Luen Tat Watch Band Manufacturer Ltd v. Stephen Liu Yiu Keung and Another
Read the full judgment text of HCMP 1567/2021 on BabelCite. This High Court CFI judgment was delivered on 7 May 2026.
1. This is the trial of the committal application brought against Mr Stephen Liu (the 1 st defendant) and Mr David Yen (the 2 nd defendant) for breach of orders made against them in the High Court proceedings in HCMP 1071/2018. The plaintiff in both this application and the HCMP proceedings is Luen Tat Watch Band Manufacturer Ltd.
Cites 10 cases
|
HCMP 1567/2021 [2026] HKCFI 2564 IN THE HIGH COURT OF THE HONG KONG SPECIAL ADMINISTRATIVE REGION COURT OF FIRST INSTANCE MISCELLANEOUS PROCEEDINGS NO 1567 OF 2021 ________________________
________________________
________________________
________________________ J U D G M E N T ________________________ 1.This is the trial of the committal application brought against Mr Stephen Liu (the 1st defendant) and Mr David Yen (the 2nd defendant) for breach of orders made against them in the High Court proceedings in HCMP 1071/2018. The plaintiff in both this application and the HCMP proceedings is Luen Tat Watch Band Manufacturer Ltd. 2.The plaintiff was formerly subject to a winding up order, which was later stayed permanently. Between 13 October 2010 and 27 November 2017, Liu and Yen, who were during that period managing directors of Ernst & Young Transactions Ltd, were appointed as its joint and several liquidators. By order made by Deputy High Court Judge To on 27 November 2017, they were removed as liquidators by reason of their misconduct. In this judgment, I shall call the removal proceedings and the judge’s written decision “the removal litigation” and “the removal decision”. 3.Shortly after the removal, in December 2017, the plaintiff requested the return of all documents kept by the defendants in their administration of the liquidation. This marks the beginning of a prolonged and protracted dispute over the return of the documents, which has eventually given rise to the present committal application. 4.In February 2018, the defendants started to return some of the plaintiff’s documents. In the course of 2018, a total of 95 carton boxes of documents and two box files of documents were returned. They included financial documents, transactional documents, liquidation documents, corporate documents and external correspondence. The last item included external emails sent or received by the defendants and their team at Ernst & Young over the seven-year period whilst the defendants were liquidators. I shall call the process of returning the documents in 2018 “the 2018 production”. 5.Not satisfied, the plaintiff commenced proceedings in HCMP 1071/2018 seeking the return of its documents. The court found for the plaintiff and made orders which form the subject matter of the present committal application. 6.On 7 October 2020, Deputy High Court Judge Le Pichon ordered the defendants to:
7.Schedule 1 read:
8.About one month after the order was made, the defendants filed a notice of appeal and applied for stay of execution. 9.On 31 March 2021, the stay application was dismissed by Recorder Houghton, SC. In the same order, a time extension of 28 days was granted. The return deadline therefore fell on 28 April 2021. 10.In this judgment, I shall call the orders in paras 6(1) and 6(2) “the production order” and “the affidavit order”, respectively, and “the orders”, collectively. The former term should be taken to mean either that order or that order as varied by the subsequent time extension order. I shall use the term “Schedule 1 documents” to refer to the documents which are ordered to be handed over under the production order. I shall refer to the proceedings in HCMP 1071/2018 as “the production litigation”, the written decision of DHCJ Le Pichon “the production decision”, and the written decision of Recorder Houghton, SC “the stay decision”. 11.On the deadline of 28 April 2021, the defendants delivered 49 boxes of Schedule 1 documents to the plaintiff. 12.On 5 May 2021, Yen filed his 6th affidavit, in which he detailed the steps taken to retrieve the Schedule 1 documents. He concluded by stating this:
13.On the same day, Liu filed an affirmation, in which he stated:
14.In this judgment, I shall refer to the defendants’ return of the documents on 28 April 2021 as “the April production”. I shall call Yen’s 6th affidavit and Liu’s affirmation “the confirmatory affidavits”. 15.The defendants contend that by making the April production, they had complied with the production order, and by making the confirmatory affidavits, they had complied with the affidavit order. 16.The plaintiff thought (and still thinks) otherwise. After some correspondence, it commenced the present committal application by way of originating summons dated 11 October 2021. (The originating summons was later amended on 9 November 2023.) 17.According to the defendants, with a view to ensuring the completeness of the production, they voluntarily handed over further documents after the deadline. They delivered 8 boxes on 24 December 2021 and a further 4 boxes on 28 July 2023. I shall call the two post-deadline handovers “the December production” and “the 2023 production”. 18.As mentioned above, in the 2018 production, various categories of documents and external correspondence had already been handed over. The production order was therefore primarily (though not solely) about retrieving and returning internal working documents and emails generated by the defendants, their team and the administrative staff at Ernst & Young in the course of the liquidation. 19.Ernst & Young did not have a centralised system for storing these internal documents. They were scattered across the employees’ laptop computers and email inboxes. And, according to the defendants, that had caused huge difficulties for them to retrieve these internal documents. In this judgment, I shall use either the term “custodian”, which is used by the defendants, or the term “employee” to refer to fee earners at Ernst & Young who had worked on the liquidation during the seven-year period. The defendants had identified them by looking for those employees who had entered time on this file. At the time of the April production, 15 of them still remained with the firm, and 36 of them had already left employment. I shall call them “the existing employees” and “the former employees”, respectively. A brief introduction of the plaintiff’s case 20.The plaintiff’s case can be outlined as follows. 21.First, as a matter of construction, the production order requires the defendants to hand over all Schedule 1 documents kept by them. That was the state of affairs which the defendants were ordered to achieve by 28 April 2021. 22.Second, however, in the April production, the defendants failed to hand over all the Schedule 1 documents. The confirmatory affidavits were therefore false. The defendants were in breach of both the production order and the affidavit order. 23.And that was because the methodology adopted by the defendants based on 85 keywords was demonstrably unable to achieve the result mandated by the production order. They had to do more and could have done more, including a manual document review which they had adopted in the 2018 production. 24.As to why the plaintiff says that the defendants’ methodology was deficient and resulted in numerous Schedule 1 documents not being produced, the following seven complaints are set out in the amended statement:
25.The plaintiff further contends that the December production and the 2023 production is indisputable proof of the defendants’ breach of the production order. 26.Third, the breach was intentional because the defendants voluntarily and consciously chose to adopt their methodology in the April production. This amounts to a state of mind punishable by contempt. 27.Fourth, even taking into account the December production and the 2023 production, as of now, the defendants have still failed to hand over all Schedule 1 documents. The committal application is therefore appropriate and necessary. Committal is indeed the plaintiff’s “last resort”. A brief introduction of the defendants’ case 28.A prominent and persistent theme running through the defendants’ case is that a huge volume of documents, among them numerous emails, were generated over the seven-year period, and that a massive electronic document review exercise was required to be undertaken in order to comply with the production order, and was indeed undertaken resulting in the April production. It involved electronically searching 1.75 million documents, and manually reviewing 60,000 documents. Against this context, the defendants’ case is briefly as follows. 29.First, as a matter of construction, the production order requires them to conduct a reasonable search of the Schedule 1 documents, rather than to produce every single document without exception. Notwithstanding the word “all” in the order, a reasonableness test applies such that as long as reasonable steps are taken, the order is not breached even if some documents are missing. 30.Second, there was no breach as their methodology, developed with input from legal, forensics and IT experts, was a reasonable one. 31.Third, the mens rea requirement is not satisfied. The defendants reasonably believed that the 85 keywords selected by them would identify all documents responsive to the production order, and did not intentionally omit any keywords which they knew would result in the exclusion of responsive documents. Any gaps in the April production did not result from intentional exclusions of documents or sources of documents. They therefore fall squarely within the “accidental” exception for contempt purposes – the gaps were not consciously excluded; the omission was accidental. 32.Fourth, the defendants’ commitment to comply with the production order was clear from the large scale of the April production. They had all along made clear that they would act on any reasonable request made by the plaintiff. They voluntarily handed over further documents in the December production and in the 2023 production. A committal application was never necessary to secure their return. Under the last resort principle as explained by the Court of Appeal in China Metal Recycling (Holdings) Ltd v Chun Hei Man [2018] 1 HKLRD 455, the committal application should not have been taken out. 33.Fifth, in any event, the production order has been complied with, at the latest, by the time of the 2023 production. If there were any prior breach, the breach has now been purged. In the circumstances, the court should not commit the defendants for contempt. 34.This is a case of civil contempt. In Hong Kong, the leading authority is the Court of Final Appeal decision in Kao, Lee & Yip v Koo Hoi Yan (2009) 12 HKCFAR 830. The court adopts a three-stage approach: para 21. It should:
35.The plaintiff and the defendants hold starkly different positions on these issues. In addition, two other issues are contentious:
36.It should be obvious that the focus of this case is the April production. The three-stage questions are to be asked in respect of this round of production as the question of whether there was a breach is to be determined with reference to the deadline of 28 April 2021. If there was a breach then, anything done after that deadline would not alter that fact. The subsequent conduct of the defendants, including the December production and 2023 production, is however still relevant in the following ways. 37.First, if the plaintiff’s construction of the production order is correct, the two subsequent productions are plainly proof that the April production was not adequate. I have been shown some samples of documents produced in the December production. Some of them are, as a matter of fact, different documents from those contained in the April production. This means that the defendants had failed to hand over all Schedule 1 documents in the April production. 38.Second, if the defendants are correct in saying that latest by the time of the 2023 production, they have returned all Schedule 1 documents, the purging of the contempt will have an impact on the ultimate order the court should make. 39.At the trial, the plaintiff called Ms Seline Li as witness. She is a director of the plaintiff. 40.At the close of the plaintiff’s case, both Liu and Yen elected to testify. They also called Mr Steven Huen as witness. Huen was an in-house legal counsel at Ernst & Young Hong Kong. He assisted the defendants in the April production and the December production. By the time of the trial, he had left the employment of Ernst & Young. He is now in-house legal counsel in another international auditing firm. 41.It is worth highlighting at the outset that this case does not, on the whole, turn on the resolution of factual disputes. There are a few discrete matters on which factual findings should be made. For instance, the existence of the adjudication summaries under the missing POD summaries issue. Another example is the defendants’ knowledge of the IT system at the time of the April production. Apart from these matters, the primary facts in this case are largely uncontroversial in the following sense. 42.The plaintiff has no first-hand knowledge of how the defendants had created, filed and stored the documents relating to its liquidation within their own system. Nor does it have any first-hand knowledge as to what the defendants actually did in order to retrieve and return the Schedule 1 documents, and what they knew (and did not know) at the time. The plaintiff’s case is premised on the defendants’ own narrative on these matters. The plaintiff is saying that based on what the defendants say about them in correspondence and in these proceedings, they were in breach of the orders with the necessary mens rea. 43.In other words, the outcome of this committal application does not really turn on the credibility of the witnesses (save for a few discrete factual disputes). The court’s task is primarily to apply the principles as explained in Kao, Lee & Yip and China Metal to the primary facts. 44.At the trial, the plaintiff was represented by Mr Wong Yan Lung, SC leading Mr Martin Kok. The defendants were represented by Mr José-Antonio Maurellet, SC leading Mr John Hui. 45.The plaintiff was part of the Luen Tat group of companies founded by the late Mr Lee Sai Nam in the 1970s. The group was in the business of manufacturing and trading watchbands and stainless-steel accessories. His three children, Seline Li, Ken Li and Richard Lee, used to work for him in the group. 46.The father was the sole beneficial owner of the plaintiff. But his shares were held by Ken Li and Richard Lee on trust for him. 47.Since around 2008, Ken Li started disputing the father’s ownership of the shares. The father was ousted from the management of the group. That led to the father commencing two actions in 2009:
48.In July 2010, the plaintiff was ordered to be wound up. 49.Following a trial in 2014, the court found in favour of the father in the father and son action. Ken Li’s appeal against that decision was dismissed in January 2017. These proceedings therefore finally resolved the ownership issue in favour of the father. 50.By order made by DHCJ To in November 2017, the winding up order of the plaintiff was stayed permanently, and the defendants were removed as liquidators on the ground of misconduct. Relevant to this committal application are the following adverse findings made against the defendants, expressed in exceptionally critical terms by the judge. 51.The defendants entertained spurious proofs of debts in favour of Ken Li and his camp. Their conduct showed their bias in favour of Ken Li and against the father. These are some of the judge’s comments:
52.The defendants decided to forego investigations and claims against Ken Li and his camp. The judge said:
53.The defendants positively assisted Ken Li in the father and son litigation. This is what the defendants did:
54.The defendants’ appeal against the removal decision was dismissed by the Court of Appeal on 27 April 2021. 55.Following the removal decision, upon requests made by the plaintiff, the defendants made the 2018 production. That was done in batches during a period of seven months from February to September 2018. 56.The plaintiff was not satisfied with that. By July 2018, in relation to emails, the defendants had only managed to hand over those up to January 2013. This meant that at least another five years’ worth of emails had yet to be returned. The plaintiff commenced the production litigation in July 2018. (To complete the chronology, I should add here that at about the same time, the plaintiff commenced an action in HCA 1397/2018 against the defendants claiming breach of fiduciary duties. I shall call this action “the breach action”.) 57.The defendants opposed the production application. In the end, on 7 October 2020, DHCJ Le Pichon found in favour of the plaintiff and made the production order. I highlight two material matters in the production decision which are relevant to the construction issue and the last resort issue. 58.First, the court effectively accepted the plaintiff’s ground for seeking the Schedule 1 documents, which was that the relationship between a company and its liquidators is essentially that of principal and agent, the plaintiff has a broad entitlement to the documents or records retained by the agent in the course of its agency insofar as such documents or records concern or relate to the principal’s affairs, and all such documents belong to the principal and have to be delivered up on the termination of the agency: paras 18 to 20. The defendants made the submission (recorded in para 26) that the principal is not entitled to all the documents brought into existence by the work in question and it is necessary to see for what purpose the documents had been brought into existence. The court rejected that submission and held:
59.Second, the judge was highly critical of how the defendants had behaved themselves in the 2018 production. Her Ladyship said:
60.A 28-day deadline was imposed. About one month later, the defendants lodged an appeal against the decision and applied for stay of execution pending appeal. 61.On 31 March 2021, Recorder Houghton, SC dismissed the stay application. I would highlight two matters recorded in the stay decision. 62.First, the recorder took note of the defendants’ submission concerning the large scale of the production exercise: para 29. The defendants’ evidence before the court was that:
63.Second, the recorder recorded that the defendants sought a time extension of 56 days should the court refuse to grant a stay pending their appeal. The recorder acknowledged that the exercise would clearly be “substantial” but commented that it “should have been in progress since the Order was made”. He granted a time extension of 28 days. 64.The defendants did not renew the stay application with the Court of Appeal. In fact, shortly afterwards, they abandoned their appeal. 65.While the stay application was ongoing, the defendants made a proposal to the plaintiff. In Yen’s 4th affidavit, the defendants classified Schedule 1 documents into three categories:
66.Hogan Lovells, who were solicitors then acting for the defendants, wrote in a letter dated 27 November 2020:
67.The solicitors then proposed to apply a set of search terms on the email inboxes, retrieve the responsive emails and limit the review exercise to those emails:
68.The proposed search terms were listed in an annex to the letter. It contained 55 terms. I reproduce the annex to this judgment under “Annex 1 – The 55 keywords”. 69.The plaintiff’s solicitors, Joseph SC Chan & Co (“JSCC”), replied by letter dated 7 December 2020 seeking clarification on a number of matters, pointing out that the defendants had a duty to locate all documents responsive to the order, and that they knew nothing about the system of storage of emails maintained by Ernst & Young and it would be practically impossible for them to decide which would be the best or suitable method to locate the documents. 70.By letter dated 10 December 2020, Hogan Lovells made a reply. They made a point of noting the plaintiff’s “wholesale refusal to narrow down the email review and come to a reasonable methodology”. They then reiterated what they saw as the purpose of the handover, namely that it was to enable the plaintiff to ascertain the full extent of the defendants’ wrongdoing. They wrote:
71.JSCC did not reply to this letter. 72.On 28 April 2021, the defendants made the April production. 73.It is necessary to set out the methodology adopted by the defendants. Documents stored in centralised servers 74.For documents stored in centralised servers dedicated to the plaintiff’s liquidation, including, eg, draft liquidators’ reports and working papers and documents generated by the tax team, they were downloaded from the servers. However, before they could be handed over, each of them needed to be checked to see whether they contained any documents relating to other clients. They might do so because documents from other client files might have been used as templates. 75.This was done by a manual review of these documents, which came to a total of 7,922 pages. According to the defendants, the process was substantial. As of 5 November 2020, six fee earners had spent 171 manhours reviewing the documents. Internal emails of fee earners 76.There was no centralised system within Ernst & Young for storing internal emails of the fee earners who had worked on the plaintiff’s liquidation. They existed on the computers of the individual fee earners and their email inboxes. With the advice from Hogan Lovells and Ernst & Young’s forensics and IT teams, the defendants devised the following methodology for retrieving these internal emails. 77.First, these fee earners were identified by checking who had entered time on the Luen Tat file from June 2010 to the time when the defendants were removed as liquidators. As mentioned above, 15 existing employees and 36 former employees had worked on the file. 78.Second, the laptop computers of the 15 existing employees were collected. For the 36 former employees, their laptop computers would not be available for searching as they would have been reformatted and recycled for use by new joiners in accordance with the usual procedure of the firm. 79.Third, the usual document hold process then took place. A document preservation notice was issued to each of the 15 existing employees. They each responded to the notice and, six of them (including Liu and Yen) confirmed that they held documents responsive to the production order. 80.Fourth, the laptop computers of these six existing employees were imaged. Around 1.75 million electronic documents including emails were retrieved. 81.Fifth, old emails from the Lotus Notes backup were extracted for five out of these six existing employees. The sixth had no data before late 2014. In his 6th affidavit, Yen explained:
82.Sixth, all these documents were then uploaded to Relativity, a document review platform. A de-duplication exercise was then performed to remove duplicated documents. Following this process, the pool of documents was reduced to around 400,000. 83.Seventh, the forensics team searched through these documents using 85 selected keywords. The list of the 85 keywords is now reproduced in Annex 2 to this judgment. Around 60,000 responsive documents were identified. On the choice of these key words, Yen said:
84.Eighth, the 60,000 documents were then individually and manually reviewed by Yen and 10 members of staff at Ernst & Young. During the review, a number of documents containing information of other client matters were identified. For example, there was an email circulated internally containing a status update on various ongoing client matters. Such documents were redacted prior to being handed over to the plaintiff. 85.Ninth, in the process, the defendants also returned about 200 emails together with their attachments which were corrupted in that symbols were shown in place of text. Yen said in his 6th affidavit:
86.Tenth, after the review, the forensics teams helped to export the relevant emails from the document review platform. A total of 49 boxes of documents were delivered to the plaintiff. Correspondence post-April production 87.The plaintiff was not satisfied with the April production. It took the view that it was plainly deficient and inadequate. 88.Correspondence was exchanged between the two sides’ solicitors between late June and late August 2021. The significance of the correspondence is that they predated the commencement of the committal proceedings in October of the same year. It is necessary to refer to some of the contents as they are relevant to the last resort issue. By the time of the correspondence, the defendants had changed their solicitors from Hogan Lovells to Davis Polk & Wardwell. 89.On 25 June 2021, JSCC issued a 17-page letter on behalf of the plaintiff expressing their dissatisfaction. I highlight the following complaints made in the letter. 90.First, the plaintiff was of the firm view that the defendants were in breach of the production order and the affidavit order. 91.Second, not all backup emails of the fee earners who had worked on the liquidation had been searched. The letter referred to the fact that more than 30 fee earners had charged the plaintiff for work done before late 2014. Yet the defendants had not searched the backup emails of these fee earners. JSCC wrote:
92.Third, as to documents generated after late 2014, the defendants should have reviewed the documents kept at the Microsoft Outlook accounts of all the 15 fee earners who remained at the firm, instead of just the five or six fee earners. 93.Fourth, the use of the 85 keywords was not appropriate. It did not amount to a full review of the documents kept by Ernst & Young. It would only identify some but not all Schedule 1 documents. The complaints about the keywords included:
94.Davis Polk gave a short reply by their letter dated 8 July 2020. They stated that the defendants had complied with the production order and the affidavit order. They emphasised that given the large volume of electronic documents, it was necessary to implement a process to identify documents responsive to the orders and it was standard practice to apply an extensive list of keywords in the exercise. They stated that contempt was a remedy of last resort, and if there was any disagreement between the parties, the proper course was to apply to the court for directions. 95.Later, by letter dated 27 July 2021, Davis Polk gave a substantive reply. It was reiterated that the defendants’ methodology was appropriate, and that they had complied with the orders. However, without derogating from that position, they would respond to the plaintiff in the spirit that “any document production methodology could be the subject of reasonable and sensible dialogue”. They made the following suggestions:
96.The plaintiff did not respond to the letter. By letter dated 23 August 2021, Davis Polk followed up with their suggestions:
97.By letter dated 24 August 2021, JSCC commented that the keyword search “self-evidently cannot identify all the documents”. They further wrote:
98.This was followed by the exchange of two more letters. 99.By letter dated 26 August 2021, Davis Polk complained that the plaintiff was insisting on a “manual review” of “each and every one of the millions of documents in the laptops and/or Lotus Notes backups of the 36 additional custodians now identified”, which was an untenable position to take. But they would continue to work on “a final production” as previously suggested. 100.By letter dated 31 August 2021, JSCC expressed their disagreement, repeating their view that any production based on the proposed keyword search approach would be “fundamentally deficient, defective, and incomplete”. They therefore had no alternative but to consider taking out committal proceedings. 101.About two and a half months after the commencement of the committal application, the defendants made the December production. Eight A4 boxes containing 1,026 documents were handed over. An email and its attachment were treated as a single document for this purpose. 102.In this section, I shall set out:
103.The documents were retrieved with the assistance of Ernst & Young’s IT and forensics teams, adopting the following methodology. 104.First, the Lotus Notes backup emails were restored for the 36 former employees. This resulted in the retrieval of 1.1 million documents. After de-duplication, the pool was reduced to 750,000 documents. 105.Second, as regards the nine existing employees whose laptop computers were not imaged in the April production, the following steps were taken:
106.Third, 112 additional keywords were chosen. Both these new keywords and the 85 original keywords were then applied to the emails retrieved in this round, and the additional keywords were applied to the original data. This process identified 88,000 documents. 107.Fourth, Davis Polk then undertook a number of steps on these 88,000 documents, including:
108.The above steps resulted in the identification of 2,890 documents for production. They amounted to 1,026 documents if an email and its attachment were counted as one document. 109.Of these 1,260 documents, Huen described 696 of them as “near duplicates”, meaning that they are documents which are at least 80% similar to documents already produced in the previous rounds of production. Huen said:
110.Of the remaining 330 documents, Huen described them as “newly” produced documents:
111.In his affidavit evidence, Yen contrasted the extensive exercise conducted with the few documents which were retrieved as a result. He said:
112.Further, in his oral testimony, Yen supplemented that of the 696 “near duplicates”, a vast majority were in fact slightly different versions of documents previously produced. He gave an example that one email might be retrieved from two sources – from the server and from an email inbox. Two versions might be generated as a result, with the only difference being the time appearing in the email. His understanding was that the email retrieved from the server might show UK time whereas the same email retrieved from the email inbox might show HK time. For the December production, Yen also mentioned in cross-examination that some of them might involve emails between secretaries about booking of conference rooms, which he did not consider to be required to be returned. 113.On the other hand, the plaintiff criticises the defendants for downplaying the significance of the documents retrieved in the December production. 114.During the trial, the plaintiff compiled a bundle containing 57 emails selected from the 330 “newly produced documents”. They concerned tax matters, CCB matters, investigation documents, PRC legal matters etc. It is the plaintiff’s contention that it is self-evident that the contents go beyond emails between secretaries. 115.Separately, after the close of evidence, as directed, the parties jointly submitted a bundle of documents, classifying the documents handed over in the December production into broad categories and including examples under each category. The purpose of the joint bundle was to give the court a more concrete idea of the contents of the documents so that it could (1) verify whether Yen’s oral testimony was accurate, and (2) see for itself how similar these documents were to those produced in the 2018 production and the April production. 116.On the one hand, the defendants point out that it is now agreed that, of the 1,026 documents, 635 either had been “wholly provided” before the December production or were identical to documents produced before save for non-content difference, such as time stamps, time zone, file name or format. That is more than half of the total production. Of the remaining 391 emails, while they had some additional or different content, they were either near duplicates or otherwise inconsequential. For instance:
117.On the other hand, the plaintiff says that the defendants are now trying to trivialise the documents. It takes the position that there were substantively new documents. The plaintiff’s written closing submissions contain a selection of emails drawn from the joint bundle to support that position. It is unnecessary to recite those emails in this judgment. I consider the most illustrative one to be an email between two Ernst & Young staff in which the sender attached a three-page document entitled “Notes on interview with ex-employees”. This was not copied to either Yen or Liu. According to the plaintiff, these interview notes are highly relevant to Ken Li’s diversion of the plaintiff’s business to companies controlled by him and, hence, the plaintiff’s case of wrongdoing against the defendants in the breach action. 118.I am not in a position to say whether the plaintiff is correct about the significance of the interview notes. However, having looked at the contents and the length of the interview notes, I consider that the email and its attachment cannot quite possibly be described (as the defendants have done) as “near duplicates” or otherwise inconsequential. 119.Furthermore, the recipient of the email was a fee earner called Choong Sze Hua. She was one of the 36 former employees. In cross-examination, Yen accepted that, apart from him and Liu, Choong can be said to be the key member in the plaintiff’s liquidation team. 120.Hence, the plaintiff says that the very fact of the December production is indisputable proof that the defendants were in breach of the production order. 121.In cross-examination, the issue of why the defendants did not cause a search of the Lotus Notes backup emails of the 36 former employees in the April production was explored. Yen revealed for the first time that at that time, he was aware of the need to search such backup and had specifically requested the in-house legal team to do so but was informed that the backup of the former employees could not be retrieved. 122.Here is the relevant exchange:
123.Yen was then referred to JSCC’s letter dated 25 June 2021, in which one of the plaintiff’s major complaints was the failure to search the Lotus Notes backup emails of the 36 former employees.
124.In cross-examination, Yen accepted that the defendants should have caused a search of the Lotus Notes backup emails of the 36 former employees in the April production, but he insisted that he had asked but had been told they could not be retrieved.
125.In the trial, Huen was immediately called after Yen’s testimony. He confirmed that he was the only person in the in-house legal team handling the production. Contrary to Yen’s testimony, Huen said:
126.I reproduce Huen’s evidence below:
Was the December production still not sufficient? 127.It is the plaintiff’s contention that the methodology for the December production was still inadequate. 128.First, even with the 112 additional keywords, the keyword search exercise was still fundamentally under-inclusive. This is self-evident as, for instance, the word “Ken” was not included as a keyword. 129.Second, in respect of the steps taken by Davis Polk (see para 107 above), only sample reviews were undertaken. The exclusionary search terms were not disclosed to the plaintiff. In the end, only a small fraction of the documents were manually reviewed by Davis Polk. It is inevitable that such exercise would not identify all the Schedule 1 documents. 130.Third, as regards the six existing employees whose laptop computers had not been imaged in the April production, the defendants should have caused their computers to be searched as they had entered time. In fact, it is the defendants’ own evidence that at least one of them had recorded time of up to 31.5 hours. 131.On 28 July 2023, the defendants made the 2023 production. Four A4 boxes containing 541 emails were handed over. 132.This round of production was prompted by the plaintiff’s complaint, made in Seline Li’s 2nd affirmation filed in April 2022. She said that there were overlapping emails between those handed over in the 2018 production and those produced in 2021, and on a comparison, some of the ones produced in 2021 were in incomplete or corrupted form whereas the same emails were produced in complete or uncorrupted form in 2018. The defendants instructed Ernst & Young’s forensics team to investigate the causes of the discrepancies, who managed to retrieve some of the emails in uncorrupted form from the Microsoft Exchange server located in Mainland China. 133.In this section, I shall set out:
134.Based on the advice of the forensics team, it is the understanding of the defendants that:
135.After the forensics team found out that there might be discrepancies between the data stored locally on the laptops and that on the server, they proceeded to retrieve the data from the server. 136.According to the defendants, the process was time-consuming because the server located in the Mainland. Yen explained:
137.In the end, 748,000 documents were extracted from the server. Keywords were applied and the resulting 72,000 documents were de-duplicated against the emails previously produced. 11,000 responsive emails were identified through the process. A manual review by Davis Polk of each of these 11,000 emails identified 938 emails, which were then manually compared against the documents previously produced. This eventually resulted in the return of 541 emails in the 2023 production. 138.The defendants say that many of the 541 emails are not substantively new emails but contain minor variances.
139.On the other hand, the plaintiff criticises the defendants for downplaying the significance of the 2023 production. The plaintiff says that the 541 emails contain substantive and significant contents which are never produced before and which are relevant to the plaintiff’s case of misconduct against the defendants. According to the plaintiff, many of the emails provide compelling evidence of the defendants’ wrongful conduct, in particular in relation to their biased stance in favour of Ken Li and even collusion between them. 140.The plaintiff’s written submissions contain a selection of emails drawn from the 541 emails to support that position. The emails sent in December 2015 or shortly afterwards are specifically highlighted as that was when judgment was handed down in the father and son litigation. The plaintiff says that the emails show that even after the court ruled against Ken Li, the defendants were still positively assisting him by feeding him with documents, even litigation and strategic advice. While the contents of some of the emails may be brief (eg “ok”), when read in context, they confirmed the specific action of the defendants taken at specific times. They will therefore be highly probative of the plaintiff’s case in the breach action. The plaintiff calls these emails “implicating” emails. 141.I would make a similar comment to the one made above. I am not in a position to determine whether the plaintiff is correct about the significance of the new emails or whether they are indeed “implicating” emails. What I accept, however, is that some of them can properly be regarded as emails with new contents which have not been produced before. 142.As to why the Microsoft Exchange server was not searched in the April production, the defendants’ position is this. 143.First, as to the knowledge of the forensics team at the time, they believed that the server would fully synchronise with the individual laptops (as it was always intended to do). In fact, the forensics team still cannot explain why they were in fact not fully synchronised. 144.Second, as for the knowledge of Yen at the time, he said that he was not aware of any difference between the data on the server and on the laptop computers. He explained:
145.At that time, he had asked the legal team if the corrupted emails could be restored. The response was that they would make every effort to recover the originals, though it was acknowledged that some files were beyond recovery. 146.I reproduce his evidence in cross-examination:
147.In his evidence, Yen stressed that he was not an IT expert and he could only rely on the expertise of Ernst & Young’s IT and forensics teams to conduct the search. 148.In response to this, the plaintiff says that the theory of incomplete downloading is completely incredible and contrary to common sense. The reason is that many of the emails involved multiple recipients, and it would therefore require each and every one of these recipients to suffer from storage shortage on their laptop computers such that the email could not be retrieved from any of their laptops. 149.In fact, the plaintiff seems to go further in their criticism and make a positive allegation that the defendants were withholding the “implicating” emails in the April production. For instance, it points to a chain of emails dated 16 January 2015 to which the only correspondents were Liu and Ken Li. It shows, the plaintiff contends, that Liu was improperly giving advice to Ken Li. These emails therefore should have been found in the laptop computer of Liu and could have been handed over in the April production. But, inexplicably, they were not. The plaintiff asks – were they deliberately excluded in the April production? If not and if they truly could not be found on the laptop of Liu, the compelling inference is that these emails had been deleted from the laptop. 150.This was put to Liu in cross-examination and was denied by him.
Was the 2023 production still not sufficient? 151.It is the plaintiff’s contention that the defendants still failed to hand over all the Schedule 1 documents from the Microsoft Exchange server. 152.This is because the retrieval process, as described by the defendants themselves, was subject to further filtering process in addition to the keyword search which was previously used. It points out that the defendants did not make clear what keywords they had used in this round. 153.As remarked earlier above, there are only a few factual disputes in this trial. I shall deal with the following in this section and they concern what the defendants did (or did not do) and knew (or did not know) at the time of the April production:
154.All three issues are relevant to the mens rea question. 155.I find that Yen’s allegation that in the lead-up to the April production, he was aware of the need to search the Lotus Notes backup emails of the former employees and specifically requested the in-house legal team to do so but was informed that such backup emails could not be retrieved to be incredible for the following reasons. 156.First, this allegation was made for the first time in cross-examination. It had never been made before either in correspondence or in the affidavit evidence.
157.Second, Yen’s new revelation was directly and unreservedly contradicted by Huen’s oral evidence. Huen has no self-interest in these committal proceedings. He was an in-house counsel at Ernst & Young but has now left the firm. He has no obvious motive to lie about this incident. I find him to be an impartial witness. His evidence that the focus at that time was the existing employees and the former employees were not considered is inherently probable. 158.I therefore reject Yen’s evidence that he had asked the legal team to search the Lotus Notes backup emails of the former employees in the lead-up to the April production. I accept Huen’s evidence that such backup emails had not been considered at that time. 159.I consider that the defendants’ suggestion that the discrepancies between the data on the Microsoft Exchange server and the data on the individual laptop computers of the fee earners were likely due to incomplete downloading owing to the storage capacity of the laptop computers being exceeded to be credible. That suggestion is made by the forensics team at Ernst & Young, who should be taken to have the expertise in these matters. And it is inherently plausible. I am prepared to accept that at face value. 160.As the matter appears to be highly technical, I find Yen’s assertion that at the time of the April production, he was not aware of any difference between the data on the server and the data on the individual laptop computers to be inherently plausible and credible. I accept his evidence as set out in paras 144 to 147 above. 161.As regards the third issue, Mr Maurellet submits that this point is not open to the plaintiff as it was not pleaded. I agree. This is a serious allegation and is not set out as a ground of contempt in the amended statement. 162.In any event, the evidence before me is not sufficient to show, beyond reasonable doubt, that there was a deliberate attempt on the part of the defendants to conceal the “implicating” emails in April 2021. 163.The evidence relied on by the plaintiff is basically that there were “implicating” emails and that some of them were not disclosed in the April production but only disclosed in the 2023 production. That is, as I see it, the extent of the plaintiff’s evidence. 164.The defendants have now put forward an explanation as to why the 541 emails were only handed over in the 2023 production but not before, which I have found to be credible. 165.A further counter-argument is that in the 2018 production, the defendants had already produced documents which may on their face be prejudicial to them. Why would they want to withhold other similarly prejudicial documents in the April production? Lastly, as pointed out by Mr Maurellet, the entire document production process was overseen by Ernst & Young’s in-house legal team, with support from the IT and forensics teams. Outside law firms were engaged. It seems far-fetched to suggest that these employees and external lawyers would collude with the defendants to suppress documents. Insofar as it is suggested that the defendants withheld documents by themselves, there is no evidence before me as to how they managed to do so. STAGE 1 – CONSTRUCTION OF THE ORDERS 166.I now address the stage 1 question. The plaintiff’s construction is that the production order requires the defendants to hand over all Schedule 1 documents. The defendants say that they are only required to conduct a reasonable search. 167.There is no dispute as to the proper approach for interpreting a court order. A court order should be given the meaning it conveys to a reasonable person having all the background knowledge reasonably available at the time when the order was made: see, eg, Alan Chung Wah Tang and Hou Chung Man v Lee Siu Fong [2022] HKCFI 3683, para 26. In other words, in the construction exercise, proper regard must be made to the context against which the order was made. 168.In my view, there are two important contexts here. 169.First, the basis of the production order. The order was made after a contested hearing. The reasoning contained in the production decision provides the most important context for the order. 170.I have set out the reasoning of DHCJ Le Pichon in para 58 above. In essence, it was held that all the documents generated by the defendants in the course of the liquidation belong to the plaintiff, which have to be delivered up on the termination of the relationship between the plaintiff and the defendants. Mr Wong submits, and I agree, that in the production litigation, the plaintiff was seeking to recover its own documents from the defendants. The application was granted on that basis. Importantly, the judge explicitly rejected the defendants’ suggestion that there are documents which they are entitled to retain. 171.Second, at the time when the production order was made, it must have been known or was indeed known that the return exercise was going to be a massive one, requiring the input of substantial manpower, time and costs. While this matter was not referred to by the judge in the production decision, it must have been reasonably obvious that the exercise was going to be a huge one as the liquidation spanned over a long period of seven years. In the stay decision, Recorder Houghton, SC expressly acknowledged that the exercise would be substantial. 172.Against the above contexts, it is plain and clear to me that the word “all” in the production order should be given its literal meaning. “All” means all. The defendants are effectively ordered to return to the plaintiff what belongs to it. It is difficult to see why a reasonableness test should be introduced which will allow the defendants to return only some, but not all, of the plaintiff’s own documents. This interpretation would go against the legal basis of the production order. At the same time, it must have been known or was indeed known to the court that the scale of the exercise would be a large one. With this piece of information, the court nonetheless proceeded to make an order using the word “all”. It must be taken to mean that all documents are to be returned, notwithstanding that a substantial exercise would have to be carried out. The defendants’ analogy with specific discovery orders 173.Mr Maurellet invites the court to draw an analogy with specific discovery orders made in civil litigation. He submits that the analogy is appropriate because the language used in the production order closely mirrors that in specific discovery orders. He cites a long list of cases where the compliance of the discovery order in question requires a large-scale review exercise. It is said that in those cases, the courts have construed the orders to mean that as long as reasonable steps are taken, the orders will not be breached even if some documents are missed. The same approach, Mr Maurellet submits, should apply to the production order. 174.It is unnecessary for me to discuss those cases. The short answer to the argument is that the context for the production order here and that for specific discovery orders are different, and hence the analogy is not a proper one. As explained above, the production order is about the return of what belongs to the plaintiff to the plaintiff. The defendants are ordered to hand over the plaintiff’s own documents. 175.By contrast, a specific discovery order is about the disclosure of documents which are relevant to the disputes and necessary for the disposal of the case or for saving costs. A respondent in a discovery order is required to hand over documents not because those documents belong to the applicant, but because those documents are relevant and necessary. And that is why considerations of proportionality, efficiency and oppressiveness would come into play. It is not difficult to understand why in cases where a large-scale production exercise is anticipated, the court would be prepared to subject the respondent’s obligation to a reasonableness test. If it does not, the exercise may then become disproportionate, inefficient and oppressive. The defendants’ impossibility argument 176.Mr Maurellet mounts another argument in support of the “reasonableness” construction, which I shall refer to as “the impossibility argument”. 177.It is submitted that if the production order is given its literal meaning, it would be impossible for the defendants to obey it, as it would be impossible for them to produce every single Schedule 1 document by the deadline of 28 April 2021. The sheer magnitude of the production exercise is clearly shown in the defendants’ evidence. There were 1.75 million emails to start with (ie those emails retrieved from the laptop computers of the six existing employees). To conduct a review of this mass volume of emails with zero error would be impossible. Mistakes are inevitable. Even with genuine efforts, some documents may be missed. To require the defendants to produce literally all Schedule 1 documents without exception is to require them to do the impossible. This would be a most unreasonable result which the court could not have intended. No reasonable person, aware of the complexities involved, would have read the production order in such an onerous and draconian manner. 178.The impossibility argument raises a prior issue – is this an argument properly raised at stage 1? Or should the argument be dealt with at stage 2? Or even stage 3? 179.In a committal application, it is well established that the plaintiff bears the burden to prove beyond reasonable doubt that the defendant was in breach of the order in question. In doing so, the plaintiff also bears the burden to prove that it was within the power of the defendant to comply with it: China Metal at para 68. There will be no finding of contempt if the burden is not discharged. 180.As held by the Court of Appeal in that case, at para 75:
181.This is a stage 2 question which goes to actus reus. 182.It is also well established that the plaintiff bears the burden to prove beyond reasonable doubt that the breach was accompanied by the requisite mens rea. In the context of a mandatory injunction, mens rea is negatived if the omission to do the act mandated by the court order is involuntary. One such instance would be where the injunction was impossible of performance. In such a situation, the omission would be regarded as unintentional: Kao, Lee & Yip at paras 41 and 48. The Court said, at para 56:
183.This is a stage 3 question which goes to mens rea. 184.Hence, the impossibility argument raised by Mr Maurellet may potentially be dealt with at stage 1, stage 2 or stage 3. In fact, this is reflected in counsel’s submission. He says that the plaintiff’s construction should be rejected but submits, alternatively, that if the construction is adopted, the committal application should fail for impossibility. 185.I do not think there is any hard and fast rule for determining at which stage of a committal application an argument based on impossibility should be resolved. It must depend on the circumstances of each case. 186.Here, for the reasons set out above, I am of the clear view that the legal basis of the production order provides the most important context for construing the order, and the essence of the order was to require the defendants to return the plaintiff’s own documents to the plaintiff, and that must mean all documents. The impossibility argument is therefore irrelevant in the construction exercise. But that does not mean that the defendants will be committed for contempt for failing to do something which is impossible (if that is proved), as the impossibility argument can be properly raised at stage 2 and/or stage 3 and, if upheld, there will be no finding of contempt for lack of either actus reus or mens rea (or both). 187.To conclude, as regards the stage 1 question, I hold that the defendants are obliged to hand over all Schedule 1 documents under the production order. I reject the defendants’ argument that the obligation is subject to a reasonableness test. 188.I should add, by way of footnote, that the reasonableness argument was only raised for the first time in the defendants’ opening submissions. It is apparent that before that, they had been taking the view that all Schedule 1 documents were required to be handed over “regardless of how miscellaneous and inconsequential” they are: see Hogan Lovells’ letter sent shortly after the production order was made (para 66 above). 189.Based on the above construction, I now address the stage 2 question concerning actus reus – were the defendants in breach of the production order and the affidavit order? I shall first deal with the question of whether it has been shown that it was within the power of the defendants to comply with the orders. If so, I shall proceed to determine whether the plaintiff has made out its case on the deficiencies in the defendants’ methodology under the seven heads of complaint. Was it within the power of the defendants to comply? 190.The burden is on the plaintiff to show that it was within the power of the defendants to have produced all Schedule 1 documents by 28 April 2021. 191.The defendants say that the plaintiff has failed to demonstrate how they could guarantee the production of every single Schedule 1 document given the enormous scale of the exercise and the inevitability of human errors. In Yen’s evidence, it was also suggested that while the defendants could deploy staff within their own insolvency team to conduct the review exercise, they otherwise had to rely on Ernst & Young’s legal, forensics and IT teams to execute the process, and it was the firm who decided how much resources would be given to the exercise. 192.There are two aspects to the production order. First, the substantive aspect. Was it possible to hand over all Schedule 1 documents? Second, the timing aspect. Was it possible to do so by the deadline of 28 April 2021? 193.As to the first aspect, as a matter of analysis, the production order requires a certain state of affairs to be achieved. It is for the defendants to find out the proper means or come up with a proper methodology to achieve that state of affairs. 194.This was explained in Kao, Lee & Yip. That case concerned an order requiring the defendant (a solicitor) to maintain an account of all work which was undertaken by him in relation to any former client of the plaintiff (the defendant’s former law firm). The order there and the production order here are of a similar nature in that each of them is clear in terms of what is required of the defendant but does not specify how the defendant should go about achieving the final result. In such a case:
195.In the present case, the material question to ask, in my view, is whether it was possible for the defendants to come up with a methodology, if properly executed, would be effective in retrieving all the Schedule 1 documents. 196.On the issue of methodology, there has been a long-standing dispute between the parties as to whether a manual “eyeball” review should be undertaken to go through the electronic documents or whether a review based on the use of keywords could be adopted.
197.The evidence before the court shows that there was a massive amount of electronic data which needed to be reviewed in order to retrieve the Schedule 1 documents. The data pool was simply enormous. For that reason, I think it is unrealistic to expect that a manual “eyeball” review would be effective in retrieving all the documents as the process is bound to be vitiated by human errors. I accept that the defendants were entitled to come up with an alternative methodology. 198.In fact, the evidence before me shows, and I hold, that it was within the power of the defendants to devise a methodology which, if properly executed, could effectively retrieve all the Schedule 1 documents. (This is subject to a caveat which I will come back to shortly.) 199.I say this because I consider that the steps taken by the defendants in the April production, the December production and the 2023 production collectively were comprehensive and capable of retrieving all the Schedule 1 documents. More specifically, I find the following:
200.It is in fact the defendants’ own case that latest by the time the 2023 production was made, all the Schedule 1 documents were handed over to the plaintiff. 201.It is true that the plaintiff bears the burden of showing that it was within the power of the defendants to hand over all Schedule 1 documents. However, the plaintiff has no personal knowledge of how the documents were created and stored by the defendants. It is clear from Kao, Lee & Yip that it was for the defendants to find an appropriate means to obey the production order. Here, the defendants’ own evidence demonstrates that they had (in the end) adopted and executed a proper methodology which, in my view, was capable of retrieving all Schedule 1 documents. The plaintiff’s burden is therefore discharged. 202.Earlier, I refer to a caveat and it is this. The defendants argue that even with a properly devised methodology which is executed up to a reasonable standard, it is still impossible to guarantee that every single document is retrieved as there are bound to be mistakes and gaps in a large-scale production like the present one. I think the answer to this is that if that happens, as will be seen below, there may be no contempt found as any such omission may fall within the “accidental and unintentional” exception at stage 3 (the mens rea stage). In any event, as a matter of common sense, notwithstanding the potential gaps in the production, it is still correct to conclude, as demonstrated by the evidence, that it was within the power for the defendants to hand over all Schedule 1 documents. 203.I now turn to the timing aspect. I am satisfied that it was possible for the defendants to comply with the production order by the deadline of 28 April 2021. It is not necessary to make any finding as to the resources which the defendants could command for the production exercise. The objective fact is that they in fact had substantial assistance and support from Ernst & Young. Top-tier international law firms were engaged for the exercise. I have not been provided with precise figures of the total number of manhours put in and the total legal costs charged by the external law firms in the three rounds of production. But it is plain that Ernst & Young had devoted vast resources to the exercise. Insofar as it is now suggested that it was impossible to comply with the production order for want of resources, that is not tenable as being contrary to the objective fact as disclosed by the defendants’ own evidence. 204.As to timing, the short point here is that in the stay application, the defendants themselves asked for 56 days if their stay was refused, and it was remarked by the recorder that the defendants should have started the process since October 2020. This shows that had the defendants attempted to take all the steps which were taken in the three rounds of production in one go before 28 April 2021, it should have been within their power to hand over all Schedule 1 documents by the deadline. 205.I now turn to the seven heads of complaint raised by the plaintiff. 206.Under the keyword search issue, the plaintiff says that the use of the keyword search approach, as opposed to a manual review, is in principle objectionable as it would screen out responsive documents. I do not agree with this contention. I have already held above that a keyword search approach, if properly devised, should be capable of capturing all responsive documents. 207.Next, the plaintiff contends that the use of the 85 keywords in the April production was inherently defective as they were chosen with a view to identifying only the documents relevant to the defendants’ wrongdoing as alleged by the plaintiff. See the plaintiff’s objection set out in the correspondence, in particular JSCC’s letter dated 25 June 2021 (para 93(1) above). 208.Mr Maurellet submits, and I agree, that it is not open to the plaintiff to make this contention because it was not “pleaded”. The plaintiff’s contention is a positive and independent case which the defendants need to meet at trial. It should have been “pleaded” as a ground of contempt with proper particulars. That was not done. It would not be fair to allow the plaintiff to run this point in the trial. 209.The plaintiff further contends that the 85 keywords were self-evidently under-inclusive. I agree with that contention. The 85 keywords did not contain any Chinese character, did not include a number of legal proceedings relevant to the plaintiff’s liquidation, and did not include some of the relevant individuals, such as Ken Li’s former solicitors. The insufficiency of the 85 keywords is also apparent when compared with the 112 additional keywords: see Annex 2 and Annex 3. I also agree with Mr Wong’s submission that the very fact that 112 additional keywords were subsequently applied resulting in the production of new documents is indisputable proof that the initial list of 85 keywords was inadequate. 210.I therefore hold that by making use of merely the 85 keywords in the April production, the defendants were in breach of the production order in that the Schedule 1 documents not captured by those keywords were missed out. 211.Under the Lotus Notes backup issue, the plaintiff says that the defendants should have searched the Lotus Notes backup emails of the 36 former employees for the April production but did not do so. This resulted in the failure to capture those Schedule 1 documents which were stored in these backup emails but not elsewhere. 212.This is a valid complaint. In fact, in cross-examination, Yen accepted that such a search should have been done. As I remarked at the trial, it was an obvious hole in the production exercise. The search was subsequently done in the December production. I accept the plaintiff’s contention that there were new documents disclosed in that round (for which, see paras 114, 117 to 119 above). 213.I therefore hold that by omitting to search the Lotus Notes backup emails of the 36 former employees in the April production, the defendants were in breach of the production order in that those emails found only in such backup data were missed out from the April production. 214.Under the 15 custodians issues, the plaintiff says that the document hold process was deficient as it was solely dependent on the custodians’ own recollection and confirmation as to whether they still retained Schedule 1 documents. As the 15 existing employees had entered time on the file, the defendants should have caused a search on the laptop computers of all of them and should not have confined the search to the six who responded positively to the document preservation notice. 215.I have held above that the document hold process was in principle an effective means to identify responsive documents. 216.However, according to Huen’s evidence, in the December production, the search on the laptops of two additional existing employees (who each recorded more than 50 hours of time on the file) yielded a discovery of about 90 documents which were considered Schedule 1 documents. 217.For that reason, I accept that the defendants were in breach of the production order by omitting to produce those 90 documents in the April production. However, I should add at the same time that the 90 documents comprised only a very small number when compared with the 49 boxes of documents which were delivered in the April production. The Microsoft Exchange server issue 218.Under the Microsoft Exchange server issue, the plaintiff complains that the defendants failed to search the server with respect to the emails of the existing employees for the April production. In the end, the server was searched in the 2023 production which led to the production of 541 emails, some of which contain new contents which had not been previously disclosed (for which, see paras 140 and 141 above). 219.I agree that the 2023 production is indisputable proof that the defendants failed to produce all Schedule 1 documents. I hold that the defendants were in breach of the production order in that the documents found only on the Microsoft Exchange server were missed out from the April production. 220.The corrupted documents issue is in essence the same as the Microsoft Exchange server issue. This complaint is therefore made out. The Category C documents issue 221.Under the Category C documents issue, the plaintiff contends that the defendants failed to hand over emails and documents of the support and administrative staff in the April production. 222.In the December production, it is common ground (as shown in the joint table submitted by the parties) that 54 documents were handed over which are described as emails “of an administrative nature (e.g. room booking, scheduling, delivery, printing and record keeping matters)”. The plaintiff submits, and I agree, that this is indisputable proof that the April production was inadequate in that these documents were missing. 223.I therefore hold that the defendants were in breach of the production order by failing to hand over these Category C documents in the April production. The missing POD summaries issue 224.Under the missing POD summaries issue, the plaintiff contends that the defendants failed to hand over the adjudication summaries relating to five proofs of debts, which are numbered 45, 46, 49, 50 and 51, and, remarkably, these proofs were submitted by Ken Li and his associates, which coincided with the “spurious proofs of debts” referred to in the removal decision and criticised by the judge as showing the defendants’ bias in favour of Ken Li: see para 51 above. 225.The plaintiff refers to the fact that in the course of the liquidation, the defendants had devised standard form adjudication summaries of the plaintiff’s proofs of debts. The summaries would record the details of, among other things, the supporting documents, the actions taken by the defendants, their analysis of the evidence, and the internal recommendation on what action should be taken. The plaintiff points to the fact that the adjudication summaries existed for all other proofs of debts, including for a claim of as little as $4,211.50. Hence, the adjudication summaries for the five proofs of debts must have existed. 226.The defendants’ response is that the alleged summaries did not exist. In his affidavit evidence, Yen explained that preparation of adjudication summaries was not a requirement, both as a matter of law and as a matter of practice. As the five proofs of debts were more complicated and controversial in nature, the reasons and records for the adjudication were addressed in correspondence rather than in standalone adjudication summaries. 227.In cross-examination, Yen added that they had searched all the servers but nothing was found and that should show that no such adjudication summaries had been prepared in the first place. He denied that these documents were concealed by the defendants. 228.This ground of contempt turns on the factual issue of whether the adjudication summaries for the five proofs of debts existed in the first place. If they did, then there was a breach of the production order. 229.In my judgment, the evidence before me is not sufficient to support a finding on the criminal standard that the adjudication summaries existed. The only evidence in support is the fact that similar summaries had been prepared for all other proofs of debts and the only ones missing remarkably related to those proofs submitted by Ken Li and his camp. It is true that one may legitimately argue that the inherent likelihood is that adjudication summaries had been prepared for all proofs of debts, and that there was prima facie a motive on the part of the defendants to hide them because the documents might be prejudicial to them in the breach action. But that is really the extent of the plaintiff’s evidence and argument. 230.On the other hand, the objective fact is that by the time of the April production, the defendants had already produced other potentially prejudicial documents. This would seem to undermine the force of the motive point. The further objective fact is that the April production was overseen by Ernst & Young’s in-house teams, with the assistance of outside lawyers. This would go to undermine the likelihood that the defendants could have concealed the documents. 231.I therefore find that the plaintiff has failed to show that the adjudication summaries for the five proofs of debts existed in the first place. This ground of contempt is not made out. 232.In summary, as regard the stage 2 question, I find that the defendants were in breach of the production order. The plaintiff has successfully made out the breaches under all the heads of complaint save for the missing POD summaries issue. 233.As the defendants failed to hand over all the Schedule 1 documents by 28 April 2021, the confirmatory affidavits, which were to the contrary effect, were false and the defendants were therefore in breach of the affidavit order. 234.At stage 3, the question is whether the breaches, as found, were accompanied by the requisite state of mind to establish contempt. 235.There are fundamental differences between the parties on the legal question of what qualifies as the requisite state of mind in the context of an order to produce documents. 236.On the law, I am in general agreement with the submissions made by Mr Wong. The guiding principles are to be found in Kao, Lee & Yip. In addition, I consider that the Hong Kong decision in Lo Po Wai, Harry v Wong Yee Chuk [2024] HKCFI 724 and the English decision in Bird v Hadkinson [1999] BPIR 653 provide guidance on the issue. I reject the formulation put forward by Mr Maurellet on the requisite mental state. His reliance on the Court of Appeal decision in Ip Pui Lam Arthur and Ip Pui Sum v Alan Chung Wah Tang and Alison Wong Lee Fung Ying CACV 214/2016, 16 February 2017 and the first instance decision in Alan Tang is, in my view, misplaced. 237.In civil contempt, the mental element is proved if it is shown that the defendant knew the facts which are said to make his act or omission a breach of the court order in question. It is not necessary to prove that he appreciated that it did breach the order. It is not necessary to show that there was a direct intention to disobey the order or otherwise to interfere with the administration of justice: Kao, Lee & Yip at paras 43 to 45. 238.The rationale was explained by Lord Wilberforce in Heatons Transport (St Helens) Ltd v Transport and General Workers’ Union [1973] AC 15 at 109E, quoted in Kao, Lee & Yip at para 44:
239.The “casual or accidental and unintentional” exception exists because no injustice is done when the breach of the court order is occasioned by such a cause: Kao, Lee & Yip at para 53. On a general level, there is no good reason, whether as a matter of principle, policy or justice and fairness, for punishing a defendant for breaching an order either casually or accidentally and unintentionally. It has been said that it would be an abuse of language to say that a man who honestly tried to obey an order but by mistake was in fact in breach of it, is showing “contempt”: Adam Phones Ltd v Goldschmidt [1999] 4 All ER 486 at 494. 240.Whether the “casual or accidental and intentional” exception applies must depend on the facts of each case. Adam Phones is often cited as an example which falls within the “unintentional” exception. In Hong Kong, ZQA v SCC [2021] 3 HKC 458, [2021] HKCA 194 is a recent decision of the Court of Appeal in which the “unintentional” exception was upheld: para 5.22. While Mr Wong and Mr Maurellet have each discussed this case at some length in their submissions, I do not find it necessary to do so here, as the facts there are very different. 241.Here, we are concerned with a mandatory order requiring a defendant to achieve a certain state of affairs. As remarked above, it is for the defendant to find out the proper means or devise the proper methodology to achieve the result. If the methodology come up with by the defendant was ineffective in delivering the result mandated by the order, that would be a breach. Actus reus would be established. On mens rea, two material questions need to be answered:
242.As regards the first question, the answer, again, is to be found in Kao, Lee & Yip. If the defendant consciously and voluntarily adopted the methodology, mens rea is established. 243.I recite the ruling in Kao, Lee & Yip at para 54:
244.As regards the second question, I think the answer is that there is a prerequisite which the defendant needs to satisfy before he can invoke the “accidental and unintentional” exception. Mr Wong submits that the defendant needs to demonstrate that he had taken all the steps which were reasonably required of him in the circumstances when devising his methodology and he had executed it with reasonable care. It is only after that is shown that he should be allowed to argue that he should not be committed for contempt because he had in good faith believed that what he had done was good enough to obey the order (when it was not). 245.I agree with that submission. I think it is correct as a matter of principle and is supported by the authorities. (I should however make it clear that where the prerequisite is met, whether the exception can be invoked still turns on the specific facts of the case.) 246.The rationale of the accidental exception, as we have seen above, is that, conceptually speaking, it is just not right to commit a person to prison for breaching an order accidentally and unintentionally. However, there will be cases in which the defendant genuinely thought he was obeying an order but his efforts in complying in fact fell short of what a reasonable person would do in his circumstances. In these cases, contempt should be found as otherwise the usefulness of the committal process as a tool to enforce orders and uphold the authority of the court would be much diminished. This is therefore a good policy reason for imposing a prerequisite on a defendant who wishes to rely on the exception. 247.On authorities, I find support in Kao, Lee & Yip, Bird v Hadkinson and Lo Po Wai. 248.The above proposition is consistent with what the Court of Final Appeal said in Kao, Lee & Yip. It was held, at para 55, that the plea that the defendant “did his best” did not lead to the conclusion that the breach was “casual or unintentional or accidental”. It was further held, at para 63, that a mistake about the legitimacy of action to comply with a court order is no excuse for non-compliance. If it were otherwise, obedience to court orders would be problematic as it would be dependent on the correctness of legal advice about what the defendant might lawfully do to obey it. In my view, what was said there supports the proposition that whether the exception applies does not turn solely on the subjective state of mind of the defendant, and that there is an objective standard which he needs to meet. 249.Next, Bird v Hadkinson. This case concerned a disclosure order made in aid of a Mareva injunction. The respondent there was ordered to answer questions about the movement of some funds. Neuberger J (as he then was) addressed the question of whether a disclosure order requiring the respondent to give information would be satisfied by merely an honest answer as to belief or whether it would require an accurate answer: 657E-F. Even though the content of the disclosure order there is different from the production order before me, one can immediately see the parallel between the two scenarios. And I think the answer and reasoning given by his Lordship applies equally to the present case. 250.The answer was that if the respondent gives an inaccurate answer, that is not in compliance with the order. However, if the inaccurate answer is given in good faith, and is given after investigating the matter in a way which is reasonable in all the circumstances, either there is contempt which is of a most technical nature or there is no contempt:
251.The last authority is Lo Po Wai, a case relied on by Mr Wong. That case concerned an order that the plaintiff be allowed to inspect categories of documents specified in the order. It was held that two documents, which were attachments to an email and referred to in the decision as “the DY-2018 and DY-2019 Reports”, were within the scope of the order and the defendant had failed to disclose them. The defendant might not have actual contemporaneous knowledge of their existence. Notwithstanding that, the “accidental and unintentional” exception did not apply. This was because it had been within his power to discover the documents by checking all the emails “carefully and diligently” but he had failed to do so: paras 71(h) to (i). 252.K Yeung J ruled that the failure to “carefully and diligently” check the emails amounted to the requisite state of mind to find contempt, and the exception did not apply, at para 71(k):
253.As for the defendants, it is submitted by Mr Maurellet that in the context of a document production order, the requisite mental element is an intention not to produce the document. He relies on Ip Pui Lam (at para 4.2) and Alan Tang (at para 25.4) in support. In both cases, it was stated that the burden is on the plaintiff to show beyond reasonable doubt that:
254.Mr Maurellet then submits that where there are gaps in the production, when determining mens rea, the correct question to ask is whether those gaps were consciously excluded as opposed to merely being accidental, taking into account what was within the defendant’s knowledge and power. Applying that to the present case, the pertinent question is whether the defendants knew of and intended the omissions rendering the methodology in breach of the production order. For instance, under the keyword search issue, the question is – was there a conscious decision to exclude keywords which the defendants knew would identify additional documents? If not, then mens rea is not established. 255.I do not think that Mr Maurellet’s formulation represents the correct test. First, his formulation cannot be found in Ip Pui Lam and Alan Tang. The two authorities simply contain the one-line statement without any elaboration. Second, as a matter of principle, his formulation plainly goes contrary to Kao, Lee & Yip. It is plain to me that the one-line statement was meant by the courts in the two decisions to be applied in accordance with the general principles explained in Kao, Lee & Yip. 256.I should, however, add that in addition to the above formulation, Mr Maurellet also makes the submission that the defendants had taken all reasonable steps within their knowledge and control to comply with the orders. They formulated an extensive methodology developed with input from legal, forensics and IT exerts. After applying the keywords to the original data pool, they deployed a large team to manually review tens of thousands of documents. Any gaps in the April production should fall squarely within the accidental exception. 257.This submission is, in principle, in line with Mr Wong’s submission which I have accepted above, namely that the defendants should demonstrate that they had taken all reasonable steps to comply as a prerequisite before invoking the accidental exception. 258.In my judgment, mens rea is established for the breaches under:
259.This is because in the April production, the defendants consciously and voluntarily adopted the methodology which made use of the 85 original keywords only, and focused on the search of the emails of the six existing employees, without addressing their minds to the 36 former employees. The “accidental and unintentional” exception does not apply because it was unreasonable for the defendants to confine the search to the 85 keywords and not to consider searching the emails of the former employees. (Yen himself accepted at the trial that the former employees should have been considered.) In the circumstances, they should not be allowed to take advantage of the “accidental and unintentional” exception even though it appears to be the case that they honestly considered that they had done enough to comply with the order at the time.[1] 260.On the other hand, I am not satisfied that mens rea is established for breaches under:
261.As regards (1), it was a conscious and voluntary decision on the part of the defendants not to search the laptop computers of the nine existing employees who gave a negative response to the document preservation notice. As explained above, I consider that the document hold process was an effective means to identify fee earners who were still in possession of responsive documents. It was reasonable for the defendants to rely on the reply of the nine existing employees and, based on that, not to search their laptop computers in the April production. It turns out that two of them in fact did hold a very small number of responsive documents. There was therefore a breach but the breach was accidental and unintentional and should not be punishable by contempt. 262.As regards (2) and (3), similarly, the defendants made a conscious and voluntary decision to conduct the search on the Microsoft Outlook emails on the laptop computers of the existing employees, without at the same time looking into the Microsoft Exchange server. On the other hand, I have found above that Yen was not aware of any difference between the data on the server and the data on the individual laptop computers of the fee earners at the time. In light of that, it would not be reasonable to expect him to cause a search on the Microsoft Exchange server. I hold that while the failure to search resulted in 541 emails being omitted from the April production and was therefore a breach of the production order, the breach was accidental and unintentional in the circumstances.[2] 263.The last resort principle is well established. 264.Contempt proceedings should only be pursued as a last resort. In cases of civil contempt, the prime consideration is the enforcement of the court order in question. While there are cases where it is necessary (and indeed imperative) that the court should exercise its power in the punishment of contempt to achieve compliance, there are other cases where lesser options should be explored before one is to resort to the draconian power of committal. In cases where it can be demonstrated that an alleged contemnor has been making genuine and serious efforts in complying with the order and non-compliance by a certain deadline is due to circumstances beyond his control, it would be rare that the court will deem it appropriate to consider the option of committal. It is wrong to bring contempt proceedings without regard to the efforts and continued efforts by the alleged contemnor even though he has failed to completely fulfil its requirement within the time prescribed: China Metal at paras 60 and 78. 265.It is submitted on behalf of the defendants that it was not necessary for the plaintiff to apply for committal as other options were available. Mr Maurellet highlights the objective fact that despite its deficiencies (as found by the court), the April production was a huge production. After it was made, the defendants made clear their willingness to meet the plaintiff’s concerns through retrieving documents from further sources and applying additional keywords. On this, see the correspondence between the parties’ solicitors from June 2021 onwards, as quoted in paras 94 to 99 above. Mr Maurellet submits that the correspondence is clear proof that the defendants were willing to perform whatever additional steps or procedures that could be reasonably agreed or otherwise directed. It was not necessary to bring the committal application. 266.On the other hand, Mr Wong contends that the committal process was indeed appropriate and necessary to secure full compliance of the production order. It is notable, he submits, that in the post-April production correspondence, the defendants maintained (and in fact still maintain) that there was no breach, and the additional steps were proposed without prejudice to that position. 267.Further, the background of the case is highly relevant. From the plaintiff’s perspective, the documents should have been returned promptly after the removal decision in late 2017. But that did not happen. In the production decision, DHCJ Le Pichon was highly critical of the conduct of the defendants in the return process (the documents being returned “in dribs and drabs”, and the defendants having conducted themselves “disgracefully”). 268.The frustration of the plaintiff is captured by Seline Li’s oral testimony. Given the above background, the frustration is well understandable:
269.In my view, in light of the background of the case, upon receiving the documents in the April production which were still incomplete and suffered from a number of apparent deficiencies, it was appropriate for the plaintiff to consider committal. It is true that the defendants made clear that they were willing to address the plaintiff’s concerns. In theory, the plaintiff had an option to wait and see whether the defendants would rectify the problem. However, given everything that had happened, in my view, it was not a reasonable option which should be imposed on the plaintiff. In the end, the objective fact is that it was only after the commencement of the committal application and further complaints made in affirmations filed in these proceedings that the December production and the 2023 production took place. 270.I therefore find that the committal application has been properly brought under the last doctrine principle. 271.The defendants submit that if the court finds that they are in contempt, the contempt has by now been purged by the December production and the 2023 production. 272.I accept that submission on the evidence. I refer to the discussion in para 199 above. The steps taken in the three rounds of production, when viewed collectively, were extensive, elaborate, and capable of retrieving all Schedule 1 documents. 273.The plaintiff says that the two subsequent productions are still not good enough for the reasons set out in paras 128 to 130 and 152 above. 274.I do not consider that any further search based on yet more keywords (especially generic ones) or a wholesale search of the large data pool (as opposed to sample reviews) may in fact be counter-productive as they are bound to generate even more false positives. I also find that the steps taken by the in-house legal team as set out in para 105(3) to be an effective means to confirm the existence of responsive documents. 275.In the end, the two subsequent productions generated only a very small number of new documents, when compared to the overall size of the April production. 276.I am satisfied that the defendants have by July 2023 handed over to the plaintiff all the Schedule 1 documents. The contempt has been purged. The next question is what effect the purge has on the outcome of the committal application. 277.At the trial, there was a dispute as to whether the purge goes to mitigation only or it may be taken into account at the liability stage as well. 278.Mr Wong submitted that the issue of purge should not affect liability. He relied on Alan Tang in which Cheng J remarked, at paras 57 and 66, that any further consideration of the question of purging should be left for consideration at the stage of the determination of the proper penalty for contempt. On the other hand, Mr Maurellet argued that the court has a discretion at the liability stage to take into account a purge of contempt, and decide the issue of liability accordingly. 279.I agree with Mr Maurellet’s submission, which I consider to be sound as a matter of logic and correct as a matter of authorities. 280.There is no good policy reason why a mechanistic approach, as contended for by Mr Wong, should apply such that once the three-stage questions are answered in the positive, a committal order should automatically follow. The court should retain a general discretion to decide whether the order is appropriate, giving effect to the general objectives of ensuring compliance of court orders and upholding the authority of the court with reference to the individual circumstances of each case. 281.On authorities, I do not think that Alan Tang can be regarded as authority for the proposition that the issue of purge only goes to sentencing. There was no such discussion in that case, and her Ladyship was simply making a statement that that was what she would do in that case. 282.On the other hand, the Court of Appeal decision in Re Mahesh Roy [2017] 5 HKLRD 830 shows that the issue of purge can be considered at the liability stage. The case concerned an order to summarily punish a person for contempt under section 42 of the Labour Tribunal Ordinance, Cap 25. The conviction was set aside on appeal. One of the grounds was that the contempt had been purged by the defendant’s apology, and it was held that it was wrong in principle for the defendant to have been committed in the circumstances. Lam VP (as he then was) held, at para 47:
283.Lastly, I also draw support from Atkin’s Court Forms (2nd ed) Vol 10(2) (2021 Issue). In Form 47, the template contains wording to the effect that while the court is of the opinion that the defendant has committed a breach of a court order and is therefore in contempt of court, the court makes no order on the application given that the contempt has been purged. 284.After the trial, both parties have written to the court confirming their agreement that the court does have the power and discretion to grant the order as set out in Atkin’s Court Forms. 285.In the present case, the three-stage elements of contempt have been established. I am also satisfied that it was necessary and appropriate for the plaintiff to bring the committal application and there is no contravention of the last resort principle. The defendants are therefore in contempt of court. 286.But, as found, the contempt has by now been purged. I consider that there is no need for the court to make any order now in order to coerce compliance with the original orders. Further, this is not a case where the defendants deliberately disobeyed the orders at the first instance. Vast resources were subsequently devoted to purge the contempt. For these reasons, I do not see fit to consider imposing any penalty on the defendants save for on costs. I therefore would exercise my discretion not to make any order on the committal application. ORDERS 287.In conclusion, I am of the opinion that the defendants were in breach of the production order as made out under (1) the keyword search issue, (2) the Lotus Notes backup issue, and (3) the Category C documents issue. I am of the further opinion that they were in breach of the affidavit order. They are accordingly in contempt of court. 288.However, the contempt has been purged through the December production and the 2023 production. 289.In the circumstances, I make no order on the amended originating summons. 290.On costs, I make an order nisi that the defendants do pay the plaintiff’s costs of the amended originating summons, including all reserved costs, to be taxed on an indemnity basis if not agreed, with a certificate for two counsel.
Mr Wong Yan Lung, SC and Mr Martin Kok, instructed by Joseph S.C. Chan & Co., for the plaintiff Mr José-Antonio Maurellet, SC and Mr John Hui, instructed by Davis Polk & Wardwell, for the 1st and 2nd defendants The 55 keywords
The 85 keywords
[The remaining part of the page is deleted] The 112 additional keywords
[1] If Yen had indeed requested the legal team to search the Lotus Notes backup emails of the former employees, he might be able to invoke the “unintentional” exception as he had to rely on the legal team to carry out the search but was told that no backup could be retrieved. These (hypothetical) facts would bring this case closer to the facts in ZQA in which the “unintentional” exception was upheld. [2] It may be said that this factual scenario is analogous to the one in Adam Phones: see the discussion of the case in para 48 of Kao, Lee & Yip. | |||||||||||||||||||||||||
Cases cited in this judgment