Re-Investigating the degree of persistence of U.S. economic policy uncertainty using the Fourier non-linear quantile unit root test

ABSTRACT

This research tests the mean reversion properties of the U.S. EPU (economic policy uncertainty) index and its 11 sub-categories using the Fourier non-linear quantile unit root test over the period 1985M1-2020M12. The results indicate the following. (1) The U.S. EPU index responds asymmetrically to shocks, in which positive shocks are more long-lasting than negative shocks. (2) Monetary, fiscal, and trade policies and also national security and regulation series behave like unit root processes in all quartiles, whereas by contrast financial regulation and sovereign debt exhibit stationary behaviour over all quantiles. (3) Government spending, health care, national security, and entitlement programmes present stationary processes in low/high quantiles.

KEYWORDS:

• Economic policy uncertainty
• unit root
• quantile regression
• Fourier expansion

JEL CLASSIFICATION:

• C22
• E60
• D81

Disclosure statement

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

Notes

1 For more details, please see Baker, Bloom, and Davis (2016).

2 In addition to empirical findings for the U.S. economy, Wang and Lee (2022) found China’s policy uncertainty is transmitted to domestic and international crude oil markets, and the effects on real crude oil returns are time-varying and non-linear.

3 Our result about the stationary behavior of the U.S. EPU index is in line with the findings of Gil-Alana and Payne (2020), Solarin and Gil-Alana (2021), and Solarin and Stewart (2021). Other results herein are not in line with findings of Gil-Alana and Payne (2020), which is the only empirical study on stationary properties of the U.S. EPU series and found shocks to 11 U.S. sub-categories are persistent, but mean-reverting.

4 Lee, Wang, and Pan (2022) found that promoting financial inclusion and research and development activities could offset the harmful effects of uncertainty on economic performance, where the effect of research and development is statistically significantly greater than the impact of financial inclusion.