AGILEVC My idle thoughts on tech startups

January 30, 2009

An interesting conversation has been going on for the better part of a week at PEHub on ways in which the venture capital business might be radically reinvented.  [NOTE: PEHub typically archives stuff behind a paywall but at least for now this article is still available].  The underlying premise for various reform approaches is that VC model as it exists today is largely broken.  But before contemplating how the venture biz might be reinvented, I wanted to first think about whether the VC model does indeed need dramatic change.

[UPDATED:  PEHub has now archived this article.  At a high level, it proposed one radical approach for reinventing VC in which individual "star" investors essentially become sole proprietorships working with a single large LP.  The idea being that individual VCs would then stand up on their own merit (as opposed to the collective partnership or the established brand) and LPs could construct and actively manage (i.e. hire new, replace underperformers, etc) their VC allocation thru this roster rather than making partnership level bets.
There are lots of different points of view on this question, but as best I can tell they cluster into a handful of categories:
1) Everything’s copasetic - ignore the lack of exits, the weak average industry returns, the dramatic growth in fund sizes, strategy creep of various forms (international, growth equity / buyout, etc).   Things will rebound in the not too distant future and everything will be fine.  
–> My take:  I personally hear very few people espousing this view today, and the tiny handful that do usually have a material interest of one type or another in maintaining industry status quo.
2) VC industry is broken and should be junked - VCs invested in too many crazy Web 2.0 ideas, had build to flip mentalities, most LPs have gotten limited distributions over last decade, most VC professionals are jerks, startups are far more capital efficient today and can be built solely thru bootstrapping / angel funding, etc.  Let the VC biz go the way of the dodo.
–> My take:  I do hear a nontrivial number of folks with some variation of this point of view, and certain portions of their arguments hold some truth.  But most of the time when I hear this POV it’s based largely on anecdotal evidence, i.e. person had a small number of bad experiences (either as an entrepreneur or occasionally an LP) or because their perspective comes from a limited part of the overall VC industry.
3) VC model just needs a “back to basics” approach - lots of innovation still occuring, lots of capable entrepreneurs to back, good risk/reward balance for the asset class going forward.  But funds have grown too large to invest in startups very profitably and raise challenges like LP/GP incentive alignment (as mgmt fees grow larger in absolute terms), strategy creep, and managing large partnerships (in terms of headcount).  A world composed largely of funds in the $100-300M range, with teams in the 3-6 partner range means a healthy and more sustaintable future for the business.
–> My take:  there is probably some merit to this argument.  Whether by choice or by market forces, we’ve seen firms like CRV, Atlas, et al taking this approach over the last several years and a number of new firms setting out with it.
4a) VC as an asset class is simply overfunded - reduce the aggregate LP investment in VC and everything will get better.  Underperforming VC firms, whether they’re new ones never able to prove themselves or established ones who can’t repeat past success for whatever reasons, will go away over time.  The remaining ones will be more disciplined about valuations, will fund fewer “me too” companies, and aggregate returns for the industry will increase.
–> My take: some of this will happen one way or another.  Yes… given 10yr fund lives and such this process takes awhile.  But it’s already started to happen quietly.
4b) VC as an asset class is simply overfunded - academic research by Josh Lerner at HBS and others indicates the top 10-20% of VCs account for a disproportionate share of the returns generated by the industry as a whole.  LPs should only invest in VC if they can get access to the long established, brand name firms.  So if LPs reduce the aggregate investment in VC and concentrate it in the small handful of firms successful over the prior 15-20 yrs, they will be more disciplined about valuations, will fund fewer “me too” companies, and aggregate returns for the industry will increase. 
–> My take:  if you point to firms like Kleiner and Sequoia there is obviously some merit to this argument.  But this research doesn’t form a particularly illuminating insight for LPs or VC biz overall.  By and large it’s based at looking at performance data of individual funds as opposed to firms across the sweep of time.  Even the most respected of firms have had some poor performing funds, and in fact a number of VCs with exceptional returns from 1990s vintage funds have funds from the current decade that look rather abysmal.  Why would one expect VC performance (at a fund level) to be a normal distribution anyway?  Lots of things bear closer resemblence to a Pareto distribution and given the very wide spectrum of outcomes for individual startup investments, it’s not really a surprise that VC fund performance is not normally distributed.  It does not necessarily follow, however, that firm level performance from here into the future is obviously concentrated.  
5) VC needs radical reform - star VCs operating as sole proprietorships (as suggested in the PEHub article), drastic overhaul of the 2 & 20 model, algorithmic trading strategies for investing in tech startups, and more.  While VC investing is largely driven by innovation in technology and business models, the industry itself is heterogeneous in terms of strategies, operating models, pricing, etc.  
–> My take:  unclear today whether radical reinvention is necessary for the VC industry to be more successful overall.  I do welcome the innovation that’s occuring or at least being contemplated and it will be interesting to see what approaches succeed in the longer run.