top of page
Search

Delay, Disclosure, and Disruption

  • Writer: Aişe Gül Akkoyun
    Aişe Gül Akkoyun
  • Oct 10, 2025
  • 3 min read

Updated: 5 days ago

Three developments in October 2025, and what they suggest about where the strain in investment arbitration actually sits.

Enforcement delayed, again

In Smurfit Holdings B.V. v. Bolivarian Republic of Venezuela (ICSID Case No. ARB/18/49), an ad hoc annulment committee decided on 25 August 2025 to continue the stay of enforcement of the award.

The underlying figures are worth stating precisely. The tribunal ruled in favour of Smurfit on 28 August 2024, ordering Venezuela to pay USD 468.7 million in damages plus USD 4.5 million in costs, with interest running from 31 May 2024. The claim was filed in December 2018 following the occupation of the claimant's Venezuelan operations earlier that year.

So the sequence runs: measure in 2018, claim in 2018, award in 2024, and as of late 2025 the award remains unenforceable while annulment proceeds. Seven years in, the claimant has a judgment and no money.

Annulment under the ICSID Convention is meant to be exceptional and narrow. It is not an appeal, and the grounds are limited. But a stay of enforcement pending annulment is close to automatic in practice, and the combination converts a narrow remedy into a general delay mechanism. Venezuela's outstanding obligations under ICSID awards are, by external estimates, well into the billions.

Confidentiality under pressure

A report published this year, Beyond Dispute, found that close to 90 per cent of ICSID cases became public during the proceedings, and that around 18 per cent surfaced before they were formally filed.

The second figure is the striking one. A case that is public before registration is public because someone chose to make it public: a listed company's disclosure obligation, a government's political need to explain a measure, or a party's tactical interest in framing the dispute first.

That has consequences for research design. If the visibility of a dispute depends on the incentives of the parties rather than on any institutional rule, then samples drawn from public sources are not random. They over-represent disputes involving listed claimants and politically salient measures, and under-represent everything settled quietly.

Transparency is usually discussed as a legitimacy question. It is also a sampling question, and the sampling question is the one that determines whether the numbers we publish mean anything.

Energy and climate claims on the rise

ICSID's caseload statistics for 2025 show extractive and energy-related disputes taking a growing share of new registrations, with Latin American respondents, particularly Colombia and Peru, prominent among them. UNCTAD's parallel count records that disputes concerning extractive activities and energy supply rose to more than half of the 58 cases filed in 2024, against roughly a third across the period from 1987 to 2023.

That is a real compositional shift rather than a blip, and it maps onto the same regulatory pressure visible in the energy transition: hydropower licensing, mining permits, environmental review, and the reversal of earlier concessions.

What connects the three

At first glance these are three unrelated items: an enforcement decision, a transparency report, a caseload statistic. The connection is that each describes a gap between what the system is designed to do and what it observably does.

Annulment is designed to be exceptional and functions as delay. Confidentiality is designed as the default and functions as a filter on what researchers can see. Caseload statistics are designed to describe the system and describe only its visible part.

What would be worth measuring

  • Stay frequency: in what proportion of annulment applications is enforcement stayed, and for how long on average?

  • Annulment outcomes: UNCTAD records eight annulment decisions in 2024, of which only one, Agility v. Iraq, resulted in partial annulment. If the success rate is that low, what is the stay actually protecting?

  • Disclosure trigger: for cases that became public before filing, what made them public, and does that correlate with claimant listing status?

  • Sector drift: how much of the rise in extractive and energy cases is new disputes, and how much is better visibility of disputes that would previously have stayed confidential?

The open question

Each of these mechanisms is defensible on its own terms. A State facing a large award should be able to seek annulment. Parties to a commercial dispute have legitimate confidentiality interests. Institutions can only publish what they know.

The difficulty is cumulative. Taken together, they produce a system in which the time from measure to payment is measured in decades rather than years, and in which the evidence base for evaluating that system is assembled from whichever cases the parties found it convenient to disclose.

So the question is not whether annulment or confidentiality should exist. It is whether anyone is measuring what they cost, and whether the reform debate can be conducted honestly without that measurement.

Sources

Comments

bottom of page