The economists’ whole city

Edward Glaeser, Giacomo Ponzetto, and Yimei Zou have a fascinating new working paper in which they attempt to determine whether it is preferable, from the point of view of economic development, to follow a course of urban development which prioritizes the massification of ‘mega-cities’ or one which prioritizes the linking-together of smaller cities in ‘urban networks.’ And this, of course, is a question predicated on some working definition of what constitutes a ‘single’ city—the same question which I am trying to answer from the perspective of historical geography. What is the difference between several areas which are geographically distinct yet tightly interlinked by social and economic ties (the ‘urban network’), and an area whose social and economic ties are so substantively complete that its territory forms a single entity (the ‘mega-city’)?

Leave aside, for a moment, the several objectionable generalizations which these economists make in their search for a model that can be resolved into mathematical terms. (The mystification of idea-transfer mechanisms, for example, betrays a fealty towards the heroic myth of the entrepreneur-capitalist, and the breezy sketch history of urban development from Mesopotamia to modern China over-emphasizes the role of marketplaces and trade entrepôts in dictating urban development.) What is left is an interesting abstraction of two models of urban intensification, one monocentric and the other polycentric. If we follow the authors in assuming that there are compounding returns to densification in terms of entrepreneurial activity, firm creation, and job opportunities (this is one of Glaeser’s core assertions), how best can we nurture these effects while recognizing the countervailing problems—like congestion, housing scarcity, and inequality—that also come with increased density?

The authors conclude that context matters. Where it is easy to add new housing quickly and cheaply, as in the American Sunbelt, the monocentric model is superior, while in regions like Europe and the American Northeast, where an earlier pattern of urbanization has led to tighter restraints on new construction, the polycentric model is preferable—as it is, too, in places like China where super-cities like Beijing and Shanghai have grown so large that additional core density is impractical. The authors also recognize that the economic advantages of densification in mega-cities are more likely to accrue to skilled workers than unskilled workers, since the former earn a higher premium from both their labor mobility and their access to entrepreneurial networks.

But wait: it is worth pausing for a moment to stop and wonder what these idealized models (of mega-cities and urban networks) actually look like in real life. It is easy to imagine them diagrammatically: the mega-city is a very large dot getting much bigger, while the urban network is a constellation of medium-sized dots connected together by thick lines getting thicker. It is also easy to imagine them naïvely, in terms of places that we already know. The Dutch Randstad, for example, which the authors repeatedly invoke as a canonical urban network, is not (in our naïve definition) one place by itself, but rather several distinct places closely connected together: Amsterdam, Rotterdam, Utrecht, and the Hague. By contrast, Beijing is a canonical mega-city, massively and symmetrically arranged around a single focal point.

Yet it is often not so easy to characterize places according to these two diagrammatically perfect schemes. Referencing Jean Gottman, the authors note that “it is debatable whether the U.S. Northeast corridor is an urban network or the BosWash megalopolis” (p. 11). In other words—is the northeastern Megalopolis many interconnected place or is it just one place? And it is not just at massive scales like Megalopolis that we find a difficulty distinguishing between singleness and multiplicity. New York City seems at first to be one of the United States’ natural mega-cities. But less than 150 years ago, what we now call New York City was an urban network, made up of half a dozen closely interrelated communities centered on Manhattan, Brooklyn, and Newark. Even today, London, another supposed mega-cities, is a warren of independent jurisdictions, some, like the City of London, that retain an independence founded on an idiosyncratic medieval legacy.

So on what logic do we call the Dutch Randstad an urban network while calling London a single mega-city? Simply because there are strips of “open” (i.e., morphologically “non-urban”) land separating the built-up areas of the Randstad cities? Simply because Amsterdam and Utrecht have political sovereignties which Westminster and Hackney do not? What evidence exists to show that Rotterdam and The Hague are not more economically and socially fused-together than Islington and Croydon? Here again, the authors admit: mega-cities “are made up of sub-city neighborhoods and can resemble urban networks” (p. 2).

The authors make three assumptions which they use to distinguish, for their purposes, a unitary versus a networked geography. First, low-skilled workers are assumed to move frictionlessly within a mega-city, but not amongst the several nodes of an urban network. Second, “network amenities” are assumed to be homogenous within a mega-city, but not amongst the nodes of an urban network. Finally, they assume “idea exchange” to be more homogenous within the mega-city than amongst the nodes (pp. 11–12). Still, the authors confess that all three of these definitional assumptions “can be debated in either direction” (p. 12).

Because of the debatable character of all three of these definitions, there is something of an ontological sleight-of-hand going on here, one which I have attempted to highlight by emphasizing the words within and amongst in the paraphrases above. The character of space in a mega-city is of a within character: when one has moved from one location to another inside of a mega-city, it turns out one has not really moved at all! In the ontologically single mega-city, space is flat and homogenous—it is all the same place. By contrast, the character of space in an urban network is of an amongst character: when one has moved from one node to another, one really has moved. In the ontologically plural urban network, space is differential amongst the nodes—it is a collection of different places. But it is important to note: each of the nodes of the urban network is itself unitary, and it is only by leaving one node and arriving at another that we accomplish a heterogenous spatial transition.

So: if we leave the Bronx and turn up in Staten Island, have we gone anywhere? No: we are still in the same place, the same homogenous mega-city. If we leave Boston and turn up in Lowell, have we gone anywhere? Yes: we are in a new node of this urban network. What about if we leave Dorchester and turn up in Charlestown? No, now we have not gone anywhere: we are still in the same node of the urban network.

This, of course, is somewhat absurd. It is absurd because these definitions of geographic homogeneity and difference are under siege by two truths. First, every place is different from every other one. This holds true at even the tiniest scale. The other side of the library in which I am now writing is not perfectly flat and homogenous to this side. Second, every place is connected to every other one. This does not mean all places are connected equally strongly, but it does mean that there are no perfect discontinuous borders in nature itself. (The authors of the paper recognize this point, assuming “that networks have clear borders solely for expositional ease” [p. 11].) In the recognition of these two truths, the elegant diagrammatic perfections of the mega-city and the urban network can never be achieved in actuality.

Now, I do not mean, in the face of this challenge, to say that it is perfectly meaningless to speak of any place as a “single” entity. Quite to the contrary: I think it is even more interesting to discover what mechanisms are in effect when we speak of somewhere as having a unitary geography. What shifts is that we must cease to assume that geographic units merely exist, and instead confront the fact that they are made to exist. This allows us to leave behind a deceptively empirical geography and draw into our analysis a consideration of social relations, ideologies, and political beliefs. Because they are assertive and not coolly descriptive, these constructions of units bear the marks of how we think about the world.

In Glaeser, Ponzetto, and Zou’s case, unitariness is defined by roughly homogenous capital and labor-market conditions. For instance, their first mark of distinction between mega-cities and urban networks is that people can move frictionlessly and homogeneously within a mega-city. That’s a claim which would justifiably strike any critical urban geographer as preposterous. You need only look at the U. S. racial dot map in order to see just how difficult it is for certain groups to relocate within the boundaries of a single city. But for these economists, what is important is not residential micro-geography but a single labor market. “Though they may live in neighborhoods that are largely segregated by income,” the authors admit, workers “can work in businesses throughout the mega-city” (p. 2). That genericization of workers into a single, homogenous labor market thus becomes the way of defining geographic unitariness: in their history of New York, for example, it is the construction of the Brooklyn Bridge which “turned Manhattan and Brooklyn from distinct cities, with clearly separate central business districts, into a single labor market” (p. 7).

The working definition of geographic unitariness which enables the operative distinction between mega-cities and urban networks is therefore, for these authors, the territory over which capital relations are roughly homogenous. Though I can think of many objections to that definition, it is not so different from, say, an environmental historian’s definition of the city which is based on its metabolic inflows of water and food. Making geographic wholes involves muting differences which we do not care about, and amplifying the discontinuities which we do care about.