If you think talking a venture capital firm into funding your startup is hard, try getting one to share its secrets with you.
That’s the challenge Stanford Graduate School of Business finance professor Ilya Strebulaev took on when he founded the Stanford Venture Capital Initiative, which has been steadily amassing a deep and unprecedented database designed to figure out how the VC world really works.
Strebulaev and his co-researchers have already mined two high-profile papers out of the data. The first, “How Do Venture Capitalists Make Decisions?” was almost anthropological in nature, based on surveys answered by some 900 professionals at more than 650 different VC firms. It found that the most important factor driving VC investment decisions was not the potential of the product being pitched but the quality of the team behind it.
The second study raised eyebrows when Strebulaev’s team discovered that VC-backed startups with valuations over $1 billion — so-called “unicorns” — were uniformly reporting valuations well above their true market value. The paper, “Squaring Venture Capital Valuations with Reality,” analyzed 135 unicorns founded after 1994 and concluded that every one of them was overvalued, some by more than 100%.
Although gathering the data for such research has been a challenge, Strebulaev says VCs are becoming increasingly willing to help the project, and he and his team hope to produce more groundbreaking studies soon. “This is just a start,” he says.
Stanford Insights recently sat down with Strebulaev to find out what he’s already learned and what he hopes to learn about an investment sector that continues to have a disproportionate impact on innovation worldwide.
Why study venture capital firms?
The VC world is interesting because the truth is that it’s very small in terms of available funding. One large pension fund or sovereign fund is bigger than the whole VC industry, but its relative impact, of course, is huge. Even though it is such a small industry, of the 1,300 or so companies that became public in the U.S. over the past four decades, 40% were backed by venture capital and they accounted for 82% of the research and development expenditures by all those 1,300 firms. Of the 10 that went on to become the biggest, as measured by market capitalization, eight began with VC funding.
And yet we know very little about how VCs make decisions, or how the economics of their funds really works. What are the best contracts to incentivize entrepreneurs? What’s the best way to add value to these fledgling firms? We have a huge list of unanswered questions.
Is that because VCs are notoriously secretive?
There are a lot of secrets, yes, but that’s a generic problem of private enterprise, not just VCs. They’re not required to file many documents and make them available in a way that public companies have to do, so there’s just not enough data.
How did the effort begin?
It started about three or four years ago. I was teaching the Venture Capital class, which turned out to be very popular, and I was actively engaged in researching venture capital. I talked to the Stanford GSB dean at the time, Garth Saloner, and we created the Stanford Venture Capital Data Initiative [recently renamed the Stanford Venture Capital Initiative]. We began by approaching the National Venture Capital Association, some alumni, and other people in the VC industry. We got a lot of support, and the data actually started coming through.
What kind of data?
One of the most important data sets we have is thousands of contracts between VC firms and the companies they invest in. It’s the paperwork that basically carves out the relationship between shareholders. Some of it is publicly available via the articles of incorporation that every company has to file, but for the most part it’s very difficult to get ahold of this stuff. Once you get the contracts, that’s when the real work begins, because they’re hard to read. Each contract might have hundreds of variables that have never been collected in a consistent manner.
How many have you collected?
The total number is in the tens of thousands, but so far we’ve analyzed a little less than 1,000 contracts. We had to build a whole infrastructure with lawyers, data scientists, and dozens of research assistants who help us read them.
Is it hard to create apples-to-apples comparisons?
It’s very, very difficult. There is no standardized legal language, because each one is basically the result of much negotiation and bargaining between the contractual parties. You see a lot of stuff that’s unique to a specific contract. But once you’re able to link the contracts through various data sets, it begins to get interesting. An important example is that we were able to use it to determine values of existing companies that had been backed by venture capital.
This is the unicorn study.
Yes. It consisted of two parts. The first was the framework that we developed to value these private companies. But the second part was getting dirty in the data, reading every single contract very carefully and understanding the implications for cash-flow rights and preferences of various shareholders — basically, who is going to get what in any eventual outcome, whether it’s liquidation or a sale or an IPO. And that took a lot of effort. A lot of effort.
I imagine some people weren’t happy with your conclusions.
Absolutely. If I say that Company X is overvalued by 100%, people at that company are not pleased. I heard from some of their general counsels.
Was that worrisome?
No. I’m very confident in the framework we developed, and I’m confident that what we did was right. I replied to every communication and welcomed them to give us all the data about their company, because there could be some private documents that we haven’t seen that might affect our estimate of value.
If we’re inaccurate, help us become accurate.
Did that work?
One company provided some further information that elucidated their contract. In all the other cases, we haven’t received any follow-up information, which suggests that they agreed with the way we read and interpreted their contracts.
What audience do you have in mind when you’re deciding what kind research to perform on the data?
We have four audiences in mind. The first one, obviously, is students — our students here at Stanford GSB and students around the world — who are just learning how to become VCs, how to become entrepreneurs, and how to become investors in innovation more generally. The contracts that founders and VCs sign with each other are very important and it is truly critical for everybody to understand the economics of what is going on there. Similarly, the contracts that investors sign with the fund managers drives the economics and returns of those funds. The second audience is academics who are trying to understand this world of innovation and venture capital. The third audience is practitioners — those who are already VCs, already limited partners, already investors in VC funds, already corporate executives. There’s a lot of value in showing them best practices and how to improve.
And then the fourth audience is policymakers. There’s a lot of misconception among policymakers, both here and around the world, about what VCs do and what innovation really is. People in Washington need to understand the difference between an entrepreneur who opens a laundry shop in Missouri and an entrepreneur who launches a tech startup in Silicon Valley. They face very different kinds of risks and have very different potential impacts. I don’t mean in any way to demean the entrepreneurs opening laundry shops. Entrepreneurship in general is really important. But companies that are funded by VCs have much more potential to impact the entire economy and millions of lives, and I think it’s important for policymakers to understand that and also to appreciate that these startups and the entire innovation ecosystem required a different approach.
Is there any data out there that you wish you had access to that you haven’t been able to get?
[Laughs] Yes. A lot. If any of your readers have access and are ready to share data, we will be very happy to receive it. We already have access to a lot of confidential information that we get under NDAs, so we’re very well positioned to work with anonymized data. It shouldn’t be a problem. The good thing about being an academic is that people understand that we’re doing this for the benefit of the community and that, at the end of the day, science is about finding the truth.
Where are the specific data gaps?
For one, we don’t have good data on the employment contracts of people who work for firms funded by venture capital. What are the vesting agreements and how they are structured for various stages and various firms? How are the employment agreements structured and what is their economics? We don’t really have a good grasp on that yet. That would be number one.
Second, we would love to work more with limited partners — the funders of these funds, essentially — to understand better how they choose which VC funds to invest in.
So you want information from people at opposite sides of the spectrum.
Exactly. We actually are starting to have a good grasp of what goes on in between, but less so of those two ends.
It seems that the measurements used to determine success in the venture capital world all have to do with generating wealth. Are there other metrics that you think might be more important?
This is just another economic industry, so the measures of success are really the same as in any human endeavor. From the finance point of view, it’s about generating value and wealth, but it’s also about fostering innovation and generating employment. Do VCs care that much about employment per se? Probably not, because above all they have fiduciary duties to their investors. But as a byproduct, they are generating innovation and employment. And at the end of the day, their products can make life easier and better for consumers — or at least different.
Most people don’t realize that without venture capital, we would have never had iPhones, because Apple was backed by VCs. We would have never had computers, because the semiconductor industry was backed by VCs. We would have never had search engines, and so on and so forth. At the very least, one can reasonably claim that it would not have happened in such a short period of time.
Some of us might not like all of the innovations, but there’s no question that this industry has had an outsized and underappreciated impact on the economy and on humanity. That’s why I’m so excited to study it.