In order to sharpen, and reify, the issues, consider the United States-China relationship as an example, presenting two relevant types of existential security threats: in addition to the threats of significantly debilitating or ballistic cyberattack discussed above, security competition between these countries includes the risk that acquisition by a competitor of technology or technological capabilities can make a significant difference in the balance of military power. The United States-China relationship is unprecedented in modern history. In the Cold War between the United States and the Soviet Union, there was little need for economic engagement, the Soviet Union was a relatively small economy, and there were few recognized cybersecurity risks. By comparison, the United States and Chinese economies are highly integrated, and decoupling would cause mutual pain.
Because it is mutually economically destructive to decouple these security competitors, it has become necessary to imagine how to maintain trade and investment, while (i) avoiding cybersecurity risks and (ii) avoiding excessive technological empowerment of a security competitor. This article is concerned with creating legal and institutional mechanisms to maximize the combination of gains from international investment on the one hand and security on the other hand. But it is important to recognize that these goals are not easily commensurable. While investment exists in the world of absolute gains—I wish to maximize my gains even if my counterparty also gains—some of the security issues must be viewed in the context of relative gains—I abstain from potential absolute gains to me if they would come at the expense of an advance in the relative position of my adversary.
The United States has recently expanded its national security-based regulation of foreign investment. In 1988, the United States established an interagency committee, the Committee on Foreign Investment in the United States (CFIUS), to review foreign investment pursuant to the Exon-Florio Amendment to the Defense Production Act of 1950. CFIUS examines three main threats, the second and third of which are directly relevant to the concerns addressed in this paper:
1. Denial or manipulation of access to supplies;
2. Leakage (referring to sales of goods or technology, especially of a military nature); and
3. Sabotage or espionage.
This law was amended by the Foreign Investment Risk Review Modernization Act of 2018 (FIRMMA) to provide intensified review of foreign investment in high-technology activities in the United States. Under FIRMMA, the range of transactions subject to pre-acquisition review was expanded to include even minority investments in certain sensitive industries, including investments that provide specified types of access to any industry that (i) “produces, designs, tests, manufactures, or develops . . . critical technologies”; (ii) “owns, operates, manufactures, supplies, or services critical infrastructure”; or (iii) “maintains or collects sensitive personal data” of U.S. citizens threatening national security. Critical technologies include “emerging and foundational technologies” controlled by the 2018 Export Control Reform Act (ECRA).
CFIUS is required to consider “whether a covered transaction involves a country of special concern that has a demonstrated or declared strategic goal of acquiring a type of critical technology or critical infrastructure that would affect United States leadership in areas related to national security,” implicitly targeting China. China has been a growing target of CFIUS scrutiny in recent years. In 2017, then President Trump blocked a Chinese firm from acquiring a U.S. semiconductor chipmaker, Lattice Semiconductor Corp., on the basis of a CFIUS finding that the transaction “pose[d] a risk to the national security of the United States that cannot be resolved through mitigation.” In 2018, President Trump prohibited the proposed acquisition of U.S.-based Qualcomm by Broadcom, based in Singapore, because of concerns regarding Broadcom’s potential adverse effect on the development of U.S. 5G technology. In 2018, one of the authors of FIRMMA, Senator John Cornyn, said that “FIRRMA would help stop China and other bad actors from eroding our military advantage by stealing or otherwise acquiring sensitive dual-use technology by investing in U.S. companies.”
Under CFIUS, the United States may block a foreign acquisition if the president concludes that other U.S. laws (including export control laws) are inadequate to protect national security and determines that there is “credible evidence” that the acquisition threatens to impair U.S. national security. Upon that finding, “the President may take such action . . . as the President considers appropriate to suspend or prohibit any covered transaction that threatens to impair the national security of the United States.” The statute provides that the president’s findings and actions are not subject to domestic judicial review.
This article examines the extent to which international investment law provides sufficient policy space for states to protect themselves through domestic mechanisms like CFIUS from cybersecurity threats and technology appropriation threats. Conversely, it examines the extent to which commitments made in international investment law may be inappropriately undermined by excessive claims of security threats in these areas. In Part II, I examine the types of measures a potential host state may take to protect itself from these threats and compare them to restrictions in international investment law, as well as to the scope of security exceptions in international investment law. I find that these exceptions may be either too narrow or too broad in certain important contexts. In Part III, I elaborate on a contract theory-based approach with regards to these specific types of security threats in the investment law context. In Part IV, I suggest how some institutional responses may address the issues of underinclusiveness and overbreadth of security exceptions, informed by contract theory. In Part V, I conclude.
II. Sovereignty, Schmitt, and the Underinclusiveness and Overbreadth of Security Exceptions
Security exceptions are in tension with the rules to which they allow derogation. The central questions are (i) What parameters will be necessary to allow an exception? and (ii) Who decides? The latter question, in a sense, subsumes the former; as Carl Schmitt observed, “sovereign is he who decides on the exception.” Schmitt’s argument can be understood in terms of incomplete contracting: the law cannot anticipate every circumstance, so, in extreme cases, or in circumstances that the sovereign determines to be extreme, the sovereign decides on its application or disapplication.
While at the limit, the rule of law does not definitively answer social questions, the rule of law may influence, to a greater or lesser degree, decisions of those in power. Therefore, the parameters agreed to in advance, and the institutional structure to make such determinations, are significant prior to the limit. This is true in domestic law and in international law. In international law, states make rules to achieve particular goalsand also recognize that other values may surpass those goals at particular points. In anticipation of these points, they attempt to incorporate exceptions of greater or lesser generality in their rules.
This is because states have a variety of goals. One goal, related to foreign investment, is to promote inbound foreign investment and also to secure opportunities and protection reciprocally for outbound foreign investment. Foreign investment treaties, most often in the form of bilateral investment treaties (BITs) or investment chapters of preferential trade agreements, have grown in number. Each of these treaties has different language and a different context and purpose. These treaties may themselves contain explicit exceptions, including those for national security. In addition, these treaties exist within broader customary international law and treaty law milieu, which supplements, and, in some contexts, may reduce the scope of, the obligations contained in these treaties.
At the time of this writing, China and the United States have not concluded a bilateral investment treaty. There have been many issues that have made it difficult to conclude negotiations. Some difficult issues include the CFIUS review process, national security exceptions, and the correct perception that, at least in terms of market access, in the current era, investors from China would be treated differently and less favorably by the United States than investors from Western countries. If negotiations are to be concluded, it is likely that China would be required to accept that, as a security competitor of the United States, it would find the possibility of investment in certain industries foreclosed. Presumably, China would have a similar set of concerns vis-à-vis the United States. Alternatively, as discussed in Part IV, a more highly articulated approach that relies on establishment of trust with individual firms regardless of their home state or provides for appropriate supervision and verification as conditions for investment may allow more extensive investment to continue.
A. The International Investment Law Context
BITs (including, for these purposes, investment chapters of preferential trade agreements) generally protect foreign investment originating in the counterparty (home) country and sometimes also include market access guarantees providing permission for entry of foreign investment from the home country. The market access guarantees often are framed as requirements of national treatment with respect to the establishment of the investment.
BITs that include these market access guarantees may raise issues regarding whether a host state may exclude certain foreign investors from certain industries in order to carry out a cybersecurity program or establish cybersecurity-based conditions for market entry. In addition, changes in technology or perception result in changes in cybersecurity concerns that may lead to ejection of, or the imposition of more stringent conditions on, foreign investors.
These measures may raise issues under BITs provisions that protect foreign investment from discrimination or mistreatment after establishment. The discrimination may violate national treatment or most-favored-nation treatment obligations. The mistreatment may violate obligations not to expropriate, or, under the so-called customary international law minimum standard, or under a higher standard specified in a BIT, to provide “fair and equitable treatment” and “full protection and security.”
While each BIT is different, some states, such as the United States, have model BITs with which they begin negotiations. For purposes of illustration, I will discuss the relevant provisions of the 2012 U.S. Model BIT. Article III (National Treatment) of the 2012 U.S. Model BIT provides as follows:
1. Each Party shall accord to investors of the other Party treatment no less favorable than that it accords, in like circumstances, to its own investors with respect to the establishment, acquisition, expansion, management, conduct, operation, and sale or other disposition of investments in its territory.
2. Each Party shall accord to covered investments treatment no less favorable than that it accords, in like circumstances, to investments in its territory of its own investors with respect to the establishment, acquisition, expansion, management, conduct, operation, and sale or other disposition of investments.
Note that Article III(1) provides for national treatment as to establishment—this is a commitment to market access for investment. Article III also provides for national treatment for foreign investors and their investments—treatment no less favorable than that accorded domestic nationals. So the question raised in connection with security-based restrictions on investment is whether exclusions or special conditions applied to foreign investors or their investments would constitute less favorable treatment. Current jurisprudence is somewhat uncertain as to the circumstances under which differential treatment can be justified in a way that avoids its characterization as “less favorable.” But where the basis for the different treatment is founded merely on different nationality, as opposed to different risk characteristics, it would be unlikely to withstand national treatment scrutiny. Similar issues would arise under the most-favored-nation treatment obligation of Article IV, in cases where a state determines to treat foreign investors or investments from different countries differently in connection with cybersecurity risk.
In addition, Article V of the 2012 U.S. Model BIT provides that “each Party shall accord to covered investments treatment in accordance with customary international law, including fair and equitable treatment and full protection and security.” The scope of “fair and equitable treatment” and of “full protection and security” may restrict security-based restrictions on foreign investments, including when a foreign investor or investment is subjected to costly requirements or restraints based on cybersecurity concerns. While the 2012 U.S. Model BIT provides relatively restrictive definitions of fair and equitable treatment and full protection and security—referring to due process and police protection—other investment treaties do not delineate the scope of these obligations in this way.
Finally, practically all BITs contain restrictions on the right of a state to expropriate foreign investments, sometimes including extensive definitions of expropriation, as well as specifications about the level of compensation required in connection with expropriation.
In the case of Global Telecom Holding S.A.E. v. Canada, Global Telecom claimed that a measure by Canada to modify its regulations relating to ownership violated the obligations in the relevant BIT for fair and equitable treatment. Specifically, Global Telecom argued that Canada used a statutory national security review as a pretense to inappropriately gather information regarding Global Telecom’s business plan and to force the sale of its subsidiary. It argued that the national security review lacked a legitimate basis and was, therefore, arbitrary and unreasonable. Canada argued that states must be accorded deference in their national security affairs, citing the award in Devas v. India discussed below:
An arbitral tribunal may not sit in judgment on national security matters as on any other factual dispute arising between an investor and a State. National security issues relate to the existential core of a State. An investor who wishes to challenge a State decision in that respect faces a heavy burden of proof, such as bad faith, absence of authority or application to measures that do not relate to essential security interests.
The tribunal found that Global Telecom failed to meet the relevant burden of proof.
1. Security Exceptions
Many BITs include clauses making the protection of essential security interests a basis for exception from their obligations,justifying an action of the state that would otherwise be prohibited.
These security exceptions may be overbroad in that they are sometimes framed as “self-judging” and may also provide excessive scope for state action in other dimensions. To the extent that these clauses are self-judging, and perhaps in other ways as well, they may provide excessive latitude for states to use pretextual claims of security concerns to avoid their obligations or may allow states to defect from their commitments, imposing costs on their counterparties where they could incur a smaller cost to themselves to mitigate the harm.
For example, Article XVIII (Essential Security) of the 2012 U.S. Model BIT contains the following security exception:
Nothing in this Treaty shall be construed:
1. to require a Party to furnish or allow access to any information the disclosure of which it determines to be contrary to its essential security interests; or
2. to preclude a Party from applying measures that it considers necessary for the fulfillment of its obligations with respect to the maintenance or restoration of international peace or security, or the protection of its own essential security interests.
This model has a self-judging or subjective feature based on the “that it considers” phrase. While these types of clauses are not common in investment agreements, their use has grown, and, in recent years, most new international investment agreements contain such clauses. It is not clear whether the self-judging feature extends to the definition of “essential security interests.” Recently, a World Trade Organization (WTO) panel limited the scope of the self-judging aspect of a similar provision in Article XXI of General Agreement on Tariffs and Trade (GATT). In any event, most BITs that contain security exceptions do not contain language such as “that it considers,” with the result that whether a measure is necessary for the protection of the acting state’s essential security interests is an objective question and is not self-judging.
These clauses are also sometimes underinclusive, insofar as they limit the availability of the security exception to circumstances that do not necessarily sufficiently include cybersecurity threats and technology appropriation threats as defined here. For example, the draft EU-China Comprehensive Agreement on Investment (CAI), ratification of which was suspended in the summer of 2021, largely tracks the text of the GATT Article XXI security exception with its self-judging feature but also limits the security purposes similarly. The relevant language of CAI, Section VI, Article X, reads as follows:
Nothing in this Agreement shall be construed:
(b) to prevent a Party from taking an action which it considers necessary for the protection of its essential security interests: (i) connected to the production of or traffic in arms, ammunition and implements of war and to such production, traffic and transactions in other goods and materials, services and technology, and to economic activities, carried out directly or indirectly for the purpose of supplying a military establishment; . . . ; or (iii) taken in time of war or other emergency in international relations.
This language appears to be based on Article XXI of GATT. Like Article XXI of GATT, these limits would seem to exclude many types of measures designed to protect against both cybersecurity threats and technology appropriation threats. First, cybersecurity would require restrictions that go beyond production or traffic of weapons or materiel to support a military establishment, so item (i) does not cover cybersecurity-based controls. Second, investment relating to AI or robotics may or may not be seen as related to implements of war or may be seen as too indirectly related to implements of war or material to support a military establishment. Third, much of the motivation for these cybersecurity and technology appropriation motivated controls relates to a future war or emergency, not a current one.
A BIT security exception was considered in connection with arbitration cases relating to Argentina’s 1999-2002 economic crisis. Article XI of the Argentina-United States BIT provides:
This Treaty shall not preclude the application by either Party of measures necessary for the maintenance of public order, the fulfilment of its obligations with respect to the maintenance or restoration of international peace or security or the protection of its own essential security interests.
Note that this provision contains no indicator that it is self-judging, and, indeed, the tribunals that considered it indicated that, without explicit language making the security exception self-judging, it is not.
In CC/Devas (Mauritius) Ltd. v. The Republic of India, a tribunal seated at the Permanent Court of Arbitration examined claims by Devas that India expropriated its investment and violated its obligation of fair and equitable treatment under the India-Mauritius Bilateral Investment Promotion and Protection Agreement (BIPA). An arm of the Indian government had terminated a frequency allocation contract based on claims that it prejudiced India’s essential security interests. Article 11(3) of the BIPA provided as follows:
The provisions of this Agreement shall not in any way limit the right of either Contracting Party to apply prohibitions or restrictions of any kind or take any other action which is directed to the protection of its essential security interests, or to the protection of public health or the prevention of diseases in pests or animals or plants.
The tribunal found that this language was not self-judging because it does not contain specific language making it self-judging. The tribunal also pointed out that the provision does not contain a “necessity” qualifier, and so none applied. The respondent was required, as a condition for the availability of the exception, to establish that its measure “related to” its essential security interests. Here, the tribunal accorded a wide measure of deference to the respondent state.
In a number of the Argentinian economic crisis cases, the issue emerged whether economic crisis could be a basis for invocation of this type of security exception. This question is important to the question of the availability of a security exception for cybersecurity or technology appropriation threats. Several of the tribunals rejected the argument that Article XI was only applicable in circumstances amounting to military action and war. One tribunal stated that to find that a severe economic crisis could not constitute a national security issue was “to diminish the havoc that the economy can wreak on the lives of an entire population and the ability of the Government to lead.” For the same reasons, this type of provision might be interpreted to be invocable in order to avoid cyberattack-based havoc or perhaps technology appropriation vulnerability. The Argentina tribunals varied with respect to their interpretation of the degree of severity of disruption that would be necessary in order to invoke the security exception. It is important to note that Article XI does not provide a limited set of security circumstances that can give rise to the exception, as found in GATT Article XXI and in CAI.
Where a treaty includes no security exception, or perhaps even where a security exception is by its terms inapplicable, a customary international law defense of necessity, based on security needs, may still be available. But under Article 25 of the Articles on State Responsibility, the customary international law necessity defense requires that the non-compliance “(a) is the only way for the State to safeguard an essential interest against a grave and imminent peril; and (b) does not seriously impair an essential interest of the State or States towards which the obligation exists, or of the international community as a whole.” These limitations may prevent the use of Article 25 in the typical cybersecurity threat or technology appropriation threat case, in part because the peril may not be “grave and imminent.”
In these cases, customary international law—such as the customary international law reflected in Article 25 of the Articles on State Responsibility—may be available to supply a default rule of excuse for necessity. But the Article 25 rule may not be broad enough to include these threats in full, for example, because the threat does not constitute a grave and imminent peril or because the restriction on investment is not the only way to safeguard the security interest.
This brief review suggests that there may be circumstances in which a state feels bound to take measures to protect against cybersecurity threats or technology appropriation threats but may not have a security exception available. This may be because these threats do not necessarily fit within the relevant language. Furthermore, some BITs do not contain security exceptions at all. Given the concerns described above, that defenses against cyberattacks or technology appropriation threats through investments may violate other provisions of investment liberalization treaties, states may wish to review their policies.
III. The Incomplete Contracts Approach to Security Exceptions in Connection with Cybersecurity Threats and Technology Appropriation Threats
Treaty provisions or customary international law doctrines providing exceptions may be understood in different ways. First, an exceptional provision provides a hierarchy of values. That is, compared to the normal obligations, when the exception applies, the value incorporated in the exception eclipses the value incorporated in the normal obligation. Note that the very idea of a hierarchy of values limits the ability to commensurate and balance among those values. Second, an exceptional provision allocates costs in the sense that, where the exception applies, the person to whom the normal performance is owed is denied that right. In both dimensions, exceptions may fail to maximize net social welfare by providing an optimal combination of effectuation of the relevant policies. By providing a strict hierarchy of values, they fail to optimize the combined value of the multiple values that may be at stake. They are also possibly distributively unsatisfactory and fail to provide incentives for optimal behavior, insofar as one party bears all the costs of the eventuation of the exceptional circumstance, with the possible adverse effect that the other party may not have sufficient incentive to reduce those costs.
Investment law contains multiple types of obligations, and the likely costs of violation will vary significantly across the different types of obligations. Perhaps most often, security exceptions will apply ex ante to prevent establishment or acquisition. In that case, the reliance costs, sunk costs, or asset specificity may be relatively small, and the investor’s loss is largely a loss of opportunity costs. Therefore, the equities may weigh in favor of granting greater autonomy to states to exclude foreign investment that presents security risks. But sometimes, security exceptions may be applied to existing investments, for example, where an existing investment is found by the host state to present a security threat. The result may be expropriation of a valuable asset, failure to comply with full protection and security, or fair and equitable treatment obligations with respect to an investment, possibly imposing costs on the investor as high as the total value of the investment.
On the other hand, security concerns may vary in magnitude. For our purposes, we will put aside ordinary privacy interests or consumer protection concerns and focus on (i) high risk cybersecurity, where the investment establishment itself, or the goods or services it produces, may be weaponized through cyberattack, and (ii) weapons denial restrictions, where an investment might be utilized to gain access to technology or materials that can be used to advance weapons systems in areas such as AI or robotics.
Thus, in ordinary circumstances of low costs to the foreign investor and high potential costs to the potential host state, perhaps the simplest approach is to allow the host state to limit foreign investment. But there will be circumstances in which the costs to the foreign investor are higher, while the benefits to the host state are less compelling. This results in a more difficult calculus and raises the possibility that a more nuanced approach might be valuable. This seeks to induce or structure more efficient use of security exceptions such as an efficient breach.
Efficient breach would occur in this context where the value of violation of the general provision is less than the value of the security interest protected. A rule of efficient breach would ordinarily require the breaching state to pay damages equal to the value to the foreign investor of the breaching state’s compliance. Also, a rule of damages designed to allow efficient breach might cause the breaching state to structure its breach to minimize costs to the foreign investor in order to minimize its obligation to pay damages.
It is worth noting that the international law of expropriation already contains a mechanism that may be understood as akin to efficient breach. The duty to pay compensation would ensure that the security concern prompting state action would outweigh the cost of paying compensation. This proposition depends in part on the way in which compensation is calculated. Furthermore, an act prompted by valid security concerns would not necessarily violate obligations of fair and equitable treatment or full protection and security, especially if investors are compensated for the action taken.
In his 2015 article on economic necessity in international investment law, Alan Sykes develops a contract theory approach to the problem of exceptional clauses referencing economic necessity. The economic necessity context is somewhat distinct from the security exception context because, in the types of security contexts discussed here, the state faces an existential threat. This makes it difficult to commensurate between the lost economic welfare from investment on the one hand and the risk of security catastrophe on the other hand. The existential threat context is one where we might consider the parties engaged in a competition for relative advancement vis-à-vis the adversary (a relative gains context), whereas, in the economic necessity context, each party seeks to maximize its own absolute gain without regard for the gains of its counterparty (an absolute gains context).
Interestingly, Sykes, writing in 2015, assumes that states would have sufficient reputational incentives not to abuse discretion accorded pursuant to security exceptions. In her commentary on Sykes’ paper, Anne van Aaken questions the effectiveness of these reputational effects, especially in the BIT context. The subsequent experience of the United States and other states suggests that security exceptions may be abused.
Sykes explores the question of how an efficient breach regime might work in connection with breaches of international investment law obligations and explores the extent to which the existing regime may be characterized as permitting efficient breach. Efficient breach is permitted by a contract or treaty structure that allows actors to violate prior obligations when the benefit of violation exceeds the cost of violation. One way to ensure that this is the case is to require the violating obligor to compensate the harmed obligee in an amount equal to the value to the obligee of the obligor’s performance. Under this rule, an obligor will only utilize the capacity to breach if it is efficient in the sense that the value to the obligor of breach is greater than the value to the obligee of performance. Under an efficient breach incentive structure, the obligor honestly values its own interest in security and balances that value against the damage to the investor. But this is not the rule specified in most BITs. Sykes explores whether this is the rule provided by customary international law.
As noted above, under Article 25 of the Articles on State Responsibility, unless the parties have agreed otherwise, a defense of necessity may be invoked to preclude the wrongfulness of an act if it “is the only way for the State to safeguard an essential interest against a grave and imminent peril.” This test seems too stringent for the context of cybersecurity or technology appropriations security risks from foreign investment, which are not necessarily grave and imminent in a conventional sense. While the peril may be grave in either case, it may not be imminent. Furthermore, there may be other very costly means to avoid the peril. So it seems appropriate for investment treaties to specify a more lenient standard for a necessity defense.
Sykes notes that Article 27 of the International Law Commission’s Articles on State Responsibility provides that the authority to act under necessity is “without prejudice” to the right of the injured party to compensation. The Articles on State Responsibility may be understood to supplement otherwise applicable customary international law and otherwise applicable treaty law, at least where the parties are not understood to have sought to achieve a different legal rule as a matter of lex specialis.
A relative gains context—such as that posed by a high-risk cyberattack or advancement of the relative position of the security adversary—might be comparable to the domestic common law doctrine of physical or legal impossibility because a host government finds it politically or morally impossible to accept the existential threat posed by the possibility of high-risk cyberattack or excessive advancement of the relative position of its security adversary. Under the U.S. common law impossibility doctrine, when performance becomes impossible, the obligor is excused from fulfilling its obligation and is not required to compensate the obligee.
In the WTO context, although there seems to be no case of a state being required to provide compensation where it relies on a general exception or security exception, in the Asbestos litigation, the panel and the Appellate Body interpreted Article XXIII(1)(b) of GATT as allowing claims for compensation for non-violation nullification or impairment, even in cases where the action satisfied the requirements for an exception under the general exceptions provision of Article XX of GATT.
There are four problems with a necessity exception that does not include a compensation obligation.
- First, such an obligation may be more prone to abuse because it may be difficult to verify true cases of “security impossibility.”
- Second, if the obligation is structured as a flat prohibition, without an effective ability to act while compensating the injured party (a property rule rather than a liability rule), states are likely to be reluctant to allow judges or others to determine whether their security concerns are a suitable basis for an exception, and, even if they do agree to third-party determination ex ante, they may refuse to accept the determinations ex post.
- Third, a duty to compensate will provide appropriate incentives for host states to identify means of protecting their national security interests that may be less costly than the costs occasioned by the proposed violation of their investment law obligations. Where judicial identification of “less restrictive alternatives” may be difficult or may be viewed as excessive interference in sovereign prerogatives, a duty of compensation incentivizes the actor itself to identify less restrictive alternatives.
- Fourth, while law and the economics impossibility doctrine suggest placing the costs on the most efficient cost avoider—the person who is the best risk-bearer—it may be difficult to measure the lost profits or other benefits that might have accrued from non-violation of the investment law obligation. But it would be even more difficult to measure the security losses avoided, especially in advance. Therefore, it is impossible to know which party is the best risk-bearer for these losses. Thus, there seems to be little reason to impose the full costs of the violation on the investor home state. An obligation of compensation, limited to the minimal costs to the home state, would provide both the home state and host state appropriate incentives to minimize costs.
This all argues for imposing a compensation obligation on a host state that invokes the security impossibility exception. Of course, as discussed below, this type of obligation is subject to problems of calculation and enforcement. Furthermore, it may be possible to devise less restrictive alternatives to claiming an exception, allowing states, through cooperation, to maximize joint welfare.
IV. Less Restrictive Alternatives and Adjuncts to Security Exceptions: Trust and Verification
The degree to which international investment law may restrict the ability of states to establish security restrictions addressing (i) risks of cyberattack and (ii) risks of technological appropriation competitive detriment is somewhat uncertain. The language of these exceptions was not well-designed for application in this context. As discussed in Part II, if the exceptions are interpreted broadly, then state restrictions on investment will not be required to be precise or well-designed to address the trade-off between investment and security. Thus, the level of restrictions on investment may be overbroad. If the exceptions are interpreted more narrowly, for example, to exclude concerns based on threats of technology appropriation, they may not provide the ability of states to protect their security.
So, unless the jurisprudence develops to provide more specific constraints on the use of the security exception, or states are able to agree on a legal and institutional arrangement beyond the existing investment law systems to verify the security of establishments and their products and services, some excessive security restrictions will be imposed, and some states may be required to choose between taking efficient security action and complying with their international investment law obligations. Therefore, it is useful to imagine what the structure of a legal and institutional arrangement to verify security of investments might look like and how it might interact with the existing investment law regime.
It might appear that the bargaining power in negotiating such a regime lies with the host states, but, because investment is often valuable to host states, both sides can be expected to be interested in establishing an appropriate regime. Furthermore, many relevant states will be both a capital importer and a capital exporter, so that reciprocity will be an important incentive. This is the case with the United States and China. Finally, states engaged in geo-economic competition may be willing to utilize an appropriate regime to ameliorate the security competition, in the way that arms control agreements have been used. It is important to note that verifying satisfactory security from cyberattacks and verifying that the technologies associated with the investment will not be used to gain military advantage are two different things.
Thus, as suggested by the analysis in Part III, security exceptions, as they presently are framed in international investment treaties and customary international law, have two limitations. First, they are not specific enough regarding the types of security risks involved and the role of risk mitigants, such as mechanisms for selectivity regarding types of investors, transparency, surveillance, and enforcement that may reduce risk to acceptable levels. Second, they do not provide for the possibility that the host state would compensate the home state, or its investors, for lost opportunities or rights due to the assertion of a security exception. Both limitations may be addressed in investment treaty negotiations by establishing more specific rules and relevant institutions to determine with more precision the availability of security exceptions.
States may refuse to comply with or accept international investment law obligations that require them to accept an investment that would place them in danger. Security is often a paramount concern. It will then be useful to identify possibly revised or reinterpreted obligations that provide appropriate incentives for states not to abuse the security exception and to minimize the scope and cost of assertions of security exceptions.
A. Trust and Verification: Cybersecurity
This subsection examines how trust and verification may be used to address cybersecurity concerns. One major obstacle to permitting foreign investment in high-risk technology sectors, where the establishment itself, its products, or its services may pose cyberattack dangers to the host country, is that, by definition, a very high degree of confidence of safety is required. This confidence can only be supported by high levels of trust or high levels of verification.
While trust is often associated with nationality, or at least nationality of beneficial ownership and control of the investor, this basis for trust is both over-inclusive and underbroad. First, it is necessary to look beyond the owner or producer to evaluate the trustworthiness of its supply chain and employees. Furthermore, if production or other relevant activities take place outside the territorial jurisdiction of the host state, then the cooperation of other relevant states must be examined as well. Therefore, trust can only be established by a detailed due diligence analysis of the persons and activities involved at all phases of investment activity that present vulnerabilities and continued surveillance and supervision of these activities. It may be possible to appoint or deputize a trusted person to make an enterprise more trustworthy. I discuss this possibility in connection with CFIUS mitigation measures below.
The core technical problem of ensuring cybersecurity in connection with investment lies in evaluating the initial and updated software utilized in the establishment or in its goods or services. This can be achieved through trust or verification, although both will be subject to hacking:
- Trust. As to trust, the focus is on evaluating the person controlling or obtaining access to the software. This is not a simple question of the nationality of suppliers but is complicated by the need to include (i) ownership, (ii) supply chains, and (iii) employees in the determination of trust.
- Verification. Verification focuses on the software itself, rather than the controlling person. It involves establishing a sufficiently trusted evaluator with the technical capabilities needed to evaluate software to be installed initially or upon update and also with the ability to ascertain whether the software actually works. Verification is extremely difficult for complex software. This problem is compounded by the problem of ensuring that the software actually utilized is the same as the one that was tested. States may decide to establish rules for transparency or verifiability of software. Software utilized in production, products, or services must be sufficiently alike to that which has been evaluated.
- Hacking. One critical facet of verification will involve ascertaining whether the level of vulnerability of the investment facility and its products to hacking of software and hardware is acceptable.
The obstacle to investment in goods or services facilities that presents high cybersecurity risks is that, by definition, a very high degree of confidence of safety is required. This confidence can only be supported by high levels of trust or high levels of verification.
Verification with respect to the activities of a foreign investment may vary depending on the degree to which trust in the controlling person is established. Verification will involve assessment of the vulnerability of the software and hardware comprising the facility or its products or services, including vulnerability to hacking by third parties.
States will be reluctant to give up the flexibility of the security exception that allows them to block investment in high-risk facilities entirely or block investment in situations where the investment is controlled by persons that, in the importing state’s view, do not sufficiently merit trust. Under these circumstances of high risk and low trust, strong verification is the only basis for investment.
B. Trust and Verification: Technology Appropriation
The problem with technology appropriation in connection with investment involves ensuring that technological knowledge or opportunities derived from investment in the host state are not available to enhance the military capabilities of the home state.
There are two phases of technology appropriation. The first involves actual weapons or dual use technologies that can be utilized as weapons, or “direct threat items.” The second involves goods, services, or technologies that will assist the development of technological capabilities of the security adversary so that the adversary will be able, indigenously, to develop AI, robotics, or other advanced technologies that will advance its security competition position, or “indirect threat items.”
Direct threat items are covered, to some extent, by export controls in countries like the United States that prohibit the transfer or disclosure of these goods or technologies to specified countries. The modern U.S. export control system focuses on “end-user analysis”—requiring that the end-user be, in effect, a trusted person.
There is still the question of what types of investment can be permitted while ensuring compliance with these prohibitions. If a Chinese company acquires a U.S. company that holds restricted technologies, how can assurance be provided so that the technologies will not be transferred or disclosed to Chinese government officials or their agents? This question is especially difficult because Article 7 of China’s National Intelligence Law of 2017 requires that “any organization or citizen shall support, assist, and cooperate with state intelligence work according to law.”
China’s National Intelligence Law may apply extraterritorially to activities of Chinese organizations or citizens outside of China, including in the United States. On the other hand, U.S. law would prohibit some of these actions. It is uncertain whether the “according to law” language in Article 7 would limit activities that would violate host country law.
Indirect threat items raise more profound issues because the investment limitations they suggest may be far broader and because these limitations would not naturally fit within conventional security exceptions.
Depending on the types of direct threat items or indirect threat items involved, foreign investment by potential security adversaries may be prohibited or may be subjected to safeguards. In the U.S. CFIUS regime, “mitigation measures” are often required as a condition for investment. These mitigation measures are highly detailed and context-specific and are intended to limit the extent to which security threats are posed by foreign investment.
Section 721(l) of CFIUS, as amended by FIRRMA, provides that the Committee may require measures to mitigate national security risks as a condition for concluding a transaction. In 2020, CFIUS required mitigation measures with respect to twenty-three notified covered transactions, which it estimated comprised approximately twelve percent of the total number of notified transactions that year. The statute requires CFIUS to base its determination “on a risk-based analysis, conducted by the Committee, of the effects on the national security of the United States of the covered transaction, which shall include an assessment of the threat, vulnerabilities, and consequences to national security related to the transaction.”
Mitigation measures utilized under CFIUS may have several different components, including the following:
- Limiting the transfer or sharing of certain intellectual property, trade secrets, or technical knowledge;
- Ensuring that only authorized persons have access to certain technology;
- Ensuring that only U.S. citizens handle certain products and services and that certain activities and products are located only in the United States;
- Establishing a Corporate Security Committee and other mechanisms to ensure compliance with all required actions, including the appointment of a U.S. Government-approved security officer and/or member of the board of directors and requirements for security policies, annual reports, and independent audits; and
- Periodic reporting and on-site audits and continued monitoring by the CFIUS committee.
Furthermore, as Reinsch and Benson suggest and elaborate:
A new approach to export controls, although it has antecedents, has been to utilize new software and hardware technology to better track the use of exported items and to restrict access to authorized users. This approach could potentially guarantee that access to certain exported advanced technology would be restricted to authorized end users, and it could give enforcement authorities the capability of knowing if items were being used consistent with the terms of the license authorizing their export.
This approach would include elements that use technological means to restrict access to a sensitive product only to verified end-users and review data to monitor compliance with end-user and end-use based restrictions. This approach may also allow mitigation of technology appropriation risk in connection with foreign investment.
As noted above, presidential determinations and actions under CFIUS are not subject to judicial review under U.S. law. For the same types of reasons that motivate security exceptions in international economic law, U.S. national security authority to the president often limits judicial review. But, in order to ensure that these measures are not used beyond their intended purposes, or arbitrarily or capriciously, it would be reasonable to impose national judicial review and international review to determine compliance with a revised security exception.
International review could be carried out by arbitral tribunals, by another international tribunal, or by a special investment security court and would be based, of course, on a revised national security exception. This revised national security exception should permit a “right to regulate” in the national security interest, along the lines of what would be expected from a good faith application of CFIUS rules. van Aaken suggests that a good faith test can maintain effective obligation while preventing opportunism. By “good faith,” I mean an objective requirement of a reasonable and reasoned explanation of the magnitude of the risk to essential security interests, the possibility of mitigation at a proportionate cost, and the proportionality of the measure to the risk. This seems to be the approach taken by the CFIUS process in most cases, and so, from a practical and political standpoint, the United States might be expected to accept international disciplines of this nature.
V. Toward a Revised Security Exception Relating to Market Access and Forced Disposition in International Investment Law
This article has argued that existing national security exceptions in connection with international investment law are both under-broad and overinclusive. To review, first, existing national security exceptions do not necessarily fully encompass the types of insidious, non-immediate, yet great threat of cyberattack through foreign investment. Nor do they provide permission for controls on access to weapons or dual use goods, services, or technology or technology appropriation-based controls on access to technologies, or technological development that might facilitate growth of a potential adversary’s weapons capabilities. Second, and conversely, these exceptions sometimes provide excessive discretion for abuse of security exceptions to serve protectionist purposes. This is especially true when the exceptions are expressed in self-judging terms. Interestingly, if the under-breadth is remedied, there will be less need for this overinclusiveness because all legitimate security purposes would be adequately permitted to be fulfilled. How might a revised security exception respond to these concerns?
First, exceptions should encompass a broader set of rationales, including cybersecurity based on potential for high-risk attacks and those drawn from weapons-denial (including dual use) based export controls.
Second, while deferring to state determinations of essential security interests and necessity, exceptions should require good faith national assessments of high risks to essential security interests with detailed explanations and should also require that the national measure taken be proportional to the risk. Less restrictive alternatives such as mitigation should be utilized where they are effective and not excessively costly in relation to the benefit.
Fourth, where a private investor is damaged by utilization of the security exception, the investor should be compensated for actual damages caused by reasonable reliance on expectations of market access. Reasonable expectations would not exist where the host state has provided sufficiently specific ex ante limits on investment. On the other hand, requiring an established foreign investor to exit for security reasons would ordinarily give rise to larger reliance damages.
Here is a model provision to provoke discussion:
Article ____: Essential Security
Nothing in this Treaty shall be construed:
1. to require a Party to furnish or allow access to any information, the disclosure of which it determines to be contrary to its essential security interests;
2. to preclude a Party from applying measures that it considers necessary for the fulfillment of its obligations with respect to the maintenance or restoration of international peace or security; or
3. to preclude a Party from applying measures that it considers necessary in good faith for the protection of its essential security interests, including, without limitation, security against cyberattack and avoidance of enhancement of the weapons development and production capabilities of potential security adversaries, provided (as to this paragraph 3 only) that, prior to its action:
(a) it provides a specific and reasoned explanation showing in good faith a substantial risk to its essential security interests;
(b) its measure is proportional to the articulated risk to its essential security interests, including, without limitation, that its measure includes mitigation measures where effective and not excessively costly in relation to their benefit; and
(c) the investor is compensated for actual damages caused by reasonable reliance on expectations of market access.
