Learning to respect property by refashioning theft into trade

Agent-based simulations and human-subject experiments explore the emergence of respect for property in a specialization and exchange economy with costless theft. Software agents, driven by reciprocity and hill-climbing heuristics and parameterized to replicate humans when property is exogenously protected, are employed to predict human behavior when property can be freely appropriated. Agents do not predict human behavior in a new set of experiments because subjects innovate, constructing a property convention of “mutual taking” in 5 out of the 6 experimental sessions that allows exchange to crowd out theft. When the same convention is made available to agents, they adopt it and again replicate human behavior. Property emerges as a social convention that exploits the capacity for reciprocity to sustain trade.