The first of a series of international workshops on `Empirical Modelling of North-South Macroeconomic Interactions' was held on 2/3 February. The purpose of these workshops is to bring together researchers on this topic, particularly those from the North with those from the South. The Rockefeller Foundation will provide financial support for the workshops, with additional support from the World Bank. The February workshop was organized by Ralph Bryant (The Brookings Institution) and by CEPR Research Fellows David Currie (London Business School) and David Vines (University of Glasgow).
The first day of the workshop was devoted to the review of global models, focusing on their ability to analyse North-South interactions. On the second day, the workshop discussed the development of models for the South that were oriented towards studying adjustment with growth.
In the workshop's opening paper, David Vines began by identifying the principal North-South linkages. He described a model constructed jointly with Anton Muscatelli, designed to simulate North-South interactions. Vines stressed the need to model the hitherto neglected supply-side linkages between the South and the North. One linkage in particular is controversial: Southern investment and its financing could be modelled in more than one way. Investment could be determined in a neo-classical framework, with external finance adjusting to accommodate it, as in the Vines-Muscatelli model, or the supply of external finance could be constrained and Southern investment accommodating, as in the IMF's MULTIMOD model. Vines went on to illustrate the results of a fiscal expansion in the North. Simulations of the Vines-Muscatelli model, as well as of the empirically based MULTIMOD, revealed that such a fiscal expansion was harmful to the South in the long run. Vines concluded by describing the empirical research which he and Currie are directing in these areas.
Some participants felt that the results of Vines's simulation experiments were counter-intuitive or ex aggerated, and attributed this to Vines's assumption that the South alone produced primary commodities. Other participants suggested a more general specification of investment/debt linkages, which allowed different countries to be modelled differently according to the potential supply of external finance. It was also argued that the effects on the South of exchange rate movements in the North needed more attention from modellers.

Next Christopher Gilbert (Queen Mary College, London, and CEPR) and Andrew Hughes Hallett (University of Newcastle-upon-Tyne and CEPR) described the motivation for and salient features of their new research on `Global Macroeconomics of Commodities, Oil and Debt'. They sought to build a model of the LDCs, based on demand and supply equations for five commodity groups and five LDC regions. The model explicitly linked payments of interest and principal on debt to production and consumption decisions. The authors noted the econometric difficulties of modelling debt service, as data were available only since 1982; even if data were available for the earlier period, structural changes made it unusable. In the ensuing discussion, workshop participants identified other empirical difficulties, including the problems of taking into account the effect of Southern debt service problems on Northern financial markets and of modelling flows of foreign private investment. It was difficult and often impossible to develop models of individual LDCs; Gilbert and Hughes Hallett, like other researchers, had aggregated countries into regions. There was a long discussion of the possible criteria for constructing homogeneous regions. Participants suggested a number of criteria, such as degree of indebtedness, level of per capita income, commodity composition of exports, political regimes and nature of reactions to shocks. Many modellers construct regions on a geographical basis: this practice found advocates, who argued for its stability over time, its comparability, and consistency with historical patterns of trade and currency linkages, as well as its possible overlap with other criteria.
Christian Petersen (World Bank) then presented a taxonomy of global models in terms of their capacity to generate forecasts, disaggregation of the Southern economies, and degree of detail in model specification for the South. The models ranged from the simple global accounting frameworks currently being developed at the World Bank to the elaborate and long established LINK model. Petersen noted that several models were used at the Bank for global forecasting and the analysis of alternative scenarios; the choice for any given application was based on the relative strengths of the models.
The afternoon session began with Michael Beenstock (Hebrew University of Jerusalem and CEPR) reporting his progress in building a North-South model and outlining the extensions he had made to the modelling of the South by endogenizing growth and inflation. Beenstock emphasized the importance of defining some key variables carefully: total trade, not just merchandise trade; net trade (after netting out intra-South trade), not gross trade; net, not gross debt. Beenstock also described the methods that he had devised to surmount data difficulties, in particular in computing net trade deflators and Southern exchange rates. The discussions that followed concentrated on the reliability of International Financial Statistics and World Tables data because of `breaks in series'. The World Bank and IMF representatives acknowledged this difficulty and noted that in their work they often rely on improved data series (which are confidential). Participants argued that the Bank or the IMF could be asked to make available to researchers these improved data series in aggregated form (perhaps using aggregation criteria specified by the researchers). This would allow researchers access to better data while at the same time protecting the confidentiality of data at the country level.
The afternoon session concluded with a series of shorter presentations describing current research on global models: Paul O'Brien (OECD) on INTERLINK, Simon Wren-Lewis (National Institute of Economic and Social Research and CEPR) on GEM, Peter Pauly (University of Pennsylvania) on LINK, Hiromi Kato (Economic Planning Agency, Japan) on the EPA world model, Kirit Parikh (India Gandhi Institute of Development Research, Bombay) on the IIASA's global agricultural model, and Ralph Bryant on the networking of research activities at the Brookings Institution.
The workshop's second day began with Peter Montiel (IMF), who described research at the IMF on macroeconomic adjustment and growth. First, he described the Polak model, which was used for financial programming of adjustment, and the RMSM of the World Bank, which was added to the Polak model to give the analysis a `growth' dimension. Montiel then described the features of an `ideal' model, which among other things would incorporate forward-looking behaviour, distinguish households which were credit-constrained, take account of the complementarity of public investment and allow for quantitative restrictions on credit, trade and capital flows. Montiel outlined a simulation model currently being developed at the Fund based on the Mundell-Fleming model, which he described as a half-way house between the simple Polak-RMSM model and the ideal model. Jaime Ros (UN Commission for the South, Geneva), in his discussion of Peter Montiel's paper, questioned whether it was appropriate in a short-run analysis to assume that capacity was fully utilized. Ros argued that a modified version of Montiel's model could generate different and more realistic outcomes: a credit contraction, for example, could affect both actual output and the growth of capacity. Workshop participants noted the importance of a variety of other issues, such as the wide differential between debtor and creditor interest rates in LDCs and the need to allow for both backward- and forward-looking expectations in simulation models.

Joaquin Vial (CIEPLAN, Chile) reviewed the experience of building empirical macroeconomic models in Latin America. Models tended to be unreliable tools for policy analysis: estimated multipliers varied widely in magnitude, aggregate demand shocks had a surprisingly small inflationary impact and nominal shocks appeared to be non-neutral. Reacting to a question on the validity of aggregation of Latin American economies, Vial remarked that there were at present strong structural similarities which justified aggregation but that he could not be sure whether this would be so in the future.
Shantayanan Devarajan (Harvard University) discussed Computable General Equilibrium (CGE) models and illustrated, by means of a two-sector model, how the CGE approach could be used to study adjustment policy in the medium term. The discussions which followed Devarajan's presentation revealed considerable enthusiasm for combining this two-sector approach with more conventional macroeconomic models. Some reservations about the CGE approach were voiced: the `calibration techniques' used to set the values of the models' parameters were often based on just one set of observations. It was also noted that CGE models ignore inflation, an essential feature of the adjustment process.
David Vines presented a research agenda for LDC macroeconomic analysis. He described how a typical LDC model would differ from a model of an OECD economy in the treatment of the domestic capital market, the inflationary process, external capital flows, the role of labour in production and its inherent dualism. Vines argued that the simple two-sector CGE model presented by Devarajan could be brought closer to the macromodelling tradition by suitable relabelling of sectors and by incorporating a Phillips curve relation.
Vittorio Corbo (World Bank) presented a historical review of research at the World Bank on adjustment and growth. He presented a simple algebraic model to help organize the research agenda, which began with simple accounting schemes and extended to the building of quite large, `modern' macroeconomic models of a number of countries and a study of the political economy of adjustment.
The workshop's concluding session was partly devoted to a discussion of `networking' arrangements to allow researchers to share data, models and software. Suggestions for topics for the next meeting were also discussed. Arrangements to create and share a reliable and comprehensive databank could take several years to complete. Vittorio Corbo volunteered to circulate to all the participants information about the various databases currently maintained at the World Bank, and CEPR offered to act as an information exchange and facilitate continued contact among the participants. Access to some global models, such as GEM and LINK, was already available to researchers and the OECD would allow visiting researchers to use the INTERLINK model. Ralph Bryant noted that the Brookings Institution hoped to be able in due course to provide support for scholars, particularly from LDCs, to spend some time at leading institutions to learn about global models.

The group decided that its next meeting would be combined with a seminar on balance of payments modelling for heavily indebted economies, to be held in late 1989 at the Catholic University in Rio de Janeiro. Some participants suggested that for the next meeting modellers should conduct common policy experiments and communicate their results to participants for discussion at the workshop. Alternatively a policy issue such as debt forgiveness could be the theme of the meeting, and different models could be used to explore the issue. Other topics were suggested as well, such as modelling highly inflationary debtor economies (HIDEs), NICs and technology transfer issues

Some of the research discussed at this workshop is described in more detail in:
Macroeconomic Interactions Between North and South, edited by David Currie and David Vines.
This volume, published by Cambridge University Press in September 1988, was based on an earlier conference organized by CEPR with the International Economics Study Group.