Mean Reversion of Stochastic Convenience Yields For CO2 in ETS
Mean Reversion of Stochastic Convenience Yields For CO2 in ETS
www.elsevier.es/srfe
Article
a r t i c l e i n f o a b s t r a c t
Article history: This paper examines the mean-reversion property and volatility features of stochastic convenience yields
Received 11 February 2012 for CO2 emissions allowances by using ADF, ECM-GARCH and ECM-TGARCH models. Empirical results
Accepted 10 January 2013 show that the convenience yields for CO2 emissions allowances exhibit time-varying trends when differ-
Available online 6 March 2013
ent maturities are considered, and that convenience yields exhibit a linear mean-reverting process. We
also find that the volatility of convenience yields exhibits a mean-reversion process and asymmetric lever-
JEL classification: age effect using ECM-GARCH (1,1) and ECM-TARCH (1,1) models. Unfavorable market information has a
C1
higher impact on this volatility than favorable market information, and unfavorable market information
C32
G19
has a lower effect on the long-term volatility of convenience yields.
Q29 © 2012 Asociación Española de Finanzas. Published by Elsevier España, S.L. All rights reserved.
Keywords:
CO2 emissions allowances
Convenience yields
Mean-reversion
Asymmetric clustering
ECM-GARCH
ECM-TARCH
2173-1268/$ – see front matter © 2012 Asociación Española de Finanzas. Published by Elsevier España, S.L. All rights reserved.
http://dx.doi.org/10.1016/j.srfe.2013.01.001
40 K. Chang et al. / The Spanish Review of Financial Economics 11 (2013) 39–45
interest rates are important variables. Liu and Tang (2011) show allowances. Section 4 proposes mean-reversion empirical method-
that the volatility of convenience yields is heteroskedastic for ology. Section 5 estimates and discusses the empirical results.
industrial commodities, and propose that the heteroskedasticity Section 6 provides a brief conclusion.
of the convenience yields predicts an upward sloping implied
volatility smile. These signs are of central importance to com-
2. Data description
modities pricing and commodities options value. The stochastic
convenience yield is a significant variable for commodity pricing,
The EU ETS is the largest greenhouse gas (GHG) emissions trad-
and mean-reversion convenience yields are central for forecast-
ing system in the world. It has experienced two phrases: the Pilot
ing commodity price and estimating hedging returns. Accordingly,
phase (2005–2007) and the Kyoto phase (2008–2012). The CO2
in this paper we examine the mean-reverting properties and the
emission rights, called EU allowances (EUA), allow for the right to
volatility features of stochastic convenience yields for CO2 emis-
emit one ton of CO2 into the atmosphere under the European Union
sions allowances.
emissions trading scheme (EU ETS). The minimum trading volume
Mean reversion is a tendency toward return to long-run aver-
for each standard futures contract is 1000 tons of CO2 equivalents.
age value over time. Previous studies find that convenience yields
In this paper, we select data samples from the settlement spot and
show a mean-reverting process in the two-factor and three-factor
futures prices in the BlueNext and ICE exchange platform. Spot trad-
commodity futures pricing models. Generally speaking, mean-
ing in the BlueNext exchange was introduced on June 24, 2005, and
reversion behavior in convenience yields is expected because of
now BlueNext has become the most liquid platform for CO2 spot
the strong tendency, and short-term random convenience yields
trading. ICE has become the most liquid platform for CO2 futures
converge to their mean values in the long run. After tempo-
trading since its introduction on April 22, 2005.
rarily deviating from their equilibrium value, convenience yields
After the European Union banned out-of-phase banking and
always revert toward their equilibrium value, hence the process is
borrowing, the spot price for CO2 emissions allowances fell down to
mean-reverting. This property of convenience yields is an impor-
zero from October 2006 to December 2007 (see Chevallier, 2010).
tant hedge and risk management factor for commodity producers,
The trading of futures contracts with vintages December 2013
hedgers, financial intermediaries and other market participants.
and 2014 was introduced on April 8, 2008. We select data sam-
Immature emissions allowances markets bring about the over-
ples from time-varying settlement prices on EUA futures contracts
reaction in spot and futures prices (see Montagnoli and Vries,
with different maturities from December 2010 to December 2014.
2010; Zhang and Wei, 2011). In the weak-effective emissions
Considering the availability and continuity of EUA futures prices,
allowances markets, spot and futures prices have greater upward
we choose these data samples to cover the period from April 8,
risk and downward risk trends, the obvious market risk changes
2008 to December 20, 2010 in the Kyoto phase. Here we choose as
bring market participants about tremendous uncertainty in assets
the constant free-risk rate, the average coupon rate of 3.06% which
portfolio between spot and futures for emissions allowances. The
was the rate for three-year government bonds issued in 2010 in the
convenience yields are potential benefits implied from emissions
European Union.
allowances markets and the above early literatures on emissions
In Fig. 1, S denotes spot price for CO2 emissions allowances,
allowances do not propose empirical results in mean reversion
F1 denotes the EUA futures contracts that are closest to matu-
property and volatility features of convenience yields. Mean rever-
rity, F2 denotes the second closest to maturity, and F3 , F4 , F5 are
sion property and volatility features of convenience yields are
defined similarly. From Fig. 1, we obviously observe that CO2 price
central to accurately predicting futures options pricing and making
series for both spot and futures contracts with different maturi-
correct assets portfolio hedging policies.
ties exhibit strongly time-varying trends throughout the sample
The main innovations of this paper are that we capture mean-
period.
reversion property and asymmetric leverage effects in convenience
yields for emissions allowances by using the ECM-GARCH and ECM-
TARCH models. These empirical results are helpful for capturing 3. Convenience yields for CO2 emissions allowances
market price behavior and explaining the spread between spot and
futures prices. They are also helpful for accurately adjusting assets Convenience yields are defined as the immediate benefit or risk
portfolio sizes between spot and futures and achieving the greater premium associated with holding underlying products or physical
assets portfolio revenues. commodities at hand. Spot holders can achieve potential bene-
The remainder of our paper is organized as follows. Section 2 fits due to price volatility, but the holders of futures contracts
describes the sourcing of data samples. Section 3 analyzes the sta- cannot attain such benefits (see Working, 1949; Brennan, 1958).
tistical analysis results in convenience yields for CO2 emissions The prices of CO2 emissions allowances exhibit random trends,
40
35
30 S
F1
25
Price (Euros)
F2
20 F3
F4
15
F5
10
5
Date
0
04-08-08
06-08-08
08-08-08
10-08-08
12-08-08
02-08-09
04-08-09
06-08-09
08-08-09
10-08-09
12-08-09
02-08-10
04-08-10
06-08-10
08-08-10
10-08-10
12-08-10
Fig. 1. Time series in spot and futures prices for EUA emissions allowances.
K. Chang et al. / The Spanish Review of Financial Economics 11 (2013) 39–45 41
Table 1
Descriptive statistics in convenience yields for CO2 emissions allowances (×100%).
so spot holders of emissions allowances suffer the largest mar- the convenience yields are larger positive, thus convenience yields
ket risk. Price fluctuation for emissions allowances is induced exhibit a leptoturtosis disttribution.
by many factors, such as change in regulation policy for emis-
sions control, energy price volatility, innovation and application 4. Empirical methodology for convenience yields
in low-carbon technology, extreme climate change, etc. Conve-
nience yields fully reflect the spot holders’ expectation of price Based on Granger and Teräsvirta (1993), the nonlinear unit root
changes for emissions allowances. Short supply induces the market test used in this paper is a basic extension of the traditional linear
scarcity in quantity of CO2 emissions allowances. Greater mar- ADF unit root test. In order to investigate whether convenience
ket scarcity pushes up the price for emissions allowances, so spot yields for CO2 emissions allowances involve linear mean reversion,
holders can achieve the highest convenience yields for emissions we use the Augmented Dickey–Fuller (ADF) unit root methodology.
allowances.
m−1
The simple cost of carry model describes an arbitrage relation
among the futures price, spot price, and the cost of carrying the xt = a + bxt−1 + ci xt−i + εt (2)
assets (see Heaney, 2002). Assuming no arbitrage, transaction cost i=1
and storage cost for CO2 emissions allowances, based on the cost Here xt denotes the time series of the convenience yields for CO2
of carry model, the futures and spot prices are linked through emissions allowances, and shows an m lag vector. xt = xt − xt−1
F(ı, t, T ) = St e(r−ı)(T −t) . Here St denotes spot price for emissions is the basis of non-stationary variables xt , and εt is a white noise
allowances at time t, F(t, T) denotes price of futures contracts for vector. Rejection of the unit root null hypothesis H0 : b = 0 against
maturity T at time t, r is the continuously compounded risk-free the stationary linear alternative hypothesis H1 : b < 0 implies that
interest rate, and ı is the convenience yield for CO2 emissions the series is linear mean reverting.
allowances (see Working, 1949; Brennan, 1958). Accordingly, the Engle and Granger (1987) demonstrate that the error-correction
convenience yield is equal to (Table 1) term must be mean reverting if two underlying assets prices
1
F(t, T ) ln F(t, T ) − ln St show co-integration. The error correction term reflects short-
ı(t, T ) = r − ln =r− (1) term deviation from the long-term equilibrium. Accordingly, the
T −t St T −t
error-correction model (ECM) considers non-stationary time series,
In Fig. 2, cy1 , cy2 , cy3 , cy4 , cy5 denote the convenience yields for long-term equilibrium and short-term dynamics for the under-
CO2 emissions allowances futures contracts with different maturi- lying financial assets. The volatility in the convenience yields is
ties from December 2010 to December 2014. Among them, cy1 is the significantly variable for the holders of both spot and futures
convenience yield with the futures contract that is closest to matu- contracts. In order to capture the volatility behavior, we use error-
rity, cy2 is the convenience yields with the futures contract that is correction general auto-regression conditional heteroskedasticity
second closest to maturity, and cy3 , cy4 , cy5 are defined similarly. (ECM-GARCH).
From Table 2, we detect the obviously time-varying trends for all
the convenience yields, and standard deviation in the convenience xt = ext−1 + fxt−1 + (3)
yields for CO2 emissions allowances declines with the increase of i j
time-to-maturity. Mean value in the convenience yields decreases
t2 = g + 2
˛i t−1 + 2
ˇj t−1 (4)
with the increase of time-to-maturity. Higher negative conve-
p=1 q=1
nience yields in the Kyoto period show that market participants
reach higher yields by holding futures contracts. All skewnesses in where ˛, ˇ are all positive, and volatility in the convenience yields is
the convenience yields are less than zero, thus convenience yields time-varying. The conditional volatility in the convenience yields
exhibit left-skewed and asymmetric distribution. All kurtoses in follows a GARCH (1,1) process, whereby the conditional volatil-
cy1
6
cy2
4
Convenience yields
cy3
2
cy4
0
cy5
–2 Date
–4
–6
–8
04-08-08
05-08-08
06-08-08
07-08-08
08-08-08
09-08-08
10-08-08
11-08-08
12-08-08
01-08-09
02-08-09
03-08-09
04-08-09
05-08-09
06-08-09
07-08-09
08-08-09
09-08-09
10-08-09
11-08-09
12-08-09
01-08-10
02-08-10
Fig. 2. Convenience yields for CO2 emissions allowances with different maturities (×100%).
42 K. Chang et al. / The Spanish Review of Financial Economics 11 (2013) 39–45
Table 2
Empirical results of Augmented Dickey–Fuller (ADF) tests for convenience yields.
Notes: 1. Lag length is equal to 1. We estimate coefficients by the following equation xt = a + bxt−1 + c1 xt−1 + εt .
2. At confidence levels of 99%, 95%, 90%, the critical values of the ADF test with intercept are −3.4396, −2.8655, −2.5689.
3. p-Value denotes the ADF test statistic with intercept, p-value denotes the first-difference ADF test statistic with intercept, the statistic value in parentheses is the standard
error.
*
Estimated coefficients are significant at the significance levels of 90%.
**
Estimated coefficients are significant at the significance levels of 95%.
***
Estimated coefficients are significant at the significance levels of 99%.
2
ity at time t depends on the last period’s squared innovation t−1 brings ˛ +
r multiple impacts on the conditional volatility t2 .
2 . GARCH time-varying variance
and the last period’s volatility t−1 If
r > 0, the convenience yields for CO2 emissions allowances
models are an alternative for modeling the heteroskedasticity often exhibit a leverage effect, and the last period’s asymmetric impact
found in time series. The above ECM-GARCH model describes the term makes the conditional volatility increase. If
r > 0, the last
aggregated behavior of volatility in the convenience yields for CO2 period’s asymmetric impact term makes the conditional volatility
emissions allowances. decrease.
Time-series prices for many financial assets often exhibit a
leverage effect, and unfavorable information in the financial assets 5. Empirical results
price is associated with higher volatility than favorable informa-
tion. The spot and futures prices for emissions allowances are 5.1. Mean-reversion of the CO2 convenience yields
determined by many uncertain factors, and the spot and futures
prices may exhibit a lead–lag relationship. In order to exam- In Table 2, the ADF statistical values in the convenience yields
ine the asymmetry dynamics in the volatility of convenience cy1 , cy4 , cy5 are smaller than the critical value −2.5689 at the
yields for CO2 emissions allowances, we select the asymme- 90% confidence level. These show that the time series in the con-
try ECM-TARCH model by Zakoian (1994) and Glosten et al. venience yields cy1 , cy4 , cy5 are stationary, while the remaining
(1993). series are non-stationary. Statistic values with first-difference ADF
in five convenience yields are far less than the critical value at
i j k
the 99% confidence level, and the first difference series in five
t2 = h + 2
˛i t−1 + 2
ˇj t−1 +
r u2t−1 It−1 (5) convenience yields are stationary. All estimated b coefficients are
p=1 q=1 r=1 significant at the 95% confidence level, and b < 0 implies that the
convenience yields series exhibit a linear mean-reversion process.
Asymmetry in conditional variance is introduced through the The linear mean-reversion property of convenience yields suggests
term u2t−1 , and the impact of such asymmetries is captured by mean reversion to a long-run equilibrium that might change ran-
It . Here It−1 is a dummy variable vector, It−1 is equal to 1 when domly over time. These results of convenience yields are similar
t−1 < 0, otherwise It−1 is equal to zero when t−1 ≥ 0. The term to previous results of many scholars, such as the Schwartz (1997),
r u2t−1 It−1 in the conditional volatility Eq. (5) is the asymmet- Pindyck (1999), Schwartz and Smith (2000), Cortazar and Schwartz
ric impact term or TARCH term. The term t2 in the conditional (2003), and Bernard et al. (2008).
volatility equation depends on the last period’s squared residual The early empirical results show many financial products mar-
2 , the last period’s conditional volatility 2
t−1 and the asym-
t−1 kets and commodities markets have prices spillover and futures
metric impact term
r u2t−1 It−1 . From Eq. (5), we propose that mispricing trends (see Taylor, 2004; Bilson et al., 2005; Mcmillan
volatility in the convenience yields exhibits a shock effect when and Ulku, 2009). Arouri et al. (2012) find that carbon spot and
emissions markets receive favorable and unfavorable informa- futures have asymmetrical and nonlinear relationships and sug-
tion. The convenience yields for CO2 emissions allowances indicate gest the usefulness of nonlinear models in pricing and forecasting
asymmetric effect when
= / 0. When t−1 > 0, the asymmet- carbon allowances prices. Our empirical results find that the con-
ric impact term It−1 = 0, favorable information in the emissions venience yields series exhibit a linear mean-reversion process as
allowances markets brings ˛ multiple impacts on the conditional similar as other financial products markets. A number of com-
volatility t2 . When t−1 < 0, the asymmetric impact term It−1 = plex factors such as political decision, energy prices volatility,
1, unfavorable information in the emissions allowances markets stock markets and extraordinary temperature bring about the low
Table 3
Empirical results of convenience yield volatility using the ECM-GARCH (1,1) model.
e 0.9785*** (0.0101) 0.9871*** (0.0064) 0.9873*** (0.0053) 0.9936*** (0.0029) 0.9957*** (0.0025)
f −0.4571*** 0.0361 −0.3989*** 0.0328 −0.2941*** 0.0312 −0.1973*** 0.0395 −0.1611*** 0.0412
g 0.0070 0.0020 0.0003 0.0001 0.0003
˛1 0.2273*** (0.0288) 0.1445*** (0.0220) 0.0942*** (0.0141) 0.0789*** (0.0070) 0.0970*** (0.0098)
ˇ1 0.7602*** (0.0306) 0.8088*** (0.0299) 0.8938*** (0.0158) 0.9173*** (0.0072) 0.8894*** (0.0112)
Note: cy1 –cy5 denote the convenience yields for CO2 emissions allowances with different maturity from December 2010 to December 2014.
*Significance of the estimated coefficients at the significance levels of 90%. The numbers in parentheses are standard error values.
**Significance of the estimated coefficients at the significance levels of 95%. The numbers in parentheses are standard error values.
***
Significance of the estimated coefficients at the significance levels of 99%. The numbers in parentheses are standard error values.
K. Chang et al. / The Spanish Review of Financial Economics 11 (2013) 39–45 43
5
4.5
4
3.5
3 cyv1
cyv
2.5
2
1.5
1
0.5 Date
0 04-10-08
06-10-08
08-10-08
10-10-08
12-10-08
02-10-09
04-10-09
06-10-09
08-10-09
10-10-09
12-10-09
02-10-10
04-10-10
06-10-10
08-10-10
10-10-10
Fig. 3. Convenience yield volatility (cyv1 ) using the ECM-GARCH (1,1) model.
efficiency and overreaction in the emissions allowances market Figs. 3 and 4 show convenience yield volatility with dif-
(see Zhang and Wei, 2011; Montagnoli and Vries, 2010). These com- ferent maturity form December 2010 to December 2014.
plex markets factors exert greater market prices shock; however, Figs. 3 and 4 show that convenience yield volatility for CO2 emis-
prices shock effectiveness has a tremendous difference in time and sions allowances exhibits a time-varying trend, especially during
channel between emissions allowances spot and futures prices in the peak time from October 2008 to May 2009. Short-term volatil-
the immature emissions allowances markets. Current emissions ity is almost higher than long-term volatility in convenience yields.
allowances markets are weak effective markets, and they exist in Finally, the convenience yield volatility cyv1 increased sharply at
market bias, transaction cost and market overreaction. Unexpected the end of its maturity.
market information will have a different change speeds for spot Oh et al. (2008) study the long-term memory in various stock
and futures prices, emissions allowances markets exhibit a lead–lag market indices and foreign exchange rates using Detrended Fluc-
relationship between spot and futures. Based on the cost of carry tuation Analysis (DFA), and their results imply that the long-term
theory, the theoretical and actual futures prices have a dynamics memory property of the volatility time series can be attributed to
market deviation trends. When the above prices deviations are big- the volatility clustering observed in the financial time series. All the
ger than transaction cost, markets participants can achieve extra coefficients ˛1 + ˇ1 are approximately equal to 1. The lower ˛1 and
arbitrage revenues by adjusting assets portfolio between emissions higher ˇ1 indicate that convenience yields will eventually return
allowances spot and futures. to the equilibrium value. However, the volatility of convenience
yields will continue for a long time. These results show that the
5.2. Mean-reversion tests for CO2 convenience yield volatility convenience yields for emissions allowances exhibit strong volatil-
ity clustering effects which have an obvious persistent property.
Table 3 presents empirical results for CO2 convenience yield The above signs indicate that emissions allowances markets are
volatility using the ECM-GARCH (1,1) model with Gaussian pro- immature enough, so market participants frequently have to take
cess. Previous convenience yields and error-correction terms are speculative factors into account. As a result, emissions allowances
significant factors for current convenience yields at the significance assets exhibit a greater transaction risk. Asymmetric market infor-
level of 99%, and estimated coefficients slowly increase as time-to- mation exerts the unexpected market scarcity which pushes up
maturity increases. Current convenience yields exhibit a positive the higher volatility of spot and futures prices, so the emissions
correlation with previous convenience yields and a negative cor- allowances markets are difficult to eliminate the clustering effects
relation with error-correction terms. Estimated coefficients of of convenience yields.
convenience yields volatility are significant at the significance lever In order to achieve market arbitrage, market participants opti-
of 99%, so convenience yield volatility exhibits a mean-reversion mize their assets portfolio policies between spot and futures
process. The coefficients of the GARCH model are significant, and using the clustering effects of convenience yields. Wang et al.
the standard errors are approximately equal to 0. These results (2012) find that convenience yields have a significant call and put
show that convenience yields for emissions allowances exhibit options feature, and market participants achieve distinct options
obvious ARCH effects. values by flexibly adjusting assets portfolio policies between spot
0.3 cyv2
cyv3
0.25 cyv4
cyv5
0.2
cyv
0.15
0.1
0.05
Date
0
04-10-08
06-10-08
08-10-08
10-10-08
12-10-08
02-10-09
04-10-09
06-10-09
08-10-09
10-10-09
12-10-09
02-10-10
04-10-10
06-10-10
08-10-10
10-10-10
12-10-10
Fig. 4. Convenience yield volatility (cyv2 –cyv5 ) using the ECM-GARCH (1,1) model.
44 K. Chang et al. / The Spanish Review of Financial Economics 11 (2013) 39–45
Table 4
Empirical results for convenience yield volatility using the ECM-TARCH (1,1) model.
h 0.0070*** (0.0027) 0.0014*** (0.0005) 0.0003** (0.0001) 0.0002** (0.0001) 0.0003** (0.0001)
˛ 0.1484*** (0.0543) 0.0467** (0.0287) 0.0374** (0.0192) 0.0335** (0.0181) 0.0321** (0.0169)
ˇ1 0.7514** (0.1032) 0.8571*** (0.0353) 0.9189*** (0.0207) 0.9260*** (0.0163) 0.9175*** (0.0184)
1 0.2073** (0.0462) 0.1321** (0.0573) 0.0636** (0.0342) 0.0579** (0.0334) 0.0636** (0.0331)
**
Significance of the estimated coefficients at the significance levels of 95%. The numbers in parentheses are standard error values.
***
Significance of the estimated coefficients at the significance levels of 99%. The numbers in parentheses are standard error values.
and futures. Our empirical results show that the convenience overreaction in spot and futures prices push the greater conve-
yields have obvious time-varying trends and clustering effects. nience yields volatility, and then distort convenience yields of
The overreaction of emissions allowances markets brings markets emissions allowances. Based on the higher degree of liquidity in
participant about many arbitrage opportunities, so short-term mar- the spot market relative to the futures market, these pessimism
ket speculations are active. When convenience yields are positive, overreactions have more quick transition speed of spot prices rela-
the convenience yields are call options, and market participants tive to futures prices. The higher convenience yield volatility brings
make assets portfolio policy of holding spot contracts while sell- market hedgers about the greater arbitrage returns through adjus-
ing futures contracts. When convenience yields are negative, the ting assets portfolio between spot and futures, so these uncertain
convenience yields are put options, and market participants make factors promote more rampant speculation. In summary, favorable
assets portfolio policy of holding futures contracts while selling and unfavorable market information has an asymmetric impact on
spot contracts. The above assets exchange policies can achieve extra the convenience yield volatility.
arbitrage revenues.
6. Conclusion
5.3. Asymmetric behavior of CO2 convenience yield volatility
CO2 emissions allowances markets in the EU ETS are emerg-
Daal et al. (2007) propose asymmetric GARCH-Jump models that ing financial markets. Previous results show that mean-reverting
synthesize autoregressive jump intensities and volatility feedback convenience yields are central to forecasting commodity prices
in the jump component, and then capture several distinguishing and estimating hedging returns. Based on the simple cost-of-carry
features such as more volatility persistence, less leverage effects. model, we examine the properties of convenience yields for CO2
Yamamoto (2010) documents the empirical research on asymmet- emissions allowances futures contracts with different maturities
ric volatility and volatility clustering in stock markets, and the from December 2010 to December 2014. We find that the conve-
empirical results show that herding matters for volatility clustering nience yields for CO2 emissions allowances exhibit a time-varying
while a borrowing constraint intensifies the asymmetry of volatility trend, and standard deviation in the convenience yields declines
through the herding effect. with an increase in time-to-maturity. All skewnesses and kurtoses
All the estimated coefficients in Eq. (5) are significant at the 95% show that convenience yields exhibit left-skewed and leptotur-
confidence level, and all the standard errors are minor. The con- tossis distribution trends.
ditional volatility in the CO2 convenience yields mainly depends We test mean-reversion and volatility in the stochastic con-
on the last period’s residual errors, the asymmetric impact term venience yields for CO2 emissions allowances by using the ADF,
and the last period’s conditional volatility. Especially favorable ECM-GARCH and ECM-TARCH models. Empirical results from the
and unfavorable market information has an asymmetric impact ADF tests show that all estimated b coefficients are significant
on the conditional volatility. In Table 4, all the
1 coefficients of at the 95% confidence level, and b < 0 implies that all the conve-
asymmetric terms are obviously bigger than zero, the convenience nience yields follow a linear mean-reverting process. Empirical
yield volatility (cyv) has an obvious leverage effect, so the conve- results from the ECM-GARCH (1,1) model show that the last period’s
nience yield volatility exhibits asymmetric dynamics. Positive
1 convenience yields and error correction terms significantly affect
coefficients show that asymmetric leverage effects increase the current convenience yields, and that convenience yield volatility
convenience yield volatility. Take cyv1 for example, when emis- exhibits a mean-reversion process. We also detect that convenience
sions allowances markets receive favorable information, t−1 ≥ 0, yield volatility exhibits asymmetric dynamics by using the ECM-
and these favorable information brings 0.1484 times impacts on TARCH (1,1) model. Asymmetric leverage effects increase volatility,
the cyv1 . When emissions allowances markets receive unfavor- unfavorable market information has a greater impact on volatility
able information, t−1 < 0, these unfavorable information brings than favorable market information, and unfavorable market infor-
cyv1 an impact of 0.3557 times. Unfavorable market information mation has a lower effect on long-term volatility in convenience
has a higher impact on convenience yield volatility. Estimated yields.
1 coefficients of cyv1 –cyv4 decline with an increase in time-to- The leaner mean-reversion property and asymmetric volatility
maturity, and unfavorable market information has a lower impact of convenience yields provides much enlightenment for govern-
on the long-term volatility of convenience yields. ment regulators. Effective macro-control and macro-regulation
The complex factors such as Strict quota allocation rules and is necessary to improve markets efficiency in the emissions
regulation policies of carbon emissions, long-term dependency of allowances markets. Macro-controlling failure, uncertain decision
fossil energy and extreme temperature are favorable market infor- and inactive trading volumes have significant impacts on the
mation, the complex factors such as negotiation failure of global leaner mean-reversion phenomenon and asymmetric volatility
greenhouse emissions, fossil energy prices volatility, developing of convenience yields. Government regulators should strengthen
block in low-carbon technology are unfavorable market informa- international governments’ operation and communication, estab-
tion. The favorable market push spot and futures prices upwards, lish scientific emissions reduction plan and strict emissions quota
increasing speeds of spot prices are higher than futures prices. The allocation rules. EU Government should make consistent and
unfavorable market information brings about greater pessimistic systematic decision in the greenhouse emissions reduction,
demands in the weak effective emissions markets. The pessimistic support regime-switching behaviors among different emissions
K. Chang et al. / The Spanish Review of Financial Economics 11 (2013) 39–45 45
markets, and strengthen international cooperation in the green- Daal, E., Naka, A., Yu, J.S., 2007. Volatility clustering, leverage effects, and jump
house emissions reduction. EU government should present an dynamics in the US and emerging Asian equity markets. J. Bank. Finance 31,
2751–2769.
international information network system of emissions trading Daskalakis, G., Psychoyios, D., Markellos, R.N., 2009. Modeling CO2 emission
markets, market participants timely capture demand and supply allowance prices and derivatives: evidence from the European trading scheme.
information, motoring and checking emissions reduction informa- J. Bank. Finance 33, 1230–1241.
Engle, R.F., Granger, C.W.J., 1987. Cointegration and error correction: representation,
tion, etc., and they can achieve fair, transparent and symmetric estimation and testing. Econometrical 55, 251–276.
emissions markets. Gibson, R., Schwartz, E.S., 1990. Stochastic convenience yield and the pricing of oil
Emissions investors, hedgers, arbitragers should be aware of risk contingent claims. J. Finance 45, 959–976.
Glosten, L.R., Jagannathan, R., Runkle, D., 1993. On the relation between the expected
reduction between emissions allowances spot and futures, regime
value and the volatility of the Nominal excess return on stocks. J. Finance 48,
switches and threshold effects when attempting to forecast both 1779–1801.
spot and futures prices. Market participants can flexibly adjust Granger, C.W.J., Teräsvirta, T., 1993. Modelling Nonlinear Economic Relationships.
Oxford University Press, Oxford.
their assets portfolio trading policies between spot and futures
Heaney, R., 2002. Approximation for convenience yields in commodity futures pri-
and then achieve additional arbitrage returns by using the conve- cing. J. Future Markets 22, 1005–1017.
nience yields. An active hedging strategy involving spot and futures Liu, P., Tang, K., 2011. The stochastic behavior of commodity prices with
markets seems to be of interest in asymmetric leverage effects of heteroskedasticity in the convenience yield. J. Empirical Finance 18,
211–224.
convenience yield volatility in order to protect against unfavorable Mcmillan, D.G., Ulku, N., 2009. Persistent mispricing in a recently open emerging
movements in emissions allowances assets prices. index futures market: arbitrageurs invited. J. Future Markets 29, 218–243.
Our empirical results help market participants to capture the Miltersen, K.R., Schwartz, E.S., 1998. Pricing of options on commodity futures with
stochastic term structures of convenience yields and interest rates. J. Financ.
optimal hedge ratios and adjust portfolio sizes between spot and Quant. Anal. 33, 33–59.
futures, so market participants can enhance their risk management Milunovich, G., Joyeux, R., 2010. Market efficiency and price discovery in the EU
capabilities for emissions allowances. Future research on emis- carbon futures market. Appl. Financ. Econ. 20, 1–28.
Montagnoli, A., Vries, F.P., 2010. Carbon trading thickness and market efficiency.
sions allowances should analyze the term structure of convenience Energy Econ. 32, 1331–1336.
yields, construct optimal hedge ratios by using convenience yields, Oh, G., Kim, S., Eom, C., 2008. Long-term memory and volatility clustering in high-
and address GARCH option pricing for emissions allowances. frequency price changes. Physica A 387, 1247–1254.
Pindyck, R.S., 1999. The long run evolution of energy prices. Energy J. 20, 1–27.
Schwartz, E.S., 1997. The stochastic behavior of commodity prices: implications for
References valuation and hedging. J. Finance 52, 923–973.
Schwartz, E.S., Smith, J.E., 2000. Short-term variations and long-term dynamics in
Arouri, M.E.H., Jawadi, F., Nguyen, D.K., 2012. Nonlinearities in carbon spot-futures commodity prices. Manage. Sci. 46, 893–911.
price relationships during chase II of the EU ETS. Econ. Model. 29, 884–892. Seifert, J., Homburg, M.U., Wagner, M., 2008. Dynamic behavior of CO2 spot prices.
Benz, E., Truck, S., 2006. CO2 emission allowances trading in Europe-Specifying a J. Environ. Econ. Manage. 56, 180–194.
new class of assets. Prob. Perspect. Manage. 4, 30–40. Taylor, N., 2004. Trading intensity, volatility, and arbitrage activity. J. Bank. Finance,
Benz, E., Trück, S., 2009. Modeling the price dynamics of CO2 emission allowances. 1137–1162.
Energy Econ. 31, 4–15. Wang, S.S., Chang, K., Liu, Y., Li, Z.C., 2012. The analysis of convenience yields and
Bernard, J.B., Khalaf, L., Kichian, M., McMahon, S., 2008. Forecasting commodity options value for carbon emissions. J. Syst. Manage. 21, 552–558.
prices: GARCH, jumps, and mean reversion. J. Forecast. 27, 279–291. Working, H., 1949. The theory of the price of storage. Am. Econ. Rev. 39, 1254–1262.
Bilson, C., Brailsford, T., Evans, T., 2005. The international transmission of arbitrage Yamamoto, R., 2010. Asymmetric volatility, volatility clustering, and herding agents
information across futures markets. J. Bus. Finance Account. 32, 973–1000. with a borrowing constraint. Physica A 389, 1208–1214.
Brennan, M., 1958. The supply of storage. Am. Econ. Rev. 48, 50–72. Zakoian, J.M., 1994. Threshold heteroskedastics models. J. Econ. Dyn. Control 18,
Chevallier, J., 2010. Modelling risk premium in CO2 allowances spot and futures 931–955.
prices. Econ. Model. 27, 717–729. Zhang, Y.J., Wei, Y.M., 2011. Interpreting the mean reversion of international carbon
Cortazar, G., Schwartz, E.S., 2003. Implementing a stochastic model for oil future futures prices: empirical evidence from the EU ETS. Syst. Eng. Theor. Pract. 31,
prices. Energy Econ. 25, 215–238. 214–220.