Research Briefs in Economic Policy No. 159

Political Discretion and Antitrust Policy: Evidence from the Assassination of President McKinley

By Richard B. Baker, Carola Frydman, and Eric Hilt
April 17, 2019

For the past 40 years, the level of concentration within many sectors of the American economy has increased substantially. This rise in concentration has been associated with a decline in both competition and labor’s share of income. It has also been associated with a slowdown in aggregate output. The failure of antitrust authorities to restrain these developments has provoked concerns that existing antitrust statutes may no longer offer regulators adequate tools for policing anti-competitive behavior. Yet government agencies often hold significant discretion over regulatory enforcement, and it has been argued that stronger enforcement of existing statutes could have reigned in anti-competitive forces. Disentangling the effects of existing statutes from the efforts exerted to enforce them is quite difficult because enforcement efforts are typically not easy to measure or even observe.

We study an extraordinary episode from the Gilded Age when the enforcement of antitrust statutes was suddenly strengthened, and we show that political discretion over antitrust enforcement can have meaningful consequences for the economy. No period in American history witnessed a more significant consolidation of economic activity into large firms than the Great Merger wave of 1895–1904. William McKinley, who was elected president in 1896, was generally friendly toward business interests and did not attempt to use the Sherman Act to challenge these mergers. His assassination by an anarchist in September 1901 presents a unique opportunity to study the effects of a change in the president’s attitude toward enforcement of antitrust laws at a time when all other institutions remained unchanged. In contrast to McKinley, Theodore Roosevelt, who succeeded him as president, was openly critical of big business. The sudden accession of a well-known progressive reformer to the presidency likely shifted expectations regarding the aggressiveness with which antitrust laws would be enforced.

We use the stock market’s reaction to the McKinley assassination to measure the expected impact of this change in the president’s preference for antitrust enforcement. The quasi-random nature of the assassination enables us to estimate the market’s reaction in a way that election outcomes, which are generally well anticipated, do not. The assassination did not coincide with any other major changes; the composition of Congress, the courts, and even the attorney general remained unchanged. But a president who wanted vigorous antitrust enforcement unexpectedly replaced one who did not. In response to the shooting of McKinley, the value of firms traded on the New York Stock Exchange fell by an average of 6.2 percent. To put the magnitude of this fall in perspective, the stock market declined by an average of only 1.6 percent after each of the four other presidential assassinations and after two assassination attempts that nearly succeeded. Importantly, the change in aggressiveness with which antitrust laws were expected to be enforced meant that firms that had engaged in mergers prior to the assassination were more likely to be vulnerable. We find that following McKinley’s shooting, firms involved in recent mergers saw declines in their abnormal returns that were 1.5–2 percentage points greater than those of other firms. We also identify a group of firms that were likely expected to benefit from stronger antitrust enforcement, and we show that the decline in their abnormal returns was about 2–3 percentage points smaller than that of other firms. These results suggest that investors expected a change in antitrust enforcement and that they anticipated that these new policies would have meaningful impacts.

A possible source of skepticism regarding these estimates is that the effects that the change from McKinley to Roosevelt had on the stock market might be the effects of a presidential assassination and not a change in antitrust enforcement. For example, the fact that an anarchist shot the president might have been perceived by the stock market as a sign of rising political instability. Yet the experience of the stock market following the McKinley assassination offers a unique opportunity to address this skepticism. President McKinley initially survived the shooting, and three days later his doctors announced that they expected him to make a “full recovery.” When that prognosis was announced, the losses experienced following his shooting were largely reversed, and firms particularly vulnerable to antitrust enforcement saw differentially large gains. Then, seven days following the shooting, it was suddenly announced that McKinley was in fact near death. Upon receiving this news, the market reversed again with an overall fall in share prices of similar magnitude. Since the effects from political unrest should have been reflected in prices on the day of the shooting, this latter decline in stock prices suggests that investors reacted instead to expected policy changes that would result from Roosevelt becoming president. Finally, when Roosevelt took the oath of office, his statements defied investors’ expectations and signaled that he would follow McKinley’s policy agenda; the result was differential gains for firms that would have more likely been targets of renewed efforts to enforce antitrust statutes.

An additional source of skepticism is that the transition from McKinley to Roosevelt might have been regarded as harmful to particular firms for reasons unrelated to antitrust enforcement. For example, Roosevelt might have been perceived to be less friendly to corporations affiliated with major donors to McKinley’s campaign, such as those connected to J. P. Morgan & Company or the founders of Standard Oil, and to be friendlier toward firms whose executives or directors were personally connected to him. If the firms that were closely affiliated with McKinley and his donors happened to be part of mergers or cartels, that could explain the effects we observe on firms that had engaged in recent mergers. Yet our results are robust to controlling for affiliations with those donors or ties to Roosevelt. It is also possible that Roosevelt’s policy agenda might have differed from McKinley’s on issues beyond antitrust enforcement. But we show that the two most plausible policy differences—which were related to tariffs and labor relations—are unlikely to be the main drivers of our results.

Once in office, Roosevelt violated his pledge to follow McKinley’s agenda and began to enforce the Sherman Act more aggressively. To validate our approach to estimating the role of antitrust issues in the market’s reaction to the assassination, we use an event study methodology to analyze the stock market’s response to the announcement of his first antitrust suit. On February 19, 1902, Roosevelt’s attorney general announced that he was going to file suit against the Northern Securities Company, an enormous holding company formed by J. P. Morgan in 1901 that controlled several major competing railroads. Plans for this suit were kept secret, which enables us to observe the market’s assessment of the expected change in antitrust doctrine that would result from the suit. Our analysis indicates that many of the firms whose shares performed worse in response to bad news regarding McKinley’s health also suffered differentially low abnormal returns following the announcement of the suit.

The analysis of stock market returns only allows us to study the effects expected by investors over a very short horizon. To provide insights into the long-run effects of stricter enforcement of antitrust statutes, we construct a panel of accounting data for railroads listed on the New York Stock Exchange and investigate whether the profitability of those that were most likely affected by Roosevelt’s policy changes declined differentially. The results suggest that the railroads that were differentially vulnerable to stricter antitrust enforcement did indeed see their profitability fall by about 10 percent more than that of other firms. Although we cannot use this difference-in-differences approach to assess the welfare impact of the change in policies, our findings at least suggest that stronger enforcement of antitrust laws might have had strong and persistent redistributive consequences.

This research brief is based on Richard B. Baker, Carola Frydman, and Eric Hilt, “Political Discretion and Antitrust Policy: Evidence from the Assassination of President McKinley,” NBER Working Paper no. 25237, November 2018,

Download the Full Research Brief in Economic Policy

Richard B. Baker, The College of New Jersey; Carola Frydman, Northwestern University; and Eric Hilt, Wellesley College.