These results shed an interesting light on debates about more planning vs. less planning; urban ecological evolution; neighbourhood formation and stability; private communities; and many other contemporary urban issues. It's main substantial point is that neighbourhoods can develop from bi-lateral action between households that are motivated by individual welfare maximisation. Areas of common consumption and production behaviour governed by informal rules and norms that might appear to be a spatial expression of collective action can emerge through bi-lateral action. The simulation also shows that neighbourhoods can be understood to be areas in which individuals agree on behavioural norms and voluntary regulation regimes for mutual self-interest. There is no community spirit (altruistic behaviour) built into the model. This means that neighbourhood stability is fragile and susceptible to individuals breaking faith. There is no incentive for this in the model once a stable area has evolved but exogenous factors might break the equilibrium and cause knock-on effects in neighbourhood boundaries throughout the city. Wu and Webster (1998) demonstrate something like this and show how random changes of land use at the boundaries of same-use zones cause a constant shifting of natural (market) land-use zones across the city over time in pursuit of economic efficiency. The fragility of neighbourhoods formed by informal contracts might be supposed to lead to a demand for more formal neighbourhood contracts as the city evolves. Returning to the propositions introduced in section II, this is the more likely the higher value of the nuisance problems (the higher the e curves in Figure 1 and the larger the triangle E) and the higher the risk of reneging neighbours. Two outcomes can be predicted: demand for government legislation that grants more powers of control (property rights) over shared neighbourhood facilities to residents; and a rising demand for private neighbourhoods. From one point of view, all forms of micro-neighbourhood governance including devolved municipal powers and proprietary communities (gated suburbs, condominiums, shopping malls, leisure complexes and industrial parks) might be thought of as a response to the capricious nature of voluntary agreements. Even where there are strong planning and environmental nuisance laws, successful residential, commercial and industrial neighbourhoods rely on a good deal of voluntary compliance in the use of public domain attributes. Where this cannot be relied upon there will be a natural evolution towards clearer property rights assignment and greater control whether this is via legal contracts and the market or via government regulation and policing.
There is a rich vein of ideas at the boundaries between the economic theory of property rights; the social science paradigms of emergence; the natural science paradigm of self-organising systems; and the economic analysis of land and property markets including the analysis of alternative levels, styles and tools of intervention. Fruitful lines for further research in the tradition of the work reported in this paper include the following. First, the simulation presented above does not model mobility. This is a limitation in the light of the importance of mobility in neighbourhood formation and evolution. An interesting extension would to allow relocation of households offering unsuccessful good-neighbour bids. Second, the model focuses only on the benefits from mutual restraint and does not allow neighbours to consume-produce beyond their own efficiency margin in retaliation to bad-neighbour gestures. It would be of interest to allow downward or degenerative evolution in this respect. In the extreme, some neighbourhoods may become dominated by asymmetric exchanges in which positive gestures yield equal reciprocal positive responses but negative gestures yield greater negative responses. Such neighbourhoods might turn into welfare black holes as the rent dissipated by producing and avoiding anti-social behaviour approaches the welfare gained by locating in the neighbourhood. The sum might even be allowed to become negative, simulating neighbourhoods in terminal decline -- something analogous, perhaps, to the tragedy of the commons in which the consumer surplus of everyone in the neighbourhood has been used up and there is no choice but to seek to exit. Third, and less bleakly, the simulation described in this paper only allows bilateral contracts. It would be of interest to simulate institutional growth under multi-lateral contracts. This might include an exploration of thresholds in the number of good-neighbour bidders; the introduction of history into the simulation such that institutions evolve as a function of past institutional configurations not just present; and an analysis of transaction costs and prisoner dilemma dynamics. There is also scope for more sensitivity analysis. In the simulations reported, a critical threshold of approximately 14 percent to 15 percent initially seeded good neighbour bids seems to hold. Above this range, the city eventually turns bright red (efficient). The higher the percentage of initial bids, the speedier the convergence to global social efficiency. Below the threshold, cities fragment into a mixture of efficient, inefficient and unstable neighbourhoods. The lower the percentage, the lower the proportion of efficient neighbourhoods at equilibrium. Fourteen to fifteen percent is the tipping threshold. The particular size of the threshold is at one level, a function of the model's cell geometry. Whether there is any relationship between the threshold-geometry association in modelling space and the equivalent association in real city space is an intriguing empirical question that addresses many familiar issues in environmental psychology, urban design and urban sociology.