Networks , Dialogue or One-Way Traffic ? An Empirical Analysis of Cross-Citations Between Ten of Europe ’ s Highest Courts

Courts play a decisive role in most legal systems. Thus, it is interesting to explore empirically how courts work and how their judgments are drafted. In the present article we analyse a particular question of court-reasoning, namely: do highest courts from different countries interact with each other, for instance, by way of reading and citing each other’s case law?1 Previous research often found that courts rarely look abroad.2 However, in a globalising world, even in law, which has traditionally been a prerogative of the sovereign nation-state, some cross-border interaction seems likely. For instance, Anne Marie Slaughter contemplates that nowadays:


Introduction
Courts play a decisive role in most legal systems.Thus, it is interesting to explore empirically how courts work and how their judgments are drafted.In the present article we analyse a particular question of court-reasoning, namely: do highest courts from different countries interact with each other, for instance, by way of reading and citing each other's case law? 1 Previous research often found that courts rarely look abroad. 2However, in a globalising world, even in law, which has traditionally been a prerogative of the sovereign nation-state, some cross-border interaction seems likely.For instance, Anne Marie Slaughter contemplates that nowadays: 'judges see each other not only as servants and representatives of a particular polity, but also as fellow professionals in an endeavour that transcends national borders.' 3 This statement was made in the context of transnational litigation, but it may also be a reflection of a more general trend.At the most basic level, the development of a 'one-way traffic' situation is conceivable, when a court of a relatively small jurisdiction closely follows the jurisprudence of a larger one.With the development of a more intense interaction, this may become a dialogue if the latter jurisdiction begins to take some interest in the case law of the former country. 4Going further, if countries belong to the same group of countries (for instance, the EU or the same legal family), it may be the case that the communication between highest courts forms part of a formal or informal network. 5uch interaction may take place in various forms.There is some direct transnational collaboration and communication between highest court judges, 6 but in this article we focus on cross-citations as a form of influence.Of course, the citation of a foreign court does not necessarily mean that foreign ideas were really a decisive consideration for the outcome of a case.Nonetheless, cross-citations can show to what extent courts use foreign law as a justification for a judicial decision, be it a positive or negative example.In the project from which this article derives, 7 we have collected data on how often and in which circumstances ten European highest courts cite each other.We managed to get access to the full text of almost all decisions of these highest courts for the period between 2000 and 2007.In total we have considered 636,172 decisions and we have found 1,430 cross-citations.
The remainder of this article is structured as follows: Section 2 summarises the data considered and the search methodology used.Section 3 presents bar charts on cross-citations.Network presentations of the data follow in Section 4. In Section 5 we distinguish between outgoing and incoming citations.Section 6 concludes.

Population and search methodology
Table 1 presents the list of countries and courts examined, the databases used, and the subject matter jurisdiction of the ten highest courts.It also indicates how many decisions the highest courts have published between 2000 and 2007 and how this translates into the number of decisions per 1,000 inhabitants.In an associated article we describe the choice of countries and courts in detail. 8In this article we also make it clear that our project does not aim to examine why, according to Table 1, not only the absolute number of cases but also the decisions per capita are very disparate.It would be the topic of a separate empirical study to explore this question.For instance, it possibly matters that there are differences in subject matter jurisdiction.Moreover, many further factors may influence the number of highest court decisions, such as appeal requirements and procedures, as well as ease of access to lower courts, availability of self-help and out-of-court settlement, differences in substantive law, legal culture etc.9 The actual data used in our project are the cross-citations between these ten courts.We managed to get access to the full text of (almost) all decisions of these highest courts for the period between 2000 and 2007. 10In order to locate citations to foreign courts covered by our study, we compiled an extensive list of search terms. 11Then, in all countries, we first looked at the actual decisions.Where they were available, we also included opinions by the reporting judge or the advocate general. 12Inclusion of these documents was necessary because in some countries the legal justification of a decision that in other systems would be found in the decisions themselves, will appear only in the opinions of the reporting judge or advocate general. 13e checked all citations and classified them according to the reason why foreign courts had been cited: (a) case history and jurisdictional issues; (b) an underlying European or international legal basis; and (c) purely comparative reasons. 14Citations of type (a) are the ones which a court usually cannot avoid.This type of citation is not exactly what we were looking for because such citations have no bearing on a possible transnational dialogue between the courts, or the influence of foreign legal arguments.Therefore, the following sections only report the total number of citations in categories (b) and (c), which have in common that judges have freedom of choice as to which foreign court (if any) to cite. 15

Figure 1 Absolute number of cross-citations
had to rely on a random sample of decisions because citations to English courts do not always reveal whether the cited court is really the Court of Appeal (for details see Gelter & Siems, supra note 8).12 Opinions of the advocates general were included for Belgium and the Netherlands.For France we only got access to selected opinions of the advocate general and the reporting judge.13 For further details see Gelter & Siems, supra note 8.
There are different ways in which our data can be displayed.In this section we present bar charts that show how often the ten 'citing courts' have made reference to the ten 'cited courts' .Figure 1 is based on the absolute number of cross-citations.Figure 2 shows the cross-citations per all decisions of a particular court and Figure 3 the cross-citations per all of its cross-citations. 16t can be observed that the citations from Austria to Germany and from Ireland to England (and Wales)17 dominate the picture: Austria has cited Germany 459 times, and Ireland has cited England 456 times.The other relationships trail behind these two by one order of magnitude: 58 and 45 citations from the Netherlands and Switzerland to Germany, 41 citations from Belgium to France, and 34 citations from Germany to Austria.

Figure 2 Cross-citations per all decisions of a particular court
A problem with Figure 1 is that it does not consider that the total number of highest court decisions varies widely between the ten countries (see Table 1, above).Figure 2 tries to rectify this problem but without much success since the distribution of the ten bars in Figure 2 is even more unequal than in Figure 1.The High Court of Ireland has cited the Court of Appeal of England and Wales in about 19% of all of its decisions, well ahead of the next relationships -Austria and Netherlands to Germany in 1.6% and 0.6% respectively of all of the decisions of these courts.This could make sense because of differences between common law and civil law countries: when common law countries cite each other, this is not seen as an 'import' of foreign law, but as a way to identify the common legal rules and principles (the 'common law is a whole'18 ).The fact that England does not cite Ireland very frequently may be regarded as a rare exception since it can be shown empirically that the Court of Appeal of England and Wales frequently cites courts from Australia, Canada and New Zealand. 19However, a closer view has to lead to the conclusion that Figure 2 is not the most helpful presentation of our data.It is not clear to what extent the number of cross-citations is affected by the total number of decisions.One might suspect that cross-citations typically occur in the most important cases, in which there will be an appeal to the respective highest court in every country.Thus, theoretically, the total number of decisions need not affect the number of cross-citations.
Moreover, there are many factors other than the total number of decisions that may determine whether and how often foreign courts are cited.On a general level, it possibly matters who the highest court judges (and advocates general) are: for instance, how in each country judges are trained, appointed and promoted, 20 and what 'judicial mentality' they have. 21It also matters how (and for which audience) 22 judgments are drafted: while, for example, common law judges or the courts in German-speaking countries often write comparatively long opinions, French decisions tend to be short and written in an idiosyncratic formulaic style. 23Furthermore, a low number of citations may simply reflect differences in citation style between the ten courts: in some countries it may be completely acceptable (or even expected) for judges or advocates general to look for inspiration from other countries and to indicate such findings in their opinions, whereas in other countries there may be social or legal restrictions on citing foreign law. 24hus, in order to control for such unobserved differences between citing courts, Figure 3 is based on the citations per all cross-citations of the particular court.In contrast to Figures 1 and 2, this figure cannot be used in order to compare the differences in the total number of cross-citations (by definition, this is always 100%).Yet, Figure 3 is the most informative way of showing which of the cited courts each of the citing courts prefers to cite.
It can be seen that seven out of the ten courts have a favourite court accounting for more than 50% of its foreign citations.These are the Irish citations to England (98%), the Austrians to Germany (94%), the Spanish to Germany (88%), the Germans to Austria (83%), the Belgians to France (70%), the Swiss to Germany (70%) and the Italians to France (67%).In contrast to this, the citations of the English, French and Dutch highest courts are more evenly split.In an associated paper we have used regression analysis in order to determine which factors account for these differences in cross-citations.We found that language skills, membership of the same legal family, cultural, political and economic indicators, and the population size of the cited country all matter for which countries are cited.The most important of these factors are language skills, with the possible policy implication that countries should provide English translations of their highest court decisions.25

Network presentation of cross-citations
Network analysis has become increasingly popular in the last three decades. 26It started in sociology but it has also been used in politics, economics, business, psychology, anthropology and, more recently, law. 27he main interest of social network analysis is to identify, visualise, compare and analyse the relationships between individuals or entities.In the terminology of network analysis the individuals are called 'nodes' and the relationships are called 'ties' or 'edges' .
In the present case the 'nodes' are the 10 countries and the 'ties' are the cross-citations between them.In Figure 4 all ties with more than five cross-citations (one way) are displayed.The strength of the ties, the size of the arrow heads and the closeness of the countries in the chart 28 is determined by the logarithm of the absolute number of cross-citations.We have used the logarithm of the citations since a network presentation with the absolute numbers would not have been very revealing: Austria and Germany, and Ireland and England would have merged to one dot each, whereas the other pairs of countries would only have been connected by very thin lines.The logarithmic transformation has the advantage that it reduces the high numbers more than the lower ones, making the range of values more manageable.

Figure 4 Network based on logarithm of absolute number of cross-citations
The first point to observe from Figure 4 is that we have no unconnected parts.In particular, it is worth noting that common and civil law countries, and Francophone, Germanophone and Anglophone countries are not completely unconnected.Two countries, the Netherlands and Germany, are in the centre of the network since they are connected with five other countries.More isolated are Italy, Ireland and Spain which are just connected with one of the other countries.
Almost all the arrow heads of Figure 4 show in one direction only (an exception is, for instance, the tie between the Netherlands and England).Thus, in terms of the direction of cross-citations (see Section 1 above), there seems to be mainly 'one-way traffic' .However, as noted already (see Section 3 above), the problem is that many unobserved factors influence the decision as to whether a particular highest court cites foreign courts at all.As in Section 3, it is therefore preferable to consider the citations per the total number of cross-citations of this court.
Figure 5 displays all ties which denote more than 4% of all cross-citations of this court.Ireland is again relatively isolated.Germany and France are in the centre since they are cited by many other countries, confirming that big countries are cited more often than smaller ones (see Section 3 above).It is also interesting to see that there are triangles between Italy, France and Belgium on the hand, and Austria, Germany and Switzerland on the other, indicating the relevance of common languages and legal cultures (see also Section 3 above). 29Finally, Austria and Germany are connected by a clear, dual-headed tie because both courts are the favourite cited courts of each other. 30This seems to be different for the other relationships.Thus, again, we do not appear to find evidence of a judicial dialogue.However, this should not be our final word on this issue because the following section will provide a more sophisticated treatment of the association between outgoing and incoming citations.

Outgoing and incoming citations
Table 2 reports the correlation between the nine outgoing and incoming citations of each of the ten courts, as well as the correlation between all ninety outgoing and incoming citations.The correlation coefficients of the absolute data (1) are problematic since unobserved factors influence the decision of the highest courts as to whether they cite any foreign court at all (see Sections 3 and 4 above).Thus, the option which uses the cross-citations per all cross-citations of the court in question (2) is again preferable.Of course, the previous sections have also shown that some countries are more 'popular' (in terms of citations) than others, and that there are also differences in the frequency of citing foreign courts.So, despite positive correlation coefficients in Table 2, outgoing and incoming citations are not perfectly symmetrical with each other.Again, network analysis comes to our help, because it enables us to identify the core outgoing and incoming countries.This is done by way of a core-periphery model. 31Such a model presupposes that, in terms of the present study, some countries are more popular outgoing countries and others are more popular incoming countries.Consequently, we cannot use the percentage form of our dataset (see Figures 3 and 5, above) since, here, by definition, all outgoing citations of each court add up to 100%.Thus, despite its previously mentioned problems, Table 3 is based on the absolute numbers of cross-citations (similar to Figure 1, above).The '*' , '**' , and '***' denote whether there are single-, doubleor triple-digit citations between two courts.The core cited and citing courts are in the shaded field.

Table 2 Correlation between outgoing and incoming citations (1) Basis: absolute number of cross-citations (2) Basis: cross-citations per all cross-citations of a particular court
It follows from Table 3 that England, France and Germany are the core incoming countries, and that the Netherlands, Belgium and Switzerland are the core outgoing ones.This is an interesting result.England, France and Germany are often regarded as the three 'origin countries' whose legal systems have heavily influenced legal systems all around the world.32Thus, it is not unexpected that these countries are most important in terms of incoming citations, with France and Germany being slightly ahead of England, possibly, because six out of seven of the remaining countries are civil law jurisdictions.The main explanation for the core outgoing countries seems to be linguistic diversity: Belgium and Switzerland are multilingual countries, and the Dutch advocates general frequently cite English, French, German and Dutch materials in their original languages.Thus, these judges and advocates general may be more cosmopolitan than the ones of the other countries, as reflected in a high number of citations to different foreign courts.

Caveats and conclusions
There is a general trend to use quantitative methods in order to compare court proceedings and judgments across countries. 33For instance, the World Bank's Doing Business Report employs various indicators in order to measure 'the efficiency of the judicial system in resolving a commercial dispute' . 34The European Commission for the Efficiency of Justice (CEPEJ) 'has undertaken a regular process for evaluating judicial systems of the Council of Europe's member states' . 35The Netherlands Council for the Judiciary has assigned a study to 'design and implement a method to periodically compare the judiciary system of the Netherlands with that in other countries' . 36And, at universities, a recent project of the University of Oxford has collected data on the costs and funding of civil litigation in various jurisdictions, 37 and a handbook of three Dutch universities has developed a methodology for measuring access to justice. 38t is clear that such quantitative research can always be challenged as being a rather simplistic way of comparing how courts actually work. 39Thus, in this article we not only reported our results but also indicated where and to what extent the counting, in our case the counting of cross-citations between ten highest courts, has its limitations.
Overall, we observe some one-way traffic, some dialogue and some networks.We were also able to identify why there is no mutual interaction between all of these highest courts: courts are more likely to cite each other when they are from the same legal tradition and the same language group.Furthermore, we found that larger jurisdictions are cited more frequently than smaller ones.Given these constraints, we would not expect legal systems to freely pick and choose 'the most efficient solution' (however defined), as had been suggested by Ugo Mattei. 40ross-citations are important because they show to what extent courts use foreign law as a justification for a judicial decision.We do not argue that these citations have actually resulted in legal transplants.Indeed, a comparative analysis may also be used to explain the differences and reinforce national solutions rather than leading to uniformity. 41Nevertheless, frequent cross-citations are evidence of some diffusion of ideas. 42In this respect, our data may be seen as an empirical confirmation of Daphne Barak-Erez's observation that 'learning from other legal systems has always been a major technique in the development of law' . 43here can also be dialogue between highest courts that is not reflected in cross-citations.Highest court judges are increasingly involved in transnational networks with the aim of fostering collaboration and communication. 44It may also be the case that in Europe ideas are often exchanged through intermediaries such as, say, the ECJ, its advocates general and the ECHR, 45 or with references to European 'model laws' such as the forthcoming instrument on European contract law. 46A further caveat is that we do not know how much foreign case law may matter behind the scenes all the while that judges or advocates general do not mention it explicitly in their opinions. 47onomics 14, pp.3-19.Similarly, M. Andenas & D. Fairgrieve, 'There is A World Elsewhere -Lord Bingham and Comparative Law', in M. Andenas & D. Fairgrieve (eds.),Tom Bingham and the Transformation of the Law: A Liber Amicorum, 2009 ('international market place for judgements').41 Canivet (2006)  2006), supra note 2, p. 1398 ('Because the methods of courts ought in all circumstances to be transparent, it is obvious that the comparative analysis which the judge intends to include among the elements under examination ought to be (…) explicit').
It follows from Table2, model (2), that, overall, courts do 'return the favour' because there is a statistically significant positive relationship between incoming and outgoing citations.Austria, Ireland and Italy have very high correlation coefficients which can be explained by the fact that Austria almost only cites and is cited by Germany, Ireland almost only cites and is cited by England, and Italy almost only cites and is cited by France.With the exception of Spain, however, the other countries too have positive correlation coefficients.