Tools & Methods

Our analysis tool: the E3ME-FTT-GENIE1 integrated assessment model

Analysing the dynamics of the Energy-Water-Food nexus requires evaluating outcomes of the present-day and future condition of the global economy and environment system. This requires using large amounts of data about the economy, the energy system and the environment from all around the World, in order to attempt to predict the future evolution of the system as a whole. In BRIDGE, we use an integrated simulation model that explores scenarios of the future, in order to assess the most likely impact and implications of various possible hypothetical choices of policies, in all relevant sectors of decision-making. We use the model to determine whether chosen sets of policies achieve or not their goal (e.g. reducing carbon dioxide emissions in order to reach climate targets), and whether or not they have unintended consequences (e.g. worsening energy poverty or access to food). Our model allows to analyse policies in many sectors of the economy, in 59 countries or regions of the world.

The E3ME-FTT model is maintained and based at Cambridge Econometrics, Ltd (Cambridge, UK), www.camecon.com. The GENIE1 model is maintained and based at The Open University, Milton Keynes, UK.

What is a simulation-based Integrated Assessment Model (IAM)?

An IAM is a model that features detailed representations of interacting global human and environmental systems. They typically at least analyse the global energy and industrial system, in order to correctly account for global greenhouse gas emissions, and the carbon cycle and climate system, in order to evaluate climate impacts of increases in emissions. However IAMs often include more components than this: agriculture and land-use, ecology and biodiversity, transport, etc. In BRIDGE, we cover comprehensively the global macroeconomy, the carbon cycle, the climate system, the global energy and electricity sectors (supply and end-use), the road transport sector, household heating, and finally, we are developing a model of agriculture and land-use.

The E3ME model: global non-equilibrium macroeconomics

Figure 1: Example of global industrial and fossil fuel emissions calculated by E3ME-FTT. The legend shows the classification of the types of fuel users in the model.

The E3ME model (www.e3me.com) is a global macro-econometric model that simulates the evolution of the global economy. It has a high sectoral and regional disaggregation, which means that it is able to analyse the economy with a high level of detail.

E3ME features:

  • The whole world in 59 countries/regions, including all EU member states and all G20 countries
  • 69 (EU) or 44 (other regions) industrial sectors
  • 22 types of users of fuel
  • 12 types of fuels
  • It features around 30 econometric relationships making a closed economic system. For example, it evaluates employment, investment, industrial production, material use, energy use, etc.
  • Regression parameters are evaluated using data from 1970 to 2010, and projections run from 2010 up to 2050.
  • It is calibrated to calculate accurate global greenhouse gas emissions, which can be used to calculate climate change implications of economic, industrial and energy policies.

E3ME is a non-equilibrium model, which means that the model’s theory does not assume that the economy is in equilibrium. In theories of the economy in equilibrium, production and investment is chosen as that which is optimal, and all resources are fully utilised (full employment). This implies properties such as there being no involuntary unemployment of the labour force. It also means that capital resources are utilised optimally, which implies that investment in low-carbon technology necessarily takes away investment in other parts of the economy, which therefore reduces GDP. The E3ME model does not take these assumptions; instead, spare resources are available in the economy, and finance is created by financial institutions according to the credit-worthiness of entrepreneurs. For more details, see our paper in  Pollitt & Mercure (2017) and Mercure et al. 2016 (open access publications in Climate Policy and Global Environmental Change), as well as our working paper Mercure et al. (2017) C-EERNG working papers.

Examples of policy analyses that have been done using E3ME include assessments for the European Commission on employment impacts of the Energy Roadmap 2050, and employment and social impacts of energy efficiency policy. See also Pollitt et al. (2014) and Barker et al 2015.

The FTT family of technology models


Figure 2: Technology transitions, innovation and market competition. Figure adapted from Mercure (2015).

When studying technological change, projecting technology adoption is not accurately done using regressions on time series. Technological change is highly non-linear. For that reason, for sectors where technological change is important (e.g. the electricity sector and transport), we use a different novel method that we have developed (Mercure 2012). This method takes a cross-sectional cost dataset and evaluates investor or consumer preferences for various technologies in competition in a market (e.g. different types of cars, different types of electricity generation technologies). It uses a theory of diffusion of innovations, in which innovations lie in niches with small numbers of applications, until the right conditions emerge to allow them to diffuse to a wider level of use. This is sometimes expressed as a technological transition. It is not possible to build detailed FTT models to cover all technologies in E3ME. Instead, we focus on the most environmentally-relevant sectors (the sectors with highest emissions: power, road transport, household heating, industry, land-use), and expand the technology resolution of E3ME by replacing these sectors by a detailed FTT model. Three models currently exist, FTT:Power (Mercure et al 2014), FTT:Transport (for road transport, Mercure et al 2017), and FTT:Heating (for household heating, Knobloch & Mercure, in preparation, 2017). Two additional models are under development: FTT:Industry and FTT:Agriculture.

The GENIE1-PLASIM integrated climate carbon cycle model

PLASIM-GENIE is a three-dimensional intermediate-complexity atmosphere-ocean global climate and carbon-cycle model.

Figure 3: Illustration of the carbon cycle in the GENIE model.

The physical (climate) components of PLASIM-GENIE describe the atmosphere, ocean, sea ice and land surface. The PLASIM atmosphere has the same dynamics as state-of-the-art climate models, but it is run at a low (500 km) resolution to increase computational speed. PLASIM includes the most important atmospheric processes that are found in fully complex models, but uses simplified representations. These processes include cloud formation, convective and large scale precipitation, and the interactions of incoming sunlight and outgoing infra-red radiation with water vapour, clouds and greenhouse gases. The ocean model is GOLDSTEIN, which has similar dynamics to fully complex models, but with approximations (and resolution) that increase computational speed. The sea-ice model includes ice growth and decay, determined by local heat fluxes, and ice transport, governed by surface ocean currents and diffusion.

The biological (carbon-cycle) components of PLASIM-GENIE represent both marine and terrestrial ecosystems. The ocean is modelled by BIOGEM, which represents ocean biology and chemistry through nutrient “tracers” (most importantly carbon, oxygen and phosphate). Equations describe the conversion between inorganic and organic compounds. Organic matter is created by photosynthesis in the well-lit surface and is oxidised back to minerals as it sinks to the ocean bed. Tracers are moved around by the ocean currents simulated by GOLDSTEIN. Vegetation and soils are modelled with ENTS, describing the accumulation of plant matter through photosynthesis, depending upon temperature, moisture availability and atmospheric CO2. Leaf litter from vegetation feeds the soils, which are in turn respired back to CO2 by bacteria. The land and ocean carbon reservoirs are connected through a well-mixed “box-model” atmosphere that neglects chemistry.

PLASIM-GENIE is a computationally demanding model, requiring 2 computing days to simulate 100 years of real time. While this is often practical, in some cases it is either impractical or unnecessary. When it is unnecessary, for instance when detailed climate projections are not required, PLASIM is replaced with a simplified representation of the atmosphere, in a configuration called GENIE-1, with which we can simulate 100 years of real time in 20 minutes. When even this is not fast enough, for instance for comprehensive evaluation of model uncertainty or for “live” projections, or when both speed and detailed climate are required, then we use emulators. These are statistical approximations to the models that provide effectively instantaneous outputs.

Agriculture, commodity trade and land-use change modelling

In BRIDGE, we are developing a new generation land-use model. Based on the FTT method, we are building a system in which we model decision-making, by heterogenous agents, for land-use in agriculture. In other words, we model what different types of farmers decide to do with the land they own, according to a context of prices for their agricultural products. Models always involve a reduction of reality to a tractable mathematical and computational methodology, which uses data we can access. Here, FTT:Agriculture makes use of (1) food balance tables, integrated to E3ME to calculate the demand for agricultural products, and (2) costs for producing these commodities, and (3) maps of what the land is currently used for, and maps of what the land is good for producing. We do not wish to model what the land should be used for; rather, wish to model what the land is likely going to be used for.

Food balances: E3ME currently features energy balances for all of its 59 regions and countries, expressing how much of each energy product various fuel users are using or transforming. Using econometric regressions, this enables to evaluate the future demand for 12 fuels according to the energy needs of up to 70 sectors of all 59 economies. We will develop the same for food balances: the demand for meat, cereals, dairy, etc. Some of these commodities are directly consumed by humans; however, a large amount is used for other purposes (e.g. producing biofuels) or for feeding animals. This is expressed in food balances made available by FAOSTAT. As global economies evolve, and income grows, countries of the World change diets, and thus the demand for agricultural products changes.

Land productivity: FTT will feature data obtained from large simulations of plant growth on the land surface, including the model LPJml from the Potsdam Institute for Climate Research (PIK). These data, forming grid points on a global map, provide us with the productivity of the land for various crops or forestry (i.e. the rate at which crops grow), linked to the climate and other parameters. As the climate changes, this data enables us to know how the productivity of the land is changing. This, in turn, will enable us to simulate how the crops being grown in various areas of the World may change in the future, as the climate changes and as the global economy changes its demands.

Land allocation: FTT:Agriculture will determine what cropland is used for: it will allocate the demand for produce to areas of land in scenarios of future economic, climate and policy evolution. Depending on what land is available for producing various crops (e.g. rice, wheat, maize, etc), production costs will evolve, and so will prices, and thus the demand for these products. In a dynamical interaction with E3ME, FTT:Agriculture will determine the land allocation and the price of commodities, while E3ME will evaluate their demand.

Competition for land between various uses: 

A key issue in Nexus research is land-use change, which often stems from conflicting demand for land for different types of activities. For example, competition could arise for fertile land between energy and food crops (e.g. corn). Alternatively, competition for specific agricultural commodities (e.g. soybeans) could also arise between energy and food purposes. These demand originate from completely different sectors of the economy, and in a model, they must be looked at simultaneously (e.g. how much ethanol is demanded in road transport, how much feed is required in the meat industry). E3ME-FTT, with support from GENIE-PLASIM, will explore these issues simultaneously, such that the outcome of these competition effects can be explored and better understood. This, in turn, enables to design better policies, whether they are of regulatory or economic nature.