Is MIGPUMA using 2000 or 2010 Census definitions within the 2020 ACS 5-Year Public Use Microdata Sample (PUMS)?


I am attempting to obtain some demographic information on those migrating to Forsyth County, NC. For this, I was planning to leverage the MIGSP and MIGPUMA variables available in the 2020 ACS 5-year Public Use Microdata Sample (PUMS). While reviewing the data, I noticed some PUMAs listed within the MIGPUMA variable that I am familiar with seemed to be based on 2000 Census definitions not 2010 as described in the 2019 PUMS 1-Year Data Dictionary. I made an assumption that all PUMAs listed within the MIGPUMA variable were based 2000 Census definitions, and joined it with the 2000-2010 PUMA Crosswalk spreadsheet. I found that some of the PUMAs listed within the MIGPUMA variable were not present in the 2000 column of the linked spreadsheet but were in the 2010 column, and vice versa. In short, I am a little confused what the MIGPUMA variable is referencing. 

A minimal example script detailing the issue can be found here:

Any guidance would be much appreciated!

  • hi Daniel--

    Generally, Census Bureau creates MIGPUMAs as the least common denominator grouping of counties (group can be as few as 1) that are entirely coincident with a group of PUMAs (as few as 1). 

    So... multiple counties (maybe rural?) grouped into a single PUMA means the PUMA and MIGPUMA are same.

    But multiple PUMAs nested within a county become one MIGPUMA.  (Example: Dakota County MN has 400,000 people and 3 PUMAs, but those are all collapsed into a single MIGPUMA.)

    Why is it this way?  It is because Census Bureau codes ACS migration responses with the state and county (and maybe principal city??) that moving persons came from. ACS does not know the exact tract where people moved from.

    That help?

    --Todd Graham

    Metropolitan Council Research,

    PS: There is a similar story w.r.t. Place Of Work PUMAs.

  • Hi Todd - Yep that makes sense, thank you for the explanation!

Reply Children
No Data