Macroeconomic Impacts of Carbon Capture and Storage in China

Abstract

Carbon capture and storage (CCS) is a key technology for reducing greenhouse gas emissions. But a CCS facility consumes vast amounts of energy and capital. With this in mind we analyze macroeconomic consequences of a large scale introduction of CCS in China. We modify and extend the DRC-CGE, a macroeconomic CGE model of the country that is used for long-term planning and policy analyses. We analyze an internal finance scenario of domestic funding, and an external finance scenario of international funding. In the external finance scenario CCS is installed on 70 % of all power plants by 2050. This increases demand for coal in 2050 by one fifth and import of coal by one fourth. The strain on coal resources may be an important political concern for China. In the internal finance scenario coal resources are not strained since this scenario introduces a price on carbon that lifts prices of energy. Moreover, the price on carbon cuts across the board and the internal finance scenario is much more effective at reducing $\hbox {CO}_{2}$. On the other hand, in this scenario GDP goes down about 4 %, which also raises political concern.

Appendix

1.1 Splitting Electricity Production by Fuels

Electricity is most often generated at a power station by electromechanical generators, primarily driven by heat engines fueled by chemical combustion or nuclear fission, but also by other means such as the kinetic energy of water flows and wind. There are many technologies that can be and are used to generate electricity such as solar photovoltaics and geothermal power.

However, in order to utilize China’s IO data² for CCS analysis, it is necessary to split power generation by technology (fuel source). Based on the available data we introduce five categories for power generation: coal-fired electricity, non-coal fossil electricity, nuclear electricity, hydro-electricity and other electricity.

Because the IO data base must satisfy a numbers of key equilibrium conditions in all 42 sectors, we cannot simply go into the data base and alter a given set of flows without destroying one or more of these equilibrium conditions. According to energy data and cost data for power generation from IEA, a cross entropy method is used to share the single electricity production out among different technologies.

1.2 Methodology

The cross entropy method is an approach that originates from information theory (Shannon, 1948) and then was brought to economics (Theil 1967). In information theory, the cross entropy between two probability distributions measures the average number of bits needed to identify an event from a set of possibilities, if a coding scheme is used based on a given probability distribution, rather than the “true” distribution.

In the 1990s, the cross-entropy method was applied to estimation of input–output tables and social accounting matrixes. Golan et al. (1994) use the method to estimate the coefficients of an input–output table. Robinson et al. (2000) and Robinson and El-Said (2000) use the cross entropy to update and estimate a social accounting matrix. In general, the cross entropy method is a method of solving underdetermined estimation problems, using only and all the information available.

In this study we follow the works of Golan et al. and Robinson et al. to estimate the input matrix for electricity production by technology.

1.3 Estimating Disaggregate Electricity Data

Table 5 shows a schematic production matrix of electricity by fuel for China. $M_{s\times el}$ is the input matrix for different power generation technologies. The index el covers all types of electricity generation: coal-fired electricity, non-coal fossil electricity, hydro-electricity, nuclear electricity and other electricity. Index $s$ refers all inputs: 42 intermediate input commodities,³ 2 factors and 1 net product tax (production tax minus production subsidy). We just know the aggregate cost structure of electricity $(R)$ with China’s IO table and do not know $M$ and $C$. $C$ will be calibrated first. Then cross entropy method is used to estimate $M$.

Table 5 a schematic production matrix of electricity by fuel for one country

Fig. 9

Composition of electricity production by fuel source (2007). Source: NBS (2011). (Color figure online)

1.4 The Estimation of Output for Electricity by Fuel

With the IEA database and China’s energy statistical yearbook, it’s easy to find the electricity production in terms of physical volume, by country and fuel. Figure 9 shows the composition of electricity production by fuel type in 2007.⁴ The following formula is used to calibrate the output of electricity in China from different fuels:

$$\begin{aligned} C_{el} =\frac{\delta _{el} E_{el} }{\sum _{el} {\left( {\delta _{el} E_{el} } \right) } }X^{ely} \end{aligned}$$

(1)

Here, $E_{el}$ is the production of electricity by fuel in terms of physical unit (MWh) and comes from IEA database. $\delta _{el}$ is the cost disparity coefficient for different electricity categories, see Table 3 in the main text. $X^{ely}$ is the aggregated output of electricity in the IO table. $C_{el}$ is the output of electricity by fuel in terms of value (US$).

1.5 Estimating the Production Matrix

Because the cross entropy method begins with prior information we should find some additional information except C and R. It would be ideal to find the same cost structure data for each type of electricity generation. Unfortunately, we just find simple cost structure data for each type of electricity generation. According to the data, the unit cost of electricity generation includes three parts: investment, operation and maintenance (O&M) and fuel costs. In order to utilize the cost data to estimate the production matrix of electricity by fuel, it is necessary to bridge the gulf between two classifications (IEA and China’s IO table). We set up the correspondence between these two classifications (see Fig. 10).

Fig. 10

the correspondence between IEA cost and China’s IO table

Now the cost share data from IEA can serve as the prior information in our estimation. The cross entropy method problem is to derive the estimated production matrix that minimizes the Kullback–Leibler measure of the “cross entropy” distance between the prior cost information and the estimated cost information. The equilibrium conditions in SAM should also hold. The cross entropy problem is as follows:

$$\begin{aligned} \min \,\, \textit{entropy} = \sum _{el} {\sum _{ct} {\left[ {\gamma _{el} \cdot \bar{{\alpha }}_{el,ct} \cdot \ln \left( \frac{\alpha _{el,ct} }{\bar{{\alpha }}_{el,ct} }\right) } \right] }} \end{aligned}$$

Subject to:

$$\begin{aligned}&\sum _s {M_{s,el} } =C_{el}\end{aligned}$$

(2)

$$\begin{aligned}&\sum _{el} {M_{s,el} } =R_s\end{aligned}$$

(3)

$$\begin{aligned}&\sum _{f:s\rightarrow ct} {\frac{M_{s,el} }{C_{el} }} =\alpha _{el,ct}\end{aligned}$$

(4)

$$\begin{aligned}&M_{s,el} \ge 0 \end{aligned}$$

(5)

The index ct covers three cost categories: investment, operation and maintenance (O&M), and fuel costs. $f:\;s\rightarrow ct$ is the image between the IEA categories (ct) and input–output categories (s) defined in Fig. 10. The output share for each electricity category in total electricity output $(\gamma _{el} =\frac{C_{el} }{\sum _{el} {C_{el} } })$ is incorporated into the cross entropy as weights. $\bar{{\alpha }}_{el,ct}$ is the “prior” cost share and is calculated based on IEA cost data.

As for constraint (2), the sum of all inputs must in the base year be equal to the output for each electricity category. In the constraint (3), for a given input (intermediate commodities or factors), the sum of inputs over all electricity categories must be equal to the aggregate input in the IO table. Equation (4) gives the definition of cost share in terms of investment, O&M and fuel, based on the estimated production matrix. The last constraint means that each cell in the production matrix is non-negative. To solve the problem, prior cost information and the aggregate cost structure of electricity in IO table are used to initialize $M_{s,el}$. With GAMS we solve the cross entropy problem.