What are the starting points? Evaluating base-year assumptions in the Asian Modeling Exercise☆

The need to understand the impacts from climate change and the costs and benefits associated with different mitigation policies has made long-term energy and emission scenario analysis increasingly important. Numerous Integrated Assessment (IA) and energy-economic models have been developed for analyzing and understanding the dynamics of economic development, energy use, and environmental outcomes.

The fundamental characteristics of these long-term scenarios –population growth, economic growth, energy consumption, greenhouse gas emissions, and so forth – can vary substantially among (and even within) models and modeling groups. This variation is a result of differences in assumptions regarding the future character of underlying demographic, economic and technological forces, as well as differences in assumptions embodied in different model structures. These uncertainties about the long-term future, while not necessarily welcome by decision makers and other consumers of long-term scenarios, must be accepted as a legitimate result of our lack of knowledge about how the future may unfold.

What may be less well understood, however, is that variation across models applies not just to the long-term future, but also to the base-year values used by these models; that is, the models often do not even agree about recent history. A natural question is the degree to which this variation is a result of real uncertainty about history or methodological differences among modeling teams.^1,2 As we will discuss, methodological differences are, indeed, an important component of the variation. This includes differing base-years and regional definitions. However, there are also multiple data sources used by the IA modelers that also contribute to the variability over base year values.

Base-year variations are particularly important for modeling in Asia. Historically, the developed countries have been the primary consumers of energy, and relatively reliable data, based largely on stable and consistent collecting and reporting methodologies, are available for important variables for most of these countries. In the future, however, the role of developed countries in shaping energy scenarios is bound to decrease because most of the future population and economic growth is expected to be concentrated in Asian, African and Latin American regions. Therefore, the quality of recent historical data for these regions is a source of concern for global and regional modeling teams and the decision makers that seek to formulate domestic and international policy using insights gleaned from these models.

The current paper focuses on the primary sources of historical data and the varying usage of these data in the models that participated in the Asian Modeling Exercise (AME),³ a multi-model scenario comparison exercise focusing on the role of Asia in long-term energy and emission scenarios. The paper focuses on the following research questions: (1) how different are the base-year numbers across models, (2) what are some potential explanations for these differences, (3) where does the data in various databases used by IA modelers come from, and how big are the differences among these sources; and (4) what are the implications of the use of different base year data for long term scenarios produced by IA models?

The fundamental result of the exploration is that differences in base-year data can be traced to differences in methodologies for aggregating regions, choices of which base-year to use, different ways of adjusting the time-value of economic output, and differences in the sources of base-year data among models. The differences among base-year sources are of particular importance because these differences indicate that even if a standardized methodology was used by all modelers to incorporate base-year data, the values would still not necessarily match. The past is, indeed, uncertain.

The analysis has been conducted for four sample countries: China, India, Japan and Korea. As per IEA (2007a), the share of Asian GDP for these regions in 2005 was 52% (Japan), 20% (China), and 7% (for India and Korea), and their share in Asian primary energy supply in 2005 was 14%, 46%, 14% and 6% respectively. In many of the global energy assessment models that participated in the Asia Modeling Exercise, these regions are modeled separately, reflecting a view that these four countries will be important in defining the future Asian energy and emission pathways. Because many of the teams model each of these regions, it is feasible to compare key variables across models. In addition, these countries cover a range of economic development, enabling exploration of whether differences in base year data are unique to developing or developed countries or more widespread.

The remainder of the paper proceeds as follows. Section 2 begins with an overview of the degree of variation in 2005 numbers across models for the four target Asian countries: China, India, Korea, and Japan. Section 3 then discusses the main data sources used for IA models and important issues related to data collection and processing. Section 4 discusses the factors that might lead to different base year numbers across models. The implications of different base year numbers on the future scenarios are discussed in Section 5. The paper concludes, in Section 6, with a discussion on important insights and implications from the analysis.