The (W)hole Story: On the Origin of Gaps in Commercial Contracts

Through choice-of-forum provisions such as arbitration clauses or court selection clauses, contractual parties have the opportunity to determine where and how their disputes should be solved ex ante. They appear to be among the most basic and also essential clauses in commercial agreements. They are typically taught within the first year of a US legal education, are simple to understand, short and easy to draft. And yet, it turns out that some of the most sophisticated commercial actors routinely fail to include choice-of-forum provisions in their agreements. Indeed, only about half of all contracts publicly registered companies have submitted to the SEC as part of their disclosure requirements specify any type of dispute settlement forum, whether it be a court or arbitration (a number that is only slightly higher for international contracts than it is for domestic agreements).

In a recent article, ‘Stickiness and Incomplete Contracts’, I empirically examine the likely causes for this phenomenon. My starting point is a hypothesis previously introduced in the work of Stephen Choi, Mitu Gulati and Robert Scott. In a series of studies, they have found that the evolution of contract language can be highly path-dependent and ‘sticky’, with future agreements only rarely improving upon previous ones. Choi, Gulati and Scott identify that template-driven drafting practices are a source for the stickiness: lawyers copy-paste language from previous contracts without much reflection on the actual meaning or purpose of individual terms. The authors focus on the pari passu clause, which, in the sovereign debt context, allows a small minority holdout creditors to prevent a debt restructure that a majority of bondholders agrees to, effectively forcing the sovereign to either pay the holdout creditors in full or to default. Choi, Gulati and Scott show that pari passu clauses have long been perceived as obscure by drafters, with hardly anyone anticipating the importance they may have for sovereign debt restructuring. As such, some amount of unreflective copy-pasting may not be too surprising. In contrast, choice-of-forum provisions are simple to understand, easy to draft and forum shopping has been shown to have very important consequences for the outcome of contractual disputes, even if it happens strictly within the US federal court system.  Is it plausible that the stickiness rationale can similarly explain the presence or absence of something as basic as choice-of-forum provisions?

Using natural language processing to analyze roughly half a million contracts submitted to the SEC between 2000 and 2016, I find ample evidence to support that this is indeed the case. There are a series of findings in the paper but perhaps the most striking result relies on a comparison of the first draft circulated between the parties to the final contract: If the first draft includes a choice-of-forum provision, there is a 90% chance to see one in the final agreement submitted to the SEC. However, if the first draft does not include a choice-of-forum provision, there is only a 16% probability that the final agreement includes one. This suggests that a gap in the initial template almost always makes it into the final contract without being challenged or bargained over by either side.

There are at least two plausible interpretations for the findings. On one hand they may be a case of unintentional oversight on behalf of transactional lawyers. If true, this implies that current drafting practices reflect somewhat poorly on the legal profession and that improving the practices is likely to benefit clients in the long run. On the other hand, the results may suggest that forum choice is simply not important enough to warrant the costs of negotiating and drafting the clause. In support of this explanation, publications such as the ABA Deal Point Studies have repeatedly shown that many provisions in M&A agreements are carefully tailored to reflect the relative bargaining power between the parties. Although my study does not offer hard empirical evidence to settle this debate, some qualitative evidence may help inform our assessment.

The unanimous response in an informal survey of transactional lawyers on their thoughts about missing choice-of-forum provisions is that such an omission would be ‘negligent’ and ‘borderline malpractice’. Interestingly, even those who demonstrably left the forum undefined in their own M&A agreements report that a choice-of-forum provision would be absolutely ‘essential’ in their deals. Alternative explanations, such as the idea that parties intentionally leave the forum unspecified in order to indicate their trustworthiness and desire for an amicable solution of future disputes, were dismissed by practitioners as ‘academic’ constructs with little descriptive accuracy for transactional practice.

To be sure, nothing in the article supports that unreflective copy-pasting characterizes the majority of the transactional work that US lawyers do for their clients. Indeed, publications such as the ABA Deal Point Studies have repeatedly shown that many provisions in M&A agreements are carefully tailored to reflect the relative bargaining power between the parties. To the degree that stickiness is a problem, an intriguing question for future research is to identify the characteristics of terms that garner the most attention by negotiators, and to assess the extent to which these characteristics correlate with the terms’ economic significance. In that respect, a recent paper by Adam Badawi and Elisabeth de Fontenay is of particular interest. The authors find that the first draft of an agreement has a greater influence on the final contract for terms that are difficult to monetize, such as material adverse change clauses. In contrast, easily monetizable terms such as those on merger breakup fees are heavily negotiated and the first draft’s influence is marginal at best. As advances in computational text analysis allow us to analyze increasingly large contractual data sets with accuracy and granularity, it will be interesting to see what other patterns emerge.