The harmonized version of this variable has an error which is currently being fixed by the IPUMS USA team for 2018 and 2019. It should be released in the next few months. In the meantime you can merge the source variable onto your extract using the files available at data.census.gov/mdat. The variable is called TAXAMT.
The Census file can be matched to the IPUMS USA extract by merging the Census ID variables SERIALNO and SPORDER with the corresponding IPUMS ID variables CBSERIAL and CBPERNUM. The IPUMS household ID (CBSERIAL) is formatted slightly differently than the original Census ID. The Census variable is a 14-character string with “HU” or “GQ” in positions 5-6. The IPUMS variable is a 14-digit number, with position 5-6 being “00” for housing units (“HU”) and “01” for group quarters (“GQ”). You can convert one of the variables to be compatible with the other, or you can skip that step by including the source variables for CBSERIAL in your extract. No conversion is necessary for CBPERNUM.