Political Accountability and the 1918 Influenza Pandemic
By Yuri M. Zhukov and Jacob Walden
When the 1918 influenza A (H1N1) pandemic hit San Francisco and Philadelphia, a tale of two cities emerged in the response of political actors. San Francisco’s mayor ordered an island in the bay quarantined; streets, streetcars, and public phones disinfected; shops, churches, and public spaces closed. Plainclothes police arrested hundreds of people for failing to wear masks and heavily fined others, often aided by other public employees like streetcar operators. Mayor James Rolph, elected as an anti-corruption, pro-business Republican after a series of scandals in city government, was swift in his response, even in the face of protests from local businesses and an “Anti-Mask League” with thousands of supporters. By November 1, San Francisco’s Health Officer Dr. William C. Hassler told a local newspaper that the city was a month ahead of its eastern counterparts in recovery from the brutal second wave of the pandemic.
In Philadelphia, political decisions did not align as closely with health experts’ recommendations. While San Francisco severely restricted public behavior with force, Philadelphia’s response focused on keeping business and key facilities, like the Navy Yard, open. Tens of thousands of troops serving on the European front in WWI passed through Philadelphia’s ports, and the shipyards were a major base of political support for the party machine that controlled the Philadelphia region. At the height of the outbreak — and despite objections from local physicians — Philadelphia Department of Public Health Director Wilmer Krusen allowed a large parade promoting government bonds to be held in the city. Krusen made his decision under pressure from entrenched local politicians from the city’s machine, under the pretext of nationalism, wartime patriotism, and economic activity. The parade became a mass-spreader event, which led to many thousands of deaths.
Why do some political incumbents adopt aggressive measures to slow the spread of infectious diseases while others do not? Public health emergencies raise fundamental questions on the public goods and civil liberties that citizens may expect from their government. In a pandemic, protecting citizens’ health and safety may require mass restrictions on public behavior — from nonpharmaceutical interventions (NPIs) like stay-at-home orders and bans on mass public gatherings, to pharmaceutical interventions (PIs) like testing and vaccines.
These interventions have been a fact of our daily existence over the past two years, but there is nothing fundamentally new about them, or the political backlash that often ensues. Looking at past pandemics can help us better understand why some incumbents are more willing than others to impose such restrictions in the name of public health. While sitting under lockdown last year, Jake Walden and I decided to do just that, by comparing public health responses during Covid-19 to those implemented during the 1918 influenza A (H1N1) “Spanish Flu” pandemic.
Similarities and differences between the 1918 flu and Covid-19
There are striking parallels between the dynamics of the 1918 flu and Covid-19 pandemics, and between the government responses to them.
The symptoms of the two diseases are similar (e.g. fever, sore throat, headache, nasal congestion, coughing, fatigue, muscle pain). Like H1N1, severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) — the virus that causes Covid-19 — spreads through respiratory droplets, aerosols, contaminated objects and surfaces. The two viruses initially had comparable rates of transmission, with basic reproduction numbers (i.e. average number of secondary infections produced by a single infectious case) in the range 2-4 (see here and here).
These similar epidemiological characteristics call for similar methods of mitigation (e.g. social distancing, masking). During both pandemics in the U.S., NPI responses remained largely decentralized, with states, counties, and municipalities implementing and lifting local stay-at-home orders, mask mandates, business closures, and gathering restrictions.
There are also important differences. Unlike H1N1, SARS-CoV-2 became more transmissible over time, with reproduction numbers reaching 5-8 for the Delta variant in 2021, and potentially higher still for the Omicron variant. Covid-19 has a longer period of pre-symptomatic and asymptomatic spread than the flu, further complicating efforts to slow transmission. While the 1918 flu affected many healthy young adults, Covid-19 is most severe in older patients with underlying medical conditions.
The federal response was also more robust in 2020-21 — federal mask mandates at airports, national eviction moratoria, stimulus payments, and federalized vaccine rollouts are four examples among many. By contrast, President Woodrow Wilson made no public mention of the flu pandemic in 1918, and there was no national strategy to fight the spread. There are also other important differences in the political context (e.g. peacetime vs. wartime, presidential vs. midterm election year, diminished dominance of city machine politics), which may affect the incentives facing political incumbents.
More political competition, more aggressive public health response?
Jake and I were particularly interested in what the experience of 1918 can tell us about political accountability. Not all jurisdictions have the same capacity to act. Epidemics require competent public health experts, responsive bureaucracies, and the machinery to enforce regulations. Yet even capable governments may fail to act. Why? If we assume that incumbents generally pursue policies that they expect to maximize their chances (or their party’s chances) of staying in office, then there are two potential answers to this question.
One possibility is that incumbents implement more aggressive public health measures when they face relatively less political competition. In the trade-off between closing the economy and mitigating disease, politicians and voters may have different time horizons on their preferences. NPIs are economically and socially disruptive in the short term, but their public health and economic benefits may not be realized until much later. Voters may be unaware of these future benefits (or may not attribute them to NPIs), but they certainly know that lockdowns are costly to endure. Vulnerable incumbents may therefore prioritize minimizing economic and social disruptions, by taking a more “hands-off” approach and adopting fewer NPIs. By this logic, NPIs are a luxury that only autocrats and entrenched machine politicians can really afford.
An alternative possibility is that political competition increases incentives to intervene. When incumbents’ risks of losing office are relatively low, their incentives to change policy to maximize constituents’ well-being are also relatively low. Political competition changes this calculus. The threat of losing office drives incumbents to more faithfully represent their constituents by implementing policies that are responsive to public demands. In competitive environments, challengers are likely to publicize actual or perceived missteps, making constituents more aware of failed policies. To the extent that the marginal voter might punish incumbents for inaction, political competition should increase incentives for politicians to act (or at least appear to do so).
What the data tell us
Which of these two patterns more closely aligns with the historical record? To assess the impact of political competition on local policy responses (and local severity of the outbreak), we collected municipal and county-level data on four types of public health outcomes in 1918: (1) NPIs and PIs, including the timing of adoption, duration, and enforcement, (2) daily reported cases and deaths, (3) excess mortality, and (4) secondary infections like bacterial pneumonia, which accounted for the majority of flu-related deaths in 1918. We combined these data with information on local electoral competitiveness, based on the outcomes of pre-pandemic congressional, gubernatorial and mayoral elections in the United States. We collected a parallel set of data on Covid-19, to enable comparisons between then and now.
Our analysis, which you can read in full here, suggests that more vulnerable incumbents face stronger political incentives to slow a virus’ spread. How incumbents have responded to these incentives, however, has changed in the last 100 years.
Data from the 1918 H1N1 influenza outbreak suggest that more politically competitive constituencies (i.e. those where winning margins tend to be small, and incumbents are less secure) imposed more nonpharmaceutical interventions, kept them in place longer, promoted them more aggressively and enforced them more often. More competitive localities also saw fewer deaths directly attributable to the flu, fewer overall excess deaths in 1918, and fewer cases of pneumonia.