Regime-Dependent Effect of Crude Oil Price on BRICS Stock Markets

ABSTRACT

In this study, the dynamic relation between global crude oil prices and stock prices is investigated in terms of crude oil-exporting and -importing countries. The relationship between crude oil prices and stock prices is examined for BRICS countries (Brazil, Russia, India, China, and South Africa) for the periods of January 1995 to December 2016 by means of the Markov Switching Vector Autoregression (MS-VAR) model. The impulse-response analysis results suggest that the responses of the stock market to an oil price shock vary over the regimes for all countries. Specifically, we find that the responses of the stock market to an unexpected oil price shock are positive and statistically significant in the high-volatility regime in all countries except for China, and these results suggest that the increase in oil prices may be evaluated by demand-side shock in these countries.

KEY WORDS:

• MS-VAR
• oil price
• stock returns

Supplemental Material

Supplemental data for this article can be access on the publisher’s website.

Notes

1. See the Supplementary Material, available online.

2. The global oil price represents the West Texas Intermediate crude oil price. Real oil price is calculated by using the US Consumer Price Index.

3. See Pierce and Enzler (1974), Rasche and Tatom (1977), Mork and Hall (1980), Darby (1982), Brown and Yücel (2002), and Lardic and Mignon (2006).

4. Although Ciner (2001) used the nonlinear causality test, the test method is based on estimation of the linear VAR model.

5. The grid setting is: p₂₂ = 0.4, 0.5, 0.6, 0.7, 0.8, 0.9; μ₂ = −0.3, −0.2, −0.1, 0, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6; σ₂ = 1.50, 1.65, 1.80, 1.95, 2.10, 2.25, 2.40, 2.55, 2.70, 2.85 and it has 3600 points.

6. It should be noted that WTI oil price has been widely used in the literature as global crude oil price, and Wang, Wu, and Yang (2013) indicated that it is highly correlated with other crude oil prices such as Brent and Dubai.

7. See the Supplementary Material, available online.

8. Additionally, whether the MS-VAR model yields better results in representing the data than the linear VAR model was researched using the LR test. Both the LR test statistics and Davies (1987) upper-bond p-value indicate that the MS-VAR model yields better results than the linear model.

9. Because the aim of the study is to examine the dynamic relation between stock markets and oil prices for the BRICS countries, the causality test for the US stock market and the impulse-response analysis results are not reported.

10. The impulse-response analysis based on Cholesky decomposition is very sensitive to the order of the variables. Therefore, while carrying out impulse-response analysis, the variables are ordered as oil price, the US stock returns, and stock returns of the BRICS countries in the MS-VAR model. To test the robustness of the results, impulse-response analyses are employed by changing the order of the variables, and the results showed no significant difference.