Problems with incwage in 2011

Dear staff member,

I am having big problems with the CPS-ASEC data in 2011. The topcoding procedure changed and this has crucial implications for the research I am doing. I tried to use the swap values provided by the Census Bureau but the biggest problems persist. I tried a procedure where I replace all top coded observations by the top code values that were used in 2007 (this is the first year of my analysis) and when I do this all the problems dissappear so it’s definitely an issue with top coding.

The two most serious problems I am encountreing are:

  • There is a huge drop in the value of incwage only for the top of the distribution (coming from inclongj) in 2011. Even though it seems strange, this could in principle be a “real” drop. However, as I said, when I change the top coded values to be all the same, equal to the 2007 value, this drop does not occur anymore.

  • There is a huge increase in the value of taxes in 2011, again, only for the top of the distribution. Same comments as for incwage apply here.

Can you please explain to me what is happening and how can I fix this? It has very important implications for my results so I need to understand whhere this is coming from and find a “proper” procedure to use.

Thank you very much,


I am unable to replicate your issue with INCWAGE in 2011. It appears that INCWAGE increases for the top of the distribution beginning in 2011, as expected with the new rank proximity swapping procedure.

The increase in the value of taxes in 2011 is due to a change in the topcoding of taxes. Specifically, the topcode was increased from 99997 to 999997.

Hope this helps.