Every year, millions of Americans experience Medicaid enrollment states that do not align with their eligibility for the program. Status misalignment has economic, health, and social costs for individuals, State governments, insurers, and employers. As an entitlement program, Medicaid uses a means test to assess eligibility for the program; if an individual is determined to be eligible, they are entitled to the benefits of the program. The fact that misalignment occurs regularly and across all State programs, however, indicates that implementing the means test is not straightforward. Despite its importance to citizens, States, and industry, we still have a poor understanding of why enrollment misalignment is prevalent or how it occurs.
Existing research frames phenomena of enrollment misalignment (i.e., missed program take-up, churn, and fraud) as classification errors that represent unintended consequences of implementation or idiosyncratic individual behavior. The underlying Cartesian assumption that policy effects can be reduced to their constituent parts and traced back to prime causes fundamentally limits the ability of this treatment to provide insights about program enrollment. Evidence from health services research and social policy indicate that demographic characteristics, State program structures, and economic and social context each contribute to State variation in program take-up and churn. However, these studies fail to explicitly account for the interdependencies and interconnections among these factors that produce irreducible, complex phenomena.
This dissertation research uses the case of enrollment phenomena resulting from implementation of Medicaid’s means test to explore broader questions about the complex phenomena that characterize collective action systems. Drawing on insights from complex systems science and critical science studies, this research seeks to understand how complex policy systems work and where surprising patterns come from. Using both conceptual and computational systems models, this research explores how decisions of program design (e.g., eligibility criteria) and implementation (e.g., application and eligibility determination procedures) affect how beneficiaries and potential beneficiaries accumulate and move through the Medicaid system.
The primary contribution of this research is a conceptual framework for studying collective action programs as complex adaptive systems. Policy cybernetics integrates insights from systems science, complexity science, and implementation studies to guide inquiry concerning sources of policy resistance in dynamic systems such as Medicaid. The framework provides a logically consistent explanation for how complexity – interdependencies, feedback, emergence, and context – can lead to policy resistance and hamper progress toward stated goals of a program. Assumptions, concepts, and expectations are applicable to other public program contexts, including other social welfare programs, as well as law enforcement, criminal justice, and workforce development. In describing public programs as complex human systems, this research aims to explore and assess strategies to improve program performance across the many criteria against which we measure it (e.g., effectiveness, equity, efficiency).
As a proof of concept, policy cybernetics is applied to the problem of program enrollment in Medicaid. A set of models of Medicaid enrollment system illuminate the feedback structure among agents, rules, and environment that produces dynamic behaviors (e.g., churn) in implementation of the program’s means test. A simulation of Medicaid’s enrollment mechanism permits experimentation with several program interventions in a virtual world. These experiments examine the tradeoffs inherent in public program enrollment, and identify administrative strategies that increase the performance of the program enrollment mechanism (i.e., enrolling eligible individuals, not enrolling ineligible individuals) and minimize costly movement on and off the program (i.e., churn) among beneficiaries over relevant time horizons. These models illustrate that the program’s design, administrative structure, and individual attributes interact to produce program enrollment outcomes, both `intended’ (i.e., take-up) and `unintended’ (i.e., churn). Policy resistance – the intervention-dampening patterns that arise in response to the system itself) – is endemic to this complex system of collective action.
Policy cybernetics helps policy scholars and practitioners alike understand the multidimensional roots of policy resistance and the mechanisms by which it operates. It also provides a framework for exploring alternative strategies to deal with “unintended consequences”, and tradeoffs in policy objectives. Simulations allow decision makers to alter underlying assumptions and mechanisms to assess the consequences of their decisions across long time horizons in silica with zero social costs. Thus, a collectively developed simulation can be used to clarify values, make assumptions explicit, and make updated use of the evidence based in program planning, administrative decision-making, and performance evaluation.