On the determinants of cost performance and the decomposition of returns to scale measures in the presence of quasi-fixed inputs a comment on Morrison and Schwartz