The Matthew effect in science funding