Available online at www.sciencedirect.com ScienceDirect Procedia Economics and Finance 15 ( 2014 ) 1396 1403 Emerging Markets Queries in Finance and Business International crude oil futures and Romanian oil companies: volatility, correlations and causality Mihai-Cristian Dinic a, *, Erica Cristina Balea a a Bucharest University of Economic Studies, Bucharest, 010374, Romania Abstract The understanding of causal relationships between oil and stock markets is an important issue in portfolio management and energy hedging. The paper analyzes the causality between the Brent crude oil futures returns and the returns of the largest Romanian oil company, OMV Petrom (SNP). Using a GARCH model, we also examine the volatility spillovers between the two financial instruments. 2014 The Authors. 2013 Published by by Elsevier Elsevier B.V. Ltd. This Selectio is an open on access and/or article peer-review under the under CC BY-NC-ND responsibility license of Emerging (http://creativecommons.org/licenses/by-nc-nd/3.0/). Markets Queries in Finance and Business local organization Selection and peer-review under responsibility of the Emerging Markets Queries in Finance and Business local organization Keywords: volatility; causality; volatility spillovers; stock markets 1. Introduction One of the effects of globalization is the increasing financial integration between different economies and types of instruments or assets. Thus, the financial markets become more correlated and because of the high degree of correlation between the financial markets it appears the phenomenon of contagion and volatility spillovers. One of the most important commodities markets for the global economy is the crude oil, influencing the cost of production, transportation and heating. Therefore, a great number of studies discuss the impact of crude oil price fluctuations on inflation expectations, consumer confidence, spending or other macroeconomic variables: Sadorsky (1999), Hooker (2002), Hamilton and Herrera (2004), Cunado and Perez de Garcia (2005), Cologni and Manera (2008) and Park and Ratti (2008)). * Corresponding author. Tel.: +40 726 318 341; E-mail address: mihai.dinica@gmail.com 2212-5671 2014 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/3.0/). Selection and peer-review under responsibility of the Emerging Markets Queries in Finance and Business local organization doi:10.1016/s2212-5671(14)00604-2
Mihai-Cristian Dinică and Erica Cristina Balea / Procedia Economics and Finance 15 ( 2014 ) 1396 1403 1397 The links between the evolution of crude oil prices and stock markets are investigated in several papers. Boyer and Filion (2007) show that the returns of the Canadian oil and gas companies are positively related with the increases in oil prices. Nandha and Faff (2008) show that the rise in oil prices has a negative impact on stock returns from all industries, with the exception of oil and gas sector. Malik and Ewing (2009) found significant transmission of return and volatility shocks between oil prices and five stock sector indices from United States. Arouri and Nguyen (2010) emphasized that there are important benefits from diversification by adding crude oil to a portfolio of stocks. Arouri et al. (2011) examine the volatility spillovers between oil and European and U.S. stock markets. Their findings show that there are significant volatility cross effects between oil and different sector indices and in Europe the transmission is greater from oil to stocks. Our paper examines the links between the evolution of the Brent crude oil futures prices and the shares prices of the largest individual oil company listed on Bucharest Stock Exchange, Petrom. We find significant and positive correlations between the returns and the volatilities of the two financial instruments. Also, the Granger causality test suggests that oil returns affect the returns of the company and not vice versa. Another finding is that significant volatility spillovers exist between the oil market and Petrom, the high degree of integration in the financial markets fostering the contagion of volatility shocks. 2. Database and methodology The database used consists in daily spot prices of Brent crude oil futures and Petrom shares during the period 01.10.2007 28.09.2012. The futures price is given by the price of the nearby futures contract of the Brent crude oil traded on Intercontinental Exchange (ICE). Petrom (symbol SNP) is the largest oil company traded on the Bucharest Stock Exchange (BSE), with activities in exploration, production and oil refining. Given the activity of the company, it is expected that its shares prices to be correlated with the evolution of the international oil market. The first step of our methodology consists in analyzing the evolution of the two prices during the period. Also, we compute and discuss the following descriptive statistics for prices and daily returns: mean, median, maximum, minimum, skewness, kurtosis, standard deviation and annualized volatility. The daily returns are computed as logarithms of the ratio of two consecutive daily prices, and are given by the following formula: (1) The annualized volatility is computed by multiplying the standard deviation of the daily logarithmic returns with the squared root of the number of business days in a year. We consider 252 business days in a year. Because using non-stationary data in regressions can lead to spurious results, in order to check if the data series are characterized or not by unit root processes, we applied the Augmented Dickey-Fuller (ADF) test for price levels series and logarithmic returns. Following, we computed the Pearson correlation coefficients (noted ) between the returns of Brent (BRT) and Petrom (SNP) prices, given by: (2) In order to capture the dynamics of the correlation between the returns, we computed time varying correlation coefficients by two methods and analyzed their evolution. The first method consists in using an expanding period for calculation, by adding each new observation in the sample. The second method consists in using 3 months rolling calculation periods, by simultaneously adding each new observation in the sample and
1398 Mihai-Cristian Dinică and Erica Cristina Balea / Procedia Economics and Finance 15 ( 2014 ) 1396 1403 dropping the oldest one. After analyzing the correlation among the variables, it appears the question if causal relationships exist between them. And if these causal relationships exist, which variable causes the other. To answer these questions, we applied the Granger causality test with three lags for the logarithmic returns. The relationships used to perform the test are the following: (3) (4) Next, the focus of our analysis moved to the volatility of the two prices. In order to estimate the volatility and to give a greater weight to the recent periods, we used an autoregressive AR (1) GARCH (1,1) model. The mean and the variance equations are given by: (5) (6) Where r is the return of the price of the instrument (Petrom share or Brent futures contract), is the error and is the conditional variance at time t. The standard deviation (), that is the square root of the variance, represents the estimation of the volatility. After analyzing the evolution of volatility in time, we computed and discussed the evolution of the correlation between the volatilities of Brent and Petrom s prices. Using the same methodology like in the case of returns, we calculated expanding and rolling coefficients of correlation of the standard deviations ( ), given by: (7) The final step of the methodology consists in testing the existence of volatility spillover between the Brent market and the price of Petrom s shares. Initially, an autoregressive AR (1) model is estimated for the returns of the Brent s futures. (8) In order to test the volatility spillover, we estimate an AR (1) GARCH (1,1) model for returns of Petrom price shares. (9) (10) Where represents the contemporaneous innovations from the Brent market that affects the volatility of the Petrom price. There innovations are the squared errors estimated by the equation (8). The parameter characterizes the spillover effects and its significance would suggest that there exist volatility spillovers from the Brent crude oil market to the volatility of the price of Petrom shares.
Mihai-Cristian Dinică and Erica Cristina Balea / Procedia Economics and Finance 15 ( 2014 ) 1396 1403 1399 3. Results Figure 1 shows the evolution of the two prices, expressed in USD (per barrel, respectively per share) during the analyzed period. It can be seen that the price evolutions are strongly correlated, both instruments facing a sharp decline in price during the financial crisis and experiencing a recovery starting with the year 2009. Also, it can be noticed that after the financial crises, the prices had a higher correlation degree. Fig. 1. Price evolution for Brent and Petrom (SNP) In Table 1 are depicted the descriptive statistics of the prices and returns in the case of Brent and Petrom for the analyzed period. The mean and median values are closed in all cases. The extreme values of the prices suggest a high degree of variability in the price evolution. For instance, in the case of SNP, the maximum price reached 0.2416 USD/share, while the minimum was reached at 0.0339 USD/share, the ratio between them being 7.126. In the case of Brent, although the difference between the maximum and minimum price is very high, the ratio between them is lower than the one of SNP. Also, the extremes values of returns show a great volatility. The skewness and kurtosis statistics computed for the returns series evidence that the distribution is not normal, being asymmetric and leptokurtic. The volatility levels show that the Petrom s prices encounter a higher degree of variability than the prices of Brent crude oil. Table 1. Descriptive statistics SNP BRT Level Return Level Return Mean 0.1163-0.05% 91.9646 0.03% Median 0.1037 0.00% 92.6000 0.06% Maximum 0.2416 18.87% 146.0800 20.66% Minimum 0.0339-17.81% 39.5500-16.31% Skewness 0.9143-0.21-0.2387 0.12 Kurtosis 3.2279 8.36 2.2705 10.02 St. dev. 0.0451 3.16% 23.5327 2.49% Volatility 50.17% 39.59% In order to check if the data series are stationary, we applied the ADF test for price levels and logarithmic
1400 Mihai-Cristian Dinică and Erica Cristina Balea / Procedia Economics and Finance 15 ( 2014 ) 1396 1403 returns series. The results (described in Table 2) show that the prices are unit root processes, but the returns are stationary series. Further in the analysis, the returns series are used. Table 2. ADF test results t-stat p-value SNP Price -2.176 0.215 Return -33.591 0.000 Brent Price -1.444 0.562 Return -38.334 0.000 The evolution of the correlation between the daily returns of Brent futures and SNP shares is evidenced in Figures 2 and 3. The expanding correlation coefficient experiences an important increase in the first part of the sample, during the financial crisis, and converges after to the 0.3 level. The three months rolling correlation coefficient is characterized by a high variability during the analyzed period, ranging between a minimum of - 0.088 and a maximum of 0.562. It is important to notice that, with few exceptions, the time varying correlation coefficients had positive values during the entire period..4.6.5.3.2.1.4.3.2.1.0.0 -.1 EXPANDING_CORRELATION ROLLING_CORRELATION Fig. 2. Expanding correlation between returns of BRT and SNP Fig. 3. 3M rolling correlation between returns of BRT and SNP After analyzing the correlation among the variables, it appears the question if causal relationships exist between them. And if these causal relationships exist, which variable causes the other. The results of the Granger causality test, synthesized in Table 3, show that the returns of Brent futures prices Granger cause the returns of the Petrom shares prices. Also, the null hypotheses that the SNP returns do not Granger cause the returns of Brent cannot be rejected. The results are logical from an economical point of view, because the international crude oil market is much bigger than the market of an individual stock. Table 3. Granger causality test results Null hypothesis F-statistic Probability BRT return does not Grange cause SNP return 4.35686 0.00462
Mihai-Cristian Dinică and Erica Cristina Balea / Procedia Economics and Finance 15 ( 2014 ) 1396 1403 1401 SNP return does not Grange cause BRT return 1.60427 0.18662 We estimate the volatility through the conditional standard deviation from a AR(1) GARCH(1,1) model. The evolution of the Petrom and Brent volatilities is presented in Figures 4 and 5. It can be observed that both volatilities alternate periods of high and low variability. As expected, the highest volatility is observed in the first part of the sample, during the financial crisis. Also, it can be noticed that the volatility of Petrom is higher than the volatility of Brent..12.07.10.06.08.05.06.04.04.03.02.02.00.01 Conditional Standard Deviation Conditional Standard Deviation Fig. 4. Conditional standard deviation for Petrom Fig. 5. Conditional standard deviation for Brent Next, we computed the expanding and three months rolling correlation coefficients between Petrom and Brent conditional standard deviations. Their evolutions are evidenced in Figures 6 and 7. Like in the case of returns, the expanding correlation coefficient between volatilities experiences a sharp increase in the first part of the sample, during the financial crisis, and converges after to the 0.6 level. Also, it can be observed that the volatilities of Petrom and Brent are more correlated during the analyzed period than their returns. However, the rolling correlation coefficient exhibits a high variability, showing that on the short term, the volatilities can be positive or negative correlated. 1.0 0.8 0.6 0.4 0.2 0.0 1.0 0.5 0.0-0.5-0.2-0.4-1.0 EXPANDING_CORREL_STDEV ROLLING_CORREL_STDED
1402 Mihai-Cristian Dinică and Erica Cristina Balea / Procedia Economics and Finance 15 ( 2014 ) 1396 1403 Fig. 6. Expanding correlation between volatilities of BRT and SNP Fig. 7. 3M rolling correlation between BRT and SNP volatilities The final step of the analysis consists in testing the existence of volatility spillovers between the Brent market and the price of Petrom s shares. Initially, an autoregressive AR (1) model is estimated for the returns of the Brent s futures. The contemporaneous squared errors from the Brent AR(1) model are then used as exogenous variables in the AR(1) GARCH(1,1) model for SNP. The results are depicted in Table 4. The sum of the ARCH and GARCH coefficients is close to the unit value, suggesting that the volatility shocks are persistent for Petrom. Also, the coefficient is positive and strongly significant, showing that spillover exists between the international crude oil market and the volatility of Petrom s share price. Thus, the high degree of integration in the financial markets fosters the contagion of volatility shock in the commodity market (in this case, the crude oil market) to the evolution of individual shares prices of the oil companies. Table 4. Results of the volatility spillover test AR(1) for BRT AR(1) - GARCH(1,1) for SNP Coefficient Probability Coefficient Probability c 0.000327 0.6436 0.000579 0.4016-0.086694 0.0023 0.041793 0.1743 - - 0.000021 0.0005 - - 0.099234 0.0000 - - 0.841621 0.0000 - - 0.058499 0.0005 4. Conclusions The understanding of the relationships between crude oil and stock markets is an important issue in portfolio management and energy hedging. The paper analyzed the relationships between the evolution of Brent crude oil futures price and the price of the largest Romanian oil company s shares. During the period of five years analyzed, the price evolutions are strongly correlated, both instruments facing a sharp decline in price during the financial crisis and experiencing a recovery starting with the year 2009. Also, it can be noticed that after the financial crises, the prices had a higher correlation degree. With few exceptions, the time varying correlation coefficients between the daily returns had positive values during the entire period. The results of the Granger causality test show that the evolution of the Brent s price affects the returns of Petrom, but the opposite is not valid. The volatilities of Petrom and Brent prices alternate periods of high and low variability and Petrom prices exhibit a greater volatility. The correlation degree between the volatilities of the two instruments is higher than the one between daily returns. The results also show that between the international crude oil market and the volatility of Petrom s share price exist volatility spillovers. Thus, the high degree of integration in the financial markets fosters the contagion of volatility shocks, with important implications for portfolio management. Acknowledgments This work was co-financed from the European Social Fund through Sectorial Operational Programme Human Resources Development 2007-2013, project number POSDRU/107/1.5/S/77213, Ph.D. for a career in
Mihai-Cristian Dinică and Erica Cristina Balea / Procedia Economics and Finance 15 ( 2014 ) 1396 1403 1403 interdisciplinary economic research at the European standards. This work was financed through IOSUD ASE (Institution for Doctorate Studies, Academy of Economic Studies). References Arouri, M. E. H., Jouini, J., Nguyen, D. K., 2011. Volatility spillovers between oil prices and stock sector returns: Implications for portfolio management, Journal of International Money and Finance 30, p. 1387 1405. Arouri, M., Nguyen, D.K., 2010. Oil prices, stock markets and portfolio investment: evidence from sector analysis in Europe over the last decade, Energy Policy 38, p. 4528 4539. Boyer, M.M., Filion, D., 2007. Common and fundamental factors in stock returns of Canadian oil and gas companies, Energy Economics 29, p. 428 453. Cologni, A., Manera, M., 2008. Oil prices, inflation and interest rates in a structural cointegrated VAR model for the G-7 countries, Energy Economics 38, p. 856 888. Cunado, J., Perez de Garcia, F., 2005. Oil prices, economic activity and inflation: evidence for some Asian countries, The Quarterly Review of Economics and Finance, 45, p. 65 83. Hamilton, J.D., Herrera, A.M., 2004. Oil shocks and aggregate macroeconomic behavior: the role of monetary policy, Journal of Money, Credit and Banking 36, p. 265 286. Hooker, M.A., 2002. Are oil shocks inflationary? Asymmetric and nonlinear specifications versus changes in regime, Journal of Money, Credit and Banking 34, p. 540 561. Malik, F., Ewing, B.T., 2009. Volatility transmission between oil prices and equity sector returns. International Review of Financial Analysis 18, p. 95 100. Nandha, M., Faff, R., 2008. Does oil move equity prices? A global view. Energy Economics 30, p. 986 997. Park, J., Ratti, R., 2008. Oil price shocks and stock markets in the U.S. and 13 European countries, Energy Economics 30, p. 2587 2608. Sadorsky, P., 1999. Oil price shocks and stock market activity, Energy Economics 21, p. 449-469.