Do Energy Prices always Affect EU Allowances? Evidence Following the Copenhagen Summit 1

Transcription

1 Journal of Contemporary Management Submitted on 29/04/2015 Article ID: Xin Lv, Weijia Dong, and Qian Chen Do Energy Prices always Affect EU Allowances? Evidence Following the Copenhagen Summit 1 Dr. Xin Lv Faculty of School of Management and Economics, Center for Energy and Environmental Policy Research, Beijing Institute of Technology 5 South Zhongguancun Street, Haidian District, Beijing, , CHINA lvxin2311@bit.edu.cn Dr. Weijia Dong (Corresponding author) Faculty of Graduate School of Economics, Nagoya University Furo-cho, Chikusa-ku, Nagoya, , JAPAN dongweijia0316@163.com Dr. Qian Chen Faculty of School of Public Finance and Public Policy, Central University of Finance and Economics 39 South College Road, Haidian District, Beijing, , CHINA water111716@hotmail.com Abstract: This paper re-examines the impact of energy prices (oil and gas prices) on the EU Emission Trading System (ETS) market and some heteroscedastic characteristics of the EU ETS after the 2009 Copenhagen Summit. We aim to explain how the failure of the 2009 Copenhagen Summit caused the change in the EU ETS and its driving factors. Utilizing the Markov Switching model and GJR-GARCH model, we find that the oil prices no longer affected the EU allowance (EUA) price, and the impact of gas price on the EUA price became relatively smaller after the 2009 Copenhagen Summit. In addition, the relationship between energy prices and EU ETS market risk was investigated innovatively, and we find that oil prices could affect the EU EST market risk linearly and asymmetrically. Furthermore, the empirical results show that the EU ETS market still retains some basic financial asset market characteristics; on the other hand, our findings also indicate the EU ETS market s uncertainty and riskiness increased after the Copenhagen Summit. Finally, a non-linear relationship between gas prices and EUA prices was found under different market regimes. Keywords: Energy price; EUA price; Markov Switching model; EU ETS market risk JEL Classifications: C58, G12, Q54 1 This research is funded by the Beijing Institute of Technology Fund Program for Yong Scholars, No ~ 13 ~

2 ISSN: (Print); (Online) Academic Research Centre of Canada 1. Introduction The EU Emission Trading Scheme (ETS) market, the most successful cap and trade market, is the biggest carbon trading market with a share of nearly 80% of the entire carbon market. Because the emission allowance permits the emission of one ton of CO 2 to be traded in the market, and not only by emission regulated firms, but financial sectors are also allowed to buy and sell such allowances; thus, it always trades as a new financial asset. Therefore, the existing literature attempts to prove the EU ETS market demonstrates similar characteristics with other financial markets. Abundant literature finds that the EU ETS market price also demonstrates financial asset characteristics of heteroskedastic dynamics and changes in the volatility of the underlying stochastic price process. (Paolella & Taschini, 2008; Benz & Trück, 2009; Daskalakis et al., 2009; Chevallier, 2010). While other articles demonstrate that the EU allowance (EUA) price is driven by some fundamental factors (Mansane-Bataller et al., 2007; Convery & Redmond, 2007; Fehr & Hinz, 2006; Alberola et al., 2008; Hintermann, 2009; Zhang & Wei, 2010; Bredin & Muckley, 2011; Reboredo, 2013; Lutz et al., 2013; Hammoudeh et al., 2014a; Hammoudeh et al., 2014b; Koch et al., 2014) 2. Moreover, the most cited variable among these fundamental factors for EUA pricing is the energy price, especially the oil price (Mansane-Bataller et al., 2006; Kanen, 2006; Convery & Redmond, 2007; Reboredo, 2013; Lutz et al., 2013; Hammoudeh et al., 2014a; Hammoudeh et al., 2014b) and the natural gas price (Mansane-Bataller et al., 2006; Alberola et al., 2008; Hintermann, 2009; Lutz et al., 2013; Hammoudeh et al., 2014a; Hammoudeh et al., 2014b), finding that oil prices and gas prices have significant impacts on the EUA price. Unfortunately, rarely does research notice the crucial feature of learning by doing for the EU ETS market, which means that the EU ETS mechanism dynamically changes as an emerging financial market. In recent years, the foundation of the EU ETS market has been gradually changed after the failure of 2009 United Nations Climate Change Conference (known as the Copenhagen Summit), which means there are no legal treaty regulates global CO 2 emissions after the expiration of the Kyoto Protocol. This fact has changed the foundation of the EU ETS market in at least two ways. First, some countries that no longer commit themselves to CO 2 emission abatement targets will meet lower emission costs than the other countries, and the intrinsic value of the EUA is certainly changed. Second, the failure of the Copenhagen Summit brings more uncertainty to the EU ETS market, and it changes the expectations of market investors. The changes of investor sentiment will always cause processes of financial asset (EUA) pricing changes (Baker and Wurgler, 2006; 2007). Thus, it is worth studying the behavior of the EU ETS market following the Copenhagen Summit. However, most of the previous literature (Bredin & Muckley, 2011; Reboredo, 2013; Lutz et al., 2013; Hammoudeh et al., 2014a; Hammoudeh et al., 2014b) ignores this potential market structure change in the EU ETS market. Therefore, the main purpose of this paper is to re-examine the financial asset characteristics (heteroskedastic dynamics) of the EU ETS market, and the relationship between the EUA price and energy prices after the Copenhagen Summit. The Markov Switching model (Hamilton, 1989; 1994) and GJR-GARCH model (Glosten et al., 1993) help us to achieve the objective. First, we choose a two-regime Markov Switching model suggested by Benz & Trück (2009) and re-examine whether the different regimes of the EU ETS always appeared as a high-mean, low variance state (bull market) or a low-mean, high variance state (bear market). Second, the GJR-GARCH model is employed to re-examine EUA price dynamics and investigate both the liner and asymmetric impacts of oil and gas prices on the EUA price. In addition, the GJR-GARCH model also helps us explain how changes in energy prices affect EU ETS market risk 2 Based on the above literature, the fundamentals of EUA pricing include oil prices, gas prices, coal prices, electricity prices, extreme temperatures, macro-economy and some other factors. ~ 14 ~

3 Journal of Contemporary Management, Vol. 4, No. 3 measured by the volatility of allowance return. Overall, this paper makes four major contributions to the current literature. First, to our knowledge, this study is the initial work that considers potential EU ETS market changes after the 2009 Copenhagen Summit and innovatively re-examines EU ETS market empirical experiences in both the financial asset aspect and the relationship with driving energy factors. We find oil prices have no longer affected the EUA price since 2009 and the natural gas price has had a positive impact on the EUA price, but the effect is very minor. Second, this paper innovatively investigates whether energy price changes affect EU ETS market risk measured by conditional heteroskedastic variance. Our empirical results demonstrate that oil price volatility will negatively affect EU ETS market risk, but the relationship between gas prices and EU ETS market risk does not show any statistically significant characteristics. At the same time, we also find asymmetric effects of oil prices on EU ETS market risk. This result demonstrates that bad news can always lead to larger volatility in the EUA price than can good new, which benefits investors risk management behavior. Third, we employ the Markov Switching model (Hamilton, 1989; 1994) to re-examine the risk-return relation of the EU ETS market under different market regimes. The Markov Switching model divides the EU ETS market since the Copenhagen Summit into bull markets and bear markets, defined as high means-low variance and low means-high variance consistent with Benz & Trück (2009). However, the bull market return demonstrates negative value, which indicates that market collapses are due to uncertainty and risky situations, and this finding is quite different from other financial markets as in Benz & Trück s (2009) conclusion. Finally, we find new evidence of asymmetrical impacts from oil price changes on the EU ETS market. The findings show non-regular asymmetric effects of gas prices on the EUA price. The paper is structured as follows: Section 2 demonstrates Empirical methodology including the Markov Switching model and the GARCH-type model. Data and empirical results are presented and discussed in Section 3 and Section 4, respectively. Finally, concluding remarks are provided in Section Econometric Methodology 2.1 Markov Switching Model Structural changes always occur in the financial market, and Hamilton (1989, 1994) introduced the Markov Switching model to capture these structural changes in stock returns. In this paper, following the idea of Benz & Trück (2009) who specified there are two regimes in the EU ETS market, the EUA price or returns are assumed to display either low mean-high volatility (a bear market) or high mean-low volatility (a bull market) in each time t. The regime variable is designed to be an unobserved latent variable that governs the switching probability between regimes. Furthermore, we assume no AR lag in R t in the Markov model, consistent with Benz & Trück (2009). Then, we can specify the Markov Switching model as follows: in which denotes the market return, represents the different market regimes and is assumed to take the value of 1 or 2; and represent the state-dependent mean and variance of ; innovatively follows the GED distribution suggested by Hamilton (1994). The parameter is a regime-dependent parameter. Thus, the model can be written separately: ~ 15 ~ (1) (2)

4 ISSN: (Print); (Online) Academic Research Centre of Canada According to the property of the Markov-chain, the current regime past through the most recent regime value: (3) depends only on the PROB = (4) where is defined as the probability of switching from regime i in time t-1 to j in time t, and in this paper it is assumed that i, j= (1, 2). Then, the transition matrix for takes the form as follows: =, (5), (6). (7) Because the density of is designed to be conditional on its own lagged values and the current and previous p values of the regime, Hamilton suggested employing the maximum likelihood estimation method to obtain inferences on the unobserved regime variables. Thus, based on Hamilton s calculation method, we can obtain the smoothed probability for each state as follows: ; (8). (9) In this paper, this filter probability serves as the reference for market classification. For the EU ETS market regime, it is simply decided by comparing the value of filter probability. For example, the market regime is specified to be Bull Market means P 1 is bigger than P 2 in this period. 2.2 GJR-GARCH Model In this paper, GJR-GARCH (Glosten et al., 1993) models are modified and applied to the study of the EU ETS market. The reason to apply GJR-GARCH model is that Engle & Ng (1993) proved that the GJR-GARCH model is the best GARCH-type model in capturing the characteristic of the asymmetry effect and estimating the parameters. For the analysis of the relationship between energy prices and allowance prices, we introduce the energy price (oil or gas price) and market regime dummy variable into the model. The modified GJR-GARCH mean model is derived as follows: (10) (11) ; (12) In the new models, R represents the return of allowance traded in the EUETS market; while EP denotes the oil price or natural gas price, while q represents the lagged period 3 of impact caused by the energy price; ER denotes the return of oil prices or natural gas prices. is a dummy 3 Through trial operation of the models, it is clearly found that the impact energy price casts on allowance never appears immediately but a few days later. In this paper, a week s lagging will be calculated with q=3. ~ 16 ~

5 Journal of Contemporary Management, Vol. 4, No. 3 variable that equals 1 if the EU ETS market is a bull market 4, otherwise, equals 0. In this modified model, we add energy variable and dummy variable. Based on the modified GJR-GARCH model, the coefficients and measure the impact of energy price changes on the EU EST market and asymmetric relationship between them in bull or bear markets, respectively. Furthermore, the effects of energy price changes on EU ETS market risk (measured by variance) is depicted by coefficient. Other estimators further describe some characteristics of heteroskedastic dynamics of the EU ETS market. When we estimate the model, we assume the error term follows the GED distribution, consistent with the Markov Switching model. 3. Data In this empirical analysis, we follow the idea of Rittler (2012) and Reboredo (2013) to employ daily EUA future prices and energy future prices. Before the 2009 Copenhagen Summit, there were international negotiations four times in that year, and the new draft agreement was reached in the second international negotiation during June 1 st to 12 th. In addition, this new draft agreement was the basis of the Copenhagen Summit. Therefore, our calendar commenced on June 1 st, 2009 and ended on April 14 th, For the energy market, we utilize the daily Brent oil futures price as the oil price ($/barrel) and the daily NYMEX natural gas price ($/million Btu) as the natural gas price. The two categories of energy will be utilized in our econometric models. EUA future data are obtained from the website of the Intercontinental Exchange (NYSE: ICE) 5. The energy price data are collected from the US Energy Information Agency database. 4. Empirical Results 4.1 Descriptive Statistics and Unit Root Test Table 1 below presents the descriptive statistics of the observed data. OILRETURN and GASRETURN denote the changing rates of the prices of these two types of energy. The available data for analysis number approximately 1,215 in each category. Table 1. Descriptive Statistics Statistic EUARETURN OILRETRUN GASRETURN Mean Median Maximum Minimum Std. Dev Skewness Kurtosis Observations Market regime (Bull or Bear) was selected by the Markov Regime model. 5 Data was downloaded from the website ~ 17 ~

6 ISSN: (Print); (Online) Academic Research Centre of Canada In this paper, the Augmented Dickey-Fuller unit root test (1979) and Phillips-Perron unit root test (1988) are applied to test the change rate of the EUA price, oil price return and natural gas price return data. According to the results presented in Table 2, the ADF statistic and PP statistic are significant at the 1% level, indicating the null hypothesis that assumes there is unit root should be rejected. Hence, the carbon allowance return data and changing rate of energy prices can be employed in model estimation. Table 2. Result of Unit Root Test Test EUA RETURN OIL RETURN GAS RETURN Augmented Dickey-Fuller (P-statistic) *** (0.0000) *** (0.0000) *** (0.0000) Phillips-Perron (P-statistic) *** (0.0000) *** (0.0000) *** (0.0000)) Notes: All price data has been applied with a first order operation. The upper term in the grid is t-statistic while the p value is written in parentheses. The marks *, **and*** represent the 10%, 5% and 1%level of significance, respectively. 4.2 Results of Markov Switching Model By estimating the two-regime Markov Switching model introduced in sub-section 2.1, the stage-dependent means and, variance of allowance return, and the filter probability of each regime for the EU ETS market are generated. The results of the Markov Switching model are listed in Table 3, Figure 1 and Figure 2. In Table 3, the estimated stage-dependent means of bull and bear markets are and , respectively, while the corresponding variances are calibrated as and This is co-inherent with the assumed properties of the bull market and bear market, defined as high means-low variance and low means-high variance (Benz & Trück, 2009). However, our results indicate that the bull market does not demonstrate positive returns after the 2009 Copenhagen Summit, which is quite different from Benz & Trück s (2009) finding. The negative value of EU ETS market return in both bull and bear markets implies that the market declines because of the uncertain and risky situation after the 2009 Copenhagen Summit. This result also demonstrates the EU ETS market has changed enormously after the Copenhagen Summit. Therefore, it is necessary for us to analyze the current property of the EU ETS market further. It is obvious from Figures 1 and 2 that P1-the filter probability of a bull market- always increases accompanied with higher return mean and lower variance, while the opposite situation can be observed in the graph of P2, which denotes the filter probability of a bear market. In addition, the number of switching filter probability P11 and P22 in Table 3 are much larger (P11=0.90 and P22=0.87), which indicates if the EUA state is bull (or bear) market in time t and in time t+1, the bull (bear) market occurs with a probability of 90% (or 87%). This result demonstrates that both the bull and bear market state will persist for quite a long time. Then, the two markets could be divided into different regimes, either bull market or bear market, along the time axis. According to Hamilton s method, the market division can be achieved by comparing the filter probability of each regime, and if the filter probability P1 is greater than P2 at time t, then time t is defined as a bull market, otherwise, a bear market. For example, in the case of EUA price data, because the probability of a bull market (P1) on June 1 st, 2009 is 57.41%, which ~ 18 ~

7 Journal of Contemporary Management, Vol. 4, No. 3 is larger than the 42.59% of a bear market (P2), then the period can be considered to be a bull market. We adopt this result to define the dummy variable of bull and bear markets, and it will be employed in the GJR-GARCH model to check the asymmetric effects of energy prices on the EU ETS market. This result will be discussed in the next sub-section. Table 3. Estimated Results of Markov Switching Model for EUA AIC LogLik *** ** 0.90*** 0.87*** (0.2782) (0.73) (2.1195) (5.9180) Notes: the number in parentheses is the standard error; *,** and *** indicate 10%, 5% and 1% levels of significance, respectively /6/1 2010/6/1 2011/6/1 2012/6/1 2013/6/1 EUA Price Figure 1. EUA Prices during June 1st, 2009 and April 14th, Results of GJR-GARCH The results of the GJR-GARCH modifying the empirical relationship of energy prices and the EUA price are listed in Table 4 and Table 5. In this article, the GJR-GARCH generated the results that the relationship between energy prices and allowance prices were successfully demonstrated, though there are some lags, usually approximately 3 days before such a short-lasted relationship can be observed. The details about the relationship between each energy and carbon market under different regimes are specified as follows. In the EU ETS market, the effect of oil prices on EUA returns is not insignificant in any market regime division model shown by in Table 4 column 1-3. This result is quite different from the previous literature (Mansane-Bataller et al., 2006; Kanen. 2006; Convery & Redmond, 2007; Reboredo, 2013). Therefore, this study implies that the oil price has no longer been a driving factor of the EUA price after the Copenhagen Summit. Moreover, when we test the asymmetric effect of oil prices on the EUA market, we find that the value of coefficient is also insignificant in Table 4 column 4-6, which means the change in oil prices could not cause different changes in EUA return in either a bull market or bear market. Therefore, we find that oil prices have no impact or an asymmetric impact on the EU ETS market after the Copenhagen Summit. ~ 19 ~

8 ISSN: (Print); (Online) Academic Research Centre of Canada /01/09 10/01/09 02/01/10 06/01/10 10/01/10 02/01/11 06/01/11 10/01/11 P1 Figure 2(a) Filter Probability of Bull and Bear Markets Bull Market 02/01/12 06/01/12 10/01/12 02/01/13 06/01/13 10/01/13 02/01/ /01/09 10/01/09 02/01/10 06/01/10 10/01/10 02/01/11 06/01/11 10/01/11 P2 Figure 2(b) Filter Probability of Bull and Bear Markets Bear Market 02/01/12 06/01/12 10/01/12 02/01/13 06/01/13 10/01/13 02/01/14 However, when we examine the effect of oil prices on EUA market risk, the parameter is estimated to be negatively significant in the first day after oil price changes. That is to say, a decrease (increase) in oil prices always results in an increase (decrease) in market risk (or volatility). Moreover, referring to the property of the EU ETS price heteroskedastic dynamics characteristics, the GARCH-mean effects shown by parameter are not reflected on the EU ETS market, but a general GARCH property such as conditional heteroskedasticity ( and ) is found. Furthermore, the result that indicates asymmetry in the response of volatility to oil price changes, which means that bad news can always cause larger volatility in the EUA price than good news. These two results of the EUA financial market characteristics demonstrate that the EUA market price volatility also presents heteroskedastic characteristics after the Copenhagen Summit similar to previous findings (Paolella & Taschini, 2008; Benz & Trück, 2009; Daskalakis et al., 2009; Chevallier, 2010). ~ 20 ~

9 Journal of Contemporary Management, Vol. 4, No. 3 In the second step of this section, we test the effect of gas prices on EUA returns. The second day s lagging effect of gas prices on the EU ETS return is discovered, noting the positive values of in Table 5 column 5.This result implies that the gas market also has an impact on the EU ETS market; however, this relationship became relatively smaller after the Copenhagen Summit because the first and third days lagging effects of gas prices no longer affects the EU EST market. In addition, we find the asymmetric effects of gas prices on the EU EST market, but this asymmetric impact of gas prices on the EU ETS market under different market regimes (bull or bear market) is not regular. For example, the impact of gas prices on the EU ETS in a bull market is positive ( in Table 5colume 4) in the first day s lag but negative in the third day s lag in Table 5 column 6). That means the change in gas prices will first cause larger positive returns one day later, but larger losses three days later, a bear market rather than a bull market. These results may be caused by irrational investors in a bear market. Bear Market Effect GED PARAMET Akaike ER info criterion Log likelihood Table 4. Impacts of Oil Price on EU ETS (GJR-GARCH Model) OIL (No Markov Regime Division) OIL (Markov Regime Division) OIL(-1) OIL(-2) OIL(-3) OIL(-1) OIL(-2) OIL(-3) (0.0652) (0.0092) (0.0281) *** (0.0357) *** (0.0226) *** (0.0369) *** (0.0148) ** (0.0654) (0.0662) (0.0091) (0.0296) *** (0.0369) *** (0.0229) ()0.1194** *** (0.0383) *** (0.0152) (0.0595) (0.0657) (0.0091) (0.0295) *** (0.0370) *** (0.0231) *** (0.0383) *** (0.0152) (0.0613) (0.0653) (0.0092) (0.0283) (0.2630) *** (0.0361) *** (0.0228) *** (0.0375) *** (0.1570) ** (0.0652) (0.2613) (0.0662) (0.0091) (0.0298) (0.2628) *** (0.0369) *** (0.0233) *** (0.0383) *** (0.0156) (0.0595) (0.2624) (0.0658) (0.0091) (0.0297) (0.3038) *** (0.0372) *** (0.0235) *** (0.0381) *** (0.0153) (0.0615) (0.3026) Notes: Numbers in parentheses are the standard errors of the corresponding estimated parameters; *,** and *** indicate 10%, 5% and 1% levels of significance, respectively. ~ 21 ~

10 ISSN: (Print); (Online) Academic Research Centre of Canada Furthermore, the natural gas price could not affect the risk of the EU ETS, as proven by insignificant coefficient in Table 5. For the GARCH-type property, the relationship between natural gas and the EU ETS market is quite similar with oil s: no GARCH-mean effect, significant conditional heteroskedasticity ( and ) and asymmetric impact of news on variance ( ). Table 5. Impacts of GAS Price on EU ETS (GJR-GARCH Model) GAS (No Markov Regime Division) GAS (Markov Regime Division) Bear Market Effect GED PARAMET Akaike ER info criterion Log likelihood GAS (-1) GAS (-2) GAS (-3) GAS (-1) GAS (-2) GAS (-3) (0.0649) (0.0090) (0.0118) *** (0.0307) *** (0.0229) *** (0.0377) *** (0.0148) (0.0254) (0.0637) (0.0089) *** (0.0119) *** (0.0305) *** (0.0237) *** (0.0383) *** (0.0154) (0.0267) (0.0646) (0.0089) (0.0122) *** (0.0303) *** (0.0222) *** (0.0373) *** (0.0146) (0.0266) (0.0648) (0.0090) (0.0121) *** (0.0406) *** (0.0304) *** (0.0235) *** (0.0376) *** (0.0148) (0.0269) *** (0.0386) (0.0637) (0.0089) ** (0.0121) (0.0858) *** (0.0304) *** ( *** (0.0389) *** (0.0154) (0.0270) (0.0845) (0.0645) (0.0090) (0.0122) *** (0.1215) *** (0.0294) *** (0.0212) *** (0.0364) *** (0.0143) (0.0264) *** (0.1209) Notes: Numbers in parentheses are the standard errors of the corresponding estimated parameters; *,** and *** indicate 10%, 5% and 1% levels of significance, respectively. ~ 22 ~

11 Journal of Contemporary Management, Vol. 4, No Concluding Remarks This paper re-examines the impact of energy prices on the EU ETS market and EU ETS market financial characteristics utilizing daily data after the 2009 Copenhagen Summit. The results suggest that oil prices are no longer the driving factor of the EU ETS market and gas prices have a minor effect on the EUA price after the 2009 Copenhagen Summit. The empirical findings innovatively demonstrate the relationship between energy prices and EU ETS market risk and explain also that oil prices have some linear and non-linear impacts on EU ETS market risk. In addition, the results further suggest that the EU ETS market displays a two-regime switching process. One regime is high-mean, low variance state, which is called a bull market in this paper. The other regime is low-mean, high variance state, which is called a bear market. However, in both bull and bear markets, the market mean becomes negative, which indicates the EU ETS market became more risky and uncertain after the Copenhagen Summit. After market division, GJR-GARCH models are applied to modify the non-linear relation between energy price and allowance return and capture the property of volatility in prices. Of the findings, only gas prices have an asymmetric impact on carbon allowance returns under different market regimes. Finally, our further investigation will focus on the effects of other market fundamentals such as weather events on the EU ETS market after the Copenhagen Summit. Furthermore, it is quite meaningful to apply new econometric methodologies to investigate the impacts of fundamental driving factors on the EU ETS market because the EU ETS market structure has changed since the 2009 Copenhagen Summit. References [1] Alberola, E., Chevallier, J., and Cheze, B. (2008). The EU Emissions Trading Scheme: the effects of industrial production and CO 2 emissions on European carbon prices, International Economics, 116(4): [2] Baker, M., and Wurgler, J. (2006). Investor sentiment and the cross-section of stock returns, The Journal of Finance, 61(4): [3] Baker, M., and Wurgler, J. (2007). Investor sentiment in the stock market, Journal of Economic Perspectives, 21(2): [4] Benz, E., and Trück, S. (2009). Modeling the price dynamics of CO 2 emission allowances, Energy Economics, 31(1): [5] Bredin, D., and Muckley, C. (2011). An emerging equilibrium in the EU emissions trading scheme, Energy Economics, 33(2): [6] Chevallier J. (2010). Modelling risk premia in CO 2 allowances spot and futures prices, Economic Modeling, 27(3): [7] Convery, FJ., and Redmond, L. (2007). Market and price developments in the European Union Emissions Trading Scheme, Review of Environmental Economics and Policy, 1(1): [8] Daskalakis, G., and Markellos, RN. (2009). Are electricity risk premia affected by emission allowance prices? Evidence from the EEX, Nord Pool and Powernext, Energy Policy, 37(7): [9] Engle, RF., and Ng VK. (1993). Measuring and testing the impact of news on volatility, The Journal of Finance, 48(5): ~ 23 ~

12 ISSN: (Print); (Online) Academic Research Centre of Canada [10] Fehr, M., and Hinz, J. (2006). A quantitative approach to carbon price risk modeling, Institute for Operations Research, ETH Zentrum, [Online] Available at: math.ethz.ch/staff/maxfehr/carbon.pdf#search='a+quantitative+approach+to+carbon+price+ri sk+modeling' (June 20 th, 2015). [11] Glosten, L., Jagannathan, R., and Runkle, D. (1993). Relationship between the expected value and the volatility of the nominal excess return on stocks, The Journal of Finance, 48(5): [12] Hamilton J.D. (1989). A new approach to the economic analysis of nonstationary time series and the business cycle, Econometrica, 57 (2): [13] Hamilton J.D. (1994). Time Series Analysis. Princeton, New Jersey: Princeton University Press. [14] Hammoudeh, S., Nguyen, D.K. and Sousa, R.M. (2014a). What explain the short-term dynamics of the prices of CO 2 emissions, Energy Economics, 46: [15] Hammoudeh, S., Nguyen, D.K. and Sousa, R.M. (2014b). Energy prices and CO 2 emission allowance prices: A quantile regression approach, Energy Policy, 70: [16] Hintermann, B. (2010). Allowance price drivers in the first phase of the EU ETS, Journal of Environmental Economics and Management, 59(1): [17] Koch, N., Fuss, S., Grosjean, G., and Edenhofer, O. (2014). Causes of the EU ETS price drop: Recession, CDM, renewable policies or a bit of everything? New evidence, Energy Policy, 73: [18] Lutz B J, Pigorsch U, RotfußW. (2013). Nonlinearity in cap-and-trade systems: The EUA price and its fundamentals, Energy Economics, 40: [19] Mansanet-Bataller, M., Pardo, A., and Valor, E. (2007). CO 2 prices, energy and weather, The Energy Journal, 28(3): [20] Paolella, M., and Taschini, L. (2008). An econometric analysis of emission allowance prices, Journal of Bank and Finance, 32(10): [21] Reboredo, J.C. (2013). Modeling EU allowances and oil market interdependence. Implications for portfolio management, Energy Economics, 36: [22] Rittler, D. (2013). Price discovery and volatility spillovers in the European Union emissions trading scheme: A high-frequency analysis, Journal of Banking and Finance, 36(3): [23] Zhang, Y.J., and Wei, Y.M. (2010). An overview of current research on EU ETS: Evidence from its operating mechanism and economic effect, Applied Energy, 87(6): ~ 24 ~