FL-2022-000014 - [2025] EWHC 2631 (Ch)

FINDINGS ON CONTESTED FACTUAL AND EXPERT ISSUES

In this section I shall resolve some contested factual and expert issues. It helps to do this at this stage, as some of them are relevant to more than one element of the claims. I shall address them under a number of headings:

Much of the relevant history is addressed in Section B above but I make some further findings here.

From at least July 2020 BaFin required the Greensill Group to reduce its exposure to GFG. Representatives of SBIA were aware of these requirements, as shown in internal emails from August and October 2020.

On 8 December 2020 BaFin required an acceleration of the reduction in exposure to the GFG group in a manner that was not considered by GCPL to be sustainable. Mr Greensill recognised that this would cause significant disruption.

By 18 December 2020 the Greensill Group was unable to settle an intra-day trade concerning notes in the Fairymead Note Programme. Mr Greensill accepted that the group did not have liquidity to settle this trade.

GCPL disclosed the BaFin requirements to the prospective investors in its capital raising programme. On 23 December 2020 TDR and other planned investors told Mr Greensill that they were no longer willing to commit funds to GCPL, at least on the then proposed terms.

Mr Greensill immediately contacted SoftBank to discuss whether SoftBank would be prepared to provide a bridge facility of $1.5 billion. Mr Greensill spoke to Mr Misra.

The minutes for the board meeting on 31 December 2020 stated that:

“LG has been informed his morning that Softbank was not prepared to provide a bridge facility. It was, however, prepared to consider making a further equity investment of up to $250m provided it was alongside investments from other parties”.

During his evidence Mr Greensill confirmed that this accurately recorded what had happened. However, his evidence was not entirely consistent. Mr Greensill had said in the LG transcript that he had many calls with SoftBank in the period up to 29 December 2020 when he learnt that SoftBank would not be able to advance the loan and that this had resulted in him asking A&O for recommendations for insolvency practitioners.

For his part, Mr Cheung described in his evidence a call from Mr Greensill on Christmas Day 2020 about the BaFin requirements and the position of TDR. Mr Cheung understood at that stage that Mr Misra was thinking about making a loan. Both Mr Cheung and Mr Greensill said in evidence that the funding had to be there by 31 December 2020. Mr Greensill also spoke to Mr Misra on Christmas Day concerning the $1.5bn bridge. Mr Cheung said that he understood that the loan was not something Mr Misra supported.

On 26 December 2020 Mr Cheung sent an email to Mr Misra saying that he had delivered the message that this was not something SoftBank supported. He gave evidence that he did not understand the position to change after the Boxing Day call and that he did not ever communicate that they would provide the $1.5 billion loan. Mr Cheung also said that Mr Misra was considering something smaller.

There was another message from Mr Greensill to Mr Misra on 26 December asking for an urgent call “as the alternate next steps have consequences that you and I need to discuss”. At one point in his evidence Mr Greensill accepted that it seemed that SoftBank had said by this point that it would not provide the loan he had sought. Mr Greensill also asked to speak to Mr Son on 26 December.

Mr Greensill also gave evidence, however, that nothing had been settled as he was still reaching out to try to contact Mr Misra and he also said that he thought he probably spoke to Mr Misra on the morning of 31 December 2020 when he was told that SoftBank was not going to proceed with the bridge. He said at another point in his evidence that it is likely that the minutes of 31 December 2020 which recorded that SoftBank had that morning refused the bridge loan were accurate.

On the other hand, at other stages in his evidence he said that he did not know the date on which Mr Misra communicated the negative decision and said that it was a blur. He also said at one point that it was Mr Cheung who had delivered the news, rather than Mr Misra. He also reiterated that he had formed the view on his birthday of 29 December 2020 that it was time to call in insolvency advisers. He called it the worst birthday of his life.

It is not surprising that Mr Greensill was unable to give entirely consistent or clear evidence about these matters. Taking the evidence as a whole, I find as follows. Mr Greensill sought a bridging facility of $1.5 billion from SoftBank on 23 December 2020. By 26 December 2020 he had had at best a lukewarm response, albeit it seems probable that he thought the door was not completely shut and that there was the possibility of a different amount being offered. Mr Greensill continued to seek to contact Mr Misra and Mr Son until the end of the month, which suggests that he still hoped to arrange a substantial bridging loan. By 29 December 2020 no such loan had been arranged and Mr Greensill and the other directors decided that they should seek the advice of insolvency practitioners. That was consistent with Mr Greensill continuing to seek to persuade SoftBank to advance a bridging loan. I conclude that is more likely than not that he was only given a firm refusal on the morning of 31 December 2020, as recorded in the board minutes of GCPL. That is the best evidence of the timing of the refusal and it was not suggested by the claimants in their cross-examination of Mr Greensill that he had lied to the Board of GCPL in making that statement.

The SBDs contended that there was an agreement or understanding between the Vision Funds and Mr Greensill on behalf of the Greensill companies that the $440 million injected into GCPL under the CLNs would be used to redeem or purchase the Fairymead Notes.

There was debate at the trial as to whether the SBDs’ case was properly pleaded or whether it had shifted. The claimants contended specifically that the SBDs had changed their case as to whether the Greensill companies had to use the same funds (i.e. the very $440 million advanced on 10 November 2020) for this purpose; about the natural persons said to have made this agreement or arrangement; and about the timing and terms of any obligation on the Greensill companies (or understanding) to redeem or acquire the Notes.

In my judgment the case advanced by the SBDs is sufficiently clearly pleaded. The pleaded case is the amount of $440 million provided under the CLN and paid to GCUK on 10 November would provide the Greensill companies with liquidity to repurchase or repay the Fairymead Notes. This was said to be agreed between Mr Romeih and Mr Cheung on the one side and Mr Greensill on the other side. The SBDs confirmed during their opening oral submissions that they were not alleging that the Greensill companies had to use the self-same $440 million for the alleged purpose; and that they did not make any allegation about the precise date by which, or means by which, the repurchase or redemption had to occur. They did however maintain that Mr Greensill had agreed that the $440 million would be used for the purpose of acquiring or redeeming the Fairymead Notes. They also alleged that the SBDs understood that GL could not lawfully have released the obligations under the RPA without securing the repayment of redemption of the Fairymead Notes. They contended that this was understood by the SBD representatives. They did not however advance a positive case that Mr Greensill shared this understanding. I reject the claimants’ argument that the case was not sufficiently pleaded.

I shall address in due course a legal issue of what constitutes the relevant “transaction” for the purposes of the claim under section 423. However at this stage I am concerned with the factual issue of what was understood by the SBDs on the one side and Mr Greensill on the other; and whether there was an agreement or arrangement about this, in the sense of a common understanding. In the following passages, for ease of reading, I shall sometimes refer to the various groups of companies as “Katerra”, “Greensill” and “SoftBank” when there is no need to distinguish between the various entities constituting those groups.

A helpful starting point for approaching the factual findings about these issues is to determine the catalyst for the $440 million injection. The evidence of Mr Cheung was in essence that the catalyst for the various negotiations which led to the $440 million investment was a call from Mr Greensill in October 2020. As the chronology set out in section B shows, by that date the Greensill Group was engaged in seeking to arrange pre-IPO funding, based on a much higher valuation than the valuation at which SVF1 had made its investments. Mr Cheung said in evidence that Mr Greensill explained to him that the financial problems facing Katerra put the pre-IPO fundraising at risk. If Katerra went into bankruptcy, the default would jeopardise the investment of the Credit Suisse SCF funds, which represented Greensill’s main source of liquidity. That would put the entire fundraising exercise at risk. Mr Greensill therefore wished to find a way to remove the Katerra default risk. In his evidence Mr Cheung described this as a bomb that needed to be defused.

The claimants, by contrast, contended in broad terms that the catalyst for the negotiations which led to the agreements of November 2020 (including the CLNs) was quite different. They argued that the documents showed that Mr Greensill had been contending for many months that Mr Son had orally agreed that SBG would provide credit support in respect of the Katerra funding, as part of the CEP (as summarised in section B above). The Greensill companies contended in these emails that they had relied on Mr Son’s representations that SBG would provide such credit support and that they were now obliged to provide it. The claimants said that it was essentially this dispute that led to the November 2020 agreements, including the Omnibus Deed. They noted in support of this contention that the Omnibus Deed expressly released SBG from any liabilities relating to credit support in relation to Katerra.

More specifically, the claimants submitted that the discussions that had taken place in December 2019 about the CEP concerning the Katerra Group were key to understanding the events of Autumn 2020. Several documents dated between 12-18 December 2019 referred to SBG’s and the Greensill Group’s efforts to set up the joint venture and provide a guarantee in advance of Katerra being funded. Mr Greensill sent an internal email recording his understanding that Mr Son had given him his personal commitment to issue the guarantee. Mr Cheung accepted in evidence that Mr Greensill had said that he believed that Mr Son had given this commitment. This was reflected in the arrangements between Katerra and GL: it was indeed a condition precedent for the requirement on GL to provide funding under the RPA that SBG or an affiliated company would provide credit support to GL.

There were further discussions between SBG and the Greensill Group during 2020 about formalising the CEP joint venture. Mr Greensill gave evidence that this dragged on and SBG stopped engaging with him about it.

Mr Cheung said in evidence - and I find - that there was a difference of view between the Greensill Group and SBG as to whether SBG had committed to providing credit support in respect of the Katerra facility. Mr Greensill contended that Mr Son had made a firm commitment. SBG contended that there was no binding agreement about the CEP.

Mr Greensill accepted in evidence that there was no written agreement and explained in his evidence that the Greensill Group was naturally advancing the strongest position it could in the continuing negotiations over the CEP.

In an email of 14 September 2020 to Mr Misra, Mr Greensill said that Greensill had been funding Katerra on the understanding that it was part of the CEP, and he listed a number of documents which he said demonstrated this. Mr Greensill gave evidence that it seemed reasonable to conclude that he spoke to Mr Misra in September 2020, at about the time of this email, and that Mr Misra told Mr Greensill about SBD’s awareness of Katerra’s problems.

Mr Greensill also explained that he was anxious about Katerra’s financial position in October, particularly given the unresolved issue of the CEP.

On 30 September 2022 Mr Greensill sent a message to Mr Cheung saying, “We could ramp up the Katerra Facility and SBG could buy the increased amount – very quick and elegant”. Mr Greensill accepted in evidence that by this time he was aware of financial issues within Katerra and that he was starting to consider proposals for a solution. Mr Cheung agreed that by this stage he and Mr Greensill had probably been discussing Katerra.

On 2 October 2020 SBG informed Greensill that it was terminating the process for establishing the JV for the CEP. This was naturally treated as very unwelcome news within the Greensill Group. Mr Greensill emailed Mr Son on 11 October 2020 in terms similar to his email of 14 September 2020, but with a section called “Legal Position Summarised”.

Separately, from September 2020, the Greensill group was engaged in the pre-IPO fundraising process managed by Credit Suisse. The documents show that the Credit Suisse investment banking department was hoping to make $40-50 million fees from the fundraising over the next 12-18 months.

As already explained in section B above, Katerra had experienced financial problems throughout 2020. It had breached various covenants in the RPA, requiring standstill agreements on 30 March 2020 and 1 June 2020 (and again on 12 November 2020). Mr Greensill accepted in his evidence that he knew about this. As also explained above, in May 2020 Katerra replaced Mr Marks with Mr Kibsgaard as its CEO.

In early September 2020 Katerra’s financial position worsened further and it requested further forbearance from Greensill or an amendment to the covenants. SBIA was kept updated. Mr Greensill accepted in evidence that he was considering making proposals to SoftBank to deal with the risk. He accepted that by October 2020 he was becoming anxious about Katerra’s financial position. He was also in communication with Mr Marks, who remained a shareholder in Katerra, about a potential bankruptcy and solutions to avoid it. He indeed told Mr Marks on 9 October 2020 that he was speaking to Mr Misra about this topic.

Mr Cheung said in evidence (and I find) that in early October 2020 he had a series of telephone calls from Mr Greensill and Mr Misra concerning the fact that Katerra was having financial problems and might enter into bankruptcy. Mr Greensill said in those discussions that if Katerra went bankrupt Greensill would be significantly exposed.

Mr Greensill said in evidence that he thought at the time that if Credit Suisse suffered a loss on the Fairymead Notes and gated the SCF funds it would probably be a company-ending event for Greensill, since the SCF funds were the principal source of liquidity. He gave evidence that the SCF funds had provided over $10 billion of funding. I find that Mr Greensill thought at that time that a default on the Fairymead Notes would have been existential for Greensill.

SBIA shared this view. An internal presentation of 2 November 2020 referring to the “Greensill/Katerra Transaction” explained that Katerra required recapitalisation in order to preserve going concern status. The recapitalisation comprised two aspects: (a) a new equity injection of $380 million, with SVF’s share being $200 million and new investors contributing $180 million; and (b) “new equity” being used to restructure the $440 million Katerra indebtedness at “40% of par”. The document also referred to a loss on the Katerra debt as having significant adverse consequences for SVF1’s investment in Greensill, including that it would “derail” Greensill’s current funding round, which was expected to be 68% up in value on the last round. It would also adversely impact “Greensill’s institutional funding access going forward - Katerra exposure held within a CS Fund”. The link was therefore made to the liquidity being provided by the SCF funds more generally. The document also explained a proposal under which SVF2 would purchase 100% of the economic interest in the $440 million Katerra debt. It also stated that the recapitalisation of Katerra needed to be agreed within the next 24-48 hours and that if there was no agreement the Katerra Board would have to file for bankruptcy. It said that otherwise the “knock-on impact on Greensill, particularly on its funding sources, will be very significant.”

Mr Cheung gave evidence that the threat to Greensill’s funding process from a Katerra bankruptcy set off a fire alarm. In his view this was a bomb that was going to detonate. He explained that SVF1 would stand to benefit very significantly from the proposed funding round (which would place a much higher valuation on Greensill) and that this would be lost if Katerra went down. I accept his evidence in this regard.

Mr Romeih gave similar evidence (which I accept) about a conversation he had with Mr Greensill in about October 2020 in which Mr Greensill said that he needed assistance in maintaining the pre-IPO process.

On 11 October 2020 Mr Greensill asked SBG to solve the problem posed by a potential Katerra default by purchasing the Fairymead Notes. A presentation was prepared within Greensill, and it was pitched by Mr Greensill in a call with Mr Son on 11 October 2020.

One of the slides in that presentation summarised a proposal under which SBG would purchase the Katerra Notes for $440 million. It also suggested that in the event that there was no transaction, SVF1 stood to lose $1.45 billion as a 50% write down on its equity investments in Greensill. Mr Greensill explained in evidence (and I find) that the presentation reflected his belief that a failure to perform would be company-ending for Greensill.

The negotiations with the SBDs ensued after this presentation. Some of the communications have been referred to in section B above. Mr Greensill gave evidence that SoftBank “strong-armed” him or that it was holding a gun to his head. He also explained (and I find) that in saying this he was referring to the terms of the proposed transaction, including SoftBank’s insistence on the provision of a personal guarantee and the sale of private aircraft.

The claimants contended that Mr Greensill’s evidence that he was strong-armed showed that the SBDs procured and directed the Greensill companies to enter the various transactions. I reject that conclusion. I find that it was Mr Greensill who sought to persuade SBDs to inject the $440 million to prevent a potentially disastrous default under the RPA, and that he did this in order to salvage the proposed Greensill fund-raising. This conclusion is supported by Mr Greensill’s evidence that the SBDs were ultimately persuaded to introduce the $440 million of new equity, and that it was the terms and conditions that were imposed by SoftBank as the price for this investment that Mr Greensill found to be painful.

I am unable to accept Mr Greensill’s evidence that SoftBank unexpectedly rang him on a Sunday evening at around this time and presented him with terms he was required to take or leave. The documents show that the initial idea of SBG buying the Notes for $440 million came from Mr Greensill. The documents also show that there were then further negotiations about the structure of the deal; and that its terms evolved over time. There was no take-it-or-leave-it moment. Indeed the structure of the deal was partly developed through a process of negotiation between the external lawyers. I accept the evidence of Mr Romeih that while SBIA helped to co-ordinate the November agreements, they did not have the ability to impose them or direct Greensill to do anything. The agreements were the result of commercial negotiations.

As already mentioned, the claimants contended that the essential catalyst for the negotiations which led to the agreements of November 2020 (including the investment of $440 million for the CLNs) was Mr Greensill’s contention that Mr Son had orally agreed that SBG would provide credit support in respect of the Katerra funding, as part of the CEP. It is correct that the negotiating documents and the final forms of the agreements entered into in November 2020 show that the settlement of the CEP dispute was one of the components of the November transactions. The dispute concerning the CEP clearly played a part. Numerous emails in the period leading up to the Omnibus Deed refer to the dispute about the CEP, and the Omnibus Deed addressed aspects of the CEP in terms. Moreover, it was natural that the Greensill side should emphasise its position that there was a binding commitment about the CEP concerning Katerra. This was a way of seeking to negotiate the best terms.

On the other hand, the troubling financial circumstances played a more important part. Katerra appeared to be to close to bankruptcy and its collapse would imperil the pre-IPO funding. It would also probably lead to the general withdrawal of liquidity from the Credit Suisse supply credit finance funds, which would threaten Greensill’s very existence. I accept the evidence of Mr Cheung that from the SoftBank perspective these latter concerns created an urgent need to find a solution to the Greensill exposure to Katerra.

This sense of urgency about the GCPL fundraising exercise is shown, for instance, in a message from Mr Greensill to Ms Chan on 6 November 2020 where he said that Mr Son had “to tell people to get it [the finalisation of the agreements] done today. Else it puts our capital raise at risk”. Mr Greensill continued to press for the deal to be done as urgently as possible.

For these reasons, I accept the evidence of Mr Cheung that while the dispute about the CEP played a part, it was the threat to the fundraising that was the spur for the Greensill’s proposal that SBG should inject $440 million for the purpose of acquiring the Fairymead Notes. As noted, I accept his evidence that it was the potentially disastrous consequences of a Katerra default that principally generated the need for a transaction. I also note that the parties saw the need for a very urgent solution because of the immediate risk of a Katerra bankruptcy if the company was not restructured, and the knock-on impact of that on Greensill’s fundraising efforts. I also accept the evidence of Mr Romeih that the CEP was not mentioned in his conversations with Mr Greensill about the potential Katerra default.

I therefore find that, as part of the negotiations, the parties did compromise their continuing dispute about the CEP, but that the negotiations were primarily driven by the pressing need to rescue Katerra from collapse and to salvage the Greensill fundraising; and that this required Katerra’s indebtedness under the RPA to be addressed.

The documents set out in Section B show how the negotiations developed from the initial proposal for SBG (and then SVF1 or SVF2) to acquire the Fairymead Notes for $440 million into a transaction under which the SBDs would inject $440 million into GCPL under the CLNs.

As they developed in the period until 10 November 2020, the proposed transactions had two main elements: first, the restructuring of Katerra’s balance sheet by injecting equity and removing or reducing the indebtedness under the RPA and, second, by injecting the sum of $440 million to ensure that the SCF Subfund would not suffer any impairment in respect of the RPA. These dual purposes of the November transactions were explained in the oral evidence of Mr Cheung, Mr Romeih and Mr Greensill.

I find that, though the structure of the transactions changed, SBIA and the SBDs continued to understand that the sum of $440 million to be injected under the CLNs by Greensill would be used to repurchase the outstanding Notes from the SCF Subfund and that this would shift the liabilities under the Notes to Greensill’s balance sheet. The parties referred to this process in a number of communications in terms of Greensill “assuming” or “internalising” the risk of the indebtedness. Examples are given above at paragraphs [‎103] (2 November 2020) and [106‎] and [‎113] (both 5 November 2020). In an email of 27 October 2020 Mr Cheung explained to Mr Misra that Mr Greensill had “crafted a CLN structure with embedded Katerra note risk and Greensill equity”. In another email of 2 November 2020 Mr Greensill referred to the agreement they had discussed which included a subscription by SVF2 of $440 million for CLNs and then said “Greensill assumes all risk on the Katerra Notes (current notional $440 million) and manage their recoveries. All recoveries will be remitted to accounting for the same to SVF(2)”.

As to the Katerra side of the transaction, the letter of intent dated 2 November 2020 from the New Money Consortium (para [110] above) contained an express condition precent to the investment of $180 million that the indebtedness under the RPA would be compromised for no more than $176 million. The evidence suggested that Mr Greensill did not see the letter itself. However I find that he was kept updated on the substance of this aspect of the proposals. Ms Chan’s notes refer to keeping him updated and when the New Money Consortium dropped out Mr Romeih called to give him a heads up.

As explained in section B above, the lawyers were sent the commercial terms and they then negotiated the documents. Mr Funder’s email of 4 November 2020 is set out at para [120] above. He said that,

“SVF is putting Greensill in funds to the amount of $440m and Greensill will be able to use this to fund the buy back of the CS notes. That is why Greensill has the obligation to remit any funds recovered.”

This email supports the conclusion that the parties understood that one of the Greensill companies would use the $440 million injection to buy the Fairymead Notes and take the risk onto its own balance sheet.

The way that the $176 million payment by Katerra to Greensill was to be funded was explained in an SBIA Risk Dept Summary dated 4 November 2020. This stated:

“SVF are to contribute $200m of a $380m round, with ~$176m in proceeds funding an anticipated negotiated paydown of the $440m Greensill facility (upon which the facility shall be considered fully paid/discharged). Other debt obligations are to be similarly restructured.”

As noted in section B, the transactions in their then form were approved in principle by the relevant SBDs on 5 November 2020. The Katerra investment of $200 million was approved by the IC for SVF1 that day on the basis of an investment memorandum. The Greensill $440 million CLN was approved by the IC for SVF2 the same day on the basis of a second memo. SBG, as the sole LP in SVF2, was also presented with a memorandum of the same date. The latter document stated:

“By purchasing the Katerra Notes from the CS Fund, Greensill will be able to internally manage (a) all risks relating to the Katerra Facility including repayment and default risks and (b) recoveries from the Katerra Facility. All recoveries will be remitted to SVF2.”

Each of the three IC memoranda stated under the heading “Use of Proceeds” that:

“It is anticipated that proceeds will be used to purchase the Katerra notes from external investors and manage the risk internally”.

Mr Romeih explained that these documents would have been provided to the lawyers to draft the transaction documents.

There is further contemporaneous evidence that the SBDs understood that the proceeds of the CLNs would be applied to acquire the Fairymead Notes. On 19 November 2020 the interlinked elements of the transactions were summarised by an SBIA partner for the benefit of SBG’s CFO. Slide 10 referred to the current balance on the RPA as $440 million and stated that Greensill agreed to a 40% payment of $176 million. Next to a box representing Greensill the text stated:

“Use $440m to re-purchase notes (at par) currently held by external investors with exposure to Katerra facility (Greensill will take on full risk)”

I find that the SBDs and SBIA used the language of Greensill “managing the risks” internally or “taking on the risk” (and like phrases) as a way of referring to Greensill buying or redeeming the Fairymead Notes – thereby removing that risk from the Noteholders. I find, based on the documents just mentioned, that the SBDs and SBIA believed when they approved the November 2020 agreements that the $440 million would be used to buy or redeem the Notes.

As explained in section B, various documents were executed on 10 November 2020 including the CLNs and the Omnibus Deed. The CLN payment of $440 million was made that day.

I find that the SBDs and SBIA’s understanding that one of the Greensill group companies would buy back the Fairymead Notes using the $440 million was also reflected in the terms of the recitals and clause 3 of the Omnibus Deed (see paras [131] and [137] above). The reference in the recital to Greensill assuming the losses reflects the parties’ understanding that Greensill would take the Notes onto its own balance sheet. (As already mentioned, similar language, of “assuming” losses or risks, appears in a number of the internal SBIA documents in which the transactions were explained.) Clause 3 of the Omnibus Deed indeed presupposes that a company in the Greensill Group would become the owner of the Notes: it would not have been possible for a Greensill company to account to SVF2 for recoveries under the Katerra Programme (i.e. the RPA) if the Notes remained with the Noteholders.

I find that the SBDs and SBIA understood the existing securitisation structure, and that under that structure, GL could not have extracted the proceeds of the RPA. The end date of the obligation under clause 3 was 31 December 2020.

As explained in section B above, on 18 November 2020 the New Money Consortium dropped out. I find that this was an unwelcome development for the SBDs. It required amendments to be made to the deal. On 23 November 2020 Mr Greensill checked with Mr Romeih about whether Greensill could liaise with Katerra about the writing off of the RPA (see para [149] above). Mr Romeih explained that further authorisation would be required before this could happen. The claimants submitted that this showed that the SBDs were procuring Mr Greensill to do their bidding, and that the SBDs unilaterally imposed the transactions on Greensill. I do not accept that submission. Rather, the change was unwelcome for the SBDs as they would potentially have to inject more cash to effect the recapitalisation. Mr Romeih was simply saying that the write-off of the RPA could not occur until the funding gap had been resolved.

The funding gap was, as explained above, ultimately filled by SVF2 agreeing to forgo the recovery of $176 million, in return for equity in Katerra. The terms of the CLNs were also amended, so that the stake in GCLP would be increased. Mr Cheung explained, and I accept, that it was important for SVF2 to receive consideration for the additional $176 million.

The revised commercial deal was agreed between the parties by 29 November 2020 (see the emails set out at [155] to [156] above). An updated SBIA memo for SVF2 explained that the use of the proceeds remained “to purchase the Katerra notes from external investors and manage the risk internally”. This is consistent with the earlier memos concerning the original structure. This was also referred to in the minutes of an Investment Committee meeting on 3 December 2020.

The proposed investment by SVF2 was approved by Mr Romeih and Mr Misra on 30 November 2020 with Mr Son recusing himself. The SVF1 investment committee did not vote on the revised proposal as this was not needed.

After the email exchange of 29 November 2020 confirming the commercial terms of the revised deal, the lawyers for the parties again got down to work on negotiating the terms of the legal agreements. Mr Cheung and Mr Romeih accepted that they probably did not read the terms of the formal agreements.

On 1 December 2020 Mr Funder of A&O emailed Mr Grubb-Sharma of MoFo to communicate that there had been some changes to the commercial terms agreed in the Omnibus Deed, and attached a draft amended and restated Omnibus Deed.

On 4 December 2020 the Board of GCPL approved the entry of the Amended Omnibus Deed and the Amended CLN and any other documents completed by or ancillary to those documents. There were no material changes to the terms of these agreements between 4 December 2020 and their execution on 23 December 2020.

On 9 December 2020 a draft of the CEA was sent by Katerra’s lawyers, K&E, to the other lawyers acting on the transaction.

On 11 December 2020 Greensill provided the signature pages of Mr Greensill, GCUK and GCPL to the Amended Omnibus Deed to be held in escrow. These were to be held by A&O to be released once SBIA had given its consent to release the Vision Funds’ signature pages.

On 12 December 2020 Katerra, Greensill, SBIA, and Wolff concluded the agreement of indicative terms as to the settlement between Katerra and Wolff. One of the conditions precedent was the extinguishment of the RPA and the investment of $200 million by SVF1.

On 22 December 2020 HSR approval was given for the investment by SVF1 into Katerra.

On 23 December 2020 SBIA provided its consent to the release of its signature pages on the Amended Omnibus Deed and the Amended CLN.

On 24 December 2020 Greensill sent K&E the signature pages of the CEA and the TA, signed by Mr Lane, to be held in escrow.

On 29 December 2020 Greensill confirmed that its signature pages could be released on simultaneous release of the counterparties’ signatures.

On 30 December 2020 a “Katerra closing call” took place. After this Weil sent K&E and others a pack of documents, which included executed versions of the PSPA and the TA. Later the same day K&E sent Greensill an update stating that the transaction had successfully closed that morning and attaching a copy of the executed CEA.

There are several indicators in the evidence that the SBDs and Mr Greensill believed that the injection of $440 million was part of a broader commercial transaction. First, the SBD documents consistently referred to “the Greensill/Katerra Transaction” and the internal SBIA documents set out the various elements of the transactions together.

Second, Mr Greensill repeatedly referred in his evidence to the overall or broader transaction or overall deal.

Third, both Mr Romeih and Mr Cheung gave evidence that closing out the RPA was part of an overall transaction. Mr Cheung explained that this was reflected in clause 3 of the Omnibus Deed.

In my judgment, notwithstanding the changes to the commercial deal after the withdrawal of the New Money Consortium, the SBDs continued to believe that the Greensill companies would use the $440 million injection to buy the Fairymead Notes. Mr Cheung, in particular, understood the securitisation structure and believed throughout that the Greensill companies would have to take control of the Notes before GL would be in a position to compromise the RPA. Mr Cheung gave evidence, which I accept, that the $440 million was going to go into Greensill for the Note repurchase. Mr Romeih said (and I accept) that he never thought that the capital they put into Greensill (the $440 million) would be used for any purpose other than buying the Notes, since the whole point of the arrangement was to enable Mr Greensill to resolve the Katerra bankruptcy risk.

I also accept Mr Cheung’s evidence that he and others at SBIA understood that the $440 million investment on 10 November 2020 would not have occurred unless they had understood that the Greensill group required the funds in order to buy the Fairymead Notes and thereby internalise the risks under the RPA.

Mr Cheung explained in evidence (and I accept) that the commercial decision to inject the funds was taken in order to protect SVF1’s existing investment in Greensill against the damage that would be inflicted on the pre-IPO fundraising by a bankruptcy of Katerra and the consequent termination of Greensill’s main source of liquidity.

Mr Cheung accepted that the SBDs did not inject the $440 million from a sense of duty to the Noteholders. But that evidence is consistent with his evidence that the catalyst for $440 million investment was the need to protect Greensill’s fundraising - and that the SBDs would not have made that investment unless they had understood that Greensill would acquire the Notes.

As to this, Mr Cheung gave evidence that he understood that Mr Greensill would take the necessary steps to execute the repurchase of the Notes and proceed with the fundraising while ensuring that Greensill complied with its obligations to SoftBank, Credit Suisse and Katerra. The obligations to Credit Suisse included the obligation not to collapse the securitisation while the Notes remained outstanding. He said that in the discussions that took place during December 2020 he did not contemplate whether the Notes had actually been repurchased. Mr Cheung also said (in paragraph 93 of his witness statement) that he believed that repurchasing the Notes was a necessary condition precedent to GL forgiving the Katerra lending line. I accept this evidence. I find that Mr Cheung believed that the repurchase of the Notes had to occur before GL could forgive the RPA debt.

For his part, Mr Greensill gave evidence that it was the understanding of all the parties that the $440 million would be used at some stage to repay the Noteholders. He said that this was the intention from the very beginning.

I find that Mr Greensill communicated to the SBG and SBIA that he intended to use the $440 million to redeem or purchase the Fairymead Notes at the time when the transactions were first agreed with the SBDs in early November 2020 and that he never told them anything else as the deal changed in late November/early December 2020.

The claimants advanced a number of reasons for submitting that the SBDs and Mr Greensill did not have this understanding about the purposes of the $440 million injection.

The claimants’ first reason was the absence of any express agreement requiring the use of the $440 million for the specific purposes of repaying or redeeming the Notes. They pointed out that there was no “use of proceeds” clause. This is correct, but of limited weight. Mr Cheung convincingly explained that Mr Greensill had come to SoftBank seeking funds to enable him to remove the risk of a default on the Katerra facility, as it would seriously damage the pre-IPO funding efforts. Mr Greensill wanted to neutralise the risk. That could have been achieved by the SBDs buying the Notes themselves or by Greensill doing so. The first of these routes was considered initially but was then rejected. The second route, of internalising or adopting the risk, was then pursued. Mr Cheung said in evidence (and I find) that that he understood that the second route required the Notes to be bought by Greensill, as otherwise the RPA could not properly have been cancelled.

I have concluded, having heard Mr Cheung’s evidence in the light of the surrounding documents, that since Mr Greensill had sought $440 million for this very purpose and had stressed its urgency, it did not occur to Mr Cheung that Mr Greensill would not ensure that this had not carried into effect. Mr Cheung also explained that, in his experience, use of proceeds clauses were rarely used in structured finance deals. Indeed, the one example of such a clause the claimants were able to identify required only that the proceeds of an advance be used for general corporate purposes. The inclusion of a clause of that kind would have made no difference to the outcome in the present case.

The claimants also submitted that Mr Greensill said in evidence, and the SBD witnesses agreed, that the $440 million itself was fungible. This point is essentially an elaboration of the first. Mr Cheung and Mr Romeih did not suggest in their evidence that they thought that the self-same funds as were paid over on 10 November 2020 were going to be used to purchase or redeem the Fairymead Notes. Their evidence was however that this was the purpose of Mr Greensill asking the money and that they therefore thought that it would be used for that purpose.

The claimants also contended that there was no agreement about a mechanism for or timing of any purchase or redemption. Again there is only limited weight in this point. I find that the SBDs left it to the Greensill companies to decide on the manner and timing of the purchase or redemption. The technicalities were a matter for Greensill. However I am satisfied from the evidence of Mr Cheung and Mr Romeih that they understood from Mr Greensill that the situation was urgent and that they believed that it was necessary for such purchase/redemption to have taken place before the release of the RPA. As Mr Cheung put it in evidence, that was part of the chain. I also accept the evidence of Mr Cheung that he believed from the 30 Dec WSJ Article that Greensill must have bought in the Fairymead Notes, as otherwise they would not have been able to compromise the RPA.

The claimants’ next point was that the SBDs knew on about 2 December 2020 that the Fairymead Notes remained outstanding and they did not complain. That is correct. However I accept the evidence of Mr Cheung that the SBDs understood that Mr Greensill had been seeking the amount of $440 million in order to defuse the potentially disastrous impact of a Katerra bankruptcy, and that they understood that the only way the debt under the RPA could properly be released was if Greensill had acquired the Notes by the date of the release. As Mr Cheung put it in evidence the $440 million completed the chain, meaning that it allowed Greensill to compromise the RPA. It was his evidence that he understood that the risk had to be taken onto Greensill’s balance sheet before the release of the RPA could happen. I also accept the evidence of Mr Cheung that once the payment of $440 million had been made he thought that the immediate problem had been addressed – the bomb had been defused – and that Katerra’s financial issues no longer posed a threat to Greensill’s pre-IPO fundraising. He said, and I accept, that after the $440 million liquidity was introduced into Greensill his focus was on Greensill’s fundraising efforts and that he paid little attention to the details of the legal agreements.

The claimants also relied on what they said was the muted reaction of the SBDs when they discovered in January 2020 that the Notes had not been purchased or redeemed. As to this, I accept the explanations given by the SBDs witnesses. Mr Cheung in particular explained that he was concentrating on Mr Greensill’s demands for even more money. Having agreed the $440 million injection from the SBDs, in late December 2020 Mr Greensill was seeking yet further funding. The impression I took from his evidence was that from late December 2020 onwards Mr Cheung was becoming exasperated by Mr Greensill’s continuing attempts to raise more money from SoftBank entities. I also accept his evidence that his main concern in this period was Greensill’s continuing efforts to promote its fundraising, and that from his point of view the Fairymead Note issue had been addressed through the $440 million injection. He simply did not think about this aspect very much after the investment was made. I also accept Mr Cheung’s evidence that when he discovered in January 2021 that Greensill had not completed the acquisition of the Fairymead Notes he assumed that there must have been some arrangements with the Credit Suisse funds for deferred completion of the acquisition of the Notes.

As to the claimants’ submissions that the $440 million investment was essentially the price paid to compromise the dispute with SBG about the CEP in relation to Katerra, I have already found that while that dispute was part of the background to the negotiation of the November 2020 transactions - and was, indeed, expressly compromised in the Omnibus Deed - the principal catalyst for the injection of $440 million was the perceived need to remove the risk to the pre-IPO funding from a Katerra default. I accept the evidence of Mr Cheung and Mr Romeih that Mr Greensill approached them in October 2020 explaining that the financial issues at Katerra were jeopardising the pre-IPO fund-raising and that Greensill had an urgent need for $440 million in order to repurchase or redeem the Fairymead Notes and remove that risk. This was the essential catalyst for the series of transactions.

The claimants also contended that Mr Greensill sought the permission of the SBDs to disclose the existence of the Omnibus Deed to Credit Suisse and that the SBDs had refused such permission. The claimants contended, relying on some evidence of Mr Greensill, that the SBDs imposed a “cone of silence” on him, preventing him from telling CS about the transactions. The claimants argued that this undermined the evidence of the SBDs that they understood that the SBDs would apply the $440 million in acquiring the Fairymead Notes. As to this, the cross-examination of Mr Cheung proceeded on the basis that the SBDs had not given consent for the disclosure of the Omnibus Deed to CS in the context of the fundraising. The claimants did not squarely suggest to either Mr Cheung or Mr Romeih (who gave evidence second) that the SBDs had prevented Mr Greensill from disclosing the nature of the arrangements concerning the $440 million from the separate Credit Suisse team dealing with the Fairymead Notes. The cross-examination therefore did not squarely raise the case advanced in closing to the effect that the SBDs had refused to allow Mr Greensill to discuss the Omnibus Deed with the representatives of the Credit Suisse funds. Moreover, in their evidence Mr Cheung and Mr Romeih said (and I accept) that they were unaware of any confidentiality terms concerning the RPA and that there were no discussions with Mr Greensill about this. Furthermore, the references in the contemporaneous documents to Greensill seeking permission to disclose the Omnibus Deed relate to disclosure to CS in the context of the fundraising. There was no evidence that Mr Greensill had asked for permission to discuss any of the agreements with Credit Suisse in the context of the Fairymead Notes structure. I have concluded that the contemporaneous documents show that such discussions as there were to disclosure of the Omnibus Deed related only to disclosure in the context of the fundraising efforts.

In reaching these conclusions I have given careful consideration to the claimants’ submissions that an adverse inference should be drawn from the failure of the SBDs to call Mr Son and Mr Misra.

In this regard, Mr Greensill gave evidence, which I accept, that Mr Son was the principal decision maker at SBG. The documents generally show that Mr Misra was heavily involved in the transactions. The documentary evidence also shows that Mr Greensill had direct access to Mr Son and Mr Misra. However the current question is whether the court should draw an adverse inference from their absence as witnesses. The claimants sought an adverse inference concerning their allegation about the waiver of confidentiality in the Omnibus Deed. They contended that Mr Cheung and Mr Romeih were not involved in the discussions concerning waiver of the confidentiality obligations in the Omnibus Deed. However, as just explained, the documents establish that the Greensill companies only sought a waiver of these obligations in order to make disclosures to potential investors in Greensill under the pre-IPO fundraising. There is nothing in the documents to show that Mr Greensill sought permission from any SoftBank entity to discuss the CEA or TA with the claimants.

Rather, I find that Mr Greensill took a calculated risk not to tell the claimants what was going on because he hoped to be able to raise funding by one means or another to satisfy the claimants’ claims. I do not therefore consider that the claimants have established a case for the SBDs to answer in this regard. I also note that there was no pleaded case that Mr Greensill asked the SBDs for permission to discuss the CEA or TA with the claimants or that the SBDs refused to give permission. The only pleaded case was in relation to the Omnibus Deed and the discussions about that were in relation to the pre-IPO fundraising. It is not appropriate to draw any adverse inferences on this issue from the absence of Mr Son or Mr Misra.

The claimants also invited the court to draw, from the absence of Mr Son and Mr Misra as witnesses, an adverse inference about the SBDs’ understanding of the purposes of Mr Greensill in causing GL to enter the CEA and TA and their knowledge or failure to satisfy themselves as to whether the Fairymead Notes had been repaid. I do not consider this would be appropriate. There was ample evidence before the court on this issue on the basis of which I am able to reach factual conclusions. Mr Cheung and Mr Romeih were able to give evidence about their understanding of the transactions. They were involved in the details. There was also the documentary record set out in detail above. The pleaded case about knowledge was that all of Mr Cheung, Mr Son and Mr Misra had the relevant knowledge. I consider in all the circumstances that the SBDs were justified in calling Mr Cheung and Mr Romeih as witnesses and do not consider that I should draw adverse inferences from the absence of others.

Drawing the threads together, I find that the SBDs understood that the Greensill companies would use the $440 million injected under the CLNs for the purposes of redeeming or purchasing the Fairymead Notes and that, had they not understood that to have been the purpose for the payment, the SBDs would not have injected the $440 million into GCPL. From the outset Mr Greensill explained to them that this was why he was seeking the money and this was their consistent understanding, as reflected in the documents.

I also find that the SBDs believed that the repurchase or redemption of the Notes must have happened by the time of the CEA and the TA as they believed that Greensill would not otherwise have been able to compromise the RPA. Mr Cheung’s evidence to this effect was not challenged.

My findings as to Mr Greensill’s understanding are these. He thought that there were two preconditions to buying back the Fairymead Notes. The first was that the need for HSR approval, which was itself a precondition to the recapitalisation of Katerra by SVF1. The second was that the injection of $380 million into Katerra had to occur so that Katerra had the resources to pay the $176 million to GL for onward transmission. It appears therefore that Mr Greensill did not necessarily anticipate applying the amounts of $440 million to buy or redeem the Fairymead Notes until after the recapitalisation had been consummated. Indeed the SBDs did not advance a positive case at the trial that Mr Greensill shared their understanding that the Notes would have to be repaid before the security was released. Their position was that Mr Greensill did not apply his mind on 30 December 2020 to whether he was acting in accordance with the alleged understanding.

In my judgment, the fact that Mr Greensill thought this was not inconsistent with my conclusion that the SBDs thought that once the payment of $440 million had been made the Greensill companies would promptly buy or redeem the Notes in order to internalise the risk. The SBDs did not believe it was necessary to agree a mechanism or final date for this (though the Omnibus Deed referred to 31 December 2020) because they understood it was urgent. As already explained, the evidence of Mr Cheung (which I accept) was that he understood that Greensill had to acquire the Notes (or otherwise secure the position of the Noteholders) before GL could properly release the RPA.

Hence I find that the parties did not share the same understanding about the arrangements. The SBDs thought that the Greensill group urgently required the money to enable them to internalise the risk from the Fairymead Notes. They thought that Mr Greensill would promptly use the money for that purpose and did not consider that it had to await the other steps. They also thought that the acquisition of the Notes had to occur before the RPA could be released. Mr Greensill appears to have regarded the injection of funds as improving the group’s position and providing him with the necessary liquidity to buy in the Notes once the recapitalisation of Katerra had occurred.

On the pleadings it is common ground that the Secondary Trade was entered into on 31 December 2020 and cancelled on 7 January 2021. The evidence showed that the trade was orally agreed by Mr Greensill and Mr Degen at 10.30am on 30 December 2020.

The SBDs submitted that by 31 December 2020 Credit Suisse knew about the cancellation of the RPA at about the time it happened. They relied on Mr Greensill’s evidence and the terms of the WSJ article of 30 December 2020 which stated that “Greensill Capital agreed to cancel around $435 million in debt owed by Katerra in exchange for a roughly 5% stake in the company”.

For the following reasons, I find that Credit Suisse was not aware of the cancellation of the RPA either at the date of the Secondary Trade or its cancellation on 7 January 2021.

As noted in [271] above, on 30 December 2020 Mr Degen referred to the news article and asked Mr Greensill: “Not new for you... means you swap in to eq after the buy back I assume...?” Mr Greensill replied noting: “Correct, Michel. Warmest regards, Lex”. In my judgment Mr Greensill’s message was opportunistic and misleading. Mr Degen’s email showed Mr Greensill that Mr Degen did not believe the debt could already have been released. The Fairymead Notes remained outstanding, though Mr Degen had agreed with Mr Greensill by then that they would be bought by one of the Greensill companies. Mr Degen therefore assumed that the exchange (and therefore the cancellation) would take place once the Secondary Trade had settled. Instead of explaining that the RPA had already been cancelled by the CEA, Mr Greensill agreed with Mr Degen’s message. Mr Greensill was unable to explain this at all convincingly in evidence, falling back on saying that he had written it in the early hours when he was not at his best. That was to my mind a tacit acceptance that the email was misleading.

That this was Mr Degen’s contemporary understanding of the position is supported by the email from Mr Mathys to Mr Degen timed at 12:48pm on 31 December 2020. Under the subject line “vorschlag” (suggestion), he said,

“Up to December 31, we have executed additional sell orders in notes related to Vision Fund in order to bring the remaining exposure in line with the agreed internal investment guidelines. All notes from View as well as the entire Katerra multi-obligor program were sold (after transfer of the Katerra notes, the program will be cancelled by Greensill thereafter in exchange for an equity stake in the company). As required by Greensill, value date of the transactions is January 14th”.

This email again refers to the cancellation of the programme in exchange for the equity stake to take place after the transfer of the Notes to Greensill. This is consistent with Mr Degen’s email of 30 December 2020.

In his email to Mr Varvel on 4 January 2021, Mr Greensill said:

“Katerra has been restructured, however the fund does not have credit exposure to Katerra. As a multi-obligor receivables programme the credit risk is on multiple customers of Katerra. These notes all run off within the next 90 days and the programme is 100% insured. There will be no performance impact on the SCF Fund”.

Mr Greensill did not explain in this message that the RPA had already been released (so that there was no longer anything backing the Fairymead Notes). In my judgment this email too was misleading. It gave the impression that the Notes continued to be backed by receivables.

Mr Greensill’s initial evidence was that this sentence contained a typo and that it should have said the credit risk “was” on multiple customers of Katerra. That was simply not credible. The purpose of the wording of the passage was to reassure the claimants about the security for the Notes. To tell the claimants that the credit risk had been, but was no longer, placed on multiple customers would have given investors no assurance at all. On the contrary, under questioning by the SBDs Mr Greensill gave different evidence, accepting the suggestion that the wording emanated from Mr Degen. That did not however adequately explain why Mr Greensill was willing to confirm it.

Mr Greensill’s answers about this email (including his change of position) led me to view some of his evidence as opportunistic and has bolstered my view that I should take a cautious view of his evidence more generally.

The SBDs relied on email exchanges showing that Credit Suisse was involved in drafting some of the wording of this email, and in particular the reference to the credit risk. But in my judgment this point led nowhere. Mr Greensill knew the true position and the content of the email was misleading. Indeed in my view the prior exchanges with Credit Suisse about the contents of this email support the conclusion that Credit Suisse was unaware that the RPA had already been released; the premise of the email was that there were still assets backing the Notes, hence the reference to “credit risk”. It is most unlikely that Credit Suisse, which wanted to be able to disclose this information to its own investors, would have originated language which it knew to be false.

As explained in section B above, on 5 January 2021 Mr Greensill approached Credit Suisse to suggest the cancellation of the Secondary Trade. Instead it was agreed that the remaining Fairymead Notes would be redeemed over a longer period. In his communications with Credit Suisse he did not refer to the CEA or the release of the RPA. It appears to me inherently improbable that Credit Suisse would have been prepared to reach this agreement had it known that the RPA had already been released and that the Notes had no backing.

The SBDs also relied on an email of 14 January 2021 in which Ms Warner asked Mr Varvel to provide more color on exactly how “…our $440M in the Virtuoso fund relates to the “forgiven debt” from Greensill and the $200M infusion by Softbank Vision fund in December”. The SBDs submitted that this showed that the highest levels in Credit Suisse were aware of the release of the debt under the RPA. I do not accept this. The use of inverted commas around the phrase “forgiven debt” shows to my mind that Ms Warner was unsure of the position. It has also to be read in the context of the earlier emails from Mr Greensill which had given the impression that the equity-debt exchange would happen once the Fairymead Notes had been repurchased. It is inherently highly improbable that Credit Suisse’s internal communications would have been as relaxed as this if they had understood that the assets backing the Notes had already been released.

The SBDs also relied on an email of 9 March 2021 which, they submitted, showed Ms Warner said that she was told by Mr Greensill about the release of the debt and the receipt of the $440 million. They invited me to reach the conclusion that Mr Greensill had told her those things in late 2020. As to this, Mr Warner referred in the email to the press reports in December 2020 (i.e. the 30 December 2020 article). She then explained that Mr Greensill had told her that he had received the $440 million (she actually referred to $435 million but that does not matter) in November 2020 and that it should have made its way to Credit Suisse’s account when the receivables became due in March 2021. The bankruptcy had supervened and the funds were trapped. The email did not however say when Mr Greensill had told Ms Warner. On the balance of probabilities I find that Mr Greensill told Ms Warner about his arrangements with SoftBank after Greensill entered insolvency. There is nothing in the documentary record to suggest that she was aware of the release of the RPA or any arrangements concerning the $440 million injection before then.

By the time of the trial, the only area of disputed US law expert evidence concerned a specific category of future receivables (“FRs”).

The parties divided the FRs into three categories. Type 1 FRs were those where Katerra had undertaken the relevant work or services but had not invoiced or billed for it. Type 2 FRs were where there was an existing contract between Katerra and a customer, but the relevant work or services had not yet been performed. Type 3 FRs were where there was no existing contract between Katerra and a customer but it was hoped that such a contract would be entered. The FRs covered by the RPA included each of these three types. For completeness, Actual Receivables (or “ARs”) are those already invoiced or billed by the Katerra Sellers.

The parties agreed that the RPA gave GL an effective security interest over ARs and Type 1 FRs, and that it gave no such interest in Type 3 FRs. The remaining area of debate therefore concerned Type 2 FRs.

The SBDs contended (a) that any security interest over Type 2 FRs under the RPA did not “attach” under NY UCC §9-203 at the point when the Katerra Sellers had entered into the underlying contracts but only when the Katerra Sellers’ right to payment arose; and (b), in any event, that rights to payment for Type 2 FRs did not constitute “proceeds” of the underlying contract from which they arose for the purposes of US Bankruptcy Code (“Code”) §522(b)(1).

As to the first issue, under NY UCC §9-203(a) a security interest “attaches to collateral when it becomes enforceable against the debtor with respect to the collateral”. For this purpose, “collateral” means “the property subject to a security interest”: NY UCC §9-102(12). As Professor Schaffer accepted, Type 2 FRs can in principle be “collateral” if the other requirements of §9-203 are met.

Enforceability for the purposes of §9-203 is governed by §9-203(b). The experts agreed that the only requirement in issue in relation to Type 2 FRs was whether a debtor “has rights in the collateral or the power to transfer rights” under §9-203(b)(2), which is governed by NY common law.

Both experts accepted that the Restatement (Second) of Contracts (“the Restatement”) could reasonably be regarded as an authoritative source for these purposes.

Professor Schwarcz explained in his first report that under the Restatement §321 the assignment of a right to a future payment expected to arise under an existing contract takes effect as at the date of the entry into the underlying contract. He also said that Illustration 1 of Restatement §321(1) shows that such an assignment is both effective and not defeated by the assignor’s bankruptcy even if the work generating the right to payment is carried out during the bankruptcy.

Professor Schaffer explained that he had not considered §321 in preparing his first report, but accepted that Illustration 1 was “consistent with” the terms of §321(1) (as is perhaps obvious).

In the event, it emerged that Professor Schaffer’s objection to any reliance on §321(1) was derived from his own interpretation of the RPA. His view was that under the RPA there was no assignment of Type 2 FRs and that it covered only ARs and Type 1 FRs.

The experts agreed that the interpretation of the RPA is an objective exercise, and the experts did not identify any principles that are materially different from the familiar English law ones concerning commercial contracts.

On the issue of interpretation of the RPA:

I also prefer the analysis advanced by Professor Schwarcz that a right to payment under an existing contract and which is contingent on work or services being performed is capable of being immediately and enforceably assigned before such performance has actually taken place.

On this basis, I find that GL obtained an enforceable security interest in the Type 2 FRs at the date of formation of the Katerra Sellers’ underlying contract with its obligors. GL thereby acquired a valid and effective security interest in Type 2 FRs with effect from the same date.

As to the second issue, the issue is whether under Code §552 GL had a perfected security interest in Type 2 FRs. As to this I accept Professor Schwarcz’s view that GL’s security interest would not have been cut off by a Katerra bankruptcy because GL had acquired pre-petition property (i.e. the contractual rights expected to arise under the underlying contracts between the Katerra Sellers and their obligors). It was therefore not property acquired after bankruptcy within the US Bankruptcy Code §552(a). It follows that under Code §552(a) a bankruptcy would not have affected the security. I did not consider that Professor Schaffer had any real answer to Professor Schwarcz’s analysis.

I also accept the claimants’ alternative argument based on §552(b). Where an immediate payment obligation in respect of Type 2 FRs arises post-petition, that obligation would constitute “proceeds” of the Type 2 FRs for the purposes of the additional exception in Code §552(b)(1). As Professor Schaffer accepted, the question under this provision is whether a right to payment arising post-petition may be considered the “proceeds” of rights assigned pre-petition. He also accepted that this involves a fact-specific enquiry dependent on the terms of the security arrangements. The US federal courts have not alighted on a single test to determine whether such rights are “proceeds” under Code §552(b1). Furthermore, I conclude that there is also no rule or principle that payments received in connection with work undertaken post-petition are incapable of being “proceeds” of rights assigned pre-petition.

I prefer the views of Professor Schwarcz that the case law establishes that rights to payment arising from post-petition work can constitute “proceeds” of pre-petition collateral.

In my view, Professor Schaffer’s contrary opinion ultimately turned to his taking the narrower interpretation of the RPA, to the effect that “Purchased Receivables” under the RPA could not include Type 2 FRs, not on the meaning or operation of Code §552(b)(1). As explained above, however, I reject Professor Schaffer’s narrowly drawn interpretation of the RPA.

I find that the RPA contained a broad definition of “proceeds” which was wide enough to capture the proceeds of “rights arising out of [the] collateral” as defined in UCC §9-102.

For these reasons I accept the submissions of the claimants that GL had security over Type 2 FRs under the RPA which would have been valid and effective in a bankruptcy of Katerra.

The value of the Purchased Receivables as at 30 December 2020 depends on a number of variables, including the actual make up of those receivables, the risk of Katerra entering into a bankruptcy, and the impact that would have had on recoveries.

Before turning to the parties’ positions, it helps to refer to some relevant concepts. Under the US Bankruptcy Code a corporate bankruptcy may fall under Chapter 7 or Chapter 11. Chapter 7 is a form of liquidation. Chapter 11 bankruptcies take various forms, including what the experts called a “going concern” bankruptcy, where the company continues to trade as a debtor in possession (“DiP”) and may ultimately emerge from bankruptcy, or “an orderly wind down” where there is more protection than under Chapter 7, but the company ceases to trade. Both types of Chapter 11 bankruptcy will involve DiP funding, which is typically a form of super-secured lending.

The claimants’ position in their pleadings was that the value of the receivables over which GL had security as at 30 December 2020 was $440 million.

The claimants’ position in their closing submissions was that:

This case was not set out in the pleadings. There was just a general pleading that the value of the receivables was $440 million or such amount as the court may determine.

The SBDs’ position in closing was in summary as follows:

The first question is whether Katerra would have gone into bankruptcy absent the CEA and TA.

I find as a fact that had the CEA not been entered into the PSPA would not have been entered into either. The PSPA expressly referred to the execution of the CEA agreements as a closing condition, and the execution of the PSPA took place as part of the same closing process. Moreover the PSPA and the CEA were two elements of an overall restructuring of the balance sheet and capital base of Katerra. The purpose of the two transactions was to save Katerra from bankruptcy. This was indeed explained by Katerra’s CEO to journalists: see the WSJ article of 30 December 2020.

I reject the claimants’ submissions that in assessing the risks of bankruptcy and valuing the receivables the court should proceed on the basis that the PSPA would have been entered into even if the CEA had not been entered.

Their first submission was that the PSPA should not be included as an element of the “transaction” they have pleaded for the purposes of the claim under section 423. Since any order to be made under section 423(2) is designed to restore “the position to what it would have been if the transaction had not been entered into” only the CEA and the TA need be disregarded in the relevant hypothesis.

I shall assess this submission on the assumption that the relevant “transaction” comprises the (combined) CEA and TA (an assumption I shall rule on below). Even making this assumption I am unable to accept the claimants’ submission. The hypothesis required by section 423(2) requires one to identify the position that would have occurred had the transaction not been entered into. The identification of this counterfactual state of affairs is a question of fact. I have found that if there had been no CEA there would also have been no PSPA. The claimants’ argument conflates the identification of “the transaction” with the separate (counterfactual) question of what the position would have been had the transaction not occurred.

The claimants’ second argument was that the PSPA was not a condition precedent of the CEA. That is so, but the PSPA was expressly contingent on the CEA being entered into. They were part of the recapitalisation of Katerra.

The claimants’ third argument was that SoftBank may still have considered investing in Katerra if the CEA had not been entered into. There was no evidential basis for that contention. The internal SBG and SBIA documents point the other way. The injection of $200 million was part of the Katerra recapitalisation and the two transactions were closely interconnected. I do not consider that is an argument which had any realistic basis in the evidence.

This leads to the question of what would have happened had the CEA and the PSPA not occurred. In my judgment it is clear beyond doubt that Katerra would have entered bankruptcy at the end of December 2020 or in early January 2021. Katerra was in serious, indeed perilous, financial difficulties during the second half of 2020. At several stages during that period the threat of bankruptcy was imminent. A rescue was required. Katerra itself announced on 30 December 2020 that the injection of $200 million and the cancellation of the Katerra debt had allowed it to avoid bankruptcy. I am in no doubt that, had the restructuring not occurred, the directors would immediately have filed for bankruptcy.

I reject the claimants’ further arguments that, absent the CEA, Katerra would have continued as a going concern. Their first argument was that the SBDs continued to support Katerra after 30 December 2020 and that this shows that the SBDs considered Katerra to be a going concern. These historical events, however, took place in a real world, where both the CEA and the PSPA were entered, that is, one in which Katerra had been restructured and recapitalised. The fact that SoftBank continued to support the restructured Katerra tells one nothing of the hypothetical world in which the restructuring is assumed not to have happened.

The claimants next relied on contemporaneous valuations which posited lower probabilities of bankruptcy. On 6 January 2021 Duff & Phelps assumed a 20% probability of Katerra going into bankruptcy. On 1 April 2021 Houlihan Lokey assumed that Katerra was a going concern on the basis of SoftBank’s continued support. On 11 January 2021 SoftBank assumed a 50% probability of liquidation but expressed confidence in Katerra’s new CEO Paal Kibsgaard to solve Katerra’s liquidity issues and to reach cashflow breakeven in Q4 2022. However each of these valuations and reports were made in the real world, in which both the cancellation of the $440 million owed under the RPA and the injection of $200 million under the PSPA happened. As already explained, they are uninformative in the hypothetical world in which it is assumed that neither occurred.

For these reasons, there is no warrant for the probabilistic approach advanced by the claimants in their closing submissions. The 20% and 50% figures are based on the documents referred to in the paragraph immediately above. These real world events do not assist in conducting the exercise the court must conduct, which is to consider, counterfactually, what the position would have been had the CEA not been entered into. As explained above I find that there is no realistic doubt that Katerra would not have continued as a going concern. It would have entered bankruptcy in December 2020 or early January 2021.

I find as a fact that such a bankruptcy would have been commenced under Chapter 11. It would probably have been an orderly wind down rather than a going concern bankruptcy. There was no evidence before the court to support the latter possibility and the claimants did not argue for it in their closing submissions. A going concern bankruptcy would probably have required far more substantial DiP funding than an orderly wind down and there was no evidence before the court that such DiP funding would have been available.

It is convenient to note at this point that the first expert report of Mr Brown, which addressed the valuation of the receivables, proceeded on the unarticulated assumption that Katerra would have continued after 30 December 2020 as a going concern. Mr Brown accepted in oral evidence that his going concern assumption was based on the $200 million received by Katerra under the PSPA. Mr Brown ultimately accepted that, absent financing at the end of December 2020, Katerra would have entered bankruptcy.

The court’s guidance for experts requires them to explain significant assumptions. Mr Brown did not do so. He appears to have reached his opinion on the basis that, since he was instructed to reach an opinion on the basis that the CEA and TA should be disregarded in the counterfactual, he should assume that the PSPA should not be disregarded. In other words he appears to have thought that he was required to assume that the PSPA would have been entered into even if the CEA had not. In his first report he did not explore the reality or reasonableness of this assumption. He did not comment on it. In my judgment his approach fell below the standards to be expected of an independent expert. When approaching the valuation of the receivables on the counterfactual assumption that there was no CEA, it was self-evidently relevant to ask, first, whether the PSPA would have been entered into, and, secondly, what would have happened if the PSPA had not been entered into.

I turn next to the probable rate of recovery of receivables in the event of a Katerra bankruptcy. It was common ground in closing submissions that the rate of recovery in a Katerra bankruptcy falls to be calculated by reference to a structured, wind-down, Chapter 11 bankruptcy.

The expert witnesses took different approaches to this question. Mr Brown produced an analysis of anticipated liquidation recoveries in 14 bankruptcies, which he contended were comparable to Katerra. The anticipated recoveries were in almost all cases given as a “low” estimate and a “high” estimate. These produced an average (mean) range of 37% (low) to 50% (high).

Mr Farrell took a different approach. He explained how a bankruptcy of Katerra would have been likely to affect the recoverability of receivables. He said that debtors would have been likely to contend that, by failing to perform its obligations, Katerra had caused them losses, which they would have been likely to seek to set off against the receivables. Katerra’s customers would have stopped paying amounts due. More specifically:

The SBDs also relied on the actual course of Katerra’s bankruptcy (which commenced in June 2021). They contended that the actual bankruptcy experienced comparatively low recoveries. I shall return to this below.

I turn to my assessment of the competing expert evidence.

As already noted, Mr Brown analysed the estimated high and low rate of recoveries in 14 sample bankruptcies. These were derived from estimates that were produced at the time of the relevant bankruptcy proceedings. In Chapter 11 bankruptcies there is a requirement to show that the outcome will be at least as good as would occur in Chapter 7 liquidation, and therefore, at an early stage in the Chapter 11 proceedings, estimates are given of expected recoveries in such a liquidation. Mr Brown’s analysis was derived from public filings containing these estimates.

The SBDs challenged Mr Brown’s analysis of comparable bankruptcies on a number of grounds. They contended as follows.

First, the range of expected recoveries set out in the table of sample bankruptcies is very wide. In the 14 comparable bankruptcies, the range of estimated recoveries is from 0% (the lower bound for Gramercy) to 100% (for Welded Construction). This itself casts doubt on the statistical utility of the exercise. Moreover the calculation is based on only 14 sample bankruptcies. There have been at least tens of thousands of other construction bankruptcies in the period covered by the sample. Mr Brown does not apply any statistical methods to test the reliability of his approach.

Second, the method is statistically susceptible to outliers. A small debt with a high recovery rate disproportionately alters the average recovery rate. There are several outliers which distort the average recovery rate. These are Welded Construction, TNT, Unitek, McDermott and Dixie Electric.

Third, the analysis included companies which Mr Brown accepted were “less comparable”. The bankruptcies which Mr Brown accepted in evidence were “more comparable” to Katerra were IES, Carpenter, Blanton, Maquire, KPH, CBC and RCR. The average recovery rate of those companies was 33% (low) to 44% (high).

Fourth, the even the “more comparable” companies were generally much smaller than Katerra. Other than IES, all the “more comparable” companies were owed between $3 million and $7 million of accounts receivables. They are therefore poor comparators.

Fifth, the analysis included subcontractors as well as general contractors. IES, Carpenter, RCR and CBC were subcontractors. Mr Brown accepted that it was plausible that subcontractors would probably be engaged on shorter term contracts, and would probably be subject to lower damages claims for non-performance. Katerra was a general contractor.

Sixth, even the general contractors are not good comparators. Blanton, Maquire and KPH are the “more comparable” general contractors in Mr Brown’s opinion. None was comparable to Katerra, which was heavily loss-making and built modular products in its factories.

The claimants supported Mr Brown’s comparative approach and took issue with each of the criticisms. They submitted that the correct rate of recoveries of ARs and Type 1 FRs was 50%. This was at the top of a range of 37% to 50% advanced by Mr Brown in his expert evidence. They argued for the higher end because Mr Brown’s estimate was prepared with reference to expected outcomes in Chapter 7 liquidations, which typically allow for lower recoveries than Chapter 11 bankruptcies.

The claimants submitted that Mr Brown’s evidence was to be preferred to that of Mr Farrell for a number of other reasons.

They argued, first, that Mr Brown had greater experience of bankruptcies. While Mr Farrell has experience in the construction industry he has little experience in relation to construction bankruptcy.

Second, Mr Farrell did not explain in his reports whether his proposed range of 10-15% was based on evidence of comparable bankruptcies and did not indeed identify any specific cases from which any experience he had was derived.

They argued, third, that as a matter of methodology Mr Brown’s comparative analysis was a more secure empirical basis for assessing the likely recoveries.

In support of Mr Farrell’s approach, the SBDs submitted that Mr Farrell had extensive experience of the construction industry and had identified, by reference to Katerra’s specific business and contracts, the practical reasons why Katerra’s customers would have ceased to pay and recoveries would have been very low. These have been summarised in [478] above.

They also relied on what occurred in the actual bankruptcy of Katerra, the Chapter 11 filing which took place on 6 June 2021.

This is a convenient point at which to resolve a dispute between the parties as to whether the evidence about Katerra’s actual bankruptcy is admissible and (if so) helpful.

The SBDs contended, first, that when considering the counterfactual it is appropriate to have regard to what actually occurred, provided that intervening events do not make a material difference: see Assetco plc v Grant Thornton UK LLP [2021] Bus LR at [210]; and the discussion at [603‎] to [604‎] below. Second, there are no such intervening events in the present case. The obvious possible intervening event is Greensill’s own bankruptcy, which caused concern among bond insurers. However, if the CEA and PSPA had not occurred, then Katerra would still have encountered liquidity issues, being burdened by the $440 million RPA debt. This would have caused the same concern among bond insurers.

The SBDs also contended that ARs are earned amounts and not dependent on future performance. Therefore, although the reasons for bankruptcy in June 2021 may have differed from those in December 2020, those reasons would have had no bearing on the recovery of ARs.

The claimants submitted that, while in appropriate cases the court may have regard to subsequent events where the value of an asset at the valuation date is uncertain, it cannot do so where there has been a material change in the circumstances such that the later event throws no helpful light on the position as at the earlier date. Here the documents filed in the Katerra bankruptcy itself show that the intervening bankruptcy of Greensill was a major and material factor leading to the bankruptcy of Katerra.

On this issue, I have concluded that the fact that Katerra went into bankruptcy in June 2021 does not assist in assessing the question whether Katerra would have gone into bankruptcy in December 2020 if the CEA had not been entered into. I accept the claimants’ submission that the intervening bankruptcy of Greensill materially damaged the business and prospects of Katerra as third parties (including bonding companies) were more reluctant to do business with a company so closely associated with Greensill. I have indeed reached the clear conclusion that Katerra would have entered bankruptcy without being influenced by the evidence that it actually went into bankruptcy in June 2021.

On the other hand, it appears to me that what happened in respect of the recovery of receivables in the actual bankruptcy of Katerra is capable of throwing some helpful light on the counterfactual exercise of determining what would have been recovered in a notional bankruptcy assumed to have commenced some months earlier. Given the common ground that the notional bankruptcy would have been a wind-down bankruptcy, rather than a going concern bankruptcy, there is no reason to suppose that the level of recoveries in the notional bankruptcy would have been materially different from that in the actual one. In this regard, I find that there is no reason for concluding that, in the actual bankruptcy, the intervening collapse of Greensill materially affected the level of recoveries (see further below).

The SBDs drew attention to the following features of the actual bankruptcy of Katerra.

First, while Katerra’s revenue for 2020 averaged $145 million per month, in the first three months after bankruptcy it was $20.4 million, $12.1 million, and $12.4 million respectively. The SBDs contended that this supports Mr Farrell’s opinion that in a bankruptcy customers would have ceased paying to a material extent.

Second, from June to August 2021 accounts receivables decreased from $332 million to $272 million, and those 90 days overdue increased from $28 million to $95 million. In the same period, aggregate operating receipts amounted to c. $14 million, which implies that the decrease was not due to collections. The SBDs contended that this supports the view that Katerra was facing claims for damages from customers which eroded the value of receivables.

Third, the expected plan distributions predicted a distribution to unsecured creditors of $38 million to $75 million, with prior payment of $5 million of secured claims and $11 million-18 million of priority claims. So far, the actual realisations have been $80.345 million, which includes $45.31 million paid in fees, $2.2 million of administrative claims, and $9.9 million of priority claims. There has been no payment to unsecured creditors. These low rates of recovery support Mr Farrell’s views.

As to Mr Farrell’s reliance on the rates of recovery from Katerra’s actual bankruptcy, the claimants submitted that this exercise is not useful for several reasons.

First, the bankruptcy is ongoing and has not yet completed, so it is not known what the overall recovery will be.

Second, Greensill’s intervening bankruptcy had an adverse influence on Katerra’s rates of recovery in its bankruptcy.

Third, even if a comparison is properly to be made even in the actual bankruptcy, Katerra’s expected recovery on 1 September 2021 was $54.1 million to $98 million in respect of accounts receivable of $272.72 million, which is a rate of recovery of 20% to 36%. This is higher than Mr Farrell’s estimated rate.

I come to set out my conclusions about the probable level of recoveries in the counterfactual bankruptcy of Katerra.

In my judgment, while the evidence of both experts was of some assistance, each had serious shortcomings.

As to Mr Brown’s analysis, I consider there is real force in several of the criticisms levelled by the SBDs.

First, I had serious reservations about the statistical reliability of the analysis. The numbers exhibit an extremely wide spread of expected recoveries (in both the low and high cases) from 0% to 100%. Mr Brown has not carried out any analysis of their distribution to determine whether the mean is a reliable measure of the central case. Moreover in a number of cases the high expected recovery is the same as the low expected recovery, but this is not explained.

In addition the sample size of 14 is small when compared to the much larger body of construction bankruptcies during the period covered by his list. Mr Brown’s evidence did not to my mind adequately provide a principled basis for the selection of the 14 comparables.

Second, Mr Brown has not applied a weighting to the average. This means that a small debt with a high recovery rate has a large impact. For instance, if the 100% recovery for Welded Construction’s $600,000 debt is excluded, the re-calculated average recovery rate falls to 32%-46%.

Third, as already noted, Mr Brown accepted in his evidence that seven of the companies were “less comparable” companies. The more comparable ones were IES, Carpenter, Blanton, Maguire, KPH, CBC and RCR. The average recovery rate of those companies is 33%-44%.

Fourth, most of the companies were much smaller than Katerra and they were likely to have less complicated contractual arrangements. Other than IES (with receivables of $135 million) Mr Brown’s “more comparable” companies were owed between $3 million and $7 million of accounts receivables. I consider that this casts doubt on their usefulness as comparators.

Fifth, I had real doubts about the inclusion of subcontractors as well as general contractors as comparators. Of the “more comparable” companies, IES, Carpenter, RCR and CBC were subcontractors. Mr Brown accepted that it was plausible that subcontractors would probably be engaged on shorter term contracts and would probably be subject to lower damages claims for non-performance. The average recovery rate for the subcontractors IES, Carpenter and RCR is 45% to 63%, whereas for Blanton, Maguire and KPH (the general contractors within the “more comparable” group) it is 30% to 36%. Katerra was a general contractor.

Sixth, I consider that even among the general contractors there are real questions over comparability. None of Blanton, Maguire and KPH are readily comparable to Katerra. Katerra was a very large company which sought to apply technology to construction. A large part of its business was the production of modular products in its factories, which it then installed. By contrast Blanton was a small local builder which owed only $6.3 million. KPH was a small general contractor which entered bankruptcy due to litigation over a hotel development. Its receivables were $4.5 million. Maguire served federal, state and municipal governments, which are likely to be less risky than private counterparties, and their recovery estimates assumed 75% to 95% project completion. Its receivables were $4.7 million. None are comparable to Katerra in nature of business, types of customers or scale of receivables.

Seventh, there are a number of outliers which tend to distort the average recovery rate. Specifically:

For these reasons I concluded that Mr Brown’s comparables analysis must be approached with a considerable degree of caution.

The SBDs accepted that, if a comparative approach is to be used, IES was probably the closest comparator for Katerra. Mr Brown said in evidence that it was the closest to Katerra. In its bankruptcy IES had an expected recovery rate on accounts receivable of 22% (low) to 35% (high). The SBDs submitted that, if the comparable approach is appropriate at all, the lower end of the range should be taken, since IES was an electrical subcontractor, not a general construction contractor.

On the other hand, I also had serious reservations about Mr Farrell’s conclusions. I was able to accept much of his evidence about the general problems a trustee in bankruptcy would have met with in collecting receivables. These included the fact that customers of Katerra would have terminated contracts for non-performance and would have brought claims for damages. In this regard it is relevant that a major part of Katerra’s business involved the manufacture and installation of modular products and that customers would face difficulties in accessing substitute performance from other contractors to complete unfinished projects. There is no doubt force in each of these general considerations.

However, Mr Farrell’s recovery range of 10-15% was not derived from empirical data or experience. Mr Farrell did not base this range on any identified comparable bankruptcies which could be scrutinised by the claimants or the court. Though Mr Farrell has considerable experience of the construction industry, his personal experience of bankruptcy is limited. It appeared to me that his range was based more on feel or educated instinct rather than comparable empirical experience.

I also take into account my general observations about the expert witnesses. As recorded above, I concluded that Mr Brown was inclined to act as an advocate for the claimants and appeared unwilling to engage constructively with questions properly posed in cross-examination. In some of his answers about his analysis of the comparative recoveries of receivables it appeared to me that he was seeking to downplay the differences between Katerra and his selected comparables.

I have concluded that Mr Farrell sought to assist the court but, as explained above, I had concerns about his limited experience of bankruptcies and the extent to which his general points about the problems in making recoveries in a bankruptcy could be translated into an evidence-based anticipated rate of recovery.

I have also decided that it is appropriate to take into account the experience of the actual recoveries during the actual bankruptcy of Katerra. That experience suggests that recoveries were likely to have been comparatively low. However, it is also correct to note that the Katerra bankruptcy is not complete and that, accordingly, the overall rate of recovery is not yet known.

For these reasons, the expert evidence on both sides was less helpful than might have been hoped or expected. However, the assessment of what would have happened in the counterfactual can rarely be precisely estimated and the court's task is to do the best it can with the available material. I have concluded that the appropriate rate of recovery to apply is 22%, which falls at the lower end of the anticipated recovery range for IES, the closest comparable about which there was evidence before the court. In reaching this conclusion I have given some weight to the actual experience in the Katerra bankruptcy, which tends to support a lower rate. I have also given some weight to Mr Farrell’s evidence about the difficulties that would be faced by the trustee in bankruptcy of Katerra in making recoveries. I have also given weight to the material differences between Katerra and IES, including that IES was an electrical subcontractor, not a general contractor.

The rate of 22% applies to both ARs and Type 1 FRs. As already explained, it was common ground that the recovery rate for Type 2 FRs is nil.

The next issue is the value of ARs as at 30 December 2020.

It was common ground that the face value of the ARs in the 22 December 2020 Purchase Request File was $330 million.

The claimants contended that that was the correct figure to take for ARs. They said that some amounts may have been collected by Katerra between 22 December and 30 December 2020 but contended that they would have had effective security over such collections.

The SBDs, supported by the evidence of Mr Farrell, approached the value of the ARs differently.

Mr Farrell adopted two approaches. The first was to consider the consolidated balance sheet for Katerra dated 31 December 2020, which gave a figure of $408 million for ARs and $85.9 million for Type 1 FRs. Mr Brown accepted that the balance sheet was potentially an appropriate basis for analysis.

Mr Farrell then deducted from the balance sheet figures the value of non-US receivables and US retainage (being sums for work done by Katerra withheld by Katerra’s US customers until performance of the whole contract). This was because such receivables were not included within the definition of the Purchased Receivables under the RPA.

Mr Farrell’s calculations for non-US Receivables and US retainage were based on a document called the December Go-Forward Plan. The Go-Forward Plan recorded non-US receivables as 10.53% of the total accounts receivable figure and US retainage as 28.32% of the total accounts receivable figure. Applying these percentages to the total of $408 million in the 31 December 2020 balance sheet, the totals to be deducted are $43 million and $115 million respectively.

The claimants took issue with these figures. Specifically, Mr Brown gave evidence that typical retainage under a construction contact is significantly lower than 28% and is normally no more than 10-15%.

On Mr Farrell’s approach, with these deductions made, eligible ARs would have been $250 million.

Mr Farrell also considered a second approach, which he described as a cross-check for the first. The second method started with the 22 December 2020 Purchase Request File, which (as already noted) listed $330 million of ARs. Mr Farrell then examined the available underlying documentation relating to the Receivables listed in the 22 December 2020 Purchase Request File. He identified two categories of deductions which he said fell to be made from the headline ARs figure: (a) duplicate ARs and (b) already paid ARs.

As to (a) duplicate ARs, Mr Farrell explained that, in respect of several Payment Applications, there were other Payment Applications relating to the same work and projects (generally including the same project names, project numbers and payment application numbers). In total some $11,897,730 worth of ARs were duplicated by further ARs. These figures are calculated from documentation relating to 87% of the ARs, as documents were not available for the remaining 13%. Mr Farrell grossed up the $11.9 million figure to give a figure for duplicate ARs of $13.8 million.

As to (b), payments already made, Mr Farrell identified documents showing that several Payment Applications underlying the ARs in the 22 December 2020 Purchase Request File had already been paid. In total Mr Farrell identified Payment Applications totalling $74,264,602 where the documents appeared to show that the Receivable had been paid. Again the documents covered only 87% of the ARs. Mr Farrell grossed this up to give a figure for amounts already paid (and which would therefore potentially amount to Collections) of $85.8 million.

To give an example of (b), the 22 December 2020 Purchase Request File included an amount of approximately $5.6 million due from Lennar Multifamily Communities, LLC. However Payment Application #16 dated 30 September 2020 included a document called “Unconditional Lien Waiver” stating that Katerra had received $5,557,122.78 in respect of “Requisition 16”. The amount therefore appears to have been paid before 30 September 2020 (but not remitted to Greensill on any later settlement date (in contravention of the RPA)).

When the duplicates amount of $13.8 million and already paid amount of $85.8 million are deducted from the $330 million of ARs in the 22 December 2020 Purchase Request File, the resultant figure is $230 million. Mr Farrell said that this second approach therefore provided support for his first approach, which had led to the conclusion that there were $250 million of ARs as at 30 December 2020.

Mr Farrell also explained that the Collections amount had to be reduced to $29 million, since $214.9 million of Katerra’s total cash of $243.9 million as at 30 December 2020 was in a general Katerra Wells Fargo bank account over which Greensill had no security. While this may have been a breach of the RPA, in a bankruptcy of Katerra, Greensill would only have been able to recover from accounts over which it had security.

The SBDs accepted that the $230 million produced by Mr Farrell’s second approach may be an underestimate on two grounds: (i) some Future Receivables listed in the 22 December 2020 Purchase Request File might have become ARs by 30 December 2020, and (ii) the extrapolation from the documented 87% of cases is not a completely reliable process. They contended however that it provided corroboration for the $250 million figure produced by Mr Farrell’s first approach.

The SBDs therefore submitted that the correct value of the ARs as of 30 December 2020 was $250 million, and that Greensill also had security over Collections of $29 million.

The claimants challenged the approach adopted by Mr Farrell.

They submitted, first, that insofar as Mr Farrell’s valuation was based on the draft unaudited 31 December 2020 Katerra balance sheet and the Go-Forward Plan, it was flawed, including for the following reasons:

Insofar as Mr Farrell’s valuation was based on his secondary method, it was also wrong, for the following reasons:

As to Mr Farrell’s approach to Collections, the claimants note that he reduces his valuation of the collected receivables from $85.8 million to $29 million for two reasons: (a) of the $243.9 million of cash on Katerra’s 31 December 2020 balance sheet, $175 million was attributable to the PSPA; (b) of the $243.9 million, $214.9 million was held in a Wells Fargo account over which GL did not hold Account Security. The claimants contended that this approach was wrong for three reasons:

I come to my conclusions about the amount of the ARs.

Overall it appears to me that, while it suffers from imperfections, Mr Farrell’s approach was to be preferred. My reasons follow.

First, the ultimate question is the value of the ARs as at 30 December 2020 and the consolidated balance sheet for 31 December 2020 is the closest dated reference document.

Second, the claimants did not seriously challenge his adjustment to the balance sheet figures for non-US work.

As for the figures for retentions, there was some force in their point that the figures appear higher than that usually found in construction contracts, but its force is somewhat tempered by the observation that the retention percentage tends to increase as contracts approach completion. There is also some force in the claimants’ contention that the way in which the figure for retention was calculated was mechanical (being the product of a formula in a spreadsheet). But that was nonetheless a figure calculated by Katerra and I do not think that it can readily be dismissed. As to the apparently high level of retentions compared to other construction contracts, I accept Mr Farrell’s evidence that while typical retainage under such contracts is 10%-20%, the percentage of retainage against ARs tended to rise as projects approach completion.

Third, I considered there was force in Mr Farrell’s opinion that the Purchase Request File was unreliable in relation to at least some duplicates and amounts already received by Katerra. The examples given by Mr Farrell showed persuasively that there was duplication and prior collections of amounts included in the File.

Fourth, Mr Farrell’s approach involved the two approaches described above. The second was a cross-check for the first. The resultant numbers were reasonably close. The claimants’ criticisms of the second method had some force and Mr Farrell accepted that the second method was imperfect. Specifically Mr Farrell accepted that he has not had access to some categories of underlying documents, which might throw more light on the position. He also accepted that he had extrapolated from a sample. He said that extrapolation was a standard method in accounting and financial analysis, but accepted it was not perfect. Taking account of the imperfections in the method, I concluded that the secondary method gave some support as a cross-check for Mr Farrell’s first approach.

Fifth, I do not consider that Mr Farrell’s approach to the reliability of the Purchase Request File involves the conclusion that Katerra was engaged in deliberate wrongdoing. There may be a number of explanations of duplication of invoices and the inclusion of amounts already paid. These include accounting and billing errors. Katerra was a group made up of several business units and it is possible that there were internal reporting errors. In any event, the evidence showed that Katerra banked amounts received from customers into general bank accounts over which GL had no security. It appears therefore that Katerra in fact conducted its business contrary to the terms of the RPA.

Sixth, I return to Mr Brown’s general approach as an expert witness. I have explained that I was not always confident that he was seeking to assist the court. It appeared to me that he tended to see his role as advancing the claimants’ case. While he was highly critical of Mr Farrell’s approach he did not appear to me adequately to address the shortcomings in the Purchase Request File or justify its use as the source of the figures for ARs. By contrast, I concluded that Mr Farrell was doing the best he could to assist the court. He recognised the limitations of the available evidence and the shortcomings of some aspects of his approaches. Nonetheless the two approaches he took were a reasonable attempt to address the valuation issue.

For these reasons, having weighed all the evidence of both experts, I consider that the appropriate figure to take for the face value of the ARs as at 30 December 2020 is $250 million.

As to the question of the extent of the security held by GL over amounts already collected, I prefer the analysis advanced by Mr Farrell. The documents showed that Katerra was paying amounts into general banking accounts over which GL held no security. This was a breach of the RPA. Mr Farrell’s approach, which took the 31 December 2020 balance sheet as one of the reference points, was potentially imperfect in that there could have been payments made between 30 December (when $175 million was received under the PSPA) and 31 December 2020. Nonetheless it appears to me that his approach was reasonable and was the best that could be done with the available information. Hence there were secured collections as at 30 December 2020 of $29 million.

As to the face value of the FRs as at 30 December 2020, the claimants’ position is that Type 1 FRs were worth $22 million, and Type 2 FRs were worth $216 million. They base this on the fact that Katerra’s balance sheet as at 31 December 2020 recorded “costs and estimated earnings in excess of billings” of $85.9 million. Using the October Go-Forward Plan, this would have comprised non-US Type 1 FRs (or underbillings) of $64 million. Therefore the value of eligible Type 1 FRs was $22 million. There is no evidence that Katerra sold Type 3 FRs. Hence the remaining $216 million of FRs in the 22 December 2020 Purchase Request File must be Type 2 FRs.

The SBDs’ position is that Katerra Type 1 FRs were worth $5.6 million. They contended that the most reliable approach is to take the FRs contained in the 22 December 2020 Purchase Request File (which reflected the position on 15 December 2020) and to consider what quantity of those FRs were likely to have been converted to Type 1 FRs by reason of work done between 15 December and 30 December 2020. Mr Farrell’s approach was as follows:

The SBDs noted that the claimants’ figure of $22 million, which is derived from the October Go-Forward Plan, draws on forecasted “costs and earnings in excess of billings” which include amounts relating to all 199 Katerra customers, while only 13 customers were included in FRs. The SBDs submitted that the claimants’ approach, which asks what work was done subsequent to the November 2020 billing date but before 15 December 2020, involves the wrong question.

The claimants made a number of criticisms of Mr Farrell’s methodology for valuing Type 1 FRs:

I come to my conclusions about Type 1 FRs.

In my judgment, the claimants are right to say that if the exercise is to value which of the totals for “Future Receivables” in the 22 December 2020 Purchase Request File fall within Type 1, the correct question, in respect of each project, is how much work had been done in the period between the most recent prior billing date and 15 December 2020. That is because the receivable is only an AR once invoiced. Hence the ARs in the Purchase Request File are those which had been invoiced by the most recent billing date before 15 December 2020. Therefore the Type 1 FRs in the Purchase Request File were those classified as FRs in the Purchase Request File, but in respect of which work had been done up to the 15 December 2020.

That is correct so far as it goes. But the ultimate goal of the exercise is to value which Future Receivables listed in the Purchase Request File had become Type 1 FRs by 30 December 2020. That exercise would include any Future Receivables in respect of which work was done after 15 December 2020 (as well as where work was done before then). If the billing date fell before 30 December 2020 they would have become ARs by that date and reported in “Accounts Receivable” in the 31 December 2020 Katerra balance sheet. In order to conduct that exercise, Mr Farrell had sought to calculate the ARs on the basis of the 31 December 2020 balance sheet, and then calculate the value of Type 1 FRs by that date.

There is, however, in my judgment considerable force in the claimants’ submission that Mr Farrell’s methodology may have resulted in his disregarding possible Type 1 FRs which relate to contracts under which no invoices had been sent out. However there is an evidential gap in this regard.

It may also be the case that Mr Farrell has underestimated Type 1 FRs by taking the end of the month in some cases to be 30 December 2020 rather than 31 December 2020. Mr Farrell did not adequately explain the inconsistent approach he had taken in this regard.

There are other potential flaws in Mr Farrell’s methodology for estimating the amount of work that would have happened in the relevant periods. He has extrapolated from earlier periods. But, as the claimants submitted, the previous experience did not support the view that work was carried out at a consistent rate. Again it appears to me that substantial adjustments would need to be made Mr Farrell’s conclusions to reflect the risks of underestimation of the Type 1 FRs.

I have concluded overall that the value of the Type 1 FRs should be generally based on the approach of Mr Farrell, which I prefer to that of Mr Brown. However in order to guard against the risks of underestimation in Mr Farrell’s approach, as identified by the claimants, it seems to me appropriate to make significant upward adjustments. Doing the best I can I have concluded that the appropriate figure for Type 1 FRs, including such adjustments, is $10 million.

To summarise, I have concluded that as at 30 December 2020 there were ARs with a face value of $250 million and Type 1 FRs with a face value of $10 million. The 22% recovery rate gives a value for the $57.2 million ($260m x 0.22). In addition there were secured collections of $29 million. The total value of GL’s rights as at 30 December 2020 was therefore $86.2 million.