Can Technological Change Weaken the Robustness of Common-property Regimes?

Maija Halonen-Akatwijuka (University of Bristol)
Evagelos Pafilis (King's College London)

Abstract : We examine the effect of technological change on the incentives to cooperate in the provision of common-pool resources (CPRs). We focus our analysis on CPRs that require investments in improvement and maintenance, such as irrigation systems. We find that major technological improvements, such as replacing a primitive irrigation system with a modern system, risk compromising cooperation as the temptation to freeride on other farmers’ investments is increased. By contrast, minor technological improvements within an existing irrigation system, such as strengthening water diversion devices, do not hinder incentives to cooperate. In our analysis, an irrigation system can be well-managed for a long period of time during technological progress when changes are minor. When technology changes are major, cooperation can be maintained if the community is patient and initially their discount factor is well above the critical level for cooperation. However, when the threshold is reached, any further major technological improvement will lead to a breakdown of cooperation and collapse of investments in the irrigation system.

Silly Rules Improve the Capacity of Agents to Learn Stable Enforcement and Compliance Behaviors

Raphael Koster (DeepMind)
Dylan Hadfield-Menell (UC Berkeley)
Gillian K Hadfield (University of Toronto)
Joel Z Leibo (DeepMind)

Abstract : How can societies learn to enforce and comply with social norms? Here we investigate the learning dynamics and emergence of compliance and enforcement of social norms in a foraging game, implemented in a multi-agent reinforcement learning setting. In this spatiotemporally extended game, individuals are incentivized to implement complex berry-foraging policies and punish transgressions against social taboos covering specific berry types. We show that agents benefit when eating poisonous berries is taboo, meaning the behavior is punished by other agents, as this helps overcome a credit-assignment problem in discovering delayed health effects. Critically, however, we also show that introducing an additional taboo, which results in punishment for eating a harmless berry, improves the rate and stability with which agents learn to punish taboo violations and comply with taboos. Counterintuitively, our results show that an arbitrary taboo (a "silly rule") can enhance social learning dynamics and achieve better outcomes in the middle stages of learning. We discuss the results in the context of studying normativity as a group-level emergent phenomenon.

Mass Political Action and Revolutionary Capital

Weijia Li (Monash University)
Gerard Roland (University of California, Berkeley)
Yang Xie (University of California, Riverside)

Abstract : Why are most mass political action, when they occur, of very short duration? Why does the onset of revolution and mass protests, seem often so unpredictable? We provide unified answers to these questions by building a model based on the concept of revolutionary capital, a state variable that evolves stochastically and endogenously over time, and the stock of which is the basis for a revolutionary leader's decision to engage intertemporally in mass action. The key feature of the equilibrium dynamics is the hazard rate of the status quo to the scale of action. The predictions of the model are consistent with preliminary empirics of reported anti-government protests across 162 countries over 1990--2018.

Do a Few Bad Apples Spoil the Barrel?: Community Enforcement with Incomplete Information

Takuo Sugaya (Stanford GSB)
Alexander Wolitzky (MIT)

Abstract : We study the repeated prisonerĂ­s dilemma with random matching when some players may be "bad types" who never cooperate. We establish an anti-folk theorem: with anonymous players, cooperation is impossible in large groups under a smoothness assumption on the distribution of the number of bad types. Communities may avoid this grim outcome by segregating themselves into smaller sub-groups, at the cost of forgoing some gains from trade. Making players' identities observable does not help much: cooperation remains impossible in groups whose size N is large relative to the discount factor delta, in that (1-delta)*(sqrt of N) going to infinity. However, allowing within-match cheap talk supports cooperation in much larger groups: those where (1-delta)*(log of N) going to 0. Thus, in contrast to the situation where all players are rational, communication is essential for supporting cooperation in large groups in the presence of a few bad apples