Greenhouse Gas Emissions Forecasting: Learning from International Best Practices

The challenges associated with estimating emissions reductions from climate policy measures and actions are well documented.^[10] On the methodological side, these include the development of baseline assumptions^[11] and the impact of specific policy measures, assumptions about financial costs and consumers’ preferences, assumptions about the direction and rate of technological change, costing individual actions versus integrated actions, assumptions about macroeconomic costs, the effect of policy measures on cost incidence and total costs, and the full range of costs and benefits of GHG emission reduction policies. There is no simple method to account for all these actions.

Perhaps less technically complex, but just as challenging, are governance issues related to effective emissions forecasting. The NRTEE’s 2007 KPIA Response emphasized the importance of transparency and clarity with respect to key assumptions and methods, and the consideration of important sensitivities and uncertainties. It also emphasized the importance of consistency in approach across various departments, programs, and measures, and the need to integrate findings in a holistic framework. This report will explore how two other countries address both sets of these challenges and determine, to the extent possible, if it is appropriate to apply their methodology and governance approaches to the Canadian context.

3.1 What is Forecasting?

In this report, forecasting is defined as a depiction -- an economic model -- of how a system will evolve in future both with and without policy intervention. Since the future is of course unknown and thus uncertain, we cannot say that one forecast made today is better than another. What we can do, however, is assess alternative forecasting methods in terms of the following criteria:

1. past accuracy, which may or may not bode well for future forecasts;
   
2. sound representation of current and emerging system dynamics,^[12] which should increase the probability of a better forecast;
   
3. greater transparency, which increases the ability for outsiders to examine and critique all key assumptions, and perhaps test alternatives; and
   
4. the ability to conduct and record sensitivity analyses, which should improve the understanding of the critical forecast model assumptions and related key uncertainties.

3.2 Modelling Approaches

Emissions forecasts are calculated using energy-economy models. These models are designed to forecast the effects of policies on energy-related GHG emissions. The structure of most energy-economy modelling approaches ranges from detailed bottom-up models reflecting engineering-economic details of a wide menu of technologies in each sector to top-down models of the whole economy calibrated on historic data from up to hundreds of sectors. Hybrid models -- those that combine the strengths of the bottom-up and top-down approaches -- are considered by many modelling authorities as optimal approaches to forecasting.^[13] Canada’s E3MC model^[14] is an example of a hybrid model, along with the U.S. Energy Information Administration’s (EIA) model (described in section 4.4.2)^.[15] In the later discussion of specific countries’ best practices approaches to emissions forecasting, we will assess what kinds of conclusions can be drawn as to the effectiveness of forecasts from the various modelling approaches used.

10 Examples include Jaccard, M., J. Nyboer and B. Sadownik, The Cost of Climate Policy (2002); Bernstein, S., "International institutions and the framing of domestic policies: The Kyoto Protocol and Canada’s response to climate change," Policy Sciences, 35 (2002); and Nordhaus, R. and K. Danish, Designing a mandatory greenhouse gas reduction program for the U.S. (2003).

11 At the very least, baseline assumptions should include supporting data and data consistency with other agencies.

12 In this case, system dynamics refers to how an energy-economy model captures changes in technologies, costs, and behaviours over time.

13 Dowlatabadi, H., D.R. Boyd, J. MacDonald (2004). Model, Model on the Screen, What’s the Cost of Going Green? Washington, D.C.: Resources for the Future, p. 10.

14 E3MC stands for Energy-Economy-Environment Model for Canada. A description of it is found in section 4.3.1.

15 Please refer to Appendix A for a discussion of modelling approaches.