E0: COVID-19 Political, Economic & Social Ramifications featuring David Friedberg

Chamath’s claim on 15 March 2020 was that SARS‑CoV‑2 would not be contained in the U.S. but would “roll through the population,” with remaining uncertainty only about the timing and management of that spread. Subsequent data show widespread, population‑level penetration of the virus in the United States, consistent with that prediction. By February 2022, the CDC estimated that about 58% of Americans had antibodies indicating prior SARS‑CoV‑2 infection, including roughly three‑quarters of children. (aha.org) Serosurveys of U.S. adults from August 2021–May 2022 found over 90% had anti‑spike antibodies from vaccination and/or infection, and about 42% had anti‑nucleocapsid antibodies indicating documented past infection, with higher infection prevalence in younger adults. (cdc.gov) Blood‑donor data further show that by July–September 2022, about 96% of donors aged 16+ had SARS‑CoV‑2 antibodies from infection or vaccination, with nearly half having hybrid immunity. (pubmed.ncbi.nlm.nih.gov) These findings, along with repeated national waves of transmission and COVID‑19’s current endemic status in the U.S., demonstrate that the virus did spread broadly through the general population rather than being contained, matching the core substance of the prediction. The quote does not specify an exact timeline or policy details, only that widespread spread was inevitable and that what remained uncertain was speed and management; on that narrower claim, available evidence confirms it was accurate.

The antecedent of Chamath’s conditional prediction clearly occurred: on March 27, 2020, Trump signed the $2.2 trillion CARES Act, widely described as the largest economic stimulus package in U.S. history, passed with overwhelming bipartisan support. (en.wikipedia.org)

However, both implied consequences are not borne out:

1. Historical judgment of his COVID‑19 handling – Early scholarly and expert retrospectives characterize the U.S. federal response under Trump as slow, mismanaged, and marred by political interference in health agencies, despite some notable achievements like Operation Warp Speed. Academic analyses describe the U.S. as having “responded badly” to COVID‑19 and emphasize failures of executive leadership. (pubmed.ncbi.nlm.nih.gov) Public opinion during and after 2020 also skewed negative: majorities consistently disapproved of Trump’s handling of the coronavirus outbreak and judged the U.S. response as less effective than that of other wealthy countries. (pewresearch.org) In broader historical evaluations, Trump has been ranked near the bottom of U.S. presidents overall and last in a prominent 2024 “Presidential Greatness Project” survey, indicating that historians and political scientists do not, on balance, view his crisis leadership favorably. (businessinsider.com) This contradicts the claim that “history will judge that he will have done a good job” on the crisis.

2. Effect on his reelection odds – Trump lost the November 2020 election to Joe Biden. (en.wikipedia.org) More importantly for the probabilistic claim, multiple political‑science studies using county‑level data find that higher local COVID‑19 case counts reduced Trump’s vote share and that, all else equal, he likely would have been reelected had the pandemic been somewhat less severe (e.g., 5–10% fewer cases), implying that the pandemic hurt rather than helped his reelection chances. (pmc.ncbi.nlm.nih.gov) Pew’s synthesis of 2020 public opinion likewise concludes that the pandemic’s impact on the election was “hard to overstate” and that Biden supporters overwhelmingly cited Trump’s handling of COVID‑19 as a major reason for voting against him. (pewresearch.org) This evidence runs directly counter to the prediction that, conditional on a big stimulus, his odds of reelection would be higher than in a world without the coronavirus crisis.

Since the condition (securing an enormous stimulus) was met, but both projected outcomes (favorable historical judgment of his crisis performance and improved reelection odds relative to a no‑COVID counterfactual) are contradicted by subsequent evidence, the conditional prediction is best classified as wrong.

Assessment of the “lockdown → China‑style curve flattening” part

• In many major U.S. metros (e.g., New York City, Seattle, Detroit), stay‑at‑home orders and closures were imposed in the third week of March 2020 and widely followed for several weeks. New York’s “PAUSE” order began March 22, and similar statewide or citywide stay‑at‑home orders were issued in Washington and Michigan on March 23–24. (en.wikipedia.org)
• In New York City, reported COVID‑19 cases rose exponentially through mid‑March but began slowing the week of March 22, with case rates for most adult age groups peaking in the week of March 29—roughly two weeks after March 14—and then declining. (cdc.gov)
• That pattern—sharp growth followed by a peak and flattening within a few weeks of aggressive NPIs—matches the qualitative pattern seen earlier in China, though with local differences in timing and severity.

Assessment of the “IgG antibody tests will be shipped here and clarify how many were infected” part

• The FDA authorized the first COVID‑19 IgG/IgM serology test (Cellex qSARS‑CoV‑2 IgG/IgM Rapid Test) on April 1, 2020, and by mid‑April multiple additional serology assays (including some from Chinese manufacturers such as Autobio Diagnostics) had Emergency Use Authorizations. (fda.gov)
• By late March through May 2020, the CDC and others were already running large U.S. seroprevalence studies using IgG tests to estimate how many people had been infected, which showed infections far exceeding reported case counts and materially improved understanding of the epidemic. (jamanetwork.com)

Conclusion

Both key elements of Friedberg’s forecast—the conditional early flattening of city epidemic curves given sustained lockdowns, and the rapid rollout and use of IgG antibody tests in the U.S. within a few months to quantify cumulative infections—occurred broadly as described, albeit with uncertainties and imperfections. Overall this prediction was directionally accurate.

Evidence shows that no clinical trial results for remdesivir in COVID‑19 were published by late March 2020.

• The first substantial human dataset on remdesivir in COVID‑19 was the New England Journal of Medicine study of 53 patients treated under a compassionate‑use program. This was not a randomized trial and was e‑published on April 10, 2020, more than three weeks after March 14.(pubmed.ncbi.nlm.nih.gov)
• The first randomized, double‑blind, placebo‑controlled clinical trial of remdesivir (Wang et al., Hubei, China; NCT04257656) was published in The Lancet with an Epub date of April 29, 2020.(pubmed.ncbi.nlm.nih.gov)
• In addition, Gilead’s own February 26, 2020 press release about its Phase 3 SIMPLE trials stated that results from the ongoing trials in China were expected in April, not March, indicating that even internal expectations were on a longer timeline than “within the next two weeks” from mid‑March.(gilead.com)
• NIH’s large NIAID ACTT‑1 randomized trial reported preliminary positive results publicly and via press in late April 2020, with peer‑reviewed publication in NEJM in May 2020—again, well after the late‑March window.(xinhuanet.com)

Because no remdesivir clinical trial results (randomized or otherwise) were published by late March 2020, Friedberg’s timeline prediction that such results would be published "in the next two weeks" from March 14, 2020 did not come true.

Friedberg argued that, given adequate beds/ICU/ventilators and use of emerging treatments like remdesivir, the COVID infection fatality rate (IFR) in well‑resourced settings (e.g., the U.S.) could be pushed down to ~0.2%, similar to “Rest of China” outside Wuhan, rather than ~4% as seen in Wuhan.

1. His China benchmark (~0.2%) was already too low. Early analyses of China’s first wave show that outside Hubei, the crude case fatality rate (CFR) was about 0.4%–0.9%, not 0.2%. An ecological study found fatality rates of 0.49–0.88% in the “Rest of China,” while detailed CDC modeling estimated a crude CFR of 0.35–0.43% for China excluding Hubei. (pubmed.ncbi.nlm.nih.gov) IFR (which is lower than CFR because it includes undiagnosed infections) has been modeled for China overall at roughly 0.6%, again well above 0.2%. (arxiv.org) So his reference point itself understated the true fatality risk.

2. Empirical IFR in high‑income countries was many times higher than 0.2% in the pre‑vaccine era, even with modern hospital care. An Imperial College analysis in October 2020 estimated an overall IFR of 1.15% (95% PI 0.78–1.79) in high‑income countries, versus 0.23% in low‑income countries with much younger populations. (imperial.ac.uk) A global meta‑analysis of early IFR studies (through mid‑2020) found a pooled IFR of 0.68% (0.53–0.82%) across populations. (pubmed.ncbi.nlm.nih.gov) US‑specific work using county‑level and seroprevalence data estimated IFRs around 0.8–1.0% for the first wave (e.g., ~0.86% for New York City, with similar national estimates at or below 1%). (pmc.ncbi.nlm.nih.gov) These are 3–5× higher than his 0.2% target.

3. Even over the full pandemic, with vaccines and improved care, population‑level IFR in the U.S. stayed well above 0.2%. As of 2025, CDC‑linked summaries and excess‑mortality analyses estimate roughly 1.2–1.23 million COVID‑19 deaths in the United States and more than 313 million infections by November 2022, with modeling suggesting that by late 2022 about 94%+ of Americans had been infected at least once and 97% had some immunological exposure. (en.wikipedia.org) Taking ~1.2 million deaths over ~313–>330 million infections implies a cumulative IFR on the order of 0.3–0.4%, roughly double Friedberg’s 0.2% claim, even after accounting for later waves with lower severity and widespread immunity.

4. The treatment‑based mechanism he relied on (beds/ventilators + remdesivir) did not deliver such a large IFR reduction. While the ACTT‑1 trial showed remdesivir shortened time to recovery with at best a modest, statistically uncertain mortality effect, (acc.org) the much larger WHO Solidarity trial found little or no impact of remdesivir on 28‑day mortality. (who.int) Major mortality reductions instead came later from corticosteroids, anticoagulation, non‑invasive oxygen strategies, and especially vaccination and prior infection, none of which were the main levers Friedberg emphasized. Even with these advances, age‑structured IFRs in older adults remained far above 0.2%.

5. His conditional (“assuming enough beds/ICU/ventilators”) caveat does not rescue the forecast. Many high‑income settings had periods where hospitals were strained but not completely overwhelmed, yet their inferred IFRs from serology and deaths still clustered around 0.5–1%, consistent with the Imperial and meta‑analytic estimates above and inconsistent with a generalized 0.2% IFR for all infections in a population with a rich‑country age structure. (imperial.ac.uk)

Given that:

• (a) the “Rest of China” fatality rates were significantly higher than 0.2%,
• (b) high‑income and U.S. IFR estimates in 2020 were around 0.7–1.2%, and
• (c) even with vaccines and improved care, the cumulative US IFR has stayed around 0.3–0.4%, not 0.2%,

Friedberg’s prediction that well‑resourced settings could generally drive COVID’s infection fatality rate down to ~0.2% using hospital capacity and treatments like remdesivir did not come true.

Outcome of Trump’s March 13, 2020 COVID‑19 test
Multiple contemporaneous reports state that President Trump took a COVID‑19 test on March 13, 2020 and that the White House physician, Dr. Sean Conley, announced on March 14–15 that the result was negative.

• CNBC reported on March 14, 2020 that, according to the White House physician, “President Donald Trump has tested negative for the coronavirus.” (cnbc.com)
• TIME likewise summarized that the president’s physician confirmed Trump’s COVID‑19 test came back negative in an official memo. (time.com)
• NPR member stations (e.g., KEDM, WHRO) carried the same Conley statement: “This evening I received confirmation that the test is negative.” (kedm.org)
• A summary on the White House COVID‑19 outbreak notes that in March 2020 Trump’s physician announced he had tested negative after exposure at Mar‑a‑Lago. (en.wikipedia.org)

Assessment
Chamath’s prediction was: When the White House announces the results of President Trump’s COVID‑19 test taken on March 13, 2020, they will state that his result is negative.

The documented White House announcement indeed said the test was negative, matching the prediction exactly.

Therefore the prediction is right.

What was predicted
Chamath speculated that around Trump’s March 13, 2020 COVID test, the White House would:

1. Test a larger group of ~50 White House personnel at the same time as Trump.
2. Label samples so that only a tiny inner circle knew which vial was the president’s, and lab technicians would not know which was his.
3. If any of those samples were positive, they would quarantine the entire tested group, rather than publicly isolate only Trump.

This comes directly from the All-In “E0” transcript: he describes testing 50 people in the White House, anonymizing the vials so no lab tech knows which is which, and quarantining everyone if there is “even one positive result.” (podscripts.co)

What we can verify about Trump’s March 13, 2020 test

Public reporting and the White House physician’s memo show only that:

• Trump was tested for COVID-19 on March 13, 2020 after exposure to members of the Brazilian delegation. (ktep.org)
• Dr. Sean P. Conley, the physician to the president, stated in a memo that Trump elected to proceed with testing and that on March 14 he received confirmation the test was negative. (wgbh.org)
• The memo and contemporaneous news coverage describe Trump’s negative result and his remaining “symptom-free,” but do not describe any special protocol involving 50 staffers, pooled or anonymized samples, or a plan to quarantine a whole group based on any positive result. (wgbh.org)

Later reporting on the White House’s COVID practices (including the much-documented White House outbreaks in fall 2020) focuses on inadequate masking, incomplete contact tracing, and limited disclosure of testing, not on any sophisticated pooled/anonymized testing protocol around the president. (en.wikipedia.org) None of these sources retroactively describe the March 13 test as being conducted via a 50-person anonymized group or followed by a “quarantine everyone if any positive” policy.

Why this is classified as ambiguous, not right or wrong

• The internal handling of Trump’s March 13 sample—how many others were swabbed along with him, how vials were labeled, what the lab technicians were told—is not publicly documented in available sources. The Conley memo and news reports simply say he was tested and tested negative. (wgbh.org)
• Chamath’s contingent claim about what would happen if any sample in the group were positive (quarantine all 50 people) is also untestable for that episode, because the White House publicly reported Trump’s test as negative and did not announce any positive White House staff tests or mass staff quarantine tied to that batch at that time. (reviewjournal.com) If no one in such a hypothetical group was positive, the quarantine rule would never have been triggered.
• While later behavior of the Trump White House (e.g., during the 2020 White House outbreak) suggests they were less cautious than Chamath’s scenario assumes—there was no practice of quarantining everyone in a possibly exposed cohort—those later episodes do not directly reveal the precise protocol used for the March 13 test. (en.wikipedia.org)

Because:

• There is no direct evidence that the White House did use a 50-person anonymized testing pool around Trump’s March 13 test, and
• There is also no explicit evidence that they definitely did not use such an internal procedure (those details have not been reported or leaked),

we cannot conclusively determine whether Chamath’s specific, operational prediction about that test protocol came true.

Hence the proper classification is "ambiguous": enough time has passed, but the relevant internal details are not available in the public record, so the truth of the prediction cannot be determined from current evidence.

Evidence indicates that Friedberg’s prediction was broadly accurate on both timing and severity for Tom Hanks and the first wave of NBA players, with some minor caveats.

Tom Hanks timing and characterization

• Hanks and Rita Wilson announced on March 11, 2020 that they had tested positive, describing their symptoms as feeling tired, with body aches and “like we had colds” and mild fever – i.e., explicitly cold‑like rather than severe illness. (heart.co.uk)
• On March 23, 2020, Hanks tweeted: “Two weeks after our first symptoms and we feel better,” emphasizing shelter‑in‑place and that this would pass. This is roughly nine days after the March 14 podcast date and fits the “out of acute illness in ~10 days” window. (time.com)
• Subsequent coverage stated they had been discharged from hospital and were recovering in self‑isolation, and by March 27 they were reported back in Los Angeles after recovering – no reports of a severe, long‑lasting condition from Hanks himself. (economictimes.indiatimes.com)
• Hanks does not appear to have used the exact words “bad cold,” but he publicly framed the symptoms as cold‑like and then quickly “feeling better,” matching the prediction’s qualitative thrust (mild, non‑lasting illness rather than something dramatic or chronic).

NBA players’ outcomes and timing

• Early NBA cases included Rudy Gobert and Donovan Mitchell of the Utah Jazz (announced March 11–12), and Christian Wood (March 14), among others. (en.wikipedia.org)
• Christian Wood’s agent reported him “fully recovered” by March 25, 2020, 11 days after his positive test and essentially matching the ~10‑day window from the March 14 prediction. (en.wikipedia.org)
• The Utah Department of Health cleared the Utah Jazz (including Gobert and Mitchell) of COVID‑19 on March 27, 2020 – about 16 days after their positive tests, still a relatively quick, uncomplicated recovery for young, highly conditioned athletes. (en.wikipedia.org)
• Later reporting shows Gobert had lingering loss of smell months afterward, but notes that he “quickly recovered from the virus” in terms of acute illness; the smell deficit was the main residual effect. (washingtonpost.com)

Assessment

• Core claims – that Hanks and the first wave of NBA players, being healthy and well cared for, would recover from acute COVID‑19 relatively quickly (on the order of ~10 days from mid‑March) and that Hanks would publicly describe it as a cold‑like, non‑catastrophic illness – are supported by the record.
• The exact 10‑day timing is approximate rather than perfect (some players were cleared a bit later), and Hanks didn’t literally say “bad cold,” but he did portray it as akin to having colds and soon after said he and Wilson felt better.

Taken together, the real‑world outcomes line up closely enough with Friedberg’s probabilistic prediction that it should be judged right, with only minor deviations in wording and a few lingering but non‑severe symptoms in one NBA player (Gobert).

Global data strongly indicate that direct COVID-19 deaths far exceed deaths attributable to the pandemic’s second- and third‑order economic and social effects. Estimates that account for under‑reporting and excess mortality place the total number of deaths caused by the COVID‑19 pandemic (overwhelmingly from infection itself) in the tens of millions: by early 2023, analyses compiled by WHO and others suggested roughly 19–36 million total COVID-related deaths worldwide, compared with about 7 million officially reported, with WHO and subsequent methodological reviews explicitly noting that most of the gap between reported deaths and excess mortality is due to uncounted direct COVID infections, not indirect effects. (en.wikipedia.org) Detailed decompositions in high‑quality data settings point the same way. For the United States from March 2020 to April 2021, one study attributed about 90% of 666,000 excess deaths directly to SARS‑CoV‑2 infection, with only about 10% arising indirectly (mainly increased external‑cause deaths such as overdoses and injuries). (pubmed.ncbi.nlm.nih.gov) Another national analysis of 2020–21 multiple‑cause mortality found widespread excess deaths from non‑COVID causes (diabetes, kidney disease, hypertensive heart disease, etc.), but still concluded that COVID‑labelled deaths accounted for roughly 70–75% of excess mortality, with much of the remainder likely including under‑recognized COVID rather than purely economic or social sequelae. (link.springer.com) Estimates of clearly indirect deaths from health‑system disruption and broader fallout, while substantial in absolute terms, are comparatively small. For example, WHO’s malaria program attributes about 63,000 additional malaria deaths in 2020–21 to COVID‑related service disruptions, (who.int) and TB modelling and surveillance suggest on the order of hundreds of thousands of extra tuberculosis deaths globally between 2020 and 2022 compared with the pre‑pandemic trajectory, not millions per year. (who.int) Early modelling of maternal and child health disruptions in low‑ and middle‑income countries projected up to roughly 1–1.2 million additional under‑5 deaths and tens of thousands of additional maternal deaths in extreme scenarios, but these were hypothetical upper bounds and still well below the scale of direct COVID mortality. (pubmed.ncbi.nlm.nih.gov) Taken together, the best available evidence suggests that indirect deaths from economic and social effects of the pandemic number in the low single‑digit millions at most, while direct COVID infection has killed many tens of millions. That makes Chamath’s qualitative claim—that more people would die from second‑ and third‑order effects than from the virus itself—unsupported by current data and therefore wrong.

How the prediction was framed
Chamath predicted that over roughly the next nine months from March 2020 there would be:

1. A widely discussed deleveraging of a multi‑trillion‑dollar global credit bubble.
2. This deleveraging would be “violent and messy,” not orderly.
3. The unwind would leave many people unemployed and many businesses out of business.

What actually happened

1. Leverage did not fall; it surged to record highs

Instead of a broad deleveraging of the global credit bubble, 2020 saw an unprecedented increase in public and private debt:

• The IMF estimates that global debt reached about $226 trillion in 2020, jumping 28 percentage points to 256% of global GDP—the largest one‑year debt surge since World War II, driven by both public and private borrowing. (aa.com.tr)
• The Institute of International Finance (IIF) and related reporting show global debt around $281 trillion by the end of 2020, with pandemic responses adding roughly $24 trillion to the global debt “mountain,” pushing total debt to roughly 355% of world GDP. (bnreport.com)
• In the U.S., federal public debt held by the public rose by about $3.1 trillion just from February to June 2020, reflecting massive deficit‑financed stimulus rather than deleveraging. (en.wikipedia.org)
• U.S. non‑financial corporate debt climbed sharply: companies rushed to issue bonds in early 2020, taking total non‑financial corporate debt to over $11 trillion by early 2021—about half of U.S. GDP—another sign of increased leverage, not net reduction. (spglobal.com)
• There was a brief, specific credit‑card deleveraging episode in Q2 2020, with reduced new borrowing and faster pay‑downs, but researchers at the St. Louis Fed describe it as short‑lived and quickly reversed—far from a broad, sustained systemic deleveraging. (stlouisfed.org)

Overall, the empirical record shows net re‑leveraging, not a multi‑trillion‑dollar deleveraging of the global credit bubble during 2020.

2. The dominant narrative was rising debt and massive stimulus, not ongoing “deleveraging”

Policy and media focus through late 2020 revolved around extraordinary fiscal stimulus, central‑bank backstops, and a growing global debt overhang, rather than a drawn‑out deleveraging process:

• The IMF and others characterize 2020 as a period of record‑high debt levels enabled by low interest rates and large central‑bank asset purchases, warning of the challenges of managing this new debt burden. (imf.org)
• Contemporary analyses from mid‑ and late‑2020 discuss the world being “drowning in pandemic debt” and highlight record debt‑to‑GDP ratios, not a major, ongoing deleveraging cycle. (lombardodier.com)

While some official commentary in early 2020 warned that forced asset sales and margin calls could trigger financial deleveraging in specific markets, this was framed as a risk and a short‑term dynamic within stressed markets, not as the sustained, economy‑wide deleveraging of a global credit bubble that the prediction envisioned. (imf.org)

3. Unemployment and business failures did spike violently

The prediction that many people would lose jobs and many businesses would fail did come true, though this was driven primarily by pandemic shutdowns and policy choices, not by a systemic debt‑reduction cycle:

• Globally, the ILO estimated that between April and June 2020 an equivalent of 400 million full‑time jobs were lost, and workers’ global income fell about 10% in the first nine months of 2020 (over $3.5 trillion). (en.wikipedia.org)
• In the United States, the unemployment rate jumped from 3.5% in February 2020 to 14.7% in April 2020, the highest since the 1940s, with over 20 million jobs lost in a single month. (en.wikipedia.org)
• Yelp’s economic impact data show that by August 31, 2020, around 163,735 U.S. businesses listed on Yelp had closed since March, with about 98,000 (roughly 60%) indicated as permanently closed—large‑scale business casualties consistent with his qualitative claim. (cnbc.com)

So the outcome of severe unemployment and many business closures was accurate, but it occurred in an environment where governments and firms increased borrowing massively to cushion the blow, rather than in a prolonged, system‑wide deleveraging of the credit bubble.

Overall assessment

• Core mechanism & framing: Chamath specifically expected a widely discussed deleveraging of a multi‑trillion‑dollar global credit bubble over the ensuing nine months. The data instead show record increases in global public and private debt and a narrative dominated by stimulus, bailouts, and rising debt, not by a sustained deleveraging cycle.
• Consequences (jobs & closures): He was directionally right that the period would be violent and messy in labor markets and for businesses—unemployment spiked and many firms failed—but those outcomes happened despite massive re‑leveraging, not because of a broad deleveraging of the bubble.

Because the central quantitative and conceptual claim—system‑wide deleveraging of the global credit bubble, widely discussed over the next nine months—did not materialize, the prediction as stated is best classified as wrong overall, even though some of the downstream hardships he anticipated (job losses and business failures) did occur for related but importantly different reasons.

Assessment of each part of the prediction (mid‑March–mid‑December 2020)

1. Claim: “Deleveraging” will be a widely-heard public term over the next nine months
   Evidence: While economists and central-bank researchers did discuss specific forms of deleveraging (e.g., a St. Louis Fed blog on credit‑card deleveraging during the COVID‑19 downturn), this was niche, technical commentary rather than broad public discourse.(stlouisfed.org) The dominant macro narrative in mainstream and policy reporting was about stimulus, rescue packages, liquidity provision, and rising debt, not about a broad deleveraging process.
   Conclusion: This part of the prediction did not materialize in the general public conversation.

2. Claim: A “violent and messy” deleveraging of a multi‑trillion‑dollar credit bubble
   Evidence on actual leverage dynamics: Multiple data sources show that global debt surged in 2020 instead of shrinking:

   • The IMF and World Economic Forum report that global debt jumped by about 28–30 percentage points of GDP in 2020, reaching roughly 255–260% of global GDP, the largest one‑year rise on record.(brookings.edu)
   • The Institute of International Finance (IIF) estimated global debt at ~$272–281 trillion in 2020, up sharply from 2019, and explicitly described the move as a "debt tsunami."(cnbc.com)
   • U.S. public debt alone rose by over $3 trillion in just a few months in 2020 as part of the fiscal response.(en.wikipedia.org)
     This is the opposite of macroeconomic “deleveraging” (which means reducing aggregate debt relative to GDP).(en.wikipedia.org)
     Conclusion: Instead of a credit bubble being forcibly unwound, the world saw massive additional leveraging enabled by aggressive monetary and fiscal support. This core mechanism in the prediction was wrong.

3. Claim: The process will leave “a lot of people unemployed”
   Evidence: This part did occur, though primarily from pandemic and policy shutdowns rather than the bursting of a credit bubble.

   • In the U.S., unemployment spiked from 3.5% in February 2020 to 14.7% in April 2020, with more than 20 million jobs lost in a single month—an historically abrupt labor‑market shock.(en.wikipedia.org)
   • Similar spikes appeared regionally; for example, the San Francisco Bay Area saw unemployment jump by over 13 percentage points in April 2020.(en.wikipedia.org)
     Conclusion: The direction of this sub‑prediction (a large rise in unemployment) was correct, but it was not caused by a deleveraging-driven credit bust as described.

4. Claim: “A lot of businesses [will be] out of business”
   Evidence:

   • Yelp’s September 2020 data showed ~164,000 U.S. businesses marked as closed on Yelp, with ~98,000—about 60%—permanently closed by late August 2020.(cnbc.com)
   • A New York Times analysis (summarized in secondary sources) estimated that up to one‑third of New York City small businesses might never reopen, implying tens of thousands of permanent closures and hundreds of thousands of lost jobs.(en.wikipedia.org)
   • A Federal Reserve–linked study later estimated roughly 200,000 excess permanent U.S. business closures in the first year of the pandemic.(foxbusiness.com)
     Conclusion: This part of the prediction—substantial business failures—did come true in scale, although again as a result of pandemic restrictions and demand collapse, not a global deleveraging.

Overall judgment

The central thesis of Chamath’s prediction was not merely that there would be a bad recession with high unemployment and many closures (which did happen), but that the mechanism would be a widely discussed, “violent and messy” deleveraging of a multi‑trillion‑dollar credit bubble.

Empirically, 2020 featured:

• Record increases in global debt and leverage, not reductions, due to unprecedented fiscal and monetary support.(brookings.edu)
• Public narratives centered on stimulus, liquidity backstops, and rising debt, rather than a prolonged, system‑wide deleveraging episode. (Specialized discussions of specific pockets of deleveraging, like credit‑card balances, were limited and short‑lived.)(stlouisfed.org)

Because the core causal story and “deleveraging” framing did not occur—and in fact the opposite (massive releveraging) happened—while only some secondary consequences (unemployment and closures) matched reality for different reasons, the prediction is best classified as:

Result: wrong

Chamath argued that in a drawdown of this magnitude, the equity market bottom would not be set within ~20 days, but instead around day 200–250, which he mapped to September–October 2020.

In reality, the major U.S. equity indices peaked in mid‑February 2020 and bottomed just over a month later:

• The S&P 500 closed at a then‑record high of 3,386.15 on February 19, 2020, then fell 34% over the next 33 calendar days, reaching a closing low of 2,237.40 on March 23, 2020.(marottaonmoney.com)
• Analyses of the COVID‑19 bear market consistently identify March 23, 2020 as the trough of that crash, after which the S&P 500 began a sustained recovery and ultimately reclaimed and surpassed its prior high.(pbs.org)

Thus, the COVID‑19 equity drawdown bottomed roughly one month after the peak, not 200–250 days later in September–October 2020. The specific timing component of Chamath’s prediction was therefore incorrect.

In the 1–2 quarters after March 2020, investors did not treat COVID-19 as a stock‑market problem “much bigger than 2008.” While the real‑economy shock was indeed projected to be worse than the Global Financial Crisis — the IMF’s April 2020 World Economic Outlook described the “Great Lockdown” as the worst recession since the Great Depression and much worse than the 2008–09 financial crisis in terms of global GDP contraction (imf.org) — equity investors quickly looked through it.

Market behavior shows this clearly. During the 2007–09 crisis, the S&P 500 fell about 57% from peak to trough over a 17‑month bear market (en.wikipedia.org). In contrast, in 2020 the S&P 500 dropped roughly 34% from its February 19 high to the March 23 low, and that bear market lasted just 33 days — the shortest on record (en.wikipedia.org). By August 18, 2020, the S&P 500 had already recovered to a new all‑time high, formally ending that record‑short bear market and signaling renewed optimism despite the pandemic’s toll (investing.com). By year‑end 2020, the S&P 500 and Dow were again at record levels, with the S&P up more than 60% from its March low, a rebound widely attributed to aggressive fiscal and monetary support and expectations of recovery (investing.com).

Earnings data from the first two reporting seasons after March 2020 also undercut the idea that investors concluded COVID‑19 was much worse than 2008 in market terms. FactSet reported that S&P 500 earnings were on track for about a 42–44% year‑over‑year decline in Q2 2020, which would be the largest drop since Q4 2008’s roughly 69% plunge — i.e., still less severe than the worst quarter of the financial crisis (insight.factset.com). Yet forward P/E multiples in April–July 2020 were above their 5‑ and 10‑year averages, indicating investors were willing to pay higher valuations despite the earnings collapse, effectively betting on a rapid normalization rather than a crisis worse than 2008 (insight.factset.com).

Overall, macroeconomic institutions did label the COVID recession deeper than 2008, but the stock market and investors — as reflected in price action, bear‑market depth and duration, and valuation behavior in the ensuing 1–2 quarters — did not broadly conclude it was a much bigger market problem than the 2008 financial crisis. That makes Chamath’s prediction, as stated, wrong.

Summary Friedberg underestimated how long shutdowns would last, but his core claim—that the small‑business economic damage from early shutdowns would show up a few months later and be clearly, severely visible by Q3 2020—matches what happened.

1. Shutdown duration: 2–4 weeks vs. reality (partly wrong) He said the shutdowns the U.S. was in during March 2020 would "probably" continue another 2–4 weeks. In fact, many states kept stay‑at‑home or equivalent orders far longer:

• Example states lifted stay‑at‑home orders only in late May or early June 2020 (Illinois May 30, New Jersey June 9, New Hampshire June 11, etc.).(en.wikipedia.org)
• Michigan’s stay‑at‑home order, begun in March, was repeatedly extended and didn’t finally expire until June 12, 2020.(en.wikipedia.org)
• New Jersey’s stay‑at‑home order was lifted June 9, 2020.(en.wikipedia.org)

So the “2–4 more weeks” part was too optimistic.

2. Q3 2020 timing and severity of small‑business impact (correct) Friedberg’s main point was about second‑ and third‑order effects: even if you “only” lost a few weeks of revenue, the full impact on low‑cash small businesses (salons, restaurants, etc.) would emerge over the next few months and by Q3 it would be “ugly.” That is what we observe in the data:

• A Yelp Local Economic Impact report covering business status as of August 31, 2020 (the end of Q3) found 163,735 U.S. businesses that had been open on March 1 marked as closed, and 60% (97,966) of those were indicated as permanently closed. Restaurants, retail, beauty, fitness, bars, and nightlife were among the hardest‑hit sectors, with tens of thousands of closures and a majority permanent—precisely the kinds of low‑margin, low‑cash businesses he was talking about.(trends.yelp.com)(cnbc.com)
• The U.S. unemployment rate, which had been 3.5% in February 2020, was still 7.8% in September 2020, more than double pre‑pandemic levels, indicating severe and ongoing labor‑market damage through Q3.(trendonify.com)(cnbc.com)
• Real GDP fell at an annualized −31.4% in Q2 2020, the sharpest quarterly decline on record, followed by a +33.1% annualized rebound in Q3. Even with the rebound, this pattern reflects an extreme shock and partial snap‑back rather than a quick return to normal; the unprecedented contraction and choppy recovery are consistent with Friedberg’s warning of non‑linear, “ugly” knock‑on effects rather than a mild, short‑lived hit.(cnbc.com)(bea.gov)

3. Overall judgment

• Wrong element: He significantly underpredicted how long shutdowns would last (many were 6–12 weeks, not 2–4).(en.wikipedia.org)(en.wikipedia.org)(en.wikipedia.org)
• Right, crucial element: He correctly foresaw that the real economic damage—especially to small, low‑cash businesses—would show up with a lag, and that by Q3 2020 the effects would be very clearly visible and severe. Data on mass small‑business closures, elevated unemployment, and historic GDP swings in mid‑2020 aligns with this forecast.(trends.yelp.com)(cnbc.com)(trendonify.com)(cnbc.com)(bea.gov)

Because the central, time‑specific economic prediction (severe, visible, small‑business damage by Q3 2020) came true, the overall assessment is “right,” with the caveat that his guess about shutdown length was too optimistic.

Summary of what actually happened

1. Total size of federal COVID support

   • The CARES Act (March 2020) provided about $2.2 trillion in spending, tax breaks, loans, and other support. (crfb.org)
   • The December 2020 package in the Consolidated Appropriations Act added roughly $900 billion in additional COVID relief. (crfb.org)
   • The American Rescue Plan Act (March 2021) added about $1.9 trillion more. (usafacts.org)
   • Analyses of all COVID bills together conclude Congress authorized roughly $6 trillion of support with a net budgetary cost well over $5 trillion. (crfb.org)
     → On overall scale, federal COVID support was multi‑trillion and actually larger than $3T.

2. How the support was structured
   A breakdown of roughly $3.4 trillion in early COVID relief (through the December 2020 package) shows: (crfb.org)

   • About $935 billion for small business grants and loans (largely the PPP and related programs).
   • About $590 billion for expanded unemployment benefits.
   • About $460 billion for direct rebate checks to individuals.
   • The rest went to health spending, state and local aid, tax relief, and other items.
     While the Paycheck Protection Program and other facilities were technically loans, they were widely designed to be forgiven and functioned more like grants than repayable credit. (crfb.org)

Comparison to Friedberg’s prediction

Prediction (normalized): “The US federal government will ultimately implement on the order of a $3 trillion support program, structured largely as loans or similar financial support that small businesses and unemployed individuals can access.”

• Correct on scale: The total federal response clearly exceeded $3T; being “on the order of $3T” is directionally right, though the actual figure ended up closer to $5T+ in net cost and about $6T authorized. (crfb.org)
• Partially right on targeting: A very large share of the money did flow to small businesses (PPP, other programs) and unemployed workers (expanded UI), matching the groups he emphasized. (crfb.org)
• Not accurate on structure (“largely as loans”):
  • Only about $935B of ~$3.4T in the early bills is categorized as small‑business grants/loans; much of the rest is unemployment benefits, direct checks, and other transfers, not loans. (crfb.org)
  • Even the marquee “loan” program (PPP) was designed for widespread forgiveness, acting more like a grant than a traditional repayable loan. (crfb.org)
  • Later support in the $1.9T American Rescue Plan leaned heavily on direct payments, tax credits, and spending, not loan structures. (usafacts.org)

Why the verdict is ‘ambiguous’

• On the core quantitative and macro point — that the federal government would need and ultimately provide multi‑trillion‑dollar economic backstops — the prediction was broadly validated.
• On the qualitative design point — that this would be “a $3T loan package” structured largely as loans or loan‑like support — the reality diverged: the support was more than $3T and dominated by grants, checks, enhanced unemployment, and other transfers, with loan programs an important but not majority share.

Because a key part of his claim (scale and target groups) was right, but another key part (being largely loans rather than broad fiscal transfers) was wrong, the overall outcome is best characterized as mixed rather than clearly right or clearly wrong.

Evidence shows that shortly after the March 15, 2020 podcast, the Trump administration did exactly what was described: implemented a multi‑trillion‑dollar stimulus with direct support to individuals and small businesses.

• On March 27, 2020, President Donald Trump signed the Coronavirus Aid, Relief, and Economic Security (CARES) Act into law, described by multiple sources as a roughly $2.2 trillion economic stimulus package to provide relief to the U.S. economy and American people during COVID‑19. (energy.gov)
• The CARES Act included $300 billion in one‑time cash payments to individuals (e.g., $1,200 per eligible adult, more for families) and hundreds of billions in support for small businesses, notably through the Paycheck Protection Program (PPP) – forgivable loans designed to keep workers on payroll. (presidency.ucsb.edu)
• Official fact sheets from the Trump White House characterized this as an “unprecedented relief package” exceeding $2 trillion, explicitly framed as providing economic relief and bolstering the economy by giving small businesses and workers the assistance they need to survive. (trumpwhitehouse.archives.gov)

The prediction’s concrete, testable claim was that the Trump administration would launch a very large, direct financial support/stimulus program of multi‑trillion scale, with loans or stipends to individuals and small businesses to protect the economy (with the political goal of helping his reelection). The CARES Act matches these features almost exactly: scale (≈$2.2T), timing (immediately after the prediction), and structure (direct payments plus massive small‑business lending/forgiveness).

While Trump’s motivation (“because he wants to save his presidency”) cannot be perfectly verified from public records, the observable policy outcome he was said to be certain about clearly occurred. Therefore, the prediction is best judged as right.

Summary Chamath correctly anticipated that COVID‑19 would trigger much larger fiscal responses, TARP‑style bailouts, and a German/EU break with austerity. But the specific path and especially the ultimate scale he predicted—global stimulus totaling one to one‑and‑a‑half years of world GDP, largely channeled via the IMF/World Bank—did not materialize and was off by roughly an order of magnitude.

1. U.S. packages after the initial $8B

• The first U.S. COVID bill in early March 2020 was about $8.3 billion in emergency health funding.
• Within weeks, Congress moved not to a mere 10x (~$80B), but to much larger packages:
  • Phase 2: Families First Coronavirus Response Act, about $192 billion, already ~23x the initial $8.3B. (forbes.com)
  • Phase 3: CARES Act, roughly $2.0–$2.2 trillion, passed on March 27, 2020—over 250x the first bill, not just 10x. (taxpolicycenter.org)
  • Later, the Consolidated Appropriations Act (Dec 2020, $868B) and American Rescue Plan (Mar 2021, $1.9T) pushed total U.S. COVID legislation to well over $3.5T. (taxpolicycenter.org)

So while he was directionally right that the U.S. response would balloon far beyond $8B, the actual sequence was much larger and faster than his “10x, then later a couple hundred billion” path.

2. TARP‑like bailouts of specific industries

Chamath predicted that “in a few months” there would be a TARP‑like package of a few hundred billion dollars focused on bailing out specific industries.

• The CARES Act did create a $500 billion facility (Title IV) for loans and loan guarantees via the Treasury’s Exchange Stabilization Fund, explicitly aimed at distressed sectors—$25B for passenger airlines, $4B for cargo airlines, $17B for “businesses critical to national security,” and the rest to backstop Fed lending facilities. This was widely described as a bailout‑type program. (banking.senate.gov)

Qualitatively, he was right that there would be a large, TARP‑style, industry‑focused facility. Quantitatively and temporally, the reality differed: it was about $500B rather than “a couple hundred billion,” and it arrived within weeks, not after a long sequence of gradually larger bills.

3. The “ultimate package” size: 1–1.5 years of world GDP

Chamath’s core macro prediction was that the global fiscal response would ultimately need to be on the order of one to one‑and‑a‑half years of world GDP, deployed across major economies and via IMF/World Bank support for developing countries.

Actual scale of global fiscal support

• World GDP in 2020 was about $85.8 trillion, and around $97.8T in 2021; 2019 was about $88T. (macrotrends.net)
  • So 1–1.5 years of world GDP implies something like $80–130 trillion in combined fiscal packages.
• The IMF Fiscal Monitor (Jan 2021) and related IMF/WEF summaries estimate that global discretionary fiscal support (direct spending, tax relief, guarantees, etc.) reached about $14 trillion by the end of 2020. (imf.org)
• IMF officials and reports in 2020 also referenced governments undertaking around $9–12 trillion in fiscal measures as the crisis unfolded, consistent with that $14T figure once later additions are included. (weforum.org)

Thus, total global fiscal support was on the order of 10–15% of world GDP, not 100–150%. That is roughly one order of magnitude smaller than his “one to one‑and‑a‑half years of world GDP” forecast.

Role of IMF/World Bank vs. domestic programs

• The IMF and World Bank did ramp up support, but at a much smaller scale relative to global GDP:
  • By late 2020, the IMF had provided about $91B to 80 countries; by 2021 it reported around $108–110B in COVID‑related assistance and emergency financing. (english.ahram.org.eg)
  • A later tally notes IMF lending commitments to 94 countries of roughly $287B plus a $650B SDR allocation in 2021—sizable, but far from tens of trillions. (cmacrodev.com)
• Most pandemic fiscal support was national, not channeled primarily through IMF/World Bank–run global programs.

Therefore, the central numerical claim—that the ultimate, globally coordinated fiscal response would approach 1–1.5 years of world GDP and be implemented largely via the IMF/World Bank—was decisively wrong in both scale and institutional structure.

4. Germany and the abandonment of austerity

Chamath also pointed to Germany as a bellwether, predicting it would abandon its austerity stance and participate in large domestic and European packages.

• During COVID, Germany suspended its constitutional “debt brake” and launched more than €450 billion in immediate fiscal stimulus (around 13% of GDP), a sharp reversal of its prior austerity ethos. (cnbc.com)
• At the EU level, member states agreed on NextGenerationEU, a €750B joint recovery fund financed by common borrowing, centered on the Recovery and Resilience Facility (about €672.5B). (consilium.europa.eu)

On this qualitative dimension—Germany dropping its strict austerity line and the EU moving to large joint borrowing—his intuition was broadly accurate.

Overall assessment

• Correct directions:

  • U.S. and global fiscal responses would explode far beyond initial small packages.
  • TARP‑like bailouts for specific industries would appear.
  • Germany and Europe would pivot away from austerity and toward large, coordinated fiscal support.

• Decisively incorrect core claim:

  • The ultimate scale of coordinated global fiscal stimulus was not on the order of 1–1.5 years of world GDP and was not primarily implemented through IMF/World Bank mechanisms. Actual global fiscal support, while unprecedented, peaked at roughly $14T (≈15% of world GDP), an order of magnitude lower.

Because this large, central quantitative prediction about the size and structure of the global response was badly wrong, the overall forecast is best classified as wrong, despite some accurate qualitative elements.

Summary: The prediction mixes one broadly correct intuition (no quick return to normal) with several concrete forecasts that turned out wrong. Overall, the detailed time-path he gave for health, macroeconomy, and the S&P 500 did not match what happened.

1. “0% chance we’re back to business as usual in a couple of months”
Chamath said the probability that COVID-19 and its economic consequences would be fully resolved and the U.S. would be back to business as usual by roughly May 2020 was 0%. In reality, the federal COVID-19 public health emergency ran from January 31, 2020 until May 11, 2023, and major social and economic disruptions persisted well past mid‑2020.(en.wikipedia.org) Early May 2020 small‑business surveys showed most firms expected more than six months before returning to normal operations, not a quick rebound within a couple of months.(census.gov) This part of his view was directionally correct.

2. Health impacts largely dealt with in 8–12 weeks (by May 2020)
He predicted that “first order effects of the disease” would mostly be dealt with within 8–12 weeks, and that by around May 2020 “we’ll be through most of the worst of it.” In fact:

• By April 30, 2020, the U.S. had about 61,000 recorded COVID deaths and 1+ million cases.(en.wikipedia.org)
• Far larger waves followed: by January 19, 2021 the U.S. had passed 400,000 deaths, and by September 2021 COVID had killed over 675,000 Americans, surpassing estimated U.S. deaths from the 1918 flu.(en.wikipedia.org)
• The pandemic continued with major surges through 2021–22; it did not become a minor residual issue after May 2020.

Given that the deadliest waves and the bulk of total mortality came after his 8–12 week window, the claim that the “worst of it” would be over by May 2020 was wrong.

3. Economic bottom in Q3–Q4 2020
Chamath forecast that the economic bottom would come in Q3 or Q4 2020. Instead:

• The National Bureau of Economic Research later dated the recession’s trough to April 2020, implying the recession ran only from February to April 2020.(nber.org)
• Real GDP fell at a record annualized rate of about –31.4% in Q2 2020, then rebounded sharply with +33.1% annualized growth in Q3 2020, and grew again in Q4.(bea.gov)

So on standard macro indicators (NBER dates, GDP), the economic bottom occurred in spring/Q2 2020, not in Q3–Q4. This timing prediction was wrong.

4. S&P 500 falling to ~2000 or lower
He expected the S&P 500 to “touch 2000 … if not lower.” The actual market behavior:

• The S&P 500’s pandemic crash bottomed on March 23, 2020, with an intraday low of 2,191.86 and a closing level of 2,237.40, then began a sustained rebound and went on to make new record highs later in 2020.(statmuse.com)
• The index never reached 2000 or below during that episode.

Thus, his specific level call for the S&P 500 was too bearish and incorrect.

Overall assessment

• Correct: Skepticism that the pandemic and its economic fallout would be fully resolved and “back to normal” within a couple of months.
• Incorrect: (a) that most of the worst health impacts would be over by May 2020, (b) that the economic bottom would be in Q3–Q4 2020 rather than in spring/Q2, and (c) that the S&P 500 would fall to ~2000 or lower.

Because the three concrete, testable sub‑predictions about health timing, economic trough, and market level all turned out wrong, the overall forecast is best classified as wrong, despite getting the general “no quick normalization” intuition right.

Evidence shows that severe COVID‑19–related shutdowns of social and economic life extended far beyond two months from mid‑March 2020 in many parts of the world.

• National lockdowns exceeded two months in multiple countries. For example, India’s nationwide lockdown, begun in late March, was repeatedly extended and continued with major restrictions through at least 31 May 2020, already more than two months after late March. 【1search13】 New Zealand’s first nationwide lockdown ran from 26 March to 27 May 2020 (about two months), but the country then used further regional lockdowns in August–September 2020 and February–March 2021, plus a prolonged nationwide lockdown from August to December 2021. 【1search15】
• International travel and tourism remained effectively shut down well beyond mid‑May 2020. As of 20–27 April 2020, 100% of global destinations still had COVID‑19 travel restrictions in place, and no destination had yet lifted or eased them; about 72% had a complete stop on international tourism. 【0search8】【0search5】 These restrictions and border controls continued, in varying forms, through the rest of 2020 and into 2021.
• Lockdowns and strong restrictions recurred internationally through 2020–2021. The global survey of “COVID‑19 lockdowns by country” documents that many countries and regions (e.g., Australian states, European countries) re‑entered lockdowns or curfews multiple times in late 2020 and 2021, well beyond the initial two‑month window. 【1search12】
• Global economic and social activity remained dramatically depressed throughout 2020. In aviation alone, a key proxy for international social and economic activity, 2020 global passenger traffic (RPKs) fell 65.9% vs. 2019, with international traffic down 75.6%, and air connectivity more than halved. 【2search3】 This indicates that cross‑border travel and related social/economic activity were far from “re‑opened” after just two months.
• Policy trackers show sustained high stringency of restrictions. The Oxford COVID‑19 Government Response Tracker’s Stringency Index records generally high levels of closure and containment measures across many countries well past May 2020, reflecting that significant constraints on social and economic activity persisted globally through much of the year. 【0search6】【0search9】

Because large parts of the world remained under substantial restrictions, with key sectors like international travel, tourism, and large gatherings effectively suppressed for many months to years, the prediction that global economic and social activity would be “effectively shut down” for only about two months starting in mid‑March 2020 materially underestimated the duration of the disruption. Therefore, the prediction is wrong.

Chamath’s normalized prediction is that global travel demand after the early‑2020 COVID shock would return to its pre‑shock level in about 19 months (i.e., around October 2021).

What actually happened:

1. By late 2021, global travel was still massively depressed.

   • UN Tourism (UNWTO) data show that international tourist arrivals in 2020 and 2021 were 72% and 71% below 2019 levels, respectively—meaning 2021 as a whole was still nowhere near pre‑pandemic demand.
   • For July–September 2021 (about 17–19 months after the initial shock), international arrivals were still 64% below 2019 globally, with Europe and the Americas also tens of percent below 2019 levels. (unwto.org)

2. Global air travel only neared 2019 levels in late 2023.

   • IATA reports that full‑year 2023 global air passenger traffic reached 94.1% of 2019, with December 2023 at 97.5% and Q4 2023 at 98.2% of 2019 levels. Another IATA release notes November 2023 demand at 99.1% of November 2019.
   • This implies that a near‑full recovery in air travel demand took roughly 3.5 years, not 19 months. (iata.org)

3. Global tourism (broader travel demand) recovered even later.

   • UN Tourism estimates that international tourism was only 63% of pre‑pandemic levels in 2022 and 88% in 2023, projecting a full recovery by end‑2024.
   • Their 2024 data confirm this: about 1.4 billion international tourists in 2024, 99% of pre‑pandemic levels—a practical return to 2019 volumes about four years after the initial shock. (unwto.org)

Given that both air travel and international tourism remained far below 2019 levels at the 19‑month mark and did not effectively recover until late 2023–2024, the prediction that global travel demand would recover in about 19 months was not borne out by the data.