Forecasting world trade: direct versus “bottom-up” approaches

In a globalised world economy, global factors have become increasingly important to explain trade flows at the expense of country-specific determinants. This paper shows empirically the superiority of direct forecasting methods, in which world trade is directly forecasted at the aggregate levels, relative to "bottom-up" approaches, where world trade results from an aggregation of country-specific forecasts. Factor models in particular prove rather accurate, where the factors summarise large-scale datasets relevant in the determination of trade flows.