I was having breakfast this morning from someone I know from the LP world. We were talking about what’s “working” in the broader VC ecosystem and this person expressed their enthusiasm that today there are a range of different models that are producing attractive returns for LPs.
Historically there was just one model… VCs had to invest in the Series A or B round of a startup, own 20%+ of the company, and be one of the primary capital supporters of it through exit. That model still works for quite a few firms, but now there are also some different models producing attractive returns. One example is seed/micro VCs who own far less than 20% (often <10%) of companies but are typically investing at the earliest stages and out of smaller funds, both of which can produce superior returns on a cash on cash multiple if done well.
There’s also select late stage VC investors, either as a dedicated late stage fund or as part of a broadly diversified large fund, again who often own far less than 10% of the companies they invest in but where they can still produce good multiples in shorter time frame again if executed well. Neither of these models is brand new, but they have emerged more distinctly in the last 5-7 years meaning there’s now realized returns to support the attractiveness of these seed or late-stage models in addition to the “traditional” own 20% early stage model.
But while there’s multiple models working in the VC ecosystem, it remains acutely true that there is no subprime model for VC. I’m fond of saying this phrase though I must give credit to Peter Thiel who coined it back in the early PayPal days 10+ years ago. The point is that unlike various consumer credit markets where subprime lending does indeed work, trying to apply a comparable model to VC is typically disastrous because startup outcomes follow a power-law distribution rather than a normal distribution.
In consumer credit (credit cards, mortgages, car loans, etc) a subprime approach works. You could argue some forms of subprime consumer credit are predatory or usurious in nature (e.g. payday lending). But while lending to borrowers who are less desirable (from a creditworthiness and ability to repay standpoint) will result in greater loan losses than lending to “prime” customers, as long as you price this credit accordingly (higher interest rates, fees, etc) and do a good job underwriting and managing the portfolio it can be a rather profitable enterprise. There are both specialty lenders and large banks that have built big, successful businesses in subprime lending. [see Note 1 below re: subprime crisis of 2008-9]
So why doesn’t the same model work in VC? The subprime analogy for venture would be to invest in the somewhat less desirable startups (in terms of team, mkt oppty, product, etc) but to price one’s investment capital appropriately. Put plainly, the subprime model of VC would be to invest in the mediocre startups at a low valuation. Or to focus solely on getting the lowest valuation as an investor, which by definition will price you out of the most desirable companies which typically have substantial demand for their equity. Or to invest in mediocre startups at a low valuation but hope to “manage” this portfolio to better than expected outcomes.
In theory this would work if the ok, not great startups had good, not great positive outcomes. But we all know that is in fact false… the power law distribution holds. The great startups produce 10X+ returns for their VC investors, the mediocre ones struggle to return capital or minimize losses. Even if you get a super low valuation, investing in a company that raises $20M in VC capital and sells for around that much doesn’t produce a good return. That’s what the middle of the startup outcome distribution tends to look like.
So if there is no subprime model for VC, is the corollary that valuation doesn’t matter or as some have said there are only 5-10 startups per year that “matter”? Well in a way investing in the best companies is in fact more important than valuation, though valuation still matters in certain respects as I’ve discussed before. And the answer to the second corollary in fact depends on the definition of what “matters”.
I can’t speak for Marc Andreessen but I suspect when he said only 5-10 startups a year matter, he was looking at it through the lens of a very large “platform” VC fund like AH or others that are several billion dollars in size. Indeed for groups like these to produce a good fund level return, they by definition must be investing in the companies that are ultimately worth $5-10B or more in enterprise value and there are only perhaps 5-10 of those per year (maybe less). But for a traditional early-stage VC with a $200-300M fund a single exit of $500M-1B really “matters” in terms of returning a huge portion of the fund. For a seed VC fund it might be even smaller depending on size and strategy. So “matters” is relative and while there aren’t hundreds of startup outcomes per year that “matter” there are probably many dozens. But there still is no subprime model that works in venture…
Note 1: Readers may feel that the subprime mortgage collapse which led to the financial crisis of 2008-2009 implies that subprime lending does not indeed work. The subprime mortgage collapse had many interrelated factors at work – easy monetary policy, lax underwriting standards, misleading credit ratings, fraud (at all levels from consumer to originators to securitization), uninformed investors buying MBSs & CDOs – all of which became self-reinforcing in a vicious cycle to inflate a credit bubble. This produced an obviously bad outcome. But the fundamental premise of subprime lending (lend to the less creditworthy but demanding higher interest rates to compensate for defaults) remains sound and is working today, just as the bursting of the Dutch tulip bubble doesn’t mean that tulips are valueless.