Go to Database Directory || Go to Bibliography || See also Franco Ferrari, What Sources of Law for Contracts for the International Sale of Goods

Reproduced from 25 International Review of Law and Economics (September 2005) 342-349.
Copyright 2005, with permission from Elsevier

The CISG's Expansion Bias: A Comment on Franco Ferrari

Steven Walt [*]
University of Virginia

The CISG is a treaty containing a set of almost wholly default rules governing contracts for the international sale of goods to which it is applicable. Among nations that have ratified it, the CISG describes a uniform sales law: a single collection of rules that apply to sales contracts within its scope. Franco Ferrari's very useful paper emphasizes the continuing role of non-CISG law (domestic or treaty) for transactions otherwise governed by the CISG. As I read it, the paper makes two claims: (1) that the CISG is a "success" when its scope is properly understood and (2) that properly understanding the CISG's scope requires [page 342] recognizing its limited application and therefore the continuing need to resort to non-CISG law. I am more skeptical than Ferrari about the CISG's "success." My evaluation is less favorable than Ferrari, based on what I identify below as a expansion bias in the CISG's interpretation: the tendency among some courts, arbitral tribunals, and commentators to apply the CISG to transactions or aspects of transactions not clearly within its scope by relying on an ill-defined and questionable presumption of coverage. I begin with Ferrari's second claim.

Ferrari is of course correct that accuracy and prudent contract planning require recognition of the CISG's limited scope. He carefully catalogues the ways in which the CISG's application is restricted. Taken together, these restrictions make apparent that the CISG does not have the systematic and comprehensive character of a civil code. There are too many limitations that require directly or indirectly resort to non-CISG law. To show this, Ferrari in effect describes the range of circumstances in which the CISG does not apply. His description thereby identifies the circumstances in which non-CISG law controls. As Ferrari rightly notes, non-CISG law,[1] domestic or treaty, controls either when (1) the CISG does not apply at all to a sales transaction or (2) a Contracting State's reservation makes all or part of the CISG inapplicable or (3) the contracting parties opt out of all or part of the CISG or (4) does not address a particular issue in the sales contract it otherwise governs or (5) gives the contracting parties rights that effectively are made conditional on domestic law recognizing them. The economic importance of the source of law applicable to a contract depends on its effect on the price of the contract. This is a matter of the content of applicable law, not whether the CISG or domestic law governs. However, the sheer number of occasions in which domestic law controls means that it can have a significant effect on the value of the sales contract to the contracting parties. Sources of law matter economically and so Ferrari's focus on the occasions when domestic law applies is useful. The CISG's most obvious limitation is on the type of transaction to which it applies. The CISG governs contracts for the sale of goods (Article 1). By its terms, it therefore does not apply to sales of immovables, or to non-sales transactions in goods. Even as to sales of goods, the Convention does not apply when the goods are bought for consumption purposes (Article 2(a)), are certain sorts of goods (Article 2(e)), or the sale is of a certain sort (Article 2(b)). Significant is the CISG's "indirect" criteria of applicability contained in Article 1(1)(b). Article 1(1)(b) makes the CISG applicable to a contract for the sale goods when the contracting parties' places of business are in different States, the forum is a Contracting State, and its rules of private international law select the law of a Contracting State. The limitation on the CISG's application follows from Article 1(1)(b). Where the parties do not have their places of business in different Contracting States, domestic law controls only when the forum's conflicts rules select the law of a non-Contracting State.

Permissible reservations under the CISG made by Contracting States further limit the Convention's scope and make domestic law potentially applicable. To date slightly more than a third of the States ratifying the CISG have made one or more reservations to it. The impact of reservations on the CISG's scope depends on whether the nature of the reservation. Reservations that make the CISG entirely inapplicable obviously affect its scope more [page 343] than reservations that only make it partly inapplicable.[2] An Article 95 reservation to Article 1(1)(b) is of the former sort. Because an Article 95 reservation considers the reserving State a non-Contracting State, the CISG is not applicable under Article 1(1)(b) when the forum's conflicts rules select the law of the reservation State's law. In the circumstances the CISG applies only if the parties' places of business are located in different Contracting States. If not, the reservation State's domestic law applies. By comparison, reservations that make the CISG partly inapplicable allow the possibility that the CISG might ultimately apply, thereby making domestic law inapplicable. Article 92(1), for instance, allows a Contracting State to declare that it will not be bound by either Part II or III of the CISG or both. Under Article 92(2), the State making the declaration is considered a non-Contracting State with respect to the Part rejected. In a dispute between parties having places of business in different Contracting States, where only one of the States has made an Article 92 reservation, CISG is applicable under Article 1(1)(a). However, whether domestic law applies depends on which State's law the forum's conflicts rules select. If the issue disputed is dealt with by the Part reserved and the forum's rules select the law of the reservation State, Article 92(2) considers it a non-Contracting State. Its domestic law therefore governs the issue. If the forum's rules select the law of the non-reservation Contracting State, all of the CISG governs the issue, not domestic law. Complications introduced by reservations made by the forum Contracting State, although important, are secondary; they affect only the particular circumstances in which an Article 92 reservation makes the CISG partly inapplicable.

The CISG's application is yet further limited even when it applies initially. For one thing, Article 6 allows contracting parties to opt out of the CISG entirely or, subject to some restrictions, in part. Under Article 6, opting out apparently can be explicit or implicit, by contractual provision, course of dealing or trade usage. In such cases domestic law alone applies to the contract. (The forum's conflicts rules determine whether domestic law selected by the parties controls.) Further, the CISG's scope is confined even when it remains applicable to the parties' contract. Article 4 restricts the CISG's scope to entitlements of the seller and buyer "arising from" the sales contract. The restriction means that domestic law regulates the rights of third parties, as well as the noncontractual rights of the contracting parties. Under Article 4(a) and (b), unless "expressly provided for," the CISG does not govern matters of "validity" or the effect of the sales contract on title to the goods sold. The limitation also is not exhaustive. Finally, the CISG sometimes allows domestic law to control the availability of rights granted by the Convention. For instance, Article 46(1), with few restrictions, gives a buyer a right to specific performance. Article 28 requires the forum court to grant this relief when it would do so for "similar contracts" governed by its domestic law. Thus, taken together, the provisions in effect makes specific relief conditional on recognition of the relief under domestic law.

Taken together, the CISG's limitations on its scope leaves significant areas in which domestic law continues to control. Ferrari therefore is right to dismiss characterizations of the CISG as a "comprehensive code governing international sales of goods" or as "exhaustively deal[ing] with all problems," even as off-hand, informal descriptions of the product. [page 344] A small criticism here is his failure to take into account the effect of judicial or arbitral application of the CISG. These applications give a more precise sense of the contours of the CISG's scope than its language gives, either contracting or expanding it. Although Ferrari carefully notes judicial or arbitral interpretations of the CISG, he might incorporate them into the description of the CISG's scope restrictions. Doing so might show that the CISG either displaces or continues to allow domestic law to apply more frequently than the CISG's text suggests.

In fact, I suggest that judicial and arbitral interpretations over time have expanded the CISG's scope. Consider two examples in which a trend in judicial or arbitral applications arguably works to expand the CISG's scope and correspondingly displaced domestic law. One concerns burdens of proof under the CISG. Article 79(1) expressly places the burden of proof in exempting performance on the party claiming the exemption. The CISG is otherwise silent on the matter. European courts and arbitrators nonetheless have held, based frequently on "general principles" recognized by Article 7(2), that the CISG implicitly assigns burdens of proof.[3] Domestic law assignments of burdens therefore are displaced. These holdings effectively expand the CISG's scope, even though the specific assignment of burdens under the CISG appears to mimic the assignment under domestic law. A second example of expansion concerns the language found sufficient to opt out of the CISG. Although Article 6 says nothing about the linguistic form effective to derogate from the CISG, fora and some commentators have determined that the CISG controls this matter. They have found that contractual language selecting the law of a non-reservation Contracting State, without more, is ineffective to derogate from the CISG.[4] The finding means that the CISG, not domestic law, regulates the form which opting-out must take. A description of the CISG's scope based on its text would not reveal this displacement of domestic law.

This brings me to Ferrari's first claim: that the CISG is a "success" if its scope is properly understood. Obviously the assessment turns on the standard by which success is judged. If the standard is one of prevalence, the CISG certainly succeeds. Its ratification by 63 countries whose combined economies account for over two thirds of world trade make it likely that the CISG regulates significant numbers of international sales contracts. The CISG clearly is a legally salient uniform sales law. But prevalence by itself is not a compelling standard for evaluating uniform sales law because prevalent law can contain undesirable provisions. In fact, a uniform sales law containing bad provisions argues for less prevalence -- less "success." So an alternative standard of success evaluates the CISG by the desirability of its provisions.

An attractive measure of desirability is efficiency: in the case of contracts, the maximization of the joint gains from contracting. Efficiency is a particularly appropriate measure in transactions between commercially sophisticated parties, as frequently will be the case for contracts governed by the CISG. Almost all of the CISG's provisions are default rules,[5] [page 345] and default rules are efficient when their application to a contract minimizes contracting costs to parties. Because default rules apply to a range of contracts, they are efficient if they minimize aggregate contracting costs for the majority of contracting parties. The trouble with the standard is that the evidence needed to evaluate the efficiency of defaults is unavailable. The preferences of contracting parties are reliable evidence of the size of contracting costs: contracting around a default rule is evidence that the contracting parties find that the rule does not minimize their contracting costs for their contract. A default rule therefore is efficient if the majority of contracting parties prefer to have it applied to their contracts. On this count there is some anecdotal evidence that parties contract around the CISG in negotiated agreements with some frequency, and that some trade associations do the same in their standard form contracts.[6] But this says nothing about the representativeness of the parties opting out in the relevant population of contractors. Parties whose contracts are subject to the default rules might find them satisfactory and so elect to have the rules continue to apply to their contracts. Thus, observed opting out activity by some parties is consistent with default rules minimizing contracting costs for the silent majority of parties. This last consideration is important for sets of default rules, such as the CISG, that apply to heterogeneous sorts of contracts. Hence, lacking the needed evidence of representativeness, I am agnostic about the efficiency of the CISG's rules.[7]

A third standard of success is adaptability that initially efficient default rules remain efficient when their application is expanded. I will argue that there is good reason to think that the CISG does not meet this standard. To begin, assume that the CISG's default rules are efficient. An unsystematic survey of case law and commentary interpreting the CISG reveals the following case outcomes or recommendations:

  1. Burden of proof: the CISG controls the assignment of burdens of proof when it is applicable to an issue arising from the sales contract. Domestic law assignments are displaced.[8]

  2. Derogation: the CISG controls the form in which parties can contract around its application. Derogation apparently can be express or implicit. Express provisions selecting the law of a non-reservation Contracting State, without more, are insufficient to opt out of the CISG.[9] [page 346]

  3. Scope: Computer information, such as computer software, is a "good" for the CISG's purposes. For the CISG to apply, the sale of computer information need not occur as part a "hybrid" sale involving computer hardware.[10]

  4. Displacement: The CISG governs liability for damage to property arising from breach of a contract covered the treaty. It displaces domestic law rules, principally the law of products liability, that otherwise govern liability in the same circumstances.[11]

Even allowing for dispute, the sparseness of relevant case law and simple error in interpreting the CISG, these case outcomes or recommendations describe a particular pattern: a systematic trend expanding the CISG's scope beyond its text, easy inferences from text and diplomatic history. The CISG is being interpreted to cover more transactions, or issues within a transaction, than its text or inconclusive diplomatic history establishes. These interpretations may be "correct." Nonetheless, they expand the CISG's scope in the purely descriptive sense that they go beyond what the CISG's text and drafting history uncontroversially allow.

The vagueness of the CISG's provisions affecting its scope allows the pattern of expansive interpretation described above. The preamble to the CISG's call for "uniform rule," Article 7(2)'s requirement that matters governed by the CISG but not expressly settled by it are to be settled by unspecified underlying "general principles," the absence of defined terms generally, and Article 4(a)'s exclusion of matters of "validity" from the CISG's scope combine to make a range of novel applications of the CISG uncertain and controversial. The lack of specificity in the CISG allows later interpretators to apply it to types of transactions or issues that were not the focus of diplomatic negotiations or evident at the time of the treaty's ratification. For two reasons, a bias in favor of expanded scope is unlikely to produce efficient default rules for these applications. First, the incentives of commentators, and possibly courts and arbitrators, favor novelty, not efficiency. Gillette and Scott postulate that legal academics participating in the CISG's drafting sought to enhance their professional reputation from successfully drafting a treaty.[12] From the constraints of informal voting rules and widespread enactment, they predict the production of a uniform sales law containing vague default rules and multiple reservations. Slightly extended, Gillette and Scotts' postulate characterizes the preferences of legal commentators construing the CISG. Novel interpretations of the CISG that expand its scope make the treaty more important because more widely applicable. Such interpretations therefore enhance the commentator's academic reputation, and the vagueness of some of the CISG's scope provisions do not render them "out of bounds." Narrow, syllogistic interpretations are unlikely to win high marks. Hence a bias in favor of expansive interpretations in commentary on the CISG's scope is predictable. Because the interpretive preferences of commentators are unrelated to efficiency, it is only by accident that expansive interpretations recommends efficient rules governing scope.

The incentives of courts and arbitral tribunals are less univocal. The preferences of judges are likely to be some function of salary, leisure, non-reversal, and professional [page 347] reputation.[13] With a fixed salary, interpretation of the CISG can require a judge to tradeoff leisure against reputation and non-reversal. The well-documented "homeward trend" in the CISG's interpretation, which relies on domestic law to construe the CISG, makes the task of application easier for a court. The trend is consistent with leisure being more valuable to the interpreting court than the reputational gain from interpreting the CISG is an "autonomous" manner. On the other hand, sometimes a judge may value reputational gains more than leisure or the discounted prospect of reversal. In these instances, a novel expansive interpretation of the CISG's scope can enhance the judge's reputation more than the narrow, well-established construal. The arbitrator, whose income derives from arbitral fees and therefore is not fixed, is likely to be responsive to the preferences of actual and prospective arbitrating parties. Because she bears the financial impact of her findings, there is little risk that the arbitrator will supply terms inefficient for the parties' contract. On the other hand, arbitral awards, even when published, do not create rules that apply to non-parties to the arbitration. More important, the arbitrator's responsiveness to the preferences of arbitrating parties does not guarantee that the arbitrator's interpretation of the CISG's scope minimizes contracting costs for non-arbitrating parties. Hence the conclusion for both courts and arbitrators is the same as for commentators: their expansive interpretations of the CISG will supply efficient rules only by accident.

The second reason not to expect expansive interpretations of the CISG to be efficient is more basic. Their reasoning typically turns on matters unrelated to factors affecting the minimization of contracting costs to contracting parties. It often instead turns on a presumption of coverage justified by the prevalent ratification of the CISG or close analogies of the contract in question to a contract for the sale of goods. Because neither justification focuses on aspects of the contract in question affecting efficiency, it is coincidental if the law selected contains terms efficient for the contract. The best examples of the point involve questions of characterization. For instance, if the question is whether sales of computer software are properly considered sales of goods under the CISG, one might ask after the comparative risk reducing abilities of the parties with respect to nonconformities in software, the parties' propensity for moral hazard, and the size of consequential damages to the buyer resulting from nonconformities in the software. Answers to these often-asked questions help determine whether parties contracting for the sale of software prefer CISG's default rules to other sets of rules. Invoking a presumption of coverage, or analogies to paradigmatic types of goods, seems guaranteed to get the wrong answer in many cases.

A similar analysis is in order if the question concerns derogation: what contractual language suffices to opt out the CISG? To date the predominant answer is that selection of a Contracting State's law, without more, is ineffective to derogate from the CISG. This creates a "sticky" rule that makes opting out possible but somewhat more costly for the parties to engineer. The selection of a rule regulating opt outs might ask how most parties would prefer to have their contractual language interpreted, taking into account the likely allocation between them of various risks (e.g. risk of nonconformity in the goods, liability for consequential damages, fluctuations in delivery costs) and information available to a court. Whether or not a "sticky" rule of derogation would be selected, the presumption in [page 348] favor of the CISG's coverage or analogies to sales contracts for goods are irrelevant to the inquiry.

Or consider burdens of proof. Although Article 79(1) explicitly allocates the burden of persuasion in a specific circumstance, the CISG's other provisions do not. Thus, the CISG assigns burdens of proof only if the matter is regulated by "general principles" underlying the treaty under Article 7(1) or the demands of uniform interpretation under Article 7(1). Because neither Article is obviously applicable, judgments about efficiency might help guide the CISG's coverage of the matter. An efficient assignment of burdens of persuasion minimizes the sum of the cost of erroneous decisions and the costs of enforcing those decisions.[14] Considerations of the antecedent probability of an allegation being true, the comparative abilities of the parties to produce evidence, and incentives to bring false claims are relevant to the efficient allocation of the risk of error. If domestic law assignments of burdens of proof is judged generally inefficient, the CISG can be concluded to cover the assignment. If domestic law assignments generally are efficient, then the CISG does not control the matter. A presumption in favor of the CISG's coverage or analogies to sales contracts for goods is irrelevant to the CISG's scope because they do not bear on considerations implicating efficiency. [page 349]


* Tel.: (434) 924 7930. E-mail address: <sdw6a@virginia.edu>.

1. To avoid constant reference to "domestic or treaty law," hereinafter I will consider non-CISG law to be "domestic law." The term should be read broadly to refer to national law, including treaty law.

2. For a taxonomy that distinguishes between reservations based on whether the reservation considers the reserving State a "Contracting State," see Marco Torsello, Reservations to International Uniform Commercial Law Conventions, 5 Uniform L. Rev. 85 (2000).

3. See infra note 8; Franco Ferrari, Burden of Proof Under the CISG, 2000 Rev. Conv. Contr. Int'l Sale of Goods (CISG) 1; Peter Schlechtriem, Commentary on the UN Convention on the Sale of Goods (CISG) 47 (1998).

4. See Ferrari, conference paper: Joseph Lookofsky, In Dubio Conventione? Some Thoughts About Opt-Outs, Computer Programs and Preemption Under the 1980 Vienna Sales Convention (CISG), 13 Duke J. Comp. & Int'l L. 263 (2003); Clayton P. Gillette and Steven D. Walt, Sales Law: Domestic and International 38-39 (2nd ed., 2002).

5. Strictly, all are default terms since Article 6 allows parties to contract out of the CISG entirely. Where a ratifying State makes a reservation under Article 96 and the parties have not contracted out of the CISG, Article 12 restricts their ability to derogate from its contents. The parties' ability to derogate from Article 7 also may be restricted.

6. See Filip De Ly, The Relevance of the Vienna Convention for International Sales Contracts -- Should We Stop Contracting It Out?, 4 Bus. L. Int'l 241 (2003); Michael Bridge, The UK Sale of Goods Act. the CISG and the UNIDROIT Principles, in The International Sale of Goods Revisited 115, 117 (P. Sarcevic and P. Volken (eds.), 2001) (trade association forms for commodity contracts exclude the CISG).

7. For more skeptical assessments of the CISG's rules, see Clayton P. Gillette and Robert E. Scott, The Political Economy of the CISG, conference paper; for a similar assessment of Article 2 of the Uniform Commercial Code's default rules, see Alan Schwartz and Robert E. Scott, Contract Theory and the Limits of Contract Law, 113 Yale L. J. 541 (2003) 598-609.

8. See, e.g. Tribunale di Rimini (Italy), 26 November 2002, 2 UNILEX D.2002-9; Handelsgericht Zurich, 9 September 1993, 1 UNILEX D.1993-1922; supratext accompanying note 3. Cf. Schmitz-Werke GMBH v. Rockland, Lexis 12336 (4th Cir. 2002) (burden of proof as to nonconformity under the CISG and domestic law the same).

9. See, e.g. BP Oil International, Inc. v. Empresa Estatal Petroleous de Ecuador (PetroEcuador), 332 F.3d 333 (5th Cir. 2003); Oberster Gerichshof (Austria), 28 April 2000, 1 UNILEX D.2000-2007.

10. See, e.g. Landgericht Munchen, 8 August 1995, 1 UNILEX D.1995-3.1; Lookofsky, supra note 4; Amelia H. Boss, Taking UCITA on the Road: What Have We Learned?, 7 Roger Williams U. L. Rev. 167 (2001); Christina H. Ramberg. The E-Commerce Directive and Formation of Contract in a Comparative Perspective, 26 Eur. L. Rev. 449 (2001); Schlechtriem, supra note 3; John O. Honnold, Uniform Law for International Sales 55 (3rd ed., 1999).

11. See Schlechtriem, supra note 3; Honnold, supra note 10; for the contrary view, see Lookofsky, supra note 4.

12. See Gillette and Scott, conference paper.

13. See Richard A. Posner, What Do Judges and Justices Maximize (The Same Thing Everybody Else Does)?, 3 Supr. Ct. Econ. Rev. 1 (1993).

14. See John Kaplan, Decision Theory and the Factfinding Process, 20 Stan. L. Rev. 1065 (1968).

Pace Law School Institute of International Commercial Law - Last updated November 13, 2006
Go to Database Directory || Go to Bibliography