Corruption in investment arbitration: a move towards a more human approach to evidence?
Professor Timothy Levine is one of the world’s foremost psychologists dealing with the thorny issue of truth telling. He has famously postulated that a human being’s default assumption is their counterparty is honest. Repeated studies by Levine have shown people do much better than average at guessing when people are being truthful, and far, far worse when attempting to point out liars.
The experiments reveal that we, as a species, are only minded to turn our backs on a default assumption of truth when presented with overwhelming evidence that we are being lied to. We look everywhere for those clues – and the natural human approach is to hoover up as much evidence as we can.
Those of us looking for a toolkit to root out liars in the realm of investor-state disputes know this has not always been the approach taken by arbitral tribunals.
In particular, there has been a long running debate as to the standard of evidence required when corruption is alleged. The debate focusses on whether a higher standard of proof should be adopted, disregarding facts that are insufficiently clear – or – whether a more pragmatic approach, on the balance of probabilities, should animate a tribunal. The balance of probabilities approach is promoted by its adherents, in part, because of strong public policy arguments for deterring corruption and because the hidden nature of nefarious dealings often denies a party clear evidence.
Levine’s research suggests that, as a species, we arrive at the truth more often when we do not limit the evidence we will consider. The human approach, therefore, is to weigh as much evidence as possible.
Four recent corruption cases may show a drift in that direction.
Niko Resources v Bapex and Petrobangla
The long running Niko case saw a material decision in February 2019. The original dispute arose in relation to two contracts, a joint venture agreement and gas purchase and sale agreement, concluded between Niko and Bangladeshi state parties including Petrobangla, the Bangladeshi national oil company. The original claims were for payment due for natural gas delivered and a declaration of non-liability for damage to wells drilled in Bangladesh. In 2016, the respondents brought a claim, which had been raised previously as a defence in the jurisdictional stage of the proceedings, that the claimant had obtained the contracts through corruption. The respondents requested a reconsideration of the tribunals’ Decision on Jurisdiction. The respondents claimed the contracts should be declared void ab initio based on new evidence of corruption, and Niko’s claims should be dismissed in their entirety.
The February 2019 decision addressed the issue of the standard of proof in corruption cases. Niko argued for a “heightened” standard of proof and the respondents argued the ordinary standard, the balance of probabilities, should apply.
The tribunals ultimately adopted a “less formalistic sensibility”, saying they did not find “assistance in terms such as ‘preponderance of evidence’ and ‘heightened standard of proof’”. The tribunals decided to follow a pragmatic approach that did not filter evidence to a higher legal standard, adopting the position that to do otherwise would be to “abdicate [their burden of choice] by rote reference to an abstract ‘heightened’ standard of proof”. The tribunals said they would take into account both direct evidence as well as “circumstantial evidence, inferences, […] presumptions and indicators of possible corruption (such as ‘red flags’)”.
In the end, the tribunals found the evidence did not establish the contracts were procured by corruption.
Sanum v Laos and Lao Holdings v Laos
The Sanum v Laos and Lao Holdings v Laos proceedings ran over seven years and gave rise to several other sets of treaty proceedings as well as SIAC commercial arbitrations and battles in the US and Singaporean courts. The underlying dispute concerned investments made in casinos in the Lao People’s Democratic Republic. The respondent argued the claims should be dismissed in their entirety as the investments had been procured and operated corruptly, as the claimants had engaged in bribery, embezzlement and money laundering. The claimants’ primary allegation was expropriation without compensation, flowing from a plan to push the claimants out of Laos.
As in Niko, the parties advanced two opposing standards for weighing evidence of corruption: a heightened standard versus a balance of probabilities standard. In Sanum (and with the same reasoning in the separate award in Lao Holdings), the tribunal ruled that a higher standard should be adopted to ensure “a rigorous testing” given the “severity of the consequences” of corruption allegations. The tribunals did not find there was sufficient evidence of bribery and corruption of public officials based on the higher standard of proof.
In the subsequent reasoning, however, the tribunals did not hold rigidly to this higher standard. The tribunals went on to consider a series of allegations of gross financial impropriety against the lower standard. The tribunals concluded that a lower standard of “balance of probabilities” was relevant to the issue of the claimants’ good faith and the legitimacy of the claimants’ alleged “legitimate” expectations of fair and “equitable” treatment. The tribunals held that the “corruption of Government officials [was] established to the lower standard of ‘balance of probabilities’”.
The consequence of this was determinative. Relying on the principle that “bad faith conduct of the investor is relevant to the grant of relief under an investment treaty”, the claimants’ bad faith initiation and performance of the investments meant the claimants were not entitled to the benefit of treaty protection.
The claims were dismissed on the merits and bad faith was cited throughout the tribunals’ reasoning.
Glencore v Colombia
In Glencore v Columbia, the respondent state claimed Glencore had obtained an amendment to a 30-year coal mining concession agreement through corruption of a Colombian government official.
Colombia argued the tribunal therefore lacked jurisdiction and Glencore’s claims were inadmissible as tainted by illegality. As regards the standard of proof, Colombia argued the tribunal should not apply a heightened standard but instead embrace the entirety of the evidence on the record.
The tribunal’s decision came some six months after the Niko decision and within a month of the Sanum and Lao Holdings decisions. However, in contrast to those decisions, the Glencore tribunal saw “no reason to depart from the traditional standard of preponderance of the evidence”. The tribunal considered it had a duty to consider as wide an ambit of evidence as possible: “provided that there are prima facie grounds for suspecting malfeasance, an international arbitration tribunal has the duty to investigate the facts, even sua sponte, and to take appropriate measures under the applicable principles of law”.
The tribunal quoted the respondent’s argument favourably, that a tribunal should “adopt a methodology of starting from “red flags” (individual indicia of corruption) and “connecting the dots” to obtain a larger picture”. Upon analysing the facts, the tribunal held that the respondent failed to substantiate its allegations.
Still no clear standard – but a more human approach
Although the cases have contradictory elements, the tribunals appeared to show a desire to depart from the application of a rigid and abstract higher standard of evidence in corruption cases.
The ultimate outcome of all four awards rested on an analysis of the preponderance of evidence, with the decisions heavily influenced by a flexible approach to methods of proof.
Professor Levine might say this is only human – and, moreover, the best way to get decisions right, most of the time.
Wilson Antoon, Partner, King & Wood Mallesons
Alexis Namdar, Managing Associate, King & Wood Mallesons
 Niko Resources (Bangladesh) Ltd v Bapex and Petrobangla, ICSID Case No. ARB/10/11 and ICSID Case No. ARB/10/18, Decision on the Corruption Claim, 25 February 2019 (Schneider, McLachlan, Paulsson).
 Id, para. 802.
 Id, para. 806 quoting with approval A. Llamzon.
 Id, para. 805.
 Id, para. 806 quoting with approval A. Llamzon.
 Sanum Investments Limited v The Government of the Lao People’s Democratic Republic, UNCITRAL, Award, 6 August 2019 (Sureda, Hanotiau, Stern).
 Lao Holdings N.V. and the Lao People’s Democratic Republic, ICSID ARB(AF)/12/6, Award, 6 August 2019 (Binnie, Hanotiau, Stern)
 Sanum, paras. 86-93; Lao Holdings, paras. 88-95.
 Sanum, para. 94; Lao Holdings, paras. 96.
 Sanum, paras. 107-108; Lao Holdings, paras. 109-110.
 Sanum, paras. 138, 147, 156; Lao Holdings, paras. 139, 148, 157.
 Sanum, para. 161, Lao Holdings, paras. 162, 278.
 Sanum, para. 104; Lao Holdings, para. 237.
 Lao Holdings, paras. 278-280.
 Glencore International A.G. and C.I. Prodeco S.A. v Republic of Columbia, ICSID ARB/16/6, Award, 27 August 2019 (Fernández-Armesto, Garibaldi, Thomas).
 Id, para. 566.
 Id, para. 669.
 Id, para. 664.
 Id, para. 669.