Optimizing scrip systems: efficiency, crashes, hoarders, and altruists

Ian A. Kash, Eric J. Friedman, Joseph Y. Halpern

Abstract

Scrip systems are internal currencies designed to facilitate service exchange while preventing free-riding. However, these systems are prone to recessions and monetary crashes. This paper introduces a game-theoretic framework to model scrip systems with heterogeneous populations. By proving the existence of pure strategy equilibria through strategic complementarities and utilizing maximum entropy distributions, the authors identify the optimal money supply that maximizes social welfare without triggering a system collapse. The study also reveals the surprising roles of altruists as potential crash triggers and hoarders as stabilizing agents.

Introduction

Scrip systems are ubiquitous in both online and offline communities, ranging from "credits" in P2P file-sharing networks to the famous Capitol Hill Baby Sitting Co-op coupons. These systems aim to solve the "free-rider" problem by requiring agents to earn currency by providing services before they can consume them.

Despite their utility, these micro-economies are notoriously fragile. A shortage of money leads to a recession where no one can spend, and thus no one can earn. Conversely, an excess of money can cause a "monetary crash" where agents feel wealthy enough to stop working, leading to a total evaporation of service supply. This research addresses the challenge of modeling these systems with diverse populations to find the "Goldilocks" zone of money supply—maximizing social utility while maintaining stability.

Modeling and Methodology

The authors model the scrip system as an infinite-round game where agents adopt threshold strategies: an agent volunteers to work only if their current wealth is below a specific threshold $k$.

Strategic Complementarities

A key technical contribution is the proof that best-reply functions in these systems are monotonic. This property of strategic complementarities ensures that if other agents increase their willingness to work, an individual agent's best response is also to increase (or at least not decrease) their own willingness. This mathematical foundation guarantees the existence of equilibria in pure strategies, which can be efficiently computed using iterative algorithms.

Monotonicity of the best-reply function

Wealth Distribution and Inference

In large-scale systems, the distribution of money among agents tends toward a state that maximizes entropy, subject to the total money supply. This allows for analytical prediction of the system's wealth distribution. Furthermore, the authors demonstrate an inference technique: by analyzing the log-plot of the money distribution, a system designer can "reverse-engineer" the population to identify different agent types and their internal thresholds.

Log-plot of agent wealth distribution

Key Insights: The Roles of Altruists and Hoarders

The research yields several counter-intuitive insights into system stability:

  • The Welfare-Crash Tradeoff: Social welfare generally improves as money is added because it reduces the number of "broke" agents who cannot request services. However, this improvement is bounded by a sharp threshold; exceeding this point causes an immediate monetary crash.
  • Altruists as a Risk: While seemingly beneficial, too many altruists (who work for free) act as an uncontrolled injection of money. This devalues the currency and can disincentivize standard agents, potentially triggering a crash.
  • Hoarders as Stabilizers: Hoarders—agents who earn but rarely spend—act as a "sink" for excess currency. By removing money from active circulation, they prevent wealth bloat and help stabilize the system against crashes.

Conclusion and Future Directions

The paper proves that scrip systems with payoff heterogeneity possess stable, pure strategy equilibria and provides tools for designers to optimize money supply. However, the model currently assumes fixed prices and does not account for malicious actors or dynamic environments where service values fluctuate. Future work will explore "floating prices" as a mechanism to prevent crashes through inflation, though this may introduce new complexities like speculation.

Summary updated

BibTeX
@inproceedings{kash2007optimizing,
  author = {Ian A. Kash and Eric J. Friedman and Joseph Y. Halpern},
  title = {{Optimizing scrip systems: efficiency, crashes, hoarders, and altruists}},
  booktitle = {EC},
  year = {2007},
  doi = {10.1145/1250910.1250955},
}
Figure