Causal inference project for PUBLPOL 104: Economic Policy Analysis. In this project, we combine data from the CDC Wonder multiple-cause-of-death dataset and Census data to analyze the causal relationship between Medicaid expansion and opioid deaths. We use an interrupted-time-series / differences-in-differences approach, leveraging the fact that not all states chose to adopt the Medicaid expansion, along with the fact that different states expanded Medicaid access in different years. Below we show some key figures and results from the analysis.
Note that expansion for most states occurred in 2014.
We found significant evidence of an acceleration of opioid deaths over time after states expanded Medicaid.
We include all states in our dataset, centering expansion states around the year they expanded Medicaid, and generate quadratic curves that describe opioid death rates over time.
We test our results if instead of centering around the expansion year, we center treatment states around the year 1, 2, and 3 years before they actually expanded. We find lower R^2 values and generally decreasing significance of coefficients; however, the effects do not totally disappear. Possible reasons include anticipatory effects by insurers.
