Is VC an Old Boys’ Club? Insights From Co-Investments in ~70K Transactions Between 2010 and 2020
In this blog post, I explore ~70,000 VC transactions between 2010 and 2020, focusing on co-investments. Specifically, I explain which VCs co-invest the most with each other, which pairs of VCs have the largest amount of co-investments, and the fraction of successful investments for a given pair of VC co-investors. To understand the data source and process, feel free to read the first section. Otherwise, skip over to subsequent sections to see the uncovered insights. I would like to thank Hiba Habbat, student at African Leadership Academy, for her valuable help in retrieving and cleaning the data.
Understanding the Data
The data source
I first download all Venture Capital transactions data from Crunchbase between 2010 and 2020. This includes all continents and industries. (I use a Crunchbase Pro subscription, which allows me to retrieve detailed information about each transaction)
A transaction is defined as a fundraising round by a company. The data further contains additional information, such as the funding round (e.g., seed, series A), the amount of funding raised, the company’s valuation, the industries where the company operates, the list of investors who participated in the round, and the funding status of the company.
It’s important to take some time to explain what the funding status represents, as it will be important in later stages of this analysis: the funding stage explains the current funding stage of a given company, regardless of what round the company was raising. For instance, let’s look at Airbnb. In this dataset, there are 8 transactions related to Airbnb (8 rows each with multiple columns), from Airbnb’s Series A in 2010 to its Series F in 2017. However, for each of these 8 transactions, the funding status variable shows “IPO” (this is because Airbnb launched its IPO in December 2020). The funding status variable has multiple inputs, but can largely be thought as representing two categories:
- Late Stage Venture: companies at the following stages — IPO, post-IPO, M&A and Series C & above
- Early Stage Venture: companies at the following stages — seed, Series A, Series B
Why is the funding stage variable important? By my own definition, this is what I use to identify successful investments: those investments in startups that “made it” past-Series B are therefore large enough. The use of this variable will be clearer later on. (It’s worth noting that some transactions’ funding types are missing from the dataset, which presents a limitation of the analysis.)
Some summary statistics
This dataset contains ~200,000 transactions spanning ~119,000 companies. In Figure 1, we can see that the number of transactions has steadily increased between 2010 and 2020, reaching a peak growth rate of ~220% (from base year 2010) in 2018, before declining in 2019 and 2020.
The dataset contains ~119,000 companies (start-ups) that raised money between 2010 and 2020. As figure 2 below shows, only ~14% of startups become “Late Stage Companies” (companies that have raised rounds C & above). Another 62% are labeled as “Early Stage”, while the rest are not labeled in the data.
Of the ~200,000 transactions in the dataset, the vast majority (72%) is categorized as an early round transaction (Seed, Series A, Series B), and ~6% is categorized as a late round transaction. The rest is missing from the dataset.
Now that we have an idea about the transactions and companies represented in the dataset, let’s turn our attention to the main focus of this blog post: co-investments between VCs in the same transaction.
About 37% of the transactions are completed by multiple investors
The first notable insight is that about the majority of transactions (~63%) have only one investor. This is expected, because as we’ve seen above, the majority of transactions are early round (and typically, only one investor will take a bet on a new start-up at an early stage).
Because I am mainly interested in co-investments, I disregard all transactions that have only one investor, and I focus the next set of analysis on transactions with multiple investors.
Understanding the investors in the dataset
The dataset contains ~50,000 investors. To make sense of these investors, I categorize them by size:
- Large: Having executed more than 1,000 transactions per year, on average, between 2010 and 2020
- Medium: Having executed 500–1000 transactions per year, on average, between 2010 and 2020
- Small: Having executed 300 — 500 transactions per year, on average, between 2010 and 2020
- Very Small: Having executed less than 300 transactions per year, on average, between 2010 and 2020
Note that this categorization does not take into account the quality of the transactions, only the quantity. However, one can deduce an investor’s influence and success by looking at the amount of transactions it has executed.
Figure 5 presents summary statistics based on this categorization. We can notice that there are a few large investors (109), and a large number of very small investors.
So which VCs co-invest the most? The next section will give us a view.
VCs with the most co-investors: The top 10
Figure 7 below presents a view of pairs of investors (the top 10) with the largest number of co-investments. Indeed, between 2010 and 2020, FundersClub and Y Combinator have executed 161 transactions together (remember, transaction means they invested in the same round). On average, that means FundersClub and Y Combinator have invested together ~16 times in any given year between 2010 and 2020. The table below summarizes the rest of the co-investment pairs.
Now, let’s look at the VCs with the most co-investors between 2010 and 2020. These are VCs that executed at least one transaction with their fellow VCs. Figure 8 below shows the top 10. GV (Formerly Google Ventures) comes out on top, having co-invested with 1,859 distinct investors. To veterans and followers of this space, this list will not come as a surprise, as all VCs in this list are known to be large, prominent and well-connected in Sillicon Valley.
Just like size, I define a metric to measure the degree of connectedness of any given VC, specifically:
- Well-connected: VCs with more than 500 co-investors in total between 2010 and 2020
- Medium-connected: VCs with 100–500 co-investors
- Low-connected: VCs with less than 100 co-investors
Based on this categorization, only 127 VCs: ~ 0.3% of all VCs are considered well-connected, ~3% are considered medium-connected, and the rest (~96%) is considered low-connected.
Ultra-connected VCs: an exclusive club?
To understand whether the ultra-connected VCs form an exclusive club that tend to invest with each other more often than they do with other players in the field, I construct the heatmap below.
The numbers in the heatmap below show the average number of co-investments between VCs of different sizes: A large VC and a fellow large VC have an average number of co-investment of 6.36, compared to 1.42 between a large VC and a very small VC. That is, a large VC is ~5 times more likely to co-invest with a fellow large VC than with a very small VC.
The above heatmap confirms what is already widely known in the industry: that a relatively small number of firms are large, prominent and mutually reinforce each other.
Which co-investment partnerships are most successful?
Now let’s look at which co-investment partnerships (in pairs of investors) are the most “successful”. By “successful”, I mean a transaction in a company that has subsequently reached a late stage status (and has thus survived the perilous first stages of the startup cycle). To understand how I derive this metric, I refer you to the first section of this blog post.
Figure 10 below presents the top 10 of the most successful pairs of co-investors (only for total transactions > 50). To make sense of the figure, consider the first row: Andreessen Horowitz (a16z) and Greylock together have the highest success rate, at 84%: of the 56 transactions that they have executed together, 47 were in companies that have reached late fundraising rounds.
Another interesting insight is in the second row, where one VC firm has the most track record with its affiliate in Israel. Moreover, we can see that SV Angel does a good job with its co-investments, as it appears in the top 10 3 times.
Bringing it all together: How can this be useful?
On a macro level, this analysis might be eye-opening and might raise questions about the exclusivity of the Venture Capital industry. That is an important topic that should be further explored.
On a micro level (say, at the firm level), the analyses above can be combined in Figure 11: an example of a “Network Map” for co-investments. In the figure below, I choose the most connected VC firm: GV (formerly Google Ventures), and I map the VCs with which GV has executed the most transactions. Then, I present the success rate as defined above. As the figure shows, GV has the most successful investment track record with Accel.
Why is this useful?
At the firm level, I can think of two use cases for the Network Map above:
- At the fundraising stage, VCs can use this data as a marketing tool (hey LP, we intend to co-invest only with our most successful VC partners, and those are Accel, First Round, NEA, etc)
- At the investment stage, VCs can use this data to convince startups (hey startup, when we and Accel invest together, our successful track record is 83%!)
Like what you just read? Want to see the Network Map for your firm? Email me at firstname.lastname@example.org
Thank you for reading all the way! I am always open to feedback and engagement with my posts!