Screening for Security: What Could Armenia Actually Review without Breaching its Investment Treaties?
Author: Davit Khachatryan (international law expert and lecturer, American University of Armenia, Russian-Armenian University)
Investment screening is, in its conventional justification, a national security instrument. A foreign acquisition of critical infrastructure, a sensitive technology, or a chokepoint asset can transfer leverage to a hostile principal in ways an ordinary commercial transaction does not. Where screening regimes have proliferated over the past decade, from the European cooperation framework to the expanded remit of the Committee on Foreign Investment in the United States, the organising premise has been the same. The state reserves the right to examine inbound capital and to intervene where national security or public order is engaged.
Armenia has so far stood outside this trend. A state actively rebuilding its security posture has obvious reasons to ask whether it should also acquire the capacity to police the capital entering its strategic sectors. The question this post examines is narrower and more technical than the policy debate usually allows. If Armenia were to build a screening mechanism on national security grounds, what could that mechanism lawfully do, given the investment protection regime the state already owes its existing and prospective investors? The answer turns less on security policy than on the architecture of investment protection itself, and it points toward a regime that can do far less than its proponents tend to assume.
A dormant clause
Armenia does not currently operate a formal foreign investment or national security review regime. There is no centralised authority that assesses inbound investment by reference to an investor’s origin, ultimate control, or the strategic character of the target. Sectoral approvals apply in banking, media, civil aviation, and agricultural land, and ordinary competition review applies to concentrations, yet none of these functions as a security filter.
The legislative raw material for a security-based regime does exist, in dormant form. The 1994 Law on Foreign Investments permits legislation to designate territories where the activity of foreign investors may be limited or prohibited on national security grounds (Article 6). That clause addresses geographic restriction rather than transaction review. As a doctrinal foundation for a modern call-in mechanism, it is thin. However, it does establish that a security-based limitation of foreign investment is contemplated within the Armenian legal order rather than alien to it.
The impetus for change is recent and domestic. The 2025 United States Investment Climate Statement records that ownership changes in several key enterprises have prompted debate over whether Armenia needs a formal screening mechanism to deter potentially malign influence. A small state in a contested neighbourhood, reorganising its security relationships under pressure, has reason to want visibility and a measure of control over who acquires its sensitive assets. The difficulty is that Armenia would be introducing such a power against the backdrop of commitments designed precisely to constrain it.
What screening does
A screening mechanism does not, as a rule, close sectors to foreign capital in advance. A negative list approach, as operated in Mexico or China, bars foreign investment in defined sectors ex ante and by category. Screening operates differently: it subjects individual transactions to review and allows the state to clear, condition, or block them according to an assessment of risk. Nothing is excluded by design. The state retains discretion exercised case by case.
The consequence that matters for Armenia is where in the lifecycl of an investment that discretion is exercised. A screening power can kick-in at the point of entry, governing the admission of a new investment, or it can reach established investments through post-establishment review, including the increasingly common call-in powers that allow authorities to examine or unwind transactions after completion. The investment protection regime treats these two moments differently, and the gap between them largely determines what Armenia can and cannot lawfully build.
Pre-establishment
At the point of entry, a host state’s regulatory latitude is at its greatest. Most of the bilateral investment treaties Armenia concluded in the 1990s and 2000s follow the admission model: they extend substantive protection to investments made in accordance with the host state’s laws, and they do not confer a right of establishment. The Armenia-Germany BIT of 1995 is representative: under Article 2(1) each party undertakes to promote and “permit such investments in accordance with its laws,” with no commitment to admit foreign investment in the first place. Investment protection applies once an investment has been admitted under domestic law.
A screening mechanism operating purely at the entry phase of a foreign investment, therefore, sits for the most part upstream of the treaty protections. An investment that is refused admission on security grounds, under published criteria applied through due process, has not yet acquired the status from which fair and equitable treatment, the expropriation standard, or the transfer guarantees are claimed. The legality requirement common to investment treaties, the stipulation that protected investments be made in accordance with host state law, cuts in the same direction: a screening statute becomes part of the law in accordance with which admission must occur.
Two qualifications keep this from being a clean slate. First, where a treaty contains pre-establishment commitments, extending national treatment or most favoured nation (MFN) treatment to the establishment phase itself, the admission decision no longer wholly falls outside the treaty, and a screening regime that discriminates between investors at entry can engage those standards directly. Some modern investment agreements concluded by the United States, Canada, and Japan expressly extend national treatment and MFN obligations to the establishment, acquisition, and expansion of investments. Such clauses become relevant where a screening requirement burdens foreign investors more heavily than comparable domestic investors. This point should not be overstated. Not all screening regimes are framed as foreign investor regimes. For instance, the United Kingdom, and the Netherlands, define their scope primarily by reference to the sector, asset, or transaction risk, and may also require notification by domestic acquirers. The treaty issue becomes sharper where the screening mechanism differentiates based on nationality, origin, or control. Since Armenia’s treaties protect admitted investments, a pre-entry screening regime would be safer. The main risk, then, is discrimination against foreign investors, especially where a treaty also protects establishment or acquisition. That the first publicly known investor-state proceeding to challenge a negative screening decision, Global Telecom Holding v Canada, arose under such a treaty illustrates the exposure, even as the tribunal’s dismissal of the claim shows how far states have insulated these regimes through sectoral carve-outs and reservations (analysis).
Second, the MFN clause is a persistent solvent. A screening regime calibrated to treat investors of particular origins less favourably risks importing, through MFN, the more generous admission or treatment standards a state has granted elsewhere. States that wish to retain a free hand have learned to draft around this: Australia, for instance, listed its foreign investment policy as a non-conforming measure exempt from national treatment and MFN obligations under its treaties with the United States and China, the latter paired with a reservation for measures considered necessary to protect essential security interests. A security screen designed around the nationality of capital sits awkwardly with a treaty network built on non-discrimination, and the awkwardness is doctrinal rather than merely political, though, as the carve-out practice shows, it is one that careful treaty drafting can contain.
Post-establishment
The picture inverts once an investment has been admitted. From that point, the full apparatus of protection is engaged, and a screening power exercised against an existing investment is where treaty liability accumulates.
Fair and equitable treatment (FET) is the first constraint. A foreign investor that was admitted under a regime of openness forms expectations against the legal framework in force at the time of entry. The protection of legitimate expectations, due process, and freedom from arbitrary treatment all bear on a subsequent decision to review, condition, or unwind that investment. A call-in power applied to capital that entered years earlier, when no screening regime existed, is vulnerable on precisely this ground.
The expropriation standard is the second constraint. A forced divestment of a strategic asset is the most obvious case in which a screening measure may amount to an expropriation. More subtle interventions such as conditions on continued operation, restrictions on control, or compelled changes in governance, raise the question of indirect or regulatory expropriation, where the measure’s effect on the investment rather than its form determines the analysis. States facing such claims invoke the police powers doctrine, under which bona fide, non-discriminatory regulation adopted for a legitimate public purpose does not amount to compensable expropriation. The doctrine offers genuine protection, but its application to security- driven intervention against an established owner will depend on the specific facts, including the measure’s proportionality, its non-discriminatory character, and the observance of due process.
This is the doctrinal reason a state cannot simply seize control over assets already in foreign hands. For a country whose sensitive assets largely changed ownership in the privatisations of earlier decades, a new investment screening mechanism would arrive too late to control those transactions at the point of entry.
Security exceptions and the limits of the carve-out
Between the host state’s regulatory power and the investor’s protection sit the doctrines that determine how much intervention justified through securi a treaty will tolerate. The decisive question is how far an investment treaty allows the state to rely on security justifications when it intervenes in a protected investment. For Armenia, the answer depends heavily on the wording of each treaty.
The first is the essential security exception. Some investment treaties include a clause preserving measures a party considers necessary for the protection of its essential security interests, modelled on Article XXI of the GATT. Where such a clause exists, two questions follow: whether it is self-judging, leaving the necessity of the measure to the state’s own determination, and how far a tribunal will review an invocation of it. The arbitral record arising from the United States–Argentina treaty in the wake of Argentina’s financial crisis is divided. The early awards in CMS, Enron, and Sempra declined to read Article XI as self-judging, collapsed the treaty clause into the customary necessity defence of Article 25 of the Articles on State Responsibility, and applied a demanding standard with little deference to the state’s policy choices.1 LG&E found that Argentina had established a state of necessity for a defined period,2 and Continental Casualty went furthest toward deference: it confirmed that Article XI was not self-judging, yet held that, because the provision’s text mirrored Article XX of the GATT, WTO necessity jurisprudence was the more appropriate interpretive reference.3 The annulment committees in CMS and Sempra then criticised the conflation of the treaty exception with the customary defence. The annulment committee in CMS found a manifest error of law but declined to annul the award. The annulment committee in Sempra went further and annulled the award because the tribunal had applied Article 25 of the Articles on State Responsibility instead of Article XI of the treaty.4 The two provisions operate as a primary rule and a secondary rule, respectively, on distinct legal standards.5
The lesson for a drafter is that the protective value of a security exception depends on its precise wording, and that a clause intended to be self-judging must say so unambiguously. The difficulty for Armenia is that many older treaties contain no essential security exception at all. The difficulty for Armenia is that many older treaties contain no essential security exception at all. If Armenia intervenes against a protected investment under such a treaty, it may have to rely on the customary defence of necessity under Article 25 of the Articles on State Responsibility. That defence is stringent and tribunals have read it narrowly. A treaty security exception and the customary necessity defence are doctrinally distinct. A state should not assume that Article 25 will save a measure that a treaty exception would have protected.
The second mediating consideration returns to the legality and transparency demands embedded in fair and equitable treatment. A screening regime resting on vague or unpublished criteria, or applied without procedural regularity, is exposed independently of any security justification, because arbitrariness and opacity are themselves breaches. A regime that is prescribed by law, public in its criteria, and disciplined in its procedure is on stronger ground, both in fitting the requirements of the law and in resisting a fair and equitable treatment challenge.
What follows for design
The doctrinal map yields a reasonably clear set of implications for any Armenian regime.
A screening mechanism is most defensible as a pre-establishment instrument operating on new entrants under published, security-grounded criteria. There, the protection regime is least engaged, future investors form their expectations against the regime as they find it, and the legality requirement works in the state’s favour.
Post-establishment powers are where ambition should be most disciplined. If the concern is cybersecurity, continuity of supply, or operational resilience, Armenia should first consider whether sector-specific regulation can address the risk more precisely than investment screening. Cybersecurity obligations, security of supply rules, licensing conditions, and emergency step-in powers may sometimes do the work without reopening the legality of an already admitted investment. If a call in power is nevertheless introduced, it should apply prospectively, from the entry into force of the legislation, and the statute should define the conditions for review with precision. Conditioning will usually be safer than prohibition. Mitigation arrangements, governance undertakings, security of supply commitments, and carefully framed emergency powers intrude less drastically than forced divestment, provided they are nondiscriminatory, proportionate, and procedurally regular.
Most importantly, a screening statute cannot be drafted in isolation from the treaty network. The protections that constrain government intervention live in the bilateral treaties and the Energy Charter Treaty, not in domestic law, and a security-minded reform that leaves the treaty architecture untouched will have secured the front door while the structural exposure remains where it always was. An audit of the network, attentive to the presence or absence of essential security exceptions, to denunciation and survival clauses, and to the reach of most-favoured-nation treatment, is the necessary companion to any screening proposal. The screening question, properly understood, is as much a treaty question as a security one.
Armenia is in the unusual position of designing this capacity from a blank page, unencumbered by an installed screening bureaucracy and its reflexes. That is an advantage, on the condition that the design begins from the protection regime the state already owes rather than from the security anxieties that make a screening law politically attractive. A screening regime that ignores the treaties will discover its constraints in the least forgiving forum available. A regime with built-in awareness of them can do useful, lawful work, within bounds the policy debate too rarely acknowledges.
Notes
1 CMS Gas Transmission Company v Argentine Republic (Award) ICSID Case No ARB/01/8 (12 May 2005) paras 304-394, esp paras 331 (necessity conditions not cumulatively met) and 373-374 (Article XI not self judging; review not confined to good faith); Enron Corporation and Ponderosa Assets LP v Argentine Republic (Award) ICSID Case No ARB/01/3 (22 May 2007) paras 288–345; Sempra Energy International v Argentine Republic (Award) ICSID Case No ARB/02/16 (28 September 2007) paras 325-397.
2 LG&E Energy Corp v Argentine Republic (Decision on Liability) ICSID Case No ARB/02/1 (3 October 2006) para 226, and paras 256-266 (state of necessity established for the period 1 December 2001 to 26 April 2003).
3 Continental Casualty Company v Argentine Republic (Award) ICSID Case No ARB/03/9 (5 September 2008) para 187 (Article XI not self-judging) and para 192 (GATT Article XX jurisprudence the more appropriate reference, in preference to the customary standard applied in CMS, Enron, and Sempra).
4 CMS Gas Transmission Company v Argentine Republic (Decision on Annulment) ICSID Case No ARB/01/8 (25 September 2007) para 130 (manifest error of law in the interpretation of Article XI; conflation of the treaty exception and the customary defence) and paras 132-136; Sempra Energy International v Argentine Republic (Decision on Annulment) ICSID Case No ARB/02/16 (29 June 2010) paras 200-209 (Article XI as the applicable primary norm; Article 25 offering no guide to its interpretation; award annulled).
5 See Sempra (Decision on Annulment) para 209; Continental Casualty (Award) para 168 (referring to the customary rule only insofar as it assists in interpreting Article XI).
