions. Research has demonstrated that voters often find it difficult to assign credit or blame to incumbent politicians, even when politicians are directly involved in the creation or implementation of policy. Harding and Stasavage (2013) suggest that political leaders deliberately chose policy options which are directly attributable to them to overcome this concern. This same notion of attributability is echoed by Harding (2015). Other researchers have also demonstrated that voters also assign credit to political leaders for programs and policies for which they played no role (Healy and Malhotra 2013; Labonne 2013). Moreover, the same can be said for assigning blame. The literature on voting behavior demonstrates that voters frequently assign blame to incumbent politicians for events that 35 are seemingly out of their control (Achen and Bartels 2016). Theoretical work has suggested that voters often blame incumbent politicians for matters that are far beyond their control as they use these instances to determine an incumbent politician’s “type” (Ashworth, Bueno de Mesquita, and Friedenberg 2018). Beyond these notions, a national health insurance scheme, is highly attributable to government action. Though the final distribution of healthcare via a national insurance scheme involves a wide variety of actors in potentially both the public and private sphere (doctors, nurses, hospital administrators) – a point which I return to in a later chapter - these policies themselves are distinctly governmental in nature. Indeed, eligible beneficiaries regularly register for these programs at government offices, domestic funds are often collected via specific taxes, and bureaucrats and administrative institutions are responsible for operating these programs. This notion of attributability clarifies for citizens the link between these programs and government. With government ownership clearly defined, citizens may be likely to apportion blame for the performance of a national health insurance scheme on incumbent leaders as they are impersonally connected to it through their association with the government in power. 1.6 NON-PROGRAMMATIC DISTRIBUTION OF HEALTHCARE: PERFORMANCE, AND ACCOUNTABILITY While healthcare can be distributed programmatically via a national health insurance scheme in-line with the expectations above, it often is not. Here, I focus on developing theoretical expectations related to the non-programmatic distribution of healthcare. Non-programmatic distribution comprises any form of distribution in which public formal rules and procedures dictating said distribution are absent (Stokes et al 36 2013). One important feature of non-programmatic distribution is that absent formal rules, incumbent politicians have discretion over the distribution of goods. For example, a national health insurance scheme has formalized and public rules dictating which individuals are eligible to receive a set basket of subsidized or free goods. As formal rules dictating access exist, individual incumbents do not control the distribution of goods through this system. Whereas a politician who subsidizes a specific constituent’s healthcare costs with cash handouts represents a distributive mechanism which is non-programmatic in nature. There are no public and formal rules dictating which specific constituents should receive cash handouts for healthcare services and distribution is at the discretion of the individual politician. Non-programmatic distribution in sub-Saharan Africa often involves political leaders targeting private goods to individuals. This type of non-programmatic distribution is usually coupled with informal political attachments, often in the form of a quid pro quo; private goods are targeted to individuals in exchange for political support (clientelism). If political support is withdrawn, the transfer of goods ends. Stokes et al. (2013) astutely note that there may be no actual quid pro quo involved in the distribution of goods in a non- programmatic, individually-targeted and partisan manner. Political leaders may extend, non-programmatic benefits to individuals for the purpose of gaining or maintaining political support. In this form of non-programmatic distribution, which Stokes and coauthors term non-conditional partisan bias, benefactor politicians spurn the use of brokers whose job it is to ensure that quid pro quo agreements are adhered to, and citizens do not necessarily risk losing access to resources if they do not vote for a specific politician. Rather, In these relationships, the receipt of goods is determined by access to individual 37 politicians, rather than support. If the benefactor politician loses office, the transfer of goods ends. Regardless of whether this non-programmatic, individual targeting of goods falls under the umbrella of clientelism, or is associated with non-conditional partisan bias, the receipt of goods is dependent on relations with individual politicians. This notion lies in contrast to programmatic distribution, whereby individual politicians do not have discretion over who is able to access what. Given the extent to which politicians informally target goods to individuals in sub-Saharan Africa (van de Walle 2003; Wantchekon 2003; Lindberg 2010), I now describe how this form of non-programmatic distribution shapes performance evaluations and incumbent support in relation to healthcare provision. The informal targeting of healthcare to individuals poses a quandary – How do incumbent politicians distribute healthcare to individual citizens? The concern here is that healthcare involves products and services which incumbent politicians may not have access to, and thus may not be able to distribute to individuals. The most likely way in which political leaders distribute healthcare to individual citizens is by providing them with cash handouts with which they can purchase services. As noted previously, in the 1980s, and 90s many states in sub-Saharan Africa began shifting the burden of healthcare costs onto citizens and continue to do so. Moreover, national insurance schemes see broad participation in only a select few countries, while citizens largely do not have private insurance outside of South Africa. For many African citizens, healthcare costs are the main barrier to treatment (Gros 2016). The notion that African citizens frequently turn to political leaders for assistance in covering their healthcare costs has been reported in Ghana (Lindberg 2010; Wahab 2019), and the broader sub-Saharan region (Agyepong and Adjei 2008: Gros 2016) 38 How does this form of healthcare distribution impact citizens perceptions of service delivery and incumbent performance? Individually targeted cash handouts which are distributed at the discretion of individual politicians differ greatly from goods distributed through a programmatic national insurance scheme. The primary difference between these two forms of distribution relates to the factor around which citizens develop expectations and evaluate performance. As mentioned in the previous section, the policy design features of programmatic distribution through a national health insurance scheme (contribution structure in the form of specific taxes, premium payments, and registration fees; deferred nature of benefit distribution) lead citizens to develop expectations regarding the goods that will eventually be acquired through such a program. Upon the receipt of goods, these same citizens evaluate the goods proffered relative to their expectations and form performance preferences accordingly. In this context, the evaluation of goods is the source of incumbent support or opposition. Citizens are more likely to convert performance evaluations into support or opposition for incumbents because the goods distributed are distributed programmatically and are not tied to individual politicians. The source of performance evaluations in non-programmatic distributive schemes, in which political leaders informally target monetary resources to individual citizens for the purpose of healthcare provision is focused on whether monetary resources were acquired from incumbents. In cases of targeted monetary distribution such as this, poor performance is not dictated by whether or not the healthcare a citizen receives is of poor or superior quality, but whether or not the resources required for healthcare provision were distributed by incumbents. Once this distribution has taken place, the role of the political leader in distributing healthcare has ended. If an incumbent political leader has (not) 39 provided resources for an individual in this manner, that same leader has performed adequately (inadequately) according to this distributive schema and will be evaluated positively (negatively). Figure 1.2 highlights this relationship. Whereas with programmatic distribution through a national health insurance scheme, evaluations are linked to the performance of the program, individually targeted distribution links evaluations to individual politicians. 40 FIGURE 1.2 EXPECTATIONS AND THE QUALITY OF GOODS: INDIVIDUALLY TARGETED, NON-PROGRAMMATIC HEALTHCARE DISTRIBUTION High quality – Reward incumbent Yes Actual quality of goods received Expectations me Low quality – Expectations Reward regarding Cash handout? incumbent individual handout No Sanction incumbent Expectations not met 41 With the basics of these arguments established, it remains to be seen what exactly the specific expectations of citizens are in relation to national insurance schemes in sub- Saharan Africa and whether or not they are being met. The remainder of this research program focuses on deciphering these beliefs and how they relate to theoretical expectations in the West African country of Ghana, a country with a relatively mature and extensive national health insurance scheme. The next chapter describes the Ghanaian National Health Insurance scheme and its applicability to the aforementioned argument linking the design features – payment contribution structure and deferred benefits - of a national health insurance scheme with expectations and performance evaluations. As expectations are central to this theory, qualitative data presented in chapter three describes the beliefs and expectations Ghanaian citizens have regarding the goods provided through the NHIS, their experiences utilizing goods derived from the program, and how these factors relate to notions of accountability. 42 CHAPTER 2: A NOVEL FORM OF GOVERNMENT HEALTHCARE PROVISION - THE GHANAIAN NATIONAL HEALTH INSURANCE SCHEME The previous chapter described how a national health insurance scheme is a form of programmatic distribution which possesses unique characteristics that inform performance-based evaluations concerning the distribution of healthcare. The goods derived from a national health insurance scheme are not conditional on political support or attachment to a particular politician – health insurance schemes provide benefits to a specific group (enrollees) regardless of the party or individual in power. Additionally, the benefits derived from a health insurance scheme are deferred - citizens enroll in a specific program and expect that specific benefits will be provided as needed in the future; benefits derived through an insurance scheme might not be utilized at all, if a citizen so chooses or does not have the need. National health insurance schemes are also often associated with direct pre-payment for services through registration fees, premiums, and specific taxes. As noted in the previous chapter, these policy characteristics are expected to influence accountability relationships between citizens and political leaders by engendering expectations about the quality of goods to be received. The West African country of Ghana was one of the first in the sub-Saharan Africa to adopt this form of social insurance and continues to have one of the highest enrollment rates in the region (Aryeetey et al. 2016). Initially, this chapter describes why governments create social policies such as the NHIS. Thereafter, the implementation and design of the Ghanaian National Health Insurance Scheme (NHIS) is described with reference to the specific characteristics discussed in the previous chapter. 43 2.1 THE EXPANSION OF SOCIAL POLICY IN THE DEVELOPED WORLD AND SUB-SAHARAN AFRICA Though this research program focuses on the NHIS as a particular mode of goods distribution, an evaluation of a social policy such as this would be incomplete without a discussion of its origin. Governments in the sub-Saharan region are increasingly developing and implementing social insurance programs as a means with which to increase their population’s access to healthcare. Why would governments pursue this strategy and what enables them to do so? Social policies, such as the social insurance programs mentioned above, can be defined as “collective and public efforts aimed at protecting the social well- being of people within a given territory” (Adesina 2009, pg. 28). A broad literature describes the distributional outcomes and sources of social policies in developed countries (Esping-Andersen 1990; Goldsmith 1995; Huber and Stephens 2001). As regards developed countries, researchers have proposed theories regarding the structural (industrialization and economic openness) sources of social policy (Carnes and Mares 2007). This literature often suggests that the emergence and extension of the welfare state stems from economic growth and development. Economic growth brings about new social needs, such as the increasing level of education required for workers to be employed in the industrial sector; economic growth also enhances the ability of governments to meet these new needs. In this case, the development of the welfare state results from the broader process of modernization (Huber and Stephens 2001). Quantitative studies related to economic development and the expansion of social policy often identify correlations linking levels of industrialization to aggregate social spending (Wilensky 1975). Other researchers highlighting the structural determinants of social policy expansion cite economic openness or trade volatility as important determinants of welfare state expansion. In liberal 44 economies, governments enact protective social policies to shield domestic workers from external competition and shocks (Cameron 1978; Rodrick 1997; Garrett and Mitchell 2001). Researchers have also suggested the importance of societal and class sources of welfare state expansion. The power resource perspective stipulates that levels of social spending are linked to the power of labor organizations relative to that of conservative political actors and organizations (Esping-Andersen 1985). As the organizational capacity of workers, unions, and labor parties increases, taxation and social spending increase as well as class struggles are transposed into the political sphere, which in a democratic system numerically favors workers over employers (Skocpol and Amenta 1986). Quantitative studies focusing on the power resource theory often link factors such as union density or the share of seats held by social democratic parties to social policy expenditures (Huber and Stephens 2001). Whereas power resource theory highlights the conflict between labor and employers, additional authors note the potential for cross-class alliances whereby labor and employers may both have incentives to support social policy expansion (Swenson 2002; Mares 2003). Studies such as these often seek to describe the conditions under which both labor and employers support certain social policies, as well as the general political conditions associated with cross-class alliances. State-centric approaches seeking to explain the origin of social policies focus on the impact of state structures and bureaucrats in influencing social policy development. Authors often describe the determinative impact of bureaucrats’ experiences with previous policies as an important influence on state preferences towards social policies. For example, Weir and Skocpol (1985) argue that the experience of British policymakers with a 45 limited unemployment insurance program prior to the Great Depression led to their unwillingness to enact large scale-programs during the Great Depression. The authors argue that this was not the case in the United States and Sweden, where bureaucrats lacked this previous policy experience and were more amenable to a dramatic change in policy resulting in larger increases in public works expenditures. Researchers have also described how existing social policies can impact subsequent political development. The “policy feedback” approach describes how the design of existing policies can impact the preferences and bargaining power of different actors in the political sphere by endowing them with resources and incentives (Campbell 2002; Pierson 2004). Other authors focus on the potential for bureaucrats to form preferences regarding social policies based on access to different policy ideas. Weyland (2005) describes how the Chilean model of pension privatization rapidly diffused throughout the globe. Weyland argues that the implementation of pension privatization in Chile provided bureaucrats around the world with heuristics; the Chilean reform caught the attention of other policymakers who generalized the successes of the reform to their own countries and only implemented slight and peripheral deviations from the Chilean model in their domestic reforms. While structural and societal/class-based theories of social policy development are perhaps most applicable to the developed world, state-centric approaches to social policy implementation may have more leverage where labor and employer organizations are weak and bureaucrats exert a broad influence on the policymaking process. Whereas the majority of the aforementioned theories regarding the origin of social policies stem from examinations of the developed world and Europe, the question remains as to why governments would enact these policies in developing contexts, and in sub- 46 Saharan Africa particularly? Two main factors are often associated with the introduction and expansion of social and welfare policies in developing countries: democratization and the influence of external actors (Lavers 2019). In terms of the political sources of social policy in developing contexts, the process of democratization and democratic regime type are often suggested to be main explanatory factors behind the creation of social protection and the size of the state in the developing world. These studies often argue that political leaders in democratic states seeking to appeal to a broad electoral base will make more extensive policy commitments; electoral competition leads to social policy expansion as candidates compete to reach broad groups of voters (Carnes and Mares 2009). Additionally, democratic freedoms allow for the formation of interest groups that can pressure political leaders to increase social spending (Haggard and Kaufman 2008). Political competition can often provide the arena in which novel policy ideas can be espoused. As to the impact of democratization on social policy, Wong (2003) describes how the political dynamics in Taiwan and South Korea changed after their respective transitions from authoritarian rule, such that both countries adopted and refined universal social health insurance policies. Wong argues that the national health insurance programs in both countries stemmed from the institutionalization of electoral competition as incumbent and opposition parties sought to capture or maintain popular support. Quantitative evidence has often supported the notion that democracy is positively associated with increased social spending and resultant human capital (Baum and Lake 2001; Przeworski et al. 2000; Avelino et al. 2005). It should be noted that non-democratic regimes have also been linked to certain types of welfare policies (Esping-Andersen 1990; Magaloni 2006; Desai et al. 2009). Non-democracies have implemented social and welfare policies that are highly 47 variable in scope of coverage and the degree and types of benefits (Carnes and Mares 2009). Scholars often suggest that non-democracies institute welfare policies so as to ensure regime survival (Haggard and Kaufman 2008). Leaders in non-democratic states may institute welfare policies as a means of cooptation in order to target critical groups of supporters – those who both support the regime and who would destabilize the regime if their support was retracted - with benefits and provide credible commitments regarding the continuation of said benefits (Knutsen and Rasmussen 2018). Autocratic leaders may also implement and utilize welfare policies to distribute benefits to the broader public to shield them from economic insecurity and ensure support during economic downturns (Han 2021). As regards sub-Saharan Africa, democratization and competitive elections have also been linked to social policy expansion and improved basic service provision in the areas of health and education. This notion is based on the fact that developing countries often have broad segments of population that are low-income; in this context, social policies will be a prominent concern for citizens and become key issues for politicians operating in democratic rules who are seeking to create electoral support (Carbone 2011). Studies citing the influence of democracy on social spending and social policy in sub-Saharan Africa often see competitive elections and the consequent need for broad-bases of support as significant drivers of expansion (Carbone 2011; Stasavage 2005a). In sub-Saharan Africa, notions of broad winning coalitions and policy-driven campaigns are often complicated by relatively recent democratic transitions, as well as low information, and low credibility environments (Keefer 2007). Grepin and Dionne (2013) evaluate the conditions under which governments in sub-Saharan Africa implement universal health policies. Grepin and 48 Dionne suggest that a democratic transition and the mere implementation of democratic institutions does not directly incentivize political leaders to propose and implement universal social policies in sub-Saharan Africa. These authors argue that along with democratization, citizens’ perceptions regarding the extent of democratic competition and the degree to which electoral competition is meaningful determine whether or not governments, in pursuit of broad electoral support, implement and adopt health policies that benefit large segments of the public. In contexts where democratization has nominally occurred but electoral competition is fragmented and citizens have low perceptions of democracy, governments are more likely to implement more targeted health policies. The influence of democratization on the implementation and expansion of social policies in sub-Saharan Africa has often focused on how electoral candidates garner support in these low-information environments, as well as how voters condition their support in this same context. Harding and Stasavage (2014) suggest the role democratization plays in influencing service provision lays in the degree to which electoral promises can be verified by voters – promises made regarding certain state policies are more verifiable than others, and it is in these policy areas that provision is expanded under democratic elections. These authors find that African democracies have higher rates of primary school attendance than non-democracies and that the abolition of primary school fees in African states is particularly likely in the aftermath of competitive presidential elections, yet democracies were no more likely than non-democracies to be associated with increases in other education inputs (e.g. hiring more teachers). In this case, democratization leads to social policy expansion on dimensions under which outcomes can be clearly linked to executive action - the abolition of user fees. In contrast, 49 democratization has little impact on policy dimensions in which outcomes are less attributable to executive action (hiring more teachers), owing to implementation problems in low-capacity environments. Researchers have further noted the potential for competitive elections in Africa to lead to expanded service provision specifically for majority populations. This notion is in-line with theories relating democratic development and electoral competition to candidates’ pursuit of broad-based electoral support. Stasavage (2005b) argues that political leaders subject to electoral competition are more likely to favor the social policy preferences of rural majorities, such as primary education spending. Harding (2019) sees further evidence of the pro-majority effect of electoral competition in finding that competitive elections significantly increase access to primary education and lower infant mortality rates for children in rural areas in sub-Saharan Africa, a relationship that is conditional on urbanization. Beyond the influence of democratic institutions and elections, the influence of external actors and ideas is often cited as an additional important driver of social policy expansion in the developing world and sub-Saharan Africa (Lavers 2019; Lavers and Hickey 2016; Nino-Zarazua et al. 2011). Indeed, in light of the social costs of structural adjustment programs, the World Bank launched the Social Dimensions of Adjustment program in the late 1980s, which espoused the use of safety nets and social insurance to address the negative externalities associated with structural adjustment and enhance welfare (Adesina 2011). External forces which can lead to social policy developments revolve around the ideas and practices espoused by donors, international institutions, and the development community, as well as the financial resources they offer. Donors have often played key roles in the design and implementation of social protection policies that 50 governments have later institutionalized and administered. Development institutions have been involved in the implementation and design of social policies such as cash-transfer programs in Ethiopia and Zambia (Lavers 2019; Adesina 2011), and health insurance schemes in Rwanda, Ethiopia, and Nigeria (Chemouni 2018; Lavers 2016; Onoka et al. 2015). Beyond technical and policy-design assistance, external monetary aid has also been linked to the expansion of social and welfare-enhancing policies in the developing world (Mosley et al. 2004; Morrissey 2009). Aid is often aimed at poverty reduction via the improvement of educational, health, and other social services. Donor aid is often utilized through budget support instruments, which provide additional funds to social protection schemes (Dani 2008). Quantitative research in this frame often finds positive links between external aid and state welfare expenditures (Morrissey 2009; Gomanee et al. 2005). In contrast to this notion, additional authors suggest the potential for sector-specific external aid to actually decrease the likelihood of welfare expansion, as governments are incentivized to divert revenue and effort elsewhere while donors assume welfare provision functions (Berens 2015). In terms of the expansion of social policy in Ghana, the implementation of the NHIS is broadly related to the influence of democratic institutions and elections. The transition to democracy in the early 1990s provided a means through which public healthcare reform could take place. A broad literature exists on the potential for electoral competition to elevate the capacity of citizens to influence policy makers by holding them accountable for policy performance; competitive elections create incentives for politicians to provide public goods (Brown and Hunter 1999; Bueno de Mesquita et al. 2002; Stasavage 2005; Lake and Baum 2001; Nelson 2007). In Ghana, the influence of voters and civil society on 51 political leaders and policymakers was increasing election-by-election before the 2000 contest (Graham et al. 2017; Gyimah-Boadi 2009). Popular demands for the reformation of the public health sector affected electoral competition between the NPP and the NDC; healthcare policy became a key issue in which the opposition (NPP) could confront the ruling party (NDC) (Carbone 2011). Indeed, the NPP did just that when it unseated the NDC in 2000 by proposing to address a particularly salient issue among Ghanaian citizens (Wahab 2019). Ghanaian MPs were constantly reminded of the public’s dissatisfaction with the existing healthcare system, as citizens were increasingly demanding that their representatives pay for their healthcare expenses themselves (Wardle 2008). As a result, MPs often gave cash handouts to cover their constituents’ healthcare costs, thereby conferring further incentives for them to support healthcare policy reform and the NHIS legislation (Lindberg 2010). At the structural level, it has also been suggested that the law- making powers of Ghana’s executive branch were pivotal in enabling the passage of the NHIS bill (Assensoh and Wahab 2008). 2.2 THE IMPLEMENTATION OF THE GHANAIAN NHIS In both the 1996 and 2000 general elections, the New Patriotic Party (NPP) sought to differentiate itself from the then-incumbent National Democratic Congress (NDC) by extensively campaigning that an NPP government would bring an end to the “cash and carry” system overseen by Jerry John Rawlings (Assensoh and Wahab 2008). Indeed, in 1996, the NPP manifesto called the healthcare system under the NDC “callous and inhuman” and promised to bring a “promising and equitable healthcare financing system” to the Ghanaian people through a national health insurance scheme (NPP Manifesto 1996: 52 36-37). The “cash and carry” system was enacted as part of the International Monetary Fund and World Bank-promoted Structural Adjustment Program adopted by President Rawlings in the mid-1980s to reduce state expenditures. In particular, the Hospital Fees Regulation of 1985 greatly increased out-of-pocket payments, as it sought to recover 15% of recurring costs (Carbone 2010). Under this much-maligned system, healthcare utilization in Ghana fell, as citizens could not afford to pay requisite fees at the point of delivery (Aryeetey and Goldstein 2000). Given these conditions, many Ghanaians sought handouts from political leaders to cover healthcare costs (Lindberg 2010). During this period, many turned to self-medication and herbal medicine practitioners for their healthcare needs (Asenso-Okyere et al. 1998). Rawlings attempted to address the public’s concern about user fees by announcing in a 1997 presidential address that fee exemptions would be expanded, yet this measure proved unimpactful, and public concern over the healthcare system persisted (Carbone 2011). It is in this context in which, in the leadup to the 2000 general elections, the opposition NPP made the reduction of healthcare expenses via the introduction of a national health insurance scheme a focus of its campaign (Gros 2016). In particular, the NPP sought to repeal and replace the existing cash-and-carry system with a national insurance scheme. In 2003, after its victory in presidential elections and its securing of a near-majority in the Parliament, the NPP government passed the National Health Insurance Act (Act 650), which was signed into law by President John Kufuor after the NPP took the presidency (Assensoh and Wahab 2008). Implementation of the NHIS began in early 2005 with the goal of increasing access to quality healthcare through lowering out-of-pocket expenses and other associated healthcare costs. While the NHIS was initially championed by the NPP, 53 both the NPP and NDC now include the NHIS as a primary social policy in their manifestoes. Initially, the NDC was determined to fight NPP policy regarding the NHIS within Parliament – prior to the 2000 election, the NDC had proposed a healthcare plan involving a “mix of insurance schemes” that would be coupled with an “improved” cash-and-carry system (NDC 2000). Yet gradually, upon the induction of the NPP government and introduction and implementation of the scheme, the popularity of the NHIS eroded the NDC’s resistance to the point that both parties are now in favor of the program, though their preferences for the actual administration and financing of the scheme differ (Assansoh and Wahab 2008). 2.3 THE DESIGN OF THE GHANAIAN NHIS Similar to national health insurance schemes in developed countries, the NHIS nominally provides members with a basic package of care at nationally accredited hospitals and clinics (both public and private). Nominally, the NHIS covers 95% of disease conditions (NHIS 2020). In addition to the following services, all medical emergencies are covered, as well as some dental and eyecare services. Outpatient services covered include: • Treatment for acute infections (malaria, typhoid, respiratory infections, ulcers, etc.) • Treatment for hypertension, diabetes mellitus, asthma • Laboratory services (x-rays, ultrasound scans) • HIV/AIDS treatments; antiretroviral drugs are not covered 1 • Outpatient surgery • Prescription medicines included on the NHIS Medicines List 1 Antiretroviral drugs are heavily subsidized and administered by a separate program under the Office of the President – the Ghana AIDS Commission 54 • Antenatal care Inpatient services covered include: • General/specialist in-patient care • Laboratory services (x-rays, blood tests, ultrasound scans) • Cervical and breast cancer treatment • Cancer diagnoses • Surgical operations • Accommodation in general ward • Prescriptions medicines included on the NHIS Medicines List • Deliveries and post-natal care The National Health Insurance Act requires all Ghanaian citizens to belong to a health insurance scheme, but membership is optional for non-formal sector (e.g. government) workers, who comprise the vast majority of the population (Witter and Garshong 2008). All but the poorest Ghanaians pay yearly premiums at NHIS district offices, with premium prices stratified by income level. As of 2020, premium prices ranged from a minimum of 7.2 Ghanaian cedis (1.24 USD) to a maximum of 48 cedis (8.25 USD) (NHIS 2020). In 2016, average premium prices were 35 Ghanaian cedis (12 USD) (Gros 2016). Enrollees fall into two categories, informal and exempt, with only the informal group required to pay premiums – enrollees are interviewed upon registration, and exempt status is determined by NHIS staff (NHIS 2020). The scheme provides premium exemptions for citizens aged 70 and above, formal sector employees and other individuals who contribute to the Social Security and National Insurance Trust (SSNIT), SSNIT pensioners, 55 pregnant women, and Livelihood Empowerment Against Poverty (LEAP) beneficiaries (Jehu-Appiah et al. 2012).2 Non-enrollees largely have no health insurance, with a small number of Ghanaian citizens possessing private health insurance (NHIA 2015). The yearly premium amount a Ghanaian citizen is required to pay is determined by the broad classification system seen below, with the lowest categorization (“core poor”) being exempt.3 TABLE 2.1 NHIS INFORMAL SECTOR CLASSIFICATION 4 Social Group Class Definition Adults who are unemployed A Core Poor and receive no identifiable income and therefore are unable to support themselves financially. Adults who are unemployed, B Very Poor but receive identifiable and consistent financial support . Adults who are employed, but C Poor receive low returns for their efforts and are unable to meet basic needs. Adults who are employed, and D Middle Income receive incomes which are just enough to meet basic needs. Adults who are able to meet E Rich their basic needs and some of their wants. Adults who are able to meet F Very Rich their basic needs and most of their wants. 2 LEAP is a pro-poor social program administered by the central government. 3 Classifications determined at enrollment interview. Per the Ministry of Health, “Different methods of classification may be adopted depending on local circumstances.” (Pg. 14). 4 Ghana Ministry of Health. 2004. National Health Insurance Policy Framework for Ghana. Accra: Ministry of Health 56 Additionally, NHIS members must re-enroll on a yearly basis at NHIS district offices to ensure good standing. Non-exempt individuals must also pay yearly fees upon re- enrollment. Registration and re-enrollment fees stood at 30 cedis for those aged 18-69 years old as of 2019 (Kipo-Sunyezhi et al. 2019). Individuals are given NHIS membership cards with relevant information after enrolling or re-enrolling (NHIA 2015). Apart from expectant mothers and children under five years of age, new subscribers (i.e. those not renewing) serve a three-month waiting period before they can utilize their NHIS card to access care (NHIS 2020). In 2014, the most recent year for which official figures have been released, 40% of Ghanaian citizens possessed an NHIS card (NHIA 2015). 5 The NHIS is operated and managed by the National Health Insurance Authority (NHIA). Among other duties, the main responsibilities of the NHIA are registering enrollees, issuing NHIS cards to enrollees, accrediting providers, and managing the National Health Insurance Fund (NHIF). The NHIF funds the NHIS and generates revenue from the following sources:6 • National Health Insurance Levy: 2.5% levy on goods and services collected under the Value-Added Tax.7 • 2.5% of Social Security and National Trust (SSNT) monthly contributions • National Health Insurance Authority (NHIA) investment dividends 5 See “Enrollment” section below for details on 2015-2018 enrollment. 6 Christmals et al., 2021. “Implementation of the National Health Insurance Scheme in Ghana: Lessons for South Africa abd Low- and Middle-Income Countries,” Risk Management and Healthcare Policy. 13: 1879- 1904. 7 In March of 2021, the Majority Leader in parliament announced that this levy will soon increase to 3.5%. 57 • Premiums paid by enrollees 2.4 NHIS ENROLLMENT NHIS enrollment has evolved over time as data collection techniques and modes of registration have been updated. Data on NHIS enrollment was collected by the NHIA and published in yearly reports from 2005 to 2014; these reports include the portion of the population that are active members (citizens with up-to-date NHIS status), as well as the specific number of active members enrolled in each of Ghana’s geographic regions. These data also include categorical profiles of active members (whether enrollees are in the formal sector, informal sector, pensioners, under 18 years of age, etc.) – though the member profile data is only measured in percentages. Enrollment data can be divided into two time periods: 2005-2009 and 2010-2014. As seen in the table below, there is a significant reduction in enrollment in data derived from the 2010 NHIA Annual Report. Prior to 2010, active enrollment was calculated by subtracting the number of all expired NHIS cards from the implementation of the scheme from the sum of all NHIS cards issued and renewed since the inception of the scheme (NHIA 2010). This calculation is likely to have inflated enrollment numbers, as it does not account for individuals who had engaged in multiple registrations, and enrollees who died while active members; additionally, the number of individuals with expired NHIS cards was not accurately tracked during this time period (NHIA 2010). The methodology used in the 2010-2014 enrollment figures is based on the sum of the number of newly registered members and the number of renewals made in a given year – this new methodology also 58 incorporated ICT, whereas the 2005-2009 data was based on the submission of manual reports (NHIA 2010). TABLE 2.2 NHIS ACTIVE ENROLLMENT: 2005-20098 Year Active Enrollment Percentage of Population 2005 1,348,160 6.31 2006 3,867,862 17.68 2007 8,184,294 36.56 2008 12,518,560 54.66 2009 14,511,777 61.97 TABLE 2.3 NHIS ACTIVE ENROLLMENT: 2010-20149 Year Active Enrollment Percent of Population 2010 8,163,714 34.00 2011 8,227,823 32.40 2012 8,885,757 34.18 2013 10,145,196 38.12 2014 10,888,000 40.00 Enrollment data for 2015 to the present is currently unavailable. This is likely due to two important developments during this time period. In the mid-2010s, the NHIS nominally began issuing biometric cards at the point of registration; prior to this there was often a waiting period (anywhere from several weeks to several months) before a new enrollee acquired a membership ID. However, the transition to the new NHIS card has been slow and fragmented, with some new enrollees still waiting months to get their cards (Thiel 2020). Additionally, the NHIS began piloting mobile phone-renewal systems around this same time period, with mobile renewal officially launched in 2018. Most researchers describe the enrollment period between 2015-2020 as “static” (Christmals and Aidam 8 Sources: National Health Insurance Scheme Annual Report. 2005; 2006; 2007; 2008; 2009. Accra: National Health Insurance Authority 9 Sources: National Health Insurance Scheme Annual Report. 2010; 2011; 2012; 2013; 2014. Accra: National Health Insurance Authority 59 2020; Nsiah-Boateng and Aikins 2018). Bolstering this notion, in an interview conducted with the NHIA Director for Policy Monitoring, Research, and Evaluation in 2019, 36% percent of the population, or roughly 10,951,200 individuals, were actively enrolled in the scheme.10 While it is difficult to analyze enrollment for the years 2005-2009, and thus to establish trends, it is likely that enrollment in the scheme experienced a significant increase in the first few years of the program. The literature on voluntary participation in schemes such as the NHIS suggests that individuals are often more willing to join these programs in the initial stages, as they are hopeful that they will accrue promised benefits (Nsiah- Boateng and Aikins 2018). Once this point is reached, apathy will often set in if citizens’ expectations are not met in regards to benefits – this notion may be gleaned from the more recent data. At the systemic level, researchers have found that poor quality services, long wait times at registration centers, and shortages of medical supplies are associated with low enrollment and renewal rates in social insurance schemes such as the NHIS (Dror et al. 2016; Mladovsky 2014). In the following chapter, interview and survey data from both enrollees and non-enrollees address some of these concerns in further detail. 2.5 EFFECTIVENESS OF THE NHIS The goal of the NHIS program is to improve healthcare access and health outcomes by providing financial risk protection against the cost of standard quality healthcare for all Ghanaian residents (NHIA 2010). In terms of effectiveness and success, the important aspects of a healthcare policy such as the NHIS center on healthcare access, health 10 Author interview. June 2019. National Health Insurance Authority Head Office. Accra, Ghana. 60 outcomes, enrollment, and finances (provider payment and program financial stability). Broadly, the NHIS has proved successful in positively impacting healthcare access and healthcare outcomes in Ghana. FIGURE 2.1 HEALTHCARE UTILIZATION IN GHANA 2005-2013 (MILLIONS OF VISITS)11 As seen in figure 2.1, outpatient service utilization has experienced a significant increase in since 2005, when the NHIS program was implemented. Inpatient services have seen a slight increase over the same period. Inpatient service utilization experienced a slight increase during this period. A relatively stable frequency of inpatient services present along with growth in outpatient services is often indicative of the benefits of preventive care – as more citizens are able to access preventive care, inpatient visits should be relatively in-line with population growth. In terms of healthcare improving healthcare 11 Sources: National Health Insurance Scheme Annual Report. 2005 - 2014. Accra: National Health Insurance Authority 61 access, the NHIS has proven effective; NHIS members acquire more consistent preventative care than non-members (Mensah et al. 2010). In terms of health outcomes, the NHIS has also proven positively influential in certain areas. Insured women are more likely than uninsured to give birth in hospital and receive prenatal care (Acharya et al. 2013; Mensah et al. 2010). Evidence also demonstrates that enrollees are more likely than non-enrollees to obtain prescriptions and seek formal care when they are sick (Blanchet et al. 2012). In light of these successes involving healthcare utilization and access, increased demand has impacted healthcare service quality (Fusheini 2016). While utilization has increased, there has not been a corresponding increase in health infrastructure, human resources, and medical equipment (Alhassan et al. 2015; Mohammed and Seidu 2014). While healthcare utilization and aggregate population health outcomes have improved under the NHIS, in order to achieve universal healthcare coverage, it is essential that citizens register, and continuously re-enroll. Yet data from recent years has indicated that active enrollment may have reached a plateau in the mid 2010s. The aforementioned data constraints do not allow for an adequate analysis of enrollment figures, but there are potential reasons as to why enrollment targets have not been met. The premium price and enrollment fees for non-exempt individuals have been suggested as a barrier to enrollment for some individuals (Christmals and Aidam 2020), while the difficulty in determining individuals as exempt adds to this problem (Aryeetey et al. 2012). The quality of services available through the NHIS also likely plays a role. Jehu-Appiah et al. (2011) found that both non-renewal and not registering were associated with poor perceptions of the quality of care and drugs received at NHIS accredited care centers. The subsequent chapter 62 elucidates and expounds on many of the reasons for enrollment and non-enrollment via interview data. Financially, the sustainability of the NHIS program was in question from the onset of the program. With its (nominally) generous benefits package and low user fees, concerns have continuously been voiced over provider payment. A successful health insurance scheme must reimburse providers in a timely manner to earn the trust of said providers and ensure that they are willing to provide standard quality care to enrollees. In this regard, the NHIS has run at a deficit every year since 2009, as the growth of claims expenditures has outpaced NHIS revenue, significantly delaying reimbursement and thus NHIS operations (World Bank 2017). Multiple studies suggest that delays in reimbursement have been a significant obstacle in assuring enrollees receive quality care at accredited facilities, and that clinics and hospitals have the means with which to provide quality care (Sakyi et al. 2012; Owusu-Sekyere and Bagah 2014; Gros 2016). In 2015, claim payments were 1.07 billion Ghana cedis (183,323,002 USD), while 300 million cedis (51,398,972 USD) were in arrears (World Bank 2017). The next chapter describes in detail how delays in reimbursements greatly impact the attitudes of both enrollees and non- enrollees towards the scheme. 2.6 THE NHIS IN CONTEXT The NHIS is often considered as one of the leading and most successful examples of national health insurance schemes in sub-Saharan Africa, yet other countries in the region have sought to achieve universal health coverage through similar schemes as well, or plan to do so in the future (Gros 2016; Blanchet et a. 2012). Beyond Ghana, the most prominent 63 and affective national insurance schemes are found in Rwanda, Gabon, Burundi, Kenya, and Tanzania. Newly implemented and/or piloted schemes, with far lower levels of enrollment, can be found in Ethiopia, Nigeria, and Senegal amongst others (Chemouni 2018). National health insurance schemes in sub-Saharan countries can broadly be divided into two types: social health insurance schemes (SHI) and community-based health insurance schemes (CBHI). Social insurance schemes are managed by a public organization, with funding generated through general taxation and mandatory contributions from certain groups (non-exempt enrollees). CBHI programs are voluntary and much more decentralized in nature, as they are controlled by individual communities and entail pre-payment for healthcare services along with government contributions (Chemouni 2018). The World Health Organization (WHO) has often touted CBHI programs as most suitable for developing contexts with low tax-bases. Insurance schemes in Gabon, Burundi, Tanzania and Kenya are most similar to the Ghanaian system in that they are social insurance schemes; whereas the most predominant CBHI scheme is in Rwanda. The only sub-Saharan country with substantial rates of enrollment in private insurance schemes is South Africa. 12 Enrollment in the aforementioned schemes is varied. Among countries with established SHI schemes, Ghana and Gabon have far higher rates of active enrollment (30- 40% population coverage) than those in Kenya, Burundi, and Tanzania (10-20% population coverage) (Chemouni 2018). Apart from Rwanda, countries utilizing CBHI programs often have enrollment rates falling below 10% population coverage. Far and away, Rwanda is the country with the highest rate of active enrollment on the continent, 12 South Africa has long sought to address disparities in health equity with a national insurance scheme. The South African NHI scheme is currently being implemented in phases over the course of 14 years and is expected to be fully implemented by 2026. 64 with 81.6% of the country enrolled in CBHI schemes, according to the Rwanda Social Security Board (RSSB 2016). The CBHI scheme in Rwanda has often been recognized as a rare success story for programs of this type in sub-Saharan Africa (Allegri et al. 2009). Part of the reason for this significantly high enrollment level, beyond the fact that the program has proven relatively successful in Rwanda, is the fact that the Rwandan government made enrollment in CBHI schemes compulsory in 2006 (Chemouni 2018). In terms of increasing access to healthcare, the Rwandan CBHI scheme has proven successful, with the average citizen accessing healthcare facilities at a rate of 1.1 times per year in 2015 (RSSB 2015). Though Ghana and Rwanda are often hailed as success stories when it comes to increasing access to care through state health insurance schemes, nearly all of the aforementioned countries’ healthcare systems face financial sustainability problems, shortages of supplies, and lack of investment (Gros 2016). Despite this, the WHO continues to urge governments to implement and develop social insurance schemes as a means through which to increase population health in the sub-Saharan region. 2.7 THE NHIS AS A FORM OF PROGRAMMATIC DISTRIBUTION With this understanding of the NHIS as a significant healthcare policy in Ghana, both in terms of politics and in practice, I now describe the NHIS as a distinct mode of programmatic distribution that is significantly different from those previously studied in the African politics literature. As described in the theoretical section of the previous chapter, the characteristics of national health insurance schemes, such as the NHIS differentiate these programs from other modes of public goods provision in sub-Saharan Africa. The NHIS is a type of programmatic policy, and thus the receipt of benefits is non- 65 conditional. In the case of the NHIS, all individuals enrolled in the scheme have the potential to benefit from the policy, regardless of which party or politicians they support. Second, the benefits associated with the NHIS are deferred. The public good that the NHIS offers to enrollees (healthcare) is one which is to be acquired in the future. Lastly, direct payment is required of most enrollees in the form of yearly premiums and registration fees. In this sense, goods are pre-purchased by most enrollees. In addition to premiums and registration fees, all Ghanaian citizens contribute to the NHIS through a specific tax, the National Health Insurance Levy. These traits differentiate the NHIS from other forms of government distribution; this section focuses on the nature of these characteristics within the NHIS. When a policy is programmatic, political incumbents have little say as to the delivery of benefits, this denotes the non-conditional nature of goods delivered through this type of distributive mechanism. Citizens are able to acquire benefits based on objective, publicly stated, criteria (Kitschelt and Wilkinson 2007); programmatic policies provide access to goods that cannot be taken away if an individual does not support a particular political leader. A policy is programmatic if its goods distribution is formalized, according to a certain set of rules for those in a defined category (Stokes 2007). In the case of the NHIS, the policy distributes goods (healthcare) to those in a defined category (enrollees) based on a certain set of rules (registration and premium payment), regardless of political affiliation. Whereas any individuals can access healthcare at a public clinic or hospital in Ghana, only NHIS enrollees acquire the benefit of free care for certain products and services. 66 The distribution of goods through a national health insurance scheme also differs from other forms of distribution discussed in the literature as concerns time horizons. Goods provided through an insurance scheme are deferred; citizens enroll in a particular program and expect that specific benefits will be provided in the future. In the case of the NHIS, this notion is strengthened by the fact that enrollees must wait three months to utilize their insurance card once it is in their possession – if an individual is in ill health and not registered with the NHIS, they are highly unlikely to go to the registration office, apply for and acquire their card, then wait three months to utilize it to acquire healthcare. Further, the time it takes to actually acquire the NHIS card is highly variable. Nominally, once individuals have their NHIS card they are able to utilize the program after a three- month waiting period. However, some NHIS members have reported difficulties in obtaining the physical card, which then delays the start of the three-month waiting period. Some individuals in this study described how they registered at a particular district office and were then told to return at a later date to receive their physical card (author interview 2019). This notion demonstrates how the NHIS is dependent on the belief among citizens that future healthcare services will be provided. The payment structure of the NHIS ensures that most enrollees (those deemed non- exempt) pre-pay for services directly in the form of yearly premiums and registration fees. Though direct premium payments are nominally stratified by income, in many cases a flat premium is charged due to the difficulties in assessing household income levels (Kwarteng et al. 2019). In relation to exemptions, some studies have also found that exempt individuals have reported paying premiums to register for the NHIS (Kwarteng et al. 2019). As previously noted, Ghanaian citizens also contribute to the NHIS through the 2.5% 67 National Health Insurance Levy (NHIL) tax. Though all Ghanaians pay the NHIL, the majority of enrollees use their own resources to contribute directly to the NHIS through registration fees and yearly premiums. As mentioned in the theoretical chapter, these design characteristics engender beliefs regarding goods that will be provided in the future. The next chapter evaluates the relationship between these design characteristics and citizens’ expectations regarding the performance of the NHIS. Do citizens believe that the goods proffered by the scheme are of the quality they expect? If the services provided by such a policy are below the expectations of citizens, how does this influence how they view the program and the incumbent government? To preface the findings described in the next chapter, the citizens included in this study have formed lofty expectations about the goods that the NHIS should provide (completely free healthcare), while also stating their concerns that they are being taxed and contributing premiums to pay for goods and services they believe are not being adequately delivered. The fact that the goods associated with the NHIS are not immediately available to citizens and are goods for which they have pre-paid, both through taxation and direct contributions via premiums, gives citizens the perception that they have done their part, and the government must now fulfill its promise. Under meaningful democratic conditions, citizens who have high expectations regarding future benefits, and are monetarily invested in a government policy are likely to hold political leaders accountable for the inadequate production of expected goods. 68 2.8 CONCLUSION Throughout its existence, the NHIS has been a highly salient policy for both citizens and political leaders, both politically and in terms of population health. Prior to its implementation, the proposal of the soon-to-be NHIS program arguably won the election for the NPP, and to this day politicians and policymakers from both the major Ghanaian political parties espouse the preservation and expansion of the program as means to achieve universal health care. For citizens, the hardships experienced under the cash-and- carry system in the 1980s and 1990s are not soon forgotten, and they have responded in kind by enrolling in the scheme at a substantial rate. In turn, the program has improved health outcomes for Ghanaian citizens and increased access to care. Though the program has and continues to experience problems of financial sustainability and quality care, from the view of improving population health, the NHIS has proved impactful. The characteristics of the NHIS policy – that it is a programmatic means of distribution, that it is funded through contributions and a specific tax, and that the goods it offers are associated with future time horizons make this policy novel and distinct from other forms of goods provision evaluated in the literature on distributive politics. These policy characteristics elevate the importance of the views and expectations of citizens in determining the success or failure of the NHIS. Yet indicators of policy outcomes, rates of enrollment/re-enrollment, and distributive characteristics are not the entire story. Ghanaian citizens have real experiences with the policy, how it performs, and how it is administered. The following chapter seeks to answer the questions posed in the previous section utilizing interview and survey data to in an effort to discern how Ghanaian citizens view and respond to the NHIS program, with 69 particular attention focused on the expectations citizens have in regards to the final goods produced by the policy. 70 CHAPTER 3: CITIZEN PERCEPTIONS OF THE NHIS Whereas the previous chapter described the implementation, administration, and effectiveness of the NHIS from top-down and structural perspectives, this chapter describes the views of those who are perhaps most familiar with the administration and effectiveness of the policy: ordinary Ghanaian citizens. The goal of this chapter is to determine whether the design characteristics of the NHIS which were argued to be influential in engendering performance evaluations are indeed impactful in this regard. Utilizing in-depth interviews and survey data, this chapter explores multiple areas of enquiry. What are citizens’ expectations regarding the benefits the NHIS distributes? Has the NHIS lived up to citizen expectations regarding access to quality healthcare? Given that most enrollees pay for NHIS services, has the government held up its end of the bargain in providing said services? If the services provided by such a policy are below the expectations of citizens, how does this influence how they view the program and their government? The theoretical discussion in chapter two described aspects of national health insurance schemes which structure expectations related to the goods they offer – their association with deferred benefits and direct contributions in the form of premiums and specific taxes. This chapter highlights the influence of these factors in molding enrollees’ expectations regarding the goods that they expect the NHIS to deliver, and how these expectations compare to the actual goods the program produces. As I will discuss throughout this chapter, though the NHIS has at times positively impacted the citizens included in this study, there are serious problems with the program that negatively impact the views of citizens (both enrollees, and non-enrollees) as they pertain to the NHIS and public healthcare system in Ghana. Multiple themes emerge 71 regarding the views of those participating in this study. Most important for the theoretical expectations developed in chapter 3, there is a broad disconnect between what enrollees believe they are entitled to, what the government says the NHIS provides, and the final product the program produces for enrollees. The government has told citizens that if they pay their registration fee and yearly premium, the broad services the NHIS nominally provides should come at no charge; however, the data describe broad variations in exactly what goods and services can be acquired for free at points of care. Because of this, many citizens believe the NHIS program is not working as intended, as it covers but a small portion of their healthcare costs – often the least expensive goods and most minor services; citizens expect it to cover virtually everything. Second, there is a strong perception of inferiority among enrollees in acquiring quality care vis-à-vis non-enrollees. As described in detail below, nearly all enrollees perceive they are at a disadvantage when they utilize their NHIS card to acquire care and believe the care they receive is of lower quality than non-enrollees’. Additionally, many NHIS members report being stigmatized vis-à-vis non- enrollees at points of care and often report being asked by caregivers to pay cash if they want to receive adequate services. Lastly, many enrollees and non-enrollees are aware of the program’s difficulty in reimbursing hospitals. There is a strong sense among individuals in this study that program administrators, doctors, and nurses are not at fault for the current state of the program; respondents believe political leaders have allowed the program to faulter and underperform. As suggested in the theoretical section, this fact is bolstered by the views of respondents that they have paid into the NHIS system through premium contributions, registration fees, and the NHIL tax. 72 The remainder of this chapter describes the methodology with which the survey and interview data were collected within the Accra Metropolitan Region, as well as the demographic profiles of enrollees and non-enrollees. Thereafter, qualitative and survey data highlight and expand upon the themes mentioned above with reference to the theoretical expectations described in chapter 1. 3.1 RESEARCH DESIGN An original oral survey with embedded vignette experiment, as well as semi- structured interviews (conducted by an enumerator in the respondent’s preferred language – Twi, English, Fante, and Hausa), was conducted throughout the Accra Metropolitan Region of Ghana from June to September 2019. Over the course of research, 150 1-2 hour interviews were carried out with citizens within the Accra Metropolitan Region in five randomly selected districts, incorporating nine individual municipalities and villages.13 In addition to these interviews, elite-level interviews were also conducted with directors and administrators at the NHIA head office in Accra, as well as at NHIS district offices. The survey was executed using a random-walk strategy beginning in a central location (police station, post office, place of worship, etc.) within a specific town or locality in each selected district within the Accra Metropolitan Region. From this central point, the enumerator would proceed to sample every third home, business, or shop and and ask potential respondents for their permission to be included in this study. Over the course of 13 Districts included: Accra Metropolitan, Ayawaso East, Adenta, Ga East, La Nkwantanang Madina, and Shai Osudoku. 73 this research study, 157 individuals were selected for inclusion in the study, and seven individuals declined to take part, resulting in 150 study participants. Localities included in this study range from urban impoverished and informal settlements (Nima, Accra New Town, Adabraka), to working- (Adenta, Dome, Madina) and middle-class (Legon) urban areas, as well as rural areas outside of Accra (Dodowa).14 In total, 80 individuals were interviewed in urban areas; with ten individuals being interviewed in eight separate localities within four districts within the Accra Metropolitan Region. To account for variation among urban and rural respondents, 70 respondents resided in rural areas near the small town of Dodowa in the Shai Osudoku district, forty to fifty kilometers from the Accra city center. The sample is random, but not designed to be representative of the Accra Metropolitan Region or Ghana. The original survey contained 50 closed-ended questions broadly pertaining to beliefs and perceptions related to the NHIS, healthcare utilization and quality of care, government performance, and Ghanaian politics. In addition to these 50 questions, three open-ended questions were included that focus on perceptions of the NHIS, what it provides, and what it should provide. Lastly, the survey contained a vignette experiment, which is described in the next chapter.15 The demographic profile of the respondents included in this study can be seen in Table 3.1. 14 Localities included: Adabraka, Adenta, Ayawaso, Dodowa, Dome, Legon, Madina, Accra New Town, Nima 15 See full survey in Appendix A for specific survey questions and the vignette experiment. 74 TABLE 3.1 DEMOGRAPHIC PROFILE OF STUDY PARTICIPANTS Gender Sample (%) N Male 39 59 Female 61 91 Age 18-25 13 20 25-30 7 11 30-35 22 33 35-40 18 27 40-45 11 17 45-50 13 19 50-55 9 13 55-60 2 3 60+ 5 7 Residence Urban 53 80 Rural 47 70 Education No formal schooling 15 22 Some primary school 14 21 Primary school complete 17 25 Some secondary school 13 19 Secondary school complete 27 41 Some university 9 14 University complete 5 8 Occupation Agricultural/farming/fishing 2 3 Artisan/Skilled manual labor 20 30 Clerical 1 1 Homemaker 2 3 Mid-level professional 5 8 None/unemployed 5 7 Pensioner 1 1 Retail/shop 30 45 Security services 3 4 Trader/vender 27 41 Student 2 3 Unskilled manual labor 2 3 Upper-level professional 1 1 75 As seen above, urban and rural areas were surveyed at roughly equal rates, while women included in the study outnumber men. Most of the respondents also lay in the 30- 50 age group. In terms of educational attainment, 71% had completed primary school or higher, with the largest portion (27%) having completed secondary school. The occupations of study participants are highly skewed towards traders and shop owners in the informal sector. Beyond occupational classes, a lived poverty index was constructed to shed further light on individuals’ economic profiles. Respondents were asked how many times in the past year they had gone without food, water, or a cash income. Responses to these queries include “never”, “once or twice”, “several times’, or “many times.” See Table 3.2 for this economic profile. TABLE 3.2 LIVED POVERTY INDEX Gone without food Sample (%) N Never 85 127 Once or twice 14 21 Several times 1 1 Many times 0 0 Gone without water Never 88 132 Once or twice 11 16 Several times 1 1 Many times 0 0 Gone without cash income Never 61 92 Once or twice 24 36 Several times 10 15 Many times 4 6 76 In addition to these demographic profiles, respondents were also asked about their political profile – whether or not they identified with a political party, and what particular party they identified with. Of the respondents who chose to answer this question, 60 (40%) indicated that they were members of a political party. Of this partisan group, 35 (58%) identified with the incumbent NPP, whereas 25 (42%) identified with the opposition NDC. 3.2 DETERMINANTS OF ENROLLMENT Among the 150 respondents, 43% (65) were NHIS members who were currently registered in the program and considered active enrollees. The demographics of this group of study participants can be seen in Table 3.3. The fourth column represents the percent of the total study population of each demographic category that is actively enrolled in the NHIS. TABLE 3.3 NHIS ACTIVE ENROLLEES Gender Sample (%) N % of Total Male 31 20 34 Female 69 45 50 Age 18-25 9 6 30 25-30 8 5 45 30-35 20 13 39 35-40 23 15 56 40-45 18 12 71 45-50 9 6 35 50-55 6 4 31 55-60 5 3 100 60+ 2 1 50 Residence Urban 45 29 36 Rural 55 36 51 77 TABLE 3.3 (cont’d) Education No formal schooling 12 8 36 Some primary school 17 11 79 Primary school complete 12 8 32 Some secondary school 15 10 53 Secondary school complete 26 17 41 Some university 9 6 43 University complete 8 5 63 Occupation Agricultural/farming/fishing 3 2 1 Artisan/Skilled manual labor 17 11 37 Clerical - - - Homemaker 3 2 1 Mid-level professional 8 5 63 None/unemployed 5 3 43 Pensioner - - - Retail/shop 35 23 51 Security services 2 1 0 Trader/vender 23 15 37 Student 3 2 1 Unskilled manual labor - - - Upper-level professional 2 1 100 _________________________________________________________________________________________________________ As can be seen in Table 3.3, the active enrollees in this study exhibit a broad range of demographic features. The vast majority of NHIS enrollees included in this study are considered informal enrollees. As noted in the last chapter, informal enrollees are individuals who are employed outside of the state sector, and those not receiving social security payments. Per the most NHIA recent data, individuals employed in the public sector and social security recipients comprise only around four percent of the total population of enrollees (NHIA 2014). 69% of the enrollees in this study were women. The higher proportion of women enrollees may be due to the fact that pregnant women are enrolled in the scheme at no cost and receive three months of pre- and three months of post-natal care covered by the NHIS. The notion that women are more likely to be enrolled 78 also stems from the fact that children under 18 are enrolled in the scheme at no cost as well (Dake 2018; Salary et al. 2019). Additionally, a larger proportion (55%) of enrollees lived in rural areas outside of Accra. The literature is inconclusive on whether rural-dwelling individuals are more likely to enroll in the NHIS, with some finding urban dwellers are more likely to enroll (Dake 2018), and others finding rural residents more likely to enroll (Agyepong et al. 2016; Amu and Dickson 2016). To better ascertain the differences among enrollees and non-enrollees, we know look at the determinants of enrollment. Table 3.4 highlights the correlates of NHIS enrollment for the 150 individuals participating in this study. The dependent variable in this analysis (NHIS enrollee) indicates whether or not a respondent is an active NHIS enrollee. A logit model, with standard errors clustered at the district level, is utilized to evaluate several factors influencing whether or not an individual is an active NHIS enrollee. Explanatory factors include the respondent’s Age, Gender, residence (Rural), highest level of education (Education), whether or not an individual felt close to a political party (Partisan), and a lived poverty index (Poverty).16 Additionally, a second logit model was run including whether or not an individual is a member of the NPP, the party that implemented the NHIS program in Ghana. 16 For Gender, female=1. The variable Poverty is a an additive index compiled via questions regarding how often a respondent had gone without food, water, or a cash income in the previous year. Higher values denote more instances of shortages. 79 TABLE 3.4 DETERMINANTS OF NHIS ENROLLMENT NHIS enrollee NHIS enrollee Age 0.067 0.072 (0.058) (0.060) Gender 0.623 0.598 (0.415) (0.426) Rural 0.611** 0.584** (0.046) (0.041) Education 0.061 0.065 (0.075) (0.075) Partisan -0.420** - (0.108) Poverty -0.350** -0.335** (0.122) (0.113) NPP - -0.095 (0.279) Constant -0.962 -1.145 (0.504) (0.413) Pseudo R-squared 0.08 0.07 N 150 150 **p < 0.01 As can be seen is column 1 above, active NHIS enrollment within this study is associated with the variables indicating rural residence, party membership, and lived poverty experience. Rural residence is associated with an increased likelihood of active enrollment. As mentioned previously, studies are inconclusive as to whether rural residents are more likely to enroll in the program. Among the interviewees included in this study, several participants mentioned that they receive faster and better care when 80 utilizing their NHIS cards at rural hospitals and clinics as opposed to facilities in Accra, a fact that may play a role in this finding. Moreover, some participants also stated that nurses working at a clinic within the rural localities included in this study encouraged them to enroll in the scheme – something not mentioned by urban residents. In addition to rural residence, the lived poverty index is negatively associated with NHIS membership; individuals who have higher values on the lived poverty index are less likely to be active NHIS members. As the NHIS was intended to alleviate financial burdens to care, this is a troublesome finding that has also been found in other studies associated with NHIS enrollment (Salari et al. 2019; Akazili et al. 2014; Kusi et al. 2014; Agyepong et al. 2016). There are multiple reasons why poverty may be associated with a lower likelihood of enrollment. As it stands, individuals who are exempt from paying premiums (those classified by the NHIS as “indigents”) are those who are enrolled in the Livelihood Empowerment Against Poverty (LEAP) program, which provides bi-monthly cash payments to program registrants experiencing extreme poverty. If an individual is not a part of this program, they do not qualify as premium-exempt. This notion marks a change from earlier periods of the NIHS, in which community leaders would designate individuals within their communities as exempt owing to extreme poverty. These community leaders would then report individuals as exempt to NHIS/NHIA authorities. Owing to abuses within this system of exempt status designation, the NHIA moved to ensure that only LEAP members could register as indigent.17 Thus, exempt status depends on LEAP enrollment, which is determined by the Ghana Statistical Service through ranking areas of need (regions, districts, and communities) based on the Ghana Living Standards Survey. 17 Author interview with NHIA Director of Policy Planning, Monitoring, and Evaluation. June 2019. 81 Individuals within the highest-ranked areas of need are then interviewed by a third-party organization, which sends household data to LEAP to determine eligibility. 18 Based on these criteria, there are large numbers of individuals within Ghana who do not meet LEAP thresholds based on their residence. Even for those individuals who qualify for reduced premiums based on their NHIS registration interview, the yearly premium can be a high financial burden – not to mention a registration fee is also collected at the time of enrollment. As a respondent in the small market town of Dodowa stated when asked about their NHIS status: “I have never been on the scheme and have never registered and won’t. I am not working, and I cannot afford the enrollment fees. I think it is the right thing for a Ghanaian to do. Whenever I do get some money, I will enroll in the scheme.” This individual is one of many Ghanaians who may lay somewhere between the extreme poverty levels associated with LEAP assistance and the broader poor who are unable to afford even reduced NHIS fees. In addition to fees, numerous participants in this study who were not enrolled in the program claimed that they could not take the time to register at their local NHIS district office as they would miss out on vital potential income. Participants (both enrollees and non-enrollees) consistently cited the length of time involved in registering and enrolling at NHIS offices, with some individuals told to return at a later date because the NHIS office was lacking supplies. One participant in Adenta East, the capital of Adenta district, had this to say about their experience attempting to enroll: 18 http://leap.gov.gh/eligibility-criteria/ Accessed Jun 25, 2021. 82 “I do not have the card. When I came to Accra I made a couple of attempts at getting it, but I could not because of the long queue. I work so I do not have much time and I cannot be wasting time at the NHIS office trying to do it.” Another participant in Legon, despite being unwell, deemed it not worthwhile to attempt to register based on a past experience: “I am not well now, but I do not have the money to pay, even if I had the money I would not use it to get the NHIS card. The last time I attempted to, I was told that they were short equipment at the registration center. Since then I have not gone back and I will not waste my time there.” In this study, individuals who were not active members were asked why they had not enrolled in the NHIS. 35% of the respondents in this group stated that enrollment and premium fees were too high or they could not neglect their work for loss of daily income. Moreover, of the participants who were enrolled at one time but were not presently enrolled at the time of the study, 25% stated their reason for not re-enrolling was that they could not afford the fees or could not lose income to take the time to re-enroll. Beyond rural residence and poverty, we see in column 1 of Table 3.4 that the Partisan indicator is negatively associated with likelihood of being actively enrolled in the NHIS, indicating that individuals who felt close to a political party were less likely to be active enrollees. Based on this finding, column 2 of Table 3.4 includes an indicator of NPP partisanship to determine the impact of specific partisanship orientations (in this study, all individuals who identified as partisan felt close to either the NPP or NDC). However, we see that NPP members are no more likely to be active NHIS enrollees than NDC members, as 83 the variable NPP fails to achieve significance. This result may be surprising given the politicized environment in which the NHIS came to be. As mentioned in the previous chapter, both of the main political parties in Ghana support the NHIS – the program is bipartisan in nature. Survey responses reflected this. Respondents were asked whether or not both the NPP and the NDC supported the NHIS, or if only one of these parties did. 92% (138) of respondents (including both enrollees and non-enrollees) stated that both parties supported the program. Among partisan individuals, 85% (51) stated the same. Moreover, respondents were also asked if they agreed or disagreed with statement, “The NHIS does not change much regardless of which party is in power.” 65% (98) of the respondents included in the study stated that they “strongly agreed” or “agreed” with this statement. The issue of whether or not programmatic policies elicit partisan effects (i.e. whether or not certain parties or incumbent leaders are rewarded for enacting programmatic policies) has been hotly debated in the political science literature (Pop-Eleches and Pop-Eleches 2012; De La O 2015; Larreguy et al. 2015; Imai et al. 2020). However, for the individuals included in this study, it is apparent that the NHIS as a programmatic policy has moved beyond partisan politics, not only because of its programmatic nature, but because both of Ghana’s major parties support it. Having determined the factors associated with active NHIS enrollment in this study, we now briefly evaluate the relationship between NHIS enrollment and healthcare utilization. Respondents were asked if they had visited a health clinic in the previous year. 105 respondents (70%) responded in the affirmative. Table 4.5 highlights the correlates of visiting a health clinic in the past year, with NHIS enrollee the main independent variable. 84 Other explanatory factors include Age, Education, Gender, Rural residence, and a lived Poverty index. Again, a logistic regression with standard errors clustered at the district level was utilized for this analysis. As individuals were not asked about their specific health status or conditions, Age is the best proxy available for these explanatory factors. TABLE 3.5 DETERMINANTS OF HEALTHCARE UTILIZATION Visited Health Facility in Past Year NHIS enrollee 1.604** (0.412) Age 0.087** (0.036) Gender -0.089 (0.365) Rural -1.161** (0.212) Education 0.210** (0.042) Poverty -0.067 (0.110) Constant 0.148 (0.269) Pseudo R-squared 0.16 N 150 **p < 0.01 As can be seen in the table above, NHIS enrollee, Age, Rural residence, and Education are all associated, at a statistically significant level, with the likelihood that an individual within this study had visited a health facility in the year prior. As age is likely associated 85 with increased need for care, this relationship is not surprising. The negative impact of rural residence is also expected, as the rural dwellers in this study had far fewer health facilities at their disposal, and the myriad clinics and hospitals of Accra were around 40 kilometers away. Importantly, we see that NHIS enrollment is positively and significantly associated with visiting a health facility in the previous year. This finding is similar to that of the nationally representative relationship between the NHIS program and healthcare utilization mentioned in the previous chapter, though the caveat remains that some of these individuals may enroll in the scheme because they are in need of more frequent healthcare. The variable Age may capture some aspect of this notion, it is by no means a fully sufficient proxy. As described in the theoretical chapter, the characteristics of the NHIS program – the deferred nature of the goods it distributes, along with the contributions citizens make through direct premium payments and specific taxation are expected to structure expectations regarding the quality of goods the program offers and elicit performance evaluations. Having acquired this picture of the Ghanaian citizens taking part in this study, the next section utilizes interview and survey responses to link NHIS enrollees and their government by elucidating the expectation citizens have regarding the program and its benefits and how their experiences with the NHIS relate to their broader attitudes relating to political accountability. 3.3 THE NHIS – PERCEPTIONS AND EXPERIENCES We now evaluate how the individuals participating in this study view the NHIS, based on their own experiences and views of the program. Across demography and NHIS 86 membership, the sample of respondents included in this study were cognizant of the NHIS program and had strong opinions about it. As this portion of analysis proceeds, multiple themes present themselves and will be described in detail. Namely, there is a broad disconnect between what enrollees believe they are entitled to as members of the program, what the government says the NHIS provides, and the final good enrollees receive. Moreover, enrollees perceive that they receive inferior care vis-à-vis non-enrollees, as nearly all enrollees believe they are at a disadvantage when they utilize their NHIS card to acquire care. Lastly, many individuals (enrollees and non-enrollees) are cognizant of the program’s difficulty in reimbursing hospitals. There is a strong sense that, despite their association with many of the negative perceptions of the program, NHIS administrators, doctors, and nurses are not at fault for the current state of the program, it is the country’s current political leaders that have allowed the program to falter and underperform. We begin with reviewing what citizens believe they are entitled to as NHIS enrollees. There was the perception amongst respondents that NHIS membership entails completely free healthcare. Respondents were asked an open-ended question inquiring what benefits the NHIS provides to enrollees. 75% (113) of the sample stated that they believed that NHIS members were entitled to completely free healthcare. Of the individuals who stated that NHIS members are entitled to completely free care, 48% (54) were current enrollees. Thus, 83% of the active NHIS members included in this study were of the perception that they should not be paying for medical services when using their NHIS cards. As noted in the previous section, the NHIA claims that the NHIS covers 95% of the disease conditions afflicting Ghanaian citizens (NHIA 2014). The respondents in this study have broadly taken this to mean that they are entitled to free healthcare if they are NHIS 87 members. This active enrollee from Madina, La Nkwantanang Madina District describes this notion: “I went to the hospital and I paid over 180 cedis when I have health insurance. I think it is not working. The purpose of the NHIS is to access free healthcare, however this is not the case in Ghana.” Thus we see the notion that the respondents in this study broadly equate the NHIS card with free healthcare. Yet, is the NHIS actually covering what it claims to cover based on the experiences of enrollees when utilizing the insurance? Despite the laudable coverage nominally offered by the NHIS, when we look at respondents’ actual experiences using their NHIS cards, we see the many enrollees believe the NHIS is not living up to this claim. Respondents’ experiences associated with using their NHIS cards suggests wide variation in what is actually covered. Perhaps accordingly, among non-enrollees, perceptions of what services are covered by the scheme are often bleak. In addition to the interview question concerning what NHIS members are entitled to, respondents were also asked about their experiences using their NHIS cards – the care they received and the services and medications they were provided. Though the NHIA claims that the NHIS covers 95% of disease conditions, many respondents reported that the scheme failed to cover many aspects of their care. Individuals were asked what they think the NHIS should cover that it currently does not based on their own experiences. A wide variety of responses were reported. The vast majority of respondents reported that, though they expect them to be provided free of charge, what they deemed “expensive drugs” were not covered by the scheme. Though these may be drugs that are simply not covered by the scheme, there may be broader issues here - this concern is examined in more detail below. 88 Beyond the notion that more drugs should be covered by the scheme, many respondents reported, based on their experiences utilizing the NHIS, that the scheme should cover services and products that it nominally does. Respondents often mention that they wish the scheme would cover things like a hospital bed, registration materials at hospitals, vitamins, x-rays, and ultrasounds – these are all services that the NHIA claims are covered by the scheme. Indeed, 18 (28%) active enrollees in this study recalled specific instances within their enrollment period (within the previous 12 months before being interviewed) in which they had used their NHIS card and were required to pay for a hospital bed, hospital registration fee, or x-ray for themselves or their children. Though some medications that individuals were told they had to pay for are likely not covered by the NHIS, many respondents reported paying for aspects of care that the NHIA claims are covered, as this active enrollee in Dodowa, Shai Osudoku district stated: “You cannot say the NHIS is working. I have used it for three different pregnancies over a period of close to ten years. I pay for everything…I pay for everything during all these pregnancies. I spent one week after the delivery and I paid for everything even though I know that the health insurance is supposed to cover three months after the delivery. Just for this one week, I had to pay for the costs that were there for me and my baby. Can you even imagine that? On one occasion when I was there, the line that was used to set the drip for me, I paid for it. I paid for the cotton that was used. I paid for everything that the nurses used.”19 19 This respondent is correct, pregnant women are registered for free and are not required to pay any premium. Three months of pre-natal, and three months of post-natal care are completely covered by the scheme. 89 As the vast majority of respondents in this study believe the NHIS covers nearly all medical care, many were often surprised when they were presented with bills after utilizing the card. These notions present a disconnect between what citizens believe they are entitled to as NHIS members, and what they are actually provided with. This context gives many the impression that the scheme is not working at all – even though the scheme is indeed covering at least some of their costs. Though this context may exist owing to citizens’ knowledge concerning what the scheme is actually designed to cover, most respondents stated that they are told by their government and political leaders that the NHIS will cure all their financial woes concerning access to healthcare, yet their expectations are simply not fulfilled. A resident of East Legon and long-time NHIS member clarified this notion: “I think they should just come clean. The government should just let us know – “when you go to the hospital it covers 30% of everything you get from the hospital”…50%, fine…so that we will know. But you can’t just say, you can’t make it look like everything is supposed to be free. Then you go, and they tell you “it doesn’t cover this, it doesn’t cover that”. I think it is disrespectful…Everything should be made clear.” A pharmacist and owner of a local pharmaceutical store, who was very familiar with the workings of the NHIS and enrollee concerns also echoed this statement regarding expectations of coverage: “The government over the years has clearly not explained to citizens what the health insurance covers and what it cannot cover. There are certain tests that the insurance will never be able to cover. These are things that are very expensive and the government 90 cannot subsidize. But, the government has been silent. People go to the hospital and they expect this to work, this should not be the case.” The perceptions of individuals who had never been enrolled in the scheme are most often based on the experiences of others, with particular emphasis given to negative experiences. As most non-enrollees believe the scheme provides free healthcare, hearing that a friend or family member had to cover some of their own costs despite having NHIS insurance leads them to question the utility of the scheme, as this respondent in Dome, Ga East Municipal District suggested: “I don’t have it, and I don’t think I will get it. Because most of the time, the complaints when they come back (from the hospital or clinic)…it only covers paracetamol and some minor drugs. It is a shame that people think the health insurance is working when they go to the hospital, and then they come back with bills to pay and other things to do. That doesn’t make people want to register because eventually they will have to pay certain money. This shouldn’t be the case.”20 Based on these excerpts and survey findings, it is clear that citizens have high expectations for the scheme, which are either not in-line with what the scheme actually covers, or are based on experiences where the scheme did not cover what it was intended to. Moving beyond expectations of coverage, there is a broad notion among respondents that enrollees are treated as second class-citizens at points of care, and are given lower- quality products and services vis-à-vis non-enrollees. Indeed, respondents were asked whether or not they agreed with the statement, “NHIS enrollees receive the same quality of 20 Paracetamol (acetaminophen) reduces fever and is a pain reliever. It is the active ingredient in Tylenol. 91 care at health clinics as non-enrollees.” 79% (118) of the individuals included in this study disagreed or strongly disagreed with this statement. 78% (51) of active enrollees gave these responses. Even among individuals who were not enrolled, 78% (67) held these beliefs as well. The respondents in this study suggested the presence of two dimensions of “quality care” associated with utilizing the NHIS program: the quality of medication given to enrollees, and their treatment by hospital administrators, nurses, and doctors. In terms of the medication NHIS enrollees are entitled to, there is the perception that individuals with NHIS cards are candidates for lower-quality drugs. Individuals describe common themes whereby they visit accredited health clinics or hospitals that accept NHIS insurance, only to be told by hospital staff that the medications they have been prescribed are not available, or that they must return with cash to acquire medical attention. These issues are often due to a lack of resources on the part of the NHIA; hospitals and points of care prefer immediate remuneration for provided services and medication, whereas the NHIA may take several months (or longer) to reimburse them. Thus, even if some drugs are covered by the scheme and are available, the perception is that hospitals and clinics prefer to give these drugs to paying individuals rather than wait for the NHIA to reimburse them. This context creates the perception among individuals that if an enrollee attempts to acquire care for the same condition as a non-enrollee, they will be given minor or inferior drugs that may not cure their ailment, while the non-enrollee will be given appropriate medication – adequate drugs are reserved for those with cash. An NHIS enrollee from Legon gives an apt description of this situation: 92 “The people with the card are treated differently from the people who are paying for the service. If, two of us, both want the same drug; I have the card and you don’t have the card. They will give it to you, they won’t give it to me. Because, if they give it to me the scheme will take years to gather funds before they can come and pay for it. So if you are having the money now why not. Sometimes the pharmacies complain that people come and get drugs with a card, and when it is time for the government to pay there are delays. So, they are not happy giving out the drugs when you go with the card. That is why they don’t give out the drugs. Even the hospitals, they don’t give out drugs, because the government does not pay them on time, so they are not happy about it. So you can’t really feel entitled to get free drugs just because you are holding the card. Someone has to do something, and that person is not doing his part, so you can’t really blame the hospitals. It is not their fault. The person who is supposed to pay them on our behalf is not doing it, so then why should they bother themselves giving you expensive drugs, so you get paracetamol and vitamin C.” Some respondents even reported the belief that this notion goes beyond medications to actual medical procedures. This respondent from a rural area outside of Dodowa stated such: “In my case, I had a complication that required that a surgery be performed on me. I was transferred to Ridge Hospital in Accra and in one day I had to undergo two surgeries. The NHIS did not cover for these surgeries. In fact, I was happy it did not. My fear was that if it did end up covering them that I would not get the right operation – I would just get something and then I wouldn’t come back and I would die. It didn’t cover it, so I had to pay for it and they had to do the procedure well. Because I paid for it the procedure was done 93 well, and within a few days I was on my feet. If that had been under NHIS, I know there would have been complications.” Over the course of this research, multiple respondents were in the healthcare industry; nurses, a pharmacist, and pharmacy technicians were interviewed in the random sample. Their statements about the NHIS confirm many of the beliefs held by other respondents. This hospital administrator noted that their hospital had stopped accepting NHIS insurance: “I am a worker at a hospital and I have private insurance there so I don’t require the NHIS. My experience is that people who come to my hospital and use the NHIS, the government is not paying us on time. So over the years our hospital has lost interest in it and we have decided not to accept the NHIS card anymore because we need money and the money was not forthcoming.” Thus, we see that there exists a common perception that, owing to concerns of reimbursement, the drugs and services that are given to NHIS enrollees are of inferior quality, and that care providers prefer to give covered drugs and services to individuals with cash. Beyond the quality of medication and services, there is a perception that enrollees are often stigmatized at points of care. A common refrain among respondents is that NHIS enrollees are sequestered from paying individuals at points of care, and looked down upon by nurses and hospital administrators. Some enrollees even reported being humiliated and disrespected by hospital staff. These problems even induce some active NHIS enrollees to claim they did not actually have the insurance when approaching hospital administration, as this individual from Nima describes: 94 “When you go there with health insurance they try to make a distinction between those that hold the card and those that don’t have it. The services that they give to you…you feel bad that you came there with the card because they make you understand that the people that come with cash are more important to them because the clinic gets an income right away from them, they don’t get this with the insurance people. Because of that…this puts me off. I just stopped using it even though every year it is renewed. I feel so bad because there is a clear signal from them – they want people who come without NHIS.” Notions such as this, combined with concerns over medication, leave many with the perception that the scheme is not working and not worth having. However, there are some individuals who do feel there are benefits associated with the NHIS. Some enrollees simply accept that they will have to bring cash with them when they use their NHIS cards. These individuals are often those who seek medical care multiple times within a calendar year and believe that whatever the coverage benefits are, the amount of money saved by the program more than covers what they paid to register and their yearly premium. There are also individuals who note that, if nothing else, at least their registration fee is covered at hospitals and clinics. Of all the active enrollees included in this study, 32 reported that, though there are problems with the scheme, they are glad to be enrolled in it. However, it should be noted that 10 of these individuals had never used their NHIS card. The majority of this group of respondents was young and decided to register because they viewed the program as an aspect of Ghanaian citizenship, as this individual from Adabraka, Accra Metropolitan District stated: 95 “I registered because I am a Ghanaian, I know it is important to register…it is something that is Ghanaian. I needed to get involved because as a Ghanaian it is important to do that. That is my motivation.” Beyond these 10 individuals who had never used their NHIS card and reported that they were satisfied with the program, three additional individuals stated they were satisfied with the program only because the NHIS card is accepted as an official identification document with which one can use to open a bank account or register for a SIM card. Taking account of these two groups of respondents, 29% (19) of active enrollees reported positive experiences and satisfaction with the scheme. Among these individuals, several noted that they were satisfied with the scheme and able to use it because they had higher incomes and could afford to also pay for services that are not covered, they often express concerns that the scheme would not work for enrollees with little or no disposable cash. As noted in the previous section, rural respondents were more likely than urban dwellers to be active NHIS enrollees. The vast majority of enrollees who reported that they were satisfied with the NHIS were from rural areas. Rural enrollees often report better treatment by hospital and clinic staff, and better coverage for prescription medication. This respondent from Dome clarifies this point: “The NHIS is working well. I have been on the scheme for over 10 years now. I always take it to my village in Winneba and there I have wonderful experiences. I am very happy with the NIHS there. In Accra, the nurses look down on us and they do not attend to us in time. I have taken the firm decision never to go to any hospital with my health insurance in Accra. So anytime I feel that I am going to fall sick, I go back to my village in Winneba, where I go to the hospital where they take very good care of me. Last September, I was 96 unwell and I went to the hospital at Winneba, I took several injections and three drips, and I was there for three days. On the fourth day when I was discharged, I was asked to pay five Ghana cedis for the entire period. This would never happen in Accra. This will not happen. So because of that, I prefer to take my card and go to Winneba where I can access better healthcare. So for me, health insurance has been the best, it is always working for me.” One may suggest that the negative experiences and perceptions respondents have concerning the program may be the result of the actions of specific hospitals and clinics. Indeed, nearly all respondents’ complaints associated with the NHIS program center on using the card at points of care and their misgivings with the actions of doctors, nurses, and hospital administrators. The NHIS is a complex program that involves multiple layers of administration and actors, both in the public and private sectors. One may expect that this notion may complicate the ability of citizens to directly link their experiences with the program to government action. Though the program itself is directly attributable to the government, the process by which it administers goods is less clear. Attribution of policy outputs to government action is an important aspect of citizens’ accountability linkages with their leaders and governments (Stasavage and Harding 2014; Harding 2015). The NHIS is decentralized in terms of actors and responsibilities, with hospital administrators, doctors, nurses, and government officials all contributing to the final good produced by the program. During interviews, respondents were queried about the source of their misgivings with the NHIS. The vast majority of citizens in this study placed the blame for the inadequate functioning of the NHIS on the incumbent government. Indeed, most did not fault doctors, nurses, and hospital administrators for inadequate provision of medical care and demands for cash. A prominent theme in the sample was that 97 the disadvantages NHIS members experience are directly linked to the current administration. Many statements echo this theme, succinctly described by a resident of a village outside of Dodowa who reported consistent demands for payment before using his NHIS card: “This is my major problem. This problem is not coming from the hospital doors…the service providers believe that the government administration is not fast in paying them. This is this government’s fault – if our government is collecting money from people, they should be able to pay the providers on-time so that the providers can do whatever they need to do and there will not be any sort of problem.” The notion of ascribing blame to the current administration is even mentioned by multiple incumbent party (NPP) supporters, as this resident of Dome demonstrates: “The insurance is not working and that is why some of us are not motivated. I am an NPP member, in fact I am strong here, in the neighborhood they know me to be an NPP woman. But this is the truth…Nana has disappointed us more than any other, because the stakes were high, our aspirations and hopes were very high with it. He has not met our expectations and we are so disappointed.” The combination of citizens’ high expectations (free healthcare) and the ultimate good produced by the policy (perceived poor quality medication and mistreatment at points of care) has led individuals to believe that the government is not upholding its end of the bargain and the NHIS is performing poorly. Respondents, both enrollees and non- enrollees, are well aware that a broad portion of the funding for the NHIS stems from direct premium payments and the NHIL tax. Indeed, when one is given a receipt in Ghana, there is explicit mention of the amount of NHIL (National Health Insurance Levy) tax associated 98 with a given purchase. Respondents most often expressed dissatisfaction with the services they were provided in light of their financial contributions to the scheme, particularly premium payments. As argued in the theoretical section of chapter 1, resources contributed directly to the program by enrollees inform expectations enrollees have regarding what the scheme offers. Enrollees pay into the scheme, and their expectations for the goods the scheme delivers are grounded in this contribution. Moreover, the resources contributed by enrollees inform their conclusions regarding where to place the blame for the perceived poor performance of the NHIS. Though not all NHIS enrollees pay premiums, all but two of the respondents in this study reported paying premiums at the time of registration. These notions convince respondents that they are pre-paying for services that are not being adequately provided by the government. The fact that there is known taxation element involved with the scheme links all Ghanaian citizens with the program. This respondent in Adenta, Adenta Municipal District, who has never been enrolled in the scheme describes frustration with this aspect: “I have not enrolled because I have no confidence in the system. It is the government’s fault. We pay taxes but the NHIS does not cover us, then what is the essence of the tax that we pay? This government has increased the amount of money people need to register for the national health insurance scheme. They are only interested in taking money in taxes, and they don’t even do anything with it. I cannot believe it. I will never consider registering for the NHIS, I prefer to pay.” These excerpts demonstrate that, despite the multiple actors and layers of administration associated with the NHIS scheme, citizens attribute the functioning of the program to the government in power and recognize that the responsibility of the scheme 99 lays with incumbent political authorities. As the interview and survey data have demonstrated, respondents who are financially contributing to the scheme have high expectations for the NHIS, which are not being met by the government. Having described the expectations citizens have regarding the goods produced by the NHIS and how they relate to the goods the scheme proffers, utilizing an experimental vignette, the next chapter explores whether or not the non-conditional nature of programmatic distribution through the NHIS links performance evaluations to perceptions related to incumbent performance. This context is juxtaposed with perceptions of incumbent support associated with the non- programmatic distribution of healthcare via individually targeted transfers. 100 CHAPTER 4: THE POLITICAL IMPACTS OF PROGRAMMATIC AND NON-PROGRAMMATIC DISTRIBUTION: HEALTHCARE ALLOCATION AND ACCOUNTABILITY Whereas the previous chapter utilized qualitative and survey data to evaluate how design features of the NHIS structure the perceptions of NHIS performance among study participants as well as how these factors contribute to beliefs associated with accountability within the program, this chapter analyzes how programmatic and non- programmatic modes of healthcare distribution affect how goods performance influences incumbent support. In doing so, this chapter revisits the theoretical expectations of chapter 1, which describes the process by which citizens reward or punish incumbent politicians when healthcare is distributed programmatically (via a national insurance scheme) and non-programmatically (through individually targeted monetary handouts). As noted in previous chapters, the NHIS is a unique form of public goods distribution in that its benefits are not linked to individual politicians nor conditional on political support. This type of distribution is quite different from non-programmatic forms of distribution in which discretionary politicians target particularistic goods to specific citizens, a common occurrence in Ghana. Whereas the latter context is often associated with poor public goods provision and the electoral retention of poorly performing political leaders, the manner in which the NHIS distributes goods (e.g. healthcare) is shown to be associated with far different outcomes. Utilizing a vignette experiment embedded within the survey, the following explores how citizens’ attitudes toward political accountability respond to contexts in which healthcare is procured through the NHIS, and when it is procured through an individually targeted cash handout from an incumbent politician. The results here suggest that goods 101 distributed through the NHIS are associated with different attitudes toward political accountability vis-à-vis a non-programmatic context in which private goods are offered at the expense of public. Specifically, NHIS distribution is associated with the notion that incumbent members of parliament be held accountable for poor policy performance and inadequate healthcare provision. I argue that this result stems from the non-conditional nature of the NHIS, which ties incumbent support to evaluations regarding the quality of goods received through the program. When healthcare is acquired via an individually targeted cash handout, incumbent politicians maintain support despite inadequate healthcare provision. I argue that this result stems from the fact that individually targeted healthcare involves incumbents distributing cash handouts, which are then used by recipients to pay for healthcare services. In this distributive context, citizens link incumbent support not to their evaluations regarding the quality of healthcare goods they eventually access, but on the ability of incumbent leaders to proffer a handout. The latter finding also highlights the extent to which targeted transfers tied to individual politicians can undermine public goods provision. Public goods provision suffers when citizens evaluate leaders on their performance in distributing individually targeted goods. Together, these findings demonstrate both the possibility that a unique programmatic policy (the NHIS) can elicit performance-based voting, as well as the propensity for citizens to favor private goods over public goods in certain contexts. 4.1 DISTRIBUTION AND PERFORMANCE-BASED ACCOUNTABILITY The goods citizens accrue through programmatic policy are starkly different in nature from private goods gained through non-programmatic, targeted transactions; these 102 differences have significant ramifications that alter the relationship between citizens and their representatives. Theories of democratic governance postulate that citizens should reward incumbent politicians who provide public goods and punish those who do not. Yet when political leaders distribute particularistic goods targeted to individuals, citizens are often incentivized to support poorly performing leaders who might distribute few public goods, because they instead offer private goods (Wantchekon 2003; Golden and Min 2013). Whereas goods provided via programmatic policy might encourage citizens to vote based on how politicians implement or manage these programs, when the distribution of private goods is favored, citizens’ votes are de-linked from broader policy performance and the production of collective goods, significantly undermining the relationship between public goods performance and political support (Hicken 2011). As discussed previously, for a good to be considered distributed in a programmatic manner, the criteria for its distribution must be public, and clearly defined formal rules of distribution must shape the actual distribution of benefits (Stokes et al. 2013). Moreover, the delivery of goods must not be conditioned on political support or linked to an individual politician. When a good is distributed in this way, both political supporters and detractors maintain receipt of these goods regardless of the group in power so long as they fit the criteria for access. Entitlements to programmatic goods are not subject to the discretion of individual political leaders (Bardhan and Mookherjee 2017). The lack of political conditionality in relation to the accrual of goods falling under the rubric of programmatic distribution is what differentiates this form of distribution from those linked to discretionary politicians (Stokes 2007). Under programmatic distributive conditions, citizens are more likely to vote according to their perception of the quality of 103 goods produced, rather than for other considerations. As described in chapter 1, the non- conditional feature of programmatic distribution enables beneficiaries to translate performance evaluations into voting intentions. If a recipient’s wellbeing is not dependent on their support for or association with an individual incumbent politician, citizens are more likely to align their voting intentions with their evaluations of goods production performance (Carlson 2021). Because goods are distributed programmatically and are free from the interventions of discretionary politicians, beneficiaries’ evaluations of goods are the source of incumbent support or opposition in the case of programmatic distribution. When citizens deem that programmatic distribution conveys quality (poor quality) goods, incumbent leaders are rewarded (punished) due to their perceived association with the government in power. These expectations are particularly appropriate in relation to the NHIS, as it is a programmatic means of distribution. The good provided by the NHIS (healthcare) is available to all those within a defined category (enrollees), and access is governed by formal rules (members must be registered, and/or contribute yearly premiums). Access to the goods distributed through the NHIS are not conditional on political support or membership in particular identity groups. Goods distributed through some forms of non-programmatic means are unlikely to be associated with these same outcomes. Though substantial numbers of Ghanaian citizens are enrolled in the NHIS program, for a variety of reasons highlighted in the previous chapter a sizeable portion of the population are not NHIS members and must acquire healthcare through other means. It is likely that the NHIS coexists with non-programmatic, individually targeted distribution in relation to the acquisition of healthcare in Ghana. Researchers have often noted have often noted the extent to which citizens seek out 104 monetary resources from political leaders to specifically address their healthcare needs. This phenomenon has been described in Ghana (Agyepong and Adjei 2008; Lindberg 2010; Wahab 2019), as well as sub-Saharan Africa more broadly (Gros 2016). These notions are indicative of the presence of non-programmatic forms of healthcare distribution, in which political leaders target cash handouts to individuals who then utilize these funds to acquire healthcare. No distinction is made at this point as to whether or not Ghanaian citizens who access healthcare in this way are enmeshed in the hierarchical quid pro quo, patron-client relationships associated with clientelism, this point is addressed later in the chapter. The important distinction between non-programmatic, individually targeted handouts and programmatic distribution through the NHIS relates to the source of performance evaluations. Under the distributive conditions of the NHIS, performance is evaluated based on the quality of goods distributed by the program, yet when monetary resources are targeted to individuals in pursuit of healthcare goods, performance is evaluated based on the act of political leaders in providing handouts. For the politician doling out resources, it does not matter whether or not recipients are acquiring quality healthcare; their role in distribution is simply to provide monetary resources. In these instances, incumbent support is related to acquiring a handout, rather than to the quality of goods received. These propositions relating incumbent performance, healthcare quality, and mode of distribution are evaluated in the following section. 4.2 METHODOLOGY AND EXPERIMENTAL VIGNETTE An experimental vignette was embedded in the survey to determine the impact of NHIS distribution on attitudes toward political accountability in relation to the acquisition 105 of healthcare services. The experimental design randomly varied whether a hypothetical Ghanaian citizen unsuccessfully attempting to access healthcare was utilizing the NHIS (programmatic distribution) as a means with which to acquire healthcare or monetary assistance from their MP (non-programmatic distribution). Half of the subjects were randomly assigned to hear a vignette associated with NHIS distribution, while others heard a vignette associated with a non-NHIS member who relies on cash handouts from their MP to acquire medical care (a common recourse for the uninsured requiring medical care in Ghana). The inclusion of this context is apt given the history of healthcare in Ghana and the extent to which targeted redistribution continues to occur in the country (Lindberg 2012). The hypothetical individuals in both vignettes encounter a lack of medication upon requiring medical care. Thus the quality of the good being received (medication) is held constant. Variation lays in how the good is provisioned, either through a programmatic policy (the NHIS), or non-programmatically via individual cash handouts. The different versions of the vignette experiment are described in Table 1. As seen in the vignette, there is variation in how resources (in this case medication) are distributed: programmatically (through the NHIS) or non-programmatically (through a handout); this enables a comparison of accountability as it pertains to these two different modes of distribution in the context of poor performance. 106 TABLE 4.1 EXPERIMENTAL VIGNETTES Introductory sentence: Now I am going to read you a description of an individual in a specific situation and ask you a few questions regarding this individual… Treatment 1: Non-Programmatic distribution – Cash handout Kojo does not have any type of health insurance and is in poor health. Usually, when Kojo needs to pay for medical care, he goes to his member of parliament to ask for assistance. Yet when he went to the local health clinic with money given to him by the MP, there was no medication available that day and he was told by clinic staff to return 2 weeks later to receive his medication. Treatment 2: Programmatic distribution - NHIS Kojo is an active NHIS member and is in poor health. Usually, when Kojo needs medical care, he uses his NHIS card to receive medical services. Yet when he went to the local health clinic there was no medication available that day and he was told by clinic staff to return 2 weeks later to receive his medication. Following the vignette, a series of attitudinal outcomes were measured. Respondents were queried regarding the likelihood that the vignette subject would vote for a challenger to his current MP in the next election. Respondents were asked: ‘How likely, on a 1 to 10 scale, with 1 being extremely unlikely and 10 being extremely likely, do you think it is that Kojo will vote for a candidate other than his current MP in the next election?’ This variable was utilized to evaluate the extent to which accountability standards are impacted in both the NHIS and the non-programmatic distributive contexts. Given poor performance 107 (i.e., the clinic did not have the medication Kojo needed), do respondents think the individual is more likely to punish an incumbent when a good is proffered via a programmatic policy, or when goods are accrued through a cash handout? To supplement the above analysis, the same respondents were asked how likely, on the same one to ten scale, “do you think it is that Kojo will consider the issue of a lack of medicine at his health clinic when he votes in the next presidential election?” This variable was used to further evaluate the extent to which these two modes of distribution affect vertical accountability more broadly. Given the lack of medication, do respondents think the individual is more likely to consider this problem when voting in the next presidential election when the good is distributed through a programmatic policy, or when the good is distributed through a cash handout? It should be noted that this type of experiment has limitations. Survey respondents are asked to determine how they expect the individual within the vignette to respond to certain situations, as well as assign likelihoods as to the actions of this hypothetical individual. It is possible that some respondents may have imparted some of their own perceptions or beliefs on the individual within the vignette. As can be seen in Table 4.3, several covariates were included in analyses evaluating the impact of the vignette treatment to capture sociodemographic and partisan variation amongst respondents which may have influenced their responses to the vignette. Additionally, it is impossible to capture reality in vignette format. However, though some of the particulars of the hypothetical situations described in the vignettes may slightly diverge from real world occurrences, I am confident in external validity based on the existing literature that the interactions and situations highlighted are associated with healthcare distribution in Ghana 108 and the experiences of Ghanaian citizens. It has often been suggested that citizens in Ghana seek out political leaders to acquire resources for their personal healthcare needs (Lindberg 2010; Wahab 2019). While drawbacks exist, vignettes are often useful in limiting concerns surrounding social desirability bias. Respondents who may have been reluctant to discuss the political aspects of healthcare distribution may have been more forthcoming given their separation from the content of the vignettes. 4.3 RESULTS Table 4.2 presents the average treatment effect of NHIS distribution on the likelihood of an opposition vote within the vignettes. Those who heard that Kojo utilized the NHIS as a means with which to acquire healthcare thought it was extremely likely he would vote against his MP; the average outcome on the 1-10 scale for those assigned to that vignette was 8.8 (std error=1.2). The results were quite different for those who heard that Kojo relied on monetary handouts from his MP to acquire healthcare. These respondents were, on average, much more likely to think Kojo would not vote for their MP’s opponent (p<.01). The average response on the outcome scale for this group was only 1.8 (std error=.16). TABLE 4.2 VIGNETTE RESULTS: OPPOSITION MP NHIS Member vs. NHIS Member Non-Member Non-Member P-value Opposition MP vote 8.773 1.800 6.973 0.000 Standard error 1.248 0.158 1.090 109 TABLE 4.2 (cont’d) Observations 75 75 150 In total, 67% of respondents associated with the NHIS distribution vignette ranked the likelihood that Kojo would vote against his MP as 6 or greater on the 1-10 scale, with 35% of respondents ranking the likelihood as 10. Of the respondents exposed to the non- member vignette, 98% ranked the likelihood that Kojo would vote against his MP as 5 or less on the 1-10 scale, with 55% ranking the likelihood as 1. Following the vignette, we now utilize a series of ordered logit models to mitigate concerns about clustering in the data structure. Additionally, a series of control variables (measured via pre-treatment survey questions) are included to improve precision on estimates. Table 4.3 reports coefficient estimates for four model specifications predicting the likelihood of an opposition vote within the vignette. Treatment is a dichotomous variable representing the hypothetical NHIS distribution treatment, with Treatment holding a value of ‘1’ if the vignette was associated with the NHIS distribution and ‘0’ if it was associated with the non-programmatic context. Control variables include the respondent’s reported age, sex (Female), education, rural residence area, and lived poverty.21 An additional model in column 2 incorporates the respondent’s partisanship (NPP), and column 3 adds a variable distinguishing whether or not the respondent is an 21 Poverty is an additive index describing how often (never, once or twice, several times, many times, always) the respondent has gone without food, water, and a cash income in the previous year. Education is an ordinal measure of educational attainment (no formal schooling, informal schooling only, some primary school, primary school completed, some secondary school, secondary school completed, some university, university completed). 110 NHIS member (NHIS CARD).22 Column 4 includes jackknifed standard errors clustered by district to demonstrate that the results are not being driven by a single district within the sample. Lastly, column 5 incorporates random effects at the town/city/village (i.e. municipal level). TABLE 4.3 TREATMENT EFFECTS: ORDERED LOGIT MODELS Model (1) (2) (3) (4) (5) Opposition Opposition Opposition Opposition Opposition MP Vote MP Vote MP Vote MP Vote MP Vote Treatment 5.815** 5.817** 5.816** 5.837** 5.937** (0.659) (0.659) (0659) (0.344) (0.675) Age 0.045 0.056 0.044 0.055 0.081 (0.079) (0.075) (0.075) (0.067) (0.084) Female -0.626 -0.624 -0.636 -0.643 -0.685 (.349) (0.349) (0.355) (0.218) (0.353) Education 0.094 0.081 0.090 0.094 0.114 (.089) (0.074) (0.099) (0.064) (0.101) Rural 0.112 0.121 0.106 0.083 0.203 (0.324) (0.324) (0.326) (0.334) (0.550) Poverty -0.338** -0.338** -0.334** -0.321** -0.329** (0.123) (0.123) (0.125) (0.104) (0.129) NPP - 0.064 0.069 0.067 0.186 (0.371) (0.370) (0.293) (0.376) NHIS - - 0.052 0.053 0.064 member (0.333) (0.129) (0.132) Observations 150 150 150 150 150 22 NPP=1 denotes that the respondents is a member of the current incumbent party. Of respondents who expressed partisanship in the survey, all were members of either the NPP or the NDC. 111 In accordance with the findings reported in Table 4.2, the treatment effect remains positive and statistically significant in all models, indicating that respondents who received the NHIS distribution vignette were more likely to say that the hypothetical individual in the vignette would vote for an opposition MP in the next election. Incorporating controls for the respondent’s age, sex, education level, rural residence, partisanship, and NHIS membership does not affect the impact of the Treatment variable. Moreover, the incorporation of jackknifed standard errors clustered on district alleviates concerns that results are driven by any given district in the sample. Additionally, the Treatment variable is unaffected by the incorporation of random effects at the municipal-level. Poverty – the respondent’s lived poverty experience – remains negative and statistically significant in all models. This finding suggests that individuals within the sample who had experienced a lack of food, adequate drinking water, or cash income were more likely to state that the hypothetical individual within the vignette would not vote against the incumbent. This finding makes sense given that individuals with limited resources would be less likely to vote against an incumbent MP who is proffering any form of resource. We now turn to results concerning whether or not the lack of medicine described in the vignette influences attitudes towards politicians in a broader electoral context. Table 4.4 presents the average treatment effect of NHIS membership on the likelihood that the individual in the vignette would consider the lack of medicine when voting in the next presidential election. 112 TABLE 4.4 VIGNETTE RESULTS: PRESIDENTIAL ELECTION NHIS Member vs. NHIS Member Non-Member Non-Member P-value Consider lack of 7.66 6.12 1.54 0.000 medicine in presidential vote Standard error 0.294 0.287 0.01 Observations 75 75 150 The above results indicate that those respondents who received the vignette stating that Kojo was an active NHIS enrollee believed he would be more likely to consider the problem of a lack of medication when he voted in the next presidential election than those respondents who heard Kojo that to relied on a cash handout for his healthcare (p <0.01). While the averages for both vignettes indicate that the respondents believed there was a positive likelihood of this occurring, 73% of respondents associated with the NHIS vignette rated the likelihood that the lack of medicine would be considered in the next presidential election as a 6 or above on the 1 to 10 scale. Additionally, 39% of these respondents rated this likelihood as a 10. 46% of respondents associated with the non-programmatic vignette rated this same likelihood as a 6 or greater on the 1 to 10 scale, with only 16% rating the likelihood as a 10. The average rating of 6.12 for the non-programmatic vignette owes to the large number of respondents (40%) who rated the likelihood a 5 on the 1 to 10 scale. Despite the fact that the clinic failed to provide needed medication in both vignette scenarios, respondents who were exposed to the non-programmatic context were significantly less-likely to state that the individual in the vignette would consider this fact 113 when voting in an upcoming presidential election. This result, in combination with that of the prior vignette, indicate that the incumbent politician associated with the non- programmatic vignette is not the one performing poorly, rather it is the incumbent politician associated with healthcare distributed through the NHIS that is associated with poor performance. As these vignette results show, the respondents randomly assigned to vignette treatments associated with the hypothetical citizen utilizing the NHIS as a means with which to acquire healthcare and encountering a lack of medication were much more likely to expect the individual included in the vignette to vote against his MP in the next election, and consider the poorly functioning healthcare system when voting in an upcoming presidential election. These findings suggest that individuals are punishing the MP for the poor performance of the NHIS policy, and believe it likely that poor experiences with the NHIS can play a role in influencing vote presidential choice. The MP is not rewarded for the existence of the NHIS policy, but is sanctioned for its lack of effectiveness. Respondents randomly assigned to the non-NHIS context believed the hypothetical individual in their vignette would be unlikely to vote against the incumbent in a future election. This finding demonstrates the impact of favoring the targeting of benefits to individuals over universal distribution and demonstrates how this phenomenon undermines broader goods performance. Despite the fact that Kojo has not received the particularly important goods he requires, the cash handout from his MP ensures his support. Poor performance, or “failure” is associated with different outcomes in each of these two scenarios. In the context of NHIS distribution, poor performance is denoted in the fact that the government did not uphold its end of the bargain in delivering a programmatic 114 good. The MP is impersonally connected to this failure through their connection to government and it is logical for them to be punished for the program’s failure. In the context of non-programmatic distribution via a cash handout, poor performance would mean the failure of the MP to provide material resources to their constituent. As the relationship between the constituent and the incumbent is personal, rather than programmatic, the MP is not connected with the failure of the clinic to provide medication. The hypothetical individual in the vignettes does not receive the medical care they need under either condition of redistribution, yet the MP associated with NHIS distribution is perceived to have performed poorly, while the MP associated with the non-programmatic context has performed adequately. These results clearly demonstrate how political leaders can maintain support by distributing targeted goods despite performing poorly in the production of public goods. The findings also suggest that the NHIS can elicit behaviors consistent with theories of democratic accountability. Based on these results, a valid concern could be raised that the vignette associated with the cash handout provided cues to survey respondents such that they assumed the cash handout was transactionally distributed as part of a well-established clientelistic relationship, in which the individual within the vignette would face myriad pressures not to defect and vote against their patron. However, existing evidence suggests that the targeted distribution of cash assistance by Members of Parliament in Ghana is widespread, and not necessarily suggestive of a well-established clientelistic relationship. Lindberg (2010) interviews Ghanaian Members of Parliament to assess the accountability pressures they face and how these pressures shape their behavior. The MPs interviewed unanimously report that 115 citizens within their constituencies most hold them accountable for personal requests for monetary assistance. The MPs interviewed by Lindberg also claimed that personal assistance of this type was providing fewer and fewer returns during election season, as constituents avoided loyalty to supposed patrons. In relation to healthcare, MPs claimed that pressures for monetary assistance were so pervasive that they created strong incentives for them to focus on the production of collective goods. Many of the MPs interviewed by Lindberg claimed this fact led them to create the NHIS. If cash handouts are ubiquitous enough so as to lead Ghanaian MPs to create legislation to address the root cause of the issue, it is unlikely that MPs have well-established patron-client relationships with all of those who were seeking assistance. Moreover, the fact that MPs in Lindberg’s study reported that citizens’ demands for cash were their primary sources of accountability pressures indicates that these monetary transfers are not necessarily clientelistic in nature, but may be more associated with incumbent politicians seeking to gain and/or maintain support. Lindberg suggests a situation whereby citizens hold MPs accountable for their performance in distributing cash handouts. This context inverts the conventional terms of clientelistic exchanges, whereby citizens are held accountable by politicians if they renege on the terms of a clientelistic exchange and withdraw their support. This situation is reminiscent of the non-programmatic, targeted distribution of private and club goods which Stokes et al. (2013) associate with “non-conditional partisan bias.” Stokes and coauthors suggested that incumbent leaders may distribute private and club goods to citizens to gain support, without the expectation that recipients will be monitored by brokers to ensure that they adhere to a clientelistic quid pro quo. 116 4.4 DISCUSSION Competitive elections allow the means through which citizens in democratic polities can hold their governments accountable. Yet in many elections across sub-Saharan Africa, elections often fail to provide this mechanism, owing to confounding factors such as clientelism, vote-buying, and ethnic voting. Whereas several authors have demonstrated that this is not the full story--African voters do hold governments accountable for public goods provision under certain contexts--this chapter demonstrates that a programmatic social policy can engender attitudes that are consistent with democratic notions of accountability based on policy performance vis-à-vis the archetypal context associated with the targeted distribution of material benefits. The NHIS represents a commitment on behalf of the government to its citizens. When this commitment is not satisfactorily realized, citizens are willing to punish underperforming political leaders. These findings also shed light on how effective the targeted distribution of cash handouts is in disrupting the relationship between public goods performance and accountability. The vignette experiment demonstrates how unlikely these voters are to punish political leaders for poor public goods performance and consider goods performance in electoral decision-making. Healthcare is considered a chief concern among African citizens, and is essential to wellbeing, but even in this context, respondents demonstrated that a private handout was more determinative of political support than the adequate functioning of a public health system in the vignette experiment. Many of the Ghanaians that took part in this experiment spoke as to this point, as this citizen did in Adenta: 117 “At least Kojo is getting something…(respondent laughs). He (Kojo) will find something to do with it. Cash is cash. Most times people get nothing from them (the politicians). So, he will vote for him.” Beliefs such as this demonstrate the political reality facing many citizens in developing contexts in sub-Saharan Africa and beyond. However, the findings associated with performance-based assessments related to the NHIS policy provide evidence that goods provided through such a programmatic policy can encourage citizens to punish poorly performing politicians. 118 CONCLUSION This research program has sought to evaluate the programmatic distributive aspects of national health insurance schemes, decipher how these traits differ from non- programmatic, targeted forms of distribution, and the political outcomes associated with goods distribution via both these mechanisms of distribution as they pertain to citizens’ evaluations of government performance in producing healthcare goods. The first chapter denoted that national health insurance schemes differ non-programmatic forms of distribution in that the receipt of goods is not conditional on political support or attachment to a particular politician. Programmatic distribution through a national health insurance scheme also conveys goods, which are deferred. Finally, in order to acquire these goods citizens often have to make direct pre-payments in the form of registration fees, premiums, and/or specific taxes. The subsequent theoretical section highlighted how these features lead citizens to form expectations regarding future goods and how these expectations factor into performance evaluations and incumbent support. These theoretical expectations were juxtaposed with those associated with a non-programmatic mode of distribution, whereby healthcare goods are acquired via monetary transfers from incumbent politicians targeted to individuals. I argue that programmatic distribution through a national health insurance scheme focuses citizens’ performance evaluations of incumbents on the quality of goods received. When healthcare is distributed in a non- programmatic manner through a cash handout, citizens evaluate incumbents based on their providing said handout, rather than on the quality of goods received. Later chapters focusing on the NHIS in Ghana broadly supported these theoretical expectations. The NHIS members in this study hold lofty expectations (be they accurate or 119 not) regarding the future benefits they expect to receive through the program because they are monetarily invested in it through their premium and tax contributions. In practice, these expectations are rarely met. As the actual goods delivered do not align with expectations, citizens are under the perception that the NHIS is performing poorly. Despite the fact that many respondents who are NHIS members reported instances whereby individuals outside of government demanded payment for services, denied medicines, or treated them poorly, respondents associate the poor performance of the system with the government in power. Many of these respondents cited their contribution of direct payments to NHIS officials as a reason for this belief – NHIS enrollees had pre-paid into this governmental program and had formed beliefs and expectations regarding the goods the program would produce. These expectations were largely not met by the incumbent government. The last chapter utilized a vignette experiment to evaluate healthcare distribution in Ghana more directly, by comparing accountability outcomes linked to both the programmatic distribution of healthcare through the NHIS and the non-programmatic distribution of healthcare through an individual cash handout. The results of the experiment are stark. Respondents exposed to the NHIS distribution vignette were far more likely to claim that the individual in the vignette would punish their incumbent MP for the failure of the program to provide a healthcare good. I suggest this result stems from the non-conditional nature of programmatic modes of goods distribution, which enable beneficiaries to evaluate incumbent performance based on the quality of goods distributed. The non-programmatic healthcare distribution of healthcare via a cash handout is associated with significantly different results. Respondents exposed to the a vignette 120 associated with this form of distribution were far less likely to believe that the individual in the vignette would punish their MP for the inadequate provision of healthcare goods. In this case, despite the fact that needed healthcare was not acquired by the individual in the vignette, respondents deemed that the incumbent MP had performed well, as a handout was given. I argue that this result stems from the fact that under this form of distribution, political leaders are evaluated solely on their provision of a handout, rather than on the quality of goods acquired with said handout. This research program makes several contributions to the literature focusing on distributive politics, social policy, and accountability in sub-Saharan Africa. First, through my incorporation of programmatic distribution, I broaden the literature on goods allocation in the region. Whereas there exists a dearth of studies focusing on goods distribution in relation to clientelism and ethnic politics, researchers have largely neglected to evaluate distribution through programmatic means. Moreover, authors who do analyze programmatic distribution in sub-Saharan Africa almost exclusively focus on conditional cash-transfer programs (Lavers 2019; Hickey et al. 2019). This neglect of programmatic distributive mechanisms is troubling given the theoretical potential for these types of policies to replace clientelistic linkages between citizens and their political leaders by shifting the focus from targeted to collective goods. I also contribute to the literature on distributive politics in Ghana in particular. My research provides valuable insights into this context, whereby programmatic distribution takes place alongside the more targeted distributive mechanisms conventionally associated with African politics. I also add to the literature evaluating voter preferences in sub-Saharan Africa by introducing the production of goods via programmatic policy as an influential force that 121 plays a role in how citizens judge incumbent performance. Whereas a few researchers have analyzed public goods production and voting intentions (Harding 2015; Bleck 2015), goods derived through programmatic policy have yet to be evaluated in this regard. By focusing on programmatic distribution through a distinct social policy, my research also contributes to the recent literature on social policy in sub-Saharan Africa (Adesina 2009; Aryeetey and Goldstein 2000; Wahab 2019), my research on the NHIS demonstrates citizens’ actual experiences utilizing these types of policies. Indeed, as more governments in the developing world turn to national insurance schemes to increase access to care and control healthcare costs, it is essential to understand how citizens view and respond to these types of social policies. Though there are myriad reasons why an individual chooses to vote the way that they do in sub-Saharan Africa, the final chapter highlights one of the important outcomes of programmatic distribution – it has to the potential to encourage citizens to vote in accordance with their evaluations of goods production. In light of this finding, future research should focus on other forms of programmatic distribution on the continent. Multiple states in sub-Saharan Africa possess social security programs, and while the political ramifications of social security benefits have been the focus of much academic research as pertains to the developed world, these programs are significantly understudied in developing contexts. Moreover, multiple conditional cash transfer programs, a particular form of programmatic policy, have been piloted or implemented in sub-Saharan Africa. These programs that often blur the line between clientelistic and programmatic politics are ripe for analysis in the political science literature. 122 Additionally, national health insurance schemes in the region provide multiple avenues for future research. As noted previously, when it comes to national health insurance schemes in sub-Saharan, Ghana and Rwanda are the main cases, as these two states possess the most extensive and longest-lived schemes on the continent. Whereas the research included in this program focuses on the democratic context of Ghana, Rwanda presents a distinctly more authoritarian regime context. How citizens respond to the distributional aspects of health insurance schemes in Rwanda may be significantly different than in Ghana. Moreover, whereas the Ghanaian scheme was conceived so as to address a public concern (discontent over the “cash and carry system”), the health insurance in Rwanda has been described of as being associated with developmental legitimacy (Chemouni 2018). These two contexts may have significant ramifications for how citizens view these programs. Lastly, in Rwanda, membership in a health insurance scheme is compulsory, and based on the high enrollment numbers seen in Rwanda, there is a degree of enforcement. This is not the case the Ghana. How do African citizens view these distributive programs in a compulsory vs non-compulsory context? 123 APPENDIX 124 APPENDIX The following pages contain the original survey utilized during fieldwork in the Accra Metropolitan Region of Ghana during the Summer/Fall of 2019. The two vignettes are presented at the end of the survey for reasons of space. When the survey was conducted each individual survey contained one randomly assigned vignette. Filled-out by interviewer: Date: _________________ Time: _________________ Gender of respondent: _________________ City/town/village name: _______________________ Urban/rural locality: _____________________ Interview language: _____________________________ Audio recorded: ____________________________ Consent and Introduction (see consent form) Note: The individual (18 years of age or older) must give his or her informed consent by agreeing to participate (see consent form). If participation is refused, walk away from the respondent and continue convenience sampling. Controls/Demographics: All Respondents: Thank you for agreeing to this interview, let’s begin with a few facts about yourself… 1. How old are you? 2. What is your highest level of education? a. No formal schooling b. Informal schooling only c. Some primary schooling d. Primary school completed e. Some secondary school f. Secondary school completed g. Some University h. University completed i. Don’t know 3. What is your current occupation? a. None/unemployed b. Student 125 c. Homemaker d. Agriculture/farming/fishing e. Trader/vender f. Retail/shop g. Unskilled manual labor h. Artisan/skilled manual labor i. Clerical j. Supervisor k. Security services l. Mid-level professional (teacher, nurse, mid-level government employee) m. Upper-level professional (doctor, lawyer, banker, engineer, senior-level government employee) n. Don’t know o. Other (fill-in):__________________________________ 4. Over the past year, how often, if ever, have you or anyone in your family gone without enough to eat? a. Never b. Once or twice c. Several times d. Many times e. Always f. Don’t know 5. Over the past year, how often, if ever, have you or anyone in your family gone without enough clean water for home use? a. Never b. Once or twice c. Several times d. Many times e. Always f. Don’t know 6. Over the past year, how often, if ever, have you or anyone in your family gone without a cash income? a. Never b. Once or twice c. Several times d. Many times e. Always f. Don’t know All respondents: Now, I would like to ask you some questions about the National Health Insurance Scheme… 7. What benefits does the NHIS give to enrollees? (fill-in) 126 8. Please tell me whether you agree or disagree with the following statement: NHIS enrollees are not required to pay for services at health clinics: a. Strongly agree b. Agree c. Neither agree nor disagree d. Disagree e. Strongly disagree f. Don’t know 9. At what type of health facility are you allowed to use your NHIS card to get health care? a. Government clinics only b. Private health clinics only c. Faith-based/NGO clinics d. Any health facility of my choosing/I choose my health care provider 10. Who runs the NHIS? a. The government/state/National Health Insurance Authority b. Political party/the president c. NGO/CSO d. INGO e. Don’t know 11. What do you think the NHIS should provide to citizens that it currently does not? (fill- in) 12. How do you feel about how the NHIS performs/is run? (fill-in) 13. I am going to read you a statement, which statement is closest to your views? Choose Statement A or Statement B. Statement A: The government should provide financial protection for health care to all citizens because health care is the responsibility of the government. Statement B: The government should NOT provide financial protection for health care to all citizens because this is expensive and the government does not have the funds to do this. a. Agree very strongly with A b. Agree with A c. Agree very strongly with B d. Agree with B e. Don’t know 14. Do you have an active National Health Insurance Scheme (NHIS) card? a. Yes (if “yes”, skip to question 17) b. No c. Don’t know 15. Have you ever been enrolled in the NHIS? a. Yes (if “yes”, DO NOT ASK question 16) 127 b. No (if “no”, skip to question 16) c. Don’t know 16. Why did you not re-enroll in the NHIS? a. Could not afford renewal payment b. Re-enrollment office too far away c. Not satisfied with clinic care when using NHIS card d. No confidence in the NHIS/Note satisfied with NHIS offices e. Did not need medical care f. Acquired different type of health insurance g. Don’t know h. Other (fill-in): 17. Why have you not enrolled in the NHIS? (After asking, skip to question 20) a. Could not afford enrollment fee b. Did not have required documents c. Enrollment office too far away d. No confidence in the NHIS e. Did not need medical care f. Have other type of health insurance g. Don’t know h. Other (fill-in): 18. Do you/Did you (if “yes” for question 14) pay a yearly premium? You know, an annual fee for your NHIS card? a. Yes b. No c. Don’t know 19. How long have you been/were you (if “yes” for question 14) an NHIS member? (fill- in): ___________________________________________ 20. How satisfied are you/were you (if “yes” for question 14) with the NHIS as a whole? a. Very satisfied b. Fairly satisfied c. Not very satisfied d. Not at all satisfied e. Don’t know All respondents: Now, I would like to ask you some questions about NHIS district offices, the locations where you enroll, re-enroll, and seek information about the NHIS… 21. Have you ever been to an NHIS district office? a. Yes b. No (if “no”, skip to question 22) 128 c. Don’t know if “don’t know”, skip to question 22) 22. How satisfied are you with your experiences at NHIS offices when enrolling, re- enrolling, raising a concern, or seeking information? a. Very satisfied b. Fairly satisfied c. Not very satisfied d. Not at all satisfied e. Don’t know 23. In your opinion, how often are NHIS enrollees treated unfairly by NHIS officials at NHIS offices? For example, how often do some NHIS enrollees receive faster service at NHIS offices than others? a. Never b. Rarely c. Often d. Always e. Don’t know 24. What do you do/would you do (if “no” or “don’t know” for question 20) if you have a complaint about some aspect of the NHIS? a. File a complaint at my local NHIS district office b. Contact a traditional or religious leader c. Contact a family member d. Contact a government official (non-NHIS official) e. Contact my local government representative f. Nothing/there is nothing I can do g. Don’t know h. Other (fill-in) 25. Have you ever filed a complaint at an NHIS district office? a. Yes b. No (if “no”, skip to question 26) c. Don’t know (if “don’t know”, skip to question 26) 26. How satisfied are you with how your complaint was handled? a. Very satisfied b. Fairly satisfied c. Not very satisfied d. Not at all satisfied e. Don’t know 27. Please tell me whether you agree or disagree with the following statement: NHIS officials care about the opinions/concerns of enrollees. a. Strongly agree b. Agree 129 c. Neither agree nor disagree d. Disagree e. Strongly disagree f. Don’t know All respondents: Now, I would like to ask you some questions about your experiences at health clinics… 28. Have you been to a health clinic in the past year? a. Yes b. No c. Don’t know 29. What type of health clinic do you usually go to? a. Government/public clinic b. Private clinic c. Faith-based/NGO clinic d. Other (fill-in): 30. How would you rate the quality of care you receive at health clinics? a. Very good b. Fairly good c. Neither good nor bad d. Fairly bad e. Very bad f. Don’t know 31. Please tell me whether you agree or disagree with the following statement: NHIS enrollees receive the same quality of care at health clinics as non-enrollees. a. Strongly agree b. Agree c. Neither agree nor disagree d. Disagree e. Strongly disagree f. Don’t know 32. Please tell me whether you agree or disagree with the following statement: Exempt (non-paying) NHIS enrollees receive the same quality of care at health clinics as NHIS enrollees that pay a yearly fee for their card. a. Strongly agree b. Agree c. Neither agree nor disagree d. Disagree e. Strongly disagree f. Don’t know 130 33. I’m going to read you a list of groups. For each, please tell me whether you think they would receive good quality care or poor quality care at a health clinic: a. NHIS enrollees? ___________________ b. Non-insured individuals (paying out-of-pocket for services at health clinics)? ___________________ c. Individuals with private insurance? ___________________ d. Members of the ruling political party? ___________________ 34. How often, if ever, have you been asked to pay a fee or a bribe at a health clinic? a. Always b. Often c. A few times d. Never e. Don’t know All respondents: Now, I am going to ask you some questions about government and politics in Ghana… 35. How easy or difficult is it for an ordinary person to have their voice heard between elections? a. Very easy b. Somewhat easy c. Somewhat difficult d. Very difficult e. Don’t know 36. In the past year, how often, if at all, have you made a complaint to a government official, for example, by going in person or writing a letter? a. Never b. Once or twice c. Several times d. Many times e. Don’t know f. Did not have a complaint in past year 37. How likely is it that you would join with others to try to make your assemblyman listen to your concerns about a matter of importance in your community? a. Very likely b. Somewhat likely c. Not very likely d. Not at all likely e. Don’t know 38. In your opinion, how often do politicians keep their campaign promises after elections? a. Never b. Rarely c. Often 131 d. Always e. Don’t know 39. Do you agree or disagree with the following statement: A good citizen should always voice their concerns to elected officials when they are not happy with public services. a. Strongly agree b. Agree c. Neither agree nor disagree d. Disagree e. Strongly disagree f. Don’t know 40. I am going to read you a statement, which statement is closest to your views? Choose Statement A or Statement B. Statement A: It is most important that the government gets things done, even if citizens have no influence over what it does. Statement B: It is most important that citizens influence what the government does, even if it takes a while for the government to get things done. a. Agree very strongly with A b. Agree with A c. Agree very strongly with B d. Agree with B e. Don’t know 41. Do you agree or disagree with the following statement: Politics and government sometimes seem so complicated that it is hard to understand what is going on. a. Strongly agree b. Agree c. Neither agree nor disagree d. Disagree e. Strongly disagree f. Don’t know 42. How interested are you in politics? a. Very interested b. Somewhat interested c. Not very interested d. Not at all interested e. Don’t know 43. When you get together with friends or family, how often would you say you discuss political matters? a. Frequently b. Occasionally c. Never d. Don’t know 132 44. Can you tell me the name of the MP from your constituency? a. Incorrect guess b. Correct c. Know, but cannot remember d. Do not know 45. Do you agree or disagree with the following statement: Citizens should be responsible for their own well-being; a citizens’ well-being should not be the responsibility of the government. a. Strongly agree b. Agree c. Neither agree nor disagree d. Disagree e. Strongly disagree f. Don’t know 46. Did you vote in the most recent national election in 2016? a. Yes b. No c. Was not registered d. Don’t know/can’t remember 47. I’m going to read you a list of individuals, please tell me which individual(s) you would turn to if there was a problem with public services in your community. a. Local assemblyman? __________________________ b. Member of parliament? __________________________ c. Political party official? __________________________ d. Religious leader? __________________________ e. Traditional leader? __________________________ 48. I’m going to read you a list of individuals, please tell me how often you have contacted any of these individuals about a personal problem in the past year (Very often, often, once or twice, never, don’t know): a. Local assemblyman? __________________________ b. Member of parliament? __________________________ c. Political party official? __________________________ d. Religious leader? __________________________ e. Traditional leader? __________________________ 49. Who is most responsible for making sure that, once elected, Members of Parliament do their jobs? a. The president b. The parliament c. Their political party d. Voters e. Don’t know 133 50. Have you attended a protest or demonstration in the past year? If not, would you do this if you had the chance? a. No, would never do this b. Would do this if had the chance c. Once d. More than once e. Don’t know 51. In your opinion, do both of the two major political parties support the NHIS, or does only one party support the program? a. Yes, both parties support the NHIS (if “yes”, skip question 51) b. No, only 1 party supports the NHIS c. Don’t know (if “don’t know”, skip to question 51) 52. Which party? a. NPP b. NDC c. Don’t know 53. Do you agree or disagree with the following statement: The current opposition party could run the NHIS better than the party in power. a. Strongly agree b. Agree c. Neither agree nor disagree d. Disagree e. Strongly disagree f. Don’t know 54. Do you agree or disagree with the following statement: The NHIS does not change much regardless of which party is in power. a. Strongly agree b. Agree c. Neither agree nor disagree d. Disagree e. Strongly disagree f. Don’t know Vignettes: Now I am going to read you a description of an individual in a specific situation and ask you a few questions regarding this individual… Vignette A: Kojo does not have any type of health insurance and is in poor health. Usually, when Kojo needs to pay for medical care, he goes to his member of parliament to ask for assistance. Yet when he went to the local health clinic with money given to him by the MP, there was 134 no medication available that day and he was told by clinic staff to return 2 weeks later to receive his medication. How likely, on a 1 to 10 scale, with 1 being “extremely unlikely” and 10 being “extremely likely”, do you think it is that Kojo will go to a government official to address the issue of a lack of medicine at his health clinic? _________________________ How likely, on a 1 to 10 scale, with 1 being “extremely unlikely” and 10 being “extremely likely”, do you think it is that Kojo will go to a traditional or religious leader to address this issue? _________________________ How likely, on a 1 to 10 scale, with 1 being “extremely unlikely” and 10 being “extremely likely”, do you think it is that Kojo will consider the issue of a lack of medicine at his health clinic when he votes in the presidential election? _________________________ How likely, on a 1 to 10 scale, with 1 being “extremely unlikely” and 10 being “extremely likely”, do you think it is that Kojo will discuss his problem with other citizens? _________________________ How likely, on a 1 to 10 scale, with 1 being “extremely unlikely” and 10 being “extremely likely”, do you think it is that Kojo will vote for a candidate other than his current MP in the next election? _________________________ Partisanship: Now, please answer these two final questions: 1. Do you feel close to any particular political party? a. Yes b. No (if “no”, end interview) c. Refused (if “refused”, end interview) d. Don’t know (if “don’t know”, end interview) 2. Which party? a. New Patriotic Party (NPP) b. National Democratic Congress (NDC) c. Convention People’s Party (CPP) d. Progressive People’s Party (PPP) e. Democratic People’s Party (DPP) f. Refused g. Don’t know h. Other (fill-in) ____________________________________ Vignette B: Ama is an active NHIS member and is in poor health. Usually, when Ama needs medical care, she uses her NHIS card to receive medical services. Yet when she went to the local 135 health clinic there was no medication available that day and she was told by clinic staff to return 2 weeks later to receive her medication. How likely, on a 1 to 10 scale, with 1 being “extremely unlikely” and 10 being “extremely likely”, do you think it is that Ama will go to a government official to address the issue of a lack of medicine at her health clinic? _________________________ How likely, on a 1 to 10 scale, with 1 being “extremely unlikely” and 10 being “extremely likely”, do you think it is that Ama will go to a traditional or religious leader to address this issue? _________________________ How likely, on a 1 to 10 scale, with 1 being “extremely unlikely” and 10 being “extremely likely”, do you think it is that Ama will consider the issue of a lack of medicine at her health clinic when she votes in the presidential election? _________________________ How likely, on a 1 to 10 scale, with 1 being “extremely unlikely” and 10 being “extremely likely”, do you think it is that Ama will discuss her problem with other citizens? _________________________ How likely, on a 1 to 10 scale, with 1 being “extremely unlikely” and 10 being “extremely likely”, do you think it is that Ama will vote for a candidate other than her current MP in the next election? _________________________ Partisanship: Now, please answer these two final questions: 3. Do you feel close to any particular political party? a. Yes b. No (if “no”, end interview) c. Refused (if “refused”, end interview) d. Don’t know (if “don’t know”, end interview) 4. Which party? a. New Patriotic Party (NPP) b. National Democratic Congress (NDC) c. Convention People’s Party (CPP) d. Progressive People’s Party (PPP) e. Democratic People’s Party (DPP) f. Refused g. Don’t know h. Other (fill-in) ____________________________________ 136 BIBLIOGRAPHY 137 BIBLIOGRAPHY Acharya, Arnab, Sukumar Vellakkal, Fiona Taylor, Edoardo Masset, Ambika Satija, MargaretBurke, and Shah Ebrahim. 2013. “The Impact of Health Insurance Schemes for the Informal Sector in Low- and Middle-Income Countries: A Systematic Review.” The World Bank Research Observer 28: 236-266. Achen, Christopher H., and Larry Bartels. 2016. Democracy for Realists: Why Elections do not Produce Responsive Government. Princeton NJ, Princeton University Press. Adesina, J. 2009. “Social Policy in sub-Saharan Africa: A Glance in the Rear-View Mirror.” International Journal of Social Welfare 18(1): 37-51. Agyepong I., D. Abankwah, A. Abroso, C. Chun, J. Dodoo, and S. Lee. 2016. “The “Universal” in UHC and Ghana’s National Health Insurance Scheme: Policy and Implementation Challenges and Dilemmas of a Lower Middle Income Country.” BMC Health Serv Res 16:504. Akazili, James, Paul Welaga, Ayaga Bawah, Fabian Achana, Abraham Oduro, John K. Awoonor-Williams, John E. Williams, Moses Aikins, James F. Phillips. 2014. “Is Ghana’s Pro- Poor Health Insurance Scheme Really for the Poor? Evidence from Northern Ghana.” BMC Health Services Research 14: 637. Alhassan, RK, SO Duku, and W. Janssens. 2015. “Comparison of Perceived and Technical Healthcare Quality in Primary Health Facilities: Implications for a Sustainable National Health Insurance Scheme in Ghana.” PLoS One 10(10): 1-19. Allegri, M., R. Sauerborn, B. Kouyate, and S. Flessa. 2009. “Community Health Insurance in Sub-Saharan Africa: What Operational Difficulties Hamper its Successful Development?” Tropical Medicine and International Health 14(5): 586-596. Amu H., and S.K. Dickson. 2016. “Health Insurance Subscription Among Women in Reproductive Age in Ghana: Do Socio-Demographics Matter?” Health Economics Review 6(24): 1-8. Aryeetey, Genevieve, Caroline Jehu-Appiah, Ernst Spaan, Irene Agyepong, and Rob Baltussen. 2012. “Costs, Equity, Efficiency, and Feasibility of Identifying the Poor in Ghana’s National Health Insurance Scheme: Empirical Analysis of Various Strategies.” Tropical Medicine and International Health 17(1): 43-51. Aryeetey, Ernest, and Markus Goldstein. 2000. “Ghana: Social Policy Reform in Africa.” in Reforming Social Policy, ed. Daniel Moralez-Gomez, Necla Tschirgi, and Jennifer Moher. Ottawa, Canada: International Development Research Center. 138 Aryeetey, Genevieve, Justice Nonvignon, Caroline Amissah, Gilbert Buckle, and Moses Aikins. 2016. “The Effect of the National Health Insurance Scheme on Health Service Delivery in Mission Facilities in Ghana: A Retrospective Study.” Globalization and Health (12): 32. Ashworth, Scott, Ethan Bueno de Mesquita, and Amanda Friedenberg. 2018. “Learning About Voter Rationality.” The American Journal of Political Science. 62(1): 37-54. Assensoh, Akwasi B. and Hassan Wahab. 2008. “A Historical-Cum-Political Overview of Ghana’s National Health Insurance Law.” African and Asian Studies. 7: 289-306. Asenso-Okyere, W.K., A. Anum, I. Osei-Akoto, and A. Adokonu. 1998. “Cost Recovery in Ghana: Are there any Changes in Health Care Seeking Behavior?” Health Policy and Planning 13(2): 181-188. Avelino, G., D. Brown, and W. Hunter. 2005. “The Effect of Capital Mobility, Trade Openness, and Democracy on Social Spending in Latin America” American Journal of Political Science 49(3): 625-641. Bardhan, Pranab, and Dilip Mookherjee. 2017. “A Theory of Clientelistic Politics versus Programmatic Politics.” Working paper. The Institute for Development Economics Working Paper Series. Boston University Department of Economics. Barkan, J.D. and M. Chege. 1989. “Decentraliziing the State: District Focus and the Politics of Reallocation in Kenya.” Journal of Modern African Studies 27(3): 431-453 Baum, M., and D. Lake. 2001. “The invisible Hand of Democracy: Political Control and the Provision of Public Services.” Comparative Political Studies 34(6): 587-621. Berens, Sarah. 2016. “Crowding Out or Welfare Promotion? How Foreign Aid Affects Social Expenditures in Latin American Welfare Systems” Social Policy and Administration 50(3): 353-378. Besley, Timothy, and Robin Burgess. 2002. “The Political Economy of Government Responsiveness: Theory and Evidence from India.” Quarterly Journal of Economics 117(4): 1415-1451. Blanchet, NJ, G. Fink, and I. Osei-Akoto. 2012. “The Effect of Ghana’s National Health Insurance Scheme on Health Care Utilization.” Ghana Medical Journal. 46(2): 76-84. Bleck, Jamie. 2015. Education and Empowered Citizenship in Mali. Baltimore, MD: Johns Hopkins University Press. Bratton, Michael. 2007. “Are You Being Served? Popular Satisfaction with Health and Education Services in Africa,” Afrobarometer Working Paper 65. Retrieved from http://www.afrobarometer.org/papers/AfropaperNo65.pdf 139 Bratton, Michael. 2008. “Vote Buying and Violence in Nigerian Election Campaigns.” Electoral Studies 27(4): 621-632. Bratton, Michael, and Eric C. C. Chang. 2006. “State-Building and Democratization in sub- Saharan Africa: Forwards, Backwards, or Together?” Comparative Political Studies 39(9): 1059-1083. Bratton, Michael, and Carolyn Logan. 2006. “Voters but Not Yet Citizens: The Weak Demand for Political Accountability in Africa’s Unclaimed Democracies.” Afrobarometer Working Paper 63. Retrieved from http://www.afrobarometer.org/papers/AfropaperNo63.pdf. Briggs, Ryan. 2012. “Electrifying the Base? Aid and Incumbent Advantage in Ghana.” Journal of Modern African Studies 50(4): 603-624. Brown, David, and Wendy Hunter. 1999. “Democracy and Social Spending in Latin America.” American Political Science Review 93(4): 779-790. Bellina, Severine, Dominique Darbon, Stein Sundstol Eriksen, and Ole Jacob Sending. 2009. The Legitimacy of the State in Fragile Situations. Oslo: Norwegian Agency for Development Cooperation. Bueno de Mesquita, Bruce, James D. Morrow, Randolph Siverson, and Alastair Smith. 2002. “Political Institutions, Policy Choice, and the Survival of Leaders.” British Journal of Political Science 32(4): 559-590. Cameron, D. 1978. “Expansion of the Public Economy: Comparative Analysis” American Political Science Review 72(4): 1243-1261. Campbell, A. 2002. “Self Interest, Social Security, and the Distinctive Participation Patterns of Senior Citizens” American Political Science Review 96(3): 565-574. Carapinha, Joao, Dennis Ross-Degnan, Abayneh Tamer Desta, and Anita Wagner. 2010. “Health Insurance Systems in Five Sub-Saharan African Countries: Medicine Benefits and Data for Decision Making.” Health Policy 99(3): 193-202. Carbone, Giovani. 2011. “Democratic Demands and Social Policies: The Politics of Health Reform in Ghana.” The Journal of Modern African Studies 49(3): 381-408. Carbone, Giovanni. 2012. “Do New Democracies Deliver Social Welfare? Political Regimes and Health Policy in Ghana and Cameroon.” Democratization 19(2): 157-183. Carlson, Elizabeth. 2015. “Ethnic Voting and Accountability in Africa: A Choice Experiment in Uganda.” World Politics 67(2): 353-385. Carlson, Elizabeth. 2021. “In Defense of a Divided Opposition: Programatic Distribution and Ethnic Minor Party Support.” Political Science Research and Methods 9: 745-759. 140 Carnes, Matthew, and Isabela Mares. 2009. “Social Policy in Developing Countries.” Annual Review of Political Science 12(93): 93-113. Carrin, G., O. Doetinchem, J. Kirigia, I. Mathauer, and L. Musango. 2008. “Social Health Insurance: How Feasible is its Expansion in the African Region?” Development Issues 10(2): 7-10. Chemouni, Benjamin. 2018. “The Political Path to Universal Health Coverage: Power, Ideas, and Community-Based Health Insurance in Rwanda.” World Development 106: 87-98. Christmals, Christmal Dela, Kizito Aidam. 2020. “Implementation of the National Health Insurance Scheme in Ghana: Lessons for South Africa and Low- and Middle-Income Countries.” Risk Management and Healthcare Policy 13: 1879-1904. Corbridge, Stuart. 2005. Seeing Like a State: Governance and Governmentality in India. Cambridge: Cambridge University Press. Dake, Fidelia. 2018. “Examining Equity in Health Insurance Coverage: An Analysis of Ghana’s National Health Insurance Scheme.” International Journal for Equity in Health 17(85): 1-10. Dani, A. 2008. “Putting the “Social” Back in Public Policy – New Frontiers of Social Policy.” in: A. Dani and A. de Haan, eds. Inclusive States: Social Policy and Structural Inequalities. Washington, DC: World Bank. de Kadt, D., and E. S. Liberman. 2020. “Nuanced Accountability: Voter Responses to Service Delivery in Southern Africa.” British Journal of Political Science 50(1): 185-215. de la Cuesta, Brandon, Lucy Martin, Helen Milner, and Daniel Nielson. 2013. “Do Indirect Taxes Bite? How Hiding Taxes Erases Accountability Demands from Citizens.” Unpublished manuscript. Retrieved from: https://scholar.princeton.edu/sites/default/files/hvmilner/files/ownership_paper.pdf. De La O, Anna. 2012. “Do Conditional Cash Transfers Affect Electoral Behavior? Evidence From A Randomized Experiment in Mexico.” American Journal of Political Science 57(1): 1- 14. De La O, Ana. 2015. Crafting Policies to End Poverty in Latin America: The Quiet Transformation. Cambridge: Cambridge University Press. de Leeuw, Evelyn, Carole Clavier, and Eric Breton. 2014. “Health Policy – Why Research it and How: Health Political Science.” Health Research Policy and Systems 12(55): 1-10. Desai, R., A. Olosgard, and T. Yousef. 2009. “The Logic of Authoritarian Bargains” Economics and Politics 21: 93-125. 141 Diaz-Cayeros, Alberto, Federico Estevez, and Beatrice Magaloni. 2009. “Welfare Benefits, Canvassing, and Campaign Handouts.” In Consolidating Mexico’s Democracy, ed. Jorge I. Dominguez, Chappel Lawson, and Alejandro Moreno. Baltimore, MD: Johns Hopkins University Press Dionne, Kim Yi, and Karen A. Grepin. 2013. “Democratization and Universal Health Coverage: A Case Comparison of Ghana, Kenya, and Senegal.” Global Health Governance. 6(2): 1-27. Dror, DM, SAS Hossain, A. Majumder, TL Perez, D. John, PK Panda. 2016. “What Factors Affect Voluntary Uptake of Community-Based Health Insurance in Low- and Middle-Income Countries? A Systematic Review and Meta-Analysis.” PLoS One 11: 1-31. Esping-Andersen, G. 1985. Politics Against Markets: The Social Democratic Road to Power. Princeton: Princeton University Press. Esping-Andersen, G. 1990. The Three Worlds of Welfare Capitalism. Princeton: Princeton University Press. Fearon, James. 1999. “Electoral Accountability and the Control of Politicians: Selecting Good Types versus Sanctioning Poor Performance.” in Democracy, Accountability, and Representation, ed. Adam Przeworski, Susan Stokes, and Bernard Manin. Cambridge, MA: Cambridge University Press. Fenny, Ama Pokuaa, Robert Yates, and Rachel Thompson. 2018. “Social Health Insurance Schemes in Africa Leave out the Poor.” International Health 10: 1-3. Fusheini, Adam. 2016. “The Politico-Economic Challenges of Ghana’s National Health Insurance Scheme Implementation.” International Journal of Health Policy and Management 5(9): 543-552. Garret, G., and D. Mitchell. 2001. “Globalization, Government Spending, and Taxation in the OECD.” European Journal of Political Research 39(2): 145-177. Golden, Miriam, and Brian Min. 2013. “Distributive Politics Around the World.” Annual Review of Political Science 16: 73-99. Goldsmith, Michael. 1995. “The Growth of Government.” In The Scope of Government, eds Ole Borre and Elinor Scarbrough. Oxford: Oxford University Press. Gomanee, K., O. Morrissey, P. Mosley, and A. Verschoor. 2005. “Aid, Government Expenditure, and Aggregate Welfare.” World Development 33(3): 355-370. Graham, Emmanuel, Ransford Gyampo, Pamphilious Faanu, and Eric Yobo. 2017. “The Third Peaceful Transfer of Power and Democratic Consolidation in Ghana.” Journal of Pan African Studies 10(5): 99-128. 142 Gottlieb, Jessica. 2016. “Greater Expectations: A Field Experiment to Improve Accountability in Mali.” American Journal of Political Science 60(1): 143-157. Gros, Jean-Germaine. 2016. Healthcare Policy in Africa: Institutions and Politics from Colonialism to the Present. Maryland, US: Rowman and Littlefield. Gyimah-Boadi, Emmanuel. 2009. “Another Step Forward for Ghana,” Journal of Democracy. 20(2): 138-152. Haggard, S., and R. Kaufman. 2008. Democracy, Development, and Welfare States: Latin America, East Asia and Eastern Europe. Princeton: Princeton University Press. Harding, Robin. 2019. “Who is Democracy Good For? Elections, Rural Bias, and Health and Education Outcomes in sub-Saharan Africa.” The Journal of Politics 82(1): 241-254. Harding, Robin. 2015. “Attribution and Accountability: Voting for Roads in Ghana.” World Politics 67(4): 656-689. Harding, Robin, and David Stasavage. 2014. “What Democracy Does (and Doesn’t Do) for Basic Services: School Fees, School Inputs, and African Elections.” The Journal of Politics 76(1): 229-245. Hern, Erin Accampo. 2017. “Better than Nothing: How Policies Influence Political Participation in Low-Capacity Democracies.” Governance 30(4): 583-600. Hicken, Allen. 2011. “Clientelism.” Annual Review of Political Science 14: 289-310. Hickey, S., Lavers, T., Nino-Zarazua, M., and J. Seekings. 2019. “The Negotiated Politics of Social Protection in sub-Saharan Africa.” WIDER Working Paper No 34. Retrieved from https://www.wider.unu.edu/publication/negotiated-politics-social -protection-sub- saharan-africa. Huber, E., J.D. Stephens. 2001. Development and Crisis of the Welfare State: Parties and Policies in Global Markets. Chicago: Chicago University Press. Imai, Kosuke, Gary King, and Carlos Velasco Rivera. 2020. “Do Non-Partisan Programmatic Policies Have Partisan Electoral Effects? Evidence from Two Large-Scale Experiments.” The Journal of Politics 82(2): 714-730. Jehu-Appiah, Caroline, Genevieve Aryeetey, Ernst Spaan, Thomas de Hoop, Irene Agyepong, and Rob Baltussen. 2011. “Equity Aspects of the National Health Insurance Scheme in Ghana: Who is Enrolling, Who is Not, and Why?” Social Science and Medicine 72: 157-165. Keefer, Philip. 2007. “Clientelism, Credibility, and the Policy Choices of Young Democracies.” American Journal of Political Science 51(4): 804-821. 143 Keefer, Philip, and Razvan Vlaicu. 2008. “Democracy, Credibility, and Clientelism.” Journal of Law, Economics, and Organization 24(2): 371-406. Kipo-Sunyezhi, Daniel D., Martin A. Ayanore, Daniel K. Dzidzonu, and Yakubu A. Yakubu. 2019. “Ghana’s Journey Toward Universal Health Coverage: The Role of the National Health Insurance Scheme.” European Journal of Investigating in Health, Psychology and Education 10: 94-109. Kitschelt, Herbet. 2012. “Linkages Between Citizens and Politicians in Democratic Polities.” Comparative Political Studies 33(6-7): 845-879. Kitschelt, Herbert, and Steven Wilkinson. 2007. Patrons, Clients, and Policies. New York: Cambridge University Press. Kitschelt, H. and S. Wilkinson. 2007. “Citizen-Politician Linkages: An Introduction.” in Patrons, Clients and Politics: Patterns of Democratic Accountability and Political Competition, ed. H. Kitschelt and S. Wilkinson. Cambridge, MA: Cambridge University Press. Knutsen, Carl Henrik, and Magnus Rasmussen. 2018. “The Autocratic Welfare State: Old- Age Pensions, Credible Commitments, and Regime Survival.” Comparative Political Studies. 51(5): 659-695. Koter, Dominica. 2013. “King Makers: Local Leaders and Ethnic Politics in Africa.” World Politics 65(2): 187-232. Kramon, E. and Daniel Posner. 2016. “Ethnic Favoritism and Education in Kenya.” Quarterly Journal of Political Science 11(1): 1-58. Kumlin, Staffan, and Bo Rothstein. 2005. “Making and Breaking Social Capital: The Impact of Welfare-State Institutions.” Comparative Political Studies 38(4): 339-365. Kusi, Anthony, Ulrika Enemark, Kristian Hansen, Felix Asante. 2015. “Refusal to Enroll in Ghana’s National Health Insurance Scheme: Is Affordability the Problem?” International Journal for Equity in Health 14(2): 1-14. Labonne, Julien. 2013. “The Local Electoral Impacts of Conditional Trash Transfers: Evidence from a Field Experiment.” Journal of Development Economics. 104: 73-88. Lake, David, and Matthew Baum. 2001. “The Invisible Hand of Democracy: Political Control and the Provision of Public Services.” Comparative Political Studies 34(6): 587-621. Larreguy, Horacio, John Marshall, and Lauran Trucco. 2015. “Breaking Clientelism or Rewarding Incumbents? Evidence from an Urban Titling Program in Mexico.” Working paper. Harvard University. http://goo.gl/P4YXk. 144 Lavers, Tom. 2016. “Social Protection in an Aspiring “Developmental State”: The Political Drivers of Community-Based Health Insurance in Ethiopia.” ESID Working Paper No. 71. Manchester: The University of Manchester. Lavers, Tom. 2019. “Social Protection in an Aspiring “Developmental State”: The Political Drivers of Ethiopia’s PSNP.” African Affairs 184: 646-671. Lavers, Tom, and Sam Hickey. 2016. “Conceptualizing the Politics of Social Protection Expansion in Low Income Countries: The Intersection of Transnational Ideas and Domestic Politics.” International Journal of Social Welfare 25: 388-398. Layton, Matthew and Amy Smith. 2015. “Incorporating Marginal Citizens and Voters: The Conditional Electoral Effects of Targeted Social Assistance in Latin America.” Comparative Political Studies 48(7): 854-881. Lindberg, Staffan. 2010. “What Accountability Pressures do MPs in Africa Face and How Do They Respond? Evidence from Ghana.” The Journal of Modern African Studies 48(1):117- 142. Lindberg, Staffan and Minion Morrison. 2008. “Are African Voters Really Ethnic or Clientelistic? Survey Evidence from Ghana.” Political Science Quarterly 123(1): 95-126. Lindberg, Staffan. 2012. “Have the Cake and Eat It: The Rational Voter in Africa.” Party Politics 19(6): 945-961. MacLean, Lauren M. 2011. “State Retrenchment and the Exercise of Citizenship in Africa.” Comparative Political Studies 44(9): 1238-1266. Magaloni, Beatrice. 2006. Voting for Autocracy: Hegemonic Party Survival and its Demise in Mexico. New York: Cambridge University Press. Majumdar, Sumon, Anandi Mani, and Sharun Mukand. 2004. “Politics, Information, and the Urban Bias.” Journal of Development Economics 75: 137-165. Mares, Isabela, and Matthew Carnes. 2009. “Social Policy in Developing Countries.” Annual Review of Political Science. 12: 93-113. Mani, Anandi, and Sharun Mukand. 2007. “Democracy, Visibility, and Public Good Provision,” Journal of Development Economics 83: 506-529. Martin, Lucy. 2016. “Taxation, Loss Aversion, and Accountability: Theory and Experimental Evidence for Taxation’s Effect on Citizen Behavior.” Working paper. Mcloughlin, Claire. 2015. “When Does Service Delivery Improve the Legitimacy of a Fragile or Conflict-Affected State?” Governance 28(3): 341-356. 145 Mensah, Joseph, Joseph Oppong, and Christoph Schmidt. 2010. “Ghana’s National Health Insurance Scheme in the Context of the Health MDGs: An Empirical Evaluation Using Propensity Score Matching.” Health Economics 19: 95-106. Mettler, Susan and J. Stonecash. 2008. “Government Program Usage and Political Voice.” Social Science Quarterly 89: 273-293. Mladovsky, P. 2014. “Why Do People Drop Out of Community-Based Health Insurance? Findings from an Exploratory Household Survey in Senegal.” Social Science and Medicine 107: 78-88. Mohammed, M., and M. Seidu. 2014. “Assessing the Quality of Primary Health Care Provided by Accredited Service Providers in the Upper East Region, Ghana.” International Journal of Economics, Commerce, and Management 11(10): 1-32. Morrissey, O. 2009. “Aid and the Financing of Public Social Sector Spending.” in K. Hujo and S. McClanahan, eds. Financing Social Policy: Mobilizing Resources for Social Development. Basingstoke: Palgrave Macmillan. Mosley, P., J. Hudson, and A. Verschoor. 2004. “Aid, Poverty Reduction, and the “New Conditionality.” The Economic Journal. 114(496): 217-243. National Health Insurance Authority. 2015. Ghanaian National Health Insurance Authority 2014 Annual Report. Accra, Ghana: NHIA. National Health Insurance Authority. 2014. Ghanaian National Health Insurance Authority 2013 Annual Report. Accra, Ghana: NHIA. National Health Insurance Authority. 2013. Ghanaian National Health Insurance Authority 2012 Annual Report. Accra, Ghana: NHIA. National Health Insurance Authority. 2012. Ghanaian National Health Insurance Authority 2011 Annual Report. Accra, Ghana: NHIA. National Health Insurance Authority. 2011. Ghanaian National Health Insurance Authority 2010 Annual Report. Accra, Ghana: NHIA. National Health Insurance Authority. 2010. Ghanaian National Health Insurance Authority 2009 Annual Report. Accra, Ghana: NHIA. National Health Insurance Authority. 2009. Ghanaian National Health Insurance Authority 2008 Annual Report. Accra, Ghana: NHIA. National Health Insurance Authority. 2008. Ghanaian National Health Insurance Authority 2007 Annual Report. Accra, Ghana: NHIA. 146 National Health Insurance Authority. 2007. Ghanaian National Health Insurance Authority 2006 Annual Report. Accra, Ghana: NHIA. National Health Insurance Authority. 2006. Ghanaian National Health Insurance Authority 2005 Annual Report. Accra, Ghana: NHIA. National Democratic Congress. 2000. Manifesto 2000: Spreading the Benefits of Development. Accra: NDC. Nelson, Joan M. 2007. “Elections, Democracy, and Social Services.” Studies in Comparative International Development 41(4): 79-97. NEPAD – African Union Development Agency. 2014. AUDA-NEPAD 2014 Annual Report. Midrand, ZA: AUDA-NEPAD Nino-Zarazua, M., A. Barrientos, S. Hickey, and D. Hulme. 2012. “Social Protection in Sub- Saharan Africa: Getting the Politics Right” World Development 40(1): 163-176. Nsiah-Boateng, Eric, and Moses Aikins. 2018. “Trends and Characteristics of Enrollment in the National Health Insurance Scheme in Ghana: A Quantitative Analaysis of Longitudinal Data.” Global Health Research and Policy 3(32): 1-10. Onoka, C., K. Hanson, and J. Hanefeld. 2015. “Towards Universal Coverage: A Policy Analaysis of the Development of the National Health Insurance Scheme in Nigeria.” Health Policy and Planning 30(9): 1105-1117. Ortega, D and M. Penfold-Becerra. 2008. “Does Clientelism Work? Electoral Returns of Excludable and non-Excludable Goods in Chavez’s Misiones Programs in Venezuela.” Unpublished paper, Inst. Estud. Superiores de Administracion, Caracas, Venezeula. Owusu-Sekyere, E., D. Bagah. 2014. “Towards Sustainable Healthcare Financing in Ghana: Is National Health Insurance the Solution?” Public Health Research 4(5): 185-194. Paler, Laura. 2013. “Keeping the Public Purse: An Experiment in Windfalls, Taxes, and the Incentives to Restrain Government.” American Political Science Review 107(4): 706-725. Pierce, Jon, Tatiana Kostova, and Kurt Dirks. 2001. “Toward a Theory of Psychological Ownership in Organizations.” Academy of Management Review 26(2): 298-310. Pierson, Paul. 1994. Dismantling the Welfare State? Reagan, Thatcher, and the Politics of Retrenchment. Cambridge Studies in Comparative Politics. Cambridge: Cambridge University Press. Pop-Eleches, Cristian, and Grigore Pop-Eleches. 2012. “Targeted Government Spending and Political Preferences.” Quarterly Journal of Political Science 7(3): 285-320. 147 Posner, Daniel. 2005. Institutions and Ethnic Politics in Africa. Cambridge, UK: Cambridge University Press. Przeworski A., M. Alvarez, J.A. Cheibub, and F. Limongi. 2000. Democracy and Development: Political Institutions and Well-being in the World, 1950-1990. Cambridge Studies in the Theory of Democracy. New York: Cambridge University Press. Rodrick, D. 1997. Has Globalization Gone Too Far? Washington DC: Institute for International Economics. Sakyi, EK, RA Atinga, and FA Adzei. 2012. “Managerial Problems of Hospitals Under Ghana’s National Health Insurance Scheme.” Clinical Governance: An International Journal 17(3): 178-190. Salari, Paola, Patricia Akweongo, Moses Aikins, and Fabrizio Tediosi. 2019. “Determinants of Health Insurance Enrollment in Ghana: Evidence from Three National Household Surveys.” Health Policy and Planning 34(8): 582-594. Sandbu, Martin. 2006. “Natural Wealth Accounts: A Proposal for Alleviating the Natural Resource Course.” World Development 34(7): 1153-1170. Skocpol, T., and E. Amenta. 1986. “States and Social Policies.” Annual Review of Sociology 12: 131-157. Skocpol, Theda. 1992. “State Formation and Social Policy in the United States.” American Behavioral Scientist 35(4-5): 559-584. Skocpol, T., and M. Weir. 1985. “State Structures and the Possibilities for “Keynesian” Responses to the Great Depression in Sweden, Britain, and the United States.” In Bringing the State Back In. eds P. Evans, E. Rueschemeyer, and T. Skocpol. New York: Cambridge University Press. Soss, Joe. 1999. “Lessons of Welfare: Policy Design, Political Learning, and Political Action.” The American Political Science Review 93(2): 363-380. Soss, Joe and Sanford Schram. 2007. “A Public Transformed? Welfare Reform as Policy Feedback.” American Journal of Political Science 101(1): 111-127. Stasavage, David. 2005. “Democracy and Education Spending in Africa.” American Journal of Political Science 49(2): 343-358. Stasavage, David. 2005. “The Role of Democracy in Uganda’s Move to Universal Primary Education.” The Journal of Modern African Studies 43(1): 53-73. Stokes, Susan. 2007. “Political Clientelism.” in Oxford Handbook of Comparative Politics, ed. C. Boix and S. Stokes. Oxford, UK: Oxford University Press. 148 Stokes, Susan, Thad Dunning, Marcelo Nazareno, and Verlia Brusco. 2013. Brokers, Voters, and Clientelism: The Puzzle of Redistributive Politics. Cambridge: Cambridge University Press. Streefland, Pieter. 2005. “Public Health Care Under Pressure in Sub-Saharan Africa.” Health Policy 71(3): 375-382. Swenson, P. 1991. “Bringing Capital Back In, or Social Democracy Reconsidered: Employer Power, Cross-Class Alliances, and Centralization of Industrial-Relations in Denmark and Sweden.” World Politics. 43(4): 513-544. Thiel, Alena. 2020. “Biometric Identification Technologies and the Ghanaian Data Revolution.” Journal of Modern African Studies 58(1): 115-136. United Nations. 2016. The Sustainable Development Goals Report: 2016. New York: United Nations Vaughan, Megan. 1991. Curing Their Ills: Colonial Power and African Illness. Palo Alto: Stanford University Press. van de Walle, Nicholas. 2003. “Presidentialism and Clientelism in Africa’s Emerging Party Systems.” Journal of Modern African Studies 41(2): 297-321. Wahab, Hassan. 2019. “The Politics of State Welfare Expansion in Africa: Emergence of National Health Insurance in Ghana, 1993-2004.” Africa Today 65(3): 91-112. Wang, Huihui, Nathaniel Otoo, and Lydia Dsane-Selby. 2017. Ghana National Health Insurance Scheme: Improving Financial Sustainability Based on Expenditure Review. Washington DC: World Bank. Wantchekon, Leonard. 2003. “Clientelism and Voting Behavior: Evidence from a Field Experiment in Benin.” World Politics 55(3): 399-422. Wardle, Pete. 2008. “The Cost of Politics: Synthesis Report.” Westminster Foundation for Democracy. Weitz-Shapiro, Rebecca. 2012. “Curbing Clientelism: Politics, Poverty, and Social Policy in Argentina.” Unpublished manuscript. Brown University. Weitz-Shapiro, Rebecca. 2014. Curbing Clientelism in Argentina: Politics, Poverty, and Social Policy. New York: Cambridge University Press. Weyland, Kurt. 2005. “Theories of Policy Diffusion: Lessons from Latin American Pension Reform.” World Politics 57(2): 262-295. 149 Wilensky, H. 1975. The Welfare State and Equality: Structural and Ideological Roots of Public Expenditures. Berkeley and Los Angeles: University of California Press. Witter, Sophie, and Bertha Garshong. 2009. “Something Old or Something New? Social Health Insurance in Ghana.” BCM International Health and Human Rights 9(20). Wong, J. 2006. Healthy Democracies: Welfare Politics in Taiwan and South Korea. Ithaca: Cornell University Press. World Health Organization. 2018. World Health Statistics 2018: Monitoring Health for the Sustainable Development Goals. Geneva: World Health Organization World Health Organization Regional Office for Africa. 2018. The State of Health in the WHO African Region: An Analysis of the Status of Health, Health Services and Health Systems in the Context of the Sustainable Development Goals. Brazzaville: WHO Regional Office for Africa. Zucco. Cesar Jr. 2013. “When Payouts Payoff: Conditional Cash Transfers and Voting Behavior in Brazil 2002-10.” The American Journal of Political Science 57(4): 810-822. 150