ABOUT THE MAP
The Patient Protection and Affordable Care Act of 2010, a.k.a. "Obamacare," makes significant and wide-ranging changes to the U.S. health care system. Among the most important of these is the law's restructuring of the individual market for health insurance: the market used by people who shop for coverage on their own, instead of having that coverage purchased on their behalf by their employers or the government.
The Obamacare Rate Map, a product of the Manhattan Institute's Center for Medical Progress, is a tool for policymakers, researchers, and ordinary Americans to understand the ways in which the Affordable Care Act affects the individual market for health insurance. The Map presents three categories of information on a state-by-state basis: (1) the gross impact of the ACA on health insurance premiums in the individual market; (2) the percentage of overall and uninsured Americans who qualify for subsidized coverage on the ACA's insurance exchanges; and (3) the net impact of premium changes and subsidies on Americans of different ages.
The Obamacare Rate Map was produced by Yevgeniy Feyman, a fellow at the Manhattan Institute; Avik Roy, a senior fellow at the Manhattan Institute and editor of The Apothecary at Forbes.com; and Paul Howard, senior fellow at the Manhattan Institute and director of the Manhattan Institute's Center for Medical Progress.
In order to adequately assess the information required to develop the Obamacare Impact Map, we used three separate approaches for the three separate sets of data presented in the map.
In order to document rate changes, we first gathered pre-ACA insurance rates using the federal government's finder.healthcare.gov website. Our pre-ACA dataset consists of the five least expensive plans (by monthly premium) for the most populous zip code in every county. To cover a significant age range we collected rates for 27, 40, and 64-year old male and female non-smokers. We adjusted these rates to take into account those who are denied health insurance coverage as well as those who receive a surcharge. Using the "denial rate" and "surcharge rate" from the federal government's repository, we assumed that those who are surcharged pay 75 percent more and those who are denied, find insurance elsewhere at three times the original rate. We used this to develop a weighted average of the five least expensive insurance plans for every zip code we identified. To develop a state-wide average, we took the state-wide average for every age-gender combination.
For ACA rates, we created state-level averages by averaging rates for the five cheapest plans across all counties in a state. The data was sourced from ValuePenguin and because the ACA bans denials based on pre-existing conditions, there is no need to develop a weighted average of these rates.
In order to tabulate the total number of potential subsidy beneficiaries we used the University of Minnesota Population Center's Integrated Public Use Microdata (IPUMS) for the Census Bureau's Current Population Survey (CPS). We used the 2012 March Economic Supplement as the primary data source.
It is important to note that we opted not to use replicate weights - instead we used the standard person-weights while using states as our strata. The reason for this is that the improvement in standard errors from using replicate weights was relatively minor given the additional computation time required to implement them.
In determining who would eligible for subsidies, we assumed a complete Medicaid expansion - and therefore excluded those below 139 percent of the federal poverty level (the ACA expands Medicaid to 133 percent with a five percent income disregard, making it effectively up to 138 percent). While the ACA will use 2013 income to determine eligibility, we used 2012 income and therefore used 2012 cutoffs to determine the federal poverty level as well. This should not make a material difference in the final results - as income grows in 2013, so does the federal poverty level.
The data is reported in two groups - the first column shows the number of uninsured, subsidy-eligible individuals as a percentage of the total uninsured population for each age group. The second column goes a step further - because individuals currently in the individual market may end up transitioning to the exchanges, we identify those currently purchasing individual health insurance who would be eligible for subsidies. We add this number together with the number of total uninsured eligible for subsidies and report it as a percent of the entire population for each age group. Put simply, the numerator and denominator for the first pie chart are: the total number of uninsured potential subsidy beneficiaries and the total number of uninsured above 138 percent of FPL, respectively (to exclude those eligible for the Meaboutdicaid expansion). For the second pie chart, the numerator is the total number of insured eligible individuals in the individual market plus the number of uninsured potential subsidy beneficiaries; the denominator is the total population. In most cases this does not include government programs - in the case of New York, however, the Healthy New York population will be transitioned to the exchanges and thus was included in the count.
Additionally, we report the data along five age groups - 20 and under; 21 to 30; 31 to 40; 41 to 50; and 51 and up.
Lastly, we use Adjusted Gross Income (rather than Modified Adjusted Gross Income) to determine subsidy eligibility; this should not make a material difference in the findings.
The focus of the third dataset was to address the effects of premiums on the cost of insurance. To do so, we used 2012 American Community Survey (ACS) data (also sourced from IPUMS) to find median household incomes for three ages - 27, 40, and 64-year olds - as well as the median household size and found the subsidy-adjusted net cost of insurance under the ACA for each representative household, using the average of the five cheapest plans for this calculation (household here is defined as a "family unit"). In addition, we built "breakeven points" (more precisely, critical points) which reflect the household income level, for a given representative individual, where a subsidy-adjusted rate increase (if there is one) turns into a subsidy-adjusted rate decrease.
This analysis was broader than in previous sections primarily because we took into account the entire population of each of the three ages studied. Taking into account the entire population allows for "churn" in insurance markets. That is, a 27-year old who is currently insured through his employer may eventually end up on the exchanges at some point. The caveat with this approach, of course, is that the focus is on the population as a whole rather than just the uninsured.
A few important notes are also in order. The breakeven point is only available for those states where we found rate increases for a median income household with subsidies. In states with a decrease, the inflection point is labeled with as "0."
*Methodological Caveats: As of this writing, we have been unable to find data for Hawaii for datasets 1 and 2. Additionally, the average of 5 cheapest plans in Alabama ends up being more expensive than the second-cheapest silver plan (which is used to calculate subsidies). Thus, in Alabama, we assume, for the purposes of developing a net cost of insurance for the median individual, that the individual purchases the second-cheapest silver plan.