Venture Fund Performance - Sample Size
The comments to my Venture Fund Performance series have been excellent. In particular, VC Data Junky’s contributions have been terrific. He’s mentioned the falloff in sample size in the Venture Economics data on several occasions and I thought it would be useful to look at that.
The first thing you notice about this graph is that the venture business experienced a huge number of new funds in the mid 90s. That was also the time period when the business was putting up really strong numbers. That’s how markets work. Success brings more competition. And that’s what we got in the venture business in the mid 90s. Which, among other reasons, ultimately led to lower returns.
But I want to focus all of you on the recent years, 2002 and beyond. It is true that the number of funds that are reporting to Venture Economics has fallen to levels below where they were in 1990. On the other hand, Cambridge Associates’ sample size is back up to wehre it was in the mid 90s.
This highlights the difference between the two collection methodologies. Venture Economics relies on self reporting whereas Cambridge gets their data largely from the firms that their clients invest in.
And so as a result for the 2002 vintages and beyond, the Cambridge data should be much more accurate and less subject to selection bias. Very interesting.
For sure the Cambridge data is better, but still vastly under-reported. According to VentureEconomics the average number of funds raised per year was:1990-96: 1101997-2000: 4132001-2006: 226So, to have the same sample depth you’d need double the number of firms reporting post bubble as pre-bubble, and you certainly don’t. (Obviously I’m oversimplifying correct sampling methodologies, but this isn’t “real” statistics – which means it might actually be true).An interesting twist on the bias arguement (which everyone assumes means upwardly biased) is perhaps more recently the top tier guys are not reporting (dragging the metrics down) as they have notoriously dropped public reporting LPs and generally done their best to obscure returns from public disclosure. Non-reporting doesn’t mean bad, it just means non-reporting. Anyone who regularly talks with LPs knows there are more recent funds with very good returns that don’t seem to be in these numbers.