On the Measurement of Intra-Industry Trade
讨论产业内贸易的度量问题，指出分类聚合的未知影响是主要难点，并评估了不同度量方法的优劣，对研究贸易调整成本和国际交换决定因素的学者有参考价值。
countries (or more specifically newly industrialising countries) increased in importance as exporters of manufactures over this period, some 6o % of the overall expansion of trade in manufactures comprised trade within Western Europe and North America.' A phenomenon which appears to have increased in importance as a consequence of the growth of trade between manufacturing nations is intra-industry trade, i.e. the simultaneous import and export of commodities classified in the same 'industry' or product group. Although there are considerable difficulties associated with the computation of a meaningful summary statistic of the importance of intra-industry trade, relative to total trade, a substantial amount of evidence now exists to suggest that intra-industry trade has increased in importance to the point where it may account for in excess of 6o % of total trade in most developed market economies.2 Intra-industry trade is important for at least two reasons. First, accurate measurement of pure intra-industry trade can give some indication of the importance of determinants of international exchange other than relative factor proportions. Recent work has, for example, emphasised the role of preference diversity and decreasing costs. Despite attempts to generalise the HeckscherOhlin model, neither of these factors can be comfortably accommodated. Consequently, the Heckscher-Ohlin postulate (that countries specialise in and export commodities which use their abundant factor intensively) is of little value in explaining intra-industry exchange. Secondly, there is a possibility that adjustment to trade expansion may be easier when the expansion takes the form of an increase in intra- as opposed to inter-industry exchange.3 In order to explore either of these issues empirically or to test any of the models of intraindustry exchange which have emerged in recent years,4 intra-industry trade has to be measured as accurately as possible. The principal complication with measurement is the unknown influence of categorical aggregation. This paper discusses alternative approaches to assessing the influence of categorical aggre