I keep coming across interesting articles, and I keep emailing them to myself to read later. I haven't found a tool I like that allows me to save these articles and jot down notes. My hope for the new year is that posts here will both unburden my inbox and force me to read/write/think a bit more. Here goes ...
I'm still trying to wrap my head around the details of the math in this, but this paper seems awesome at first glance -- a model for VC returns that is intuitive and replicable. The general ideas and my takeaways:
- Each round of a VC investment is just an option on the company
- A start-up's full life cycle can be modeled as a compound option. After the initial investment: do we continue with a follow-on or liquidate? And if we continue: if the start-up is valuable, IPO; if not, do a M&A (at a discount).
- In the past few decades, "VC" really means tech, so we can model VC exposure as a levered NASDAQ-100 bet, with ~2.2x beta in the first two years of the start-up, ~1.6x beta in the next 2 years, and a 1.4x beta thereafter.
- If you use the levered NASDAQ model to replicate VC cash flows, venture funds in aggregate have underperformed this since 2000
- It's not clear if this is because the NASDAQ has consistently overperformed, or if VC funds have underperformed, or if the start-ups' beta has decreased since 2000, or something else entirely
These results seem simple and interesting enough to try to replicate -- I hope to dig in more. (For example: I'm not sure how much this would differ from, say, a simple 2x levered NASDAQ long with a "pacing plan" for contributions and withdrawals.) I think it would ultimately be cool to be able to take this same model and reverse it: given the valuation of the NASDAQ, would this options-based VC model say it's a poor time to allocate to tech VC? Of course,
all models are wrong, but it'd still be interesting to see how the numbers play out. I'd also love to see if this model could break out other types of VC funds (e.g. biotech) with the same results. It follows one of the trends that seems to be playing out: the lines between asset classes are blurring, and so is it possible to get a similar exposure to biotech VC by just using a levered, paced biotech index?
ECON 252 - Financial Markets - Lecture 6: David Swensen (Feb 2, 2011)
Old but good, almost a brief restatement of Swensen's big ideas. (1) An allocator's tools are asset allocation, market timing, and security selection; asset allocation is far and away the most important. (2) Swensen uses the gap between top and bottom quartile managers as a proxy for market efficiency (although to me, it seems like a poor proxy -- more below*). (3) Asset allocation should be at one extreme or the other -- very active (a la Pioneering Portfolio Management) or largely passive (a la Unconventional Success). (4) Measures of "risk" are still insufficient in the funds management world.
*High variance in funds alone seem like a poor measurement: (a) venture and private equity firms typically hold fewer assets (higher idiosyncratic risk) and (b) it's cheaper and easier to spin up a venture capital fund, meaning more funds that shouldn't exist are allowed to.
- For (a): a good experiment could be to compare private equity firms' variance to 10-stock portfolios (equal-weighted) with valuation marked quarterly -- would these portfolios show large interquartile variation?
- For (b): this is harder to account for -- perhaps re-do the measurement only with funds that are not the VC's first fund, to try to weed out the shooting stars? (This obviously introduces a survival bias, though ...)
I found this paper in the book Abundance (2025) by Ezra Klein and Derek Thompson. Background: to get grand funding, academic researchers can apply to the NIH (large but risk-averse, $28.4 billion budget in 2007) and Howard Hughes Medical Institute (HHMI, and more willing to bet on “people,
not projects"). HHMI's mandate “urges its researchers to take risks, to explore unproven avenues, to embrace the unknown – even if it means uncertainty or the chance
of failure"; they do this by offering more research freedom, quicker review times (6 weeks), a shorter application, better feedback, and more. They find that HHMI's program in aggregate rewards longer-term success and leads to more "breakthrough" innovation (e.g. top percentile papers). To be fair, the NIH is federally funded and so any "errant" research is bound to be lambasted in the halls of Congress as "frivolous, fraudulent research," a waste of taxpayer money ... so politics definitely drives the NIH's risk aversion.
The crossover from moonshot academic research (this paper) to moonshot start-ups (i.e. VC) is apt. Perhaps in this comparison, the small-business loans administration (SBA) is the NIH, and VCs as a whole are HHMI. My most immediate thought, though, is what VCs are willing to fund: the industry seems to coalesce around certain themes or trends, then fund them to death. (Today, it's "agentic AI.") Perhaps this truly is the next big thing, but these bets feel increasingly risk-averse -- more NIH in nature -- and perhaps less "innovative" for both start-up and VC. (VC is tough, though: early consensus is great, building consensus is good, but late consensus is bad!) My parting question: the VC industry has gotten so large that it begs the question, is VC consensus driving what founders are building, or are great founders driving what VCs invest in?
On the docket:
- Dive into the numbers for Yale PE's returns, based on Ludo Phalippou's research and recent post
No comments:
Post a Comment