I understand that Larrimore et al. (2008) produced a solution to the top-coded income variables in the ASEC files for 1975 through 2002 (cell means). Moreover, Census provides a method of value swapping that applies to 1975 onward. I would like to backcast one or both of these through 1967, but no documentation on this site indicates how I might do this. In a more recent paper, Armour, Burkhouser, and Larrimore (2016) produce estimates that extend back to 1967.
All available IPUMS CPS documentation on these top-coding solutions can be found on this page. Here you’ll find links to the cell-means and swap-value data files for 1976-2000 and 1976-2010, respectively. There is also some documentation on how these adjustments are made. More detailed documentation can be found in the paper’s by Larrimore et al. (2008) and Armour et al. (2016). These resources should provide helpful detail for extending this series for 1967 and onward.
Hi Jeff, thanks very much for your response! I apologize that my confusion is persisting, but it is my understanding that the methodology outlined in those papers requires access to the internal data to replicate. Given that I only have access to the public data, how can I backcast using their methodology through 1967?
Yes, you are correct. The methods discussed by Larrimore et al. use internal restricted access data in order to construct cell-means above the income top-codes for each income component. I am not familiar with a method for performing such a task with only access to the public use files. In the Larrimore et al. (2008) article there is a discussion of how researchers address this incomparably over time. None of the solutions are perfect (hence the need for Larrimore et al.'s method), but they can be helpful. You can apply for access to the restricted data via a Census Bureau Research Data Center, if performing such a task is critical to your research.