Towards macroeconomic analysis without microfoundations: measuring the entropy of simulated exchange economies

This paper demonstrates through computer simulations that the entropy of exchange economies can be empirically measured using a calorimetric approach, validating the framework of thermal macroeconomics by confirming that entropy acts as a concave, path-independent state function even in complex systems where traditional microfoundations are infeasible.

The Gist

Here is an explanation of the paper "Towards Macroeconomic Analysis Without Microfoundations," translated into simple, everyday language with creative analogies.

The Big Idea: The "Black Box" Economy

Imagine you are trying to understand how a massive, chaotic crowd moves through a city.

• The Old Way (Microfoundations): You try to track every single person. You ask: "What is John thinking? Is Sarah hungry? Is Mike in a rush?" You try to build a giant computer model of every individual's brain, their social norms, and their mood. The problem? It's impossible. People are too complex, and the math gets too messy to solve.
• The New Way (Thermal Macroeconomics): Instead of looking at individuals, you look at the crowd as a whole. You don't care who John is; you only care about the total number of people, the total heat of the crowd, and the pressure they exert on the buildings.

This paper argues that we can do the same thing with money and goods. We don't need to know exactly how every single person decides to buy a coffee or sell a stock. We can treat the whole economy like a thermodynamic system (like steam in an engine) and measure a hidden quantity called "Economic Entropy."

---

The Core Concept: Economic Entropy

In physics, entropy is a measure of disorder or the number of ways a system can be arranged. In this paper, the authors define Economic Entropy as a measure of the "value" or "potential" of an economy's total resources.

Think of it like this:

• Physics: If you have a cup of hot coffee and a cup of cold milk, you know they will mix to become lukewarm. You don't need to know the path of every milk molecule to predict this. You just need the laws of thermodynamics.
• Economics: If you connect two economies (like two countries trading), money and goods will flow until they reach a "balance." The authors claim this balance is governed by Entropy. If you know the Entropy of an economy, you can predict prices, the value of money, and how the economy will react to changes, without ever asking a single person what they want.

The Problem: How Do You Measure Something You Can't Calculate?

In simple economies (where everyone is identical and rational), you can calculate Entropy using math formulas. But in the real world, people are messy. They have different rules, they make mistakes, and they care about what their neighbors have.

So, how do you measure Entropy in a messy, complex economy?

The Solution: The "Economic Calorimeter"

In physics, to measure the heat capacity of a strange new metal, you don't need to know its atomic structure. You just put it in contact with a known substance (like water) and watch how heat flows. This is called calorimetry.

The authors did the same thing with computers:

1. The Test Subject: They created a complex, messy simulated economy (the "strange metal").
2. The Meter: They attached a simple, known economy (the "water") to it.
3. The Exchange: They let them trade money and goods.
4. The Measurement: By watching how the "meter" economy reacted (how prices changed), they could calculate the Entropy of the messy economy.

They did this step-by-step, like walking a grid, measuring the "temperature" and "pressure" of the economy at every step to map out the Entropy function.

The Experiments: Testing the "Black Box"

The authors ran computer simulations with three types of tricky economies to see if this "calorimeter" method worked:

1. The "Satiated" Agents: Imagine people who only want a little bit of a good (like a specific type of cheese). Once they have enough, they don't want more, even if it's free.

   • Result: The math was too hard to solve from the bottom up. But the "calorimeter" successfully measured the Entropy, and it turned out to be a smooth, predictable curve.

2. The "Picky" Agents: Imagine people who only buy if the price is between $0.90 and $1.10. If it's outside that range, they refuse to trade.

   • Result: Again, the math was a nightmare. But the method worked. The Entropy was measurable, and the economy behaved exactly like a thermodynamic system.

3. The "Social" Agents: Imagine people who care about what their neighbors have. If their neighbor has more, they feel poorer.

   • Result: This is the most complex scenario. The authors couldn't prove the math worked. But when they ran the simulation, the Entropy still measured perfectly, and the economy followed the rules of thermodynamics.

The "Aha!" Moment: Path Independence

The most exciting part of the paper is a concept called Path Independence.

Imagine you are hiking up a mountain.

• Path A: You take a steep, rocky trail.
• Path B: You take a long, winding dirt road.
• The Result: If you start at the bottom and end at the top, your altitude is the same, no matter which path you took. Altitude is a "state function."

The authors tested if Economic Entropy is like altitude. They moved the economy from Point A to Point B using different trading routes (different sequences of trades).

• The Finding: In every single case, the final Entropy value was the same, regardless of the route taken.
• Why it matters: This proves that Entropy is a real, fundamental property of the economy, just like altitude is a real property of a mountain. It doesn't matter how the economy got there; the "state" is what matters.

Why This Changes Everything

1. We Can Stop Trying to Model Every Brain: We don't need to understand the complex psychology of every human to predict inflation or trade flows. We can treat the economy like a gas in a box.
2. It Works for Messy Systems: The method works even when the rules are weird, people are irrational, or they care about their neighbors.
3. It's Concave: In all their tests, the Entropy curve was "concave" (shaped like a bowl). This is a specific prediction of the theory, and the data confirmed it. This suggests the theory is robust.

The Bottom Line

This paper is a "proof of concept." It shows that we can measure the "health" and "potential" of an economy (Entropy) by observing how it trades with a simple reference economy, just like a physicist measures heat.

It suggests that macroeconomics can stand on its own feet, without needing to be built on the shaky, impossible-to-solve foundations of individual human behavior. We can study the weather (the economy) without needing to track every single water molecule (the individual).

In short: You don't need to know how the engine works to know how much fuel it needs; you just need to measure the heat. The authors have shown we can do this for money, too.

Technical Summary

Here is a detailed technical summary of the paper "Towards Macroeconomic Analysis Without Microfoundations: Measuring the Entropy of Simulated Exchange Economies" by Y. Luo, R.S. Mackay, and Nick Chater.

1. Problem Statement and Motivation

The paper addresses a fundamental limitation in modern economics: the reliance on microfoundations (deriving aggregate phenomena from individual agent behaviors).

• The Challenge: Constructing micro-founded models is inherently difficult because individual agents are complex, boundedly rational, and influenced by social norms. Standard economic models often require radical simplifications (e.g., perfect rationality) to remain analytically tractable, which limits their realism. Conversely, agent-based simulations allow for realistic complexity but often lack a unifying macro-level analytical framework.
• The Proposed Solution: The authors propose Thermal Macroeconomics (TM), an approach analogous to classical thermodynamics. Just as 19th-century thermodynamics described macroscopic systems (pressure, temperature) without needing to know the molecular details of gases, TM aims to describe aggregate economic phenomena using an entropy function derived from macro-axioms, independent of specific micro-foundations.
• The Gap: While TM theory suggests that entropy can be measured empirically (analogous to calorimetry in physics) even when micro-foundations are intractable, this had not been computationally demonstrated for complex economic systems. The paper aims to prove the feasibility of measuring economic entropy in simulated exchange economies where analytical derivation is impossible.

2. Methodology

The authors employ Agent-Based Modeling (ABM) combined with a numerical integration technique inspired by physical calorimetry.

A. Theoretical Framework

• Axioms: The system assumes the economy satisfies the axioms of TM (based on Lieb and Yngvason's axiomatic thermodynamics).
• State Variables: The economy is defined by the aggregate amounts of Money ($M$) and Goods ($G, H$).
• Entropy Definition: The core hypothesis is that the economy possesses an entropy function $S(M, G, H)$ such that for quasi-static changes:
  $$dS = \beta dM + \nu_g dG + \nu_h dH$$
  Where $\beta$ is the value of money (inverse of "economic temperature") and $\nu_j$ are the values of goods.

B. Measurement Protocol (Economic Calorimetry)

To measure $S$ without knowing the micro-rules:

1. The "Meter": A reference economy with known properties (a Cobb-Douglas economy) is attached to the target economy $E$.
2. Equilibration: Money and goods are exchanged between $E$ and the meter until equilibrium is reached. At equilibrium, the "values" ($\beta, \nu_j$) equalize.
3. Data Collection: The values $\beta$ and $\nu_j$ are measured in the meter economy (as time-averages of utility parameters) for various states of $E$.
4. Numerical Integration:
   • A reference state $(M^*, G^*, H^*)$ is chosen with $S=0$.
   • The system is stepped through a grid of states $(M, G, H)$.
   • $S$ is approximated by integrating the measured values:
     $$S(M', G', H') \approx S(M, G, H) + \frac{1}{2}(\beta + \beta')dM + \frac{1}{2}(\nu_g + \nu'_g)dG + \frac{1}{2}(\nu_h + \nu'_h)dH$$
5. Validation:
   • Path Independence: The calculated $S$ must be a state function. The authors quantify the deviation from path independence using the Residual Sum of Squares (RSS) relative to the Total Sum of Squares (TSS). A low RSS/TSS ratio indicates a valid entropy function.
   • Concavity: The resulting entropy surface is checked for concavity, a key prediction of TM.
   • Agreement: For cases where analytical solutions exist, the measured $S$ is compared to the theoretical $S$.

3. Key Contributions

1. Proof of Concept for Macro-Analysis: The paper demonstrates that an entropy function can be successfully measured for simulated economies even when the underlying micro-foundations are too complex for analytical derivation.
2. Validation of Path Independence: The study confirms that for a wide range of economic models, the measured "economic entropy" is a state function (path-independent), validating the applicability of thermodynamic axioms to these systems.
3. Extension to Intractable Systems: The method is applied to systems where micro-foundational analysis is mathematically impossible (e.g., agents with non-linear utility thresholds or social interdependencies), showing that TM principles still hold.
4. Autonomous Macro-Level Explanation: The results suggest that macroeconomic behavior can be understood and predicted via entropy without needing to resolve the "black box" of individual agent complexity.

4. Results

The authors tested three categories of economies:

A. Known Analytical Cases (Verification)

• Cobb-Douglas (CD) Economies: Both homogeneous and heterogeneous agents.
  • Result: The measured entropy matched the theoretical logarithmic formula with high precision (Goodness of Agreement $\approx 10^{-6}$ to $10^{-7}$).
• Substitutes and Complements: Agents with utility functions based on sums or minimums of goods.
  • Result: Excellent agreement with the theoretical entropy derived via Legendre transforms of free energy functions.
• Mixed Economies: Combinations of CD, substitute, and complement agents.
  • Result: The method successfully recovered the complex entropy surface, confirming path independence.

B. Intractable Cases (Proof of Concept)

• Satiable Agents: Agents with utility functions that peak and decline (non-monotonic). The partition function was analytically intractable.
  • Result: The measured entropy was path-independent and concave.
• Price-Sensitive Agents: Agents who only trade if the price falls within a specific threshold. This introduces non-reversibility in some configurations.
  • Result: Despite the complexity and potential non-reversibility, the system exhibited a well-defined, concave, path-independent entropy function.
• Interdependent Agents: Agents whose utility depends on the possessions of their neighbors (social comparison).
  • Result: Even with complex social dependencies, the entropy function was measurable, path-independent, and concave.

General Observation: In all tested cases, the entropy function was concave, and the Goodness of Fit (path independence) was extremely high (RSS/TSS ratios typically $< 10^{-5}$).

5. Significance and Implications

• Decoupling Macro from Micro: The paper provides strong evidence that macroeconomic laws (specifically the Second Law of Thermodynamics applied to economics) may hold even when micro-foundations are unknown, complex, or non-standard. This challenges the necessity of "perfect rationality" assumptions for macro-prediction.
• Robustness: The entropy function appears robust to changes in micro-parameters (e.g., encounter rates, specific utility shapes), suggesting the existence of autonomous macro-level generalizations.
• Practical Application: While currently applied to simulations, the methodology offers a blueprint for analyzing real economies. If "economic calorimetry" could be applied to real-world data (by observing responses to injections/removals of money or goods), it could allow policymakers to predict the direction of economic flows and equilibrium states without needing a complete model of individual behavior.
• Limitations and Future Work: The authors note that real economies may violate strict axioms (like extensivity) and that equilibrium timescales must be considered. Future work aims to extend TM to include production, consumption, and financial markets, and to test systems where condensation (herding) might break the axioms.

In conclusion, the paper successfully bridges the gap between thermodynamic theory and economic simulation, demonstrating that entropy is a measurable, state-dependent quantity in exchange economies, offering a powerful new tool for macroeconomic analysis that bypasses the intractability of micro-foundations.