The Rise And Fall Of The Venture Business

This is the third installment in my venture capital fund performance series that I am doing this week. Today I want to start looking at the ugly years (1999 to 2003). As I’ve done in past posts, I like to focus on the "realization ratios", the cash on cash returns, instead of the IRRs. IRRs are useful, but I get paid on gains and that’s what I always want to look at.

This chart shows distributions over paid in capital (DPI) and total value over paid in capital (TVPI) from the late 1960s to now. The data comes from Venture Economics. These numbers are called "pooled average" which means they take all the funds they have data for, pool them all together, and calculate one single number. This is an estimate of what you’d get if you invested in every fund that was raised in a particular vintage year.


You’ll note that DPI and TVPI are basically identical until the mid 90s. That’s because funds raised before the mid 90s are pretty much fully distributed now. The gap between TVPI and DPI is the value of the unrealized investments.

This chart is dramatic. The venture capital business was a great investment from the mid 80s through the mid 90s. You could get 2.5x to 5x on your money if you bought the entire industry.

But by the end of the 90s, starting with the awful 1999 vintage, the business has not been an attractive place to put money.

Frankly I am surprised that the 2001, 2002, and 2003 vintage years aren’t showing better numbers. Those were years when you could buy quality companies at very attractive valuations and many of the companies that were funded in that period have since been liquidated at much higher prices. I don’t understand why that is not showing up in the data.

My expectation is that this chart will look at lot differently in three or four years and we’ll see a return to a money making business. But I doubt the numbers will get to where they were in the early/mid 90s.

Next up – I am going to dig a little deeper into the "ugly years", from 1999 to 2003, to see exactly what is going on in those vintage years.

#VC & Technology

Comments (Archived):

  1. Steven Kane

    am i reading this right? that 1998 vintage fund — as of today, about nine years later — have a TVPI of 2.5X? which means investors got/are getting 8.5% annual return? if thats the case, i’m amazed – i though 1998 vintage funds were much better performers overallfred, when you say your expectation in 3-4 years will see a return to money making, what exactly do you mean? that the asset class of VC, after fees, will handily beat, say, the major public market indexes?

    1. fredwilson

      i can’t predict what public markets will deliver or frankly what VC will deliver, but i do think that VC will start producing1.5x to 2x TVPIs.the thing you are leaving out of your math steve is VC capital calls come slowly and distributions happen starting in years 4, 5 and 6, so the average duration of a VC investment is more like 5 years than 10 years.fred

  2. leehower

    As others have pointed out, it’s not easy to translate a fund multiple (measured here in TVPI) to an IRR. So it’s true that if the IRR calc is based on a 10 years of compounding, the net return would be 8.5% for a fund multiple of 2.25x.But an LPs commitment to a VC fund doesn’t work like that from a cash flow standpoint… they don’t put in their full $X million on the date of the initial fund closing nor do they get all the distributions 10yrs later when the fund is dissolved. VCs typically call capital on an ongoing basis (either deal by deal, or on annual/quarterly schedule), so an LPs *invested* capital will have an average life shorter than 10yrs and similarly the timing of distributions they receive back will typically be based on when various portfolio companies have exits.So as was highlighted in the comments from prior posts in this series, the actual time period that an LPs dollars are invested (which matters for IRR as the compounding period) is typically 5-7yrs. So for 2.25x multiple example from ’98 vintage, the IRR would be 12.3% if you assume an avg of 7yrs that LPs investment is actually “at work” and 17.6% if you assume 5yrs.

    1. fredwilson

      i will publish some IRR data as part of this series as i can see people want to see that.

  3. VC Data Junky

    Back to the bias question. You may have a sort of reverse answer to your question “Frankly I am surprised that the 2001, 2002, and 2003 vintage years aren’t showing better numbers. Those were years when you could buy quality companies at very attractive valuations and many of the companies that were funded in that period have since been liquidated at much higher prices. I don’t understand why that is not showing up in the data.”Looking at the data of reporting funds vs funds raised (by #) you are below 20% for 2000 forward with less than 20 funds reporting / year from 2002 forward. Might be the reverse issue of selection bias. Also, many of the notable exits recently were funded pre-bubble (97-99). E&Y/VentureSource data shows the 2006 median time to M&A from first funding as 6.0 years with 2007’s data showing 6.6 years with the same data for IPO showing 6.3 and 6.8 years respectively.Combine those dates with a 2-3 year investing window for the average fund and most 2001-03 funds’ deals aren’t “mature” yet (but we hope real soon).Other point, still not sure about using pooled average as no LP can provide money into all the funds on a cap weighted basis and we know the bigger funds have the best returns (reason the pooled average is above the median). If you put equal dollars into every fund you’d hit closer to the average, which is generally below the pooled average and above the median.

    1. fredwilson

      Great discussion going on here. I love it.Your point that many of the recaps done in ’01/’02’03 were done in the ’99 vintage funds is true. And its strange that the ’99 vintage funds are not back above water as a result.But I also think the valuation environment that existed for the ’01/’02/’03 vintages should start producing good return numbers. I am surprised we don’t see it yet.As to selection bias, I will look at Cambridge data. They get all the funds that their clients invest in so it should include the better funds.fred

      1. VC Data Junky

        Not really sample bias versus simply a small sample. Looking at the most recent (Q2-07) Cambridge data it is certainly richer (double to triple the number of funds in the sample). The numbers appear marginally better, but not much unfortunately.As for ’99 funds making money, you have to play (and play heavily) in recaps or big down rounds (with mult. liq. prefs) to make any money and I think a lot of funds were too busy backing off in 01-03 to take advantage.

        1. fredwilson

          I am biased by my experience. Our ’99 vintage flatiron fund is now well above water and we took some huge hits in ’00 and ’01Fred

  4. Bob Warfield

    It seems to me that a lot of funds made another change: distribution could happen well after the perceived liquidity event of IPO. The last company I was at, Callidus, IPO’d in 2003, yet many of the original VC’s still haven’t distributed all of the funds 4 years after the event. Hence there are hidden returns that won’t be reflected until all those shares are distributed. How many other IPO’s are still not reflected in these numbers?At some point these VC’s evidently wanted to make even more money on the appreciation post-IPO as they watched portfolio companies soar after an IPO. There may be other motivations at work too.Cheers,BW

  5. Venture Capitalist

    I am a VC and I know from experience that generally speaking if a deal doesn’t produce a good return (multiple or IRR) by around year 5 or 6 it is probably doomed. Therefore, I find it surprising not that the IT 2000, 2001, 2002 vintages haven’t returned better results, but that they haven’t produced WORSE results. I am sure they will…

  6. SF VC

    Also, what the hell is a VC in NYC!? Every single NYC VC that I have met is more or less a small time player…and a joke…

    1. fredwilson

      Have we met?

  7. Don Jones

    I wonder what the correlation would be for breaking out IPO exit returns vs. M&A exit returns. I bet the good IPO years would correlate much more closely.

  8. PierreBradshaw

    Where did that comment come from!? Catfight!!!Is the rivalry between Silicon Valley and Silicon Alley VC’s that hardcore? HahaNice chart breakdown Fred. I’m a stats junkie, this is very interesting. I’m looking forward to seeing more details from 99 to 03.

  9. ron pruett

    Fred,Always more interesting to me is why VC’s did so well in the early days – smaller funds and real relationships with business-builders. They were also generalists vs. specialists though with an eye toward technology. Maybe it will come full circle?

    1. fredwilson

      I think supply/demand was in the VCs favor back then. Not so much anymoreFred

  10. akcchoi

    Fred, thanks for the putting the chart together. What I find intriguing was the rapid decline (around from the beginning to around 1982) and the subsequent recovery (1984 – 1996) in VC returns. I wonder if anyone can suggest what are the drivers leading to both the decline (technology stagnation, US macro-economic slow down, too many VC firms, etc?) and the recovery (sudden shift in technology, reduced number of VC firms?). Also, are we seeing similar drivers which can point to a VC recovery in the near term?

    1. fredwilson

      My guess is that is related to the market drop in ’87 which took a while to recover from. Early 80s funds did not have a great IPO market to liquidate into whereas mid 80s/late 80s funds were less sensitive to that

  11. dintersmith

    I used to spend way too much of my life analyzing the capital inflows and outflows of the venture business. Now, thankfully, I’m traveling around the world with my wife and two children (, having the time of our life. But I have followed your blogs on VC performance, and wanted to add my comments from our hotel overlooking the Taj Mahal at sunrise.The dirty little secret of the venture industry is that we spend 99% of our time tracking and following investment and fund-raising data (generally hard to get), and almost no time tracking or reporting on distributions to LPs (fairly easy to get). The industry reports quarterly on investment levels, and will often say things like “The industry invested $6.3 billion this quarter, showing great discipline and keeping investment levels at a long-term sustainable level.” Yet the data, if you dig, shows that nothing could be further from the truth.I was on the NVCA Board of Directors for four years, and pushed the organization to report on distributions data, in conjunction with investment data. Without data on distributions, it’s like a portfolio company telling us in a Board meeting all about expenses, but failing to talk about revenues. The NVCA, I believe, is reluctant to make this distribution data visible, since it would hurt the ability of many of its members to raise funds (and pay annual dues), but they do have the data.If you are an NVCA member, there is a section in Member Resources on industry data. Believe it or not, slide 276 has data on investment levels and distributions. The data shows that, with the exception of the two bubble years (1999-2000), we have for many, many years been distributing anywhere from $10 to $18 billion annually back to LPs. If the industry invests $25 billion a year (which it’s done, more or less, since 2002, it needs to return $50 billion a year in “steady state” to be a remotely attractive investment class. Only in the year 2000 (not even the bubble-fueled 1999) exceeded this amount. Now, obviously, dollars invested today generate returns down the road, and returns today reflect investments much earlier in time, but a steady state model shows the dramatic unsustainability of current investment levels.The fact is that LPs keep shoveling money at our industry, and pay venture capitalists large fees to deliver mediocre (and often negative) returns. Each year, there will be a supernova investment (Google, Skype, at some point Facebook), and everyone will talk about how “If we can just get one Google in our portfolio, we’ll be in great shape,” but they won’t.The reason our industry is so inefficient has to do with two things. First, there is a long (a decade long) lag between investment decisions and realized results for the industry’s Limited Partners. It’s highly likely that an LP committing to a new venture fund today will be off to some other career assignment by the time the dust settles on the merits of the investment. And the LPs, by and large, have become large enough to have private equity (and often venture capital) specialists, who would effectively be turning in their resignation if they went to their boss and said, “This industry is unattractive and we should get out of it.” But they should.For the past five years, I was on the investment committee of a university, and one of our members was in charge of private equity investing for a very recognizable college. When we discussed investing in venture capital, he said, “I wouldn’t invest a nickel of my own money in the sector. But I do for my employer, because that’s what they’re paying me to do.” So we have LPs paid to invest in venture, venture capitalists getting large fees to invest in portfolio companies, sectors trampled with competitors, and money stuffed into portfolio companies beyond what they need. And so far, nothing seems to be bringing the situation back into balance.Ted Dintersmith

    1. fredwilson

      tedthanks for taking time out of an amazing trip to comment. i am jealous.and thanks also for saying what many in our industry won’t’s tough to make money in this business and getting tougheri believe in our model: keep it small, invest early, and have a lot of discipline.i think we can generate the kinds of returns our investors are expecting.but it’s certainly not easyfred