metarea acs and metare in cps

Good afternoon,

I would like ask the differnce between the code for metarea in the acs data and in the cps. I can see that defintion for acs has changed in the 2012 but is it the same for cps? is it possible to use or find the same code for both datasets?

You are correct that IPUMS-USA began using a new metro area definition (2013 OMB MSA standards) beginning with the 2012 sample and the introduction of PUMAs based on the 2010 Census. IPUMS-CPS, however, switched to the 2013 MSA codes beginning with its May 2014 sample.

Keep in mind that metro areas have not been directly identified by the Census Bureau in the IPUMS-USA microdata since the 1980 5% sample. Instead, IPUMS constructs metro areas using the smallest available geography, such as PUMA in the ACS data. This leads to some metro areas being incompletely identified in the microdata (see here for a list). In contrast, the Census Bureau does directly report the respondent’s metro area in the CPS microdata, avoiding the IPUMS-USA issue of incomplete identification. Due to the smaller sample size of the CPS relative to the Census/ACS, estimates for metro areas with populations under 500,000 calculated from the IPUMS-CPS data should be used with caution because of the relatively large sampling variability associated with these estimates.

For the reasons mentioned above, METAREA/MET2013 are not completely comparable between IPUMS-USA and IPUMS-CPS. Since the degree to which a metro area differs between data sources varies depending on the metro area of interest, I recommend consulting the list of incompletely identified METAREAs (provided above) and the MET2013 match error tables. Ultimately, it is left to the individual researcher to determine whether comparing metro areas between the IPUMS-USA and IPUMS-CPS data sources is appropriate.

Hope this helps.

While Tim’s answer certainly has helpful information about a) incompletely identified MSA’s in IPUMS-USA data and b) CPS estimates not being representative of small MSA’s, these seem like general comments that would apply to any question related MSA’s. A relatively small fraction of the answer seems to be devoted to the actual question, which was about how to merge MSA-level information between IPUMS-USA data and IPUMS-CPS data. The part of the answer focused on that question was helpful, but ommitted a lot of information.

Indeed, it seems there’s no metro code that is directly comparable between IPUMS-USA and IPUMS-CPS. For instance, in the pre-2014 CPS, metarea takes codes (with value labels) like:

  1. New York-Northern New Jersey-Long

  2. New York, NY

  3. Los Angeles-Long Beach-Santa Ana,

  4. Los Angeles-Long Beach, CA

  5. Chicago-Gary-Lake IL

  6. Chicago-Naperville-Joliet, IL-IN-

whereas the pre-2012 ACS/Census use metarea codes like:

  1. New York, NY-Northeastern NJ

  2. Los Angeles-Long Beach, CA

  3. Chicago, IL

So the obvious thing is that they can’t be directly compared because they’re different digits. But, it seems like the first 3 digits of the IPUMS-CPS-metarea should correspond to the IPUMS-USA-metarea. To my knowledge, this is not documented or explained anywhere on any IPUMS site (including the descriptions of the metarea variables), and I think something like this is the direct answer to the question asked. It would be nice to have a more explicit discussion from IPUMS staff, who is far more knowledgable than me.

In particular, it’s not clear what one can do after IPUMS-USA switched to met2013. The corresponding met2013 codes are:

  1. New York-Newark-Jersey City, NY-

  2. Los Angeles-Long Beach-Anaheim,

  3. Chicago-Naperville-Elgin, IL-IN-

and it doesn’t look like there’s any clear relationship with IPUMS-CPS metarea codes from the same years (2014-). While Tim said that IPUMS-CPS switched to the 2013 OMB codes starting in May 2014, it’s not clear what to do with this or how to “use or find the same code for both datasets?” which was the original question.

This is especially true if one wishes to look at many MSA’s (as researchers normally do), when simply looking for similar names seems inadequate. For instance, if I want to calculate an MSA-level variable using IPUMS-USA’s 2015 ACS and then merge it into the 2015 CPS (which seems like a very sensible research task), that’s surprisingly difficult. Aside from the quite general recommendation of watching out for incompletely identified MSA’s or the generally useful knowledge that IPUMS-CPS MSA information is provided by BLS, whereas IPUMS-USA’s is constructed from smaller geographies (both of these are points regularly and repeatedly made by IPUMS staff), it would be nice to have some guidance on how best to handle this task.

METAREA codes are based primarily on the 4-digit OMB codes of 1990 metropolitan areas but with adjustments to reflect hierarchical relationships among metro areas across time. For any group of metro areas that were at any time defined to be components of a single metro area together, the 4-digit METAREA codes were adjusted to use the same first 3 digits. This system enables users to obtain relatively consistent samples for “top-level” metro areas across time by grouping records according to the first 3 METAREA digits. (From the METAREA description.)

Therefore, when comparing pre-2012 IPUMS USA METAREA to IPUMS CPS METAREA, the first three digits can be used.

When making comparisons using the IPUMS USA variable, METAREA2013, the IPUMS CPS variable, METFIPS may be useful. For example, this page provides the 2014 onward metropolitan area codes. The documentation notes the following: some metropolitan areas are identified using the earlier (June 30, 2003) definition, care should be taken when tallying smaller areas, such as smaller cities, counties and metropolitan areas during the time frame of May 2014-July 2015, and that some areas will need two codes to identify them in their entirety during the period May, 2014 to July, 2015.

As Tim mentioned in his response, it is ultimately the choice of the researcher to determine whether comparing metro areas between the IPUMS-USA and IPUMS-CPS data sources is appropriate.

I hope this provides some additional clarification.