Creating state level household benchmarks from ACS to match FDIC unbanked/underbanked supplement

I am trying to use the ACS to create household level benchmarks for the state of California. I want household distribution by age/race-eth/household structure where race-eth is non-Hispanic White, non-Hispanic Black, non-Hispanic Asian, Hispanic, and non-Hispanic other and household structure is married, unmarried female, unmarried male. These benchmarks will be used for comparison of the FDIC unbanked/underbanked supplement.
I assume I need to restrict to household heads as that’s the unit in the FDIC supplement. My confusion is on how to get the household structure from the ACS to line up with the FDIC unbanked/underbanked supplement. I appreciate anyone’s insight or guidance!

My understanding of your question is that you would like to define household types using ACS data that use the same definitions as the CPS Un(der)banked Supplement. IPUMS CPS and IPUMS USA both include variables that define family types, but they differ slightly. I will describe both briefly below and then address your question about restricting to the householder/head of household.

IPUMS CPS provides the variable GQTYPE, which categorizes households in terms of the family type. You can see the original categories in the CPS in the unharmonized variables UH_HHTYPE_A1 and UH_HHTYPE_B2. Note that GQTYPE is not specific to the Un(der)banked Supplement, but part of the Basic Monthly Survey (BMS), the core portion of the survey asked of supplement respondents along with any supplements that are fielded in that month. This means that respondents’ BMS responses are linked to their supplement responses, so the value of GQTYPE reflects the household type at the time of the Un(der)banked Supplement.

In IPUMS USA, which provides data from the ACS, we offer HHTYPE, which, like GQTYPE in IPUMS CPS, classifies households in terms of the family type. The categories are not identical to those in GQTYPE, but you could construct very similar categories between the two surveys.

Depending on how important it is for these definitions to match, you could look at household composition (i.e., using RELATE) and define your own family types.

In both the ACS and the CPS, the household type variables are household-level variables, and are therefore applied to every member of the household. All members of a household will have the same value of HHTYPE (in IPUMS USA) or GQTYPE (in IPUMS CPS). The Un(der)banked Supplement variables are also household-level, and will be the same for each member of the household. For household-level analyses, you should restrict your records to one unit per household (e.g., PERNUM == 1, or possibly use RELATE to retain only the householder) and apply the household weight.