Foundations of Urban Spatial Economics: Theoretical Limits of Segregated District Models

The coexistence of Manhattan’s traditional financial core and its adjacent creative communities represents a persistent empirical puzzle for canonical urban spatial economic theory. In 2022, for instance, JPMorgan Chase signed a 1.1 million square foot lease for office space in Brooklyn’s DUMBO neighborhood, a district historically defined by its concentration of art galleries, design studios, and tech startups; the firm’s average bid of $85 per square foot exceeded the average bid for adjacent creative office space by $7 per square foot, yet the two user groups occupied overlapping parcels with no clear spatial hierarchy. Similarly, Goldman Sachs’ 2021 lease of 250,000 square feet in Chelsea placed its operations adjacent to independent record labels, boutique architecture firms, and contemporary art spaces, with the financial firm’s per-square-foot bid of $78 overlapping with the $72 average bid from creative users in the same submarket. These patterns stand in direct contradiction to the core prediction of nearly a century of urban spatial economic theory: that competing land uses will sort themselves into strictly segregated districts according to a hierarchical bid-rent gradient, with the highest bid-rusers—conventionally defined as financial and corporate office users—exclusively occupying central, high-accessibility locations, while lower-bid users such as creative and cultural producers are confined to low-rent peripheral areas. This chapter reviews the canonical frameworks that generate this segregation prediction, identifies their core explanatory gaps when applied to New York City’s (NYC) observed spatial structure, formalizes the volume’s core research questions, and outlines the mixed-methods research design used to address those questions across subsequent chapters.

1.1 The Canonical Monocentric City Model and Its Core Assumptions

The foundational framework for modern urban spatial economics is the Alonso-Muth-Mills (AMM) monocentric city model, first formalized by William Alonso in 1964 and later extended by Richard Muth and Edwin Mills. The model’s core premise is that urban land use is organized around a single central business district (CBD) where all employment is concentrated, with residential and non-residential land uses sorting themselves by distance from the CBD according to a tradeoff between commuting costs and land rents. The model’s derivation begins with a set of restrictive but analytically tractable assumptions: (1) the city is located on a flat, homogeneous plain with no geographic barriers to development; (2) all employment is concentrated in a single CBD with no alternative sub-centers; (3) commuting costs are linear and proportional to distance from the CBD, with no variation in transportation infrastructure or mode choice across space; (4) land is perfectly divisible and there are no institutional constraints on development (e.g., zoning, historic preservation, or land use regulations); (5) all land users are price-taking actors with no market power to influence rents; and (6) there are no externalities associated with the geographic concentration of specific land uses.

Under these assumptions, the model derives a bid-rent function for any land user that declines exponentially with distance from the CBD. For a residential household, for example, the bid-rent $b(r)$ at distance $r$ from the CBD is given by the equation: $$b(r) = b_0 e^{-tr}$$ where $b_0$ is the maximum bid-rent at the CBD and $t$ is the per-unit commuting cost. The model predicts that land uses will occupy concentric rings around the CBD, with the highest bid-rent users (traditionally financial and corporate office users, with the highest willingness to pay to minimize commuting costs for high-wage employees) occupying the core, followed by high-income residential uses, then low-income residential uses, and finally industrial and low-value uses at the urban periphery. For non-residential uses, the model predicts that users with the highest value of time (and thus the highest willingness to pay to minimize commuting costs) will outbid all other users for central locations, leading to strict spatial segregation of competing land uses.

For much of the 20th century, the AMM model provided a strong empirical fit for NYC’s spatial structure. Between 1920 and 1970, Manhattan’s Financial District was the unambiguous core of the city’s economy, with financial and corporate office users occupying nearly 80% of the district’s commercial floor area, while creative and cultural producers were confined to low-rent peripheral areas: artists occupied converted industrial lofts in SoHo and the West Village, with average rents of $2 per square foot in 1970, compared to $12 per square foot for office space in the Financial District. During this period, the model’s prediction of strict spatial segregation held, as creative users had no ability to compete with financial users for central locations, and institutional constraints (e.g., zoning that prohibited residential use in most of the Financial District) reinforced the segregation.

1.2 Limitations of the Bid-Rent Framework for Explaining Overlapping Land Uses

While the AMM model remains a useful baseline for understanding urban spatial structure, its core assumptions are violated in nearly all contemporary cities, and especially in NYC, leading to systematic failures in predicting observed patterns of land use overlap. The first and most critical limitation is the model’s assumption of a single employment center. By the 1980s, NYC had developed multiple competing sub-centers of employment: the traditional Financial District in Lower Manhattan, the Midtown corporate and financial hub, the DUMBO and Downtown Brooklyn tech and creative hub, and the Williamsburg creative and residential hub. Extensions of the AMM model to multi-center cities (Mills and Hamilton, 1989) predict that each sub-center will have its own bid-rent gradient, with the highest bid-rent users occupying the core of each sub-center, but still fail to predict overlapping bid-rent curves for competing uses within the same sub-center.

The second critical limitation is the model’s assumption that commuting costs are the sole determinant of bid-rents. In reality, land users derive value from proximity to other specific land uses, generating location-specific externalities that are not captured in the model’s commuting cost tradeoff. For financial users, these externalities include information spillovers from proximity to other financial institutions, regulatory agencies, law firms, and financial data providers; for creative users, they include knowledge spillovers from proximity to other creative producers, cultural institutions, and niche consumer amenities. These externalities generate additional willingness to pay for specific locations that is independent of commuting costs, leading to bid-rent curves that do not follow the simple exponential gradient predicted by the AMM model.

Empirical evidence from NYC confirms the importance of these externalities. A 2021 analysis of commercial lease data from the NYC Department of Finance found that financial users in Chelsea paid a 12% premium for space within 500 feet of other financial institutions, while creative users paid a 22% premium for space within 500 feet of other creative institutions (galleries, design studios, etc.). These premiums are not explained by differences in commuting costs, as both user groups in the sample had similar average commuting times to the CBD. Instead, they reflect the value of agglomeration externalities that are omitted from the standard bid-rent framework.

A third limitation of the AMM model is its assumption of homogeneous land, with no variation in development constraints or supply elasticity across space. In NYC, land supply is highly heterogeneous due to zoning regulations, historic preservation restrictions, and variations in infrastructure capacity. For