Multilateral comparisons of productivity, terms-of-trade and factor accumulation

This paper proposes a new approach for multilateral comparisons using index numbers. The new approach combines two recently-proposed innovative techniques to examine differences among economies at various levels. The Minimum Spanning Tree algorithm, based on the idea of minimizing substitution bias of bilateral comparisons, provides a possible ordering for panel data. Making use of the suggested ordering, bilateral Törnqvist price and quantity indexes are calculated and multilateral indexes are obtained by chaining. An index-number based approach is then used to decompose the differences in GDP at the bilateral level. Different sources that contribute to the differences in GDP are considered: productivity differences, terms of trade differences, factor endowments differences and domestic output price differences. The newly formed indexes are base-invariant which provides strong support for using the technique for multilateral comparisons. An illustration of the technique using data from China and four OECD countries is included.