TL;DR
This paper tackles the challenge of fairly distributing a budget among public projects in participatory budgeting (PB) by extending fairness concepts from committee selection to PB. The authors introduce new proportionality axioms and analyze existing ones, providing a systematic overview of how fairness can be achieved in PB.
What Problem Does This Paper Address?
Participatory budgeting (PB) is a democratic process where citizens decide how to allocate public funds to projects. A key aspect of making PB fair is ensuring that the allocation of the budget is "proportional," meaning it reflects the preferences of the voters. While fairness, particularly proportionality, has been extensively studied in the context of selecting a committee of representatives (multi-winner voting or MWV), extending these ideas to PB, where projects have varying costs and budgets are allocated, presents unique challenges. This paper aims to systematize the understanding of proportionality in PB by defining new axioms and analyzing how existing ones, originally from MWV, behave in the PB setting.
Methods
The authors extend existing proportionality axioms from multi-winner voting (MWV) to participatory budgeting (PB). They introduce two novel axioms: Proportional Justified Representation (PJR) for PB and Laminar Proportionality (LP) for PB. PJR, in its new form, ensures that for any group of voters who are "cohesive" (agree to a certain degree on a set of projects), the winning allocation provides enough utility to at least one member of that group. LP, on the other hand, deals with situations where the election instance has a "laminar" structure, allowing for proportional allocation across different sub-groups or stages. The paper then analyzes how these new axioms, along with existing ones like "core" and "priceability," apply to well-known PB rules such as Phragmén, Proportional Approval Voting (PAV), and Rule X. They investigate the logical implications between these axioms and assess the properties of these rules.
Datasets
Not discussed in the paper. The paper uses theoretical examples and profiles to illustrate concepts and prove theorems, rather than relying on real-world datasets.
Key Insights and Ideas
- Bridging MWV and PB: The core insight is that fairness concepts from committee selection (MWV) can be meaningfully adapted to participatory budgeting, but require careful redefinition due to differing structures (e.g., varying project costs).
- New Axioms for PB: The introduction of PJR and LP for PB provides more nuanced ways to capture proportionality in budget allocation scenarios.
- Rule X's Versatility: Rule X emerges as a promising rule, satisfying several proportionality axioms, including a "proportional justified representation up to one project" variant, suggesting its robustness.
- Phragmén's Strengths and Weaknesses: Phragmén is shown to satisfy PJR and priceability, but not LP, highlighting a trade-off in its fairness properties.
- PAV's Limitations: PAV, while popular, struggles with some key proportionality axioms in the PB setting, particularly PJR.
- Systematic Overview: The paper provides a valuable systematization of proportionality axioms in PB, clarifying logical relationships and identifying gaps in current understanding.
Key Results
- Phragmén: Satisfies PJR and priceability in PB. It does not satisfy EJR or the core in PB (or MWV). It also does not satisfy LP in PB.
- PAV: Does not satisfy PJR in PB. It also does not satisfy the core, priceability, or LP in PB.
- Rule X: Satisfies PJR-up-to-one and EJR-up-to-one in PB. It also satisfies priceability and LP in PB.
- Axiom Relationships: The paper establishes several novel logical implications between proportionality axioms in PB, contributing to a clearer understanding of their hierarchy. For instance, EJR implies PJR in PB.
- Priceability vs. PJR: The paper demonstrates that priceability does not imply PJR in PB, and vice-versa, showing these are distinct fairness criteria in this context.
Limitations and Future Work
- Focus on Approval/Utility-Based: The current analysis primarily focuses on approval-based and utility-based PB settings. Future work could extend this to ordinal preferences.
- Specific Axioms: The paper focuses on a subset of proportionality axioms. Exploring other fairness criteria and their interactions would be beneficial.
- Richer PB Settings: The authors suggest extending the study to more complex PB scenarios, such as those with diversity constraints, project groups, negative attitudes, or different resource types.
- Laminar Proportionality Nuances: While LP is shown to imply priceability in laminar PB-instances, the paper also notes that laminar proportional bundles don't always satisfy core, EJR, or PJR, suggesting further investigation into conditions that guarantee these properties.
Summary updated