How to deal with topcoded income values of 99997 in CPS 2011 onward?

In CPS for 2011 onward, the topcode tables (here) indicate that “income values greater than or equal to swap value were systematically swapped with other topcoded cases.” However, for many income variables (INCASIST, INCCHILD, etc), there are both topcoded values in the data (that is, there are values equal to “99997” or “999997,” respectively) as well as values greater than the topcode/swap value.

For instance, for INCASIST in 2011, the topcode/swap value is $30,000. But in the data - I’m using ASEC - there are many observations with INCASIST values greater than 30,000 (which indicates that values were “swapped,” as indicated on the topcode table page) AND there are observations with the top-code “99997.” When were the codes “99997” applied, rather than swap values? What am I supposed to do with observations that have the “99997” code? This problem seems to occur across a whole range of income variables, and for all years 2011 and onward.

1 Like

In the ASEC survey, respondents can provide a value for financial assistance income (see FIN-VAL question) between $0 and $99,999. The 2011 topcode swap value for INCASIST is $30,000, but the IPUMS data distinguishes between ASEC responses of $30,000-$99,998 and responses of $99,999. The latter response is essentially already topcoded by the CPS survey question. As a result, values of $99,999 are coded as $99,997 (topcode) in the IPUMS microdata and values between $30,000 and the highest value below $99,999 (e.g. $84,000 in 2011) are systematically swapped with each other. The same methodology is used for all of the topcoded income values beginning with the 2011 ASEC sample.

Hope this helps.