When Criminal Proceedings Become a Treaty Question

Aişe Gül Akkoyun
Aug 26, 2025
4 min read

Updated: Aug 5

What Shell's threatened claim against Bolivia would have to establish, and why the forum is the hard part.

Background

Shell has signalled that it may bring an investment treaty claim against Bolivia over measures affecting its oil and gas interests. Two elements of the grievance stand out from the reporting: the annulment of a commercial arbitration award in Bolivia, and what the company characterises as malicious criminal proceedings brought against its local employees.

No claim has been filed at the time of writing. What follows is not an assessment of the merits, which are not public, but an attempt to set out what such a claim would have to establish and why it sits at an unusually awkward intersection of domestic criminal law, commercial arbitration and treaty protection.

The forum problem comes first

The most immediate obstacle is not substantive but jurisdictional, and it is specific to Bolivia.

In May 2007, Bolivia became the first State ever to denounce the ICSID Convention. It subsequently terminated a large number of its bilateral investment treaties, reported at around twenty-one, including agreements with the United Kingdom, Spain, France, China and the United States. That is not a detail; it removes the default forum that most investors in extractive sectors would expect to use.

What survives, and for how long, depends on the sunset clauses in each terminated treaty. These provisions typically preserve protection for investments made before termination, often for ten to fifteen years afterwards. So the threshold questions in any Shell claim would be: which treaty, made when, covering which investment, and is the relevant survival period still running?

Those are unglamorous questions. They are also, on the evidence of the last decade of ISDS outcomes, where a large share of claims actually resolve.

Criminal proceedings as a treaty issue

The allegation that criminal proceedings were used improperly against local employees is the more distinctive feature, and it is not without precedent in Bolivia.

In Quiborax v. Bolivia (ICSID Case No. ARB/06/2), the tribunal ordered the suspension of domestic criminal proceedings on the basis that they risked a chilling effect on witnesses in the arbitration. That case ended with an award of approximately USD 48.6 million to the claimants in September 2015; Bolivia's application for annulment was subsequently dismissed by an ad hoc committee.

The precedent matters less for its outcome than for what it shows about how tribunals handle this category of allegation. A State prosecuting individuals is exercising a core sovereign function. Tribunals are, quite properly, reluctant to characterise prosecution as a treaty breach. But they have been willing to intervene where criminal process appears to be operating on the arbitration itself, by deterring witnesses or pressuring a claimant into abandoning a claim.

That distinction is where the argument lives. Not whether criminal proceedings occurred, but whether they were directed at the dispute rather than at the alleged conduct.

The annulled award

The second element, annulment of a commercial arbitration award by domestic courts, raises a different and older problem: when does a national court's treatment of an arbitral award become an internationally wrongful act?

An award set aside at the seat is, in most jurisdictions, an award that has ceased to exist for enforcement purposes. The investor's recourse against that outcome is not an appeal but a denial-of-justice claim, and the threshold for denial of justice is deliberately high. It is not enough that the annulment was legally wrong. The conduct must fall below a minimum standard: a manifest failure of process, a decision no competent court could have reached, or a pattern indicating that the courts were not operating as courts.

States win most denial-of-justice claims. That is the base rate a claimant has to overcome.

What this looks like as data

Cases combining these two elements are rare, and that rarity is the interesting property. Most ISDS claims involve a regulatory measure: a licence revoked, a tariff changed, a concession cancelled. Claims where the alleged breach consists of how the State's own courts and prosecutors behaved sit in a much smaller category, and one that is under-documented relative to its doctrinal importance.

Several things would be worth coding across that category:

Sequence: did the criminal proceedings precede the arbitration, follow it, or track its procedural milestones?
Target: are the proceedings directed at the corporate entity, at local employees, or at both?
Tribunal response: provisional measures, adverse inference, or no intervention?
Outcome: how often do denial-of-justice claims succeed relative to regulatory-measure claims?

The fourth question is the one with policy weight. If denial-of-justice claims almost never succeed, then the practical protection available to an investor whose award has been annulled at the seat is considerably thinner than the treaty text suggests.

The open question

There is a real tension here that the standard framing tends to obscure.

A State that prosecutes individuals connected to a foreign investment is doing something it is entitled to do, and international law should not make prosecution costly simply because a foreign company is affected. Equally, a State that uses prosecution to make an arbitration unwinnable has found a way to defeat treaty protection without ever passing a measure that looks like expropriation.

The line between these is not drawn by doctrine so much as by evidence: what a claimant can actually show about timing, targeting and effect. Which makes this, in the end, an empirical problem wearing doctrinal clothing.