We develop a criterion to distinguish two dominant paradigms of international trade theory: homogeneous-goods perfectly competitive models, and differentiated-goods monopolistically competitive models. Our analysis makes use of the pervasive presence of home-biased expenditure. It predicts that countries' relative output and their relative home biases are positively correlated in differentiated-goods sectors (the "home-bias effect"), while no such relationship exists in homogeneous-goods sectors. This discriminating criterion turns out to be robust to a number of generalisations of the baseline model. Our empirical results, based on a world-wide cross-country data set, suggest that the differentiated-goods model fits particularly well for the machinery, precision engineering and transport equipment industries, which account for some 40% of sample manufacturing output.