Consistent measure hours worked and weeks worked (no intervals)

USA
1

I want to use variables related to hours worked in a week and amount of weeks worked in a year non-intervalled.
It seems to me that:

  • number of weeks worked in a year is available only in intervals following 2007, is this true? Many papers refer to using weeks worked in a year after 2007 from ACS without specifying is available only in intervals…
  • If I want to use hours worked without intervals it looks like I need to rely on HRSWORK1 (hours worked last year) until 1990 (and with the exception of 1960-70). However, this variable is not avaialble later on, and it is substituted by a non-intervalled variable of usual hours worked per week (UHRSWORK) available from 1980 onwards. Also in this case I have found papers saying the use usual hours worked before 1980…

Many thanks for your help!!

2

The number of weeks worked in the past 12 months is available in the variable WKSWORK1. The intervalled form is WKSWORK2.

You’re right that there is not a completely comparable measure of weekly hours throughout the entire span of the IPUMS USA samples. Although UHRSWORK and HRSWORK1 are not entirely comparable, they are not likely to give significantly different answers in aggregate. I suggest studying the relationship between the variables in the 1980 and 1990 samples, when both are available, and just be aware of the definitional change between 1990 and 2000. You can also use the intervalled version of HRSWORK1, which is HRSWORK2, and is available for 1940-1990 censuses.

I’m not sure about pre-1980 papers reporting on usual hours worked. Can you link to an example? If you want to double-check the availability of a question in IPUMS USA, you can see the original enumeration forms for each census at this page.

3

Matthew, thanks a lot for your quick reply!

  1. regarding weeks worked I can certainly use WKSWORK1 if I dont want it in itervalled, but do you confirm that this non-intervalled measures in not available after 2007? And that there is no other variable after 2007 which reports weeks worked not in intervals?

  2. Thanks for your suggestion. Where can I find the exact definitional change between 1990 and 2000?
    With respect to the paper, I had in mind the following one: https://www.ddorn.net/papers/Autor-Dorn-LowSkillServices-Polarization.pdf
    As you can see in the footnote of table 1 they report that “Hourly wages are defined as
    yearly wage and salary income divided by the product of weeks worked times usual weekly hours”. Also in the footnote they say the table is based on the following samples, which cover prior decades to 1980: “Source: Census 1 percent samples for 1950 and 1970; Census 5 percent samples for 1980, 1990, 2000; American Community Survey 2005.”

4

  1. Weeks worked has only been collected in interval form by the Census Bureau since 2008. You can see this in the ACS questionnaire forms, for example item 39 (under Person 1) in 2018. It appears that the non-intervalled form made a return in 2019.

  2. Regarding definitional changes: I was just referring to the availability of “last week” hours worked only up to 1990, and usual hours starting in 1980. So you actually could make the break in 1970-80, 1980-90 or 1990-2000 depending on whether you want to prioritize usual or weekly hours. You can see the exact question text for each sample by looking at the Questionnaire Text page for UHRSWORK and HRSWORK1.

Regarding the paper: I suspect they use weekly hours prior to 1980 and just don’t detail this in the paper. I suggest checking with the authors on this.