Sharing the proceeds from a hierarchical venture when agents have needs

1. Problem Statement

The paper addresses a resource allocation problem within a linear hierarchy of agents engaged in a joint venture.

• Context: Agents are organized in a hierarchy (e.g., $M = \{1, \dots, m\}$), where agent 1 is the lowest rank and agent $m$ is the highest.
• Inputs: Each agent $i$ generates a revenue $r_i$ and has an individual need $z_i$.
• Constraint: The model assumes initially that individual revenues cover individual needs ($r_i \geq z_i$), though later extensions relax this to a collective condition ($\sum r_i \geq \sum z_i$).
• Objective: Distribute the aggregate revenue $\sum r_i$ among the agents such that:
  1. The total allocation equals the total revenue (Balance).
  2. Every agent receives at least their need ($x_i \geq z_i$).
  3. The distribution respects the hierarchical structure (revenue "bubbles up" or is shared based on position).

This extends the work of Hougaard et al. (2017), who studied revenue sharing in hierarchies without individual needs, by introducing the Needs Lower Bound as a fundamental axiom of distributive justice.

2. Methodology and Axiomatic Framework

The authors employ axiomatic analysis to characterize allocation rules. They define a set of axioms that capture fairness, consistency, and structural properties of the hierarchy.

Key Axioms:

• Needs Lower Bound: Every agent receives at least their need ($x_i \geq z_i$).
• Lowest Rank Consistency: If the lowest-ranked agent leaves after allocation, the remaining agents' allocations in the reduced problem (where the residual revenue is passed to the next agent) remain unchanged.
• Highest Rank Independence: The allocation of agents below the top rank is unaffected by changes in the top agent's revenue or needs (provided the top agent's need is covered).
• Highest Rank Splitting Neutrality: If the top agent splits their revenue and needs into two agents (one remaining at the top, one added above), the allocation to the original hierarchy members remains unchanged.
• Bilateral Linearity: Linearity holds for two-agent problems regarding revenues and needs.
• Equal Treatment of Equals: Agents with identical revenues and needs receive identical allocations.
• Hierarchical Order Preservation: Higher-ranked agents receive at least as much net surplus (allocation minus need) as lower-ranked agents.
• Canonical Bilateral Fairness: In a two-agent hierarchy where only the bottom agent is productive, the net surplus is split equally.
• Superior Independence & Canonical Symmetric Fairness: Strengthened versions of independence and fairness applied to groups of top agents or scenarios with multiple non-productive superiors.

3. Key Contributions and Results

A. The Family of Need-Adjusted Geometric Rules

The primary contribution is the characterization of a family of rules denoted as $\{\phi^\lambda\}_{\lambda \in [0,1]}$.

• Mechanism: The lowest agent receives their need plus a share $\lambda$ of their surplus ($r_1 - z_1$). The next agent receives their need plus $\lambda$ of their own surplus plus the remaining surplus from the previous agent (which is $(1-\lambda)$ of the previous surplus). This process "bubbles up" the net revenue.
• Formal Definition:
  $$x^\lambda_i = z_i + \lambda(r_i - z_i) + (1-\lambda)x^\lambda_{i-1}$$
• Characterization (Theorem 1): A rule satisfies Needs Lower Bound, Lowest Rank Consistency, Highest Rank Independence, Highest Rank Splitting Neutrality, and Bilateral Linearity if and only if it is a need-adjusted geometric rule.

B. Three Focal Rules ("The Three Musketeers")

The authors isolate three specific rules within the geometric family by adding single axioms:

1. Need-Adjusted Full-Transfer Rule ($\lambda = 0$): The lowest agent keeps only their need; all surplus bubbles up to the highest agent.
   • Characterized by: Highest Rank Independence + Equal Treatment of Equals.
2. No-Transfer Rule ($\lambda = 1$): Each agent keeps their own revenue (no redistribution).
   • Characterized by: Highest Rank Independence + Equal Treatment of Equals (Note: The paper notes a slight nuance in the proof logic, but essentially this rule implies no transfer).
3. Need-Adjusted Balanced-Transfer Rule ($\lambda = 0.5$): The surplus is shared such that the immediate predecessor's surplus is weighted by $1/2$, the one before by $1/4$, etc.
   • Characterized by: Lowest Rank Consistency + Highest Rank Independence + Highest Rank Splitting Neutrality + Canonical Bilateral Fairness.

C. The Need-Adjusted Serial Rule ("D'Artagnan")

The paper introduces a rule outside the geometric family, inspired by the serial cost-sharing rule (Moulin & Shenker, 1992).

• Mechanism: Each agent keeps their need and shares the residual surplus equally with all their predecessors (including themselves).
• Formula:
  $$\phi^s_i = z_i + \sum_{j \leq i} \frac{r_j - z_j}{n - j + 1}$$
• Characterization (Theorem 3): This rule is uniquely characterized by Lowest Rank Consistency, Superior Independence, and Canonical Symmetric Fairness.
• Significance: This rule treats non-productive superiors symmetrically, ignoring the hierarchical "distance" in favor of collective contribution to the structure.

D. Extensions and Domain Variations

1. Collective Needs ($Z^*$): When individual $r_i < z_i$ is allowed (as long as $\sum r \geq \sum z$), the Needs Lower Bound axiom becomes too restrictive, uniquely forcing the Full-Transfer rule. However, using a Weak Needs Lower Bound (applying only when $r \geq z$), the full characterization of geometric rules is recovered.
2. Null Needs ($Z^0$): When $z_i = 0$ for all $i$, the model collapses to the Hougaard et al. (2017) model, recovering their geometric rules and the serial rule (Ju et al., 2025).
3. Two-Agent Case: In the two-agent scenario, the Need-Adjusted Balanced-Transfer Rule and the Need-Adjusted Serial Rule coincide. This is termed the "Folk Solution." The paper highlights that extending this solution to $n > 2$ agents yields different results depending on whether one prioritizes "splitting neutrality" (geometric rules) or "symmetric fairness" (serial rule).

4. Significance and Implications

• Theoretical Advancement: The paper bridges the gap between hierarchical revenue sharing and claims problems with needs. It provides a rigorous axiomatic foundation for distributing resources in organizations where lower-level employees have basic subsistence requirements.
• Policy Relevance: The distinction between the Geometric Rules (which preserve some hierarchy-based incentive structures) and the Serial Rule (which emphasizes collective responsibility and equality among superiors) offers policymakers and organizational designers a spectrum of allocation mechanisms.
  • Geometric rules might suit sales teams or multi-level marketing where "bubbling up" incentives are crucial.
  • Serial rules might suit public goods or cooperative ventures where the hierarchy is merely administrative and all members contribute equally to the structure's existence.
• Methodological Insight: The paper demonstrates how relaxing the "individual rationality" constraint ($r_i \geq z_i$) fundamentally changes the axiomatic landscape, forcing a shift from a family of rules to a single extreme solution unless the axiom is weakened.

In summary, the paper successfully generalizes existing hierarchy models to include individual needs, characterizing a flexible family of geometric rules and a distinct serial rule, thereby offering a comprehensive toolkit for analyzing distributive justice in hierarchical economic settings.