Interpretation of inefficiency determinants in logarithms, SFA

[HA Borrero]

I am implementing a 'true fixed effect' model (Greene, 2005) using the 'sfpanel' Stata command. I'm mostly interested in analyzing the determinants of technical inefficiency (the z_it variables), so I assume that u_it is a non-negative truncation (at zero) of the normal distribution with mean z_itδ and variance σ². Now then, most studies introduce the z_it variables in their original units of measurement. Nonetheless, I wonder how the interpretation would change if I introduce them in logarithms [as ln(z_it)]. Could they be interpreted as 'elasticities'?