Constructing HH income before 1968 and family income before 1964

CPS
1

Dear folks–

For years prior to 1968, there is no HHINCOME variable. Is there any reason (besides top-coding) that I can not just sum the INCTOT values by SERIAL number and get a variable equivilent to HHINCOME for those years? And similarly,can I sum incomes for FAMUNIT = 1 to get the Census family incomes, FTOTVAL, for 1962 and 1963?

And finally, can I sum INCTOT values for famunit FAMUNIT value other than one to get incomes for the non-householder families, the ones that Census publications generally ignore?

Andrew

2

We see no reason why your methods shouldn’t work as an approximation of the various total income measures, however, as you noted, in some cases the data have been top-coded which will effect estimates. Also, it is important to notice that in 1962 and 1963, INCTOT does not have data for persons in ROTATE4 and 8, which will effect estimates. Missing values were coded as 99999998 for INCTOT, so be sure to account for this in your recoding process.

I hope this helps.