In the CPS March Supplement (1976-2018), why does INCWAGE have non-zero values for observations that have WKSWORK1 = 0?
I’ve looked into this issue a bit. These observations only occur between 1976 and 1989. Most cases involve individuals in the armed forces. WKSWORK1 universe only includes civilians so it appears those in the military were assigned a value of 0 for this variable. For the remaining case, most were in the military in the previous year. There are a relatively small number of cases, mostly in the 1970s, where there is no indication of affiliation with the armed forces and yet the pattern you saw is still present. In all such cases, the amount of wage income reported was less than $1000. I suspect this is largely due to people reporting wages paid in January for work done the previous year. There could also be some errors in the original data, especially since this was back in the days of paper surveys. You may want to search through the original codebooks for any more clues to explain these cases.