On Line Supporting Material
Extended Description of GCAM and Approach
For this analysis we use the Global Change Assessment Model (GCAM, Calvin, et al., 2011). GCAM is a member of a class of models called integrated assessment models (IAMs). IAMs simulate the behavior of a range of economic agents making decisions about the supplies of and demands for various resources. GCAM, like other IAMs, has energy, agriculture, land use, land cover, macro-economy and climate modules. Each of the modules in the human system contains a description of the physical flows of resources such as coal, oil, gas, uranium, wind, solar insolation, or geothermal, as they move from a primary state through various transformation, for example, refined fuels, electricity or hydrogen steps until they reach their final uses, such as building services (e.g. heating, cooling, cooking, water heating,), industrial services (process heat, steam, mechanical drive, direct electrical services), or transport of goods or passengers. Those flows are motivated at every step by the prospect of economic gain and self-interest. Bioenergy, is produced as either a purpose-grown crop (e.g. corn, palm oil, switch grass, woody biomass) competing with other land uses such as food and fiber crops, pastures, forests, urban areas or other unmanaged ecosystems. The role of the IAM is to project the demand for and supply of resources at each step along the way. Demands and supplies are reconciled by the market, which sets a price such that total demands match total supplies at every step along the way. Since these systems are highly non-linear, much of the work of an IAM goes into finding the set of prices that brings all of the systems into balance.
Since we will focus much of our attention in this paper on the production of electricity, a short description of how GCAM determines the amount of electricity to produce and the choice of fuels with which to produce it. The total amount of electricity to be produced depends on the price of electricity, which in turn determines the amount consumers are willing to buy. Total production comes from both existing vintages of power production capacity and from investments in new plant and equipment. Throughout their lifetimes the historical vintages always produce as long as the price of electricity is sufficiently high to cover operations, that is the capital costs are assumed to be “sunk costs” and therefore irrelevant to the production decision for existing plants. The portfolio of new investments however depends on the expected cost of producing power, including a normal return on investment to the purchasers. Less expensive sources of power receive a larger share of new investment. But, GCAM employs a probabilistic approach to projecting the share of the market and the lowest cost source of new power does not capture the entire investment portfolio, merely the largest share.
Demands for and supplies of all of the resources in the model depend on the external conditions provided to the IAM from outside. The most important externally provided information supplied to an IAM can be grouped into the following categories: the size and composition of the population, the scale of economic activity (GDP), the suite of availability technologies, and policies. Given those externally provided inputs, the model projects the state of the economy in terms of supplies of primary energy, transformation of those primary energy forms into forms useful to consumers, and the uses to which the energy was put. Outputs of GCAM include a long list of energy detail, ranging from primary energy production, by type and region of the world, production of intermediate energy forms such as electricity, refined petroleum products, and hydrogen, including the inputs to their production and total production, and finally fuel use for the wide range of end use applications such as transport of goods and passengers, building services and industrial processes, by region and year. The detailed technology representations included in GCAM are used to calculate detailed emissions projections for greenhouse gases (CO2, CH4, N2O, and manufactured gases), aerosols (sulfur, black and organic carbon), and short-lived pollutants (CO, NO2, and other short-lived species). GCAM’s default time step is five years and it is frequently used to explore scenarios to the year 2100, though its time step can be shorter or longer as can its time horizon.
The purpose of this paper is to explore how the introduction of a specific policy, namely a set of rules governing the rewards for undertaking specific actions under a hypothetical and generic “offsets” program affects the behavior of economic agents in the model, and beyond that, how those changed behaviors play through the larger economic system. GCAM has sufficient detail to allow analysis of the behavior of agents at the scale of power generators undertaking actions under an offsets program, but also a fully described system in which the individual decision makers interact with each other and the larger set of economic agents.
This allows us to take each agent and present them with alternative program designs and prices in which they can participate as much or as little as they would like. So, for example, an electric utility in China might be offered the opportunity to be paid a price for every kWh of new power generation for which CO2/kWh is lower than the average for the power sector in the GCAM reference scenario (without offsets). Depending on the price to be paid for expanded deployment of such facilities, the utility will determine how much new program compliant capacity to bring onto the system. Obviously the payment affects the cost of producing power from facilities in compliance with the offsets program and the amount of electricity that can profitably be sold. In the sections that follow we explore varying offset prices and program rules for reimbursement.
We start with the GCAM reference scenario (without offsets) and use the set of assumptions about population, GDP, technology availability, and policy taken from Thomson, et al. (2011) as our set of externally given conditions. The only assumptions that go into creating the reference case that we will change in this paper are the policies governing offsets.
We will use the GCAM reference scenario to provide the crediting baseline, we then calculate three sets of results corresponding to various offset prices:
1. Offsets credits supply from regions which we assume have no emissions limits and eligible to create offsets with certified emissions reductions (CER) certificates (China, India, Other South and East Asia, Mideast, Africa, Latin America).
2. Global emissions with the offsets program in place, from which net emissions abatement are computed relative to the Reference Case.
3. Economic potential emissions mitigation using the MAC approach a la Richels, et al. (1996).
We initially assume an offsets program based on the “Operating Margin” (OM) method and will then compare this approach to the “Combined Margin” (CM) method in Section 4. The OM method is summarized in Table 1. The value of the emissions reductions to the seller is the product of the calculated reductions and the price of carbon. This value is passed on to the consumer in the form of lower energy prices. We assume that monitoring, verification and transaction costs are negligible.
We assume that offsets can only be credited for actions that would not have been undertaken without the offsets program. We estimate the magnitude of CERs that might be introduced into an offsets market by region in the year 2020 for a program initiated in 2015.
Table 1: Offset Crediting Protocol for 2020 Using Operating Margin MethodSectoral Coverage / Electric Utilities
Eligible Activities / New power generation facilities for which CO2/kWh is lower than the average for the power sector in the GCAM reference scenario (without offsets). Only technology deployment above the GCAM reference scenario is eligible.
Offset Supplying Regions / China, India, Other south and east Asia, the former Soviet Union, Mideast, Africa, Latin America
Program Start Date / January 1, 2015
Year for which potential supply is estimated / 2020
Offset Calculation / Difference between facility emissions (CO2/kWh) and power sector average in the reference scenario, times the facility’s power produced.
Alternative baseline formulations are explored in Section 5.
Microeconomic Illustration of the difference between an offset subsidy and tax on electricity price, production, and factor input selection
Another way to understand the forces driving our result is from a microeconomic perspective. Assume for the moment that there are only two inputs to produce electricity, renewable energy and fossil fuel energy. Figure 1, Panel A shows the cost-minimizing combination of renewables and fossil fuels, which is found where the electricity production isoquant (combination of all sets of renewable and fossil fuel inputs that can deliver the same level of electricity production) is just tangent to the equal cost line (TCref, all combinations of renewables and fossil fuel inputs that have equal cost), point R is the cost minimizing combination of renewable and fossil technologies, Rref and Fref for producing electricity output, Elec 2.
Panel B shows an analysis for a fixed output. The effect of the subsidy on renewable energy changes the relative cost of producing power between renewable and fossil inputs, shifting in favor of renewable energy. This reduces emissions by the amount fossil energy input use declines (times its emissions coefficient). The amount of credit is the increase in renewable energy deployment times its crediting rate. While there need not be a one-to-one correspondence between crediting rate and emissions reductions, the energy system shifts in the right direction.
Panel C shows the consequence of imposing a carbon tax on the fossil fuel. The tax changes the slope of the total cost curve in favor of renewable energy such that for the same total cost (TCref) reducing the amount of fossil fuel that can be purchased and electricity that can be produced for a given total cost (TCtax). The cost has not changed, but less electricity can be produced at that same cost (Elec 1 as opposed to isoquant Elec 2). Thus relatively more renewable energy is used, however, less electricity is used, reducing the input of both renewable and fossil fuel energy in absolute terms. Of course, there is no reason to believe that total cost will be exactly the same as initially. It could be higher or lower, depending on the price elasticity of electricity demand response and competition in end-use sectors between electricity and the direct use of fossil fuels. However, if the carbon tax is applied to the power sector alone, the negative slope of the aggregate electricity demand function guarantees that less power will be produced than in the reference scenario.
Panel D shows the comparative static consequences of offering a credit for deploying renewable energy. As was the case with the carbon tax, the offset credit shifts the relative cost of renewable versus fossil fuel inputs in favor of renewable energy. However, the relative cost shift is accomplished by lowering the renewable energy price, which in turn lowers the cost of producing electricity. For any given total cost more power can be produced but the cost-minimizing combination of fossil and renewable energy shifts in favor of renewable energy (Elec 2 to Elec 3 in Panel D). The loser cost of power generation resulting from the subsidy will increase the demand for electricity, if the subsidy is applied only to the power sector, because the electricity demand curve has a negative slope, though whether total cost is higher or lower than in the reference scenario is an empirical question. Credits are generated by the incremental increase—there is no credit given for deployments that would have occurred in the reference scenario. (See green arrow in Panel D.) However, the expansion in electricity use results in deployment of more fossil fuels as well as renewable energy, and since emissions depend only on fossil fuel deployment, emissions increase.
Panel E combines Panels A, C, D and E into one figure so that all of the comparative statics can be seen at once. Note that we have chosen the points S and T for ease of exposition as they have equal total cost and make clear that the cost per kWh of electricity declines. In general, total cost could be either higher or lower, but the electricity-price effect shifts the system to less electricity (lower isoquant) under a tax applied to that sector alone, and more electricity (higher isoquant) under an offset program applied to that sector alone[1].
The point that Figure 1 makes abundantly clear is that offsets are determined by the shift along the renewable input axis, while the change in emissions is determined by a shift along the fossil fuel input axis. There is no reason to believe that offset creation and emissions reductions should be equivalent except in very special circumstances—namely that fossil fuel and renewable energy are perfect substitutes. In the special case in which renewable energy and fossil energy are perfect substitutes, one would anticipate a corner solution in which the cheapest source was purchased exclusively.
Figure 1: Microeconomic illustration of the shift in fuel mix under a reference scenario, a carbon tax, and an offsets payment equivalent to the carbon tax.Panel A: Reference scenario production mix
/ Panel B: Production Mix, Fixed Output
Panel C: Shift in Production Mix With Carbon Tax
/ Panel D: Shift in production mix with offsets payment
Panel E: Comparison of three production mix choices under the reference, tax and offsets / Notes:
Elec 1, Elec 2, Elec 3: Three different levels of electricity production isoquants, which show all possible combinations of Fossil fuels and Renewable energy inputs that product the same level of net electric power, with Elec 1<Elec 2<Elec 3.
TCref, TCOffset, TCTax = Isobudget line showing all combinations of Fossil fuels and Renewable energy inputs with equal Total cost.
Fref, FOffset, FTax = Level of fossil fuels used in power generation under the Reference, Offsets, and carbon tax scenarios respectively.
Rref, ROffset, RTax = Level of renewable energy used in power generation under the Reference, Offsets, and carbon tax scenarios respectively.
R = cost minimizing combination of fossil fuels and renewable energy in the reference scenario.
T = cost minimizing combination of fossil fuels and renewable energy in the carbon Tax scenario.
S = cost minimizing combination of fossil fuels and renewable energy in the Offsets scenario.
[1] Note that the indirect effect on fossil and renewable energy input demands associated with the offset subsidy to renewable power is very similar in nature to the “income effect” that occurs in consumer theory.