The Rise of Anger — Emotions and Policy Views

Core finding: Anger in U.S. political discourse is at its highest level in 150 years, and it causally shapes policy views — not just engagement, but the substance of what citizens want from government.

What the paper documents

Algan, Davoine, Renault, and Stantcheva (2026) measure eight discrete emotions across nine datasets spanning 1874–2025: congressional floor speeches (66M sentences), tweets from members of Congress, tweets from voters matched to electoral registries, Reddit comments, and campaign speeches.¹

Demand side (citizens): Anger is the dominant emotion across every policy topic. From 2013 to 2025, the share of angry sentences rises 35% in voter tweets, 79% in climate tweets, and 46% in Reddit comments. This is not a change in who is speaking or what they are talking about — the same users become angrier over time even when discussing the same topics. The rise appears across parties, age, gender, education, and urban/rural groups.²

Supply side (politicians): Anger rises 35% in official party tweets, 72% in congressional tweets, and 50% in floor speeches. Unlike citizens, politicians’ anger follows political cycles — the party losing the White House becomes angrier and remains so until the next election. But members of Congress are consistently angrier online than on the House/Senate floor, and this online amplification is more pronounced among Republicans.³

Engagement incentive: Angry tweets by members of Congress or voters receive ~60% more retweets than neutral tweets. Hope, joy, pride, and gratitude get fewer. This provides a demand-driven explanation for the growing reliance on anger in political rhetoric.⁴

Causal effects on policy views

Two nationwide experiments (9,128 respondents) induce emotions via videos and framings that contain no information about the policies being asked about. Any effect therefore comes from the emotional state itself, not from new information.

Negative emotions (anger/fear/sadness):

• Increase support for protectionism (trade), restrictive immigration policies, and redistribution
• Do NOT increase support for populism (contrary to correlational literature)
• Effects range from 11% to 15% of the Trump–Harris voter gap⁵

Positive emotions (joy/tranquility):

• Improve perceptions of trade and immigration but leave policy views unchanged (the mirror image of negative emotions on immigration)
• Reduce pro-populist attitudes, mainly by reducing the view that expert governance is bad⁶

Anger vs. fear (Survey B, climate):

• Anger strengthens belief in human-caused climate change, increases support for climate policies, and increases willingness to take private actions (effects 13%–27% of the Trump–Harris voter gap)
• Fear has no detectable effect on most outcomes
• “Anger mobilizes, fear paralyzes”⁷

The ACTORC mechanism

The authors build on Bordalo et al.’s (2026) ACTORC framework: emotions change which “category” comes to mind when people think about a policy problem. The category directs attention toward different features of the problem. Anger cues blame/grievance and action; fear cues threat/vigilance and avoidance.⁸

Repeated exposure to anger makes anger-related categories more accessible. This creates a feedback loop: people see anger because others communicate in angry terms, which makes grievance categories easier to retrieve, which makes them more likely to communicate angrily themselves. This mechanism could be contributing to the persistent rise in voter anger since 2016.⁹

Open questions and connections

• Why the phase change? The long-run data shows a sharp rise after the mid-2010s, not a slow creep. Nathan’s hypotheses from the journal discussion: not-so-latent racism triggered by Obama’s election that metastasized; economic regime change from the Great Recession; or temporary shocks that became a vicious cycle via the internet (internet as lubricant, not cause).¹⁰
• Cross-domain links:
  • early-1970s-structural-break — the 1970s wage-productivity decoupling as another candidate structural break; both eras show a discontinuity in the data that resists easy explanation
  • abundance-agenda — Klein/Thompson’s supply-side liberalism as a proposed antidote to zero-sum grievance politics; if anger cues protectionism and restriction, does abundance framing cue the opposite?
  • bullshit-jobs — Graeber’s “managerial feudalism” as a rival diagnosis of the same malaise
  • moral-economy-price-tag — fixed price as moral technology vs. dynamic pricing as the removal of a reference point; anger may be the emotional register of a moral economy that has lost its anchor
  • nudges-vs-prices — List et al.’s finding that nudge effects can be priced as equivalent taxes/subsidies; emotions may be the unpriced complement to cognitive nudges
  • data-center-backlash — the anger series’ emotional register in a single live fight: acyclical, bipartisan, and aimed at the physical plant of the AI boom

Sources

• 2026 — Algan Davoine Renault Stantcheva Emotions Policy Views
• 2026 — Everything Today Feels Angry — Stefanie Stantcheva Twitter Thread

Footnotes

1. 2026 — Algan Davoine Renault Stantcheva Emotions Policy Views ↩

2. 2026 — Algan Davoine Renault Stantcheva Emotions Policy Views ↩

3. 2026 — Algan Davoine Renault Stantcheva Emotions Policy Views ↩

4. 2026 — Algan Davoine Renault Stantcheva Emotions Policy Views ↩

5. 2026 — Algan Davoine Renault Stantcheva Emotions Policy Views ↩

6. 2026 — Algan Davoine Renault Stantcheva Emotions Policy Views ↩

7. 2026 — Algan Davoine Renault Stantcheva Emotions Policy Views ↩

8. 2026 — Algan Davoine Renault Stantcheva Emotions Policy Views ↩

9. 2026 — Algan Davoine Renault Stantcheva Emotions Policy Views ↩

10. 2026 — Everything Today Feels Angry — Stefanie Stantcheva Twitter Thread ↩