If you are referring to IPUMS-USA data, then it is not possible to create panel data. The ACS PUMS data is composed of a random sample of respondents from each year. Since there is no effort made to include the same respondents in multiple samples, with the ACS actively avoiding including the same households in consecutive samples, constructing a panel of respondents or households over modern samples is not possible (although historical Census samples have been linked).
If you are referring to IPUMS-CPS data, then it is possible to follow a person/household for a maximum of 16 months (4 consecutive months in the sample, 8 consecutive months out, 4 consecutive months back in). You should reference these previous answers for more information on utilizing the panel aspect of the CPS: here, here, here, and here.
I hope this helps.