There has been a lot of excellent blogging and valuable dialog recently about the Venture Capital Model Being Broken.
Everyone is trying to figure out what happened to the VC industry, and many are making suggestions on how it might be fixed. Unfortunately, a lot of this work is based on bogus data.
Erick Shonfeld's post "A Scary Line Has Been Crossed for VCs" on TechCrunch picked up on some provocative writing by Adeo Ressi, founder of The Funded.com, in his talk at Harvard Business School. This graphic from The Funded got a lot of attention and coverage in the old media.
Fred Wilson wrote an excellent post in response to Claire Cain Miller's article in the New York Times titled "Venture Capital Returns Dip Below Zero." Part of Fred's argument is the data below, from Thomson Reuters and the National Venture Capital Association. Like Fred, I have referred to this data many times. It is regarded as the best available data on many financial asset classes.
The perspectives illustrated in the graphics above seem contradictory. So what's really going on?
In my opinion, nobody really knows because the data on VC industry returns is bogus.
When you look at the returns on an index like the NASDAQ or S&P 500 you can trace back all of the underlying source data to accurate financial information submitted by the public companies as required by law. You can be reasonably confident that every company in these indexes has submitted recent, accurate financial information.
The same is not true of the venture capital industry.
When I ran a family of VC funds, I would get calls all the time from Thomson asking us to submit our data. At first, I dutifully submitted our information. Then I got busy with our portfolio, so I delegated it to one of our junior team members. I was never sure they actually knew how to complete the information forms.
Then we got even busier and nobody in our organization had the time to report all of the detailed information that Thomson was asking for. Thomson would call often. They would track me down at conferences and offer to buy me drinks. I believed in what they were doing and sincerely wanted to help, but I just couldn't find the resources to submit the information they needed.
It bothered me and I occasionally wondered how they could aggregate meaningful data if funds like ours didn't submit their information. I asked one of the Thomson team at a conference once. They looked at the floor and mumbled something about "that being a problem in their industry."
Then one day at a national VC conference the truth hit me like a hammer between the eyes.
I was watching a presentation on the latest statistics on the Canadian industry's dismal investment returns. I kept thinking to myself that the numbers couldn't be that bad. Then the main researcher said something to the effect that there was a bias in the numbers.
The glass-half-full-entrepreneurial-optimist in me leapt on his words and I thought to myself: "Ah, ha! I knew it couldn't be as bad as those numbers up on the screen."
Then the researcher went on to explain that the bias came from the fact that "under performing funds tend not to report". My brain churned for a second as I tried to assimilate that statement. Then it dawned on me. The bias in the data meant that the actual returns were worse than what the numbers showed - possibly much worse.
Since then, I've asked several of my VC friends about this. Some had never wondered about the accuracy of the industry statistics or worried about this bias. Many weren't submitting their data either. There is no question that human nature makes it much less likely that an underexposing fund will report, but some of my friends with good results were often not reporting either simply because of the huge amount of effort it requires.
There has been widespread criticism of the venture industry for not publicly reporting their returns. As the private equity asset class has exploded over the past decade or two, the regulators have tried to improve disclosure. Until we have full disclosure, we will never really know how the venture capital industry is actually doing.
If any of you think I am missing something here, please add a comment.
Angelblog propagates best practices for entrepreneurs and angel investors. [more]
Basil:
After doing a similar study in 2005 that looked at all jurisdictions in Canada and the US, state by state, province by province, we concluded that there was no direct correlation between the amount of VC invested in a locale and its ability to generate successful exits (measured as value at acquisition and value at the pricing of the IPO). If there was a positive correlation, then Quebec would have created the most exit wealth in Canada. When you normalized by population, GDP or money invested, BC was the winner in the time period we looked at and Quebec was dead last... which brings up a very important point: You can't measure exits and money invested, or raised for that matter, in the same year. The slide you show is completely misleading. Companies don't exit the year they get investment!!!
We chose to do the funding period from 1995 to 2002 and the exit period from 1999 to 2005 in the study. The follow-up study by UBC's Thomas Hellman and Jim Brander did a much better job of spreading the timeline by finding average time to successful exit from initial funding (In Canada it averaged around 76 months, or 6.3 years).
So the theory that the model is broken and that the data is poor I cannot disagree with... but the premise of the slide and the article in the NY Times is completely misleading... what they need to do is take the funding and investment dates back about 6 or 7 years to compare to exits in the current year.
Basil, I am not sure what the point is that you are making.
1. There is widespread consensus that the traditional VC model is not suited to the new world of information technology for a bunch of reasons (new tools and technology stacks, open source, offshore, agile, internet distribution, yada yada). You shouldn't get much argument on that.
2. The data on VC returns is sketchy, for the reasons that you laid out, as well as the reasons Brent laid out above. This makes sense. (I suspect though that as flawed as the VC data is, it is probably better than the data on angel returns).
Sooo, what is the punchline again....?
Basil,
I was and still am a keen follower of this data and did more than my fair share of producing it and working on it for the Canadian industry.
When you speak to data collectors they to admit to issues with data coming from VCs, then say they correlate with data coming from LPs. How true this is I don't know.
In Canada this is a big problem because of how small the sample size is. Look through the list of reporting funds and decide for yourself if it is representative. Take into account that these are dollar-weighted IRRs - VW's numbers move the results a lot more than Greenstone's (sorry, couldn't resist Brent - note I didn't say which direction). Further, the big funds a likely more accurate as they have staff / finance folks doing this as opposed to over worked investment officers who don't really care. I went through the data every quarter for VW and was often correcting mistakes from the Thomson system - does everyone do that I don't know, but I doubt it.
My simple rule of thumb was to ignore everything short of 10 year returns, or better yet, only returns on fully realized funds. Also, the US numbers from the shear size and number of funds statistically is going to be more accurate. That said, if you look at the reporting percentages (something I did in the past) it has been falling precipitously for the last few vintage years (compare reporting funds to funds raised in a given vintage year with the Thomson data). IIRC back in the 90's you got 50%+ reporting, and as of 2007 you were down into the high-teens low 20s for % of funds reporting for vintage years post 2000. On the flip side, if all the good funds report, then the vast majority of the dollars is reporting and the dollar-weighted IRR will be correct. Based on research Gilles DuRuffle has done, the good funds in the US have the most money, so if the good funds report, than the bulk of the money is also reporting.
Oh, one other note about horizon returns - they are just that, horizon returns. Starting value, plus/minus changes during the period, then ending value. Right now 10 year horizons start with 1998, a healthy year for valuations and end with 2008 and not too bad year, but not 1998. Try next year's horizons, 1999-2009 - you think that may not look so good? ;) To be fair, that's as much inflated 1999 values as it is poor performance in the intervening years. Horizon returns are an artifact of people wanting to compare PE to Equities - not a real test, but a test that is easy for LPs to understand.
I can go on about this topic ad nauseum. The only people with real data are the LPs themselves and I encourage people to go look up individual fund returns from the various pension funds with public disclosure rules (Calpers, Calstrs, WSIB, UTIMCO, etc.). If people are having trouble finding the data let me know and I can track down links.
I agree with Brent, that the Industry is using current economic conditions as an excuse for failed investments made 5-8 years ago. It's a convenient scape goat for managers to use.
The question I asked at the end of the last VEF was aimed directly at this point too. I challenged the panel of VCs to answer how current economic times effects investment velocity, since they plan for 5-8 years to exit a new invest made today. Do they honestly believe that the market will still be recession 5 years from now? Logic dictates that they should be stepping up investing in early stage companies during this time of lower valuations and less competition. Let's see if they agree with me.... and Warren Buffet :-)
"Be fearful when others are greedy, be greedy when others are fearful."
A well-written article Basil, and equally cogent responses, esp from David M and Brent. Fairly long experience around, with and in venture capital and financing leave me to conclude that the 2 following essential principles guide most, perhaps, all financing activity, ultimately:
#1 - What gets rewarded gets done. What is punished is avoided.
#2 - What isn't visible can be rearranged to suit #1.
Judy B
