Equilibrium structure of a bidimensional asymmetric city (Q2472897)

This paper aims at explaining the internal structure of a city endogenously, as an equilibrium solution between two different possible uses of land, for residence and for business. In the model, the driving force for concentration, and indeed for the existence of cities in the first place, is a positive externality of labor; the more workers are concentrated at a given place, the more productive each of them becomes. The model differs from the preceding ones in several respects. The lot size (amount of land occupied by one residence) is not assumed to be constant, as in former articles, but is determined endogenously at each location by the equilibrium condition . The cost \(c(x,y)\) of commuting from location \(x\) to location \(y\) is not assumed to be linear in the distance \(|x-y|\); in fact, no specific functional form is assumed, all that is required is that \(c\) is a continuous functions of \(x\) and \(y\). Finally, and perhaps most importantly, the model is truly bidimensional: the (prescribed) shape of the city can be any bounded domain \(\Omega\) in the plane. In the preceding literature, the city is assumed either to be ''long and thin'', in fact one-dimensional, or to have a circular structure, so that the given boundary is a circle and every location is fully characterized by its distance from the center. The main result is an existence theorem: it is proved that there is an equilibrium structure of the city, for any prescribed utility level \(\overline{u}\) and shape \(\Omega\). Very general (and, to us at least, very natural) assumptions on the utility function of consumers, the production function of firms, the externality of labor and the transportation costs are make. As a price to pay for this generality, people living at the same location \(x\) do not all commute to the same location \(y=s(x)\) for work: there is, for each location \(x\), an equilibrium probability \(P_x\) on \(\Omega\), so that \(P_x(y)\) is the probability of working at \(y\), conditional on living at \(x\). To have all individuals living at the same location \(x\) commute to the same location \(y=s(x)\) for work requires an additional assumption on the transportation cost, for instance that \(c(x,y)=C(x-y)\), where \(C\) is a smooth and strictly convex function. It will also be shown, under quite general assumptions, that, at equilibrium, some agents do have to commute: the autarchy situation where every agent works where he/she lives cannot be an equilibrium. It is not clear how to adapt presented method to the closed-city model, where the total population, and not the utility level, is prescribed. On the other hand, we believe that our method can solve other problems, such as a multi-sectorial industry, inhomogeneous terrain, or residence externality between households. Roughly speaking, an equilibrium in the present framework is a configuration (residents/jobs densities, wages/revenues, fraction of land devoted to business use, transportation plan and productivity) in which each agent, firm, landowner solves his pointwise maximization program, wages, revenues and the commuting plan are consistent with free-mobility of labor and finally, productivity reflects production externalities. This yields the formal definition. The basic theorem (Theorem 6.2) and the general approach of the present paper, exhibit some features that seem to be novel. Firstly, the result is really two-dimensional and does not require any radial symmetry assumption. Precisely, because here is considered a really two-dimensional case, standard methods based on ordinary differential equations just do not work here. For special (quadratic) cost functions however, the existence of equilibria is, at least formally, linked to some fully nonlinear partial differential equations of Monge-Ampère type. Secondly, the cost functions are quite general: they do not need to be smooth, to depend only on distance, or to have particular symmetry or convexity/concavity properties . Finally, using the optimal transportation framework also allows to deduce commuting plans in an easy and general way from the wages and revenues patterns. The main result and constructions remain valid if one imposes such zoning restrictions, and the proofs are actually simpler. In fact, zoning restrictions are one of the mathematical artefacts that is used to construct approximate equilibria. It should then be clear to the reader from the proof that Theorem 6.2 still holds in models with additional zoning restrictions.