Foreign sovereign immunity and comparative institutional competence.

• Author: Chilton, Adam S.
• Position: Continuation of III. Evaluating the Claims: Findings A. Before the FSIA: Factors Influencing the State through Conclusion, with footnotes, p. 448-486

1. Bivariate Analysis: The Influence of Individual Legal and Political Factors

   Figure 1 graphically presents the results of fifteen bivariate logit regressions that estimate the influence of a wide range of legal and political factors on foreign sovereign immunity decisionmaking. (149) Each box represents a separate regression. (150) The circles are estimates of the effects of the indicated legal and political variables on the likelihood of immunity when those variables change from 0 to 1 (for binary variables) or from their 25th to their 75th percentile values (for continuous variables). (151) Because no statistical estimate can be absolutely certain, we also include a line that indicates each estimate's 90% confidence interval: in 90% of applications of the same sampling procedure, the actual effect will lie within that interval. (152) The circle and line are solid when there is at least 90% confidence of a positive or negative effect on the likelihood of immunity. Otherwise, the circle is open and the line is dotted. Thus, circles show the estimated percentage change in the likelihood of immunity for each variable, black lines to the left of zero mean that the variable indicates a statistically significant lower likelihood of immunity, and black lines to the right of zero indicate a statistically significant higher likelihood of immunity.

   The results do not reveal evidence that political factors systematically influenced the State Department's immunity decisions. The only variable that achieves statistical significance at the 90% level is a legal variable, Corporate Defendant. In cases where the foreign state defendant was a corporate entity, the State Department was an estimated 36% less likely to grant immunity. (154) As explained above, customary international law reflected, and the State Department officially recognized, the restrictive theory of foreign sovereign immunity, according to which foreign states are not immune from suits based on their commercial activity. (155) The negative effect of the Corporate Defendant variable suggests that the prevailing legal standard did indeed systematically influence the State Department's immunity decisions: immunity was less likely when the foreign state was a corporate entity (and thus presumably engaged in commercial activity).

   The fourteen other variables presented in Figure 1 do not have a statistically significant influence on the likelihood of immunity. This suggests that formal relationships, political importance, economic power, and military power did not systematically affect the State Department's immunity determinations. This was true for the variables presented in Figure 1 and for a range of other political variables that we coded as part of this project. (156) Although these results do not prove that there were no political influences on the State Department's immunity decisions, they challenge the conventional account by suggesting that any such influences were not consistent and systematic. (157)

2. Multivariate Analysis: The Combined Influence of Multiple Legal and Political Factors

   Although the bivariate analysis did not reveal systematic political influences, it is possible that some political influences may become apparent after controlling for the effects of other variables. To explore this possibility, we used multivariate logit regression, which controls for a range of political and legal factors simultaneously. The results are presented in Figure 2. The figure presents four separate regression analyses or "models." The circles are estimates of the effects of the indicated legal and political variables on the likelihood of immunity when those variables change from 0 to 1 (for binary variables) or from their 25th to 75th percentile values (for continuous variables), while all other variables are held constant at their mean.158 As with our bivariate analyses, circles and lines are solid when there is at least 90% confidence of a positive or negative effect on the likelihood of immunity. Otherwise, the circles are open and the lines are dotted.

   Model 1 includes the legal variable Corporate Defendant, the formal relationship variable U.S. Ally, the political importance variable UNSC Member, the economic power variable GDP Per Capita, the military power variable CINC, the political system variable Democracy, and the decisionmaker's ideology variable Decisionmaker's Party. Model 2 includes these same variables, but adds the variables U.S. Contacts and US. Plaintiff. Model 3 contains the same variables as Model 1, but adds the variables Borders Communist and Communist State. Finally, Model 4 contains all of these variables. Unfortunately, Models 2 and 4 contain only a small number of observations (thirty-eight) due to the lack of information about U.S. contacts and the plaintiff's nationality for many State Department decisions. This reduces these models' sensitivity to the effects of legal and political factors.

   Consistent with the bivariate results, there is evidence that legal factors influenced the State Department's decisions. Corporate Defendant has a negative effect on the likelihood of immunity in all four multivariate models. The effect is statistically significant in Models 1 and 3, and barely misses significance at the 0.1 level in Models 2 and 4. (160)

   In contrast, the results provide little evidence that political factors systematically influenced the State Department's immunity decisions. However, there is some evidence that as the foreign state's GDP Per Capita increases, the likelihood of immunity increases. This effect is statistically significant in Models 1 and 3. The other variables designed to assess whether the State Department was systematically rewarding allies did not achieve statistical significance in any of the model specifications. This is not only true of the models presented, but also for a range of alternative models that we investigated. (161)

   Overall, the results of these bivariate and multivariate logit analyses suggest that at least one key legal factor--Commercial Activity--influenced the State Department's foreign sovereign immunity decisions. (162) On the other hand, our analyses do not reveal consistent evidence of systematic political influences on the State Department's foreign sovereign immunity decisions. This runs counter to the perceptions expressed in the FSIA's legislative history. (163) Of course, the absence of systematic influences does not preclude the possibility that political factors influenced certain individual decisions. Nevertheless, our analysis raises doubts about the FSIA's underlying assumption that the State Department was insufficiently competent to make immunity decisions based on legal rather than political factors, and that a transfer of authority to the judicial branch was necessary to depoliticize foreign sovereign immunity decisionmaking. This is one side of the comparative institutional competence equation.

   2. After the FSIA: Factors Influencing the Courts' Foreign Sovereign Immunity Decisions

   The other side of the equation is the competence of the judicial branch. The FSIA was motivated not only by the perception that the State Department's foreign sovereign immunity decisions were excessively influenced by political rather than legal factors--a perception, as we have just seen, that may have been exaggerated--but also by a comparative institutional competence claim that the judiciary would be better able to make immunity decisions based on law, not politics. We empirically assess the validity of this claim by applying the same methods used above. While we find evidence of legal influences for certain types of foreign sovereign immunity decisions by the U.S. district courts, we also find strong evidence of political influences. Thus, these findings--together with our findings regarding the State Department presented above--do not support the claim that the courts are more institutionally competent than the State Department to make immunity decisions independent from political factors.

1. Bivariate Analysis: The Influence of Individual Legal and Political Factors

   We again begin with bivariate analysis to explore the potential influences of individual legal and political factors on immunity decisions. Figures 3A, 3B, and 3C graphically depict the results of sixteen bivariate logit regressions that estimate the effects of legal and political variables on the likelihood that a U.S. district court will grant immunity.

   Figure 3A analyzes all district court decisions in the FSI Dataset together. One legal factor--U.S. Contacts--has the expected negative effect on the likelihood of immunity. However, while the bivariate regression results for the State Department's decisions did not reveal evidence of systematic political influences, the same cannot be said of the U.S. district courts' decisions. A wide range of political factors appear to influence outcomes: U.S. Ally, Diplomatic Representation, UNSC Member, GDP Per Capita, U.S. Exports, and OECD Member are all associated with states being between 10% and 30% more likely to receive a grant of sovereign immunity. In addition,

   Judicial Common Space has a statistically significant negative effect on the likelihood of immunity, suggesting that more conservative judges are less likely to grant immunity than more liberal judges.

   The commercial activity exception is the FSIA's most frequently litigated exception to immunity. (165) The FSIA defines the scope of the exception with some specificity. (166) It provides that "[a] commercial activity' means either a regular course of commercial conduct or a particular commercial transaction or act." (167) It does not define the term "commercial" other than to specify that "[t]he commercial character of an activity shall be determined by reference to the nature of the course of conduct or particular transaction or act, rather than by reference to its purpose." (168) The Supreme...