Why some HHINCOME does not equal the sum of INCTOT? (2010 data)


#1

Hi,

According to IPUMS documentation (https://usa.ipums.org/usa-action/variables/HHINCOME#description_section), HHINCOME should equal the sum of all household member’s individual income, coded as INCTOT. So I was checking the 2010 census data against this fact.

However, in 2010, it seems like the HHINCOME does not include the income of unrelated household members (RELATE>=10). For instance, observations where SERIAL = 8, the household member is unrelated to the hh head (RELATE=12) and earns income (INCTOT=1200), but hhincome is N/A. Another case is, for instance, SERIAL=898615. The HHINCOME=1, but it does not reflect the unrelated household member’s INCTOT.

Codes I used:

* replace n/a income *;

replace hhincome=. if hhincome==9999999

replace inctot=. if inctot==9999999

replace ftotinc=. if ftotinc==9999999

* manually generate tot to see if tot=hhincome *

egen tot=sum(inctot),by(serial)

list serial numprec hhtype pernum relate sex age hhincome tot inctot ftotinc slwt in 1/50

* compare the two variables: hhincome tot *

compare hhincome tot

list serial numprec hhtype pernum relate sex age hhincome tot inctot ftotinc slwt if hhincome != tot & hhincome !=.

list serial numprec hhtype pernum relate sex age hhincome tot inctot ftotinc slwt if hhincome != tot in 1/500


#2

I think the instances where HHINCOME does not equal the sum of INCTOT can be explained either by HHTYPE or by the special codes included in INCTOT. HHTYPE classifies households as either family or nonfamily households. For instance, for SERIAL==8, the HHTYPE==N/A. Additionally, for SERIAL==898615, INCTOT==1 for each of the members of the household. This code does not necessarily indicate a personal income of $1, but rather signifies “break even” income. So the sum of three individuals “breaking even” is still “break even”. I hope this helps, sorry for any confusion.