My crosstabulations of people covered by medicaid and % FPL show upwards of 39% of those covered by medicaid are above 138% of the FPL. I calculated %FPL as a ratio of OFFTOTVAL and OFFCUTOFF, and subset my population to those who have a positive response in HIMCAID and negative response in HIMCARE. I am working in R using the survey package to incorporate the wtsupp weighting scheme in my tables.
There are a number of complexities around medicaid eligibility and how it is measured in the CPS. First of all, the HIMCAID variable is a part of the ASEC supplement, which uses the entire previous year as the reference period. HIMCAID simply identifies if the individual was ever covered by Medicaid in the previous year. Since the income variables are also annual figures, it is possible that the individual was under the income cut off at one or more points during the year, but aggregated over the whole year, the income values exceed the limit. Furthermore, while the variable is primarily based on self-report, as noted in the HIMCAID variable description, “The data in HIMCAID were edited by the Census Bureau and assign Medicaid coverage to some persons who did not report such health insurance during questioning.” The CPS technical documentation gives the full definition used for identifying Medicaid coverage on page 9-8 (pdf page 112) and also mentions the final complicating factor, that Medicaid eligibility can be different between states.
I hope this helps.