Interpreting coefficients in log-form when their unit is percent

[Noemi Seng]

Hello all,

in my regression, the dependent variables is log of GDP growth and the main explanatory variable of interest is log of aid to GNI. So the unit of both variables is percent. When interpreting the regression coefficient, I am thus not sure whether to interpret it in percent or in percentage points. Also, with the log-form, the effect of aid/GNI on GDP growth should depend on its average level, right? How can I calculate the effect then? I would be very thankful for some help.

[George Ford]

you should not log GDP growth as it can take negative values. nor would I log the ratio aid/GNI.

[Noemi Seng]

Okay thank you for the advice! Maybe, for my learning, could you still help me on how I would interpret the coefficients in the log form?

[George Ford]

It's an elasticity. %changeY/%changeX