In the decades surrounding the dawn of the 20th century, the dominant rail line in the northwest was James J. Hill’s Great Northern Railway (GN), which linked St. Paul, MN, with Washington and Oregon, spanning hundreds of miles of rich agricultural, mineral, and timber land. Included in this span was a lucrative market for grain shipments running from the Red River Valley across the northern part of North Dakota, where the GN held a virtual monopoly that included a roughly 60-mile band of feeder lines running north-south along its main east-west line (which connected Grand Forks, Devils Lake, Minot, and Williston, and many points in between).
There were two small feeder lines that did not threaten GN’s market and were therefore allowed to operate without challenge, and for years Hill was allowed to rake in the grain and the cash from the many elevators and towns that developed along his lines. At least until 1904, when the GN received intelligence that a “friendly” Twin Cities competitor—the Minneapolis, St. Paul, and Sault Ste. Marie Railroad, colloquially known as the Soo Line—was sending surveyors to northern North Dakota to scope new lines that would tap into the GN’s bounty.
The resultant competition between the GN and the Soo Line became known as the North Dakota Railroad War of 1905, which resulted in nearly 500 miles of new track and the development of more than 50 new towns by the end of the year. This extraordindary battle for market share between a near-monopolist and an upstart entrant serves as a unique historical laboratory for an analysis of spatial strategic competition: There are many forms of potential strategic behavior among incumbent firms and potential entrants into a market. Depending on conditions, incumbents may have an incentive to respond pre-emptively to threatened entry or in a particular manner once entry has occurred. In such cases, the incumbent’s intent is to favorably shape the structure of market competition. The (potential) entrant, in turn, faces a decision as to how to best respond to the incumbent’s behavior, and indeed the incumbent’s action likely builds expectations of this reaction into their initial strategy. Spatial models can provide insights into resource allocation and price determination. , offering empirical insights into an enduring question within Industrial Organization research.
The Question
Researchers have long investigated the strategic behavior of incumbent firms: a business already established in a market or industry, especially one with a significant market share that typically brings several advantages over competitors. when threatened by potential entrants: a business that attempts to sell a particular product or service in a market dominated by an incumbent firm. . Incumbents, for example, may be induced to respond pre-emptively through expansion that discourages newcomers, while entrants, on the other hand, need to consider those actions when planning their strategy. Will an incumbent’s actions deter an entrant or merely reshape an incumbent’s strategy, resulting in a series of chess moves that ultimately reshape the market? And how do incumbents and entrants make such moves across space, weighing not only consumer preferences but the costs and benefits of protecting or challenging established market share?
One theoretical framework for analyzing this issue involves product proliferation: a strategy used by companies to increase the number of products they offer in a market, often to ward off entrants. This can involve introducing new products or making minor modifications to existing products. as an incumbent’s response, meaning that a competitive threat induces the incumbent to absorb the “product space” by introducing new products and thereby reducing the entrant’s expected profitability. If successful, such a strategy can preserve market share and pricing power for incumbents. A textbook example is when incumbent cereal manufacturers release differentiated products to make inroads into new markets and, thus, ward off potential entrants.
Research into this market dynamic has been limited by opportunities to observe actual behavior by incumbents and entrants. This paper addresses that gap by studying the case of the North Dakota Railroad War of 1905, which offers a clear example of spatial strategic competition. This paper is likely the first to offer a detailed economic examination of this episode; in doing so, Chicago Booth’s Chad Syverson not only offers lessons into the behavior of firms, but also into broader social welfare effects.
The Market
As described above, the setting for this analysis is the railroad market across northern North Dakota surrounding the turn of the 20th century. Figure 1 (1b from the paper) shows GN lines in black and Soo lines in red, with the dashed red line representing the Soo’s intended invasion into GN territory. This new line, known as the Wheat Line, would connect the Soo’s existing lines extending from Minneapolis and into Canada, roughly 25 miles north of the GN’s east/west line and tapping into the gaps between the GN’s branch lines.
Based in Minneapolis, the Soo Line was not bashful about infringing on GN’s established territory. This was true even though leaders of the Soo Line were long-time friends of James Hill and his family. This friendship, though, did not materialize into any known forms of collusion; the companies remained dogged competitors. In many cases, the Soo Line was the only competitor with the GN, meaning that when the Soo Line made its forays into GN’s markets, it was entering into a mostly monopolist market. This is key to the Syverson’s analysis: If we assume that GN’s route network was profit-maximizing for a monopolist, that means that potential additional lines for the GN were necessarily unprofitable, meaning that the resulting cannibalization of its own traffic would outweigh, net of costs, any profits from new traffic. However, under the right circumstances, an additional line could be profitable for an entrant. Was that the case for the Soo? Did the GN’s monopolistic business strategies leave it open to competition? To address these questions, the Syverson assumes that the GN and Soo had similar demand and cost structures, which allows him to estimate the Soo’s expected profitability from the Wheat Line.
Before describing the actions taken by the GN and the Soo, a brief word about catchment areas. Product differentiation in these rail markets was spatial—hauling grain was costly for farmers, so proximity to an elevator/town was important. For rail companies, efficient placement of towns meant maximizing the amount of grain received from farmers without overlapping on another town’s market space, or catchment area. Rail companies used something less than 9 miles as a practical limit for catchment areas; elevators/towns were built at 6- to 7-mile intervals on main lines and 7- to 8-mile intervals on branches. As we will see, these infrastructure considerations meant that building new lines had ramifications beyond a rail company’s profit and loss statement, to also include the effects of town-building on the fortunes of business owners, workers, and speculators. For those interested in a more detailed description of catchment areas, including a series of maps, please see the working paper.
The Choice
To begin, it is important to note that prior to 1905, assuming that it was an effective monopolist, the GN believed it had built the optimal number and length of rail lines in North Dakota; that is, it neither overbuilt (it had not abandoned any lines) nor underbuilt (it had no plans to expand). Then along came the Soo with plans to build 306 miles of new track, perhaps invoking a present-day maxim associated with Amazon founder Jeff Bezos: “Your margin is my opportunity,”1 and the Soo was suddenly poised to add lines in direct competition with the GN.
But was the GN’s margin worthy of competition? Had GN truly left that much business on the table? According to Syverson’s careful analysis, the answer is likely “no.” He finds that under the most optimistic scenarios, the Soo line would be only marginally profitable. Indeed, product proliferation by the GN through the addition of just one branch (at a relatively low cost) would plunge the Soo’s expected profitability into negative territory and preserve about 1,000 square miles of monopoly profit for the GN.
Given these estimates, one would assume that the GN did, in fact, begin laying track in advance of the Soo to stifle any possible competition. But this was not the case. Instead, despite the evidence that the Soo was planning to build, the GN chose to wait until the Soo had committed to laying track. What explains this curious choice not to engage in what would later be termed as spatial strategic competition? Why not stop the Soo threat before it even starts? Syverson offers three possible answers, with roughly equal weight attributed to each:
- It is possible that the GN may have overestimated the cost of deterrence by overestimating the Wheat Line’s profitability (and here Syverson concedes that he may similarly be in error). On the other hand, such estimations are possibly too low. One thing is certain: These estimates were likely crucial to informing the GN’s actions.
- The GN may have considered itself a player in a “repeated game,” whereby the Soo might lay track at various locations along the GN’s network, keeping the GN on the defensive. In doing so, the Soo would develop a “disruptor” reputation that could work to its advantage when confronting an incumbent firm. But this raises the question as to why the GN would not counter this game by developing its own reputation as an aggressive player against competitive entry. The best way to deter entry, in other words, is to squash it before it begins.
- A third possibility is that the GN did not consider or understand the deterrence scenario; rather, it waited for the Soo to build because it saw this as the first best option. Why waste precious resources guessing where the Soo might build; rather, it is best to “wait and see” before responding.
The Outcomes
New towns, similar struggles
Regardless of motivation, the GN ultimately did build aggressively in response to the Soo’s encroachment, and the resultant competition led to the founding over 50 towns in one year on a fraction of an otherwise sparsely populated state. Ads for the Soo Line promised “A Chance for Every Man to Carve Out a Fortune.”2 The operative word there is “chance,” and many were willing to play those odds, hoping that their grain elevator, their dry goods store, their hotel, their plot of land, would bring wealth for years to come. Few achieved that dream, and many towns passed away almost as quickly as they were built. (See Sidebar: “First in line, but still too late.”)
First in line, but still too late
How intense was the competition to quickly build new towns, establish businesses, and thereby preserve market share? Examples abound, but the battle between McCumber (a GN town) and Rolette (Soo), provides a useful case study. One factor determining a town’s success was the public sympathy accorded the Soo Line. Locals seemingly resented that the GN only built new lines when forced by the Soo; if the GN truly cared about local farmers and businesses, it would have built new lines and towns on its own.3
The element of public opinion weighed profoundly on the outcome of the McCumber vs. Rolette saga. Heavily promoted, the tracks to McCumber were two months ahead of the Soo’s tracks into nearby Rolette (the towns were so close they were in view of each other). McCumber had four elevators in place, a promised flour mill, and a bank and other businesses ready to go. Even so, local farmers and businesses let the GN know that this was too little too late, and they informed GN officials of their support for Rolette as the county seat at a forthcoming election. This was before a single property was platted in Rolette.4
The townsite sale for Rolette in August 1905 yielded more than 30 proposed businesses, and the effect on McCumber was immediate, as businesses quickly moved to Rolette. In early 1906, a local election placed the school in Rolette, and within a year just seven businesses remained in McCumber. The GN town would eventually become a ghost town while Rolette thrived, gaining status as the largest of the 1905 War Towns.5
Syverson conducts an accounting of 51 towns built in 1905 (25 by the Soo and 26 by the GN), residing in “War Counties” (those that the Wheat Line crosses that do not contain the GN main line), and “GN Main Line Counties” to assess their longevity and to measure their social welfare. He finds the following:
- Railroad War Towns grew more quickly, increasing from a combined population of 6,200 in 1910 to 7,900 in 1930, about a 25 percent growth rate. This compares to 8 percent in GN Main Line Counties.
- However, this growth trend did not prevail. Population in War Towns declined in every decade post-1930 to 2020, with their combined population falling 40 percent. Many shrank, some died.
- These declines were not peculiar to War Towns but followed a general rural trend over that same period. County populations in 2020 were half those of 1930.
- This decline was not experienced homogeneously: Compared to towns that existed before the Rail War, the War Towns underperformed.
- New Soo towns outperformed new GN towns, with their aggregate population over time always higher than GN towns. Further, 84 percent of Soo towns were incorporated, compared to just 42 percent for GN towns.
- However, no towns escaped the challenges facing rural towns throughout the 20th century and beyond. Ranging in size from a few hundred to a few thousand, none of the War Towns emerged as regional or local service/shopping centers. Economists use the term hysteresis to describe phenomena that exist (in this case towns) beyond their founding factors, and a likely reason that War Towns persist is the lower cost of housing they provide.
As with War Towns, so with War Lines
Syverson’s analysis of the War Railroad Lines reveals the following:
- The main rail lines of the GN and Soo exist today, with the GN’s branches now owned by the Burlington Northern Santa Fe, and the Soo by the Canadian Pacific, its former partner. Business along these lines remains robust enough to support continued operation by the largest category of rail carriers.
- However, most of the branch lines within the “war zone” are abandoned. Even the Wheat Line has suffered: a 65-mile swath in the middle of the line is deserted.
- In the end: the competition between the GN and the Soo produced more rail lines and towns than the market could bear.
To conclude, this work offers a unique economic exploration into the North Dakota Railroad War of 1905. In doing so, it addresses an empirical gap in the research surrounding spatial strategic competition, bringing insights not only to the companies involved, but also to the social welfare effects of the competition.