Here’s how a pandemic-driven recession could affect elections up and down the ballot this November.
How we did our research
Using election returns over the past five decades and economic data from the Bureau of Economic Analysis, we compared election-year economic performance and votes for presidents, governors, senators and members of the House of Representatives.
In the graphs below, you can see how growth or loss in average wages per worker and increase or decrease in the incumbent president’s party’s share of the vote are correlated during presidential election years, for all four types of elections. When wages increase, the president’s party tends to do better at all four levels. On average, over the past 45 years, a 1 percent change in wages is associated with about a 1-percentage-point change in votes for the incumbent president’s party.
But how can you separate the influence of the economy from everything else?
Of course, many factors affect national elections. For instance, Democratic candidates did poorly in 1980 — but it’s hard to tell how much that was because of the poor economy and how much was because of the Iran hostage crisis. Similarly, Republicans did poorly in 2008 — but how much of that was because of the Great Recession and how much because of the unpopular war in Iraq?
To figure that out, we looked at local variation in the economy within particular states in a given election year. If a particular county’s economy was doing better or worse than its neighbors’ within the same state while other national conditions remain the same — if, say, “Springfield” is in a recession but “Middletown” is thriving — we can see whether people in Springfield and Middletown voted differently in the same year.
Using this strategy in a new study, we compare administrative data on county-level economic conditions from 1969 to 2018 with election results for U.S. local, state and federal offices.
Just as the graphs above suggest, we find that over the past five decades, voters hold the president’s party accountable for the economy — at nearly all levels of government. The graph below shows the county-level relationship between changes in wages and changes in Democrats’ average vote shares in presidential, House and Senate elections under a Democratic president (in blue) and a Republican president (in red). Our model accounts for year- and state-specific factors that could affect elections and shows the relative changes in wages and share of the vote.
The results in the graph show the average Democratic vote share across the entire range of economic performance. The president’s party tends to do about 0.15 percent better for every 1 percent improvement in the local economy — and about 0.15 percent worse for every 1 percent decline in the local economy.
And that’s true across all federal elections: Candidates from the president’s party are rewarded and punished for changes in the local economy.
So how might a pandemic-prompted recession affect this fall’s elections?
These results suggest that if the virus prompts a 3 percent decline in local wages, Republican candidates at the federal level would get a 1 percent smaller share of the vote than they would if there were a 3 percent increase in local wages. Local and state candidates from the Republican Party would have similar declines in their share of the vote as well.
Some analysts have argued that the economy doesn’t matter anymore. Voters today perceive the economy (and so much else) through a strongly partisan lens, goes this line of thinking. Even when there’s unambiguous information about the economy, as was true during the Great Recession, people attribute blame based on their partisan views.
Surveys do show that whether people approve of the president’s performance is increasingly decoupled from what they think of the economy, which TMC editor John Sides argues may be because our politics are getting more polarized. But we find that the president’s party still gets credit for a strong local economy and blame for a weak one. For instance, over the past decade, areas with a growing economy were more likely to support Democrats when Barack Obama was president, while they were more likely to support Republicans when Donald Trump was president.
Shouldn’t Trump get the benefit of the overall economic growth since 2016? That’s not how it works. Like other political scientists, we find that voters judge politicians based on the last year before they cast their ballots, not over the full four years of a president’s term. This suggests that Trump and Republicans could lose votes for 2020’s economic distress, despite earlier growth.
Politicians seem to understand this, which could explain Trump’s recent proposal for a payroll tax cut through Election Day. Of course, both the pandemic and the economy could change dramatically between now and November. But if not, Republicans may be unhappy with November’s results — not just for the presidency but in the Senate, House, governors’ seats and state legislatures.
Justin de Benedictis-Kessner (@jdbk) is an assistant professor at Boston University. His research addresses accountability, representation, local politics and public opinion.
Chris Warshaw (@cwarshaw) is an assistant professor at George Washington University. His research focuses on political representation, public opinion and elections.