The Reputational Impact of Investor-State Disputes

ABSTRACT

To what extent do alleged violations of international commitments damage state reputation? This article explore this question with specific reference to investor-state disputes arising under the protection of international investment agreements. Its main contributions are threefold. First, building on the political institutions literature, the study places the theoretical importance of information about the rules of the game, and the actions of the participants at the center of analysis. Second, in contrast to prior empirical research, the study systematically analyzes the costs of state involvement in investment treaty arbitration by examining all known disputes. Third, the study addresses the impact of investment disputes on both foreign investment flows and state reputational rankings. We show that the consequences of investment disputes vary with the transparency of the investor-state dispute settlement process. The central implication of these findings for the broader body of literature on international institutions is that reputational mechanisms for effective treaty enforcement cannot be taken as given but instead need to be explored on the basis of a nuanced approach that addresses the pivotal issues of institutional design and information costs.

KEYWORDS:

• BITs
• FDI
• ICSID
• international institutions
• reputation

Supplementary material

Supplememtal data for this article can be accessed here.

Notes

¹ See, for example, Allee and Peinhardt (2011); Büthe and Milner (2014); Büthe and Milner (2008); Elkins et al. (2006); Simmons (2000); Tomz (2007).

² See, for example, Fearon (1997) and Simmons and Danner (2010).

³ See Tomz (2007) and Allee and Peinhardt (2011).

⁴ UNCTAD (2015).

⁵ Gaukrodger and Gordon (2012:8).

⁶ ICSID (2015:7,10).

⁷ UNCTAD (2015).

⁸ North (1990:60).

⁹ Tomz (2007).

¹⁰ Simmons (2000:819).

¹¹ Büthe and Milner (2008:746).

¹² See Büthe and Milner (2009) and Elkins et al. (2006).

¹³ Tomz (2007:86–94).

¹⁴ Schwenzer and Hachem (2011:426).

¹⁵ Allee and Peinhardt (2011:414).

¹⁶ Ibid.: 429.

¹⁷ Parish, Newlson, and Rosenberg (2011:236).

¹⁸ Europe Cement Investment & Trade S.A. v. Republic of Turkey. 2009. ICSID Case No. ARB(AF)/07/02: 177.

¹⁹ Cavallaro and Brewer (2008); Guzman (2008b); Guzman (2008a); Staton and Moore (2011).

²⁰ Knight (1992:59); North (1990:54–60).

²¹ Franck (2005).

²² For example, see Egli (2006); Franck (2005); Kim (2011). For an example of a specific case compare the rulings issued in CME Czech Republic B.V. v Czech Republic and Lauder v Czech Republic: “Final Award in the Matter of an UNCITRAL Arbitration” 2001; “UNCITRAL Arbitration Proceedings CME Czech Republic B.V. (The Netherlands) vs. The Czech: Final Award” 2003.

²³ Amendments to the ICSID’s arbitration rules in 2006, however, mandate the Centre to publish excerpts of the legal reasoning applied by arbitration tribunals in reaching their decisions in specific cases (Antonietti 2006). UNCITRAL has also adopted new rules on transparency effective April 1, 2014, but they only apply to treaties concluded prior to that date at the agreement of the disputing parties. For UNCITRAL disputes brought under treaties concluded at a subsequent date, exceptions to the new rules require the agreement of both disputing parties (UNCITRAL 2013:33–40).

²⁴ North (1990:57).

²⁵ Allee and Peinhardt (2011:419–420).

²⁶ ICSID (2015:7, 10).

²⁷ For each of these models we have 29 observations for every country.

²⁸ We also run a model across the full panel of 112 lower-income countries, including both countries that had disputes and those that did not using fixed effects for each country. The regression coefficients on the dispute variables are consistently positive, rather than negative, and significant at a 95% confidence interval.

²⁹ We gather each of these measures from the World Bank (2013).

³⁰ Political Risk Services Group (2013).

³¹ Chinn and Ito (2008).

³² Specifically, we use the Polity 2 score from the Polity IV project developed by Marshall, Gurr, and Jaggers (2013).

³³ Political Risk Services Group (2013).

³⁴ This is the same exclusion criterion used by Allee and Peinhardt (2011). Specifically, we follow their case selection rule of excluding those countries that were members of the OECD at the beginning of the time period of the analysis. Applying different case selection rules, such as removing upper-income countries as defined by the Word Bank, leads to similar results.

³⁵ The documentation of codings for treaty-based ICSID disputes is available on request from the authors.

³⁶ Given the contrast between our findings and those reported by prior research, we conducted a replication of the study of Allee and Peinhardt (2011), which to our knowledge is the most prominent work on this subject. We find that the significant negative effects of ICSID disputes on FDI flows reported in their work are a result of a coding error. A full discussion of this issue and its effect on their results are presented in the Appendix to this article.

³⁷ For example, see Multilateral Investment Guarantee Agency (2011).

³⁸ Additionally, scholars such as Kerner (2014) and Lipsey and Sjöholm (2005) have noted that FDI data that are commonly used do not always accord well with the theories that political scientists are interested in testing. The specific critique that is put forward by Kerner on the use of investment flow data is that foreign capital investment takes different forms and the level of political risk varies accordingly. In this article, we are interested in foreign investor confidence, however, not vulnerability to political risk. Fixed capital may be more vulnerable than more volatile forms of investment to political risk, but that is not the focus of our work here. All kinds of investment flows, whether they end up in the stock market or investment in land, are indicators of business confidence.

³⁹ We do not advance any hypotheses about the effects of winning or losing disputes because a third of registered treaty-based disputes have not been concluded, and there is no reason to think that those belonging to the concluded set are representative of the broader universe. “Concluded” is also a potentially misleading label even as applied to ICSID cases inasmuch as a dispute decided by an arbitral panel at one point in time is subject to further revision and annulment proceedings as well as supplementary annulment proceedings. Distinction between “concluded” and “pending” cases is accordingly rather blurry.

⁴⁰ Antonietti (2006); Yackee and Wong (2011).

⁴¹ Political Risk Services Group (2013).

⁴² UNCTAD (2013).

⁴³ World Bank (2013).

⁴⁴ Ibid.

⁴⁵ Ibid.

⁴⁶ Political Risk Services Group (2013).

⁴⁷ Chinn and Ito (2008).

⁴⁸ Specifically, we use the Polity 2 score from the Polity IV project developed by Marshall et al. (2013).

⁴⁹ While there are other variables that might be considered theoretically relevant to the study of investment reputation, we have opted for those utilized in prior IPE research with broad country coverage and limited overlap with other variables.

⁵⁰ A simple bivariate analysis shows that the effect is highly uncertain across countries as well.

⁵¹ This corresponds to three for ICSID disputes over the past 2 years, 6 for disputes over the past 5 years, and 10 for the cumulative number of disputes.

⁵² We begin our period for analysis here at 1994 because the infrequency of disputes before that date leads to cases in which no country had a dispute within the last 2 years.

⁵³ As a check on the results from this series of pooled models, we add a binary variable that equals one after 2007 and zero otherwise to each of the models shown in Table 4. We then add an interaction between the binary variable and the dispute measure for each model. In each case, we find that the binary variable and the interaction term have a significant negative effect, at a 95% confidence interval, on reputation. We choose, however, to present the results from the series of pooled models to more clearly highlight the increasingly negative effect of disputes over time.

⁵⁴ As with our analysis involving the ICRG investment profile measure, we also run this analysis using the full data set but interacting the disputes measures with a binary variable that equals one after 2007 and zero otherwise, and here again we find that the interaction and its constitutive terms have a significant negative effect. Additionally, the simulation analysis that was shown in Figure 5 for the ICRG investment profile variable returns similar results for the Heritage Foundation property rights measure. For the sake of space, however, we choose to only present the coefficient estimates for the dispute variables.

⁵⁵ Here again by running this model using the full panel but with an interaction effect, we find that the interaction and its constitutive terms have a significant negative effect. Additionally, a simulation-based analysis assessing the effect of disputes shows similar results as what was shown in Figure 5.

⁵⁶ Beck and Katz (1995).