# How does the top-coding of individual income variables affect aggregate variables to which they sum?
**URL:** https://forum.ipums.org/t/how-does-the-top-coding-of-individual-income-variables-affect-aggregate-variables-to-which-they-sum/2198
**Category:** CPS
**Created:** [March 7, 2018, 2:22am UTC](https://forum.ipums.org/t/how-does-the-top-coding-of-individual-income-variables-affect-aggregate-variables-to-which-they-sum/2198 "2018-03-07T02:22:00Z")
**Posts on this page:** 2
**Page:** 1
### Author:  [@AHoerner](https://forum.ipums.org/u/AHoerner)
#### Post date: [March 7, 2018, 2:22am UTC](https://forum.ipums.org/t/how-does-the-top-coding-of-individual-income-variables-affect-aggregate-variables-to-which-they-sum/2198/1 "2018-03-07T02:22:00Z")
In the IPUMS-CPS, for income variables that are the sum of other income variables, such as INCTOT, HHINCOME, and FTOTVAL, I notice that you do not provide top-coding thresholds or replacement values. Do you have a recommended procedure for recognizing whether one of these values has been affected by top-coding?
How do the summing variables relate to the top-coded components? In the years before 1996, is the summary value created by summing the components, treating top-coded values as if they earned the top-codeing threshold? In the years from 1996 to 2010, are they summed as if the top-coded value were replaced by the mean?
---
### Author:  [@JeffBloem](https://forum.ipums.org/u/JeffBloem)
#### Post date: [March 7, 2018, 6:14pm UTC](https://forum.ipums.org/t/how-does-the-top-coding-of-individual-income-variables-affect-aggregate-variables-to-which-they-sum/2198/2 "2018-03-07T18:14:00Z")
The income variables that are defined by summing up other, more specific, income values (such as [INCTOT](https://cps.ipums.org/cps-action/variables/INCTOT#comparability_section), [HHINCOME](https://cps.ipums.org/cps-action/variables/HHINCOME#codes_section), and [FTOTVAL](https://cps.ipums.org/cps-action/variables/FTOTVAL#codes_section)) are all defined conditional on the top coded values. Therefore, in years prior to 1996, these variables take the top coded values as if it is the actual amount earned by the individual. In years later, this same method is used with replacement values.