We build a compact global macroeconometric model capable of generating point and density forecasts for a core set of macroeconomic factors using recent advances in the analysis of cointegrating systems. We do so for a set of countries/regions and explicitly allow for interdependencies that exist between national and international factors. In an unrestricted VAR(p) model in k endogenous variables covering N countries, the number of unknown parameters will be unfeasibly large. We deal with this issue by first estimating individual country/region specific vector error-correcting models, where the domestic variables are related to corresponding foreign variables constructed to match the relevant international trade pattern. The country models are then combined to generate forecasts for all the variables in the world economy simultaneously. We estimate the model using quarterly data from 1979Q1 to 1999Q1, and investigate the time profile of the transmission of shocks of one variable to the rest of the world to show the degree of regional interdependencies.