Top coded monthly gross rent values seem to vary more widely than expected


I’m using the 2014-2018 5-year ACS and trying to calculate a rent burden variable, which is the ratio of rental costs to household income. For rent burden I’m using monthly gross rent variable (rentgrs). However, it seems that it is topcoded.

Following the documentation for monthly contract rent variable on which the monthly gross rent variable is based, I followed the link to inspect the topcodes for each sample. Variables with rent values higher than the top code would be coded as the mean value for the state in which the observation resides.

When I inspected the topcodes for just 2018 in New York State above $3900, I found that there were several cases with different values. See attached output. I would have expected there to just be just one value for all cases in New York State in 2018 above the top code, since the mean value for the state replaces the actual value when it exceeds the threshold.

I cross-referenced these values with their respective contract rent value and there was only one value for each year as I would expect. I think the values on rentgrs vary because the gross rent variable is calculated from adding utility variables after the mean state value above the topcode has been added and every unit has different utility expenses. Otherwise, I would expect them all to have the same state mean above the topcode value.

Can someone confirm this for my peace of mind?

Thanks. Best,

You’re correct that in the ACS, the component values are topcoded (when applicable), and then summed to calculate RENTGRS. There is no additional topcode for RENTGRS.

1 Like

Great. Thanks for the help, Matthew.