Facts, Evidence and the Burden of Proof in the World Bank Group Sanctions System

Thelargeandgrowingnumberofinternationaljudicialandquasi-judicialbodieshasmadeinternationalprocedurallawanimportantpartofpublicinternationallaw.Thisarticleexamineshowproceduralrulesofacertaintype—provisionsrelatedtofacts,evidence,andtheburdenofproof—havebeendesignedintheWorldBankGroupSanctionsSystem.Themainconclusionisthatsuchrulesplayacentralrole,andthatconsiderableeffortshavebeenmadeduringthelasttwodecadestodevelopawell-functioningbodyofproceduralprovisions.However,thearticlealsoarguesthatcertainpartsofthesystemcouldbedevelopedfurther,inordertomakeitasclearaspossiblewhatisexpectedfromthedifferentactorsintheproceedings.


INTRODUCTION
When describing the international legal landscape of today it is difficult to overstate the importance of international judicial and quasi-judicial bodies: courts, tribunals, panels, boards, and so forth. The main role of such bodies is to interpret and apply substantive rules within different fields of international law. However, if there is a judicial or quasi-judicial body, there is also always some sort of procedural rules, stating inter alia how proceedings can be initiated, the rights of the parties, and how evidence should be treated. In domestic legal systems, there is usually a well-developed body of procedural rules, interpreted in case law, and analysed in the writings of legal scholars, during a long period of time. In international law, on the other hand, the situation is often rather unclear. Treaty negotiators and international judges have their backgrounds in different types of domestic systems: common law, civil law, and others. And to a large extent, they rely on domestic principles and concepts when procedural provisions are negotiated and interpreted. 1 Thus, there is always a risk for the introduction of procedural elements not necessarily well-suited for the international context, or not fitting together. An important research task is therefore to examine the special character of international proceedings, and to analyse how principles and concepts originating in domestic systems are incorporated, interpreted, and applied in international legal regimes. International procedural law has become a research field more relevant than ever before. 2 The aim of the present article is to contribute to a better understanding of how a certain type of procedural rules-rules related to facts, evidence, and the burden of proof-has been designed in an important, but from a scholarly point of view still rather unexplored, international system with strong quasi-judicial features: The World Bank Group Sanctions System (WBGSS). 3 Each year, the World Bank Group (WBG) provides loans and other types of financial assistance, worth billions of dollars, to developing countries around the world, for investments in areas such as education, health, and infrastructure. 4 Fraud and corruption are well-known problems in WBG-financed projects, and the WBGSS has been developed as a key component in the fight against these problems. Like in other international legal systems, it has been relevant to examine domestic legal systems, and to distill general principles of law under international law. 5 And like in other legal systems, rules related to facts, evidence, and the burden of proof play important roles, since they determine inter alia which facts the quasi-judicial body has to be persuaded about, which party shall carry the heaviest burden of persuading the body, and to what extent the body must be persuaded.
After this introduction, section II provides a brief description of the WBG and the historical background of the WBGSS. Since almost all cases handled in the WBGSS so far concern projects related to the two WBG bodies referred to as the World Bank (WB)-the International Bank for Reconstruction and Development (IBRD) and the International Development Association (IDA)-the article will focus on the provisions applicable in this type of cases. 6 Section III analyses a number of fundamental issues related to facts, evidence, and the burden of proof in the WBGSS, in light of how such issues can be understood from a more general, theoretical perspective. Finally, section IV contains a summary of the main conclusions in the article.
Officer (SDO) in the rules applicable in WB cases. Another important change was the expansion of the procedure in 2006, to cover the MIGA, the IFC, and the Bank's Partial Risk Guarantee (PRG) activities, and to cover fraud and corruption more generally, not only in procurement cases. 10 Today, there is a large body of case law concerning the different sanctionable practices, and since 2012 all full-text decisions of the SB are published. There are also two editions of the SB Law Digest, setting out legal principles and core holdings in cases between 2007 and 2019. 11 The 'normative architecture' of the system is described in the Bank Directive of 2016. 12 Five sources of law are mentioned: Articles of Agreement, Policy Framework, Operational Legal Framework, Authoritative Interpretation, and General Principles of Law. These sources are also defined in more detail in the document. The jurisdiction of the WBGSS, and the persons and entities that may be subject to a sanction, are described in the Bank Directive and the Sanctions Procedures. 13 Compared to other international sanctions regimes, the WBGSS has a rather limited scope. While for example the United Nations Security Council may decide on a broad range of sanctions under Article 41 of the UN Charter, sanctions within the WBGSS are limited to a few specific forms, as a response to a few specific sanctionable practices within bank-financed projects. However, regimes similar to the WBGSS can be found in other development banks, for example the African Development Bank Group and the Asian Development Bank. 14

II. FACTS, EVIDENCE, AND THE BURDEN OF PROOF IN THE WBGSS
A. Fact-finding in adversarial proceedings A distinction can be made between inquisitorial and adversarial proceedings. To put it simple, in the first type of proceeding, the judicial or quasi-judicial body is responsible for finding answers to questions of fact. In the second type, this responsibility rests on the parties. The reason why a system is designed in one way or the other can be described as a choice between 'truth' and 'fairness' . 15 In an inquisitorial system, the main goal is to come as close as possible to the truth. The idea is that this can be achieved if the body, as a neutral actor, has the responsibility to investigate factual issues. In an adversarial system, on the other hand, the main goal is a fair result. The idea is that this can be achieved if the body remains passive, and leaves to the parties to present the facts. 10  Common law systems are often described as adversarial and civil law systems as inquisitorial. But in reality, it is probably difficult to find any pure adversarial or inquisitorial systems. 16 Many proceedings-including international proceedings-contain both inquisitorial and adversarial elements, but are of a mainly adversarial character, in the sense that the main responsibility for answering questions of fact rests on the parties. 17 Typically, one of the parties will claim that the other has violated a rule. The judicial or quasi-judicial body may have certain possibilities to investigate factual issues. But it will usually primarily be up to the parties to make claims about facts and produce evidence. Thereafter, the body will determine which facts have been proved and whether the rule has been violated. 18 Like proceedings in other international organizations, the design of the WBGSS has been inspired by proceedings in domestic legal systems. The system is often described as administrative in nature, similar to suspension and debarment processes found in certain countries, but it also contains elements from a range of other procedural models. Since the sanctionable practices constitute conduct considered criminal in most countries, it has been relevant to benchmark the system against criminal proceedings. Elements related to contract and tort law have also been relevant to examine. Moreover, the international character of the system has made it relevant to take into consideration both common and civil law traditions, and arrangements in other international organizations, primarily other multilateral development banks. 19 The WBGSS is adversarial in character, in the sense that there are two parties in the proceeding-the INT and the Respondent-carrying the main responsibility for claiming facts and presenting evidence. Allegations of sanctionable practices are investigated by the INT, performing a function similar to that of a prosecutor, or other type of agent for the government, in a domestic legal system. 20 The INT has an obligation to consider all relevant evidence, also evidence that would reasonably tend to exculpate the Respondent or mitigate the Respondent's culpability. 21 If the investigation shows sufficient evidence to support a finding of one or more sanctionable practices in connection with a WB-financed project, the INT may seek to initiate proceedings. 22 The INT will then submit a 'Statement of Accusations and Evidence' (Statement) to the SDO, which is the first tier in the two-tier system. 23 Moreover, if the SDO finds sufficient evidence in support of the accusations, it will issue a 'Notice of Sanctions Proceedings' (Notice) to the Respondent. 24 One part of the Notice is the recommendation of one of the following five sanctions: reprimand, conditional nondebarment, debarment, debarment with conditional release, or restitution. 25 After this, the Respondent is given thirty days to submit a 'written explanation' (Explanation), with evidence and arguments in support of a withdrawal of the Notice or a revision of the recommended sanction. 26 The SDO will withdraw the Notice if there is a 'manifest error or other clear basis for supporting a finding of insufficiency of evidence against the Respondent' , and that would be the end of the proceeding. 27 In cases where the Respondent does not contest the accusations within ninety days, the proceeding will also end here, and the sanction will enter into force. 28 However, the Respondent may also contest the case by submitting evidence and arguments in a 'written response' (Response) to the SB, which is the second tier in the two-tier system. 29 The INT is then given the opportunity to submit a 'written reply' (Reply) with evidence and arguments. 30 At this point, the Respondent and the INT may also request a hearing, where they can present their cases orally for the SB. 31 After having made a de novo review of the case, the SB will make a 'written decision' (Decision), including a recitation of the relevant facts, the determination as to the culpability of the Respondent, any sanction to be imposed and the reasons therefor. 32 If the SB finds that the Respondent has been engaged in sanctionable practices, the decided sanction will be imposed. Otherwise, the proceeding is terminated. The Decision is final, binding on the parties, takes effect immediately, and is without prejudice to any action taken by any government under its applicable law. 33 As can be seen, the system implies that the SDO and the SB perform the function of neutral bodies, examining evidence and arguments presented by the parties. Formal rules of evidence do not apply. 34 Hence, the parties may present any type of evidence, and the SDO and the SB may use any kind of evidence-including hearsay evidence-in support of their conclusions. Moreover, it is entirely up to the SDO and the SB to determine the relevance, materiality, weight, and sufficiency of the evidence presented. 35 However, beside the above-described adversarial elements of the system, there are also certain inquisitorial elements, especially when a case has reached the SB. The SB has the authority to call and question witnesses, and to question the representative of the INT and/or the Respondent or the representative of the Respondent. 36 If a party refuses to answer, or fails to answer truthfully or credibly, this may be construed against that party.
Systems of a mainly adversarial character, like the WBGSS, need burden and standard of proof rules. Such rules stipulate which party shall carry the heaviest burden of persuading the judicial or quasi-judicial body about the facts (the burden of proof) and to what extent the body must be persuaded (the standard of proof). The burden and standard of proof in the WBGSS will be examined below. However, it can be noted that, while these rules determine how questions of fact shall be answered in the proceeding, the interpretation and application of the rules themselves are questions of law. And in most legal systems-including the WBGSS-the responsibility to answer questions of law rests on the judicial or quasi-judicial body. 37

B. Facts
In a proceeding where one party has claimed that another party has violated a rule, the claim is really about facts that must be proved. A simple example: Rule X stipulates that Act Y is prohibited. The act is described in the rule. If one party-Party A-claims that the other party-Party B-has violated Rule X, the judicial or quasi-judicial body wants to know: (1) how to interpret the description of Act Y in the rule (a question of law), and (2) whether Party B has performed this act in reality (a question of fact). A description of an act in a rule can be referred to as an 'abstract legal fact' , and a claim that this act has been performed in reality a 'concrete legal fact' . 38 The latter is a fact that needs to be proved in the proceeding (factum probandum). 39 When trying to persuade the body about such fact, a party may also claim a fact that is not itself a concrete legal fact, but that is offered in evidence as proof of a concrete legal fact (factum probans). 40 This can be referred to as an 'evidentiary fact' . 41 The concrete legal facts and the evidentiary facts will look different in different cases. The INT will claim facts that are supposed to show that the rule has been violated, and the Respondent facts that are supposed to show that the rule has not been violated. But the possible abstract legal facts in a system are usually the same, since they are part of the substantive rules. In the WBGSS, the relevant abstract legal facts can be found in the provisions describing the sanctionable practices. The WBG Policy contains the following general definition of a Sanctionable Practice: (i) collectively and individually, corrupt practice, fraudulent practice, collusive practice, coercive practice, and obstructive practice, as such practices may be defined in the legal agreements or other documents governing the WBG operation, including any relevant instrument prepared thereunder, in connection with which such practices may occur; and (ii) specific and limited violations of procedures in connection with the Sanctions System as the Board may determine. 42 Even though there is a possibility for the Board-meaning the Executive Directors of the IBRD, IDA, IFC, and/or the Directors of the MIGA-to specify violations, the main descriptions of the sanctionable practices can be found in the first subparagraph: 'corrupt practice' , 'fraudulent practice' , 'collusive practice' , 'coercive practice' , and 'obstructive practice' . Moreover, in the Sanctions Procedures, the term Sanctionable Practice is defined in the following way: (i) with respect to any case under sub-paragraph 1.01(c)(i) of Section III.A, a corrupt, fraudulent, coercive, collusive, or obstructive practice, as such terms are defined in the Anti-Corruption Guidelines, Procurement Guidelines or Consultant Guidelines, as the case may be, under which such case is being brought (see Annex A); (ii) with respect to any case under sub-paragraph 1.01(c)(ii) of Section III.A, a corrupt, fraudulent, coercive, collusive, or obstructive practice, as defined in the World Bank Vendor Eligibility Policy in connection with the Bank's corporate procurement; (iii) with respect to any case under sub-paragraph 1.01(c)(iii) of Section III.A, a violation of a Material Term, as defined in the VDP Terms & Conditions; and (iv) with respect to any case under sub-paragraph 1.01(c)(iv) of Section III.A, a violation of sub-section 11.05 of Section III.A. 43 Sub-paragraphs (iii) and (iv) refer to special types of sanctionable practices, namely breaches of material terms in Voluntary Disclosure Programs (VDP) and breaches of confidentiality provisions in the Sanctions Procedures. But the main sanctionable practices can be found in sub-paragraphs (i) and (ii). These are the same as the five 40  practices mentioned in the WBG Policy. The Sanctions Procedures refer to the definitions in the Anti-Corruption Guidelines, the Procurement Guidelines, the Consultant Guidelines, and the World Bank Vendor Eligibility Policy. 44 These definitions are also normally incorporated by reference into the Bank's legal agreements. 45 Moreover, four of the definitions-corrupt practice, fraudulent practice, coercive practice, and collusive practice-are not only used within the WBGSS. The WBG has agreed with a number of other multilateral development banks, and the International Monetary Fund and the European Investment Bank Group, to apply the same standardized definitions (the Uniform Framework). 46 As can be seen, it is not entirely easy to find the relevant abstract legal facts. One has to look in the relevant guideline or policy. Even though the definitions of the sanctionable practices are not exactly the same, word-by-word and comma-by-comma, in the different documents, the important abstract legal facts are the same. The definition of 'fraudulent practice' can be used as an example. 47 In the most recent guidelines and policy documents, and in the Uniform Framework, the definitions are more or less the same. The following definition can be found in the Uniform Framework: ' A fraudulent practice is any act or omission, including a misrepresentation, that knowingly or recklessly misleads, or attempts to mislead, a party to obtain a financial or other benefit or to avoid an obligation' . 48 There are a number of abstract legal facts in this sentence. The first part is that there has to be an 'act or omission' , and it is specified that this includes 'misrepresentations' . Moreover, the act or omission has to 'mislead' or 'attempt to mislead' another 'party' . And this has to be done 'knowingly or recklessly' , and to 'obtain a financial or other benefit' or to 'avoid an obligation' . In certain documents, some of these abstract legal facts are defined further in footnotes. 49 Hence, in a specific case, the INT will have to claim concrete legal facts matching all of the above abstract legal facts. As can be seen, the facts include both objective and subjective elements. The INT will for example have to prove that there has been an act or omission that can be objectively observed, but also that the Respondent has had certain intentions. The Respondent, on the other hand, will claim concrete legal facts showing that the requirements in one or several abstract legal facts are not fulfilled. And both parties will probably claim evidentiary facts supporting their concrete legal facts.
C. Burden of proof In reality, the judicial or quasi-judicial body will rarely reach full certainty about the concrete legal facts in a case. If we look at the example above, we can see that the parties in the proceeding will claim two different concrete legal facts, where one is the negation of the other. Party A will claim that Party B has performed the act (Y), and Party B that it has not performed the act (∼Y). Hence, Party A will have to persuade the body about Y, and Party B about ∼Y. When we refer to the 'burden of proof' , we usually refer to what could be described as the 'greater risk of non-persuasion' , or simply the 'burden of persuasion' . 50 Each party will carry a risk of not persuading the body, but one of them will carry a greater risk than the other. If the judicial or quasi-judicial body finds, based on the evidence presented, that the claim of Party A seems as probable as the claim of Party B, it still has to make a ruling. One of the parties must lose, and the losing party is the party carrying the burden. Hence, in proceedings of this type it is unavoidable to allocate the burden to one of the parties. 51 The WBGSS includes a proceeding of a mainly adversarial character, and it has therefore been necessary to handle burden of proof issues when developing the system. Already in 2000, in the First Thornburgh Report, it was recommended that the SC should develop 'written policies and statements for determining the sufficiency of proof of wrongdoing' . 52 When the Sanctions Committee Procedures of August 2001 were adopted, there was no regulation of the burden of proof, only a rather vague regulation of the standard of proof. It was stated that the SC should recommend a sanction if it found that the evidence was 'reasonably sufficient' to support a finding that the Respondent had engaged in sanctionable practices. 53 But it was not stated whether the burden should rest on the INT or the Respondent. However, in practice, the burden was normally imposed on the INT. 54 In the Second Thornburgh Report, burden and standard of proof issues were examined rather thoroughly. It was stated that in virtually all judicial and administrative proceedings, there is a requirement that the party initiating the proceeding should carry the burden. It was also stated that in many proceedings, once the initiating party has made out its case, the burden shifts to the other party to overcome the evidence against it. 55 Hence, the recommendation in the report was that the Sanctions Procedures should be changed and that it should be clearly stated that the burden should rest on the INT, but shift to the Respondent when the INT had presented evidence establishing that the Respondent had engaged in sanctionable practices. 56 These recommendations were adopted, and are now part of the Sanctions Procedures applicable today. The WBGSS consists of a two-tier procedure, where the first step is the SDO, and the second step the SB. The INT will submit an accusation to the SDO, and the SDO will issue a Notice if it determines that the accusation is supported by sufficient evidence. After that, the Respondent will be given an opportunity to submit an Explanation. And then the SDO may decide to withdraw the Notice, if there is a manifest error or other clear basis for supporting a finding of insufficiency of evidence. However, the Sanctions Procedures do not contain a specific regulation of how the burden should be allocated during this first step of the proceeding. There is only a specific regulation for the second step. With regard to decisions of the SB, the burden of proof is defined in the following way in the Sanctions Procedures: INT shall have the burden of proof to present evidence sufficient to establish that it is more likely than not that the Respondent engaged in a Sanctionable Practice. Upon such a showing by INT, the burden of proof shall shift to the Respondent to demonstrate that it is more likely than not that the Respondent's conduct did not amount to a Sanctionable Practice. 57 When looking at how the burden of proof is regulated in the WBGSS, and the background documents, it seems clear that the system has adopted the principle actori incumbit probatio (the claimant carries the burden). This is a widely recognized principle, applied in many legal systems, both at the national and the international level. 58 However, when the principle is applied, it is not always clear who should be considered the 'claimant' . The term could refer to the party initiating the proceeding, or to the party making a claim in respect of a certain fact. It appears that the first-mentioned view has been adopted in the WBGSS. In the Second Thornburgh Report it was stated that it is usually the party initiating the proceeding who should carry the burden, and this was followed when designing the regulation in the Sanctions Procedures, where it is stated that the burden rests on the INT. In many situations, the above distinction does not play an important role, since the party initiating the proceeding is also the party claiming the relevant concrete legal facts. But this is not always the case. If one takes a broader view on how burden of proof issues are handled in legal systems applying the above-mentioned principle, it seems reasonable to understand the 'claimant' as the party making a claim in respect of a certain fact. This follows for example from the fact that the principle is often combined with 56  another burden of proof principle: quicunque exceptio invocat eiusdem probare debet (the party invoking an exception carries the burden). 59 If the party initiating the proceeding has made certain claims in relation to a general rule, the respondent may have the possibility to invoke an exception. And the respondent will then be the 'claimant' in respect of the concrete legal facts related to the exception. 60 An additional example is the burden with regard to evidentiary facts. In many legal systems the burden to prove such facts rests on the party claiming the fact, regardless of whether this is the party initiating the proceeding or the respondent in the proceeding. 61 It appears that these latter burden of proof principles have not been discussed when developing the WBGSS. However, this type of broader understanding of the 'claimant' could be useful in discussions in case law, and in future reforms of the system. 62 Furthermore, another important part of the WBGSS burden of proof rules is the provision saying that the burden shall shift to the Respondent when the INT has reached the applicable standard of proof, and that the Respondent will then have to reach the same standard. This provision may cause some confusion in light of the rather established view that, in basically all legal systems, the burden does not shift when it has been allocated to a certain party. 63 So the question is how the provision should be understood. Statements in background documents and case law indicate that the provision does not concern the 'burden of persuasion' , but rather the 'burden of production' . 64 The main difference between the two burdens is that, while the former becomes relevant at the end of the proceeding, the latter is relevant during the proceeding. At least in theory, one could imagine a point in time in the proceeding when the party carrying the burden of persuasion has presented enough evidence to reach the required standard of proof. At this point, the burden of production shifts to the other party, which has to produce evidence to persuade the judicial or quasi-judicial body. But at the end of the proceeding, the body will have to consider the totality of the evidence presented and determine whether the party carrying the burden of persuasion has reached the required standard of proof.
However, the burden of production only becomes relevant in a proceeding where the judicial or quasi-judicial body actually determines whether the burden has shifted. If both parties are allowed to present evidence more or less simultaneously, and the body assesses all evidence together, there is no need to talk about a burden of production. In the WBGSS, the first tier-the SDO-contains different steps. The INT will submit a Statement, and the SDO will then assess the evidence. If the SDO finds sufficient evidence-that is, if the INT has reached the required standard of proof at this pointit will issue a Notice to the Respondent. And the Respondent will then be given the opportunity to present evidence in an Explanation. Here, it seems reasonable to talk about a burden of production. The Notice will be issued when the SDO has determined that the burden of production has shifted to the Respondent. After the Explanation has been submitted, the SDO will examine all evidence together and determine whether the party carrying the burden of persuasion-the INT-has reached the required standard of proof.
When we come to the second tier-the SB-the situation is different. The Respondent will submit a Response and the INT a Reply. There may also be a hearing, and after that the SB will make a Decision. But if one looks at the Sanctions Procedures, there is no apparent room for the SB to determine whether the burden of production has shifted. All evidence will be assessed together, and the SB will only determine whether the party carrying the burden of persuasion-the INT-has reached the required standard of proof. 65 This implies that, with regard to the part of the proceeding where it seems relevant to talk about a shift in the burden of production-the SDO-there is no specific regulation of that burden, or the burden of persuasion, in the Sanctions Procedures. And with regard to the part of the proceeding where it seems irrelevant to talk about a shift in the burden of production-the SB-there is a specific regulation referring to such shift.

D. Standard of proof
The standard of proof is the level of persuasion required in the proceeding, with regard to a certain fact. As described above, the parties will usually claim different facts, and both of them have to persuade the judicial or quasi-judicial body to different degrees. 66 The party carrying the burden of persuasion has to persuade the body to a higher 65 In certain cases, however, the reasoning of the SB seems to imply that the evidence is examined in two steps. degree than the other party. In the Second Thornburgh Report, it was noted that several members of the SC were of the opinion that the standard applicable at the time in the WBGSS-that the evidence should be 'reasonably sufficient'-was unclear. 67 Therefore, the report discussed different standards in different types of proceedings. It was observed that it is common to apply the high 'beyond a reasonable doubt' standard in criminal proceedings, and the significantly lower 'preponderance of evidence' standard in civil proceedings. Two arguments in favour of a low standard were presented. 68 The first was that the primary purpose of the WBGSS is to protect the Bank's assets from misuse or abuse in the future, not to punish the Respondent or to compel the Respondent to make restitution. In light of this, the report argued that it would be neither effective nor efficient to adopt a high standard. The second argument was that, to a large extent, the INT has to rely on the Bank's own records and those provided voluntarily by third-parties or the Respondent. It was mentioned in the report that, unlike investigators in domestic legal systems, the INT does not have law enforcement powers to obtain evidence. And unlike parties in certain domestic legal systems, it does not have the right to compel the production of evidence or witness testimony. The report argued that, because of the difficulties the INT may have to produce evidence, it would be unrealistic to apply a high standard. 69 Hence, the report recommended that the 'reasonably sufficient' standard should be replaced with a more descriptive standard, such as 'more likely than not' . 70 This recommendation was adopted, and is now part of the Sanctions Procedures applicable today. The main ground for initiating sanctions proceedings is if the INT, after an investigation, believes that there is 'sufficient evidence' to support a finding of sanctionable practices in connection with a WB-financed project. 71 Moreover, the SDO will only issue a Notice if it determines that the INT's accusations are supported by 'sufficient evidence' . 72 And the INT may withdraw the Notice if it, after having examined the Respondent's Explanation, finds that there is an 'insufficiency of evidence' . The Sanctions Procedures contain the following general definition of the term 'sufficient evidence': ' . . . evidence sufficient to support a reasonable belief, taking into consideration all relevant factors and circumstances, that it is more likely than not that the Respondent has engaged in a Sanctionable Practice.' 73 The term 'sufficient evidence' is the only specific regulation of the standard of proof for the first step of the proceeding-the SDO-in the Sanctions Procedures. However, a general definition of the standard, covering both the SDO and the SB, can be found in the WBG Policy: 'Sanctions are imposed through sanctions proceedings only if the SDO, the relevant EO or the Sanctions Board, as the case may be, after considering the whole of the evidentiary record provided to them, determines that it is more likely than not that the sanctioned party has engaged in, or bears responsibility for, a Sanctionable Practice.' 74 This definition is also in line with the specific regulation of the standard of proof under the SB proceedings that can be found in the Sanctions Procedures: The Sanctions Board shall determine whether the evidence presented by INT, as contested by the Respondent, supports the conclusion that it is more likely than not that the Respondent engaged in a Sanctionable Practice. "More likely than not" means that, upon consideration of all the relevant evidence, a preponderance of the evidence supports a finding that the Respondent engaged in a Sanctionable Practice. 75 Hence, unlike the burden of proof, the standard of proof is described in several different places, not only in the Sanctions Procedures, but also in other documents. As mentioned in the Second Thornburgh Report, a common term for this standard is the 'preponderance of evidence' standard (or the 'balance of probabilities standard'). 76 This is the lowest possible standard for a party carrying the burden of persuasion.
The arguments presented in the Second Thornburgh Report, as to why this particular standard should be adopted, have to be evaluated in light of the general reasons for allocating the burden to a certain party and for applying a certain standard. It seems that the most important reason is the desire to avoid erroneous outcomes. 77 An erroneous outcome is a ruling, which the judicial or quasi-judicial body would not have made if it would have had full knowledge about the concrete legal facts in the case. In our example, there are basically two types of possible erroneous outcomes. The body may find that Party B has violated Rule X, even though it has not, or that Party B has not violated the rule, even though it has. The first type may be referred to as a Type I error, and the second as a Type II error. 78 If the burden is allocated to Party A (the claimant), the risk of Type II errors will be higher than the risk of Type I errors. The higher the standard, the higher the risk. 79 And if the burden is allocated to Party B (the respondent), the risk of Type I errors will be higher than the risk of Type II errors. The higher the standard, the higher the risk. Hence, a higher standard than the preponderance of evidence standard should basically only be applied if one type of error is considered more problematic than the other. 80 A classic example is the 'beyond reasonable doubt' standard often applied in criminal proceedings. 81 In many legal systems the prosecutor will have to reach this very high standard since it is considered more desirable to avoid Type I errors (to convict an innocent person) than to avoid Type II errors (to acquit a guilty person). If one type of error is not considered more problematic than the other, it usually makes sense to apply the preponderance of evidence standard, since this is a way of reducing erroneous outcomes in general. 82 The judicial or quasi-judicial body will simply rule in favour of the most probable claim.
The first argument in the Second Thornburgh Report was that the primary purpose of the WBGSS is to protect the Bank's assets from misuse or abuse in the future, not to punish the Respondent or to compel the Respondent to make restitution. This argument has to do with what could be referred to as the 'norm goal' . It is often possible to see that one type of error affects the goal of the regulatory framework-in case of the WBGSS to protect the Bank's assets-more negatively than the other. An argument related to the norm goal implies that the burden of proof rules shall be designed so that the risk of this type of error is reduced. In the WBGSS, one could argue that Type II errors affect the norm goal more negatively than Type I errors, since they entail that actors who have been engaged in sanctionable practices are not sanctioned. Hence, if the norm goal is considered more important than other aspects, it may make sense to allocate the burden of persuasion to the Respondent. It may also make sense to adopt a high standard. The Second Thornburgh Report did not go this far. Instead, it used the argument to recommend a low standard for the claimant. A reason for this could be another argument presented in the beginning of the report, pointing in another direction. 83 It was observed that, often, projects funded by the WB are largescale, or specialized in nature. Thus, there may be only a few actors with the necessary capabilities. Therefore, it could be in the economic interest of the Bank to avoid results where the Respondent is erroneously found guilty of a sanctionable practice (a Type I error). This is an argument related to the norm goal, but which, unlike the argument above, can be used in favour of allocating the burden to the claimant.
However, as part of the first argument it was also mentioned that the primary purpose of the system is not to punish. Even though the purpose is not to punish, a sanction could still be seen as a punishment, in the sense that it has negative consequences for the Respondent. A sanction usually also implies that the Respondent has been found guilty of a conduct that is considered criminal in many domestic legal systems. 84 Thus, the decision of the SDO and/or the SB may have other negative consequences than the sanction. Here, we find another category of arguments, namely arguments related to the 'risk aspect' . When taking into account such arguments, the relevant question is not whether the purpose of the system is to punish, but rather how serious a Type I error should be considered for the Respondent. The negative consequences for the Respondent could have been used as arguments in favour of adopting a higher standard than the preponderance of evidence standard for the INT.
The second argument in the report is also a type of argument relevant to use when designing burden of proof rules, and it could even have been used to allocate the burden of persuasion to the Respondent. The argument has to do with the 'evidentiary possibility' . If it is significantly easier for one of the parties in the proceeding to produce evidence, it may be reasonable to allocate the burden to this party, and sometimes also to adopt a high standard. The idea is that, if the judicial or quasi-judicial body wants to know as much as possible about the facts in the case, it shall force the party with the best evidentiary possibility to produce evidence. This way, the general risk of erroneous outcomes will be reduced. But in order to know whether the argument makes sense, one has to look both at the general level and the situation in specific cases. While the INT may have certain general difficulties to produce evidence, compared to investigators in domestic legal systems, the evidentiary possibility of the Respondent may differ depending on who the Respondent is. In cases with a 'weak' Respondent, the evidentiary possibility could be an argument not only in favour of allocating the burden to the INT, but also in favour of adopting a higher standard.

III. CONCLUDING REMARKS
A large and increasing number of international judicial and quasi-judicial bodies have made the field of international procedural law an important part of public international law. The aim of the present article has been to contribute to a better understanding of how procedural rules of a certain type-provisions related to facts, evidence, and the burden of proof-have been designed in the WBGSS. The main conclusion is that such rules play a central role in the system, and that considerable efforts have been made during the last two decades to develop a well-functioning body of procedural provisions. The WBG has come a far way. The Sanctions Procedures contain detailed provisions, explaining the mainly adversarial character of the system, where the INT and the Respondent carry the main responsibility for presenting and proving facts. Issues related to evidence, and the burden and standard of proof, are regulated, and have been discussed in different types of background documents over the years.
However, the article has also shown that, in certain respects, there is still room for improvement. One part of this is how the procedural provisions, and the substantive provisions containing the relevant abstract legal facts, are structured. The abstract legal facts-what needs to be proved in a certain case-can be found in the definitions of the sanctionable practices. The WBG Policy refers to the definitions in the legal agreements governing specific projects, and such agreements usually incorporate definitions from other WBG documents, primarily the different guidelines. Moreover, the Sanctions Procedures refer to these other documents. Hence, it is not entirely easy to follow the references, and to find the relevant abstract legal facts. To collect a uniform set of definitions in one place, and to make sure that the central procedural document-the Sanctions Procedures-contains a clear reference to this place, would be an improvement.
Furthermore, the provisions related to the burden and standard of proof are rather straightforward, in the sense that it is clear which party shall carry the burden, and which standard shall be applied. But there is only an explicit regulation of the burden in respect of the second step of the procedure (the SB), not the first step (the SDO). And regulations of the standard can be found both in the WBG Policy, and in several places in the Sanctions Procedures. A clearer structure would be to collect the provisions concerning the burden and standard of proof in one place in the Sanctions Procedures, and explicitly state that they are applicable for both steps of the procedure.
When it comes to the content of the burden and standard of proof provisions, two remarks can be made. The first has to do with the fact that the Sanctions Procedures state that, when the INT has reached the standard, the burden 'shifts' to the Respondent. This is a provision that could be difficult to understand and apply. The established view is that the burden-the 'burden of persuasion'-does not shift during the proceeding. It seems that, what the provision in the Sanctions Procedures refers to is rather the 'burden of production' . When the INT has produced evidence and reached the standard, the Respondent has to do something in order not to lose the case. Evidence pointing in the other direction has to be produced; the burden of production has shifted. However, it is only relevant to talk about a burden of production if there is a point in the proceeding when the judicial or quasi-judicial body will actually determine whether the burden has shifted. This seems to be the case during the first step of the proceeding (the SDO), but not the second (the SB). Hence, if the system is reformed in the future, it would probably be a good idea to carefully consider the use of the word 'shift' . As the system looks now, it could be reasonable to use it for the first step of the proceeding, but not the second.
The second and final remark has to do with the reasons for allocating the burden of persuasion to a certain party, and for applying a certain standard. Today, the WBGSS contains the burden of proof principle actori incumbit probatio, and the preponderance of evidence standard. Different reasons for this have been presented in background documents, primarily the Second Thornburgh Report. An attempt in the present article has been to place these reasons in a more general framework-using aspects such as the 'norm goal' , the 'risk aspect' , and the 'evidentiary possibility'-to discuss why it can make sense to allocate the burden in a certain way or to adopt a certain standard. The conclusion here is that, to a large extent, the reasons presented in the background documents support the approach taken in the system. But there are also arguments for allocating the burden in a different way and for applying a different standard. Moreover, it should be noted that there is nothing saying that a legal system has to adopt the same burden of proof principle, and the same standard, for all types of cases and all abstract legal facts. Hence, in future reforms of the system, it could be useful to further examine the above type of arguments, and to explore whether it would make sense to adopt a more differentiated approach.