It’s an approximation that needs to be imputed, and can’t be measured directly. It doesn’t need units to be a useful measure against itself at different periods, to measure the rate of change of how environmental variables effect a nation or firms production function. I think it’s cool and useful that someone tried to quantify how much infrastructure affects productivity by abstracting away all the immeasurable variables into one term. It’s an extremely important concept that deserves more work because of how closely it is tied to advances in standard of living. It’s not perfect but it’s very interesting and could be ripe for a revamp given how much data we are now collecting from all sources

My point about units is that (depending on the exact model) if you change the units, your conclusion may change, which is absurd.