I have encountered some examples that are seemingly inconsistent with the stated topcodes for the ORG earnings variables.
For example, in August 1990, the individual with CPSIDP of 19900500409301 reports the following:
HOURWAGE: 69.99
PAIDHOUR: 2
EARNWEEK: 5
UHRSWORKORG: 28
Based on the hourly earnings topcode for 1990, HOURWAGE seems like it should be topcoded at 68.68 based on working 28 hours per week, not 69.99.
I also do not understand the EARNWEEK value. According to the EARNWEEK description, since this individual is paid hourly (PAIDHOUR=2), it seems like EARNWEEK should equal HOURWAGE * UHRSWORKORG = 69.99 * 28 = 1,959.72 (or 68.68 * 28 = 1,923.04 when topcoded), not 5.
Is there something I am missing, or some other variable(s) I should look at to make sense of this?
This case, and a handful of other cases in that and nearby samples, is an anomaly in the original CPS data. We do not have a good explanation for this. This type of thing affecting small numbers of cases is very common in the CPS microdata, especially in the earlier years. The IPUMS CPS team generally does not edit the originally reported values in situations like this, where a limited number of cases are affected by a (seemingly) non-systematic error. That being said, the number of cases affected is few, and the actual topcode is generally close to the one that you’d expect. I also didn’t find any issue like this after 1991.