Marginal effects of log transformed variable

[Serena Gallo]

Hi everyone,
I have some issues.
I ran logistic regression with some independent variables in log-transformed (ln one-plus).
I ran marginal effects with the following command:

margins, dydx(*)

My question is: How Can I interpret the margins for log variables?
For example, if margins for ln(bank size) is 0.034: one-unit increase in ln bank size increases the probability of bank failure by 3.4%, Is it right?

[George Ford]

if both Y and X are ln, then it's an elasticity. So 10% increase in bank size is 0.34% increase in Y.

[Richard Williams]

I like how the mcp command lets you plot relationships involving logged variables. See pp. 7-10 of

https://www3.nd.edu/~rwilliam/xsoc73994/Margins03.pdf