Matching individuals on a monthly base and avoiding discontinuity in 2004


In 2014 it was not possible to link IPUMS-CPS across samples (How do you link/join a tobacco supplement from Jan to a march sample with income data?). My question is whether this is possible today.

I know it is possible to identify individuals using HRHHID and HHRID2, linea, age, sex and race. At the moment I am facing a discontinuity in 2004 (from April to May) using these variables. This might due to the variables HRHHID and HRHHID2 but I do not know why.

Evaluating the descriptions of HRHHID and HHRID2, I firstly thought it might be possible to only use HRHHID2 since it was additionally backwards created from 1994-May 2004, but then the description of HRHHID says that one need both, HRHHID and HRHHID2.

So the question is how is it possible to link individuals between 1999-2013 and which variables do I have to use to uniquely identify the individuals? The goal is to match these individuals later on a yearly based month-to-month match, meaning from Jan 1999 to Jan 2000, Jan 2000 to Jan 2001…, Feb 1999 to Feb 2000 and so on.



The linking variable for CPS is now available (see CPSIDP and CPSID). I also recommend looking through the Drew, Flood, Warren (2014) paper linked in those variable descriptions as it goes into detail about some of the difficulties associated with creating these linking keys.

I hope this helps!