Kaldor's ‘technical progress function' and Verdoorn's law revisited