By pooling five years of the Medical Expenditure Panel Survey (MEPS), 2005-2009, this paper provides information from the recent past about the population of adults with MAGI between 100 and 138% of FPG to help inform state decisions in the near future.  Specifically, what is the pre-ACA insurance status of this population, including those who are currently eligible but not enrolled in previous Medicaid expansions?  How much might they spend if enrolled in Medicaid or private insurance through the exchanges?

            Both simulation and regression methods are used.  Medicaid eligibility under current rules is simulated using detailed, state-specific eligibility rules and details collected in the MEPS about family relationships and types and amounts of personal and family income.  Expanded Medicaid eligibility and exchange eligibility for adults are simulated based on final federal regulations.  In particular, they are not Medicare beneficiaries and not eligible for Medicaid due to disability.  We estimate that, at a point in time, 8.6 million adults were in the population that could be eligible for the exchanges or Medicaid.  Of these, 52% were uninsured, 32% had employment-related insurance or Tricare, 12% were enrolled in Medicaid, and 4% had other public or private insurance.  Most (55%) had children, who will be eligible for Medicaid. 

            Regression analysis is used to predict the potential effects of public versus private insurance on out-of-pocket spending for medical care on the uninsured (N=5,696).  For the exchanges, out-of-pocket spending is estimated using those with MAGI below 200% of FPG and insurance through large employers for the entire year (N=5,000).  While the benchmark plans for the exchanges are small employers, the actuarial values of the plans for which this population will be eligible will be much closer to those of large employers.  For Medicaid, adults enrolled in Medicaid for the entire year, and who are not eligible due to disability or through pathways with limited benefits are the estimation sample (N=3,500).  Similarly, their uninsured children’s Medicaid spending is predicted from children enrolled in Medicaid the entire year (N=12,000).  The regressions are estimated using two part models.  The first parts are the probability of incurring any out-of-pocket spending.  The second parts are generalized linear models of the level of expenditures among those with expenditures.  Regressions control for health status, attitudes, health behaviors, socioeconomic and geographic characteristics, and year fixed effects.  Total out-of-pocket spending in the exchanges includes spending on premiums, 2% of MAGI for this population.