O'Sullivan (2009).

             relates the utility of a worker with the total workforce within a city,.

             shown in Figure 1.

             From this example it can be shown that cities may be too large, but.

             not too small. The utility curve reaches its maximum with 2 million workers .

             in a city so a region with 6 million workers will maximize utility with 3 cities,.

             where utility has adjusted until workers are indifferent between the two cities.

             If cities are too small, so for example if there are 6 small cities each.

             with 1m workers, there is an unstable equilibrium because the utility curve.

             is positively sloped at this point. If a worker moves from one small city.

             to another they create a utility gap as the population in one has decreased.

             - and hence utility - while the other increased. This gap encourages even.

             more movement, and as self-reinforcing effects generate extreme outcomes,.

             the extreme outcome is that everyone will move from one city to another,.

             making the city 'disappear'. However when you have two large cities, with.

             3m workers each, when a worker moves and a utility gap is created, utility.

             is actually higher in the smaller city (perhaps due to congestion and.

             overcrowding in the now larger city) making migration self-correcting.

             rather than self-reinforcing. Migrants will then either move back to the.

             smaller city, or existing inhabitants of the larger city will move to the.

             smaller one until a stable equilibrium of 3 million workers.

             The hukou system's restrictions will limit worker's ability to.

             migrate between cities and it is almost certain that an optimum equilibrium.

             state will not being reached. Workers will be contained within the region.

             they currently reside and while it may be possible for equilibrium to be.

             reached within the region, it limits the probability of it being optimal.

             dramatically. This reflects Andes'(1995) view that political forces, more.

             so than economic ones, drive urban centralization, hence cities are induced.