The wide-ranging impacts of energy transitions in the economy cannot be captured by one single model or approach (Chang et al., 2021). Models of lower complexity are often argued to be more appropriate for decision-making due to their apparent transparency. For instance, in the 1990s, the Dynamic Integrated Climate Economy (DICE) model attempted to understand the interplays of economy and climate change by considering only a few variables (Nordhaus, 1992). Although this model was originally designed for illustrative purposes, it has been used for policy purposes in some countries (Mercure et al., 2016). Models with higher complexity may be more difficult to use, but often present more realistic representations of economic and energy systems. Over and above these considerations, two main approaches for the assessment of macroeconomic aspects in energy transitions can be summarized.
The first approach relies on a descriptive analysis of historical data expressed through a set of indicators and is particularly suitable for an informative and quantifiable overview of the current situation. These “basic” (EPA, 2011) or “analytical” (Hardt and O’Neill, 2017) approaches can be used to obtain broad estimates. Models can be considered analytical if they contain few equations that can be solved numerically by iterative techniques, that is, they can be explicated in terms of equations that describe the relationship between different parameters. However, with no attempt to represent the underlying complexity between macroeconomy and energy systems, these models are characterized by relatively simple formulations and data.
While analytical approaches require minimal input data, time and technical expertise, they can also be used for preliminary analyses purposes (EPA, 2011). The Job and Economic Development Impact (JEDI) assessment is an example of a model to estimate the economic impacts of constructing and operating power plants (EPA, 2011). Models based on Input-Output (IO) tables can provide a snapshot view of the economy and are commonly applied to analyze interactions and feedback effects between mutually interdependent industrial sectors (Berg et al., 2015). One analytical model was proposed to determine the effects of transitioning from fossil fuels to renewable energy on GDP per capita (D’Alessandro et al., 2010).
In contrast, the second approach relies on more complex “numerical” models (Hardt and O’Neill, 2017). Numerical models rely on computer-based simulations and often contain a larger number of equations and assumptions. Modelling techniques in numerical approaches, such as Computable General Equilibrium (Vrontisi et al., 2019), Hybrid Models (Ghersi, 2015) and Econometrics (Režný and Bureš, 2018; Garcia-Casals et al., 2019), are useful when analytical rigour is desired and when sufficient data, time and resources are available. In order to conduct a numerical-based macroeconomic analysis of energy transitions, a representation of interactions between the energy systems and the rest of the economy is needed to capture how energy, socioeconomic and environmental aspects interact with each other (EPA, 2011; Lutz et al., 2014). Computational modelling was already used for energy planning as early as the mid-1970s to understand the implications of the first oil embargo (Nakata, 2004).
However, the outcomes of numerical models are tied to implicit theoretical aspects. For instance, conventional equilibrium models imply that all resources are currently allocated in the economy in the most productive way they can be, even though this assumption cannot be verified empirically. Other studies are criticized for their limited treatment of societal actors and socio-political dynamics, while numerical models based on cost-benefit analysis and cost-optimization have several shortcomings (Hardt and O’Neill, 2017). These considerations cast doubt on the realistic representations of numerical approaches and their ability to provide reliable evidence for policy processes (Mercure et al., 2016). It is thus critical to lay out the assumptions and theoretical aspects of numerical approaches in a way that allows for the interpretation of results beyond uncertainties and limitations.
Overall, analytical and numeric approaches have different levels of complexity and vary in scope, yet both rely on approximations and trends. Analytical approaches often use simplified reproductions of macroeconomic dimensions, while challenges in predicting technology adoption and diffusion (IEA, 2020b), change in human behaviour (D’Aprile et al., 2020; Larson et al., 2020), and policy effectiveness and implementation (Mercure et al., 2019) can be observed in numerical approaches. The key consideration for deciding which specific approach should be used thus lies in whether science-policy interfaces can use the results from these approaches to inform specific decision-making processes.