Mathematical models of urban spatial structure in cities with decentralized employment involve the interaction between a minimum of three markets: land, labor and capital. I study the interaction between the land and the labor markets carefully. Standard theory assumes that when firms decentralize, firms located away from the Central Business District (CBD) offer lower wages to their workers. The continuous downward adjustment in wages as firms move further from the CBD is characterized as the wage gradient. By this approach, following firm decentralization, the adjustment in wages is sufficient to leave everyone indifferent between the monocentric city and the latter case. Consequently, no one relocates and the city is left essentially unchanged.I refute this argument on the grounds that it does not hold in a model with a full labor market. First, I consider the wage gradient approach, where labor demand is infinitely elastic. Next, I look at the case where labor demand becomes vertical. I consider several special cases, and find that when proportionally more employment moves further out, everyone in the city is better off. Finally, I include a labor market where firm movements take place due to movement of capital, a complimentary factor of production. I show that the wage gradient posited by standard urban theory breaks down. Instead, as firms decentralize in a city with fixed population, an upward sloping utility gradient appears as firms move further out.Finally, I consider a case where there are a fixed number of high skilled and low skilled workers in the city with a full labor market. I study where they live, given the proximity and strength of the employment subcenter. When the CBD and the subcenter locate close to each other, the two groups stay segregated, all low skilled workers living close to the center and all high skilled workers occupying a low density suburban area. As more jobs move further out, this pattern changes, and low skilled residential areas and high skilled living areas intermingle. In all, I find five possible patterns of residential location.
