UBI as Automation Catalyst in a Small Open Economy: An SFC-ABM Approach
Mateusz Maciaszek, 2026
We reverse the standard causality: Universal Basic Income does not respond to automation — it causes it. Through a double cost shock (rising reservation wages + rising interest rates), UBI makes labor-intensive business models mathematically unsustainable, forcing firms into a nonlinear technological leap. The resulting "Endogenous Technological Deflation" — where AI-driven productivity growth outpaces monetary expansion — resolves the inflationary dilemma of Modern Monetary Theory.
| Metric | UBI = 0 PLN | UBI = 2 000 PLN | UBI = 3 000 PLN |
|---|---|---|---|
| AI + Hybrid adoption | 12.9% | 61.9% ± 16.4% | 32.8% |
| Inflation (r/r) | −22.6% | −13.4% | +19.4% |
| Unemployment | 78.7% | 39.6% | 19.4% |
| Exchange rate PLN/EUR | 3.29 | 4.66 | 5.08 |
| Market wage | 4 000 PLN | 5 331 PLN | 6 487 PLN |
| Public debt | −0.83 mld | 12.58 mld | 15.23 mld |
Bimodality at UBI = 2 000 PLN confirmed (Hartigan dip test: p = 1.7 × 10⁻⁵), indicating a phase transition with three attractor states identified by GMM (K=3, BIC-optimal).
- SFC-ABM: 10 000 heterogeneous firms across 6 sectors (GUS 2024 calibration)
- Network: Watts-Strogatz small-world (k=6, p=0.10) — demonstration effect diffusion
- Monte Carlo: 100 seeds × 3 scenarios × 120 months
- Balance sheets: Full stock-flow consistency across 6 accounting blocks
- Sectors: BPO/SSC (σ=50), Manufacturing (σ=10), Retail/Services (σ=5), Healthcare (σ=2), Agriculture (σ=3), Public (σ=1)
├── simulations/
│ ├── scala/
│ │ └── simulation_mc.sc # Ammonite SFC-ABM simulation (865 lines)
│ ├── scripts/
│ │ └── run_sweep.sh # Parameter sweep runner (21 UBI levels)
│ └── results/
│ ├── baseline_*.csv # UBI = 2 000 PLN (N=100)
│ ├── nobdp_*.csv # UBI = 0 PLN (N=100)
│ ├── bdp3000_*.csv # UBI = 3 000 PLN (N=100)
│ ├── gus/ # GUS-calibrated runs
│ └── sweep/ # 21-point parameter sweep (0–5 000 PLN)
├── analysis/
│ └── python/
│ ├── mc_charts.py # Main Monte Carlo panel (6 subplots)
│ ├── mc_welfare.py # Welfare analysis charts
│ ├── sweep_analysis.py # Bifurcation diagram + inverted-U
│ ├── diptest_analysis.py # Hartigan dip test + GMM bimodality
│ └── gus_charts.py # GUS dual paradox + sector comparison
├── figures/ # 9 generated plots (PNG)
├── latex/
│ ├── esej.tex # Paper source (XeLaTeX + biblatex)
│ ├── esej.pdf # Compiled paper (~50 pages)
│ ├── references.bib # Bibliography
│ └── figures/ # Figures embedded in paper
├── Makefile
└── LICENSE
- Ammonite (Scala scripting)
- Python 3 + matplotlib, seaborn, scipy, scikit-learn, pandas
- XeLaTeX + biblatex (for paper compilation)
# Full pipeline: simulation → figures → paper
make all
# Or step by step:
make simulate # ~45 min (3 × 100 seeds × 10 000 firms × 120 months)
make figures # Generate all plots from CSVs
make paper # Compile LaTeX → PDFcd simulations/scala
BDP=2000 SEEDS=10 PREFIX=quick amm simulation_mc.sc
Main panel. Six macroeconomic variables across 120 months for three UBI scenarios (0, 2000, 3000 PLN). Mean lines with 90% confidence bands over 100 seeds. The shock at month 30 triggers fundamentally different trajectories depending on UBI level.
Bifurcation across 21 UBI levels. Four panels: adoption, inflation, adoption variance, and unemployment vs BDP. The variance peak at BDP = 2000 PLN marks the critical point of the phase transition.
Phase transition evidence. Adoption histogram shows a bimodal distribution at UBI = 2000 PLN (three attractor states via GMM, K=3). Per-sector bars reveal BPO/SSC and Manufacturing respond in opposite directions. Phase space scatter confirms regime separation.
Formal bimodality test. Hartigan dip test rejects unimodality (p = 1.7 × 10⁻⁵). BIC model selection identifies K=3 as optimal. KDE overlay shows bimodality is unique to the critical UBI level.
The Acceleration Paradox in two panels. Adoption vs UBI shows an inverted-U — moderate UBI maximizes automation, while higher UBI triggers inflation that chokes investment. This is the core finding of the paper.
Per-sector adoption trajectories at UBI = 2000 PLN. BPO/SSC (σ=50) races to near-complete automation while Manufacturing (σ=10) stalls — the double cost shock hits high-elasticity sectors hardest.
Welfare trade-offs. Real consumption, Gini coefficient, and a composite dashboard across scenarios. UBI = 2000 achieves the highest productivity but at the cost of maximum inequality at the transition point.
GUS 2024 sector structure. All 6 sectors including Public and Agriculture. The 45% weight of Retail/Services (σ=5) in the Polish economy dampens aggregate adoption compared to the simplified 4-sector model.
The Dual Paradox. BPO/SSC accelerates (+21 pp) while Manufacturing decelerates (-14 pp) under the same UBI. The inflation channel explains the divergence: high-σ sectors benefit from cost pressure, low-σ sectors are crushed by it.
- Paper-02: Monetary Regimes
- Paper-03: Empirical σ Estimation
- Paper-04: Phase Diagram & Universality
- Paper-05: Endogenous Technology & Networks
- Core engine
MIT