We identify the effects of part-time employment, study time at home, and attitudes in school, in the production function for educational performance among UK teenagers in compulsory education. Our approach identifies the factors driving differences between the reduced form 'policy effect' of in-school employment, and its direct effect or 'production function parameter'. Part-time employment is shown to reduce performance among girls but not boys, because employment crowds out both study time at home and positive attitude in school to a greater extent for girls than boys. Part-time work also induces earlier initiation into risky behaviours for girls than boys.