Are International Merchants Stupid? Their Choice of Law Sheds Doubt on the Legal Origin Theory


  • Stefan Voigt

    Corresponding author
    1. Institutional and International Economics, Department of Business Administration and Economics, Philipps University Marburg
    Search for more papers by this author

  • I thank Anne van Aaken, Lorenz Blume, Michael Ebeling, Fernando Gómez-Pomar, Janina Satzer, Nguyen Quoc Viet, and Jan Wagner for stimulating discussions; Henry Hansman, an anonymous referee of this journal, and, in particular, Giesela Rühl, for excellent written comments; Natacha Pinon from the Research Department of the International Court of Arbitration for some data; and Manfred Heider from the International Arbitral Center of the Austrian Federal Economic Chamber for valuable background information. I am grateful to participants of the Sixth Workshop on Institutional Analysis at Pompeu Fabra (Barcelona) as well as the Law & Economics Workshop at the University of Kassel for further suggestions. I bear the sole responsibility for all remaining errors.

*Chair for Institutional and International Economics, Department of Business Administration and Economics, Philipps University Marburg, Barfüßertor 2, D-35032 Marburg, Germany; phone: +49-6421-282 3080; email:


In economics, there is currently an important discussion on the role of “legal origins” or “legal families.” Some economists claim that legal origins play a crucial role even today. Usually, they distinguish between common-law, French, Scandinavian, and German legal origin. When these legal origins are compared, countries belonging to the common-law tradition regularly come out best (with regard to many different dimensions) and countries belonging to the French legal origin worst. In international transactions, contracting parties can choose the substantive law according to which they want to structure their transactions. In this article, this choice is interpreted as a revealed preference for a specific legal regime. It is argued that the “superiority-of-common-law view” can be translated into the hypothesis that sophisticated and utility-maximizing actors would rationally choose a substantive law based on common-law tradition, such as U.K. or U.S. law. Although exact statistics are not readily available, the evidence from cases that end up in international arbitration courts (such as the International Court of Arbitration run by the International Chamber of Commerce in Paris) demonstrates that this is not the case. Hence, this evidence sheds some doubt on the superiority-of-the-common-law view.

I. Introduction

Until recently, the conjecture that the legal origins of a country determine the quality of its government spread like an epidemic in economics. Although the debate concerning the relative merits of common versus civil law had been going on for decades and legal comparativists have traditionally distinguished between a number of legal systems or legal families, the literature was given a real boost with a number of papers authored by La Porta, Lopez-de-Silanes, Shleifer, and Vishny (LLSV) in the late 1990s (1997, 1998, 1999). Indeed, LLSV has become somewhat of a brand name by now. The number of citations is indeed stunning: on April 6–7, 2007, the academic search engine “” contained 1,930 electronically available papers that had cited the 1997 paper. The 1998 paper was cited 2,914 times, and the 1999 paper still a very noteworthy 1,040 times. These numbers are truly astonishing. For comparison, Coase's (1960) paper on social cost, often considered the most cited economic paper of all times, had 3,469 citations on that same date.

Distinguishing between common law, socialist law, and three different civil-law families (French, German, and Scandinavian), LLSV find (1999:261):

Compared to common law countries, French origin countries are sharply more interventionist (have higher tax rates, less secure property rights and worse regulation). They also have less efficient governments, as measured by bureaucratic delays and tax compliance, though not the corruption score. French origin countries pay relatively higher wages to bureaucrats than common law countries do, though this does not buy them greater government efficiency. French origin countries fall behind common law countries in public good provision: they have higher infant mortality, lower school attainment, higher illiteracy rates, and lower infrastructure quality. . . . As predicted by the political theory then, the state-building intent incorporated into the design of the French legal system translates, many decades later, into significantly more interventionist and less efficient government, less political freedom, and evidently less provision of basic public goods.

Hence, the results are at least as stunning as the number of citations. They can be summarized as the “superiority-of-the-common-law view” or, equivalently, as the “inferiority-of-the-French-law view.”

This article relies on a very simple idea to test some aspects of this view empirically. In international trade, the contracting parties are free to choose the substantive law that suits their interests best. We assume that rational actors are interested in the best possible ex ante protection of their property rights. If French law does not adequately protect private property rights, we would not expect international traders to choose it as their substantive law. Correspondingly, if common law offers them a better basis to conduct business, we would expect them to choose one substantive law out of this legal family to structure their interactions. In this article, I analyze a subset of all cases in which the parties have the right to choose substantive law, namely, those cases containing an arbitration clause. The bulk of international arbitration cases is managed by only a dozen organizations spread around the globe. It is hence, at least potentially, quite easy to get a complete picture. Yet, almost all these organizations are reluctant to publish detailed data concerning choice-of-law questions. Still, the little available empirical evidence is sufficient to shed some doubt on the hypothesis that follows from the legal origin discussion. American law is chosen less frequently than expected. On the other hand, French, but also Swiss, law are chosen more frequently than could be expected. On the basis of the available empirical evidence, the following conclusion seems warranted: if merchants choose the substantive law that suits their interests best and U.S. law is chosen less frequently than French or Swiss law, this indicates that U.S. law is perceived as less efficient than its competitors.

The article is organized as follows. Section II describes the legal origin hypothesis in a little more detail, spells out some of the underlying assumptions as well as the implications, and criticizes some aspects of the argument. Section III presents some of the relevant institutions and organizations used by international merchants to structure their interactions and contains a description of their preferences regarding substantive law based on their revealed preferences. Section IV contains a discussion of the results. Section V concludes.

II. The Legal Origin Hypothesis: Assumptions, Effects, Implications, Critique

The most important claim of the legal origin discussion as presented by LLSV seems to be that the legal past of a country determines the present quality of its government. This section serves to present this argument in greater detail, to spell out some of the assumptions and implications, and also to criticize the argument.

In their paper on the quality of government, LLSV (1999) basically distinguish between three law families, namely, socialist, common, and civil law. That socialist law tends to be correlated with an interventionist and grossly inefficient state is in line with our intuitions. What is more interesting is the difference between common- and civil-law families. In civil-law countries, it is parliamentarians that can “make” law, whereas in common-law countries, it is the judges who “find” it. Due to the doctrine of stare decisis, precedents are a source of law in common-law countries. In civil-law countries, there is no equivalent doctrine and precedents do not have the binding effect that they enjoy in common-law countries. That these differences in the production of law could have important consequences for the ensuing quality of law is a conjecture that has been discussed for decades.

In the 1960s, for example, Hayek (1960) and Leoni ([1961] 1991) argued that common law would be superior to civil law because the latter can be radically changed overnight, whereas the former would be more stable. This would, in turn, allow private actors to form long-term expectations and to act on them, that is, make long-term investments that should, ceteris paribus, generate higher growth rates in common-law countries. In the 1970s, it was argued (Rubin 1977; Priest 1977) that common law was more efficient because parties that could deal with resources more efficiently would resort to the courts until precedent was changed to the more efficient allocation of resources.

Legal scholars seem to agree that the difference between the two families is often largely exaggerated. Posner (2002:38), for example, describes the two traditions as “convergent” (for a similar evaluation by comparativists, see, e.g., Zweigert & Kötz 1998). Although differences between the two families seem to be disappearing, in particular with regard to substance and outcomes, LLSV's results seem to imply that a country's present is largely determined by its membership in a specific legal family. Yet, LLSV dig a bit deeper with regard to civil-law origins because they distinguish between French, German, and Scandinavian families. We have already cited the results concerning French law. With regard to German legal origin, LLSV (1999:262) find that “these countries are rather similar to those of common law origin.” With regard to Scandinavian legal origin, they find that they are sharply more interventionist though not less efficient. These results are interesting—and potentially undermine LLSV's more general argument: if the general distinction is between common and civil law and the more fine-grained one within the civil-law family, then one would expect all members of the civil-law family to do worse—or better—than the other families. That two out of three members of the civil-law family do just as well as common law might be an argument in favor of the more fine-grained distinction but lets the more general distinction between common and civil law appear questionable.

The somewhat contradictory LLSV results may be a consequence of the difficulty in precisely and consistently coding countries' legal systems. Several such difficulties exist.1 First, they must assume that legal transplants are possible, no matter whether transplantation is by conquest, colonization, or voluntary adoption. Former British colonies are almost always coded “English” legal origin,2 former French, Spanish, Portuguese, and Belgian colonies always “French” (except where they are socialist as is the case, e.g., for Vietnam). This coding appears plausible, but questions arise nevertheless: If the effects of legal systems are “very long term,” should the countries not be coded according to their precolonial system? And if the very long-term effects are so important, why not code some of the formerly socialist countries in central and eastern Europe as German instead of socialist?

Second, LLSV do not sufficiently distinguish between legal origin and a country having been a colony. The former British and French colonies obviously play an important role in the results. However, having been a British or French colony and having a British or French private-law system is not necessarily the same thing. It is, at least theoretically, conceivable that a country has adopted French law without ever having been a French colony. As Persson et al. (2004) recently found, “French colonial origin is associated with less corruption, counteracting the positive effect on corruption of having a French legal system.” Corruption is, of course, only one of the many aspects dealt with by LLSV. Yet, Persson et al. argue that it made a difference if a country has a French legal system only or if it was—in addition—a French colony. If the French exported their implementation know-how, at least one variable of interest seems to have had offsetting effects. Joireman (2004) recently found that a sample including all common-law countries does not score significantly better than countries of other legal origins with regard to the rule of law. This was only the case if the sample was reduced to former colonies. This is, thus, another study confirming the relevance of having been a colony. A more fine-grained mapping (legal origin? colonial history?) might thus be in order.

Third, LLSV may not sufficiently take into account the differences between the British and the French as colonial powers. Zweigert and Kötz (1998) assert that the British did not try to replace Islamic, Hindu, or unwritten African law. In India, the English courts were instructed to apply Islamic or Hindu law depending on the religion of the parties in cases of inheritance, marriage, caste, and so on. In Africa, judges were to apply English law only to the extent that local circumstances permitted. The British thus did not insist on the material content of their law but did export the procedures, which enabled the emergence of a variety of common laws after the colonies had become independent.

The French, in contrast, strove to “improve” people in the colonies and lift them up to French standards of civilization (Zweigert & Kötz 1998). They thus attempted to implement the material content of the Code Civil in all their colonies, even if there were serious conflicts between it and, say, Islamic laws. It could thus be argued that it was not French or civil law, per se, that was inapt to induce economic growth and development, but the attempt of the French to implement the material content of their law even against resistance, whereas the English refrained from such attempts.3 Whereas the French (tried to) export material law, the English confined themselves to the export of procedural law, which could be “filled up” according to local custom and tradition. This would, then, indicate that the French approach was much more likely to create conflicting property rights systems than the British approach. In other words: successfully transplanting procedural law seems to be much easier than successfully transplanting material law. Although we have identified some flaws in the approach of LLSV, it has nevertheless enabled us to produce a hypothesis, namely, that the French attempt to implement their material law has had far-reaching effects on the quality of government that remain significant to this day.

One could also argue that it is not origin that counts but the way in which law is transplanted. This hypothesis has been promoted by Berkowitz et al. (2002). They argue that the way the law was initially transplanted and received is a more important determinant than affiliation with a particular legal family. They claim that the origin of a legal system as identified by LLSV (1998, 1999) might be a good predictor for what I would call de jure legality, but not for the factual implementation of the law. For de facto legality, the way the law was transformed according to the specific situation of a society, whether it was imported voluntarily or enforced by a colonial power and so forth, would be much more important. According to their findings, countries that have developed legal orders internally or adapted transplanted legal orders to local conditions and/or had a population that was already familiar with basic legal principles of the transplanted law have more effective legality than countries that received foreign law without any similar predispositions.

Glaeser and Shleifer (2002) have endogenized the choice of legal systems. Comparing England with France, they argue that the different structural characteristics of both systems could have been adapted to the different environments; both choices might thus have been efficient. One environmental condition receives special attention, namely, the amount of bullying observed in a country. They (2002) write: “In contrast, the transplantation of rules designed for a system with little bullying into a system with a greater amount can lead to the breakdown of the rule of law.” The French system is better suited for countries with a high amount of “bullying” than the English system. Given that “bullying” was higher in the former colonies, would this not imply that the transplantation of French law is better able to protect the rule of law than English law? But is this implication not entirely at odds with LLSV?

To sum up, although their results are quite impressive, the story told by LLSV suffers from a number of shortcomings. The coding of some countries appears questionable and possible interdependencies between legal origins and former colonies are not explicitly dealt with.

III. Revealed Preferences Regarding Preferred Substantive Law

Section II focused on a description of the LLSV approach toward legal origin and its critical evaluation. The critique was largely on a conceptual level. In this section, the LLSV theory will be approached from a different angle, namely, by analyzing what substantive laws are chosen by sophisticated actors when they can freely choose their most preferred law. This can be interpreted as an empirical test of the LLSV view of the world. If civil law and, in particular, French law is clearly inferior to common law, then rational actors who have the possibility of opting out of French law—and into a more efficient legal system—should be expected to do so. Whether they actually do so will be analyzed in this section. Before getting to the numbers, some decisions that the contracting parties usually make are mentioned. If the contracting parties originate from two different legal regimes, they will usually choose (1) a substantive law and (2) a forum in case a dispute arises. Here, we analyze only cases in which the contracting parties opt against state courts and in favor of private arbitration (although a substantive law can also be chosen if they opt in favor of state courts).

There are a number of reasons why nonstate court arbitration is frequently preferred over state court arbitration in international contracts.

  • 1When private-law subjects from different jurisdictions negotiate the details of a contract, they are often afraid of the so-called home bias of state courts, that is, the tendency of state courts to decide in favor of the party originating from their jurisdiction.4
  • 2Many states are reluctant to enforce the decisions of courts of other states as this is perceived as an undue acceptance of other legal systems and would involve an—unwanted—abdication of one's own sovereignty. But all states that have ratified the New York Convention have pledged to enforce decisions made by nonstate courts. Ex ante, enforcement is thus more likely if nonstate court arbitration is chosen.
  • 3Past a certain threshold concerning the monetary claims involved, nonstate arbitration is cheaper with regard to out-of-pocket costs.
  • 4The sequence of courts is often reduced to one in nonstate arbitration, which means that judicial decisions can be reached faster than in state courts. This increases speed, which is highly valued by businesspeople.
  • 5In private arbitration, the judges do not necessarily have a legal background but can be engineers, physicists, and so forth. Many businesspeople believe that it is an advantage to have issue experts judging their cases.5
  • 6Arbitration awards usually remain unpublished. This allows the losing party to save face. It could increase the probability of retaining a business relationship with the winning party. More generally, the losing party need not expect to suffer losses in reputation that would make it more difficult to enter into contracts with other—third—parties.

All these reasons explain why a major share of international contracts do not rely on state courts at all, but on nonstate arbitration that is often administered by specialized organizations such as international chambers of commerce.6 Opting out of state court administered jurisdiction does, however, not imply opting out of substantive state law. Although it is theoretically possible to do so and to refer, for example, to the lex mercatoria, more than 80 percent of those international contracts that contain a choice-of-law clause are concluded on the basis of the substantive law of a state (International Court of Arbitration 2004).7 By now, it is well known that all contracts are by necessity incomplete, that is, it is impossible to anticipate all possible contingencies completely and to specify the legal consequences accordingly. If a situation arises that has not been dealt with in the contract, the chosen substantive law becomes relevant: it serves as a subsidiary law that is to “fill the gap” left in the contract. Derains (1995:10) claims that it is “not possible to overemphasize the importance of the law applicable to the contract.” The question then is: What factors determine its choice? This is the question that will be dealt with in the next paragraphs.

From an economic point of view, one would expect each of the contracting parties to propose that law to its contracting partner that promises the highest expected benefits to itself. Aspects that possibly play a role here include:

  • 1The familiarity with a private-law system. A high degree of familiarity involves a low degree of additional transaction costs that have to be incurred.
  • 2The quality of the law in the sense that some legal order promises to regulate specific transactions in ways that suit all contracting partners better than other legal orders, for example, because it is highly developed.
  • 3The predictability of decisions made by arbitrators based on the respective law. The predictability, in turn, depends on the precision of the relevant laws. Predictability further depends on the stability of the relevant law.
  • 4The perceived neutrality of the relevant laws. Ex ante, parties do not know whether they will ever be plaintiffs or defendants, which means that they have incentives to consent to a law that does not unduly favor one of the parties.

Suppose all available substantive laws were of equal quality and there were neither differences in predictability nor in perceived neutrality. Under these assumptions, we would, due to the transaction cost argument spelled out as item 1 above, expect all contracting parties to propose the law of their own country as the law applicable to the contract. If there is a power asymmetry between the parties, we can thus expect that the “home” law of the more powerful party will be chosen as the applicable law. If there is no power asymmetry, it seems reasonable to expect that the contracting parties will agree on some law of a third state that satisfies items 2 through 4.

Comparing legal systems with regard to item 3, it could be that common-law systems have a systematic disadvantage here because the law is “discovered” by judges, implying that it is not sufficient to know legislation in order to know the currently valid law, but also many court decisions. This can cause two different types of costs, namely, (1) higher transaction costs in ascertaining the currently valid law and/or (2) higher transaction costs in trying to spell out as many contingencies as possible in the contract, which makes contracts more complex, lengthier, and so forth.

If a substantive law is subject to the danger of heavy government intrusion into private property rights, then this substantive law will not be chosen by international business representatives, simply because it would reduce predictability. According to LLSV, the protection of private property rights in legal systems belonging to the French legal family is very low. At the margin, this lack of protection could be so severe that it dominates all other criteria, implying that even firms that are familiar with the system would prefer to opt out of it into a system that they know less well. This hypothesis could be termed LLSV extreme. It would imply that no international contracts should be based on French law because even French firms—or, more generally, firms from countries belonging to the French legal tradition—prefer to opt into other substantive laws. They thus expect that the higher transactions costs they need to incur due to their lower familiarity with other legal systems will be more than offset by the expected efficiency gains.

But LLSV do not need to imply this “corner solution.” It could be that some firms based in countries belonging to the French legal origin would prefer to stick to their legal origins, whereas others would prefer to opt out. The number of contracts according to which French law is chosen as the applicable law then needs to be compared to the number of contracts one would expect to observe given that all parties strive to conclude contracts according to their own law. More specifically, the number of contracts concluded relying on substantive law belonging to French legal origin is hypothesized to be lower than its expected share according to where the contracting parties originate. Inversely, the number of contracts concluded that rely on substantive law belonging to the common-law legal family is hypothesized to be higher than its expected share based on the geographical origins of the contracting parties. This hypothesis will be dubbed LLSV light.

Additional considerations that could have an impact on the choice of law are (1) the home base of internationally active law firms. Our hunch here is that Anglo-American law firms are more important than the share of internationally traded goods and services from those countries would lead one to expect. This could be an advantage for the English-law origin countries as we expect lawyers to advise their clients to choose a law with which the lawyers—not the clients themselves—are familiar.

How to test these hypotheses empirically? There are a number of associations that have specialized in providing arbitration services to internationally active firms. We contacted quite a few of them asking for case statistics and, in particular, for the chosen substantive legal systems. The answers were disappointing. Most of the organizations replied that they do not produce such statistics and that they were not able to produce them expressly for us. Table 1 gives an impression of the number of cases that these various organizations have received over the last dozen years.

Table 1.  The Case Load of International Arbitration Courts
  • *

    The statistics include domestic as well as international arbitrations.

  • **

    As a result of Severe Acute Respiratory Syndrome, the caseload for 2003 was down compared to 2002.

  • +

    Cases completed.

  • Note: AAA—American Arbitration Association; CIETAC—China International Economic and Trade Arbitration Commission; HKIAC—Hong Kong International Arbitration Centre; ICC—International Chamber of Commerce; JCAA—Japan Commercial Arbitration Association; KCAB—Korean Commercial Arbitration Board; Kuala Lumpur—Kuala Lumpur Regional Centre for Arbitration; LCIA—London Court of International Arbitration; SIAC—Singapore International Arbitration Centre; Stockholm—Arbitration Institute of the Stockholm Chamber of Commerce; Vancouver—British Columbia International Commercial Arbitration Centre; Vienna—International Arbitral Centre of the Austrian Federal Economic Chamber.

  • Source: Hong Kong International Arbitration Centre at and own modifications.

Kuala Lumpur4387357101112436

The only organization that provided at least some additional data was the International Chamber of Commerce in Paris, which runs the International Court of Arbitration (ICA). This is why we present this court in a little more detail. Information is based on the Statistical Report of the Court for the year 2003. The court deals only with conflicts that have an international and commercial background. In 2003, the court had 114 members from 78 countries and met on 60 occasions, 12 times in plenary session and 48 times as a committee. Decisions on single cases are, however, not prepared by the court but by arbitrators. As Table 1 indicates, 580 cases were filed in 2003; 1,584 parties were involved in these cases (implying that there were quite a few multiparty cases). Those parties came from 123 different countries, the highest number of parties coming from the United States and Canada (220). Next in line were France (127), Germany (112), Italy (92), and the United Kingdom (71).

Choosing the ICA is not equivalent to choosing Paris as the place to hear the case. In 2003, cases were heard in 47 different countries, the most frequently selected cities being Paris (112 cases), London (48 cases), Geneva (38 cases), New York (33 cases), and Zurich (26 cases). Eighty-two percent of the contracts underlying the disputes referred to the ICA in 2003 contained a choice-of-law clause specifying the applicable law. In 80.4 percent of the contracts, the parties opted for a national law. The amounts in dispute ranged between less than US$50,000 to more than US$1 billion. Almost 35 percent of the cases involved amounts between US$1 and 10 million.

According to the statistics for 2003, “the laws of England, Switzerland and France” were the most commonly chosen ones. The Documentation and Research Department of the ICC informed me that this still held true in 2004. The percentage points are 24 percent for English law, 20 percent for Swiss law, and 19 percent for French law, whereas U.S. and Canadian law were chosen only 10 percent of the time.8 LLSV code Swiss law as German legal origin. The Swiss Civil Code was indeed written by legal scholars from the German part of Switzerland who had been trained in Germany (Zweigert & Kötz 1996:165ff.). Yet, subsequent to the French Revolution, French influence had been considerable, particularly in the western and southern cantons of the country. The Code Civil was valid in the cantons of Geneva and Berner Jura because they belonged to the French Republic. Even after the two cantons joined the Swiss confederation in 1815, the Code Civil remained valid there. When many of the cantons passed their own civil codes during the 19th century, the French Code Civil served as a model in many French-speaking cantons, such as Vaud, Fribourg, Valais, and Neuchâtel, but also in Italian-speaking Ticino (Zweigert & Kötz 1996:102). Even though most observers seem to agree in classifying Switzerland as belonging to German legal origin, there is no doubt that French influence is not negligible.

Clearly, the numbers just cited are not in accordance with LLSV extreme because not all firms originating from French civil-law countries opt out of French law. But how about LLSV light? To know, we need a theoretically derived benchmark with which we can compare the actual numbers. A rather straightforward benchmark could look like this: every party to a contract will try to make its “home law” the law applicable to the contract. It was conjectured that the more powerful party could have an advantage here. Since we do not know anything about the relative powers of contracting parties, a reasonable assumption seems to be that all contracting parties have an equal chance of making the other party agree to its home law. Taking multiparty cases into account, the number of cases is then not divided by 2 (which we would do if all cases were two-party cases), but by 2.73 in order to arrive at a benchmark. This number is the result of the division of 1,584 (the number of parties that filed a request for arbitration with the ICC in 2003) by 580 (the number of cases filed in 2003).

The statistical report of the ICC contains the number of parties ordered according to geographic origin. We thus know how many parties came from the four jurisdictions mentioned above and can divide the number by 2.73. The result is our benchmark, that is, the number of cases that we would expect to be conducted under a substantive law of a nation-state given that no other factors, such as predictability, certainty, or anything else, come into play.

Unfortunately, our information concerning the use of substantial law is highly restricted. We know this number only as a percentage for England, France, Switzerland, and the United States and Canada. This is why the benchmark is calculated for only these four jurisdictions. The resulting ratio between the expected number of contracts under a substantial law and the factually observed one is easy to interpret: values larger than 1 in Table 2 indicate that the substantive law of a country is chosen more often than was to be expected, whereas values smaller than 1 indicate that the substantive law is chosen less often than expected.

Table 2.  Expected and Observed Proportions of 580 Contracts Selecting a Country's Law
CountryExpected Proportion of Contracts Choosing Country's Law (Party N = 1,584; Contract N = 580)Observed Proportion of Contracts Choosing Country's Law (N = 580)Ratio of Observed to Expected Proportion of ContractsSignificance (Binomial)
  1. Source: ICC 2003 Statistical Report, private communication, own calculations. For calculating the implied number of cases, the entire caseload of the ICC (i.e., 580 cases) was taken as the base.


The results seem to indicate that Swiss law is most attractive, followed by English law, that is, a law belonging to the common-law family. If LLSV light were right, we would expect French law to receive a coefficient of less than 1, which is not the case. What is even more astonishing is that U.S. law appears to be rather unattractive. Its coefficient is by far the smallest among the countries for which data are available. What is more, its coefficient is clearly below 1, indicating that North American firms are opting out of their substantive law.

IV. Discussion of the Results

Of course, these results should be treated with caution: we only have data from one single arbitration court and we only have data for a very limited number of countries. Yet, the Document and Research Department of the ICC has confirmed that the year 2003 was not a year that was particularly disadvantageous to the United States. If anything, it was favorable to the United States, as in 2002, only 7.7 percent of all cases dealt with by the ICC used U.S. law as substantive law; this would, ceteris paribus, mean that the ratio for U.S. law would have been 0.55 instead of 0.72.

Based on revealed preference of sophisticated international actors, we can thus conclude that LLSV extreme is clearly refuted by the available data. LLSV light is not confirmed by the available data either. Although English law was chosen more frequently than expected according to our benchmark, this was also the case for French as well as for Swiss law, with Swiss law being the most attractive, even though it has been heavily influenced by French law as discussed above. The data further seem to indicate that U.S. law is perceived as unattractive by international businesspeople. The data are corroborated by anecdotal evidence according to which U.S. law is often perceived as bizarre.

It is noteworthy, however, that the performance of common-law countries is pointing in different directions: whereas English law is chosen much more frequently than expected, U.S. law is chosen significantly less frequently than expected. It would be interesting to know whether U.S. firms use this as a bargaining chip, that is, agree on English law after it has become clear that their partners are not willing to consent to the use of U.S. law.

Switzerland is clearly overrepresented, which might also be due to the construction of the benchmark. In its construction, it was assumed that the substantive law chosen is always the home law of one of the contracting parties. Yet, our theoretical considerations have led us to hypothesize that in cases of approximately equal strength of the contracting parties, they might just as well choose the substantive law of a third country. Here, Switzerland clearly has the advantage of being small: due to its small population, the likelihood that one of the contracting parties is Swiss is lower than, say, the likelihood of finding a British, German, or American at the negotiating table, which means that Swiss law has a higher a priori likelihood of being chosen if negotiations advance to this stage. The perceived neutrality of Swiss law, as well as the declared political neutrality of Switzerland, might be additional reasons. Yet, laws of countries with a similar population size, such as Austria, are selected less frequently.9 One reason for this could be that the Swiss are able to work in a number of different languages and that law commentaries and so forth for that country are available in at least four languages (namely, German, French, English, and Italian).

One possible counterargument with regard to the figures could be that it is small wonder that French law was chosen so often given that the seat of the organization providing for arbitration services was Paris. For two reasons, this counterargument is, however, not convincing: (1) the choice of location and the choice of substantive law are not only uncorrelated on logical and judicial grounds, but also seem to be factually uncorrelated (Corinne Truong 2001); (2) the choice of Paris as the location of the arbitration organization indicates that the set of rules and procedures applied there seem to be what many parties prefer over other organizations at other locations that could, in principle, also be chosen.

V. Conclusions and Outlook

The hypothesis that countries with common-law origins have a number of advantages concerning the quality of their government vis-à-vis countries with civil-law origins was briefly presented and critically discussed. It was argued that if the common law is indeed better suited to structure business transactions, businesspeople would, if given the choice, prefer to conduct their transactions under substantive law of the common-law family rather than of the civil-law family. In international transactions, businesspeople have precisely this choice. The choice actually used by them thus constitutes a natural experiment to test the hypothesis concerning all the claimed advantages of the common law.

The results presented here are based on cases conducted under the auspices of the International Court of Arbitration. They show that French law, which is generally supposed to be the worst branch within the civil-law family, is frequently chosen, just as is Swiss law, which has been substantially influenced by French law. Moreover, U.S. law is chosen less frequently than expected according to the benchmark here constructed.

These results can, however, only be preliminary. For future studies, broader data sets covering more countries, more arbitration organizations, and longer time periods are desirable. In addition, some additional variables could be of interest. For example, it could be the case that the average contract length has substantially increased over time, which could be interpreted as an attempt of the contracting partners (or their attorneys) to reduce the number of contingencies and thus to reduce the likelihood of ever having to resort to a state-created substantive law. This would, of course, mean that the relevance of the choice of law has decreased over time.

On a broader scale, it is noteworthy that LLSV never tested the significance of legal origins for explaining variation in income and growth. Our own tests in conjunction with judicial independence (Feld & Voigt 2003, 2006) showed that only socialist legal origins have a significant (negative) impact on growth rates. This points toward the necessity to “unbundle institutions” as proposed by Acemoglu and Johnson (2005). It appears plausible that differences in public law are a lot more relevant than differences in private law if one wants to explain differences in the growth paths of different countries.


  • 1

    In the meantime, an entire cottage industry of recoding particular aspects of LLSV has emerged. In particular, the way LLSV coded the rights of minority shareholders is subject to dispute. A number of studies have proposed modified coding for single countries (e.g., Braendle 2006 with regard to Germany). Spamann (2006) made the Herculean attempt to recode LLSV's entire sample. Based on the recoded indicators, he finds no significant differences between civil and common law nor any predictive value of the index for stock market outcomes.

  • 2

    Egypt and Mauritius, which used to be British colonies but are coded as belonging to French legal origin, seem to be the only two exceptions.

  • 3

    But see, e.g., Mahoney (2001), who seems to argue that the French law is “bad” whereas the common law is “good.”

  • 4

    Whether such a bias really exists is disputed. The factual existence is, however, irrelevant as long as international contractors are afraid of it. Clermont and Eisenberg (1996) and Thiel (2000) do not find evidence for home bias with regard to U.S. courts.

  • 5

    Although over the past couple of decades, the process seems to have shifted more and more toward technocratic adjudication (Bühring-Uhle 1996; Schneider forthcoming). Arbitration may now be just as lengthy and costly as trials in state courts.

  • 6

    Precise numbers are hard to come by, however. There are estimates that between 80 and 95 percent of all international contracts contain mandatory arbitration clauses (Kropholler 1975). Van den Berg et al. (1988) report mandatory arbitration clauses in as many as 90 percent of all international contracts. Casella (1996) cites officials of the Netherlands Arbitration Institute whose estimate is that more than 80 percent of all international contracts have arbitration clauses. In a recent study focusing on publicly held companies and relying on information contained in SEC filings, Eisenberg and Miller (2006) found that only some 20 percent of all contracts with one non-U.S. partner had arbitration clauses. Although only a fraction of the former estimates, this is still a major share, sufficiently large enough to ask what substantive laws were chosen in these cases.

  • 7

    More precisely, 82 percent of the contracts underlying the disputes referred to ICC arbitration in 2003 contained a choice-of-law clause specifying the applicable law. In 80.4 percent of these cases (or in 65.93 percent of all cases), the parties opted for a national law.

  • 8

    The United States and Canada are reported in conjunction here because the numbers provided by the ICC are only available for both countries together. As we are interested in the choice of common-law versus civil-law jurisdictions, this does not constitute a major problem as Canada belongs to the common-law tradition (with the exception of Quebec).
    In a study confined to an analysis of international distribution contracts, Corinne Truong (2001) finds choices of substantive laws that seem broadly in line with the more general figures provided here. She finds that civil law is chosen much more frequently than common law, which might, however, be due to the fact that distribution was usually carried out in Europe. The rank order of the most attractive civil laws is French, Swiss, German, Italian, Spanish, Belgian, Dutch, Danish, and Swedish. Note that Italian, Spanish, Belgian, and Dutch laws are all part of the French tradition.
    On a more informal level, the secretary of the Vienna-based International Arbitral Centre of the Austrian Federal Economic Chamber has confirmed that more than half the cases decided there use Austrian, German, or Swiss substantive law.

  • 9

    Traditionally, Austria has also been neutral.