Editor’s note: David Teten is a partner with ff Venture Capital and founder and chairman of Harvard Business School Alumni Angels of Greater New York. Follow him on Twitter @dteten.
I recently had an extensive debate with Tom Grossi, partner with New Enterprise Associates, about why and whether small funds (like mine) tend to outperform large funds (like his). He and I wanted to share the conversation with you.
For some time, the conventional wisdom in the VC industry has been that small funds have better performance profiles than large ones. See the TechCrunch posts by my partner John Frankel and Professor Robert Wiltbank, my recent post on the quality of angel returns data, as well as reports from the Silicon Valley Bank and Kauffman Foundation. An article on the New Enterprise Associates blog by Tom, however, makes an interesting point: most of the megafunds were raised since 1999, a period in which the entire industry did poorly. Though the megafunds did underperform in absolute terms, they may have outperformed in relative terms. While I don’t ultimately agree with the article’s conclusions, I think it warrants consideration.
I agree with Tom on the principle that there is a natural upper limit to fund size before returns tend to suffer, but we are at odds as to where that limit is. I argue that, just as with hedge funds and mutual funds, the larger the venture capital fund, the more difficult it is to generate strong returns. Moreover, VC funds on average earn approximately two-thirds of their revenue from fixed fees.
For larger funds whose partners can earn seven-digit compensation before producing any profits for investors, incentives are not fully aligned. They’re like Bally’s Atlantic City, which The Onion recently featured as having had “‘an unbelievable night‘ Monday”; “Bally’s is in the midst of an impressive winning streak, coming out ahead an astonishing 6,753 nights in a row.” Partners at smaller funds, by contrast, have to hustle before they can cover their mortgage.
Just as with hedge funds and mutual funds, the larger the venture capital fund, the more difficult it is to generate strong returns.
Regardless of whether large funds should or should not be performing better than small ones, the more pertinent question is whether they actually are. Let’s look at each of Tom’s major claims in his article in turn.
He begins by observing that larger funds are drawing a higher percentage of industry capital commitments than was historically the case. He surmises that LPs aren’t buying the argument that large funds don’t perform. Beyond the fact that LP capital commitments don’t prove anything about returns, however, large funds are likely much more resilient to a few bad years than small funds are. The Economics of Private Equity Funds demonstrates that the VC industry survives mostly on fee-based income (of which larger funds have a proportionally larger amount).
Therefore, large funds are beneficiaries of survivorship bias, given the inherent conservatism of institutional LPs. Moreover, LPs have been losing money in VC for years. The Kauffman Foundation points out several reasons why they choose to keep pouring capital into the industry: the J-curve narrative, VC investment allocation mandates (which should disproportionally benefit large funds), the “relationship business” philosophy, and potentially misleading return metrics (such as IRR).
Tom’s next general observation is that although smaller funds have the potential to return much larger multiples, we should only concern ourselves with the performance of the VC asset class as a whole (both from the perspective of LPs as a class and for the purposes of public policy). At a macro level, after all, LPs are definitionally invested in all funds. I think this argument is not relevant for the great majority of investors who only invest in one to five VC funds. Such investors investors are far more impacted by individual investments than industry-wide returns.
Second, I think it absolutely does matter that the best small funds outperform the best large funds. Many researchers have found that past performance is predictive of future performance in the VC industry. I’m obliged to say that sentiment is not universal. Dr. Susan Woodward, formerly Chief Economist of the U.S. Securities and Exchange Commission, and Partner at Sand Hill Econometrics, observed, “Among the scientists who have studied the performance of venture capital and buyouts, there is no consensus that there is persistence in performance by fund.
Past performance says that LPs will achieve better marginal returns by putting their money into top-performing smaller funds.
The data to speak to this issue are hard to come by. The worst-performing funds fall out of the data altogether. Some big famous old funds share no data whatsoever and are extremely good at eradicating the history of their failures from the Internet (leading me to suspect they aren’t doing quite so well as they would like the world to believe). Counting IPOs gets us only a short distance to an Answer. This is a difficult question to study.”
Assuming you agree (as do most LPs) that past performance is at least somewhat predictive of future results, past performance says that LPs will achieve better marginal returns by putting their money into top-performing smaller funds. LPs aren’t putting all their capital into predictable, low-return fixed income funds because they understand that a higher return dispersion promises higher returns for savvy investors who can invest in top-quartile funds. The same reasoning explains why it is strategic to put money into smaller funds.
Tom agrees with my conclusion and notes that, “I completely agree with you that LPs with access to top-tier smaller VC funds should take advantage of that. But the reality is that most top-tier small VC funds are essentially closed to new capital. Just think about it: if the fund size never expands, the only way for new LPs to enter is if others drop out. But why would any LP ever drop out of such a fund? So the practical question becomes: if an LP is presented with a choice between a large fund with a top-quartile track record and a smaller fund without one, which should it choose? The data suggests the track record is what matters, and the large fund is the far better bet in that situation.”
Tom draws on two metrics to demonstrate that larger funds are, in fact, more likely to be top performers. The first is that the ThomsonOne database shows capital weighted average returns consistently beating simple average returns for vintage years 1980-2008, implying that larger funds do better than smaller funds. The problems with this evidence are threefold. First, it isn’t very precise. It shows that, generally speaking, larger funds do better, but we can’t define “large” any more than we can formally define “small.” Therefore, the claim won’t extrapolate to any specific brackets. Second, credibility (fund reputation) is a huge confound. It is a variable explaining why large funds are able to raise so much and why some small funds remain small.
If top-performing funds tend to scale up, large funds might be doing well in spite of their size rather than because of it. At the very least, ThomsonOne’s numbers do not prove a causal relationship between fund size and performance. Tom agrees on this point: “I tend to agree with, but can’t necessarily prove, that it is probably the case the large funds are doing well in spite of their size not because of it.” Third and finally, other datasets fail to corroborate the result. Cambridge Associates reports that funds with more than $150M capitalization only generated higher pooled returns than those with less in 11 of the 30 vintage years between 1981 and 2010. Kauffman found that their top performers are mostly small funds. Silicon Valley Bank used absolute divisions (>$250M and $50-$250M) and found that the small funds do better, across the board, in terms of multiples.
The second metric from ThomsonOne shows that a greater proportion of the largest 5 percent and 25 percent of funds are in the top quartile of funds from vintage years 1980-2008 than of the smallest 75 percent. Aside from credibility remaining a confound and other datasets disagreeing, I have two other concerns. “Large fund” is a moving target; the average fund size has crept up since the late ’90s. Additionally, we already know that the top performing small funds do better than the top performing large funds, so the story would likely be different if we looked at the (arguably more important) smallest quintile or decile.
If top-performing funds tend to scale up, large funds might be doing well in spite of their size rather than because of it.
Tom’s article notes two caveats that I suspect are responsible for some of the fuzziness regarding correlations between fund size and performance. Firstly, “fund size” can be misleading as some firms prefer to raise a large aggregate amount split between several concurrent funds. Secondly, none of the VC return studies to date, including the ones we discuss here, are definitive.
To some degree, they are all hampered by data scarcity problems. I think that probably the best data source is the Kauffman Foundation study cited above, because: they are highly sophisticated investors who have access to almost any fund they want; they can calculate precisely the idiosyncratic fee structures of all the funds in which they invest; and, unlike almost every other data source, they understand precisely the dynamics of the limited partners because they are one.
In response, Tom writes, “The Kauffman data set is exceptionally small in comparison to the others and inherently biased (not in a nefarious way, but in the statistical sense that they had an active management policy that selected their investments and those are the funds for which they have return data). I grant you that the data sources I have access to are not that great either, but it was all I have. But the paper I linked to by Steve Kaplan, et al. examined four separate sources using a variety of return metrics. It is by far the most comprehensive and least biased analysis, which is why I linked to it. And it was done by an expert academic in this space.
The results were that 1) fund size doesn’t matter a whole lot; 2) but to the extent it does, it seems there is a slight correlation toward larger funds having slightly better returns. Now, it is true, if you search you’ll find another Kaplan study that shows increasing fund size among sequential funds by the same manager has a negative correlation with return.”
Personally, I think that the Kauffman Foundation, a financially motivated industry insider, represents the motivations and returns of a typical LP more than an outside academic likely does.
Tom also notes (and I agree), “One other point: My points are all macro-level and from an LP’s perspective. Ironically, it may be true that it is better for the VCs themselves to raise smaller funds because the more volatile a fund the more valuable the carried interest (which has, after all, the exact same payout structure as a call option). A cynical observer might argue this is why some firms raise many concurrent small funds… then they have both large management fee streams and high volatility on each given carried interest pool. But I’m sure that has nothing to do with why those firms structure that way :-).”
Thanks to Dartmouth intern Matt Joyce for help researching and drafting this and for Tom Grossi’s and Dr. Susan Woodward’s thoughtful comments.