Uninsured New Yorkers After Full Implementation of the Affordable Care Act: Source of Health Insurance Coverage by Individual Characteristics and Sub-State Geographic Area

© 2013, The Urban Institute Health Policy Center • www.healthpolicycenter.org Uninsured New Yorkers After Full Implementation of the Affordable Care Act: Source of Health Insurance Coverage by Individual Characteristics and Sub-State Geographic Area January 2013 Fred Blavin, Linda Blumberg, Matthew Buettgens The Urban Institute developed a New York state-specific version of its Health Insurance Reform Simulation Model (HIPSM) for use in providing analytic support to the state in its effort to assess the implications of the implementation of the Patient Protection and Affordable Care Act of 2010 (ACA). Initial findings from this work were made available in March of 2012.1 As the state develops and designs its health benefits exchange consistent with the requirements of the ACA and stakeholders prepare for the full implementation of the reforms, the need arose for additional detail on the characteristics and areas of residence of those that are likely to be gaining coverage. The tables presented here provide sub-state analyses, consistent with our previous findings, focusing on those without insurance coverage of any kind prior to reform, the baseline uninsured. We show the share of uninsured expected to gain coverage under the ACA, whether through public insurance (Medicaid or the Children’s Health Insurance Program, CHIP) or through private coverage (via the new exchange or outside it). The tables include the distribution of characteristics (age, income, race/ethnicity, health insurance unit type,2 health status, language, gender, and education) for those anticipated to gain insurance of each type whenever sample sizes allow. For these results, we assume the small employer size threshold is set to 100 employees and the small group and direct pay markets remain separate for premium rating purposes. The 100 employee threshold is consistent with the fully phased-in federal requirement, and the separate markets are consistent with the state’s most recent policy decision. Together, these design features are consistent with the “Alternative #1” assumptions in our March 2012 report. The post-reform estimates of the uninsured in the March 2012 report differ somewhat form those presented here. The estimates in the earlier report are the net effects of many uninsured New Yorkers gaining coverage and a small number of other New Yorkers becoming uninsured post-reform. The estimates in the present analysis focus exclusively on those uninsured prior to reform, thus they do not include any adjustment for the small share of individuals expected to go from insured to uninsured post-reform. Methods The HIPSM-NY model that we have used to analyze the effects of the ACA in earlier reports was based on two years of the Current Population Survey (CPS).3 There were enough observations of New Yorkers in the two-year merged CPS data for statewide analysis, but not for analysis of detailed areas within the state. For that, we reproduced the results of our model on three years (2008-2010) of the American Community Survey (ACS), which yielded a far larger sample size. We augmented the ACS with several data elements that are important in simulating the effects of the ACA:  Modified Adjusted Gross Income (MAGI) was computed as specified in the ACA. This required the imputation of unemployment compensation, which was not directly available on the survey.  Immigration status was imputed following a methodology originally developed by Passel.4  Eligibility for Medicaid and CHIP was imputed using current state rules.5  The presence of an offer of employer-sponsored insurance was imputed using the methodology developed for HIPSM.6 1 © 2013, The Urban Institute Health Policy Center • www.healthpolicycenter.org We used the previously published results of HIPSM-NY to predict changes in coverage on our enhanced ACS data. The New York ACS records were reweighted to simultaneously achieve two objectives. First, in developing HIPSM-NY, we had access to state-specific data that allowed us to determine enrollment in various types of health insurance coverage more accurately than survey responses could. This was particularly true of Medicaid, CHIP and individual (nongroup) market enrollment. Second, effects of the ACA, such as Medicaid and exchange enrollment, were calibrated to be consistent with earlier HIPSM-NY results. This multi-constraint reweighting was based on entropy maximization.7 We were then able to tabulate the results by sub-state area. The smallest geographic areas available on the public-use ACS are Public Use Microdata Areas (PUMAs) created by Census. These do not necessarily follow county boundaries, so we were not able to separate results for all counties. We present results for those counties that we could separate and divide the rest of the state into aggregations of PUMAs (Super-PUMAs). In cases where we provide separate tables with county specific estimates, observations for that county are