Hello, I am trying to replicate the identification strategy used by Card (2001) here: https://davidcard.berkeley.edu/papers/immig-inflows.pdf. In order to do this, I need the total population of immigrants by country of origin residing in the US in a baseline year (1990 in my case), the distribution of these immigrants across geographic areas in 1990 (so, for instance, the counties/PUMAs in which these immigrants lived in 1990, always by country of origin), and the annual national inflow of immigrants by country of origin in later years (2000-2018). As Card (2001) uses Census Data to do this, I imagine that it is possible to infer these quantities by using IPUMS microdata and person weights. Is this correct? Which variables would it be most appropriate to use to obtain the annual inflow of immigrants by country of origin for the period 2000-2018? Are there alternative data sources I should consider?
Thank you.
The total population of immigrants and their country of origin can be identified using the detailed version of BPL (Birthplace). Annual inflow of immigrants can be estimated using YRIMMIG (Year of Immigration) or YRSUSA1 (Year in the United States). The person-level weights PERWT or REPWTP should be used to estimate nationally-representative statistics with these variables. As you noted, you will need to use PUMAs in some cases since not all counties are available (see this blog post for a more detailed discussion about missing U.S. counties).