🔬 physics

Relations Between the Inequality Indices Gini, Pietra and Kolkata: Theory and Data Analysis

This paper rigorously establishes theoretical relations and validates them through empirical data analysis across US income, Bollywood earnings, and Nobel laureate citations, demonstrating that while the Pietra index is consistently slightly larger than the excess wealth fraction $2k-1$ (with deviations under 5%), simple analytic approximations linking the Gini, Pietra, and Kolkata indices hold only for low inequality values.

Original authors: Asim Ghosh, Bikas K. Chakrabarti

Published 2026-03-17

📖 5 min read🧠 Deep dive

CC0 1.0

Original authors: Asim Ghosh, Bikas K. Chakrabarti

Original paper dedicated to the public domain under CC0 1.0 (http://creativecommons.org/publicdomain/zero/1.0/). ✨ 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 measure how "unfairly" a pie is being shared among a group of people. Some people get huge slices, while others get crumbs. Economists and scientists have invented three different rulers to measure this unfairness. This paper is like a detective story where the authors compare these three rulers to see how they relate to each other and if they tell the same story.

Here is the breakdown of the paper in simple, everyday language:

1. The Three Rulers of Inequality

The paper looks at three specific ways to measure inequality, all based on a famous graph called the Lorenz Curve (think of this as a map showing who has what).

The Gini Index (The "Area" Ruler):
Imagine the Lorenz curve is a squiggly line on a graph. The Gini index measures the empty space between that squiggly line and a straight diagonal line (which represents perfect equality). The bigger the empty space, the more unequal the society is. It's like measuring how much the pie distribution deviates from a perfect, flat table.
The Pietra Index (The "Robin Hood" Ruler):
This measures the biggest single gap between the squiggly line and the straight line. It answers the question: "What is the maximum amount of money we would need to take from the rich and give to the poor to make everyone equal?" It's the "Robin Hood" number—the amount of wealth redistribution needed to fix the system.
The Kolkata Index (The "80-20" Ruler):
This is the newest ruler (introduced in 2014). It asks a specific question: "What fraction of the total wealth is owned by the top fraction of people?"
- If the answer is 0.8, it means the top 20% of people own 80% of the wealth. This is the famous "Pareto Principle" or the 80-20 rule.
- The Kolkata index finds the exact point where this happens. If the index is 0.8, it means the richest 20% hold 80% of the pie.

2. The Big Discovery: The "Robin Hood" Connection

The authors found a fascinating mathematical link between the Pietra Index (Robin Hood) and the Kolkata Index.

The Theory: If the top 20% (1 - 0.8) own 80% (0.8) of the wealth, they are holding 60% more than their "fair share" (since a fair share for 20% of people would be 20%).
- Mathematically, this "excess" is calculated as $2k - 1$ .
- The authors proved that the Pietra Index (the amount needed to redistribute) should theoretically be exactly equal to this excess amount ( $2k - 1$ ).
The Reality Check: When they looked at real-world data, the Pietra Index was almost exactly equal to this number, but slightly higher (by less than 5%).
- Analogy: Imagine you calculate you need to take $50 from the rich to feed the poor. The math says you need exactly $50. In reality, you might need $52. It's very close, but not a perfect match. The paper confirms that the "Robin Hood" amount is basically the "excess wealth" held by the top tier.

3. Testing the Rulers on Real Life

The authors didn't just do math on paper; they tested these rulers on three very different types of "pie":

US Income (IRS Data): They looked at tax returns from 1983 to 2022.
- Result: Inequality has been growing. The "Kolkata" number has gone up, meaning the top 20% (or fewer) are grabbing an even bigger slice of the pie over time.
Bollywood Movie Income: They looked at how much money Indian movies made from 1999 to 2024.
- Result: A few super-hits make almost all the money, while most movies make very little. The inequality here is extreme, and the rulers agreed on the pattern.
Nobel Prize Citations: They looked at how often papers by Nobel winners are cited by other scientists (2020-2025).
- Result: A tiny number of scientists get the vast majority of the attention (citations). This is a "winner-take-all" scenario.

4. What About the "Simple Formulas"?

The authors tried to use simple, straight-line formulas to predict how these rulers relate to each other (e.g., "If Gini goes up by 1, Pietra goes up by 0.75").

The Verdict: These simple formulas work okay when inequality is low (like in a very fair society). But as inequality gets high (like in the US tax data or Hollywood), the simple formulas break down.
The Lesson: You can't use a simple ruler to measure a complex, twisted shape. When the gap between rich and poor gets huge, the relationship between these indices gets messy and non-linear.

Summary: The Takeaway

This paper is a confirmation that our mathematical tools for measuring inequality are consistent.

The Kolkata Index tells us who holds the wealth (e.g., "The top 15% hold 85%").
The Pietra Index tells us how much needs to be moved to fix it.
The paper proves that the amount needed to fix it (Pietra) is almost exactly equal to the "extra" wealth the rich are holding (derived from Kolkata).

In a nutshell: The authors showed that while the math gets complicated when inequality is high, the core idea remains simple: The "Robin Hood" amount needed to fix society is directly tied to how much the top tier is hoarding beyond their fair share. And unfortunately, in the real world (US taxes, movies, and science), that hoarding is getting bigger every year.

1. Problem Statement

The paper addresses the lack of rigorous theoretical connections between three prominent inequality indices derived from the Lorenz function $L(x)$ :

Gini Index ( $g$ ): Measures the normalized area between the line of perfect equality and the Lorenz curve.
Pietra Index ( $p$ ): Also known as the Robin Hood or Hoover index, it measures the maximum vertical distance between the equality line and the Lorenz curve. It represents the fraction of total wealth that must be redistributed to achieve equality.
Kolkata Index ( $k$ ): A newer index (introduced in 2014) defined as the non-trivial fixed point of the complementary Lorenz function $L_c(x) = 1 - L(x)$ . It represents the fraction of wealth possessed by the richest $1-k$ fraction of the population (generalizing Pareto's 80-20 law where $k=0.8$ ).

The authors aim to determine if strict mathematical relationships exist between these indices, specifically testing theoretical derivations against extensive empirical data across economics, entertainment, and academia.

2. Methodology

Theoretical Analysis

The authors employ two main theoretical approaches:

Geometric Rigor: They analyze the geometric definitions of the indices. Since the Kolkata index $k$ implies that the richest $1-k$ fraction holds wealth $k$ , the "excess" wealth held by this group over an equal distribution is $k - (1-k) = 2k - 1$ .
Polynomial Expansion: To derive approximate linear relationships for low inequality, they assume a minimal polynomial form for the Lorenz function: $L(x) = Ax + Bx^2$ (where $A+B=1$ ). This allows for analytical derivation of $g$ , $p$ , and $k$ in terms of the coefficient $B$ .

Numerical Data Analysis

The theoretical predictions are tested against three distinct datasets spanning different domains and time periods:

US Income Data (1983–2022): Analyzed using IRS data for both total individual income and tax returns.
Bollywood Movie Income (1999–2024): Box office earnings for Indian films, representing market-driven inequality in the entertainment sector.
Nobel Laureate Citations (2020–2025): Citation counts for 40 Nobel Prize winners in Economics, Physics, Chemistry, and Medicine, representing inequality in scientific impact.

3. Key Contributions and Theoretical Derivations

The paper establishes the following theoretical relations:

The Pietra-Kolkata Lower Bound:
The authors rigorously prove that the Pietra index must be greater than or equal to the excess wealth fraction defined by the Kolkata index:
$p \geq 2k - 1$
Reasoning: The Pietra index is the maximum vertical distance between the equality line and the Lorenz curve. At the specific point $x=k$ , the vertical distance is exactly $k - L(k) = k - (1-k) = 2k - 1$ . Since $p$ is the maximum over all $x$ , $p$ must be at least this value.
The Robin Hood Interpretation:
The paper clarifies that if the Robin Hood index is interpreted strictly as the fraction of wealth the rich must transfer to the poor to equalize incomes, this value is exactly $2k - 1$ . However, the standard Pietra index ( $p$ ) is defined geometrically as the maximum distance, which is slightly larger than the specific distance at $x=k$ .
Small-Inequality Approximations:
Using the minimal polynomial expansion $L(x) = Ax + Bx^2$ , the authors derive linear relationships valid for small Gini values ( $g \to 0$ ):
1. Pietra-Gini Relation: $p = \frac{3}{4}g$
2. Kolkata-Gini Relation: $k = \frac{1}{2} + \frac{3}{8}g$

4. Results

Empirical Validation of $p \geq 2k - 1$

Analysis of all three datasets confirms the theoretical inequality $p \geq 2k - 1$ .

The ratio $p / (2k - 1)$ is consistently greater than unity.
The deviation is remarkably small, never exceeding 5% across all datasets and years.
This suggests that while the Pietra index is technically the maximum distance, the maximum occurs very close to the Kolkata fixed point $x=k$ in real-world distributions.

Failure of Linear Approximations for High Inequality

The derived linear relations ( $p = 0.75g$ and $k = 0.5 + 0.375g$ ) derived from the minimal polynomial assumption hold only for low inequality values.

Deviation: As $g$ and $k$ increase (which is observed in the data, particularly in US tax data and Nobel citations where $k$ approaches or exceeds 0.87), the linear relations deviate significantly from observations.
Trend: In high-inequality scenarios (e.g., US tax data, $k \approx 0.82$ ; Nobel citations, $k \approx 0.85-0.95$ ), the actual values of $p$ and $k$ diverge from the simple linear predictions, indicating that the Lorenz function in these systems is more complex than a simple quadratic form.

Observations on Specific Datasets

US Income: Inequality has generally increased over time (1983–2022), with $k$ values rising from ~0.67 to ~0.72 (income) and ~0.74 to ~0.82 (tax returns).
Bollywood: Shows a trend of increasing inequality, with $k$ values rising from ~0.64 to ~0.87.
Nobel Citations: Exhibits the highest inequality, with $k$ values often exceeding 0.85 and approaching 0.95, suggesting a "self-organized critical" state where a small fraction of papers dominates citations.

5. Significance

Unification of Indices: The paper provides a rigorous geometric link between the older Gini/Pietra indices and the newer Kolkata index, showing they are not independent but bound by strict inequalities.
Refinement of the Robin Hood Concept: It clarifies that the "Robin Hood" redistribution amount is theoretically $2k-1$ , while the Pietra index is a slightly looser upper bound ( $p \geq 2k-1$ ).
Universality vs. Specificity: The study highlights that while simple analytic models (like the quadratic Lorenz curve) work for low inequality, real-world systems (especially those with self-organized critical dynamics like scientific citations) exhibit higher inequality levels where simple linear approximations fail.
Data-Driven Validation: By analyzing four decades of diverse data, the authors demonstrate that the relationship $p \approx 2k - 1$ is a robust empirical law, holding true across vastly different economic and social systems with high precision.

In conclusion, the paper successfully establishes that the Kolkata index provides a precise lower bound for the Pietra index ( $p \geq 2k-1$ ) and that while simple linear relations exist for low-inequality regimes, real-world high-inequality systems require more complex modeling.