Marginal Abatement Cost Curves: Combining Energy System Modelling and Decomposition Analysis

Abstract

Marginal abatement cost (MAC) curves are a useful policy tool to communicate findings on the technological structure and the economics of CO₂ emissions reduction. However, existing ways of generating MAC curves do not display consistent technological detail and do not consider system-wide interactions and uncertainty in a structured manner. This paper details a new approach to overcome the present shortcomings by using an energy system model, UK MARKAL, in combination with index decomposition analysis. In addition, this approach allows different forms of uncertainty analysis to be used in order to test the robustness of the MAC curve. For illustration purposes, a sensitivity analysis concerning fossil fuel prices is applied to the transport sector of the UK. The resulting MAC curves are found to be relatively robust to different fuel costs at higher CO₂ tax levels. The new systems-based approach improves MAC curves through the avoidance of an inconsistent emissions baseline, the incorporation of system-wide interactions and the price responsiveness of demand.

Appendix

The general decomposition formula of the Divisia index in the logarithmic mean specification is defined as follows:

$$ \Delta \;{\text{C}}{{\text{O}}_{{2,x}}} = \sum\limits_i {\frac{{{\text{CO}}_{{2,i}}^C - {\text{CO}}_{{2,i}}^0}}{{\ln \;{\text{CO}}_{{2,i}}^C - \ln \;{\text{CO}}_{{2,i}}^0}} \times \ln \left( {\frac{{x_i^C}}{{x_i^0}}} \right)} $$

(5)

In this context, x is a factor that drives CO₂ emissions, e.g. fuel intensity, and i is a criterion for structural differentiation. The superscript 0 and C represent a base scenario and a CO₂ reduction scenario.

In the example for the UK transport sector, the activity effect is calculated as follows:

$$ {\text{Activity}}\;{\text{effect}} = \sum\limits_{{i = {\text{vehicle}}\;{\text{type}}}} {\sum\limits_{{j = {\text{technology}}}} {\frac{{{\text{CO}}_{{2i,j}}^C - {\text{CO}}_{{2i,j}}^0}}{{{\text{ln CO}}_{{2i,j}}^C - {\text{ln CO}}_{{2i,j}}^0}} \times { \ln }\left( {\frac{{a_i^C}}{{a_i^0}}} \right)} } $$

(6)

The activity effect describes reductions in CO₂ emissions related to changes in the demand for energy services. As demand is assumed to be price elastic, it will decrease if the price for an energy service increases as a result of carbon policies. For example if the demand for car travel decreases, the emissions saved will fall under this category.

The structure effect is calculated as follows:

$$ {\text{Structure}}\;{\text{effect}} = \sum\limits_{{i = {\text{vehicle}}\;{\text{type}}}} {\sum\limits_{{j = {\text{technology}}}} {\frac{{{\text{CO}}_{{2i,j}}^C - {\text{CO}}_{{2i,j}}^0}}{{{\text{ln CO}}_{{2i,j}}^C - {\text{ln CO}}_{{2i,j}}^0}} \times { \ln }\left( {\frac{{s_{{i,j}}^C}}{{s_{{i,j}}^0}}} \right)} } $$

(7)

The structure effect as specified in Eq. (7) highlights the CO₂ emission reduction due to a shift in technologies satisfying transport demands. Emission savings related to this effect occur due to a reduction of the relative part of carbon intensive measures in the technology mix. However, it is more interesting to see what technologies are chosen instead of the carbon intensive ones. Therefore, the emission reduction associated with the reduced use of a carbon intensive technology is redistributed to less carbon intensive technologies satisfying a higher part of transport demands. In an example where 5 % of all petrol cars are substituted for electric cars, the emission reduction is not attributed to the lower use of petrol cars, but to the higher use of electric cars.

The fuel intensity effect is calculated as follows:

$$ \matrix{ {{\text{Fuel}}\;{\text{intensity}}\;{\text{effect}}} \hfill &{} \hfill \\ {} \hfill &{ = \sum\limits_{{i = {\text{vehicle}}\;{\text{type}}}} {\sum\limits_{{j = {\text{technology}}}} {\frac{{{\text{CO}}_{{2i,j}}^C - {\text{CO}}_{{2i,j}}^0}}{{{\text{ln CO}}_{{2i,j}}^C - {\text{ln CO}}_{{2i,j}}^0}} \times { \ln }\left( {\frac{{f_{{i,j}}^C}}{{f_{{i,j}}^0}}} \right)} } } \hfill \\ }<!end array> $$

(8)

The fuel intensity effect represents the impact of changes in the fuel efficiency of technologies on CO₂ emissions. If a more efficient car type, which still uses the same fuel, but can travel the same distance with less fuel is chosen by the model, the resulting decrease in emissions is recorded as a fuel intensity effect.

The carbon intensity effect is calculated as follows:

$$ \matrix{ {{\text{Carbon}}\;{\text{intensity}}\;{\text{effect}}} \hfill &{} \hfill \\ {} \hfill &{ = \sum\limits_{{i = {\text{vehicle}}\;{\text{type}}}} {\sum\limits_{{j = {\text{technology}}}} {\frac{{{\text{CO}}_{{2i,j}}^C - {\text{CO}}_{{2i,j}}^0}}{{{\text{ln CO}}_{{2i,j}}^C - {\text{ln CO}}_{{2i,j}}^0}} \times { \ln }\left( {\frac{{c_{{i,j}}^C}}{{c_{{i,j}}^0}}} \right)} } } \hfill \\ }<!end array> $$

(9)

The carbon intensity effect refers to changes in the carbon intensity of fuels. The carbon intensity of diesel can be decreased by blending in biodiesel and the carbon intensity of electricity can be reduced by switching towards CCS technologies, renewables or nuclear in the electricity sector.