The Impact of Health Insurance in Low- And Middle-Income Countries

Chapter 10: Main Findings, Research Issues, and Policy Implications

178 Main Findings, Research Issues, and Policy Implications Maria-Luisa Escobar, Charles C. Griffin, and R. Paul Shaw This book contains rich and varied analyses of the impact of health insurance in different socioeconomic and organizational settings. It begins with a comprehensive literature review that distills findings on prior studies that examine causal effects between health insurance and health outcomes. This is followed by seven country case studies, most of which use advanced statistical techniques and new data sources to shed light on how health insurance improves health outputs and outcomes. This chapter summarizes the main findings, the methodological issues that can understate or diminish the estimated impact of health insurance on health, and the country scenarios that illustrate the art of the possible for policymakers interested in scaling up well designed health insurance programs. To a large extent, selecting the countries was opportunistic, as explained in chapter 1. In view of pressing policy concerns in many low-income countries, as well as major gaps in our knowledge of the impacts of health insurance, we chose countries where scaling up health insurance aims to be more inclusive of the poor and where available data permit explorations of the impact on health status. No pretense is made, therefore, that findings reported here come from a uniform dataset or research methodology applied to all countries or that all studies satisfy the “gold standard” for empirical robustness as described in chapter 2. Instead, several caveats and qualifiers should be kept in mind. Chapter 10 Main Findings, Research Issues, and Policy Implications179 Above all, we have learned that health insurance is not a homogeneous product (like an approved oral medicine or vaccine). It tends to be heterogeneous in the entitlements to medical goods and services created for health insurance members; the quantity, quality, and distribution of providers where members can access services ; the extent that copayments and deductibles affect out-of-pocket spending by members; and so on. This heterogeneity cannot be controlled or made uniform. It shapes the impact that health insurance has on outputs and outcome measures in one country versus another, so that the measured effects of health insurance on access, service uptake, and out-of-pocket spending vary widely across countries. The temptation to generalize findings across countries must be tempered accordingly. We have also learned that the extent to which health insurance succeeds in being pro-poor has more to do with a purposive effort to design health insurance in a way that benefits the poor than with any presumption that health insurance is automatically and intrinsically pro-poor or anti-poor. Health insurance has important design features that can benefit low-income households, such as pooling contributions by rich and poor households, then paying for treatment of illnesses that disproportionately fall on the poor. But if health insurance fails to enroll the poor or extend services to them, the distributional impacts of health insurance on equity will likely be muted. Accordingly, to generalize that health insurance does, indeed, contribute to greater equity in health care consumption will be conditional on successful pro-poor design features. This caveat also applies to generalizations we might be inclined to make about distributional impacts of health insurance on women and children. Pooling risks is equity improving within the risk pool, but who benefits depends on who is in the risk pool and how it is designed to function. Finally, we have learned that the robustness of empirical analysis varies across case studies. Robustness depends on the comprehensiveness and quality of ­ datasets, availability of appropriate measures of impact, success in controlling for endo­ geneity, and appropriate application of statistical models. This caveat forms the backbone of chapter 2, where checklists of key methodological concerns are used to score the quality of a wide variety of studies and the robustness of the empirical estimates they have reported. Managing these problems well is critical if empirical findings are to be taken seriously. With these caveats in mind, the collective findings in this book do lend themselves to some conservative generalizations that advance not only the evidence base but also contribute information to current policy debates on the desirability of scaling up health insurance. The next section considers the accumulation of evidence in six areas. 180 Chapter 10 Six general findings In this volume the estimated benefits of health insurance among the insured are measured relative to conditions affecting the uninsured. In this...