CIRCEE: The CIRCular Energy Economy model

General description

CIRCEE, a deterministic Dynamic Stochastic General Equilibrium (DSGE) model augmented with material flow analysis with mass balance and has a one-way soft-link to the IAM WITCH model (Emmerling et al., 2016), that analyzes the circular economy by integrating material, energy, and economic flows. The model simulates interactions among producers, consumers, government, and the external sector, with a particular focus on consumer-focused circular strategies (sharing, repair, recycling, sufficiency) and climate change mitigation goals. This model captures the trade-offs and synergies between different sustainability objectives. Additionally, there’s a version called « CIRCEE_LIFE » that soft-links CIRCEE in a two-way with the lifestyle model LIFE model (Pettifor et al., 2023).

Project structure

Main files

CIRCEE_PF.mod

Role: Main model file containing the full economic structure. The file contains the declarations of variables and parameters, as well as the main equations and FOCs of the model. For a comprehensive overview of the equations, please refer to the Appendix of Corbier et al. (2025). It’s important to note that the model’s equations have undergone slight modifications since Corbier et al. (2024 and 2025).

Contents:

• Macro definitions: Selection of region (JPN/EU27/CN), model type (B2C/C2C) and SSP scenarios, and sectors, materials, and lifestyles definition.

Users can choose between B2C and C2C sharing business models.

Users can select different regions, such as Japan, and the EU27 coming in October and mainland China in 2026.

Users can only choose the scenario SSP2 among all available scenarios. However, other SSP scenarios beyond SSP2 will be added in the next following months.

Users can only run CIRCEE without any lifestyle changes.The parameters related to lifestyle changes are referred to as « modifiers » and have a default value of 0 in the public version of the file. If users wish to run CIRCEE-LIFE (sufficiency, repair, and sharing lifestyles), they can request this feature at [email protected].

• Endogenous variables (300+):

  • Firms’ decision variables (output, capital, labor, energy and material inputs)
  • Household variables by lifestyle (consumption, saving/investments, energy-using and other durable goods)
  • Government variables (budget, taxes, subsidies)
  • International trade variables
  • Material stocks and flows
  • Waste flows
  • CO₂ emissions

• Predetermined variables: Capital stocks, energy-using and other durable goods stocks, material stocks

• Exogenous variables: Technological efficiencies, resource prices, fiscal policies

• Parameters: Substitution elasticities, preference parameters, depreciation rates

• Model equations:

  • Household behavior (by lifestyle: low-carbon, cautious, constrained)
  • Sectoral production functions (consumer and capital goods, materials, circular services: repair & sharing)
  • Market-clearing conditions
  • Physical constraints (material balances, emissions)
  • Trade with Armington CES

Features:

• Uses the Dynare preprocessor with @#for loops to auto-generate equations
• Supports two variants: B2C (business-to-consumer sharing) and C2C (consumer-to-consumer sharing)
• Perfect-foresight solution with or without anticipation errors over 82 periods (2018–)

---

CIRCEE_RunFile.m

Role: Main execution and post-processing script

Execution steps:

1. Setup

   • Select mode ('classic' only available)
   • Select model (B2C/C2C)
   • Choose SSP scenario (SSP2 only available, SSP1/SSP4/SSP5/NoGrowth in later versions)

2. Generate calibration files

   • CIRCEE_calibration.m: Structural parameters from calibration.csv
   • CIRCEE_baseyear_values.m: Initial values (2018 base year) for exogeneous variables
   • CIRCEE_shocks.m: Paths of exogenous shocks from shocks.csv. To introduce specific exogenous or policy variables (varexo in CIRCEE_PF.mod) with a particular growth rate, users can follow the instructions in section 3.1.2. of "CIRCEE_RunFile.m". Alternatively, they can specify specific absolute values in shock.csv or in CIRCEE_PF.mod. If they desire a fully deterministic model with no error of anticipation, they can use shock.csv. Conversely, if they want a deterministic setup with anticipation errors, they can include the line "@#include CIRCEE_shocks.m " in CIRCEE_PF.mod. Also, you need to insert a terminal value for the shock in "terminal conditions" in CIRCEE_steadystatemodel.m. For further information of the shock structure, please refer to the Dynare manual book at https://www.dynare.org/. One piece of advice : avoid using too many shocks and to be cautious about the magnitude of the shocks or you will bump into infeasibilities and corner solutions.
   • CIRCEE_endvalues.m: Terminal values for convergence

3. Run Dynare

   • Call dynare CIRCEE_PF.mod
   • Compute steady state from the CIRCEE_steadystatemodel.m
   • Solve the transition path

4. Post-processing of results

   • Rescaling: Multiply by population × labor productivity

   • Unit conversion:

     • Energy in EJ (÷10¹²)
     • Employment in millions of people (÷10⁶)

   • GDP calculation: In MER and PPP

   • CSV generation:

     • CIRCEE_output_levels.csv: Scaled absolute levels
     • CIRCEE_output_deviation.csv: Deviations relative to 2020
     • CIRCEE_output_levels_efficientlaborunit.csv: Per effective labor unit
     • CIRCEE_output_deviation_efficientlaborunit.csv: Relative deviations
     • Thematic files for CIRCEE_LIFE: Electrification, Sharing, Repairing, Lowering_Expenditures

5. Welfare analysis

   • Compute steady state consumption equivalent by lifestyle
   • Compare 2018 vs 2050

---

CIRCEE_steadystatemodel.m

Role: Analytical steady-state computation. If you modify any equation in CIRCEE_PF.mod, you must also adjust the model’s resolution for the steady state in CIRCEE_steadystatemodel.m.

Sections:

1. Energy prices: Derived from WITCH trajectories

2. Parameter aggregation: Lifestyle-weighted averages

3. Numeraire: Fix the price of capital goods

4. Price aggregators: Composite (CES) prices for each sector

5. Investment decisions: Capital and household durables

6. Production sectors:

   • Consumer goods (nondurables, semi-durables, durables)
   • Capital goods
   • Materials (virgin, recycled)
   • Services (repair, sharing)

7. Household consumption: By lifestyle and good type

8. Market clearing: Supply–demand equalization

9. Employment: Sectoral labor allocation

10. Aggregates: GDP, emissions, material flows, waste

11. Government budget: Revenues, expenditures, primary balance

12. Trade: Imports, exports, trade balance

Method: Recursive solution starting from normalized ratios, then computing absolute levels

---

Calibration files

calibration.csv

• Columns: Variable, Scenario, Model, Value
• Contains all structural parameters and most exogeneous variables of the model
• Filtered by SSP scenario and model type (B2C/C2C)

Example parameters:

• Substitution elasticities (σ)
• Distribution shares (α)
• Depreciation rates (δ)
• Fiscal parameters

shocks.csv

• Columns: Variable, Year, Value

• Time paths for 2019–2080 for:

  • Energy and material efficiencies (WITCH)
  • Share of each lifestyle in total population
  • Resource prices evolution (WITCH)
  • R&D, energy efficiency and power generation sector investments (WITCH)
  • Behavioral modifiers
  • Emission factors

---

Key model variables

By production sector

• Y_s: Sectoral output
• K_f_s, L_s: Capital and labor
• M_s, E_s: Materials and energy
• El_s, Nel_s: Electricity and fuels
• p_s: Wholesale price

By household lifestyle

• C_h: Aggregate consumption
• D_h or D_lowuse_h: Stock of energy-using durable goods
• X_h, SD_h: Nondurables and semi-durables
• ES_h: Energy services
• Inv_d_new_h, Inv_d_repair_h: New and repaired durable investment

Government

• Revenues, Expenses, PS: Public budget
• Carbon_budget, EPR_budget: Tax redistribution

Trade

• IMP_s, EXPORT_s: Sectoral trade
• TB: Trade balance
• Bonds_foreign: External debt

Environment

• CO2: Total emissions
• MW, IW: Municipal and industrial waste
• DMC, DMI: Domestic material consumption and inputs
• M_stock: Total material stock
• Gross_additions_stock and Net_additions_stock: Gross and net additions to stock
• Material_balance : Ensures mass balance is verified

---

Usage

Standard mode

CIRCEE_RunFile('classic', 'C2C')

Parameters

• mode: 'classic' (default) or 'calibration' (for CIRCEE_LIFE)
• modelType: 'B2C' or 'C2C'

---

SSP scenarios

In its current version, the model supports SSP2 and will support other population and economic growth scenarios:

• SSP1: Sustainable development
• SSP2: Middle-of-the-road trends (default)
• SSP4: Inequality
• SSP5: Rapid growth
• NoGrowth: Demographic stationarity

---

Outputs

Generated CSV files (in /Output)

1. Absolute levels and relative deviations
2. Variables normalized per unit of effective labor
3. Thematic files for circular strategies

---

Dependencies

• MATLAB R2016b or later. A version on Julia will be made available in later months.
• Dynare 4.6 or later (https://www.dynare.org/).
• Toolbox: Statistics and Machine Learning.
• If you encounter any issues, please contact [email protected]. Also, any comments or suggestions for model development are more than welcomed.

---

Technical notes

• Time horizon: 2018–2100. We add 20 more years (82 periods in total) to avoid end of the year spikes. Users should discard results from 2081 to 2100.
• Frequency: Annual
• Solution method: Perfect foresight with anticipation errors or without anticipation errors. Refer to the Dynare manual for more information.
• Expectations handling: Support for expectation errors (learnt_in)
• Convergence: Automatic steady-state verification

---

Author and references

If you use CIRCEE, please cite:

• Corbier, D., Pettifor, H., Agnew, M., & Drouet, L. (2024). CIRCEE, the CIRCular Energy Economy model: Bridging the gap between economic and industrial ecology concepts. Journal of Industrial Ecology, 28(6), 1996-2011.

• Corbier, D., Pettifor, H., Agnew, M., & Nagashima, M. (2025). Shaping sustainable consumption practices: Changing consumers’ habits through lifestyle changes and Extended Producer Responsibility schemes. Resources, Conservation and Recycling, 217, 108214.

Other references :

• Emmerling, J., Drouet, L., Reis, L. A., Bevione, M., Berger, L., Bosetti, V., ... & Havlik, P. (2016). The WITCH 2016 model-documentation and implementation of the shared socioeconomic pathways (No. 42.2016). Nota di Lavoro.

• Pettifor, H., Agnew, M., & Wilson, C. (2023). A framework for measuring and modelling low-carbon lifestyles. Global Environmental Change, 82, 102739.

---