Paper: arXiv 2404.18467

Authors: Yuyu Chen, Taizhong Hu, Ruodu Wang, Zhenfeng Zou

Abstract

We study stochastic dominance between portfolios of independent and identically distributed (iid) extremely heavy-tailed (i.e., infinite-mean) Pareto random variables. With the notion of majorization order, we show that a more diversified portfolio of iid extremely heavy-tailed Pareto random variables is larger in the sense of first-order stochastic dominance. This result is further generalized for Pareto random variables caused by triggering events, random variables with tails being Pareto, bounded Pareto random variables, and positively dependent Pareto random variables. These results provide an important implication in investment: Diversification of extremely heavy-tailed Pareto profits uniformly increases investors’ profitability, leading to a diversification benefit. Remarkably, different from the finite-mean setting, such a diversification benefit does not depend on the decision maker’s risk aversion.

Complexity vs Empirical Score

  • Math Complexity: 8.5/10
  • Empirical Rigor: 2.0/10
  • Quadrant: Lab Rats — high math complexity, limited empirical testing

Why this score: The paper is highly mathematical, using advanced probability theory (majorization order, stochastic dominance) and complex proofs, but lacks any empirical backtesting or implementation details, focusing purely on theoretical properties of infinite-mean distributions.

Research Flowchart

  flowchart TD
  A["Research Goal:<br>Diversification effect for<br>infinite-mean Pareto models"]
  B["Methodology:<br>Majorization Order & Stochastic Dominance"]
  C["Inputs:<br>iid extremely heavy-tailed<br>Pareto random variables"]
  D["Computation:<br>Analyze portfolios of size n<br>using FSD conditions"]
  E{"Generalization check"}
  F["Key Finding:<br>More diversified portfolios<br>strictly increase FSD"]
  G["Outcome:<br>Diversification benefit<br>independent of risk aversion"]

  A --> B
  B --> C
  C --> D
  D --> E
  E -- Yes --> F
  F --> G

Evidence Card

What the authors claim, on what data, and who has checked it. Read from the paper's own text by a language model: every quote is verbatim, nothing was re-run. How to read this card

Proposes

The paper proposes that a more diversified portfolio of iid extremely heavy-tailed (infinite‑mean) Pareto random variables dominates less diversified portfolios in the sense of first‑order stochastic dominance, and this result is generalized to Pareto variables with triggering events, bounded tails,

Source sentence
With the notion of majorization order, we show that a more diversified portfolio of iid extremely heavy-tailed Pareto random variables is larger in the sense of first-order stochastic dominance. This result is further generalized for Pareto random variables caused by triggering events, random variables with tails being Pareto, bounded Pareto random variables, and positively dependent Pareto random variables.

Assumes 5

  • The random variables are independent and identically distributed Pareto with tail parameter α∈(0,1] (infinite mean). model
    Source sentence
    Let X = (X1, . . . , Xn) be a vector of independent components with Xi ∼Pareto(αi) for all i ∈[n] where 0 < α1 ≤· · · ≤αn ≤1.
  • The preference functional ρ is weakly monotone (and mildly monotone) with respect to stochastic order. other
    Source sentence
    For the rest of our discussions, we make minimal assumptions of monotonicity on ρ in the following two forms. (a) Weak monotonicity: ρ(X) ≤ρ(Y ) for X, Y ∈Xρ if X ≤st Y .
  • Triggering events A₁,…,Aₙ are independent of the Pareto variables X₁,…,Xₙ. data
    Source sentence
    Let X1, . . . , Xn be iid Pareto(α) random variables with α ∈(0, 1], and A1, . . . , An be the respective triggering events of X1, . . . , Xn such that A1, . . . , An are independent of X1, . . . , Xn.
  • Positive dependence is modeled via a Clayton copula. model
    Source sentence
    Theorem 3 deals with extremely heavy-tailed Pareto random variables that are positively dependent and modelled by a Clayton copula.
Show 1 more
  • Pareto random variables may have an upper bound (bounded Pareto). model
    Source sentence
    Proposition 6 studies extremely heavy-tailed Pareto distributions with an upper bound.

Data

No dataset named in the text.

Costs

Not applicable The paper reports no trading or execution result.

Finds 5

  • A more diversified portfolio of iid infinite‑mean Pareto profits uniformly improves preferences for any monotone preference, establishing a diversification benefit. theoretical
    Source sentence
    With the notion of majorization order, we show that a more diversified portfolio of iid extremely heavy-tailed Pareto random variables is larger in the sense of first-order stochastic dominance. These results provide an important implication in investment: Diversification of extremely heavy-tailed Pareto profits uniformly increases investors' profitability, leading to a diversification benefit.
  • The diversification benefit does not depend on the decision maker’s risk aversion, unlike the finite‑mean case. theoretical
    Source sentence
    Remarkably, different from the finite-mean setting, such a diversification benefit does not depend on the decision maker's risk aversion.
  • The optimal diversification strategy under these conditions is the equal‑weight (1/n) rule. theoretical
    Source sentence
    As such improvements are strict, the optimal investment strategy is to assign an equal weight to each investment in the portfolio.
  • Risk‑averse expected‑utility agents also prefer more diversified portfolios of iid Pareto profits, regardless of the tail parameter. theoretical
    Source sentence
    This strategy, known as the 1/n rule, is also optimal for risk-averse expected utility agents if the profits have finite mean. The following result shows that risk-averse expected utility agents always prefer a more diversified portfolio of iid Pareto profits, regardless of the tail parameter α. Proposition 7.
  • Theorem 1 establishes first‑order stochastic dominance of weighted sums when weight vectors satisfy the majorization order. theoretical
    Source sentence
    Now, using Lemmas 1 and 2 and the fact that first-order stochastic dominance is closed under convolution (e.g., Theorem 1.A.3 (b) of Shaked and Shanthikumar (2007)), we obtain the following theorem for n > 2 independent extremely heavy-tailed Pareto random variables. Theorem 1. Let X = (X1, . . . , Xn) be a vector of independent components with Xi ∼Pareto(αi) for all i ∈[n] where 0 < α1 ≤· · · ≤αn ≤1.

Checked by others

No other paper in the archive has been found to support or contradict these findings (compared with 19 related papers). Coverage is partial, so this is neither replication nor refutation.

Cited by 1 paper in the archive (1 by independent authors): Diversification and Stochastic Dominance: When All Eggs Are … (uses its method or data)

Code

No repository by the authors found in the paper.

Extracted 2026-11-14 · same card as JSON · A language model chose these items: check the source sentence before relying on one.

Cited References (34)

  1. Heavy-Tailed Distributions and Robustness in Economics and Finance (2015) — link
  2. New majorization theory in economics and martingale convergence results in econometrics (2005)
  3. Correlation and Dependence in Risk Management: Properties and Pitfalls (2002) — link
  4. Modelling of extremal events in insurance and finance (1994) — link
  5. Capitalism and democracy in the 21st century : proceedings of the International Joseph A. Schumpeter Society Conference, Vienna, 1998 “Capitalism and socialism in the 21st century” (2010)
  6. THE CANONICAL MODEL SPACE FOR LAW‐INVARIANT CONVEX RISK MEASURES IS L1 (2012) — link
  7. The Modelling of Operational Risk: Experience with the Analysis of the Data Collected by the Basel Committee (2004) — link
  8. Quantitative Risk Management: Concepts, Techniques, and Tools (2006) — link
and 26 more
  1. Increasing risk: I. A definition (1970) — link
  2. General Proof that Diversification Pays (1967) — link
  3. Uncertainty and the size distribution of rewards from innovation (2000) — link
  4. Residual Life Time at Great Age (1974) — link
  5. Coherent Measures of Risk (1999) — link
  6. Risk Management: Value at Risk and Beyond (2003) — link
  7. Statistical Inference Using Extreme Order Statistics (1975) — link
  8. Beyond Gaussian averages: redirecting international business and management research toward extreme events and power laws (2007) — link
  9. On peakedness of distributions of convex combinations (1998) — link
  10. Portfolio diversification and value at risk under thick-tailedness (2009) — link
  11. The Dual Theory of Choice under Risk (1987) — link
  12. Nondiversification Traps in Catastrophe Insurance Markets (2008) — link
  13. A Clustering Approach and a Rule of Thumb for Risk Aggregation (2017) — link
  14. Extremal Dependence Concepts (2015) — link
  15. Risk aggregation in multivariate dependent Pareto distributions (2016) — link
  16. Stochastic finance. an introduction in discrete time (2004) — link
  17. Power-law distributions of corporate innovative output: evidence from U.S. patent data (2019) — link
  18. The size distribution of innovations revisited: an application of extreme value statistics to citation and returns measures of patent significance (2007) — link
  19. Statistical Review of Nuclear Power Accidents (2012) — link
  20. Robustness and sensitivity analysis of risk measurement procedures (2010) — link
  21. PORTFOLIO SELECTION* (1952) — link
  22. Inequalities: Theory of Majorization and Its Applications (2010) — link
  23. Technical Note—An Unexpected Stochastic Dominance: Pareto Distributions, Dependence, and Diversification (2024) — link
  24. Risk exchange under infinite-mean Pareto models — → 2403.20171
  25. Some remarks on the effect of risk sharing and diversification for infinite mean risks — → 2411.10139
  26. Convex combinations of random variables stochastically dominate the parent for a new class of heavy tailed distributions (2025) — link

2 of 34 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2596 linked papers.