About the Global VAR (GVAR) Modelling

The Global Vector Autoregressive (GVAR) approach, originally proposed in Pesaran et al. (2004), provides a relatively simple yet effective way of modelling interactions in a complex high-dimensional system such as the global economy. Although GVAR is not the first large global macroeconomic model of the world economy, its methodological contributions lay in dealing with the curse of dimensionality (i.e. the proliferation of parameters as the dimension of the model grows) in a theoretically coherent and statistically consistent manner. Other existing large models are often incomplete and do not present a closed system, which is required for simulation analysis, see Granger and Jeon (2007) for a recent overview of global models.

The GVAR model was developed in the aftermath of the 1997 Asian financial crisis to quantify the effects of macroeconomic developments on the losses of major financial institutions. It was clear then that all major banks are exposed to risk from adverse global or regional shocks, but quantifying these effects required a coherent and simple-to-simulate global macroeconomic model. The GVAR approach provides a useful and practical way of building such a model, and, although developed originally as a tool for credit risk analysis, it soon became apparent that it has numerous other applications. For an extensive survey of the latest developments in GVAR modelling, both the theoretical foundations of the approach and its numerous empirical applications, see Chudik and Pesaran (2016).

The GVAR can be briefly summarized as a two-step procedure. In the first step, small-scale country-specific models are estimated conditional on the rest of the world. These models are represented as augmented VAR models, denoted as VARX* and feature domestic variables and weighted cross-section averages of foreign variables, which are also commonly referred to as ‘star variables’, which are treated as weakly exogenous (or long-run forcing). In the second step, individual country VARX* models are stacked and solved simultaneously as one large global VAR model. The solution can be used for shock scenario analysis and forecasting as is usually done with standard low-dimensional VAR models.

The GVAR approach has been applied to a number of diverse problems. Individual units need not necessarily be countries, but could be regions, industries, goods categories, banks, municipalities or sectors of a given economy, just to mention a few notable examples. Mixed cross-section GVAR models, for instance linking country data with firm-level data, have also been considered in the literature. The GVAR approach is conceptually simple, although it requires some programming skills since it handles large data sets. Fortunately, an accessible and easy-to-use package (the GVAR Toolbox), with no background knowledge of MatLab or Excel required, has been developed by Smith and Galesi (2014). The GVAR Toolbox together with the publicly available GVAR quarterly dataset set, covering the period 1979–2016 for 33 countries, can be used for the application of the GVAR methodology.

GVAR Toolbox
Copyright L. Vanessa Smith and Alessandro Galesi (2014)

About the GVAR Toolbox: The GVAR Toolbox 2.0 is a collection of MatLab procedures with an Excel-based interface, designed for the purpose of GVAR modelling. It is primarily tailored to policy analysis and forecasting, though can be easily customised for other purposes. It is an accessible and easy-to-use package, with no background knowledge of MatLab or Excel required. In order to use it, both Microsoft Excel and MatLab have to be installed on the user's computer. No specific MatLab toolboxes are required for running the program. The GVAR Toolbox was originally launched in December 2010 with the release of version 1.0, sponsored by the European Central Bank. Version 1.1 subsequently followed in July 2011. The GVAR Toolbox 2.0 was released in August 2014 and is available to download, free of charge, from the GVAR Toolbox webpage.

Download the GVAR Toolbox: Visit the GVAR Toolbox webpage to download the lastest version of the Toolbox as well as a detailed document describing how to run and use the GVAR Toolbox 2.0, with the aim of building a global VAR model that allows for global interlinkages. In addition, the document contains details of the underlying econometric and computing methods.

Threshold-augmented Global VAR (TGVAR)

The Threshold-augmented Global VAR (TGVAR) is developed in Chudik et al. (2020) "A Counterfactual Economic Analysis of Covid-19: Using a Threshold Augmented Multi-Country Model", Cambridge Working Papers in Economics No. 2088.

You can download the TGVAR codes and data for the following two papers:

• Click here for the data as well as the Matlab files needed to replicate the empirical findings in Chudik et al. (2021) Covid-19 Fiscal Support and its Effectiveness", Economics Letters 205, pp. 109939/1–5.



• Click here for the data as well as the Matlab files needed to replicate the empirical findings in Chudik et al. (2020) "A Counterfactual Economic Analysis of Covid-19: Using a Threshold Augmented Multi-Country Model", forthcoming in Journal of International Money and Finance. .




Updated: Global VAR (GVAR) Quarterly Dataset, 1979Q2-2019Q4

·         Compilation, Revision and Updating of the Global VAR (GVAR) Database, 1979Q2-2019Q4, Kamiar Mohaddes and Mehdi Raissi (2020), University of Cambridge: Judge Business School (mimeo).

About the GVAR Dataset: This is the latest version of the GVAR dataset. It includes quarterly macroeconomic variables for 33 economies (log real GDP, y, the rate of inflation, dp, short-term interest rate, r, long-term interest rate, lr, the log deflated exchange rate, ep, and log real equity prices, eq), as well as quarterly data on commodity prices (oil prices, poil, agricultural raw material, pmat, and metals prices, pmetal), over the 1979Q2 to 2019Q4 period. These 33 countries cover more than 90% of world GDP. It would be appreciated if use of the updated dataset could be acknowledged as: “Mohaddes, K. and M. Raissi (2020). Compilation, Revision and Updating of the Global VAR (GVAR) Database, 1979Q2-2019Q4. University of Cambridge: Judge Business School (mimeo)”.

Download the GVAR quarterly dataset: You can download the data, as well as a description of the compilation, revision and updating of the GVAR dataset, from here.



GVAR Handbook and Survey Paper

·         The GVAR Handbook: Structure and Applications of a Macro Model of the Global Economy for Policy Analysis, edited by Filippo di Mauro and M. Hashem Pesaran (2013), Oxford University Press, Oxford.

About the GVAR Handbook: The GVAR model was initially developed in the early 2000 by Professor Pesaran and co-authors, for the main purpose of analysing credit risk in a globalised economy. Starting from mid-2000 the model was substantially enlarged in the context of a project financed by the European Central Bank (ECB), to comprise all major economies and the Euro area as a whole. The purpose of this version was to exploit the rich modelisation of international linkages in order to simulate and analyse global macro scenarios of high policy interest. The rich, yet manageable, specification of international linkages has stimulated a vast literature on the GVAR. Since early 2011, the basic model - and its database - has also been publicly available. Moreover, the dedicated GVAR Toolbox webpage provides an easy-to-use interface allowing practical applications by an extended audience, as well as more complex analysis by the expert public. The book provides an overview of the extensions and applications of the GVAR which have been developed in recent years. Such applications are grouped in three main categories: 1) International transmission and forecasting; 2) Finance applications; and 3) Regional applications. By using a language which is accessible to not econometricians, the book reaches out to the extended audience of practitioners and policy makers interested in understanding channels and impacts of international linkages.

·         Theory and Practice of GVAR Modeling, Alexander Chudik and M. Hashem Pesaran (2016), Journal of Economic Surveys 30, pp. 165–197.

Abstract: The Global Vector Autoregressive (GVAR) approach has proven to be a very useful approach to analyse interactions in the global macroeconomy and other data networks where both the cross-section and the time dimensions are large. This paper surveys the latest developments in the GVAR modelling, examining both the theoretical foundations of the approach and its numerous empirical applications. We provide a synthesis of existing literature and highlight areas for future research.