Here are some facts about venture capital that might surprise you if you’ve learned some economics:
VC firms’ returns tend to persist over time (Kaplan and Schoar 2004). In other words, as an investor, you could have “alpha” (market-beating returns) by investing with the currently top-performing VC firms. Almost 75% of a VC fund’s rate of return is explained by the performance of their last 2 funds (Cochrane 2001).
The top few VC firms have particularly crazy track records. For instance, in the 2004-2014 decade, while only 14% of venture capitalists participated in a single billion-dollar exit, Sequoia, Greylock and NEA participated in 8 each (CB Insights 2013). The top decile of VC funds has a median return of 200% (Janeway 2010). Combine this with the previous point: you should be able to predictably get a 150% annualized return by investing with someone whose last two funds were top-decile.1
Y Combinator gets an incredible deal on the startups they fund. As of 2015, they’d invested $120k in each of 940 companies (Altman 2015, Altman 2014), for a total of about $120m. In return, they got 7% of these companies, which are now valued in total at $65 billion. Assuming that YC got diluted by 3x, they’ve turned their initial $120m into $1.5b already, a >10x return. (This is probably a huge underestimate of their eventual return, because the larger recent batches, where most of the money went, still have a lot of growing to do—the biggest winners so far are mostly from 2010 or earlier (CB Insights 2015).)
It looks like you might even get alpha by investing in the nominees of the TechCrunch “best startup” award, whose nominees are chosen by… the vote of TechCrunch readers.
It’s hard to get actual hard data on fund returns, so these data points are by necessity somewhat speculative. But I think they’re enough to provide strong evidence that something weird is going on in the market for venture capital—that (a) if you were able to invest in any startup of your choice, at what seems like the market rate, you’d be able to get substantial free money; and (b) the market isn’t “clearing”—the supply of capital is much bigger than the demand for it.
This is confusing!
If you believe in the efficient market hypothesis, this probably sounds pretty weird. Your normal models would predict things like:
If a startup is such an obviously good deal that the average TechCrunch reader knows it, they should be able to bid VCs against each other and increase their valuation until it’s no longer an obvious win.
If a VC firm looks like it’s obviously going to beat the market, they’ll get swamped with investor interest and increase the size of their fund until their return goes down to market rates. (This is what people always claim hedge funds do.)
If Y Combinator’s seed investments look like a slam dunk on average, either (a) their investees would demand better terms, or (b) YC would accept more companies, until it stops looking that way.
Why doesn’t that happen here?
My guess at the explanation has two parts.
First, startups can be predictably undervalued because they care more about optimizing their chance of success than their valuation. That’s because startups are driven by power-law dynamics—the top couple “winners” in a space are likely to be as valuable as everyone else combined—so increasing your chance of being a winner is much more important than anything else.
Here’s an (incomplete) list of things that startups sacrifice their valuation for:
Minimizing distraction. Fundraising is excruciatingly distracting. Adding more players (to start a bidding war to bid up the valuation) makes it 10x worse. I’ve seen startups cut their valuation by 1.5-2x in order to raise from existing, low-maintenance investors instead of having to shop around and manage a bidding war.
Getting good investors. The best investors really do add an enormous amount of value by giving good advice, introducing you to the right people, and so on. This was really unintuitive to me at first—how could a single intro email from a single person really be worth that much? But Wave has this effect all the time, where a single face to face meeting with a certain regulator/bank director/credit card processor saves us weeks or months of setup time. Which explains why investors like Ron Conway would be worth leaving quite a lot of money on the table for.
Minimizing risk of a down round. Down rounds (raising at a valuation below the valuation of your last round) put startups in a disproportionately bad position: they dilute common stock by a lot, hurt employees’ morale, and generally involve raising from crappy adversarial investors. This is important enough that Sam Altman calls it “the prime directive of fundraising strategy”:
The badness of a down round is difficult to overstate; in fact, the threat of that is the best reason not to take a super high price when you’re offered one. If you raise at such a price, everything has to go perfectly in order for your next round to be an up one.
But even if we can explain why some startups are predictably undervalued, that’s not sufficient to explain why some VCs predictably earn excess returns. Why don’t the best venture capitalists expand their funds until their average return isn’t market-beating? Here’s one theory:
Since startups don’t care that much about valuation, VCs have to get startups to want them to invest in other ways: for instance, good advice, connections, or not distracting the founders very much.
These things are really hard to evaluate, so they mostly get rolled up into how good of a “brand” a venture capitalist has. Startups judge a VC’s brand by (a) how famous the partners are, and (b) how successful are the other startups they invested in.2
- That means a good venture capital brand doesn’t scale. If a good VC makes their fund bigger, it decreases the average quality of their companies and partners, which lowers their social capital and makes it harder for them to compete for the top deals.3
Y Combinator is the only organization to have plausibly solved the scaling problem. My guess is that they “solved” it by completely monopolizing their stage of the fundraising pipeline. If you’re raising a Series A there are lots of equally super-famous investors to pick from, so if one of them dilutes their brand you can go to someone else. Meanwhile, if you’re looking for an accelerator, your second best choice can’t tell “literally” from “literately”.4
Where’s my free money?
Normally, when an asset class is inefficiently priced, you can buy it and make free money. But in this case, you can’t. You can’t invest in the best startups because they’re already snapped up by the best VCs, and you can’t invest in the best VCs because they can’t increase their fund size without diluting their brand. In Yudkowskian terms (Yudkowsky 2017), even though the market is inefficient, it’s not exploitable (unless you’re a top-decile VC firm).5 Too bad for you!
Altman 2014: The New Deal
Altman 2015: YC Stats
CB Insights 2013: The Exceedingly Rare Unicorn VC
CB Insights 2015: The 13 Y Combinator Startups Worth Over $50 Billion
Cochrane 2001: How High are VC Returns?
Kaplan and Schoar 2004: Private Equity Performance: Returns, Persistence and Capital Flows
Mowshowitz 2015: In a world… of venture capital
Wikipedia: Sequoia Capital
Yudkowsky 2017: Inadequate Equilibria: Where and How Civilizations Get Stuck
Technically, for this to be true it requires some additional assumptions about the shape of the joint distribution of (past performance, current performance). ↩︎
The brand is also its own asset, separate from whatever value-adds it’s based on, because the best-brand VCs have their pick of startups, and so if they invest it sends a strong signal that the startup is likely to succeed. See Mowshowitz 2015 for a more cynical take that this is most of the reason for the market inefficiency. ↩︎
The brand theory also explains why so many actors and sports people become venture capitalists. It’s not that they have good business sense. Instead, merely by being famous, they become a Schelling point that the best companies and advisers coalesce around. ↩︎
Actually, startup equity is Yudkowsky’s first example of an inefficient-but-inexploitable market, although I’m demonstrating a different type of inefficiency (predictable price increases), and for different reasons (VCs not competing on price). ↩︎