Table 12

Robustness of 2SLS Estimates of the Effect of Medicaid Eligibility

Dependent Variable
Change to AnalysisMedicaidIHS(Number of Vacancies)Number of Skills
Main specification0.234***
(0.022)		0.894***
(0.186)		−2.354***
(0.574)
No demographic controls0.238***
(0.027)		0.957***
(0.233)		−2.478***
(0.656)
Unweighted0.243***
(0.009)		0.507***
(0.181)
Only years 2010 and 20160.282***
(0.026)		1.264***
(0.276)		−3.218***
(0.700)
Controls for capacity0.234***
(0.021)		0.888***
(0.184)		−2.339***
(0.561)
State-level0.210***
(0.032)		0.619***
(0.164)		−2.432***
(0.801)
Alternative instrument0.227***
(0.042)		0.918***
(0.270)		−2.508***
(0.777)

  • Notes: The data come from job vacancies posted online, 2010–2016. The unit of observation for Columns 1 and 2 is a year–area combination, and the baseline sample contains a total of 5,082 observations from 726 CZs. The unit of observation for Column 3 is a vacancy posting, and the baseline sample contains a total of 14,278,072 observations. Each cell is the effect of Medicaid eligibility from separate regressions. All regressions in Columns 1 and 2 include year fixed effects, area fixed effects, and the annual area-level covariates described in the text. All regressions in Column 3 include occupation-by-area fixed effects, occupation-by-year fixed effects, and the annual area-level covariates described in the text. Unless otherwise indicated, regressions are weighted by population. Standard errors are clustered at the area level and are shown in parentheses below the estimates. The geographical level of the analysis is CZ in the first five rows and state in the bottom two rows.