Comparing change with quadratic term in regression.

I have a model that relates the log of wage to expertise, expertise squared, looks, and gender.

Where expertise is in years, looks is an ordinal scale 1->5, and female =1 if gender is female.

I want to test that the change in wage with respect to expertise is statistically different to the change in wage with respect to looks. At first I thought to test B1+B2=B3, but am struggling to make any sense of it. How do I reparameterize the model to test this?