Aristocrats in Arbitration: Did Class Affect Inter-state Arbitration Before or After the 1899 Hague Peace Conference?

Abstract

This article explores whether the aristocratic status of arbitrators or disputants affected the outcome of inter-state arbitrations either before or after the 1899 Hague Peace Conference. This article takes a longue durée approach to the topic by including all inter-state arbitrations between 1794 and 1989. The research shows a statistically significant relationship between respondent-appointed aristocratic arbitrators on a tribunal and the outcome against the aristocratic party, as well as a statistically significant relationship between no respondent-appointed aristocratic arbitrators and outcome in favor of the claimant, but only in inter-state arbitrations before the Conference for both of these relationships. This article brings into question whether aristocratic arbitrators from before the Conference were fully committed to the international rule of law. Moreover, it highlights how such arbitrators stopped making their decisions based on class after the Conference, which should reassure users of inter-state arbitration that worry about such biases.

1 Introduction

1.1 The Connection Between Aristocratic Arbitrators and Outcome

The 1899 Hague Peace Conference widely is seen as a watershed moment for inter-state arbitration, with its creation of the Permanent Court of Arbitration, among other developments during this period.¹ This article shows that the Conference marks a point in time when certain arbitrators changed their behavior in certain situations. Using a population census of all inter-state arbitrations between 1794 and 1989,² as well as some coding from a previous project,³ this project set out to identify and code indicators of the aristocratic status of arbitrators in Alexander Stuyt’s influential book Survey of International Arbitrations 1794–1989 and to quantify the involvement of aristocratic arbitrators in these inter-state arbitrations before attempting to determine correlation and causal relationships. This article provides that analysis, along with the other analysis relating to class and arbitral outcome undertaken, even though most did not lead to a determination of a statistically significant relationship. The existence of some statistically significant relationships makes this article noteworthy. In particular, the research at the heart of this article shows that there is a statistically significant relationship between respondent-appointed aristocratic arbitrators on an arbitral tribunal and outcome against the aristocratic party, as well as a statistically significant relationship between no respondent-appointed aristocratic arbitrators and outcome in favor of the claimant, but only in inter-state arbitrations before the 1899 Hague Peace Conference for both of these relationships. There are no such statistically significant relationships observed in the data after this point in time.⁴ While not based on causal inference analysis, it nevertheless was observed in the data that aristocratic parties tended to appoint aristocratic arbitrators throughout this entire period. Such class-related observations have never been noted in the literature on international judicial resolution before now,⁵ nor has the impact on aristocratic arbitrator behavior been noted in the literature on the Conference,⁶ which makes this article even more noteworthy. These observations make this article especially relevant to the focus of this journal, inasmuch as it brings into question the commitment of some arbitrators to the rule of law, at least before the Hague Peace Conference.

Pinpointing one cause for these changes in behavior—such as the weakening of a particular class, contextual changes for aristocrats in particular regions of the world, or the normalization of inter-state arbitration at around this time—would require a different dataset and different analytical tools than those used here. Instead, it is important to acknowledge from the beginning that it likely was a multivariate change, with too many possible variables, including the relationship between different arbitral issues. The popular judicial decision-making models concerning the US judiciary have indicated that sociological, psychological, and political factors may impact judges’ decision-making process.⁷ It is possible to consider arbitrators in inter-state arbitrations arbitrators as facing influences similar to decision-making factors in domestic courts. The list of possible factors can never be exhausted, as with all other social science research. If forced to identify only one likely cause for these quantitative research outcomes, the increase in public engagement with international law and international affairs during this time period seems like a promising candidate. Previous commentators have documented the increase in public attention to international law and international affairs leading up to and during the Conference, thanks in part to the help of peace movements and sensational journalism at the time.⁸ A relationship between public attention and change in domestic judicial behavior has been demonstrated elsewhere,⁹ and there is no reason to think that that might not be at play here at the international level. If such an explanation were proven to be the correct explanation, then the observed changes might have nothing to do with the class status of the arbitrators or the disputants. However, neither the dataset nor the analytical tools used allow for a determination of a single explanation. That would require a chronological dataset of public attention on inter-state arbitration by state and a due time series analysis, in addition to causal inference analyses. Such a task is left for future researchers. This article is delimited by only identifying the statistically significant relationships between aristocratic status of the arbitrators and the outcome of the arbitration.

1.2 Definitions and Delimitations

Before providing that analysis, a few definitions and delimitations are needed in order to understand the data-driven research at the heart of this article. Concerning definitions, first, a definition of inter-state arbitration is needed. Arbitration is a judicial form of international dispute settlement that relies on law to decide the matter, with the parties having a large measure of freedom to choose such aspects as their arbitrators and the applicable rules.¹⁰ Not only is that the definition of inter-state arbitration now, but President and Reporter for the Conference’s Committee on Examination Descamps also explained it in this manner in the committee’s report.¹¹ Since inter-state arbitration involves the application of public international law to resolve disputes between sovereigns, it has the alternative title of public international arbitration, as opposed to private international arbitration where at least one private entity is a party.¹² Section 1.2. below elaborates on the characteristics of inter-state arbitration in the context of analyzing the data from Stuyt’s book.¹³

Second, a definition for “aristocratic” must be provided. Obviously, “aristocratic” means more than just an individual’s social standing based on title or name, or whether the state where that individual was born or lived was ruled by a monarch, for example. Indeed, monarchs and aristocrats often have been seen as separate and even competing in post-classical and modern history, especially over property during medieval times.¹⁴ Thomas Hobbes famously favored strong monarchs over aristocracies,¹⁵ while Niccolo Machiavelli saw them as being close to one another on a political spectrum between monarchy and republic at the extremes.¹⁶ Monarchies usually involve rule by one person and one family over time due to some event that took place in the distant past, with aristocracies sharing power among those who possess special merit that society values, which power can then become hereditary.¹⁷ It must be recognized that the relationship between monarchs, aristocrats, the bourgeoisie, and all others has been changing gradually from the sixteenth century until the second half of the nineteenth century, when societies modernized to the point where feudal structures had limited control and the remaining monarchies consolidated power into a centralized government.¹⁸ Nevertheless, at their core, monarchs (and their supporting officials) and aristocrats collectively represent a broad feudal class, along with princes and similarly privileged groups, who generally owned land in a pre-capitalist world.¹⁹ The corresponding author of this article has juxtaposed “aristocratic” to “republican” elsewhere because it was a widely accepted term in the peace-related literature.²⁰ In this largely quantitative study on aristocrats and inter-state arbitration, it is more helpful to juxtapose “aristocrat” with “non-aristocrat,” just to remind the reader that this article is using a MECE (mutually exclusive, collectively exhaustive) measure, which is needed with the independent variable—namely, aristocratic participation on an arbitral tribunal in an inter-state arbitration—if the analysis in Part III is to be possible. Of course, there are complexities involved with this definition, including the existence of constitutional monarchies in the United Kingdom and elsewhere, as well as aristocrats who fought for republican (or rather non-aristocratic) values in the peace movements of the nineteenth century, such as Baroness von Suttner,²¹ much like Lafayette during the French Revolution, for example. Nevertheless, the binary classification of “aristocrat” and “non-aristocrat” enables a useful approximation of the relationship between different variables using the tools described in Sect. 3.1. below, much like how calculus can determine the area under a curve with approximations by drawing mutually-exclusive rectangles of space.²² These types of limitations are to be expected with such historiometric studies as this one, which use the Bayesian approach to statistical analysis of events that could have happened in a different way, like inter-state arbitrations between 1794 and 1989.²³

Finally, it must be acknowledged that aristocracy is not perfectly consistent either across time or across different countries and societies, even though official aristocratic registries like the Almanach de Gotha make it appear that way.²⁴ At the same time, aristocracy noticeably has some consistent characteristics, such as general privilege, landownership, and a degree of supra-nationality. This article is delimited by adopting an assumption of consistency for the purpose of quantification and for comparing populations in order to identify patterns and to approximate relationships, again, similar to how calculus relies on approximations.

1.3 Outline

This article has four parts, including this introduction and an equally brief conclusion in Parts 1 and 4, respectively. Part 2 starts by providing some empirical observations about inter-state arbitration between 1794 and 1989, based on a comprehensive review of the information contained in Stuyt’s book. At a minimum, these descriptive statistics show that there has been immense diversity in inter-state arbitration from 1794 to 1989, much like the diversity observed with international organizations,²⁵ although patterns nevertheless can be seen in their structure and functioning. Part 3 quantifies aristocratic participation in inter-state arbitrations between 1794 and 1989 and provides suitably advanced statistical analysis of whether such participation makes a statistically significant difference on the outcome of inter-state arbitrations. In particular, this part explores whether there is a correlation and statistically significant relationship between the percentage of aristocratic participation on an arbitral tribunal in an inter-state arbitration (the independent variable) and the success rate of the aristocratic party (the dependent variable). Again, as highlighted in the introductory section above, this article shows that—with the dataset used—there was no statistically significant effect of class type on arbitration outcome in general, either by the class of the parties or the class of the arbitrators. However, there was a statistically significant relationship between respondent-appointed aristocratic arbitrators on a tribunal and outcome against the aristocratic party, as well as a statistically significant relationship between no respondent-appointed aristocratic arbitrators and outcome in favor of the claimant, but only in inter-state arbitrations before the 1899 Hague Peace Conference. There was no such statistically significant relationship observed in the data after this point in time. This article provides the findings from that analysis and explores some of the explanations for those findings.

1.4 Legal Realism as the Theoretical Foundation

In exploring the biases and potential biases of aristocratic arbitrators in inter-state arbitration from 1794 to 1989, this article challenges legal dogma concerning what is considered when arbitrators decide on a dispute, and it supports an alternative paradigm presented by legal realism. The notion of legal realism has suffered a fragmented existence on account of the fact that it has had many rather imprecise definitions over the years.²⁶ For example, commentators sometimes see legal realism as a type of methodology that forms an integral part of comparative approaches to the study of law, inasmuch as judges and lawyers alike naturally look outside their own system to see how others tackle similar problems.²⁷ Despite the variation in definitions, the red thread throughout all of them is the point that law depends on what judges and lawyers actually think and do, not necessarily what they should do, which can be noticeably different at times. In this regard, legal realism can be seen as being similar to the post-positivist positions observed in more mainstream social sciences.²⁸

Bias and emotions feature prominently in some definitions of legal realism. Frederick Schauer refers to the involvement of judges’ emotions and other extraneous factors like bias that may drive them to a particular decision as a “tamed” version of legal realism when these factors are relied on only when the law is unclear, with the “untamed” version driving them to a particular decision even when the law is clear on its face.²⁹ Either way, emotions and bias form an important part of an arbitrator’s decision-making processes under this theory. Numerous studies within the US-based literature have taken this approach to studying judicial decision making, to the point that it seems possible to say that it represents a significant, if not dominant, way of analyzing domestic court cases. However, this approach has been slow to spread beyond US domestic law, with this thesis representing the most recent effort to expand its applicability. While this study does not look at arbitrators on an individual level, as is common with other studies with legal realism as the theoretical foundation, looking at groups of arbitrators in the aggregate allows this article to see patterns of behavior never noted before. Making normative assessments of such behavior is left to other studies. Instead, this article’s quest to understand the impact of class on the outcome of inter-state arbitration pushes the research into the realm of positivism, or the belief that reality can be understood through careful observation.³⁰ This thesis explores reality through systematic and comprehensive inductive analysis of inter-state arbitration between 1794 and 1989. Without such an analysis of reality, it becomes difficult to make meaningful reforms to legal doctrine and to the functioning of institutions in this area in the future.

1.5 Methodology

This article’s research methods are important because they remove effects that may distort the results. Some of these effects may include the type of arbitration involved, the politics of the era and the litigant, as well as characteristics of the government which one might consider as having an undue influence on the decision of a state’s arbitrator. This section on methodology highlights the three stages used, which were the classification of parties into aristocratic and non-aristocratic parties, coding of Stuyt’s book for a host of elements related to aristocratic status, and quantifying aristocratic participation before engaging in statistical analysis.

Given this article’s reliance Stuyt’s book, an explanation of its reliability first must be provided. Stuyt took great pains over more than 50 years to collect data on a range of inter-state arbitrations between 1794 and 1989.³¹ Admittedly, Stuyt’s book is not perfect, with some arbitrations repeated,³² as well as some deleted arbitrations so that there are gaps in the sequential numbering of the arbitrations in his survey.³³ However, there is no evidence from other surveys of inter-state arbitrations that Stuyt’s is missing any inter-state arbitrations that reached a decision between 1794 and 1989.³⁴ In fact, international arbitration scholars see Stuyt’s book as comprehensive and exhaustive.³⁵ This is so even though international arbitration generally is seen as having confidentiality as one of its defining characteristics.³⁶ This characteristic presumably would make it difficult to put together an exhaustive survey of international arbitrations, if not impossible. However, inter-state arbitration is different, with some commentators asserting that decisions in these types of arbitrations are “always published,”³⁷ at least when decisions have been made. The reason for this difference presumably is because inter-state arbitration impacts a state’s population to a larger extent than private international arbitration, and so the need for transparency far outweighs the need for confidentiality.³⁸ Without any reliable evidence to the contrary, this article treats Stuyt’s survey as a population census of inter-state arbitrations between 1794 and 1989.

1.5.1 Determining a Party’s Aristocratic Status

The first stage involved the determination of when all of the states that were a party to an arbitration in Stuyt’s book stopped being an aristocratic state, usually when it became a republic. Some parties have never become republics, such as Belgium, The Netherlands, Sweden, and the United Kingdom, while some were born out of revolution, such as the United States. In the case of a few states, such as France, Spain, and Mexico, this stage required a determination of the dates for successive republics when monarchs returned to power and then fell from grace again. That information was found in the World Factbook and the Encyclopaedia Britannica. Using this information, each party was assigned a binary classification of aristocrat/non-aristocrat for each arbitration. Each arbitration had all of the parties to the dispute attached along with all of their aristocratic/non-aristocratic affiliations. An example would be Arbitration No. 66 in Stuyt’s book, which was a 1860–1864 arbitration between Argentina and France.³⁹ Argentina was coded as a non-aristocratic party to the dispute with a vested interest in the arbitration’s outcome. On the other hand, France had a monarchical government at that time, and so it was coded as an aristocratic party. Since these parties had different political affiliations, the arbitration was classified as a mixed arbitration.

Special care was taken to ensure that the classification was accurate at the time of the start of the arbitration and throughout the duration of the arbitration until the rendering of the award. There are three main difficulties in this exercise, either where there was revolution or decolonization during the arbitral proceedings. The first is Arbitration No. 44 between France and Spain in 1851–1852, during which time France’s Louis Napoleon Bonaparte transitioned from being France’s first president to its last emperor when he made himself emperor.⁴⁰ The second involves Arbitration No. 275 between The Netherlands and Portugal in 1904–1914, during which time a revolution in Portugal led to its first republic in 1910.⁴¹ The third is Arbitration No. 413a between India and Pakistan in 1948–1950,⁴² with India breaking away from the United Kingdom in early 1950 and becoming a republic. In all three, the status of the transitioning at the end of the arbitration was seen as more important as the status it started off with because it was at the end when the arbitrators actually were making their decisions, which presumably is when prejudices would have manifest themselves, if at all. Nevertheless, such particularities affect this study, which readers must keep in mind when assessing the reliability of this article’s analysis.⁴³

1.5.2 Coding Stuyt’s Survey

The second stage involved coding the inter-state arbitrations that Stuyt surveyed that reached a decision. Eric Posner and John Yoo kindly shared the database that they created for their 2005 article “Judicial Independence in International Tribunals,” which included columns for the parties, the subject matter of the dispute, the structure of the arbitration (whether a commission, an arbitrator or an institution), the date of the law giving jurisdiction to the arbitral tribunal, whether the law was created before or after the dispute arose, the date of the award, the winner of the arbitration, the outcome, whether a head of state served as arbitrator, and whether the parties complied with the award.⁴⁴ To this database has been added: a binary designation of aristocratic status of each party; a designation of the winner as an aristocrat, non-aristocrat or a mix in the case of partial awards; an indication of whether the parties were both aristocrats, both non-aristocrats or a mix; the number of arbitrators overall; whether a monarch was the sole arbitrator; the number of arbitrators for the claimant(s) and respondent(s); the number of aristocratic arbitrators overall and for each party; whether an aristocratic arbitrator was the president; a breakdown of how those arbitrators were aristocratic; and the percentage of aristocratic participation in relation to non-aristocratic arbitrators.

Some of these categories naturally lend themselves to statistical analysis. For example, the year an inter-state arbitration commenced, the year an award was rendered, the number of years needed for the losing party to comply with an arbitration’s terms all are elements that are naturally countable. Whether a party won a particular arbitration represents an obviously numeric variable inasmuch as it is a binary outcome allowing for a binary variable. Other aspects of these arbitrations had to be categorically classified. For example, many of the arbitrations involved the same type of subject matter. These include arbitrations involving disagreements over the terms of previous treaties and agreements, the right to use particular bodies of water, claims of damages from armed conflict, border disputes and so forth. This project had to generalize the topic matter of these arbitrations into 15 categories. Some types of claims, such as personal claims, made dichotomization between an aristocrat as a representative of the state and as a natural party to a claim, or as a physical human being arbitrator, particularly interesting. The same applies to non-aristocratic cases. It falls outside the scope of this introduction to the project’s methodology to describe all the interesting ways these dichotomies appeared in the analysis.

Peaking at other types of potentially interesting data in Stuyt’s survey also became required when determining aristocratic status. Keeping in mind the formal definition of “aristocratic” provided in Sect. 1.2. above, this study used the data provided in Stuyt’s book to identify and classify those arbitrators who were (1) monarchs, (2) government officials (including ministers, ambassadors, consuls, judges, or king’s or queen’s counsel), (3) military officials of monarchies (including all military titles), (4) possessors of aristocratic titles (including all titles), and (5) possessors of aristocratic names, or a combination of any of these categories. The fundamental assumption of these categories is that all of these demonstrate aristocratic affiliation. While there presumably are degrees to aristocratic affiliation or commitment, this project must assume that all in that population are equal in aristocratic affiliation or commitment since there is no reliable or accepted graduated measure for that, at least not at this time. Of course, titles can be purchased, and striving parents can construct the appearance of aristocracy by the names they select for their children and even themselves. In order to group these individuals into populations for proper statistical analysis, it must be emphasized how monarchs and aristocratic governments typically regulate the adoption and passing on of aristocratic names and titles in their jurisdictions,⁴⁵ presumably to preserve a large measure of the status quo. Such predictability and consistency give such names and titles relevance to the analysis provided in this article. In order to ensure that an aristocratic name possesses aristocratic significance, the nationality and birth year for the arbitrators in question were determined and compared with the years of monarchy in their state of nationality, both at the time of the arbitrator's birth and at the time of the commencement and conclusion of the arbitration in question. Aristocratic names rarely were the basis of determining aristocratic status of individual arbitrators, since their position with an aristocratic state’s government or military was made clear in Stuyt’s book. All of this information was entered into an Excel spreadsheet.

Critics will assert that using Stuyt’s classifications for this project causes problems because this article’s formal definition of “aristocrat” in Sect. 1.2. indirectly relies on approaches taken (and thus definitions adopted) in Stuyt’s book. However, this should not be much of a concern because this article and Stuyt’s book used similar methods to identify and classify the persons being studies—namely, the formal nation-state parties to an arbitration, the persons participating in those arbitrations, and the arbitrators who made the arbitral decisions. Admittedly, what is more controversial is when aristocratic status of an arbitrator was not provided in Stuyt’s book. In those instances, which again were limited in number, then exhaustive archival research for the arbitrator’s name and particulars was conducted among the available lists and biographies of arbitrators, lawyers, peace advocates, and government officials between 1794 and 1989. That limited set of documents and information represents one of the greatest potential limitations to keep in mind when assessing the reliability of this article’s analysis. The archival research for this was extended as far as humanly possible to ward off critiques of shoddy workmanship. However, readers must keep in mind such limits to this and all work involving historical archival analysis.

1.5.3 Quantifying Aristocratic Participation

By collecting and coding the data, and entering that information into the Excel spreadsheet, it became possible to give arbitral tribunals a percentage of aristocratic participation, even when the arbitrator had no official position or title in connection with a monarchic government. Quantitative analysis then can tell us about correlations and whether statistically significant relationships exist between the percentage of aristocratic participation and awards in favor of the aristocratic party, among other factors.

The approach adopted in this article towards quantifying aristocratic participation in an arbitral tribunal represents one of the most serious flaws in this methodology. As described previously, the authors combed the original archival material in order to see which parties represented themselves using aristocratic titles, ranks or other affiliations. When determining the state the aristocrat supposedly served, one could rather easily assess the political nature of a party to an arbitration. Statistically speaking, one cannot easily describe a state as 30-percent aristocratic or determine that a political plurality is sufficiently large to warrant categorizing a particular jurisdiction in one way or the other. The existence of a monarch in that state was used as a short cut for adopting a binary approach towards such a status for states. Aristocratic status (or not) of the persons statistically studied in this article was analyzed as carefully as possible. Again, in some cases, establishing such status proved straight forward. Persons of aristocratic lineage were well known, such as in the newspapers or in official registries like Almanach de Gotha. In other cases, use of a formal title served as sufficient evidence. One would hardly expect an arbitrator to use a particule (establishing aristocratic lineage of some kind) or sign as a Duke if one could not claim possession of these titles. As official documents, all parties to an inter-state arbitration would have taken care to ensure all due formalities and protocols were followed. One would never expect the systematic and non-random fraud that would bias a statistical study. A mad noble may have entered into inter-state arbitrations throughout even decades. However, his or her participation in the process would not leave behind the types of bias and imprecision that would affect a study like this one.

With actual data leading the way at every stage and with the aid of the Pearson’s correlation coefficient and the Analysis of Variance (ANOVA) test, this article’s findings support the notion that something at the 1899 Hague Peace Conference softened aristocratic arbitrators towards aristocratic parties and claimants. Although pure speculation, it might have been the increased competition in the realm of international arbitration from republicans and the republican peace movement at the Conference that shocked aristocratic arbitrators into ceasing their prejudices against aristocratic parties and claimants in particular. Legal diplomacy posits that aristocratic arbitrators in inter-state disputes might gravitate towards aristocratic solutions when faced with aristocratic parties in an effort to avoid alienating aristocratic parties, or at least to soften the impact on aristocratic parties from their decisions,⁴⁶ especially if the efforts at institution building with the PCA and other arbitral institutions were markedly republican, as this article’s corresponding author has argued elsewhere.⁴⁷ The following part reports on relevant data in relation to these types of themes.

2 Descriptive Statistics

Even without the aid of advanced analytical tools, it is possible to make a number of insightful observations with inter-state arbitration from the data contained in Stuyt’s book, beyond those that already have been made elsewhere.⁴⁸ This part starts with a few general observations about the arbitrations themselves, followed by more specific observations relating to arbitrators and then the parties. These descriptive statistics were a helpful guide in figuring out where to focus the efforts with inferential statistics using more advanced methods, as reported in Part 3 below.

2.1 General Observations About the Arbitrations

This article focuses on 316 inter-state arbitrations between 1794 and 1989 surveyed in Stuyt’s book that ended in a decision,⁴⁹ which are the units of analysis for this article. At a minimum, this article adds something new to the literature by slicing and dicing the Stuyt data in various ways in an effort to find informative patterns concerning inter-state arbitration, its users, and its operators.

2.1.1 Variation in Timing

Starting with a simple bar graph, Fig. 1 divides the data by year. In this way, one can see how arbitrations clustered around time periods. This figure shows that the number of arbitrations generally increased leading up to the Conference in 1899, with that number declining soon thereafter. The dotted vertical line in the middle indicates the point in time when the 1899 Hague Peace Conference took place.

Fig. 1

Bar graph showing chronological distribution of the number of inter-state arbitrations between 1794 and 1989

A chronological distribution of the number of inter-state arbitrations between aristocratic states and republican states in unveiled in Fig. 1. From both the bars and the curved line, it is possible to divide of the period into three windows: the development window from 1798 to about 1850s, the peak window from 1850 to 1950s, and the final window from 1950s to 1990. Most of inter-state arbitrations settling disputes between aristocratic states and republican states were captured in these windows. The growing trend reached its peak at the year of 1899 Hague Peace Conference and several years after that. Since then, states from both camps started to have less and less issues brought to inter-state arbitration until reaching the plateau of final window.

Myriad factors likely influenced this drop in inter-state arbitrations. Explaining the drop falls outside this article’s scope. Suffice it to say that the two world wars combined with the 1918–1920 flu pandemic, the rise of alternative fora for legal resolution of inter-state disputes like the Permanent Court of International Justice, and the growth of mixed tribunals like investor-state arbitration that decreased the importance of state espousal of citizens’ claims through diplomatic protection all likely contributed to the decrease in inter-state arbitrations.

2.1.2 Variation in Parties

A large percentage of the 316 arbitrations of the study involved at least one aristocratic party. Out of 316 arbitrations, 148 of them had at least one aristocratic party (or 46.84 percent). In other words, participation by aristocratic parties seems completely random, since they roughly had an equal chance of involvement as not. In more statistical language, the standard deviation around the average proportion of their participation was equivalent to a coin flip (Fig. 2).

Fig. 2

Summary statistics for inter-state arbitration database

2.1.3 Variation in Disputes

The arbitrations of this project involved disputes on a wide range of topics. The topics were limited to 15. The y-axis of Fig. 3 below shows the number of arbitrations dealing with that particular subject. As shown along the x-axis, border disputes, personal claims, and freedom of the seas represented the 3 main topics. On average, each category involved 21 cases—subject to the variation shown in Fig. 3. The top 5 subjects consisted of about half of all inter-state arbitrations. There is not sufficiently detailed data to determine which types of cases involved which classes of parties—for example, whether a border dispute involved an aristocratic claimant (such as a duke seeking more territory), a non-monarch (such as seeking political support from an irredentist faction) or the class of a member/head of the arbitration panel (such as a pope brokering a deal between two monarchs).

Fig. 3

Issues addressed by inter-state arbitration

Border disputes represent seemingly obvious—yet surprising—reasons for inter-state arbitration. During the nineteenth and twentieth centuries, the modern international system as known today slowly evolved. During the ending part of our time period, the balance of power guaranteed by the Treaty of Westphalia—combined with the inter-marriage of dynamic royal clans—was a way of ensuring stability. By 1899, international trade and the peace brought by a kind of mandate or trustee view of administering the colonies brought about a peace which only skirmishes (or world-wide conglomerations) tested. Freedom of the seas represented another area where inter-state arbitration had been sealing off the end of the high colonial era. By 1800, most conflicts at sea, rights at sea, and territorial claims involving large bodies of water already had been resolved. The United Kingdom continued its dominance of the high seas. Arbitration made only major adjustments to the general system of triangular trade that was powering the world’s economy until the First World War. With the exception of an arbitration involving the Rainbow Warrior, the latest arbitration involving freedom of the seas appeared in 1944.

2.1.4 Variation in Compliance

The amount of non-acceptance and non-compliance in the database makes any interesting analysis almost impossible. Parties almost always complied with arbitration decisions—whether under the PCA regime or not. Aristocrats seemed to accept arbitral decisions far less often than their non-aristocrat peers. However, again, because the rate of acceptance is so high, the differences in these rates seem to matter less. A statistician would want to know two things. First, are these rates of accept really different, or just an accident reflecting random events occurring at the time of each acceptance? The few differences that exist in the data are small that common sense leads one to conclude that they likely are similar. Second, do they reflect chance? Would a betting person really “beat the house” by betting on non-aristocrats accepting their arbitrations all night long?

The variation in the decisions accepted in Fig. 4 shows just how fragile the international system of arbitration has been for the past two centuries. The variation of both acceptance rates by aristocrats and non-aristocrats shows that 51 percent could feasibly lie within the parameter’s estimation zone. In other words, even though there has been 74 percent of the non-aristocratic states accepting arbitral decisions in the past centuries, it is possible to assume enough variation exists in this data to make 50-percent acceptance a real possibility. The paucity of data and the variation in the outcomes of each case make more precise averages impossible to obtain.

Fig. 4

Acceptance and compliance rates by aristocrats and non-aristocrats

Compliance and acceptance did not vary significantly over time, type of arbitration or group involved in the arbitration. One could speculate that—unlike private arbitration—public international arbitration involves an implicit threat of violence. The alternative to arbitration definitely must have been higher than the arbitration observed. Otherwise, one logically would assume parties to take the easier or cheaper route. Compliance agreements often have penalties and other disincentives, which clearly must have worked historically. Looking at compliance, one sees that the 1899 Hague Peace Conference had no effect at all on changing the already almost complete compliance with arbitral decisions. Interestingly, the variation in the data suggest that—had other events happened or negotiations changed in some way—compliance figures could have fallen greatly. Figure 5 reflects the standard deviation in the compliance data. Because these represent pure statistical abnormalities, taken from a possible “range of past outcomes,” it is impossible to know why this data appears or if it is correct.

Fig. 5

No effect of PCA on compliance in inter-state arbitration

2.1.5 Variation in Valuation

The value of the arbitrations might obviously interact with—or influence—the way an aristocrat acts during an arbitration. High stakes arbitrations naturally would have had some sort of an effect on the aristocrat’s personal wealth and professional wealth. High stakes arbitrations also may have had political ramifications. Figure 6 attempts to show the approximate dollar range for many of the arbitrations that could be monetized. In some instances, arbitrations dealt with issues of small monetary value but some with larger value. The diversity of these values is too wide to describe here. As show, most of the value of these arbitrations fell into the range of US$100,000 to US$1,000,000. These values definitely would be much higher if they were adjusted for price inflation or exchange rate changes over time. Without such adjustments, $1 of a reported award shows up in the figure as $1.

Fig. 6

Distribution of inter-state arbitration pay-outs (unadjusted for discounting or exchange rate volatility)

Several trends emerge from this data. First, the low value of arbitrations represents the most obvious issue arising from this data. As discussed, the stated values have not been discounted (when available) or approximated (when expressed in a unit exchangeable in gold, for example).⁵⁰

The question of which social class earns the higher awards arises. Aristocratic parties—namely, the legal person and social organization, not the natural person acting as the claimant or respondent—definitely earned higher arbitral awards. Figure 7 shows the value of awards won by these parties. Aristocratic states won—when non-discounted monetary values could be computed—roughly $1 million more than non-aristocratic states, such as republics and communist states. As explained earlier, when a regime changed from aristrocratic to non-aristocratic, the award’s categorization also would have changed. In several arbitrations, several parties joined the arbitration. The sums in the figure would reflect the total amount awarded to all the winners, if they belonged to the same winning social class.

Fig. 7

The value paid highest and most variable for mixed class arbitration

Two other points bear noting from the data. First, the award granted to socially different kinds of state applicants participating in the same, (or “mixed” applicants) and on the same side, roughly exceeded those by $400,000 on average over the entire period. In other words, the United States and the United Kingdom seeking arbitral relief from any another kind of adversary would have won about 3 times more than the non-aristocratic party-state seeking redress and roughly 12 percent more from the aristocratic adversary or comparator. Second, this figure draws bands reflecting the standard deviation in the data to reflect the differences in these outcomes over the years. In other words, “mixed” parties would have had more variation in the size of their awards (higher values when high and lower values when low). Naturally, the error bands look unsymmetrical because it was not possible to observe negative awards.

Statistics, however, might use a different approach to estimating the value of awards won by parties of various regime types (aristocratic or not) than lawyers might. Figure 8 above showed the best estimates of the value of the award, without tampering with exchange rates of interest rates in order to discount the value of awards. A second approach might consist of finding the proportions of artistocratic regime victories as use that as some frequentist measure of the probability of their likely, future success. Figure 6 shows these actual data. Artistocratically-defined regimes won roughly 40 percent of the time. The non-aristocratic party won about 45 percent and mixed regimes made up the difference. The red box shows this data

Fig. 8

Artistocrats likely winning “By Chance” (not due to systemic features of thearbitration regime)

This section has highlighted some of the general observations concerning the 316 arbitrations at the heart of this study. More relevant for the purposes of this article is the involvement of aristocrats as arbitrators, which is the topic of the following section. This analysis starts with a focus on the aristocratic arbitrators themselves, and then it proceeds to focus on the aristocratic parties, with the aim of identifying patterns and developing hypotheses to test with the statistical analyses in Part 3 below.

2.2 Observations About the Aristocratic Arbitrators

This section focuses on the arbitrators in the inter-state arbitrations surveyed by Stuyt’s book. Of the 306 arbitrations that provide at least some information about the arbitrators,⁵¹ the mean number of arbitrators for these 306 arbitrations is 2.54 arbitrators, with a median and mode of 3, with a range of 1 to 13, and outliers of 7, 9, 10, and 13 arbitrators. When focusing on the 135 arbitrations before the Conference, the mean decreases to 2.34 arbitrators, with the median becoming 2 and the mode 1, with the range remaining 1 to 13, and the outliers decreasing to 7 and 13. This means that the tribunals tended to be smaller before the Conference, with a mode of 1 indicating that sole-arbitrator situations were the most common out of all of the permutations from this period.⁵² When focusing on the 171 arbitrations after the Conference, the mean increases to 2.7 arbitrators, with a median and mode of 3, with a range of 1 to 10, and outliers of 7, 9, and 10 arbitrators. This means that the tribunals tended to get larger after the Conference, with the 3-member tribunals being the most common out of all of the permutations from this period, with the variety in the number of arbitrators decreasing after the Conference. These results suggest that the creation or recognition of certain norms at the Conference may have had an influence on these changes.

This article looks at the level of participation of aristocratic and non-aristocratic arbitrators, and this section focuses specifically on aristocratic arbitrators. Of the 306 arbitrations where something was known about the arbitrators, 187 had some degree of aristocratic participation on the arbitral tribunal (or 61.11 percent). With those 187 arbitrations, the mean of aristocratic arbitrators was 1.58 arbitrators, with the median and mode being 1 arbitrator, and a range of 1 to 7, with outliers of 4, 5, 6, and 7 arbitrators. Broken down into time periods, there were 88 before the Conference, with 99 after the Conference, indicating that there was a relatively small increase in the number of arbitrations in which an aristocratic arbitrator was participating. Before the Conference, the mean of aristocratic arbitrators was 1.51 arbitrators, with the median and mode staying at 1 arbitrator, and the range staying at 1 to 7, with 5, 6, and 7 arbitrators being the outliers. After the Conference, the mean increased to 1.65 arbitrators, with the median and mode staying at 1 arbitrator, and the range decreasing to 1 to 5, with 4 and 5 arbitrators being outliers. These figures suggest that aristocratic arbitrators were in more arbitrations after the Conference and there were, on average, more aristocratic arbitrators in those arbitrations after the Conference. However, these figures only provide a partial picture of what is going on in these arbitrations.

Instead of absolute numbers, the relative numbers in terms of the percentage of aristocratic arbitrators on an arbitral tribunal in comparison with the non-aristocratic arbitrators is more important in terms of gauging the potential for influence by the aristocratic arbitrators. For the 187 arbitrations with an aristocratic arbitrator, the overall mean was 71.13 percent aristocratic arbitrators, with a median of 71.43 percent aristocratic arbitrators and a mode of 100 percent aristocratic arbitrators, with a range of 14.29 percent (or 1 aristocratic arbitrator in 7) to 100 percent, with no outliers. All of the aristocratic sole arbitrators highlighted in Sect. 2.3. below skewed these results towards the higher percentages. Of the 88 with aristocratic arbitrators in arbitrations before the Conference, the mean was 77.29 percent aristocratic arbitrators, with a median and mode of 100 percent, and a range of 16.67 percent (or 1 aristocratic arbitrator in 6) to 100 percent, with no outliers. Of the 99 after the Conference, the mean was 65.78 percent aristocratic arbitrators, with a median of 66.67 percent and a mode of 100 percent, with a range of 14.29 percent (1 aristocratic arbitrator in 7) and 100 percent, again, with no outliers. All of this indicates that, while the number of arbitrations with an aristocratic arbitrator increased from before to after the Conference, the number of aristocratic arbitrators decreased in relation to non-aristocratic arbitrators between these time periods. This suggests that aristocratic arbitrators may have had a lesser chance of influencing the outcome of the arbitrations after the Conference, compared to before, even though aristocratic arbitrators continued to get appointed after the Conference. The following subsections analyze the different types of aristocrats on these arbitral tribunals.

2.2.1 Governmental Officials

Concerning the breakdown of types of aristocratic arbitrator, of the 187 arbitrations that had some degree of aristocratic participation on the arbitral tribunal, the most common was monarchical government officials, including those inside and outside of the diplomatic corps, at 134 instances. Besides a general division between diplomatic corps (including ambassadors, chargés d’affaires, envoys, and lower level diplomats) and non-diplomatic government officials (also included ministers of non-diplomatic agencies, judges and king’s or queen’s counsel), there were too many variations in the types of positions that arbitrators held to keep track of with a reasonable amount of effort. Of those 134, 103 had at least one arbitrator who was a non-diplomat government official (or 76.87 percent), while 44 had at least one arbitrator who served in the diplomatic corps (or 32.84 percent), with 13 arbitrations having both a diplomat and a non-diplomat serving on the tribunal (12.62 percent). Before the Conference, there were 57 arbitrations, 39 of which had at least one non-diplomat (or 68.42 percent), 24 with diplomats (or 42.11 percent), and 6 having both diplomats and non-diplomats (or 10.53 percent). After the Conference, there were 77 arbitrations, 64 of which had at least one non-diplomat (or 83.12 percent), 20 with diplomats (or 25.97 percent), and 7 having both diplomats and non-diplomats (or 9.09 percent). While monarchical government officials can be seen throughout both periods, non-diplomats seemed to enter their own between 1903 and 1930, when their presence dominated many of the arbitrations during that period.⁵³ All of these numbers indicate that non-diplomat government officials played a much greater role in inter-state arbitrations after the Conference, while the level of involvement by aristocratic diplomats stayed roughly the same. While diplomats are connected to their home government through diplomatic cables and correspondence, not to mention loyalty, identity, and culture, it is not difficult to see this increase in involvement of non-diplomat government officials as an effort on the part of the appointing parties or the monarchs asked to appoint arbitrators to choose people with a stronger connection to their respective governments and therefore would be in a better position to mobilize political power to reach an acceptable outcome. These results might also suggest a professionalization of arbitrators, with those with status based on appointment having a role as arbitrator more often than those without aristocratic status through appointment. However, since Part 3 below shows that there was no statistically significant relationship between aristocratic status of arbitrators and the outcome of an arbitration in favor of the aristocratic party after the Conference, it perhaps is more probable that these non-diplomat government officials turned out to be more conservative than anticipated, and intentionally avoided doing anything that might embarrass their respective monarchs, such as letting aristocratic prejudice enter into their calculus when deciding which party wins an arbitration. As reiterated in the conclusion of this article, more research is needed to understand what actually is happening here.

2.2.2 Military Officers

Concerning military officials, of the 187 arbitrations that had some degree of aristocratic participation on the arbitral tribunal, there were a total of 12 arbitrations who had at least one arbitrator with aristocratic status from the arbitrator’s position as a military official of a monarch’s military, with 8 before the Conference and 4 after. This represents a significant drop in the number of arbitrations that military officers were involved in between these time periods, although both numbers are relatively small. However, these number are deceptive due to the fact that there often were multiple arbitrators on a tribunal, and any one arbitration could fit under multiple categories in this section. Individual arbitrators could have multiple ways of being an aristocrat, such as a government position, an aristocratic title, and an aristocratic name all at the same time. For the sake of simplifying the statistical analysis in the following part, one aristocratic individual has been counted under only category, with the categories other than title and name taking priority, and with the status of government official taking priority over military status, based on perceived strength of that status vis-à-vis the others under the aristocratic system generally. Here, there only were two instances where an arbitrator had military status in addition to another type of aristocratic indicator, both involving a government position and both from before the Conference.⁵⁴ In addition, there were three arbitrations with multiple military officers of aristocratic status, adding four more arbitrators to the total.⁵⁵ Of these 18 aristocratic arbitrators with military officer status, there were five arbitrators that had the rank of general or admiral, one arbitrator who was a lieutenant general, one who was a major general, two who were captain, one who was a commander, seven who were colonel or lieutenant colonel, and one who was a first lieutenant. Of the 12 aristocratic arbitrators with military officer status from before the Conference, three were generals or admirals, one was a lieutenant general, one was a major general, none were captains, six were colonels or lieutenant colonels, one was a commander, and none were first lieutenants. Of the 6 aristocratic arbitrators with military status after the Conference, there were two generals or admirals, no lieutenant generals, major generals or commanders, two were captains, one was a colonel or lieutenant colonel, one was a first lieutenant. These numbers indicate that military officers were much less involved with arbitrations after the Conference, with a decrease in all categories except for the category of first lieutenant. Therefore, not only were the absolute numbers decreasing, but the average rank of the aristocratic arbitrator with military officer status decreased. As mentioned in the previous section, such observations may signal the professionalization of the arbitrator corps within inter-state arbitration, inasmuch as military officer may have not been especially well suited for such sensitive matters as legal proceedings.

2.2.3 Individuals with an Aristocratic Title

Concerning individuals with aristocratic titles, there was at least one arbitrator with an aristocratic title as the only indictor of aristocratic status in 32 out of the 187 arbitrations with some degree of aristocratic participation (or 17.11 percent). Of those 32, 15 were before the Conference,⁵⁶ with 17 after,⁵⁷ which essentially is evenly divided between these two periods. Again, these numbers are deceptive, for the reasons stated in the previous section. The relatively low numbers of aristocrats identified by the title should not be interpreted as these individuals not having had significant influence in monarchical governments or on inter-state arbitrations, but rather that they were so well connected that their government-official status was used much more often to determine aristocratic status than by using their title. If absolutely all aristocratic titles were to be counted for all arbitrations that had some degree of aristocratic participation, there were 69 arbitrations that had at least one arbitrator with an aristocratic title, out of the 187 arbitrations with some degree of aristocratic participation (or 36.90 percent). Before the Conference, there were 32 such arbitrations, with 37 afterwards, which represents a slight increase in arbitrations with at least one arbitrator with an aristocratic title. There were at least 99 times that there was an arbitrator with a title in total, with 57 arbitrators before the Conference and 42 arbitrators afterwards, indicating that there were fewer with aristocratic titles after the Conference.⁵⁸ Of these 99 arbitrators, 3 had the title of prince or duke, 4 had the title of marquis, 11 had the title of count or earl, 28 had the title of viscount or baron, 7 had the title of baronet or lord, 43 had the title of sir, knight, or chevalier, and 3 had the title of esquire, or their foreign equivalents. Of the 57 before the Conference, 3 had the title of prince or duke, 4 had the title of marquis, 7 had the title of count or earl, 19 had the title of viscount or baron, 2 had the title of baronet or lord, 20 had the title of sir, knight, or chevalier, and 2 had the title of esquire, or their foreign equivalents. Of the 42 after the Conference, none had the title of prince, duke, marquis, 4 had the title of court or earl, 9 had the title of viscount or baron, 5 had the title of baronet or lord, 23 had the title of sir, knight, or chevalier, and 1 had the title of esquire, or their foreign equivalents. These numbers show that the only aristocratic titles that increased in involvement in inter-state arbitration after the Conference were baronets, lords, sirs, knights, and chevaliers, with all of the others dramatically decreasing between these periods, except for esquires, which only slightly decreased. This suggests that aristocrats had to settle for the involvement of lower level aristocrats on these arbitral tribunals after the Conference, perhaps to avoid antagonizing their non-aristocratic counterparts. Alternatively, individuals with the other titles may have become preoccupied with other matters than inter-state arbitration or individuals with those titles became less common, although there is little from the historical record to support that either of those scenarios actually took place.

To look at this group of 99 arbitrators with aristocratic titles in a different way, 58 arbitrators also had aristocratic status through a different way than through aristocratic title, such as government-official status, with 34 before the Conference and 24 after. Of these 58 arbitrators, 40 had the status of a governmental official of a non-diplomatic nature (or 68.97 percent), 14 had the status of a governmental official of a diplomatic nature (or 24.14 percent), and 4 had the status of a military official (or 6.90 percent). Before the Conference, of the 34 arbitrators, the primary aristocratic status came from being a governmental official of a non-diplomatic nature on 18 occasions (or 52.94 percent), a governmental official of a diplomatic nature on 13 occasions (38.24 percent), and a military official on 3 occasions (or 16.67 percent). After the Conference, of the 24 arbitrators, the primary aristocratic status came from being a governmental official of a non-diplomatic nature on 22 occasions (or 91.67 percent), a governmental official of a diplomatic nature on one occasion (or 4.55 percent), and a military official on one occasion (or 4.55 percent). These numbers show that arbitrators with an aristocratic title also fit in fewer aristocratic categories after the Conference than before. The number of governmental officials of a non-diplomatic nature with an aristocratic title at the same time increased after the Conference, while the number of governmental officials of a diplomatic nature with a title decreased significantly. It is difficult to infer meaning from these changes, given how the numbers for governmental officials of a non-diplomatic nature and of a diplomatic nature were moving in opposite directions. Nevertheless, those changes could be explained by those selecting arbitrators caring more about an aristocrat’s perceived proximity to monarchical power than by titles, even though it is possible that certain titles also showed proximity to power. Part 3 below explores whether any relevant patterns can be found in this area.

2.2.4 Individuals with an Aristocratic Name

Finally, a few arbitrators had aristocratic status on account of their name having indications of such status, including such prepositions as “de” and “von,” with the individual being born into and having the nationality of a state with a monarch and maintaining that nationality at the time of the arbitration. Again, if an arbitrator was an official of a monarchical government, a military officer of a monarch’s military, or possessed an aristocratic title at the time of the arbitration, those were used as the basis to determine their aristocratic status, instead of their aristocratic name. Of the 187 arbitrations with some degree of aristocratic participation, there was at least one arbitrator with an aristocratic name as the only indictor of aristocratic status in only five arbitrations, with two from before the Conference,⁵⁹ with three after.⁶⁰ Again, these numbers are deceptive inasmuch as there often were multiple arbitrators on a tribunal, and any one arbitrator could fit under multiple categories in this section, as already explained above. To these five instances where an arbitrator’s name was the basis for determining aristocratic status can be added the one arbitration from before the Conference where there were two arbitrators with an aristocratic name being the only indicator of aristocratic status,⁶¹ as well as the three instances where an arbitrator had an aristocratic name in addition to another indicator of aristocratic status, such as a military officer in a monarch’s military,⁶² and a governmental official of a diplomatic nature.⁶³ Concerning the two instances where an arbitrator had both an aristocratic title and name but no other aristocratic distinction,⁶⁴ the arbitrator was listed under the category of having an aristocratic title, in order to facilitate the appropriately advanced statistical analysis provided in Part 3 below.

Noticeably absent from all of this analysis is any mention of monarchs, who actually played a significant role in inter-state arbitration, both before and after the Conference, but only as sole arbitrators. The following subsection on sole arbitrators highlights this involvement of monarchs, as well as the other instances where other types of aristocrats served as a sole arbitrator, with the sheer number of these types of arbitrations justifying their separate treatment.

2.3 Observations About Sole Arbitrators

Of the 306 inter-state arbitrations with information about the arbitrators, 120 involved a sole arbitrator. Coincidentally, there were exactly 60 sole-arbitrator situations before the Conference and 60 after. Therefore, the Conference does not appear to have influenced use of sole arbitrators. Before the Conference, 41 of those 60 sole arbitrators were aristocrats (or 68.33 percent), whereas there were only 19 after the Conference (or 31.67 percent), representing a significant decline in aristocratic sole arbitrators after the Conference, although this is not especially meaningful for determining aristocratic influence overall because aristocratic participation could have increased in multi-member tribunals, as discussed above. As explained in the methodology provided above, these aristocrats were made up of monarchs, monarchical government officials including non-diplomats (including judges, and king’s or queen’s counsel) and diplomats, military officials, individuals with aristocratic titles, and individuals with aristocratic names who were born in an aristocratic state and had the nationality of an aristocratic state at the time of the arbitration. The following subsection looks at the types of sole arbitrators in the abstract, with the next subsection focusing on the types of sole arbitrators in light of the aristocrat/non-aristocrat identity of the parties, both with the aim of finding patterns in these groupings.

2.3.1 Sole Arbitrator Based on Type

This subsection looks at the types of sole arbitrators that were appointed, overall as opposed to relating to the types of parties involved, which is addressed in the following section. Starting with monarchs, not surprisingly, monarchs never served on tribunals, although a few tribunals were royal in establishment, with the UK monarch being the most active, with three.⁶⁵ Princes occasionally served on multi-member tribunals,⁶⁶ although such involvement is inherently different from monarch involvement, inasmuch as princes simply are not sovereigns, although they would have been aristocrats. The involvement of monarchs as sole arbitrators is notable, with 26 out of 120 sole-arbitrator situations (or 21.67 percent), especially from before the Conference with 18,⁶⁷ compared to after with 8.⁶⁸ Before the Conference, Spain’s monarch was the most active as an arbitrator, serving on five separate instances between 1857 and 1898, with Queen Isabelle II serving once, King Alphonso XII serving twice until his death when Queen Regent Maria Christina took over, King Alphonso XIII serving once, and Queen Regent Maria Christina serving again as sole arbitrator.⁶⁹ All but one party in those arbitrations were a former Spanish colony, which suggests that at least some states were willing to let bygones be bygones in allowing their former colonial masters continue their influence, or perhaps there were elements of neo-colonialism at play. Russian and Belgian monarchs each had three arbitrations before the Conference, with Tsars Alexander I, II, and III each having their own arbitration,⁷⁰ and Leopold I having three arbitrations between 1858 and 1863, each involving the United States or the United Kingdom as one of the parties.⁷¹ Pope Leo XIII had two arbitrations, while The Netherlands’ King William I and King William II had one each.⁷² The Austrian, British, French, German, and Prussian monarchs all had one arbitration each.⁷³ This distribution of inter-state arbitrations between states might explain why Russia and Belgium had leadership roles in the promotion of international arbitration and the creation of a permanent arbitral institution during the Conference,⁷⁴ because their monarchs had considerable experience with it. Despite Spain’s involvement in inter-state arbitration prior to the Conference, it did not have much of a role in the Third Commission of the Conference where international arbitration was discussed, presumably on account of it having lost its empire by losing the Spanish-American War in 1898, and with that its confidence on the international level.

The period after the Conference saw the involvement of a few new monarchs, including those of Sweden and Saudi Arabia.⁷⁵ Italy’s Victor-Emmanuel III also served on three arbitrations.⁷⁶ Despite having lost its empire, Spain’s King Alphonso XIII maintained his high pace of handling arbitrations after the Conference, serving on two, again, all involving at least one former colony from Spanish-speaking Latin America.⁷⁷ Likewise, the monarchs of Belgium, Germany, and the United Kingdom, as well as Pope Pius X, maintained their involvement after the Conference, each with one arbitration.⁷⁸ The last monarch to serve on a tribunal was in 1931,⁷⁹ which presumably marks the further professionalization of international arbitration.

Of these 26 arbitrations involving monarchs as the sole arbitrator, eight involved disputes between only aristocratic states,⁸⁰ five involved disputes between only non-aristocratic states,⁸¹ and 13 involved disputes between a mix of aristocratic and non-aristocratic states in the dispute.⁸² Before seeing those figures, one might have assumed that monarchs would be the most biased in favor of the aristocratic party, given their presumed patronage arrangements with other aristocratic systems and aloofness to concepts like the rule of law. When considering how many non-aristocratic parties agreed to a monarch being the sole arbitrator, either where only non-aristocratic states were involved in the dispute or there was a mix of aristocratic and non-aristocratic states in the dispute, it seems that non-aristocratic parties were not worried about such prejudice, or else they would not have consented to have the monarch serve as the sole arbitrator. As Part 3 below shows, suitably advanced statistical analysis of these figures shows that monarchs were not prejudiced in favor of aristocrats in a statistically significant way. As the following paragraphs and other parts of Part 3 show, the same might not be able to be said about the other types of aristocrats who served on arbitral tribunals, at least concerning before the Conference.

Even more common than monarch involvement as a sole arbitrator is the appointment of a representative of a monarch as a sole arbitrator, which takes myriad forms. The first example involves where the arbitrator is an aristocratic government as a whole, which was not common in the inter-state arbitrations between 1794 and 1989. In practice, it presumably fell upon one of the government’s designated representatives to serve as the sole arbitrator, as opposed to the whole body acting as the arbitrator. The aristocrat-dominated Senate of Hamburg handled three arbitrations between 1855 and 1864, all of which involved the United Kingdom as the claimant, with Portugal as the respondent in two of them and Peru in one.⁸³ The Government of Spain also played that role in an arbitration between 1886 and1888.⁸⁴ More common representation of a monarch involved government-related individuals. Non-diplomat government officials served as the sole arbitrator on eight occasions,⁸⁵ with four involving two aristocratic parties,⁸⁶ and three involving just one aristocratic party.⁸⁷ All eight involved European states, except one between Russia and the United States, which suggests that European states largely trusted ministers from other European states, as one might expect with the uniting-Europe aristocratic peace movement seen at end of the nineteenth century.⁸⁸ Members of the diplomatic corps, including ministers of foreign affairs and staff, ambassadors, envoys, and colonial governors, served as the sole arbitrator nearly twice as often, with 15 sole arbitrations,⁸⁹ with parties coming from more diverse parts of the word. In terms of noteworthy observations, two must be made here. First, up until the end of the First World War, all but one arbitration handled by the diplomatic corps had at least one non-European state as a party,⁹⁰ including five without a European party at all.⁹¹ Second, this involvement of the diplomatic corps as the sole arbitrator would increase for European states and decrease for non-European parties after the First World War, where four of the five arbitrations after the First World War handled by a member of the diplomatic corps would have both parties as European states,⁹² with the only exception being a dispute between Panama and the United States in 1915–1916.⁹³ Both observations about the reliance on members of the diplomatic corps to handle sole arbitration could be explained by their international connections and experience, at least vis-à-vis their non-diplomatic counterparts. Alternatively, it could have been the increased aristocratic status of members of the diplomatic corps, inasmuch as many of them were aristocratic in their own right, either by title or name, in addition to their appointments from aristocratic governments.⁹⁴ Military officers rarely served as sole arbitrators, with that happening only once, in a 1912–1914 arbitration between two monarchic states Bulgaria and Serbia also over their border.⁹⁵ Although pure speculation, this suggests that military officials might not have been appointed more often due to a perceived inability to act as a neutral umpire in relation to aristocrat and non-aristocratic status, at least not without the tempering influence of other arbitrators being involved. Individuals with the role of king’s or queen’s counsel were even less common in the role of sole arbitrator, with only one arbitration between 1896 and 1901 being noted, involved the United Kingdom as the claimant and Siam as the respondent,⁹⁶ suggesting a degree of trust of such individuals by the United Kingdom and indifference by at least that respondent state. However, the numbers are too small to draw any reliable inferences.

This subsection has focused on the sole-arbitrator situations in general. The following subsection looks at sole-arbitrator situations in light of the aristocrat and non-aristocratic classification of the parties, with the aim of seeing if there are any insightful patterns to explore further with more advanced methods.

2.3.2 Sole Arbitrator Based on Classification of the Parties

Of the 120 arbitrations involving a sole arbitrator, as identified in the previous subsection, 53 of them involved mixed aristocrat/non-aristocratic parties, with 37 involving just non-aristocratic parties and 30 involving just aristocratic parties. Therefore, it would appear that non-aristocratic parties overall preferred sole arbitrators more than aristocratic arbitrators preferred them. More surprising is that 17 of the 37 instances where only non-aristocrats were the parties in the arbitration chose an aristocratic sole arbitrator. Even more surprising is that examples are spread through the entire period, not just in the earlier years when colonial ties might have been first severed. Of course, more monarch sole arbitrators were seen during the period before the Conference, with five,⁹⁷ compared to two after.⁹⁸ Moreover, three of the five had the king or queen of Spain serve as the sole arbitrator, for disputes involving Colombia and Venezuela between 1881 and 1891, Honduras and Nicaragua between 1894 and 1906, and Chile and Peru starting in 1898,⁹⁹ all of which had a colonial connection with Spain in the past. Of the two after the Conference, one had a UK monarch as the sole arbitrator in a dispute involving the United States.¹⁰⁰ The rest were consistent before and after the Conference, with there being three monarchical government officials (two non-diplomats and one diplomat before the Conference,¹⁰¹ as well as one non-diplomat and two diplomats after the Conference¹⁰²), and one judge before and one after.¹⁰³ Besides the greater number of monarchs before the Conference, the other main difference between before and after was that there was one aristocratic arbitrator based on his military position in a monarch’s military after the Conference,¹⁰⁴ with another based on aristocratic title.¹⁰⁵ Again, the main noteworthy point is that non-aristocratic parties agreed to an aristocratic sole arbitrator at all, when they presumably could choose whomever they wanted. These results suggest that non-aristocratic parties did not see aristocratic status as relevant when choosing their sole arbitrators, or perhaps not relevant enough for that factor to be determinative vis-à-vis the other factors. This raises the question of whether this article is making too much of aristocratic status when analyzing inter-state arbitrations between 1794 and 1989. Part 3 explores whether the aristocrat/non-aristocratic dichotomy was a meaningful one, based on statistical analysis of that data.

For those 30 arbitrations where both were aristocratic parties, 22 of them appointed an aristocratic arbitrator (or 73.33 percent). The arbitrator’s aristocratic status was unknown in three of those arbitrations where both were aristocratic disputes.¹⁰⁶ Therefore, in only five instances did two aristocratic parties not choose an aristocratic arbitrator, all of which involved the United Kingdom: two between the United Kingdom and Portugal,¹⁰⁷ another between Belgium and the United Kingdom,¹⁰⁸ a fourth between the United Kingdom and Spain,¹⁰⁹ and a fifth between the United Kingdom and Greece.¹¹⁰ However, all of these parties appointed an aristocratic sole arbitrator in another arbitration,¹¹¹ so it does not appear that they were committed to not appointing aristocratic arbitrators, or perhaps the other two arbitrators were able to hide their aristocratic status or the historical record is incomplete in that regard. Of those 22, ten were monarchical government officials (seven non-diplomats,¹¹² combined with three diplomats¹¹³) (or 45.45 percent), seven appointed a monarch (or 31.82 percent),¹¹⁴ two appointed an individual with an aristocratic name (or 9.09 percent),¹¹⁵ one involved a king’s or queen’s counsel (or 4.55 percent),¹¹⁶ and one a judge (or 4.55 percent).¹¹⁷ This suggests that aristocratic parties tend to trust monarchs and those from within monarchical governments when both of the parties are aristocrats, as one might have expected, unlike with arbitrations involving both non-aristocratic parties, as discussed in the previous paragraph.

With the remaining 53 instances of a sole arbitrator where there were mixed aristocrat and non-aristocratic parties, a similar conclusion can be reached as with the instances where both parties were non-aristocrats—that non-aristocratic parties generally were not opposed to having aristocratic sole arbitrators. In particular, of the 53, 31 arbitrations involved the mixed parties appointing an aristocratic sole arbitrator (or 58.49 percent). Again, these instances were spread throughout the periods before and after the Conference, with 18 before the Conference (or 33.96 percent of the total),¹¹⁸ with 13 after (or 24.53 percent).¹¹⁹ Interestingly, 8 of the 18 from before the Conference involved the United States as a party,¹²⁰ whereas none were observed after the Conference. In fact, before the Conference, the United States never was in a mixed aristocratic and non-aristocratic dispute where the sole arbitrators was a non-aristocrat. That changed after the Conference, with the United States being in three such arbitrations.¹²¹ This suggests that the United States might have been agnostic towards aristocrats as sole arbitrators before the Conference, but that changed after the Conference. An alternative explanation was that the United States arrogantly thought it could win arbitrations even if the arbitrator was an aristocrat,¹²² but that would not explain why the United States selected no aristocratic sole arbitrators after the Conference, given that the United States presumably had no reason to decrease its arrogance after the Conference. A more likely explanation is that the US negotiating power to influence the selection of a non-aristocratic sole arbitrator changed at around the time of the Conference, which would be supported by the fact that the United States won the Spanish-American War in 1898 and came into its own on the world stage at around this time. In US Delegate Frederick Holls’ passionate speech during the Conference to convince recalcitrant states to agree to create a permanent arbitral institution, he referred to the United States as a “new Power,” in comparison to “great European Powers,”¹²³ emphasizing this point that the United States was starting to see itself as a great power around this time. Determining exactly which explanation is the correct one falls outside the scope of this study. In reality, it might be a combination of a few explanations.

Latin American states had a similar change of heart towards the appointment of aristocratic sole arbitrators in their mixed aristocrat and non-aristocratic disputes, with seven instances before the Conference,¹²⁴ with only three after.¹²⁵ In only three instances before the Conference was such a state able to appoint a non-aristocratic sole arbitrator in a mixed aristocrat and non-aristocratic dispute: one involving an 1864–1870 arbitration between Argentina and the United Kingdom over personal claims,¹²⁶ another in an 1894–1897 arbitration between Colombia and Italy over arbitrary acts,¹²⁷ and a third in an 1899 arbitration between the United Kingdom and Honduras over maritime seizures.¹²⁸ Only four instances after the Conference were when Latin American states were able to negotiate a non-aristocratic sole arbitrator: one involving a 1900–1903 arbitration between Italy and Peru over treaty interpretation,¹²⁹ another involving a 1901–1904 arbitration between Brazil and the United Kingdom over a border,¹³⁰ a third involving a 1922–1923 arbitration between Costa Rica and the United Kingdom over a contract,¹³¹ and a fourth involving an 1932–1933 arbitration between the United Kingdom and Panama over the international responsibility of the United Kingdom over the acts of certain policemen. All of these examples of non-aristocratic sole arbitrators in mixed aristocrat and non-aristocratic disputes suggest that Italy and the United Kingdom were willing to appoint non-aristocratic sole arbitrators in their mixed disputes, suggesting open-mindedness on their part, or arrogance or a lack of negotiating power to succeed in getting an aristocratic sole arbitrator. Alternatively, it could have been the Latin American states’ strong negotiating power that led to this result, although it may have had to have been stronger than the U.S. negotiating power before the Conference to get those types of results, which might not be reasonable to assume. Finally, France was the main non-aristocratic state that saw an increase in reliance on aristocratic sole arbitrators with mixed aristocrat and non-aristocratic disputes between before the Conference,¹³² versus after,¹³³ presumably because it had been switching between having an aristocratic and non-aristocratic regime quite often before the Conference. However, again, the numbers are too small to make any reliable inferences on this point.

Of all of these with an aristocratic sole arbitrator in a mixed dispute, monarchs were favored overall, with particular emphasis before the Conference,¹³⁴ as opposed to after.¹³⁵ Monarchical government officials were a close second, with a closer split between the two time periods of before,¹³⁶ compared to after.¹³⁷ Aristocratic judges,¹³⁸ members of the diplomatic corps,¹³⁹ and individuals with an aristocratic title made up the remainder,¹⁴⁰ with all three categories seeing an increase between the periods. Noticeably absent were instances where the parties selected military officers as the sole arbitrator, suggesting that they might have been seen as the most prejudiced against the non-aristocratic party, at least from the perspective of the non-aristocratic party. These numbers also suggest that the non-aristocratic parties seemed to trust monarchs and non-diplomatic government officials to serve as the sole arbitrator, presumably due to their possession of or proximity to political power to reach a decision that ultimately would be respected. Part 3 below explores whether monarchs exhibited any prejudice for aristocratic parties or against non-aristocratic parties in terms of outcome of arbitrations. Regardless, the non-aristocratic parties likely would not have had access to this type of aggregated data or descriptive statistics when deciding on their sole arbitrators. It remains unknown whether they intuited this result, guessed it or selected monarchs and government officials for other reasons, such as colonial connections.

These are the observations that can be made about the sole arbitrators based on the types of sole arbitrators and the aristocratic and non-aristocratic composition of the parties, again, without the aid of advanced software or statistical analysis. While this section has looked at the involvement of sole arbitrators in light of the mix of aristocratic and non-aristocratic parties, the following section looks at the mix of aristocratic and non-aristocratic parties in general, to see if there are any useful patterns for this article’s purposes.

2.4 Observations About the Mixed Parties

The inferential statistics in Part 3 below focus on the instances where there were mixed aristocrat and non-aristocratic parties, inasmuch as that part tries to determine if different types of arbitrators might have favored one type of party over the other in a statistically significant way. Instances of mixed aristocrat and non-aristocratic parties are especially important for this project because one cannot detect aristocrat and non-aristocratic prejudice when parties have the same status as aristocrats or non-aristocratic. Therefore, a few general observations on those particular situations are provided here. Of the 306 inter-state arbitrations between 1794 and 1989 with information about the arbitrators and an arbitral award, 148 arbitrations had a mix of aristocrat and non-aristocratic parties. Of those 148, 99 arbitrations had at least one aristocratic arbitrator (or 66.89 percent), with 44 before the Conference and 55 after. Of those 99 arbitrations, 73 of them ended with an award (or 73.74 percent of the 99 arbitrations and 49.32 percent of the 148 arbitrations), 37 of the awards before the Conference and 36 after. Again, those arbitrations that had both aristocratic parties or both non-aristocratic parties, as well as those arbitrations that did not end in an award, have been excluded from the core statistical analysis contained in the following section because those arbitrations tell us nothing about how arbitrators decide in favor of one party over the other. These 73 remaining arbitrations are the focus of the statistical analysis in the following section.

Before proceeding with the statistical analysis of those arbitrations, it is interesting to note that 40 of the 73 awards (or 55 percent) were at least partially in favor of the aristocratic state, without taking into consideration the percentage of participation of aristocratic arbitrators on the tribunal. These descriptive statistics are meaningful because they show that the data is relatively equal, which means that the analysis need not worry about how to balance the bias naturally caused by unequally distributed Boolean data. From a broader perspective, these types of descriptive statistics suggest, again, without taking into account the exact percentage of aristocratic participation on the arbitral tribunal, that the aristocratic party had a roughly even chance of winning.

In terms of relative participation of aristocratic arbitrators vis-à-vis non-aristocratic arbitrators on average, of the 148 arbitrations overall, the mean was 67.75 percent aristocratic participation, with a median of 66.67 percent and a mode of 100 percent, with a range of 14.29 percent (or one aristocratic arbitrator in 7) to 100 percent, with no outliers. For before the Conference, the mean was 72.19 percent, with again a median of 66.67 percent and a mode of 100 percent, with a range of 16.67 percent (or one aristocratic arbitrator in 6) to 100 percent, with no outliers. For after the Conference, the mean was 64.28 percent, with yet again a median of 66.67 percent and a mode of 100 percent, with a range of 14.29 percent (or one aristocratic arbitrator in 7) to 100 percent, with no outliers. These numbers could explain why aristocratic arbitrators stopped being able to influence the outcome of arbitrations in favor of aristocratic parties after the Conference, as demonstrated in Part 3 below—because the participation of aristocratic arbitrators decreased a considerable degree after the Conference. However, these numbers do not explain why non-aristocratic arbitrators also stopped influencing or being able to influence the outcome after the Conference, since their percentage participation increased on average after the Conference. It could be that non-aristocratic arbitrators became less assertive or ambitious after the Conference, although there is nothing in the historical record to suggest how this could have happened. The simplest explanation seems to be that the prejudice of non-aristocratic arbitrators towards non-aristocratic parties observed before the Conference was merely a reaction to the prejudice of the aristocratic arbitrators towards aristocratic arbitrators during that same period, and the decrease of aristocratic prejudice or influence after the Conference naturally led to a decrease in non-aristocratic arbitrators favoring non-aristocratic parties. In light of the well-known concept of Ockham’s Razor, which states that “when there are multiple possible explanations for an event or result, the simplest is almost always correct,”¹⁴¹ this article posits that that might be the actual explanation for these results. Of course, there is the problem that non-aristocratic arbitrators—and non-aristocratic parties, for that matter—would not have had access to the aggregated data or descriptive statistics contained in this article in order to detect that the aristocratic arbitrators no longer would or could influence the outcome in favor of the aristocratic party. Astute observers at the time might have sensed that the winds were shifting at the Conference, away from aristocrats and towards non-aristocrats. However, it is unliked that even the best observers at that time would have had enough sense, let alone evidence, to muster support for the idea that non-aristocratic arbitrators now could give up their prejudice in favor of non-aristocratic parties because the Conference had leveled the arbitral playing field. Part 3 explores these types of topics using statistical analysis.

2.5 Conclusion for this Part

This part has provided the descriptive statistics relating to inter-state arbitration gleaned from Stuyt’s monumental book. Occasionally, there were a small number of examples for a particular point, in which case the strength of the observations were relatively weak. With a larger number of examples, the observations were stronger, although the methodology of this part limits the observations to mere suggestions. The following section uses statistical analysis to test some of these observations, as well as to uncover other patterns and relationships that otherwise would have remained hidden without it.

Before proceeding to that analysis, however, it is important to note that all of the data in this part has failed to show any advantages for aristocratic parties from having aristocratic arbitrators involved in their arbitration. Such a result should not shock, particularly given the incentives of aristocrats highlighted in the introduction to this article. One would imagine that non-aristocrats would have a much stronger incentive to use arbitration, inasmuch as it arguably is cheaper, fairer, less risky, more cosmopolitan, and more media friendly when it comes to resolving inter-state disputes, compared to most of its competitors. Yet, the data and the analysis provided in this part has failed to bear out this conclusion. The following part applies statistical analysis of the expanded database that was created for this article in order to determine if aristocrats somehow had an advantage when engaging in inter-state arbitration, and if the probabilities of various forms of aristocratic participation influenced the outcome in favor of the aristocratic party.

3 Inferential Statistics

Statistical analysis can tease out from the data subtler relations between class type of the parties and arbitrators on the outcome of inter-state arbitrations between 1794 and 1989. This part starts with an explanation of the statistical methods used for analyzing the correlation and causal inference contained in this part. It must be noted that some detailed statistical determinations are not shown in the paragraphs below, for the sake of streamlining the analysis for the reader. After providing a brief statistical analysis of the general data provided in Part 2, this part provides the correlation results and the results relating to causal inference.

3.1 Statistical Methods Used

In order to demonstrate the robustness and reliability of analyses provided in this part, this section explains the statistical methods used for analysis of correlation and causal inference. This explanation starts with the assumptions needed to apply the ANOVA test, with a brief discussion of the alternatives to ANOVA, before explaining the significance of p-values and degrees of freedom.

There are three requirements to fulfil before truly applying the Analysis of Variance (ANOVA) test. First, the variables being tested must be distributed normally and in bell curves.¹⁴² Second, the variances of grouped arbitrations must be equal.¹⁴³ Third, each inter-state arbitration must be independent from the others.¹⁴⁴ Based on the nature and procedure of arbitration, it is not difficult to satisfy the third since arbitrations clearly are independent from one another, notwithstanding any reliance on precedent in between arbitral tribunals.¹⁴⁵ The answers to the first two can be provided by Shapiro–Wilk’s test and Levene’s test. If failing those, the Kruskal–Wallis H Test replaces the position of the ANOVA test in the analyze.¹⁴⁶

Other than ANOVA, determining the strength of a causal relationship between one factor and another requires a two-sample t-test. Utilizing that test also requires meeting the conditions of normal distribution and equal variance.¹⁴⁷ If the variances of both variables are not equal, Welch’s t-test is the alternative for the general t-test. Furthermore, when the assumption of normality on distribution is not met, a Mann–Whitney U test can take place for causal inference analysis. It is important to note that the comparison of the correlation coefficient provided in the next section is not simply value comparing. That requires a Fisher’s z transformation and then application of a t-test to confirm if the change is significant or not based on the p-value awarded. This part is fully aware that mentioning numeral statistical test names would only lead to additional understanding difficulties for non-statisticians. Therefore, it was decided to only use general names in this section even when sometimes variants of tests were used. The statistical determinations discussed above on which statistical test shall be utilized were done behind the scenes, and the results are not specifically mentioned in these paragraphs to avoid becoming overly technical for non-statisticians.

While this part still relies on p-values, it is crucial to be aware of the limitation on statistical significances addressed by p-values. Amrhein et al. strongly condemned the wrong interpretation of statistical significance: “[W]e should never conclude there is “no difference” or “no association” just because a P value is larger than a threshold such as 0.05 or, equivalently, because a confidence interval includes zero.”¹⁴⁸ For that purpose, it is important to keep in mind that any results shall not be simply interpreted as no association where more appropriate interpretations are provided in the sections located below. For readers who are unfamiliar with these concepts, a brief explanation of p-value and degrees of freedom might be helpful. Since this article is about aristocracy, using a story about a tea party to explain the meaning of p-value and the application of its related Central Limit Theorem seems apropos. Based on Ronald Fisher’s original, a young patrician at a teatime party argued that adding milk or tea first would significantly change the flavor of a cup of milk tea, which could be determined by the taste.¹⁴⁹A statistician at that party quickly thought of asking the woman to do a blind test of the milk tea to see whether the milk or the tea was added first.¹⁵⁰ However, with only one test, there was a 50–50-percent chance for such the gentleman to luckily but arbitrarily point out the right one.¹⁵¹ If that gentleman had to answer correctly 50 times without getting it wrong once, there would be a 8.88 × 10^–14 percent chance that the gentleman would be lucky enough to get it right every time.¹⁵² That percent is the meaning of the p-value, and an increasing number of trials to confirm the hypothesis is the meaning of degree of freedom (df), the meaning of sample size (N) in statistics, and the application of Central Limit Theorem. Degree of freedom (df) simply aims to add the reliability of statistical analysis and to explain the variability of the statistical tests used. Such information is provided in this part for those readers who have substantial hypothesis-testing knowledge and would question the reliability of the analysis provided. All others can ignore this information. As an aside, these are the fundamental reasons why sample size of 35–50 is so essential for statistical analysis. With these points in mind, the following sections provide the statistical analyses of the data.

3.2 Analysis of General Data

The obvious first step is to investigate the distribution of aristocratic arbitrator involvement in arbitrations throughout this period. Figure 9 provides a bar graph that shows that the percentage of aristocratic arbitrators’ involvement is rather bimodal. However, the overall distribution is in a flat manner, referencing the lines for kernel density estimate. On this ground, there is no significant clustering or concentration of such a percentage around specific values or intervals, meaning that for a dispute between non-aristocratic and aristocratic states, the percentage did not significantly lean on either side of arbitrators’ involvement across the history of inter-state arbitration from 1794 to 1989. The analysis also observed a slight decline in aristocratic arbitrators’ involvement after the 1899 Hague Peace Conference. It might be interesting for future researchers to determine the reason for such a drop and whether that drop is significant.

Fig. 9

Distribution on percentage of aristocratic arbitrators

The only other basic information from Part 2 above that might benefit from more sophisticate analysis would be Fig. 1 above. Figure 10 below displays the chronological changes on percentage of aristocratic arbitrators’ involvement in inter-state arbitral tribunals between aristocratic and non-aristocratic states. The blue line in Fig. 10 uses a method of exponential moving average that smoothened the original, orange line graph. That assists in having a better view of the kernel of short-term trends and reducing the impact of noise or random fluctuations, which in the case of Fig. 10 were those undulated orange waves from 0 to 100 percent during all analyzed years. Although the blue line is continuous, readers must be cautious about the discontinued parts of the orange line, which indicate that there was no inter-state arbitration during such a period of annual time. Figure 10 seems to illustrate that states began to actively consider the insertion of aristocratic arbitrators into arbitral tribunals near the year of 1860. Yet, keeping in mind that around the year 1860, as shown in Fig. 1 above, inter-state arbitrations were not frequently utilized to settle disputes between aristocratic and non-aristocratic states with only 16 cases documented in Stuyt’s records. Hence, it should rather be explained as aristocratic states and non-aristocratic states during this period not sufficiently using the function of inter-state arbitration. After that window of years, states from both camps put more aristocratic arbitrators into the arbitral tribunals, leading to the average percentage of more than 50 percent. Moving afterwards to the years near 1915, the average percentage of aristocratic arbitrators’ involvement started to be less than 50 percent, as before. With these general observations in mind, the rest of this section is divided into reporting the correlation results and another on the results from causal inference analysis.

Fig. 10

Chronological distribution of percentage of aristocratic arbitrators on tribunal

3.3 Correlation Results

This section emphasizes the correlation between different factors. Admittedly, there are an overwhelming number of variables involved with this project in respect to outcomes and aristocratic participation. Given these multiple variants of correlation coefficients, this section relies heavily on Spearman’s correlation coefficient (r_s) between arbitral outcomes in favor of aristocratic parties and factors related to aristocratic arbitrators in those arbitral tribunals. Spearman’s correlation coefficient has its advantage in analysis concerning ordinal variables, which in this case is the variable identifying whether aristocratic parties won or not. In brief, this coefficient allows one to explore whether one factor’s increase has a synchronized positive or negative impact on another factor in a monotonic function, which is different from the linear relationship illustrated by the Pearson’s correlation coefficient (ρ).¹⁵³ With such a test, the r_s value of 1 and − 1 means that there is an absolute positive and negative monotonic relationship, respectively, where the r_s value of 0 tells the reader that there is no relationship whatsoever between those two factors. This subsection starts by looking for overall correlation and then its breakdown into before and after the 1899 Hague Peace Conference.

3.3.1 Overall Correlation and Break Downs

This project started with the hypothesis that there would be a correlation between the percentage of aristocratic participation on a trial and a result in favor of the aristocratic party for the entire period of 1794 to 1989. With a correlation coefficient of − 0.19 (in Column One Row One of Fig. 11 above), it must be concluded that the opposite of the hypothesis is correct—it is more likely to see the increasing percentage of aristocratic participation in the arbitral tribunal leading to the decrease of the aristocratic party’s winning rate, based on the data provided in Stuyt’s book.¹⁵⁴ This overview test, nevertheless, shows there exists a weak positive monotonic relationship of 0.19 between outcomes from aristocratic countries and the involvement of claimant-appointed aristocratic arbitrators (in Column One Row Five of Fig. 11 above), and on the other hand, moderately negative monotonic relationships of − 0.32 for the one for respondent-appointed aristocratic arbitrators (in Column One Row Seven of Fig. 11) and − 0.25 for the one for presiding aristocratic arbitrators. The following two figures break down the time period into the years before and after the Conference, looking for correlation coefficients as the previous figure provided (Figs. 12, 13).

Fig. 11

Overall correlation matrix on aristocratic participation in the arbitral tribunal and results favoring the aristocratic party

Fig. 12

Correlation matrix on aristocratic participation in the arbitral tribunal, results favoring the aristocratic party and other related factors for before the 1899 Hague Peace Conference

Fig. 13

Correlation matrix on aristocratic participation in the arbitral tribunal, results favoring the aristocratic party and other related factors for after the 1899 Hague Peace Conference

These results originally were surprising in that the percentage of aristocratic participation on an arbitral tribunal is weakly negatively correlated (r_s = – 0.21) with results favoring the aristocratic parties before the 1899 Hague Peace Conference (Fig. 12), but it changed to an extremely weak negative monotonic relationship or merely no correlation (r_s = − 0.0052) after the 1899 Hague Peace Conference (Fig. 13). Those results introduce a scientific view of which the aristocratic arbitrators no longer induce any negative effect on the arbitral results for same-type parties after the 1899 Hague Peace Conference. Yet, to reiterate the meaning of the correlation coefficient, this statement does not imply causality between one factor to the other factor. Although all said correlation coefficients except for the ones of respondent-appointed aristocratic arbitrators still were less than the statistically critical values computed by the sample size,¹⁵⁵ these values show an improved association between these two factors in favor of this project’s original hypothesis. Such a result requires further investigation. While this study focuses on the percentage of aristocratic participation in an arbitral tribunal and the outcome in favor of the aristocratic party, other factors cannot be ignored. In particular, the variables concerning appointments from either party need further investigation. For example, when the number of arbitrators on an arbitral tribunal increased, the number of aristocratic arbitrators strongly synchronously increased, and vice versa. At the same time, when selecting aristocratic arbitrators, parties tended to appoint them based on their aristocratic titles, as well as aristocratic status owned via positions in the government and military before the 1899 Hague Peace Conference. However, the appointments after the cutoff would be more likely to be based on aristocratic status earned through government positions, aristocratic names, and aristocratic titles. Parties no longer wished to appoint aristocrats from the military.

3.3.2 Party-Related Variables and Results for Aristocratic Parties

When more detailed variables associated with aristocratic arbitrators related to arbitral parties were taken into account, this analysis found that claimant-appointed aristocratic arbitrators had a weak positive effect on arbitral awards in favor of aristocratic parties no matter whether it was before (r_s = 0.13) or after the 1899 Hague Peace Conference (r_s = 0.16), as reflected in Figs. 14 and 15 below. At the same time, the respondent-appointed aristocratic arbitrators had their moderately negative effect on such award results prior to 1899 (r_s = − 0.34) and afterwards (r_s = − 0.27). Presiding aristocratic arbitrators, from their former weak negative attitude (r_s = − 0.19), turned to stay in a neutral position (r_s = − 0.079) to the aristocratic party after the cutoff of 1899. By comparing the focused correlation coefficients found in Fig. 14 and in Fig. 15, it showed that the arbitral awards issued by inter-state arbitrations no longer have any relationship with aristocratic arbitrators’ participation from both parties, but a game between claimant-appointed and respondent-appointed aristocratic arbitrators appears. In particular, the more claimant-appointed aristocratic arbitrators, the more favorable to aristocratic state parties, and the opposite condition—respondent-appointed aristocratic arbitrators—may face the opposite results.

Fig. 14

Correlation matrix on party-related variables before the 1899 Hague Peace Conference

Fig. 15

Correlation matrix on party-related variables after the 1899 Hague Peace Conference

This project found that there was a correlation between those discussed factors. This raised one more question: did the state types of claimants and respondents have any association with their intention to appoint aristocratic arbitrators? To reply to that question, this project made a percentage on how claimants or respondents are close to aristocratic states using the state’s status as aristocratic or non-aristocratic,¹⁵⁶ and further in this subsection provided two correlation matrices adopting the Pearson’s correlation coefficient as all variables in those matrices were no longer ordinal.

The state types of claimants were correlated with their appointed aristocratic arbitrators both before (ρ = 0.43) and after the 1899 Hague Peace Conference (ρ = 0.49). Simultaneously, the ones for respondents also were correlated before (ρ = 0.28) and after the 1899 Hague Peace Conference (ρ = 0.1). Furthermore, even taking into account the number of arbitrators, the percentages concerning state-appointed aristocratic arbitrators remained at a similar level with their numbers. Finding that both said variables of claimants and respondents were correlated, where the ones of claimants even crossed the statistically critical values calculated from the sample size, the answers given by this subsection also need to be examined by causal inference analysis. The following section provides that analysis.

3.4 Results Relating to Causal Inference

Correlation does not imply causation. For example, if correlation analysis identifies that there was a positive relationship between illiteracy among dinosaurs and their extinction, then causal inference is going to verify if illiteracy caused the dinosaurs’ extinction. After viewing the correlation, the statistical analysis continues with ANOVA tests and t-tests.

Since this project focuses on more than two scenarios regarding results awarded by aristocratic parties, the ANOVA test is capable of handling that. Simply put, ANOVA is a comparison of means on their difference, which is preliminary in determining whether there is a statistically significant causality between the percentage of aristocratic participation in arbitration and the outcome in favor of an aristocratic party. This is looking at all inter-state arbitrations from 1794 to 1989, including those that had all parties being aristocratic, all parties being non-aristocratic, and those arbitrations that had a mix of the two. Once there were results found to be interesting, applicable two-sample tests were applied to conduct a further analysis on whether certain variables can induce a winning or losing of the aristocratic party.

3.4.1 From Aristocratic Arbitrator Involvement to the Winning of the Same-Type Party

To discuss whether the percentage of aristocratic arbitrators led to the aristocratic parties winning for the entire period of 1794 to 1989, this subsection first applied the ANOVA test. Looking at the statistics given by the ANOVA test, a p-value of 0.59303 showed that there was no statistically significant causation between those two factors. To put it in a different way, if there were 100 arbitrations during the analyzed period, 59.303 percent of those would not see the aristocratic participation in the arbitration as leading to a significant change of a particular outcome, including the one favoring the aristocratic party.¹⁵⁷ Admittedly, this is not enough of a result to be deemed sufficiently statistically significant. Therefore, density plots are needed in order to explore the reasons further.¹⁵⁸

As mentioned, the ANOVA test only considers the means of data, which in Fig. 16 refers to those three dashed lines. Therefore, it is necessary to look at the statistics as a whole, with a graph of distribution being the best at showing the meaning of abstract numbers. Looking at the overlapping area of instances where the aristocratic parties were both “winner” and “loser,” as well as the graphical view in their shapes, it is possible to see those arbitrations where aristocratic participation of the supermajority would not make the outcome lean in favor of the aristocratic party, whereas such a bias cannot be seen where aristocratic participation was less than nearly 65 percent. That is indeed the cause of the negative correlation between the two facts shown in the previous section. Again, the ANOVA test only checks whether the differences of means are statistically different enough. Nevertheless, the graph shows that there are overall differences, even though they are relatively small. The underlying reason causing such a negative correlation and the result of ANOVA behind the correlation requires a detailed investigation of other related variables, especially the factors for aristocratic arbitrators appointed by claimants and respondents.

Fig. 16

Density plot of distribution of percentage of aristocratic participation in arbitral tribunals

Yet, the overall picture is not clear enough to understand the impact incurred by the 1899 Hague Peace Conference. The following two figures use the same methodology as above but break down the time period using April 18, 1899, as the date for the last inter-state arbitral award before the Conference as the key point in time: 1794 to April 18, 1899 and April 19, 1899 to 1989. The ANOVA result for the period before the 1899 Hague Peace Conference is 0.32466, which still is less than the statistical significance threshold of 0.05. The ANOVA result for the post-Conference period is 0.99923, which also shows that such differences are not statistically significant. Therefore, density plots are used to further explore the reasons for these results.

The distribution for the pre-Conference period (Fig. 17) is similar to the overall distribution (Fig. 16), inasmuch as it is a bimodal peak at the left side of the graph. According to Fig. 17, the area framed by the blue solid line and x-axis, which reflected the probability of aristocratic states winning the arbitration, is noticeably larger. Locating the “dead cross” of the orange and blue solid lines, it is possible to find if non-aristocratic states have a higher winning rate once aristocratic participation in the arbitral tribunal is higher than about 36 percent. Due to that peak of distribution at the left-hand side in Fig. 17, the mean of “winner” is skewed to that side, making that higher than the “loser” line on that graph. Again, this is why the correlation coefficient computed and the ANOVA’s result were worse for the original hypothesis.

Fig. 17

Density plot before the 1899 Hague Peace Conference

Figure 18 below—with a focus on the post-Conference period—shows a direct result: There was no relationship, both for the correlation discussed in the previous section and causation, between the percentage of aristocratic participation in an arbitral tribunal and an outcome in favor of the aristocratic party. No matter from the ANOVA test result of 0.99923, or from the proximity of the dash lines, the involvement of aristocratic arbitrators appointed by both claimants and respondents is not associated with the outcomes favoring either type of state. The following subsection turns to the loyalty of aristocratic states towards aristocratic arbitrators.

Fig. 18

Density plot for after the 1899 Hague Peace Conference

3.4.2 Loyalty of Aristocratic States to Aristocratic Arbitrators

This section continues to address two causality questions corresponding to the correlation analysis on party-related variables and results for aristocratic parties. The first question is: were the claimants more likely to be aristocratic and the respondent’s more likely to be non-aristocratic, and, at the same time, were claimants more likely to be aristocratic states than respondents were to be aristocratic states? The second question is: did aristocratic arbitrators appointed by either party cause significant influence on the outcome of arbitration favoring either side?

In response to the first question, this subsection continues the analysis by using one-sample t-tests companying visualization of such via density plots. The statistical tests indicated that in an overall manner across all reviewed years of 1794–1989 (Fig. 11), a claimant in an inter-state arbitration was very likely to be an aristocratic state (p = 1.78 × 10^–32, df = 146), where a respondent in an arbitral tribunal was very likely to be non-aristocratic (p = 7.35 × 10^–34, df = 146) (Fig. 19). That implies that it is extremely unlikely to see any outliers out of all available arbitrations coded by Stuyt with an aristocratic state sitting in the respondent seat, and vice versa. In particular, non-aristocratic states were more likely to be brought before an arbitral tribunal to respond to the claim raised by aristocratic states. However, the breaking down into time periods showed that before the 1899 Hague Peace Conference (Fig. 20), there was a clear indication for claimants to be aristocratic (p = 4.34 x. 10^–23 df = 62) and respondents to be non-aristocratic (p = 3.21 × 10^–22, df = 62), but no longer after the 1899 Hague Peace Conference (Fig. 21). However, both claimants (p = 1.75 × 10^–16, df = 83) and respondents (p = 1.38 × 10^–13, df = 83) still had statistical significance to indicate that their state type were specific in aristocratic or non-aristocratic form. Both non-aristocratic and aristocratic states after the 1899 Hague Peace Conference started to raise their issues before arbitral tribunals in place of the conditions before the 1899 Hague Peace Conference where only aristocratic states were willing to initiate arbitral proceedings. Based on those statistical results, this subsection recalls the results given in the previous subsection, and finds echoed changes occurring in aristocratic arbitrators’ involvement in arbitral tribunals as well.

Fig. 19

Density plot for distribution of percentage of state types closing to aristocratic status

Fig. 20

Density plot for distribution of percentage of state types closing to aristocratic status before the 1899 Hague Peace Conference

Fig. 21

Density plot for distribution of percentage of state types closing to aristocratic status after the 1899 Hague Peace Conference

Moving to the second question, two-sample t-tests were done in order to examine the causality concerning both claimant-appointed and respondent-appointed aristocratic arbitrators with the arbitral results favoring aristocratic states, respectively. From the perspective of claimant-appointed aristocratic arbitrators, though claimant-appointed aristocratic arbitrators were statistically unlikely to cause the arbitral victory for aristocratic parties before the 1899 Hague Peace Conference (p = 0.57332, df = 39) or after the 1899 Hague Peace Conference (p = 0.093806, df = 42), the probability of aristocratic parties’ winning was increasing slightly. That change happened since a clear cut of the percentage of claimant-appointed aristocratic arbitrators between winning and losing situations appeared after the 1899 Hague Peace Conference (Fig. 22) in lieu of synchronously before the 1899 Hague Peace Conference (Fig. 23).

Fig. 22

Density plot for distribution of percentage of claimant-appointed aristocratic arbitrators after the 1899 Hague Peace Conference

Fig. 23

Density plot for distribution of percentage of claimant-appointed aristocratic arbitrators before the 1899 Hague Peace Conference

Attention now turns to the side of respondent-appointed aristocratic arbitrators. Before the 1899 Hague Peace Conference, respondent-appointed aristocratic arbitrators significantly caused the loss of aristocratic states (p = 0.03121, df = 40). Figure 24 below illustrates that only when there was no respondent-appointed aristocratic arbitrator could the claimant states achieve their victory in arbitration. After the 1899 Hague Peace Conference, that causation no longer reached statistical significance (p = 0.10832, df = 43) (Fig. 25). The notable decline of influence from the respondent-appointed aristocratic arbitrators was implied by such results.

Fig. 24

Density plot for distribution of percentage of respondent-appointed aristocratic arbitrators before the 1899 Hague Peace Conference

Fig. 25

Density plot for distribution of percentage of respondent-appointed aristocratic arbitrators after the 1899 Hague Peace Conference

In sum, this subsection has indicated that aristocratic states were more likely to be in the claimant position, and respondents were more likely to be non-aristocratic states, where this condition especially appeared before the 1899 Hague Peace Conference but not quite after then. It also found that claimant-appointed aristocratic arbitrators gave favorable results to aristocratic parties, and respondent-appointed aristocratic arbitrators would be in favor of non-aristocratic parties. There exists a change after the cutoff brought by 1899: while the claimants’ aristocratic arbitrators started to show their bias towards aristocratic states, the aristocratic arbitrators appointed by respondents no longer had an extremely strong preference for non-aristocratic states. That said, aristocratic states still remained their relative loyalty to their appointees by appointing them in inter-state arbitration. That, in addition to the situation of non-aristocratic states starting to be claimants and aristocratic states sitting in opposite seats, may jointly be the reasons behind the changes in correlation between aristocratic arbitrators’ involvement from slightly negative to neutral.

In the previous paragraphs, this project has found that there was a moderately positive, significant correlation between party-appointed aristocratic arbitrators and the aristocratic state’s arbitral results. Changes also occurred in the parties’ state types, and the relationship between aristocratic arbitrator involvement and the winning of a same-type party analyzed causality in the previous subsection. With all due analyses of the above pieces of information, this part confidently can assert that the aristocratic states would appoint more loyal aristocratic arbitrators to give at least part of an arbitral award favoring them, and the same thoughts existed in the minds of non-aristocratic states. That hidden game suddenly stopped in 1899, when non-aristocratic states were no longer only satisfied to sit in the respondent seats after that cutoff. In light of that point, instead of looking at aristocratic arbitrators’ involvement from a general perspective, future researchers might want to continue to separate them into the claimant-respondent dimensions.

4 Conclusion

In a way that has aimed to be accessible to non-statisticians, this article has shown that there is a statistically significant relationship between respondent-appointed aristocratic arbitrators on a tribunal and outcome against the aristocratic party, as well as a statistically significant relationship between no respondent-appointed aristocratic arbitrators on a tribunal and outcome in favor of the claimant, but only in inter-state arbitrations before the 1899 Hague Peace Conference for both of these relationships. To reiterate, no such statistically significant relationship was observed in the data after this point in time, suggesting that something at or around the time of the Hague Peace Conference changed their collective attitude. Non-statistical qualitative work by the corresponding author elsewhere suggests that that change might have been due to an increase in competition in the inter-state-arbitration space coming from non-aristocratic elements in society, mostly right before, during, and after the 1899 Hague Peace Conference that created the Permanent Court of Arbitration,¹⁵⁹ although that is irrelevant for the purposes of this article. What is relevant for this article is that aristocratic arbitrators were biased before the Conference and that that behavior changed after the Conference.

This research boldly stands for the notion that the 1899 Hague Peace Conference was important for international relations, both then and in the future, contrary to the argument of some scholars.¹⁶⁰ This article shows with empirical evidence and statistical analysis some of what observers of the Conference were feeling intuitively at that time in terms of the Conference changing international dispute settlement:

Through the Hague convention of 1899, for the first time by a general treaty, nations in effect agreed that under certain circumstances, at least, they were morally bound, as were ordinary corporations or mere private individuals, to submit the merits of their disputes to impartial examination. The old doctrine that the king, as the representative of Deity, could do no wrong and the newer fiction that national governments were sovereign beyond the ordinary gauges of right and wrong and were their own courts of last resort upon the rightfulness of their actions toward other governments, subject only to the arbitrament of war, were measurably impaired, the signatory nations admitting fallibility and agreeing that, composed as they were of an aggregate of individuals, like their component parts they might err, and that the question as to whether they had erred or not could fairly be determined by other human beings, perhaps no wiser, but certainly more impartial than themselves.¹⁶¹

Contrary to this view, though, statistical analysis does not support the notion that those arbitrators who were affiliated with the aristocratic order were biased in favor of parties from their class before the Conference. The opposite appears to have been true—that certain aristocratic arbitrators were biased in favor of certain non-aristocrats before the Conference. What is correct about the above quote is that the Conference was a watershed moment for both international relations in general and inter-state arbitration in particular.

Future researchers might want to explore the time period between 1990 and the present, since the dataset used in this study involved only inter-state arbitrations from 1794 to 1989.¹⁶² For that, reliance on volumes 22 to 34 of the UN Reports of International Arbitral Awards will be helpful in creating a dataset of those arbitrations.¹⁶³ Admittedly, there seems to be some ambiguity over whether the UN Reports are exhaustive enough to create a population census of these arbitrations, unlike with Stuyt’s book. For example, the UN Codification Division’s website gives contradictory signals concerning exhaustiveness of the UN Reports. The second paragraph there provides, “Prior to the publication of the first four volumes of the United Nations Reports of International Arbitral Awards, there existed no systematic collection of such decisions.”¹⁶⁴ One could interpret this to mean that the UN Codification Division became systematic about collecting decisions after the first four volumes, and so the reports are comprehensive after that point. However, the next-to-last paragraph provides the following:

At the time of the preparation of the first volume of [Reports of International Arbitral Awards] in 1948, the decision was made to exclude arbitral awards contained in highly authoritative collections which were easily accessible at the time. However, with the passage of time, the accessibility of the awards in these collections has diminished since many of them are out of print and no longer available, despite the continuing historical and legal significance of the awards. Therefore [sic] forthcoming volumes of the publication will be devoted to a collection of international decisions rendered from the late eighteenth century to the mid-twentieth century.¹⁶⁵

One could interpret that to mean that these UN Reports were not exhaustive for a time, leaving out some awards that appeared elsewhere, but that the UN Codification Division is filling (or has filled) those gaps now that these other publications have stopped publishing some awards. Surprisingly, they mention "from the late eighteenth century," which does not relate to the decision not to publish some awards that appeared elsewhere after the first volume in 1948. More importantly, it is unknown when this second-to-last paragraph was written, so it is unclear when “forthcoming volumes” started to fill the gaps in the past and whether those gaps actually have been filled. Some scholars note that the UN Reports are not comprehensive, although they do not mention the United Nations’ efforts to fill the gaps.¹⁶⁶ Future researchers will want to clarify these types of matters before relying exclusively on these UN Reports as the source of their data.

The late nineteenth century and early twentieth century saw many states with aristocratic systems change dramatically, either through bloody revolutions or more slowly through evolution, as wars drained coffers and populations demanded better living conditions. While aristocratic involvement in inter-state arbitration did not significantly decrease in absolute terms after the 1899 Hague Peace Conference, aristocratic arbitrators’ behavior changed in unanticipated ways. This article has identified some of those changes. Again, further analysis is needed to determine the causes for these changes in behavior. Nevertheless, this article has started the research down the path of greater understanding of the impact of class on inter-state arbitration.