US Venture Capital Report 2013 — A Mark Is Not a Price

2013 is when the billion-dollar private company acquired a label, and the label became a goal.

The underlying phenomenon was real and follows directly from the US Venture Capital Report 2011: once large capital was available privately and the compulsion to list was gone, companies stayed private at valuations that would previously have required a listing.

What the label obscured is that the number attached to these companies is not the same kind of object as a public market capitalisation.

A public price has properties that make it informative:

• Many participants transact, so it aggregates dispersed information.
• It is continuous, so it updates as information arrives.
• It is the price of a defined security whose terms are public.
• And it is executable — you can transact at approximately that level.

A private valuation has none of these:

• One buyer negotiated it, so it reflects that buyer's view and constraints, not a consensus.
• It is a point in time and does not update until the next round, which may be two years later.
• It is the price of a preferred share with undisclosed rights, applied to all shares including common — the distortion set out in the US Venture Capital Report 2011.
• And it is not executable. No one can sell at that level, as the private secondary market's discounts demonstrate.

Calling this a valuation and comparing it to a market capitalisation is a category error, and it drove a great deal of the period's confused debate about whether valuations were justified.

The year's second structural development is that late-stage private investing separated into its own asset class, with different investors, different return expectations and different risks from the venture capital it grew out of.

Why a mark behaves differently

The distinction between a mark and a price has consequences that are easy to overlook and quite large.

A mark is sticky downward, and asymmetrically so.

• When a company performs well, a new round is raised at a higher price, and the mark rises promptly. There is an eager party — the company — with every incentive to establish the higher number.
• When a company performs poorly, no round is raised at all. The mark stays where it was. Nobody has an incentive to establish a lower number, and there is no mechanism that forces one.

So marks rise on good news and stay flat on bad news, until the company either raises at a lower price or fails outright.

Three consequences follow:

Reported volatility is understated. A portfolio of private marks displays smooth returns because the downside is not recorded until it is undeniable. This is a measurement artefact, and it is frequently mistaken for genuine diversification — private assets appear to have low correlation with public markets partly because they are not being measured at the same frequency.

Reported returns lead reality on the way up and lag it on the way down. The peak of a private portfolio's reported value comes well after the peak of its actual value.

And the correction arrives in discrete steps, since the reset happens at a financing event. A portfolio can report stability for two years and then reprice all at once, which is what the US Venture Capital Report 2022 documents.

Private assets look less volatile than public ones in the same way a photograph looks less volatile than a film. The subject is moving either way.

The practical adjustment is to treat reported private volatility and correlation as lower bounds, and to look through to the underlying exposures — a private software company and a public software company are exposed to the same things, whatever the reported correlation says.

When growth over profit is correct

The period's dominant operating doctrine — prioritise growth, accept losses, defer profitability — is neither right nor wrong in general. It is right under conditions that can be specified and checked.

The underlying logic: if acquiring a customer produces more value over that customer's life than it costs to acquire, then spending on acquisition creates value, and spending more creates more. Profitability now would mean forgoing that value.

The conditions required for this to hold:

The lifetime value must genuinely exceed the acquisition cost, calculated honestly — with real retention data rather than assumed retention, with gross margin rather than revenue, and discounted for the time over which it arrives.

Retention must be measured, not projected. A company with eighteen months of history is projecting a five-year customer life from a curve it has not observed. This is the single largest source of error, and it is systematically optimistic because early cohorts of any product are its most enthusiastic users.

Acquisition cost must not rise with scale. The cheapest customers are acquired first. If cost per customer rises as the easy segments are exhausted, the economics that justified the spending deteriorate exactly as the spending increases.

The payback period must be financeable. A customer who repays acquisition cost in six months can be funded from operations; one who takes four years must be funded from investors for four years, which is a financing risk independent of whether the arithmetic works.

And a durable advantage must accrue from being early — a network effect, switching costs, or scale economics. Without one, the growth buys revenue that a competitor can take back with the same tactic. This is the condition most often assumed and least often present.

When the conditions hold, growth-over-profit is straightforwardly correct. When they do not, it converts investor capital into revenue at a loss and calls the result traction.

The reason the doctrine spread beyond its valid domain is that it is unfalsifiable in the short run. All five conditions concern the future — retention not yet observed, costs not yet at scale, advantages not yet tested. A company can be wrong about all of them for years while reporting excellent growth. The US Venture Capital Report 2022 covers the reckoning.

Late stage becomes its own asset class

The capital entering large private rounds was doing something structurally different from venture capital, and conflating the two caused persistent misunderstanding.

Traditional venture capital:

• Invests early, when failure is the base case.
• Expects most investments to fail entirely and returns to come from a small number of extreme outcomes.
• Holds concentrated positions with governance rights and active involvement.
• And accepts a decade-long horizon.

Late-stage private investing:

• Invests in established companies with substantial revenue and identifiable market positions.
• Expects most investments to return capital, with returns from moderate multiples rather than extreme ones.
• Holds smaller stakes with limited governance, often no board seat.
• And expects a shorter horizon, frequently underwritten to a listing within a few years.

These are different activities with different risk profiles, and the second is much closer to public growth equity investing than to venture capital.

Why the distinction matters for allocators:

• The return distribution is different. Venture is tail-driven; late-stage is not. A late-stage portfolio's outcome depends on the median investment, which means diligence on individual companies matters far more than it does in a seed portfolio.
• The risk is different. Venture risk is that the company fails; late-stage risk is principally that the entry valuation was too high. The second is a pricing risk and requires pricing discipline, which is a different skill from company selection.
• And the liquidity assumption is different. Late-stage investors underwrote to an exit within a few years — an assumption dependent on public market conditions entirely outside their control.

The last point is where the model was most exposed, and the US Venture Capital Report 2022 documents what happened when the assumed exit window did not open.

What naming a category did

The threshold label had effects beyond description, and they are a good illustration of how a metric becomes a target.

As a descriptive term it was useful — it identified a genuinely new phenomenon, since large private companies had previously been rare.

As a target it distorted behaviour:

• The threshold became a goal for founders, since achieving it conferred status, press coverage and recruiting advantages entirely disproportionate to the difference between just below and just above.
• And the threshold could be reached through structure rather than performance. A company unable to raise at the target valuation on clean terms could reach it by granting the investor stronger protections — a higher liquidation preference, a ratchet, guaranteed returns on a listing.

This is the mechanism that makes the headline number least reliable exactly when it is highest:

An investor indifferent between a lower valuation with clean terms and a higher valuation with strong protections will take whichever the founder prefers. Founders preferred the higher number, because the number was public and the terms were not.

So the valuation and the protections moved together in the wrong direction. The companies with the most impressive headline numbers were disproportionately those that had paid for them with structure, and structure is precisely what destroys common shareholder value in a mediocre outcome.

The information problem is that only one side of the trade was visible. The valuation was announced; the terms were not disclosed. An observer could not distinguish a company that raised at a billion on clean terms from one that raised at a billion with a guaranteed multiple attached — and these are enormously different situations.

The US Venture Capital Report 2014 covers the structure arms race that followed, and the 2022 report covers the outcomes.

The network effect that was assumed

The period's most-cited justification for aggressive spending was the network effect, and it was invoked far more often than it applied. The concept is precise and the usage was not.

A genuine network effect means the product becomes more valuable to each user as more users join. A communications network is the clean case: a service with two users is nearly useless and with a million is nearly essential, and the change comes from the users themselves, not from anything the company did.

What was frequently described as a network effect but is not:

• Scale economics. Lower unit costs at volume are a real advantage but a different one — they help the company's margins, not the user's experience. A competitor with capital can match them by buying scale.
• Brand recognition. Being well known helps acquisition. It does not make the product better for existing users, and it is purchasable through advertising.
• A large content library. More inventory is genuinely more useful, but if the content is licensed rather than user-generated, a competitor can license it too. The advantage lasts as long as the exclusivity does.
• Data advantages. More usage producing better recommendations is real, but the returns diminish quickly — the difference between a hundred thousand and a million data points is usually far larger than between ten million and a hundred million.

The distinguishing test is whether a competitor with equal capital could replicate the position. If yes, the advantage is a head start rather than a moat, and the spending buys time rather than exclusivity.

Two structural qualifications that were routinely omitted:

Network effects are frequently local rather than global. A marketplace connecting local buyers and sellers has a network effect within each city — being large in one city confers little advantage in another. A competitor can therefore enter city by city, which is exactly what happened in several of the period's most capital-intensive categories. The company must win each market separately, which is a completely different capital requirement from winning once.

And they saturate. Beyond a threshold, additional users add little — a social network with most of a user's contacts already present gains little from more. After saturation, the effect stops defending the position and the company competes on ordinary product quality.

A network effect is a property of the market's structure, not a property of the company's ambition. It can be checked before the capital is committed, and it rarely was.

The US Venture Capital Report 2014 covers the multi-homing condition specifically, which is the practical test of whether an effect is exclusive or merely present.

What an allocator could act on

Treat private marks as lower bounds on volatility and correlation. Marks rise on financing events and stay flat on bad news, so smooth reported returns are a measurement artefact rather than diversification.

Look through to underlying exposure rather than reported correlation. A private software company and a listed one face the same demand and discount rate; only the measurement frequency differs.

Check the five conditions before accepting growth-over-profit. Honest lifetime value, observed rather than projected retention, non-rising acquisition cost, a financeable payback period, and a durable advantage from being early. All five concern the future, which is why the doctrine survives so long when wrong.

Underwrite late-stage private positions on the median outcome, not the tail. The return distribution is not venture-like, so individual company pricing discipline matters far more than portfolio breadth.

Assume nothing about terms from a headline valuation. Structure and valuation move together, so the most impressive numbers are disproportionately the ones that were purchased with protections.

Ask what a late-stage investment assumes about the exit window. A few-year horizon depends on public market conditions the investor does not control, and that assumption is usually implicit rather than stated.

What 2013 established

• A private valuation is a mark, lacking the multiple-participant, continuous, defined-security and executable properties that make a price informative.
• Marks are asymmetrically sticky, understating volatility and delaying corrections into discrete steps.
• Late-stage private investing became a distinct asset class, closer to public growth equity than to venture capital.
• Growth-over-profitability is conditionally correct, and the conditions are checkable but concern the future.
• A named threshold became a target achievable through structure, making the headline number least reliable when it is most impressive.