Oil Market Dynamics and US Monetary Policy Uncertainty: Evidence from Explainable Artificial Intelligence Models

Abstract

We utilize explainable artificial intelligence techniques to examine the main link between oil market dynamics and US monetary policy uncertainty. Our analysis unveils key findings. First, higher oil prices and rising oil market uncertainty are significantly associated with elevated levels of monetary policy uncertainty. Second, option-implied oil market volatility (an indicator of oil market uncertainty) emerges as the most important predictor of monetary policy uncertainty. Third, option-implied oil market volatility frequently interacts with option-implied gold market volatility while oil prices mostly interact with fluctuations in the US dollar. The findings offer valuable insights into the understanding of monetary policy uncertainty.

Keywords:

• Explainable artificial intelligence models
• Macroeconomic risks and uncertainties
• Monetary policy uncertainty
• Oil market uncertainty
• Oil prices

JEL CLASSIFICATION:

• C60
• E52
• E58
• Q43

Data availability

The data that support the findings of this study are available from the corresponding author upon request.

Disclosure statement

The authors have no conflict of interest to declare.

Notes

1 There are other monetary policy uncertainty indices, including monthly data, which can be found at https://www.policyuncertainty.com/monetary.html. However, in this work, we prefer to use the MOVE index because we utilize daily data to better capture immediate risk and uncertainty transmissions.

2 Implied volatility represents the 30-day option-implied volatility in the relevant market.

3 To maintain consistency, US data is utilized for the analyses.

4 The descriptive statistics of the data suggest that the variables do not follow a normal distribution. The findings are available upon request.

5 Tuned hyperparameters and optimal values for machine learning models are available upon request.

6 For the sake of brevity, we avoid detailed technical explanations. For a comprehensive exposition of the methodological procedures, please refer to Breiman (2001), Chen and Guestrin (2016), Ke et al. (2017), Prokhorenkova et al. (2018), and Lundberg, Erion, and Lee (2018).

7 To ensure robustness, we made changes to the training and test datasets. We used the CatBoost model (the best-chosen model) and performed an analysis with an 80% training and 20% testing data split, yielding highly similar results. The findings are available upon request.

8 To improve accuracy, we incorporate further specifications for both the mean and variance equations. The primary outcomes remain unchanged across the different specifications. Diagnostic checks, including ARCH LM test statistics, squared residuals, and Ljung-Box Q statistics with various lag specifications, indicate that our model is correctly specified. To save space, EGARCH estimation results are provided upon request.

9 We included the well-known geopolitical risk index (GPR), introduced by Caldara and Iacoviello (2022), in our analysis for a robustness check. According to the results of this analysis, our findings remain robust, as we obtained very similar results. The findings are available upon request.

10 The steeper slope observed in the line regarding the impact of CVOILF in Figure 6 can be attributed to the descriptive statistics of CVOILF. The mean, minimum, and maximum values of CVOILF are 3.03, 1.04, and 324.95, respectively. All descriptive statistics are available upon request.