Hello,
In ACS2022 1Y microdata, I see that there are about 9k HH observations where mortamt1==0 & mortamt2 != 0. As far as I can tell, this should not happen based on this phrase “By definition, respondents could not have a second or junior mortgage if they reported no first mortgage (see MORTAMT1).” See IPUMS USA: descr: MORTAMT2. There is some discussion that the Census allocated responses when this happened. I just want to make sure IPUMS is operating as expected here. For my current project, I have downloaded ACS 1Y files for years 2006, 2007, 2009, 2010, 2018, 2019, 2021, 2022, but I only see this for 2022.
In 2020, 2022, 2023, and 2024 1-year ACS samples, I do see cases where there is no first mortgage amount given in MORTAMT1 but MORTAMT2 gives a second mortgage amount. I do not see any such cases in other 1-year ACS samples. This reflects what is present in the original ACS data—the source variables, which you can also include in your data extract, show the same differences between samples.
While looking into your question, I noticed that the documentation for some of our mortgage-related variables could be improved. Our IPUMS USA team will work to update universe statements and codes to better reflect the original ACS documentation and data where necessary. Please let me know if you have any additional questions or comments.