Why is the respondent not in the labor force

Hey IPUMS forum,

While spempstat details whether the spouse of the respondent was retired, disabled or unable to work at the time of ATUS survey, empstat does not offer this information regarding the respondent. Is there a way to understand whether the respondent is disabled, retired etc., and therefore not in the labor force?

As far as I understand, the retired variable’s universe only includes respondents that have previously reported being retired (when? in the CPS survey?). Therefore, crosstabing it with empstat shows that a lot of the cases that are not in the labor force in empstat are niu of retired (and also quite a few un/employed cases that are “retired” according to the retired variable).

I think looking isn’t much help either because although it is supposed to include all 15+ non-employed cases in its universe, a large portion of the non-employed sample is defined “niu” here too for some reason (any idea why?). Additionally, looking and retired do not correspond at all.

Any information on the matter will be much appreciated!

The variable SPEMPSTAT is created using responses to multiple questions that ATUS respondents are asked about their spouses during the ATUS interview. In the 2023 ATUS Data Dictionary, you can see on page 49 that respondents are asked whether their spouse or unmarried partner had a job in the last seven days. The response options are: Yes, No, Retired, Disabled, and Unable to Work. Then, respondents are asked whether their spouse or unmarried partner did any work for pay or profit in the last seven days. The response options are: Yes, No, Retired, Disabled, and Unable to Work.

The responses to these two questions are used to determine whether an ATUS respondent’s spouse is coded in SPEMPSTAT as:

  • Employed - at work (had a job in the last seven days and did any work for pay or profit in the last seven days),
  • Employed - not at work (had a job in the last seven days and did not do any work for pay or profit in the last seven days),
  • Not employed (did not have a job the last seven days),
  • Retired,
  • Disabled, or
  • Unable to work.

The variable EMPSTAT is created using responses to multiple questions that ATUS respondents are asked about their own work. In the 2023 ATUS Data Dictionary, there are several questions that solicit the information contained in EMPSTAT. On page 36, the variable TUABSOT reports whether the respondent had a job in the last seven days. The response options are: Yes, No, Retired, Disabled, and Unable to Work. On page 42, the variable TUFWK reports whether the respondent did any work for pay or profit in the last seven days. On page 43, the variable TULAY reports whether a respondent was on layoff from their job in the last seven days. On page 44, TULK reports whether a respondent has done anything to try to find work in the last four weeks. Respondents who are retired, disabled or unable to work are coded as Not in labor force (EMPSTAT=05).

The variable RETIRED reports whether the respondent was retired at the time of the ATUS interview. The universe of the variable is ATUS respondents who were retired in the eighth CPS interview, as reported in EMPSTAT_CPS8 (code 5). All respondents who do not have a value of 5 for EMPSTAT_CPS8 are not-in-universe for RETIRED. I will note a small number of individuals who were retired in the eighth CPS interview according to EMPSTAT_CPS8 are coded as not-in-universe for RETIRED; small inconsistencies between the documented universe of a variable and the individuals who have valid responses for the variable are not uncommon in survey data such as the ATUS.

The variable LOOKING reports whether the respondent has been seeking work during the past four weeks, and if not, the reason why. The universe of the variable is non-employed ATUS respondents age 15 and over. The universe does not include people on layoff (those with EMPSTAT=3). Our documentation does not seem to reflect this, so I will ask our IPUMS ATUS team to update it.

Note that the universes of these variables (EMPSTAT, LOOKING, RETIRED) are restricted to ATUS respondents. In the ATUS, each sampled household has just one ATUS respondent selected, who completes a time diary and answers a large number of survey questions. Some basic information is collected on non-respondent household members in sampled households, and information on these non-respondent household members collected during the eighth CPS interview is also included in the ATUS. If your data extract includes both respondents and non-respondents, this may be a source of confusion regarding why some individuals are NIU. You can use the variable RELATE to check which individuals in your dataset are ATUS respondents, and which are not.

1 Like

Hey Isabel,

Thank you for all the information regarding these three variables!

I have a few follow up questions if that is okay, because I am still missing something here:
Can we as users access TUABSOT, TUFWK, TULAY, and TULK through IPUMS somehow?
If I understand correctly EMPSTAT=5 includes all ATUS respondents that are not in the labor force - i.e., replying retired, disabled, unable to work in TUABSOT, but also those answering they don’t have a job (TUABSOT=“No”) provided they are not looking for work and are not on layoff.
If this is the case and IPUMS users don’t have access to the underlining questions, the bottom line is that we can’t tell the reason behind an ATUS respondent not being in the labor force for two reasons: First, all the retired, disabled and unable to work respondents are lamped together in empstat=5 (unlike the case of SPEMPSTAT). Second, respondents that are empstat=5 can also be non-retired and able workers that are not in the LF for some other reason (e.g., stay at home dads or mums that are not looking for work).
Since RETIRED is only asked if the respondent was retired in the CPS interview it is not of great help - respondents that became retired during the period between the two surveys will not be included in it (by the way, in my extract including all years, there are 3,278 cases that are niu of RETIRED although they are EMPSTAT_CPS8=5, out of 48,604 such cases. This is not an insignificant amount I think. Other variables I have worked with have a much more “clear cut” universe (e.g., CLWKR)).
As for LOOKING, as far as I can tell, using only relate=10 respondents, I see that 55,012 of respondents that are not in the labor force (EMPSTAT=5) are defined NIU in LOOKING (in addition to the respondents that are EMPSTAT=3). This is a huge proportion of the respondents that are not in the labor force. What am I missing here?

Thank you again so much for your help and time! In general, this forum is such a great tool to help deal with complexities of the data!

IPUMS does not provide original (source) variables from the ATUS. However, you can access the original data and variables from the Bureau of Labor Statistics directly.

To determine the reason that all out-of-labor-force respondents are not in the labor force, you would need to use EMPSTAT in combination with another variable. Using EMPSTAT, LOOKING, and RETIRED together is inadequate to determine the reason for labor force inactivity for the full universe of respondents, as you noted. This is because the universes of the variables LOOKING and RETIRED are more restrictive than stated in our documentation, and do not include all respondents who are out of the labor force. I have included more details on the universes of these variables, based on the original 2023 ATUS questionnaire, at the end of this post.

I suggest using the variable FWK in the original ATUS data from the BLS. The universe of this variable is all respondents, and asks, “In the LAST SEVEN DAYS, did you do ANY work for [pay/either pay or profit]?” Response options include:

  • Yes
  • No
  • Retired
  • Disabled
  • Unable to work

Combining information from this variable from the recoded employment status variable (EMPSTAT), you should be able to determine, for each respondent, whether they were in the labor force, and if not, why not. While the IPUMS variables RETIRED and EMPSTAT only record individuals as retired if they were also retired in the final CPS interview (due to the universes of the variables in the original data), FWK allows any respondent to state that they were retired. I will ask our IPUMS ATUS team to provide FWK on our platform in the future.


Additional information about the universes of RETIRED and LOOKING

Individuals are asked whether they did any work for pay or profit in the last seven days before they have the opportunity to be asked about retirement. Individuals who report they did work for pay or profit in the last seven days are not asked whether they are still retired, even if they reported being retired in the last CPS interview. Additionally, the age restriction on being asked the retirement question means that some individuals who were retired in the eighth CPS interview were not asked if they are still retired, due to their age. The original ATUS variable RET asks, “The last time we spoke to someone in this household you were reported to be retired. Are you still retired?” This question is asked of the following people:

  • Retired (final CPS interview) AND when asked if they did any work for pay or profit in the last seven days, they said no, retired, or disabled (ATUS interview) AND age 65+ (ATUS interview) AND does not have a business or farm (ATUS interview)
  • Retired (final CPS interview) AND when asked if they did any work for pay or profit in the last seven days, they said no, retired, or disabled (ATUS interview) AND age 65+ (ATUS interview) AND there is more than one household member (ATUS interview)
  • Retired (final CPS interview) AND when asked if they did any work for pay or profit in the last seven days, they said no, retired, or disabled (ATUS interview) AND they have a business or farm but did not work on it in the last seven days (or did not know whether they worked on it, or refused to say) (ATUS interview) AND age 65+ (ATUS interview)

Overall, the universe statement for the IPUMS variable RETIRED in IPUMS ATUS seems to lack some of the detail implied by the ATUS questionnaire. I will discuss this with our IPUMS ATUS team, who will determine whether any updates are necessary.

The universe of IPUMS variable LOOKING also seems to be more complex than indicated by our documentation. In the ATUS questionnaire linked above, the original ATUS variable LK (on page 41) asks, “Have you been doing anything to find work during the last 4 weeks?” The universe of the variable is “LAY6M = 2, Don’t Know, Refused or DP didn’t work and didn’t have a disability that prevented DP from working or DP indicated retired but wanted a job full or part time.” The original ATUS variable LAY6M asks, “Have you been given any indication that you will be recalled to work within the next 6 months?” and a code of 2 is a no response.