I read quite a lot of papers about finance and investing, but I can't remember the last time I came across a 52-page paper which I simply devoured, avidly, reading every word, and even following the footnotes. But such is the latest publication from the Kauffman Foundation, a truly wonderful report on the foundation's own experiences in the world of venture-capital investing. This is required reading for all institutional investors with any kind of exposure to VC, and I sincerely hope that it succeeds, at least at the margin, in forcing those institutional investors to behave a bit more like investors, and a bit less like chumps being bullied into throwing millions of dollars into a series of opaque black boxes delivering decidedly subpar returns.
The Kauffman Foundation was created to encourage entrepreneurship; its endowment currently stands at some $1.83 billion. Of that, $249 million is invested in VC and growth equity funds; the foundation has been investing in VCs for 20 years now. As a rich, long-term institutional investor devoted to the cause of early-stage companies, the Kauffman Foundation is - or should be - pretty much the perfect LP as far as VC funds are concerned. And indeed, over the years, it has invested in 100 such funds, and therefore now has a spectacular real-world backward-looking dataset of VC returns from an LP perspective.
This is the kind of dataset that money, literally, can't buy: VC funds' investment agreements have such tight confidentiality clauses that Kauffman and other institutional investors would never be allowed to share this information with anybody else. But by anonymizing their data, and by self-critically coming clean on their own returns from venture capital, Kauffman's investors have managed to put together a detailed and compelling report with a very simple conclusion: venture capital is not much of an asset class, and insofar as it is an asset class, it's very, very broken.
Over the past 20 years, net of fees, Kauffman has been paid out 1.31 times, on average, the amount that it invested in any given fund - well below the standard "venture rate of return" of twice committed capital. The payout is meant to come after no more than 10 years, but the 10-year figure is honored mainly in the breach: Kauffman alone has 23 funds more than 10 years old, and eight funds more than 15 years old. One fund, at age 19, still retains more than 20% of the capital that Kauffman committed way back in 1992.
Here's the bigger picture: the total amount returned by all of Kauffman's funds. The red line is at that 1.31x average multiple:
This is very much a short-head, long-tail dataset, with the short head having high returns and the long tail being decidedly disappointing. What's more, the really high returns in this chart - the ones which achieve that "venture rate of return" of 2x committed capital - come exclusively from funds with less than $500 million committed: something which is very rare among top-tier VCs these days. What's more, most of them also come from funds raised before 1995. If VC funds were good investments once upon a time, they're not any more:
In reality, VC returns have been dismal for the past 15 years:
During the twelve-year period from 1997 to 2009, there have been only five vintage years in which median VC funds generated IRRs that returned investor capital, let alone doubled it. It's notable that these poor returns have persisted through several market cycles: the Internet boom and bust, the recovery, and the financial crisis… In eight of the past twelve vintage years, the typical VC fund generated a negative IRR, and for the other four years, barely eked out a positive return.
I'm all in favor of investment strategies which display low-volatility returns, but only when those returns are actually positive; in reality, according to this report, the average VC fund returns less money to investors than they invested in the first place.
The Kauffman report makes a very strong case that the best way to look at VC returns is to benchmark them against a liquid public stock index: they like the Russell 2000 small-cap index. If you do that, you get a chart which looks a bit like this. The x-axis is PME, or public market equivalent: if a fund is lower than 1, as most of these are, then you would have been better off just investing in the Russell 2000. And while there are certainly funds which massively outperformed that index, in order to get that kind of performance you really need to invest your money before 1995, to take full advantage of the dot-com bubble.
Given the high fees and the illiquidity and the inherent risks of venture-capital investing (the Russell 2000 can't go to zero, a venture fund can), it's reasonable to expect a venture fund to return at least 3% per year over and above the Russell 2000. Using that metric, 78% of the funds that Kauffman invested in have failed. And remember that Kauffman has more access to the top-tier firms than most VC investors.
But the really interesting thing about the Kauffman paper is that it doesn't blame the VCs for this underperformance. Instead, it pins the blame squarely on the LPs - the investors who irrationally invest money with VCs.
The first thing they do is have a "VC bucket," which then turns them into "bucket fillers." VCs love bucket fillers, because an LP with a bucket is an LP with no ability to invest in something better instead, like the Russell 2000.
GPs we interviewed are very aware of "bucket filling" behavior, and said LPs with VC mandates act like the money is "burning a hole in their pockets." They just need to spend it. Institutional investors governed by mandates presumably attempt to get into the ten to twenty top-tier VC funds; but if they can't, they're left to choose from second- and third-tier funds-a strategy that nearly guarantees returns unlikely to exceed a low-cost, liquid, small cap public index.
If you look at the performance of VC funds during the golden years of 1986-1999, it turns out that once you strip out the top-performing 29 funds, the rest - more than 500 - collectively invested $160 billion, and managed to return $85 billion to investors. If you can't get into one of the best funds - and everybody knows which funds those are - then there's really no point investing in venture capital at all. But what happens is that some investment board looks at VC returns inclusive of the best funds' returns, and then mandate a certain investment in VC which assumes they'll have some kind of access to those top-tier funds. And that's an extremely dangerous assumption to make, because most of the time it won't be true.
The second thing that LPs do is that they pay VCs to raise lots of money, rather than pay them for returns. Most of the income in the VC industry comes from the 2 bit of 2-and-20, rather than the 20 bit. Which means, as the report puts it, that "while a select group of VCs remain focused on delivering great investment performance to their investors, too many are compensated like highly-paid asset managers."
As such, these GPs have every incentive to exaggerate early-year returns in their funds, despite the fact that the whole point of the way the funds are structured is to allow long-term investments which don't pay off until many years down the line. Here's the most astonishing chart from the report:
What you're looking at here is the self-reported returns from all 100 of the Kauffman foundation's funds, plotted on a time zero axis. In theory, if you believe the VC industry's hype, the returns should look a bit like the green line: negative in early years, as you make investments which won't pay off for a long time, and then positive by year 10.
In reality, reported returns peak very early on, in month 16 - which just happens to coincide with the point at which the GPs tend to start going out on sales calls, trying to raise their next fund. (The blue line shows total fund returns, while the red line shows returns net of fees - the money which actually goes to LPs.) Of course, at month 16, none of the returns is realized: they're driven instead by increases in portfolio-company valuations, and those valuations are set by the GPs themselves.
If GPs were incentivized mainly by their 20% performance fee, then you'd expect something like the green line, or at the very least you'd expect the performance to rise over time, as the fund's illiquidity premium manifested itself. If GPs were incentivized mainly by their 2% management fee, however, then you'd expect something much more like the real-world red and blue lines, where performance figures are used more to raise new funds than to make money.
Here's how the report explains GP math:
Data from the Foundation's portfolio indicate that the median time to the first capital call of a subsequent fund is 26.6 months. As a thought experiment, assume a VC partnership raises a $250 million fund. Early in year three, exhibiting early positive IRRs, the firm raises a subsequent $350 million fund. Demand for Fund III remains strong, and the GPs raise another $500 million fund later in year five. Each new fund adds a fresh income stream to the residual fees older funds continue to generate over the ten-year life. Without visibility into the firm financials, LPs don't see the total cumulative management fees the firm receives, and, more importantly, don't know where those fees go. In this theoretical example, a moderately successful VC firm raises three smaller-sized funds within the investment period of the first fund; and the operating income climbs to more than $19 million by year five. Our experience would indicate that VCs may somewhat increase fixed costs like additional staff with subsequent funds, but in most cases expand very conservatively.
The Kauffman foundation has lots of bright ideas about how this misalignment of incentives could be fixed: for instance, investors in VC funds could ask for much more transparency on the costs of running those funds, pay a budget-based management fee instead, and then pay a performance fee which rises sharply when the fund genuinely outperforms the public markets. For instance, it could be 20% when the public markets are outperformed by 3%, 25% if the outperformance is 5%, and 30% if the outperformance is 7%.
Ian Charles ran an intriguing experiment in 2008, when he asked LPs how much of a performance fee they would be willing to pay to get into a top-performing, $250 million fund. He got bids as high as 72.5%, and a market-clearing level of 42.5%. Clearly it's possible, in theory, for top VCs to charge much higher performance fees than they're charging right now, and some of them are OK with budget-based management fees, too.
But here's the weird thing: it's the LPs, much more than the VCs, who are averse to any change on the compensation front. Investors, it turns out, are extremely conservative, and don't want change even if it's in their best interest.
One startling fact from the report explains why that might be the case:
Our own casual review of institutional private equity investors reveals median job tenure of around three years. There is much turnover at the LP investor level, which creates little incentive for any one investor to fight hard on behalf of their institution for better economic terms, especially if they are not getting paid for that outcome and views such behavior as harmful to industry 'relationships,' which the investor likely will find beneficial in landing their next job.
What this says to me is that the VC industry, as a whole, is being incredibly successful at extracting rents from dumb institutional investors. These investors wouldn't dream of investing in a public company where there was no transparency as to basic questions like how much money the principals were being paid, but they happily invest in venture capital funds where the founders cream off so much of the income that younger top performers end up leaving the firm. And in general, VCs are incredibly good at playing fear off against greed: would-be investors really want massive VC returns, and they really don't want to be left out in the cold. Even Kauffman does that: "We have chosen to stand down on terms," they write, "when faced with an investment decision in a top-tier fund."
The big picture, here, is that an enormous number of institutional investors are still chasing VC returns which haven't really existed in the industry since the mid-90s. So long as all that money is chasing a relatively small number of opportunities, and especially now that valuations for early-stage tech companies are going through the roof, the chances that the average LP will make any money at all in VC are slim indeed. Venture capital is sexy, and it makes a lot of money for GPs. But investors need to take a very hard look at the asset class, and ask whether it's worth it from an LP perspective. After reading this report, they should probably conclude that it isn't.