Is top code for 'EARNWEEK' the same for all years?

Hi, re: the documentation for EARNWEEK: “2003+: Usual weekly earnings greater than or equal to $2884.61 are assigned a value of $2884.61.”

Is this the same top code for all years, regardless of inflation?

Thanks much, SG

The top code of EARNWEEK is 2884.61 in all ATUS samples. I do see that while the original ATUS documentation from the BLS states that the maximum value (top code) of EARNWEEK is 2884.61 in the 2024 sample, in the actual data, there are larger values present.

Thanks Isabel. This could be a serious issue for those of us using this variable: In all years other than 2024, the earnings are top coded to 2884.61. In 2024, there are more than 150 observations with values greater than that:

-------------±-------------------------------------------------------—————————————-
earnweek | 167 mean 6095.979 sd 3360.391 min 2920 max 11952.53

That’s 4% of the observations in 2024. If that proportion applies to other years, more than 10,000 observations have had their earnings top coded. More likely the proportion of top codes has changed over the years due to inflation.

Can we get the raw data before the top-coding?

Thanks much, SG

Our team found BLS documentation stating that the top code for EARNWEEK has been changed from 2024-onward to correspond to the top 3 percent of earners in a given month. We will update our documentation to reflect this change.

Top codes are present in the original data provided by the BLS. IPUMS does not have access to data without these top codes, nor do we impose any top codes. I would suggest reaching out to the BLS directly to inquire about the possibility of accessing data without income top codes. This IPUMS User Forum post from 2024 discusses your concern about a static income top code and you may find some of the resources linked in that thread useful.

Hi again, just noted that same issue for 'SPEARNWEEK’–not top coded for 2024.

Thank you for letting us know. I have asked our IPUMS ATUS team to update the documentation for SPEARNWEEK to reflect the new top code in the 2024 sample. The documentation on EARNWEEK has been updated.

It may be good to modify the comparability note also for both EARNWEEK and SPEARNWEEK–SG

1 Like