ABM models of inflation representative firm → firm network

A stock-flow-consistent agent-based model on a UK firm-to-firm production network reconstructed from public ONS data. An energy or demand shock propagates through both the input-output network and the income–expenditure circuit; the central bank’s response trades inflation against unemployment and corporate defaults. Methodology in docs/METHODOLOGY.md.

Most systemically critical UK sectors

Firm-level economic systemic risk (ESRI), aggregated to sectors over an ensemble of reconstructed graphs. The ranking is robust to reconstruction noise (Spearman ≈ 0.99).

Shock & policy simulator

Shock window = periods 2–9. The economy starts at the exact Leontief steady state, so any movement is the shock and the policy response.

Price level (% above pre-shock)

Real GDP (% deviation)

Unemployment (%)

Firms defaulted (%) & policy rate (%)

The dilemma frontier

Sweep central-bank aggressiveness φ against the current energy shock: peak inflation falls while peak unemployment and defaults rise — with a tipping point where tightening triggers a default cascade through the network.

Peak inflation vs unemployment (each dot = one φ)

Firms defaulted vs φ

Network reconstructed from ONS Input-Output Analytical Tables (2023) and UK Business Counts (2025). The browser runs a JavaScript twin of the Python engine (web/js/firmnet_abm.js ↔ src/firmnet/abm.py) on the shipped, pre-reconstructed network — deterministic, so it matches the Python results exactly.