A GVAR Model with Different Weights for Analysing the Financial Channel of Real Shock Propagation

Abstract

External shocks affect economies through a variety of transmission channels. A widely used approach for assessing macroeconomic responses to such shocks is the global vector autoregressive (GVAR) model. In most applications, dimensionality is reduced through aggregation techniques, typically relying on foreign trade weights. In some cases, capital flows are used instead to capture financial linkages. This paper develops a GVAR model that jointly incorporates real and financial sectors by employing two types of weights: those based on foreign trade and those based on foreign direct investment (FDI) flows. We compare the results from this extended specification with those from a baseline model that includes only the real sector and trade-based weights. Impulse response functions of key macroeconomic indicators to a China output shock and an oil price shock are analysed. The results show that incorporating the financial sector and FDI-based weights leads to stronger and more pronounced responses in the short to medium term. The model produces dome-shaped impulse responses, consistent with findings in the empirical literature. However, differences between the two model specifications diminish over the long run, suggesting that the financial channel primarily influences short- and medium-term adjustment dynamics.