Is it correct to take logs of a variable, e.g. tax revenue, already measured in percentage form?

e.g.
(1) ln(GDP) = c + B1ln(physical capital) + B2ln(human capital) + B3TaxRevenue

vs


(2) ln(GDP) = c + B1ln(physical capital) + B2ln(human capital) + B3ln(TaxRevenue)


How would the interpretation change between (1) and (2) - I seem to be getting better and more appropriate results when I use (2) - but wondering if it is permissible to take logs of a variable already in percentage format (e.g. 25.8)

Paul

**It may help by saying that I am looking to interpret my answer as a percentage point increase in e.g. tax revenue...**