Understanding the Link Between Aggregated Industrial Production and the Carbon Price - Green Energy and Efficiency: An Economic Perspective - page 119

carbon prices with a lag, due to the speci c institutional constraints of this envi-

ronmental market. Besides, the Markov regime-switching model captures most of
the shocks identi

ed on the carbon market (January-April 2005, April-June 2006,
October 2008, and April 2009-present). In line with previous studies, no statisti-
cally signi cant impact going from the carbon price to industrial production could

be detected (i.e., there is no effect). The results are robust to the

introduction of energy dynamics (e.g., Brent, gas, coal).
While there seems to remain considerable uncertainties regarding the evolution
of this carbon-macroeconomy relationship in phase III of the EU ETS with the shift
to auctioning and the need to meet the EC 20/20/20 targets, the bottom line of this
work can be summarized as follows. The carbon-macroeconomy relationship seems
adequately captured by two-regime threshold error-correction and two-regime
Markov-switching VAR models compared to linear models as main competitors.
Finally,

“ bounce back ”

“ mixed equity/industrial production ” econometric strategies can also

be found:

￿ Bredin and Muckley [ 5 ] have used the industrial production index computed by

Eurostat to capture the in uence of economic activity in their equilibrium model

of phase II carbon prices (including as well energy prices, equity prices and
temperatures deviations). The nancial markets indicator used is the Eurex Dow

Jones Euro Stoxx futures contract. According to the authors, the motivation for
including this variable is that it offers an up-to-date indicator of expectations on
both nancial and economic conditions at the required daily frequency. Further,

given the

nancial nature of the underlying asset, they consider including such a
proxy informative. We can certainly agree upon that statement concerning the
bene ts of a mixed nancial/macro approach.

Mansanet-Bataller [ 24 ] have used the industrial production index calculated by
Tendances Carbone , in conjunction with

￿

nancial indicators such as the EU
Economic Sentiment Index, the slope of the Euro area yield curve, the Reuters
momentum variable concerning the EUA market, and the CBOE VIX volatility
indicator. During phase II of the EU ETS, statistical signi

cance could only be

found for the EUA momentum variable.

Table 2 provides a useful summary of the categories of indicators used in
previous studies as proxy for economic activity:

4 Empirical Analysis

In our empirical analysis, we wish to develop a Threshold VAR model (TVAR)
applied to the carbon-macroeconomy relationship. The focus here is to study the
inter-relationships between the EU 27 industrial production index and the price of
CO 2 in a nonlinear framework.
The necessity to adopt such a methodological viewpoint compared to early
studies

— which were essentially based on linear regressions — has been further