Americans vote on Tuesday in an election that remains extraordinarily close. Kamala Harris and Donald Trump are competing across a small group of battleground states that will determine whether Democrats retain the presidency or Trump completes an extraordinary return four years after losing it. The immediate arguments are familiar: inflation and immigration favour Trump; abortion and concerns about Trump’s character and democratic conduct favour Harris; both campaigns insist the stakes are unusually high. Yet beneath the polling lies a more serious institutional problem. Large parts of both political coalitions increasingly believe that losing this election would represent something more dangerous than an ordinary policy defeat. Many Democrats believe another Trump presidency threatens constitutional constraints and democratic institutions. Many Republicans believe government agencies, courts, media organisations and electoral institutions have become systematically hostile towards their movement. Democracy becomes considerably harder to sustain when elections stop being competitions over who governs and begin being interpreted as contests over whether the political system itself can still be trusted.
This matters because democratic stability requires something counterintuitive: citizens need to care enormously about winning while remaining willing to lose. Elections only work when losers believe another opportunity to compete will remain available. If defeat means permanent exclusion, persecution or destruction of the political system, accepting defeat becomes irrational. Political scientists sometimes describe this through losers’ consent. The legitimacy of democracy is demonstrated not principally by whether winners like the system but by whether losers continue accepting its rules.
The United States experienced a severe breakdown of this norm after the 2020 election. Trump continues to claim that election was stolen despite courts, state officials and his own administration failing to establish evidence of fraud capable of changing the result. The culmination was 6 January 2021, when supporters attacked the Capitol while Congress was certifying Joe Biden’s victory. The institutional importance of that episode goes beyond the violence itself. A sitting president attempted to overturn or obstruct the ordinary mechanism through which electoral defeat becomes succession.
That should remain part of how voters evaluate Trump.
But there is another danger if Democrats interpret every Trump voter as rejecting democracy consciously. Millions of people support him primarily because of immigration, prices, taxation, cultural politics or dissatisfaction with Biden’s administration. Some believe the prosecutions against Trump themselves demonstrate politicisation of institutions. Others dislike Trump’s conduct but rank it below economic concerns.
This creates an interpretation gap. Democrats see voters tolerating anti-democratic behaviour. Trump voters see Democrats using the language of democracy to delegitimise their preferred candidate.
Both sides therefore increasingly believe they are defending the political system against the other.
That is a dangerous equilibrium.
The economic backdrop explains part of Trump’s resilience. By conventional macroeconomic measures, the United States has performed considerably better than many expected. Unemployment remains low, economic growth has been relatively strong and inflation has fallen sharply from its 2022 peak. The Federal Reserve began reducing interest rates in September.
Yet households do not experience an economy through GDP growth.
They experience prices.
The cumulative rise in food, rent, insurance and other everyday expenses since 2021 remains highly visible. Inflation falling towards 2–3% does not reverse those increases. It merely slows additional increases.
This distinction between the inflation rate and price level may be one of the most important political-economic lessons of the election. Policymakers celebrate falling inflation because the rate of deterioration has slowed. Voters compare today’s prices with the prices they remember.
A household whose grocery bill increased from $150 to $200 does not feel that the problem disappeared because the bill is now rising only slowly.
This creates a political asymmetry.
Governments receive blame when prices rise rapidly.
They receive much less credit when prices merely stop rising rapidly.
The original price increase remains visible every time someone purchases the product.
That may explain why incumbents globally have struggled after the pandemic inflation shock regardless of ideology. Inflation produces a persistent political scar even after the macroeconomic variable improves.
Immigration creates another state-capacity problem. Border encounters reached extraordinary levels during Biden’s presidency before declining more recently following policy changes and Mexican enforcement. Republicans argue that the administration allowed the border to become uncontrolled. Democrats argue that Trump helped derail a bipartisan border bill earlier this year because continued dysfunction benefited him electorally.
Both arguments reveal something important: immigration policy becomes politically toxic when citizens believe government cannot implement whichever rules formally exist.
A country can choose high immigration.
It can choose lower immigration.
What destabilises confidence is the impression that the actual number entering results less from deliberate policy than from administrative incapacity.
This is why border control and immigration openness are not necessarily opposites. Credible control can make legal immigration more politically sustainable because citizens know admission levels reflect collective decisions.
Trump’s strongest political insight has been recognising that many voters interpret immigration through sovereignty and control rather than economic optimisation alone.
His proposed solutions, including mass deportations and much more restrictive enforcement, create enormous legal and economic questions. But political competitors cannot answer those proposals effectively by implying the underlying concern is illegitimate.
The same principle applies to manufacturing and trade. Trump has proposed substantial tariffs as part of a broader economic-nationalist programme. Economists generally warn that tariffs raise prices and reduce some efficiency. Trump responds that efficiency is not the only objective: domestic production, industrial employment and reduced Chinese dependence possess strategic value.
This argument has moved substantially closer to the economic mainstream since 2016.
Biden maintained many Trump tariffs.
His administration imposed additional restrictions on Chinese technology and subsidised semiconductor and clean-energy manufacturing.
Both parties now accept some version of economic security.
Trump therefore shifted the policy equilibrium even while outside office.
The disagreement concerns how far protection should go.
A targeted semiconductor subsidy can be justified through national-security externalities.
A universal tariff on almost every imported product is harder to defend because consumers pay higher prices across sectors without equivalent strategic benefits necessarily emerging.
Protection should therefore be evaluated according to additional productive capacity created per unit of consumer cost.
That is an economic test rather than an ideological one.
Harris faces a different structural problem because she is simultaneously an incumbent and a replacement candidate. Biden withdrew only in July after his debate performance intensified concerns about his age. Harris inherited his campaign infrastructure and achievements but also inherited public dissatisfaction with his administration.
She therefore needs to argue that the government performed better than voters believe while simultaneously presenting herself as change from that government.
That is a difficult message.
Trump’s message is much simpler.
You remember my presidency.
You dislike the current one.
Return to the previous one.
Political communication rewards clear counterfactuals, even when those counterfactuals are incomplete.
The Trump economy before COVID benefited from trends that began earlier. Biden inherited pandemic disruption. Global energy and supply shocks affected inflation. Presidents do not control every macroeconomic outcome.
Voters nevertheless evaluate presidents partly through association.
That is retrospective democracy.
It is imperfect.
It is still one of the mechanisms through which governments remain accountable.
The abortion issue pushes in the opposite direction. The Supreme Court’s 2022 Dobbs decision, enabled partly by Trump’s three appointments to the Court, eliminated the federal constitutional protection established under Roe v. Wade and returned abortion regulation largely to states. Since then, abortion-rights positions have performed strongly in several referendums, including in conservative states.
This demonstrates another complication in American politics: voters can prefer Republican candidates while supporting policies Democrats favour.
Party choice bundles issues together.
A voter cannot select Trump’s immigration policy, Harris’s abortion policy, Biden’s industrial policy and a libertarian tax system separately.
They receive packages.
Elections therefore produce bundled preference aggregation.
This is why interpreting victory as a mandate for every policy is analytically weak.
If Trump wins, some voters will have chosen him despite his abortion legacy.
If Harris wins, some will have chosen her despite disliking Democratic immigration policy.
A presidential election determines who receives executive authority.
It does not reveal a national referendum result on every issue simultaneously.
The Electoral College complicates this further because the presidency depends on state-level outcomes rather than the national popular vote. In 2016, Trump won despite receiving fewer votes nationally than Hillary Clinton. The system is constitutionally established, so such a victory is legally legitimate. But repeated divergence between popular and electoral outcomes can create political legitimacy concerns because citizens intuitively expect national offices to reflect national numerical majorities.
Defenders argue that America is a federal union and presidential candidates therefore need geographically distributed support.
Critics argue that a vote in Wisconsin should not possess more practical electoral importance than a vote in California simply because one state is competitive.
Again, this is a federalism-versus-majoritarianism trade-off.
There is no neutral electoral system.
Every rule changes political incentives.
The important requirement is that rules remain stable enough that actors know them before voting and that results are counted according to those rules afterwards.
This brings us back to Tuesday.
The most important institutional event may not be who wins.
It may be how the loser responds.
If Trump loses, will he concede if evidence demonstrates a Harris victory?
If Harris loses, will Democrats accept that voters knowingly returned Trump despite their warnings about him?
Both questions matter.
Democracy requires political actors to distinguish “this result is dangerous” from “this result is illegitimate.”
A result can be terrible in your judgement and still legitimate.
That distinction is the foundation of peaceful alternation.
The United States has spent four years arguing about whether it still possesses that norm.
Tuesday will test it again.
America’s institutional strength remains enormous.
Courts operate.
States administer elections independently.
Civil society is active.
Media remain pluralistic.
Federalism distributes authority.
The military remains outside partisan electoral administration.
Those are substantial safeguards.
But institutions ultimately depend on citizens and leaders accepting constraints when those constraints produce outcomes they dislike.
No constitution can automate democratic restraint completely.
The stakes therefore extend beyond Harris and Trump.
Nearly half the country will lose.
The real democratic test begins when they do.
If voters can say “I believe this government will be harmful, but I accept that it won and will organise to defeat it next time,” the system remains resilient.
If defeat becomes proof that the system itself must have been fraudulent, every election becomes progressively harder to resolve peacefully.
America does not need political consensus on Tuesday.
It needs something more modest and more important.
A credible count.
A legitimate winner.
And a loser willing to remain inside the system long enough to compete again.
