On Twitter, Robin Hanson asked a question yesterday I have given a lot of thought to, which is why the bigger venture capital firms dominate the market and make oversize profits while the other firms taken together make almost no profits. Most industries do not work that way, so this requires explanation. I attempted to explain my view on Twitter, but was largely unsuccessful. Marc Andreessen did a better job of explaining a fully compatible position, but Twitter still is not the best medium for a question this important and complex, so I am going to consider the question here in more detail. I agree with everything Marc said, but my view of the problem is that it goes deeper, and is more insidious, than his observations imply.
Consider a start-up whose business plan requires venture funding. Most companies that require venture funding need to get such funding, or at least are better off seeking such funding, multiple times; there will be seed money, then a series A, then a series B, and so forth. It would not make sense in the first funding round to give the company enough money to see itself all the way through to stable profitability, since most companies will fail long before that happens and the company is not yet worth enough to take that large an investment.
A goal of every start-up is to avoid having ‘down’ rounds where the valuation of the company declines. These rounds are essentially doom for a Silicon Valley style technology-based startup, as they play havoc with the equity table, reduce the available incentive pie and send the signal that the company is doing poorly and could not avoid sending the signal that the company is doing poorly. Often the company more or less falls apart rather than attempting to soldier on, to free up valuable human capital.
Being a start-up is essentially a giant signaling game played on multiple meta levels; starting a business is not about doing business. Starting a new business is about raising investment. MetaMed is a great example of this phenomenon. We made the deadly mistake of thinking of MetaMed as a business that was trying to turn a profit and grow organically, rather than a start-up whose ‘profits’ would come from selling its stock to investors at a higher price. Even in the explicit context of a profitable case, this was still true. If we had closed an additional $100,000 case and made a marginal profit of $50,000, then we would have added $50,000 to our bank account, which is great, but the valuation of the company likely would have gone up by a million dollars or more and our ability to close investment at all would have gone up substantially. If MetaMed had raised another funding round, it would have sold between 10% and 20% of itself to those investing, which means $100,000 to $200,000 in direct additional cash and another $800,000 of equity to distribute.
The only good reasons to charge your customer money is to test what will cause customers to pay money, and to signal that you can successfully charge your customer money! The purpose of a start-up is to turn itself into a future business, and the potential profits of that future business are what is valuable and what everyone is working to create. It raises money by convincing others of these potential profits, then selling off a portion of those future profits, and then uses that money to enhance its ability to signal its future profits. The system works, to the extent it works, when the signal is both correlated to the company’s ability to succeed (better founders with better ideas can signal more effectively) and the signaling actions themselves serve to create a real company. Founders who understand the dynamics involved will be hill climbing in order to send the best signals possible and raise the most money, so it is very important that their doing so results in actual companies that hopefully do actual valuable things for people.
A venture capitalist is listening to a pitch from a start-up founder. The start-up founder is explaining why his company is exciting and the venture capitalist should invest. What criteria does the venture capitalist use to decide whether to invest?
Fundamentally, the venture capitalist is asking: will this company be able to raise money in future rounds? If the answer is yes, and they are a good fit with the company, then the capitalist will likely invest. If the answer is no, the capitalist will almost certainly not invest. Even if the capitalist believes that the company would, if it was funded, be likely to succeed, the capitalist will refuse to fund it now because it will not be able to get funding in the future. If the founder pitches the company poorly, that has a primary effect that a poor pitch is unlikely to sway the capitalist, but it has an even bigger effect of signaling this founder does not know how to give a good pitch. As a result, in the next round, he will give a poor pitch, thus signaling a poor pitch in the third round as well, and therefore he will not get funded, so I should save my money.
The founders’ potential partners and employees are thinking the same thing. They know the company’s success depends on raising money, so the easier it looks to be to raise money, and the more money has been successfully raised, the more eager everyone is to work with you, to partner with you, and so forth. Nothing succeeds like success, except the expectation of future success. It is often more important to have raised and be able to raise money, and to have raised your money from high-quality sources at the correct prices, than it is to actually be in possession of the money. Peter Thiel’s investment in MetaMed gave us $500,000 to make the company succeed, but the gift of being able to say that he had invested in the company, and on generous terms, was worth far more than the money. If that money had come from a random investor, even if we had gotten substantially more money, we would never have made it half as far as we did.
All the venture capitalists are trying to predict what other venture capitalists will do, and mimic those future decisions. One of their most valuable assets is their reputation for having a strong reputation. If your reputation is that your reputation is strong, then I expect others to expect the companies you invest in to do well, which means those companies will do well. If you instead go against the central tendency of the venture capital market, not only will the companies they invest in find it harder to raise money, but your own reputation will suffer. Others will look and say: Peter invested in a company that could not raise funds from others, so when Peter invests in a future company, I have to worry that he did it for his own quirky reasons rather than because he is predicting the actions of other investors. Peter’s decisions become less correlated with the market’s future actions, and the signal weakens.
When you have a strong meta-reputation, the signal you send by investing is strong, and enhances the value of the company. Everything the companies you invest in have to do becomes easier, and the shared belief that you can enhance the value of the company causes the company’s value to actually become enhanced. That also means that founders beat a path to your door. If one of them has a hot new idea or a strong pitch, they will go to you first, and prefer to take your investment rather than someone else’s investment. In order to get your name and reputation, they will not only take your money first, they will give you better terms, which is pure profit. Then you have them over a barrel, because once you have invested once, your failure to invest again in the future would be a horrible signal, so you can now drive a hard bargain and capture even more of the gains.
The biggest venture capitalists with the strongest reputations get the best founders at the best prices, and their names and reputations are hugely valuable to the companies they invest in, to the extent that people will (quite rationally) flat out give them stock in some cases in order to name-drop them or have them make a few phone calls. Meanwhile, others in venture capital play the game at a huge disadvantage, and on net get very poor returns, possibly negative returns. The good news for them is two-fold. One is that some of them will succeed, build a reputation, and be able to collect these reputational rents themselves. Two is that often they are investing other people’s money, which means that they can earn a fine living without having to get good returns.
The tragedy of all of this is that everyone gets caught in a meta-reputational meta-signaling trap that allocates resources extremely poorly and forces founders to focus solely on activities that can help them raise funds until the point where they have to get ready to approach the actual stock market, and thus need to build a real company. Deviating from this plan gets you punished on multiple meta-levels. The system as a whole does continue to be profitable, because it sucks in huge quantities of high-value human capital both in terms of founders and employees, pays everyone largely out of potential future profits, and hands them large amounts of money and powerful connections. Occasionally one of them build a highly profitable company, and much more often one of them proves that they have managed to hire high-quality human capital and sells that asset to Google or another similar company, since high-quality human capital employees are usually underpaid by orders of magnitude relative to their marginal productivity.
I wish I knew how to shift us away from this equilibrium and get us our flying cars, but for now, we are stuck with a hundred and thirty characters.
If I understand you correctly, your argument is that superstar VCs get all the profits because the signalling value their superstar status bestows on their portfolio companies is a helpful ingredient for the success of such companies due to path-dependent effects on hiring and capital raising.
I think this can be true without implying that the VC world is a ridiculous meta game. Because the frictions in allocating capital are very low, all the best companies will prefer to deal with the #1 VC even if its advantage over other VCs is extremely slight. A tiny inherent advantage will be magnified into a huge difference in outcome… and we see similar “winner take all” effects in lots of industries with similarly low frictions. E.g. the top couple athletes get a huge proportion of the endorsement dollars, the top couple musicians get a huge proportion of concert and recording revenue, the top couple private equity firms get a huge proportion of the assets, etc.
And of course it is possible that the best VCs are genuinely better at picking the companies with inherent potential to become huge and profitable. At least in some cases this will have nothing to do with signalling. An easy extreme case is where only one “round” of fundraising is likely to be needed. Another easy extreme case is when the businesses is immediately massively successful. E.g. Instragram would have worked out about the same irrespective of who their investors were. (And all VCs are afraid of missing out on future Instagrams.)
I agree that the overside profits in no way inherently require the path-dependent effects on hiring and raising capital. If those effects did not exist, but founders still preferred to deal with top VCs due to the top VCs having other perceived advantages, however slight, then the top VCs would still achieve oversize profits, and if the top founders reliably gave them first pick, this effect alone would be quite large. I do not know how big a percentage of the edge would still be there from this effect alone.
However, I also have first hand experience with MetaMed that says that the effects of superstar status are both real and large, as well as second-hand knowledge from talking to others who have experienced such markets. I believe strongly that a large part of the preferential treatment the superstars get comes from the desire to get the superstar halo, and I believe the superstar halo is quite valuable – to the extent that (to continue the concrete example) one can fully explain Peter Thiel’s oversize returns at Founders’ Fund simply through the value his name adds to the companies involved, even if his decision algorithm and pool of investment options were both the same as an average VC’s.
Is it possible that the “best” VCs are better at picking companies? If you define best as most skilled, it is definitely true. Better VC is better! If you define better as having a better reputation or more funds, it is possible, but I would guess that this effect is mostly quite small once you exclude the true suckers at the table. There are some people/firms that wander into VC with no idea what they are doing on a whole different level, and are effectively dead money, and the big VCs can avoid doing that. But my guess is that a combination of the dominance of reputation effects, which discourages top VCs from using their unique judgment, combined with the randomness of the acquisition of big wins and reputation effects, leads to the choice effect being quite small.
In the case of something like an Instagram (I don’t know how they were funded, but presume that they were pre-revenue for a long time so they had multiple rounds), it is possible that the reputation effects I discuss had little effect on the company’s results, but I doubt it; Instagram has strong network effects so the generation of buzz and expected success likely helped its growth curve quite a bit, possibly boxing out competition, and leading to a higher final price. It also certainly enhanced the interim valuations, and therefore the share of the company retained by the founders, and reduced the time they needed to spend in fundraising, allowing them to focus on building up the company.
If a company only needs one round, that is a special case where these effects are much smaller; for a while MetaMed was trying to jump through that hoop even though on reflection such an attempt was massively unwise. This case certainly happens, but is a very small share of the overall VC marketplace.
Just to clarify: The point of the Instragram example was that they built it to a $1B acquisition with only $7.5M of capital raised (a seed and an “A round”, and had extremely visible instant success shortly after launching.
Ah, sure. I did not know that they only needed two rounds before getting acquired; certainly quite fortunate for them. Then again, even now I don’t really “get” instagram as a product.
Pingback: 1p – In a world… of venture capital | Profit Goals
Pingback: 1 – In a world… of venture capital | blog.offeryour.com
Pingback: 1 – In a world… of venture capital | Exploding Ads
Your set up is wrong. Almost all industries work this way, per Jack Welch. Profits are all at the top.
It’s been years since I did any contracting work for Metamed, but to this day I still think back and try to learn from the experience.
Embarrassed to say I did not understand the “down round” problem until watching the show Silicon Valley, and while I think additional problems contributed to the company’s problems, getting $500,000 from the top Venture Capitalist in the valley was an intimidating feat to followup.
Pingback: What Is Rationalist Berkley’s Community Culture? | Don't Worry About the Vase
Pingback: Exploring Premium Mediocrity | Don't Worry About the Vase
Pingback: Expanding Premium Mediocrity | Don't Worry About the Vase
Pingback: Best of Don’t Worry About the Vase | Don't Worry About the Vase
Pingback: Book Review: The Elephant in the Brain | Don't Worry About the Vase
This can’t be entirely correct as there are investment institutions with sufficient capital to offer a complete funding package to a large enough sample of startups to receive close to the expected profit.
Thus, some large fund with lots of money should just go around offering startups with promise that fail out an *exclusive* funding deal (e.g. we fund you on the terms that only we fund you in the future…or only we modulo some tiny fraction used to set the prices of those future rounds). Or they could even just buy a majority stake in the company outright leaving the founders with what amounts to substantial stock options to ensure they retain the same profit motive. The founders should be happy to agree to this when the alternative is likely dissolution (even early on I presume a good fraction of founders realize they are unlikely to play the funding game effectively).
On your theory of resources being deployed very inefficiently this should be a massively profitable approach. However, as you note it doesn’t seem to be.
I think a more likely analysis is that while everything you said is true the fact that there are frequent opportunities to simply sell companies outright means the resource allocation is still decently efficient (obviously not perfect but pretty good and within the bounds one would expect given the informational asymmetries).
I mean sure, playing this game well is important but given everyone knows this anyone who gets involved can also hire or partner with someone who can make these good pitches so while who gets rich may not closely track who offers the most value it’s quite plausible that what ideas get funded doesn’t allocate resources particularly poorly.
Or to make the point more quickly: Just because who gets funded depends largely on these signals that aren’t directly related to fundamentals doesn’t mean they aren’t very good indicators of who will succeed.
I mean SAT questions and elite university diplomas have very little direct relevance to what it will take to succeed in most fields but jumping through those hoops is still very good evidence of future success.
The fact that there is a non-standard financial arrangement that would be a win for both parties does not count as much evidence, to me, of much of anything. Non-standard arrangements like this approximately never happen, even if they are obvious win-win deals. There just are lots of opportunities for someone with the ability to fund companies to make very good investments, given sufficient understanding and willingness to buck social reality. Of course, that doesn’t mean companies in such a spot would have the time and bandwidth to get such information to such funders, were they to exist, which they don’t.
But there seems little question that such deals would often be win-win absent social pressure.
I even have a true story of someone offering, unprompted, an exclusive win-win funding deal, me agreeing to it as offered on the spot, and it (predictably, in slow motion) getting revoked due to social pressure. People just won’t do weird things.
Also note that my model says that the signals do predict who will succeed, because they predict who will be funded, which is success. And yes, these signals do contain real information about who is competent to do things, but that would be true of almost anything if it was known to be the basis of funding decisions. Testing people on arbitrary tasks tells you who can do things.
I come back to this comment a lot, and its one of the most personally depressing things I’ve ever read. Ironically more depressing than breakdowns of existential risk. It’s like people just don’t like to be happy.
Pingback: An Eclectic Mix – Dissent
Pingback: Additional arguments for NIMBY | Don't Worry About the Vase
Pingback: Approval Extraction Advertised as Production | Compass Rose
Pingback: Stripping Away the Protections | Don't Worry About the Vase
Pingback: Create a Full Alternative Stack | Don't Worry About the Vase
Pingback: Mazes Sequence Roundup: Final Thoughts and Paths Forward | Don't Worry About the Vase
Pingback: Can crimes be discussed literally? | Compass Rose
The simplest answer to the initial question: “On Twitter, Robin Hanson asked a question yesterday I have given a lot of thought to, which is why the bigger venture capital firms dominate the market and make oversize profits” is that the largest venture capital firms have a larger, in a nonlinear sense, capacity to identify and invest in the handful of firms that turn into unicorns.
Pingback: On Robin Hanson’s “Social Proof, but of What?” | Don't Worry About the Vase
Pingback: Zvi’s Thoughts on the Survival and Flourishing Fund (SFF) | Don't Worry About the Vase
This is very similar to private equity. Like private equity VC is largely about deal flow i.e. getting access to (lots of) good opportunities – not only b/c you get to pick good deals but you get a better sense of market (i.e. private information). You also have a better chance to raise capital etc in the first place and offer other things to investees.
Form an abstract “model” perspective I’d say this is classic platform markets + some learning by doing effects. The first has massive tendencies to concentration in the market (see ) and the second has some.
If you are interested in market concentration i recommend this masterpiece https://mitpress.mit.edu/books/sunk-costs-and-market-structure which has both rich modelling and lots of data. Note Sutton doesn’t have anything on platforms which, if anything, lead to even greater concentration. If you wonder what we are going to do about this level of concentration – given that fixed/sunk costs and platforms are ubiquitous in an information economy – check out the Open Revolution https://openrevolution.net/
What do you think about what Tiger Global has been doing?
By doing, I mean: https://randle.substack.com/p/playing-different-games
Pingback: The Debtors’ Revolt | Compass Rose
Pingback: Ukraine #3: Decision Theory, Madman Theory and the Mafioso Nature | Don't Worry About the Vase
Pingback: Covid 4/14/22: China Stays the Course | Don't Worry About the Vase
Pingback: Book Review: Talent | Don't Worry About the Vase
Pingback: Are venture capital markets inefficient? (2018) - My Blog