Investment Disputes and Abnormal Volatility of Stocks
IInvestment Disputes and Abnormal Volatility of Stocks ∗ Jozef Barun´ık † , Zdenˇek Dr´abek , and Matˇej Nevrla CERGE-EI Charles University and Czech Academy of Sciences
Abstract
Dramatic growth of investment disputes between foreign investors and host states rises seriousquestions about the impact of those disputes on investors. This paper is the first to explain in-creased uncertainty of investors about the outcome of arbitration, which may or may not leadto compensation for damages claimed by the investor. We find robust evidence that invest-ment disputes lead to abnormal share fluctuations of companies involved in disputes with hostcountries. Importantly, while a positive outcome for an investor decreases uncertainty backto original levels, we document strong increase in the volatility of companies with negativeoutcome for the investor. We find that several variables including size of the award, politicalinstability, location of arbitration, country of origin of investor or public policy considerationsin host country explain large portion of the investor’s uncertainty.
Following the rapid growth of foreign direct investment (FDI) over the last 30 years or so, invest-ment disputes between investors and host country states have also increased dramatically. Thisis serious since FDI have been the key factor of global economic growth during the period, andinvestment disputes can jeopardize economic prosperity, global fight against poverty, diseases andother public mal (Gonzalez and Kusek, 2018; Newman et al., 2015; Baldwin and Lopez-Gonzalez,2015). Investment disputes have many origins ranging from direct expropriation of investors as-sets to disputes about various regulatory interventions impeding business of the investors. Theyall can seriously affect the profitability of FDI. The resolution of disputes takes place under thelegal provisions of different international investment agreements (IIA) and, most frequently, underbilateral investment agreements (BIT), which stipulate, inter alia the rules and procedures for theresolution of disputes through international arbitration. The main purpose of these agreements isto protect foreign investors even though some, especially some of the more recent ones include ∗ Authors gratefully acknowledge support from the Czech Science Foundation under the GA 18-04630S project.Authors order is alphabetical. † [email protected] a r X i v : . [ q -f i n . GN ] J un lso provisions to improve market access for foreign firms (Dolzer and Schreuer, 2012; Juillard,2000). Investment disputes can be very costly (Hajzler, 2012). Host countries may suffer adversereputational costs, high litigation costs and high costs of compensations. For many poor countries,the costs of litigation can be excessive and the arbitration is sometimes accessible only with thehelp of foreign aid.
Pari passu , an involvement of investors in disputes with foreign governmentscan increase uncertainties about the conduct of their business, and the market may perceive the firmas a greater risk. Should investors lose the arbitration, the disputes will lead not only to litigationcosts but also to lost income. The relative costs to investors will depend on the value of the claimrelative to the size of the balance sheet of the firm and the actual outcome of the dispute.The empirical literature dealing with investment disputes is extremely limited. It has beenalmost exclusively concerned with the analysis of the effects of IIA on FDI. Most of the researchis not strictly about investment disputes because IIA are not only about investment protection butalso about market access, as noted above. Economic studies of investment disputes are few and allare concentrated on three specific issues. First, the research has been entirely concentrated on hostcountries and their benefits from FDI. The second important stream of the empirical literature hasinvolved an assessment of the performance of international arbitrations (Franck, 2007; Van Harten,2011; Schultz and Dupont, 2014). The literature provides an empirical assessment of the litigationexplosion, focusing on (1) identifying the parties to disputes, and the subject of the arbitration;(2) increases in awards, (3) analysis of win/loss rates, (4) of amounts claimed and awarded, (5)arbitration costs. (6) Attempts were also made to compare different dispute resolution processes,(7) the role of nationality and gender of arbitrators and (8) assess the fairness and independence ininvestment arbitration. They also ask about the costs of arbitrations whether the costs are excessivein relation to benefits and how they relate to the levels of income. Third, an alternative approach tothe study of investment disputes is a recent attempt to link macroeconomic indicators and disputes(Bellak and Leibrecht, 2020). On the other hand, there is no empirical literature that would examine the impact of FDI dis-putes on investors. This is unfortunate because the disputes can affect the value of firms as wellas their incentives to relocate abroad. In addition, as disputes are very likely to affect the sharesof investor’s firm, they will also impact their trading in the market. We intend to explore this newarea of research.The aim of our study is to assess the impact of investment disputes on the volatility of thecompany’s shares. The risks and uncertainties originating in investment disputes are likely to The level of protection also varies among BITs (Alschner and Skougarevskiy, 2016). Note that the breach ofcontract can “normally” only be initiated by investors and host countries typically do not have the right to challengebreaches by investors. The latter are subject to standard commercial courts. The literature has been reviewed, for example, in Pohl (2018) and the empirical evidence is synthesized in Bradaet al. (2020). Bellak and Leibrecht (2020) also review the relevant literature. The only exception is a very recent attempt of Brada et al. (2020) who look in their event study at the effectsof arbitral decisions on the real returns of investing companies but the study has not yet been published. The studycurrently exists as a working paper and is under review for publication. The authors find that “in cases where thearbitrators rule in favor of the MNC and award it damages, the stock price and the value of the firm increase by3 percent over the ten-day period following the announcement of the arbitrators decision.... In cases, where thearbitration tribunal ruled against the MNC and awarded no monetary damages, the stock price of the firm declined by2 percent”.
2e reflected in changes in the prices of shares of investors in the market. One approach to theevaluation of the effects of disputes on investors is to estimate real rates of return of the investors.Our approach is different. We document the impact of an investment dispute on the uncertaintyand on the stock price measured by the second moment, its volatility. As uncertainty rises aboutbusiness operations and hence about the risk of the stock, stock price volatility is likely to increasewith the emergence of disputes. Positive resolution of disputes in favor of the investor can beexpected to lead to a reduced uncertainty and, consequently, to lower stock volatility.Academics, policy makers, traders and other financial practitioners have been trying to un-derstand the uncertainty regarding future price fluctuations measured by volatility for decades. Whereas origins of price movements may have good economic sense or can be explained by eco-nomic fundamentals or behavioral psychology, its ultimate causes are poorly known in all spec-ulative markets. Rapidly changing fluctuations can hardly be attributed to a single cause. Theeconomic sources of movements typically include fundamental shocks affecting the economy, theshocks to technology, to consumer preferences, to demographics, natural resources, monetary pol-icy or other instruments. On the other hand, volatility is often attributed to changes in opinionsbased on news, or psychology connected to the confidence or speculative enthusiasm.Uncertainties about the outcome of investment disputes is the key towards understanding volatil-ity of shares of firms involved in disputes that we aim to study. We view this uncertainty as anunusual driver of volatility in excess of “normal” market uncertainty and volatility associated withit as experienced by investors. Hence, as a proxy for the usual price fluctuations, we use broad S&P500 index and we assume that it will help us control all the economic and psychological effects.In addition, it is a widely accepted stylized fact that volatility is dynamically changing over timeEngle (1982); Bollerslev (1986) hence, by controlling for the time dynamics, we aim to test theexcess volatility around the outcome date.Following Białkowski et al. (2008), we employ the GARCH model, which allows us to analyzetime varying volatility during a period preceding the announcement window. We shall examine thepattern of volatility over a long period of time in which the critical day is the day when the outcomeis announced to the market. We make a distinction between disputes won by investors and thosethat went in favor of host states. We also allow for the magnitude of compensation to see whetherand how it affects the volatility. Clearly, an arbitral award that brings in compensation for damagesin excess of the amounts claimed by the investor is likely to lead to “rewards” by the market andvice versa. In addition, the data matching disputes with share prices are relatively limited. Smallsamples create statistical problems, and we are pleased that we were able to work with a relativelylarge sample.The paper is divided into seven sections. The following section 2 provides a brief review of thetheoretical literature with the view of identifying specific origins of uncertainties related to arbitra-tions. Section 3 covers our detailed description of data, its sources and the way the data has beenused in our work. Section 4 deals with methodological issues – especially a detailed descriptionof the GARCH model and the way it is used by us in the paper. The results of estimations are For a comprehensive early survey of the literature see Shiller (1992). It could be argued that poor judgement on the part of investors in the market could be another explanation ofvolatility (e.g. Shiller (2015); Simon (1991); Kahneman (2003)), Fisman and Zitzewitz (2019) have recently proposeda method of capturing the imperfections in in the investment behavior. In other words, one could theoretically dis-tinguish between uncertainty about investment dispute and uncertainty pertaining the mind and decisions making ofinvestors. We clearly are working with the former rather than the latter.
Do disputes matter? Do they matter especially for investors? Investment disputes and their effectshave been discussed at length in the legal literature and typically only in the context of litigationsaddressed by courts in the United States. International disputes originating in international tradeand investment have so far received far less attention in the literature, and among those disputes,trade disputes have dominated the debate ( e.g. Mavroidis and Sykes (2005)).For investment disputes to matter for investors, the disputes must affect – positively or nega-tively – the value of the firm as a result of the dispute. From the literature dealing with litigationsin the United States we know that litigations/disputes matter for investors because they may indeedaffect the value of the firm (Arena and Ferris, 2017). It is only through the understanding of theway disputes are resolved that the market can assess the risks emanating from the dispute for theinvestor.Ultimately, the question about disputes is – what is the value of awards? Until the decisionis taken, the value is uncertain. The value will, first of all, depend on whether the investors willbe at all able to receive compensation for his/her loss of current and future income. Note thatdamages in the ISDS system cover not only losses of income in the past but also losses comingfrom expected income in the future. Compensations in investment disputes are provided in theform of cash payment rather than retaliation as it is the case in trade disputes. This means thatthe award has a clear monetary value – in contrast to awards in the form of retaliations where thevalue is far more complex and ambiguous. On the other hand, the fact that damages in investmentdisputes are assessed as past losses as well as the value of future harms. This makes the evaluationof the damages ambiguous for different reasons. The arbiters must have a solid command not onlyover highly technical details of markets and the balance sheets of the investors but they will alsohave to agree on the costs of money and other matters of discounting. The list of complications in assessing the value of awards does not end with the descriptionof challenges for economists and accountants. The legal literature points to another issue impor-tant for evaluating cash payments: quality of courts and their ability to assess damages. Clearly,courts in different jurisdictions do work differently and differences in their judgements cannot beexcluded.Perhaps the most complex issue related to the evaluation of awards concerns the likely effectof the award on welfare of the plaintiff and the defendant state. There are two, basically opposingviews about the effect. One view, originating in the legal literature, is that litigation leads to welfarelosses to both plaintiff and the defendant. The alternative view is that international arbitrationsmay, in fact, help increase welfare. It will do so through the instrument of IIA which are seen as a For other differences see Ossa et al. (2020). As noted by Brada et al. (2020), the general premise of many studies is that there are joint wealth losses for theplaintiff and the defendant corporations but that the biggest change in value is experienced by the firm being sued(Bhagat et al., 1994; Bizjak and Coles, 1995; Bhagat et al., 1998). By winning the dispute, investor will reduce hislosses. Losing the disputes, his losses will augment. The critics of international arbitration explain the welfare benefits for investors yet from an-other perspective. For example, for Brada et al. (2020), international arbitrations are less risky thanlitigations in national courts (presumably US) and they are biased.... This argument echoes simi-lar views that have been expressed in the literature on trade disputes (e.g. Horn et al. (2005)). Thedifferences suggest that arbitration under a BIT carries less risk and possibly greater rewards forthe plaintiff than does inter-firm litigation in national courts. These views are more or less sharedby other critics such as Sachs and Johnson (2019); Van Harten (2011); Rodrik (2017) and others. In contrast, investment disputes also generate uncertainties about the International System of Dis-pute Settlement (ISDS) . To begin with, ISDS of investment disputes is different from that usedin the resolution of trade disputes (Ossa et al., 2020), they are subject to different provisions ofIIA and those provisions have been evolving over time (Pohl, 2018). This makes it more compli-cated for market participants to evaluate and anticipate the outcomes of arbitrations. Internationalarbitration can be subject to different interpretations of agreed rules, and this, again, increases theuncertainty and unpredictability of arbitrations. The jurisprudence of international investment lawis evolving and may sometimes lead to different interpretations of the same rule (Howse and Levin,2020).The differences in the views about international arbitration have serious implications for economistsand their attempts to assess empirically the effect of disputes on investors. The different percep-tions about expected welfare gains complicate the choice of the dependent variable to be measured– especially the choice between the absolute level of real rate of return and volatility of stock.Given the theoretical differences in the expected changes of welfare, the expected level of the realrate of return is ambiguous. Does the rate have to be necessarily above a benchmark or would someother values be still be acceptable? This ambiguity is one of the reasons for the choice of volatilityas our dependent variable in this study. Other benefits of international arbitration to investors canalso be noted. They range from the value of government commitments to protect future profits ofinvestors to benefits from various signaling functions of government commitments or the benefitsfrom a better control of moral hazard.However, the risks to investors are also considerable. The “classical” problem faced by in-vestors is the expropriation risk , leading to costs of what is known as police carve–outs (Aisbettet al., 2010a): Police carve/outs represent a range of regulations giving the host country govern-ment certain degree of flexibility and allowing it to address a policy issue. The regulation may See. Ossa et al. (2020); Kohler and Staller (2019); Horn and Tanger˚as (2019). To quote Brada et al. (2020), “The pool of arbitrators is made up mostly of individuals from developed countries,which are likely to be the homes of many MNCs, while respondents are frequently developing countries. If there issome arbitrator bias in favor of the legal systems of developed countries or of the rights of investors from developedcountries, there is the possibility that plaintiffs are likely to obtain more favorable awards due to such bias on the partof the arbitra-tors. Of course, the fact that arbitrators may reduce awards below the amounts claimed by plaintiffs doesnot disprove that they may favor foreign investors in their decisions because plain-tiff investors have strong incentivesto overstate their claims, both to get large judgments and to encourage respondent states to settle claims rather than torisk large losses in arbitration. Thus, even if arbitrators do reject some of the damages claimed by the plaintiffs, theirawards may still be more favorable to plaintiffs than the law and the facts of the case warrant.” It is not clear why costs and benefits of national litigation are comparable with costs and benefits of internationalarbitrations when we discuss the effects of the latter on investors. Investors are not typically facing a choice betweenlitigation in their national courts or in national courts abroad. International arbitration is their only option in thosecases. in addition to the value the country generates for itself by addressingthe public policy issue. Another reason for high costs of investors are excessive regulatory takings. Host countrygovernment’s regulations may be “weak” or “strong”. In the latter case, we say that regulatorytakings are excessive because they disregard the (welfare) interests of investors. As noted byAisbett et al. (2010b),the lack of regard for the investor’s welfare may lead to inefficient over-regulation by the host even in the absence of a “hold-up” motivation. The same argument is madeby Horn and Tanger˚as (2019) in their paper on optimal investment agreements.The third reason why the costs of investors may be understated is the fact that the costs of hostcountry’s regulations are seen by host country governments as being “external”. In other words,host country governments do not typically internalize the costs that local regulations impose onFDI. Those costs have to be born by the investor and even though the investor may attempt to passthe costs onto consumers in the country they may not be able to do so due to competition. The fourth risk faced by investors is the enforceability of the arbiters’ decision even when thedecision is in their favor. In theory, the decision can be enforced only if the host country hasthe resources to pay for the damages and when its acceptance of the decision is driven both byits respect for the court and by its concerns about its reputational costs. The latter could be highand affect the host country’s attractiveness to future business. Even though the costs of sovereigndefaults have lately increased Schumacher et al. (2018), and hence their willingness to complywith the decision of the court, the risk of non-payment for damages remains high.Finally, there are other costs that need to be mentioned such as currency risk that can be signif-icant in countries with unstable currencies. National competitors do not normally take such risks.In addition, considering differences in linguistic, cultural and historical backgrounds, it is arguablewhether the competitive conditions in the market are on a level playing field even if national treat-ment and ( and MFN) are formally a part of the conditions of establishment and operations. Thereason is that NT and MFN provisions primarily focus on taxes and other financial instrumentsand completely disregard the role of social, cultural and historical factors. Moreover, the legallanguage of IIA is often vague. E.g. The NT provision, which calls for the same conditions “inlike circumstances” is clearly one example of many that have created many controversies in theWTO DSB mechanism and in international arbitrations. In conclusion, the adjudication of claims made by investors and the evaluation of awards areclearly a daunting task for arbiters of investment disputes. The task is complicated from momentwhen the arbiters have to decide whether and which investor’s property right has been violated.They then have to demonstrate that the right has been violated by the host country policy interven- Much of the recent literature on optimal investment agreements looks at ways of designing an optimal compensa-tion mechanism in cases of indirect expropriation under IIA. For a review of the literature, see Aisbett et al. (2010a). The literature on regulatory takings originates in the paper by Blume et al. (1984) and it would be outside scopeof this paper to review it. As pointed out by Aisbett et al. (2010b), cost-internalization is the leading justification among legal scholarsfor the U.S. Fifth Amendment, which states that private property shall not be taken for public purpose without justcompensation. Aisbett et al. (2010b) makes a similar point when they argue that “.. the treaties’ investor to state provision and theuse of international tribunals create an asymmetry between domestic and foreign investors, causing their circumstancesto be inherently “unlike”. Thus, there is a tension between the expropriation and the non-discrimination clauses.”
The research on investment treaty arbitration lacks a comprehensive data set of all disputes filed orresolved. Transparency, the quality and public availability of investment disputes is improving butthe data must still be read with caution. There are various sources of data on investment disputes -some are related to data on arbitrations produced by Oxford University Law Center and publishedby Oxford University Press or by Kluwer Arbitration, other sources originate in data on IIAwhich cover, inter alia, information about dispute settlement mechanism. The most frequently usedsources of data on investment arbitration are documents released by the United Nations Conferenceon Trade and Development (UNCTAD), and International Centre for Settlement of InvestmentDisputes (ICSID). UNCTAD reports only “known” cases and will, therefore, exclude cases thatare not publicly known. In contrast, the ICSID statistics are also under-inclusive because they onlyrely on data from cases covered under the ICSID jurisdiction. According to one estimate, theICSID cover about 70 percent of all disputes. Thus, the UNCTAD data are more “inclusive” thanthe ICSID data but are also likely to be incomplete. In addition, neither of the two sources coverseach reported case with full information about the arbitral process, and we had to combine the twosources. Despite the weaknesses, the documents are the best efforts representation of trends ininvestment arbitration and should not significantly skew our analysis. There is also a recent effortto provide a comprehensive list of all known cases in the form of publicly available dataset fromPITAD Investment Law and Arbitration Database. This source aims at providing an exhaustivedata set of all cases but, on the other hand, it lacks some of the information regarding the individualcases which are provided in other sources such as UNCTAD or ICSID.We work with the versions of the documents from January 2019. We scraped the data from therespective web pages and had to merge the information when constructing our database in order toobtain the information necessary for our estimations. On the one hand, the ICSID database containsmore detailed information regarding some of the variables. For example, it contains the exactdate of the registration of the dispute, which is needed as a control date in our experiments. Theinformation is not available in the UNCTAD source. On the other hand, ICSID does not explicitlycontain information regarding the final decision of the court and the amount claimed/awarded,another important piece of information in our analysis. As a check of our merged data, we verified OUP Press Investment Claims: http://oxia.ouplaw.com Kluwer Arbitration : http://kluwerarbitration.com https://investmentpolicy.unctad.org/investment-dispute-settlement https://icsid.worldbank.org/en/Pages/cases/AdvancedSearch.aspx For more discussion of differences between both sources, see Shirlow (2015). https://pitad.org/index . See also With the exception of China, all of the investor’s countries are developed coun-tries. As for the host countries, seven cases come from transition countries and thirty three fromemerging markets.Information regarding the stock prices was obtained from Yahoo Finance. In order to isolatethe effect of investment dispute resolution from general market movements, we removed the effectof the aggregate market from the stock price of the investor’s firm. We use the S&P500 index asa proxy for general market changes because most of the companies in our sample operate interna-tionally and the S&P500 market is a good proxy for the stock market globally. By using one singlebenchmark, we also limit the effect of differences in stock price volatility in different securitiesmarkets. Stock prices employed are either direct prices of the company that made the investment,or prices of the parent’s company. The size of the sample was constrained by a number of other factors. In some arbitral proceedings, the arbitratorsdeclined jurisdiction over the dispute. Many cases could not be used in our sample because the plaintiffs (individualsor firms) were not listed on any stock exchange. Moreover, MNC often use affiliates as a vehicle for their FDI, andthose affiliates are also not listed in stock exchanges. https://finance.yahoo.com a b l e : L i s t o f ca s e s i n c l ud e d i n t h ea n a l y s i s . R e g i s t r a ti on y ea r O u t c o m e d a t e P l a i n ti ff A pp li ca b l e II AD ec i d e d i n f a vou r o f H o s t c oun t r y N a ti on a lit y o f i nv e s t o r A m oun t c l a i m e d ( m il U S $ ) A m oun t a w a r d e d ( m il U S $ ) - - A s t a l d i , S . p . A . A l g e r i a -I t a l y B I T ( ) S t a t e A l g e r i a I t a l y0 . - - HO C H T I E F A k ti e ng e s e ll s c h a f t A r g e n ti n a - G e r m a ny B I T ( )I nv e s t o r A r g e n ti n a G e r m a ny157 . . - - I m p r e g il o S . p . A . A r g e n ti n a -I t a l y B I T ( )I nv e s t o r A r g e n ti n a I t a l y119 . . - - S u ez , S . A . A r g e n ti n a - F r a n ce B I T ( )I nv e s t o r A r g e n ti n a F r a n ce . . - - S e m p r a E n e r gy I n t e r n a ti on a l A r g e n ti n a - U n it e d S t a t e s o f A m e r i ca B I T ( )I nv e s t o r A r g e n ti n a U S A . . - - S i e m e n s A . G . A r g e n ti n a - G e r m a ny B I T ( )I nv e s t o r A r g e n ti n a G e r m a ny462 . . - - C M S G a s T r a n s m i ss i on C o m p a ny A r g e n ti n a - U n it e d S t a t e s o f A m e r i ca B I T ( )I nv e s t o r A r g e n ti n a U S A . . - - D a i m l e r F i n a n c i a l S e r v i ce s AGA r g e n ti n a - G e r m a ny B I T ( ) S t a t e A r g e n ti n a G e r m a ny243 . . - - R e p s o l , S . A . A r g e n ti n a - S p a i n B I T ( ) S e ttl e d A r g e n ti n a S p a i n10500 . . - - I m p r e g il o S . p . A . A r g e n ti n a -I t a l y B I T ( ) S e ttl e d A r g e n ti n a I t a l y250 . - - F r a n ce T e l ec o m S . A . A r g e n ti n a - F r a n ce B I T ( ) S e ttl e d A r g e n ti n a F r a n ce . - - S a i p e m S . p . A . B a ng l a d e s h -I t a l y B I T ( )I nv e s t o r B a ng l a d e s h I t a l y12 . . - - P i ng A n I n s u r a n ce ( G r oup ) C o m p a nyo f C h i n a , L i m it e d B LE U - C h i n a B I T ( , ) S t a t e B e l g i u m C h i n a . . - - M u r phy O il C o r po r a ti on NA F T A I nv e s t o r C a n a d a U S A . . - - O cc i d e n t a l P e t r o l e u m C o r po r a ti on E c u a do r- U n it e d S t a t e s o f A m e r i ca B I T ( )I nv e s t o r E c u a do r U S A . . - - C h e v r on C o r po r a ti on E c u a do r- U n it e d S t a t e s o f A m e r i ca B I T ( )I nv e s t o r E c u a do r U S A . . - - E n C a n a C o r po r a ti on C a n a d a - E c u a do r B I T ( ) S t a t e E c u a do r C a n a d a . . - - E d e n r e d , S . A . F r a n ce - H ung a r y B I T ( )I nv e s t o r H ung a r y F r a n ce . - - E l ec t r i c it e d e F r a n ce ( E D F )I n t e r n a ti on a l S . A . T h e E n e r gy C h a r t e r T r ea t y ( )I nv e s t o r H ung a r y F r a n ce . . - - T e l e no r M ob il e C o mm un i ca ti on s A S H ung a r y - N o r w a y B I T ( ) S t a t e H ung a r y N o r w a y152 . . - - A E SS u mm it G e n e r a ti on L i m it e d T h e E n e r gy C h a r t e r T r ea t y ( ) S t a t e H ung a r y U n it e d K i ngdo m . . - - A I G C a p it a l P a r t n e r s , I n c . K aza kh s t a n - U n it e d S t a t e s o f A m e r i ca B I T ( )I nv e s t o r K aza kh s t a n U S A . . - - S t a n s E n e r gy C o r p . C I S I nv e s t o r R i gh t s C onv e n ti on ( )I nv e s t o r K y r gy z s t a n C a n a d a . . - - F r a n ce T e l ec o m F r a n ce - L e b a non B I T ( )I nv e s t o r L e b a non F r a n ce . . - - E VNAG T h e E n e r gy C h a r t e r T r ea t y ( ) ; A u s t r i a - M ace don i a B I T ( ) S e ttl e d M ace don i a A u s t r i a . - - A b e ngo a , S . A . M e x i c o - S p a i n B I T ( )I nv e s t o r M e x i c o S p a i n70 . . - - A r c h e r D a n i e l s M i d l a nd NA F T A I nv e s t o r M e x i c o U S A . . - - C o r n P r odu c t s I n t e r n a ti on a l , I n c . NA F T A I nv e s t o r M e x i c o U S A . . - - W a s t e M a n a g e m e n t , I n c . NA F T A S t a t e M e x i c o U S A . . - - W a s t e M a n a g e m e n t , I n c . NA F T A S t a t e M e x i c o U S A . . - - T a t e & L y l e I ng r e d i e n t s A m e r i ca s , I n c . NA F T A I nv e s t o r M e x i c o U S A . . - - I m p r e g il o S . p . A . I t a l y - P a k i s t a n B I T ( ) S e ttl e d P a k i s t a n I t a l y450 . - - B ea r C r ee k M i n i ng C o r po r a ti on C a n a d a - P e r u F T A I nv e s t o r P e r u C a n a d a . . - - E l ec no r , S . A . P e r u - S p a i n B I T ( ) S e ttl e d P e r u S p a i n20032007 - - F r a po r t AG F r a nk f u r t A i r po r t S e r v i ce s W o r l d w i d e G e r m a ny - P h ili pp i n e s B I T ( ) S t a t e P h ili pp i n e s G e r m a ny425 . . - - S G SS o c it G n r a l e d e S u r v e ill a n ce , S . A . P h ili pp i n e s - S w it ze r l a nd B I T ( ) S e ttl e d P h ili pp i n e s S w it ze r l a nd140 . - - D e u t s c h e B a nk A . G . G e r m a ny - S r i L a nk a B I T ( )I nv e s t o r S r i L a nk a G e r m a ny60 . . - - C a n f o r C o r po r a ti on NA F T A S e ttl e d U n it e d S t a t e s o f A m e r i ca C a n a d a . - - P h ili p M o rr i s P r odu c t s , S . A . S w it ze r l a nd - U r ugu a y B I T ( ) S t a t e U r ugu a y S w it ze r l a nd22 . . - - S a i n t - G ob a i n P e rf o r m a n ce P l a s ti c s E u r op e F r a n ce - V e n ez u e l a B I T ( )I nv e s t o r V e n ez u e l a F r a n ce . . - - T e n a r i s , S . A . B LE U - V e n ez u e l a B I T ( )I nv e s t o r V e n ez u e l a L ux e m bou r g299 . . - - F l ugh a f e n Z r i c h , A . G . S w it ze r l a nd - V e n ez u e l a B I T ( )I nv e s t o r V e n ez u e l a S w it ze r l a nd82 . . - - E xxon M ob il C o r po r a ti on N e t h e r l a nd s - V e n ez u e l a B I T ( )I nv e s t o r V e n ez u e l a U S A . . - - T e r n i u m , S . A . B LE U - V e n ez u e l a B I T ( ) D i s c on ti nu e d V e n ez u e l a L ux e m bou r g130 . - - T h e W illi a m s C o m p a n i e s , I n t e r n a ti on a l H o l d i ng s B . V . N e t h e r l a nd s - V e n ez u e l a B I T ( ) S e ttl e d V e n ez u e l a U n it e d K i ngdo m - - C E M E X C a r aca s I nv e s t m e n t s B . V . N e t h e r l a nd s - V e n ez u e l a ( ) S e ttl e d V e n ez u e l a N e t h e r l a nd s . . Measurement of Abnormal Volatility Around the OutcomeDate
Volatility, as a proxy for the uncertainty in the markets, has been one of the most important phe-nomena studied in financial economics. In his seminal work on stock market volatility, Shiller(1981) argues that the observed volatility is not consistent with the predictions of the present valuemodels and that the high intertemporal variation cannot be rationalized on that basis. The failureof standard valuation models to explain the high magnitudes of stock market fluctuations openedserious challenge to financial economists, and literature started the search for the volatility driversdifferent from conventional dividends and earnings. One strand of the literature examines whetherthe variability of returns can be linked to macroeconomic variables, financial leverage, or tradingvolume (e.g. Schwert (1989); Binder and Merges (2001); Engle et al. (2013)).A different route is explored by event studies connecting the excess volatility with some event. From early efforts summarized by Yadav (1992), researchers subsequently studied the reaction ofvolatility to political events such as national elections (Białkowski et al., 2008), mergers and ac-quisitions (Balaban and Constantinou, 2006) or terrorist attacks (Essaddam and Mnasri, 2015).Abnormal volatility, to the best of our knowledge, has not yet been connected to the investmentdisputes.To account for strong time variation in volatility, we need to step forward from traditionalevent study techniques. Estimating the model that will capture the time varying volatility duringthe window period immediately preceding the outcome window and including the market volatilityas control variable for volatility induced by market, we will first estimate an expected, or an usualvolatility under the null hypothesis of no reaction of the market to the event. Then to see if theresolution of the dispute induced changes in the volatility, we will compare this expectation to themeasured volatility in the time window around the outcome date. The key aspect of the method-ology used is to capture the usual time varying volatility, which allows us to measure abnormalvariation in response to the event. Moreover, following Białkowski et al. (2008), we construct abootstrap procedure to our tests in order to mitigate small sample bias.Let us discuss the procedure in detail. The key date is the outcome date t O . The announcementdate of the investor’s claim is not taken as a “surprise date” as market will have the informationalready processed from media. The surprise cannot be measured and timed, but the outcome daycan be timed. The data are measured over the time period t ∈ [1 , . . . , T ] so that t = 1 is the startingdate and t = T is end of that period, so that outcome date t O is inside of the interval defined as < t O < T . This allows us to capture the specifics of each stock. Since uncertainty about theoutcome will probably result in an abrupt volatility changes even before the announcement day,we focus on the time window around the outcome date that includes certain period of time before The approach also involves less stringent assumptions than those required in standard event studies. The lattertypically look at the effect of investment disputes on the absolute level of real returns. To do so, they need to considerthe relationship between the absolute level of award in arbitration and the change in the firm’s value in order to assessthe real damages to investor. This strict assumption is not necessary in studies of volatility. Yet another approach hasrecently been suggested – judgement of investors in the market. As Fisman and Zitzewitz (2019) suggest, investors’poor judgement may be the reason for volatility. We assume in this paper that poor judgement does not characterize all market participants and that individual mistakes do not dominate the market. Note that the dates of claims and the dates of decisions differ from case to case. Note also that the announcementis given as an announcement window, which we define further below.
10s well as after the outcome. Formally, the window of interest is defined as [ t O − N B , t O + N A ] with N B and N A being a fixed number of days before and after the outcome day t O respectively, and wewill refer to it as an announcement window. We start by estimating the case-specific component of variance for all companies in the sample i = 1 , . . . , K with generalized auto-regressive conditional heteroscedasticity (GARCH) modelover time period t ∈ [ t O − N B − , . . . , t O − N B ] immediately before the event window as r i,t = α + βr S&P 500 t + (cid:15) i,t (1) (cid:15) i,t ∼ N (0 , σ i,t ) (2) σ i,t = ψ + ψ σ i,t − + ψ (cid:15) i,t − (3)where r i,t is logarithmic return of the i th company and r S&P 500 t is stock market return representedby the S&P 500 market index.The model described by equations 1 - 3 is a GARCH(1,1) model and is one of the most popularand useful frameworks for capturing the time varying volatility. Engle (1982); Bollerslev (1986)noted that returns described by equation 1 display strong time variation in its second moment,volatility, and proposed to describe it by latent dynamics in 3 assuming the residuals 2 beingnormally distributed. We chose to estimate the model for the period of 500 days immediatelybefore the announcement window. This should allow the period to be reasonably long in orderfor the market to absorb the information and to address the possibility of the small sample bias(Hwang and Valls Pereira, 2006).Being interested in measurement of abnormal volatility, we consider variation in (cid:15) i,t aroundthe outcome date (i.e. outcome window) in relation to the usual volatility estimated in the periodbefore the outcome window. In other words, model described by equations 1 – 3 will indicate thelevel of expected usual volatility under the null hypothesis of no reaction to the announcement ofthe resolution. Comparison of the modelled volatility to the volatility measured by the actual datawill then indicate the abnormal volatility. Abnormal volatility is further controlled for stock marketvolatility. Note that this benchmark volatility captures time dynamics in the volatility as well asvolatility induced by S&P500 that is by estimating β coefficient, and including market returns intothe model directly. Formally, we define it as a k -step-ahead forecast of the variance conditional onthe information available on the last day of the period before the announcement window E t (cid:104) σ i,t + k |I t (cid:105) = (cid:98) γ k − (cid:88) j =0 ( (cid:98) γ + (cid:98) γ ) j + ( (cid:98) γ + (cid:98) γ ) k − (cid:98) γ σ i,t + ( (cid:98) γ + (cid:98) γ ) k − (cid:98) γ (cid:15) i,t (4)Importantly, the distribution of the residuals during the announcement window can be described as (cid:15) i,t ∼ N ( AR t , M t E t (cid:2) σ i,t + k |I t (cid:3) ) where the residuals will on average hold abnormal return AR t , In real world, the announcement window could be “contaminated” by “other events”. We proxy the contaminationby “other events” by looking at the excess volatility controlled by returns (and volatility as well) of S&P500 index andassume that S&P500 controls for this effect as we discuss the best practice option in the text. Note that GARCH(1,1) is powerful benchmark that captures the time variation in volatility well. Large literaturetrying to extend the model in various ways does not find significant improvements of out-of-sample forecasts. Aninteresting survey of Hansen and Lunde (2005) summarizes and compares 330 extensions and find no evidence thatGARCH(1,1) is outperformed by more sophisticated models. Hence we chose the model as a best parsimoniouschoice. M t is the multiplicative effect of the announcement-induced volatility. Under the null hypoth-esis that investors are not surprised by outcome of the dispute, the parameter M t = 1 . Givenour interest in capturing the effect of investment disputes on volatility, announcement-inducedvolatility M t is of primary interest and needs to be estimated. By noting that residuals de-meaned using cross-sectional average are normally distributed with zero mean and announcement-independent variance, we can estimate the M t as the cross-sectional variance of demeaned resid-uals by announcement-independent standard deviation (Boehmer et al., 1991) (cid:99) M t = 1 / ( K − K (cid:88) i =1 (cid:32) (cid:16) K (cid:98) (cid:15) i,t − (cid:80) Kj =1 (cid:98) (cid:15) j,t (cid:17) K ( K − E t (cid:104) σ i,t |I t (cid:105) + (cid:80) Kj =1 E t (cid:104) σ j,t |I t (cid:105) (cid:33) (5)with (cid:98) (cid:15) i,t = r i,t − (cid:98) α − (cid:98) βr S&P 500 t and t ∈ [ t O − N B , t O + N A ] .Since residuals follow standard normal distribution under the null hypothesis when M t = 1 ,abnormal change in volatility is M t − , and hence the cumulative abnormal volatility CAV (cid:16) N B , N A (cid:17) can be computed as CAV (cid:16) N B , N A (cid:17) = (cid:32) t O + N A (cid:88) t = t O − N B (cid:99) M t (cid:33) − ( N A − N B + 1) . (6)The null hypothesis of no impact of investment dispute on volatility H : CAV (cid:16) N B , N A (cid:17) = 0 (7)is equivalent to H : t O + N A (cid:88) t = t O − N B M t ( K −
1) = ( N A − N B + 1)( K − , (8)and can be tested noting that under the null, M t is a variance of K independent N (0 , randomvariables implying the test statistic for the hypothesis is χ distributed as t O + N A (cid:88) t = t O − N B ( K − (cid:99) M t ∼ χ K − N B − N A +1) . (9)Moreover, inference based on the asymptotic values may be biased in case of non-normally dis-tributed residuals, cross-sectional dependence, or autocorrelation structures that are likely to beobserved in the data. As suggested by Białkowski et al. (2008), we additionally construct a boot-strap test based on Efron (1992). The cumulative abnormal volatility during the announcementwindow is compared with the empirical distribution of CAV (cid:16) N B , N A (cid:17) simulated under the nullhypothesis. The procedure for generating the empirical distribution can be described as follows: Note that we are making a distinction between two situations to be considered by the market – event or no event.We do not differentiate among different outcomes of the event. The assumption seems reasonable since it is impossiblefor the market to predict the precise outcome of the arbitration.
12. Draw the case and date to match the number of outcomes in the original sample randomlywith replacement from the whole dataset.2. Compute
CAV (cid:16) N B , N A (cid:17) from the randomly generated sample from step (1).3. Repeat step (1) and (2) 5000 times, sort the estimated CAV (cid:16) N B , N A (cid:17) in order to obtainempirical distribution.The bootstrapped p -value is then computed as the number of bootstrapped CAV (cid:16) N B , N A (cid:17) thatexceed the CAV (cid:16) N B , N A (cid:17) calculated on the original election date, divided by the number ofreplications. Note this is one-sided test to test the null hypothesis against the alternative H A : CAV (cid:16) N B , N A (cid:17) > , but in case the number of bootstrapped statistics below the value of statisticsestimated on data is used, alternative hypothesis H A : CAV (cid:16) N B , N A (cid:17) < can be tested. Inother words, we can confirm if the potential volatility connected to the dispute is increasing ordecreasing. We chose the outcome date to be the main event for the choice of the announcement window. Wedivide the sample into cases in which the decision was made in favor of the state, cases with thedecision in favor of investor, as well as cases that have been discontinued or settled. These threegroups are distinct in their potential impact of the decision on the plaintiff company, and hence weexpect a very different reaction of investors. While our expectation is that the decision in favor ofstate will result in increased volatility, capturing increased uncertainty about the prices since thedecision may lead to large costs for the company. On the other hand, cases with decisions in favorof investor as well as those that were settled are likely to show the opposite pattern of volatility.Figure 1 shows the estimates of the cumulative abnormal volatility (cid:91)
CAV (cid:16) N B , N A (cid:17) for thethree samples. We choose to include one month before the outcome date as well as two monthsafter the outcome day in the announcement window to see the increased uncertainty before theannouncement date.As can be seen from Figure 1, we document a surge in the volatility of cases decided in favorof state. The increase in the volatility started even before the outcome date ( about two weeks priorto the date of the outcome), presumably in the anticipation of the results or news leaks. After theoutcome day, the volatility continues to surge steadily every single day throughout the two monthsperiod and even beyond. A very different pattern is documented for the cases when the decisionhas been made in favor of investor. Here the volatility decreases gradually. Cases that have beensettled or discontinued show no abnormal volatility while volatility decreases slightly at the end ofthe reaction window.To test the significance of the results, we compute the cumulative abnormal volatility (cid:91) CAV (cid:16) N B , N A (cid:17) statistics for different window windows of 2 days, 1 week, 2weeks, 1month and 2 months before13igure 1: Volatility reaction to the Dispute Outcome C u m u l a t i v e A bno r m a l V o l a t ili t y − − −1 month Outcome Day +1 month +2 months In favour of StateIn favour of InvestorDiscontinued/Settled
Notes : The figure shows cumulative abnormal volatility (cid:91)
CAV (cid:16) N B , N A (cid:17) around the outcome day. Reaction of companies with positive outcomeare captured by dashed line, reaction of companies with negative outcome (in favor of state) in black, the ones settled out of court or undecided arein dotted line. The window is one month before and two months after the announcement day. as well as after the outcome day. Specifically, Table 2 displays the cumulative abnormal volatilityin the first column, theoretical as well as bootstrapped p-values in the third and fourth columns re-spectively. In addition, the second column of the Table 2 holds percentage change of the volatilityrelative to the benchmark.The results confirm the significant increase of abnormal volatility of the cases decided in favorof state in longer than 2 weeks period. The percentage volatility increase is growing with the pe-riod, and abnormal volatility of cases decided in favor of state cumulated by 60.5% in the 1 monthbefore – 2 months after the outcome day window. In contrast, cumulative abnormal volatility inthe cases decided in favor of investors decreased over time, resulting in a significant decrease inabnormal volatility. Investors thus reacted to the “good news” with smaller than the usual volatility– by a decrease of -17.4%. Discontinued or settled cases show a similar pattern to those decidedin favor of investor with decrease in abnormal volatility of -14.5%. In brief, volatility of the com-pany’s shares was increased as uncertainty about the outcome set in. When the outcome decisionwas in favor of investor, the uncertainty decreased back to its “normal” level, and hence the ex-cess – abnormal volatility disappeared. On the other hand, if the decision was in favor of state,it increased the uncertainty about the impact of the decision on the balance sheet of the companyreflecting partly the loss of income of the investor and partly the cost of litigation.14able 2: Cumulative abnormal volatility around Decision Day Announcement window (cid:91)
CAV % vol p -val p -val boot Panel A: Decision in favour of Investor (-2 days,2 days) -1.010 -0.202 0.054 0.118(-1 week,1 week) 2.863 0.260 0.003 0.083(-2 week,2 weeks) 1.624 0.077 0.116 0.327(-1 month,1 month) -7.387 -0.145 0.000 0.003(-1 month,2 months) -13.190 -0.174 0.000 0.000
Panel B: Decision in favour of State (-2 days ,2 days) 1.141 0.228 0.129 0.337(-1 week,1 week) 1.686 0.153 0.129 0.402(-2 weeks,2 weeks) 27.653 0.317 0.000 0.000(-1 month,1 month) 19.763 0.388 0.000 0.015(-1 month,2 months) 45.949 0.605 0.000 0.000
Panel C: Discontinued/Settled (-2 days,2 days) -1.691 -0.338 0.031 0.055(-1 week,1 week) -3.700 -0.336 0.003 0.007(-2 week,2 weeks) 3.279 0.156 0.000 0.285(-1 month,1 month) 0.444 0.009 0.000 0.708(-1 month,2 months) -10.859 -0.143 0.000 0.002
Notes : The table contains cumulative abnormal volatility (cid:91)
CAV (cid:16) N B , N A (cid:17) aroundthe outcome day. Panel A of the table reports results for the cases decided in fa-vor of Investor, Panel B reports the results for the cases decided in favor of state,and Panel C reports cases that were discontinued or settled. The implied percentagechange (% vol) relative to the benchmark, asymptotic p -values as well as p -valuescomputed on 5000 bootstraps are reported and test the null hypothesis of no reaction H : (cid:91) CAV (cid:16) N B , N A (cid:17) = 0 . The previous section presented our estimates of abnormal volatility based on the GARCH model,which serves as the benchmark for volatility without the event. We have isolated the case – specificcomponent of variance within GARCH. In order to avoid a bias in estimating abnormal volatility,we have estimated the variance conditional to the prior dates of the event. To recall, the variancewas estimated by the event independent demeaned standard deviation (EIDSD). As we have alsonoted, potential complications may arise from non-normality, cross-sectional dependence, or au-tocorrelation of the regression residuals. To circumvent those problems, the statistical significanceof the impact of disputes was additionally tested using the bootstrap methodology. We shall nowconsider additional issues in order to test the robustness of our results. We begin with a few com-ments about the likelihood of additional sources of biasedness of our results. In addition, we shalldeeper explore the sources of the abnormal volatility in order to provide a stronger justification forour hypothesis.
Comments on biasedness . Given the size of our sample, it could be argued that our result arebiased due to the relatively small size of our sample. We have, therefore, taken additional steps toaddress the issue. We have chosen the time window during which the volatility is measured and15ssessed long enough in order to minimize the possibility of a small sample bias.
Pari passu , itcould be argued that our sample is biased by volatility of shares of a large company that may bedominating our sample. Once again, such concerns are not justified. To recall, we have estimatedthe excess volatility of each stock in our sample, produced 46 excess volatilities and we reportthe median. We should also note that reporting the mean does not influence the main results.We have checked the estimates of each case and have found no major deviation among them. Inaddition, there is no strong theoretical reason to believe that volatility of large firms should bedifferent – larger or smaller – than volatility of shares of smaller companies. Finally, there is asolid econometric ground to argue that our results cannot be driven by “excessive” volatility ofshares of single large company. Using the GARCH model, it is clear that an “excessive” volatilityof shares of a large firm would affect not only our sample but also the volatility of our benchmark.
Origins of Volatility . Why should we observe these particular patterns of volatility? Are thereany reasons beyond saying that the volatility was brought about by uncertainty about the dispute?Can we justify linking the abnormal volatility to particular features of the host countries and/orfeatures of the arbitration process? To explore these issues, we have built a model in which wehave tested the influence of factors that might have determined the link between uncertainty andvolatility of share prices. First, the size of award should clearly be important. It obviously matterswhether compensations to investors are small or large in relation to damages or to claims fordamages. We expect the level of compensation to have a negative coefficient. The larger theproportion of the award from the original claim, the less uncertainty about the recovery of damagesincurred by the investor and hence the better the impact of the award on the financial position ofthe investor. Second, many disputes involved host countries with several cases against them. Thefact that a country is facing multiple disputes is serious problem; it tarnishes the reputation of thecountry and makes it more risky. In addition, if the host country in question is a poor country,the exposure to disputes clearly becomes a financial problem for the government. Third and paripassu, the uncertainty is likely to be affected by political instability in host country (Kurz et al.,2005; Irshad, 2017). For example, the government claiming that the conditions of the BIT wereagreed and signed by the previous government may not be acceptable and may not accept thedecision of the court in favor of investor. Fourth, according some international legal scholars, thelocation of arbitration and/or the choice of arbitral rules may also matter (Pohl et al., 2012; Howseand Levin, 2020). Fifth, volatility may also be affected by the country of origin of investor. As itis well known from the financial literature, the appetite for risk is to a large extent determined byculture (e.g. Li et al. (2013); Kreiser et al. (2010); Mihet (2013)). Finally, a very large proportionof disputes between states and investors involve protection of public interest (e.g. Schill and Djanic(2018); Johnson et al. (2015)). Governments have sometimes taken measures to protect domesticpublic interest, and those measures have led to legal challenges by foreign investors. The rangeof actions (or inactions) by host governments has been large, and it would be beyond the scopeof this paper to test the significance of each intervention or policy stance. As a first step, weshall try to capture the role of public interest in the litigation of investment disputes by looking As noted in Section 2, the level of compensation is important particularly because it should cover not only lostprofits in the past but also profits that investors could reasonably expect in future. As argued by Johnson et al. (2015), “Available evidence regarding the approximately 600 known ISDS suits filedto date indicate that investors can use the mechanism to contest a virtually unlimited range of actions (or inactions),including measures relating to taxation, environmental regulation, tariffs for water and electricity, health insuranceregulation, and health and safety regulations of pharmaceutical imports, among others.”
16t the role of two most contested rules from IIAs – indirect expropriation and “full protection andsecurity” of investors. Both represent two areas in which public interest and policy are said to havebeen at the origin of breaches of commitments under those agreements. The lack of clarity of bothprovisions is undoubtedly an important origin of uncertainty in the markets. Unfortunately, thereis a strong reason to believe that, the two factors are also strongly correlated, and we had to limitour analysis to “indirect expropriation” only. Moreover, the unpredictability of decisions underthese two provisions is further enhanced by disagreements among economists about the impact ofregulatory measures on investment. Our model takes the following form: AV i = α + β SA SA i + β PIC
PIC i + β RL RL i + β PIC , RL PIC i × RL i + β AR AR i + β CO CO i + β IE IE i + (cid:15) i (10)Where AV t is abnormal volatility calculated as the volatility ratio constructed by dividing thevariance of real returns over the event window by the variance of real returns in the pre-eventwindow. Our independent variables are SA for the size of the award, PIC for political instabilityand corruption, RL for rule of law, AR for the type of arbitral rules, CO for country of origin, IEfor “indirect expropriation” as the reason for the breach of investor’s rights. On the other hand, wealso expect a positive effect on volatility in the case of the multiple dispute country and politicalinstability variables.Unfortunately, we were also constrained by the size of our sample. As noted above, our originalsample included 46 observations but as we had to restrict our analysis to cases actually won byinvestors, we have ended up with 29 observations. 17 observations were “lost” since they referredto disputes settled out of court or to the disputes that were abandoned. Suspecting some degree ofcorrelation, we have included in the model an interaction term between the variables captured by β PIC , RL , which turns out insignificant but it leaves all other estimates fundamentally unchanged.The terms provides, therefore, an important statistiacal control.As we can see from Table 3, we can explain a great deal of the abnormal volatility in our sampleof foreign investment disputes. The level of compensation as a proportion of the investor’s claimhad a negative impact on the change in volatility even if the estimate is only significant at the 5percent level. Both corruption and rule of law are significant and with the right sign. As expected,the results are also affected by the country of origin of investors. The results are sensitive as towhether investors come from the US or not. Interestingly enough, the results are also sensitive totwo particular features of the arbitration system – (i) whether the legal proceedings are handled un-der UNCITRAL or not and (ii) whether the subject of the alleged breach is “indirect expropriation”or not. Surprisingly, length of the process turned out to be insignificant. The adjusted R-squareexplains 32% of the abnormal volatility.The greater sensitivity of the market to dispute proceedings handled under UNCITRAL mayhave different explanations. For example, it is widely accepted that UNCITRAL arbitral tribunalshave historically been more permissive with respect to allowing the claims of dual nationals. Many cases of “indirect expropriation” were also classified as breaches of “full protection and security”, presum-ably reflecting plaintiffs’ charges as violations of both provisions. We tried but were unable to appropriate econometricadjustments to deal with this correlation problem. Economists disagree about whether police-powers carve-outs promote foreign investments or not. “Weak” policycarve-outs may promote FDI and ,with guaranteed compensation, may lead to overinvestment. Strong policy carve-outmay lead to underinvestment. However, Aisbett et al. (2010a) argues that “underinvestment” may happen even withguaranteed compensation. The debate has led to the literature on optimal investment agreements. See, for example,Horn and Tanger˚as (2019).
Estimate Std. Error t value P-valIntercept 1.044 0.174 6.004 0.000 *** β SA -0.073 0.040 -1.813 0.082 * β RL -0.365 0.166 -2.193 0.038 ** β PIC β AR -0.184 0.105 -1.747 0.093 * β CO -0.214 0.118 -1.813 0.082 * β IE β PIC , RL adjR Notes : The table presents results of regression linking the abnormal volatil-ity induced by investment disputes to independent varianbles SA – size of theaward, PIC – political instability and corruption, RL – rule of law, AR – thetype of arbitral rules, CO – country of origin, IE – “indirect expropriation”as the reason for the breach of investor’s rights.. The dependent variable isdefined as natural logarithm of the volatility ratio constructed by dividing thereturn variance computed over the event window by the variance of returns inpre-event window.
UNCITRAL tribunals are also less likely to refuse jurisdiction on the basis that a dispute arisesfrom an economic activity that does not qualify as an investment. UNCITRAL awards may faceadditional scrutiny at the level of State courts, which may be either positive or negative. Disputesabout host country interventions that are considered by investors as measures leading to “indirectexpropriation” are clearly a highly destabilizing factor in the market. The main source of unpre-dictability of disputes about the rule is uncertainty about the interpretation by the host countrygovernment of what the government considers the legitimate police-powers carve-out.
Our main aim in this paper is to assess the impact of international investment disputes on the valueof the investors’ firm. We do so by analyzing the second movement of the impact on shares –on their volatility. Using the GARCH model, we estimate the cumulative average volatility ofshares in our sample and compare our estimates against the benchmark. The key aspect of themethodology is to capture the usual time varying volatility, which allows us to measure abnormalvariation in response to an event. In contrast to standard event studies, we do not use a single daywindow but a window of few days.We find that investment disputes lead to abnormal volatility of shares of companies involved indisputes with host countries. Markets do respond to information about the emergence of disputesbetween those companies and host countries. We associate the increased volatility with increaseduncertainty in the market about the outcome of arbitration which may or may not lead to com-pensation for damages claimed by the investor. Our second important result is that the outcomeof arbitration is also very important in explaining the volatility of shares of both “winning” and Choosing ICSID or UNCITRAL Arbitration for Investor-State Disputes, 29/06/2018 by Aceris Law LLC
References
Aisbett, E., L. Karp, and C. McAusland (2010a). Compensation for indirect expropriation ininternational investment agreements: Implications of national treatment and rights to invest.
Journal of Globalization and Development 1 (2).Aisbett, E., L. Karp, and C. McAusland (2010b). Police powers, regulatory takings and the efficientcompensation of domestic and foreign investors.
Economic Record 86 (274), 367–383.Alschner, W. and D. Skougarevskiy (2016). Mapping the universe of international investmentagreements.
Journal of International Economic Law 19 (3), 561–588.Arena, M. and S. Ferris (2017). A survey of litigation in corporate finance.
Managerial Finance .Balaban, E. and C. T. Constantinou (2006). Volatility clustering and event-induced volatility:Evidence from uk mergers and acquisitions.
The European Journal of Finance 12 (5), 449–453.Baldwin, R. and J. Lopez-Gonzalez (2015). Supply-chain trade: A portrait of global patterns andseveral testable hypotheses.
The World Economy 38 (11), 1682–1721.Bellak, C. and M. Leibrecht (2020). Economic crises and investor-state arbitration cases: Anempirical analysis for the first thirty years of treaty-based dispute settlement. UnpublishedManuscript. 19hagat, S., J. Bizjak, and J. L. Coles (1998). The shareholder wealth implications of corporatelawsuits.
Financial Management , 5–27.Bhagat, S., J. A. Brickley, and J. L. Coles (1994). The costs of inefficient bargaining and financialdistress: Evidence from corporate lawsuits.
Journal of Financial Economics 35 (2), 221–247.Białkowski, J., K. Gottschalk, and T. P. Wisniewski (2008). Stock market volatility around nationalelections.
Journal of Banking & Finance 32 (9), 1941–1953.Binder, J. J. and M. J. Merges (2001). Stock market volatility and economic factors.
Review ofQuantitative Finance and Accounting 17 (1), 5–26.Bizjak, J. M. and J. L. Coles (1995). The effect of private antitrust litigation on the stock-marketvaluation of the firm.
The American Economic Review , 436–461.Blume, L., D. L. Rubinfeld, and P. Shapiro (1984). The taking of land: When should compensationbe paid?
The Quarterly Journal of Economics 99 (1), 71–92.Boehmer, E., J. Masumeci, and A. B. Poulsen (1991). Event-study methodology under conditionsof event-induced variance.
Journal of Financial Economics 30 (2), 253–272.Bollerslev, T. (1986). Generalized autoregressive conditional heteroskedasticity.
Journal of Econo-metrics 31 (3), 307–327.Brada, J., C. Chunda, J. Jingyi, and K. Ali (2020). Does bilateral investment treaty arbitration haveany value to multinational corporations? Technical report, Helsinki, BOFIT.Brada, J., Z. Drabek, and I. Iwasaki (2020). Does investor protection increase foreign investment?Unpublished Manuscript.Dolzer, R. and C. Schreuer (2012).
Principles of international investment law . Oxford UniversityPress.Efron, B. (1992). Bootstrap methods: another look at the jackknife. In
Breakthroughs in statistics ,pp. 569–593. Springer.Engle, R. F. (1982). Autoregressive conditional heteroscedasticity with estimates of the varianceof united kingdom inflation.
Econometrica , 987–1007.Engle, R. F., E. Ghysels, and B. Sohn (2013). Stock market volatility and macroeconomic funda-mentals.
Review of Economics and Statistics 95 (3), 776–797.Essaddam, N. and A. Mnasri (2015). Event-study volatility and bootstrapping: an internationalstudy.
Applied Economics Letters 22 (3), 209–213.Fisman, R. and E. Zitzewitz (2019). An event long-short index: Theory and applications.
AmericanEconomic Review: Insights 1 (3), 357–72.Franck, S. D. (2007). Empirically evaluating claims about investment treaty arbitration.
NorthCarolina Law Review 86 , 1. 20onzalez, A. and P. Kusek (Eds.) (2018).
Global Investment Competitiveness Report 2017/2018:Foreign Investor Perspectives and Policy Implications , Chapter Overview. World Bank: Wash-ington, D.C.Hajzler, C. (2012). Expropriation of foreign direct investments: sectoral patterns from 1993 to2006.
Review of World Economics 148 (1), 119–149.Hansen, P. R. and A. Lunde (2005). A forecast comparison of volatility models: does anythingbeat a garch (1, 1)?
Journal of Applied Econometrics 20 (7), 873–889.Horn, H., P. C. Mavroidis, and H. Nordstr ¨om (2005). Is the use of dispute settlement system biased.
The WTO and international trade law/dispute settlement. Elgar, Cheltenham .Horn, H. and T. Tanger˚as (2019). Economics of international investment agreements.
Centre forEconomic Policy Research Discussion Paper No 11 879 .Howse, R. and S. Levin (2020). The right to regulate and investor – state dispute settlement:Renewable energy and the fair and equitable treatment norm.
New York University, workingpaper .Hwang, S. and P. L. Valls Pereira (2006). Small sample properties of garch estimates and persis-tence.
The European Journal of Finance 12 (6-7), 473–494.Irshad, H. (2017). Relationship among political instability, stock market returns and stock marketvolatility.
Studies in Business and Economics 12 (2), 70–99.Johnson, L., L. E. Sachs, and J. D. Sachs (2015). Investor-state dispute settlement, public interestand us domestic law.
CCSI Policy Paper .Juillard, P. (2000). Freedom of establishment, freedom of capital movements, and freedom ofinvestment.
ICSID Review 15 (2), 322–339.Kahneman, D. (2003). Maps of bounded rationality: Psychology for behavioral economics.
Amer-ican Economic Review 93 (5), 1449–1475.Kohler, W. and F. Staller (2019). The economics of investment protection.
Journal of InternationalEconomics 121 .Kreiser, P. M., L. D. Marino, P. Dickson, and K. M. Weaver (2010). Cultural influences on en-trepreneurial orientation: The impact of national culture on risk taking and proactiveness insmes.
Entrepreneurship Theory and Practice 34 (5), 959–984.Kurz, M., H. Jin, and M. Motolese (2005). Determinants of stock market volatility and risk premia.
Annals of Finance 1 (2), 109–147.Li, K., D. Griffin, H. Yue, and L. Zhao (2013). How does culture influence corporate risk-taking?
Journal of Corporate Finance 23 , 1–22.Mavroidis, P. C. and A. O. Sykes (2005).
The WTO and international trade law/dispute settlement .Edward Elgar Publishing. 21ihet, R. (2013). Effects of culture on firm risk-taking: a cross-country and cross-industry analy-sis.
Journal of Cultural Economics 37 (1), 109–151.Newman, C., J. Rand, T. Talbot, and F. Tarp (2015). Technology transfers, foreign investment andproductivity spillovers.
European Economic Review 76 , 168–187.Ossa, R., R. W. Staiger, and A. O. Sykes (2020). Disputes in international investment and trade.Technical report, National Bureau of Economic Research.Pohl, J. (2018). Societal benefits and costs of international investment agreements.
OECD WorkingPapers on International Investment No. 2018/01 .Pohl, J., K. Mashigo, and A. Nohen (2012). Dispute settlement provisions in international invest-ment agreements: A large sample survey.
OECD International Investment Working Paper .Rodrik, D. (2017).
Straight talk on trade: Ideas for a sane world economy . Princeton UniversityPress.Sachs, L. and L. Johnson (2019). Investment treaties, investor-state dispute settlement and inequal-ity: How international rules and institutions can exacerbate domestic disparities.
Investor-StateDispute Settlement and Inequality: How International Rules and Institutions Can ExacerbateDomestic Disparities (September 11, 2019) .Schill, S. W. and V. Djanic (2018). Wherefore art thou? towards a public interest-based justificationof international investment law.
ICSID Review-Foreign Investment Law Journal 33 (1), 29–55.Schultz, T. and C. Dupont (2014). Investment arbitration: promoting the rule of law or over-empowering investors? a quantitative empirical study.
European Journal of InternationalLaw 25 (4), 1147–1168.Schumacher, J., C. Trebesch, and H. Enderlein (2018). Sovereign defaults in court.
EuropeanCentral Bank, Working Paper Series, No.2135 .Schwert, G. W. (1989). Why does stock market volatility change over time?
The Journal ofFinance 44 (5), 1115–1153.Shiller, R. J. (1981). Do stock prices move too much to be justified by subsequent changes individends?
The American Economic Review 71 (3), 421–436.Shiller, R. J. (1992).
Market volatility . MIT press.Shiller, R. J. (2015).
Irrational exuberance: Revised and expanded third edition . Princeton uni-versity press.Shirlow, E. (2015).
Looking behind the Statistics for Investment Arbi-tration . http://kluwerarbitrationblog.com/2015/02/24/looking-behind-the-statistics-for-investment-arbitration .Simon, H. A. (1991). Bounded rationality and organizational learning. Organization Science 2 (1),125–134. 22an Harten, G. (2011). The use of quantitative methods to examine possible bias in investmentarbitration. in Yearbook on International Investment Law & Policy 859 .Yadav, P. K. (1992). Event studies based on volatility of returns and trading volume: A review.