In the antebellum period, the American national state played a much smaller role in citizens' lives than it does currently. The national state did possess a range of critical powers, including a substantive capacity to tax and raise a military.Footnote ¹ However, even in policy areas where the national state possessed authority, such as interior management or the postal service, its capacity to pursue independently developed polices was often limited.Footnote ² The limited character of national power also extended to seemingly logical uses of federal authority, such as infrastructure activities. National debates over the constitutionality of using federal resources to fund local projects, a lack of congressional mechanisms (such as standing committees) for cooperation on large bills, and southern concerns that a federally built infrastructure was the first step in the national abolition of slavery all stalled progress on a federal internal improvements system.Footnote ³ But this lack of national administrative strength prior to the Civil War should not suggest that the antebellum United States was entirely stateless or without strong governance.

Instead, local political institutions, from state legislatures to city councils, stepped into the gap left by national inactivity to provide and administer public goods.Footnote ⁴ Through federalism, state governments in particular possessed considerable power to raise revenues, police citizens, and engage in ambitious public works projects.Footnote ⁵ As a consequence, antebellum American state building was decentralized, with local actors providing necessary goods and services. In federal systems, this decentralized approach to state building stands as an alternative to relying on the national state to achieve political development. At the same time, however, such a localized approach to governance raises the specter of free-riders and shirkers. In other words, with a reliance on subnational governments and without a central administration, there is a persistent risk that not all state legislatures will engage in state development, undermining decentralized state building.

While state-level action was important in numerous policy areas in the antebellum United States, no localized policy arena was of more critical importance to national capacity building than subnational internal improvement programs. State-sponsored internal improvement programs greatly enhanced travel and shipping across the nineteenth-century United States, which was still very much a frontier nation. Additionally, these early American infrastructure projects produced myriad overflow benefits beyond their basic function of improving citizens' and goods' mobility. The construction of antebellum infrastructure systems increased land values, enhanced agricultural opportunities, developed manufacturing cites, and improved defense.Footnote ⁶ Thus, state governments were deeply interested in road, canal, and rail projects in the antebellum era. Of these early internal improvement projects, state legislatures' involvement in rail promotion was especially vital to national capacity building.Footnote ⁷ Though the railroad's dominance was never guaranteed, railroads ultimately offered new possibilities in transportation with which other transport systems, including canals, could not complete. Specifically, railroads offered the advantage of swift and regularized transportation to the nation. Through doing so, railroads reduced the impact of geography on travel.Footnote ⁸ At their core, railroads were an undeniably modern and industrial mode of transportation. The impact of railroads was felt in numerous ways, as railroading produced larger corporations, a more homogenous nation, and an increasingly public life where Americans traveled with and interacted more.

often with strangers than ever before.Footnote ⁹ While many railroads were built by private capitalists, political institutions also exerted a considerable influence over rail planning.Footnote ¹⁰ In particular, the lack of immediate federal action on railroads allowed state governments a great deal of latitude in how they approached rail promotion, planning, and regulation within their own borders.

Realizing the need for subnational state action to promote local rail development, many state legislatures stepped forward to stimulate rail construction within their own borders. Some states were especially aggressive in their rail promotion, building railroads directly with state funds, while other states were more measured in their efforts. However, not all states were active rail promoters, raising the critical question of what led to this variation in local rail promotion policies. Local policy development often occurs due to the spreading of new policies from innovator states outward to proximate states. This diffusion of new policies is often attributed to social learning, where state legislatures observe the actions of nearby states and borrow novel policies for implementation within their own borders. However, parallel thinking, where states converge on similar policies due to shared circumstances, also plays a major role in policy diffusion. In parallel thinking cases, diffusion results from similar local circumstances, rather than direct learning occurring over state lines. While social learning placed internal improvements on neighboring states' political agendas during the antebellum era, shared local conditions, rather than active social learning, was most critical in determining whether or not a state legislature enacted its own state rail promotion program.

Significantly, state rail promotion sheds light on the role of local political actors in American capacity building. While the degree of subnational rail promotion varied greatly, state policies were nonetheless an important facet of American rail development, laying the foundation for a national transportation network.Footnote ¹¹ However, what remains unclear is whether enough states engaged in local action on railroads to provide national capacity, particularly national capacity that would rival federal action. In fact, the lack of policy diffusion on rail promotion schemes suggests that local action was not reliable enough to compensate for centralized action in state building. Additionally, beyond offering insight into American political development and local political processes, my analysis here stresses the importance of geography to state building activities. In order to develop the argument that the spread of local rail policies resulted from parallel thinking instead of social learning, I outline the relationship between the state and American federalism, enumerate the advantages of state action on railroads, and then provide an overview of the diffusion literature. Then, I present a theory of policy diffusion via parallel thinking, discuss my methods, and offer the results of my statistical analyses. Finally, I conclude by considering the implications of both political geography and parallel thinking on American political development theories, notably by stressing how the shortcomings of local state building contributed to American political centralization.

THE PROBLEM OF THE STATE AND AMERICAN FEDERALISM

The first step toward investigating how local governments contributed to national state building is to outline what is meant by the term “the state.” It is particularly important to consider the relationship between both the state and federalism, as well as the connection between the state and the built environment. In many ways, the argument that local governments provided services or legal frameworks for national-level state building seems to challenge basic notions of what the state is. The state, as a construct, is generally recognized through autonomy. A governing body becomes a “state” when that agency gains its own interests, beyond those demanded by its constituents. These interests generally originate from a state's competition in a field against other states. For instance, a state can develop a particular interest in public health, far beyond what voters or citizens are demanding. The state's interest in public health does not reflect a maternalistic desire to provide the best possible life for its citizens. Rather, this interest in nutrition and cleanliness originates from the state's need to have healthy laborers and soldiers. This development of discrete interests, as well as a capacity to effect change in society and economy, is the traditional hallmark of the modern state.Footnote ¹²

The United States, particularly in the years from the founding through at least the Civil War, was without many of the features that define the modern state. In the late eighteenth and early nineteenth centuries, the American state lacked the obvious central bureaucracies, large budgets, or sizable personnel that defined its European contemporaries. The early American state is often portrayed as weak, lacking the centralized administration that would allow it to pursue its own interests or significantly redirect social processes in the manner that is required to be defined as a state.Footnote ¹³ It is only with later crises, including industrialization and the Civil War, that the American state—in the traditional European model of the state—begins to appear, complete with larger budgets, more bureaucrats, and agencies to enforce rules.Footnote ¹⁴ In the American context, the study of state development emphasizes the “durable shift in governing authority”.Footnote ¹⁵ Picking up on the notion that studying the state involves studying an autonomous and influential institution, American political development considers how the American state transformed from being underdeveloped to finally becoming a modern state in its own right.

However, as noted above, the claim that the early American state was wholly absent is increasingly coming under critique. More and more, American political development theorists are stressing that the American state has always been present, just perhaps not in ways that are instantly recognizable. First, it is critical to underscore that the Constitution itself creates a strong central state, as indicated by the manner in which the constituent aspects of the Union (the states) are locked out of direct political representation in the national body. Specifically, state governments lack representation in the national government: while state legislatures selected early senators, legislatures nonetheless lacked the power to direct how senators voted or to recall those senators for votes the legislature disliked. From its earliest days, the American central state possessed its own power and legitimacy, wholly distinct from the constituent states.Footnote ¹⁶

Additionally, numerous scholars stress how active the national state was during the early decades of the Republic. Richard John highlights the federal government's vital role in building post roads and delivering mail, a critical function of nation building in the country's first years.Footnote ¹⁷ Echoing this sentiment, Brian Balogh argues the early American state was quite active, as evidenced by the federal government's role in collecting tariffs, conducting land surveys, investing in lighthouses, managing large tracts of public land, and using the court system to build and define economic rules. Often, this early, active national state is overlooked because it was intentionally kept out of sight, so as to not disconcert nineteenth-century Americans who were wary of a strong central state.Footnote ¹⁸ Yet, despite its invisibility, the national state was present and behaving as a state: gently guiding the economy and society, as well as pursuing its own, unique interests. Similarly, the early American national state was also involved in American public health by providing health care to merchant marines.Footnote ¹⁹ Finally, the antebellum national state possessed the traditional state powers of being able to maintain a peacetime army and exert broad taxation.Footnote ²⁰ While this range of powers was not fully deployed in the early days of the Republic, they nonetheless speak to a strong, capable national administration.

Another critique of the “early stateless” thesis of American history highlights federalism and the role of local governments. Federalism and local policies were critical in shaping the young nation's burgeoning economy.Footnote ²¹ While the national state was not obviously proactive during the nineteenth century, the same cannot be said for state and local governments. State governments had a great deal of power, and were regularly involved in public health, internal improvements, and economic regulation.Footnote ²² The American nineteenth-century state was decentralized, with American state power distributed across many actors, including local governments and private agencies.Footnote ²³ In this displaced model of the state, local actors perceive the need for particular policies or state agencies. Then, one by one, the state governments enact the necessary policies. Instead of a centralized body developing its own interests and acting toward fulfilling those ends, the development and pursuit of “state interests” is dispersed. In this federalized state building, local actors resemble ants in a colony: each ant/state works on its own, not fully aware of its participation in the whole, but nonetheless contributing to an overarching function. Furthermore, state capacity, understood as the state's ability to accomplish ends, is still enhanced through these decentralized activities. For instance, in the case of the military, local militias could be called into action and forts could be constructed. At later times, those same troops would be released, the forts pulled down, and military strength reconstituted elsewhere.Footnote ²⁴ As a consequence of federalism, the early American state was not at all absent. To the contrary, the nineteenth-century American state as enshrined in local actors was flexible, quick, and perhaps easily missed, but nonetheless still very much present.Footnote ²⁵

This dispersed American state was especially skilled at the development of infrastructural power.Footnote ²⁶ Infrastructural strength itself refers to the “power to coordinate civil society”.Footnote ²⁷ Infrastructural power allows for the spread of ideas and resources; it empowers coordination. Without a strong system of infrastructure, a state cannot function or govern effectively. Infrastructure covers a range of resources, including communication networks, agricultural production, education, and military organization. Also critical is transportation infrastructure, the material networks along which ideas, people, and goods travel. Without an ability to overcome physical distance, governance is slowed to the point of ineffectiveness. When space is a barrier, laws cannot be enforced, defense is difficult, economic growth slows, and information trickles along too slowly for maximum benefit. Thus, space matters as a medium to connect institutions with one another. However, at the same time, space is also political and an institution itself. Space is permeable, and there are few fixed or natural barriers that cannot be altered or overcome in some manner. Decisions about how to change space, which are fundamentally decisions about which actors to connect and which to separate, are always subject to choice and conflict. Further, much like any institution, the spatial relationships built by infrastructure exert a long-term impact on social relations.Footnote ²⁸ Thus, spatial development is critical to state building, as decisions about infrastructure will determine which areas are defended, economically successful, culturally relevant, and politically influential.

The role of space, and the development of transportation infrastructure, was especially critical in the postrevolutionary days of the United States. In the early nineteenth century, much of the nation was still a frontier, leaving it isolated and difficult to traverse. Travel was often slow and difficult, relying on unpredictable waterways and rutted, poorly maintained roads.Footnote ²⁹ Despite repeated groundswells for national action on turnpikes and canals throughout the late eighteenth and early nineteenth centuries, American transportation infrastructure remained very much a provenience of local governments in the decades before the Civil War. In this period, federal support was limited to post roads, engineering assistance in planning canals and railroads, as well as the clearing of harbors and rivers.Footnote ³⁰ Though important projects, these minimal federal actions were simply insufficient to meet the transportation demands of the growing frontier nation. As a result, the majority of major internal improvement projects, such as the Erie Canal and Ohio's dense network of canals, were undertaken by state governments.Footnote ³¹ Therefore, local governments' role in building and coordinating local internal improvements projects into a national system were especially salient, both for decentralized American state building as well as the rationalization of American space.

Yet, while the presence of local governance in the absence of national action during the nineteenth century is well established, the processes by which state governments became involved in new policy areas and the impact of local policies remain open questions. This decentralized approach to state building is a unique perspective on the growth of state capacity that challenges the notion of a singular, central state that develops and pursues its own discrete interests. Decentralized state building, where local actors each fulfill policy obligations within their own territories, requires many legislatures and bureaucrats to implement similar polices at approximately the same time in order to be effective. If many states are choosing to not participate, or pursuing wildly different agendas through a range of techniques, then the result is a chaotic jumble of policy environments rather than cohesive state building.Footnote ³² It is this danger of poor local choices that produces the threat of “race to the bottom” welfare policies, a danger that leads critics to argue that local control over some policy areas—such as economic redistribution—actually generates disadvantageous outcomes.Footnote ³³ Decentralized state building is a Gestalt process, where many actors must take action without any centralizing body coordinating. The lack of a central coordinator means that states could always opt out of building a piece of infrastructure or supporting a particular regulatory body, leaving a gap in the national patchwork of local services. If many states opt out of a regulation or service, then it is not clear how well local actions actually compensate for national inaction. While local governments possessed the capacity to provide the policies that compensated for a lack of national action, the question remains: Was this approach to state building even feasible, and if so, how did it work?

THE VALUE OF PUBLIC ACTION

Through federalism, there is the possibility that decentralized local actions cohere to build national capacity. However, for that process to function, local policies actually need to offer some benefit. Which raises the question: Did local rail promotion programs in the antebellum United States actually improve local railroads and even offer the possibility of contributing to a greater national whole? Though not all state rail promotion programs were unqualified successes (and some nearly bankrupted their respective state coffers), the overall impact of state action on railroads was positive. Two critical benefits flowed from state action on railroads: 1) rail development within individual states occurred at a considerably faster rate and 2) railroads penetrated isolated, but valuable, sectors of individual states more quickly.

The first, and most anticipated, result of state money being applied to railroads was the more rapid expansion of the indigenous rail network. Maryland's aid to a local rail company in 1828 is a crucial example of state support dramatically improving a local rail system. The Baltimore and Ohio Railroad, chartered by the Maryland state legislature in 1827, was one of the earliest American railroads constructed. Initially, the railroad was supported solely by private capital, along with a grant from the city of Baltimore. However, in 1828, the state of Maryland acquiesced to requests for state aid and provided support by subscribing to $500,000 of the rail company's stock. In doing so, Maryland became one of the first states to directly support local railroad development, behind Pennsylvania's investment in the Main Line.Footnote ³⁴ Eventually, this small contribution would increase and the state would end up becoming one of the company's major stockholders. Less than a decade later, in 1836, the Baltimore and Ohio boasted a complete route to Washington, D.C.; revenues of $260,000 a year; seven locomotives; over a thousand freight cars; and forty-four passenger cars. Freight shipping on the railroad was so successful that the company was actually unable to keep up with demand.Footnote ³⁵ With the benefit of assistance from the state of Maryland, America's first railroad was an overwhelming success.

Further, the positive role Maryland's state legislature served by investing state funds in a local rail company was hardly isolated. In order to illustrate more generally how states' rail promotion policies positively impacted local rail growth, Table 1 summarizes an ordinary least squares (OLS) regression demonstrating the relationship between state rail promotion policies and local railroad vitality. The dependent variable is the number of railroad miles in a state in 1860. The independent variables are 1850 economic, political, and geographic factors, all obtained from historical census data.Footnote ³⁶ The data are lagged a decade to allow for prior conditions to impact future events. The particular relationship of interest is rail support in the state and its impact on miles of track. The value for rail support is ordinal, ranging from one to three. At the minimum, states offered no support for local railroads. At the maximum, states offered direct, robust assistance, such as grants to companies. The median value describes states that provided indirect support, such as utilizing state credit to assist private railroad companies in attaining capital.

Table 1.

Ordinary Least Squares Regression of the Impact of State Rail Policies on Miles of Track Within a State, 1860

*denotes significance at 0.10 level.

** denotes significance at 0.05 level.

*** denotes significance at 0.01 level.

^a The dependent variable is the number of miles of railroad track within a state in 1860. The independent variable of interest is a state's rail promotion program in 1850. The ten-year lag allows for the policy decision in 1850 to exert an impact. The remaining variables are control variables. Sources: Carville et al., Historical U.S. County Boundary Files; Cline, Alabama Railroads; Goodrich, Government Promotion of American Canals and Railroads; Haines, Historical, Demographic, Economic and Social Data; Hull, Shortline Railroads of Arkansas; MacGill, History of Transportation; Million, State Aid to Railways in Missouri; Pisani, “Promotion and Regulation”; Reed, New Orleans and the Railroads; Rubin, “Canal or Railroad?”; Scheiber, “Federalism and the American Economic Order”; Taylor, Transportation Revolution; Usselman, Regulating Railroad Innovation.

The results in Table 1 underscore that states with more active rail promotion programs did possess more miles of track than their neighbors without such programs. Though the relationship between rail promotion and miles of track is not statistically significant, the finding is not inconsequential: the coefficient for rail support is both positive and sizable. In a bivariate model where rail miles were regressed solely against the variable for rail support, the resulting coefficient is in fact both positive as well as significant. Given the small sample size (N = 32) along with the controls for other economic, political, and geographic conditions, the findings from the more complete model in Table 1 suggest that states with active rail support developed considerably more robust transportation networks than states that chose to forgo railroad promotion. Hence, state action on railroads tended to be successful, and accomplished the goal of enhancing the local rail network.

However, beyond just enlarging the size of the state's rail system, legislative aid to rail companies also pushed railroads more quickly into areas they may not have otherwise reached for many years. For instance, state rail support in Georgia allowed an early rail network to build into the more isolated northwest corner of the state.Footnote ³⁷ The Georgia experience was hardly unique. In the 1830s, western states such as Illinois also engaged in rail promotion efforts to compensate for the slow trickle of private companies building local railroads.Footnote ³⁸ Through state promotion, railroads pressed into less-populated areas, where private capital may not have invested for a long period of time. This success of state rail action begs the question of how rail promotion programs in one state influenced contiguous states' legislatures and why all states did not adopt their own rail promotion programs.Footnote ³⁹

THE CAUSES OF LOCAL POLICY DEVELOPMENT

State rail promotion efforts can produce denser rail systems, suggesting that local actors are capable of contributing to national state building in a meaningful manner. However, a state building process would have worked better if many, or all, states adopted rail promotion efforts. If more states adopted rail promotion programs, national capacity would have slowly improved. Through such a patchwork process, capacity building could have occurred via local, alternative means. But, in the antebellum United States, this is not what occurred. Instead, there was considerable local variation, as some states chose to engage in rail promotions while other states abstained. The antebellum reliance on local state building along with local variation in approaches to rail promotion raises the conundrum of what drives local policy adoption. In the present analysis, I apply contemporary theories of local policy development to the spread of nineteenth-century rail policies to untangle whether early American states' rail development strategies were driven by social learning or parallel thinking.Footnote ⁴⁰ Through clarifying the cause of local policy development, a clearer picture of how local governments contributed to national state building can be developed.

Local policy development in the nineteenth century was driven by a variety of forces, many of which still shape contemporary legislation. In the nineteenth century, state legislators were motivated by a range of local interests, including constituent demands, rural/urban conflicts, partisanship, and ethnic ties.Footnote ⁴¹ Intriguingly, railroads were particularly susceptible to local pressures. Rail companies made choices about technological improvements based heavily on the proximity of materials as well as local usage patterns.Footnote ⁴² However, both in the nineteenth century as well as the present day, local policy development appears to also be shaped by neighboring states' policies. Policy diffusion, or the spread of a policy innovation from one jurisdiction to another, is a common occurrence in the United States.Footnote ⁴³ Through diffusion, state legislatures seek to improve local administrative strength and thereby their own competitive advantage vis-à-vis contiguous states.

There are two theoretically distinct processes that account for policy diffusion. The first mechanism driving policy diffusion is social learning, which is the predominant mechanism explaining contiguous neighbors' sharing policy proclivities. In social learning theory, states share similar policies because they actively borrow from one another. In more detail, social learning diffusion occurs when state legislatures observe policy experiences in nearby states. From observing neighbors' programs, state legislatures can vicariously experience the successes and pitfalls of a novel policy agenda. After evaluating the overall value of a new policy, based on a bordering state's experience, a state legislature can reach an informed decision on whether or not to adopt a similar policy of their own.Footnote ⁴⁴ Social learning diffusion predicts that successful policies should be adopted by many states, moving in a wave across the country. If this occurs, state governments supply necessary services and regulatory activities, effectively providing necessary national state functions through decentralized activity.

Increasingly, social learning diffusion is perceived as a complex social process, involving a series of discrete stages. Andrew Karch argues that social learning diffusion is actually a multistep process, encompassing agenda setting, information gathering, customization, and enactment. In agenda setting, a policy proposal receives attention within a state due to activity in another locale and is thereby considered by activists, interest groups, and legislators in a state. During information gathering, nonprofits and think tanks use policy success in other states to push for a policy innovation in a state lacking a program found elsewhere. Customization involves legislators altering diffusing policies to fit local conditions. Finally, enactment is the passage of a law that resembles a neighbor's policy by a state legislature. Karch emphasizes that social learning diffusion is not merely the enactment of a policy that nearby states have adopted. Rather, social learning diffusion has an arc, with contiguous states influencing one another from the initial phase of policy consideration through the final step of a governor singing a bill into law.Footnote ⁴⁵

Karch's multistep diffusion process allows for a more nuanced discussion of how states influence one another. Notably, Karch's multistep diffusion theory provides for circumstances where an increased number of neighboring states enacting a policy may help set the agenda for a particular state without necessarily leading that state to actually enact that new policy. Further, while agenda setting, information gathering, and customization are important to state building, enactment is the critical action for subnational states' successful participation in American political development. If more neighbors' engaging in novel policies cannot push a state to adopt an innovation, the problem of gaps and policy differences among states limits states' policies' contribution to national capacity.

The social learning diffusion process, as described by Karch, drives the spread of numerous policies among state governments. The most common occurrence of social learning among state governments is the spread of welfare and social safety net policies among contiguous neighbors. For state governments, redistributive policies, which move funds from haves to have-nots, are often perceived as risky propositions by legislators. Redistribution policies tend to lead to higher taxes, and they offer little direct economic benefit to states that engage in them. States with higher taxes find it more difficult to attract new businesses and residents. Essentially, the state places itself at a competitive disadvantage to other states by raising taxes to fund redistribution. Further, state legislatures always fear that by raising welfare benefits, unemployed and impoverished individuals from nearby states will relocate in order to access the higher benefits. The danger of high taxes and the relocation of benefit recipients has led to a “race to the bottom” in welfare spending, where state legislatures are not only hesitant to increase local welfare payments, but will also decrease welfare benefits whenever their neighbors lower their benefits.Footnote ⁴⁶ Reductions in local welfare spending, based on neighbors' lowering welfare budgets, exemplify local policy development driven through social learning diffusion.

This active social learning among states is not due to the natural intellectual curiosity of state legislators, but is rather driven by the structure of American federalism. More specifically, social learning diffusion is driven by competition among states. This subnational competition is fueled by the porousness of American federalism at the subnational level. State governments have no real control over their own borders, a situation that allows people and capital to flow easily between states.Footnote ⁴⁷ The result is that as states compete with one another for capital and population, legislatures assume policies similar to their neighbors in order to maintain a competitive advantage and thereby sustain the interest of investors and skilled populations.Footnote ⁴⁸ Competition, as it occurs between states, is an effort on the part of state legislatures to insulate the local economy from interlopers. Within the broader field of American federalism, state legislatures are attempting to collect more resources than their neighbors, and then prevent those resources that are already contained within the state from leaving for other locales. Thus, competition among states is not necessarily about producing more open internal markets so much as it is about advancing local development and frustrating rivals' economic growth.

This competition among states, termed “rivalistic state mercantilism” by Harry Scheiber, was as common in the nineteenth century as it is today.Footnote ⁴⁹ In the antebellum period, state governments were granted responsibility over a wide range of policy areas, making them particularly powerful political actors.Footnote ⁵⁰ Given the power of antebellum state governments, the strength of state rivalries, and internal improvement's centrality to indigenous economic development, we would anticipate that neighbors' policies would have a significant impact on local policy development, particularly in the realm of railroad promotion.Footnote ⁵¹

In many ways, an examination of the historical record would seem to bolster the claim that nineteenth-century transportation policies diffused from state to state. For instance, states regularly altered the toll rates on their canals in order to attract traffic from competitors. Even when states received help from Army engineers in the planning of early transportation systems, local leaders pushed infrastructure systems that stressed local density over regional connectivity.Footnote ⁵² Hence, regional competition among states colored early transportation polices. For instance, the spread of canal plans among the Atlantic seaboard states, and even into interior states such as Ohio, suggests social learning diffusion driven by competition.Footnote ⁵³ In this vein, Julius Rubin stresses that the construction of the Pennsylvania Main Line system of railroads and canals, as well as rail systems in both Maryland and Massachusetts resulted as direct and purposeful responses to New York's Erie Canal.Footnote ⁵⁴ As a consequence of this historical evidence, local policy convergence on rail promotion programs in the antebellum United States might be anticipated.

However, this same historical record also raises some questions over mechanisms of social learning diffusion, and even questions of whether social learning diffusion actually occurred at all. Rubin notes that the success of the Erie Canal set off a panic among other major Atlantic shipping cities, which feared that New York's dominance of the western market would lead their own economies to permanently lag. In response to the Erie Canal, the other major ports of Philadelphia, Baltimore, and Boston all became entangled in their own internal improvement plans. However, Rubin also notes that each state undertook a very different approach from the other two in their infrastructure development. Notably, Pennsylvania rushed into a canal-railroad hybrid, Maryland also moved quickly on internal improvements but preferred a massive risk on new railroad technology over canals, and Massachusetts bided its time before setting forth on its own railroad investment project.Footnote ⁵⁵ Thus, states engaged in different types of policies and not necessarily at the pace that intense state rivalries might lead us to expect. When this historical account is brought into conversation with Karch's multistage theory of social learning diffusion, it is evident that states did influence one another. However, that influence appears to have occurred during the agenda setting phase of social learning diffusion, and neighbors seemed to exert less impact on one another during the actual policy enactment phase of diffusion. If this is the case, and neighbors cannot push one another to enactment, the contribution of state governments to overall national capacity building comes into question.

However, if states that are near one another do not lead each other to enact similar policies, how do we explain the spatial clustering of policies, especially across such policy areas that range from welfare benefits to railroad construction? While social learning sustained by local competition informs most diffusion theories, there is an alternative explanation for the clustering of similar policies in contiguous states. In this second theory of diffusion, states adopt policies primarily due to their own circumstances and policy experiences.Footnote ⁵⁶ In this account, the fact that contiguous states adopt similar policies has little to do with observation and learning. Instead, the clustering of novel policies among states that share a border results from states in the same region facing comparable problems under related circumstances. In other words, clustering is the result of parallel thinking, rather than social learning. From this perspective, state policy evolves from internal political and economic characteristics, rather than a regional zeitgeist. If parallel thinking occurs, state policies reflect regional problems and resources—and fail to be consistent across the nation. Should parallel thinking occur, state governments' policies are less uniform and therefore less effectively contribute to national capacity. When parallel thinking is the cause of local policy development, state governments' ability to compensate for national policy activity is limited.

As a driving force of policy development, parallel thinking has received less attention. Recently, however, there have been numerous efforts to disentangle when social learning is and is not occurring in local policy development. There is increasing evidence that policies do not diffuse at a constant rate, and an early period of adoption may be followed by much longer gaps in-between new appearances of a policy.Footnote ⁵⁷ This delay suggests that some factors beyond social learning, such as the development of new conditions or opportunities within a state, are driving policy development. In addition, Frederick Boehmke and Richard Witmer have analyzed how the length of time a state possesses a policy impacts neighbors' adoption of that policy. In considering the time a state has been implementing a policy and its impact on neighbors' policies, Boehmke and Witmer argue for the consideration of policy success in evaluating social learning. Through examining duration in this manner, Boehmke and Witmer effectively assess whether social learning, or some other competitive process, is driving policy adoption.Footnote ⁵⁸ Yet, analyses of parallel thinking in policy diffusion remain the exception rather than the rule in studies of state politics. In part, this lack of focus on the role of a policy's success in diffusion is due to the deficit of strong diffusion measures. In order to fully assess whether social learning or parallel thinking is occurring, the value of a policy must be ascertained, by considering its output or longevity. However, the overall utility of a policy is difficult to accurately measure, and current research tends instead to simply consider the presence of a policy when evaluating the diffusion process. The overall utility of a local policy, and how any local benefits impact the spread of that new policy to nearby jurisdictions, is a question that is still open in diffusion studies.

Finally, it is worth noting that, despite limits on travel and communication in the nineteenth century, policy convergence through social learning was feasible in the antebellum period. For social learning diffusion to occur, legislators must be aware of policies in other states. Thus, there must be some mechanism for legislators to learn about neighboring states' programs, as well as the costs and benefits of those programs.Footnote ⁵⁹ State legislatures gain their information about other states' policies from numerous sources, but most notably via social networks and interpersonal connections. Specifically, state legislators are connected to each other across state borders through professional associations, partisan ties, or even personal relationships. Through such connections, legislators can share information about local policies and their utility with distant actors in other states.

During the antebellum period, communication about railroads across state lines was common and surprisingly easy. Numerous national merchant journals, such as Debow's Review, provided information on railroad development in other states. Additionally, there were many national and regional railroad conferences, which brought together investors and elected officials to discuss railroading.Footnote ⁶⁰ With these conditions, there was ample opportunity for railroad policies to spread among political and business leaders. Thus, despite the seeming barriers to communication that could have frustrated social learning in the antebellum period, there were actually sufficient, if not ample, opportunities for state legislators to share their experience with state rail promotion. In the end, both social learning and parallel thinking are plausible explanations accounting for the diffusion of rail promotion programs in the antebellum period. In order to parse between these rival explanations, the impact of both the clustering of rail policies in space as well as the success of contiguous neighbors' rail programs must be evaluated.

LOCAL CONTEXT AND STATE ADOPTION OF RAIL PROMOTION POLICIES

Though the clustering of rail promotion policies in the antebellum United States suggests social learning diffusion, my analysis will argue that rail promotion program enactment actually originated more from parallel thinking than neighbors' influence. Neighboring states' policies may have placed rail promotion on the local agenda of a particular state, but the ultimate decision as to whether or not to enact the policy resulted from the economic and political conditions within a state. The spatial clustering of rail policies resulted not from social learning, but rather from neighboring states sharing similar internal conditions. Internal conditions shaped local policies more than other states' rail programs because it was these indigenous features, and not neighbors' actions, that determined the need and risk of rail investment. State governments that engaged in rail promotion faced the difficulty of managing and funding a major infrastructure program. These immediate, tangible costs and benefits weighed more heavily for state legislatures than the competitive threat a neighbor's own rail program presented.

Though potentially both profitable to the state as well as a boon to the local economy, state-sponsored railroad development was always a gamble. Railroad promotion entailed high costs, long construction times, and uncertain profitability. There was always a risk that rail development could flounder, or even fail. If failure occurred, railroad evelopment programs would backfire: when rail programs failed, they decimated rather than lifted up the state's economy.Footnote ⁶¹ Further, rail development was a time-consuming and essentially permanent process, since tracks were not easily rerouted. States were chained to their rail development programs, and could not easily extract themselves from rail investment.Footnote ⁶² Though states competed with one another for resources, the presence of rail policy in nearby states was simply insufficient to push legislatures to invest in local railroads. Given the risks of long-term commitment rail investment involved for states, the factors pushing state legislatures toward rail promotion hinged primarily on local context. It is these differences in local conditions that determined whether a state engaged in railroad promotion. Local political and economic conditions, such as tax revenue, the presence of policy entrepreneurs, or the size of local obstacles, regularly impact state policy decisions.Footnote ⁶³ The two critical internal factors in shaping local motivation for rail promotion were local economic conditions and a state's extant transportation access. Since I argue that states preferred to avoid bearing infrastructure costs whenever possible, I expect that evidence will show that states were wary of burdening themselves with rail construction costs. The exceptions would be states with stronger economies, which were willing to hazard the risk of rail promotion for the potential benefit to their indigenous rail network. With vibrant economies, strong tax bases, and large amounts of private capital, manufacturing states were especially well-positioned to gamble on rail promotion. Additionally, manufacturing states possessed a vested interest in enhancing the local rail network. Manufacturing states required dense rail networks, since manufacturing required access to both raw materials as well as a means to ship finished products. The inherently diffuse nature of manufacturing means that any state relying on manufacturing as a major sector of its economy requires a strong infrastructure network.

In contrast to manufacturing states' support for rail promotion, I would expect that agricultural states and states with smaller population densities avoided government action to dodge the risk such large capital investments presented. In agricultural states, the economy was often fragile, and the mismanagement of state resources could be a disaster. Historically, many rural states faced bankruptcy due to poorly conceived and managed rail promotion plans.Footnote ⁶⁴ In addition, rural states often lacked the year-round traffic, either in population or bulk goods, to justify large rail programs. In manufacturing states, both raw materials and finished products were moved year round. With their larger populations, manufacturing states also presented a healthier passenger population with which to nurture nascent railroads. In contrast, rural states contained smaller populations, and only moved large amounts of product in the fall, during harvest. The underutilization of rural railroads is best observed in conflicts over rate charges. Low traffic rural railroads often led to high fees to move goods short distances, an economic situation that angered farmers and would directly contribute to the founding of the Grange in the late nineteenth century.Footnote ⁶⁵ Given these conditions, rural state legislatures were reluctant to invest in rail promotion schemes.

In addition to local economic conditions, I also anticipate that local infrastructure quality impacted state legislatures' evaluation of whether government action on railroads was necessary. Much as with local economic conditions, local infrastructure quality directly informed how great the need for state railroad promotion actually was. For instance, states with particularly underdeveloped transportation systems were more likely to have greater demand for state action to maintain their competitive advantage over other states. To gauge how infrastructure quality impacted a state's rail promotion decisions, I included measures of both rail miles within a state as well as a state's water access in the model. If states already possessed large rail networks built through private means, state action on rail promotion became less necessary. State legislators would not endorse spending public funds for promotion efforts if private actors were already meeting local rail needs. Along these lines, when states possessed numerous waterways, I argue that local connectivity was already high. In addition, states with greater water access faced considerable costs in maintaining water routes, as well as vested interests that challenged aid for new, competing infrastructure projects. For example, in New York, early attempts to bring the state into railroad promotion were initially defeated by canal interests.Footnote ⁶⁶ In both situations, when infrastructure within a state was already well developed, state legislators considered it unnecessary to spend resources on additional infrastructure. Therefore, I anticipate that states with poor water access and fewer railroads had greater interest in rail promotion programs, and therefore were more likely to engage in such policies.

Significantly, my argument posits that state legislators' decision to enact a local rail promotion program was not impacted by rail promotion activities in neighboring states. In many theories of local policy development, neighboring states' actions are regularly credited with profoundly influencing local policy choices. For instance, new school funding systems or lottery provisions regularly produce regional competition and inspire the spread of policies from state to state.Footnote ⁶⁷ However, while the geographic clustering of rail programs occurred, legislators' final choice on public investment in railroads was unrelated to the success of neighboring states' rail programs. Policy success, instead of the mere presence of a policy in a nearby state, is a significant factor for explaining policy diffusion through social learning, yet it is rarely directly evaluated. Neighbors' rail programs did place the rail promotion issue on the local agenda, making railroads a salient political issue. In the end, though, rail promotion programs in contiguous states, successful or not, were insufficient motivation to prod local action on the railroad development issue. Though states compete with one another, state legislatures are not “irrational adopters.” Clusters of nearby states adopting similar policies are observed in many policy arenas. However, this clustering results from states facing similar problems with a comparable set of policy tools in a commensurable political context. In other words, clustering results from similar context rather than through diffusion through social learning.

While my primary emphasis is analyzing policy development through considerations of internal economic features and neighbors' rail policies, I also include a number of theoretically relevant controls in my analysis. In particular, the local political culture could exert a substantial effect on state legislatures' approaches to rail promotion. If the local political climate was not predisposed to state intervention in the economy, then state rail promotion policies would be less likely. Therefore, I include a measure of local political conditions. While demand for infrastructure was bipartisan in the antebellum period, Whigs (and later Republicans) were particularly insistent in advocating an active government, especially on internal improvements such as railroads.Footnote ⁶⁸ As a result, I would expect that states with Republican and Whig governors were more likely to pass railroad promotion programs than states with Democratic political leaders.Footnote ⁶⁹

Additionally, I included a measure for a state's slave population in my analyses. The local slave population reflects both a state's economic resources as well as the political values of the state. Given the role of slavery in the antebellum United States, the impact of a state's slave population on rail promotion cannot be overlooked.Footnote ⁷⁰ Now that I have outlined the basic theoretical framework, I will consider how to measure and evaluate the factors that led some states to adopt rail promotion efforts while other states shied away.

ASCERTAINING STATE RAIL STRATEGIES

To analyze the political conditions that led some state legislatures to adopt rail promotion policies, I utilize statistical models along with geographic analyses. Though historical cases are addressed, I emphasize statistical analysis to gain a broader and more systematic view over the spread of rail promotion policies. I analyze the period 1821–1860. This time frame captures the period just prior to the first state rail promotion legislation, which was in Pennsylvania in 1827, until the outbreak of the Civil War. With the large changes in state building and federal power that occurred as the result of the war, 1860 is an ideal theoretical endpoint.

My analysis is built around an event history model, which estimates the factors that led a state to “fail” and adopt a railroad promotion program. An event history model can be thought of as an epidemiological approach to political study. An event history model attempts to predict what leads a “patient” (in this case an antebellum state government) to “die” or “become infected” (engage in a particular action or policy). In interpreting an event history model, a positive coefficient indicates a factor that led to failure, in this case the adoption of a rail promotion policy. A negative coefficient occurs when an increase in the given variable reduced the probability of a state's adopting rail promotion policies.Footnote ⁷¹

For the present analysis, I estimate a Cox-Snell proportional hazards model, with clustering by state. Unlike other event history functions, such as the Weibull or log- logistic models, there is no underlying assumption about the data's hazard rate in Cox-Snell models.Footnote ⁷² The hazard rate assumption gauges how quickly other patients (states) will fall ill (adopt a rail promotion program) once the first patient exhibits symptoms. In the case of railroad development, there is no clear theoretical guideline to estimate how quickly rail promotion will spread. Thus, the Cox proportional hazard model's flexibility makes it the ideal function for historical estimation. In addition, though states could engage in a range of rail promotion programs, a failure in this model is defined as any rail promotion effort.Footnote ⁷³ Further, I use the Breslow method to account for ties, in order to clarify the relationship between the independent variable and the dependent variables when multiple states adopted rail promotion programs in the same year.

The model estimated is state i's railroad policy in time t, as determined by political and economic conditions at time t -1 as well as geographic conditions at time t. The time frame utilized is year. The dependent variable is a binary measure of whether rail support was supplied in that year, with a value of zero or one. At zero, states offer no rail support, while a value of one signifies a state adopting a rail promotion plan. Utilizing secondary historical sources, each state was coded according to the presence of a rail promotion strategy between the years 1821 and 1860.Footnote ⁷⁴ Once a state adopted a rail policy, it was removed from the population and could not move between positions of supporting and not supporting railroads.Footnote ⁷⁵

(1)

To measure whether railroad policies diffused across states, the influence of other states’ policies on local rail promotion programs is measured. Neighbors are defined as all contiguous states sharing a border with the state. As is commonly done in policy innovation studies, the diffusion measures are lagged by one year, allowing for prior conditions to influence future policy decisions. Neighbors' rail policies are ascertained in two distinct manners. The first approach emphasizes only spatial clustering, and is the standard approach in diffusion studies. For this measure, neighbors' rail policies were measured in three ways, and the model was estimated three times: once for each measure of diffusion. The first two diffusion instruments measure neighboring states' rail policies, and weight all neighbors identically. First, a simple count of the number of bordering states with a rail policy was included. This measure is merely a summation of how many of a state's neighbors have any type of a rail promotion policy. This variable measures whether simply having neighbors who engage in any rail promotion was sufficient to push a state legislature to adopt its own rail promotion program. Second, in order to account for the fact that some states possess longer borders, thus providing them with more neighbors, the average value of a state's neighbors' rail policies was calculated. For the average rail policy measure, neighboring policies were again coded on an ordinal scale of increasing rail support. Then, the number of neighboring states with a rail policy was divided by the total number of a state's neighbors.

In addition to the first two measures, a third measure of neighbors' policies was developed by taking the average of neighbors' rail policies and weighting them by travel time between the states. By weighting the averages in this manner, space and communication are more directly incorporated into the model. The weighted value accounts for the role communication with political actors in other states served in local railroad policy development, thereby measuring how information flows from other localities influenced rail promotion decisions within states. The measure addresses the contention that states that are more closely connected with one another would be more likely to imitate their neighbors' railroad policy, since higher connectivity would enhance both the flow of information between states along with the potential competitive pressure among states.Footnote ⁷⁶

The averages were weighted by the travel times between states along existing transportation routes in each decade. To determine the weights, a closest facility network analysis was performed in ArcMap. A closest facility network analysis determines the fastest routes between any two points on a shared geographic network. The state shapefiles were derived from the Historical U.S. County Boundary Files, 1790–1999.Footnote ⁷⁷ To perform the analysis, first the geographic center of each state was determined.Footnote ⁷⁸ Then, the river, canal, and railroad routes, digitized from paper sources for the period of 1820–1860, were layered over the states.Footnote ⁷⁹ In the simulation, railroad travel times were set at 25 m.p.h., while river and canal travel times were held at 12 m.p.h..Footnote ⁸⁰ Finally, straight lines from the geographic centers of each state to the nearest rail, canal, and river vertexes were added.Footnote ⁸¹ Those “roads,” which linked state centers to the broader network, were assumed to facilitate travel at 7 m.p.h. With the network analysis, the length of time required to travel between two geographic points was simulated.

Network analyses were performed for each of the decennial years during the period. Since transport networks developed slowly, the same travel time weights were utilized throughout each respective decade. In order to weight the measures, the inverse square of travel time was utilized. The inverse square is a common transformation in geographic analyses, and produces data where closer events exert a greater influence on a state's railroad policies than more distant ones. Finally, the weighted averages were divided by the variable's mean, to produce more sensible coefficients.

However, as discussed previously, a simple measure of neighbors' policies does not adequately differentiate between social learning and parallel thinking. To overcome this limitation, an analysis of how neighbors' rail promotion successes shaped local rail promotion decisions was needed. Hence, a second set of models with measures that accounted for neighbors' actual policy success with railroad promotion was estimated. Policy success, especially with a program like rail development, is difficult to accurately evaluate. Possible measures include pace of track completion, revenue generated by the railroad, or the percent of the state accessible by rail. However, with a historical case such as nineteenth-century rail promotion, policy evaluation data are exceedingly difficult to come by.

As a result, to measure the success of a state's neighbor's rail policies, I measured the miles of track in neighboring states. The miles of track measure is ideal, since it is a direct and transparent measure of how dense a state's infrastructure system actually was at any given point in time. Further, while miles of track are not a perfect measure of success, prima facie, states with more rail miles per square mile were able to build a larger, denser infrastructure system. By simple virtue of offering rail access to more locales, states with more rail miles per square mile can be defined as successful. To account for size variation among states, and its impact on miles of track within a state, I specifically measured the ratio of miles of track to the area of the state in square miles. The only states included in these measures were those states that possessed rail promotion programs. As with the first set of neighbors' measures, I developed three distinct variables. First, I summed all of a state's neighbors' rail miles/area ratios into a single measure. Second, I determined the average of a state's neighbors' rail/area ratio. Finally, I also calculated the weighted average of the rail miles/area ratio using the travel times developed from the network analysis, as discussed previously. By adding this second set of independent variables, my analysis should more accurately evaluate how the success of neighbors' policies, rather the mere presence of the policies, impacted the diffusion process.

Beyond the policy convergence measures, states' internal characteristics, including economic, political, and geographic factors, were built into the model. While states were diverse, a median state interest was assumed for all internal factors. Local economic conditions were measured with historical census data. Measures for a state's manufacturing capital, slave population, agricultural output, and population density (total population divided by area in square miles) were all incorporated in the model.Footnote ⁸² Values for the years between censuses were interpolated.Footnote ⁸³ When added to the model, each of the variables were also divided by the mean, in order to produce more sensible coefficients. As with the neighbors' rail policy measures, all socioeconomic values were lagged by one year. To address local infrastructure quality, measures of a state's water access and existing rail system were also built into the model. The water access measure is the percent of a state's counties containing some navigable body of water. The railroad measure is the number of rail miles in a state, divided by the variable mean and also lagged by one year.Footnote ⁸⁴

After economic conditions, the next potential internal factor that could impact state rail polices was local political culture. To incorporate local political conditions' impact on state rail promotion adoption, a measure of state political ideology was included in the model. The proxy for state political ideology was the governor's partisanship. The governor value was binary, with Democratic governors coded as one, while Whig, Republican, and Opposition governors were coded as zero. Since the governor was both a locally elected office and captured the opinions of the entire state, the governor's partisanship was an ideal proxy for state political attitudes. Data on historical governor's partisanship was obtained from the National Governor's Association.Footnote ⁸⁵ The governor's party measure was also lagged by one year.

Finally, there are persistent concerns in event history models that observations do not possess a flat hazard rate.Footnote ⁸⁶ If this occurs, biased results are produced. In order to correct for this problem, a temporal dependence control was incorporated into the model. While there are many suggestions for addressing this problem, the solution adopted here was to include a duration dependency covariate as a control.Footnote ⁸⁷ To develop the control, a locally weighted regression on the probability of a state adopting rail support was performed on the time variable. The results of the regression were then incorporated into the model as an additional covariate.Footnote ⁸⁸ These statistical analyses, combined with mapping techniques and historical cases, allowed for a thorough examination of how railroad promotion spread from state to state.