Student achievement motivations are crucial in learning in two ways: as a determinant and an aim of learning. In this study, we focus on two related questions with regard to achievement motivations: to what extent are they subject-specific, and to what extent are they malleable? Answers to both questions are especially important when aiming to influence motivations. Malleability of motivation is studied by designing structural equation models that explain achievement motivation out of the most stable student characteristics one can think of: gender, and personality traits. Subject matter variability is studied by estimating these models for five main subject areas in a business program. The motivation construct we use is based on the expectancy-value model and distinguishes four different facets: cognitive competence, difficulty, task-value and affect. We find evidence for strong subject-specificity and considerable non-malleability of achievement motivation; part of that last aspect is a remarkably constant over-confidence gender gap present in the data of the calibration of competency beliefs and performance in all subject areas.