The health insurance unit (HIU) variables provided via IPUMS CPS are constructed by our partners at the State Health Access Data Assistance Center (SHADAC). These variables are constructed using the following IPUMS variables: SERIAL, PERNUM, RELATE, AGE, SEX, MARST, MOMLOC, POPLOC, SPLOC, and NCHILD.
The variables SERIAL, PERNUM, RELATE, AGE, SEX, and MARST all have equivalents available in the original Census Bureau data that you are using in the RDC. MOMLOC, POPLOC, SPLOC, and NCHILD are constructed by IPUMS. SHADAC provides Stata code for constructing an HIU ID variable using IPUMS ACS data. You could modify this code to use with the restricted CPS data. SHADAC provides technical information on HIU variable definitions and construction. Here are a few resources you may find helpful:
You could also reach out to SHADAC directly if you have questions that are not answered by their documentation, website, or code.
It is possible to approximately replicate the IPUMS-constructed family interrelationship pointers MOMLOC, POPLOC, and SPLOC. NCHILD simply counts up the number of children who are linked to each individual via MOMLOC or POPLOC, so this will be straightforward to construct once you have created parent pointer variables. The rules used to assign parent-child links in MOMLOC and POPLOC are described in MOMRULE and POPRULE, respectively. The rules used to assign spouse links in SPLOC are described in SPRULE. This paper by Gorsuch and Williams describes in detail how these family interrelationship pointer variables were constructed.
However, you could instead use the respondent-reported family interrelationship pointer variables, since these are available in the original CPS data. In IPUMS CPS, these are PELNMOM, PELNDAD, PELNPAR1, and PELNPAR2. They will have different variable names if you are not using IPUMS.
CPS data include a variable reporting adjusted gross income (AGI). See IPUMS variable ADJGINC. ADJGINC, like other tax-related variables included in the ASEC CPS, comes from the Census Bureau’s tax model, which simulates individual tax returns to produce estimates of federal, state, and payroll taxes. You could modify this variable to estimate MAGI, likely using some of the other imputed tax variables from the CPS. If these variables are not available to you, you will need to use information from the Census Bureau on their tax model to try to create them yourself. Here is some documentation to help you get started:
