Network Quality and Integration in the Telegraph Industry, 1852-1861

Tomas Nonnenmacher, University of Illinois

In 1852, the telegraph industry in the United States was in a state of disarray. Although some firms provided excellent service, poorly constructed infrastructure and errors on messages traveling over the lines of more than one firm were the rule rather than the exception. In this paper, I answer two questions concerning the choices of technology and contracts by telegraph entrepreneurs. Why did telegraph entrepreneurs choose to construct low quality infrastructure? What contract enforcement and monitoring costs did telegraph managers encounter when tracking messages over a non-integrated versus an integrated system? I present two models of network interaction. The first model predicts that investment in infrastructure is higher when a network operates under unified control. The second predicts that the transaction costs of tracking messages are lower in a unified network. I find support for both of these predictions in the telegraph industry. The lack of coordination in investment led to low quality lines and service in the industry prior to 1866. Moreover, poorly defined property rights over messages in a non-unified system caused high contracting costs and induced firms to track messages less than they would have in an integrated system.

By 1850 over 100 telegraph companies operating under three patents had incorporated in the United States. By 1861, the industry had integrated into six regional monopolists that were merged into Western Union by 1866. I focus on the period between 1852 and the beginning of the Civil War, as this is where the bulk of the system integration, or the integration of connecting lines, occurred. Other explanations of integration concerning market power account for much of the integration in the industry, but in this paper I focus on efficiency explanations of integration. A high quality network transmits messages to their final destination quickly and without errors. Messages traveling from St. Louis to Boston could travel over the lines of five different firms in 1850. The probability that the recipient received the correct message was influenced both by the investments in infrastructure made by firms along the route and the costs of switching messages between firms. The first problem faced by early telegraph entrepreneurs was one of uncoordinated investment and interdependent quality. The relevant question is whether a network with separately owned components would invest in a higher or lower quality infrastructure than an integrated monopolist. A simple model of uncoordinated investment in a network yields the result that a non-unified network offers lower quality infrastructure than an integrated monopolist. This basic insight is attributed to the propensity of networks to "double-marginalize" when pricing intermediate goods. In terms of telegraph quality, since no firm captured all the benefits of constructing a high quality line, they invested in a lower quality infrastructure than an integrated monopolist.

Telegraph demand was a function of the price of sending messages and the combined quality of all the firms operating in the system. For instance, many of the messages transmitted from Pittsburgh to Philadelphia were transferred to another line and sent onwards to New York. The demand for telegraphic services from Pittsburgh to Philadelphia was based in part on the probability that the message was correctly transmitted on the New York line. This gave the Pittsburgh to Philadelphia line the incentive to under-invest in quality and free-ride off the other lines in the network. According to the model, integrating a network lowers price, increases quality, and increases the profitability of the system. Although quality and profitability increased after integration, the barriers to entry that a complete integrated system offered tended to raise prices during the 1850s.

Although the basic insights from the investment model are powerful, they only explain part of the poor service provided by the industry. To explain quality further, I examine the contracting costs of transferring messages within a network. I find that the high costs of contracting between telegraph firms induced poor supervision of messages that traveled over the lines of more than one firm. The difficulty in contracting between firms was mirrored by the common law's inability to assign liability for errors in transmission.

The customer of the telegraph would not care about mistakes or delays if she was compensated for them. Two questions faced the courts. Who was liable for errors in transmitting messages, and for what were they liable? There were costs to ascertaining the source of an error on a message traveling over several lines. Furthermore, telegraph messages often related to financial matters, and errors could cost customers tens of thousands of dollars. Should firms have been liable for the damages created by an error on their lines?

These were important issues, especially given the interstate character of the telegraph. Some courts categorized telegraph firms as common carriers, others as bailors. The interpretation was important, as common carriers in England typically were liable for delivery to the final destination, unless a contract to the contrary was signed. Bailors were exempted from insuring the goods they held. One result of the lack of clearly defined ownership was that if the firm that originally received the message could prove that it correctly transmitted the message, and no downstream firm kept a record of the message, then the system effectively exempted itself from liability for errors. This lowered the cost of operating a single firm within the system, reduced the quality of network service, and decreased the demand for telegraphic services.

The result is similar to the problem of investing in a network. The costs of contracting between firms induced low quality service. Integrating the transaction within a single firm substituted the contract enforcement cost with the cost of monitoring internal behavior. The question is whether firms could transact internally more effectively. Because no firm in a non-unified network had the incentive to truthfully reveal whether it had made an error in transmitting a message, a telegraph manager needed to monitor the behavior of all other firms to ensure that they transmitted his messages correctly. The costs of monitoring were reduced when monitoring was coordinated through one centralized agent.

Even if the costs of tracking messages between nodes of an integrated versus a non-integrated telegraph network were identical, the common law maintained that the transactions were different. Integrating the network made the unified firm liable for messages to their final destination, while a non-unified network was exempt from liability due to the costs of ascertaining the source of an error. Therefore, a unified firm had more of an incentive to incur the costs of locating and correcting sources of errors in the system. Telegraph managers believed that the increase in demand for the network service more than outweighed the additional legal costs associated with integration.