Top-code change in ASEC


I’ve noticed that the top-codes have changed for some variables in the ASEC files compared to the same variables in the data downloaded a few years ago, but without a change in the documentation.

For example, when I downloaded the data a few years ago, the variable incasist had a max value (excluding 99,999) of 99,997; this was the stated top-code in the documentation.

However, in the current version of the data, the top-code appears to have been changed to 9,999,997, which seems way less appropriate given the distribution of the non-top-coded observations. In addition, the mean of the non-top-coded data are identical in the older version and the new version, so the only thing that has changed are 4 observations whose top-code went from 99,997 to 9,999,997.

Are we meant to replace those top-coded values with 99,997 and treat that as a mistake?

Thanks very much for your help!

David Ratner

It looks like the top-code value was expanded since your previous data extract to emphasize that this is a special top-code value and not an actual dollar value. This is why the only detail different from your previous data extract is the number of digits in the top-code value. Top-codes are handled differently by the Bureau of Labor Statistics across samples in the CPS data, so how you should handle these values depends on the specific samples you are using. See this page for more information on income top-codes in the CPS data.