Welcome back to the next installment of Conglomerate Junior Scholars Workshop. Our featured paper today is Understanding Anti-dilution Provisions in Convertible Securities by Michael A. Woronoff and Jonathan A. Rosen, which is forthcoming in the Fordham Law Review. Woronoff and Rosen have drawn on their own considerable business and practice experience in writing this piece. Mr. Woronoff is a partner at Proskauer Rose LLP (and former partner at Gordon's and my old haunt, Skadden) and a lecturer at UCLA School of Law, where he teaches a class on venture capital. Mr. Rosen is himself a J.D./venture capitalist and is associated with Shelter Capital Partners.
Once again, I have turned to Gordon Smith to comment on this paper. Gordon has recently completed a paper on a different aspect of venture capital contracting that will be forthcoming in the UCLA Law Review. Here are Gordon's comments:
Almost every time I teach my Law & Entrepreneurship Seminar, one of my students becomes fascinated with the anti-dilution provisions used by venture capitalists, but I have always discouraged them from attempting to focus their research papers on this topic because it seems that a student who is inexperienced in venture capital would have a difficult time saying anything interesting. In this paper, two experienced practitioners of venture capital grapple with anti-dilution provisions, and they have plenty of interesting things to say. If you are new to this topic, however, you may feel that some of the big-picture issues get lost in the detailed description of anti-dilution provisions, so I will focus my brief comments on those big-picture issues and suggest that the paper still needs to travel some distance to reach its proclaimed goal.
The general concept underlying anti-dilution provisions is easy enough to understand. Begin with the concept of dilution: when a company issues new shares of equity, the existing stockholders may be diluted in one or both of the following ways: (1) they may have a smaller ownership interest in the company after the new issuance (Woronoff and Rosen call this "percentage dilution"); and (2) their equity claims may have decreased in value as a result of the new issuance (Woronoff and Rosen call this "economic dilution"). Venture capitalists typically protect themselves against percentage dilution via rights of first refusal -- contract rights that allow the venture capitalists to purchase a specified percentage of any new equity issuances, usually with the goal of maintaining their pre-existing ownership interest. Anti-dilution provisions are aimed at preventing (or muting) economic dilution by requiring the issuance of additional shares (according to a formula specified in the contract) to investors who otherwise would be faced with economic dilution.
As Woronoff and Rosen observe, anti-dilution provisions come in various flavors, and the authors set for themselves the task of explaining the variation. Fortunately for us, they have a theory:
This article provides, for the first time, a comprehensive framework to understand various anti-dilution adjustments and to reconcile the apparent inconsistencies among various clauses. This framework is based on the observation that the scope and extent of different types of anti-dilution provisions can generally be understood as a rational response to the nature and level of the information barriers and agency costs typically confronted in the particular circumstances, and the existence or absence of other mechanisms to address these issues. (31)
So, how well does their theory hold up? In my view, the jury is still out. Consider their argument regarding venture capital, where the "information barriers are severe," thus causing great uncertainty in the pricing of investments. In this environment, the authors claim that anti-dilution provisions act as an ex post adjustment of the prior pricing agreement, using new information gathered from the present pricing negotiations. Evidence of this purpose is found, they assert, in the form of the anti-dilution provisions used in venture capital deals. In every case, those provisions use the initial investment value -- rather than the current market value -- of the preferred stock as the basis for triggering the anti-dilution protections. This is consistent with the notion that the parties are attempting to adjust the results of their prior negotiations, but it is also consistent with the fact that finding a current valuation for private companies is incredibly difficult. And drafting an anti-dilution provision that relied on a current valution that was independent of the negotiated valuation would not only be costly and cumbersome, but potentially quite disruptive to the enterprise if new and old investors begin fighting about the real value of the company.
Moreover, Woronoff and Rosen's theory cannot explain the deep structure of anti-dilution provisions in the venture capital context. Generally speaking, venture capital deals employ one of two types of anti-dilution provision: the ratchet provision or the weighted-average provision. For the details on how each of these provisions works, I refer you to the paper. For present purposes, suffice it to say that the vast majority of deals use the weighted-average provision, even though (according to the authors) the ratchet "is designed to fully protect against inaccurate valuation arising from information barriers." (20) To explain the dominance of the weighted-average provision, the authors must reach beyond their theory of information barriers and price correction. In other words, the observed practice does not seem to be driven by the theory, but by other considerations.
Again, readers are encouraged to comment on this paper by participating in the comments section to this post or by emailing the authors directly. No anonymous comments, please.
TrackBack URL for this entry:
Links to weblogs that reference Conglomerate Junior Scholars Workshop: Michael Woronoff and Jonathan Rosen on Convertible Securities:
1. Posted by Gordon Smith on August 24, 2005 @ 10:38 | Permalink
If any of our readers are interested in learning more about the terms of venture capital contracts generally, I encourage them to consult Brad Feld's excellent series of posts at Feld Thoughts. The wrap-up post, which contains all of the relevant links, is here.
2. Posted by William Henderson on August 24, 2005 @ 13:49 | Permalink
First, I would like to thank Gordon, Christine, and Vic for providing us with an new vehicle for reading and reviewing business-law scholarship. As I emerge from my first day of teaching Corporations, this is a welcomed and highly relevant diversion. Later in the semester, I will spend a day or two on convertible preferred stock because these securities are indeed the lifeblood of new businesses. And they are also an area where a transactional lawyer can (a) be very creative, and (b) add tremendous value. (Gordon’s casebook has a case study on Red Hat, Inc., which provides an vehicle for discussion.)
Michael Woronoff is a distinguished practitioner, and Jonathan Rosen works for a private equity fund (which presumably enters into lots of deals with convertible securities). So I don’t want to pretend that my academic views carry special weight. They don’t. They have tremendous firsthand experience that I lack. Nonethless, there are a couple of points I am not very clear on.
Let me start with a simple example. Founders have a great idea but insufficient funding to actually create their business. Banks won’t lend them money (not yet, anyway) and it is too early (and expensive, because of compliance costs with SEC regulations and state Blue Sky laws) to sell stock directly to the public. So they knock on the door of venture capital funds asking for money. But VC’s are pure capitalists; they are (or should be) experts at dissecting business plans and conducting the optimal level of due diligence.
From my perspective, it is unclear what the “information barriers” are at the time the convertible security is issued (pp. 13-14). It seems to me that lots of things are relevant at this stage: how good is the idea? What is the probability of success? What is the range of the economic upside? Are any other VC’s in play? Does management have a track record?
So this is my first point of confusion. Sure, there is residual uncertainty, but I don’t see any “information barriers” beyond the lack of a crystal ball. The anti-dilution protections are for the benefit of the VC’s for all the reasons outlined in Woronoff’s and Rosen’s article. But the conversion formula is (or should be) a deal point that apportions risks between the founders and the VC’s. It is (or should be) a function of negotiating leverage. (Clients who get into deals with convertible securities and anti-dilution boilerplate, which Woronoff and Rosen suggest is common, see p. 3 & nn. 9-10, may be getting under-served by their lawyers.)
If I am a VC, and I can get a “full rachet” provision to preserve or maximize my economic stake (or provide me with hold-up leverage if the deal does not unfold according to the founders’ optimistic projections), then I am going to demand one. I want the founders to bear the loss.
I agree with Gordon that the lack of current valuation provisions makes good sense ex ante. Why set yourself up for contentious bickering? There are only three outcomes that matter: (1) the deal goes off as advertised, and everyone is happy, (2) the deal goes better than expected, and the founders want to avoid the VC’s getting too much of this upside, or (3) the deal goes worse then expected (e.g., more money is needed), and the VC’s need to utilize the contractual provisions that protect their economic interests. In times such as these, anti-dilution provisions become very important.
Which brings me to my second point of uncertainty. Anti-dilution provisions often affect voting rights. Some convertible securities have voting rights from the get-go, some get voting rights upon the occurrence of certain events (e.g., failure to meet certain benchmarks), and others require conversion to common stock to get voting power (which might be necessary to throw out bad management). The formulas for voting rights also vary quite a bit–1:1, 2:1, 5:1, etc.
Perhaps I spend too much time in my Corporation class (which includes agency, partnership, and LLC’s as well) talking about control. But maybe I don’t. Control of the corporation (or a contractually specified mechanism for gaining control) affects risk, which in turn affects the economics of the deal. I know the anti-dilution provisions work in concert with other (perhaps better) control features, such as rights-of-first refusal and restrictive covenants, but anti-dilution provisions definitely can affect (and reapportion) voting rights. I would have like to see Woronoff and Rosen contextual how voting fits into their theory.
That said, thanks for the terrific read. I may use some excerpts of this article in my class; it has terrific (and clear) examples.
3. Posted by Michael Guttentag on August 24, 2005 @ 16:23 | Permalink
I learned a lot reading the Woronoff and Rosen paper, Understanding Anti-dilution Provisions in Convertible Securities. Moreover, I think the question of differences between public and private transaction practices are of critical importance in understanding how public and private markets work. My observations are closely related to those of William and Gordon, so I hope they are not unduly repetitive.
What I found frustrating about the paper was the lack of a convincing and clear statement of hypotheses and of how hypotheses would be tested. For example, I thought use of the terms information barriers and agency costs was much too general to generate testable hypotheses. The term information barriers, as used here, included both information asymmetry and valuation uncertainty. But I’m not convinced that both of these components of information barriers are reduced in public transactions as compared to private transactions. Based on my experience, there is less information asymmetry but more valuation uncertainty in private transactions as compared with public transactions. Second, I’m not quite sure what the author’s believe is the correct way to characterize an efficient set of anti-dilution protections. Maybe the goal should be to minimize transaction costs rather than to provide the most precise amount of anti-dilution protection, since once rights are delineated, securities can be priced accordingly. Stating hypotheses more precisely might have addressed these issues. Finally, I would have liked to see some effort to formally test the hypotheses once they were clearly stated.
As a side note, there were some statements about how markets work that struck me as incorrect. On page 4 they say: “Information barriers exist at the time the investment in the convertible security is initially made, and increase the risk that the investor will overvalue the enterprise.” Really? Are investors that naïve? Similarly, I didn’t know what it meant to say “Agency costs are imposed at the time of the dilutive event,…”
It is always nice to learn a lot when reading a paper. Good luck.
4. Posted by Kate Litvak on August 24, 2005 @ 17:15 | Permalink
Mike Guttentag is exactly right. Guys, your paper is much too vague and too sloppy for an academic article. Much, much too vague. It may work fine as a practitioner-oriented descriptive paper if you take out any claim of “explaining” anything (though I am not sure what you are adding much to the existing tons of practitioner literature on this subject). But it just doesn’t work as an analytical academic paper. Here is why.
We already know that in the Coasean world of perfect information, proper incentives, and zero transaction costs security design and contract terms don’t matter. In that world, there would be no convertible securities and anti-dilution provisions, or they would be put in randomly with no effect. So, when you are saying that “we observe anti-dilution provisions and convertible securities, therefore there must be some information/agency problems here,” you are not saying anything new.
The only way to actually move the ball is to say which *exactly* information/incentive problems are responsible for which *exactly* features of dilution provisions. The academic finance scholars have put out dozens and dozens of very detailed papers outlining the most minute details of the impact of information/incentive problems on the design of VC contracts. They have addressed the issue of convertibles in painstaking detail. You don’t seem to be aware of that literature (at least, you never cited any of it). And yet, this body of work is critical for you. If you go to SSRN and type in “convertible securities and venture capital” for both abstract and title, you’ll see lots of very, very precise papers on the subject that you are trying to address.
There are lots of other problems too. Since your paper is so vague, you don’t actually formulate any testable hypotheses. You merely point that things are different in venture capital than in public corporations and attribute it to “information problems.” But the VC industry employs a wide variety of anti-dilution provisions, and any theory of “information problems” would have to explain this intra-industry variation. Again, there is a vast body of empirical literature studying VC contracts, of which you are apparently not aware, since you never cited any of it. See Kaplan and Stromberg (not the Kaplan you’ve been citing…), Gompers and Lerner, Cumming…
From this empirical literature, you could pull out the patterns of the actual use of anti-dilution provisions by VCs and then ask whether these patterns fit your theory. Once you pose question this way, you will notice that your theory is so vague that you can’t actually suggest any patterns. Would older, bigger, better VCs have stronger or weaker dilution protections? Would it depend on who the next-round VC is? Would it depend on the quality of the entrepreneur? How would dilution protections interact with other contractual terms that are *also* aimed at reducing info asymmetries and agency problems? E.g., convertible securities are used precisely to deal with info asymmetries and agency problems. Does it mean that a VC who uses convertibles, rather than a combination of straight debt and equity, would be less likely to have a draconian dilution protection – because they are substitutes? Or more likely – because they are compliments? Would selection problems play into this pattern?
I could go on for ages, but I think I’ve made the main point clear. Your paper is very thorough and careful, but it is simply not a genre of academic literature. You can’t easily turn a practitioner-type descriptive paper into an analytic academic paper.
First: Get an academic economist as a co-author and try to develop a more detailed model of dilution protections. An economist will list a dozen of possible information problems and another dozen of incentives problems, and you will have to decide which of them you think are the most important in the real world and incorporate them into the model. It is mathematically impossible to incorporate everything, so you’d have to be very picky. This will be very painful, but you may come up with something interesting because you have a wealth of real-world experience that most VC scholars do not have.
Second: Forget the model and do an empirical study instead. You probably have enough contacts in the industry to get a decent sample of VC contracts (about a hundred would be a minimum). Read them, code them, see how dilution protections are distributed across VC funds. See whether dilution protections are correlated with other terms of funding agreements, with VC characteristics, industry of a funded company, etc. You may need to get an empirical economist or academic finance person as a co-author for this.
Either option would be very interesting. As things stand now, it’s just a practitioner paper masquerading for an academic paper.
Sorry for being so blunt. I am certain that you can do a lot of very good work if you get the right co-authors. If I didn't think that way, I would not be spending time writing all this.
5. Posted by William Henderson on August 24, 2005 @ 20:16 | Permalink
Well, I think we have proven that virtual workshops on the Web are not necessarily puff forums.
6. Posted by Michael Woronoff on August 25, 2005 @ 8:10 | Permalink
Thanks to Christine, Gordon & Vic for the opportunity to present our paper, and to the commentators who took the time to read it. We appreciate both the supportive comments and the suggestions. As you might suspect, I have a few thoughts about the comments.
I think many of the comments arise from the fact that the commentators attribute a much more ambitious goal to us than we actually had. There just isn’t much good literature (academic or professional) discussing anti-dilution provisions globally in any sort of rigorous detail. Yes, you can find a plethora of practitioner articles discussing the difference between ratchet and weighted average anti-dilution protection, but they all say about the same thing, most of which is how to calculate the different formulae and why ratchet is evil. Very few of these articles even acknowledge that there is another world of convertible securities outside the VC world. (Yet many of the difficulties I have encountered in this area in practice occurred when someone who practices primarily in one area tries to do a deal in the other.) Our goal was to provide a way for practioners and academics to think about anti-dilution provisions across all worlds to help better understand the provisions in general. It's not necessarily the only way, or perhaps even the best. It is, however, the first--and, as we note in our paper, academics have been searching for any comprehensive theory for a long while. We hope others explore the issues with our positions and that any discussion that ensues provides a better framework than the one we have presented.
So in my view, explaining why people actually do stuff is not really necessary. I admit we do get into that area a bit in attempting to explain why the framework may work, but as far as my primary purpose is concerned, I don't care if an invisible giant whispers in a lawyer's ear which provision to use. (Memo to commentators: I don't really believe this happens.) The main question, which Michael Guttentag and others touched on, is: does our model adequately predict behavior. I of course disagree with Michael that we didn't clearly state our framework or that it is difficult to test. I'll let others be the judge. In any event, we’ve further clarified this a little in the final version of the article being published.
With respect to Michael's comment that there is less information asymmetry but more valuation uncertainty in private transactions than in public transactions: Since, the relevant portion of our paper is discussing issuances by large public companies with active trading markets, I would argue that, absent fraud, the market price of the stock of these companies incorporates much more information than is available in the VC context, so I think I disagree with the premise.
With respect to Gordon's comments, we do address the overwhelming use of weighted-average versus ratchet anti-dilution in part III.B.1.b of our paper. While there are surely other potential explanations, it does seem to me that the fact that the ratchet method doesn't just fully protect against inaccurate valuation, but also adjusts for other events (say general changes in the market) makes this provision typically unacceptable as too much baby with the bath water. When analyzing our framework, however, the question of which form of conversion price based price protection VCs use is less interesting than the question of why price protection is so important in VC deals but almost irrelevant in most forms of public deals. This is a question few other commentators have dealt with. We also address (although not as fully) his point that coming up with a current valuation for private companies is incredibly difficult. I've got some more thoughts on this but will leave them to another day.
For William Henderson, good luck with the class this semester. To address your comments, I recommend two articles, both of which we cite: First, George Triantis's "Financial Contract Design in the World of Venture Capital," 68 U. Chi. L. Rev. 305 (2001). This is one of many articles that can introduce you to some of the information barrier issues. Second is Bill Klein's, "The Put-Up-or Shut-Up Strategy in Business Negotiations," 17 U.C. Davis L. Rev. 341 (which we cite numerous times to express some of the same points you do).
With respect to Kate Litvak's comments, perhaps I'm being defensive (happens to the best of us), but they sounded a bit, well, harsh in tone (though not as harsh as she was with this guy http://volokh.com/posts/1124490164.shtml (scroll down)). While we do cite her in the article, I of course disagree with most of her comments. Without going into everything, I do think our paper does a pretty good job of telling which anti-dilution provisions are typically used when certain problems are more likely to be present. She thinks we fail to "say which *exactly* information/incentive problems are responsible for which *exactly* features of dilution provisions." Unfortunately, our model is pretty much binary (information barrier vs. agency cost) so we don't need to get much more specific. Time will only tell if we have contributed anything useful.
I'll chalk our differences up to a belief that we just come at this from different places. Perhaps her point of view is driven by her apparent bias (expressed here: http://prawfsblawg.blogs.com/prawfsblawg/2005/08/why_size_matter.html) against having survey type papers appear alongside more serious work. Perhaps not. I will say that other academics who read the paper either didn't share Kate's view or were being incredibly kind to us. I hope that either Kate or other academics trained in economic analysis will take up the challenge of developing a more detailed model to analyze anti-dilution provisions. That exercise would add valuable information to the discussion.
I do think one point she makes is somewhat unfair and encourage readers to run the SSRN search she suggests. Rather than "dozens and dozens" of papers, if you delete doubles and the two classroom case studies, I found barely over a dozen (15 or so). Most of these deal with anti-dilution issues briefly or not at all, and at least one of them we actually cite. And, to anticipate the potential answer, other combinations of search terms won't help that much. Perhaps surprisingly, we did indeed search SSRN, among various other sources (pretty extensively), and read what we found. I think the footnotes in the article support this contention.
7. Posted by Pat Lee on September 2, 2005 @ 18:00 | Permalink
Good VCs do thorough diligence and should know the business as well as the founders, when they invest, or even better because good VCs have broad and high level connections. Information barrier is an issue only for dumb VCs who invest in businesses they don't understand or do not do their homework.
Neither the VC nor the founder can know whether the business can make it. So information barrier is nonsense. Anti-dilution clause is only a function of who has the upper hand to protect their interest, i.e., how desperate the VC wants to invest and how desperate the company needs the money.
The exceptions are evil VCs taking advantage of inexperienced founders, and dumb VCs who invest in businesses they don't understand. In those cases, the businesses most likely fail, so whatever anti-dilution there is won't matter.
8. Posted by Michael Woronoff on September 3, 2005 @ 15:09 | Permalink
Thanks for the comments. A few thoughts:
1. You apparently confuse our term “information barriers” for the more commonly used “information asymmetry.” For the difference see footnote 15 of our paper. Our term is broader, and acknowledges and takes account of the fact (which you note) that neither side may know whether the business can make it.
2. That being said, even the smartest VCs in the world have closed deals only to have a customer suddenly cancel or delay a program, a key employee quit, a beta product delayed, the cost of a key input increased, etc.--all in circumstances where the Founder had at least an inkling this might happen. It is simply not accurate that information asymmetries don’t exist for good VCs (no matter how good or broad their connections or how much due diligence they conduct).
3. With respect to your comment that: “Anti-dilution clause is only a function of who has the upper hand to protect their interest, i.e., how desperate the VC wants to invest and how desperate the company needs the money.”
First, we acknowledge the importance of bargaining power a few times in our paper. Second, we cite Bill Klein's, "The Put-Up-or Shut-Up Strategy in Business Negotiations," 17 U.C. Davis L. Rev. 341 numerous times in discussing anti-dilution protection as a way to protect the VC’s interests. Finally, bargaining power alone can’t be the explanation to the existence or absence of anti-dilution clauses, because even during the headiest times of the market, the hottest start-up companies still generally had some anti-dilution provisions in their securities.
4. Finally, I remain fascinated that you, like the other commentators, are so focused on anti-dilution in the VC world. Perhaps this is the result of this being a virtual workshop conducted on the internet. Perhaps not. In any event, as I mentioned in my previous post, the focus misses one of the things we were getting at.