How 'Economic Temperature' Explains Global Inequality
The richest 1% of Americans hold 32% of the nation’s wealth. In South Africa, it’s 71%. In Sweden, just 18%. These disparities aren’t random—they reflect deep structural differences in how economies function. But for decades, economists have struggled to explain why the mathematical signature of inequality, known as the Pareto exponent, varies so much across countries and time. Some nations have steep wealth tails—meaning extreme wealth is rare. Others have flatter tails, where billionaires are far more common. Conventional models treat this exponent as fixed, baked into the rules of exchange. But a new physics-inspired model flips that idea: the Pareto exponent isn’t fixed at all. It’s tunable—shaped by a single hidden control parameter that behaves like temperature in a gas. And when you turn this dial, the entire shape of inequality shifts.
In a 2026 preprint, physicist Ekrem Aydiner proposes a kinetic trap–diffusion model where money moves between agents not through fixed trading rules, but through thermally activated hops across a disordered economic landscape (Aydiner, 2026). The key innovation? Temperature $T$—not as physical heat, but as a measure of economic mobility. At low $T$, money gets trapped in deep wells, creating concentrated wealth. At high $T$, money diffuses freely, flattening inequality. Crucially, the model generates a range of Pareto exponents $\nu(T)$ from 0.5 to 1.5, matching real-world variation. This isn’t just a better fit to data—it suggests that inequality isn’t destiny. It’s dynamic, responsive, and potentially controllable.
The Science
Aydiner’s model draws from statistical physics, specifically the kinetic theory of gases and disordered systems like spin glasses. It imagines an economy as a one-dimensional lattice of $L=100$ agents, each sitting in a potential well of depth $\phi_i$ drawn randomly from $[0,1]$ (Aydiner, 2026). Money is represented as discrete particles that can hop between neighboring agents. But unlike ideal gas molecules, these particles face energy barriers: moving from a deep well to a shallow one costs energy.
The exchange rule is governed by the Metropolis algorithm—a cornerstone of Monte Carlo simulations. When a money unit attempts to jump from agent $i$ to $j$, the move is accepted with probability:
If the destination well is deeper ($\phi_j > \phi_i$), the move is always accepted—money flows downhill. But if it’s shallower, the jump only succeeds if thermal energy $T$ overcomes the barrier. This creates a competition: trapping (money sinking into deep wells) versus diffusion (money spreading out due to thermal agitation).
The model is conservative—total money $M = \sum x_i$ is fixed—and evolves via kinetic Monte Carlo. Each simulation runs for $5 \times 10^4$ steps, averaged over 1,000 realizations. Initial money per agent is $\delta M = 10$, and each hop transfers one unit. The system reaches a stationary distribution that depends only on $T$, not initial conditions—suggesting robustness.
Unlike traditional kinetic exchange models—such as Drăgulescu–Yakovenko’s random reshuffling or Chakraborti–Chakrabarti’s saving propensity models—this framework doesn’t assume fixed trading rules. Instead, exchange emerges from the interplay of disorder (random $\phi_i$) and mobility (controlled by $T$). And because $T$ tunes the effective diffusivity of money, it naturally generates different inequality regimes.
What They Found
The simulations reveal three striking results.
First, the stationary wealth distribution $P(x)$ follows a power law in the high-wealth tail:
But unlike in most models, $\nu$ is not constant. It varies continuously with temperature, from $\nu \approx 0.5$ at $T=0.01$ to $\nu \approx 1.5$ at $T=10$ (Aydiner, 2026). This range aligns with empirical estimates: U.S. income data often yield $\nu \approx 1.5$, while wealth distributions—more unequal—tend toward $\nu \approx 1.0$ or lower (Reed, 2001; Dragulescu & Yakovenko, 2001). The model doesn’t just reproduce Pareto’s law—it explains why its exponent isn’t universal.
Temperature Dependence of the Pareto Exponent
Pareto exponent ν(T) increases with temperature T, from 0.5 at T=0.01 to 1.5 at T=10.
| Label | Value |
|---|---|
| T = 0.01 | 0.5 |
| T = 0.02 | 0.6 |
| T = 0.05 | 0.75 |
| T = 0.1 | 0.9 |
| T = 0.5 | 1.1 |
| T = 1.0 | 1.25 |
| T = 5.0 | 1.4 |
| T = 10.0 | 1.5 |
Second, the fraction of zero-wealth agents, $f_0(T)$, peaks at intermediate temperatures. At very low $T$, money is trapped but widely distributed across deep wells—few agents are completely empty. At very high $T$, money diffuses so freely that no one accumulates much, but no one is fully excluded either. The maximum in $f_0(T)$ occurs around $T_c \approx 0.05$, suggesting a “sweet spot” for exclusion (Aydiner, 2026).
Fraction of Zero-Wealth Agents vs. Temperature
Fraction of zero-wealth agents f₀(T) peaks around T ≈ 0.05.
| Label | Value |
|---|---|
| T = 0.01 | 0.1 |
| T = 0.02 | 0.15 |
| T = 0.05 | 0.22 |
| T = 0.1 | 0.18 |
| T = 0.5 | 0.12 |
| T = 1.0 | 0.08 |
| T = 5.0 | 0.05 |
| T = 10.0 | 0.03 |
Third, the Gini index—a standard measure of inequality ranging from 0 (perfect equality) to 1 (maximum inequality)—also shows non-monotonic behavior. It dips at low $T$, rises to a peak near $T_c$, then declines again at high $T$. This implies three distinct regimes:
- Low $T$: Money is immobilized. Wealth concentrates in a few deep traps, but most agents retain some funds—moderate Gini, low zero-wealth fraction.
- Intermediate $T$: Trapping and diffusion compete. Some agents lose everything while others accumulate, maximizing inequality and exclusion.
- High $T$: Money flows freely. The system resembles a thermal gas—exponential (Boltzmann-like) distribution, low Gini, but also less accumulation at the top.
Gini Index vs. Temperature
Gini index G(T) peaks near T ≈ 0.05, indicating maximum inequality at intermediate temperature.
| Label | Value |
|---|---|
| T = 0.01 | 0.4 |
| T = 0.02 | 0.45 |
| T = 0.05 | 0.52 |
| T = 0.1 | 0.48 |
| T = 0.5 | 0.4 |
| T = 1.0 | 0.35 |
| T = 5.0 | 0.3 |
| T = 10.0 | 0.28 |
Remarkably, the model produces Pareto-like tails even under strongly nonuniform initial conditions. This robustness suggests the power-law form isn’t an artifact of setup—it’s a stable outcome of the trap–diffusion mechanism.
Why This Changes Things
For economists, inequality has long been a puzzle of origins: Why do some societies have steeper wealth tails than others? Standard answers point to institutions—tax policy, labor markets, inheritance laws. But Aydiner’s model offers a deeper, more physical intuition: inequality may reflect the mobility of capital itself.
Think of $T$ not as literal temperature, but as a proxy for financial fluidity. In a high-$T$ economy, capital moves easily—credit is accessible, markets are liquid, innovation is rewarded. In a low-$T$ economy, money is sticky—trapped in legacy assets, dynastic wealth, or monopolies. The model suggests that when mobility is too low, wealth concentrates. When it’s too high, no one accumulates enough to drive investment. The most unequal outcomes arise not at extremes, but in the tension between them.
This reframes policy. Rather than asking “How do we redistribute?” we might ask “How do we tune the temperature?” Financial inclusion, for instance, could act like heating: enabling more people to participate in wealth creation, reducing the fraction of zero-wealth agents. Conversely, excessive deregulation might overheat the system, eroding savings and long-term stability.
The model also challenges the idea that inequality is static. If $\nu$ is a function of $T$, and $T$ can change, then inequality regimes are metastable—capable of shifting with economic conditions. The 2008 financial crisis, for example, may have represented a sudden drop in effective $T$: credit froze, mobility collapsed, and wealth re-concentrated. Recovery wasn’t just about growth—it was about restoring flow.
Empirically, the model’s predictions align with observed patterns. Countries with high financial development indices—like the U.S. or U.K.—tend to have higher $\nu$ (less extreme wealth concentration in the tail), while rentier economies or those with high barriers to entry exhibit lower $\nu$. Even within countries, $\nu$ fluctuates: during hyperinflation, when money loses value rapidly, the effective $T$ may spike, flattening the distribution temporarily.
But the biggest implication is unification. Previous models explained either exponential (Boltzmann) or power-law (Pareto) distributions. This one produces both: at low $x$, $P(x)$ is nearly exponential; at high $x$, it crosses over to power law. That matches real data, where low incomes follow log-normal distributions and high incomes follow Pareto tails (Aitchison & Brown, 1957). The transition occurs around $x \sim T$, suggesting temperature sets the scale for where “rich” begins.
What’s Next
The model is idealized. It assumes a closed, one-dimensional economy with quenched disorder. Real economies are open, networked, and adaptive. Agents don’t just passively hold wealth—they invest, innovate, and influence the rules. Future work could extend the model to include:
- Network structure: Real exchange isn’t random or nearest-neighbor. Social and financial networks shape who trades with whom. Adding scale-free or small-world topologies could alter diffusion dynamics.
- Endogenous temperature: $T$ is currently exogenous. But in reality, financial mobility depends on policy, technology, and trust. Could $T$ evolve endogenously—say, through feedback from inequality itself?
- Wealth creation: The model conserves money. But real economies grow. Introducing multiplicative processes (e.g., returns on investment) could generate more realistic dynamics, perhaps linking to Kesten processes or Gibrat’s law.
- Policy testing: Could central bank interventions—quantitative easing, interest rates—be modeled as $T$-tuning? If so, the model might offer a new framework for macroprudential policy.
Validation is also needed. While $\nu(T)$ matches empirical ranges, direct calibration to real data is missing. One could estimate effective $T$ from financial velocity, credit spreads, or market liquidity indices, then test whether predicted $\nu$ matches observed wealth tails.
And of course, causality remains open. Does low mobility cause higher inequality, or is it a symptom? The model suggests the former, but real-world confounders—power, politics, history—are hard to disentangle.
Still, the core insight stands: inequality isn’t just about who has money. It’s about how easily it moves. By treating the economy as a physical system with tunable dynamics, Aydiner offers not just a new model—but a new way of thinking. Temperature isn’t in the air. It’s in the flow of capital. And if we can measure it, we might just learn how to regulate it.