Hi! I am using UHRSWORK1 for 1984-2010 and noticed that there are much more individuals with 999 (No hours or NIU) in 1984-1993 than 1994-2010. The number of individuals that report 999 drops significantly in 1994. Why is this happening? I want to assign 0 hours worked to individuals that are not in labor force and hence have no hours or are not in universe. But since this drop is happening in 1994, hours worked including zeros before 1994 is on average very close to 0 which does not seem correct.
As noted in the comparability section for UHRSWORK1, there are two things causing this sudden drop:
First: up to 1993 this question was part of the outgoing rotation group questions, so was only asked of 1/4 of the sample any given month. Those ineligible were given NIU codes.
Second: up to 1993, those with zero usual hours worked were given a 999 code. Starting in 1994 these have a value of 0.
Thanks for the answer! That makes sense! So then I can’t assume that individuals reporting 999 are not in the labor force before 1994? Essentially if I want to assign 0 hours worked to individuals not in the labor force I can only do that after 1994? Or is there a way around to do that before 1994?
Correct, the 999 code has a different interpretation before 1994. If you want to assign a 0 value to individuals out of the labor force prior to 1994, I’d recommend using the variable LABFORCE (or EMPSTAT for more detail). This gives the labor force status for all individuals over age 15. However, I would recommend dropping everyone outside the outgoing rotation groups in 1993 and earlier. You may also want to use EARNWT (instead of WTFINL) for the analysis in these earlier years.