Energy Economics 33 (2011) 1295–1312
Contents lists available at ScienceDirect
Energy Economics
j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / e n e c o
A model of carbon price interactions with macroeconomic and energy dynamics
Julien Chevallier ⁎
University Paris Dauphine (CGEMP/LEDa), France
a r t i c l e
i n f o
Article history:
Received 14 March 2011
Received in revised form 14 June 2011
Accepted 10 July 2011
Available online 29 July 2011
JEL classification:
C32
E23
E32
Q43
Q54
Keywords:
Carbon price
Economic activity
Energy prices
Markov-switching model
a b s t r a c t
This paper develops a model of carbon pricing by considering two fundamental drivers of European Union
Allowances: economic activity and energy prices. On the one hand, economic activity is proxied by aggregated
industrial production in the EU 27 (as it provides the best performance in a preliminary forecasting exercise
vs. other indicators). On the other hand, brent, natural gas and coal prices are selected as being the main
carbon price drivers (as highlighted by previous literature). The interactions between the macroeconomic
and energy spheres are captured in a Markov-switching VAR model with two states that is able to reproduce
the ‘boom–bust’ business cycle (Hamilton (1989)). First, industrial production is found to impact positively
(negatively) the carbon market during periods of economic expansion (recession), thereby confirming the
existence of a link between the macroeconomy and the price of carbon. Second, the brent price is confirmed to
be the leader in price formation among energy markets (Bachmeier and Griffin (2006)), as it impacts other
variables through the structure of the Markov-switching model. Taken together, these results uncover new
interactions between the recently created EU emissions market and the pre-existing macroeconomic/energy
environment.
© 2011 Elsevier B.V. All rights reserved.
1. Introduction
In its Energy Bill 2010/2011, the UK Department of Energy and
Climate Change explicitly recognizes that macroeconomic activity
has to be one of the main drivers of carbon prices, and that such
macroeconomic effects should be taken into account when choosing
between various paths towards a low carbon economy (with
simulations on the relative impacts of funding alternative energy
sources such as renewables in the wake of the recession). 1 Analysts
also explicitly recognize the influence of macroeconomic fundamentals. 2 The economic intuition behind the link between growth and
carbon pricing unfolds as follows. First, economic activity fosters
high demand for industrial production goods. In turn, companies
falling under the regulation of the European Union Emissions Trading
Scheme (EU ETS) need to produce more, and emit more CO2 emissions
in order to meet consumers' demand. This yields to a greater demand
for CO2 allowances to cover industrial emissions, and ultimately to
carbon price increases. This intuition is further motivated by the fact
⁎ Université Paris Dauphine, Place du Maréchal de Lattre de Tassigny 75016 Paris,
France.
E-mail address: [email protected].
1
Available at http://www.decc.gov.uk/.
2
See for instance the Point Carbon headlines on April 18, 2011: ‘EU carbon hit by
macroeconomic worries’ and on May 17, 2011: ‘Canada's emissions drop 6% during
recession’, available at http://www.pointcarbon.com/.
0140-9883/$ – see front matter © 2011 Elsevier B.V. All rights reserved.
doi:10.1016/j.eneco.2011.07.012
that supply side issues are negligible in the EU ETS, since allocation is
fixed and known in advance by all market participants (Ellerman and
Buchner (2008)).
The price of carbon is classically driven by the balance between
supply and demand, and by other factors related to market structure
and institutional policies. 3 On the supply side, the number of
allowances distributed is determined by each Member-State through
National Allocation Plans (NAPs), which are then harmonized at the
EU-level by the European Commission. On the demand side, the use
of CO2 allowances is a function of expected CO2 emissions. In turn,
the level of emissions depends on a large number of factors, such as
unexpected fluctuations in energy demand, energy prices (e.g., oil,
gas, coal) and weather conditions (temperatures, rainfall and wind
speed). The demand for allowances can be affected by economic
growth and financial markets as well, but that latter impact needs to
be further assessed in empirical work.
Considerable effort has gone so far into modeling the price dynamics
of CO2 emission allowances (see among others the early work by
Paolella and Taschini (2008), Benz and Trück (2009) and Daskalakis
et al. (2009)). Previous literature (Christiansen et al. (2005), Mansanet-
3
Blyth et al. (2009) and Blyth and Bunn (2011) underline that price formation in
carbon markets involves a complex interplay between policy targets, dynamic
technology costs, and market rules. Besides, they note that policy uncertainty is a
major source of carbon price risk. See Chevallier (2011c) for a literature review.
1296
J. Chevallier / Energy Economics 33 (2011) 1295–1312
Bataller et al. (2007), Alberola et al. (2008a,b) and Bredin and Muckley
(2011)) has also underlined the necessity to document the impact of
economic activity on carbon prices.4 This research question is of general
interest for market participants, brokers, academics and governments
alike, since with a better grasp of the relationship between economic
activity and the carbon market, better hedging strategies, forecasting
models and policy recommendations can be formulated. This topic
has been covered by a sparse academic literature to date. To our best
knowledge, only Alberola et al. (2008a, 2009) have showed that carbon
price changes react to industrial production in three sectors (combustion, paper, iron) and in four countries (Germany, Spain, Poland, UK)
covered by the EU ETS. Nonlinearities, in addition, have been little
studied in previous literature (see Chevallier (in press) for a first
nonparametric kernel regression exercise).5
Some researchers have indirectly attempted to tackle this research
question. Oberndorfer (2009) demonstrates that CO2 price changes
and stock returns of the most important European electricity corporations are positively related. This effect is particularly strong for
the period of carbon market shocks in early 2006, and differs with
respect to the countries where the electricity corporations analyzed
are headquartered. Chevallier (2009) examines the empirical relationship between the returns on carbon futures and changes in
macroeconomic conditions. By estimating various volatility models
for the carbon price with standard macroeconomic risk factors, the
author documents that carbon futures may be weakly forecasted on
the basis of two variables from the stock and bond markets, i.e. equity
dividend yields and the ‘junk bond’ premium. Moreover, Chevallier
(2011a) assesses the transmission of international shocks to the
carbon market in a Factor-Augmented Vector Autoregression model
with factors extracted from a broad dataset including macroeconomic,
financial and commodities indicators. Coherent with the underlying
economic theory, the results show that carbon prices tend to respond
negatively to an exogenous recessionary shock on global economic
indicators. Other studies are remotely connected to our research
question by focusing on competitiveness issues (see Demailly and
Quirion (2008) for a study focused on the iron and steel industry), or
on the macroeconomic costs of the EU climate policy (see Böhringer
et al. (2009) based on the policy analysis computable equilibrium
model). Declercq et al. (2011) analyze the impact of the economic
recession on CO2 emissions in the European power sector, based on a
counterfactual scenario for the demand for electricity, the CO2 and
fuel prices during 2008–2009. By drawing insights from the MERGE
model, Durand-Lasserve et al. (2011) also attempt to evaluate the
impact of the uncertainty surrounding global economic recovery on
energy transition and CO2 prices.
In recent contributions, Bredin and Muckley (2011) examined
the extent to which several theoretically founded factors including
economic growth, energy prices and weather conditions determine
the expected prices of the EU CO2 allowances during 2005–2009.
Through both static and recursive versions of the Johansen multivariate cointegration likelihood ratio test (including time varying
volatility effects), they showed that the EU ETS is a maturing market
driven by these fundamentals. Creti et al. (2011) further confirm this
4
A disclaimer is necessary here: only the European carbon market can be analyzed
and is analyzed in this paper, since it provides an adequate geographical scope in order
to measure economic activity, and it offers enough historical data since January 2005.
As for the ‘world’ price of carbon, which may be inferred from Certified Emissions
Reductions (CERs) for instance, our study would suffer from a lack of historical data
(with the first quotes recorded in March 2007) and benchmark against which to gage
the evolution of economic activity (as there is no such thing as a world GDP indicator).
5
See also the Appendix for a linear VAR model between industrial production and
the carbon price. This kind of linear model fails to detect the empirical relationship
between the macroeconomy and the carbon market. Hence, the purpose of this paper
is to resort to nonlinear econometric techniques with Markov regime-switching
models. We wish to thank a referee for this remark.
result, in a cointegrating framework by using the Dow Jones Euro
Stoxx 50 as their equity variable and by accounting for the 2006
structural break.
This paper specifies and estimates several Markov-switching VAR
models to interlink carbon price, energy and macroeconomic variables. The main contribution of this paper consists in addressing the
interactions between carbon price, macroeconomic and energy variables by allowing the underlying economic regime to change overtime.
Most importantly, differing relationships between the carbon market
and macroeconomic performance depending on the state of the
economy have not been documented yet. This finding, however, is of
crucial importance for regulatory authorities. This paper specifically
extends previous work by Benz and Trück (2009) on the univariate
Markov-switching modeling of the EUA price series. Hence, a novelty is
that we explicitly assess the dynamic behavior of industrial production
(taken here as a proxy of economic activity), energy and carbon prices
by examining alternative specifications of models that differ in the
parameters that switch across regimes. In sharp contrast to previous
work, we consider the possibility that there exist regime changes 6
behind the interactions involving carbon prices, macroeconomic and
energy variables.
We aim at modeling the interactions between the carbon price
and macroeconomic factors, but also by taking into account other
fundamental drivers of the carbon price which have been highlighted
in previous literature, i.e. energy prices (see Alberola et al. (2008b),
Bunn and Fezzi (2009), Oberndorfer (2009), Hintermann (2010),
Mansanet-Bataller et al. (2011) among others). 7 As for the carbon
price series, we work with futures prices (instead of spot prices) since
they were not contaminated by the ban on banking between 2007 and
2008 (see Alberola and Chevallier (2009) and Daskalakis et al. (2009)
on this topic).
The normal behavior of economies is occasionally disrupted by
dramatic events that seem to produce quite different dynamics for the
variables that economists study. Chief among these is the business
cycle, in which economies depart from their normal growth behavior
and a variety of indicators go into decline (Hamilton and Raj (2002)).
Since the creation of the EU ETS in January 2005, the EU economy has
been characterized by the alternance between sustained economic
growth (from 2005 until mid-2007), recession (from the end of 2007
until the summer of 2009), and a timid economic recovery since then.
These recent events provide a strong methodological rationale to
resort to the Markov-switching modeling technique, which allows to
detect changes between regimes.
Following Hamilton (1989), time series may be modeled by following different processes at different points in time, with the shifts
between processes determined by the outcome of an unobserved
Markov chain. In this framework, the presence of multiple regimes can
be acknowledged using multivariate models where parameters are
6
There are numerous explanations that might justify the presence of regime
changes. For instance, it is possible that a given period of economic expansion may
have a different impact on carbon futures depending on the initial size of the economic
shock ceteris paribus.
7
Weather factors have been omitted from the model specification. The main reason
behind this modeling choice unfolds from previous researchers' findings. Indeed,
Christiansen et al. (2005), Mansanet-Bataller et al. (2007), Alberola et al. (2008a,b)
and Bredin and Muckley (2011) have shown that temperatures per se do not have any
statistical influence on CO2 price changes. The absence of temperatures effects is due
to the fact that brokers and market operators anticipate quite well temperatures
changes, based on the decennial seasonal averages. Alberola et al. (2008a,b) have
further shown that the effect of temperatures on CO2 price changes can be captured
only during extreme temperatures events, i.e. an event that strongly departs from the
decennial seasonal averages. In order to take into account these effects, the authors
have introduced dummy variables in their multiple linear regressions. From that
perspective, introducing temperatures in levels in our setting would not have led to
meaningful statistical results. Besides, it does not appear suitable econometrically to
introduce dummy variables in VAR or Markov-switching VAR models, which are rather
based on quantitative variables. We wish to thank a referee for this remark.
J. Chevallier / Energy Economics 33 (2011) 1295–1312
made dependent on a hidden state process. Consider an n-dimensional
vector yt ≡ (y1t, …, ynt)′ which is assumed to follow a VAR(p) with
parameters:
p
yt = μ ðst Þ + ∑ Φi ðst Þyt−i + t
ð1Þ
i=1
t ∼ N ð0; Σðst ÞÞ
where the parameters for the conditional expectation μ(st) and Φi(st),
i = 1, …, p, as well as the variances and covariances of the error terms
t in the matrix Σ(st) all depend upon the state variable st which can
assume a number of values q (corresponding to different regimes).
Given the initial values for the regime probabilities, and the conditional
mean for each state, the log-likelihood function can be constructed
and maximized numerically to obtain the parameters estimates of
the model.8 By inferring the probabilities of the unobserved regimes
conditional on an available information set, it is then possible to
reconstruct the regimes in a spirit similar to the Kalman filter (Harvey,
1991).
Our statistical definition of a turning point is that proposed by
Hamilton (1989), who has suggested modeling the trends in nonstationary time series as Markov processes, and has applied this approach
to the study of post-World War II real GNP.9 Consequently, we view
economic recession as an abrupt shift from a positive to a negative
growth rate in the aggregate economic activity. Specifically, when the
economy is in expansion, growth is μ1 N 0 per month, whereas when the
economy is in recession, average growth is μ2 b 0.
The general idea behind the class of Markov-switching models is
that the parameters and the variance of an autoregressive process
depend upon an unobservable regime variable st ∈ {1, …, M}, which
represents the probability of being in a particular state of the world.
A complete description of the Markov-switching model requires
the formulation of a mechanism that governs the evolution of the
stochastic and unobservable regimes on which the parameters of the
autoregression depend. Once a law has been specified for the states st,
the evolution of regimes can be inferred from the data. Typically, the
regime-generating process is an ergodic Markov chain with a finite
number of states defined by the transition probabilities:
pij = Prob st
+ 1
= jjst = i ;
M
∑ pij = 1
j=1
∀i; j ∈ f1; …; M g
ð2Þ
In such a model, the optimal inference about the unobserved state
variable st would take the form of a probability. Conditional on observing
yt, for example, the observer might conclude that there is a probability
of 0.8 that the economy has entered a recession, and a probability of
0.2 that the expansion is continuing. The transition probabilities of
the Markov-switching process determine the probability that volatility
will switch to another regime, and thus the expected duration of
each regime. Transition probabilities may be constant or a time-varying
function of exogenous variables (see among others Hamilton and
Susmel (1994), Cai (1994), and Gray (1996)). 10
A major advantage of the Markov-switching model is its flexibility
in modeling time series subject to regime shifts. Markov-switching
models have been used in contemporary empirical macroeconomics
8
As noted by Fong and See (2002), it is common practice to assume that the
maximum likelihood estimators are consistent and asymptotically normal.
9
Other studies have concurred that this is a useful approach to characterizing
economic recessions (see among others Boldin (1994), Durland and McCurdy (1994)
and Filardo (1994)).
10
We rely on a constant specification to keep the model parsimonious, and leave the
study of more complicated specifications of the transition probabilities for further
research. Each regime is thus the realization of a first-order Markov chain with
constant transition probabilities.
1297
to characterize certain features of the business cycle, such as
asymmetries between the expansionary and contractionary phases. 11
The Hamilton (1989) model of the US business cycle has fostered
a great deal of interest as an empirical vehicle for characterizing
macroeconomic fluctuations, and there have been a number of subsequent extensions and refinements (see Hamilton and Raj (2002)
and Hamilton (2008) for an introduction). The Markov-switching
model has been tested against a linear autoregressive model by
Hansen (1992, 1996). Its interest has been confirmed by Layton
(1996) and Sarlan (2001) as a very reliable advance signaling system
for the cyclical aspects of the US business cycle turning points, and
by Krolzig (1997) to formalize what it means for the economy to go
into recession. It can also be extended to multivariate settings. For
instance, Krolzig (2001) has generalized Hamilton's model of the
US business cycle to analyze regime shifts in the stochastic process
of economic growth in the US, Japan and Europe (see among other
contributions Albert and Chib (1993), Diebold et al. (1994), Ghysels
(1994), Goodwin (1993), Kähler and Marnet (1994), Lam (1990) and
Phillips (1991)). By imposing further interpretable restrictions on
Hamilton's Markov-switching model, Bai and Wang (2011) have
identified short-run regime switches and long-run structural changes
in the US macroeconomic data. In an interesting application of
Markov-switching models to default spreads, Dionne et al. (2011)
show that macroeconomic factors are linked with various spreads
increases during the recent period, indicating that the spread
variations may be related to macroeconomic undiversifiable risk.
Earlier on this topic, based on Markov-switching models, Alexander
and Kaeck (2008) sent a warning signal by demonstrating that credit
default swap (CDS) spreads are extremely sensitive to stock volatility
during periods of CDS market turbulence.
In this paper, we choose to work with the aggregated EU industrial
production index (from Eurostat) as the variable of interest to represent economic activity. 12 While the EU 27 GDP certainly constitutes
the first proxy of economic activity which comes to mind (along with
GDI), it is only available with a quarterly frequency (at best) from
Eurostat, which would yield to an insufficient number of data points
to carry out our econometric analysis during the time period under
consideration (January 2005–July 2010). Note that we do not need an
exact replication of industrial production for EU ETS sectors only, since
the idea of the paper is to work with a proxy of the general state of the
economy (i.e. trends of economic growth). Besides, with such a proxy
covering only EU ETS sectors, the problem of re-aggregation of the
individual data to the perimeter of the scheme arises. Since there is no
unified methodology to date, arbitrary choices have to be made and
they hamper the comparison of the findings between researchers.
In our view, the industrial production index from Eurostat appears as
the most natural choice, since it benefits from an already established
methodology to work with EU 27 aggregated data. Note also that
industrial production is widely used as an indicator of macroeconomic
activity in previous literature (see for instance Bradley and Jansen
(2004) for an application to stock returns).
We carry out a preliminary forecasting exercise to assess the
predicting power of the aggregated EU 27 industrial production
index compared to other indices (such as energy or forward-looking
indicators). Across the various time series models, the industrial production index provides the best results in predicting the EUA Futures
11
As noted by Morley and Piger (in press), the ‘business cycle’ is a fundamental, yet
elusive, concept in macroeconomics which corresponds to transitory deviations in
economic activity away from a permanent or ‘trend’ level. Moreover, they define
recessions as periods of relatively large and negative transitory fluctuations in output.
Recessions are also found to be closely related to other measures of economic slack,
such as the unemployment rate and capacity utilization.
12
Insofar as emissions are monitored on real-time at the installation level, industrial
production also provides a good proxy of the demand for CO2 allowances.
1298
J. Chevallier / Energy Economics 33 (2011) 1295–1312
price. 13 Then, we include the price of energies (brent, gas, coal) that
has been shown to have a statistically significant impact on the carbon
price in previous literature (see for instance Alberola et al. (2008b),
Mansanet-Bataller et al. (2011)).
To summarize our results at the outset, this paper makes three
important contributions. First, we show in a univariate setting that
the Markov-switching model is able to capture well the underlying
dynamics of carbon market data, while revealing the main events in
time that had an impact on its price development. Second, we use a
two-regime Markov-switching VAR to capture the inter-relationships
between EU industrial production and carbon prices. As one would
expect from theoretical groundings, this model confirms that economic activity has a statistically significant impact (with a delay) on
carbon prices, but there are no ‘rebound effects’. 14 The main switches
between high-growth and low-growth regimes are found in January–
April 2005, April–June 2006, October 2008, and April 2009 until the
end of the sample period. We cautiously suggest some interpretations
based on changes in macroeconomic fundamentals and actual market
developments in the EU ETS. Using these estimates, the empirical
soundness of the Markov-switching model is demonstrated by showing
that it matches all time series well in the dimension of the mean,
variance, skewness and kurtosis. Third, these results are shown to be
robust to the introduction of other energy market shocks (coming from
the brent, natural gas and coal markets) possibly impacting the carbon
market (as in Hintermann (2010)).
The paper is structured as follows. Section 2 details the baseline
univariate Markov-switching model for the carbon price only. Section 3
contains the Markov-switching VAR of the carbon price with macroeconomic dynamics. Section 4 introduces energy dynamics. Section 5
provides a policy discussion. Section 6 briefly concludes.
For carbon prices, the EUA Futures price is gathered in daily
frequency from the European Climate Exchange (ECX). 15 The choice
of the frequency of analysis is a natural consequence of the availability
of financial data on a daily basis, 16 while macroeconomic aggregates
are published on a monthly basis (at best).
Hence, the monthly EUA Futures prices are computed as the
average value of daily observations during a given month. Note that
carbon spot prices are not used in this paper since they were affected
by the ban on banking between Phases I and II of the EU ETS (therefore
plummeting towards zero near the end of 2007, see Paolella and
Taschini (2008), Alberola and Chevallier (2009), Daskalakis et al.
(2009) and Hintermann (2010) on this topic).
The study period goes from January 2005 (i.e., the creation of the
EU ETS) until July 2010. The data sample consists of 67 monthly
observations. It constitutes the best available historical at the time of
writing. 17
The top panel of Fig. 1 shows the raw time series of EUA Futures
prices. The time series is characterized by the presence of shocks during
2005–2007 originating from institutional features of the EU ETS
(Ellerman et al. (2010)). Fig. 1 data also displays in shaded areas the
NBER business cycle reference dates, as published by the NBER's
Business Cycle Dating Committee.18 Recessions start at the peak of
a business cycle and end at the trough. Thus, during our sample period,
the ‘peak’ date is December 2007 and the ‘trough’ date is June 2009. Kim
et al. (2005) have shown that Markov-switching regimes applied to US
real GDP are closely related to NBER-dated recessions and expansions. In
the bottom panel of Fig. 1, the time series looks stationary when
transformed to logreturns. This first diagnostic is confirmed by usual
unit root tests (ADF, PP, KPSS) shown in Table 10 of the Appendix.
In the bottom panel of Fig. 1, there seems to remain some
instability in the transformed time series, 19 more especially for the EU
27 Industrial Production Index in logreturn form (between May 2008
and March 2009). Structural changes have also been detected in the
carbon market data during April–June 2006, which corresponds to
the first compliance event (Alberola et al., 2008b; Chevallier, 2011b).
To investigate further this question, we have run Ordinary Least
Squares—Cumulative Sum of Squares tests (OLS-CUSUM, Kramer and
Ploberger (1992)), which are based on cumulated sums of OLS residuals
against a single-shift alternative. These results, reported in Figure 7
of the Appendix, do not allow us to identify any kind of structural
instability. The empirical fluctuation processes stay safely within their
bounds, and do not seem to indicate the presence of structural breaks in
the data.20 This comment also applies in the remainder of the paper.
Since there is no proof of structural instability based on the OLSCUSUM test, 21 the subsequent Markov-regime switching analyses are
developed over the full sample going from January 2005 to July 2010.
Descriptive statistics are shown in Table 8 (see the Appendix). We
may observe that carbon futures prices have a mean value of €18.93
during the period. For both assets, the Jarque–Bera test indicates that
returns are not normally distributed (unconditionally), as the time
13
Note also that we do not work with emissions data in this paper, since accessing
such data is difficult (at least at the firm level) and it supposes to resort to panel-data
econometric techniques (as in Anderson and Di Maria (2011) who analyze CO2
emissions at the country level). In addition, this approach suffers from the drawback
that emissions data is available with a yearly frequency only in the Community
Independent Transaction Log (CITL) which oversees national registries in the EU.
14
It is hardly conceivable nowadays that carbon prices may have a global impact on
the economy, through for instance cost pass-through in the energy sector or greater
anticipated inflation coming from the rise in carbon prices and the price of
manufactured goods ceteris paribus. Nevertheless, this hypothesis may be explicitly
tested in our econometric framework.
15
From January to March 2005, EUA Futures prices were recovered from Spectron,
one of the major brokers in the energy trading industry, and stem from OTC
transactions (see Benz and Trück (2009) for more details), as ECX was not yet created.
The time series of EUA Futures prices were obtained by rolling over futures contracts
after their expiration date. Carchano and Pardo (2009) analyze the relevance of the
choice of the rolling over date using several methodologies with stock index futures
contracts. They conclude that regardless of the criterion applied, there are not significant
differences between the series obtained. Therefore, it is unlikely that we introduce any
bias by constructing our time series of carbon futures prices.
16
Note that carbon futures prices are also available at the intra-day frequency but,
due to a lack of availability of the data before 2008, the liquidity would be too low to
construct a reliable dataset over the period 2005–2010. See Chevallier and Sévi (2010,
2011) and Conrad et al. (in press) for such analyses.
17
The number of observations can obviously be increased in future studies. We are
not concerned about jeopardizing the large sample properties of maximum likelihood
estimation in this paper. Rather, we are thoroughly checking that the Markovswitching models are well estimated, i.e. that the various regime-switching models
estimated are able to discriminate very clearly between the two regimes, and that
enough data points fall into each regime. These standard conditions to check the
statistical congruency of regime-switching models can be found in Tsay (2010), and
also in Franses and Van Dijk (2003). We wish to thank a referee for this remark.
18
See more on the NBER Business Cycle Expansions and Contractions at http://www.
nber.org/cycles.html.
19
Stock and Watson (1996) first highlighted the importance of this problem for
macroeconomic time series. They suggested that nonlinearity and structural instability
(defined as permanent large shifts in the long-run mean growth rate of the
economies) shall not be analyzed in isolation, which leads to consider time-varying
models (see Granger and Teräsvirta (1999), Timmermann (2000) and Krolzig (2001)
for examples). However, this class of models falls beyond the scope of this article and
is left for future research.
20
Recall that the dataset for this paper has been gathered in monthly frequency.
21
The same comment applies for alternative estimation techniques that would
require to introduce a dummy variable for the recession period (such as in October
2008), or to regress in sub-samples. Note that the latter strategy would suffer from the
drawback of insufficient number of observations in the respective sub-samples.
2. Baseline model
First, we describe the data used, and second we begin our analyses
by using as a baseline the univariate Markov-switching model to
characterize the behavior of the EUA Futures price.
2.1. Data
J. Chevallier / Energy Economics 33 (2011) 1295–1312
1299
EUA Futures Price (ECX)
30
EUR/ton of CO2
25
20
15
10
5
JAN05
NOV05
SEP06
JUL07
MAY08
MAR09
JAN10
EUA Futures Price in Logreturn Form (ECX)
0.5
0.4
0.3
0.2
0.1
0
JAN05
NOV05
SEP06
JUL07
MAY08
MAR09
JAN10
Fig. 1. EUA Futures price in raw (top panel) and logreturn (bottom panel) forms from January 2005 to July 2010. NBER business cycles reference dates are represented by shaded
areas.
The source of the data is ECX.
series appear to be slightly skewed compared to the normal distribution, and consistent with the existence of excess kurtosis.
In Table 8, the Ljung–Box test statistic rejects the existence of
significant autocorrelation. The Engle ARCH test does not show significant evidence in support of GARCH effects (heteroskedasticity).
2.2. Univariate Markov-switching model of the EUA Futures price
As a standard of comparison, we initially fit the univariate Markovswitching model for carbon futures with the unobserved cyclical
factor st. The main purpose is to determine whether there is evidence
of switches between high- and low-growth when the time series of
carbon prices is modeled individually. Such an exercise will provide
us with insights on the idiosyncratic shocks impacting the carbon
market, and may be re-used as a benchmark for the multivariate
specifications in the next sections. The model estimated takes the
form:
yt = μSt + β1;St x1t + t ;
2
t ∼ N 0; σSt
ð3Þ
with St = {1, …, K} the state at time t. K is the number of states, σSt 2 the
error variance at state St, βi, St the coefficient for the explanatory
variable i at state St, and εt the residual vector which is assumed to
follow a Gaussian distribution.
The endogenous variable yt is EU AFUTRET. In the Markov-switching
specification, we choose to include one explanatory variable with
i = {EU AFUTRET(− 1)}, where β1, St captures the influence of the AR(1)
process.
By setting S = [11], both the βi, St coefficients and the model's variance
are switching according to the transition probabilities. Typically, we set
the number of states K equal to 2. Therefore, state K = 1 represents the
‘high growth’ phase, whereas state K = 2 characterizes the ‘low growth’
phase (for more details, see Hamilton (2008) and references therein).
When K = 1, the growth of the endogenous variable is given by the
population parameter μ1, whereas when K = 2, the growth rate is μ2.
As K rises, it becomes increasingly easy to fit complicated dynamics
and deviations from the normal distribution in the returns (Guidolin
and Timmermann, 2006). However, this comes at the cost of having to
estimate more parameters. As Bradley and Jansen (2004) put it, a
well-known problem with any application of nonlinear models is the
problem of overfitting. There is a trade-off between the depth of the
1300
J. Chevallier / Energy Economics 33 (2011) 1295–1312
economic interpretation which one would have available with higher
degrees for state variables, and the numerical difficulties which
accompany such an effort.
In its most popular version, which we use here, the Markovswitching model is estimated by assuming two states, while higherorder processes are much less frequently used. This choice of a twostate process is motivated by the fact that this model is intuitively
appealing to track the ‘boom–bust’ economic cycle, since these two
states may be associated with periods of ‘high-’ and ‘low-growth’. In
this configuration, it will be indeed relatively straightforward to
interpret the two regimes. Finally, in papers dealing with a higher
number of regimes (see Maheu and McCurdy (2000), Guidolin and
Timmermann (2006), and Chan et al. (2011)), it is often the case that
the two-regime model brings the best statistical results. 22
The model is estimated based on Gaussian maximum likelihood
with St = 1, 2. The calculation of the covariance matrix is performed by
using the second partial derivatives of the log likelihood function. P is
the transition matrix which controls the probability of a switch from
state K = 1 to state K = 2:
P=
p11
p12
p21
p22
The sum of each column in P is equal to 1, since they represent the
full probabilities of the process for each state.
Results are provided in Table 1. The statistically significant coefficients of the two means μ show the presence of switches between
high-/low-growth periods. During expansion, output growth per
month is equal to 0.55% on average. The time series is likely to remain
in the expansionary phase with an estimated probability equal to
90.80%. Regime 1 is assumed to last 11.25 months on average. During
recession, the average growth rate is equal to − 3.06%. The probability
that it will stay in recession is equal to 81.70%. The AR(1) process is
significant at the 10% level during Regime 2. The average duration of
Regime 2 is 5.63 months. According to the ergodic probabilities, the
time series would spend 66.91% (33.09%) of the time spanned by our
data sample in Regime 1 (Regime 2).
To further assist with the economic interpretation of the different
regimes, the associated smoothed probabilities 23 are shown in Fig. 2.
The univariate Markov-switching model identifies various kinds of
instabilities on the carbon futures market during the period under
consideration. First, the time series is characterized by switches from
low- to high-growth during September 2005–January 2006. This
episode corresponds to a period of carbon trading where agents had
heterogeneous information with regard to their actual level of emissions, and their ability to meet compliance requirements.
When dealing with the pilot phase of the EU ETS, it is useful to bear in
mind that this period was characterized by fundamental uncertainties
regarding the new rules of the game in this environmental market,
and the extent to which emissions trading in the EU was supposed to
be followed by post-Kyoto agreements and other regional schemes
(in the USA for instance). Therefore, these switches in growth may be
partially explained by the ‘youth’ of this market. During Phases I and II,
participants are gradually acquiring the necessary information to form
their expectations. Moreover, the functioning rules of the EU ETS are
22
See Psaradakis and Spagnolo (2002) and Cho and White (2007) for statistical tests
to determine the number of regimes in Markov-switching models.
23
The estimation routine generates two by-products in the form of the regime and
smoothed probabilities. Recall that the regime probability at time t is the probability
that state t will operate at t, conditional on the information available up to t-1. The
other by-product is the smoothed probability, which is the probability of a particular
state in operation at time t conditional on all information in the sample. The smooth
probability allows the researcher to ‘look back’, and observe how regimes have
evolved overtime (Fong and See (2002)). Since both plots are similar, we only
reproduce the smooth probability in the paper to conserve space. The plot of the
regime probability may be found in the Appendix.
Table 1
Estimation results of the univariate Markov-switching model for the EUA
Futures price.
Log-likelihood
μ (Regime 1)
49.42
0.0055⁎⁎⁎
(0.0015)
− 0.0306⁎⁎⁎
(0.0085)
μ (Regime 2)
Equation for EU AFUTRET
EU AFUTRET(− 1)
βi (Regime 1)
0.1808
(0.1636)
0.3768⁎
(0.2055)
0.0094
0.0143
βi (Regime 2)
Standard error (Regime 1)
Standard error (Regime 2)
Transition probabilities matrix
Regime 1
Regime 2
Regime 1
0.9080⁎⁎⁎
(0.1148)
0.0920⁎⁎
0.1830
(0.1284)
0.8170⁎⁎⁎
(0.0713)
(0.1355)
Regime 2
Regime properties
Prob.
Duration
Regime 1
Regime 2
0.6691
0.3309
11.25
5.63
Note: EUA Futures prices are taken in logreturn form. The (− 1) term into
parentheses refers to the AR(1) process. Standard errors are in parentheses.
The model estimated is defined in Eq. (3).
⁎⁎⁎ Denotes statistical significance at the 1% level.
⁎⁎ Denotes statistical significance at the 5% level.
⁎ Denotes statistical significance at the 10% level.
progressively amended by the European Commission (see Ellerman
et al. (2010)), especially with regard to allocation in Phase III. Conrad
et al. (in press) have modeled the adjustment process of EUAs to the
releases of announcements at high-frequency controlling for intraday
periodicity, volatility clustering and volatility persistence. Their findings
confirm that the decisions of the European Commission on NAPs II have
had a strong and immediate impact on EUAs.
This sub-period is followed by another switch from high- to lowgrowth as the information about market participants' net position
was revealed in April–June 2006, and the allowance price halved in a
few days (Ellerman and Buchner (2008), Alberola et al. (2008b)).
Starting in 2006, the time series enters a high-growth period again,
in a context of steady worldwide economic growth. This sub-period
ended in October 2008, which may be marked as a period of high
trading activity to sell allowances for cash in the midst of the ‘credit
crunch’ crisis (Chevallier (2009, 2011a,b), Mansanet-Bataller et al.
(2011)). 24 Moreover, this situation may be explained by an increased
sensitivity of brokers and traders to policy announcements regarding
the ‘Energy-Climate Package’ signed by EU Member States at the
Poznan Summit in December 2008. Finally, we notice two other
adjustments in the growth regime in April–May 2009 and May 2010,
which may be cautiously related to yearly compliance events on the
carbon market (see Chevallier et al. (2009) for a thorough discussion
of such calendar effects). Therefore, Fig. 2 points toward a number of
prominent examples of crises periods on the carbon market.
Note that most of our comments focus on the demand side of the
carbon market, while the supply of allowances is essentially fixed by
NAPs for each Phase, with some (but not decisive) changes in the
perimeter of the scheme between 2007 and 2008. Therefore, changing
supply between Phases I and II is expected to have little influence
(if any) on the price path and the switches observed in this modeling
exercise. The main drivers of carbon prices are rather linked to
24
See the editorial by Trevor Sikorski (Barclays Capital) in the issue #35 of the
Tendances Carbone newsletter, CDC Climat Research, Paris.
J. Chevallier / Energy Economics 33 (2011) 1295–1312
1301
1
Regime 1
Regime 2
0.9
Smoothed Transition Probabilities
0.8
0.7
0.6
Oct.08
Jun.06
May 10
0.5
0.4
0.3
0.2
0.1
0
JAN05
NOV05
SEP06
JUL07
MAY08
MAR09
JAN10
Fig. 2. Smoothed transition probabilities estimated from the univariate Markov-switching model for the EUA Futures price. Note: Regime 1 is ‘expansion’. Regime 2 is ‘contraction’.
NBER business cycles reference dates are represented by shaded areas.
demand-driven factors, which have been shown to vary nonlinearly
with other energy market shocks (see Alberola et al. (2008b), Bunn
and Fezzi (2009), Oberndorfer (2009), Hintermann (2010), MansanetBataller et al. (2011)). As the first compliance data was revealed in
April 2006, market participants gained useful information on the fact
that the allowance market was globally net long, with some sectors
being in ‘over-allocation’ and others being net short of allowances
(see Ellerman and Buchner (2008)). 25
Overall, the univariate Markov-switching modeling brings us new
insights as to when market shocks occur and actually impact the
carbon futures price. 26 Specifically, our univariate findings feature
‘high-’ and ‘low-growth’ regimes. Next, we examine the influence of
macroeconomic activity in the Markov-switching VAR environment.
Indeed, a good understanding of the dynamics of the carbon futures
price series may serve as a platform to study the linkages with
macroeconomic activity.
3. Macroeconomic dynamics
First, we develop a preliminary forecasting exercise to choose the
best economic indicator. Second, we show the estimation results of
the Markov-switching VAR of the carbon futures price with
macroeconomic dynamics. Third, we provide the associated smoothed
transition probabilities and model diagnostic tests.
3.1. Which monthly indicator is the best proxy for economic activity?
A preliminary forecasting exercise
To justify our choice of this ‘economic activity’ proxy (and keeping
in mind that it needs to capture as accurately as possible changes in
macroeconomic conditions for companies falling under the EU ETS
while being available to researchers for replicability), we evaluate in
25
This paper obviously does not deal with the future changes in the allocation
methodology as of 2012 onwards, even if some market participants' anticipations may
already be contained in the futures price series of the longest maturities (i.e.
deliverable in December 2012 for instance).
26
For the sake of brevity, model diagnostics similar to that of the Markov-switching
VAR are not reported here, but are available in the Appendix (Table 11).
a preliminary step the predictive power of the aggregated EU 27
industrial production index compared to other indicators.
We take an agnostic stance by considering various indicators in
the energy, commodity and monetary spheres, besides the obvious
indicators which come to mind in the macroeconomic sphere. 27
For the purpose of this comparative example, we consider the
following candidates 28:
• EU ESIt is the EU Economic Sentiment Index (from Eurostat). It
reflects overall perceptions and expectations at the individual sector
level in a single aggregate index. This index has been used as a
forward-looking indicator to mirror economic sectors' sentiment by
Mansanet-Bataller et al. (2011) in their analysis of EUA Phase II price
drivers.
• EU BCIt is the Euro Area Business Climate Indicator (from Eurostat).
Its inclusion follows the same logic as the EU Economic Sentiment
Index. Namely, the EU BCIt variable reflects managers' optimism
about order books and production expectations, as well as their
assessment of production trends observed in recent months.
• IN DNEWORDERt is the Industry Manufacturing New Orders Index (from
Eurostat). It may be seen as an indicator of macroceconomic activity, as
businesses pay close attention to the evolution of order books, capacity
constraints and production utilization in their respective industries.
• INDTURNOVERENERGYt is the Industry Turnover Index specific to
Energy Goods (from Eurostat). This index reflects short-term business
27
Chan et al. (2011) masterly illustrate some of the linkages that exist between
financial and commodity markets as follows: ‘During the recent global financial crisis,
strong linkages were observed among different assets. Falling housing prices in the US
contributed to the collapse of a number of banks and other financial institutions, which
triggered sharp declines in global equity markets, commodity prices and international
property markets. The 2008 calendar year was also one of the most volatile periods in the
history of oil prices. (…) In addition, the troubled US economy and fear of a global
recession led to a coordinated government stimulus response that resulted in record low
interest rates in many countries. (…) Strong demand for government bonds, particularly
in major developed countries, drove prices up and yields down substantially. Conversely,
corporate bond spreads widened appreciably. The gold price hit its (then) record high of
over $1000 per ounce in March 2008’.
28
Notice this list is deliberately not exhaustive, and merely aims at evaluating the
merits of the industrial production index retained here compared to other proxies. For
a more in-depth coverage, see the Factor Augmented VAR estimated by Chevallier
(2011a) to capture the interactions between carbon markets and a broad database of
macroeconomic, financial, and commodities indicators.
1302
•
•
•
•
•
J. Chevallier / Energy Economics 33 (2011) 1295–1312
statistics, with the objective to show the evolution of the market for
energy goods.29
INDTURNOVERINTCAPt is the Industry Turnover Index specific to
Intermediate and Capital Goods (from Eurostat). This index reflects
the evolution of the market for intermediate and capital goods.
Hence, it may be seen as a broader index than the previous variable.
S&P GSCIt is the Standard & Poor's Goldman Sachs Commodity
Indicator Total Return. This index may be used as a proxy of changes
in economic conditions that affect commodity markets. The constituent commodities and the economic weighting of this index aim
at minimizing the idiosyncratic effects of some individual commodity markets, and at responding to economic activity. 30 It has been
used by Chevallier (2011a) along with other commodity indicators
to track the interactions with carbon markets.
DJENERGYt is the Dow Jones Euro Stoxx Oil and Gas Energy Index. It
is composed of stocks representative of the energy sector in the Euro
area. The Oil and Gas Energy Index is one of the eighteen sectors
composing the Dow Jones Euro Stoxx Index. It is included here to
test whether an energy-specific indicator may provide a better
predicting power of EUAs than a broader macroeconomic indicator.
YIELD3MONTHt is the Yield Curve Instantaneous Forward rate on
bonds with 3-month maturity (from the European Central Bank). 31
The forward curve shows the short-term (instantaneous) interest
rate for future periods implied in the yield curve. A positive
(negative) value of the slope of the Euro area yield curve indicates
an upward-sloping (downward-sloping) interest rate term structure, and hence a trend to cool down (stimulate) the economy (see
Collin-Dufresne et al. (2001)). It has also been included as an EUA
Phase II price driver by Mansanet-Bataller et al. (2011). Besides, yield
curves are known to have good properties to predict recessions
(see Ang et al. (2006) among others).
YIELDZEROCOUPONt is the Zero-Coupon Yield Curve Spot rate (from
Eurostat), which represents the yield on euro bonds with one year
until maturity. It constitutes another indicator of monetary policy in
the Euro area. 32
Note that other well-known indicators of market trends (such as
the ZEW Indicators, the IFO World Economic Indicators, the IMF
Indicators, the OECD Indicators, etc.) could not be included since they
are available with a quarterly frequency at best, which is insufficient
for this study.
These new indicators may be seen in the Appendix (Figs. 9 and 10).
Descriptive statistics are also reported there in Table 12.
Recall that the main intuition behind this preliminary forecasting
exercise consists in testing whether other indicators of economic
activity may appear as more suitable than the aggregated industrial
production index in order to track the macroeconomic dynamics with
carbon prices.
In our view, these new indicators are susceptible to detect such
likely influences. For instance, the various indices of capacity constraints and order books considered here aim at representing the
29
Turnover comprises the totals invoiced by the observation unit during the monthly
reference period. This corresponds to market sales of energy goods supplied to third
parties.
30
See Geman (2005) for a more detailed analysis of the construction, the coverage,
the liquidity, and the weighting of this index.
31
A yield curve (which is known as the term structure of interest rates) represents
the relationship between market remuneration (interest) rates and the remaining
time to maturity of debt securities. The information content of a yield curve reflects
the asset pricing process on financial markets. When buying and selling bonds,
investors include their expectations of future inflation, real interest rates and their
assessments of risks.
32
A zero coupon bond is a bond that pays no coupon and is sold at a discount from its
face value. The zero coupon curve represents the yield to maturity of hypothetical zero
coupon bonds, since they are not directly observable in the market for a wide range of
maturities. They must therefore be estimated from existing zero coupon bonds and
fixed coupon bond prices or yields.
Table 2
In-sample forecasts for ECX futures without/with economic indicators.
Variable
RMSE
MAE
MAPE
EU
EU
EU
EU
EU
EU
EU
EU
EU
EU
EU
0.1212
0.1155
0.1212
0.1211
0.1211
0.1201
0.1203
0.1188
0.1187
0.1210
0.1204
0.0899
0.0833
0.0899
0.0896
0.0902
0.0891
0.0910
0.0882
0.0882
0.0902
0.0911
103.7648
101.3864
103.8335
107.2441
107.0141
118.7277
112.4975
124.1554
112.6925
116.8585
119.2826
AECXFUTRETt without economic indicator
AECXFUTRETt with EU 27INDPRODRETt
AECXFUTRETt with EU ESIt
AECXFUTRETt with EU BCIt
AECXFUTRETt with INDNEWORDERt
AECXFUTRETtwith INDTURNOVERENERGYt
AECXFUTRETt with INDTURNOVERINTCAPt
AECXFUTRETt with S & PGSCIt
AECXFUTRETt with DJENERGYt
AECXFUTRETt with YIELD3MONTHt
AECXFUTRETt with YIELDZEROCOUPONt
Note: All variables are taken in logreturn form. RMSE refers to the Root Mean Squared
Error, MAE to the Mean Absolute Error, and MAPE to the Mean Absolute Percentage
Error.
effects of relatively ‘tense’ vs. ‘idle’ industrial production constraints.
Moreover, the question of using forward- (such as the EU ESI) vs.
backward-looking (such as industrial production) indicators certainly
deserves our attention at this stage of the paper, which is why they are
included in the preliminary forecasting exercise.
However, bear in mind that the main interest of the paper does not
lie in capturing all the possibly relevant information (which would
point to factor models with hundreds of time series as in Chevallier
(2011a)) or sectoral dynamics (as in Alberola et al. (2008a, 2009)),
but rather in capturing the ‘core’ of the adjustment between carbon
prices and the macroeconomy based on the most representative time
series. These modeling strategies obviously differ in the kind of
general conclusions that may be drawn from the study.
Let us now take our analysis one step further by evaluating whether
these alternative economic indicators improve the forecast performance
of EUA futures. To do so, we regress the EUA futures price without/with
incorporating economic indicators and compare in-sample forecasts
based on the Root Mean Squared Error (RMSE), the Mean Absolute
Error (MAE), and the Mean Absolute Percentage Error (MAPE). These
criteria are used as relative measures to compare forecasts for the same
series across different models, i.e. the smaller the error, the better the
forecasting ability of that model according to that criterion.33
In Table 2, all criteria are minimized in the model incorporating the
industrial production index, which suggests that this variable is useful
for forecasting in this context. These results therefore confirm that
industrial production is relevant in order to forecast the EUA futures
price. 34
In what follows, we have selected for industrial production the EU
27 seasonally adjusted industrial production index gathered in
33
Let yt = x′tβ + t with β a vector of unknown parameters and t the error term.
Setting the error term equal to its mean value of zero, in-sample forecasts are
computed as ŷt = xt′b with b the estimates of the parameters β. The forecast error is
simply the difference between the actual and forecasted value: et = yt − x′tb. Suppose
the forecast sample is j = T + 1, T + 2, …, T + h, and denote the actual and forecasted
value in period t as yt and ŷt , respectively. The reported forecast error statistics are
computed as follows:
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffi
T + h ŷ −y 2
t
t
∑
h
T + 1
T + h ŷ −y t
t
MAE = ∑
h
t =T + 1
ŷ −y RMSE =
t
T + h
MAPE = 100
∑
t =T + 1
t
yt
h
:
34
Although we recognize that this approach also suffers from several drawbacks, one
of them being that industrial production is known to be more volatile than other
economic indicators such as GDP.
J. Chevallier / Energy Economics 33 (2011) 1295–1312
1303
EU 27 Seasonally Adjusted Industrial Production Index (Eurostat)
115
110
105
100
95
90
85
JAN05
NOV05
SEP06
JUL07
MAY08
MAR09
JAN10
EU 27 Seasonally Adjusted Industrial Production Index in Logreturn Form (Eurostat)
0.03
0.02
0.01
0
JAN05
NOV05
SEP06
JUL07
MAY08
MAR09
JAN10
Fig. 3. EU 27 Industrial Production Index in raw (top panel) and logreturn (bottom panel) forms from January 2005 to July 2010. NBER business cycles reference dates are
represented by shaded areas.
The source of the data is Eurostat.
monthly frequency from Eurostat. 35 The current base year is 2005
(Index 2005 = 100). Its perimeter covers total industry excluding
construction, i.e. it also covers non-ETS economic activity (and thereby reflects global economic activity within the geographical zone). It is
represented in Fig. 3 (with corresponding descriptive statistics in
Table 9).
As for EU industrial production, we may distinguish three distinct
phases during our study period. First, the period going from January
2005 to May 2008 may be viewed as a phase of economic growth.
Second, we notice after May 2008 an abrupt decline in industrial
production characterizing the entry of EU economies into the
35
The EU ETS included 25 Member States during the first two years, Bulgaria and
Romania having integrated the trading scheme in 2007 (see Alberola et al. (2009)).
Therefore, we consider the industrial production for the EU 27 (instead of the Euro
area) as the best proxy during our study period. The European-wide industrial
production index has been used in relation to the carbon credit price in Bredin and
Muckley (2011). Besides, to obtain the EU 27 aggregated industrial production index,
we rely on the methodology developed by Eurostat (2010, see Methodology of the
industrial production index).
recession. These events follow with some delay the developments of
the U.S. economy following the first interest rate cut by the Federal
Reserve in July 2007, which is mostly viewed as being the start of
the economic downturn as the first signs of financial distress in the
housing sector met the headlines. 36 Third, from April 2009 until July
2010, we may observe a timid uptake in industrial production.
Therefore, our study period contains an interesting mix of economic
growth, recession and recovery that we aim at analyzing jointly with
the behavior of EUA Futures prices.
Besides, we may remark that the EU industrial production data
corresponds fairly closely to the NBER classification of business-cycle
turning points.
Next, we detail the results obtained with the Markov-switching
VAR model.
36
While analysts detected anomalies in Credit Default Swaps as soon as January
2007, early concerns by the U.S. Board of Governors of the Federal Reserve System
concerning the effects of the credit crunch may indeed be related to August 2007. On
August 17, 2007 the Board approved an initial 50 basis point reduction in the primary
credit rate. See further press releases at the following address: http://www.
federalreserve.gov/newsevents/press/monetary/2007monetary.htm.
1304
J. Chevallier / Energy Economics 33 (2011) 1295–1312
3.2. Markov-switching VAR
As detailed in the Introduction, we apply the econometric
framework of the Markov-switching VAR to the analysis of the
interactions between the carbon price and the industrial production
index. Consequently, Table 3 reports the estimation results for the
two-regime Markov-switching VAR for EU27INDPRODRET and EUAFUTRET. 37 The order of the VAR has been set to p = 2 by minimizing
the AIC. 38
An examination of the coefficients of the two means (μ(st)), which
are all statistically significant, shows the presence of switches in growth
between the two regimes. In Regime 1 (expansion), output growth per
month is equal to 0.14% on average, while in Regime 2 (recession) the
average growth rate amounts to −0.48%. In line with our comments
in Section 2.2, the effects of the recessionary shock are found to be
quite strong during our study period. Besides, an AR(2) seems necessary
to describe the autocorrelation structure of EU27INDPRODRET. For
EUAFUTRET, the process seems be characterized by an AR(1).
Interestingly, the coefficient estimates suggest that the EU industrial
production (variable EU27INDPRODRET) has two kinds of delayed impacts on carbon futures: positive during Regime 1 (as φ2 is equal to 0.78
and highly significant), and negative during Regime 2 (as φ2 is equal to
−0.93 and highly significant). Therefore, these results confirm the
insights by Chevallier (2009, 2011a) concerning the delayed impact of
macroeconomic activity on carbon markets. Other coefficient estimates
do not suggest that carbon futures have any statistical impact on the EU
industrial production.
The bottom lines of Table 3 report the matrix of transition probabilities for the latent variable st (standard error in parentheses).
During an expansionary phase, the series are most likely to remain
in Regime 1 (with an estimated probability equal to 88.73%). On the
contrary, the probability that the series switch from Regime 1 to
Regime 2 is lower (equal to 12.11%). Once the economy finds itself in a
depression, the probability that it will be in a depression the following
month is estimated to be 49.27%. Finally, if the economy is in Regime 2
(recessionary phase), the probability that it will change directly to a
growth regime is equal to 51.89%. Hence, the recessionary phase has
a relatively high probability to be followed by a growth period (which
is consistent with the fact that the economy is picking up near the
end of our study period).
Let us now have a look at the average duration for each regime.
While Regime 2 is assumed to last 1.96 months on average, the
average duration of an expansionary phase is equal to 8.60 months.
Therefore, the transition probabilities associated with each regime
indicate that the first regime is more persistent, and that the economy
spends considerably more time in the ‘high-growth’ regime. Indeed,
the ergodic probabilities imply that the economy would spend about
80% of the time spanned by our sample of data in the first regime
(i.e. expansion). In contrast, regime 2 has an ergodic probability of
about 20%. Hence, these transition probabilities reveal the presence
of important asymmetries in the business cycle.
Finally, another relevant feature of this model lies in the difference
in the residual standard errors across different regimes. Regime 2
exhibits a relatively higher standard error (0.0013) than Regime 1
(0.0009), which reflects the view that recessions are less stable than
expansions.
3.3. Smoothed transition probabilities
Next, we examine the regime and smoothed probabilities generated by the bivariate Markov-switching model of the industrial
37
We have also estimated another specification with three states. We did not find
convincing statistical evidence that the data are really characterized by three separate
regimes. These results may be obtained upon request to the authors.
38
These results are not reported here to conserve space.
Table 3
Estimation results of the two-regime Markov-switching VAR for the EUA Futures price
and the EU 27 Industrial Production Index.
Log-likelihood
μ (Regime 1)
270.01
0.0014⁎⁎⁎
(0.0009)
− 0.0048⁎⁎⁎
(0.0006)
μ (Regime 2)
Equation for EU27INDPRODRET
EU27INDPRODRET
EUAFUTRET
φ1 (Regime 1)
0.1387⁎
(0.0777)
− 0.2079
(0.3383)
0.4343⁎⁎⁎
− 0.0003
(0.0043)
0.0192
(0.0213)
0.0043
(0.0136)
0.0134
(0.0559)
φ1 (Regime 2)
φ2 (Regime 1)
(0.1178)
1.3421⁎⁎⁎
(0.5488)
φ2 (Regime 2)
Equation for EUAFUTRET
EU27INDPRODRET
EUAFUTRET
φ1 (Regime 1)
1.3369
(1.1483)
− 1.5452
(8.7970)
0.7754⁎⁎⁎
0.1253⁎
(0.0760)
0.3455
(0.5497)
− 0.0313⁎⁎
(0.0151)
1.7236
(1.4546)
Standard error (Regime 1)
Standard error (Regime 2)
(0.1893)
− 0.9257⁎⁎⁎
(0.2448)
0.0009
0.0013
Transition probabilities matrix
Regime 1
Regime 2
Regime 1
0.8873⁎⁎⁎
(0.1700)
0.1211
(0.0815)
0.5189⁎
(0.3111)
0.4927
(0.4326)
φ1 (Regime 2)
φ2 (Regime 1)
φ2 (Regime 2)
Regime 2
Regime properties
Prob.
Duration
Regime 1
Regime 2
0.8029
0.1971
8.60
1.96
Note: EUA Futures prices and the EU 27 Seasonally Adjusted Industrial Production
Index are taken in logreturn form. Standard errors are in parentheses. The model
estimated is:
p
yt = μ ðst Þ + ∑ Φi ðst Þyt−i + t
i=1
t ∼ N ð0; Σðst ÞÞ
where the two-dimensional vector yt ≡ (y1t, y2t)′ is assumed to follow a VAR(2)
according to the AIC. The parameters for the conditional expectation μ(st) and Φi(st),
i = 1, 2, the variances and covariances of the error terms t in the matrix Σ(st) depend
upon the state variable st which has two regimes. The transition probabilities are
defined by:
pij = Prob st + 1 = j j st = i ;
M
∑ pij = 1∀i; j ∈ f1; …; M g
j=1
⁎⁎⁎ Denotes statistical significance at the 1% level.
⁎⁎ Denotes statistical significance at the 5% level.
⁎ Denotes statistical significance at the 10% level.
production and carbon price returns to trace how both time series
have evolved over the sample period.
Fig. 4 shows the associated smoothed transition probabilities (with
regime transition probabilities in Fig. 11 of the Appendix to conserve
space). Switches from one regime to another now have a clearer
economic meaning. They are especially perceptible during January–
April 2005, April–June 2006, October 2008 and April 2009 (until the
end of the study period). Whereas there are common effects associated
with broad macroeconomic conditions, we may also distinguish
market-specific effects. The first two significant changes of regime
may tentatively be related to early market developments in the EU
J. Chevallier / Energy Economics 33 (2011) 1295–1312
1305
1
Regime 1
Regime 2
0.9
Smoothed Transition Probabilities
0.8
0.7
0.6
0.5
0.4
Oct.08
0.3
0.2
0.1
0
JAN05
NOV05
SEP06
JUL07
MAY08
MAR09
JAN10
Fig. 4. Smoothed transition probabilities estimated from the two-regime Markov-switching VAR for the EUA Futures price and the EU 27 Industrial Production Index. Note: Regime 1
is ‘expansion’. Regime 2 is ‘contraction’. NBER business cycles reference dates are represented by shaded areas.
ETS. From January to April 2005, market agents had heterogeneous
anticipations with regard to the actual level of carbon prices in a
context of sustained EU economic growth (Ellerman et al. (2010)).
From April 2005 to April 2006, our model globally stays in the growth
regime (with associated probabilities higher than 80%). In April 2006,
carbon prices were characterized by a strong downward adjustment
due to a situation of ‘over-allocation’ compared to verified CO2 emissions (Ellerman and Buchner (2008)). This situation of high price
volatility lasted until the end of June 2006 (Alberola et al. (2008b)).
Then, the model is characterized by another period of growth (with
associated probabilities higher than 80%).
It is interesting to relate these states to the underlying business
cycle: the switches between high- and low-growth based on the
econometric inference do not match the NBER dating of the economic
recession (as detailed in Section 2). There is evidence of the recession
in October 2008, which corresponds to the first regime switch in
the carbon–macroeconomy relationship. Indeed, the EU industrial
production had been falling since July 2007 (see Section 3.1).
However, the carbon market seems to adjust to this situation only
in October 2008, when most operators were looking to sell allowances
in exchange of cash (Chevallier (2009), Mansanet-Bataller et al.
(2011)). This time period also corresponds to the arrival of a lot of
information on the carbon market (including VAT fraud) in a context
of strong macroeconomic uncertainty (with liquidity crises on the
interbank market).
Owyang et al. (2005) note that, for the US aggregate business cycle,
states differ significantly in the timing of switches between regimes,
indicating large differences in the extent to which state business cycle
phases are in concord with those of the aggregate economy. This may
contribute to explain why the switches in our two-regime system
appear later than the NBER business cycle end-of-recession date (June
2009). At the regional level, Hamilton and Owyang (in press) show
that differences across US states appear to be a matter of timing of
business cycles, with some states entering recession or recovering
before others. This may tentatively be advanced here as a justification
for the different timings of entry into the recession identified for the
industrial production and the carbon market.
Finally, other important events are recorded during the end of
our study period. They are characterized by a delayed adjustment of
most commodity markets (among which the carbon market) to the
global recessionary shock (Caballero et al. (2008), Chan et al. (2011),
Chevallier (2011a), Tang and Xiong (2011)) with various switches
from high- to low-growth regimes. 39 We also note that the ‘highgrowth’ regime (Regime 1) is considerably more persistent and
generally less volatile than the other regime. The data generating
process is generally more likely to be in Regime 1, with occasional
episodes of relatively short-lived crises (Regime 2).
As the smoothed transition probabilities become blurred near the
end of the study period, one may wonder whether the macroeconomic effects become less prevalent. One likely explanation is that
from December 2009 onwards, it is possible that the relationship is
weakening due to the failure of the COP/MOP Copenhagen Meeting,
when Member States failed to back up the Kyoto Protocol with a
broader regime. This might presumably translate into a perception
that environmental constraints will be less (legally) binding in the
near future.
Fig. 4 along with Table 3 suggest that the statistical characterization of the macroeconomic activity/carbon market business cycle
afforded by the Markov-switching VAR model is adequate, as our
regime-switching model is able to capture the dramatic changes in
the evolution of both time series highlighted in Sections 2.1 and 3.1.
As shown in the Appendix (Table 14), the results reported in this
section are robust to the selection of other proxies for macroeconomic
performance. 40
Overall, our results tend to confirm that the carbon market adjusts
to the macroeconomic environment only with a delay (see Chevallier
(2009, 2011a)). The main reason lies in its dependence on institutional
news events (Conrad et al., in press). Indeed, the EU ETS was created by
the EU Commission in 2005, and amendments to the scheme profoundly
impact its price path (Alberola et al., 2008a,b). Therefore, if recessionary
shocks can be shown to have a negative effect on the carbon market
(Chevallier (2011a)), the price of CO2 allowances is only weakly
connected to the variables which traditionally impact other equity, bond
and commodity markets, such as dividend yields, ‘junk’ bond yields, TBill rates and excess returns (Chevallier (2009)).
39
For more studies on the linkages between financial assets and commodities, see
Jones and Kaul (1996), Sadorsky (1999) and Driesprong et al. (2008) for the
relationships between oil price movements and stock returns, or Baur and McDermott
(2010) who identify gold as a safe-haven in extreme market conditions.
40
Indeed, the results obtained with the EU ESI indicator instead of the EU industrial
production index are qualitatively unchanged. The EU ESI indicator was selected,
because it provided the second best set of results in the preliminary forecasting
exercise. We thank a referee for this remark.
1306
J. Chevallier / Energy Economics 33 (2011) 1295–1312
Next, we report various robustness tests for the two-regime Markovswitching VAR.
Table 4
Robustness checks of the two-regime Markov-switching VAR for the EUA Futures price
and the EU 27 Industrial Production Index.
3.4. Models diagnostics
Markov-switching VAR
As put forward by Cecchetti et al. (1990), to assess the quality of the
Markov-switching model, we need to develop robustness checks. The
diagnostic checking of estimated Markov-switching models has been
dealt with by Hamilton (1996). The tests are LM-type tests, which have
the attractive property that their computation only requires the
estimation of the model under the null hypothesis. 41
The upper panel of Table 4 reports the results of two diagnostic
tests. The first is a test of the Markov-switching model against the
simple nested null hypothesis that the data follow a geometric
random walk with i.i.d innovations. Because the Markov-switching
model is not identified under the null of the geometric random
walk, the likelihood ratio statistic does not have the standard χ 2
distribution.
Therefore, to assess whether the difference in log-likelihood between the null and Markov-switching models is statistically significant, we compute the standard likelihood ratio statistic as twice the
difference in the maximized log-likelihood values of the null and
alternative models, but adjust the p-value of this statistic upward to
reflect the problem of nuisance parameters (see Hamilton (1989)
and Garcia (1998) for extensions). To adjust the p-value, we use the
methods developed in Davies (1977, 1987) who applies empirical
process theory to derive an upper bound for type I error of a modified
LR statistic under the null, assuming nuisance parameters are known
under the alternative. Note M the p-value from the LR test:
VMðd−1Þ = 2 e−M = 2 2−d = 2
2
Prob LR q N > M = Prob χd N > M +
ð4Þ
Γðd = 2Þ
where Prob(LR(q*) N M|H0) is the upper bound critical value, LR is
the likelihood ratio statistic, q ∗ is the vector of transition probabilities
(q ∗ = argmaxLnL(q)|H1) and d is the number of restrictions under
the null hypothesis. Based on this framework, Davies (1977, 1987)
derived a simple analytical formula assuming that there is a unique
global optimum for the likelihood function:
V = 2M
1=2
LR statistic
p-value
Symmetry test
p-value
RCM 2-state
18.226
0.001
1.698
0.047
6.5115
Distributional characteristics
EU27INDPRODRET
EUAFUTRET
Mean
Median
Maximum
Minimum
Std. Dev.
Skewness
Kurtosis
− 0.0001
0.0002
0.0156
− 0.0237
0.0098
− 0.5717
2.7062
0.0081
− 0.0126
0.3122
− 0.2777
0.1236
0.1229
2.8208
Note: Distributional characteristics are given for the Markov-switching processes
implied by the estimates in Table 3. EUA Futures prices and the EU 27 Seasonally
Adjusted Industrial Production Index are taken in logreturn form. RCM stands for the
Regime Classification Measure.
Compared to Tables 8 and 9, these values demonstrate that the tworegime model we employ matches quite well the first four central
moments of the data. We conclude that the Markov-switching model
produces both the degree of skewness and the amount of kurtosis that
are present in the original data.
Finally, Ang and Bekaert (2002) set out a formal definition of and a
test for regime classification. They argue that a good regime switching
model should be able to classify regimes sharply. Weak regime
inference implies that the regime-switching model cannot successfully distinguish between regimes from the behavior of the data, and
may indicate misspecification. To measure the quality of regime
classification, we therefore use Ang and Bekaert (2002) Regime
Classification Measure (RCM) defined for two states as:
RCM = 400 ×
1 T
∑ p ð1−pt Þ
T t =1 t
ð6Þ
ð5Þ
In Table 4, this adjustment produces a LR statistic equal to 18.226.
We reject the random walk at the 1% level. We conclude that the
relationship is better described by a two-regime Markov-switching
model than by the random walk model.
The second test reported in Table 4 is for the symmetry of the
Markov transition matrix, which implies symmetry of the unconditional distribution of the growth rates. 42 This test examines the
maintained hypothesis that p (the probability of being in a highgrowth state or ‘boom’) equals q (the probability of being in a lowgrowth state or depression) against the alternative that p b q. Table 4
reports statistics that are asymptotically standard normal under the
null. We reject the hypothesis of symmetry at the 5% level.
Next, Table 4 reports the distributional characteristics for the
Markov-switching processes implied by the estimates in Table 3.
Among others, we report the population values of the mean, standard
deviation, skewness and kurtosis computed from the point estimates
of the Markov-switching VAR for EU27INDPRODRET and EUAFUTRET.
41
See Smith (2008) for a review, which confirms that the LM tests have the best size
and power properties among several specification tests for Markov-switching models.
42
As noted by Cecchetti et al. (1990), this is a one-sided test of symmetry against the
alternative of negative skewness.
where the constant serves to normalize the statistic to be between
0 and 100, and pt denotes the ex-post smoothed regime probabilities.
Good regime classification is associated with low RCM statistic values.
A value of 0 indicates that the two-regime model is able to perfectly
discriminate between regimes, whereas a value of 100 indicates that
the two-regime model simply assigns each regime a 50% chance of
occurrence throughout the sample. Consequently, a value of 50 is
often used as a benchmark (see Chan et al. (2011) for instance).
Adopting this definition to the current context, the RCM 2-State
statistic is equal to 6.51 in Table 4. It is substantially below 50, consistent with the existence of two regimes. It is very interesting that
our estimated Markov-switching model has classified the two regimes
extremely well, which capture potentially the relevant information
over the period about the carbon–macroeconomy relationship.
In sum, there is substantial evidence of nonlinearity in the
dynamics of both the EU industrial production and carbon futures
as depicted by the regime-switching model. Therefore, we have
been successful in fine-tuning our understanding of the carbon–
macroeconomy relationship thanks to the two-regime Markov-switching
VAR.
To draw this discussion to a conclusion, the purpose of this
section has been to demonstrate that the Markov-switching model
is well-specified. In addition to its ability to capture certain prominent features of the data that linear models cannot, the added
attractiveness of the Markov-switching model for our purposes is
J. Chevallier / Energy Economics 33 (2011) 1295–1312
90
0.3
80
0.2
70
0.1
60
0
1307
50
40
30
20
JAN05
140
NOV05
SEP06
JUL07
MAY08
MAR09
JAN10
ICE Natural Gas 1 Mth.Fwd. EUR C/Therm
JAN05
0.6
NOV05
SEP06
JUL07
MAY08
MAR09
JAN10
ICE Natural Gas 1 Mth.Fwd. EUR C/Therm in Logreturn Form
0.4
120
0.2
100
0
80
60
40
20
JAN05
NOV05
SEP06
JUL07
MAY08
MAR09
JAN10
140
JAN05
NOV05
SEP06
NOV05
SEP06
JUL07
MAY08
MAR09
JAN10
0.2
0.15
120
0.1
100
0.05
80
0
60
40
20
0
JAN05
NOV05
SEP06
JUL07
MAY08
MAR09
JAN10
JAN05
JUL07
MAY08
MAR09
JAN10
Fig. 5. Energy variables in raw (left panel) and logreturn (right panel) forms from January 2005 to July 2010 (from top to bottom): Crude Oil-Brent Dated FOB, ICE Natural Gas 1
Mth.Fwd., and EEX-Coal ARA Month Continuous — Sett. Price. NBER business cycles reference dates are represented by shaded areas.
The source of the data is Thomson Financial Datastream.
its analytical tractability. Thus, a credible case can be made for the
Markov-switching VAR model to evaluate the carbon–macroeconomy
relationship.
In the next section, we evaluate the robustness of our results
to the inclusion of other potentially fundamental drivers of carbon
prices, i.e. exogenous shocks coming from other energy markets.
4. Energy dynamics
As in Hintermann (2010), we need to assess the sensitivity of our
results to further energy markets shocks possibly impacting carbon
futures during the period (besides macroeconomic shocks). In Fig. 1
for instance, we may notice that, contrary to industrial production,
carbon markets do not instantly react to the sub-primes crisis
inherited from the US (as delimited by the shaded areas). This may
imply that other factors (such as energy prices) were driving the price
of CO2 at that time.
Our goal is to check whether there is another fundamental driver
behind the relationship that the paper is trying to capture. As
highlighted in previous literature (Alberola et al. (2008b), Bunn and
Fezzi (2009), Oberndorfer (2009), Mansanet-Bataller et al. (2011)),
energy prices may be considered as fundamental carbon price
drivers. 43
Thus, we introduce the following energy variables 44:
• BRENTt is the Crude Oil-Brent Dated FOB €/BBL. The brent price is
indeed the reference price within the European Union for crude oil
market products.
43
Markov-switching models are complex in essence to estimate, and the purpose
behind the inclusion of energy variables (such as oil) is to verify that the results
obtained previously documenting the empirical relationship between the carbon
market and the macroeconomy is robust. This robustness check is conducted by
adding the energy variables which have been shown in previous literature to exhibit
the strongest link with the carbon market, i.e. oil, gas and coal. Including more
variables (such as the clean dark and clean spark spreads) would come at a high
numerical cost, i.e. not being able to estimate the Markov-switching regimes precisely
with these extra energy variables. Since our point has been made (i.e. the robustness of
the main results is verified), we do not believe that it would be useful to increase the
number of parameters in our Markov-switching model, which is already high. We
wish to thank a referee for this remark.
44
Note that we do not consider the price of electricity here, since such an analysis
would resort to the power producers' fuel-switching behavior, which goes beyond the
scope of the present paper. In addition, our choice of the number of variables entering
the Markov-switching model embodies a tradeoff between a model that completely
matches the data and one that is tractable.
1308
J. Chevallier / Energy Economics 33 (2011) 1295–1312
• GASt is the ICE Natural Gas 1 Mth.Fwd. € C/Therm. This price series is
usually retained as the reference price for natural gas futures in the
EU.
• COALt is the EEX-Coal ARA Month Continuous – Sett. Price – €/TE. It
represents the coal futures price series for delivery to the
Amsterdam–Rotterdam–Antwerp region.
ceteris paribus, positive oil price shocks trigger carbon price increases
and conversely (see Alberola et al. (2008b), Mansanet-Bataller et al.
(2011) among others). GAS is also found to impact significantly
EUAFUTRET during both regimes, while this is only the case at lag one
for COAL.
4.1. Smoothed transition probabilities
All price series have been converted to Euro by using the bilateral
exchange rates from the European Central Bank. These additional
energy time series are shown in Fig. 5. Descriptive statistics are provided
in Table 5.
Next, we fit another Markov-switching VAR of carbon market
(EU AFUTt) interactions with macroeconomic (EU27INDPRODt) and
energy (BRENTt, GASt, COALt) dynamics. All variables entering the VAR
are transformed to logreturns. The order of the VAR has also been
set to p = 2 by minimizing the AIC. Results are shown in Table 6. The
five equations of the VAR contain 100 parameters, of which 37 are
statistically significant.
In this model, switches between regimes are also apparent (and
statistically significant). During expansion periods, output growth per
month μ1 is equal to 0.15% on average. The series are most likely to
remain in Regime 1 (with an estimated probability equal to 90.08%).
During recession periods, the average growth rate μ2 is equal to
−0.19%. The probability that it will stay in recession the next period is
equal to 45.61%. Conversely, the probability to go from recession to
expansion is higher (55.39%). During the study period, the average
duration of an expansionary (recessionary) phase is equal to 9.95
(1.12) months. The ergodic probabilities imply that the economy
would spend about 87% of the time spanned by our sample of data in
expansion. In contrast, recession has an ergodic probability of about
13%.
Concerning the autocorrelation structure, while PRODRET seems
to follow an AR(2) process, EU AFUTRET may be fitted with an AR(1)
process. Other energy variables may be seen as highly persistent
processes, which conforms to previous literature on energy markets
(see Huisman (2009) for a review).
Our conclusions regarding the link between carbon prices and the
macroeconomy are globally unaffected by the introduction of other
energy market shocks. In the equation for EUAFUTRET, we may still
observe the delayed effect coming from PRODRET with the expected
signs: positive during Regime 1 (with φ2 equal to 0.38 and highly
significant), and negative during Regime 2 (with φ2 equal to −0.26
and highly significant). Note these effects are now present for the first
lag (with φ1 equal to 0.53 and − 0.22 for regimes 1 and 2, respectively,
and highly significant). The magnitude of the coefficients is lower
than in the model without energy dynamics, which may be explained
by the fact that other variables in the system carry explanatory power.
Indeed, we uncover the effects of energy markets shocks on carbon
futures prices (besides macroeconomic shocks). BRENTRET is found to
impact positively (negatively) EUAFUTRET during Regime 1 (Regime 2).
This result conforms to what has been found for industrial production:
Table 5
Descriptive statistics for energy variables.
Mean
Median
Maximum
Minimum
Std. Dev.
Skewness
Kurtosis
Observations
BRENT
GAS
COAL
BRENTRET
GASRET
COALRET
52.3040
51.5121
88.1249
29.9076
11.9501
0.7266
3.7993
67
55.4181
50.5710
134.1108
24.5806
22.7262
1.0254
4.1776
67
57.3626
52.5501
134.4931
19.9216
26.4271
1.0195
3.9827
67
0.0084
0.0279
0.1717
− 0.3757
0.0992
− 1.3593
5.8163
66
0.0032
− 0.0191
0.5346
− 0.4740
0.1841
0.3274
3.4255
66
0.0196
0.0189
0.1936
− 0.2899
0.1013
− 0.8379
4.0318
66
Note: BRENT stands for the Crude Oil-Brent Dated FOB €/BBL, GAS for the ICE Natural Gas
1 Mth.Fwd. € C/Therm, COAL for the EEX-Coal ARA Month Continuous – Sett. Price –
€/TE. RET stands for the transformation of the respective time series in logreturns.
Std. Dev. stands for standard deviation.
In Fig. 6, based on smoothed transition probabilities, the switches
from one regime to another appear during November 2005–September
2006, February 2007, March 2009 and October–January 2010. Compared
to the model without energy dynamics, we therefore uncover a first subperiod characterized by instabilities in carbon markets (see Alberola
et al. (2008b), followed by two similar (but shorter) sub-periods which
appear to coincide with compliance events (see Chevallier et al., 2009).
The last sub-period may be related with the adjustment to the financial
crisis. Similar comments arise from the regime transition probabilities
(shown in Table 12 of the Appendix to conserve space).
4.2. Models diagnostics
Model diagnostics are provided in Table 7. Similar comments to
the previous Section arise.
Overall, we have been able to demonstrate that the Markovswitching model of carbon futures and industrial production is wellspecified by introducing other energy market shocks. Besides the
stability of the coefficients for the former two variables, we have
reached the conclusion that the results concerning the link between
the carbon futures price and the macroeconomy are qualitatively
unaltered. Finally, we have uncovered interesting new results
regarding energy markets interactions. Depending on the geographical scope and the time period, the brent price may indeed be seen as
the leader in the price formation of most other energy markets. These
effects depend on the relative price of fuels and energy inputs to
production (see for instance Bachmeier and Griffin (2006) in this vast
literature).
5. Policy discussion
Going from Fig. 2 for the univariate Markov-switching model
of EUAFUTRET to Fig. 4 for the Markov-switching VAR of EUAFUTRET
with macroeconomic dynamics, and finally to Fig. 6 for the Markovswitching VAR of EUAFUTRET with macroeconomic and energy dynamics,
we may identify visually that the common shocks impacting the
economy occur during the following time periods:
• November 2005–January 2006 which corresponds to a period of
high growth, high energy prices, and relatively high carbon prices
(in absence of reliable information on which to base market
participants' expectations);
• June 2006 which corresponds still to a period of sustained economic
growth (and demand for energy commodities), while the carbon
market is impacted by various kinds of institutional uncertainties
(Alberola et al. (2008b)), Alberola and Chevallier (2009));
• October 2008 which corresponds to a period of severe downward
adjustment of the economy (and the price of energies) to the
financial crisis inherited from the sub-primes crisis in the U.S. In
this context, an increase in the volumes exchanged is recorded on
the carbon market, as some market participants are reported to sell
allowances in exchange for cash;
• April–May 2009 which corresponds to the 2008 compliance event
on the carbon market. Recall that the ‘trough’ date of the NBER
business cycle dating committee is June 2009;
• October 2009–January 2010: as the economy starts to pick up,
various shocks impact energy markets (including geopolitical events).
The carbon price fluctuates in the range of €15, as no credible sign of a
J. Chevallier / Energy Economics 33 (2011) 1295–1312
1309
Table 6
Estimation results of the two-regime Markov-switching VAR for the EUA Futures price with macroeconomic and energy dynamics.
Log-likelihood
μ (Regime 1)
520.59
0.0015⁎⁎⁎
(0.0009)
− 0.0019⁎⁎⁎
μ (Regime 2)
(0.0006)
Equation for PRODRET
PRODRET
EUAFUTRET
BRENTRET
GASRET
COALRET
φ1 (Regime 1)
0.1708⁎⁎⁎
(0.0137)
− 0.1807⁎⁎⁎
(0.0565)
0.4709
(0.3648)
− 0.3511⁎⁎⁎
(0.1168)
0.1069
(0.1766)
− 0.1000
(0.0909)
0.1174
(0.0941)
0.0102
(0.1003)
0.0894
(0.3555)
− 0.0100
(0.0096)
0.1581
(0.1960)
− 0.5534⁎⁎⁎
(0.2020)
0.0811
(0.0721)
0.0810
(0.0725)
0.1407
(0.1108)
− 0.0807
(0.0709)
0.1712
(0.1599)
− 0.1211
(0.1087)
0.2235
(0.1564)
0.3453
(0.2315)
Equation for EUAFUTRET
PRODRET
EUAFUTRET
BRENTRET
GASRET
COALRET
φ1 (Regime 1)
0.5259⁎⁎⁎
(0.1007)
− 0.2176⁎⁎⁎
(0.0400)
0.3848⁎⁎⁎
(0.1171)
− 0.2554⁎⁎⁎
0.3887⁎⁎⁎
(0.1544)
− 0.1079
(0.0946)
0.0710
(0.3399)
− 0.0408
(0.5365)
0.6169⁎⁎⁎
(0.2598)
− 0.6164⁎⁎
(0.3036)
0.2246⁎⁎
(0.1085)
− 0.2040⁎⁎⁎
− 0.2923⁎⁎⁎
(0.0552)
0.2428⁎⁎⁎
(0.0646)
0.2619⁎⁎⁎
(0.0720)
− 0.2612⁎⁎⁎
(0.0193)
(0.0556)
0.7382⁎⁎⁎
(0.2312)
− 0.1116⁎
(0.0695)
− 0.0725
(0.3411)
0.0658
(0.1356)
φ1 (Regime 2)
φ2 (Regime 1)
φ2 (Regime 2)
φ1 (Regime 2)
φ2 (Regime 1)
φ2(Regime 2)
(0.0880)
Equation for BRENTRET
PRODRET
EUAFUTRET
BRENTRET
GASRET
COALRET
φ1 (Regime 1)
− 0.7209
(0.6049)
0.7260
(0.5178)
0.5786⁎
0.0350
(0.3048)
− 0.0200
(0.5689)
0.0100
(0.0970)
− 0.0492
(0.7107)
0.2117⁎
(0.1288)
− 0.6114⁎⁎⁎
(0.2130)
− 0.9011⁎⁎⁎
(0.1010)
0.4959
(0.4700)
0.1942
(0.3430)
− 0.6543
(0.7521)
0.2797
(0.4813)
0.2710
(0.2598)
− 0.2007
(0.1855)
− 0.1032
(0.0844)
0.0952
(0.1133)
φ1 (Regime 2)
φ2 (Regime 1)
(0.2380)
0.7876⁎⁎⁎
φ2 (Regime 2)
(0.3350)
− 0.1861
(0.2674)
Equation for GASRET
PRODRET
EUAFUTRET
BRENTRET
GASRET
COALRET
φ1 (Regime 1)
0.9082⁎⁎⁎
(0.2212)
− 0.4500
(0.3510)
0.4250
(0.4660)
− 0.5120⁎⁎⁎
− 0.1080
(0.2001)
0.0800
(0.2311)
0.2601
(0.2210)
− 0.2103
(0.5390)
− 0.3215⁎⁎
(0.1585)
0.1603
(0.1792)
0.2900⁎⁎
(0.1397)
− 0.5706⁎⁎⁎
0.8836⁎⁎⁎
(0.1004)
− 0.3617
(0.3388)
0.6514⁎⁎⁎
(0.0876)
− 0.1411⁎⁎⁎
(0.1821)
(0.0159)
0.2031
(0.2008)
− 0.1748
(0.1607)
− 0.1932
(0.1639)
0.2056
(0.6850)
φ1 (Regime 2)
φ2 (Regime 1)
φ2 (Regime 2)
(0.0450)
Equation for COALRET
PRODRET
EUAFUTRET
BRENTRET
GASRET
COALRET
φ1 (Regime 1)
φ1 (Regime 2)
0.6367⁎⁎⁎
(0.1809)
− 0.2478⁎
0.1456
(0.2433)
− 0.3111⁎⁎⁎
(0.1485)
0.8790⁎
0.4950⁎
(0.2519)
− 0.3033
(0.4191)
− 0.1030⁎⁎⁎
− 0.8109⁎
(0.4390)
0.9264⁎⁎⁎
φ2 (Regime 1)
− 0.0189
(0.0200)
0.0193
(0.0907)
0.0408
(0.0715)
− 0.0417
(0.4922)
(0.1040)
− 0.5190
(0.3610)
0.4527
(0.3591)
(0.0154)
− 0.3443⁎⁎⁎
(0.1327)
0.3935
(0.3490)
φ2 (Regime 2)
Standard error (Regime 1)
Standard error (Regime 2)
(0.5101)
− 0.8609
(0.8858)
0.0009
0.0011
(0.0323)
− 0.3422
(0.3736)
Transition probabilities matrix
Regime 1
Regime 2
Regime 1
0.9008⁎⁎⁎
(0.3915)
0.0992⁎⁎
(0.0305)
0.5539⁎⁎
(0.2623)
0.4561⁎⁎
(0.2102)
Prob.
Duration
Regime 2
Regime properties
Regime 1
Regime 2
0.8660
0.1340
9.95
1.12
Note: All variables, which have been defined in Sections 2 to 4, have been transformed to logreturns. Standard errors are in parentheses. The model estimated is defined in Eq. (1).
⁎⁎⁎ Denotes statistical significance at the 1% level.
⁎⁎ Denotes statistical significance at the 5% level.
⁎ Denotes statistical significance at the 10% level.
1310
J. Chevallier / Energy Economics 33 (2011) 1295–1312
1
0.9
Regime 1
Regime 2
Smoothed Transition Probabilities
0.8
0.7
0.6
0.5
Fev.07
0.4
Mar.09
0.3
0.2
0.1
0
JAN05
NOV05
SEP06
JUL07
MAY08
MAR09
JAN10
Fig. 6. Smoothed transition probabilities estimated from the two-regime Markov-switching VAR for the EUA Futures price with macroeconomic and energy dynamics. Note: Regime
1 is ‘expansion’. Regime 2 is ‘contraction’. NBER business cycles reference dates are represented by shaded areas.
post-Kyoto agreement has been given by the COP/MOP Copenhagen
Conference in December 2009.
The key policy implications may be summarized as follows: (i) there
is a link between macroeconomic activity and carbon price changes,
(ii) this link seems to channel more precisely through the effects of
industrial production (and associated CO2 emissions) on carbon prices,
and (iii) this so-called ‘carbon–macroeconomy’ relationship is robust to
the introduction of energy market shocks. Univariate and multivariate
Markov-switching models appear adequate to capture this link, while
presenting complementary results.
For investment managers, EUAs appear to be well-suited for
portfolio diversification since they do not match exactly the business
cycle. 45 For regulatory authorities, the lag identified for macroeconomic variables to impact the carbon market suggests that this
environmental market is primarily sensitive to institutional news
announcements (Conrad et al., in press). By enforcing longer-term
targets and without amending the allocation methodology, the
regulator could establish a carbon price signal that would be less
sensitive to market-specific issues, and more linked to the macroeconomic environment. Such a long-term commitment would also
be beneficial to investors in the electricity industry, since their
planning horizon to build new plants goes at least to the medium term
(i.e. 2025).
To draw this discussion to a conclusion, these various Markovswitching models bring us some new insights on the underlying
dynamics in the macroeconomic and energy spheres. They allow us
to trace when switches between high- and low-growth periods
occur, which may then be carefully related to changes in market
fundamentals.
6. Conclusion
Arguably, a satisfactory explanation for carbon price changes lies
in the analysis of macroeconomic fundamentals. Fluctuations in the
level of economic activity are a key determinant of the level of carbon
price returns: as industrial production increases, associated CO2
emissions increase and therefore more CO2 allowances are needed
by operators to cover their emissions (see Hocaoglu and Karanfil
(2011) for further arguments linking industrial activity in the whole
45
This characteristic is shared by some other commodity markets (see Dionne et al.,
2011).
economy and CO2 emissions). This economic logic results in carbon
price increases, due to tighter constraints on the demand side of the
market ceteris paribus. Alberola et al. (2008a, 2009) investigated a
number of factors that could potentially influence carbon price
changes, and identified industrial production in EU ETS covered
sectors as the most important determinant. The purpose of this paper
is to spark the general interest for studying the link between
macroeconomic activity and carbon price changes.
In this paper, we used the approach innovated by Hamilton (1989)
in his analysis of the US business cycle. That approach consists in
fitting a Markov-switching process to a vector of economic time series
in question. We characterize the carbon price as a nonlinear Markovswitching process, and examine its dynamics in response to macroeconomic and energy markets factors over two regimes. The economic
developments during our study period (January 2005–July 2010), i.e. a
combination of sustained economic growth (in 2005–2007), and of a
profound recessionary shock in the aftermath of the sub-primes crisis
(from December 2007 to June 2009 according to the NBER), lead us to
consider the class of Markov-switching models.
Indeed, Markov-switching models have been widely used in
economics and finance since Hamilton (1989) introduced them to
estimate regime- or state-dependent variables. Regimes constructed
Table 7
Robustness checks of the two-regime Markov-switching VAR for the EUA Futures price
with macroeconomic and energy dynamics.
MS VAR
LR statistic
p-value
Symmetry test
p-value
RCM 2-state
29.813
0.001
1.965
0.032
8.9752
Distrib. Char.
PRODRET
EUAFUTRET
BRENTRET
GASRET
COALRET
Mean
Median
Maximum
Minimum
Std. Dev.
Skewness
Kurtosis
− 0.0002
0.0016
0.0244
− 0.0324
0.0117
− 0.3655
3.2110
0.0058
− 0.0222
0.4133
− 0.3165
0.1222
0.4389
4.3413
0.0076
0.0169
0.1929
− 0.3549
0.0980
− 1.3622
6.2987
0.0039
− 0.0030
0.3901
− 0.4349
0.1521
− 0.0797
3.6674
0.0168
0.0174
0.1938
− 0.2869
0.1091
− 0.8322
4.7303
Note: Distributional characteristics are given for the Markov-switching processes
implied by the estimates in Table 6. All variables have been defined in Sections 2 to 4,
and have been transformed to logreturns. RCM stands for the Regime Classification
Measure.
J. Chevallier / Energy Economics 33 (2011) 1295–1312
in this way, using Markov-switching models, are an important instrument for interpreting business cycles. They constitute an optimal
inference on the latent state of the economy, whereby probabilities
are assigned to the unobserved regimes ‘expansion’ and ‘contraction’
conditional on the available information set. Clearly, such an approach
is useful when a series is thought to undergo shifts from one type of
behavior to another and back again, but where the ‘forcing variable’
that causes the regime shifts is unobservable.
In a preliminary in-sample forecasting exercise, we first justify the
choice of the industrial production index (aggregated at the
EU 27 level) by showing that it brings the best results (in terms of
minimizing loss functions) compared to other potential macroeconomic, commodity and energy indicators to predict the EUA futures
price. Then, we follow the modeling approach of Markov-switching
VARs.
This modeling exercise provided us with fruitful results, as we are
able to uncover new relationships between the carbon, economic
and energy variables. We find that the regime-switching model picks
up most of the representative shocks identified by carbon market
analysts 46: January–April 2005, April–June 2006, October 2008, April
2009 until the end of the sample period. Moreover, our results indicate
that the carbon–macroeconomy relationship may fade for some
periods. One possible cause is changes unique to the carbon market
that diminish its ability to react to macroeconomic factors. The results
are robust to a wide range of diagnostic tests, and to the introduction of
energy dynamics (i.e. brent, gas and coal prices), which account for
other market shocks impacting the carbon market (see Hintermann
(2010)). Of particular interest will be the evolution of these relationships between macroeconomic fundamentals, energy markets and
carbon prices during Phase III, with the introduction of auctioning
rules on the supply side of the market and the 20/20/20 targets of the
‘Energy-Climate’ package.
Acknowledgements
I wish to thank warmly the Editor, Prof. Richard S.J. Tol, as well as
two anonymous referees for their detailed comments which led to an
improved version of the paper. For insightful comments on earlier
drafts, I wish to thank David Newbery, Michael Grubb, Michael Pollitt,
David Reiner, Pierre Noël, Ajay Gambhir, Andreas Löschel, Tim
Mennel, Waldemar Rotfuß, Jean-Pierre Ponssard, Anna Creti, PierreAndré Jouvet, Michel Boutillier, Alain Bernard, Guy Meunier, Vanina
Forget, Neil Ericsson, Richard Baillie, Barkley Rosser, Bruce Mizrach,
and Daniel Rittler. Helpful comments were also received from
audiences at the EPRG Energy & Environment Seminar (Electricity
Policy Research Group, University of Cambridge, UK), the CEP Seminar
Series (Centre for Environmental Policy, Imperial College London,
UK), the ZEW Research Seminar (Centre for European Economic
Research, Mannheim, Germany), the Envecon 2011 Conference (UK
Network of Environmental Economists, London, UK), the Environment & CSR Seminar (Ecole Polytechnique, Paris, France), the 19th
SNDE Annual Symposium (Society for Nonlinear Dynamics &
Econometrics, Washington DC, USA), and the EconomiX Lunch
Seminar, the 65th ESEM European Meeting (Econometric Society,
Oslo, Norway), and the 60th AFSE Annual Congress (French
Economics Association, Paris, France). Last but not least, I thank
Eurostat and ECX for providing the data. All errors and omissions
remain that of the author.
Appendix A. Supplementary data
Supplementary data to this article can be found online at doi:10.
1016/j.eneco.2011.07.012.
46
See for instance Point Carbon or IDEA Carbon.
1311
References
Alberola, E., Chevallier, J., 2009. European carbon prices and banking restrictions:
evidence from Phase I (2005–2007). The Energy Journal 30 (3), 51–80.
Alberola, E., Chevallier, J., Chèze, B., 2008a. The EU emissions trading scheme: the effects of
industrial production and CO2 emissions on carbon prices. International Economics
116 (4), 93–125.
Alberola, E., Chevallier, J., Chèze, B., 2008b. Price drivers and structural breaks in
European carbon prices 2005–07. Energy Policy 36 (2), 787–797.
Alberola, E., Chevallier, J., Chèze, B., 2009. Emissions compliances and carbon prices
under the EU ETS: a country specific analysis of industrial sectors. Journal of Policy
Modeling 31 (3), 446–462.
Albert, J., Chib, S., 1993. Bayes inference via Gibbs sampling and autoregressive time
series subject to Markov mean and variance shifts. Journal of Business and Economic
Statistics 11, 1–16.
Alexander, C., Kaeck, A., 2008. Regime dependent determinants of credit default swap
spreads. Journal of Banking and Finance 32 (6), 1008–1021.
Anderson, B., Di Maria, C., 2011. Abatement and allocation in the pilot phase of the EU
ETS. Environmental and Resource Economics 48, 83–103.
Ang, A., Bekaert, G., 2002a. Regime switches in interest rates. Journal of Business and
Economic Statistics 20 (2), 163–182.
Ang, A., Piazzesi, M., Wei, M., 2006. What does the yield curve tell us about GDP
growth? Journal of Econometrics 131 (1–2), 359–403.
Bachmeier, L.J., Griffin, J.M., 2006. Testing for market integration crude oil, coal, and
natural gas. The Energy Journal 27 (2), 55–71.
Bai, J., Wang, P., 2011. Conditional Markov chain and its application in economic time
series analysis. Journal of Applied Econometrics. Early View 26 (5), 715–734.
Baur, D., McDermott, T., 2010. Is gold a safe haven? International evidence. Journal of
Banking and Finance 34, 1886–1898.
Benz, E., Trück, S., 2009. Modeling the price dynamics of CO2 emission allowances.
Energy Economics 31 (1), 4–15.
Blyth, W., Bunn, D., 2011. Coevolution of policy, market and technical price risks in the
EU ETS. Energy Policy 39, 4578–4593.
Blyth, W., Bunn, D., Kettunen, J., Wilson, T., 2009. Policy interactions, risk and price
formation in carbon markets. Energy Policy 37, 5192–5207.
Böhringer, C., Löschel, A., Moslener, U., Rutherford, T.F., 2009. EU climate policy up to
2020: an economic impact assessment. Energy Economics 31 (S2), 295–305.
Boldin, M.D., 1994. Dating turning points in the business cycle. Journal of Business 67,
97–131.
Bradley, M.D., Jansen, D.W., 2004. Forecasting with a nonlinear dynamic model of stock
returns and industrial production. International Journal of Forecasting 20 (2), 321–342.
Bredin, D., Muckley, C., 2011. An emerging equilibrium in the EU emissions trading
scheme. Energy Economics 33 (2), 353–362.
Bunn, D.W., Fezzi, C., 2009. Structural interactions of European carbon trading and
energy prices. Journal of Energy Markets 2 (4), 53–69.
Caballero, R., Farhi, E., Gourinchas, P., 2008. Financial crash, commodity prices and
global imbalances. Brookings Papers on Economic Activity 2, 1–55.
Cai, J., 1994. A Markov model of unconditional variance in ARCH. Journal of Business and
Economic Statistics 12, 309–316.
Carchano, O., Pardo, A., 2009. Rolling over stock index futures contracts. Journal of
Futures Markets 29, 684–694.
Cecchetti, S.G., Lam, P.S., Mark, N.C., 1990. Mean reversion in equilibrium asset prices.
American Economic Review 80 (3), 398–418.
Chan, K.F., Treepongkaruna, S., Brooks, R., Gray, S., 2011. Asset market linkages: evidence
from financial, commodity and real estate assets. Journal of Banking and Finance 35 (6),
1415–1426.
Chevallier, J., 2009. Carbon futures and macroeconomic risk factors: a view from the EU
ETS. Energy Economics 31 (4), 614–625.
Chevallier, J., 2011a. Macroeconomics, finance, commodities: interactions with carbon
markets in a data-rich model. Economic Modelling 28 (1–2), 557–567.
Chevallier, J., 2011b. Detecting instability in the volatility of carbon prices. Energy
Economics 33 (1), 99–110.
Chevallier, J., 2011c. Carbon price drivers: an updated literature review. Working Paper
SSRN #1811963. Social Science Research Network, Cambridge, MA, USA.
Chevallier, J., in press. Nonparametric modeling of carbon prices. Energy Economics.
doi:10.1016/j.eneco.2011.03.003
Chevallier, J., Sévi, B., 2010. Jump-robust estimation of realized volatility in the EU
emission trading scheme. Journal of Energy Markets 3 (2), 49–67.
Chevallier, J., Sévi, B., 2011. On the realized volatility of the ECX CO2 emissions 2008
futures contract: distribution, dynamics, and forecasting. Annals of Finance 7, 1–29.
Chevallier, J., Ielpo, F., Mercier, L., 2009. Risk Aversion and Institutional Information
Disclosure on the European Carbon Market: a Case-Study of the 2006 Compliance
Event. Energy Policy 37 (1), 15–28.
Cho, J.S., White, H., 2007. Testing for regime switching. Econometrica 75 (6), 1671–1720.
Christiansen, A., Arvanitakis, A., Tangen, K., Hasselknippe, H., 2005. Price determinants
in the EU emissions trading scheme. Climate Policy 5, 15–30.
Collin-Dufresne, P., Goldstein, R.S., Spencer Martin, J., 2001. The determinants of credit
spread changes. The Journal of Finance 56 (6), 2177–2208.
Conrad, C., Rittler, D., Rotfuß, W., in press. Modeling and explaining the dynamics of the
European Union allowance prices at high-frequency. Energy Economics. doi:10.
1016/j.eneco.2011.02.011.
Creti, A., Jouvet, P.A., Mignon, V., 2011. Carbon price drivers: Phase I versus Phase II
equilibrium. Working Paper. 2011-06, Climate Economics Chair, Paris.
Daskalakis, G., Psychoyios, D., Markellos, R.N., 2009. Modeling CO2 emission allowances
prices and derivatives: evidence from the European trading scheme. Journal of
Banking and Finance 33, 1230–1241.
1312
J. Chevallier / Energy Economics 33 (2011) 1295–1312
Davies, R.B., 1977. Hypothesis testing when a nuisance parameter is present only under
the alternative. Biometrika 64, 247–524.
Davies, R.B., 1987. Hypothesis testing when a nuisance parameter is present only under
the alternative. Biometrika 74, 33–43.
Declercq, B., Delarue, E., D'haeseleer, W., 2011. Impact of the economic recession on the
European power sector's CO2 emissions. Energy Policy 39, 1677–1686.
Demailly, D., Quirion, P., 2008. European Emission Trading Scheme and competitiveness: a case study on the iron and steel industry. Energy Economics 30, 2009–2027.
Diebold, F.X., Lee, J.H., Weinbach, G.C., 1994. Regime switching with time-varying
transition probabilities. In: Hargreaves, C. (Ed.), Non-stationary Time-series Analysis
and Cointegration. Oxford University Press, pp. 283–302.
Dionne, G., Gauthier, G., Hammami, K., Maurice, M., Simonato, J.G., 2011. A reduced
form model of default spreads with Markov-switching macroeconomic factors.
Journal of Banking and Finance 35 (8), 1984–2000.
Driesprong, G., Jacobsen, B., Maat, B., 2008. Striking oil: another puzzle? Journal of Financial
Economics 89, 307–327.
Durand-Lasserve, O., Pierru, A., Smeers, Y., 2011. Effects of the uncertainty about global
economic recovery on energy transition and CO2 price. MIT-CEEPR Working Paper
#11-003. MIT, USA.
Durland, J.M., McCurdy, T., 1994. Duration-dependent transitions in a Markov model of
U.S. GNP growth. Journal of Business and Economic Statistics 12, 279–288.
Ellerman, A.D., Buchner, B.K., 2008. Over-allocation or abatement? A preliminary
analysis of the EU ETS based on the 2005–06 emissions data. Environmental and
Resource Economics 41, 267–287.
Ellerman, A.D., Convery, F.J., De Perthuis, C., 2010. Pricing Carbon: The European Union
Emissions Trading Scheme. Cambridge University Press, Cambridge.
Eurostat, 2010. Industrial Production up by 0.9% in Euro Area. Up by 0.7% in EU27.
Eurostat News Release Euro Indicators #50/2010. available at http://ec.europa.eu/
eurostat/euroindicators.
Filardo, A.J., 1994. Business cycle phases and their transitional dynamics. Journal of
Business and Economic Statistics 12 (3), 299–308.
Fong, W.M., See, K.H., 2002. A Markov switching model of the conditional volatility of
crude oil futures prices. Energy Economics 24, 71–95.
Franses, P.H., Van Dijk, D., 2003. Non-Linear Time Series Models in Empirical Finance,
Second Edition. Cambridge University Press, Cambridge, UK.
Garcia, R., 1998. Asymptotic null distribution of the likelihood ratio test in Markov
switching models. International Economic Review 39 (3), 763–788.
Geman, H., 2005. Commodity and Commodity Derivatives: Modeling and Pricing for
Agriculturals, Metals and Energy. John Wiley and Sons, Ltd.
Ghysels, E., 1994. On the periodic structure of the business cycle. Journal of Business
and Economic Statistics 12, 289–299.
Goodwin, T.H., 1993. Business-cycle analysis with a Markov-switching model. Journal
of Business and Economic Statistics 11, 331–339.
Granger, C.W.J., Teräsvirta, T., 1999. A simple nonlinear time series model with
misleading linear properties. Economics Letters 62, 161–165.
Gray, S.F., 1996. Modelling the conditional distribution of interest rates as a regime
switching process. Journal of Financial Economics 42, 27–62.
Guidolin, M., Timmermann, A., 2006. An econometric model of nonlinear dynamics
in the joint distribution of stock and bond returns. Journal of Applied Econometrics
21 (1), 1–22.
Hamilton, J.D., 1989. A new approach to the economic analysis of nonstationary time
series and the business cycle. Econometrica 57, 357–384.
Hamilton, J.D., 1996. Specification testing in Markov-switching time series models.
Journal of Econometrics 70, 127–157.
Hamilton, J.D., 2008. Regime-switching models, In: Durlauf, S.N., Blume, L.E. (Eds.), The
New Palgrave Dictionary of Economics, Second Edition. Palgrave Macmillan, pp. 1–15.
Hamilton, J.D., Owyang, M.T., in press. The Propagation of Regional Recessions. The
Review of Economics and Statistics. doi:10.1162/RESTa.00197.
Hamilton, J.D., Raj, B., 2002. New directions in business cycle research and financial
analysis. Empirical Economics 27 (2), 149–162.
Hamilton, J.D., Susmel, R., 1994. Autoregressive conditional heteroskedasticity and
changes in regime. Journal of Econometrics 64, 307–333.
Hansen, B.E., 1992. The likelihood ratio test under nonstandard conditions: testing the
Markov-switching model of GNP. Journal of Applied Econometrics 7, S61–S82.
Hansen, B.E., 1996. Erratum: The likelihood ratio test under nonstandard conditions:
testing the Markov-switching model of GNP. Journal of Applied Econometrics 11,
195–198.
Harvey, A.C., 1991. Forecasting, Structural Time Series Models and the Kalman Filter,
Cambridge University Press, Cambridge.
Hintermann, B., 2010. Allowance price drivers in the first phase of the EU ETS. Journal of
Environmental Economics and Management 59 (1), 43–56.
Hocaoglu, F.O., Karanfil, F., 2011. Examining the link between carbon dioxide emissions
and the share of industry in GDP: Modeling and testing for the G-7 countries.
Energy Policy 39 (6), 3612–3620.
Huisman, R., 2009. An Introduction to Models for the Energy Markets. Risk Books,
London, UK.
Jones, C., Kaul, G., 1996. Oil and stock markets. The Journal of Finance 51, 463–491.
Kähler, J., Marnet, V., 1994. Markov-switching models for exchange rate dynamics and
the pricing of foreign-currency options. In: Kähler, J., Kugler, P. (Eds.), Econometric
Analysis of Financial Markets. Physica Verlag.
Kim, C.J., Morley, J., Piger, J., 2005. Nonlinearity and the permanent effects of recessions.
Journal of Applied Econometrics 20 (2), 291–309.
Kramer, W., Ploberger, W., 1992. The CUSUM test with OLS residuals. Econometrica 60,
271–285.
Krolzig, H.M., 1997. Markov-switching vector autoregressions. Modelling, statistical
inference and application to business cycle analysis. Modelling, Statistical Inference
and Application to Business Cycle Analysis. : Lecture Notes in Economics and
Mathematical Systems, 454. Springer.
Krolzig, H.M., 2001. Business cycle measurement in the presence of structural change:
international evidence. International Journal of Forecasting 17 (3), 349–368.
Lam, P.S., 1990. The Hamilton model with a general autoregressive component:
estimation and comparison with other models of economic time series. Journal of
Monetary Economics 26, 409–432.
Layton, A.P., 1996. Dating and predicting phase changes in the U.S. business cycle.
International Journal of Forecasting 12 (3), 417–428.
Maheu, J., McCurdy, T., 2000. Identifying bull and bear markets in stock returns. Journal
of Business and Economic Statistics 18, 100–112.
Mansanet-Bataller, M., Pardo, A., Valor, E., 2007. CO2 prices, energy and weather. The
Energy Journal 28 (3), 73–92.
Mansanet-Bataller, M., Chevallier, J., Herve-Mignucci, M., Alberola, E., 2011. EUA and
sCER Phase II price drivers: unveiling the reasons for the existence of the EUA-sCER
spread. Energy Policy 39 (3), 1056–1069.
Morley, J., Piger, J., in press. The Asymmetric Business Cycle. The Review of Economics
and Statistics. doi:10.1162/RESTa.00169.
Oberndorfer, U., 2009. EU emission allowances and the stock market: evidence from the
electricity industry. Ecological Economics 68, 1116–1126.
Owyang, M.T., Piger, J., Wall, H.J., 2005. Business cycle phases in U.S. States. The Review
of Economics and Statistics 87 (4), 604–616.
Paolella, M., Taschini, L., 2008. An econometric analysis of emission allowances prices.
Journal of Banking and Finance 32 (10), 2022–2032.
Phillips, K., 1991. A two-country model of stochastic output with changes in regime.
Journal of International Economics 31, 121–142.
Psaradakis, Z., Spagnolo, F., 2002. On the determination of the number of regimes in Markovswitching autoregressive models. Journal of Time Series Analysis 24 (2), 237–252.
Sadorsky, P., 1999. Oil price shocks and stock market activity. Energy Economics 21,
449–469.
Sarlan, H., 2001. Cyclical aspects of business cycle turning points. International Journal
of Forecasting 17 (3), 369–382.
Smith, D.R., 2008. Evaluating specification tests for Markov-switching time-series
models. Journal of Time Series Analysis 29 (4), 629–652.
Stock, J.H., Watson, M.W., 1996. Evidence of structural instability in macroeconomic
time series relations. Journal of Business and Economic Statistics 14, 11–30.
Tang, K., Xiong, W., 2011. Index investing and the financialization of commodities.
Working Paper. Princeton University, USA.
Timmermann, A., 2000. Moments of Markov switching models. Journal of Econometrics
96, 75–111.
Tsay, R.S., 2010. Analysis of Financial Time Series, Third Edition. Wiley, New Jersey, USA.