China Policy Journal Volume 1, Number 1, Fall 2018 | Page 45

China Policy Journal
Note: price volatility is estimated by the standard deviation of price returns.
Figure 2. Price Volatility of CEA in Seven ETS Pilots
to operate better than other city-level
ETSs, with their high-level CEA prices
and active trading activities. Regarding
provincial level ETS, Guangdong
has higher CEA price, while Hubei has
lower volatility and larger weekly trading
volume. Chongqing and Tianjin
ETSs are not so successful considering
their low CEA prices and scarce trading
transactions. Therefore, we do not
include Chongqing and Tianjin in the
regression analysis, as their CEA prices
may be artificially set in the weeks with
zero transactions, not really relating to
the market demand and supply.
Table 5 contains the descriptive
statistics of crude oil price, coal price
and Shanghai Shenzhen 300 stock index
during 2013 week 25–2017 week
26, and LNG price data during 2014
week 1–2017 week 26. Shanghai Shenzhen
300 stock index has the largest
volatility because of the nature of the
stock index data.
3.2. Unit Root Tests and
Johansen’s Tests
We took logarithmic forms of all price
variables and tested their unit roots.
Table 6 displays the results. Not all of
the logarithmic forms of the variables is
integrated of order zero, meaning that
the logarithmic forms of some variables
are not stationary. The logarithmic differences
of all CEA price variables (i.e.
price returns) are stationary, and the
logarithmic differences of energy prices
and the stock index are also stationary.
Thus, we use logarithmic differences of
the price variables throughout the following
regression analysis.
We applied Johansen’s tests for
I(1) logarithmic price variables that
are integrated of order one, including
Chongqing CEA price, Hubei CEA
price, Shanghai CEA price, Brent oil
price and LNG price. We used the tests
for every two I(1) logarithmic variables
42