Most studies of European integration consider the effects of `1992' on international trade flows and therefore require empirical estimates of the relevant trade functions. The estimates used in empirical applications are often based on relatively old estimation techniques and data or derived from data on countries other than that to which they are applied. Much of the research in CEPR's project on `The Consequences of 1992 for International Trade' has concentrated on improving both the theoretical specifications and the data used in such estimations. Some of the issues arising from this research were examined in papers presented at the project's second workshop, which was held in London on 20/21 December and organized by L Alan Winters, Co-Director of the Centre's International Trade programme. This project is funded by the Commission of the European Communities under its SPES programme, and financial support is also provided by the UK Department of Trade and Industry and Foreign and Commonwealth Office.

In the first paper of the workshop, `Macro-Trade Functions with Imperfect Competition', Joaquim Oliveira-Martin (Centre d'Etudes Prospectives d'Informations Internationales, Paris, and OECD) and Joël Toujas-Bernate (Institut National de la Statistique et des Etudes Economiques, Paris) sought to incorporate developments in the new trade theory concerning imperfect competition, increasing returns to scale and product differentiation into the econometric estimation of trade elasticities. They first used a cointegration technique to test for the existence of a long-run relationship between relative prices and market shares. In 50% of cases, they found no clear relationship, and they attributed this finding to the exclusion of non-price effects from the estimation. They then estimated the trade elasticities using a translog functional form which permitted market shares to be determined by both prices and the number of competing firms.

In models of international trade with monopolistic competition and product differentiation, the ratio of any two producers' aggregate market shares is determined not only by their relative prices but also by the relative number of firms (or product varieties) in each market. The authors sought to capture the effects of the latter by constructing a composite index of data on prices (or unit values) and gross output indices to proxy the number of representative firms (products). Such a two-level functional form can be generalized to allow different degrees of differentiation both across product groups and among individual product varieties within a group. The authors reported a series of estimates undertaken by imposing different elasticities of substitution chosen to reflect the extent of the product differentiation they assumed. These results covered imports into France, Germany, Italy and the UK, for the textiles, chemicals and electrical machinery industries. They broke down supplies among domestic, other EC, and `all other' producers and showed that including non-price effects improved the model's performance in many cases. Moreover, estimates of the direct price effects based on the assumption of greater product differentiation were typically more in accordance with expectations.

In the discussion that followed, Anthony Venables (University of Southampton and CEPR) and Willem Buiter (Yale University and CEPR) suggested that the supply side of such a model could be estimated using a very similar equation to that used by the authors for the demand side, and they questioned whether the demand side was accurately identified. Buiter also suggested that European integration will lead to the homogenization of tastes that will change the underlying taste parameters in the utility function. L Alan Winters (University of Birmingham and CEPR) suggested that national taste differences are likely to become less important and that product differentiation will be increasingly driven by the underlying characteristics of the goods. He also suggested that the authors should examine the welfare implications of the completion of the internal market.

L Alan Winters (University of Birmingham and CEPR) and Paul Brenton (University of Birmingham) presented a preliminary paper entitled `Estimating Trade Functions for Exploring 1992', which outlined their project's main objectives and the theoretical structure of the model and presented some very provisional results. They intended to provide a comprehensive set of parameters to describe the price responses of EC member countries' international trade at a detailed industrial level. Instead of assuming the separability of imports and domestic supplies in demand, they suggested simultaneously estimating the allocation of sales across supply sources at broad levels of geographical aggregation, using the `Almost Ideal Demand System' as a first-order approximation to any underlying demand function. This would avoid the restrictions upon behaviour implied by simple models, while allowing the restrictions suggested by theory such as homogeneity, adding-up and symmetry to be imposed easily.

Winters and Brenton intended to estimate their model for France, Germany, Italy and the UK, using data for approximately 80 industries for the period 1970-87. Estimating systems of demand equations from data on 17 suppliers of imports will require a hierarchical structure, and proper aggregation will require that the demand of lower-level groups be homothetic, i.e. that suppliers in each group in the lower levels of that structure face the same income elasticities. A reasonable approximation might be for countries to be grouped by broad levels of per capita income. The authors intended to estimate their model for each market, and then to run five illustrative simulations of the consequences of 1992. These would explore the effects on trade flows and welfare of five different combinations of price reductions on domestic, intra-EC and extra-EC supplies.

Discussing the paper, Gernot Klepper (Institut für Weltwirtschaft, Kiel, and CEPR) argued that the use of price data based on unit values might bias the elasticities, especially in sectors such as high-technology goods where the quality of unit value data may be poor, and he suggested using price indices instead. Anthony Venables suggested that any apparently low price elasticities were probably short-run elasticities, while the corresponding long-run elasticities were likely to be higher.

In the final paper of the workshop, `Trade Between the EC and the NICs', Fabio Sdogati (Politecnico di Milano) examined the short- and long-run effects of income and relative price changes on trade flows between the European Community and the NICs, focusing in particular on South Korea. Sdogati began by noting the widespread assumption that `1992' will have two opposing effects on the rest of the world: an income effect (on demand for its exports), which is positive, and a relative price effect which is negative. Sdogati noted four sources of ambiguity that will affect the size of the net effect: reduced costs within the EC, reduced prices and hence higher real incomes, exchange rate changes (in particular between the dollar and NICs' currencies), and the NICs' higher growth rate.

Sdogati intended to examine this issue by estimating income and price elasticities by industry and supplying country. In the present paper he presented some preliminary results for South Korea. They covered total exports to France, Germany, Italy and the UK, and also for the US. These results suggested that the impact of 1992 will be small: this was not because the income and relative price effects cancel each other out, but rather because they are all small. Even if the income and relative price effects are small in aggregate, however, there may be large changes in particular industries. Sdogati estimated the income and price effects for textiles which might be considered `at risk' and his results showed that changes in relative prices have a significant impact on trade flows for France, Germany and the US.

In the discussion that followed, L Alan Winters noted that while the reported elasticities were actually quite high, their low significance made it impossible to draw firm conclusions. Alasdair Smith (University of Sussex and CEPR) suggested that the textiles industry was probably not a suitable choice for a case study, since it is subject to many quantitative restrictions, so that estimation is unlikely to pick up the parameters of the demand curve correctly. Carl Hamilton (Institute for International Economic Studies, Stockholm, and CEPR) proposed that an examination of the effects of 1992 on trade between the European Community and NICs should include the East European countries, which are likely to compete with the NICs in a number of important areas. André Sapir (Université Libre de Bruxelles and CEPR) suggested that reduced costs within the Community following the elimination of internal barriers on the flow of NICs' goods might prove important in certain sectors.

The workshop concluded with a general discussion of `1992 The Process and the Research', in which participants discussed further areas of research that might be incorporated into the programme. Issues raised included the interaction between the external impact of the completion of the European internal market and the success or failure of the Uruguay Round, the relationship between trade and industrial policies, and in particular the difficulties of defining `strategic' sectors. Participants agreed that it was inappropriate to seek to redefine the project at this stage, but that these issues might be considered further in research currently under way.