The WSJ wrote a story today on venture capital returns that makes for great headlines but misses the “mark” (pun intended) on how venture capital performance actually works. That’s an understandable outcome from a reputable source like the Journal given how arcane and obtuse venture capital valuations can be; we’ll try to demystify that in this post.
First, the headline — “Andreessen Horowitz’s Returns Trail Venture-Capital Elite” — is wrong because “marks” are not “returns” in venture capital; only real, actual cash and stock distributions matter.
The value of what VCs give to their limited partners (LPs) is deterministic — cash and stock that can be converted into cash — not based on some theoretical model about hypothetical returns in 7-10 years (so-called “marks”) that cannot accurately be compared across venture capital firms. Simply put, you can’t game cash.
A “mark” is a point in time accounting estimate that often varies by firm and does nothing to tell an LP what a company’s ultimate value at exit will be. Cash or stock actually realized and distributed to LPs is the only real, non-manipulable measure of a firm’s interim success.
The Journal published its data on realized and unrealized returns, but conflated the two and conveniently chose the unrealized metric to match the headline. Had the Journal measured progress based on realized distributions to LPs, the headline would have read “match venture-capital elite” versus “trail.” But that didn’t fit the narrative.
Let’s see how we got here.
Venture capital firms and other financial companies are required under Generally Accepted Accounting Principles (GAAP) to “mark to market” the value of their underlying company holdings on a quarterly basis. But, unlike a hedge fund, for example — where the mark is based on the actual, marketable value of a public security — venture capital marks are highly variable based on different valuation methods prescribed by different accounting firms, and on a venture capital firm’s qualitative assessment of the likely future prospects for that business. That means for every company in our portfolio, there is likely another venture capital firm invested in the same company that “marks it” at a different valuation.
These are the main “mark” valuation methods used by venture firms:
Last round valuation/ waterfall
Some venture firms value their companies by taking the last round company valuation in the private market and assigning that value to the firm’s ownership in that company. For example, if a firm owned 10% of a company and the last round valuation was $200 million, a firm that utilized the last round/waterfall method would report the value of its holdings as $20 million (10%*$200 million).
Comparable company analysis
Other venture firms, particularly for companies that have substantive revenue and/or profits, will use a public-company comparables analysis. In this method, the venture firm will devise a set of public “comparables” — companies that have a similar business model or are in a similar industry — and pick a valuation metric (often a revenue multiple) to reflect how the broader public market values this set of companies. That metric is then assigned to the portfolio company’s financials.
For example, if a portfolio company is generating $100 million of revenue and its “comparable” set of companies are valued in the public markets at 5x revenue, a venture firm would then value the company at $500 million ($100 million*5x) and reflect its holdings based on whatever percentage of the company the firm owned. If, as in the above example, the firm owned 10% of the company, it would value its holdings at $50 million ($500 million*10%).
Often, a firm will then also apply what’s known as a “DLOM” (a discount for lack of marketability) to reduce the carrying value of the company. Basically, this discount says that because the stock is private and can’t be freely traded, it’s worth less than that set of comparable public companies. DLOMs of at least 20-30% are common; in this case, a DLOM of 30% would reduce the carrying value to $35 million ($50 million*(1-30%)).
Option pricing model (OPM)
This tool in the valuation toolset is the most complicated mathematically and the newest addition to the venture capital valuation ranks — in fact, we use a third-party software tool to help us utilize it.
OPM uses Black-Scholes to value a portfolio company as a set of “call options” whose “strike prices” are the different valuation points at which employee options and preferred shares all convert into common stock. Perfectly clear, right? Of course — even though OPM valuations may or may not relate at all to waterfall or comparable company valuations!
Here’s a quick example: If our hypothetical company raises a Series C financing at $5.00 per share, the OPM says that all we know for certain is that anyone who holds Series C shares should value them at $5 … simple enough. But, if you own a Series B or a Series A share, the OPM says those are worth some fraction of $5 — why? Well, to really answer that question you’d need to have a Nobel Prize in economics (like Myron Scholes, the co-inventor of Black Scholes), but the not-too-mathematical answer the OPM generates is that those Series A or B shares could be worth many different values based upon a series of probabilistic outcomes if/when the company ultimately gets sold or goes public. So OPM will assign a value to the Series A and B that is a substantial discount to the $5 per share price assigned to the Series C. And then to calculate the value of the whole company, it adds these up.
OK, so then what?
In addition to the different methods above, many firms may combine them and assign different weights to the different methods, ending up with a compatibility matrix of valuation methods that is longer than a typical Chinese restaurant menu.
Let’s look at an example to see how all this impacts marks.
We’re privileged to be investors in Stewart Butterfield’s Slack, along with a number of other prominent investors. It’s been publicly reported that Slack raised money earlier this year at a $3.8 billion valuation, so let’s assume that to be the case for this purpose. Let’s also assume, only for the sake of illustration, that we own about 10% of the company and invested $10 million for that stake.
How could we then “mark” Slack based on the differing methods described above?
- Last round valuation/waterfall — under this method, the value of Slack would be $380 million (10%*$3.8 billion). So, on an investment of $10M that is a 38x hypothetical return.
- Comparable company analysis — this one is tough given the fact that Slack (and its growth rate) are not really “comparable” to any set of public companies. But, again using only public data, it’s been reported that Slack had annual recurring revenue (ARR) of $64 million at the end of 2015. To make the calculations simple here, assume that Slack is growing at 100%, so investors making an investment in 2016 would be forecasting an ARR of $130+ million for 2016. Looking at high growth multiples in the public markets, investors might assign a multiple of 10-15x on that ARR, resulting in a $1.3–$2 billion value for the company. If we took the high end of this range to mark Slack, we would hold it at $200 million (10%*$2 billion), or a hypothetical return of 20x — which is good, but not as great as the 38x above. And that’s without assigning a DLOM to it.
- OPM — You’ll have to trust me on this one, since the math does not easily lend itself to a paragraph summary (recall that we purchased a software tool to help us)! But, an OPM with reasonable assumptions on time to exit and volatility would yield about a 16x mark for Slack. Very different from 38 and even 20.
OPM, by the way, is the method we use.
So, by the method the Journal used for comparisons, our “results” on Slack could be nearly 3x worse than another venture capital firm who made the exact same investment at the exact same time, simply because of a different accounting methodology.
So, which accounting methodology is right?
They are all theoretically “right” in that different accounting firms would likely sign off on each of these as consistent with GAAP. But, at the same time, they are all “wrong” in that none actually tells LPs anything about what a company like Slack will ultimately be worth to the fund when it ultimately goes public or gets sold and the proceeds of those events are distributed back to LPs.
Here’s some more context to really understand VC valuations:
First, this is such a significant issue in our world that at least half of the conversations we have with our LPs is comparing our unrealized “marks” to those of other venture firms who have the same company in their portfolio. Since we primarily use the OPM model, our marks are deliberately more conservative and — according to our LPs — are lower than other firms who use different methods.
None of the benchmarking firms in the industry adjust valuations based on the differing methods used by firms, so any comparison of results is inherently apples-to-oranges.
Second, “marks” are just that: They reflect theoretical, unrealized gains as of a single point in time. As one of our LPs consistently says, “I can’t spend unrealized gains.” The only thing that matters in this business are actual, realized, and distributed returns to LPs — which is why Sequoia Capital, one of the most respected venture capital firms, specifically tells its LPs to ignore unrealized gains and focus only on the cost of the investments and the actual gains distributed to LPs.
Third, venture capital is a business with a long gestation period, more aptly depicted by what is known as the “J curve.” That is, returns in the early years of a fund are decidedly negative — because as a firm you are investing money (and thus calling capital from your LPs) in young, immature companies that may not get sold or go public for 10 years from the time of initial investment. In the harvesting phase of a venture capital fund (generally years 7+), cash flow to LPs in the form of realized distributions should be more than positive to make up for all of the early years’ investments. As the time to exit for many private companies continues to elongate – in many cases to 10+ years — the inflection point on the J curve has and may continue to push even further out.
To its credit, the WSJ correctly notes that Andreessen Horowitz just raised (in June of this year) another venture capital fund, but fails to connect that with its conclusions on the performance of our funds. The very data that the Journal reports to conclude that the firm is not performing is the exact same data that all of our existing LPs reviewed in connection with their decision to continue investing with us in our new $1.7B fund. The investment behavior of seasoned institutional investors is probably the best gauge of whether the firm is on or off track, at least until the funds are mature enough to be measured by realized returns.
Venture capital is a long game. Calling the outcome on funds that are less than 5 years old, where many of the portfolio companies haven’t even shipped their products, based on non-uniform, point-in-time valuation metrics, is like reporting the Superbowl winner based on the result of pre-season NFL games or even training camp scrimmages. At best, it’s off the mark.