US Venture Capital Report 2020 — The Reversal

In March 2020, the near-universal expectation across US venture capital was contraction. Firms issued guidance to portfolio companies to extend runway, cut costs, and assume that the next round would be harder and cheaper. Several widely-circulated memos framed the situation in terms explicitly borrowed from 2008.

That expectation was wrong, and it was wrong quickly. Deployment recovered through the second and third quarters and finished the year at levels that exceeded 2019. By the fourth quarter, conditions for founders raising capital were the most favourable in a decade.

The reason was not that the economy recovered — it did not, in any complete sense, during 2020. The reason was the policy response.

Policy rates went to near zero and extraordinary liquidity was provided. That has a specific and mechanical effect on the valuation of early-stage companies, and it is the single most important thing to understand about 2020 venture capital. The value of a company that will not produce cash for years is almost entirely the present value of distant cash flows. The discount rate applied to those cash flows is the dominant variable. When the rate collapses, the present value rises — automatically, without anything changing about the business.

Long-duration assets are the ones most sensitive to that rate. Early-stage venture is the longest-duration asset class in common institutional use.

Alongside the financial reversal, 2020 forced an operational change whose consequences outlasted the year: diligence and investment decisions moved to video. That removed a geographic constraint that had shaped where venture capital was deployed since the industry's founding, and it did not fully return.

Why near-zero rates lift early-stage valuations most

The mechanism is arithmetic, and setting it out precisely explains most of 2020 and all of 2021.

Any asset's value is the present value of its expected future cash flows. To convert a future cash flow into a present value, it is discounted — divided by a factor that grows with both the discount rate and the distance into the future.

Two consequences follow:

• A lower discount rate raises all present values, because less is subtracted for waiting.
• The effect is far larger for cash flows further out. The discount factor compounds with time, so a change in the rate has an exponentially larger effect on a cash flow ten years away than on one arriving next year.

This second point is what "duration" means, and it is why the impact of 2020's rate collapse was so unevenly distributed.

Consider the extremes. A mature business generating cash today derives most of its value from near-term flows, so a lower rate helps but modestly. An early-stage company generating no revenue derives essentially all of its value from cash flows five to fifteen years out — so the same change in the discount rate moves its valuation dramatically.

A rate cut is not a uniform stimulus. It is a transfer of relative value toward whatever pays off furthest in the future — which describes early-stage venture better than any other institutional asset class.

The same mechanism operating in reverse is the entire story of 2022, which is why it is worth understanding as arithmetic rather than as sentiment. The 2021 boom and the 2022 collapse were not two separate episodes of investor psychology. They were the same equation, run with a different rate.

A second channel reinforced the first. When safe assets yield nothing, capital moves toward risk — not from enthusiasm but from necessity. An institution with a return target it cannot meet from bonds must find return somewhere. Venture capital was one of the beneficiaries of that reallocation, as it had been in a milder form since the negative-rate episode described in the 2016 report.

The correction that was cancelled

The 2019 report describes a repricing of late-stage private valuations that had begun to propagate backward from the public markets. That process stopped in 2020, and understanding why matters more than noting that it did.

The 2019 discipline rested on a specific chain of reasoning: public markets would eventually price these companies, public markets were applying a stricter standard, therefore private investors should underwrite to that standard now.

The 2020 policy response broke every link in that chain within roughly six months:

• Public markets recovered and then exceeded prior highs, so the "eventual public price" that private investors were underwriting toward went up rather than down.
• The listing window reopened, and reopened on generous terms — including through blank-cheque vehicles, which offered a route to public markets with different disclosure timing than a traditional listing.
• Capital availability removed the forcing function. A company that cannot raise must accept a lower price. A company that can raise easily need not.

By the fourth quarter, the language of capital efficiency that had dominated late 2019 had largely disappeared from term sheet negotiations. Growth was again the primary metric.

The discipline had not been internalised. It had been imposed by scarcity, and when scarcity lifted, it lifted with it. A constraint that is enforced externally does not survive the removal of the enforcement. That is a general observation about market discipline, and it is the reason the 2022 correction had to re-teach a lesson that 2019 had already taught.

Remote diligence and the geography constraint

The most durable change of 2020 had nothing to do with valuations.

US venture capital had been geographically concentrated to a degree unusual among asset classes. That concentration had real causes rather than merely habit: in-person meetings were considered essential to assessing founders, board participation required travel, and the informal networks through which deals were sourced were local.

In 2020, all of that moved to video, because there was no alternative. The results were not what most participants expected:

• Investment decisions were made without meeting founders in person, at scale, for the first time. Whatever information the in-person meeting had been providing, its absence did not prevent decisions from being made.
• Deal velocity increased. Removing travel from the process compressed timelines substantially, since a partner could take many more meetings in a day.
• Geographic reach expanded. A firm no longer needed a local presence to participate in a market, which opened regions that had been structurally under-served.
• Founders gained optionality. A company outside the traditional hubs could access investors who would previously not have taken the meeting.

The lasting question this raised is whether the in-person requirement had been a genuine informational necessity or a selection mechanism — a filter that limited the deal flow a partner could physically process, with the incidental effect of concentrating capital where partners already were.

The evidence since 2020 suggests substantially the latter. Hybrid processes persisted well after travel resumed, and the geographic distribution of US venture funding remained wider than its pre-2020 pattern, even as it re-concentrated somewhat. That is the clearest lasting legacy of the year.

Pulled-forward adoption priced as permanent

A specific analytical error was made at scale in 2020, and it is worth naming precisely because the same error recurs whenever an external shock accelerates a trend.

Several categories experienced abrupt adoption increases: remote work software, e-commerce, digital health, online education, delivery. These increases were real and, in many cases, very large.

The error was in the extrapolation. An acceleration can be a permanent shift in level, a permanent shift in growth rate, or a pull-forward of demand that would have arrived anyway. These are very different, and they were routinely conflated.

• A level shift means adoption jumped and will stay at the new level, growing from there at the previous rate. Value increases.
• A growth rate shift means the underlying rate of adoption permanently increased. Value increases substantially more.
• A pull-forward means several years of adoption occurred in months, after which growth is slower than it would have been, because the future customers already arrived. Value may barely change at all.

Valuations in 2020 were frequently set by annualising a peak-period growth rate — treating the steepest part of the curve as the new run rate. That implicitly assumes a growth-rate shift, the most valuable of the three interpretations, in categories where a pull-forward was at least as plausible.

The subsequent years resolved this. Several categories that raised at 2020 valuations on pandemic-period growth found that growth decelerated sharply once conditions normalised — not because the business deteriorated, but because the demand had already been served.

The steepest part of an adoption curve is the worst possible place from which to extrapolate a run rate.

What the in-person requirement had been doing

The move to remote diligence was forced rather than chosen, which makes it an unusually clean natural experiment. The results are worth examining because they revealed something about how venture capital had been operating that was not visible from inside it.

The stated function of the in-person meeting was assessment. A partner meeting a founder in person was said to gain information — about conviction, judgement, presence, the dynamic between co-founders — unavailable through other channels.

What 2020 tested was whether decisions made without that information were worse. The honest answer is that it is hard to know, because outcomes take years and the 2020–2021 cohort was affected by so much else. But the immediate evidence was that decisions were made at higher volume, faster, and across a wider geography, without any obvious deterioration in the quality of what was funded.

The alternative explanation for the in-person requirement is that it was a throughput constraint. A partner can take a limited number of in-person meetings, which requires travel, which restricts the geography. That constraint had two effects:

• It limited the number of opportunities a partner could consider, which forced selection earlier in the process — by network, by referral, by proximity.
• It concentrated capital where partners already were, since a company near the partner was cheaper to assess than one far away.

If the requirement was primarily a throughput constraint rather than an informational necessity, then removing it should increase volume and widen geography without degrading outcomes — which is broadly what happened.

The evidence since is mixed but leans toward the second explanation. Hybrid processes persisted well after travel resumed. Geographic distribution of US venture funding remained wider than its pre-2020 pattern, though it re-concentrated somewhat. Neither is consistent with the in-person meeting having been carrying essential information, and both are consistent with it having been a filter whose incidental effect was concentration.

A constraint that everyone treats as a requirement is frequently just a constraint. The way to find out is to remove it, which is rarely possible deliberately and which 2020 did by accident.

What an allocator could act on

Identify the portfolio's dominant macro variable. A 2020 portfolio diversified across sectors was frequently concentrated on the discount rate, because long-duration assets across every sector respond to the same variable. Diversification by label is not diversification by driver, and the check is to ask what single variable would move most of the portfolio.

Treat pulled-forward demand as the default hypothesis, not the exception. An abrupt adoption increase can be a level shift, a growth-rate shift, or a pull-forward. Valuations in 2020 were routinely set by annualising a peak-period growth rate, which assumes the most valuable of the three. The distinguishing evidence — what growth does after conditions normalise — is not available at the time, which argues for holding the question open rather than resolving it favourably.

Do not annualise from the steepest part of a curve. It is the single worst point from which to extrapolate a run rate, and it is the point most 2020 valuations were set from.

Expect externally-imposed discipline to disappear with the enforcement. The capital efficiency standards that emerged in late 2019 evaporated within months of capital returning. Discipline that was imposed by scarcity does not survive abundance, which means the 2022 correction had to re-teach a lesson 2019 had already taught. A behavioural change that was not internalised is not a change.

Reconsider geographic constraints that were never examined. The concentration of US venture capital had causes that were assumed to be informational and were substantially logistical. Any similar constraint — on sector coverage, on stage, on deal size — is worth checking against the same question: is this a requirement or a throughput limit that has been rationalised?

What 2020 established for US venture

• The discount rate was demonstrated to dominate early-stage valuation, in a single year, in both directions.
• Externally-imposed discipline was shown not to survive the removal of the constraint that imposed it.
• Remote diligence permanently altered the geography of US venture deployment, the year's most durable change.
• The pull-forward error was made at scale and provides the clearest available case study of it.
• The 2019 correction was cancelled rather than completed, setting up the 2021 boom and the 2022 reversal.