Repayments and Refunds: Estimating the Effects of 2014 Premium Tax Credit Reconciliation

We applied our Current Population Survey (CPS) modeling work to the Survey of Income and Program Participation (SIPP) 2008 Panel to estimate the experience of tax claimants over two full calendar years. We computed each individual’s health insurance coverage, Medicaid and advance premium tax credit (APTC) poverty level, and eligibility category under the ACA (Medicaid-eligible, subsidy-eligible, coverage gap, etc.) following the methods discussed in depth in the appendices of our state estimates of the coverage gap and subsidy-eligible individuals.

The income, employment, and health insurance sections of the CPS and SIPP questionnaires include many of the same questions.  Implementing our CPS algorithm in SIPP produces similar calendar year-weighted estimates of both insurance coverage and ACA eligibility. CPS produces reliable estimates at the state-level at a single point in time while SIPP follows a cohort of individuals and families on a monthly basis over a period of four years, making SIPP the preferred microdata for estimating the dynamics of income and ACA eligibility.

We assessed tax claimants’ ability to predict their final 2014 annual income in late 2013 by shifting survey responses forward by two calendar years.  The current SIPP 2008 Panel includes a four-year, person-weighted sample of about 45,000 individuals over the 48-month period of 2009 to 2012.  Respondents’ annualized income and health insurance coverage status at the end of 2011 served as the point of initial enrollment (displayed throughout the text as 2013) and annual income collected during survey year 2012 provided amounts for the final tax reconciliation (displayed as 2014).  Both values were inflated with the Bureau of Labor Statistics factor from 2012 to 2014 when compared to 2014 premiums.

To accommodate the added dimension of time in SIPP, we imputed documentation status only at the beginning of the panel but imputed an offer of employer-sponsored insurance (ESI) for each unique job over the period.  Otherwise, both of these techniques mirrored the strategy outlined in the immigration status and offer imputation appendices of our prior work.

This analysis estimates the reconciliation experience of tax filers who were either eligible for an APTC themselves or who claimed a dependent eligible for APTC based on filing unit 2013 Modified Adjusted Gross Income (MAGI).  Additionally, that subsidy-eligible individual must have been a part of the potential marketplace population in January of 2014.  To create a tax filing unit weight, person-weights for single filers and heads of household were maintained, married couples’ person-weights were each divided by two, and all tax dependents’ weights were zeroed out.  This resulted in a starting population of approximately 11 million tax households based on an unweighted sample of 1,918 records.  Approximately ten percent of claimants experienced a change in tax filing unit structure at some point during the reconciliation year (2014) due to birth, death, marriage, divorce, or income or residence changes of a dependent relative.  Since a change in family size (and with it, monthly marketplace premiums) might precipitate the APTC recipient to report any revised income, these units were excluded from the analysis.

Starting in January 2014, we determined each individual’s monthly premium based on actual reported monthly insurance coverage.  All individuals without insurance, or with nongroup, unknown private coverage, or dependent ESI who also did not have access to an imputed offer of ESI for the month were designated as a marketplace enrollee for that month in need of coverage.  Their premiums were summed alongside others in their tax filing unit, and then capped according to their tax filing unit’s subsidy-eligibility from the point of application (2013 annual income) for a single pro-rated month.  After computing all twelve months of potential marketplace subsidies, this process was repeated using the tax filing unit’s subsidy-eligibility from the point of reconciliation (2014 annual income).  After capping based on current-law repayment limits, the difference between these two APTC amounts provided our final estimates of required repayments, overpayments, and net adjustments shown.


KFF Headquarters: 185 Berry St., Suite 2000, San Francisco, CA 94107 | Phone 650-854-9400
Washington Offices and Barbara Jordan Conference Center: 1330 G Street, NW, Washington, DC 20005 | Phone 202-347-5270 | Email Alerts: | |

The independent source for health policy research, polling, and news, KFF is a nonprofit organization based in San Francisco, California.