It seems very popular, these days, to complain about private valuations. The debate goes both ways; some argue that companies justify billions of dollars, others argue that valuations are unsustainably high. Nobody so far, though, has examined some of the very concrete financial reasons that venture-backed companies are overvalued. If we examine the actual venture process, it becomes clear — obvious, even — that headline valuations for venture deals are inflated by 25–35%. This isn’t about company fundamentals — revenue, profit, etc. — at all. It’s about the financial processes and instruments at play, and we’re going to demonstrate it.
I’m not going to make any claims about whether companies should be worth anything at all if they don’t have revenue. That’s subjective. I’m not going to make claims that we’re in a bubble. That’s been beaten to death. I’ve heard the argument that venture funds suffer from moral hazard — they make money on carry, which incentivizes them to raise as much as possible and make riskier investments, leading them to deploy into too few companies for too-high valuations. That sounds plausible, but hard to prove or quantify.
The only claim I’m going to make is that almost every venture-backed company’s media valuation is higher than its valuation would be if it traded on the market. This happens because our venture funding system is not analogous to the public markets.
Model Assumptions
To a financial quant, or anybody who prices securities or companies very rigorously, this entire analysis is going to be a little hand-wavy. There are a couple reasons for this. One is that every private venture deal is a little different. Every round has a different number of subscribers and firms trying to get in, and every round has different terms and liquidation preferences and voting powers. However, many deals do share similar traits. The other reason is that there are no perfect analogs for venture terms on the public market. Convertible preferred stock doesn’t typically trade, and voting rights, board seats, etc. aren’t easy to value.
In particular, we’ll be thinking about rounds in which the venture firm did not get to set the valuation (i.e. there was some competition for the round), and where the firm takes a 1x liquidation preference. This means that when the company experiences a liquidity event (sale, dissolution, or bankruptcy), the investors are first in line to get the money they put in back, before paying out anybody else. These are both true of most “unicorn” deals, or deals with valuations over a billion dollar valuation (and in some cases, the liquidation preference is greater than 1 — a liquidation preference of 2, for instance, means that before anybody else gets money, the investor gets 2x their money back). We’re going to assume as well that the investors are not getting participating preferred stock, in which case investors get their money back, and then the equity of what’s left on top of that; such deals are usually capped and have no good public market analogy, so they’re even harder to price.
Finally, because every estimate is going to be done using public market analogies, we should think of the 25–35% number we arrive at as a floor on the premium. That is, private companies are overvalued by at least this much, but possibly more. This is because small startups are likely to have higher volatility, and a wider distribution of perceived prices, which drives the premium up respectively. The 25–35% number assumes that startup stock behaves much like that of public companies — this isn’t true. The differences in behavior mean that we understate the premium here.
Reason 1: Highest Bidder, not Average Price
The valuation of a publicly traded company is, roughly speaking, an average of what all the opinions on the market think it should be. If many people believe AAPL should be above 100 and not many people believe it should be below, then when AAPL is at 100, many more people will buy than sell, moving the price up. The same is true in the opposite direction.
Venture capital rounds, however, are nothing like this. Instead, believers can only take one side (the buy side), while disbelievers are powerless to sell, instead resorting to tweeting that we’re in a bubble. Only a small number of venture firms have to believe that a company is worth billions, and then it is, no matter how many parties disagree.
In addition to that, private companies usually only offer 10–20% of the company equity for “purchase,” or investment. This furthers the problem; only a very small amount of capital (usually a couple hundred million dollars) goes into determining the price of the company, unlike larger publicly traded companies, where that much of the company moves on the order of days.
Let’s run through a quick example. (All numbers as of closing on May 27, 2016.) TWTR currently has an enterprise value of 8.5 billion dollars. If Twitter were to only allow 20% of that to be available for buyers who want to invest, they would have to buy back all but 1.7 billion dollars of shares (about 113 million shares at the last $15.10 closing price) and prevent anybody from taking short positions in the stock. In other words, if Twitter were raising a round, it would be like selling only 113 million shares to sate all of the current market demand.
Currently, there are over 77 million shares of short positions in the market, as well as 700 million shares outstanding, which means people who want to go long Twitter can buy from a supply of 777 million shares of the company. As a best approximation for what might happen if Twitter offered only 113 million shares to current demand, let’s have Twitter buy back everything else. If Twitter were to buy back 664 million shares today (restricting the current supply to 113 million, what they’d offer in a venture round), trading models on market impact suggest that Twitter stock would permanently rise roughly 3.4–4%. [1]
The exact number isn’t perfect, as the model is more conventional trading wisdom than a financial fact, and that wisdom doesn’t necessarily apply to volumes as massive as this. As a result, it’s hard to say exactly how big the premium is based on this supply shortage and inability to short, but 3–4% is a solid rough estimate.
To take another example, let’s look at AAPL. Apple has an enterprise value of around 574 billion dollars, so if it were to offer 20% of that in a (massive) venture round, it would be putting up around 115 billion dollars worth, or about 1.15 billion shares. To get its outstanding long positions to be that “small,” Apple would have to buy back 4.38 billion of the 5.52 billion long positions. The same sublinear extrapolation we used above suggests that this leads to a 3.3–3.9 standard deviation, or a 4.6–5.5% move in stock price.
In Apple’s case, it’s not just the ability to short that keeps prices fair (although an important thing to remember about shorting is that more people will do it as the price rises to a premium), but also that almost all of the company is available, and so any investor who wants in can get in at market price. Contrast this with instances of “bidding wars” by venture capitalists, where being willing to pay more is sometimes necessary to be able to invest at all, and it’s logical that valuations get inflated by a limited supply.
[1]: The model we use for market impact works roughly as follows. Traders believe that the price movement is proportional to the square root of the volume of shares moved, and we’re going to use this assumption. The linked research paper seems to suggest that moving a tenth of a stock’s daily volume permanently moves the price by around 10–12% of the daily volatility. (We want to avoid looking at the temporary shifts because of market shock, where the price moves in the short term as traders “back off” from large volumes in either direction.) 664 million shares is about 28 daily volumes (TWTR trades roughly 23.66 million shares daily). This is 280 times the paper’s experiment, so we conclude the price movement is around 16.7 times the paper’s result, and the sublinear extrapolation follows. The same model applies to AAPL. 4.38 billion shares is 108 daily volumes, which is 1080 times the paper’s experiment. The resulting price movement is 32.9 times the paper’s result, or 3.3–3.9 daily volatilities.
Reason 2: Convertible Preferred Stock is not Common Stock
The other problem that we often treat venture deals as common equity when determining a valuation. Almost every deal is done as convertible preferred stock, which is similar to convertible debt, but without the interest rate.
If Google Ventures and others invest a billion dollars in SpaceX, getting convertible preferred equity that can be is approximately a 10% equity stake, tech media says that Google bought 10% of Uber for 1 billion, giving it a 10 billion dollar valuation.
This isn’t quite true. If SpaceX gets acquired for only 5 billion dollars, Google Ventures stands to get their 1 billion dollars back; they have liquidation preference and thus are first in line to be paid out. Then, the only way they can lose money in event of acquisition is if SpaceX goes under 1 billion and can’t repay in full. If they IPO down, Google Ventures can stand to take a little bit of downside (although there are usually provisions for this as well), but it’s rare for companies to raise large rounds and then IPO down.
It’s very clear that this reduction in risk for the investor implies a share price that is at a premium to the actual share price. In other words, if Google buys convertible preferred stock from SpaceX that implies a valuation of 10 billion, we should think of that valuation as “too high,” and SpaceX’s actual valuation to be lower.
How big is this premium? It’s hard to say, because it’s rare that both convertible preferred stock and common stock trade liquidly. However, there is a similar instrument that is publicly traded or issued: convertible debt. Convertible debt also pays interest, and is guaranteed not to lose money barring bankruptcy, so we expect the premium to be a little bit higher, but it’s a good benchmark.
When Tesla issued convertible debt early in 2014, it did so at a 42.5% “conversion premium.” The conversion premium is how much the implied price of shares by “buying them” via conversion exceeds the actual market price. When Twitter did so later that year, the bonds implied a share price of around 77 dollars, a premium of 48% over the market price for common shares.
Bloomberg claimed that from 2009 to Tesla’s issuance, convertible debt conversion premiums have been, on median, about 30%. We should expect this premium to rise with volatility — option values rise with volatility, after all, and a convertible bond is basically a bond plus a call, and it seems well accepted that tech and growth companies have higher volatilities than the median stock. We should accordingly believe that the convertible debt conversion premium for most venture-backed companies would be north of 35%, and shave off the couple points of interest and a little bit of the downside protection (convertible debt prevents loss in the event of a down IPO, for instance). A premium in the range of 20–30%, depending on how much the extra downside protection is worth, is a solid estimate for the premium we should be attaching to this conversion optionality on venture-backed companies.
Venture backed companies’ headline valuations are at least 25–35% higher than they “should be.” What does this mean?
It makes little sense to compare private to public market valuations. Is SpaceX worth significantly more than Twitter? Well, Twitter has an enterprise value of 8.5B, and SpaceX a private valuation of 12B… but if private valuations are inflated 25–35%, then maybe SpaceX would have a public enterprise value of 9B. Maybe it would be 10B. These things are hard to compare. But what is clear is that the private valuations are higher than they should be.
Interestingly, this should have no bearing on how venture funds behave. If their decision seemed good at a “14 million dollar” valuation, it should be good at a 10 million dollar one, considering that they bought the exact same securities with the same terms in either case. Venture funds shouldn’t really care about what the media will say about “valuation” — they should be worried about the cash flows and the terms and the things that generate returns. Buying ten percent of Facebook for half a million dollars in 2004 is a good idea, whether the media says Facebook’s “headline valuation” at that time is ten dollars or ten billion dollars.
It also says basically nothing about an acquirer’s decision. Acquirers overpay for companies anyways — Valuation for Mergers and Acquisitions suggests that acquisition premiums are usually 20–30% over “market price.” Acquirers look at synergies, fundamentals, and what the target is worth to the acquirer; they also shouldn’t be focused on the headline valuation.
Where these inflated valuations start to mislead people is at the founder and employee level. If founders are adamant that their company is worth 14 million dollars because it’s comparable to some publicly traded 14 million dollar company, and so they agree to 10% for 1.4 million, but the investor gets them to throw in a 1x liquidity preference because it’s “standard,” they need to understand that this no longer is a good deal if they genuinely believe their company is worth 14 million. Terms have financial value. Terms affect your price. If your company was actually worth over 100 million dollars, venture funds would not demand liquidation preference to buy 10% for 10 million — they would just buy the straight equity.
For employees, it can be even worse. I’ve met many young engineers who price their equity based on the last headline valuation (which we just demonstrated to be inflated), and pick jobs and negotiate compensation accordingly. This clearly isn’t correct, and some companies, such as Buffer, are taking steps to educate potential employees on how their equity translates to cash at various exit sizes. More companies should be doing this.
This is not a statement of fundamental overvaluation of tech companies. It’s not a premonition of a crash and fall to come; it’s not a warning that we’re in a bubble and that everybody should get out now. In fact, most of the market factors above will remain true for the foreseeable future, so even after this is common knowledge, I wouldn’t expect the market to correct.
This is a reminder that the headline valuation is not necessarily accurate, and the media should stop dividing round size by equity taken to value companies. It’s a reminder that employees and founders shouldn’t value their common stock at the headline valuation, and that people shouldn’t draw comparisons to companies that don’t experience the same premiums. It’s a reminder that terms have financial value, and valuing private companies is hard and an imperfect art at best. Don’t get attached to your valuation; it’s probably inflated anyways.
Thanks to Chris Barber, Rishi Bedi, Camille Considine, Dennis Xu, and Devon Zuegel for reading drafts of this.
