Spillover effects of US uncertainty: does the type of US uncertainty matter?

ABSTRACT

This study investigates the spillover effects of US uncertainty shocks on 8 emerging market (EM) economies using a two-country VAR model. The study considers four different types of US uncertainty, namely, (i) financial, (ii) macroeconomic, (iii) economic policy and (iv) aggregate uncertainty. Then, the study compares the spillover effects of four US uncertainty measures on EMs. The main findings are as follows: (1) US uncertainty shocks negatively affect EMs. (2) The spillover effects rely on the source of US uncertainty and vary across EMs. (3) EM financial conditions play a significant role in transmitting the effects of US uncertainty shocks to the real economies of EMs. These findings suggest that EM policy-makers should be aware of the sources of US uncertainty shocks and react accordingly to such shocks.

KEYWORDS:

• US uncertainty
• emerging markets
• spillovers
• two-country VAR
• financial conditions

JEL CLASSIFICATION:

• C32
• E32
• E44
• F41

Acknowledgments

I thank the anonymous referees for their constructive comments. I also thank Anadolu University Scientific Research Projects Commission for the financial support. All remaining errors are my own responsibility.

Disclosure statement

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

Notes

¹ A different line of the literature on the role of US examines the spillover effects of US monetary policy on emerging and advanced economies (see, among others, Canova 2005; Kim 2001; Maćkowiak 2007).

² Kang, Ratti, and Vespignani (2020) compare the effects of global and US uncertainty shocks on the global economy.

³ From another aspects, Drobetz et al. (2021) and Liang et al. (2020) analyse the impact of US economic policy uncertainty on shipping freight rates and oil market volatility, respectively.

⁴ From another aspects, Belke and Osowski (2019) compare the spillover effects of US and Euro area uncertainty shocks on 18 countries, while Dogah (2021) compares the effects of US and China policy uncertainty shocks on ASEAN markets. In addition, Balcilar et al. (2021) identify the role of global, regional, and advanced market EPU (i.e. US, UK and Japan) on EM bond markets.

⁵ A recent growing literature examines the drivers of EM heterogenous responses to external shocks (see, for instance, Ahmed, Akıncı, and Queralto 2021; Ahmed, Coulibaly, and Zlate 2017; Mishra, N’Diaye, and Nguyen 2018; Yildirim 2016). Please see this literature for further discussions about the role of country characteristics in the heterogenous spillover effects.

⁶ Two recent studies by Yildirim and Ivrendi (2021) and Carrillo, Elizondo, and Hernández-Román (2020) employ a similar VAR model to examine the international spillover effects of shocks originating from US economy. Please see these studies for further information about the aspects of the model.

⁷ Preliminary tests in the Appendix (see Tables A1 and A2) clearly suggest that the VAR model specified in levels should be more appropriate. Because of the cointegration between the variables (see Table A2), using the first differences of variables in the VAR system may result in misspecification.

⁸ The AIC selects 4 lags for most countries in the EM sample, except for China ($l^{CH}=3$), Mexico ($l^{MX}=5$) and Brazil ($l^{BR}=5$).

⁹ This counterfactual analysis requires the additional zero restrictions on the system, resulting in an overidentified SVAR system. Thus, to perform the counterfactual analysis, we estimate an overidentified system through a two-step process. In the first step, as in the section 2.6, Equation (2)(2) $Z_t=\sum _{l=1}^p{\tilde{B}}_l{Z}_{t-l}+e_t$(2) is estimated via the SUR methods. In the second step, the impulse response functions, along with the posterior probability bands, are constructed by using the Broyden-Fletcher-Goldfarb-Sarno (the BFGS) algorithm and a Metropolis-within-Gibbs MCMC procedure. This MCMC procedure uses a combination of the Gibbs sampler and Metropolis – Hastings algorithm to draw from the posterior distribution of the structural parameters and the lag coefficients (see, Doan (2015)).

Additional information

Funding

This study was supported by Anadolu University Scientific Research Projects Commission (Grant Number: 2107E057)