Date of Award

4-2002

Degree Name

Doctor of Philosophy

Department

Public Affairs and Administration

Abstract

Recent changes in administration of the Medicaid program make it imperative that local agencies improve their ability to forecast demand for Medicaid services. In October of 1998 the State of Michigan redesigned the Medicaid specialty care in Michigan from a fee-for-service system to a capitated system. In a capitated healthcare system, financial risk is a result of unanticipated changes in the population size and mix. Numerous demographic factors, such as crime, population, unemployment, median age, income, and ethnicity may be used to improve the accuracy of predicted changes in enrollment for Medicaid. Using naive ordinary least squares models as the baseline; the study evaluates more sophisticated theory-based models incorporating demographic factors as predictors. The intent is to identify the models that maximize the accuracy of the projection models while minimizing the cost and difficulty of maintaining the models. The predictor data is restricted to data readily available to the state and local agencies that might use the models. Projection models are developed for the three primary Medicaid subpopulations. They are based on groupings of the program codes and logically grouped by general level of service required. Mental Health (MH Aged/Blind), Developmental Disabilities (DD Disabled E&P), and Temporary Assistance to Needy Families (TANF - TANF/Other). These subpopulations are based on different rules of eligibility and are often served by different agencies. Therefore, the models developed are evaluated according to their ability to project the changes in each subpopulation and as a component of a projection for the entire Medicaid eligible population. The overall Medicaid population is projected as a sum of the three subpopulations. In almost every case the conclusion is that the simplest naive models perform as well, or better than, the most complex models considered. Carefully maintained simple models will produce short-term enrollment projections accurate enough to effectively serve the business planning needs of county level agencies.

Access Setting

Dissertation-Open Access

Download

Share

COinS