I've been building a structural simulation engine for the past year that models country-level socio-economic trajectories to 2050.
The core idea: run 10,000 Monte Carlo paths across 26 indicators (GDP, inflation, unemployment, housing, demographics, energy, etc.), but constrain the paths with 100+ structural coupling rules so the outputs are economically coherent.
Each coupling rule has:
- A trigger condition (e.g., "fires when tax wedge exceeds 43% for 3 consecutive years")
- Effects on related indicators (e.g., "suppresses employment growth, erodes fiscal revenue")
- Duration, lag, persistence, decay, and mean reversion parameters
- An academic citation
Without the coupling rules, Monte Carlo produces statistically valid but structurally meaningless output. You get paths where unemployment is 2% while inflation is 15% and GDP is collapsing. The rules prevent that by enforcing known causal relationships.
Three cascade examples:
Energy shock: petrol > $2.00/L fires Fuel Pressure, which cascades into transport costs, inflation, interest rates, housing affordability, and net migration.
Demographic decline: fertility < 1.2 for 3 years fires Demographic Winter, which cascades through workforce contraction, tax revenue decline, debt increase, and fiscal tightening.
Technology growth: R&D > 2% of GDP for 3 years fires R&D Spillover, which cascades through hi-tech exports, GDP growth, revenue buoyancy, and fiscal improvement.
The rules produce endogenous business cycles (consistent with Burns & Mitchell 1946, Minsky 1986): growth triggers recovery rules, recovery pushes indicators past stress thresholds, stress dampens growth, which triggers new recovery.
Validation: backtested across 17 European countries from 2016-2024 (through COVID, Ukraine war, energy crisis). 77-89% of real outcomes fell within the P10-P90 bands. 63-86% directional accuracy.
What I'm looking for feedback on:
- Does the coupling rule approach make sense for modelling structural macro interactions over long horizons?
- For researchers familiar with specific countries: would these cascade mechanisms produce plausible trajectories?
- What structural interactions are missing that you'd expect to see in a 26-indicator macro model?
The engine is live if anyone wants to test a specific country and give structured feedback. Happy to provide access.
PS: I used AI to help me write/frame this
Combining Monte Carlo simulation with structural socio-economic coupling rules for long-horizon country modelling. Feedback / Does it make sense?
byu/YiannisPits91 inAskEconomics
Posted by YiannisPits91