# Dealing with income values of 99999 in CPS after replacing topcodes using the 1976-2010 file

**URL:** <https://forum.ipums.org/t/dealing-with-income-values-of-99999-in-cps-after-replacing-topcodes-using-the-1976-2010-file/1373>\
**Category:** CPS\
**Created:** [July 27, 2016, 4:04pm UTC](https://forum.ipums.org/t/dealing-with-income-values-of-99999-in-cps-after-replacing-topcodes-using-the-1976-2010-file/1373 "2016-07-27T16:04:00Z")\
**Posts on this page:** 2\
**Page:** 1

<div class="post-metadata">

**Author:** ![Maddie](https://avatars.discourse-cdn.com/v4/letter/m/b3f665/32.png) [@Maddie](https://forum.ipums.org/u/Maddie)\
**Post date:** [July 27, 2016, 4:04pm UTC](https://forum.ipums.org/t/dealing-with-income-values-of-99999-in-cps-after-replacing-topcodes-using-the-1976-2010-file/1373/1 "2016-07-27T16:04:00Z")

</div>

Hello,

I am using CPS personal income datafiles from 1990 to 2015, and I have set those “Not in Universe” values (99999s) as missing to restrict my sample (excluding those age \<15). Then I have replaced topcoded income values using the 1976-2010 file provided on [this page](https://cps.ipums.org/cps/income_cell_means.shtml), just to make sure that I have consistent topcodes in my datafile.

However, there still appears income values of 99999s after this (e.g., INCCHILD,INCALIM,INCOTHER, OINCBUS, OINCFARM), and these observations fall within the universe (age\>15). I suspect it is the topcoded value, but there still exists values of 99997 for these variables.

How should I deal with these values 99999s when there exists topcodes of 99997 and they fall within the universe?

Many thanks in advance!

---

<div class="post-metadata">

**Author:** ![grover](https://sea1.discourse-cdn.com/flex019/user_avatar/forum.ipums.org/grover/32/5_2.png) [@grover](https://forum.ipums.org/u/grover)\
**Post date:** [August 1, 2016, 4:51pm UTC](https://forum.ipums.org/t/dealing-with-income-values-of-99999-in-cps-after-replacing-topcodes-using-the-1976-2010-file/1373/2 "2016-08-01T16:51:00Z")

</div>

These 99999 values do appear on the source data files [provided by the Census Bureau](http://www.census.gov/housing/extract_files/toc/data/) (I was not able to find any 99997 values in our provided swap-values datasets, however). Presumably these values represent a topcode (likely representing values that exceed those allowed by the original digit-length of the variable) as you have guessed. However, the values are not directly addressed in the documentation. I would recommend contacting the Census Bureau directly for further explanation.