Non-LUNCHSUB recipients with SCHLLUNCH values


I have been conducting some research with colleagues that has us looking at the rate of recipiency and average dollar value of different transfers across household types. For this research, we are using the CPS ASEC files across years ranging from 1994-2017. I noticed in doing some analysis that across years, the average value of the SCHLLUNCH variable for recipient households without children is greater than ‘0’. This was puzzling, as the SCHLLUNCH variable measures the family market value of the free and reduced lunch program; without children, how could households receive this source of income?

Digging a bit deeper, I looked at just the 2000 ASEC and found that a sizable number of households with respondents that replied “No, children did not receive free or reduced price lunch,” and therefore received a value of ‘2’ for the LUNCHSUB variable, also had an income of greater than $0 for the variable SCHLLUNCH.

In 2000, 25,869 out of 133,710 respondents reported some value greater than $0 for SCHLLUNCH despite also stating that “No, children did not receive free or reduced price lunch.” (See attached Stata output photo.) This is 19.3% of all respondents in the ASEC!


Does the IPUMS Staff have any insight on this? Thanks in advance!

As a followup, I should mention that I also examined SCHLLUNCH recipiency for unique households in the 2000 ASEC in which there were no respondents under the age of 18. Theoretically these households should receive no SCHLLUNCH dollars because they have no children. I found a number of households (212) that both had SCHLLUNCH dollars and no children (See Stata output below; the variable ‘first_fam_val_sl’ is SCHLLUNCH dollars aggregated to the household head; hh_kids is a binary recording whether or not there were under eighteen year-olds in the household).


Significantly lower than the number of households marked ‘0’ for LUNCHSUB yet receiving SCHLLUNCH dollars.

Thanks for pointing this out. The IPUMS CPS team is looking into this issue. I’ll post here again once we have a better idea of what’s going on with this variable.

Thanks for the question–Matt is out of the office, but I will pass along a summary of his notes as well as information from the IPUMS CPS team.

First, if you include 18-year-olds in the definition of children (because there may be 18 year-olds who are eating lunch at school), you will see that there are no households without children have a positive SCHLLUNCH value . We will update the documentation to more clearly convey that in SCHLLUNCH.

Second, SCHLLUNCH reports the value of any school lunches eaten, regardless of whether or not they were free or reduced lunches. This is why you see values of LUNCHSUB == 2 alongside positive values of SCHLLUNCH. From the ASEC codebooks:

All students eating lunches prepared at participating schools pay less than the total cost of the lunches. Some students pay the “full established” price for lunch (which itself is subsidized) while others pay a “reduced” price for lunch, and still others receive a “free” lunch.

More information is in the “school lunches” entry in the glossary section of the ASEC codebooks.

I hope this helps. Please let me know if you have further questions.

1 Like

Thanks for your response @KariWilliams. This is all illuminating, especially given the lack of clear documentation; though, I see now what you’re pointing to in the ASEC codebooks.

I have a follow up question in regard to how a researcher might isolate the value explicitly of the free and reduced school lunch program for recipient households in the CPS data. Perhaps I’m missing it, but while I can see there’s a variable that establishes whether a household has a child recipient of a free or reduced lunch (FRELUNCH), there isn’t a corresponding variable that holds the value of that free or reduced lunch specifically, rather than providing the value of all hot lunches consumed by children in the household in general (which, if I understand you correctly, is what SCHLLUNCH represents). It’s my understanding, though, that if one wanted to determine the average value of the free and reduced lunch program for recipient households in a given year of the ASEC data, one would first isolate the recipient households using FRELUNCH, and then among those recipient households take the mean of the SCHLLUNCH variable. Does this sound correct? If so, I think the only flaw with this logic is that one couldn’t disaggregate the value of a school lunch (not free or reduced) from the added value of a free or reduced lunch. Presumably every recipient of a free or reduced lunch gets the value of a school lunch that all those who eat school lunches receive, plus the added value of a reduced or free lunch on top of that. If that’s also true, I’m struggling to see how one can truly measure the value of a free or reduced lunch on the level of the household using this data.

Any thoughts you have on this would be very helpful!

The variable ATELUNCH reports how many children ate a complete school lunch. You could combine this information with the number of children captured by FRELUNCH and the value reported by SCHLLUNCH and/or SPMLUNCH to get at your question. I would direct you to the literature on school lunches that leverage ASEC data; there should be many papers and reports directly from Census Bureau staff that will likely provide more targeted commentary on how to operationalize the benefit of free and reduced lunches.