Some recent studies have debated the appropriateness of semi-log earnings function
in approximating actual earnings pro® les. This study contributes further to this
debate by estimating four speci® cations of earnings function, namely, semi-log quadratic,
semi-log cubic, quadratic and cubic with cross section data for 4035 full-time
urban male earners in Iran for 1993. The analysis shows that the semi-log quadratic
earnings function ® ts the actual empirical earnings pro® les very poorly. The quadratic
earnings function provides a much better approximation to the actual earnings.
Measures of out-of-sample prediction errors also show the superiority of quadratic
function over others.