Do Local Campaign Visits by a Populist Politician Matter in Elections?
Campaign appearances are one of the most common strategies in electoral races. Personal visits and mass rallies appear to be particularly important for populist politicians, allowing them to display closeness to the ordinary people. However, before the turn of the 19th century, candidates rarely reached out to voters personally. The US presidential election of 1896 was a major turning point. In the run-up to the election, the Democratic candidate, William J. Bryan, undertook a major nationwide speaking tour. Bryan was an unknown congressman from Nebraska before his surprising nomination by the Democratic party. He embraced populism, fought for the interests of the common men who were hit by the economic depression of the early 1890s, and revolutionized campaigning by using state-of-the-art technology, the railroad, to address millions of voters. Not surprisingly, Bryan and his campaign regained attention recently when populism had a comeback in Western democracies, going so far as to portray Bryan as “The Trump before Trump”.
In a recent paper, my co-author Stephanos Vlachos and I explore the impact of Bryan´s populist campaign. Many previous studies have looked at the effect of campaign visits on voting outcomes in the 20th and 21st centuries and often found only minimal effects. What was specific about Bryan´s campaign that warrants a careful examination? First, Bryan was a populist politician with outstanding speaking abilities. Second, his campaign was novel and drew immense crowds. Most importantly, the campaign was one-sided as his opponent, the Republican William McKinley, stayed at home. Such a one-sided campaign is a situation that “occurs rarely, if at all, in presidential campaigns”. Thus, the 1896 election provides an almost unique opportunity to study campaign effects in a setting where the possibility that rival campaign visits by the opponent offset a candidate’s efforts is eliminated.
Bryan´s revolutionary campaigning strategy
The 1896 US Presidential Campaign followed the 1893 financial crisis that prompted debates about whether to end the gold standard and to switch to the coinage of silver (“Free Silver”). Bryan supported the free silver movement and the struggle of farmers and laborers against the industrial elites. Unexpectedly, he became the presidential candidate of the Democrats, endorsed by the Populist Party, but without backing from the Democratic establishment.
Unlike his opponent, the Republican William McKinley, Bryan could not draw on abundant financial resources. Considering his meager budget, he decided it would be cheaper to travel around the country and address ordinary people in person, making use of his abilities as an extraordinary speaker. His skills were widely known. “I might just as well put up a trapeze on my front lawn and compete with some professional athlete as go out speaking against Bryan.” McKinley was quoted when considering the idea of going on the trail himself. The Republican instead followed a front-porch campaign and used his funds for print propaganda.
Using the most modern transportation method, the railroad, Bryan traveled on four separate trips to 546 cities, 386 counties, and 26 states, covering 18,009 miles according to his diary (see Figure 2). In total, he gave 746 speeches in 113 days, most of which were identical. The charismatic Democrat attracted big and enthusiastic crowds that celebrated his presence with festivities and parades. In total, Bryan spoke to an estimated 4 million voters.
There was nothing sophisticated about his railroad trips. Bryan would travel on regular public trains, his schedule was only roughly planned, not rigid, and often overturned, giving his tour “a thrown together look”. Still, his campaign followed a logic: Bryan first chose states based on past state-level voting outcomes (making him focus on the North-West). Within these states he focused on locations with more voters to persuade, subject to the constraint of railroad accessibility. For winning the election, Bryan had to convince large parts of the urban population which were less supportive of free silver than the farmers.
The effect of Bryan´s campaign on voting
Did campaign visits by Bryan matter at the election booth? Matching county-level election data with information on the campaign, we attempt to identify whether speeches given by the Democratic candidate affected voting. We focus on a sample of counties connected to the railroad, and compare votes for the Bryan in 1896, conditional on county characteristics and past political preferences. In the cross-section of counties, we find that places visited by Bryan increased their vote share for the Democratic party by about 1 percentage point (Figure 3). This effect is found only in 1896, but not in elections before or after. The effect is robust to various estimation strategies aimed at improving counties’ comparability along their observable characteristics.
We further capture unobserved, time-invariant county characteristics by estimating a difference-in-differences specification that includes county fixed effects. Figure 4a shows that counties where Bryan spoke voted more for the Democratic candidate. Again, there are no pre-trends: speech counties did not vote differently in the elections before 1896.
One concern that arises from the specificity of the historical context is the fact that Bryan was not a typical Democrat, as he was supported by the Populists, too. This raises the concern that pre-trends in Democratic (and Populist) voting might not capture well this “structural break”. We address this issue by exploiting the fact that the Republican party has been more stable in its positions in the decade before 1896. Reassuringly, we find that the vote share for the Republicans significantly declined in places where Bryan spoke, and there are no trends before 1896 (Figure 4b).
Overall, we estimate a very robust effect of Bryan’s speeches on voting that does not seem to be driven by local characteristics unobservable to us. But what is the source of the additional votes for the Democratic candidate in places visited?
Who voted for Bryan in speech localities?
To investigate the origin of the increase in the Democratic vote in speech counties, we classify parties based on their support for the “free silver” or “pro-gold” movement. Our empirical results imply that the gain in the votes for Bryan came at the cost of “pro-gold” parties: Besides the vote share of the Republicans, the vote share of other “pro-gold” parties, declined as well. On the contrary, none of the other minor “free-silver” parties were affected, suggesting that Bryan was successful in winning over parts of the “pro-gold” movement.
Did Bryan convince the urban workers? As expected, in the absence of a speech, counties with a larger industrial labor population voted less for Bryan, while counties with more farmers supported him more. Crucially, however, places with more workers increased their support for the Democratic if he had given a speech. Such effect is not present in places with more farmers. This suggests that Bryan indeed achieved to win over some of the urban workers. Farmers, on the contrary, which already backed his campaign, did not change their preferences when Bryan addressed them in person.
The overall impact of the campaign
Eventually, Bryan obtained only 176 electoral votes, compared to 271 won by McKinley, losing the 1896 Presidential election. What was the overall impact of his speaking tour? Our results imply that the campaign led to an addition of 60,000 votes at the national level (roughly an increase by 0.5 percentage points in the popular vote). Thanks to his tour Bryan probably drove home a close win in South Dakota, but no other state was won because of his visits. Thus, while Bryan gained votes in counties where he spoke, his efforts were not large enough to close the gap with McKinley at the national level.
Although Bryan lost the election and went into history as “one of America’s most important losers”, we document that his speeches were quite successful in persuading voters. This finding implies that campaign effects do not need to be minimal if campaigning efforts and skills of the candidates are unbalanced. It can also help understand why in equilibrium, even nowadays, both candidates usually campaign. The 1896 campaign by Bryan also shows that lopsided campaigns matter, but that it is nevertheless improbable that campaign appearances can completely overturn an electoral result.