📈 economics

Emergent Physicalism in Economic Modeling:\From Physical Constraints to Macroeconomic Regularities

This paper introduces "Emergent Physicalism," a JAX-accelerated agent-based modeling framework demonstrating that key macroeconomic regularities and structural stability arise uniquely from physical constraints interacting with minimal institutional scaffolding, rather than from behavioral rules or exogenous shocks.

Original authors: Liutao Hu

Published 2026-07-01

📖 6 min read🧠 Deep dive

CC BY 4.0

Original authors: Liutao Hu

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine you are trying to understand why a city's traffic jams, housing prices, and job markets behave the way they do. Most economists build models like complex video games where every person (agent) is programmed with a "brain" that tries to be smart, learn from mistakes, and maximize happiness. They ask, "What if people act this way?"

This paper proposes a different approach called Emergent Physicalism. Instead of giving the people in the model a "brain," the author gives them only physical rules, like gravity or friction. The idea is that if you build a system with the right physical constraints, the complex economic patterns we see in the real world will "emerge" naturally, without anyone needing to be told to act that way.

Here is the paper explained through simple analogies:

1. The Core Idea: The "Toy Car" vs. The "Robot"

The Old Way (Robots): Imagine a model where every agent is a robot programmed to say, "I will buy a house if I think prices will go up." The modeler has to guess the robot's personality.
The New Way (Toy Cars): Imagine a model where agents are just toy cars on a track. They don't have brains. They just have:
- Wear and Tear: Their wheels get smaller over time (depreciation).
- Crowding: The more cars they have, the less speed each new car adds (diminishing returns).
- Fuel: They need a minimum amount of gas just to stay alive (subsistence).
- The Result: Even though the cars have no "plans," they naturally form traffic jams, stop-and-go waves, and clusters of fast vs. slow cars. The paper argues that real economies work the same way: the patterns come from the physics of resources, not the psychology of people.

2. The Big Experiment: The "Lego City"

The author built a digital city with 200 "agents" (people) running for 10,000 days. They didn't program the people to be greedy or smart. They just set the physical rules:

Tools rot: If you own a hammer, it gets rusty (depreciation).
Too many tools help less: Owning 100 hammers doesn't make you 100 times more productive than owning 10.
Quality is separate: You can have a rusty hammer (low quality) or a shiny one (high quality), but the number of things you make depends on how many hammers you have, not how shiny they are. This is a crucial "decoupling" trick the author invented to prevent the whole system from crashing.

3. What Emerged? (The Magic Tricks)

Even without programming "behavior," five famous economic phenomena popped up on their own:

The Kitchin Cycle (The Breathing Economy): Just like a chest expands and contracts, the economy naturally went through 3–5 year cycles of boom and bust. This happened simply because people bought new tools, the tools wore out, and they had to buy more. No external "shock" was needed.
The Financial Accelerator (The Wealth Snowball): When the economy got a little shaky, the rich got richer and the poor got poorer faster. This happened because the system naturally created a "credit loop" where those with more tools could borrow more, creating a self-reinforcing cycle of inequality.
The Lemon Market (The Trust Problem): Buyers couldn't tell if a product was high quality or "lemon" (bad). So, they paid an average price. This made sellers of high-quality goods leave the market, leaving only the bad ones. This happened purely because of information gaps, not because people were "irrational."
The Matthew Effect (The Rich Get Richer): Those who started with more tools invested more in making their tools better, widening the gap between the "employers" and the "workers."
Credit Crowding: When people started lending money to each other, it actually reduced the total amount of trading happening in the market.

4. The "Ablation" Test: The "Turn-Off" Switch

To prove these patterns weren't just accidents, the author played a game of "What if I break this part?" (Ablation).

The Test: They turned off the "lending" rule. Result: The "Financial Accelerator" disappeared.
The Test: They turned off the "quality" rule. Result: The "Lemon Market" disappeared.
The Conclusion: Each economic pattern was tied to one specific physical rule. If you remove the rule, the pattern vanishes. This proves the patterns are necessary consequences of the physics, not just random noise.

5. The "Stress Test": The "Earthquake"

The author hit the system with extreme scenarios to see if it would break:

The "Depreciation Storm": They made tools rot 5 times faster than normal. The capital stock crashed by 74%.
The Result: The system didn't die. It recovered fully. Why? Because the skills of the people and the structure of the society remained intact. Once the shock passed, they started building tools again immediately.
The "Quality Shield": Even when the economy collapsed or exploded, the quality of the goods remained almost exactly the same. This proved that the author's "decoupled" design (keeping quality separate from quantity) worked perfectly.

6. The "Institutional" Twist: The "Uniform"

The author changed the "rules of the game" (like who counts as a boss). They changed the definition of an "employer" from the top 5% of rich people to the top 40%.

The Result: Almost nothing changed. The economy looked the same.
The Meaning: This suggests that the "physics" of the economy (depreciation, resource flow) is the real driver. The "institutions" (laws, job titles) are just a thin layer on top that changes the details but not the fundamental behavior.

Summary

This paper argues that we don't need to assume people are complex, calculating geniuses to explain the economy. If we build a system with simple, unchangeable physical rules (things break, resources are limited, quality varies), the complex behaviors of the real world—cycles, crashes, inequality, and market failures—will build themselves.

The author calls this Emergent Physicalism: The economy is less like a chess game played by grandmasters, and more like a river flowing over rocks. The shape of the river (the economy) is determined by the rocks (physical constraints), not by the water's "intentions."

Technical Summary: Emergent Physicalism in Economic Modeling

1. Problem Statement

The paper addresses a fundamental methodological tension in Agent-Based Modeling (ABM) of macroeconomics. Traditional ABMs often equip agents with complex behavioral rules (utility maximization, adaptive expectations, learning heuristics) to generate macroeconomic regularities. However, this approach faces criticism regarding "parameter explosion," a lack of systematic validation, and the difficulty of distinguishing whether observed phenomena are genuinely emergent or merely encoded within the behavioral assumptions. Conversely, Dynamic Stochastic General Equilibrium (DSGE) models achieve macro-consistency through representative agents but suppress the micro-heterogeneity that drives real-world dynamics.

The central question is: Can macroeconomic regularities emerge solely from physical constraints (depreciation, diminishing returns, resource flows) interacting with minimal institutional scaffolding, without relying on encoded behavioral rules or exogenous shock processes?

2. Methodology: Emergent Physicalism

The authors propose Emergent Physicalism, a methodology where macro-regularities are generated by material production constraints rather than agent cognition or intentions. The approach is operationalized through a JAX-accelerated ABM featuring 200 heterogeneous agents over 10,000 timesteps.

Core Design Principles

Physical Constraints vs. Institutional Scaffolding:
- Physical Constraints: Direct analogues to material reality (e.g., capital depreciation $\delta$ , diminishing returns $K$ , subsistence thresholds $c_s$ , resource inflows $f$ ). These operate regardless of agent intent.
- Minimal Institutional Scaffolding: Simple rules defining economic structure (e.g., employer status based on top 20% tool holdings, a fixed savings rate of 10%, lending rates anchored to depreciation). These are not calibrated to data but chosen to be minimal.
Decoupled Architecture:
- The model breaks the "death spiral" cascade found in coupled models (where quality decay reduces output, which reduces investment, further reducing quality).
- It separates Quality (affecting price and trade income) from Quantity (affecting production volume and self-sufficiency).
- Four independent channels interact via resource flows rather than equation coupling: Quality, Quantity, Investment, and Credit.
Validation Framework:
- Ablation Validation: Systematically removing physical constraints to verify necessity (if a constraint is removed, the phenomenon must disappear).
- Parameter Sweep: Varying physical constants to test robustness and define distinct "economic universes."
- Stress Testing: Applying extreme shocks to verify structural integrity and recovery dynamics.
- Multi-Seed Verification: Ensuring results are not artifacts of stochastic initialization.

3. Key Contributions and Results

A. Five Emergent Phenomena

The study demonstrates that five distinct macroeconomic phenomena emerge endogenously without exogenous shocks:

Endogenous Kitchin Cycles: Inventory cycles (period $\approx$ 2,500 ticks) emerge from the second-order negative feedback between investment (diminishing returns) and depreciation.
Financial Accelerator: Wealth inequality (Gini) exhibits an overshoot-reversion pattern driven by endogenous credit dynamics (lending increases during recessionary gaps and stabilizes during booms), without interest rate adjustments or monetary policy.
Quality Matthew Effect: Employers (top capital holders) achieve higher tool quality than non-employers, but credit markets mitigate this gap by allowing borrowers to invest in quality.
Lemon Market Adverse Selection: Information asymmetry in quality signaling leads to a stable adverse selection metric (14.1% of trades involve below-average quality goods) and a quality premium of 1.78.
Credit–Trade Crowding: Enabling lending reduces aggregate trade volume by ~29% as resources are transferred from active traders to borrowers, reflecting a frictionless peer-to-peer credit channel.

B. Validation Findings

Traceability and Necessity: An ablation matrix (5 treatments $\times$ 10 seeds) confirms that each emergent phenomenon maps to a unique necessary physical constraint. No redundancy exists; removing a specific constraint (e.g., quality investment) eliminates specific phenomena (e.g., the Quality Matthew Effect) while leaving others intact.
Quality Decoupling Invariance: The decoupled architecture proves robust. Tool and goods quality metrics remain invariant (variation < 6%) across 39 experimental conditions, including:
- Capital collapse (tools reduced to 8).
- 74% capital destruction (depreciation storm).
- Extreme parameter sweeps (depreciation rates from 0.005 to 0.08).
- Institutional reconfigurations (employer ratio varied from 5% to 40%).
Parameter as Theory: Different parameter values define distinct "economic universes" rather than calibrations of a single economy. For example:
- Low depreciation ( $\delta=0.005$ ) creates a "high-capital society."
- High depreciation ( $\delta=0.08$ ) creates a "high-turnover society."
- Low capital scale ( $K=3$ ) creates an "artisan society"; high scale ( $K=50$ ) creates a "post-industrial society."
Resilience: The system recovers fully from a 74% capital destruction shock within ~3,000 ticks. Structural metrics (employment, Gini) remain stable under extreme stress, while quality metrics remain virtually unaffected even when capital stock collapses.

4. Significance and Claims

The paper claims to provide a falsifiable framework for emergence. By operationalizing emergence through ablation (removing a constraint $\rightarrow$ phenomenon disappears), the authors move beyond asserting that phenomena are emergent to testing that claim.

Methodological Shift: The work shifts the modeling paradigm from "which parameters best fit the data?" to "which economic physics generates the phenomena?" Parameters are treated as physical constants defining different economic regimes.
Political Economy Implications:
- Class Differentiation: Social stratification (employers vs. workers) emerges from the physical distribution of tools and diminishing returns, not just institutional property rights.
- Credit as a Conditional Equalizer: Credit can narrow quality inequality (by funding human capital investment) even as it may widen wealth inequality.
- Institutions as Modulators: Institutional configurations (e.g., employer ratios) modulate the degree of phenomena (e.g., lending volume) but do not create or destroy the underlying macro-regularities, which are rooted in physical constraints.
Limitations: The authors acknowledge that the distinction between physical constraints and institutional scaffolding requires further precision, that the model lacks spatial structure and a significant service sector, and that population turnover is too low to observe skill evolution via natural selection.

In conclusion, the paper argues that Emergent Physicalism offers a rigorous design discipline—minimizing behavioral assumptions and maximizing physical constraints—supported by a validation toolkit that can prove the necessity and robustness of macroeconomic regularities.