Fund Size: It’s About The Change, Not The Dollars

Twitter confirmed the obvious on Friday, with a short blog post about how it had raised a new round of VC funding. No dollar or valuation info, of course (Twitter is to financial info what Bill Belichick is to injuries), but a list of investors. They were: Insight Venture Partners, T. Rowe Price, Morgan Stanley, Benchmark Capital, Institutional Venture Partners and Spark Capital.

Notably missing from the list was Union Square Ventures, which had been an investor with the micro-messenger since the beginning. Fred Wilson, a co-founding partner of USV, even sits on the Twitter board of directors.

So what happened? Wilson declined to discuss it (basically because of Twitter’s neurosis – my words, not his), but my assumption is that it came down to an issue of size. Specifically, the round was just too large for USV. The deal reportedly was worth $100 million at a $1 billion valuation, while USV is currently investing out of a $155 million fund. Moreover, USV originally invested in Twitter out of a $125 million predecessor fund (not sure if subsequent deals went cross-fund).

USV’s basic calculation would be that it’s better to accept the dilution in exchange for being able to add a few new portfolio companies. You need at least a couple mega-exits to generate top-tier VC returns for a $155 million fund, so the more bites at the apple the better.

All of this brings us back to the VC fund size issue, which we last discussed vis-à-vis Bill Gurley’s column about VC industry shrinkage. Before that, we discussed it when I ran data to see if VC funds that cut already-committed money in 2001-2003 outperformed the market (the results were mixed – although perhaps I should have benchmarked them against large funds that didn’t cut).

So my basic question: Is USV’s size strategy superior to a Benchmark Capital or Spark Capital size strategy?

There are no definitive answers, of course, but we can take preliminary instruction from some unpublished research by Josh Lerner (HBS) and Antoinette Schoar (MIT Sloan). They looked at how changes to fund size and firm size affect returns, and found that there is an apparent relationship.

Lerner and Schoar first took the universe of all mature VC and buyout funds (those raised through 1999), and discovered an inverted U-shaped relationship. Basically, mid-sized funds outperformed both small and large-sized funds. Peak performance was at around $300 million (note: the researched time period is admittedly dated).

They then broke out the types of funds, and found that the inverted U-shape is sharper and most statistically significant for VC funds than for buyout funds. In this breakdown, the peak VC size was $280 million compared to $1.2 billion for buyout funds.

So is the answer to have a mid-sized fund? Well, not exactly, Lerner and Schoar didn’t actually find too much performance variance based purely on size. Sure there was peak performance, but it didn’t vary too much from the high and low-end quartile benchmarks. For example, a $5 billion buyout fund had a predicted return of just 1.2% lower than a $2 billion buyout fund.

What is important, however, is the rate at which a firm’s fund sizes grow. Lerner writes: “We find that growth has a substantial negative effect on growth: A doubling of fund size, all else being equal, leads to a fall in IRR of -5.3%. This analysis suggests that a group that had a 25% IRR in its $1 billion dollar third fund could expect a return of only a little below 20% if it went and raised a $2 billion fund next.”

The related chart is fairly remarkable, in that the negative correlation between fund size growth and performance resembles a very steep hill.

So what is ultimately important for firms like USV and Benchmark isn’t purely the number of dollars in their current funds, but rather how that compares to the dollars in its past and future funds.