用多区域因素模型预测世界贸易（英）

Nowcasting World TradeWith a Multi-RegionFactor Model Chris Jackson and Daniel Rivera Greenwood WP/26/48 IMF Working Papersdescribe research inprogress by the author(s) and are published toelicit comments and to encourage debate.The views expressed in IMF Working Papers arethose of the author(s) and do not necessarilyrepresent the views of the IMF, its Executive Board,or IMF management. 2026MAR IMF Working PaperResearch Department Nowcasting World Trade With a Multi-Region Factor Model Prepared by Chris Jackson and Daniel Rivera Greenwood* Authorized for distribution by Rafael PortilloMarch2026 IMF Working Papersdescribe research in progress by the author(s) and are published to elicitcomments and to encourage debate.The views expressed in IMF Working Papers are those of theauthor(s) and do not necessarily represent the views of the IMF, its Executive Board, or IMF management. ABSTRACT:This paper presents a nowcasting model for global trade that allows for regional dynamics andspillovers. World trade growth is driven by common global factors but also regional trends. While existing tradenowcasting models have focused on the former, we allow for the latter using a dynamic factor model (DFM)with a multi-factor block structure. By directly modeling global trends, regional variation and spillovers, weimprove on the performance of standard trade nowcasting models, particularly periods characterized byregional heterogeneity. A multi-factor regional framework may be particularly advantageous for tracking tradedevelopments in the future given a period changing trade patterns and geo-economic fragmentation. Themodel also sheds light on trade spillovers and the drivers of news in global trade: Asia, in particular, hasnotable spillovers to the global and other regional trade cycles. Nowcasting world trade with a multi-region factor model Prepared by Chris Jackson and Daniel Rivera Greenwood1 1Introduction Trade is an important part of the global economy and viewed as a leading indicator of globalactivity. The ability to track global trade growth reliably in real time is therefore of keen interestto policymakers. Doing so, however, presents several challenges. Global trade data are publishedwith a lag. A variety of more timely relevant indicators exist to help forecast trade, but they areoften noisy and their mapping to trade data is uncertain. More so than other global macroeconomicdata, trade data are also interrelated: trade growth in one region has implications for the outlookfor trade in another. Dynamic Factor Models (DFMs) present a way to deal with these challenges. DFMs are a populartool for nowcasting macroeconomic data because they efficiently utilize information from large andvaried datasets. This is achieved by modeling the common dynamics of several series as explainedby a small number of latent factors (Stock and Watson 2016). They also have the advantage ofbeing able to handle efficiently the real-time data flow and allow for missing data at the end ofsamples due to asynchronous publication lags, the so-called ‘jagged’ or ‘ragged edge’ problem(Giannone et al. 2008). As a result, the literature on nowcasting global trade has tended to focus on DFMs or related prin-cipal components analysis, including Guichard and Rusticelli 2011 and more recently Barhoumi etal. 2016 and Mart´ınez-Mart´ın and Rusticelli 2021. These papers model world trade as a function ofa common global factor, with Guichard and Rusticelli 2011 finding that this method outperformsalternative forecasting methodologies such as bridge equations. Direct forecasts of global tradehave also been found to outperform aggregated country-level forecasts (Burgert and Dees 2009),which speaks to the synchronous nature of world trade. This paper builds on this literature in using a multi-factor DFM to nowcast global trade growth.While global factors are an important determinant of world trade growth, the forecasting perfor-mance of a DFM can be improved by allowing for regional factors and spillovers. Specifically, amulti-factor block structure can account for variation due to a global cycle, regional fluctuationsand series-specific variation. As argued by Doz and Fuleky 2020, if these regional fluctuationswere not properly modeled it would appear either as weak common factors or errors that are cor-related within the same region. Indeed, Guichard and Rusticelli 2011 found that adding additionaldata, particularly regional-level data, had a diminishing effect on the forecast performance of theirsingle global factor trade DFM. This may be not because those data have limited marginal infor-mation but because the model is incorrectly specified. Spillovers from one region to another arecaptured through the loading and VAR structure of the factors. In methodology, this paper is closely related to Cascaldi-Garcia et al. 2024, who use a similarmulti-region DFM for nowcasting activity in the euro area as a whole and in its largest membercountr