Why losing a country to populism is faster than getting it back

Five countries that swung between left and right governments offer a difficult lesson about democratic recovery: the damage moves faster than the repair, the repair is rarely complete, and what gets damaged depends substantially on which kind of populist did the damaging.

If your country's government is moving in an authoritarian direction, the consoling thought is that the next election can undo the damage. It is a reasonable hope and it is partially true. But looking at five countries that have alternated between left and right governments over the past thirty years (Brazil, Argentina, Ecuador, Bolivia, and the United States), the data tells a more uncomfortable story. Populist governments tear down institutions faster than their successors rebuild them. The rebuilding is itself fragile. And the institutions they tear down are not random: left- and right-populist governments break democracies in characteristically different ways, and those differences shape whether recovery is even possible.

What the five cases look like together

I tracked twenty-seven governing periods across these five countries using the V-Dem democracy indices, which measure things like judicial independence, press freedom, electoral integrity, and the rule of law. For each presidency or coherent governing period, I asked how the indices moved between when the leader took office and when they left it (or the present, if they are still in power).

The top half of each country panel shows the institutional democracy score across thirty years. The bottom half shows two measures derived from news coverage, which I will return to in a moment. Each shaded background corresponds to a governing period, color-coded by ideology.

The pattern that emerges across all five countries is consistent. Right-populist governments produce measurable democratic decline, between two and six percent of the maximum score in a single term. Left-populist governments are more variable: some (Correa in Ecuador, Morales in Bolivia) caused major declines, others (the second Lula presidency in Brazil, Alberto Fernández in Argentina) actually improved indicators. The most distinctive feature of contemporary right-populism is consistency: every right-populist phase in the dataset produced decline. Left-populism is more heterogeneous.

When non-populist or differently-populist successors take office, the indices usually recover, but rarely all the way. The Moreno presidency in Ecuador is the only complete reversal in the dataset: he fully undid Correa's decade of damage. Lula's third term recovered most but not all of Bolsonaro's damage to Brazilian institutions. Biden recovered almost none of Trump's damage in the United States, mostly because the damage was concentrated in the federal judiciary, which a new president cannot quickly change.

Left and right populists damage different institutions

Behind the aggregate "all populists damage democracy" finding lies a more specific pattern that turns out to matter enormously for the recovery story. When I extended the analysis to thirteen populist cases worldwide (the five-country sample plus Hungary, Poland, Turkey, India, El Salvador, the Philippines, Venezuela, Nicaragua, and Mexico) and normalized for tenure length, the two ideological camps produced distinctive damage signatures.

Right-populist governments deteriorate institutions about fifty percent faster per year than left-populist ones, and the largest gap is in judicial independence. The right-populist project, in case after case, runs through the courts: Hungarian and Polish court packing, Turkish and Indian judicial subordination, the Trump administration's three Supreme Court appointments. Judicial independence drops at roughly twice the per-year rate under right-populist governments compared to left-populist ones. Media restriction comes second, with self-censorship and government censorship effort moving about thirty to seventy percent faster under right-populist regimes. These are the two domains contemporary right-populism most efficiently captures.

Left-populist governments concentrate elsewhere. The largest signature on the left side is civil society repression: NGOs framed as foreign-funded opposition agents, harassed through tax inspections, defunded, or banned. Chávez did this in Venezuela. Ortega did it more aggressively in Nicaragua. Correa did it through NGO regulations in Ecuador. The second left-populist signature is electoral manipulation: not always outright fraud, but irregularities, intimidation, and the slow erosion of free-and-fair elections. Venezuela and Nicaragua are the extreme cases, but the pattern shows up more mildly in Bolivia and Argentina's Kirchner periods.

These differences are not just academic taxonomy. They explain which kinds of damage are reversible and which are not.

The three country stories, read through the mechanism

Ecuador is the case that should give the most hope and the most caution. Correa was a textbook left-populist: a decade of governing during which civil society organizations were systematically harassed, NGO regulations tightened, electoral integrity gradually compromised, and press freedom indicators dropped. Then Moreno, who had been Correa's vice president and was expected to continue his project, publicly broke with him and proceeded to undo the entire decade of damage. He repealed the NGO restrictions. He restored electoral oversight. Press freedom recovered. By 2022 Ecuador's democracy indicators were back to where they had been in 2006.

The cautionary half of the story is that this recovery is now partly being undone. Noboa, elected in 2023 and re-elected in 2025, is pursuing a security-state agenda that V-Dem reads as the early stages of democratic backsliding. The Moreno-Lasso recovery did not protect Ecuador from a fresh authoritarianizing turn. Reversibility is real, but the reversal can itself be reversed. Crucially: Noboa's signature looks right-populist, hitting judicial and media indicators that Correa had left largely alone. The new damage is in different domains than the old.

Brazil is the case of delayed recovery. Brazilian institutions had already been damaged before Bolsonaro took office, partly by the Rousseff impeachment crisis and the Lava Jato corruption fallout. Bolsonaro's term produced additional decline but from an already-low baseline. His signature was the right-populist one: pressure on the Supreme Court (which Alexandre de Moraes and his colleagues actively resisted), attempts to undermine the federal police's independence, attacks on media credibility, threats to electronic voting integrity. When Lula returned to power in 2023, the recovery was substantial in his first year alone — V-Dem indicators rebounded sharply. By 2024 Brazil had recovered most of the damage. The Brazilian case suggests that institutional decay is genuinely reversible when a successor government actively prioritizes restoration, especially when the courts themselves were the locus of resistance during the damaging phase.

The United States is the case where the damage did not heal. Trump's first term produced a measurable decline in V-Dem indicators, with the bulk concentrated in the federal judiciary because of three Supreme Court appointments that shifted the institution rightward for a generation. This is the right-populist judicial-capture signature in its purest form. Biden's term did not recover this damage, and could not, because Biden could not change the court he inherited. By V-Dem's measurement, Biden's term registered the largest single-period drop in judicial-independence indicators in any of the five countries' recent histories, driven by the Court's own decisions during his term. This is the case that fits the pessimistic ratchet model of democratic backsliding: damage delivered through structural judicial appointments outlasts the administration that delivers it.

The three cases together suggest a pattern. Damage that runs through formally civil-society and electoral channels (the left-populist signature) is reversible by determined successors because the institutions involved respond to current governing decisions: NGO regulations can be repealed, electoral oversight can be rebuilt. Damage that runs through judicial capture (the right-populist signature) is structurally harder to reverse because judges are appointed for long terms and their decisions outlive the administrations that put them there. The asymmetry between Ecuador's full recovery and the United States' failure to recover is, at root, an asymmetry between which institutions were damaged.

What the news sounds like during these episodes

The bottom half of each country panel shows a different kind of measure. To explain it requires a short detour, because the technical version takes a paragraph and the intuitive version takes a sentence.

Imagine you could measure two things for every news article a country produces: the emotional tone of the article (positive or negative) and the conflictual content of the event it describes (cooperative or hostile). In a politically stable country, these two things track each other closely. Bad events get bad coverage, good events get positive coverage, and the news is roughly a faithful mirror of what is happening. I call this responsiveness β. When β is high, the news is reactive to events. When β is low, the news is detached from events, either because press freedom has collapsed, or because coverage has become so polarized that it ignores events in favor of personality, or because something else is happening.

You would expect, intuitively, that β should rise during democratic deterioration. If institutions are being attacked and protests are erupting, the news should get louder and more reactive. This was my prediction when I started the analysis. The data did not cooperate.

Across the five countries, β rises during every recovery period in the dataset. Lula's third term, Moreno's presidency, Alberto Fernández, Arce in Bolivia, even Biden in the United States — all six recovery periods show news coverage becoming more responsive to events. The news gets louder when democracies are climbing back up.

What about during the right-populist periods that produced damage? Mixed results. Milei in Argentina and Bolsonaro in Brazil show rising β, consistent with the original prediction. But Trump's first term and Noboa in Ecuador show β actually falling during periods of measured institutional decline. The news went quieter while institutions were being damaged.

What this means for the noisy-news puzzle

The two patterns put together suggest that β is not a deterioration alarm. It is a turbulence meter. It rises when politics is changing rapidly, whether the change is institutional renewal or institutional erosion. It only stays low when politics is stable, when press freedom has collapsed entirely, or, more interestingly, when coverage has detached from events in some other way.

The Trump case is the most puzzling example of detachment. The United States during Trump's first term produced more news events per month than any other country in the dataset, six times Brazil's volume. There was certainly no shortage of coverage. What seems to have happened is that the coverage stopped being about events in the conventional sense and became about Trump himself. Personality-driven coverage decouples tone from event content, and β falls. Noboa in Ecuador shows a similar pattern, with coverage dominated by ongoing security crisis rather than a variety of differently-toned events.

The implication is that the silence of the news is not always a sign of authoritarian consolidation. Sometimes it is a sign that coverage has reorganized around something other than events. This matters for monitoring purposes: a country whose news has gone quiet might be in the middle of severe press repression, or might simply be in a polarized stretch where the news has stopped pretending to be about events.

The honest limits of what this shows

Thirteen countries and twenty-seven phases across the five-country alternation sample is a small dataset. The patterns I describe are visible here; whether they hold globally would require expanding to fifteen or twenty more cases. The judicial-capture finding for right-populism is robust across the cases I have, but a thoughtful critic could ask whether judicial systems in the right-populist sample (Poland, Hungary, India, the United States) were already structurally vulnerable in ways that made them easier targets, rather than right-populism being uniquely efficient at capturing them. Both readings are consistent with the data.

V-Dem indicators are themselves the product of expert coders making judgment calls. When Lula returned to office, V-Dem coders revised their Brazil assessment upward partly based on their evaluation of his stated commitments, not only based on measured institutional change. Some of the recovery I describe is real institutional repair; some is the coders updating their beliefs about a new government. The data does not let me cleanly separate the two.

The news-coverage findings have their own limits. The β measure relies on automated event detection by the GDELT system, which is reliable enough for cross-country comparison but blurs important distinctions within each country's media environment. The Trump-era finding that coverage detached from events is consistent with the data but not the only possible explanation.

What survives

The cleanest empirical conclusion from this analysis is uncomfortable for anyone who hopes that a single election can reverse democratic deterioration. Sometimes it can: Moreno did this in Ecuador, Lula III did most of it in Brazil. Sometimes it cannot: Biden could not undo Trump's judicial appointments. The variable that distinguishes successful recoveries from failures is not the partisan alternation itself but two things together: what the incoming government actively does, and what institutions the previous government damaged. Moreno proactively rebuilt Ecuadorian civil society and electoral oversight, both of which respond to current governing decisions. Lula III restored Brazilian judicial space partly because the courts themselves had remained the locus of institutional resistance during Bolsonaro's tenure. Biden faced damage in the one domain where a new administration cannot act unilaterally: the Supreme Court he inherited.

If you are watching your own country move in a direction you find concerning, the comparative record is mixed news. It is possible to undo democratic damage through electoral means, when the damage is in domains that respond to electoral change. It is much faster to do the damage than to undo it. The undoing requires a deliberate institutional restoration project, not just a victory at the polls. And the kind of damage that proves most durable is precisely the kind that contemporary right-populism specializes in producing: structural judicial appointments that outlast the political moment that delivered them.

Code, data, and the longer technical version of this analysis at github.com/alejandrofeged/conflict-framing.

---

Alejandro Feged-Rivadeneira is a researcher at Tecnológico de Monterrey working on computational social science, political economy of media, and human ecology.