How conflicting definitions of"manufactures"distort output and trade statistics

This paper argues that the definition of"manufactures"used in compiling production data for industrial and developing countries is far broader than the definition used for trade statistics. This limits the analytical utility of output and trade data for studies using, say, apparent consumption or import penetration ratios. International agencies also use different definitions of manufactures when compiling trade statistics. For some countries, these definitional changes produce major differences in the value and share of manufactures exports and imports. The author assesses the analytical implications of six differentdefinitions by comparing results when each is used to tabulate exports of the"manufactures"of 72 developing countries. Sensitivity tests show that five or six products - especially refined petroleum and some processed food products - are responsible for the main discrepancies. These items are included in the UNIDO (trade) and most agencies'output definitions but are excluded from the trade definition used by UNCTAD, GATT, and the World Bank. Clearly, international organizations must resolve these inconsistencies in definitions of output and trade statistics.