Entrepreneurs Are Not Overconfident

Not too long ago, the well-known economist Robert Hall presented this paper (co-authored with Susan E. Woodward) at my place of work.  Here is the abstract:

In the standard venture capital contract, entrepreneurs have a large fraction of equity ownership in the companies they found and are paid a sub-market salary by the investors who provide the money to develop the idea.  The big rewards come only to those whose companies go public or are acquired on favorable terms, forcing entrepreneurs to bear a substantial burden of idiosyncratic risk.  We study this burden in the case of high-tech companies funded by venture capital.  Over the past 20 years, the typical venture-backed entrepreneur earned an average of $4.4 million from companies that succeeded in attracting venture funding.  Entrepreneurs with a coefficient of relative risk aversion of two and with less than $0.7 million would be better off in a salaried position than in a startup, despite the prospect of an average personal payoff of $4.4 million and the possibility of payoffs over $1 billion.  We conclude that startups attract entrepreneurs with lower risk aversion, higher initial assets, preferences for entrepreneurship over employment, and optimistic beliefs about the payoffs from their products.

During the seminar it occurred to me that these results, assuming they are correct, are evidence of an absence of overconfidence, at least among the kinds of people who leave good jobs to form high-tech startups.  The reason is that if potential entrepreneurs were massively overconfident, one would expect to see lots of entry of startups based on weak ideas, which would lead to an expected payoff so low that forming a startup would be a losing proposition for the potential entrepreneur unless he/she started out extremely wealthy and/or had very low risk-aversion.  But what the authors actually find is that forming a startup with an average-quality idea* is a break-even proposition for a potential entrepreneur with quite modest wealth and with a more-or-less standard degree of risk-aversion.

After the talk, I asked Professor Hall if he agreed with this interpretation (he seemed to), and if he would object to my posting about it on OB (he didn't).  But I will make him aware of this post, and invite him to comment if he would like, and correct any mistakes that I might have made.

*The authors have no way to distinguish the quality of an idea, so there is an implicit assumption that the marginal quality of the idea is equal to the average quality of all ideas that actually get implemented.
GD Star Rating
Tagged as: ,
Trackback URL:
  • http://www.infoaxe.com Vijay Krishnan

    Did they actually measure risk aversion in an unbiased manner? It appears to me that overconfidence could well be tucked into their lower risk aversion. In other words, part of the low risk aversion of entrepreneurs might come from them assigning a lower probability to failure than the true probability, on average.

  • http://profile.typekey.com/robinhanson/ Robin Hanson

    I’ve heard average returns from more ordinary businesses, e.g., restaurants, are negative. Any truth to that rumor?

  • Ben

    Doesn’t your conclusion require an assumption that venture capitalists choose ideas to fund at random?

  • Chad

    Ben: Being a 15 year veteran of Silicon Valley, I’d have to say that is a fairly safe assumption.

  • frelkins

    All the VCs I know go through the idea and business plan with triple-fine Japanese combs. Merely average ideas don’t survive.

    If Hall starts with already funded firms, they have much much better than average ideas and have survived intense formal marketing analyses. Am I missing something here?

  • Terren

    one would expect to see lots of entry of startups based on weak ideas

    But you do see lots of startups based on weak ideas, at least in the web space over the last 12 years or so. More recently, the buzzword “web 2.0” is practically a euphemism for “no business plan”.

  • David J. Balan

    I’ve been thinking about this some more, and there are some wrinkles to this that aren’t discussed above. I don’t think the wrinkles overturn what I said in the post, but it’s worth laying them out a little better to see if I’m right.

    The basic assumption behind the VC funding model described in the paper is that the VC has less information than the entrepreneur about which ideas are good. This is why the VC structures compensation in the form of a below-market salary and a substantial equity stake: to fix things so that only entrepreneurs with ideas that are likely to succeed want to take the deal. If you assume that VC is competitive (i.e., VC firms are currently making zero economic profits), then VC firms would not be willing to offer the same terms they offer now if the average quality of ideas was lower. The terms would have to be worse, either in the form of a lower salary of a lower equity stake.

    Despite this, I think something like the original point remains. If entrepreneurs were massively over-confident, we would expect to see more ideas funded, on terms less generous to the entrepreneur, and such that the average idea produced payoffs low enough that an entrepreneur with an average quality idea would only want to take the deal if he/she was either really rich or really tolerant of risk.

  • David J. Balan

    Vijay, they didn’t measure risk-aversion. The asked how risk-averse someone would have to be to be indifferent between being an entrepreneur and not. And they find that a pretty conventional measure of risk-aversion does it. But this does raise the interesting question of whether overconfidence takes the form of over-estimating the expected return or of under-estimating the variance of the return distribution.

    Robin, I’ve heard similar things, but I don’t know of any evidence.

    Ben, it does assume that VCs are limited in their ability to evaluate ideas. I think the model implicitly assumes that they can’t do so at all, but of course that is what economists call “stylized.”

    frelkins, the “average” idea here is indeed the average quality of the ideas that actually get funded (only these ideas show up in the data).

  • frelkins

    @David J Balan

    I would love to hear what Paul Kedrosky or Guy Kawasaki would say about this. Oh yeah. I think they would laugh.

  • Lord

    The VCs have already drummed some reality into them. Those that haven’t made it that far, not so much.

  • http://don.geddis.org/ Don Geddis

    That “average” is the tricky part. The study concludes that the average funded company is worth it for most founders (in terms of expected value), compared to salaried jobs. But that’s not a choice most entrepreneurs considering jumping from a safe job to founding a company get to make. Their decision to leave their safe jobs only gets them in the pool of all startup companies, not the much smaller pool (10%?) of successfully-funded VC-backed companies.

  • http://www.theexpertsedge.com/ Chrissy

    Me, being an ‘underconfident’ entrepreneur needs help. I found some help reading, “The Expert’s Edge” by Ken Lizotte.
    It was time well invested. The Expert’s Edge book gave me a clearer focus on concepts I knew, served as a refresher in other areas, and introduced me to new ideas that will lead me closer to not only becoming a Thoughtleader in my subject field, but also a “Thoughtleader of Choice”.

  • Eric Falkenstein

    If the mean is $4.4MM, but the payouts top out over $1B, the median is probably well under $1MM. Give standard utility functions and the opportunity costs, I don’t see how this can work without some sort of delusion. Of course, there are non-standard utility function, but they have their own problems.

    Given Sturgeon’s Law as applied to any innovative effort (ie, 90% of everything is crap), it’s a wonder we do it (eg, write books, songs, create business plans).

  • http://profile.typekey.com/davidjbalan/ David Balan

    Eric Falkenstein,
    The paper cannot distinguish the quality of the ideas, and so treats all ideas as equal (meaning all ideas are of average quality). And the results the authors get seem to show that you would not have to be particularly overconfident to want to be an entrepreneur with an average quality idea. You are certainly right that the marginal idea is going to be of lower quality than the average idea, meaning that the guy with the marginal idea may well have to be overconfident to want to go ahead with it (otherwise he/she would have to be either really rich or really risk tolerant). So the title of the post is really too strong. But the result does, I think, put a bound on how much overconfidence exists among potential entrepreneurs. If there was really massive overconfidence, you would expect a finding where even the average quality idea wouldn’t make sense to do (again, unless you were remarkably rich and/or risk-tolerant).

  • Douglas Knight

    Eric Falkenstein,
    looking at the paper, the median is low, but it’s dominated by returns of 20-50MM, not 1B. They claim that they use standard utility functions. The only think I can point to that is a clear error is trusting VC data. That’s probably not enough to discard the paper, but I’m going to assume that they make some other error I didn’t find.

  • Grant

    Venture capitalists typically fund businesses after they’ve gotten started (though they still may not be at all profitable). Angel investors (or in the case of bootstrapped businesses, the entrepreneur’s own judgment, spouse, etc) already do some “idea screening”.

    Terren, most of the questionable “Web 2.0” businesses I’ve seen were self-funded. The increasing productiveness of web development tools means more and more people can start a web-based business without needing as much outside help (and opinion, expertise, etc) as before.

  • Sonnie Bailey

    An article by Mark Simon, Susan Houghton, and Karl Aquino entitled ‘Cognitive Biases, Risk Perception, and Venture Formation: How Individuals Decide to Start Companies’ was published in the Journal of Business Venturing a few years back. It doesn’t talk about VC specifically, but it is probably relevant to your discussion.

    To quote: “The study’s findings tentatively suggest that individuals start ventures because they do not perceive the risks involved, and not because they knowingly accept high levels of risk.”

    The specific cognitive biases they investigated were overconfidence, illusion of control, and belief in the law of small numbers. In contrast to the latter two, they did not find a statistically significant relationship between overconfidence and perceived level of risk.

  • samantha

    If money wise on average an entrepreneur breaks even with less adventuresome folks then it seems there is nothing to lose and much to win. It has been said that the path of the start up is like compressing all the effort of an entire career as an employee into a mere three years in the hopes of gaining all that income much sooner.

    But actually I think this piece misses the point. To me being an entrepreneur is about having a “fire in the belly”, having something inside that one fervently wishes to see born into the world and that can only be born through business. Sometimes it is possible to bring one of these to life as an employee, but not so often. Then the choice is whether to give birth to this thing inside or not. Making a lot of money is secondary.

    Besides, it is a lot more intense.

  • http://rolfnelson.com Rolf Nelson

    Yes, restaurants et al are terrible industries to startup in compared with high-tech, see Scott Shane’s work for an overview.

    The narrowing to only analyzing VC-funded firms therefore greatly skews the sample. VC-funded firms are primarily in favorable industries, and furthermore I believe entrepreneurs in favorable industries that are not VC-funded are more likely to fail.

  • Pingback: Simoleon Sense » Blog Archive » Entrepreneurship & Overconfidence