Geopolitical risk shocks and the Brazilian economy

ABSTRACT

This paper investigates the impacts of international geopolitical risk shocks (GPR) on the Brazilian economy. The dynamic relation between Brazil and GPR is modelled using a Bayesian vector autoregressive model that accounts for large economic shocks with heavy tails and stochastic volatility. The study analyses quarterly data from Brazil, as well as the indices that account for GPR related to the US, Russia, and a global GPR index. The results indicate that real economic activity reacts more intensely to Russian risk than to global or US risk, whereas the responses of prices and financial indicators appear more sensitive to global risk.

KEYWORDS:

• Geopolitical risk shocks
• bayesian VAR
• stochastic volatility
• heavy tails
• Brazilian economy

JEL CLASSIFICATION:

• C11
• C32
• E32
• F51

Disclosure Statement

No potential conflict of interest was reported by the author(s).

Notes

¹ The literature has also documented the effects of international shocks in a general context. For example: impacts on domestic prices (Amiti, Itskhoki, and Konings 2019), transmission mechanisms (Mumtaz and Surico 2009), and the effects of international supply and demand shocks (Charnavoki and Dolado 2014; Feldkircher and Huber 2016). This study contributes to that literature by investigating the effects of international GPR shocks on a small open economy.

² We use the ‘Recent’ indices in all cases. See Caldara and Iacoviello (2022) for details on the construction of such indices.

³ This is defined as the log deviation from a trend, computed using the HP filter.

⁴ Sources are presented in the Online Appendix A. The Online Appendix B presents the descriptive statistics.

⁵ Online Appendix C presents the estimated stochastic volatility and the implied ${\lambda }_{i,t}$. There is evidence of time volatility and large shocks, features that support our chosen model.

⁶ In addition to an exercise with a VAR with more lags, Online Appendix D presents evidence that the results obtained here for the real activity are robust in a VAR with fewer variables and with a different ordering scheme.

Additional information

Funding

This work was partially supported by the National Council For Scientific and Technological Development - CNPq (Brazil).