Loading field map
Health policy evaluation has always been pulled between two distinct questions. One is normative: which policy produces the greatest social benefit? The other is positive: why do policies take the shape they do, and what determines whether they succeed or fail in practice? The frameworks that have shaped this subfield over the past six decades can be understood as different answers to these questions, each responding to the limitations of its predecessors while carving out a lasting role in the evaluator's toolkit.
Cost-Benefit Analysis (CBA) emerged in the 1960s as the first systematic framework for evaluating health policies within the welfare economics tradition. Its core commitment was straightforward: monetize all consequences of a policy—health gains, productivity improvements, even lives saved—and compare the sum of benefits to the sum of costs. If benefits exceeded costs, the policy was judged efficient. CBA gave evaluators a clear decision rule rooted in the Kaldor-Hicks compensation principle, which held that a policy was worthwhile if the winners could theoretically compensate the losers. In practice, however, monetizing health outcomes proved deeply controversial. Placing a dollar value on a life year saved required assumptions about willingness to pay that many clinicians and policymakers found ethically troubling. This tension between theoretical elegance and practical acceptability created the opening for an alternative.
Cost-Effectiveness Analysis (CEA) emerged in the 1970s as a direct normative alternative to CBA, narrowing the evaluation question while preserving the welfare-economic logic. Instead of monetizing health outcomes, CEA measured them in natural units—life years gained, cases averted, or, most influentially, quality-adjusted life years (QALYs). The analyst then calculated a cost-per-QALY ratio and compared it against a threshold. CEA did not reject CBA's efficiency orientation; it preserved the core idea of comparing costs to outcomes. But by sidestepping the monetization of health, CEA became far more palatable to health ministries and clinical guideline bodies. The two frameworks have coexisted ever since, with CBA remaining dominant in multi-sector infrastructure projects and CEA becoming the default for health technology assessment. Their rivalry is not a matter of one replacing the other but of a narrowing of scope: CEA absorbed CBA's efficiency logic while discarding its most politically vulnerable feature.
The 1980s brought a fundamental reorientation. The Political Economy of Health Policy framework shifted the evaluator's gaze from what policies should be to why they are what they are. Drawing on public choice theory and institutional analysis, this framework argued that health policies are not the product of technocratic optimization but of bargaining among interest groups—physician lobbies, pharmaceutical firms, insurers, patient advocacy organizations—operating within institutional constraints. A CEA might show that a single-payer system is cost-effective, but political economy explains why such a system is politically infeasible in a given country. This framework did not reject the normative ambitions of CBA or CEA; it simply bracketed them. Instead of asking whether a policy was efficient, political economy asked whose interests it served, how institutional rules shaped the set of feasible alternatives, and why inefficient policies often persist. The result was a productive tension: normative frameworks provided benchmarks for what could be achieved, while political economy explained the gap between the benchmark and reality.
The 1990s witnessed a methodological explosion that transformed how health policies are empirically evaluated. Microeconometrics in Health Economics, often called the credibility revolution, brought a sharp focus on causal identification. Researchers armed with difference-in-differences, instrumental variables, regression discontinuity, and later randomized controlled trials sought to answer a deceptively simple question: what is the causal effect of a specific policy on a specific outcome? The framework's defining commitment was to internal validity—eliminating confounding to isolate treatment effects. This represented a narrowing of ambition compared to the structural models that preceded it. Microeconometricians were willing to trade theoretical richness for credible estimates. A typical study might ask: what was the effect of Medicaid expansion on hospital utilization? The answer came from comparing outcomes in expansion and non-expansion states before and after the policy change, with minimal reliance on economic theory. This approach coexisted uneasily with the normative frameworks: microeconometrics could estimate effects, but it could not by itself say whether those effects were worth the cost.
Running parallel to the credibility revolution, Structural Econometric Modeling in Health took the opposite methodological path. Structural models began with explicit economic theory—often a utility-maximizing agent facing dynamic choices about insurance, treatment, or preventive care—and used data to estimate the deep parameters of that theory. Once estimated, the model could simulate counterfactual policies that had never been observed. Where microeconometrics prioritized internal validity, structural modeling prioritized external validity and the ability to predict behavior under entirely new regimes. The two frameworks have been in living disagreement since the 1990s. Microeconometricians accuse structural modelers of making untestable assumptions; structural modelers counter that reduced-form estimates are useless for predicting the effects of policies that differ from anything seen in the data. In practice, the frameworks have begun to converge: modern structural work often embeds credible identification strategies, and microeconometricians increasingly use theory to guide their choice of instruments and control variables. But the philosophical divide—between learning from variation in observed data and learning from a fully specified model—remains the central methodological debate in the subfield.
The 2000s introduced a framework that cut across both the normative and positive traditions. Behavioral Health Economics challenged the rational-choice assumptions that underlay CBA, CEA, and most structural models. Drawing on psychology and experimental economics, behavioral economists showed that patients and providers systematically deviate from the predictions of expected utility theory: they procrastinate on preventive care, are overly influenced by default options, and struggle with complex insurance choices. This framework did not replace its predecessors; instead, it transformed them. CEA now routinely incorporates behavioral parameters, such as time-inconsistent preferences, into its models. Structural modelers have begun building behavioral microfoundations—hyperbolic discounting, reference-dependent preferences—into their dynamic choice frameworks. Microeconometricians have used behavioral insights to design better experiments, testing how framing and defaults affect health decisions. Behavioral health economics thus functions as a critique and a complement, pushing every other framework to confront the gap between the rational agent of theory and the actual human being making health decisions.
Today, no single framework dominates health policy evaluation. CBA and CEA remain the workhorses of formal health technology assessment, with CEA dominant in public systems and CBA preferred in multi-sector regulatory analysis. Political economy provides the essential context for understanding why evidence-based policies are often ignored or distorted. Microeconometrics supplies the gold standard for causal evidence, while structural modeling offers the tools for forecasting the effects of novel policies. Behavioral health economics has become a cross-cutting lens that enriches all of these approaches.
What the leading frameworks agree on is that evaluation must be empirical and transparent: assumptions should be stated, data should be public, and sensitivity analysis should be routine. Where they disagree is on the role of theory. Microeconometricians tend to view theory as a source of identifying assumptions to be minimized; structural modelers see theory as indispensable for prediction; behavioral economists see standard theory as a flawed starting point that must be revised. The subfield's vitality comes from this disagreement. A student entering health policy evaluation today inherits not a settled toolkit but a set of competing frameworks, each with distinctive strengths, and the challenge of knowing when to use which.
The history of health policy evaluation is not a story of linear progress from crude to sophisticated methods. It is a story of frameworks that narrowed, broadened, challenged, and absorbed one another. CBA established the welfare-economic foundation; CEA narrowed it to avoid monetization; political economy broadened the lens to include power and institutions; microeconometrics and structural modeling split over the role of theory; and behavioral economics undermined the rational-agent assumptions that had held the whole edifice together. Each framework remains active because each answers a question the others leave unanswered. The evaluator's task is not to choose one framework forever but to understand what each can and cannot do.