The second CEPR international workshop on `North-South Macroeconomic Interactions' took place on 29/31 October 1990 in Rio de Janeiro, held in association with the Brookings Institution and the Pontifícia Universidade Católica do Rio de Janeiro (PUC-Rio). The October conference was organized by Ralph Bryant, Senior Fellow in the Economic Studies program at Brookings, Winston Fritsch, Professor of Economics at PUC-Rio, and David Vines, Research Fellow in the International Macroeconomics and International Trade programmes at the Centre for Economic Policy Research and Professor of Political Economy at the University of Glasgow. Financial assistance for this series of workshops is provided to CEPR by the Rockefeller Foundation, and further support for this workshop was provided by United Nations Development Programme.

Modelling North-South Interactions

Paul Masson (IMF) opened the first session, which was devoted to simulations of North-South macroeconomic interlinkages using the major global models, with a paper on `The Effects of Industrial Country Fiscal Policies on Developing Countries in the 1980s', jointly written with John Helliwell. He used the IMF's MULTIMOD model to show that the divergences in industrial countries' fiscal and monetary policies in the 1980s redistributed growth and current account balances among themselves but had only modest spillover effects on resource transfers between North and South. The global tightening of monetary policy had strong negative effects, however, on developing countries which suffered from falling export volumes and increasing debt-service payments. Debtor countries were forced to run current account surpluses which reduced domestic investment and output.

David Currie (London Business School and CEPR) expressed surprise at the weak impact of fiscal policies and the sluggish response of short interest rates to increased budget deficits. Also, the category of net debtor countries was very wide, and the effects on Latin America, for example, were almost certainly underestimated by the aggregation.

Chris Allen (London Business School) and David Currie (London Business School and CEPR) then presented their paper on `North- South Interactions in GEM', which simulated the impact of global economic shocks on Latin America. They used a version of the LBS/NIESR Global Econometric Model, supplemented by the Srinivasan/Vines Latin American model, which characterized industrialized countries' policies by a fixed nominal money supply rule leading to overshooting of forward-looking exchange rates. They found that the effects of higher interest rates substantially outweighed the benefits to Latin America of increased world trade. This finding contrasted both with that of Masson and Helliwell using MULTIMOD and with those of simple reduced form trade models using GEM for other developing country regions. Allen and Currie also found that an expansion in Latin America financed by foreign borrowing had substantial short-run spillover effects on the industrialized countries (especially the US) and on world trade.

Paul Masson expressed surprise at the size of the effects of expansion on the rest of the world. Allen attributed this to the capital constraint on growth, whereby any fiscal expansion would be quickly offset by a similar increase in imports.

Warwick McKibbin (Brookings Institution) presented a paper on `Macroeconomic Linkages between the OECD and the Asia-Pacific Region', written jointly with Mark Sundberg. They used a version of the McKibbin-Sachs Global (MSG) model, which included models of the Asian NIEs and ASEAN countries with broad sectoral disaggregation, in which investment behaviour was driven by intertemporal optimization. The two regions responded very differently to external shocks: for example, the NIEs gained from a US fiscal stimulation but the ASEAN countries suffered. The relative effects on the two blocs depended critically on the commodity composition and direction of their trade, as well as their exchange rate regimes and outstanding foreign indebtedness.

In the ensuing discussion, Tain-Jy Chen (Chung-Hua Institution for Economic Research, Taiwan) emphasized the increasing importance of wealth effects on consumption in the NIEs and of private capital flows in Hong Kong and Singapore.

Christian Petersen spoke on his joint work with Desmond McCarthy, Beatrice Laury, E Mick Riordan and Jian-Ping Zhou (World Bank) entitled `Global Accounting Framework (GAF) The Basic System', which reported the Bank's latest work in developing a standard data base for the public, private and foreign sectors, covering 120 series for each of 137 countries. Petersen noted that the publication of this standard world data set will greatly advance the progress of global macroeconomic modelling.

`International Economic Linkages and the Evolution of the International Debt Situation' was the subject treated by Gerd- Peter Dittus (OECD). His paper reported the results of the OECD Secretariat's macroeconomic models of debt for the four principal countries and the aggregated remainder of Latin America for 1983- 6. Potential output and investment were functions of domestic savings and inflows of real external resources, and trade volumes depended on capacity utilization and world trade. This model attributed the continued rise in debt ratios to the rise in the dollar and the worsening of Latin America's terms of trade.

Christian Petersen (World Bank) also presented his own paper on `The Middle East Crisis: Possible Consequences for the Global Economy', in which he used the GEM model to assess the effects of the oil price shock and more generally to illustrate the use of such models for policy purposes. He warned that many of the key questions in policy simulations must remain matters of judgement, and Ralph Bryant observed that different models had produced quite divergent forecasts of the impact of the oil price shock. Chris Allen noted that differences in the assumed policy regime could give very different results even in simulations on the same model.
Ralph Bryant then chaired a round-table discussion of global modelling, in which Warwick McKibbin stressed the importance of supply side and noted that disaggregation by country and commodity had proved to be more important than expected. Paul Masson argued that supply-side modelling is more important for developing than for industrialized economies, since problems on their current accounts immediately affect imports of intermediate and capital inputs and thus constrain aggregate supply. David Currie suggested, however, that supply-side spillovers among the principal industrialized countries had probably been just as important as those on the demand side. He also questioned the credibility of forward-looking parameters in macroeconomic models: exchange rates and commodity prices might well be rationally forward looking, but the argument for consumption seemed less credible. Indeed, the evidence indicates that even financial markets tend to over-react to news. David Vines pointed out the differences in the closure of developing country blocs in the various models. In MULTIMOD, the financing constraint is absolutely binding and directly determines investment; MSG permits no explicit government feedback mechanism, but current account imbalances eventually affect demand through the money supply; while GEM allows interest rates to exert direct effects on domestic demand. Jaime Ros (University of Notre Dame) noted that the governments of Latin American countries are subject to draconian MULTIMOD-style closure rules. The responses of their private sectors will depend, however, on whether they have plentiful external assets (as in Mexico and Venezuela) or face external constraints (as in Brazil).

Macroeconomic Modelling of Developing Countries

Kenneth Rogoff (University of California, Berkeley) opened the second session, on the modelling of developing countries, with his joint paper with Jeremy Bulow entitled `Sovereign Debt Repurchases: No Cure For Overhang'. He noted that debt overhang reduces the marginal value of investment to a debtor country, since a proportion of any resulting income gains goes to its creditors. Debt buy-backs may stimulate investment by reducing this distortion, but debtor countries will not gain from buy- backs at discounted market prices, since their creditors will enjoy all the efficiency gains from increased investment. Negotiated buy-backs, on the other hand, allow the debtor country to bargain over the efficiency gains. According to Rogoff, the upper bound on Mexico's gain from its 1990 buy-back under the Brady Plan compared with a hypothetical market buy-back was some $4.2 billion, or some 2% of GNP.

David Vines (University of Glasgow and CEPR) and T G Srinivasan (University of Glasgow) presented two papers based on their Latin American model. Discussing `Simulations of an Econometric Model of Latin America', Vines explained that the demand side of their model incorporating public, private and foreign sectors was conventionally Keynesian with a multiplier/accelerator process and wealth effects. On the supply side, however, they included an upward-sloping supply curve in the real exchange rate; and the trade balance satisfied the Marshall-Lerner conditions. Vines noted that a fiscal expansion was very quickly crowded out by inflation, loss of competitiveness, and reserve outflows.

Srinivasan presented their second paper, entitled `Adjustment to a Negative Export Shock: Simulations Using a Latin American Model'. They simulated the effects of a fall in export demand with no policy response and also when the government responds by cutting public expenditure, for the cases where the current account deficit is covered by a reduction in reserves and hence in the domestic money supply and by debt accumulation. They found that the introduction of government feedback rules had little effect when the deficit was covered by reserves, but that it reduced the variation in the current account considerably when this was covered by debt.

Jaime Ros welcomed the paper as major step forward in developing country modelling, but he maintained that more attention should be paid to the sustainability of the public sector deficit and the effects of government spending cuts on investment. Warwick McKibbin pointed out that excluding the domestic interest rate might underestimate the effects of a fiscal expansion on domestic investment.

Klaus Schmidt-Hebbel and Silvana Vatnick (World Bank) presented a paper on `Macroeconomic Effects of Alternative Debt Transformation Schemes: A Simulation Analysis for Brazil'. They used a detailed model of various internal and external financial assets with a simple macro-structure to examine various debt transformation schemes. Their results showed that debt reduction had only a small effect on growth and that certain domestically financed packages may even be detrimental to growth.

Eustaquio Reis (Instituto de Pesquisa Economica Aplicada, Rio de Janeiro) argued that some of these results derived directly from the assumption that bonds were not indexed, which was rarely the case in Brazil. He agreed, however, that equity swaps that brought in no new money had little effect. Kenneth Rogoff pointed out that a debt-equity swap was essentially the same as a debt buy-back combined with a more positive approach to foreign investment. In principle, the same effect could be obtained through foreign investment incentives.

Yoon-Ha Yoo (Korea Development Institute, Washington DC) presented a joint paper with Won-Am Park on `External Adjustment of the Korean Economy'. They used the Institute's model to examine the impacts of the 1973, 1979 and 1986 oil shocks, the fall in the dollar and the reduction of interest rates on Korea's balance of payments. They found that their simulation results represented a substantial improvement in the accuracy of the decomposition of balance-of-payments effects over those of conventional models.

Tain-Jy Chen argued that their simulations must underestimate the effects of interest rates, since they did not affect investment. Warwick McKibbin suggested that taking account of the effects of the oil shocks on the world economy would increase the already large estimates of their impact even further.

Trade Modelling

James Riedel (Johns Hopkins University, Washington DC) opened a session on modelling the determinants of the prices and volumes of developing country exports with a joint paper with Premachandra Athukorala entitled `How Valid is the Small Country Assumption?'. According to this hypothesis, the demand for a given country's products on world markets is infinitely elastic, and its export volumes are therefore determined solely by supply considerations. The data Riedel presented for Korea in this paper and those from his earlier work on Hong Kong provided as much support for this `supply' hypothesis as for the more conventional approach based on export demand functions. Further, if the supply hypothesis were true, then developing countries' exports would not be dependent on the expansion of world trade, and their growth rates should be primarily determined by domestic considerations.

Anton Muscatelli (University of Glasgow) then presented a paper written with T G Srinivasan and David Vines on `The Empirical Modelling of NIE exports: An Evaluation of Different Approaches'. They attributed Riedel's well-known earlier conclusion that very different results are obtained by normalizing demand using a price equation rather than a quantity equation to the omission of variables or dynamic misspecification. They controlled for the latter by estimating general autoregressive distributed lag models for both demand and supply functions under alternative normalizations, and their results clearly rejected Riedel's simple dynamic specification. Their very high estimates of world trade elasticities (around 2.5), which in principle should lead to a continuing improvement in the NIEs' terms of trade over time, may have been overestimated on account of microeconomic factors such as increasing product diversification and technology transfer.

Alvaro Zini (Universidade do Sao Paulo) presented a paper on `Import and Export Functions for Brazil', in which he estimated structural demand and supply functions in the aggregate and also for four component sectors, on the grounds that trade in manufactures and trade in commodities react very differently. He found evidence of a structural change in trade performance following the debt crisis of the 1980s, and he expressed surprise at the large size of the export income elasticities in relation to the export price elasticities. These results directly contradicted the arguments of the proponents of extensive import substitution in Brazil, who had assumed that low income elasticities and high price elasticities would lead to `immiserizing' growth as the terms of trade worsened over time.

In the ensuing general discussion of trade modelling, Winston Fritsch noted that the debate over the determinants of exports remained open, and he emphasized the importance of technology transfer and product diversification to policy in developing countries. James Riedel noted the urgent need for empirical evidence to resolve the question about the microeconomic foundations. David Currie said that product diversification ran against the general `macro' specification of all the models and could only be introduced through substantial disaggregation, while Gerd-Peter Dittus remarked that the OECD Secretariat had already taken explicit account of product diversification in their model of Korea by adding a `market potential' term to the export volume equation, which reduced the estimated export elasticity from more than 5 to around 1.2. Chris Allen proposed an alternative model in which the small country assumption might hold, but capital was mobile among the developing countries owing to the activities of multinational corporations. In this case, the structural demand and supply equations might not be properly identified, since relative price effects in the demand function might in fact represent producers' supply decisions about relative sourcing.

Commodity Prices and the World Economy

Annalisa Cristini (Istituto Universitario di Bergamo) presented a paper on `OECD Economic Performance: The Role of Primary Commodity Prices', in which she demonstrated that an increase in commodity prices increases the wage wedge and reduces employment, while terms-of-trade improvements allow developing countries to reduce their debt exposure and hence reduce pressure on world capital markets. Not surprisingly, the depressive effects on wages dominate, and estimating the commodity price equations showed that these are largely insensitive to OECD demand in the long run.

A paper on `Primary Commodity Prices and Inflation' by Christopher Gilbert (Queen Mary and Westfield College, London, and CEPR) was given by David Vines. According to Gilbert, there is no satisfactory method of measuring aggregate OECD inflation, and he therefore used separate reduced form models of inflation dependency on oil and non-oil commodity prices for the US, Japan, West Germany and the UK. He showed that the reduction in commodity prices in the early 1980s played only a small role in reducing inflation, while the effects of the 1986 oil price fall were more substantial. Further estimates of non-oil commodity price equations indicated that inflation was relatively insensitive to interest rates and to long-run OECD growth, but that commodity prices are of limited use as indicators of inflation.

David Currie said that Gilbert's model underestimated the effects of the fall in commodity prices in the early 1980s, since he measured them in absolute terms and not relative to the price of manufactures. Annalisa Cristini thought that he had exaggerated the OECD price index problem, since a price index could be derived using a basket of currencies.

Conclusions

Ralph Bryant then introduced a general discussion of the modelling issues raised at the workshop. He first noted the general consensus on the importance of clearly worked-out assumptions regarding the supply side, capital flows and the exchange rate. Other critical issues remained open, however, including the closure of models and the sustainability of policy. Trade and commodity prices were now being examined closely, and the World Bank had at last begun to address the overall problem of the availability of data.

Warwick McKibbin suggested that computable general equilibrium models could be linked with estimated time-series models and that prototype small country models could be quickly built from input- output data. Paul Masson maintained, however, that the relatively small size and comprehensibility of existing models were important advantages. David Currie welcomed the major role to be played by the World Bank in producing a consistent data base, which would be essential to all empirical modelling of developing countries.

Gerd-Peter Dittus questioned whether detailed developing country models were really required when it might be possible to use reduced form reaction functions. Chris Allen replied that experience of using the GEM model for Latin America suggested that reaction functions were inadequate to deal with the complex dynamics of different responses to external shocks.

Winston Fritsch hoped that future research would focus on the impact of structural changes in the world economy such as technology transfer and increased foreign direct investment. He noted the key importance to policy-makers in the South and particularly in Latin America of whether their economies are capital-constrained as in the Glasgow model, or characterized rather by a lack of effective demand. Christian Petersen welcomed the increased convergence between the Glasgow model and the computable general equilibrium models, as demonstrated by the recent emphasis of both models on supply-side considerations.



CEPR intends to publish a selection of the papers presented at this series of workshops in a forthcoming conference volume.