Belgium: linking the economy and the energy system

Why it was set up

Industry accounted for about 46 per cent of Belgian final energy and non-energy demand in 2019. How much industry Belgium will still host in 2050 is one of the largest uncertainties in long-term energy planning.

Most net-zero studies treat industrial activity as an assumption. Energy system models let firms choose between technologies. They do not let firms change what they produce, where they source inputs, or how much labour they use. A reduction or increase in industrial demand then enters as an exogenous assumption. The project asks what happens to the economy and the energy system when that change has an economic cause whose effect further propagates through the energy-economy system.

What it sets out to do

The project links three kinds of model that are usually run apart: an economic model, an energy system model and an input-output model of international supply chains. It works in three steps.

• Year one. Describe industrial transformation cases and group them into three storylines: GREEN-IT (core products stay, processes change), SHIFT-IT (high-value activity stays, the rest relocates) and LEAVE-IT (energy-intensive activity moves out of the EU). Published as Deliverable D1.1.
• Year two. Turn the cases into stylised shocks in the modelling system and show what they do. Published as Deliverable D2.1.
• Year three. Combine the shocks into fuller energy-economy pathways, and test policy against them.

The work runs in two streams.

• Stream 1: growth or displacement? How do large economic trends change Belgian manufacturing, and what does that mean for the energy system?
• Stream 2: to tax or to subsidise? What happens to emissions and to industrial activity under a carbon price, a subsidy for clean technology, or both?

How the models fit together

Stream 1. GreenREFORM-BE runs a scenario to 2050. It passes its energy service demand, by sector and year, to TIB3R, which finds the cheapest technologies to meet it. The economic model also passes its input-output tables to an input-output model built on Eurostat’s FIGARO tables. That model traces the environmental impact of Belgian consumption at home and abroad.

Stream 2. The energy technology model of GreenREFORM-EU is built into GreenREFORM-BE. Industries then pick from a list of dated, discrete technologies, instead of a smooth trade-off between capital and energy. The two models are solved together, in one system. The technology lists come from TIB3R, through a Python script written for the project. Steel is the first sector in the integrated model.

The Belgian version uses fourteen industries. It keeps the elasticities of GreenREFORM-EU.

What the October 2026 report shows

These are first results from stylised scenarios. The report says they are not forecasts and not policy advice. They test what the modelling system can do.

• The cause of a loss matters, not only its size. Two pairs of scenarios remove the same amount of production from energy-intensive industry. In each pair, one version carries a gain in productivity or income and the other does not. Energy demand falls by more in the version without a gain. In the trade pair, final energy demand falls by 4.1 per cent in one case and 6.7 per cent in the other. GDP ends 5.1 per cent above baseline in the first and 0.5 per cent below in the second.
• Electricity shows the widest gap. In the trade pair, electricity demand falls by 5.8 TWh against 13.3 TWh by 2050. The report notes this matters for grid planning.
• Less industry at home is not always less impact. In both trade scenarios, lower impact in Belgium is outweighed by higher impact elsewhere. The global environmental footprint may rise.
• A productivity boom in services can raise industrial energy use. In one scenario on artificial intelligence, higher wages push manufacturing towards capital and energy. Final energy demand rises by 5.5 per cent.
• Technology choice changes the shape of the response. In the steel test, emissions fall in steps tied to the dates when plants retire and new technologies become available. A higher carbon price and a clean-technology subsidy both bring investment forward. By 2050 the subsidy cuts steel emissions by 36 per cent against the baseline. Combined, the two policies do more than the sum of their parts until the late 2030s, and less afterwards.

What comes next

In its final year the project will develop its final energy-economy pathways and policy analysis. This will involve refining the models, widening the spectrum of stylised shocks and combining them into coherent scenarios. In Stream 2 it will add basic chemicals, cement and paper to the integrated model, and test existing subsidy schemes.

Relation to GreenREFORM-EU

GreenREFORM-BE is the Belgian implementation of the GreenREFORM-EU model. The Federal Planning Bureau received the model in the TSI 2024 project and has been one of the four teams in Version 2.0. CEEP-IT is a separate Belgian project with its own funding.

Who takes part

• VITO/EnergyVille. Coordinates the project. Energy system modelling and input-output analysis.
• ICEDD. Energy system modelling and the technology data.
• Federal Planning Bureau. Economic modelling with GreenREFORM-BE.
• Energy Transition Fund / FPS Economy. Funds the project.

Results

• Deliverable D1.1: Towards scenarios of industrial transformation in Belgium (December 2025).
• Deliverable D2.1: Towards a linked energy-economy system for Belgium (October 2026).

The final-year report will be listed here when it is out.

Read more

• EnergyVille project page
• VITO project page
• Commission page on the Belgian TSI project
• The TSI 2024 project