The Economic Logic of Company Towns

George Pullman built his city from scratch in 1880, twelve miles south of Chicago on the shore of Lake Calumet. The Pullman Palace Car Company needed labor; the labor market required housing workers could reach on foot; and Pullman, being a man of pronounced aesthetic opinions and unusual organizational confidence, decided he would build the housing himself rather than leave it to the market’s tendency toward squalor. The resulting town had paved streets, gas lighting, indoor plumbing, a hotel, a library, and a theater — amenities that were genuinely superior to what most American industrial workers had access to anywhere in the country in 1880. Pullman charged rent, kept it slightly above market rate, and retained ownership of everything. The arrangement worked, commercially and socially, for about fourteen years. Then came 1894.

The Pullman Strike of that year is usually taught as a labor history story, and it is that, but it’s also a lesson in the economics of a specific kind of institution. When the Panic of 1893 hit and the Pullman company cut wages by 25 percent, it did not cut rents. This was defensible by one accounting — rents were a separate business — and indefensible by every other. Workers who couldn’t pay rent were evicted from company homes. Workers who complained to supervisors about the wage cut were dismissed by supervisors who were also their landlords’ agents. The conflict of interest that had always been latent in the arrangement became suddenly, violently explicit. The strike spread to rail workers across the country, paralyzed the national freight network, required federal troops to break, and ended with Pullman’s town model discredited in the eyes of American labor and American law for a generation.

What is remarkable is how little this history slowed the recurring invention of company towns in new forms. The logic that drove Pullman keeps reappearing because it is, at the start, genuinely compelling. Attracting workers to remote locations requires offering them something beyond wages. Mining towns in Appalachia and the Colorado coalfields operated on the Pullman model through the early twentieth century. Company stores, company housing, company scrip in lieu of currency — the complete package. The rationale was always the same: the company was providing infrastructure that the market couldn’t or wouldn’t provide in that location, and it was recapturing the cost through rent and retail margins.

The same economics, stripped of its most coercive elements, built the postwar American suburb. Levittown, constructed by William Levitt beginning in 1947 on potato farms in Nassau County, New York, was not a company town in the Pullman sense because Levitt sold the houses rather than renting them. But the logic of vertically integrated residential development — developer buys land, installs infrastructure, builds identical units at production-line efficiency, captures value through sales price — was directly descended from the company town model. Levitt’s innovation was to exit before the conflict of interest emerged. Once the houses were sold, he was no longer landlord, employer, and grocer simultaneously. The residents could complain about the roof without worrying that the person hearing the complaint was also responsible for their next paycheck.

That exit mechanism is the crucial variable, and it’s one that the history of company towns keeps circling. The company town model generates value in its early phase because the integrator can solve coordination problems that markets handle poorly: who builds the roads, who runs the water system, who ensures that workers are close enough to the plant to show up reliably. But the model stores up a structural problem, which is that the residents of the town have no way to exit their dependence on the company without also losing their housing, their access to credit (if company scrip is involved), their children’s schooling, and sometimes their access to the only store within miles. The company’s interests and the residents’ interests diverge eventually, sometimes slowly and sometimes suddenly, and when they do, the residents have no leverage.

The coal mining companies of the early twentieth century discovered this the hard way through a different mechanism than Pullman did. In eastern Kentucky and southwestern Virginia, mining companies controlled so much of the local political economy that they effectively controlled the government — the judges, the sheriffs, the county commissioners. This created short-term security for the companies and long-term brittleness. When the union organizing drives of the 1930s arrived, backed by the federal government’s newly favorable posture toward labor under Roosevelt’s National Labor Relations Act, the mining companies’ political investments proved worthless. The sheriffs who had been breaking strikes in 1928 were not going to do so in 1936. The local political capture had been real, but it depended on federal indifference, and that had changed.

The postwar American version of the company town took a different form: the planned industrial community financed by federal government programs. Oak Ridge, Tennessee, built from scratch in 1942 as part of the Manhattan Project, housed 75,000 people at its peak on land that the federal government owned entirely. It was a genuine company town — residents needed security clearances, and the Atomic Energy Commission controlled housing assignments — but because it served national security rather than profit, the internal contradictions developed differently. The town was incorporated as an independent municipality in 1959, shedding the most objectionable features of managed residence.

Corporate campuses of the 2010s tried a version of this logic that avoided the housing element while retaining almost everything else. Google’s Mountain View campus, Apple’s ring building, Amazon’s spheres in Seattle: the idea was to make the workplace so comprehensively comfortable that workers wouldn’t need to leave for much of what daily life requires. Cafeterias with high-quality food, on-site gyms, dry cleaning, healthcare, haircuts. The employee need never engage with the grim ordinary world. This generated the Pullman-style critique immediately — that the goal was to capture worker time and attention, that the benefits were fundamentally paternalistic, that the company was building dependence. The critique was accurate, and the companies implementing it were aware of it, and they mostly didn’t care because the workers at those companies were choosing to be there and could leave. The conflict of interest that destroyed Pullman Town was attenuated by the fact that losing your job at Google didn’t mean losing your house at the same time.

There is a serious proposal, made periodically by economists and urban developers and occasionally by technology industry figures, to build entirely new cities governed by private entities with contractual rather than democratic authority structures. The academic literature on “charter cities” traces a lineage from 2009, when economist Paul Romer proposed them as a development policy tool. Honduras actually established “special development regions” along these lines in 2012, though the constitutional court subsequently invalidated the enabling legislation. The appeal is the same appeal that drove Pullman: integrated ownership solves coordination problems that fragmented political authority handles poorly.

The failure mode is also the same. Any system in which a single entity controls residence, employment, commerce, and legal dispute resolution has created the conditions under which the interests of that entity and the interests of its residents will diverge, and in which the residents will have no recourse when that divergence becomes acute. This is not a contingent feature of bad management; it is structural. The reason liberal democratic states developed the institutions they did — separation of powers, independent courts, protections for property that can’t be voided by executive decision, freedom of movement — is precisely because concentrated authority over multiple dimensions of life produces systematic exploitation of the people subject to it. Every generation of company town builders rediscovers this the expensive way.

What the company town experience actually reveals is something about the conditions under which markets work and don’t work. Markets require that participants have genuine alternatives. When a miner in Harlan County, Kentucky in 1930 had a choice between the company store and nothing, the company store was not competing in any meaningful sense. The price it charged was not a market price. The housing it provided was not competing with other housing. Everything about the company town model in its classic extractive form was predicated on eliminating the alternatives that make market transactions voluntary rather than coerced.

The minimum condition for a company town to avoid the Pullman failure mode is free exit: residents must be able to leave without losing everything simultaneously. This condition is harder to satisfy than it sounds. If the company town is built in a remote location — which is often why it exists — physical exit requires money and transportation that impoverished workers may not have. If the company provides credit at the company store, exit requires paying debts that may exceed wages. If the housing is rented rather than owned, exit means giving up the physical shelter over one’s family’s heads. The architecture of the classic company town was designed, whether consciously or by institutional drift, to make these exit costs prohibitive.

The pattern that repeats across every generation of company town is this: integrated vertical control generates genuine benefits in the early phase, when the coordination problems are real and the interests are roughly aligned. Then the interests diverge. Then the residents discover they lack the tools to protect themselves. Then there is either a crisis, producing reform or destruction, or there is an exit mechanism that allows the developer to shed the ongoing liability while the residents form a real community. The ones that end in crisis tend to teach nothing, because the companies involved don’t survive in recognizable form. The ones that exit cleanly tend to produce the most interesting cities, because they begin with infrastructure advantages that purely organic development lacks.

Levittown is now a dense suburban municipality with all the ordinary apparatus of local government. Oak Ridge has a functioning city council and an independent school system. The coal towns of eastern Kentucky have largely depopulated, the mining economics having failed long before the institutional problems were resolved. Pullman is now a neighborhood of Chicago, absorbed by annexation in 1889, the company having lost its independent status before the strike that became its epitaph. The built environment outlasts the institution. The streets and buildings and the basic logic of the settlement pattern persist after the company that built them is gone.

That persistence is what makes the company town a durable subject for analysis. The question is never really whether integrated private development of residential communities is economically efficient in the short term. It often is. The question is who captures the value as it accumulates over time, and what happens to the people who live there when the interests of those two parties diverge. American history answered that question repeatedly in the same way. The current generation of corporate city builders in various parts of the developing world is likely to discover the answer again, because the structure of the problem hasn’t changed and neither has the structure of the solution that’s actually stable.

The stable solution is municipal government: the transfer of control over the common infrastructure and legal environment from a single private entity to a body that represents the residents’ interests, with formal mechanisms for replacing leadership and contesting decisions. This is less efficient than centralized private control in the narrow sense that decisions take longer and require consensus. It is more stable in the larger sense that the residents cannot be simultaneously deprived of their housing, their employment, and their legal recourse in a single decision by a single company. That redundancy is not waste. It is the entire point.