Micro Foundations for International Productivity Comparisons

This paper describes the methodology and procedures of international comparisons of productivity levels for twelve manufacturing industries (producing food products, beer, soap and detergents, iron and steel, machine tools, various types of machinery, computers, audio and video equipment, industrial electronics, wired communication equipment, passenger cars and car components) between Germany, Japan and the United States. These estimates were carried out for a study on comparative productivity by the McKinsey Global Institute (1993). The starting point were estimates derived form the International Comparisons of Output and Productivity (ICOP) project at the University of Groningen (van Ark and Pilat, 1993). The original ICOP estimates on output, labour input and unit value ratios (the latter were used to convert output to a common currency) for the selected industries were adjusted where necessary to attain exact matching of activities on the basis of information derived form trade sources, companies and industry experts. Significant alterations were made to individual product PPPs to take account of differences in product mix and product quality between the countries. The extent and the direction of the corrections is different for the various industies. The biases are biggest for industies within the machinery and transport equipment branches. For manufacturing as a whole the original ICOP productivity estimates were only slightly downwardly biased by about 4-6 per cent. It is concluded that when using census data for comparisons at industry level, mix and quality problems need to be carefully considered. However, at a more aggregate level, for example for major branches in manufacturing, the original census-based UVRs are more robust.