Global Investment Outlook 2019 — The Reckoning That Almost Happened

The 2015 report identified a gap opening between private and public valuations, and noted that no mechanism existed to force the comparison. In 2019, the mechanism arrived: several of the largest private companies of the preceding decade came to public markets.

The results were mixed in a specific and informative way. Companies with clear paths to profitability generally fared acceptably. Companies with large losses, complicated governance, or business models that depended on assumptions that did not survive disclosure fared poorly. In at least one prominent case, the listing was withdrawn entirely after the disclosure process itself destroyed the valuation.

The withdrawn listing is the most analytically valuable event of the year. It demonstrated something that the completed listings could not: that a private valuation can be a function of what has not been disclosed. Public listing requires publishing audited financials, related-party transactions, governance arrangements and risk factors. When that disclosure occurred, the valuation did not survive it — and no shares ever traded.

The second theme of 2019 was a policy reversal. Central banks that had been tightening through 2018 cut rates during 2019, in response to slowing growth and trade uncertainty rather than to any acute crisis. Markets responded strongly, and the episode established a widely-held expectation about the asymmetry of policy response that persisted until inflation broke it in 2022.

The critical fact about 2019, visible only in hindsight, is that the private-market reckoning it began was interrupted. Within months, the pandemic response flooded markets with liquidity, which reopened funding and postponed the correction. The gap did not close — it reopened wider. The 2022 repricing was the same reckoning, delayed by two years and arriving with compound interest.

What public listing actually tests

The events of 2019 clarified something that is easy to state abstractly and easy to underestimate in practice: private and public valuations are produced by different processes and therefore measure different things.

A private valuation is negotiated. A company and a small number of investors agree a price. The investors may hold strategic motives beyond financial return, may receive protective terms that change their effective downside, and are typically buying a small percentage. The price applies to that transaction.

A public valuation is a continuous auction among many participants with different views, holding periods and access to the same disclosed information, any of whom can express a negative view by selling or not buying.

Three specific differences did the work in 2019:

• Disclosure. Public listing requires audited financials, related-party disclosure, governance detail and risk factors. Some private valuations depended on information asymmetry — not through deception, but simply because a small investor group did not demand the scrutiny a public filing compels.
• Structure. Private rounds frequently carry liquidation preferences and other protections. A price paid for protected preferred shares is not the same as a price for common equity, and headline valuations that multiply the preferred price across all shares overstate the whole. Listing converts everything to common stock, and the overstatement disappears mechanically.
• The marginal buyer. In a private round, the price is set by the most optimistic investor willing to transact. In a public market, the price is set where supply and demand clear across all views — including sceptical ones.

A private mark says what one optimistic buyer paid for a protected security. A public price says what the marginal buyer will pay for common stock with everything disclosed. These are not the same measurement, and 2019 was the year the difference was collected.

The policy reversal and what it taught

Having tightened through 2018, major central banks reversed during 2019 and cut rates — not in response to a crisis, but to slowing growth and trade-related uncertainty.

Markets responded strongly, and the episode reinforced a belief that had been forming since the financial crisis: that policymakers would respond asymmetrically, tolerating strength but acting quickly against weakness.

That belief had real consequences for behaviour:

• It compressed risk premia, because a perceived floor under asset prices reduces the compensation investors demand for holding risk.
• It encouraged leverage, on the reasoning that the downside was partially insured.
• It shortened investment horizons, since the relevant question became what policymakers would do next rather than what businesses would earn.

The belief was reasonable given the evidence available at the time. The evidence covered a decade in which inflation was consistently below target, which gave central banks room to support growth without cost.

That condition was the load-bearing assumption, and almost nobody stated it explicitly. When inflation returned in 2021, the asymmetry disappeared — a central bank facing inflation cannot ease into weakness without abandoning its mandate. The lesson is not that the belief was wrong, but that a pattern observed under a stable condition is a claim about that condition, not a law.

The inversion signal problem

The yield curve inverted during 2019 — short-term yields exceeded long-term ones — which historically has been among the more reliable recession indicators.

A recession did follow, in 2020. But the signal was practically useless, and understanding why is instructive.

An inverted curve means markets expect rates to be lower in the future than now, which usually implies expectations of weaker growth. The historical record of the indicator is genuinely strong.

Its limitations are equally genuine:

• The lag is long and highly variable. The interval between inversion and recession has ranged from months to roughly two years. An investor who exits on inversion may forgo substantial returns.
• The sample is small. A handful of observations over several decades supports a claim about direction, not about timing or magnitude.
• The signal describes expectations, not causes. In 2020 the recession arrived through a pandemic, which the curve was not predicting. The signal was correct about the outcome and wrong about the mechanism — which means the correctness was partly coincidental.
• Structural changes may have altered the relationship. Years of central bank bond purchases affected the long end of the curve directly, which plausibly changed what the shape means.

The general problem: an indicator with a strong historical record can be simultaneously correct and unusable, if its timing distribution is wide relative to the horizon over which decisions must be made. Signals need a stated timeframe to be actionable, and this one does not have a reliable one.

The interrupted correction

The most consequential fact about 2019 is not any single event but what did not follow.

The listings of 2019 began a repricing of private valuations. Companies preparing to list saw comparable companies trade below their private marks. Investors began applying public multiples to private holdings. Late-stage rounds became harder to raise on 2019 terms. The correction had started, through the normal mechanism: public prices propagating backward into private markets.

Then it stopped.

The pandemic response beginning in March 2020 produced extraordinary monetary and fiscal support. Rates fell to near zero, liquidity expanded dramatically, and risk assets recovered rapidly and then exceeded prior highs. Private funding reopened at higher valuations than before, and the 2021 boom followed.

The gap the 2019 listings had begun to close reopened wider than it had ever been.

This is the sequence that makes 2022 legible. The repricing of 2022 was not a new phenomenon. It was the same correction, postponed by two years of exceptionally favourable conditions and arriving against a much larger accumulated gap. Companies that would have adjusted to a moderate reset in 2020 instead raised at elevated 2021 valuations and faced a far more severe adjustment.

A correction interrupted is not a correction avoided. It is a correction deferred, usually to a moment when the accumulated gap is larger and the conditions for absorbing it are worse.

That pattern — the postponed reckoning — is among the most useful frames the archive offers, and it recurs. It is worth holding in mind when reading the 2026 outlook's discussion of the AI capital cycle, where the question of when a thesis gets tested is again the central variable.

Why corrections get deferred rather than completed

The interrupted-correction pattern is the most useful frame in this archive, and it is worth setting out as a general mechanism rather than as a description of one episode.

A correction requires a forcing function. Prices adjust when someone is compelled to transact at a price they would prefer not to accept. Absent compulsion, a holder can simply wait, and a private valuation that is never tested is never corrected.

The forcing functions in private markets are few:

• A need for capital. A company that must raise accepts the available price.
• A fund life ending. A manager who must return capital must sell.
• A liquidity need at the holder level. An institution needing cash sells what it can.
• A public listing, which subjects the valuation to an auction.

Each of these can be relieved by capital availability, which is why abundant capital defers corrections rather than merely cushioning them. A company that can raise need not accept a lower price. A fund whose LPs will accept an extension need not sell. An institution with distributions coming in need not liquidate.

And crucially, the deferral has a cost that compounds:

• The gap grows. During the deferral the company continues raising and spending against an unvalidated valuation, so the eventual adjustment is larger.
• The resources that would have absorbed the adjustment are consumed. A company that corrects early does so with runway and options. One that corrects late has neither.
• The cohort expands. Companies funded during the deferral join the cohort that will eventually need to correct.

A correction interrupted is not a correction avoided. It is a correction deferred to a moment when the accumulated gap is larger and the conditions for absorbing it are worse. That is not a prediction — it is arithmetic about what accumulates during the interval.

The pattern recurs throughout this archive. The 2016 late-stage pause was interrupted by returning capital. The 2019 reckoning was interrupted by the 2020 policy response. The 2022 bridge rounds deferred adjustments that expired in 2023, as that year's US venture report describes. In each case the deferral was individually rational and collectively costly.

What an allocator could act on

The 2019 mechanisms have unusually concrete implications, because most of them are answerable from filings.

Compute the implied common-share value. A headline private valuation multiplies the preferred price across all shares. The capitalisation table in any S-1 filing lets you compute what the common shares are actually worth given the preference stack. This is arithmetic, not judgement, and the gap between the two numbers is frequently large. An investor holding late-stage private positions can perform the same computation from their own term sheets and rarely does.

Treat a listing as a disclosure test rather than a liquidity event. The question worth asking of any private holding is: what would this business look like in a document written under legal liability for its completeness? If the answer is materially different from the current presentation, that difference is a risk that has not been priced.

Ask what an indicator's timing distribution is before acting on it. The yield curve's record is genuinely strong and its lag ranges from months to roughly two years. An indicator whose timing distribution is wider than the horizon over which decisions must be made is not actionable regardless of its accuracy. An indicator without a stated timeframe is a claim about direction, not a decision rule.

Identify what a pattern's precondition is. The policy asymmetry belief was reasonable on a decade of evidence, and the evidence covered a period of persistently below-target inflation. Almost nobody stated that precondition explicitly, which meant almost nobody noticed when it stopped holding. Any pattern observed under a stable condition is a claim about that condition, and the discipline is to name the condition at the moment you adopt the pattern.

Watch for the forcing function rather than for the correction. If a correction has begun and then paused, the useful question is which forcing function was relieved and whether the relief is durable. In 2020 the answer was capital availability, and the relief was temporary. Knowing that would not have predicted the timing — but it would have prevented the conclusion that the problem had been solved.

Assume the accumulated gap grew during the pause. A deferred correction is not frozen. Companies keep raising and spending against an unvalidated valuation throughout, so the adjustment that eventually arrives is larger than the one that was avoided. Sizing to the gap as it stood when the correction paused understates it.

What 2019 established

• Public listing was demonstrated to be a disclosure test, not merely a liquidity event.
• Private mark inflation via structure was shown to be mechanical and quantifiable rather than a matter of opinion.
• The policy asymmetry belief was reinforced, along with its unstated dependence on low inflation.
• The limits of long-lag indicators were demonstrated by a signal that was directionally right and practically unusable.
• The interrupted-correction pattern was established, and it is the single most useful frame for understanding 2022.