Carbon Price Drivers: Phase I versus Phase II Equilibrium?

Transcription

1 Carbon Price Drivers: Phase I versus Phase II Equilibrium? Anna Creti 1 Pierre-André Jouvet 2 Valérie Mignon 3 1 U. Paris Ouest and Ecole Polytechnique 2 U. Paris Ouest and Climate Economics Chair 3 U. Paris Ouest and CEPII nuary 27th

2 On Carbon Price Drivers Since the EU ETS launch, the corresponding price has shown rather turbulent dynamics: nervous reactions to policy announcements; price collapse after a visible over-allocation in Phase I in As a consequence, the question whether fundamental factors a ect the EUA price remains partially unresolved....but I am not sure that "carbon price does not tend to set according to either short-term, nor long-term fundamentals" as E. Fages concludes...

3 In Creti, Jouvet Mignon, En Econ 2012 we nd: While an equilibrium relationship exists between the carbon price and its fundamentals during Phase II of the EU ETS, such relationship can be obtained for Phase I only if the 2006 structural break occurred on the carbon market is accounted for. increasing role of fundamentals in Phase II. Calculations of equilibrium prices show that while the observed carbon price was close to its equilibrium value and then overvalued at the beginning of Phase II, it tends to be undervalued since the end of 2009.

4 Previous ndings in the carbon proce drivers literature (1) On spot or OTC prices in Phase I: OLS (with Newey-West covariance matrix estimator)-short term e ects Explanatory variable Sign Oil and gas prices + Extreme weather events + Coal price? Switching Price +/(?) Policy Proxies (Shortage) +

5 Our work We look for long-term e ects by using cointegration, in contrast with most of the previous analyses Cointegration: a vector of variables, all which achieve stationarity after di erencing, could have linear combinations which are stationary in level Co-movements among trending variables which could be exploited to test for the existence of equilibrium relationships within a fully dynamic speci cation framework. We rely on cointegration techniques to identify the carbon price determinants: on the whole period (June 24, 2005 to December 20, 2010); on the two subperiods corresponding to Phases I and II of the EU ETS. E. Fages uses the same techniques but between coal and power and gas and power, with a national focus, and then "derive an understanding of carbon price behavior"

6 Our paper extends Bredin and Muckley (2010)- who found cointegration in phase II only-along two directions. we take into account the compliance break in 2006; we derive the equilibrium prices implied by the cointegrating relationship both in Phase I and in Phase II, and unveil the main reasons for the discrepancies between the predicted and the observed price.

7 Data Dependent variable Futures contracts ECX* Explanatory variables Dow Jones Euro Stoxx 50 ICE Brent futures index Switching price *expiry dates until December 2012 We do not include power prices: high volatility di erent geographical scale: a "European" power market still to be developed (high sensitivity to the technological mix)

8 Dependent variable CARBON CARBON /06/ /08/ /10/ /12/ /02/ /04/ /06/ /08/ /10/ /12/ /02/ /04/ /06/ /08/ /10/ /12/ /02/ /04/ /06/ /08/ /10/ /12/ /02/ /04/ /06/ /08/ /10/ /12/ /02/ /04/ /06/ /08/ /10/2010

9 Explanatory variables EUREX GAS EUREX 130 GAS /06/ /08/ /10/ /12/ /02/ /04/ /06/ /08/ /10/ /12/ /02/ /04/ /06/ /08/ /10/ /12/ /02/ /04/ /06/ /08/ /10/ /12/ /02/ /04/ /06/ /08/ /10/ /12/ /02/ /04/ /06/ /08/ /10/ /06/ /08/ /10/ /12/ /02/ /04/ /06/ /08/ /10/ /12/ /02/ /04/ /06/ /08/ /10/ /12/ /02/ /04/ /06/ /08/ /10/ /12/ /02/ /04/ /06/ /08/ /10/ /12/ /02/ /04/ /06/ /08/ /10/2010 COAL BRENT COAL BRENT /06/ /08/ /10/ /12/ /02/ /04/ /06/ /08/ /10/ /12/ /02/ /04/ /06/ /08/ /10/ /12/ /02/ /04/ /06/ /08/ /10/ /12/ /02/ /04/ /06/ /08/ /10/ /12/ /02/ /04/ /06/ /08/ /10/ /06/ /08/ /10/ /12/ /02/ /04/ /06/ /08/ /10/ /12/ /02/ /04/ /06/ /08/ /10/ /12/ /02/ /04/ /06/ /08/ /10/ /12/ /02/ /04/ /06/ /08/ /10/ /12/ /02/ /04/ /06/ /08/ /10/2010

10 Cointegration and causality results We estimate the following relationship Carbon t = α + β 1 Brent t + β 2 Eurex t + β 3 switch t + ɛ t (1) where Carbon t and Brent t respectively denote the prices of carbon and oil, Eurex t is the equity price index, and switch t is de ned by Equation (??). All price series are expressed in logarithmic terms. We include a dummy variable in this relationship to account for the compliance break

11 Phase II: the estimated coe cients have all the expected, positive sign. Phase I: two remarkable di erences the sign of the stock price variable is negative: CO 2 allowances increase the diversi cation of a nancial portfolio and reduce the overall investment risk (Mansanet-Bataller et al., 2008); the switch price is non signi cant (similar to Alberola et al. 2008a)

12 Based on the estimation of the corresponding dynamic equilibrium representation (vector error correction), further results show Phase I-Phase II di erences: the carbon price is not impacted in the short run by the considered fundamentals during Phase I, but it is in uenced in Phase II by the evolution of the switching price and stock prices. the carbon price in uences both the oil and stock prices, putting forward the increasing role of the carbon market in the economy. This can explain (part of) the results by E. Fages

13 Equilibrium Carbon Price Based on the cointegrating relationship obtained for each subperiod, it is possible to derive the corresponding equilibrium carbon price. Phase I Observed and forecasted carbon prices series (in logs) 5 Di erences between the observed and forecasted series 0,4 4,9 0,3 4,8 4,7 0,2 4,6 0,1 4,5 4,4 0 4,3 4,2 0,1 4,1 0,2 4 24/06/ /07/ /08/ /09/ /10/ /11/ /12/ /01/ /02/ /03/ /04/ /05/ /06/ /07/ /08/ /09/ /10/ /11/ /12/ /01/2007 Observed Forecast 24/02/ /03/ /04/ /05/ /06/ /07/ /08/ /09/ /10/ /11/ /12/2007 0,3 24/06/ /07/ /08/ /09/ /10/ /11/ /12/ /01/ /02/ /03/ /04/ /05/ /06/ /07/ /08/ /09/ /10/ /11/ /12/ /01/ /02/ /03/ /04/ /05/ /06/ /07/ /08/ /09/ /10/ /11/ /12/2007

14 Phase II Observed and forecasted carbon prices series (in logs) 5 Di erences between the observed and forecasted series 0,4 4,8 0,3 4,6 0,2 4,4 0,1 4, ,1 3,8 0,2 0,3 3,6 0,4 3,4 01/01/ /02/ /03/ /04/ /05/ /06/ /07/ /08/ /09/ /10/ /11/ /12/ /01/ /02/ /03/ /04/ /05/ /06/ /07/ /08/ /09/ /10/ /11/2009 Observed Forecast 01/12/ /01/ /02/ /03/ /04/ /05/ /06/ /07/ /08/ /09/ /10/ /11/ /12/2010 0,5 01/01/ /02/ /03/ /04/ /05/ /06/ /07/ /08/ /09/ /10/ /11/ /12/ /01/ /02/ /03/ /04/ /05/ /06/ /07/ /08/ /09/ /10/ /11/ /12/ /01/ /02/ /03/ /04/ /05/ /06/ /07/ /08/ /09/ /10/ /11/ /12/2010

15 The sources of misalignments: Phase I The periods of over and undervaluation globally follow the observed movements of the carbon price The rst overvaluation episode occurs at the early operation of the EU ETS market, thus re ecting its immaturity. The second episode hits the market from the beginning of 2006 until April 2006, due to an upward bias in industrial energy and emission forecasts After this spell, the veri cation of the emissions did reestablish the link between anticipations and market fundamentals. The third overvaluation episode occurs in the summer of 2007: driven by an upward and contingent trend in oil prices and The last (milder) overvaluation episode has occurred toward the end of Phase I: driven by demand increase in futures prices transactions

16 The sources of misalignments: Phase II Two distinct patterns: until October 2009, there was an overall overvaluation, except a downward spike during the 2009 winter; since October 2009, a rather stable undervaluation is observed. The nature of the undervaluation episode in 2009 is di erent from the one that has determined the trend observed since the end of the misalignment between January and March 2009 can be explained by looking at the consequences on carbon demand from non-energy related sectors; the recent undervaluation is related to institutional factors in uencing the credibility of the EU ETS market and its design (carousel fraud, phishing attacks).

17 Research avenues? How to measure the credibility of environmental regulation? including other policy instrumets (green incentives) Information transmission from the EU ETS to energy markets? is the EU ETS an accelerator for the European energy markets (decoupling from the international ones?)

18 Switching price, obs. and forecast. carbon prices, Phase I Switching price, obs. and forecast. carbon prices, Phase II /01/ /01/ /02/ /03/ /04/ /05/ /06/ /07/ /08/ /09/ /10/ /11/ /12/ /12/ /01/ /02/ /03/ /04/ /05/ /06/ /07/ /08/ /09/ /10/ /11/ /12/ /12/ /01/ /02/ /03/ /04/ /05/ /06/ /07/ /08/ /09/ /10/ /11/ /11/ /06/ /07/ /08/ /09/ /09/ /10/ /11/ /12/ /01/ /01/ /02/ /03/ /04/ /05/ /05/ /06/ /07/ /08/ /08/ /09/ /10/ /11/ /11/ /12/ /01/ /02/ /03/ /03/ /04/ /05/ /06/ /07/ /07/ /08/ /09/ /10/ /10/ /11/ /12/ Observed Forecast Switch Observed Forecast Switch

19 Cointegration tests Table: Results of Johansen s cointegration trace test (p-value) Null hypothesis Whole period Period 1 Period 2 None 0*** 0*** *** At most At most At most ** At most Note: *** (resp. **): rejection of the null hypothesis at the 1% (resp. 5%) signi cance level. Accounting for the 2006 structural change leads us to recover a long-term relationship between the carbon price and its determinants, putting forward the well-known fact that breaks can bias cointegration tests

20 The cointegrating relationship Table: Estimation of the cointegrating relationships Brent Eurex switch Constant Phase I FM OLS E (6.3690) ( ) (0.0666) (7.1009) DOLS E (6.1480) ( ) (0.1310) (6.8005) Phase II FM OLS (5.7628) (3.3934) (7.4195) ( ) DOLS (5.6359) (3.1669) (7.2915) ( ) Note: t-statistics of the estimated coe cients are given in parentheses, and are based on the Newey and West (1987) s heteroskedasticity and autocorrelation consistent covariance estimator.